Sei sulla pagina 1di 2211

WILEY CRITICAL CONTENT

PETROLEUM
TECHNOLOGY
VOLUME 1-2

WILEY-INTERSCIENCE
WILEY-VCH
A JOHN WILEY & SONS, INC., PUBLICATION
Copyright # 2007 by John Wiley & Sons, Inc., and WILEY-VCH Verlag GmbH & Co. KGaA,
Weinheim. All rights reserved.

Published by John Wiley & Sons, Inc., Hoboken, New Jersey.


Published simultaneously in Canada.

No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any
form or by any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except
as permitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the
prior written permission of the Publisher, or authorization through payment of the appropriate
per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923,
(978) 750-8400, fax (978) 750-4470, or on the web at www.copyright.com. Requests to the Publisher
for permission should be addressed to the Permissions Department, John Wiley & Sons, Inc.,
111 River Street, Hoboken, NJ 07030, (201) 748-6011, fax (201) 748-6008, or online
at http://www.wiley.com/go/permission.

Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts
in preparing this book, they make no representations or warranties with respect to the accuracy or
completeness of the contents of this book and specifically disclaim any implied warranties of
merchantability or fitness for a particular purpose. No warranty may be created or extended by sales
representatives or written sales materials. The advice and strategies contained herein may not be
suitable for your situation. You should consult with a professional where appropriate. Neither the
publisher nor author shall be liable for any loss of profit or any other commercial damages, including
but not limited to special, incidental, consequential, or other damages.

For general information on our other products and services or for technical support, please contact our
Customer Care Department within the United States at (800) 762-2974, outside the United States at
(317) 572-3993 or fax (317) 572-4002.

Wiley also publishes its books in a variety of electronic formats. Some content that appears in print
may not be available in electronic formats. For more information about Wiley products, visit our web
site at www.wiley.com.

Library of Congress Cataloging-in-Publication Data is available.

Wiley Critical Content: Petroleum Technology, 2V


ISBN 978-0-470-13402-3

Printed in the United States of America

10 9 8 7 6 5 4 3 2 1
CONTENTS VOLUME 1

PART I: EXPLORATION, PRODUCTION, AND REFINING

Petroleum, Introduction 1 Catalysis 499


Exploration, Drilling, Catalyst Deactivation
and Production Engineering 43 and Regeneration 553
Enhanced Oil Recovery 167
Distillation 621
Oil Shale 194
Distillation, Azeotropic
Tar Sands 222 and Extractive 667
Drilling Fluids 250
Separations Process
Petroleum Refinery
Synthesis 734
Processes 291
Oil Refining, Environmental Liquid–Liquid Extraction 776
Considerations 334 Pipelines 851
Natural Gas 355
Sulfur and Hydrogen Sulfide
BTX Processing 427 Recovery 872
Fluid Catalytic Cracking,
Corrosion and Corrosion
Units, Regeneration 442
Control 896
Fluid Catalytic Cracking,
Catalysts and Additives 477 Bioremediation 920

v
CONTENTS VOLUME 2

PART II: REFINED PRODUCTS AND FUELS

Liquefied Petroleum Gas 1 Fuels, Synthetic, Liquid 329


Lubrication and Lubricants 14 Fuels, Synthetic, Gaseous 377
Waxes 85
Octane Enhancers 406
Asphalt 169
Petroleum Coke 205 Methyl Tert-Butyl Ether 428

Gasoline and Other Motor Fuels 223 Combustion Science


and Technology 449
Aviation Turbine Fuels 273
Heating Oil 293 Emission Control, Automotive 493

Naphthenic Acids 317 Emission Control, Industrial 529

PART III: PETROCHEMICALS

Petrochemical Feedstocks 576 Ethylbenzene 974


Hydrocarbons 591 Cyclopentadiene and
Dicyclopentadiene 991
Acetylene 652
Naphthalene and
Ethylene 727
Hydronaphthalenes 1007
Propylene 799
Anthracene 1020
Butadiene 820
Olefins, Higher 1028
Isoprene 849
Acylation and
Cyclohexane 876
Alkylation 1047
Benzene 885
Synthetic Organic Chemicals,
Toluene 914 Economic Evaluation 1114

Xylenes 944 Index 1141

vi
PREFACE

Wiley’s major reference works cover a multitude of subject disciplines and have
earned many accolades over the years for their well-structured, accessible arti-
cles, covering both fundamentals and cutting-edge topics. The Wiley Critical
Content Series was conceived in response to demand from customers who
have a strong interest in certain key topics covered within more than one of
Wiley’s reference works and who have asked to see the most relevant articles
brought together in one place. Wiley Critical Content: Petroleum Technology
is the first title to be published in this series. It features articles from two
of Wiley’s most prestigious encyclopedias—the Kirk-Othmer Encyclopedia of
Chemical Technology and Ullmann’s Encyclopedia of Industrial Chemistry.
Wiley Critical Content: Petroleum Technology provides comprehensive, up-
to-date information on all aspects of petroleum: properties and origin, explora-
tion, production and refining processes, economic issues, and environmental
and health concerns. The two volumes contain over fifty articles, organized
thematically:

- Part I: Exploration, Production, and Refining


- Part II: Refined Products and Fuels
- Part III: Petrochemicals

Written by prominent scholars from around the world, this reliable reference
will be of particular interest to petroleum engineers, researchers in industry
and academia, and other professionals and consultants in petroleum-related
industries.

vii
CONTRIBUTORS

Colin Barker, University of Tulsa, Tulsa, M. Larry Campbell, Exxon Chemical Company,
Oklahoma, United States, Petroleum, Florham Park, New Jersey, United States,
Introduction Cyclohexane
Scott D. Barnicki, Eastman Chemical Martin Carrera, Amoco Corporation,
Company, Kingsport, Tennessee, United Naperville, Illinois, United States, Propylene
States, Separations Process Synthesis Clarence D. Chang, Mobil Research and
Calvin Bartholomew, Brigham Young Development Corporation, Princeton,
University, Provo, Utah, United States, New Jersey, United States, Fuels, Synthetic,
Catalyst Deactivation and Regeneration Liquid
Jürgen Bässler, Uhde GmbH, Dortmund, W. C. Cheng, W. R. Grace & Company,
Germany, Acetylene Columbia, Maryland, United States, Fluid
Arno Behr, Henkel KGaA, Düsseldorf, Catalytic Cracking, Catalysts and Additives
Germany, Hydrocarbons T. T. Peter Cheung, Phillips Petroleum
Hartmut Behringer, Hoechst Company, Bartlesville, Oklahoma, United
Aktiengesellschaft, Werk Knapsack, States, Cyclopentadiene and
Germany, Acetylene Dicyclopentadiene
L. E. Bell, UOP LLC, Des Plaines, Illinois, P. D. Clark, University of Calgary, Calgary,
United States, Fluid Catalytic Cracking, Units, Alberta, Canada, Sulfur and Hydrogen Sulfide
Regeneration Recovery
Rosemarie Berger, Ludwigshafen, Germany, Gerd Collin, DECHEMA e.V., Frankfurt am
Liquid–Liquid Extraction Main, Germany, Anthracene; Hydrocarbons;
Ronald L. Berglund, Terracon, Kingwood, Naphthalene and Hydronaphthalenes
Texas, United States, Emission Control, Axel Commichau, Mobil Europe Gas Inc., The
Industrial Hague, The Netherlands, Natural Gas
Dieter Biedenkapp, BASF Aktiengesellschaft, Werner Dabelstein, Deutsche Shell AG,
Ludwigshafen, Germany, Hydrocarbons Hamburg, Germany, Octane Enhancers
Geoff J. Bishop, Shell Aviation Ltd., London, Rula Deeb, Malcolm Pirnie, Inc., Emeryville,
United Kingdom, Aviation Turbine Fuels California, United States, Methyl Tert-Butyl
Eckhart Blass, Technische Universität Ether
München, München, Germany, Liquid–Liquid James R. Dickey, Lubricants Consultant,
Extraction Basking Ridge, New Jersey, United States,
Klaus Boehlke, BASF Aktiengesellschaft, Lubrication and Lubricants
Ludwigshafen, Germany, Waxes Wolfram Dietsche, BASF Aktiengesellschaft,
John T. Boepple, Nexant ChemSystems, White Ludwigshafen, Germany, Waxes
Plains, New York, United States, Michael F. Doherty, University of California,
Petrochemical Feedstocks Santa Barbara, Santa Barbara, California,
John K. Borchardt, Shell Development United States, Distillation, Azeotropic and
Company, Houston, Texas, United States, Extractive
Enhanced Oil Recovery T. J. Dougan, W. R. Grace & Company,
James A. Brient, Merichem Company, Columbia, Maryland, United States, Fluid
Houston, Texas, United States, Naphthenic Catalytic Cracking, Catalysts and Additives
Acids Mary Noon Doyle, Shepherd Chemical
P. F. Bryan, Chevron Research and Technology Company, Cincinnati, Ohio, United States,
Company, Richmond, California, United Naphthenic Acids
States, BTX Processing Lawrence J. Drew, U.S. Geological Survey,
Klaus Buckl, Linde AG, Höllriegelskreuth, Reston, Virginia, United States, Petroleum,
Germany, Acetylene Introduction
Narasimhan Calamur, Amoco Corporation, Günter Ebersberg, Degussa-Hüls AG, Marl,
Naperville, Illinois, United States, Propylene Germany, Acetylene

ix
x CONTRIBUTORS Vol. 1

Jörg Fabri, Deutsche Shell AG, Hamburg, Gerd Hohner, Hoechst Aktiengesellschaft, Werk
Germany, Octane Enhancers; Xylenes Gersthofen, Augsburg, Germany, Waxes
James R. Fair, The University of Texas at Hartmut Höke, Weinheim, Germany,
Austin, Austin, Texas, United States, Anthracene; Hydrocarbons; Naphthalene
Distillation and Hydronaphthalenes
Kevin J. Fallon, Washington Group Craig A. Hoyme, Eastman Chemical Company,
International, Inc., Cambridge, Kingsport, Tennessee, United States,
Massachusetts, United States, Ethylbenzene Separations Process Synthesis
William Fruscella, Unocal Corporation, El Chang Samuel Hsu, Exxon Research and
Segundo, California, United States, Benzene Engineering Company, Annandale, New
Dorothea Garbe, Haarmann & Reimer GmbH, Jersey, United States, Petroleum, Introduction
Holzminden, Germany, Hydrocarbons Walther W. Irion, Deutsche Shell AG,
Bruce C. Gates, University of California, Hamburg, Germany, Oil Refining,
Davis, Davis, California, United States, Environmental Considerations
Catalysis Michael Kavanaugh, Malcolm Pirnie, Inc.,
Eugen Gehrer, BASF Aktiengesellschaft, Emeryville, California, United States, Methyl
Ludwigshafen, Germany, Acylation and Tert-Butyl Ether
Alkylation Roland Kettner, Mobil Erdgas–Erdöl GmbH,
Heinz-Peter Gelbke, BASF Aktiengesellschaft, Celle, Germany, Natural Gas
Ludwigshafen, Germany, Ethylbenzene Jeffrey P. Knapp, E. I. du Pont de Nemours
Ulrich Graeser, Veba Öl AG, Gelsenkirchen, & Company, Wilmington, Delaware, United
Germany, Xylenes States, Distillation, Azeotropic and Extractive
Helmut Greim, Institut für Toxikologie und Christian Küchen, Institut für Wirtschaftliche
Umwelthygiene, Technische Universität Oelheizung, Hamburg, Germany, Heating Oil
München, Freising-Weihenstephan, Germany, G. R. Lappin, Albemarle Corporation, Baton
Naphthalene and Hydronaphthalenes Rouge, Louisiana, United States, Olefins,
Karl Griesbaum, Universität Karlsruhe (TH), Higher
Karlsruhe, Germany, Hydrocarbons Maryline Laugier, Malcolm Pirnie, Inc.,
Baruch Grinbaum, IMI (TAMI) Institute for Emeryville, California, United States, Methyl
Research and Development, DSBG, Haifa Bay, Tert-Butyl Ether
Israel, Drilling Fluids George M. Long, Institute of Gas Technology,
Georg Hammer, Mobil Erdgas–Erdöl GmbH, Chicago, Illinois, United States, Pipelines
Celle, Germany, Natural Gas Eckhard Loser, Bayer AG, Wuppertal,
Scott Han, Mobil Research and Development Germany, Isoprene
Corporation, Princeton, New Jersey, United Torsten Lübcke, Mobil Erdgas–Erdöl GmbH,
States, Fuels, Synthetic, Liquid Celle, Germany, Natural Gas
R. H. Harding, W. R. Grace & Company, Dieter Mayer, Hoechst Aktiengesellschaft,
Columbia, Maryland, United States, Fluid Frankfurt, Germany, Acetylene; Hydrocarbons
Catalytic Cracking, Catalysts and Additives Helmut Meinass, Linde AG, Höllriegelskreuth,
B. W. Hedrick, UOP LLC, Des Plaines, Illinois, Germany, Acetylene
United States, Fluid Catalytic Cracking, Units, Georg Michalczyk, DEA Mineralöl AG,
Regeneration Hamburg, Germany, Waxes
Werner Hefner, BASF Aktiengesellschaft, John J. Mooney, Environmental and Energy
Ludwigshafen, Germany, Acetylene Technology and Policy Institute, Wyckoff,
Franz-Leo Heinrichs, Hoechst New Jersey, United States, Emission Control,
Aktiengesellschaft, Werk Gersthofen, Automotive
Augsburg, Germany, Waxes Patrick Moran, U.S. Naval Academy,
Robert N. Heistand, Consultant, Englewood, Annapolis, Maryland, United States,
Colorado, United States, Oil Shale Corrosion and Corrosion Control
C. L. Hemler, UOP LLC, Des Plaines, Illinois, Sanford L. Moskowitz, American Economics
United States, Fluid Catalytic Cracking, Units, Group and Chemical Heritage Foundation,
Regeneration Abington, Pennsylvania, United States,
Albert M. Hochhauser, ExxonMobil Research Synthetic Organic Chemicals, Economic
and Engineering Company, Paulsboro, New Evaluation
Jersey, United States, Gasoline and Other Eckart Müller, Frankfurt am Main, Germany,
Motor Fuels Liquid–Liquid Extraction
Vol. 1 CONTRIBUTORS xi

Richard Müller, Degussa-Hüls AG, Marl, W. J. Reagan, UOP LLC, Des Plaines,
Germany, Acetylene Illinois, United States, Fluid Catalytic
Thomas Narbeshuber, BASF Cracking, Units, Regeneration
Aktiengesellschaft, Ludwigshafen, Germany, Herta Recknagel, Ruhrgas AG, Essen,
Acylation and Alkylation Germany, Natural Gas
Paul Natishan, Naval Research Laboratory, Arno Reglitzky, Deutsche Shell AG, Hamburg,
Washington, DC, United States, Germany, Octane Enhancers
Corrosion and Corrosion Control J. R. Riley, W. R. Grace & Company,
J. R. D. Nee, W. R. Grace & Company, Columbia, Maryland, United States,
Columbia, Maryland, United States, Fluid Fluid Catalytic Cracking, Catalysts and
Catalytic Cracking, Catalysts and Additives Additives
L. H. Nemec, Albemarle Corporation, Baton Heinrich Rischmüller, Technische Universität
Rouge, Louisiana, United States, Olefins, Clausthal, Clausthal-Zellerfeld, Germany,
Higher Exploration, Drilling, and Production
Hans-Joachim Neumann, German Petroleum Engineering
Institute, Clausthal-Zellerfeld, Germany, Winston K. Robbins, Exxon Research and
Natural Gas Engineering Company, Annandale, New
Otto S. Neuwirth, UK Wesseling, Wesseling, Jersey, United States, Petroleum,
Germany, Oil Refining, Environmental Introduction
Considerations Michael Röper, BASF Aktiengesellschaft,
Franz Nierlich, Oxeno Olefinchemie GmbH, Ludwigshafen, Germany, Acylation and
Marl, Germany, Methyl Tert-Butyl Ether Alkylation
E. Dickson Ozokwelu, Amoco Chemical F. S. Rosser, UOP LLC, Des Plaines, Illinois,
Company, Naperville, Illinois, United States, United States, Fluid Catalytic Cracking, Units,
Toluene Regeneration
Barbara Paczynska-Lahme, Consultant, Michael Sakuth, Hüls Infracor GmbH, Marl,
Osterode, Germany, Natural Gas Germany, Methyl Tert-Butyl Ether
Christian Paetz, Bayer AG, Leverkusen, J. D. Sauer, Albemarle Corporation, Baton
Germany, Hydrocarbons Rouge, Louisiana, United States, Olefins,
P. Palmas, UOP LLC, Des Plaines, Illinois, Higher
United States, Fluid Catalytic Cracking, Units, Alan W. Scaroni, Pennsylvania State
Regeneration University, University Park, Pennsylvania,
Peter Pässler, BASF Aktiengesellschaft, United States, Combustion Science and
Ludwigshafen, Germany, Acetylene Technology
Wolfgang Payer, Hoechst Aktiengesellschaft, Kuno Schädlich, VEBA OEL AG,
Werk Ruhrchemie, Oberhausen, Germany, Gelsenkirchen, Germany, Octane Enhancers
Waxes Hans Schmidt, Hoechst Aktiengesellschaft,
Udo Peters, Oxeno Olefinchemie GmbH, Marl, Werk Gersthofen, Augsburg, Germany,
Germany, Methyl Tert-Butyl Ether Waxes
Mark R. Pillarella, Air Products and Peter Schug, VEBA OEL AG, Gelsenkirchen,
Chemicals, Inc., Allentown, Pennsylvania, Germany, Octane Enhancers
United States, Natural Gas Ekkehard Schulte-Körne, Oxeno Olefinchemie
Edwin M. Piper, Piper Designs LLC, Littleton, GmbH, Marl, Germany, Methyl Tert-Butyl
Colorado, United States, Oil Shale Ether
Sarma V. Pisupati, Pennsylvania State Reza Sharifi, Pennsylvania State University,
University, University Park, Pennsylvania, University Park, Pennsylvania, United States,
United States, Combustion Science and Combustion Science and Technology
Technology Wolfgang Siegel, BASF Aktiengesellschaft,
Heinrich Predel, MiRO GmbH & Company Ludwigshafen, Germany, Acylation and
KG, Karlsruhe, Germany, Petroleum Coke Alkylation
Roger C. Prince, Exxon Research and Jeffrey J. Siirola, Eastman Chemical
Engineering Company, Annandale, Company, Kingsport, Tennessee, United
New Jersey, United States, Bioremediation States, Separations Process Synthesis
Günter Pusch, Technische Universität Thomas A. Simo, Lurgi Öl Gas Chemie GmbH,
Clausthal, Clausthal-Zellerfeld, Germany, Frankfurt, Germany, Xylenes
Exploration, Drilling, and Production Domien Sluyts, Bayer AG, Antwerpen,
Engineering Belgium, Liquid–Liquid Extraction
xii CONTRIBUTORS Vol. 1

James G. Speight, Consultant, Laramie, Klaus Weggen, Hemmingen, Germany,


Wyoming, United States, Asphalt; Fuels, Exploration, Drilling, and Production
Synthetic, Gaseous; Petroleum Refinery Engineering
Processes; Tar Sands Hans Martin Weitz, BASF Aktiengesellschaft,
Knut Spitzmüller, Hamburg, Germany, Ludwigshafen, Germany, Isoprene
Heating Oil Vincent A. Welch, Washington Group
H. N. Sun, Exxon Chemical Company, Baytown, International, Inc., Cambridge,
Texas, United States, Butadiene Massachusetts, United States, Ethylbenzene
Martti Suominen, Fortum Oil and Gas Oy, Hans-Jürgen Wernicke, Linde AG,
Finland, Methyl Tert-Butyl Ether Höllriegelskreuth, Germany, Acetylene
W. A. Sweeney, Chevron Research and Peter J. Wessner, Merichem Company,
Technology Company, Richmond, California, Houston, Texas, United States,
United States, BTX Processing Naphthenic Acids
Jörg Talbiersky, Dorsten, Germany, Thomas P. Whaley, Consultant, Sun City,
Anthracene Arizona, United States, Pipelines
R. Ray Taylor, Phillips Petroleum Company, Josef Wildgruber, Hoechst Aktiengesellschaft,
Bartlesville, Oklahoma, United States, Werk Gersthofen, Augsburg, Germany, Waxes
Liquefied Petroleum Gas Uwe Wolfmeier, Hoechst Aktiengesellschaft,
Shmuel D. Ukeles, IMI (TAMI) Institute for Werk Gersthofen, Augsburg, Germany,
Research and Development, DSBG, Haifa Bay, Waxes
Israel, Drilling Fluids J. P. Wristers, Exxon Chemical Company,
L. L. Upson, UOP LLC, Des Plaines, Illinois, Baytown, Texas, United States, Butadiene
United States, Fluid Catalytic Cracking, Units, G. Yaluris, W. R. Grace & Company,
Regeneration Columbia, Maryland, United States,
Heinz-Werner Voges, Hüls Aktiengesellschaft, Fluid Catalytic Cracking, Catalysts
Marl, Germany, Hydrocarbons and Additives
J. D. Wagner, Albemarle Corporation, Baton X. Zhao, W. R. Grace & Company, Columbia,
Rouge, Louisiana, United States, Olefins, Maryland, United States, Fluid Catalytic
Higher Cracking, Catalysts and Additives
Roland Walzl, Linde AG, Hoellriegelskreuth, Heinz Zimmermann, Linde AG,
Germany, Ethylene Hoellriegelskreuth, Germany, Ethylene
PART I: EXPLORATION,
PRODUCTION,
AND REFINING

PETROLEUM,
INTRODUCTION
1. Origin

Petroleum is a naturally occurring complex mixture made up predominantly of


carbon and hydrogen compounds, but also frequently containing significant
amounts of nitrogen, sulfur, and oxygen together with smaller amounts of nickel,
vanadium, and other elements. It may occur in solid, liquid, or gaseous form as
asphalt, crude oil, or natural gas (see NATURAL GAS), respectively. The economic
importance of crude oil and natural gas has stimulated considerable interest in
their origin. Because of their fluid nature petroleum phases are mobile in the
subsurface and may have accumulated far from the place where they formed.
Herein, origin of petroleum means the combined processes that operate in nature
and culminate in the petroleum that is present as commercial accumulations in
permeable and porous reservoir rocks. The processes involved are petroleum gen-
eration, which occurs in the source rock; migration out of the source rock and to
the reservoir; and maturation and alteration which operate to change the compo-
sition of the petroleum after it has accumulated in the reservoir.
A biogenic origin for the carbonaceous material in petroleum is widely but
not universally accepted. An inorganic origin of petroleum has been proposed

1
Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.1518090702011811.a01.pub2
2 PETROLEUM, INTRODUCTION Vol. 1

(1,2) and there is a dualist theory incorporating both biological and inorganic
aspects (3). However, because inorganic processes generate racemic mixtures,
the presence of optically active compounds in oils, especially the multiringed
cycloalkanes (naphthenes), provides strong support for a biological hypothesis.
Oils also contain the so-called chemical fossils or biomarkers, compounds hav-
ing characteristic molecular structures that can be related to living systems.
The compounds include isoprenoids, porphyrins, steranes, hopanes, and
many others. The relative abundances of members of homologous series are
often similar to those in living systems. The strong odd preference in the
long-chain normal alkanes (C4) is particularly well documented (4). In addi-
tion, the lack of thermodynamic equilibrium among compounds (5), and the
close association of petroleum with sedimentary rocks formed in an aqueous
environment, suggests a low temperature origin. In this context, low tempera-
ture means less than a few hundred degrees Celsius as opposed to tempera-
tures in the 700–12008C range that characterize igneous processes involving
silicate melts. The elemental composition of petroleum (C, H, N, S, O), the iso-
topic composition of oils, and the presence of petroleum-like materials in more
recent sediments are consistent with a low temperature origin. The evidence
supporting a biological source for the material that generates petroleum is
extensive (6–8).
Organisms produce a wide range of organic compounds including signi-
ficant amounts of biopolymers like proteins, carbohydrates, and lignins, together
with a wide variety of lower molecular weight lipids (6–8). After the death of the
organism, all or part of this organic material may accumulate in aquatic envir-
onments where the various compounds have very different stabilities. Some are
metabolized in the water column by other organisms (including bacteria) and
only the biochemically resistant material is incorporated into sediments. Survi-
val of organic material depends on many factors but particularly the oxidizing or
reducing nature of the system. Preservation is strongly favored in anoxic sedi-
ments (9). However, the formation of a petroleum accumulation requires more
than just a concentration of the relatively low molecular weight hydrocarbons
that are present in more recent sediments. Although C2 –C10 hydrocarbons are
present in extremely low (parts per billion (ppb) level) concentrations in organ-
isms and sediments, these can account for up to 50% or more of the volume of
some crude oils.
Compounds that are not synthesized by organisms are also reported in
crude oils. Apparently these were formed from the available organic matter in
what is thought to be the main process of petroleum generation. As the organic
matter in sediments is buried in a reducing environment, and subjected to gra-
dually increasing temperature and pressure, petroleum is generated as an inter-
mediate in a transformation process that ultimately leads to methane and
graphite. Oxygen is first lost from the organic matter as carbon dioxide and
water. Continued low temperature (<8C) rearrangement takes place and the
immature organic matter is converted to high molecular weight, insoluble mate-
rial called kerogen. This becomes more aromatic and carbon-rich as it generates
petroleum and evolves toward graphite. In contrast, the solvent-extractable
lower molecular weight organic materials called bitumen or extractables
Vol. 1 PETROLEUM, INTRODUCTION 3

increase in hydrogen content and progress through compositions typical of crude


oils to those of gas (10).
The nature of organic material (kerogen) in source rocks controls whether
oil or gas is generated, and also controls the composition of these products.
In many cases, the type of organic matter can be recognized by microscopic
examination after removing the mineral matrix with hydrofluoric and hydro-
chloric acids. Lignin-rich wood-derived materials appear to generate gas;
surface coatings of plants that are rich in cutin and esters of long-chain
acids and alcohols produce waxy oils that frequently have high pour points;
organic matter derived from algae and other aquatic organisms seems to pro-
duce the normal range of crudes (6–8). Algal debris is frequently amorphous.
The principal organic matter types can be identified because these produce dis-
tinct tracks across an atomic H/C versus atomic O/C diagram, the so-called van
Krevelen diagram (11). With rising temperature, generally produced by
increasing depth of burial, both the H/C and O/C atomic ratios decrease as
the kerogen increases in carbon content. Much of the amorphous kerogen is
hydrogen-rich and oxygen-poor, whereas the woody type is oxygen-rich and
hydrogen-poor.
The relative amounts of the various types of organic matter is controlled by
the environment where the sediments were deposited. For example, rivers trans-
port lignin and surface coatings derived from land plants to the marine environ-
ment. As a consequence deltas are gas-prone environments (12) that frequently
contain high pour-point crude oils rich in long-chain alkanes (13,14). Source
rocks containing low percentages of organic matter generate insufficient oil to
drive migration. The oil remains in the rock and is subsequently cracked to gas,
so that poor source rocks only expel gas regardless of the organic matter type.
The ratio of bitumen to kerogen in sediments is initially low, but increases
with depth of burial. It was shown in 1965 (15) that in the Los Angeles basin the
increase becomes significant below about 3350 m, but in the adjacent Ventura
basin the increase occurs deeper at about 4575 m. The geothermal gradients
are different in these two basins, but the marked increase in bitumen:kerogen
ratio occurs at the same temperature in both of them. Because the generation
of bitumen from kerogen is the process of petroleum generation, this demon-
strated clearly the important role of temperature. It also implied a minor
(if any) role for pressure which differed by more than 50% between the two
basins. Subsequent analyses of samples from many wells worldwide have
shown similar trends. Wells have relatively low bitumen:kerogen ratios for the
shallower section and rapidly increasing values deeper (7).
The petroleum generation process can be duplicated by laboratory pyroly-
sis. Higher temperatures are needed to produce these reactions in a few hours
or days rather than the millions of years in nature (16,17). Both dry pyrolysis
and hydrous pyrolysis have been used.
The generation of petroleum is nonbiological, induced by temperature,
and influenced by available time. It follows essentially first-order kinetics
(18) where an increase of 108C roughly doubles the reaction rate at low
temperatures. The low molecular weight compounds generated from the
kerogen show none of the biological characteristics typical of compounds in
4 PETROLEUM, INTRODUCTION Vol. 1

more recent sediments. Therefore, at increasing depth, and hence increasing


temperature, the bitumen fraction loses features such as odd–even predomi-
nance in long-chain normal alkanes, optical activity, and the predominance
of four- and five-ringed cycloalkanes (11). These trends have been well-
documented in many areas. The petroleum generation process can be treated
quantitatively using models based on first-order kinetics (19). The Lopatin
approximation (20) seems to give reasonable values for the onset of oil
generation when properly calibrated (21), but more rigorous methods that
treat generation as a series of parallel first-order reactions are more reliable
(18,22–25).
An important exception to thermal generation is the bacterial formation
of methane. The bacteria are anaerobic and effective in sulfate-free, anoxic con-
ditions (26), and have long been recognized for their role in forming marsh gas.
Bacterially produced methane is isotopically light, generally more negative

than 55 parts per thousand (expressed as 55 = ) for carbon, relative to the
Pee Dee Belemnite (PDB) standard, and contains only trace amounts of ethane
and higher hydrocarbons. This methane may, however, have higher amounts of
inorganic gases. Many of the huge natural gas accumulations in Siberia,
including the giant Urengoy field which has reserves of 5.9  1012 m3 (210
Tcf), are thought to be bacterially generated. These accumulations are usually

shallow, are isotopically lighter than 59=  (27), and have only a percent or
less of higher hydrocarbons. Examples are also known from other parts of
the world (28–30).
Most petroleum is found in reservoir rocks that have high permeabilities
and porosities, where these properties have been developed by natural pro-
cesses of sorting and winnowing that remove fine-grained particles, including
organic materials. Reservoir rocks generally have insufficient organic matter to
generate commercially significant quantities of petroleum (see OIL SHALE). It is
believed that petroleum generation occurs in organic-rich source rocks, and
that part of the bitumen then migrates to accumulate in reservoir rocks
(6,7). This is the source rock concept. Clearly, migration has a critical role in
linking the organic-rich source rocks to the reservoir. Solubilities of hydrocar-
bons in subsurface waters are generally too low to be significant in the petro-
leum migration process (31,32), and most recent studies have stressed the
expulsion of a separate crude oil phase out of source rocks (33,34). It appears
that source rocks develop high internal petroleum saturation caused by petro-
leum generation and the displacement of water during compaction (35,36).
Upon continuing compaction and kerogen conversion, oil droplets are forced
out of source rocks into adjacent permeable carrier beds. High pressures
approaching the rock load (lithostatic) can develop, and induce near-vertical
fractures that are important in providing migration pathways out of the source
rock. In this case the pressure gradient that develops can overcome buoyancy,
and oil may be expelled downward and out the bottom of source rocks, as well
as out of the top. The alkanes are less strongly absorbed in the source rock and
so are preferentially expelled. In contrast, the nitrogen–sulfur–oxygen com-
pounds (NSOs) are most strongly adsorbed and thus depleted in the expelled
oil (37). Migration efficiency varies widely and appears to be dependent on
the organic matter content of the rock. High efficiencies are associated with
Vol. 1 PETROLEUM, INTRODUCTION 5

high organic contents. The controlling factor is hydrocarbon (bitumen) satura-


tion, and rocks having less than about 1.0 wt% organic matter do not generate
sufficient bitumen. As a consequence no oil migrates from these rocks and they
are not effective source rocks (18,38).
Buoyancy is the main driving force through the carrier beds and oils con-
tinue to move upward (toward shallower areas) until stopped at a slope reversal
in a structural trap, or where permeability decreases as in a stratigraphic trap
(39,40). Migration distances can be in excess of 100 km (35,39). Oil may be
remobilized after its initial accumulation in the reservoir. Although in the sim-
plest case this may involve only a simple relocation, it can lead to significant
compositional changes if both gas and oil are involved. When an anticlinal
reservoir is full to the spill point and has a gas cap over oil, any spilling off
the bottom is oil, and the next shallower trap thus accumulates oil with no
gas cap. This process of differential entrapment (41) leads eventually to oil in
the up dip (shallower) reservoirs and gas in the deeper ones. Geological exam-
ples are given (42,43).
The composition of petroleum changes and evolves in the reservoir in
response to changing conditions. Thermal maturation of crude oil is brought
about by the increasing temperature that accompanies increasing depth of burial
(44,45). Some large molecules are broken down into smaller fragments and the
trend is for an increasing percentage of the lighter fractions as the oil progresses
to lower densities in the sequence from oil, to lighter oil, to wet gas, and finally
dry gas (46). The increasing hydrogen content implied by this sequence is pro-
vided by parallel reactions involving cyclization and aromatization. These resi-
dual molecules get steadily more aromatic and larger, and as the solvent
properties of the oil change, these molecules are precipitated in the reservoir
in a process called natural deasphaltening (47). The solid precipitate also takes
with it much of the NSOs, nickel, and vanadium, so that the producible oil is of
better quality.
Large changes in petroleum composition can be produced by contact with
flowing water (48,49). As the water moves past the oil in the reservoir, most
water-soluble components are removed. These include the light ends, particu-
larly the small aromatics, leaving a tar layer at the oil–water interface. If the
water brings bacteria and oxygen into contact with the oils at temperatures
below approximately 708C, substantial changes in crude oil composition can
result (49–51). The aerobic bacteria preferentially consume normal alkanes
and may convert waxy crudes of high pour points to naphthenic crudes of low
pour points. The removal of low density compounds leaves a heavier oil
enriched in sulfur and nitrogen compounds because heavier fractions resist
bacterial degration and so accumulate in the residue. Water washing always
accompanies biodegradation.
Biomarkers form a small percentage of bitumen and crude oils, but relative
distributions and complex structures are modified by the various processes
involved during petroleum generation and accumulation. These biomarkers are
widely used for correlation studies, and for recognition and documentation of the
progress of generation and maturation (52,53).
The overall petroleum system (54) that leads to the accumulation of
oil and gas in natural reservoirs can be summarized as follows: organic
6 PETROLEUM, INTRODUCTION Vol. 1

matter is incorporated into sediments as these are deposited; possible shallow


generation of biogenic methane; organic matter is converted to petroleum-like
materials by the influence of increasing temperature with lower temperatures
partially offset by longer times; part of the lower molecular weight material
that is generated subsequently migrates from the source rock through perme-
able carrier beds to the reservoir; after the oil reaches the reservoir significant
compositional changes may be produced by increasing temperature, water
washing, and bacterial degradation.

2. Composition

Petroleum, literally rock oil, describes a myriad of hydrocarbon-rich fluids pre-


sent in source rocks and accumulated in subterranean reservoirs. Petroleum
can include three phases: gaseous (natural gas), liquid (crude oil), and solid
or semisolid (bitumens, asphalt, tars, and pitches) (55) (see NATURAL GAS).
The molecular composition of the liquid portion of petroleum contributing to
the crude oil properties and behavior is discussed in this section. Crude oils
vary dramatically in color, odor, and flow properties. These properties often
reflect the origin of the crude. Historically, physical properties such as boiling
point, density (gravity), odor, and viscosity have been used to classify oils
(2–4). Crude oils may be called light or heavy in reference to relative density
(or specific gravity). Light crude oils are rich in low boiling and paraffinic
hydrocarbons; heavy crude oils contain greater amounts of high boiling
and asphalt-like molecules. The heavy oils tend to be more viscous, higher boil-
ing, more aromatic, and contain larger amounts of heteroatoms. Likewise, odor
is used to distinguish between sweet or low sulfur, and sour or high sulfur,
crude oils.
The distribution of biomarker isomers, molecules that retain the basic car-
bon skeletons of biological compounds from living organisms, serves not only as a
set of fingerprints for oil–oil and oil–source correlation (to relate the source and
reservoir for exploration), but also to give geochemical information on organic
source input (marine, lacustrine, or terrigenous source), age, maturity, deposi-
tional environment (clay or carbonate, oxygen levels, salinity, etc), and alteration
(water washing, biodegradation, etc) (56,57).
Knowledge of the composition of petroleum allows the refiner to optimize
conversion of raw petroleum into high value products (58). Originally, petro-
leum was distilled and sold as fractions, primarily for use in illumination
and lubrication. Crude oil is sold in the form of gasoline (see GASOLINE AND
OTHER MOTOR FUELS), solvents (see SOLVENTS, INDUSTRIAL), diesel and jet fuel,
heating oil, lubricant oils, and asphalts, or it is converted to petrochemical
feedstocks such as ethylene (qv), propylene (qv), the butenes, butadiene, and
isoprene. Modern refining uses a sophisticated combination of heat, catalyst,
and hydrogen. Conversion processes include coking, hydrocracking, and cata-
lytic cracking to break large molecules into smaller fractions; hydrotreating
to reduce heteroatoms and aromatics, thereby creating environmentally
acceptable products; and isomerization and reforming to rearrange molecules
to those having high value, eg, gasolines of high octane number.
Vol. 1 PETROLEUM, INTRODUCTION 7

A knowledge of the molecular composition of a petroleum also allows envir-


onmentalists to consider the biological impact of environmental exposure.
Increasingly, petroleum is being produced in and transported from remote
areas of the world to refineries located closer to markets. Although only a
minuscule fraction of that oil is released into the environment, the sheer volume
involved has the potential for environmental damage. Molecular composition can
not only identify the sources of contamination but also aids in understanding the
fate and effects of the potentially hazardous components (59).
Crude oils contain an extremely wide range of organic functionality and
molecular size. The variety is so great that a complete compound-by-compound
description for even a single crude oil is not likely. The composite molecular com-
position of petroleum can, however, be described in terms of three classes of com-
pounds: saturates, aromatics, and compounds bearing the heteroatoms sulfur,
oxygen, or nitrogen. Within each of these classes there are several families of
related compounds. Table 1 lists some of the compounds typically found in petro-
leum crude oils. Structures are shown in Figure 1. The saturates include normal
alkanes, branched alkanes, and cycloalkanes called paraffins, isoparaffins, and
naphthenes, respectively. Alkenes (olefins) are rare to the extent of being con-
sidered an oddity. Aromatics range from benzene to multiple fused-ring analo-
gues such as naphthalene, phenanthrene, etc. Sulfur is found in both polar
(thiol) and nonpolar (thioether and thiophenic) forms (60). The nitrogen and

Table 1. Compounds Found in Petroleum Crude Oilsa


CAS Registry Molecular Structure
Compound number formula number
Saturates
n-octane [111-65-9] C8H18 (1)
2-methyloctane [3221-61-2] C9H20 (2)
propylcyclohexane [1678-92-8] C9H18 (3)
Aromatics
n-butylbenzene [104-51-8] C10H14 (4)
1-methylnaphthalene [90-12-0] C11H10 (5)
9-methylphenanthrene [883-20-5] C15H12 (6)
Sulfur compounds
propyl mercaptan [107-03-9] C3H8S (7)
methyl propyl sulfide [3877-15-4] C4H10S (8)
dibenzothiophene [132-65-0] C12H8S (9)
Oxygen compounds
phenol [108-95-2] C6H6O (10)
2-phenanthrene [40452-20-8] C15H10O2 (11)
carboxylic acid
cyclohexyl carboxylic acid [98-89-5] C7H12O2 (12)
Nitrogen compounds
quinoline [91-22-5] C9H7N (13)
carbazole [86-74-8] C12H9N (14)
2(1H)-quinolinone [59-31-4] C9H7NO (15)
a
See Fig. 1.
8 PETROLEUM, INTRODUCTION Vol. 1

CH 3
CH 2 CH 3
CH 2 CH 2 CH 2 CH 3 CH CH 2 CH 2 CH 3 CH 2
H 3C CH 2 CH 2 CH 2 CH 3 CH 2 CH 2 CH 2

(1) (2) (3)

CH 3
CH 2 CH 2
CH 2 CH 3

CH 3
(4) (5) (6)

CH 2 SH CH 3 CH 2 CH 3
CH 3 CH 2 S CH 2 S

(7) (8) (9)

COOH
OH
COOH

(10) (11) (12)

N N N O
H H

(13) (14) (15)

Fig. 1. Structures of compounds in petroleum crude oils. See Table 1.

oxygen compounds are more likely to be found in polar compounds such as pyr-
idines, pyrroles, phenols, carboxylates, amides, etc, than in nonpolar ones such
as ethers (61). The distribution and characteristics of these molecular species
account for the rich variety of crude oils.
2.1. Elemental Composition. On an atomic basis, H/C ratios range
from 1.5–2.0. The range of elemental composition of crude oil (16) may be
given as follows:

Element Composition range, wt%


carbon 84–87
hydrogen 11–14
sulfur <0.1–8
oxygen <0.1–1.8
nitrogen <0.1–1.6
Vol. 1 PETROLEUM, INTRODUCTION 9

Nickel and vanadium are also generally present from a trace amount up to 1000
ppm. The value of using sulfur for petroleum classification can be clearly seen.
Whereas the range of hydrogen and carbon are quite narrow, sulfur, the princi-
pal heteroatom in crude oil, varies significantly. The ranges for the other two het-
eroatoms (O and N) are also fairly wide, however in most crude oils these
elements are nearly an order of magnitude lower than the sulfur level (Fig. 2).
For example, the average level of nitrogen in >9000 crude oils is 0.094%, the
average sulfur content is 0.65% (7). Although >90% of crude oils have nitrogen
levels <0.2%, some crude oils have > 1.5% nitrogen. As noted in Figure 2b, the
nitrogen concentrates into the highest boiling fractions. The trace metals Ni and
V are generally orders of magnitude higher than other metals in petroleum,
unless it is contaminated with coproduced brine salts (Naþ, Mg2þ, Ca2þ, Cl)
or iron corrosion products gathered in transportation.
2.2. Molecular Classes. The molecules in crude oil include several basic
structural types (Table 1, Fig. 1). Because they may contain from 1 to 100þ carbon
atoms and may occur in combination, the statistical potential for isomeric struc-
tures is staggering. For example, whereas there are just 75 possible paraffinic

10
Sulfur, % of fraction

C
0
(a)
Nitrogen, % of fraction

1.0

A
B

0
15 210 345 565
Boiling point, °C
C5 C10 C20 C40

(b)

Fig. 2. Distribution of compounds as a function of crude oil boiling point: (a) sulfur
where A, B, and C represent high (>2%), medium (ca 1.5%), and low (<0.1%) sulfur,
respectively, and (b) nitrogen, where A and B represent high (ca 0.5%) and low (<0.1%)
nitrogen, respectively.
10 PETROLEUM, INTRODUCTION Vol. 1

structures for C10, there are >105 isomers for C20. A few structures tend to dom-
inate the distributions of each isomer group, however.
The inclusion of naphthene and other aromatic rings introduces two addi-
tional dimensions, increasing the number of hydrocarbon isomers even further.
A three-dimensional array in which the molecules could be described in terms of
the number of aromatic rings, the number of naphthenic rings, and the number
of carbons in alkyl side chains has been proposed (61). Conceptually, this
amounts to describing a three-dimensional molecular mountain for hydrocar-
bons. There is also the potential of constructing similar mountains for heterocyc-
lics. The two-dimensional image in the naphthenic and aromatic dimensions has
been projected using sidebars to indicate the variation in alkyl substituents
among crude oils (7).
Molecular characterization of a whole oil is beyond the capability of most
analytical techniques. Distillation, however, can separate petroleum into mole-
cular weight fractions that simplify the task. Pioneering work with this
approach, sponsored by the American Petroleum Institute (API) starting in
1925, has led to the identification of hundreds of individual compounds in distil-
lation fractions of a single crude oil (62). Developments in chromatography
allowed oils to be fractionated by polarity as a second dimension. Under API
sponsorship, the U.S. Bureau of Mines extended separations and measurement
techniques to heavier fractions (63). At the same time, individual compounds
have been isolated and quantified from increasingly higher boiling fractions
(64–67). Techniques have been developed that use combinations of classical
open-column adsorption chromatography, gel permeation chromatography, and
ion-exchange separations to isolate fractions in which compounds could be iden-
tified by mass spectrometry.
Whereas neither distillation nor chromatography achieves perfect separa-
tions among groups, the fractions generated are amenable to molecular charac-
terization (66,67). An elegant argument has been made for the use of
distillation for primary separation which is combined with a solubility step
to achieve the atmospheric equivalent boiling point (AEBP) scale that covers
room temperature to 13708C (68). Because the distillations simplify analytical
complexity, the bulk of available molecular compositional data on petroleum
has been generated on sets of fractions defined by boiling points as gases
(C1 –C4), naphtha (initial bp 2108C), mid-distillate (210–3458C), vacuum
gas oil (345–5658C), and vacuum residuum (>5658C). The initial boiling
point (ibp) of the naphtha depends on the amount of C1 –C5 hydrocarbons dis-
solved in the oil. For characterization purposes, naphtha is usually depenta-
nized so that the ibp is about 328C. The material that boils above 3458C is
referred to as atmospheric resid and is further distilled under vacuum into
vacuum gas oil (VGO) and vacuum resid. Whereas the AEBP of the vacuum
resid is >5658C, the actual temperature of distillation does not exceed 3458C
to avoid thermal decomposition.
Crude distillations yield different quantities in each fraction. About the
same amounts are distilled into the middle distillate and vacuum gas oil from
conventional crude oils. More naphtha is distilled from light crude oils and more
vacuum residuum is obtained from heavy crude oils (Fig. 3). The typical distri-
bution of classes of petroleum compounds shows a significant shift with boiling
Vol. 1 PETROLEUM, INTRODUCTION 11

100
A

Cumulative volume distilled, %


B
C

0
15 210 345 565
Boiling point, °C
C5 C10 C20 C40

Middle Vacuum
Naphtha Resid
distillate gas oil

Fig. 3. Cumulative volume distilled as a function of boiling point from A, light; B, inter-
mediate; and C, heavy crude oils (not including condensates).

point (Fig. 4). Whereas the lower boiling fractions are dominated by nonpolar
saturated hydrocarbons that exist in limited isomeric forms, the higher boiling
fractions increasingly contain a larger variety of classes, that have, in turn, an
increasing number of possible isomers. As the boiling point increases, aromatic
ring structures build in, first as naked rings, then more and more as rings

100
1 Ring
2 Ring
3 Ring
4+ Ring
Cumulative mass,%

50 C

0
15 210 345 565 705
Boiling point, °C
C5 C10 C20 C40

Fig. 4. Distribution of compound classes in crude oils as a function of boiling point.


Region A represents normal paraffins; B, isoparaffins; C, naphthenes; & the region of
alkyl and napthenic aromatics; and the region of polars.
12 PETROLEUM, INTRODUCTION Vol. 1

having attached side-chain and naphthene ring carbons. Polar compounds,


typically those having O and N functionality, that appear as trace impurities
in the lower boiling fractions gradually become significant components in the
higher boiling fractions. This is confirmed by the distribution of S and N in
petroleum against boiling point (Fig. 2). The S, not including H2S and the
light sulfur compounds such as mercaptans and sulfides, present in petroleum
gases, is more widely distributed than the nitrogen that concentrates in the
highest boiling fraction. Not shown is the subtle decrease that occurs in H/C
ratio with increasing boiling point reflecting the increasing number of aromatic
ring types at higher boiling point. The metals, nitrogen, and oxygen are predo-
minantly found in the higher boiling fractions rich in polars.
Analytical Approaches. Different analytical techniques have been
applied to each fraction to determine its molecular composition. As the molecular
weight increases, complexity increasingly shifts the level of analytical detail
from quantification of most individual species in the naphtha to average molecu-
lar descriptions in the vacuum residuum. For the naphtha, classical techniques
allow the isolation and identification of individual compounds by physical proper-
ties. Gas chromatographic (gc) resolution allows almost every compound having
less than eight carbon atoms to be measured separately. The combination of gc
with mass spectrometry (gc/ms) can be used for quantitation purposes when com-
pounds are not well-resolved by gc.
For the mid-distillates and vacuum gas oils (VGOs), class isolation
and measurement techniques allow speciation of many compounds. In particu-
lar, multidimensional instrumental techniques have been applied. Capillary gc
quantifies even closely related isomers. Element-selective detectors for gc are
used to identify N, S, and O heterocompounds; gc/ms techniques are used to
identify and quantify individual compounds or families of compounds. Charac-
terization of families of compounds having similar degrees of unsaturation,
expressed by z-number as in CnH2nþzX, where X represents heteroatoms, can
be done by ms quantification (69). The combination of high performance liquid
chromatography (hplc) and high resolution/low voltage ms has also been used
to resolve overlaps between aromatic hydrocarbons and sulfur analogues as
well as among isomers of alkyl aromatics and naphthenoaromatics (69). For
VGOs ms techniques give semiquantitative results owing to the lack of refer-
ence compounds for calibrating relative sensitivities. Alternative techniques
such as ultraviolet detection may be needed to distinguish among structural
isomers (70,71).
The combination of chromatographic isolation followed by ms measure-
ments has been extended well above 5658C into fractions of the vacuum resid
(72–74). For fractions not amenable to that approach, techniques that provide
average functionality data have been applied to describe petroleum composi-
tion (72). Established techniques such as titration for acidic and basic function-
alities and for sulfur types, ultraviolet (uv) spectroscopy for aromaticity,
and infrared (ir) spectroscopy for dipolar functionality have been supplemented
with alternative instrumental techniques to expand the average database.
These include nuclear magnetic resonance (nmr) for aromaticity, x-ray photo-
electron spectroscopy (xps) for chemical bonding, and extended x-ray absorp-
tion fine structure (exafs) for atomic coordination environment, as well as
Vol. 1 PETROLEUM, INTRODUCTION 13

Table 2. Compounds Found in Petroleum Naphthaa


Molecular Structure
Compound CAS Registry formula number
n-heptane [142-82-5] C7H16 (16)
2-methylheptane [592-27-8] C8H18 (17)
methylcyclopentane [96-37-7] C6H12 (18)
ethylcylohexane [1678-91-7] C8H16 (19)
1,3-dimethylbenzene (m-xylene) [108-38-3] C8H10 (20)
indan [496-11-7] C9H10 (21)
naphthalene [91-20-3] C10H8 (22)
tetrahydronaphthalene (tetralin) [119-64-2] C10H12 (23)
decahydronaphthalene (decalin) [91-17-8] C10H18 (24)
methyl mercaptan [74-93-1] CH4S (25)
3-methylthiacyclohexane [5258-50-4] C6H12S (26)
3-(methylthio)pentane [57093-84-2] C6H14S (27)
a
See Fig. 5.

improved Fourier-transform ir and Raman techniques for functionality. The


individual average techniques do not provide molecular composition (75). On
the other hand, these techniques can provide an average composition when
used in combination. For those portions of the vacuum residua that are both
nonvolatile and insoluble and hence not accessible to molecular speciation tech-
niques, the alternative techniques are helpful.
Petroleum Gases and Naphtha. Methane is the main hydrocarbon
component of petroleum gases. Lesser amounts of ethane, propane, butane, iso-
butane, and some C4þ light hydrocarbons also exist. Other gases such as hydro-
gen, carbon dioxide, hydrogen sulfide, and carbonyl sulfide are also present.
The naphtha fraction is dominated by saturates having lesser amounts
of mono- and diaromatics (Table 2, Fig. 4). Whereas naphtha (ibp to 2108C)
covers the boiling range of gasoline, most raw petroleum naphtha molecules
have a low octane number and most raw naphtha is processed further, to be
combined with other process naphthas and additives to formulate commercial
gasoline.
Within the saturates in petroleum gases and naphtha, except for a few
highly branched components in the C8 –C10 range, every possible paraffin
from methane to normal decane (n-C10) is present. Depending on the source,
one of the low boiling paraffins may be the most abundant compound in a
crude oil, sometimes reaching a concentration of several percent. The isoparaf-
fins begin at C4. Isobutane is the only isomer of n-butane. The number of iso-
mers grows rapidly with carbon number so that there are 74 isomers of C10.
Some of the individual 2-methyl isoparaffins may be present in concentrations
>1%. Using instrumental techniques every possible isoparaffin in petroleum
up to C8 has been identified. Although complete resolution of all isomers >C8
is beyond analytical capabilities, many of these compounds have also been
identified.
In addition to the aliphatic (chain) molecules, the saturates contain
cycloalkanes, called naphthenes, having mainly five or six carbons in the ring
(Fig. 5). Methyl derivatives of cyclopentane and cyclohexane are commonly
14 PETROLEUM, INTRODUCTION Vol. 1

CH 3 CH 3

CH 2 CH 2 CH 2 CH CH 2 CH 2
CH 3 CH 2 CH 2 CH 3 CH 3 CH 2 CH 2 CH 3

(16) (17) (18)

CH 3
CH 2
CH 3
CH 3

(19) (20) (21)

(22) (23) (24)

S CH 2
CH 3 CH 3 CH CH 3
CH 3 SH
CH 2 CH 3
S
(25) (26) (27)

Fig. 5. Structures of compounds in petroleum naphtha. See Table 2.

found in greater quantity than the parent unsubstituted structures and can be
present at levels above 2% (7). Fused-ring dicycloalkanes such as decahydro-
naphthalenes (decalins) and hexahydroindans are also common, but nonfused
bicylic naphthenes, eg, cyclohexyl cyclohexane, are not.
The numerous aromatics in petroleum naphtha begin with benzene, but
the C1 –C3 alkylated derivatives of benzene generally are present in larger
amounts. Toluene concentrations may reach nearly 2%; the combined xylene
isomers exceed 1%; benzene, however, rarely exceeds 1% of a whole crude oil
(7). Although present in lesser concentrations, each of the alkyl benzene homo-
logues through the 20 isomeric C4 alkyl benzenes have been isolated from
Ponca City crude, as have several of the C5-derivatives (76). Benzenes having
fused cycloparaffin rings (naphthenoaromatics) such as indan and tetralin
have been isolated along with a number of their methyl derivatives. Naphtha-
lene is included in the naphtha, whereas the 1- and 2-methyl naphthalenes and
higher homologues of fused two-ring aromatics appear in the mid-distillate
fraction.
Sulfur is the only heteroatom to be found in naphtha, and only at trace
levels (Fig. 2). A systematic study of organic sulfur compounds in Wasson
Texas crude, and to a lesser extent three other crude oils, culminated in the
identification of over 200 compounds, most of which were in the naphtha fraction
(24). However, the total of these compounds accounted for less than 1% of the
Vol. 1 PETROLEUM, INTRODUCTION 15

sulfur in the whole crude. In sour petroleum naphthas, 50–70% of the sulfur can
be found in the form of mercaptans, ie, thiols. Over 40 individual thiols have been
identified, including all the isomeric C1 to C6 compounds plus some C7 and C8
isomers and thiophenol (77). In sweet naphthas, the sulfur is distributed
between sulfides, ie, thio-ethers, and thiophenes. In these cases the sulfides
may constitute more than 50% of the sulfur compounds in the form of both linear
(alkyl sulfides) and five- or six-ring cyclic (thiacyclane) structures. Sulfur struc-
ture distribution mimics the hydrocarbons; ie, naphthenic oils having high
cycloalkanes have high thiacyclane content, etc. Typical alkyl thiophenes in
naphtha have multiple short side chains or exist as naphthenothiophenes (77).
Though rare, methyl and ethyl disulfides have been confirmed to be present in
some crude oils (78).
Mid-Distillates. As is indicated in Figure 4, saturates remain the primary
component in the mid-distillate fraction of petroleum, but aromatics, which
include simple compounds having up to three aromatic rings, and heterocyclics
are present and represent a larger portion of the total. Some raw middle distil-
lates are used directly as kerosenes, jet fuels, and diesel fuels; others are
cracked and hydroprocessed before use. Some compounds found in middle dis-
tillates are listed in Table 3. Within the saturates, the concentration of n-par-
affins generally decreases regularly from C11 to C20. The 2-methyl analogues
are sufficiently unique to be seen as distinct peaks in gc analyses of mid-distil-
lates. Few additional isoparaffins have been identified. However, the two iso-
prenoids (30) and (31) (see Fig. 6) are generally present in crude oils in
sufficient concentration to be seen as irregular peaks alongside the n-C17 and
n-C18 peaks in a gas chromatogram. These isoprenoids, believed to arise primar-
ily as fragments of ancient chlorophyll [1406-65-1] have relevance, as simple

Table 3. Compounds Found in Petroleum Middle-Distillatesa


CAS Registry Molecular Structure
Compound number formula number
n-hexadecane (cetane) [544-76-3] C16H34 (28)
2-methylpentadecane [1560-93-6] C16H34 (29)
pristane (2,6,10,14-tetramethyl- [1921-70-6] C19H40 (30)
hexadecane)
phytane (2,6,10,14-tetramethyl- [638-36-8] C20H42 (31)
hexadecane)
pentamethyldecalin [80655-44-3] C15H28 (32)
fichtelite [2221-95-6] C19H34 (33)
adamantane [281-23-2] C10H16 (34)
biphenyl [92-52-4] C12H10 (35)
fluorene [86-73-7] C13H10 (36)
phenanthrene [85-01-8] C14H10 (37)
2-methylbenzothiophene [1195-14-8] C9H8S (38)
dibenzothiophene [132-65-0] C12H8S (39)
benzothiacyclohexane [2054-35-5] C9H10S (40)
a
See Fig. 6.
16 PETROLEUM, INTRODUCTION Vol. 1

CH 2 CH 2 CH 2 CH 2 CH 2 CH 2 CH 2 CH 3
CH 3 CH 2 CH 2 CH 2 CH 2 CH 2 CH 2 CH 2

(28)

CH 3
CH CH 2 CH 2 CH 2 CH 2 CH 2 CH 2
CH 3 CH 2 CH 2 CH 2 CH 2 CH 2 CH 2 CH 3

(29)

CH 3 CH 3 CH 3 CH 3
CH CH 2 CH CH 2 CH CH 2 CH
CH 3 CH 2 CH 2 CH 2 CH 2 CH 2 CH 2 CH 3

(30)

CH 3 CH 3 CH 3 CH 3
CH CH 2 CH CH 2 CH CH 2 CH CH 3
CH 3 CH 2 CH 2 CH 2 CH 2 CH 2 CH 2 CH 2

(31)

CH 3
CH
CH 3
H 3C H 3C CH 3
CH 3

H 3C CH 3
CH 3

(32) (33) (34)

(35) (36) (37)

S CH 3 S S

(38) (39) (40)

Fig. 6. Structures of compounds in petroleum middle-distillates. See Table 3.


Vol. 1 PETROLEUM, INTRODUCTION 17

biomarkers, to the genesis of petroleum. The distribution of pristane and phy-


tane relative to the neighboring n-C17 and n-C18 peaks has been used to aid in
the identification of crude oils and to detect the onset of biodegradation (6). The
ratio of pristane to phytane has also been used for the assessment of the stage of
oxidation/reduction of the environment in which ancient organisms were con-
verted into petroleum (56).
Mono- and di-cycloparaffins having five or six carbons per ring constitute
the bulk of naphthenes in the middle-distillate boiling range, decreasing in con-
centration as the carbon number increases (7). The alkylated naphthenes appear
to have a single long side chain as well as one or more methyl or ethyl groups
(79). Similarly substituted three-ring naphthenes have been detected by gc/ms
(80). Generally, fused rings share just a single face; however, the highly
symmetric multifused molecule adamantane and its alkyl-substituted homolo-
gues have been found (6,81).
The most abundant aromatics in the mid-distillate are mono-, di-, and tri-
methyl naphthalenes. Other one- and two-ring aromatics are undoubtedly pre-
sent in small quantities as either naphtheno or alkyl homologues in the C11 –C20
range. In addition to these homologues of alkylbenzenes, tetralins, and
naphthalenes, the mid-distillate contains some fluorenes and phenanthrenes,
with traces of biphenyls (82). The phenanthrene structure is favored over
that of anthracene structure (7).
The S-heterocyclics in the mid-distillate range are primarily the thiacy-
clanes, benzothiophenes, and dibenzothiophenes. There are lesser amounts of dia-
lkyl–, diaryl–, and aryl–alkyl sulfides (81). Alkylthiophenes are scarce or absent,
but some evidence exists for benzthiacyclanes (Fig. 6). As for the naphtha frac-
tions, these sulfur species account for a minimal fraction of the total sulfur in
the crude.
Although only ppm levels of nitrogen are found in the mid-distillates,
both neutral and basic nitrogen compounds have been isolated and identified
in fractions boiling below 3458C (64). Pyrroles and indoles account for about
two-thirds of the nitrogen. The remaining nitrogen is found in the basic pyri-
dine and quinoline compounds. Most of these compounds are alkylated.
2.3. Vacuum Gas Oils. As is indicated in Figure 4, saturates contribute
less to the vacuum gas oil (VGO) than the aromatics, but more than the polars
present at percentage, rather than trace, levels. VGO itself is occasionally used
as a heating oil but most commonly it is processed by catalytic cracking to pro-
duce naphtha or by extraction to yield lubricant oils.
Within the VGO saturates, distribution of paraffins, isoparaffins, and
naphthenes is highly dependent on the petroleum source. The naphthenes
account for roughly 60% of the saturates in a normal crude oil. However,
samples can be found having paraffins from <20 to >80%. In most samples,
the n-paraffins from C20 –C44 are still present in sufficient quantity to be
detected as distinct peaks in gc analyses. Some crude oils show a nearly sym-
metric pattern of peaks such that each carbon number is present in regular pro-
gression up to a maximum around C27. Other crude oils show a similar
distribution, but have preference for odd-numbered alkanes. Both the distribu-
tion and the selectivity toward odd-numbered hydrocarbons are considered
to reflect differences in petrogenesis of the crude oils. Although n-paraffins are
18 PETROLEUM, INTRODUCTION Vol. 1

H
H 3C CH 2 CH 2 CH 3
C CH 2 CH
H 3C H H
H
CH 3
H 3C H 3C CH 3 CH CH 3

CH 3
H CH 3

H
H 3C CH 3

(a) (b)

Fig. 7. Naphthenic biomarker compounds: (a) 5a(H),14a(H),17a(H),20(R)-cholestane


[481-21-0], a C27 sterane, and (b) 17a(H),21a(H)-hopane [13849-96-2], a C30 pentacyclic
triterpane.

distinct in the gc, these usually account for only a few percent of the saturates
measured by gc.
The bulk of VGO saturates consists of isoparaffins and especially naphth-
enes (Fig. 3). A few isoprenoid compounds, such as squalane, C30; lycopane,
C40; and carotanes, C40; have been detected. Analyses of petroleum waxes iso-
lated from the saturates show a parallel lower level of 2- and 3-methyl alkanes
as the most identifiable isoparaffins. Mass spectrometry techniques show that
the naphthenes contain from one to more than six fused rings. Having an aver-
age carbon number of C32, even the six-ring naphthenes have some alkyl sub-
stitution. For mono- and diaromatics, the alkyl substitution typically involves
one long side chain and several short methyl and ethyl substituents. Some spe-
cific tetracyclic naphthenes, including steranes, and pentacyclic naphthenes,
including hopanes, have been used as biomarkers (Fig. 7). These hopanes
and steranes have also been used as nondegradable conserved internal
markers for estimating biodegradation of crude oils during bioremediation
processes (83).
The aromatics in VGO may contain one to six fused aromatic rings that may
bear additional naphtheno rings and alkyl substituents in keeping with their boil-
ing range. Mono- and diaromatics account for about 50% of the aromatics in petro-
leum VGO samples. Studies by nmr indicate an average of 3.5 substituents per
monoaromatic, whereas ms techniques show the presence of up to four fused
naphthenic rings on some aromatic compounds (71). This is consistent with the
suggestion that these species originate from the aromatization of biogenic steroids
(7,84). Although they are present at lower concentration, alkyl benzenes and
naphthalenes commonly show one long side chain and multiple short side chains.
The fused 3þ ring aromatics in petroleum include both cata- and peri-
condensed structures (see Table 4, Fig. 8). The cata-condensed species are
those structures where only one face is shared between rings, the peri-condensed
molecules are those that share more than one face. The fused ring aromatics
form the class of compounds known as polynuclear aromatic hydrocarbons
(PAH) which includes a number of recognized carcinogens in the 4þ ring
family (85). Because of the potential health and environmental impact of PAH,
these compounds have been studied extensively in petroleum.
Vol. 1 PETROLEUM, INTRODUCTION 19

Table 4. Fused-Ring Polynuclear Aromatic Hydrocarbons Found in Petroleuma


CAS Registry Molecular Structure
Compound number formula number
cata-Condensed aromatics
phenanthrene [85-01-8] C14H10 (37)b
chrysene [218-01-9] C18H12 (8)
picene [213-46-7] C22H14 (42)
peri-Condensed aromatics
fluoranthene [206-44-0] C16H10 (43)
pyrene [129-00-0] C16H10 (44)
benzo[a]pyrene [50-32-8] C20H12 (45)
benzoperylene [11057-45-7] C24H14 (46)
coronene [191-07-1] C24H12 (47)
a
See Fig. 8.
b
See Fig. 6.

The total levels of three–six ring PAH in a petroleum VGO fraction range
from 2 to 25%. However, the concentrations of individual isomers that have been
quantified are generally expressed in parts per million (ppm). Although the most
abundant reported individual phenanthrene compounds appear to be the C1 –C3
derivatives, the average carbon number for the phenanthrenes in a VGO is
C32, indicating that the average side chain actually has 18 carbon atoms. The
analytical techniques that have been used to isolate individual compounds are
biased toward the least substituted aromatic structures, ie, the techniques that
facilitate identification of a few target compounds eliminate the majority of
multiring aromatic compounds.
Within petroleum certain aromatic structures appear to be favored. For
example, alkyl phenanthrenes outnumber alkyl anthracenes by as much as

(41) (42) (43) (44)

(45) (46) (47)

Fig. 8. Structures of fused-ring polynuclear aromatic hydrocarbons. See Table 4.


20 PETROLEUM, INTRODUCTION Vol. 1

100:1. In addition, despite the bias in separation methods, alkyl derivatives


appear to be more abundant than the parent ring compounds. For larger
ring PAH, patterns exist, but are more difficult to detect owing to analytical
resolution limits for the increasing number of isomers. A survey of crude oils
by ms suggests that chrysenes are favored over pyrenes. Even for larger ring
systems, the carbon numbers show that the parent compounds are missing or
present in trace quantities. The C3 derivatives are most prevalent.
The heterocyclics are significant contributors to the VGO fraction. Assum-
ing an average molecular weight of 320 for the VGO, sulfur levels of from 0.5 to
>2.5% would indicate that from 5 to >25% of the molecules contain a sulfur atom
(78). Results of API-60 show that 60% of the sulfur compounds are thiophenic
and the remainder sulfidic. Although there is some contribution from alkylaro-
matic and diaryl sulfides, no dialkyl sulfides were detected. In contrast, from
20 to 36% of the sulfur compounds existed in thiacyclane structures of one to
eight saturated rings. Gel permeation chromatography of the aromatic fractions
coupled with ms led to identification of >30 homologous thiophenic series
including 18 having more than a single sulfur atom. Although the distribution
into these groups varied widely among crude oils, benzothiophenes, and diben-
zothiophenes, having from zero to six naphtheno rings, were the prevalent
thiophenic forms of sulfur, accounting for 30–40% of the total. Many of the
homologous series appear to be S-analogues of the hydrocarbon PAH, ie, benzo-
logues of dibenzothiophene, but no specific compounds have been isolated and
identified unambiguously.
The nitrogen levels in crude oils are generally an order of magnitude
lower than those of sulfur. In the VGO range, the nitrogen-containing com-
pounds include higher molecular weight pyridines, quinolines, benzoquino-
lines, amides, indoles, and carbazoles; and molecules having two nitrogens,
ie, diaza compounds, and three or four aromatic rings are especially prevalent
(72). Typically, about one-third of the compounds are basic, ie, pyridine and its
benzologues, whereas the remainder are present as neutral species such as
amides and carbazoles. Although benzo- and dibenzoquinolines found in petro-
leum are rich in sterically hindered structures, hindered and unhindered
structures have been found to be present at equivalent concentrations in source
rocks. This has been rationalized as geo-chromatography in which the less
polar (hindered) structures moved more readily to the reservoir (56).
Oxygen levels in the VGO parallel the nitrogen content. Thus, the most
identified oxygen compounds are phenols and carboxylic acids, frequently called
naphthenic acids. These may account for from ppm to nearly 3% of a VGO. The
presence of numerous complex naphthenic and naphthenoaromatic acid struc-
tures in crude oils, especially immature forms, has been shown (86). Among
the different structures a number of specific steroid carboxylic acids have been
identified.
2.4. Vacuum Residua. The vacuum residua or vacuum bottoms is the
most complex fraction. Vacuum residua are used as asphalt and coker feed. In
the bottoms, few molecules are free of heteroatoms; molecular weights range
from 400 to >2000, so high that characterization of individual species is vir-
tually impossible. Separations by group type become blurred by the sheer
Vol. 1 PETROLEUM, INTRODUCTION 21

mass of substitution around a core structure and by the presence of multiple


functionalities in a single molecules. Simultaneously, the traditional gc and
ms techniques require the very volatility that this fraction lacks.
Classically, pentane or heptane precipitation is used as the initial step
for the characterization of vacuum resid. The asphaltenes, the material that
precipitates from the sample, range from 1 to 25% of vacuum resids. The soluble
portion (maltenes) are then fractionated chromatographically into several poorly
resolved classes for characterization. Asphaltene separations provide a simple
way to remove some of the largest and most polar components; however, the frac-
tions generated are still so complex that techniques such as nmr, titration, xps,
or ir are mainly used to provide some average compositional detail. Asphaltenes
are thought to be aggregates of complex molecular structures of MW <2000. The
aggregation mechanism is not well understood. Evidence exists for both charge-
transfer, among the numerous heteroatom functionalities, and j–j stacking of
aromatics within the asphaltenes. The pitfalls of using average techniques for
these complex systems have been discussed (73).
Additional general compositional information has been derived for
the vacuum resid, using more sophisticated techniques that differ in approach
toward isolating fraction. One approach uses functional group separations.
Ion exchange has been used as the primary fractionation of vacuum resid
into acids, bases, and neutrals (72,84). A second approach uses distillation
into carefully defined atmospheric equivalent boiling points (AEBP)
for the initial separation (73). Fractions generated by these separations
have been subjected to a battery of techniques including liquid chromatogra-
phy, field ionization mass spectrometry (fims), nmr, and elemental analyses.
In fractions distilled above 5658C, the paraffins contribute only <2% of the
10–20% saturates. The one- to six-ring naphthenes contribute the rest, having
average carbon numbers of 48–63. In the same study, aromatics shift progres-
sively from an even distribution of mono- to tetraaromatics toward one domi-
nated by five-ring types. As evidenced by the z-series data, many aromatics
bear one or more naphtheno rings plus sufficient side-chain carbon atoms to
bring the mean mol wt to nearly 800, ie, approximately 40 side-chain carbon
atoms for a five-ring aromatic core of mol wt about 300. However, the elemental
data show that 80% of the molecules could contain sulfur. Whereas most of the
sulfur is thought to be thiophenic, studies by x-ray absorption spectroscopy have
shown that in some vacuum resids it can exist in approximately equal amounts
as sulfidic and thiophenic forms (87).
For the 5658þ fraction the levels of nitrogen and oxygen may begin to
approach the concentration of sulfur. These elements consistently concentrate
in the most polar fractions to the extent that every molecule contains >2 het-
eroatoms. At this point, structural identification is not attempted. Rather,
average characterization techniques are used to confirm the presence of func-
tionalities such as acids, phenols, carbazole, and benzoquinoline, found in
lower boiling fractions. Several models have been proposed based on the
observed functionalities, apparent molecular weight, and elemental analysis
of the fraction. These models suggest that molecules having boiling
points greater than 5658C consist of multiple units similar to lower boiling
22 PETROLEUM, INTRODUCTION Vol. 1

components linked together with carbon and sulfur bridges rather than ever-
increasing fused-ring structures (88).
The Ni and V concentrated into the vacuum resid appear to occur in
two forms. From 10 to 14% of each of these two metals can be distilled in the
565– 7058C boiling range, where they exhibit the strong visible Soret bands
associated with the porphyrin structure. This tetrapyrrole structure (48, 49),
possibly derived from ancient chlorophyll, has been confirmed by a variety
of analytical techniques.

R′ R′ R′ R′

R R R R
N N N N
Ni V O
N N N N
R R′ R R′

R′ R′ R′ R′
(48) (49)

Because the metalloporphyrins can provide insights into petroleum matu-


ration processes, they have been studied extensively, and several families of
related structures have been identified. On the other hand, the bulk of the
metals are found in the heaviest fraction. Because those highly polar fractions
do not exhibit a Soret band, they are designated nonporphyrin metals.
However, a number of techniques suggest that the metals in these Soret-
inactive compounds still exist in molecules containing tetrapyrrole structures.
Whereas other structures have been proposed (89), none has been demon-
strated conclusively (90).

3. Nomenclature

Crude oils, complex mixtures of naturally occurring organic liquids, are difficult
to characterize in detail. Thus, many of the definitions used by the exploration,
production, and refining sectors of the petroleum industry to describe petroleum
and its products often lack precision. Even the term petroleum is poorly defined.
Although often used synonymously with crude oil, petroleum is also frequently
used to include natural gas (see NATURAL GAS) and even solid hydrocarbons. Defi-
nitions of materials are commonly given in terms of the processes used to obtain
them. Gasoline, for example, is the fraction of crude oil that distills between 15
and 2008C (60 and 3928F) (see GASOLINE AND OTHER MOTOR FUELS). Further compli-
cations arise because different parts of the petroleum industry use terms in dif-
fering ways. For example, wax may refer to material made up predominantly of
long-chain alkanes, or it may refer to esters of long-chain alcohols and acids (see
WAXES). Even the term hydrocarbons is used loosely indicating all the compounds
in crude oils, whether or not these include compounds of nitrogen, sulfur, and
oxygen.
Vol. 1 PETROLEUM, INTRODUCTION 23

In nature petroleum occurs in subsurface accumulations, or reservoirs,


called fields that may be made up of one or more pools. Petroleum compositions
vary widely and range from hydrocarbon-rich gases called natural gas,
through crude oil liquids, to high molecular weight solids known as reservoir
bitumen, residual oil, or tar. Petroleum is generated from kerogen, the high
molecular weight, insoluble organic material in source rocks. High subsurface
temperatures convert the kerogen to a petroleum-like range of compounds
called bitumen. Part of this bitumen moves out of the source rock, in the
process of expulsion or primary migration, and moves through permeable
rocks to accumulate in a reservoir (secondary migration). The petroleum
engineering procedures for bringing this petroleum to the surface are called
production.
Traditionally the unit of crude oil production has been the barrel (bbl),
equal to 42 U.S. gallons, 5.61 ft3, 158.8 L, or 0.159 m3. Increasingly petroleum
reserves are given in metric tons, but because one unit is a volume and the
other a weight, there can be no unique conversion factor for a material having
a range of densities. Fields of >500  106 bbl (79.5  106 m3) of recoverable oil
(>100  106 bbl (15.9  106 m3) in the U.S.) are called giants. Oil density may
be reported in any appropriate units, and although metric units are used it
is more common to report densities as degrees API (8API) or API gravity,
where API stands for American Petroleum Institute. The relationship
between density and API gravity is an inverse one defined by the following
relationship:

API ¼ ½141:5=specific gravity at 60 F  131:5

Water corresponds to an API gravity of 10; crude oils fall between 10 and
608API. The most common crude oil values are in the 35–408 range.
Other terms relating to physical properties include viscosity; refractive
index; pour point, ie, the lowest temperature at which the oil flows; flash
point, ie, the temperature at which the oil ignites; and aniline point, ie, the
minimum temperature at which equal volumes of oil and aniline are comple-
tely miscible. These are determined under defined conditions established by
ASTM.
Natural gas production is generally given in cubic feet or cubic meters
(1000 ft3 ¼ 1 Mcf ¼ 28.3 m3). Reserves of a trillion cubic feet (Tcf) (28.3 
1012 m3) or more form a giant gas field. Natural gas is called dry when methane
is the dominant hydrocarbon, and wet if it contains more than 4 L/100m3 of nat-
ural gas liquids (>0.3 gal/1000 ft3). When gas (or oil) has a bad odor owing to high
concentrations of hydrogen sulfide and volatile sulfur compounds it is called
sour. Sweet gas has no noticeable odor. For statistical purposes gas is commonly
reported as an equivalent amount of oil based on an equivalent heating capacity.
The conversion is normally made using 170 m3 (6000 ft3 ¼ 1 bbl) and leads to a
barrel of oil equivalent (boe).
Crude oils contain a wide range of hydrocarbons including straight and
branched chains, ring compounds, and aromatics, as well as more complex com-
pounds that incorporate nitrogen, sulfur, and oxygen (often called the NSOs),
and some nickel and vanadium. The straight-chain, normal alkanes, range
24 PETROLEUM, INTRODUCTION Vol. 1

from 1 to >100 carbon atoms. These are often called paraffins in the petroleum
industry because of the useful adjective paraffinic. Branched hydrocarbon chains
that are nominally built up from repeated isoprene units (2-methyl butane struc-
ture) are called isoprenoids or terpenoids, and the 19- and 20-carbon compounds,
named pristane and phytane, respectively, are frequently present in high con-
centrations. Isoprenoids also lead formally to saturated multiring structures.
Petroleum chemists use the obsolete word naphthenes for the compounds that
organic chemists call alicyclics. A better term, cycloparaffin, is used herein, lead-
ing to the adjective cycloparaffinic rather than naphthenic. Some of the charac-
teristic structures in this group can be directly related to molecules synthesized
by organisms. Whereas these have been called chemical fossils, it is more usual
to call them biological markers or biomarkers. Common examples include the
steranes and hopanes.
Aromatic hydrocarbons form a minor but important group of compounds
in crude oils and range from single-ring to multiring compounds. The latter
are called polycyclic aromatics (PAHs). Small aromatic molecules are environ-
mentally significant and BTEX is commonly used as an abbreviation for
benzene–toluene–ethyl benzene–xylenes. Multiringed compounds containing
both aromatic and saturated rings may be referred to in the older literature
as naphtheno-aromatics. The highest molecular weight fraction of crude oils
commonly contains ashphaltenes that are dark in color, NSO-rich, and very
aromatic.
Most crude oil is refined to provide useful products and the dominant
process is distillation (qv) (Table 5). Petroleum products produced by simple
distillation without the use of pressure, cracking, or catalysts are called straight
run. Residual material that has too high a molecular weight to distill forms
a residuum, often called by such names as asphalt. Naphtha (unrelated to
naphthenes) is a distillate of petroleum having a boiling range lower than
about 200 or 2608C (even occasionally up to 3508C). As a process intermediate,
naphtha includes the components used to formulate gasoline and the lighter
grades of fuel oils such as kerosene and diesel fuel oil. As a finished product,
naphtha usually denotes a more specific type of narrow boiling range material.
The terms naphtha and solvent may be used interchangeably. For example,
Varnish Makers’ & Painters’ (VM&P) naphtha has a range of 95–1508C. The

Table 5. Generalized Distillation Ranges for Products Obtained


During Crude Oil Refining
Temperature Carbon number
Product range, 8C range
gasoline 30–210 5–12
naphtha 100–200 8–12
kerosene and jet fuel 150–250 11–13
diesel and fuel oils 160–400 13–17
atmospheric gas oil 220–345
heavy fuel oils 315–540 20–45
atmospheric residue 450 30þ
vacuum residue 615 60þ
Vol. 1 PETROLEUM, INTRODUCTION 25

majority of streams within a refinery designated as naphthas are straight-run


materials, however the term can also be used for some cracked distillates.
A number of other words that have traditionally been used in the petroleum
industry are difficult to define precisely. These refer partly to specific boiling
ranges, but also to certain intended uses. Thus, gasoline boils lower than
naphtha, and kerosenes generally higher, but these terms are applied to pro-
ducts that are intended as fuels, rather than as solvents.
Gas oil is a product boiling slightly higher (235–4258C, or sometimes
wider) than kerosene. The main feedstock to the catalytic cracking units, it
received its name from use as an enriching agent in the production of city or man-
ufactured gas. It is often used as diesel fuel.
Cylinder oil is a viscous oil used for lubricating the cylinders and valves
of steam engines (see LUBRICATION AND LUBRICANTS). It is prepared from cylinder
stock. The product from cylinder stock, when filtered and processed, is bright stock.
Cycle stock (recycle stock) denotes any product that is recycled, that is,
taken back to an earlier stage in the process. The term cycle stock is also used
for the gas oil-like product of catalytic cracking.
The word distillate is occasionally used by petroleum chemists with a spe-
cialized meaning. Although anything that has been distilled is, of course, a dis-
tillate, the term distillate is sometimes used to denote distillate fuel oil as
opposed to residual fuel oil.
In the petroleum industry the International Union of Pure and Applied
Chemistry (IUPAC) system is in widespread use for naming organic compounds.
Two points, however, regarding group names and the prefix, iso, call for
comment.
3.1. Group Names. Although the IUPAC system is effective in denot-
ing any hydrocarbon, no matter how complicated, the system does not always
result in convenient terms for groups of compounds. Because hydrocarbons
having the same number of carbon atoms are apt to have boiling points within
a small range, it would be convenient to have words that would refer to C4, C5,
C6, . . . saturates, and C4, C5, C6, . . . monounsaturates, etc. The IUPAC system,
however, goes by the number of carbon atoms in the longest straight chain.
Thus, for example, the hydrocarbon referred to by the older systematic name
of isobutane, when named in the IUPAC system is 2-methylpropane. However,
for saturated aliphatic hydrocarbons, names such as butanes, pentanes, hex-
anes, etc, are taken as names in the older system, and therefore used as
group names.
The situation is different when naming the ethylenic hydrocarbons, because
the IUPAC has provided names such as propene, butene, and pentene, which are
different from the former names ending in -ylene. However, butenes/butylenes,
pentenes/pentylenes, etc, are not truly synonymous pairs, because the IUPAC
name goes by the longest-chain rule. Isobutylene, named 2-methylpropene in
the IUPAC system, would be included under substituted propenes, but not
under butenes. Similarly the three pentylenes derived from isopentane are
methylbutenes and not pentenes. For example, if it is necessary to denote the
group of five isomeric monounsaturated hydrocarbons C5H10, the term penty-
lenes denotes this group, whereas pentenes denotes a narrower group having
only two members, the two straight-chain pentylenes.
26 PETROLEUM, INTRODUCTION Vol. 1

3.2. The Prefix Iso. In names such as isobutane, isopentane, isobutyl


alcohol, and isoamyl alcohol, the prefix iso has a precise meaning, ie, one methyl
group attached to the next-to-terminal carbon atom and no other branch. This
notation is also frequently used by petroleum chemists to have a much wider
meaning, denoting nothing more than branched-chain. If both meanings persist,
any individual use of the prefix becomes ambiguous. Herein, an effort is being
made to use branched-chain or just branched consistently for the looser meaning
of iso, so that this prefix can be kept for denoting concisely what otherwise would
require some circumlocution. An exception is made for the well-established name
isooctane, which is 2,2,4-trimethylpentane [540-84-1].

4. Petroleum Resources

Petroleum resources are distributed widely in the earth’s crust as gases, liquids,
and solids. The products derived from these naturally occurring resources
are used principally as energy sources, although substantial volumes serve as
feedstocks in the chemical, plastics, and other industries. Petroleum resour-
ces are found as natural gas, as a variety of liquids that are usually classified
as normal or heavy crude oils, and as semisolid and solid substances such as
asphalt, tar, pitch, gilsonite, and many others. The petroleum resources consid-
ered here are those liquid crude oils that can be produced through a conventional
wellbore by current primary, secondary, or tertiary (enhanced recovery) produc-
tion techniques and those unconventional crude oils that may be captured and
converted into conventional sources of crude petroleum by advancing production
technologies.
No method has been devised to estimate with complete accuracy the
amount of crude petroleum that ultimately will be produced from the world’s
conventional oil and gas fields. Degrees of uncertainty, therefore, should be
attached to all such estimates. These uncertainties can be expressed in several
ways, the most important of which is achieved by dividing a resource into var-
ious categories. Several petroleum resources classifications have been pro-
posed, and a comprehensive discussion of them (91), as well as the definition
used in the assessment of the undiscovered resources of the United States
(92), have been provided. Seven commonly used categories of resources are
given here.
Resources represent the total amount (including reserves) of petroleum that
exists in a form and amount such that economic extraction is currently or poten-
tially feasible.
Reserves constitute the petroleum that has been discovered and can be pro-
duced at the prices and with the technology that exist when the estimate is
made.
Proved reserves are estimates of petroleum reserves contained primarily in
the drilled portion of fields.
Indicated reserves constitute known petroleum that is currently producible
but cannot be estimated accurately enough to qualify as proved.
Inferred reserves are producible, but the assumption of their presence is
based on limited physical evidence and considerable geologic extrapolation.
Vol. 1 PETROLEUM, INTRODUCTION 27

This places them on the borderline of being considered undiscovered, and the
accuracy of the estimate is very poor.
Subeconomic resources constitute the petroleum in the ground that cannot
be produced at present prices and technology but may become producible at some
future date at higher prices or by improved technology.
Undiscovered resources are estimated totally by geological reasoning; no
evidence through drilling is available.
To various degrees, the conventional petroleum resources in many
parts of the world have been classified according to such a system. In certain
regions, only estimates of proved reserves are made routinely, whereas in
the United States, Canada, and certain other regions, estimates are made of
volumes of petroleum in each of these categories. In the United States and
Canada, estimates for several of these categories (in particular, proved reserves)
are made each year by governmental agencies. For other categories, eg, undis-
covered U.S. resources, many estimates have been prepared by various agencies,
committees, panels, and companies. In a similar manner, estimates for several
resource categories have been made for most other countries and for the world
as a whole.
4.1. World Reserves. Most of the large volume of crude petroleum con-
sumed in the world is extracted from only a small fraction of the total number of
oil fields discovered. The concentration of crude petroleum in a few large fields is
a consequence of the interaction of the geologic processes that create and trap
petroleum. Even though commercial quantities of petroleum have been discov-
ered in many localities around the world, there are enormous volume differences
in fields present in a single region and in the total volume of petroleum present in
different regions.
By far the largest known concentrations of conventional petroleum reserves
are in the Middle East, particularly in Saudi Arabia, the United Arab Emirates,
and Kuwait. The largest concentration of reserves is in the Burgan field (10.2 
109 m3 (64.2  109 bbl)) in Kuwait (10), which contains about 68% of that coun-
try’s reserves. The second largest concentration of reserves is in the Ghawar
field (7.4  109 m3 (46.5  109 bbl)) in Saudi Arabia (93), which is about 18% of
that country’s reserves. In some regions, a large portion of the reserves may
not be contained in the largest field. However, the largest field usually contains
more than 10% of the total reserves of a region. More than 20,000 petroleum
fields have been discovered worldwide, and more than half of the world’s proved
reserves of ca 160.1  109 m3 (1006.8  109 bbl) of petroleum are contained in only
the 51 largest fields (93).
The Energy Information Administration estimates U. S. oil and gas
reserves, but does not systematically estimate worldwide reserves. They have
published a listing of international reserves as presented in two widely circu-
lated trade publications (Table 6). The world’s oil reserves are estimated to be
1.16 trillion barrels. The United States ranks 11th in the world for proved
reserves of crude oil. World reserves increased about 4% in 2003 owing to the
addition of former Soviet Union States.
Other large reserves other than Saudi Arabia’s are located in Iraq, whose
reserves are almost five times larger than that of the United States, Venezuela
and Canada have triple the amount.
28 PETROLEUM, INTRODUCTION Vol. 1

Table 6. International Oil Reserves as of December 31, 2003, 106 barrelsa


Rankb Country Oil & Gas Journalc World Oild
1 Saudia Arabiae 261,900f 259,400f
2 Irane 125,800 105,000
3 Iraqe 115,000 115,000
4 Kuwaite 99,000f 99,375f
5 Canadae 178,893g 4,957g
6 United Arab Emirates 97,800 66,230
7 Venezuelae 77,800 52,450
8 Russia 60,000 65,393
9 Libyae 36,000 30,500
10 Nigeriae 25,000 33,000
Top 10 total 1,077,193 831,305
11 United States 22,677 22,677
12 Qatare 15,207 27,352
13 China 18,250 15,509
14 Mexico 15,674 14,597
15 Algeriae 11,314 14,000
16 Norway 10,447 9,395
17 Brazil 8,500 10,602
18 Kazakhstan 9,000
19 Angola 5,412 8,800
20 Azerbaijan 7,000
21 Oman 5,506 5,700
22 Indonesiae 4,700 5,500
23 Ecuador 4,630 4,950
24 India 5,371 4,002
25 United Kingdom 4,665 4,300
Top 25 total 1,225,546 978,689
OPEC total 869,521 807,807
World total 1,265,812 1,051,477
a
Ref. 94.
b
Rank is based on an average of oil reserves reported in Refs. 95 and 96.
c
Ref. 95.
d
Ref. 96.
e
Number of OPEC.
f
Includes one-half of the reserves in the Neutral Zone.
g
Oil and Gas Journal Canadian oil reserves include heavy (low gravity) oil.

4.2. United States’ Reserves. The United States had the following
proved reserves as of December 31, 2003: crude oil, 21,891  106 barrels; dry
natural gas, 189,0444  109 ft3; and natural gas liquids, 7,459  106 barrels (94).
Table 7 listed the estimated annual oil reserve balances since 1993.
Crude oil proved reserves, reserve changes and production for individual
states for 2003 are listed in Table 8.
Proved reserves of crude oil decreased by 786  106 barrels in 2003. Total
discoveries are those reserves attributable to field extensions, new field discov-
eries, and new reservoir discoveries in old fields.
The majority of crude oil total discoveries were in new field discoveries in
the Gulf of Mexico Federal Offshore. New field discoveries accounted for 705 
106 barrels of crude oil additions, 702  106 were in the Gulf of Mexico. Operators
Table 7. Total U.S. Proved Reserves of Crude Oil 106 barrels of 42 U.S. gal 1993–2003

New
Revisions Net of sales reservoir Change
Net and and New field discoveries Total Estimated Proved from
Adjustments revisions adjustments acquisitions Extensions discoveries in old fields discoveries Production Reserves Prior
Year (1) (2) (3) (4) (5) (6) (7) (8) (9) 12/31 (10) Year (11)
1993 271 495 766 NA 356 319 110 785 2,339 22,957 788
1994 189 1,007 1,196 NA 397 64 111 572 2,268 22,457 500
1995 122 1,028 1,150 NA 500 114 343 957 2,213 22,351 106
1996 175 737 912 NA 543 243 141 927 2,173 22,017 334
1997 520 914 1,434 NA 477 637 119 1,233 2,138 22,546 þ529
1998 638 518 120 NA 327 152 120 599 1,991 21,034 1,512

29
1999 139 1,819 1958 NA 259 321 145 725 1,952 21,765 þ731
2000 143 746 889 20 766 276 249 1,291 1,880 22,045 þ280
2001 4 158 162 87 866 1,407 292 2,565 1,915 22,446 þ401
2002 416 720 1,316 24 492 300 154 946 1,875 22,677 þ231
2003 163 94 257 398 426 705 101 1,232 1,877 21,891 786
a
Ref. 97.
b
Revisions and adjustments ¼ Col. 1þ Col. 2.
c
Net of sales and acquisitions ¼ acquisitions  sales; NA ¼ not available.
d
Total discoveries ¼ Col. 5 þ Col. 6 þ Col. 7.
e
Proved reserves ¼ Col. 10 from prior year þ Col. 3þ Col. 4 þ Col. 8  Col. 9.
Notes: Old means discovered in a prior year. New means discovered during the report year. The production estimates in this table are based on data reported
on Form EIA-23, ‘‘Annual Survey of Domestic Oil and Gas Reserve’’ and Form EIA-64A, ‘‘Annual Report of the Origin of Natural Gas Liquids Production.’’
They may differ from the official EIA production data for crude oil, natural gas, and natural gas liquids for 2003 contained in the Petroleum Supply Annual
2003, DOE/EIA-0340(03) and the Natural Gas Annual 2003, DOE/EIA-0131(03).
Table 8. U. S. Crude Oil Proved Reserves, Reserves Changes, and Production by States, 106 barrels (42 U.S. gal) 2003a
Changes in reserves during 2003

Published New Reser-


proved Adjust- Revision Revisions New field voir discov- Estimated Proved
State and reserves ments increases decreases Sales Acquisi- Exten- discoveries eries in old production reserves
subdivision 12/31/02 (þ, ) (þ) () () tions (þ) sions (þ) (þ) fields (þ) () 12/31/03
Alaska 4,678 1 168 79 0 0 35 0 0 357 4,446
Lower 48 States 17,999 162 1,318 1,313 1,107 709 391 705 101 1,520 17,445
Alabama 51 1 4 0 0 0 2 0 0 6 52
Arkansas 49 9 2 1 4 2 0 0 0 7 50
California 3,633 18 190 107 58 49 9 0 0 246 3,452
Colorado 214 4 19 4 1 2 7 0 0 16 217
Florida 73 1 0 1 0 0 0 0 0 3 68

30
Illinois 107 22 11 5 0 0 0 0 0 10 125
Indiana 15 5 1 0 0 0 0 0 0 2 19
Kansas 237 19 31 13 4 2 5 0 0 34 243
Kentucky 27b 2 2 4 0 0 0 0 0 2 25
Louisiana 501 3 57 75 46 38 31 0 6 63 452
Michigan 61 12 14 16 10 21 0 0 0 7 75
Mississippi 179 9 20 9 15 1 0 0 0 16 169
Montana 288 6 15 8 2 5 42 0 0 19 315
Nebraska 18 0 1 0 0 0 0 0 0 3 16
New Mexico 710 5 65 69 18 15 38 0 0 59 677
North Dakota 342 8 20 24 1 7 31 0 0 30 353
Ohio 67 9 4 8 1 0 0 0 0 5 66
Oklahoma 598 3 80 49 36 21 26 0 0 55 588
Pennsylvania 12b 3 6 1 0 0 1 0 0 2 13
Texas 5,015 63 362 178 639 242 77 3 2 364 4,583
Utah 241 1 7 18 24 25 1 0 0 12 221
West Virginia 13b 0 2 1 0 0 0 0 0 1 13
Wyoming 524 24 29 36 23 36 5 0 0 42 517
Federal Offshore 5,009 8 376 686 225 243 115 702 93 515 5,120
Pacific 565 1 10 2 0 0 22 0 0 30 566
(California)
Gulf of Mexico 4,088 5 289 616 201 230 93 698 91 426 4,251
(Louisiana)
Gulf of Mexico 356 2 77 68 24 13 0 4 2 59 303
(Texas)
Miscellaneousc 15 1 0 0 0 0 1 0 0 1 16
U.S. Total 22,677 163 1,486 1,392 1,107 709 426 705 101 1,877 21,891
a
Ref. 94.
b
Indicates the estimate is associated with a sampling error (95 percent confidence interval) that exceeds 20 percent of the estimated value.
c
Includes Arizona, Missouri, Nevada, New York, South Dakota, Tennessee, and Virginia.

31
Table 9. International Petroleum Supply and Consumption Summary,  106 barrels/daya
Projections

2010 2020 2025

High A Low High A Low High A


Supply and Low world world oil world oil world oil world oil world oil
Consumption 2003 oil price Reference price price Reference price price Reference price
World oil price, 27.73 20.99 25.00 33.99 20.99 28.50 36.74 20.99 30.31 39.24
2003 $/barrelb
Production
(conventional)c
Industrialized
countries
U.S. (50 states) 9.09 9.55 9.61 9.95 8.86 9.21 9.76 8.36 8.82 9.60
Canada 2.25 1.82 1.83 1.91 1.58 1.60 1.70 1.54 1.57 1.68
Mexico 3.80 4.17 4.21 4.45 4.50 4.62 4.97 4.69 4.85 5.25

32
Western Europe 6.69 6.32 6.35 6.65 5.44 5.51 5.85 4.92 5.00 5.35
Japan 0.13 0.08 0.08 0.09 0.06 0.06 0.09 0.05 0.06 0.09
Australia and 0.66 0.95 0.96 1.01 0.86 0.89 0.95 0.83 0.86 0.93
New Zealand
Total industrialized 22.62 22.89 23.05 24.06 21.30 21.89 23.32 20.40 21.16 22.90
Eurasia
former Soviet Union
Russiae 8.34 9.89 9.98 10.64 10.63 10.90 11.87 10.76 11.11 12.20
Caspian Area 1.87 3.11 3.14 3.35 5.10 5.23 5.70 6.03 6.22 6.83
Eastern Europe 0.22 0.33 0.33 0.35 0.40 0.41 0.44 0.44 0.45 0.48
Total Eurasia 10.44 13.33 13.46 14.34 16.13 16.54 18.01 17.23 17.78 19.52
Developing
countries
OPECg
Asia 1.38 1.57 1.47 1.19 1.78 1.51 1.14 1.88 1.56 1.16
Middle East 20.95 26.08 24.45 19.84 38.02 32.37 24.44 46.42 38.47 28.62
North Africa 2.99 3.67 3.44 2.79 5.21 4.44 3.35 5.77 4.78 3.56
West Africa 1.98 2.51 2.36 1.91 3.67 3.13 2.36 4.52 3.74 2.78
South America 2.85 3.56 3.34 2.71 5.21 4.44 3.35 6.27 5.20 3.87
Non-OPEC
China 3.10 3.60 3.64 3.84 3.40 3.49 3.76 3.30 3.41 3.69
Other Asia 2.59 2.62 2.65 2.80 2.64 2.71 2.92 2.56 2.64 2.86
Middle Easth 1.81 2.22 2.24 2.37 2.51 2.57 2.77 2.69 2.78 3.01
Africa 2.94 3.72 3.75 4.04 5.31 5.44 6.00 6.36 6.56 7.31
South and Central 3.93 4.48 4.53 4.83 5.77 5.91 6.44 6.22 6.42 7.05
America
Total developing 44.52 54.05 51.87 46.31 73.52 66.02 56.52 86.00 75.57 63.93
countries
Total production 77.58 90.26 88.38 84.72 110.95 104.45 97.85 123.53 114.51 106.34
(conventional)
Productionh (non-
conventional)
U.S. (50 states) 0.00 0.00 0.00 0.01 0.00 0.00 0.19 0.00 0.00 0.31
other North 0.93 1.63 1.73 1.95 2.81 3.33 3.78 2.94 3.46 4.14

33
America
Western Europef 0.04 0.04 0.04 0.05 0.03 0.05 0.06 0.03 0.05 0.06
Asia 0.03 0.03 0.04 0.05 0.02 0.05 0.08 0.02 0.07 0.10
Middle Easth 0.03 0.03 0.12 0.12 0.03 0.21 0.28 0.04 0.25 0.35
Africa 0.21 0.16 0.23 0.26 0.18 0.28 0.41 0.19 0.32 0.47
South and Central 0.57 0.82 0.82 1.23 1.27 1.48 2.43 1.18 1.50 2.79
America
Total production 1.79 2.70 2.98 3.68 4.34 5.40 7.23 4.40 5.65 6.22
(nonconventional)
Total production 79.37 92.97 91.35 88.39 115.30 109.85 105.08 128.04 120.17 114.56
Consumptiond
Industrialized
countries
U.S. (50 states) 20.00 23.23 22.98 22.41 27.34 26.32 25.47 29.55 27.93 26.85
U.S. Territories 0.36 0.40 0.38 0.35 0.48 0.43 0.39 0.54 0.47 0.42
Table 9. ðContinuedÞ
Projections

2010 2020 2025

High A Low High A Low High A


Supply and Low world world oil world oil world oil world oil world oil
Consumption 2003 oil price Reference price price Reference price price Reference price

Canada 2.17 2.39 2.30 2.15 2.68 2.62 2.39 3.16 2.80 2.55
Mexico 2.02 2.43 2.36 2.25 3.24 2.88 2.51 4.13 3.48 2.93
Western Europed 14.22 15.00 14.72 14.22 16.15 15.45 14.82 16.60 15.71 15.06
Japan 5.58 5.91 5.70 5.34 6.46 5.69 4.95 6.93 5.84 4.99
Australia and New 1.04 1.29 1.27 1.23 1.60 1.54 1.49 1.77 1.69 1.63
Zealand
Total industrialized 45.38 50.66 49.72 47.95 58.15 54.93 52.01 62.68 57.92 54.44
Eurasia
Former Soviet Union 4.18 4.46 4.39 4.26 5.94 5.74 5.55 6.73 6.45 6.24

34
Eastern Europef 1.42 1.58 1.56 1.53 1.93 1.89 1.85 2.15 2.09 2.05
Total Eurasia 5.59 6.04 5.95 5.79 7.88 7.63 7.41 8.88 8.54 8.28
Developing
countries
China 5.54 7.84 7.63 7.27 11.75 11.06 10.45 13.79 12.79 12.08
India 2.19 2.85 2.79 2.69 4.65 4.37 4.09 5.75 5.29 4.92
South Korea 2.17 2.57 2.51 2.39 2.96 2.75 2.56 3.21 2.93 2.72
Other Asia 5.74 7.37 7.28 7.11 9.76 9.47 9.20 11.07 10.66 10.35
Middle Eastf 5.58 6.90 6.83 6.73 8.49 8.34 8.21 9.28 9.08 8.93
Africa 2.72 3.16 3.13 3.07 4.26 4.13 3.99 4.85 4.66 4.49
South and Central 4.69 5.89 5.81 5.69 7.68 7.48 7.30 8.89 8.61 8.40
America
Total developing 28.64 36.57 35.98 34.94 49.55 47.59 45.80 56.83 54.01 51.90
countries
Total consumption 79.60 93.27 91.65 88.68 115.57 110.14 105.22 128.39 120.47 114.62
OPEC production 30.60 38.04 35.79 29.48 54.88 47.21 36.75 65.79 55.13 42.44
non-OPEC 48.77 54.93 55.56 58.91 60.42 62.64 68.33 62.25 65.04 72.13
production
net Eurasia exports 4.84 7.29 7.51 8.55 8.26 8.92 10.60 8.35 9.25 11.23
OPEC market share 0.39 0.41 0.39 0.33 0.48 0.43 0.35 0.51 0.46 0.37
a
Ref. 98.
b
Average refiner acquisition cost of imported crude oil.
c
Includes production of crude oil (including lease condensates, natural gas plant liquids, other hydrogen and hydrocarbons for refinery feedstocks, alcohol
and other sources.
d
Western Europe ¼ Austria, Belgium, Bosnia and Herzegovina, Croatia, Denmark, Finland, France, the unified Germany, Greece, Iceland, Ireland, Italy,
Luxembourg, Macedonia, Netherlands, Norway, Portugal, Slovenia, Spain, Sweden, Switzerland, United Kingdom, and Yugoslavia.
e
Caspian area includes Other Former Soviet Union.
f
Eastern Europe ¼ Albania, Bulgaria, Czech Republic, Hungary, Poland, Romania, and Slovakia.
g
OPEC ¼ Organization of Petroleum Exporting Countries - Algeria, Indonesia, Iran, Iraq, Kuwait, Libya, Nigeria, Qatar, Saudi Arabia, the United Arab
Emirates, and Venezuela.
h
Non-OPEC Middle East includes Turkey.
i
Includes liquids produced from energy crops, natural gas, coat, oil sands, and shale. Includes both OPEC and non-OPEC producers in the regional
breakdown.
j
Includes both conventional and non-conventional liquids production.

35
36 PETROLEUM, INTRODUCTION Vol. 1

discovered 426  106 barrels in extensions in 2003. This total was 13% less than
the previous year. New reservoir discoveries in old fields were 101  106 barrels,
34% less than the previous year.
Reserve additions are the sum of total discoveries, revisions, and adjust-
ments, and sales and acquisitions. In 2003 there were 1,091  106 barrels of
reserves additions, 48% less than the previous year.
The smaller than average volume of net revisions and adjustments and
negative net of sales acquisitions of crude oil proved reserves significantly
reduced reserves additions in 2003. Crude oil net revisions and adjustments
were 257  106 barrels, 77% less than 2002.
Production was estimated at 1,877  106 barrels in 2003.
4.3. World Petroleum Supply and Consumption. Historically, the
world’s petroleum production pattern can be related to geologic, economic,
and political factors. In the past, many countries have had large excesses in
production capacity, whereas since the 1990s, only countries in the Middle
East, such as Saudi Arabia, Kuwait, Iraq, and the United Arab Emirates,
have, in the short run, enough excess capacity to expand production of conven-
tional crude petroleum in any significant manner. In the Middle East, production
of petroleum is over five times the region’s consumption. On a much smaller
scale, Africa produces far more petroleum than it consumes. See Table 9 for
world supply and consumption projected out of 2025.
4.4. Outlook. Petroleum displaced coal as the principal source of energy
in the United States by 1948 and in the world by 1965 (99). The spectacular growth
in consumption of crude petroleum in the world during the middle and late
twentieth century is directly attributable to the ease with which petroleum can
be discovered, produced, transported, processed, and utilized (see ENHANCED OIL
RECOVERY). This growth has been so rapid that as much crude petroleum
(55.5  109 m3 (349.4  109 bbl)) was taken from the ground between 1976 and
1992 as was produced during the entire previous 119-yr period (1857–1975).
This rapid rate of expansion in production and consumption, coupled with the
finiteness of the conventional petroleum resource base, has from time to time led
some analysts to conclude that world petroleum production will peak in the near
future (100,101). Other analysts who examine such data forecast impending global
crisis as crude petroleum consumption declines and coal reclaims its former posi-
tion as the principal source of fossil energy (102).
The key factor influencing the varying interpretations is that although
there is an enormous volume of petroleum resources in the ground through-
out the world, it is found in deposits that differ in quality and quantity from
country to country. Only a small fraction of these resources are conventional
petroleum resources (160.1  109 m3 (1006.8  109 bbl)) and are in the category
of proved reserves. An additional 93  109 m3 (585  109 bbl) of conventional
petroleum is estimated to be undiscovered in the world. This estimate is of undis-
covered petroleum resources that are economic to produce by means of normal
production technology.
Although the world is not running out of petroleum, it is difficult to sum up
how much is available in the short run, as well as in the longer run, in light of
various possible future political and economic developments. The effect of a vari-
ety of sociopolitical forces now at work will be to reduce the world’s consumption
Vol. 1 PETROLEUM, INTRODUCTION 37

Table 10. Estimated World Oil Resources, 1995-2025,  109 barrelsa


Region and Proved Reserve
country Reservesb Growthc Undiscovered Total
Industrialized
United States 22.7 75.0 83.0 181.7
Canada 178.9 12.5 32.6 224.0
Mexico 15.7 25.6 45.8 87.1
Japan 0.1 0.1 0.3 0.5
Australia/ 3.6 2.7 5.9 12.1
New Zealand
Western 18.2 19.3 34.6 72.1
Europe
Eurasia
Former Soviet Union 78.0 137.7 170.8 386.5
Eastern 1.4 1.5 1.4 4.2
Europe
China 18.3 19.6 14.6 52.5
Developing
countries
Central and 98.8 90.8 125.3 314.9
South America
India 5.4 3.8 6.8 16.0
Other 11.0 14.6 23.9 49.5
Developing Asia
Africa 87.0 73.5 124.7 285.2
Middle East 726.8 252.5 269.2 1,248.5
Total 1,256.8 730.1 938.9 2,934.5
OPEC 869.5 395.6 400.5 1,665.6
Non-OPEC 396.3 334.5 538.4 1,269.2
a
Ref. 98.
b
Ref. 95.
c
Ref. 103.

of energy, in particular of petroleum; such forces include clean-air regulations in


the United States and the ever-increasing rate of taxation of petroleum use in
many countries in order to raise general revenues. Countervailing forces, such
as the advancement of exploration and production technology, can be counted
on to expand the discovery and development of additional conventional petro-
leum resources in deep-water offshore regions and in hostile arctic climates.
Also, advancements in technology, eg, in three-dimensional seismic surveys
and horizontal drilling, will increase the inventory of proved reserves through
the process of field extensions. Over the next several decades, many billions of
cubic meters of conventional petroleum will be credited to the reserves inventory
through this field-growth process. Table 10 gives an estimate of world oil
resources for the years 1995–2025 (98).
Perhaps the biggest contribution that technological advancement in
petroleum production will make is bringing large volumes of unconventional
petroleum resources, eg, heavy oil and tar sands, into a viable economic realm
by lowering the unit cost of production. Compared to the inventory of conventional
petroleum reserves and undiscovered resources, the physical inventories of such
unconventional petroleum resources are extremely large; for example, the
38 PETROLEUM, INTRODUCTION Vol. 1

Athabasca tar sands in Alberta, Canada, are estimated to contain 360  109 m3
(2250  109 bbl) of in-place petroleum (103). This volume is equivalent to
the total inventory, ie, the combined cumulative production, reserves, and
undiscovered resources, of world conventional crude petroleum. In 1992,
however, only about 10% of total in-place petroleum resources was technically
recoverable (104).
Large unconventional resources of petroleum also occur as extra heavy
crude oils in the Orinoco belt, Venezuela, and in oil shale in the western United
States. Petroleum resources in the unconventional category, such as tar sands,
heavy crude oils, and oil shales, are located mostly in the Western Hemisphere,
as opposed to the conventional resources, which are located mostly in the
Middle East. Also, the in-place resources of these unconventional resources
are about twice as large as the in-place resources of conventional crude
petroleum. Although the recovery rates from these resources are low, improv-
ing technology may capture increasing volumes of these unconventional petro-
leum resources, thereby converting them into conventional petroleum
resources.
The irony underlying the current perceptions of the world’s petroleum
resource situation is that the world is not running out of combined petroleum
resources. However, the bulk of the world’s conventional petroleum resources
that are inexpensive to produce are concentrated in the Middle East, whereas
the much larger volume of higher cost unconventional petroleum resources,
such as tar sands and heavy oils, are located in Venezuela, western Canada,
and the western United States. The uneven distribution of the conventional
resources thus causes concern that in the short run political unrest in the Mid-
dle East could cause a catastrophic interruption of supply.
The world will never ‘‘run out’’ of petroleum, simply because there is so
much of it in the ground in so many different forms. However, the resources of
conventional crude petroleum are finite. These are the petroleum resources that
are very inexpensive to produce because they flow to the wellbore either directly
or by pumping after the application of standard well completion methods. There
is a more or less general agreement among analysts that the size of the inventory
of these resources is about 350  109 m3 (2200  109 bbl); the world is consuming
these resources at about 1%/yr. The primary question that faces the world is,
‘‘Has the pattern of exponential growth in consumption of petroleum that took
place between the end of World War II and 1973 become a relic of the past, or
could growth resume as world population continues to expand?’’ Analysis of
the pattern of world energy consumption shows that the world consumption
of crude petroleum may gradually increase even with increased efficiency in
the use of energy, simply as a result of population growth. However, these deve-
lopments could be dramatically altered by an increase in the price of energy
(105,106).
Another consideration of petroleum assessment analysts is whether, and
to what degree, the vast resources of unconventional petroleum in the world
can be captured by advances in petroleum production technologies, thereby
converting them into conventional sources of petroleum. It is a simple fact
that the in-place resources of petroleum in tar sands, heavy oils, and oil
shale can guarantee the future supply of petroleum for hundreds of years at
Vol. 1 PETROLEUM, INTRODUCTION 39

the current rate of consumption, provided they can be produced at competitive


costs.

BIBLIOGRAPHY

‘‘Petroleum Origin’’ in ECT 1st ed., Vol. 10, pp. 97–109, by B. T. Brooks, Consultant;
‘‘Origin of Petroleum’’ under ‘‘Petroleum’’ in ECT 2nd ed., Vol. 14, pp. 838–845, by S. P.
Ellison, Jr., University of Texas; in ECT 3rd ed., Vol. 17, pp. 113–118, by C. Barker,
University of Tulsa; in ECT 4th ed., Vol. 18, pp. 346–352, by Colin Barker, University
of Tulsa; ‘‘Petroleum Origin’’ in ECT (online), posting date: December 4, 2000, by
Colin Barker, University of Tulsa; ‘‘Petroleum Composition,’’ in ECT 1st ed., Vol. 10,
pp. 92–97, by A. N. Sachanen, Socony-Vacuum Oil Co., Inc.; in ECT 2nd ed., Vol. 14,
pp. 845–855, by R. H. Hunt and M. J. O’Neal, Shell Oil Co.; ‘‘Petroleum–Composition,’’
in ECT 3rd ed., Vol. 17, pp. 119–131, by J. J. Elliott and M. T. Melchior, Exxon Research
and Engineering Co.; in ECT 4th ed., Vol. 18, pp. 352–370, by Winston K. Robbins and
Chang Samuel Hsu, Exxon Research and Engineering Company; ‘‘Petroleum Composi-
tion’’ in ECT (online), posting date: December 4, 2000, by Winston K. Robbins and Chang
Samuel Hsu, Exxon Research and Engineering Company; ‘‘Nomenclature in the Petro-
leum Industry’’ under ‘‘Petroleum’’ in ECT 3rd ed., Vol. 17, pp. 110–113, by C. Barker,
University of Tulsa; in ECT 4th ed., Vol. 18, pp. 342–346, by Colin Barker, University of
Tulsa; ‘‘Nomenclature in the Petroleum Industry’’ in ECT (online), posting date: December
4, 2000, by Colin Barker, University of Tulsa; ‘‘Petroleum Resources’’ under ‘‘Petroleum’’
in ECT 2nd ed., Vol. 14, 856–867, by E. R. Heydinger, Marathon Oil Co.; ‘‘Resources’’
under ‘‘Petroleum’’ in ECT 3rd ed., Vol. 17, pp. 132–142, by L. J. Drew, U.S. Geological
Survey; in ECT 4th ed., Vol. 18, pp. 469–479, by Lawrence J. Drew, U.S. Geological
Survey; ‘‘Petroleum Resources’’ in ECT (online), posting date: December 4, 2000, by
Lawrence J. Drew, U.S. Geological Survey.

CITED PUBLICATIONS

1. V. B. Porfirev, AAPG Bull. 58, 3–33 (1974).


2. T. Gold, in D. G. Howell, ed., The Future of Energy Gases U.S. Geological Survey
Professional Paper 1570, Washington, D.C., 1993, pp. 57–80.
3. R. Robertson, Nature 199, 113–114 (1963).
4. E. E. Bray and E. D. Evans, Geochimica et Cosmochimica Acta 22, 2–15 (1961).
5. M. A. Bestougeff, in B. Nagy and U. Colombo, eds., Fundamental Aspects of Petro-
leum Geochemistry, Elsevier, Amsterdam, the Netherlands, 1967, pp. 77–108.
6. J. M. Hunt, Petroleum Geochemistry and Geology, W. H. Freeman, San Francisco,
Calif., 1979, 617 pp.
7. B. P. Tissot and D. H. Welte, Petroleum Formation and Occurrence, Springer-Verlag,
New York, 1984, 699 pp.
8. M. H. Engel and S. A. Macko, eds., Organic Geochemistry, Principles and Applica-
tions, Plenum Press, New York, 1993, 861 pp.
9. G. J. Demaison and G. T. Moore, AAPG Bull. 64, 1179–1209 (1980).
10. C. Barker, AAPG Continuing Education Course Note Series, American Association of
Petroleum Geologists, Tulsa, Okla., 1979, #10, 159 pp.
11. B. Tissot, B. Durand, J. Espitalie, and A. Combaz, AAPG Bull. 58, 499–506
(1974).
40 PETROLEUM, INTRODUCTION Vol. 1

12. M. T. Halbouty, R. E. King, H. D. Klemme, R. H. Dott, Sr., and A. A. Meyerhoff,


AAPG Memoir 14, 528–551 (1970).
13. B. D. Evamy and co-workers, AAPG Bull. 62, 1–39 (1978).
14. H. D. Hedberg, AAPG Bull. 52, 736–750 (1968).
15. G. T. Philippi, Geochimica et Cosmochimica Acta. 29, 1021–1049 (1965).
16. M. D. Lewan, in Ref. 8, 419–442.
17. C. Barker and L. Wang, J. Anal. Appl. Pyrolysis 13, 9–61 (1988).
18. P. Ungerer, in M. L. Bordenave, ed., Applied Petroleum Geochemistry, Editions
Technip, Paris, 1993, Chapt. 11.5, pp. 395–442.
19. P. Ungerer in B. Durand and F. Behar, eds., Advances in Organic Geochemistry,
Pergamon Press, New York, 1990, pp. 1–26.
20. N. V. Lopatin, Izv. Akad. Nauk SSSR, Seriya Geologicheskaya 3, 95–106 (1971).
21. D. W. Waples, AAPG Bull. 64, 916–926 (1980).
22. A. K. Burnham and R. L. Braun, in Ref. 19, pp. 27–40.
23. T. M. Quigley, A. S. Mackenzie, and J. R. Gray, in B. Doligez, ed., Migration of
Hydrocarbons in Sedimentary Basins, Technip, Paris, 1987, pp. 649–666.
24. B. P. Tissot, R. Pelet, and P. Ungerer, AAPG Bull. 71, 1445–1466 (1987).
25. P. Ungerer, J. Burrus, B. Doligez, P. Y. Chenet, and F. Bessis, AAPG Bull. 74, 309–
335 (1990).
26. D. D. Rice and G. E. Claypool, AAPG Bull. 65, 5–25 (1981).
27. V. I. Yermakov and co-workers, Akad. Nauk. SSSR Dokl. 190, 196–199 (1970).
28. G. E. Claypool, C. N. Threlkeld, and L. B. Magoon, AAPG Bull. 64, 1131–1139
(1980).
29. D. D. Rice and G. W. Shurr, AAPG Bull. 64, 969–987 (1980).
30. D. D. Rice, in D. G. Howell, ed., The Future of Energy Gases U.S. Geological Survey
Professional Paper 1570, Washington, D.C., 1993, pp. 583–606.
31. C. D. McAuliffe, AAPG Continuing Education Course Note Series, Amercian Associa-
tion of Petroleum Geologists, 1978, #8.
32. L. D. Price, AAPG Bull. 60, 213–244 (1976).
33. W. A. England and A. J. Fleet, Petroleum Migration, Special Publication No. 59,
Geological Society, London, 1991, 280 pp.
34. B. Durand, Organ. Geochem. 13, 445–459 (1988).
35. J. A. Momper, in Ref. 31.
36. P. Ungerer, E. Behar, and D. Discamps, in M. Bjoroy, ed., Advances in Organic
Geochemistry, John Wiley & Sons, Inc., New York, 1981, pp. 129–135.
37. B. P. Tissot and R. Pelet, Proceedings 8th World Petroleum Congress 2, 35–46
(1971).
38. W. A. England, A. S. Mackenzie, D. M. Mann, and T. M. Quigley, J. Geol. Soc.
(London) 144, 327–347.
39. J. A. Momper and J. A. Williams, AAPG Memoir (35), 181–191 (1985).
40. T. T. Showalter, AAPG Bull. 63, 723–760 (1979).
41. W. C. Gussow, AAPG Bull. 38, 816–853 (1954).
42. D. Gill, AAPG Bull. 63, 608–620 (1979).
43. B. H. Thomas, P. Moller-Pedersen, M. F. Whitaker, and N. D. Shaw, in B. M. Thomas
and co-workers eds., Petroleum Geochemistry in Exploration of the Norwegian Shelf,
Graham and Trotman, London, 1985, pp. 3–26.
44. C. W. D. Milner, M. A. Roger, and C. R. Evans, J. Geochem. Explor. 7, 101–153
(1977).
45. W. L. Orr, AAPG Bull. 58, 2295–2318 (1974).
46. C. Barker, in E. C. Donaldson, G. V. Chilingarian, and T. F. Yen, eds., Enhanced Oil
Recovery, Elsevier, Amsterdam, the Netherlands, 1985, Chapt. 2, pp. 11–45.
47. A. J. Lomando, AAPG Bull. 76, 1137–1152 (1992).
Vol. 1 PETROLEUM, INTRODUCTION 41

48. E. Lafargue and C. Barker, AAPG Bull. 72, 263–276 (1988).


49. S. E. Palmer, in Ref. 8, pp. 511–533.
50. J. Connan, in J. Brooks and D. H. Welte, eds., Advances in Petroleum Geochemistry,
Academic Press, London, 1984, pp. 300–330.
51. W. K. Seifert, J. M. Moldowan, and G. J. Demaison, Organ. Geochem. 6, 633–
643 (1984).
52. A. S. Mackenzie, in Ref. 50, pp. 115–214.
53. K. E. Peters and J. M. Moldowan, The Biomarker Guide: Application of Molecular
Fossils in Petroleum Exploration, Prentice-Hall, Englewood Cliffs, N.J., 363 pp.
54. L. B. Magoon and W. G. Dow, The Petroleum System—From Source to Trap, AAPG
Memoir 60, American Association of Petroleum Geologists, Tulsa, Okla., 1994,
644 pp.
55. R. R. F. Kinghorn, An Introduction to the Physics and Chemistry of Petroleum, John
Wiley & Sons, Inc., New York, 1983.
56. M. Yamamoto, Adv. Org. Geochem. 19, 389 (1992).
57. K. E. Peters and J. M. Moldowan, The Biomarker Guide: Interpreting Molecular Fos-
sils in Petroleum and Ancient Sediments, Prentice-Hall, Englewood Cliffs, N.J.,
1993.
58. J. G. Speight, Fuel Science Technology Handbook, Marcel Dekker, New York, 1990.
59. R. E. Jordan and J. R. Payne, Fate and Weathering of Petroleum Spills in the Marine
Environment, Ann Arbor Science Publications, Ann Arbor, Mich., 1980.
60. M. F. Ali, M. Saleem, and S. M. Ghazali, Hydrocarbon Process. 60(9), 129 (1981).
61. L. R. Snyder, Accounts Chem. Res. 3, 290 (1970).
62. B. J. Mair, Z. Ronen, E. J. Eisenbraun and A. G. Horodysky, Science 154, 1339–1341
(1966).
63. Characterization of the Heavy Ends of Petroleum API Project 60, American
Petroleum Institute, 1971; J. E. Dooley, D. E. Hirsch, C. J. Thompson, and C. C.
Ward, Hydrocarbon Process. 53, 187 (1974).
64. L. R. Snyder, Anal. Chem. 41, 314 (1969).
65. L. R. Snyder, B. E. Buell, and H. E. Howard, Anal. Chem. 40, 1303 (1968).
66. L. R. Snyder, Anal. Chem. 41, 1084 (1969).
67. D. M. Jewell and co-workers, Anal. Chem. 44, 1391 (1972).
68. M. M. Boduszynski, Energy Fuels 2, 597 (1988).
69. C. S. Hsu, K. Qian, and Y. C. Chen, Anal. Chim. Acta 264, 79 (1992); C. S. Hsu and
co-workers, Energy Fuels 5, 395 (1991).
70. P. Cleon, M. C. Foucheres, D. Cagniant, D. Severin, and W. Holstein, Chromato-
graphia 18, 543 (1985); J. M. Schmitter and co-workers, Org. Geochem. 6, 579
(1984).
71. K. Qian and C. S. Hsu, Anal. Chem. 64, 2327 (1992).
72. J. A. Green and co-workers, Analysis of Heavy Oils: Method Development and Appli-
cation to Cerro Negro Heavy Petroleum, NIPER-452 (DE90000200, 2 vols.), IIT Re-
search Institute, National Institute for Petroleum and Energy Research (NIPER),
Bartlesville, Okla., 1989.
73. K. H. Algelt and M. M. Boduszynski, Composition and Analyses of Heavy Petroleum
Fractions, Marcel Dekker, New York, 1994.
74. M. M. Boduszynski, Liquid Fuel Technol. 2, 395 (1984).
75. C. S. Hsu and K. Qian, Energy Fuels 7, 268 (1993).
76. B. J. Mair, Oil Gas J., 130 (Sept. 14, 1964).
77. H. T. Rall, C. J. Thompson, H. J. Coleman, and R. L. Hopkins, Sulfur Compounds in
Crude Oil, Bulletin 659, U.S. Bureau of Mines, Washington, D.C., 1962.
78. V. S. Aksenov and V. F. Kamyanov, in R.Kh. Freidlina, and A. E. Skorova, eds.,
Organic Sulfur Chemistry, Pergamon Press, New York, 1980, Chapt. 1.
42 PETROLEUM, INTRODUCTION Vol. 1

79. R. J. Hood, R. J. Clerc, and M. J. O’Neal, J. Inst. Pet. 45, 168 (1959).
80. J. S. Richardson and D. Miller, Anal. Chem. 54, 765 (1982).
81. St. Lanka and Sl. Hala, Erdoel Kohle Erdgas Petrochem. 11, 698 (1958).
82. S. W. Lee, S. Coulombe, and B. Glavincevski, Energy Fuels 4, 20 (1990).
83. R. C. Prince and co-workers, Environ. Sci. Technol. 28, 142 (1994).
84. J. F. McKay, J. H. Weber, and D. R. Latham, Anal. Chem. 48, 891 (1976).
85. A. Dipple, in C. E. Searle, ed., Chemical Carcinogens, ACS Monograph 173,
American Chemical Society, Washington, D.C., 1976, Chapt. 5, 245–314.
86. W. K. Seifert and R. M. Teeter, Anal. Chem. 42, 750 (1970).
87. G. N. George and M. L. Gorbaty, J. Am. Chem. Soc. 111, 3182 (1989).
88. I. A. Wiehe, Energy Fuels 8, 536 (1994).
89. R. H. Fish and J. Komlenic, Anal. Chem. 56, 510 (1984).
90. C. D. Pearson and J. B. Green, Energy Fuels 7, 338 (1993).
91. J. J. Schanz, Jr., 1978 Oil and Gas Resources—Welcome to Uncertainty, Resources
(Special Issue), Resources For The Future, Washington, D.C., 1978, 16 pp.
92. U.S. Geological Survey—U.S. Minerals Management Service, Estimates of
Undiscovered Conventional Oil and Gas Resources in the United States—A Part
of the Nation’s Energy Endowment, unnumbered report, U.S. Geological Survey
and the U.S. Minerals Management Service, Washington, D.C., 1989, 44 pp.
93. Technical data, Petroconsultants SA, Geneva, Switzerland, 1992.
94. U.S. Crude Oil, Natural Gas, and Natural Gas Liquid Reserves, 2003, Energy Infor-
mation Administration, U.S. Department of Energy, Washington, D.C.
95. Oil Gas J. 101, 46–47 (Dec. 22, 2003).
96. World Oil, 63 (Sept. 2004).
97. U.S. Crude Oil, Natural Gas, and Natural Gas Liquid Reserves, Annual Reports
1993–2002 DOE/EIA-0126, Energy Information Administration, U.S. Department
of Energy, Washington, D.C.
98. Annual Energy Outlook, Energy Information Administration, U.S. Department of
the Energy, Washington, D.C., 2005.
99. H. Enzer, W. Dupree, and S. Miller, Energy Perspectives, U.S. Dept. of the Interior,
Washington, D.C., 1975.
100. M. K. Hubbert, in Resources and Man, National Geological Sciences, National Re-
search Council, Washington, D.C., 1969, Chapt. 8.
101. D. H. Root and E. D. Attanasi, Natl. Res. Forum 4(2), 181 (1980).
102. W. S. Fyfe, M. A. Powell, B. R. Hart, and B. Ratanasthien, Nonrenewable Res. 2(3),
189 (1993).
103. World Petroleum Assessment, 2000, U.S. Geological Survey, Reston, Va.
104. N. Alazard and L. Montadert, Nonrenewable Res. 2(3), 197–206 (1993).
105. J. F. Brookout, Episodes 12(4), 257–262 (1989).
106. ‘‘The State of the Oil and Gas Industry,’’ Policy Analysis and Statistics, American
Petroleum Institute, www.api.com, accessed March 2005.

COLIN BARKER
University of Tulsa
WINSTON K. ROBBINS
CHANG SAMUEL HSU
Exxon Research and Engineering Company
LAWRENCE J. DREW
U.S. Geological Survey
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 43

EXPLORATION, DRILLING,
AND PRODUCTION
ENGINEERING
1. Origin, Deposition Mechanisms, Exploration, and Occurrence
of Crude Oil and Natural Gas

1.1. Crude Oil. Crude oil is the name given to all organic compounds
which are liquid under reservoir conditions. They can partly solidify at the surface
after expansion and cooling. Petroleum is a mixture of hydrocarbons in varying
proportions, which are accompanied by compounds of sulfur, oxygen, phosphorus,
and nitrogen. Minor constituents are porphyrins, ash-forming metal compounds
(usually sulfides of vanadium, nickel, copper, cobalt, molybdenum, lead,
chromium, and arsenic), as well as inorganic salts, hydrogen sulfide, and – as a
typical constituent of ‘‘crude oils’’ – water in varying amounts. The differing
mixing ratios preclude uniform chemical and physical properties. Elementary ana-
lyses show average carbon contents of ca. 79.5 – 88.5 wt %, hydrogen contents of
ca. 10 – 15.5 wt %, and impurity contents of up to a maximum of 5 wt %. Most
crude oils are optically active.
The essential ingredients of the petroleum mixture are alkanes, naphthenes,
and aromatics. The oxidation products naphthenic acids, terpenes, and phenols
are also present in some cases. According to the predominant constituents, a
distinction is made between paraffin-based crude oils and naphthene- or
asphalt-based crude oils. If both hydrocarbon types are contained in considerable
amounts in crude oil, it is termed mixed-based oil. The composition of crude oils on
a worldwide basis includes more than 30 % paraffins, at least 40 % naphthenes,
and ca. 25 % aromatics. Naphthene-based crude oils – that contain resinous and
asphaltic substances – are frequently found in the upper level of the Earth’s
crust, mixed-based oils in the middle, and paraffin-based oils in the deeper zones.
Near the surface, the crude oils lose part of their volatile constituents. Dur-
ing this, with chemical processes participating, more or less solid residues are
formed. Mainly ozokerite (earth wax) is formed from paraffinic crude oils and
natural asphalts from naphthenic crude oils.
For decades, petroleum has been mainly used as raw material for the pro-
duction of engine fuels, lubricants, and fuel oil. Because of the complex composi-
tions of crude oils, the chemical industry has developed many processes through
which petroleum has become the basis of numerous plastics. In 1990, more than
10 % of world petroleum production was used in petrochemistry.
Formation. Despite a century’s discussion with many contradictory opi-
nions and intensive research, the formation of petroleum can still not be satisfac-
torily explained. The discussion on the inorganic formation of crude oils continues
to this day. Russian experts, in particular Porfirev (56) do not exclude the fact that
petroleum is formed from inorganic matter (metal carbides and superheated
steam) even though it is generally agreed that the crude oils are predominantly
of organic origin. The investigations of potential petroleum source rocks in the
past decades have contributed considerably to this opinion. All deposits in the sedi-
mentary basins contain organic residues of terrestrial, limnic, fluvial, and marine

ß2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim


10.1002/14356007.a23_117
44 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

species, whose conversion under anaerobic conditions leads – as can be demon-


strated chemically – to bitumen (soluble in organic solvents) or kerogen (insolu-
ble). Such rocks are classified as source rocks according to the amount of organic
material. In clastic sediments, a petroleum source rock should contain at least
0.5 wt % and in carbonates at least 0.3 wt % of total organic carbon (TOC). A source
rock with a TOC value of more than 4.0 wt % is classified as excellent (57).
The organic substance originates from photosynthesis that started on a
major scale 2109 years ago. WELTE (58) has calculated that in the course of
the Earth’s history 6.41015 t organic carbon (Corg) has been produced. Only
18 % of the organic carbon, however, has contributed to the genesis of petroleum,
82 % is bound in the sediment as carbonate. The ratio of Corg bound to carbonate
to Corg bound to petroleum is 11 000 : 1. The organic origin of the petroleum can
also be deduced from the observation that crude oils are cracked at temperatures
above 200 8C to methane and graphite. The simultaneous occurrence of copper,
nickel, vanadium and molybdenum (that are also detected in the sapropel) in
crude oils is also in favor of its organic origin. Sporadic occurrences of small oil
reservoirs in volcanic or in magmatic-metamorphic rocks that have repeatedly
been quoted by the ‘‘inorganic’’ protagonists as counter-evidence are always in
connection with local elevations of the basement. It can be assumed that the
crude oil has migrated from deeper-lying source rocks in intercalated volcanic
strata in a higher position between sediments or by lateral migration where
there is juxtaposition with magmatic-metamorphic reservoir rocks.
Crude oils are formed principally from phytoplankton and bacteria; up to
the Silurian period from lower plants, eg, blue and green algae, and from the
Devonian period from higher forms that occur in increasing amounts. The fats
(lipids), proteins, pigments, and amino and nucleic acids formed from them are
the starting materials for bitumens.
A prerequisite for the formation of oil is that the organic material lying
at the bottom of the sea or lake must be sedimented in the absence of oxygen,
so that it cannot be decomposed and lead to the formation of kerogen. Such
conditions exist, eg, in eutrophic calm waters at great depth where the water
in the sediment no longer contains dissolved oxygen. A reducing hydrogen
sulfide medium prevails. Ideal prerequisites for such conditions are present
in sapropels in which the organic matter is decomposed, decayed by anaerobic
bacteria. The majority of the organic matter, however, is oxidized and recycled
as CO2.
In the course of the Earth’s history, source rocks have been formed in
various anoxic zones:

1) In nonmarine lakes, a uniform warm climate with little inflow of oxygen-


containing water is the precondition for source rock formation. Lake
Tanganyika is an example.
2) In closed inland seas with a positive water balance (water inflow exceeding
rate of evaporation), anoxic conditions arise in warm climatic conditions, at
a great depth, and excess water. The Black Sea with its deep zones free
from currents and its bays free from breakers is an example.
With a negative water balance, an oxygen excess arises, preventing
source rock formation. The Red Sea and the Dead Sea are examples of this.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 45

3) The ascent of marine currents from greater depths to the surface leads to the
substantial consumption of the scarce oxygen present in these currents as a
result of the biological productivity near the surface. From the organic resi-
dues that sink in the deep water lacking in oxygen, source rocks have been
formed. Examples of this occur off the Southwest African coast (Benguela
current) and off the west American coast (Humboldt current off Peru).
4) Source rocks are formed in the open ocean as a result of global climatic warm-
ing in association with the large transgressions. The many thick source rocks
in the Jurassic and Middle Cretaceous periods are attributable to this.

An important factor in source rock formation related to the transgressions


of the oceans is the transport of the organic residues from the energy-rich zones
into regions of calm water lacking in oxygen. This is supported by the adsorption
of organic material onto fine clay particles, which sink to the sea floor in calm
regions and compress the organic material content of the sediment. Anaerobic
conditions can prevail only a few centimeters below the surface of argillaceous
deposits. Since ca. three quarters of all argillaceous and carbonatic sedimentary
rocks contain organic carbon, the applied exploration usually assumes that not
only typical sapropels but all sediments, with a fairly high content of organic
matter, are to be regarded as potential source rocks for hydrocarbons.
A high sedimentation rate with a large proportion of coarsely clastic mate-
rial, on the other hand, dilutes the content of organic carbon and leads to source
rocks of low quality, if to any at all.
Actuo-geological observations on hydrocarbon groups from recent marine
deposits off the Californian coasts and comparative infrared spectroscopic
studies on oils in the Pliocene Los Angeles Basin indicate that at the time the
oil was formed in the Pliocene epoch, water and sedimentation conditions
must have prevailed that are similar to those in the offshore regions of the sea
nowadays (59).
Even in shallow depths down to 1000 m and temperatures up to 50 8C the
organic material embedded in the sediment is converted to kerogen. This
diagenetic process intensifies with further covering and sinking as well as
with temperature rise as a result of heating from the Earth’s interior. This pro-
cess, known as catagenesis, proceeds at depths of 1000 – 5000 m and tempera-
tures up to a maximum of ca. 175 8C (Fig. 1). During catagenesis the organic
material dehydrogenates, with the formation of methane and hydrogen as
well as C13 – C30 liquid hydrocarbons with increasing temperature. Oxygen
and hydrogen escape more rapidly than the carbon and nitrogen from the
source rock and so the relative carbon content increases. The H : C ratio
decreases with increasing temperature (57).
The conversion of organic material by pressure and temperature occurs
over geological ages. The biochemical processes, which are still not known in
detail, involve the action of reducing bacteria and the catalytic action of the
rocks. The conversion can be described as a function of temperature, time, and
pressure by modeling (see (60)). The temperature is the most important factor.
According to these ideas, the conversion of the organic matter included in the
source rock starts even at low temperature with the intensive contribution of
microbes and leads to the formation of proto-bitumens. Methane is formed in
46 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 1. General schemes of hydrocarbon formation (LOM = level of organic metamorph-


ism, according to Grunau, J. Petrol., July 1983) a) All kerogen types; b) Bacterial methane,
certainly kerogen Type III; c) Mainly kerogen types I + II.

minor amounts as a byproduct of the microbial conversion. Through further set-


tling, as a result of increased sediment cover or through heating as a result of
subsurface magma intrusions, the temperature of the source rock rises and, by
liquefaction of a part of the primary bitumen, leads to true petroleum formation.
The simultaneous compaction of the argillaceous source rocks, and the
recrystallization of the clay minerals into more stable forms with lower water
content cause large amounts of pore and crystal water to be released. With
this squeezed-out dehydration water, the newly formed oil migrates according
to the geothermal gradient – at depths ranging between ca. 1000 and 3000 m
(the so- called oil window) – possibly in dissolved form or as small droplets or
bubbles, out of the source rock and into the overlying or adjacent reservoir
rock (see (61)).
This primary migration of oil droplets from the source rocks to the petro-
leum reservoir rocks can proceed laterally and, in particular, vertically. The
low-molecular, possibly gaseous hydrocarbons function as a propellant. While
lateral migration probably occurs by capillary paths, vertical migration through
impermeable strata is mainly based on fine fissures that close afterwards (59). If
the thermal stress on the oils in source or reservoir rocks proceeds further, the
degradation and break-up of the high-molecular hydrocarbons by chemial
maturation leads to ever smaller and more stable compounds. Thus, a natural
cracking of the petroleum occurs. Accordingly, paraffin-based oils are found
mainly in deeper regions of the Earth’s crust. The maturation process concludes
with the conversion to methane.
During this maturation of the oils already present, a further part of the
organic matter that has remained fixed is possibly converted into petroleum.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 47

According to (60), the temperatures necessary for the genesis and maturation of
the petroleum are closely related to the geological time available for them.
Thus, according to Russian ideas, ca. 40 – 50106 a are required for the start
of petroleum formation at an earth temperature of at least 60 8C. The Californian
deposits mentioned, though, have formed at temperatures above 100 8C in only
1 – 2106 a (62).
Formation of Deposits. Deposit formation is the concentration of infi-
nite numbers of oil droplets which form in source rocks and are then squeezed
out of them to economically exploitable accumulations. Deposit formation is the
result of oil migration. The oil particles move during primary migration from
the argillaceous source rocks into the porous reservoir rocks, which usually
lie at a higher level. A considerable fractionation of the crude oils can occur
even during this movement as a result of selective filtration, adsorption, and
condensation (63,64).
Migration. In many cases, however, the true concentration of deposit
formation occurs only through migration in the bedding or horizontal secondary
migration, ie, through movement and accumulation within the permeable
reservoir horizons. With favorable facial conditions, lateral migration can be
very important for deposit formation. It can occur over distances of 10 km to a
maximum of 100 km. With changing facies, differential migration can occur, by
which the lighter hydrocarbons migrate to the higher reservoir beds. During the
course of this, the oils experience a certain buoyancy in relation to the formation
water present in the reservoir rocks because of their lower density. The oil is
therefore concentrated in the highest zones of the reservoir horizons, which
are sealed above by mudstones or salt beds. A gravity differentation of petro-
leum over edge water within the permeable reservoir rocks then occurs which
corresponds to the different densities of the individual components. Provided
gravitational separation is not prevented by capillary forces due to small pore
diameters, the edge water limits the oil column to the botom of the reservoir
(oil – water contact). If the oils are supersaturated with gas, a gas cap forms
over the petroleum.
Possible reservoir rocks are permeable and porous sandstones and lime-
stones. They have average porosities of ca. 10 % to a maximum of 40 %. The qual-
ity of reservoir rocks is determined, aside from their pore volume, by their
permeability, which is of decisive importance for the production capacity of wells.
Petroleum and gas deposits can occur in all geological formations, starting
from the Cambrian period, provided reservoir rocks are available that are accom-
panied or underlain by source rocks of appropriate maturity. Also the reservoir
rocks must possess trap structures from the overlaying of sealing horizons,
such as clay and salt beds. It is also important that the structural traps were
already present at the time of oil formation and survived during the subsequent
geological periods.
Reservoir Rocks. The quality of deposits is greatly affected by the type of
reservoir rock.
Wind-blown sands, for example, form the good gas reservoir beds
of the Rotliegendes. They spread from England to Poland over an area of
1000 – 3000 km2 with thicknesses of up to 200 m and maximum porosity
up to 20 %. Fluvial sands of meandering rivers can have net thicknesses of
48 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

6– 75 m. The Triassic sands of Prudhoe Bay with reserves of 2.8109 t oil belong
to this type. Both alluvial fans such as the Cretaceous sands of the Sirte Basin
in Libya and the beach and barrier sands parallel to the coastlines are good
reservoir beds. The latter include the Jurassic Piper field (United Kingdom).
The marine platform sands, which occur as transgressive or regressive
sequences, are equally good reservoir beds. Deep-water sands, occurring as
so- called deep-sea fans and turbidites, certainly spread over hundreds of kilo-
meters, but are of variable quality. This reservoir type includes the Pliocene
Sands in the Los Angeles Basin with a thickness up to 700 m and the Paleocene
Sands of the Viking Graben in the North Sea.
Carbonate reservoir beds can be of very differing quality. Thus reefs or reef
limestone debris can have excellent reservoir properties as in West Texas or
Alberta/Canada, with net thicknesses of 170 – 200 m, but they are mostly of lim-
ited extent and variable quality. The chalk forms only average-quality reservoir
beds. Mainly composed of coccoliths, the chalk is thin-bedded. The production
engineering of the Austin Chalk in West Texas, the Ekofisk field in the Norwe-
gian North Sea, or the reservoir of Reitbrook (Hamburg) is difficult. Owing to
their low permeability, these deposits can be extracted profitably only in conjunc-
tion with very fine fissuring.
Trap Structures. The type of trap structure encountered strongly
depends on the worldwide or regional tectonic processes in the Earth’s crust.
Movements at the plate edges lead to individual stress patterns with a variety
of structural phenomena as the result of forces of contraction and expansion.
The structures can, therefore, be of various sizes and shapes (65,66). The prin-
ciple trap structures, as listed below, can also occur in combination.
The types of structure possible as traps for petroleum and natural gas can
be of very different origin (see Fig. 2). There are anticlines (a), tiltings, faults (d),
salt intrusions, and discordant overlappings, and sometimes synsedimentary
facies differentiations which arose during the deposition of the beds. Permeabil-
ity barriers, arising as a result of secondary recrystallization and the formation
of new minerals in the pores, laterally seal the reservoir in such a way that a
special form of the facies reservoir is formed. This can often be observed in con-
junction with faults. The most important tectonic traps are the saddle-shaped
anticlines and flat, oval mound structures as well as monoclines originating
from antithetic faults. Reservoirs in association with overthrusts are usually
arched in the frontal zone and sometimes enrolled. They are known as rollover
structures. The first reservoirs discovered were found in association with salt
domes (d), where various kinds of structural traps are possible on the flanks
and the top of the dome. In the case of discordance reservoirs (c), porous horizons
are sealed off after tilting and partial removal of horizontally deposited younger
sediments. Other typical facies reservoirs are the reefs (b), which are usually
composed of corals, as well as traps which have arisen, eg, through the argilla-
tion of a sandstone with reservoir characteristics. In addition, there is a variety
of special forms and combined reservoir types. In all of these cases, it is common
that porous reservoir horizons in their top zone are covered and transposed with
sealing by clay rocks and salt layers.
Oil Exploration. The potential for petroleum and natural gas present in a
sedimentary basin must be proved by exploratory wells. The total costs of the
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 49

Fig. 2. Most important reservoir types and trap structures a) Under an updoming (anti-
cline); b) In a coral reef; c) Under overlapping strata (discordance); d) On a salt plug tank;
e) In a conformable downthrown fault; f) In a reverse downthrown fault.

wells–which rise exponentially as a result of ever increasing depths, extreme


environmental conditions, and continually increasing costs of equipment and
personnel–require intensive and expensive scientific preliminary exploration.
Aside from the geologists, the geophysicist and, with growing importance,
the geochemist, have made themselves equally valuable as exploratory
experts. No drilling project is commenced unless these experts, after having
made their profitability calculations, have given their approval. After calcula-
tion of the risk capital for the exploration phase, an analysis must be made of
the receipts and expenditure over the whole operating life of the expected oil
field: in other words, the driving force for exploration is the return on invest-
ment. Despite all care, the risk of sinking dry wells remains high. The ratio of
discovery wells to dry wells, from 1 : 8 to 1 : 6, since the 1950s has not greatly
improved. The reason is that exploration advances into new basins (or parts of
basins) and to greater depths. The ratio is very much more favorable, for
instance, in the Norwegian North Sea, where of about 300 exploration wells,
of which about half were wildcats (ie, wells drilled in regions not known to
be productive), more than 100 were oilbearing. The success rate was 1 : 3.5.
It is therefore necessary to reconstruct the conditions of formation of
the hydrocarbons in the chronological and regional framework and to check
whether potential reservoir rocks are in trap position for oil or gas. The
50 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

seismic reflection method is the central element of modern exploration and


crucial for almost every well. This method has been operated and continually
developed for more than three decades. It is supported by a series of geological,
geophysical, and geochemical methods that are used according to the degree
of knowledge or development of a region and according to the problems. In
an already well-known region – aside from studies on geological facies and
thickness of the rock sequences – investigations start with seismic reflection
measurements. In contrast, in areas which geologically are still little explored
or with very deep parts of a basin, these measurements usually conclude the
geophysical preparation.
Geological Investigations. In areas which are still little explored,
especially on land, geological surface investigations are used for the first
assessment and possible delimitation of prospects and nonprospects. The pre-
requisite is merely that the stratigraphic sequence outcrops at least partly
at the surface and is accessible. In the present state of exploration, the number
of less explored areas has greatly diminished. In the 1980s, the number of
sedimentary basin prospects is ca. 700. Of these, 350 are at least partly
explored. In the remaining sedimentary basins, preliminary geological investi-
gations have generally been carried out which permit conclusions on the
nature and thickness of the sediments. From satellite images, precise maps
are available for all parts of the world in which geologists enter their findings.
Geodetic methods are becoming increasingly less important. In regions
with poor vegetation, the satellite image also permits conclusions on the
geology. Later–when checked and consolidated during the course of geological
field surveys by a smaller or greater number of transverse profiles–these
conclusions suffice to enable statements on the thickness, facies, and age of
the rocks. If the results of the field surveys show that criteria for an oil pro-
spect are present, special examinations of rock samples are necessary in addi-
tion to stratigraphic investigations in order to identify source and reservoir
rocks and to provide evidence of clay as well as of evaporites being important
as the seal. The evidence of oil and gas as well as asphalt traces in the rock
samples alone is not sufficient to predict the presence of oil deposits. Asphalts
are residues of originally liquid bitumens that, as a result of the break-up of a
reservoir, have moved and oxidized near the surface. Asphalt traces can,
therefore, be an indication that oil has formed and migrated. Reservoirs,
however, can be left as fairly deep structures below thick, dense sediments,
even when younger reservoirs have been destroyed, eg, by uplift, tilting, and
erosion.
Geophysical Investigations. Potential Methods. The nonseismic
geophysical methods (potential methods) include electrical, magnetic, and gravi-
metric methods. They are used both individually and combined, for the first
exploration of sedimentary basins. Since gravimetric and magnetic measure-
ments can be made from an aircraft, they are a rapid and inexpensive tool of
petroleum exploration.
Magnetic Measurements. These measurements are based on the
phenomenon that rocks differ in their magnetic susceptibility. It depends on
their content of magnetic minerals such as magnetite or ilmenite. The crystalline
basement would cause no marked anomally, owing to the uniform distribution of
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 51

magnetic minerals within them. Sediments, on the other hand, differ in their
susceptibility according to the proportion of such minerals in the strata. The
changes of the Earth’s magnetic field are recorded with the aid of proton magnet-
ometers and give, for example, indications of the shape and depth of the crystal-
line basement or of volcanic or magmatic intrusive bodies. Aeromagnetics was
developed in World War II to detect submerged submarines. The measuring
instruments are accommodated in the wingtips of the aircraft or towed behind
on cables. In their basic version, the measuring instruments are coils through
which current flows. The applied magnetic field is altered by the terrestrial mag-
netic field and the resulting anomaly is recorded.
Gravimetric Measurements. Gravimetric measurements are used to
measure the changes in the Earth’s gravitational field. Because of the different
densities of rocks, beds of relatively low density, such as salt rocks, with
negative anomalies can be demarcated from rocks of higher density, such as
intrusives or quartzites, with positive anomalies. As a result it is also possible
to detect blocks of identical rocks that are displaced tectonically by faults.
The typical measuring instrument is the spring balance, whose spiral spring
shows a different extension according to the density of the rocks. The
extension is recorded and interpreted (Fig. 3). Aerial gravimetry, combined
with aerial magnetics, is a rapid and cheap method of reconnaisance survey
for the petroleum exploration of extended areas.
Gravimetric methods are increasingly used again for the solution of special
problems. Computer model calculations can be carried out which lead to consider-
ably improved conclusions. Thus gravimetric measurements are an important
means for detecting petroleum reservoirs at the flanks of salt domes; for the defini-
tion of the drilling and casing program and the allowed deviation for deep gas wells;
and also for the interpretation of seismic surveys near salt domes, where the reflec-
tions are disturbed by the salt, in order to clearly determine the salt flanks.
Geoelectric Measurements. Geoelectric measurements of the natural
Earth currents (tellurics) in the form of magnetotellurics have not fulfilled
the hopes placed in them. The numerous interference factors in the industrial
countries, such as electrical energy in cross- country transmission lines and
on railway lines are as disadvantageous as the large spread of the results.
Magnetotelluries are therefore only rarely used.
Seismic Methods. Since the 1970s seismic methods have become the most
accurate and most frequently used exploration methods. This is mainly due to
digital recording and the many processing and interpretation programs avail-
able. If in the past, seismics was a subsidiary science of geology, today it must
be ranked of equal importance alongside it. The seismic methods permit very
accurate conclusions on the stratigraphic sequence and depth of geological
beds. These methods measure the transit time of artificially generated elastic
waves. The waves have usually been generated by the detonation of explosive
charges placed in boreholes (Fig. 4). Apart from this, other energy sources also
are used nowadays. The vibroseis method, has become established whereby a
steel plate mounted below a truck is pressed by the weight of the truck onto
the ground. This transmits controlled vibrations of a given frequency that are
produced by a vibrator installed in the truck, to the subsoil. Four to six vibrator
trucks operate simultaneously. With this method, measurements can be
52 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 3. A) Principles of gravitational surveying; Points 1 – 9 represent observation sta-


tions where the gravity meter is installed. At these points the attraction of the gravita-
tional force is measured. B) Negative gravity anomaly over a subsurface salt dome with
associated oil accumulation; a) Salt dome; b) Cap rock; C) Positive gravity anomaly over an
igneous uplift or ‘‘buried bill’’ with an oil accumulation in the overlying sediment.

performed on roads and tracks and therefore in populated areas. It is also more
rapid than explosion seismics, and cheaper.
The elastic waves spread in all directions and at bed boundaries, eg,
between claystone and sandstone or sandstone and carbonate. They are partly
refracted and partly reflected to the surface by the echo sounder principle.
The velocity of these seismic waves through the various geological strata is a
measure of the type of rock and depends on the rocks’ elasticity and density
53
Fig. 4. Reflection seismic surveying technique with two reflecting horizons, A and B.
54 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

(ca. 1000–2000 m/s in loose rocks and up to 6000 m/s in compact rocks). The more
accurately the actual velocity in each bed is known, the more accurately can the
depth of the reflection be calculated. Computer model calculations provide very
useful approximations. There are two different methods: the reflection seismic
method and the refraction seismic method, the latter was used almost exclusively
until the 1930s. The refraction seismic method is still used nowadays only for
wide-area surveying of deep-lying horizons, but is also called in for the determi-
nation of sonic velocities in the rocks and to solve special problems.
Seismic Refraction Method. In the seismic refraction method a part of
the energy waves is refracted at the refraction horizon, similar to light being
refracted at the interface between water and air. Some of the refracted waves
then continue to run along the bed boundaries and in the course of this generate
other waves that are refracted again to the surface. The measurement is of the
transit time from the detonation at the energy center until the refracted wave
impinges on the geophones set up on the surface at distances of up to ca.
25 km. The seismic refraction method requires large explosive charges with the
geophones at a relatively large distance and enables a relatively large depth of
penetration.
Seismic Reflection Method. The seismic reflection method uses, in con-
trast to the refracted wave, the waves directly thrown back by the reflecting hor-
izons. Depending on the depth of the reflecting horizons, the waves return to
the Earth’s surface after increasing time intervals and are received by groups
of geophones systematically laid out on the ground. They are then converted to
electromagnetic impulses and recorded on magnetic tapes in complicated digital
recording equipment. Recording the seismic signals on magnetic disks – having
already been pretreated in the recording truck by amplification, filtering, and
balancing of the vibrations – enables corrections to the raw data and various
investigations of the signals with simultaneous suppression of the interference
energy (noise) to be carried out at the geophysical playback centers. From the
transit time found for the emitted and reflected waves, the depth of the reflecting
bed can be determined, provided the velocity of the rock beds passed through,
which often shows rapid regional variations, is known. The depth is then half
the product of velocity and transit time. Indications of the velocity of the seismic
waves come from detailed calculations of the refraction and reflection seismic
methods, from seismic well-logging, and directly from a geophone being lowered
on a wine line in the bore hole. The geophone is positioned at the boundaries of
rock units and receives the waves produced in the usual manner off the well
site.
Aside from the compression waves which have up until now been used
almost exclusively for scanning the structural conditions in the subsoil, shear
waves also are gaining increasing importance today. They enable conclusions
to be made on the petrographic nature of the rocks and their contents. The com-
bination of compression and shear waves permits at least approximations to the
porosity of a reservoir rock and to the pore content.
Seismic Measurements in Offshore Regions. The reflection seismic mea-
surements can also be used in offshore regions. The seismic method based on
explosives previously used, has been abandoned and replaced because of
the great environmental damage, especially where a large fish stock exists; at
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 55

the same time the airgun method has established itself. In this method highly
compressed air is suddenly released. With the large number of available record-
ing vessels, seismic companies not only carry out commissions to order but, in all
seas, carry out self-financed measurements which they then offer for sale to the
oil companies for reconnaisance purposes.
Three-Dimensional (3 D) Method. In exploration seismics, the line
method has hitherto been used almost exclusively. According to the exploratory
problem definition, individual profiles spaced at a distance of 500 – 2000 m from
each other are intersected by crosslines having the same spacing. The 3 D
method, which is almost routine in the development of proved deposits, is also
being used more and more in exploration. The groups of geophones are placed
at profiles with distances of 25 m and the waves are generated by explosives
fired from various sides. The digitally recorded results are processed and
corrected, as in the line seismic method, and subsequently transmitted to work
stations. Because of the small distances between profiles and the possibility of
assembling profile lines according to one’s own ideas via shot points of neighbor-
ing lines (random line), the interpretation of the reflection horizons leads to
increased precision. Because of the high cost of equipment – several thousand
geophones are required for the geophone installation – and the large number of
staff used – up to 350 members – the method is very expensive. It pays, however,
especially during development of a productive area, since as a result of the close
surveying, false interpretations and consequently false structural descriptions
are restricted or even avoided. Characteristics of a 3 D measurement are given
below:

Area of survey 300 km2


Surveying party size 140 employees
32 vehicles
1000 tracks
Geophones 14 400
Optical-fiber cable 71 500 m
Bore holes 15 076
Drilling meterage 157 183
Scanning points 480 000
Cost ca. $ 7.5106
Cost per km2 $ 25 000

Geochemical Investigations and Determination of Coalification on


Cutting Samples. Geochemical investigations and determinations of coalifica-
tion have become increasingly important. They are used as indicators of maturity
and prospectiveness for potential source rocks and for settling questions about
migration paths and the origin of crude oils and natural gases. These determina-
tions are carried out partly with rock samples from geological field surveys or
partly with rock sequences from boreholes. Coalification investigations serve to
determine the maturity of a source rock. If it is established that source rocks are
immature then any further exploration activity is often abandoned. Coalification
studies, preferably reflection photometric measurements, are applied to vitrinite,
which is present both in coal seams and in other sedimentary rocks. Depending
56 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

on the coalification gradients, which show regional variations, the coalification


and the vitrinite reflection increase with growing depth while the volatiles con-
tent simultaneously falls. From the peat to the graphite stage, the mean reflec-
tion values range from ca. 0.2 % – 10 %. According to the carbon ratio theory, the
formation and release of oil is restricted to the previously mentioned ‘‘oil window’’
with a coalification range of ca. 0.45 – 1.25 % vitrinite reflection (subbituminous
coal to high volatile bituminous A coal stage). Below the oil window light oils and
condensates can still be formed and, at greater coalification, natural gases. By
chemical investigations on source rock samples, fairly reliable conclusions on
their degree of maturity are possible. Optical parameters are also used. Here
the change of color of the dispersed organic material in the sediment on heating
is determined. The optical studies have little significance and should be used only
together with chemical or pyrolysis data. Pyrolysis, performed on a large number
of rock samples, is very important for the analysis of kerogen. The Rock – Eval
pyrolysis tester provides rapid determination of the type and degree of maturity
of the kerogen together with a direct determination of the hydrocarbon potential.
The results of the investigation can be used as geochemical logs directly at the
well for the determination of the hydrocarbon potential and the maturity. The
comparison of several boreholes enables excellent geological/geochemical correla-
tions in a region.
Carbon-isotope analyses of natural gas components are routine in hydrocar-
bon exploration. These makes use of the close connection between the 13C : 12C
ratio in methane and the maturity of the organic matter from which the methane
has been formed. It is possible, by isotopic analysis of gases, eg, gases from cut-
tings from boreholes, to decide whether these gases have been formed indigen-
ously in the horizons just drilled through or have risen from deeper strata.
With isotope investigations, for example, it was possible to demonstrate that
the natural gases accumulated in the large fields of Northwest Germany, the
Netherlands, and southern North Sea originate from the coal seams of the Penn-
sylvanian system (67). The carbon-isotope ratios of petroleum gas also enable
reliable estimates of the stratigraphic assignment of the corresponding source
rocks (68). So-called chemo-fossils enable conclusions both on the deposition med-
ium and on the ‘‘thermal history’’ of the sediment. These geochemical fossils, also
known as biomarkers, are molecular structures of organisms in the form of lipids
(eg, fats, waxes) and porphyrins, etc., that have undergone only small changes
during their transformation to kerogen. The main features of their chemical
structure are still present.
Geochemistry has also been concerned intensively with the primary migra-
tion of crude oils, which for a long time was difficult to interpret. It is assumed
that pressure built up in the sediment due to the decomposition of kerogen which
led to the formation of microcracks, through which the hydrocarbons could
escape from the source rock. They are transported with low-molecular hydrocar-
bons as propellant in the direction of the pressure gradient to more permeable
rocks. An important instrument of exploration in future could be the geochemical
modeling of sedimentary basins. The amounts of petroleum which have formed
in a basin, migrated, and finally concentrated in a reservoir are calculated using
computer methods. This calculation is carried out in close collaboration with the
geological development and not only takes into account the sedimentation and
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 57

Fig. 5. Geotectonic classification of petroleum occurrences.

facies distribution, but also the settling and the related rise of temperature, as
well as the uplift phases with erosion of reservoirs already formed. A three-
dimensional kinetic model has attracted attention (69). It is based on a set of dif-
ferential equations for energy and mass balances that describe the course of var-
ious geological processes. The physical parameters, such as porosity and
permeability, and the thermal sequences are selected from available lithological
data and, as knowledge increases, can be adjusted iteratively.
Occurrences, Production, and Reserves. The petroleum occurrences
are associated with the sedimentary basins of the Earth. Their distribution is
in no way statistical. Even their tectonic classification is very diverse and
shows that deposit formation is not associated with only certain types of basin
(Fig. 5). The largest petroleum occurrences lie in basins that have subsided con-
siderably for instance as a result of the continental drift – approximately from
the Jurassic to the Middle Tertiary period – and, owing to a lack of contact
with oxygen-rich waters of polar origin and to a simultaneously warm climate
with high, oxygen-reducing water temperatures, have preserved large amounts
of organic material in the anoxic medium. The reserves in older geological forma-
tions are relatively small. The reason may be that there were fairly small
amounts of organic material in the shallow seas of the epicontinental region dur-
ing the flooding by the worldwide transgressions. But the smaller reserves in
older formations than those of the Mesozoic era may also be the result of thermal
evolution (conversion of oil into gas), the destruction of the reservoirs or their
sealing horizon by tectonic events and erosion. The sedimentary basins with pro-
spects are nowadays distributed both over the continents and over the regions of
the continental shelves covered by the seas (Fig. 6). The methods of exploration
and development are in both cases the same, for offshore wells the means of
transport and technical installations are adapted to the water medium. The dif-
ference lies in the greater costs for offshore wells and production installations.
Apart from the classical offshore fields in Lake Maracaibo (Venezuela) and
in Baku (Caspian Sea) that were already developed in the 1930s with modified
onshore installations, offshore exploration and production achieved increasing
importance only in the 1960s. Technical developments then became necessary
58
Fig. 6. The sedimentary basins of the Earth with the most important production areas.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 59

that enabled deposits to be developed even at a depth of more than 1000 m water.
In 1983, on the west coast of France, a well was drilled to almost 2000 m at a
water depth of 1714 m. Such depths of water cause difficulties for production,
which then is only possible by means of an underwater completion. This
means that the production equipment is totally installed on the ocean floor
and must be maintained by divers. The deepest producing field is the Jolliet
field in the Gulf of Mexico. It produces from a depth of 535 m of water and is
equipped with a tension leg platform, ie, a floating platform that is fixed to the
ground by steel wire ropes. In 1992 the Snorre field in the Norwegian North Sea
also went into production by means of a tension leg platform at a depth of up to
350 m of water. Further production installations according to the tension leg sys-
tem are being prepared up to ca. 1000 m (Auger and Marlin fields). Owing to high
investment costs, the systematic recovery of hydrocarbons at great depths of
water, in particular on the continental slope, depends on the development of
the market price of crude oil.
Production and Reserves. The prices of crude oil collapsed in the second
half of the 1980s. OPEC saw a possibility of stabilizing the prices only by fixing
production quotas. These efforts, however, had no particular success. The control
of production by OPEC as well as other economic measures in producing coun-
tries outside OPEC, influence production in such a way that the annual produc-
tion data do not give a clear picture of the productivity of a producing country.
Rather, the cumulative production gives only an impression of the geological and
technical capacities (Table 1). For better comparison, the annual production
for 1989 for the region concerned is indicated. World oil production for the
year 1989 was 3.434109 t. In 20 years, it rose only by ca. 20 %. In view of world-
wide industrial development this is only a moderate increase that is attributable
in part to energy saving efforts which started with the 1973/1974 energy crisis.
The ranking of the producing countries for 1989 shows that the Middle East
states Iran, Iraq, Kuwait, Saudi Arabia, and the United Arab Emirates lead
the list of producing countries with a production of 922106 t, followed by the
Soviet Union with 668106 t, and the United States and Canada together with
525106 t. Following behind are South and Central America (380106 t), Asia/
Oceania (365106 t), Africa (350106 t), and Europe (223106 t). The production
from offshore fields was about 25 % of world oil production (70). The proved world
oil reserves rose to ca. 164.30109 t through development, in particular, of off-
shore fields. Compared with reserves published in the 1970s (92.40109 t) this
is an increase of > 80 %. If the probable reserves determined according to the
Gaussian distribution curve are added, then there are currently 240109 t petro-
leum to be expected. Compared with the production of ca. 3.4109 t in 1990, this
gives a ratio of 72 : 1. Thus the ratio of 30 : 1 existing for decades has more than
doubled. Even taking only the proved reserves into account, the ratio is about
50 : 1. There may be various reasons for this. One could be the rapid sequence
of discovery and development of large deposits – above all offshore. These include
new discoveries in East Venezuela and in the Brazilian Campos Basin, as well as
in the Russian North Caspian Basin and in Saudi Arabia. Large reserves have
recently been developed in the Tarim Basin (West China) and in the Papua Basin
of New Guinea, in which up until now no economic oil discoveries have been
known. Another reason is certainly the nature of the recording or evaluation of
60 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Table 1. The most important crude oil producing countries; their total production up to
December 31, 1989 and their resources (assembled from various international data)
Resources

Cumulative Proven and Prospective,


production, 106 t probable, 106 t 106 t
Canada 2 353 2 050 7 950
United States 24 722 7 200 7 280

North America 27 075 9 250 15 230


Argentina 795 350 475
Brazil 429 1 700 1 625
Columbia 477 650 930
Ecuador 238 400 415
Mexico 2 639 7 100 3 975
Trinidad 397 250 160
Venezuela 6 693 4 800 4 130
Others 397 400 750

Latin America 12 065 15 565 12 460

Bahrein 127 30
Iran 5898 11 750 3 020
Iraq 3 323 16 850 5 560
Kuwait 3 800 13 700 320
Neutral Zone 713 2 200 320
Oman 493 1 100 150
Qatar 684 650
Saudi Arabia 9 157 41 500 7 950
Syria 207 300
United Arab
Emirates 1 860 10 500 800
Others 207 750

Middle East 24 469 99 330 18 120

Algeria 1 510 1 450 250


Angola 238 450 250
Egypt 747 900 300
Gabon 223 225 200
Libya 2 607 4 908 950
Nigeria 2 067 2 800 1 120
Others 29 650 1 830

Africa 7 421 11 375 4 900

Australia 509 475 325


Burma 79 30 150
China 2 226 5 000 5 100
India 445 675 400
Malaysia 668 1 200 700
Indonesia 2 194 2 100 1 250
Others 32 100 825

Far East 6 153 9 580 8 750


Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 61

Table 1. (Continued)
Resources

Cumulative Proven and Prospective,


production, 106 t probable, 106 t 106 t

Norway 572 2 500 1065


United Kingdom 1 463 2 800 1 175
Other Western
Europe 699 400 300
Eastern Europe 1 097 300 270
Europe 3 831 6 000 2 900
Soviet Union 16 630 13 250 14 625
World 1989 9 764 164 300 76 985
1990 135 000
1991 134 630

the proved reserves. The new assessment is based on the newly proven or re-
evaluated reserves in the oil fields, larger extent of the deposits, or improved
oil recovery. At the World Petroleum Congress in 1990, a rise of the reserves
by 45109 t to more than 164109 t, was reported, ie, a rise of ca. 33 %. Of
this, ca. 32109 t are apportionable to the Middle East and only 3109 t to
North and South America. Other authors quote a larger or smaller rise than
this. The reason for this may be that, in addition to the newly developed reserves,
new evaluations of the fields have been made in which, apart from the proved
reserves, probable and possible reserves have been accounted for. The reserves
according to MASTERS et al. (71) listed in Table 1 represent, despite all uncertain-
ties, a reliable order of magnitude. The size of the increase in reserves from new
discoveries also reflects the number of successful exploration wells. As stated
above, the saturation of the petroleum market since 1985 has led to a drastic
reduction of exploration activity. If in 1984 in the western countries, 1250
seismic crews and 6500 drilling rigs were in use, their numbers had fallen in
1989 to 447 and ca. 1924, respectively (1990: 2057, 1991: 1896). A speeding up
of exploration depends on the general improvement and stability of prices and
availability of reserves in the important consumer centers. The further pros-
pects for success in petroleum exploration are limited to where the assumed
profitability for foreseeable future ends or when the natural deposit potential
of liquid hydrocarbons in the subsoil of the earth is exhausted. The two aspects
are closely interlinked, since the estimation of potential reserves and even the
definition of a deposit are determined by profitability factors. The ca. 77109 t
of conventional crude oil still to be expected according to the most recent esti-
mates must be supplemented by the reserves from very heavy oil deposits and
oil (tar) sands (see Tar Sands), which can be assumed to be at least 100109 t.
An accurate demarcation from the ‘‘conventional’’ crude oils listed in Table 1
is not possible, since MASTERS et al. (71) also take into account the reserves
between 108 and 208 API, which other authors include in the low-quality heavy
oils. The heavy oil and bitumen deposits are found to the extent of 87 % in Canada
62 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

(Athabasca, Cold Lake, Peace River), the CIS (Volga – Urals, East Siberia), and
Venezuela (Orinoco). They represent a future reserve that can only be used
when, because of the depletion of the conventional fields, so- called tertiary recovery
methods are standard, eg, hot water injection or steam injection, to extract the oil
from the deposits. The prices by then must have exceeded the limit of $ 50/bbl (the
oil price in 1992 was ca. $ 20/bbl).
Regional Distribution of the Reserves. Middle East – Arabian (Persian)
Gulf. The largest reserves are in the countries of the Middle East and in the
Arabian Gulf. On 1st January, 1990, proved reserves of ca. 99109 t were estab-
lished and probable ones of 19109 t calculated (see Table 1). The first economic
oil find in recent times in this region was made in 1908 with the discovery of the
Iranian Masjid-i-Sulaiman field. Many subsequent discoveries have been added.
At present, Saudi Arabia, with more than 41.5109 t of proved reserves, followed
by Iran with 11.75109 t, Kuwait with ca. 13.7109 t, and Iraq with ca.
16.85109 t are leading. The largest deposits in the Middle East include the fields
of Burgan in Kuwait, Ghawar, and Safaniya in Saudi Arabia, Kirkuk and
Rumaila in Iraq, and the deposits of Gach Saran and Agha Jari in Iran.
Geologically, the oilfields of the Arabian Gulf are situated on the shelf area
in front of the pre-Cambrian massif outcropping in the west, and south of the
Arabian peninsula in the form of intensively block-folded crystalline. Continen-
tal and shallow-water sediments are found deposited between the Cambrian and
the Cretaceous systems on the slowly sinking shelf. The shelf is divided by
numerous North – South trends that are linked with basement highs. North
and east of the shelf, including the Taurus – Zagros – Oman mountains there fol-
lows a geosynclinal or plate-edge region. This region also received thick tertiary
sediments and during alpidic orogenesis was folded by tangential forces and
divided into structures with a northwest – southeast direction. In accordance
with the asymmetric structure of this trough, the large deposits on the Arabian
side mainly lie in Jurassic and Cretaceous beds, and the offshore deposits mainly
in Cretaceous reservoir rocks. The Persian – Iraqi fields produce in the more dee-
ply settled parts of the trough from Cretaceous and Tertiary horizons. On the
Arabian peninsula, anticline-like updomings dominate structurally, whilst in
the Persian Gulf salt structures and bending folds, and in the foothills of the Ira-
nian Zagros range anticlinal and overthrust structures can be found. The most
recent exploration activities in Saudi Arabia proved deposits with several hun-
dred million tonnes of oil in sandstones of the Paleozoic era. The Permian
Khuff, itself a reservoir bed of natural gas deposits, serves here as capping of
stratigraphic and structural traps. The oil is of Silurian origin. The Paleozoic
era could provide further potential in Saudi Arabia.
Europe. The petroleum deposits of Europe, apart from the North Sea,
are fully developed and, with production falling, are already in the phase of sec-
ondary and tertiary recovery methods. If the reserves of all onshore producing
countries are combined, proved reserves of ca. 0.7109 t can be established
together with exploratory reserves of ca. 0.5109 t. The United Kingdom and
Norway possess proved reserves of ca. 5.325109 t from offshore fields, and a
further ca. 2.34109 t exploratory reserves are expected. The comparison with
the Middle East, however, shows the small importance of the European states
as oil-producing countries. Even so, in 1974 the Federal Republic of Germany
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 63

produced more than 6106 t oil and was able to cover ca. 30 % of indigenous
requirements. Of the 823106 t oil reserves originally present, 61106 t proved
and probable oil reserves remained after a total production of 225106 t (Decem-
ber 31, 1991). The recovery factor of 35 % of the original reserves means that
537106 t remain unrecoverable. The reserves in Italy and the Netherlands,
with developed quantities of ca. 110106 t in each case and prospective reserves
of ca. 330106 t in Italy, are – on a worldwide view – of equally low importance as
those of Romania, with 190106 t and a prospective reserve of 235106 t. All
other countries on the European mainland are globally considered insignificant
as producing countries. The oil deposits of northern Germany are bound to
sediments of epicontinental floodings which occurred during the Jurassic and
Cretaceous periods. The narrow NNW – SSE striking troughs filled with Jurassic
deposits are caused by halokinetic movements. The Rhine Valley Graben as a
minor oil region dates from the ‘‘rifting’’ (continental drift) related-to young
tectonic movements. The European deposits are distributed over the foredeeps
of the Pyrenees, the Alps, and the Carpathians, although the deposits in the
Aquitaine Basin, the Molasse Basin (Bavaria), the Carpathian Basin, and
the Po Basin are not very large. The oil occurrences in Sicily, Spain, and in
the Aquitaine Basin of France are linked with Triassic, Jurassic, and Cretaceous
carbonates. The intramontane Pannonian Basin and the Paris Basin should be
mentioned because of their historic importance for the exploitation of oil in
Europe. The Lower-Saxony Basin also is of historic importance, since the
deposits of Ölheim (1881) and Wietze (1859) were discovered here almost
simultaneously with Colonel Drake’s success in the United States.
North Sea. The North Sea has been one of the most attractive exploration
regions for the international oil industry since the start of exploration in 1964.
All large oil companies, often together with recently founded local partners,
have introduced their worldwide know-how into development of this logistically
difficult region. Great depths of water, strong currents, and extremely harsh
weather conditions require the use of specially developed drilling and production
installations. Although the technology used can nowadays be regarded as
advanced, the damage to the platforms during construction and operation
shows the limits for the prospecting and production of hydrocarbons. The first
success in the North Sea was the discovery of the Ekofisk field on the Norwegian
Shelf in 1968. Two important successes in the British sector of the North Sea
were the discovery of Forties in 1970 and of Brent in 1971. About 17 oil fields
are in production in the Norwegian North Sea, and in the British North Sea
more than 100 oil and gas fields including numerous small ones. Norway’s lar-
gest fields are Ekofisk/Eldfisk, Gulfaks, and Statfjord, which have up to now pro-
vided ca. 80 % of the Norwegian oil production. In the United Kingdom the
Forties, Brent, Magnus, Ninian, Fulmar, and Statfjord (British part) fields pro-
vide more than 40 % of the British oil production. The five Danish and seven
Dutch offshore fields are negligible compared with those of Norway and the
United Kingdom. Geologically, the fields are linked with the large graben that
open up from the beginning of the Mesozoic era in conjunction with the rifting
of the Atlantic plate. The most important element is the Viking Graben which
collapsed between the East Shetland Platform and the Norwegian Vestland
Bogen. It runs for a length of ca. 455 km and has a width of ca. 50 km in the
64 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

south and 180 km in the north approximately on the midline between Norway
and the United Kingdom. In addition, there are special basins, some with
thick tertiary sediments, to which the oil is bound. The deposits are found in
sediments of the Triassic, the Jurassic, the Upper Cretaceous, and the Paleogene
periods. The traps are mostly the tilted fault blocks typical of rifting, that are
sealed by transgressively overlapping fine-grained Upper Cretaceous sediments.
The development of hydrocarbons is limited to the north by the 62nd degree of
latitude as a result of Norwegian legislation. Seismic measurements and specifi-
cally located wildcats have provided evidence that in the northern North Sea
(Tromsø, Haltenbanken) and in the Barents Sea, deposits both of oil and of nat-
ural gas can be expected. The time for specific development of these fields will
depend on the development of the reserves already developed and also on the
oil price. Even today, owing to expensive technology, a price in excess of $ 20/
bbl and tax relief by the producing countries is necessary for the continuation
of exploration in the North Sea. In 1991, the daily production of all North Sea
fields reached 192106 t.
Commenwealth of Independent States. Industrial oil recovery in the
territory of the CIS had already started at Baku in the middle of the last century
and reached its first peak around 1900, when North American production was
briefly surpassed. After a period of stagnation, a rapid upward development
started in the former Soviet Union after World War II with the development of
new oil reserves. The proved resources on January 1, 1990, were 13.2109 t, and
14.6109 t of exploratory reserves were expected. Accordingly, the total reserves
are more than 200 % higher than those of the United States. Until 1918, almost
the whole production came from the Caucasus mountains. In the 1950s the
Volga – Urals region and the Timan – Petschora Basin, adjacent to the north,
were developed. In addition, the discoveries in the North Caspian Basin
(Emba) and in the Mangyschlak trough of Turkmenia developed very rapidly,
but have passed their peak. In the 1960s and 1970s, the extended deposits
east of the Urals in the West Siberian Basin were discovered. Meanwhile 96 %
of the oil developed in West Siberia is produced in the Tjumen province. In the
southern provinces of Novosibirsk and Omsk, exploration has given disappoint-
ing results. In almost all oil provinces, as a result of technical deficiencies and
lack of oil field equipment, production is falling or remaining below expectations.
Proved reserves could be mobilized by the use of modern technology, such as the
horizontal drilling process in many reservoirs of low permeability. The opening of
the CIS to the West leads to the expectation of a strong involvement of interna-
tional oil companies which, with their know-how and their modern technology
can bring oil exploration and exploitation to a reliable high standard. There
are also justified expectations of the development of important new reserves in
Turkmenia and East Siberia, where only a small production of oil has been
achieved so far. Large oil discoveries in the pre-Caspian Depression promise
interesting exploration possibilities. In the recently discovered Thenguise,
Astrakhanskoe, and Karatshaganakskoe fields, oil is produced from carbonates
and clastics of the Paleozoic era at depths of more than 4000 m below a Permian
salt cover. With an area of 500 000 km2 and a sedimentation thickness of almost
20 000 m, good possibilities also arise here which may increase the number of
reservoirs in the CIS substantially. Geologically, the deposits in the Caucasus
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 65

mountains, the Caspian Sea, Galicia, the Fergana depression, and on Sachalin,
etc., are linked above all with Neogen horizons in prealpine depressions and
innermontane troughs. The deposits of the Volga – Ural area lie in Paleozoic
depressions of the Eastern – European Platform. The main reservoir rocks
here are Devonian and Permian reef limestones as well as sandstones of the Car-
boniferous period. In the large and oil-rich Timan – Pechora Basin, the reservoir
systems belong to Upper Devonian sandstones and chalks of the Lower Carboni-
ferous period. Mainly Permian and Mesozoic reservoir horizons are oil-bearing in
the deposits of the North Caspian Depression, in Kazakhstan, and Turkmenia.
In the deposits of the Western Siberian Platform, the hydrocarbons are assigned
to the central part of the Lower or Middle Cretaceous period. In the Molasse
Basin of the eastern Ural foreland, oil and gas deposits are linked with Upper
Jurassic sandstones and calcarenites.
Africa. In 1990, the proved and probable oil reserves of Africa were about
11.3109 t, and the exploratory ones ca. 5.0109 t. The resources are concen-
trated in Libya, Algeria, and Nigeria. There are further deposits of noteworthy
size in Egypt, Congo, Angola, Gabon, and Tunisia. The African continent was
long considered to have few prospects for hydrocarbons. The breakthrough
came in the 1950s with the almost simultaneous discoveries of the first large
deposits in Algeria and the Niger delta. The large discoveries in Libya followed
in the 1960s. Geologically, the oil deposits of Algeria and the smaller occurrences
in West Libya which are not yet producing, are linked to the Paleozoic basin
between the Precambrian Massifs of the mid-Sahara and the Atlas range. Fairly
large amounts of oil are found in the Neogene basins of the Syrte in Central
Libya and in the Niger delta. There are also older oil-producing strata, mostly
in the Cretaceous system. At the western edges of the African shield, between
several stable cratons, depression fields have been formed, on which marine
transgressions deposited thick sediments in the Mesozoic and Tertiary eras. In
some cases, between the Permo-Triassic system in Morocco and the Cretaceous
system in Gabon, salt deposits occurred that led, possibly in conjunction with the
continental drift, to diapirism. The basins not only cover parts of the current
mainland but reach far into the Atlantic. The coastal basins range as follows
from north to south: Spanish Sahara, Senegal – Mauritania, Ivory Coast,
Dahomey, Nigeria, Gabon, Congo, and Angola. In Eastern Africa, of the graben
systems which are largely linked to the rifting, only the Red Sea graben with
Miocene salt deposits and hydrocarbons below is of importance, as in the Gulf
of Suez, which is the only area of East Africa with fairly large reserves.
Libya, with 4.9109 t proved and probable reserves and ca. 1109 t pro-
spective, has the largest oil reserves of Africa. Cumulatively, ca. 2.6109 t oil has
been produced in Libya up to now. Geologically, Libya is divided into three large
basins: in the west the Murzuk basin, in the central part the Syrte basin, and in
the east the Kufra Basin. The oil originates for the greater part from reservoir
rocks of the Cretaceous (sandstones) and the Lower Tertiary (reef limestones)
systems, but also from the Cambrian and Ordovician systems of the Syrte
Basin. Libya’s only producing offshore field Bour, with a daily production of ca.
13 000 t (1991), has little importance. The development of hydrocarbons in North
Africa began with the discovery of the Zelten (now Nasser) field in 1959. Nasser
produces from Lower Eocene limestones of the Syrte Basin. With original
66 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

reserves of about 400106 t, it belongs, like Amal and Sarir, with ca. 1.2109 t, to
the largest fields of Libya. Libyan oil is low in sulfur and in great demand as a
light crude in European refineries.
Nigeria is the second largest producing country in Africa. The proved and
probable reserves are indicated as 2.8109 t and the prospective reserves as
1.1109 t oil. The fields are in the Niger Delta Basin that is filled with ca.
12 000 m of clastic rocks of the Tertiary period. In Nigeria, 15 giant oilfields
(according to U.S. definition, fields with more than 500106 bbl, ca. 80106 t)
have been proved, of which almost half lie offshore.
Algeria, with proved and probable reserves of ca. 1.4109 t, occupies the
third position among the producing countries of Africa. The prospective reserves
(ca. 240106 t) are quite insignificant. The oilfields lie in the Polignac Basin, near
the Libyan border, and in the Ghadames Basin. The fields in Polignac produce
from reservoirs in the Carboniferous, Devonian, and Cambro – Ordivician sys-
tems. In the Ghadames Basin likewise there are sandstones of the Cambro –
Ordivician, that are discordantly overlaid with evaporites.
North America. Geologically, the North American oil deposits are linked
with the large sedimentation region that appeared repeatedly since the Paleozoic
era. This basin forms the foredeep to alpine-type mountains in the west (Rocky
Mountains) and flattens to the east towards the basement area of the Canadian
shield. In the south, this sedimentary basin expands and becomes the thick Ter-
tiary Gulf basin. A lateral arm of the trough zone with Paleozoic rocks follows the
south-east edge of the Canadian shield to the mouth of the St. Lawrence river. In
accordance with this geological structure, the deposits in the more deeply sunken
western sections of the trough appear in Tertiary and Cretaceous horizons, and
in the higher-lying eastern marginal areas (mid-continent, eastern states with
Appalachians) in sandstones and limestones of the Paleozoic era. The deeply sun-
ken Gulf coast has its main deposits in Cretaceous and Tertiary strata. In the
Californian fields, the oil comes from Tertiary and Cretaceous horizons. The
new discoveries in Alaska occur mainly in Permo-Triassic sandstones. Other
oil discoveries, scarcely developed until now, in the arctic expanse of Canada
indicate several very promising basins with possibilities of reservoirs from the
Paleozoic era to the Tertiary period.
United States. On January 1, 1990, the United States had available
proved reserves of ca. 7.2109 t and prospective reserves of ca. 7.3109 t. Follow-
ing the producing countries of the Middle East and the CIS, the United States
are in the third position. The same ranking applies to the production in the
year 1990. The cumulative production of 24.7109 t crude oil was achieved
from 597 320 production wells. (The Middle East states produced ca. 26.4109 t
from only 7061 wells). The oil resources in the offshore region of Texas,
Louisiana, and California were about one quarter of the total reserves. Due to
results of exploration in the Gulf of Mexico an increase of the offshore reserves
can be anticipated from new discoveries in the deep water region, but this will
scarcely lead to compensation for the fall in reserves determined by production.
Before Texas, with the coastal strip on the Gulf of Mexico, advanced in 1930 to
become the leading petroleum province, the position of the leading producer was
occupied by Pennsylvania until the turn of the century, then by California, and
later by the mid- continent. On January 1, 1990, the order according to daily
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 67

production was Texas (ca. 306 179 t), Alaska (281 948 t), Louisiana (171 145 t),
California (152 840 t), and Oklahoma (48 900 t).
Canada. Only after World War II did Canada rise to become an impor-
tant producing country as a result of discoveries in Alberta. On January 1,
1990, the Canadian reserves amounted to ca. 2.1109 t, 7.9109 t of prospective
reserves were reported. The daily production of ca. 190 000 t is obtained from
39 159 production wells. The reserves are declining. However, numerous discov-
eries in arctic regions cannot be developed and therefore cannot be evaluated.
More than 80 % of the oil production occurs in Alberta. The hydrocarbons are pre-
sent in the foothills of the Rocky Mountains and in the Alberta Plain. The reser-
voirs for oil, as for natural gas, are present in highly stressed folds, mostly bound
to overthrusts, in limestones of the Upper Carboniferous period, more rarely in
beds of the Devonian or the Triassic periods. In the Alberta Plain, Paleozoic
reef limestones are the reservoir rocks for the oil. Occasionally Cretaceous sand-
stones are the reservoir rocks, overlaying the Paleozoic discordant.
Latin America. The most important deposits in Latin America are those
of Venezuela and Mexico; Trinidad, Ecuador, Argentina, Peru, Colombia, and
Brazil follow some way behind. The main deposits are in the foretroughs of the
cordilleras and their branches, and are bound to sediments from the Cretaceous
to the Neogene periods. In addition, fairly small coastal basins in Brazil and Argen-
tina are productive. The resource figures have increased due to new discoveries in
Venezuela and more recently in Columbia but most of all to the oil accumulations
developed in the jungle basins of Ecuador and Peru. Through the development of
the Marina – Austral Basin between Tierra del Fuego and the Falkland Islands,
the reserves of Argentina have increased considerably. The new reserves are not
yet in production because of their geographical position and high gas : oil ratio. The
high proportion of prospective reserves in Argentina of ca. 477106 t, as against
proved reserves of only ca. 350106 t still points to a good exploration potential.
As a result of the large oil deposits in the Bay of Maracaibo and on the Gulf
Coast of Mexico, the submarine reserves of Latin America reach 40 % of the
total resources. Further offshore deposits are off Trinidad and Brazil.
Far East. The most important producing regions of the Far East are
China, Indonesia, Brunei, and Australia. In addition, oil deposits are known,
for example, in Malaysia, Burma, Thailand, Japan, Taiwan, and New Zealand.
China. Only after World War II has China gained importance as a produ-
cing country. By the use of modern scientific techniques, the first small fields
were discovered in the 1950s. In three regions, groups of fields had been proved
by the end of the 1980s that originate from plate-tectonic movements. Up to
1980, 150 fields had been discovered. Of these, merely for reasons of size, only
ca. 20 fields are of importance. Even so, the proved reserves have grown to ca.
5.0109 t. The two largest groups of fields, Daqing in East China and Shengli
on the Yellow Sea, achieved annual productions of 55.6106 t and 33.5 106 t,
respectively in 1990; that is ca. 64 % of the Chinese production. In the Daqing
field, which was discovered in 1959, between 4000 and 5000 wells distributed
over an area of 1200 km2 are producing from Lower Cretaceous sandstones in
very shallow anticlinical deposits. Shengli produces in a group of more than 40
individual structures from Precambrian and Tertiary reservoir beds. Of the
numerous sedimentary basins of China, only a few are fully developed. The
68 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Tarim Basin (Sinkiang) is expected to be particularly rich in reserves. This extre-


mely large inland basin, of 560 000 km2, has a sediment thickness of 18 000 m.
Since 1983, after intensive geophysical preparation and the sinking of
200 wells – the deepest reaching 6000 m – large quantities of hydrocarbons
have been proved in both the North Tarim High and in the center of the Takel-
magan Desert. The reserves are estimated at 11109 m3 oil and almost
81018 m3 natural gas. The productive formations are sandstones of the Triassic
and Ordovician carbonates. Exploration off the coasts of China is operated as
joint ventures with international oil companies. Starting from onshore discov-
eries, in the Bohai Sea, east of Beijing (Peking), a series of fields has been
developed whose potential is estimated to be more than 150106 t crude oil. In
the occurrences discovered in the South China Sea, no important reserves are
yet known. The crude oil of the Bohai Basin originated from tertiary sediments
of various structural types, including stratigraphic traps.
Other Areas. Important new discoveries have been made in recent years,
especially in the submarine areas of Indonesia, Brunei, and South Australia
(Tasman Strait). Of the developed reserves of the Far East, offshore deposits
could account for almost one third. The most important reservoir horizons are
in the Tertiary system; Mesozoic and Paleozoic formations are secondary
possibilities. Wide shelf areas off the northwest Australian Coast as well as
on the Indonesian and Chinese Shelves are regarded as favorable for future
exploration.
1.2. Natural Gas. Natural gas has gained considerable importance
in the past decades. Enormous gas reservoirs in Siberia, the Middle East,
Algeria, and in Europe, connected by pipeline systems with the consumer
countries, contribute substantially to the energy supply. The deposits in
North America are, for environmental reasons, an important energy source
for households and industry because of the fuel’s cleanliness. Natural gas
deposits in tropical regions are at present still underprivileged, since this
gas can only be used in the industrial countries after liquefaction and trans-
portation by ship. This is economically possible only if the price level offers
an appropriate margin for this treatment of natural gas. The composition of
natural gas is extraordinarily diverse (see also Natural Gas). Aside from
the so-called ‘‘dry’’, i.e., pure methane gas, ‘‘wet’’ gases with varying pro-
portions of higher hydrocarbons occur. Inert constituents, such as nitrogen
and carbon dioxide, reduce the calorific value. Sulfur and mercury are pollu-
tants when the natural gas is used and must be removed from it (see Natural
Gas). In connection with this, the elementary sulfur removed from the gas
secures its own economically significant market value. The proportions of
helium in the nitrogen-rich natural gases are, in general, small and only in
a few cases do they allow extraction for economic utilization. The natural
gases arising in the limestones and dolomites of the Northwest German
Zechstein contain up to 20 vol % hydrogen sulfide. The natural gas in the
French Lacq field contains 15 vol % H2S; while 63 % and 88 % are known
from reservoirs in Wyoming in the United States and in Alberta, Canada.
Since the gases containing hydrogen sulfide usually occur in contact with
gypsum or anhydrite beds, formation of hydrogen sulfide is ascribed to bac-
terial reduction of calcium sulfate. Natural gas deposits with considerable
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 69

percentages of heavy hydrocarbons are known as condensate deposits. While


wet natural gases and condensates mainly lie adjacent to oil deposits or are
associated with them, dry natural gases are usually found in separate gas pro-
vinces. This already follows from the fact that the majority of dry natural
gases can be derived from the kerogen III arising from terrestrial plants.
Other dry, natural gases are formed in parts of deep basins where pressure
and temperature have exceeded the threshold value at which crude oils
form. A distinction is made between associated gas, which is closely connected
with crude oil, and nonassociated or natural gas, which is separately gener-
ated, predominantly from kerogen III. The associated gas dissolved in the
crude oil is released during production by pressure reduction. If the crude
oil is supersaturated, then part of the gas migrates upwards and usually
forms a gas cap, which under certain conditions, can be exploited separately.
The nonassociated gas stands for deposits of dry natural gases that have
been formed from kerogen III. This is methane gas (CH4) being produced from
beds of the Upper Carboniferous, the Rotliegendes, and the Zechstein systems
in the British North sea, the Netherlands, Germany, and in Poland. It is beyond
doubt that this gas originates from the solid coal seams below the reservoirs or
from vegetable residues dispersed in sediments of the lower part of the Upper
Carboniferous system. The ‘‘petroleum gas’’, as a byproduct of oil production,
plays only a subsidiary role in energy supply – apart from exceptions such as
Ekofisk. Large amounts of this gas are used in the production plants or, usually
after compression, fed to regional gas supply systems. In low-pressure reservoirs,
the petroleum gas is returned to the reservoir to raise the pressure which conse-
quently reduces the production costs by a longer-lasting eruptive phase. In the
large producing countries, for example on the Arabian Gulf and in North Africa,
enormous quantities of petroleum gas are flared, thus being both wasteful and
environmentally polluting.
Since ca. 1980, the theory of the abiogenic origin of the hydrocarbons in the
ground has gained interest again. GOLD postulates that the vast quantities of
bound (oxidized) carbon, eg, in carbonates, cannot originate from the atmosphere
but escape from the mantel of the Earth. The presence of helium, that originates
from the radioactive decay of certain minerals, is also quoted by GOLD as evidence,
as are methane inclusions in diamonds, in kimberlite, or the high methane content
in Lake Kivu, which lies over a rift fault. The theory is based on the finding of a
methane atmosphere for the outer planets and their moons. Evidence for methane
of abiological origin was to be provided in the mid-1980s by a deep borehole in
the so-called Siljan Ring in Sweden. From this enormous meteorite crater, oil
and gas traces were reported from water wells and flammable gas was detected
from lakes and pools. Gravimetric measurements showed an anomaly that was
interpreted as a zone of good porosity in the granite, created by finely divided hair
cracks and clefts due to impact by the meteorite. The borehole sunk to 6080 m from
July 1986 to January 1987 and did not establish proof of abiogenic gas. German
scientists intend to investigate the problem of the so-called mantle gas in colla-
boration with the natural gas industry and the federal ministry responsible.
As with abiogenic hydrocarbons from the Earth’s mantle, the question of
the origin of the nitrogen in natural gases, involving the theory of the inclusion
of air in the Earth’s crust, is still being discussed. More probable theories suggest
70 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

an enrichment process of nitrogen as a result of fractional differentiation during


gas migration, formation of nitrogen by the action of anaerobic bacteria
during oil generation in the source rock, and its release from vegetable materials
during coalification. For the origin of the Northwest German nitrogen enrich-
ments, the possibilities under discussion include a relatively late split off
from the Upper Carboniferous coal seams and a concentration by fractional
differentiation on the migration paths.
Origin and Deposit Formation. Natural gases in the narrower sense, ie,
the combustible hydrocarbons, are often closely related to the crude oils, both
genetically and above all with regard to deposit formation. Petroleum and nat-
ural gas frequently occur in the same geological level of a region. On the other
hand, coal and petroleum occurrences in the same level of a region are almost
entirely mutually exclusive. Dry natural gas as a derivative of coalification –
both in regions with thick coal seams as well as in those with dispersely distrib-
uted coaly material – is also often found directly adjacent to petroleum regions,
eg, in the Emsland and the Wiehengebirgsvorland of Northwest Germany. The
majority of natural gas occurrences in the world is linked with oil provinces.
Associated natural gases, often in connection with an economically uninteresting
oil seam of low thickness (eg, the enormous gas deposits in the north of the West
Siberian Platform), predominate. As a result of fractional migration effects, accu-
mulations of dry gas deposits can also occur.
Gas formation in petroleum source rocks proceeds in different ways. In the
diagenesis phase, occurring in immature source rocks, bacterial methane is gen-
erated from kerogen III. At > 60 8C, the splitting-off of wet gas from kerogen
I and II begins – in addition to oil formation – which continues to the lower
end of the oil window at 150 8C. At higher temperature dry thermal gas is
increasingly formed. The natural gas deposits of the Rotliegendes and of the
Zechstein in the natural gas belt between the south of the British North Sea
and Poland owe their formation, as already mentioned, to the solid coal deposits
in the underlying Upper Carboniferous system. The coal-bearing beds were prob-
ably subjected, already at the end of the Carboniferous period, to an early coali-
fication which was so low that 30 % of the volatiles remained in the coal. As a
result of the further sinking of the coal seams by thick Cretaceous and Tertiary
overlying strata, the temperature required for natural gas formation rose again
and the coalification process continued with release of further methane. The
majority of the gas deposits arising between England and Poland, including
the Groningen deposit, owe their formation to such processes.
Theoretically, a further cause for the formation of natural gas is its genesis
from oil shales. As in the technical recovery of gas from oil shales, including the
Kukersit shales of Estonia (see Oil Shale), natural pyrolysis processes at tem-
peratures far above 200 8C are possible, by which natural gas could have been
formed from oil shales.
Natural Gas Occurrences and Their Reserves. The importance of the
natural gas producing countries is reflected in their present reserves. The
amounts produced annually have increased as a result of the consolidation of
the international pipeline network. In addition, interest in natural gas has
increased because it is the cleanest of the fossil fuels; it is much less polluting
than coal or oil. Large reserves also justify the high investment for the treatment
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 71

and pipeline transport of the natural gas (see Natural Gas). The production
quantities reported are not a measure of the natural gas potential available
in the producing countries, which follows from the growth of the reserves. This
is evident on comparing the United States and the CIS In 1989, production in
the United States was 500109 m3, and that of the CIS 739109 m3. On
January 1, 1990, the proved and probable reserves of the United States were
ca. 8.41012 m3 and those of the CIS ca. 451012 m3. To these there must be
added exploratory reserves ca. 111012 m3 for the United States and ca.
351012 m3 for the CIS The disproportion between production and reserves in
the CIS originates in missing pipeline connections, line losses, and also marketing
problems. Other large natural gas producers are Canada, the Netherlands, Iran,
Venezuela, the United Kingdom, Romania, Mexico, China, and Germany. The
large natural gas potentials in the countries of the Arabian Gulf, in Nigeria, and
in Indonesia suffer from the distance of the markets, since a basic requirement
for transportation from remote countries to the trade centers is the liquefaction
of the natural gas. Because of the relatively small capacity of the liquid gas tankers
and the large distances to be covered, the economic use of liquid gas is often at or
below the break-even point. The natural gas resources of the world — proved and
probable — on January 1, 1990, were ca. 127.01012 m3, and the prospective
another ca. 1211012 m3. The CIS has developed its leading position through the
enormous new discoveries, above all in Western Siberia, and is followed by the Mid-
dle East, which also has a very powerful growth to record. The raising of the proved
gas reserves in these two large districts originates not only in new discoveries but
partly in the extension or higher evaluation of deposits already known and the
strengthened inclusion of petroleum gases, particularly in the Middle East and
Venezuela. While Africa and North America show a declining trend for the proved
reserves, Western Europe has improved its position in the international gas sector
by further new discoveries in the North Sea.
Regional Distribution of the Reserves. A breakdown of natural gas pro-
duction and resources according to region is given in Table 2.
Commonwealth of Independent States. The largest reserves in the
world, at 451012 m3, are those of the CIS The occurrences are widely spread
and in some cases combined with petroleum deposits.

Table 2. Natural gas production and resources up to December 31, 1989 (assembled
from various international data)
Resources

Cumulative Proven and Prospective,


production, 109 m3 probable, 109 m3 109 m3
North America 23 436 11 000 18 700
Latin America 1 734 7 000 10 600
Middle East 1 629 42 000 26 700
Africa 924 7 400 9 300
Far East 2 306 9 800 13 700
Europe 4 970 5 800 7 200
CIS 10 150 44 000 34 800
World 45 149 127 000 121 000
72 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

The first great gas boom was in the 1950s with the development of the fields
of Shebelinka in the Dnieper – Donetz Trough and Gazli in Central Asia, each
with primary resources of 450 – 500109 m3. The discoveries in the second
half of the 1960s, some of which were enormous, established the present position
of the CIS in the natural gas sector. Important deposits mentioned are those of
Krasny – Kholm (Orenburg) in the Volga – Urals district, where ca. 750109 m3
of gas have been discovered, and the Vuktyl and Layavozh deposits in the Petchora
Basin, linked with beds of the same age and each with ca. 500109 m3 of gas. In
Turkmenia, the Malai field with ca. 340109 m3, is one of the giants. Probably
the largest natural gas province in the world is in the north of the Western Siberian
platform. The gas occurring here in sandstones of the Cretaceous system (predomi-
nantly Cenomanian), in part several hundred meters thick, consists of almost pure
(97 – 99 %) methane. The structural traps are extended anticlines and mound-like
highs. About 220 fields have been developed in this area up to now.
This producing province is the most important for the future export of gas
to Europe. Its most extensive deposits include the Urengoy field, the largest
deposit proved so far with ca. 61012 m3 of initial gas reserves, and also the
large deposits of Jubileyny (ca. 21012 m3), Taz (ca. 1.151012 m3), Medvezhye
(ca. 1.05 1012 m3), and Yamburg (ca. 850109 m3) among others. At least a
further ten fields of this region have each reported primarily more than
100109 m3 of recoverable gas reserves. Two new discoveries in the Arctic
Ocean deserve attention. In 1987 in the Barents Sea, not far from the
Norwegian border, the Shtowmanowskaja field with an estimated reserve of
3 – 41012 m3 was discovered, and in 1988 in the Kara Sea, the Russanowskaja
field with 61012 m3. Both fields together are credited with 15 % of the future
production provided that the production and transportation to Nowaja Semlja
is possible under arctic conditions.
Middle East. Even if not so dominant as in the case of crude oil, the
areas on the Arabian Gulf, with proved natural gas reserves of ca. 681012 m3
play an important role in the possibility of future exports. However, a consider-
able fraction of these quantities is trapped in gas caps or is dissolved petroleum
gas linked with the large oilfields of this area and is therefore dependent on their
production. The gas cap of the Pazanan oilfield in Iran alone contains recoverable
quantities of ca. 1.41012 m3 of gas. With the recent discovery of enormous gas
accumulations in the Permian Khuff formation, very large dry gas deposits have
also been proved for the Middle East. The Kangan field in Iran is said to contain
41012 m3 of resources. The deposits available in the Permian formation are
sealed above by triadic salt beds. There are further deposits of this type, not
hitherto exploited, in Qatar. Iran also is involved in the northern part of the
Northwest Dome (North Field) – with 2.8 – 3.4 1012 m3 a ‘‘super giant’’ –
which belongs to the Permian Khuff. Notable natural gas reserves are also
known in Saudi Arabia and the Emirates.
North America. Despite higher quantities produced, the known North
American gas reserves have risen again. The reserves of 91012 m3 in the mid
1970s have risen to ca. 111012 m3 proved and probable. Of these, 8.41012 m3
in the mid 1980s were assigned to the United States alone, the traditional
land of natural gas. Its large-scale industrial use started here even before the
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 73

turn of the century. In addition to this, exploratory reserves of almost


201012 m3 can be anticipated in North America. The largest deposits in the Uni-
ted States are in Texas, Oklahoma (particularly in the deep Anadarco Basin),
Kansas, and New Mexico with the fields of Hogoton (initial reserves ca.
1.11012 m3) and Panhandle (primary resources ca. 870109 m3), which are
already in part greatly depleted. A large fraction of the resources are in the
Rocky Mountain Province with 1.251012 m3. The natural gas here is predomi-
nantly bound to Permian carbonate rocks, but is also found in sandstones,
arkoses, and weathered granite zones. Other deposits, such as the Blanco
Basin field in New Mexico and Rio Vista in California, occur in Cretaceous sand-
stones and Lower Paleozoic carbonate rocks (Pucket field, Texas).
The gas accumulations of the American Gulf Coast are mainly associated
with oil deposits or are found in separate reservoirs over the oil in Tertiary sand-
stones. The recoverable gas contents of these latter isolated deposits are only
between 100 and 300109 m3.
The North American gas deposits, especially those on the Gulf Coast, are
frequently characterized by a relatively high nitrogen content.
Africa. The proximity of Algeria and Libya to the industrial countries of
Europe favors the export of natural gas to Europe. Altogether, supply contracts
for more than 80109 m3/a have been closed, of which 12109 m3 are conveyed by
pipeline to Italy and the remainder, after liquefaction, by tanker mainly to
Germany and a further part to the United States. In addition to the gas reserves
of the oil-producing countries Libya and Algeria, that mostly occur as dissolved
gas, there are dry gas deposits, particularly in Algeria. The largest African field,
Hassi R’Mel in the Algerian Sahara, discovered in 1956, has been in production
for years. The gas that contains 86 % methane occurs in triadic sandstones in the
area of an anticline. The original resources were assessed as ca. 11012 m3, but
are probably more than double that amount.
Western Europe. The Western European deposits were of only local sig-
nificance until the discovery of the large Groningen gas field in 1959/1960 and the
subsequent discoveries in the North Sea. The small Western European gas pro-
duction comes mainly from the French field, Lacq, in the Pyrenean foredeep,
with primary reserves of ca. 200 109 m3, as well as from deposits in Northwest
Germany and the Netherlands where they are mostly bound to Zechstein dolo-
mites and from Tertiary reservoir rocks in the German Alpine approaches.
There are smaller production areas in the inner-alpine Vienna basin, in the Po
plain, and off the Adriatic coast of Italy. The Austrian and Italian reservoirs
occur mainly in Tertiary horizons. As a result of the development of the gas
deposit of Groningen in the 250 m thick Rotliegendes sands, with initial reserves
of ca. 1.91012 m3 and a methane content of ca. 82 vol %, the exploration boom for
inferred natural gas accumulations of the Rotliegendes in the North Sea and for
the neighboring region of Germany started from 1963/1964. In the North Sea a
distinction can be made between a dry gas province in the southern North Sea
with deposits off the eastern English and Dutch coasts, and the gas deposits of
the central and northern North Sea. The latter are mostly associated with oil
deposits and linked with the Dan formation but, most of all to sandstones of
the Paleogene system.
74 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

The deposits in the southern North Sea generally occur, as in the


Groningen field, in the sandstones of the Rotliegendes. The methane
gas, like most of the accumulations in Northwest Germany, is attributed to
coalification processes in the underlying seam-bearing Upper Carboniferous
system.
The natural gas reservoirs further to the north could be of similar origin, in
the broadest sense, with the oil deposits proved there. These are partly gas cap
gases and partly gas accumulations in separate reservoir horizons above oil
deposits, or above economically uninteresting oil seams.
On January 1, 1992, ca. 4400109 m3 natural gas had been discovered and
developed in the North Sea. Of this, ca. 1200109 m3 was assigned to the United
Kingdom, ca. 2700109 m3 to Norway, ca. 350109 m3 to the Netherlands, and
ca. 110109 m3 to Denmark. The results of drilling from the German North
Sea give no indication of major natural gas deposits comparable with those of
the neighboring sectors. Only one minor natural gas deposit in the extreme
part of the German North Sea with reserves of ca. 10109 m3 is under develop-
ment. Production, in the North Sea, mainly from British gas fields, increased
from ca. 50109 m3 in 1985 to 98109 m3 in 1991.
The largest known gas deposits in the North Sea include the follow-
ing fields: Frigg (Paleocene sandstone, ca. 400109 m3 resources); Leman
(Rotliegend sands, ca. 300  109 m3 reserves); Indefatigable (Rotliegend
sands, ca. 140109 m3); Viking (Rotliegend sands, ca. 140  109 m3); L-10/11
(Rotliegend sands, ca. 100 – 140  109 m3); K 13 (Rotliegend sands, ca. 100 –
140  109 m3); Hewett (Triassic sands, ca. 110  109 m3); Troll (Middle/Upper
Jurassic, ca. 1600109 m3).
With continuous exploration on the Norwegian Shelf (Arctic Ocean and
Barents Sea) further large natural gas discoveries can be expected.

2. Reservoir Engineering

2.1. Characterization of Oil and Gas Reservoirs. As described


in Chapter 1, oil and gas reservoirs are accumulations of hydrocarbons
that are stored in porous or fissured strata of the Earth’s outer crust and
contain a commercially valuable quantity of oil or gas. A precondition for
storage in the sedimentary rocks is a continuous cavity, in the form of pores
or fissures, that permits transport of the fluids into pressure sinks (wells).
Since these sedimentary strata can be found at various depths, it is
important that the rock matrix takes on the high rock compression stress
without much deformation so that when pore pressure is released, the pore
fluids can easily flow to the pressure sink, ie, the well. If large deformations
were to occur in the rock, the flow paths in the neighborhood of the well
could close upon pressure release thus resulting in the virtual sealing of the
reservoir.
Reservoir Rock. Since the reservoir rock rarely has a monomineralic
character, its mineral composition and the internal surfaces resulting from
it are very important, both for the storage capacity of the rock and for
the transmissibility of the fluids stored within it. The crystal structure and
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 75

Fig. 7. Composition of a detritral rock (72) a) Grain (texture: sand granules 1/16 – 2 mm
diameter; silt 1/256 – 1/16 mm diameter); b) Pore fillings (texture: sand granules 1/16 mm
diameter; silt 1/16 – 1/256 mm diameter; clay < 1/256 mm); c) Cement; d) Fluid.

specific internal surface of the rock-forming minerals determine the types of


porosity of a reservoir rock.
In principle, cavities are found in a mineral structure along the cleavage
planes of crystals (intercrystalline pore space), within the layer lattice of crystals
(intracrystalline pore space), and, on a larger scale, along the boundary surfaces
of particles (interparticulate pore space) or within the particle at mineral bound-
aries or growth defects (intraparticulate pore space). Hydrocarbons are stored
mainly in the interparticulate pore space of clastic sediments or the fissured
pore space of chemical sediments (limestone, dolomite). Fissures form in a rock
as a result of deformation when the shear strength is exceeded.
Figure 7 shows the typical composition of a porous clastic sediment (detrital
rock). The rock body consists of sediment particles (grain) that are mainly com-
posed of mineral components quartz, feldspar, and mica; of bioclasts; or oolites.
In most cases the rock particles are cemented together by calcites, dolomites, sili-
cates, or clay minerals. In the interparticulate pore space so formed, pore fillings
of clay minerals, quartz, and mica can lower the available porosity.
Porosity. Porosity f defines the ratio of the pore volume Vp of a reservoir
rock to the bulk volume Vb of the rock:

Vp

Vb

According to this definition, the porosity also includes the volume of isolated
pores (dead pores).
If all interconnected elements of the pore volume are taken into account, the
effective porosity is obtained. This is an indicator of the effective storage volume
of a rock. The sum of effective feff and dead fd porosities gives the total porosity:

feff þ fd ¼ f
76 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

For geometrical or hydrodynamical reasons, not all pores are available for
convective mass transport. A fixed portion of the cavity volume can be regarded
as dead-end pore volume.
The porosity depends essentially on four characteristics of the sedimentary
rock:

1) Particle shape
2) Degree of cementation
3) Particle size distribution
4) Particle arrangement

The influence of particle shape and degree of cementation is evident from


Figure 7. In general, the more edged the particles, the greater is the inter-
particulate component of the pore space. The available porosity decreases
with increasing cement component. The most important influencing factors,
however, are the particle size distribution and the particle arrangement in the
rock structure.
Figure 8 A shows a wide distribution of particle diameters which leads to a
large reduction in porosity, since the interparticulate pore space formed by the
coarse fractions is filled up by the fine components. On the other hand, a narrow
particle size distribution (Fig. 8 B) results in a high porosity. In the lower half of
the figure, the cumulative particle size distribution is also shown. Usually the
steeper the cumulative curve, the greater is the porosity of the rock.
Three special cases of particle arrangements and the corresponding poro-
sities are shown in Figure 9. In the cubic, space-centered sphere packing

Fig. 8. Grain-size distribution and porosity A) Low porosity; B) High porosity.


Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 77

A B C

90° 60°
90° 90° 60° 60°

90° 90° 90°

Fig. 9. Packing and porosity (73) A) Space-centered, cubic sphere packing (number
of contacts = 6, f ¼ 47.6 %); B) Hexagonal packing (number of contacts ¼ 8, f ¼ 39.5 %);
C) Densest packing of spheres (rhombohedral) (number of contacts = 12, f ¼ 25.9 %).

(Fig. 9 A), in which the particle midpoints form the corners of a cube, the inter-
particulate pore volume is exactly 47.6 % of the total volume. In this case each
particle has six points of contact with the neighboring particles. In the hexagonal
arrangement of particles (Fig. 9 B), the spherical particles of the upper lattice
plane are displaced by half the particle diameter relative to the lower lattice
plane. In this arrangement, the number of contacts rises from six to eight and
the porosity decreases to 39.5 %. If in the second lattice plane of the packing,
the row of spheres is displaced not only in the x-direction by d/2, but also
in the y-direction perpendicular to it, by d/2, the densest packing of spheres is
obtained (C). Here in the rhombohedral arrangement the particles of the upper
lattice plane are placed exactly in the holes of the lower plane and accordingly
each particle has twelve neighbors and the porosity is only 25.9 %. The porosity
of a natural packing of cemented sedimentary grains usually falls between the
limiting values of 39.5 and 25.9 %.
Pore Configuration Models. Although, at first glance, the shape of the
pores or fissures of a sedimentary rock has very little to do with regular geo-
metric forms, cylindrical capillaries or parallel plates are regarded as geometri-
cally similar to the pores and fissures and are chosen as models for calculating
the pore space properties. These very simple geometries are unable, however,
to reproduce the irregular variation of the pore or fissure cross section or of
the tortuosity of flow path, which are important characteristics of the pore
configuration.
In Figure 10, the commonest types of pore configuration models, up to the
most advanced representation (fractal network model), are shown. The capillary
bundle model (Fig. 10 A) is the most frequently applied pore configuration model,
mainly used for calculating the pore radius distribution, the saturation with
fluids, and the oil displacement process. A particular drawback of this model is
that a transverse flow between the capillaries is not possible. A pore configura-
tion model for fissured reservoir rocks derived from the plate model is the
78 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 10. Pore configuration models (74) A) Capillary bundle model; B) Brickstone model;
C) Intersphere network model; D) Cascade network model; E) Fractal network model
White color means pore space.

brickstone model (B), in which the individual blocks symbolize the porous matrix
and the gaps between them, the fissures. Here mass transfer between matrix and
fissure is possible. The intersphere network model (C) is used particularly for
determining the phase distribution in the pore space and the capillary forces.
With the cascade network model (D), dead-end pore effects, turbulence effects,
and mixing phenomena can be investigated.
According to the principle of self-similarity of geometric forms in nature,
relations are traceable as far as between the crystal forms in the microcosmos
and the geological outfits (morphology) of the Earth’s surface. In connection
with this, the fractal dimension is a scale-up factor between the characteristic
dimensions and the image scales. If the spherical shape (circular shape in
Fig. 10 E) is chosen as the characteristic elementary shape of the rock structure
and of the pore space, a fractal network for matrix and pore configuration can be
developed, whose specific internal surface and volume fraction are calculable and
correlatable with other physically measurable quantities such as electrical con-
ductivity and permeability.
An example of the use of pore configuration models for calculating pore
space properties is given here, based on the determination of the pore radius
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 79

distribution by the mercury injection method. If the total pore space Vp is


assumed to be represented by a bundle of capillaries of various diameters,
then on injection of a nonwetting liquid phase (ie, mercury) the individual
sizes of capillary are filled stepwise (DV ) as the pressure is raised. Since the
injection pressure for mercury pc is inversely proportional to the radius of
the capillaries, the proportions by volume DV/Vp of the individual capillary
radii can be estimated from the mercury injection curve

2sHg cosyHg
pc ¼
r

where sHg is the surface tension of mercury and y the wetting angle of
contact.
In order to reconcile this measurement with actual pore space configura-
tions, it must be taken into account that each pore has at least two characteristic
dimensions – the pore entrance and the pore widening. The characteristic cross-
section determined by the mercury injection method corresponds to the pore
entrance radius.
Compressibility of the Pore Space. The compressibility c of a substance
is the change, with external pressure, of its volume relative to its initial volume.
For a solid the compressibility is negligibly small. A porous medium which is
built up of individual particles can, however, change its particle arrangement
under pressure and, therefore, is said to have considerable compressibility.
The deformability is discussed for the different forms of packing of spherical par-
ticles in Section 2.1.1.1.1. According to the studies of GEERTSMA (75), the pore
volume compressibility cf depends on the type of pore configuration (porosity of
the medium) and the effective pressure which is the difference of external pres-
sure pe and internal pore pressure pi.

1 qVp
cf ¼ ðÞ
Vp qð pe  pi Þ

Figure 11 shows that the pore volume change reacts with increasing sen-
sitivity to an increase in the external pressure with decreasing porosity. The
porosity of sedimentary rock under in situ conditions (ie, under the influence
of the overburden pressure of the sediments) can, therefore, be up to 20 % smaller
than the porosity determined in the laboratory (77).
Physical Properties of the Pore Saturating Fluids. Gas, oil, and water
are the relevant pore saturating fluids, whose physical properties must be
known as a function of the pressure, the reservoir temperature, and the fluid’s
composition. The material properties of these multicomponent systems can
only be predicted for the gas state at low pressure with acceptable accuracy
and without an extensive experimental data base.
For liquids and phase mixtures, estimates of the material properties
are obtained in accordance with the theorem of corresponding states by
normalizing to conditions at the critical point. The state variables, pressure,
and temperature, are normalized by introducing the reduced pressure pr
80 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

14

12

10

Ct, 10−10 Pa−1


6

0
0 0.04 0.08 0.12 0.16 0.20 0.24
Φ

Fig. 11. Compressibility of the pore volume (76).

and temperature Tr.

p T
pr ¼ Tr ¼
pc Tc

Mixtures with complex structure can be described by means of pseudocriti-


cal pressure and temperature that are calculated from the molar fractions of the
individual components and the corresponding individual pseudocritical quanti-
ties. These simple correlations yield satisfactory results only when the individual
components have similar critical pressures and critical volumes.
In all other cases three-parameter thermodynamic functions of state are
used to improve the accuracy of prediction. These calculation methods are
based on the attempt to produce a functional relationship between the critical
material properties of a component and the molecular interactions. Nevertheless,
no known equation of state is able to determine all properties of hydrocarbon
mixtures equally satisfactorily.
For determining the contents of a reservoir and for calculations of recovery
and flow processes, the most important physical parameters are the density,
compressibility, and viscosity of the fluids. In the following, the methods of cal-
culation applied in practice will be considered without showing the full physical
background. The reader is referred to the references (77,78) given at the end of
this article.
Density. The density of a gas %g can be calculated directly from the equa-
tion of state for real gases

Mr p
%g ¼
zRT

where Mr is the molecular mass, p the pressure, z the real gas factor, R the gas
constant, and T the temperature.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 81

The reduced temperature and the reduced pressure serve as adjustment


parameters for the real gas factor z. For gas mixtures, instead of the molecular
mass Mr the mean molecular mass of the mixture is used together with the real
gas factor for the mixture.
For direct calculation of liquid densities, this simple form of the equation is
inadequate. Because of the higher packing density of the molecules in the liquid
allowances must be made for the inherent volume of the molecules and for mole-
cular interactions.
If the critical densities or the densities of the individual components under
standard conditions are known, the density of the mixture %f can be determined
from the molar fractions according to the rule of additive properties
P
n
z j Mr j
1
%f ¼
P
n z
j Mr j

1 %0 j

For reservoir calculations, the so- called formation volume factor B is


usually used instead of density for converting gas and liquid volumes to reservoir
or standard conditions. The formation volume factor is the ratio of the volume
under reservoir conditions to the volume under standard conditions.

Volume at pressure p and temperature T



Volume at standard conditionsð p0 ; T0 Þ

where the standard conditions for gases are: p0 = 0.1013 MPa and T0 273.15 K.
For liquids p0 = 0.1013 MPa and T0 = 293.15 K holds.
In the United States, 60 8F, corresponding to 288.65 K, is used as standard
temperature. The formation volume factor for gas Bg can be calculated directly
from the real gas equation:

zT p0
Bg ¼
pT0

From a double-logarithmic plot of the formation volume factor versus pres-


sure, a linear curve is obtained as long as the influence of the real gas factor at
elevated pressure is not noticeable (Fig. 12).
Figure 13 shows the formation volume factor for oil, Bo, as a function of
pressure at constant temperature.
For a gas-undersaturated oil, the formation volume factor Bo at first
increases with decreasing pressure as a result of the liquid-phase expansion of
the oil and reaches its maximum value at the bubble point pressure ( pb). The
greater the gradient of the branch of the curve between the initial reservoir pres-
sure ( pi) and the bubble point pressure, the more compressible is the liquid.
When the pressure falls below the bubble point pressure, the volume of the oil
and with it the formation volume factor decreases rapidly. The final value of 1
is not reached, since the thermal shrinkage of the oil from reservoir-temperature
to standard-temperature condition occurs at a final pressure of 0.1 MPa.
82 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 12. Formation volume factor for gas.

To describe the volume behavior of a two-phase mixture of gas and oil, the two
formation volume factors can be added to obtain the two-phase formation volume
factor Bt, taking the gas – oil ratio (Rs) into account.

Bt ¼ Bo þ Bg ðRsi  Rs Þ

where Rsi is the initial gas – oil ratio.


In Figure 13 the upper part of the curve is identical to the two-phase for-
mation volume factor Bt. The formation volume factor for water Bw, which is
shown in Figure 14, does not show a decrease through shrinkage resulting
Formation volume factor

1.0
0

Fig. 13. Formation volume factor for oil (78).


Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 83

Fig. 14. Formation volume factor for water (78).

from degassing below the bubble point pressure ( pb). Instead, it rises further as
a result of volume expansion on pressure release. This is due to the relatively
low dissolved gas volume in water (ca. 1 m3 gas/m3) compared with oil
(10 – 50 m3 gas/m3).
Compressibility. The isothermal compressibility of the pore fluids is
given by:
1 qv
c¼
v q pT

The pore fluids differ clearly in their compressibility behavior. While gases in
the pressure range 0.1 – 100 MPa are extraordinarily compressible (c = 105 –
108 Pa1), oils have a moderate to low compressibility (c = 108 – 1010 Pa1)
depending on their dissolved gas content, and water has a compressibility of ca.
1010 Pa1. The last figure is of the same order as the compressibility of the
rock pore volume. In calculating the compressibility of real substances, the nor-
malized thermodynamic functions of state, pressure, and temperature can be used.
For a gas mixture under real conditions, the compressibility equation
reduced to normalizedconditions is:
 
cpr 1 1 1 qz
cg ¼ ¼ 
ppc ppc ppr z q ppr

Here, instead of the critical data for a single gas, pseudocritical values ( ppc and
Tpc) are used as normalizing quantities.

p X
n
ppr ¼ ppc ¼ y j pc j
ppc 1

T X
n
Tpr ¼ Tpc ¼ y j Tc j
Tpc 1
84 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

where ppr is the pseudoreduced pressure, Tpr the pseudoreduced temperature,


and yj the molar fractions of the individual components.
For an ideal gas, whose z factor is 1, the gas compressibility is therefore
inversely proportional to its pressure. For liquids, whose critical z factor is
0.27 (most higher molecular mass hydrocarbons satisfy this condition), the
compressibility can be determined from normalized pressure – temperature
graph (78).
Viscosity. According to Newton, the dynamic viscosity m is the proportion-
ality factor between the shearing stress in the flow direction and the resulting
shearing rate change perpendicular to the flow direction. Gas, oil, and water lar-
gely behave as Newtonian substances. For the viscosity of gases at low pressure
and constant temperature, the product of the free path, mean velocity and the den-
sity of the gas molecules is a measure of internal friction. At low pressure, only the
Brownian motion velocity of the molecules is relevant for this energy transfer and
the viscosity of gases therefore increases with the square root of temperature. Only
at elevated pressure does a reduction of the internal friction and, therefore, the
viscosity, appear with gases and liquids with increasing temperature.
For gas mixtures the mixing rule definition for n components applies.
P
n
y j mg j M0:5
j
1
mg ¼
P
n
y j M0:5
j
1

To allow for the influence of pressure and temperature on the viscosity of a real
gas, the normalized representations of the reduced pressure and temperature
can again be employed.
The pressure and temperature dependence of the viscosity of reservoir oils
is shown in Figures 15 and 16, respectively. Above the bubble point pressure pb,

Fig. 15. Viscosity of oil as a function of pressure (78).


Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 85

Fig. 16. Viscosity of oil as a function of temperature a) Heavy crude oil; b) Medium crude oil.

the viscosity linearly increases with the reservoir pressure. Below the bubble
point pressure, the oil viscosity falls steeply with increasing pressure or gas
content.
The temperature dependence of the viscosity of oil is characterized by an
exponential function, so that in a semilogarithmic plot versus the reciprocal of
the absolute temperature (see Fig. 16) a straight line is obtained. The slope of
this straight line is greater the higher the oil’s content of heavy components.
For the dependence of the viscosity of water on the temperature the same math-
ematical relationships apply. In addition, the viscosity of water increases with
increasing salt content (79).
Reserves. A detailed discussion of worldwide resources of oil and gas is
given in Sections and 1.2.3,, respectively.
Resources mean the total volume of hydrocarbon stored in reservoirs. With
oil, gas, or condensate reservoirs, the initial volume of the pore contents, the so-
called original oil in place (OOIP) or original gas in place (OGIP), is the first
quantity used for evaluating a resource. Since for physical, technological, and
economic reasons the resources are not completely recoverable, a second concept
is necessary for characterizing the profitability of a discovery: the reserves.
Reserves mean those quantities of hydrocarbons that at a given time are proved
on the basis of sufficient geological and reservoir engineering information and
can, in future, be produced with currently available methods of recovery and
applicable economic conditions. Reduced to a simple formula,

Reserves ¼ Resources  Recovery factor

The recovery factors for oil reservoirs, apart from a few exceptions, are
between 20 and 45 % of the contents. For gas reservoirs, the recovery factor
can be up to 85 %. The degree of certainty in the determination of the reserves
increases with increasing development of a reservoir. According to a classifica-
tion of reserves made by the Committee of the World Petroleum Congress, ‘‘pro-
ven reserves’’ are spoken of when the probability of proof exceeds 90 %, and
‘‘unproven reserves’’ when that probability is between 10 and 90 % (80). The
86 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

unproven reserves are split by a median value of 50 % into probable and possible
reserves. The remainder are the speculative, so far undiscovered, possibly reco-
verable reserves.
Volumetric Determination of the Reservoir Contents. At the start
of development of an oil or gas reservoir, usually only the contours of the reser-
voir are known; apart from test analyses data, no adequate production and pres-
sure trends are available. The content of a reservoir can then be surveyed only by
volumetric methods. The contents related to standard conditions are calculated
from the following formulae:

Vb f ð1  Swi Þ
OOIP ¼
Boi
Vb f ð1  Swi Þ
OGIP ¼
Boi

where OOIP denotes original oil in place, OGIP denotes original gas in place, and
Swi is the initial water saturation.
The bulk volume of the reservoir Vb is determined from a geological model
of the reservoir. In this procedure the nonproductive regions of the strata must
also be taken into account such as clayey interlayers or lenses, water-bearing
belts, and the distinction between gas-, oil- and water-bearing zones (gas –
oil – water contacts) when delimiting the net rock volume from the gross rock
volume. In the evaluation of borehole measurements, so- called cut-off values
for porosity, saturation, clay content, and permeability are introduced to differ-
entiate between pay zone and nonproductive thickness and to determine the pay
zone thickness (Fig. 17). The black areas correspond to the porous, hydrocarbon-
bearing formation intervals that are included in the net layer thickness.
For displaying the geological models, various mapping methods are chosen.
For example, in Figure 18, the structural map of a reservoir is represented by
subsurface contour lines (that is in feed). For determining the net rock volume
of a reservoir, the reservoir body can be imagined to be divided into a large num-
ber of slices of equal thickness. Each of these slices can be represented by a geo-
metrically similar body such as a truncated cone, truncated pyramid, or other,
irregular shapes, allowing its volume to be calculated. A more elegant method
is the graphical evaluation of the subsurface contour planes. In this method
the areas of the contour line map determined by planimeter for the upper and
lower boundary of the reservoir are plotted against the relative depth in a system
of cartesian coordinates (77).
Furthermore, if the lower and upper boundaries of the reservoir contents –
the gas – oil and water – oil contacts, respectively — are taken into account, the
rock volume can be determined directly from the area – depth graph by means of
millimeter paper or planimetry. These computed or graphically-determined rock
volumes are inserted into the formulae for calculating the OOIP and OGIP. In
low-permeability reservoirs bearing formation water, the transition zone from
the completely water saturated zone to the oil bearing zone, containing irreduci-
ble water, is an uncertainty factor in the oil or gas resource calculation. The tran-
sition region is characterized by the capillary pressure curve.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 87

Fig. 17. Net pays from well logging (81).

2.2. Flow Processes in Porous Media. The original state of a crude oil
or natural gas reservoir is characterized by the thermodynamic equilibrium of the
hydrocarbon phases and the capillary equilibrium of the wetting and nonwetting
phases, which in turn depends on the prevailing temperature, pore space pressure,
and rock properties. Not only the composition but also the properties of the oil, gas,
and water phase show a remarkable ‘‘memory’’ for the early history of the formation
of the deposit. For example, oils that lack in certain alkanes can be recognized as
intermediate products of biological degradation from an original sapropellitic
material. Light oils which have been exposed to oxidation as a result of tectonic
88 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 18. Isomap representation of reservoirs/regular contours (82).

postsedimentary processes, might have been converted into medium-heavy oils, but
they still retain the chemical features (fingerprints) of their origins.
The initial distribution of the phases in the pore space is governed by the
equilibrium between interfacial forces and viscous forces during the migration pro-
cess. If a volatile oil enters a porous rock which is saturated with salt water and
flushes the pore space over extended periods then the resulting initial water
saturation in the deposit is normally greater than the adsorbed water saturation
(ie, the layer of water retained by physical adhesion to the interior surfaces of the
rock). If, in contrast, the migrating oil is composed of heavier hydrocarbon frac-
tions, it displaces the pore water almost down to the level of the absorbed water
saturation. After a reservoir has been filled, the materials are in dynamic rather
than static equilibrium. Diffusion and convection can influence the material equi-
librium and the distribution of the phases in the deposit.
Single-Phase Flow. If a fluid with little or no compressibility (water, oil)
is flowing in an infinitely large reservoir system towards a pressure sink (well)
then — after a transient phase — a steady-state condition is reached between the
quantity removed from the well and the quantity flowing in from the deposit.
Under these conditions, the pressure changes only as a function of space, but
not of time. However, during the transient state before the steady-state equili-
brium is reached the pressure also changes with time.
Steady-State Flow – Darcy’s Equation. In the laminar flow regime —
ie, in a homogeneously porous medium with a Reynolds number < 1 — Darcy’s
law holds; a single well in an infinitely large reservoir radial flow geometry
can be assumed

q k qp
¼v¼ ð2:1Þ
2prh m qr

where q is the flow rate, r the radius, h the thickness of the porous layer, v the
apparent velocity, k the permeability of the rock, m the dynamic viscosity, and p
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 89

the pressure. Integration of Equation (02.1) between the well radius rw and the
radius re of the area drained by the well gives the following for steady-state flow:
Boundary conditions: pe at re; pwf at rw are constant with time

2phk ð pe  pwf Þ
qw ¼ ð2:2Þ
Bm ln re =rw

where B is the formation volume factor of the fluid.


However, for a compressible medium (gas) the volume flow is not constant
along the direction of flow. If Darcy’s formula is still to be applied it is necessary
to convert the flow rate to some mean pressure pm between the system’s input
and output pressures (77). The average flow rate can be converted to the required
flow conditions in the well by using the equation of state for ideal gases.
Boundary conditions for linear geometry: pe at x = l; pwf at x = 0, where x
denotes the distance from the well.

pe þ pwf
pm ¼
2
Ak ð pe  pwf Þ
qm ¼
m l
2
Ak ð pe  p2wf Þ
qw pwf ¼ qm pm ¼ 
m 2l
2
Ak ð pe  p2wf Þ
qw ¼ 
Bm 2 pwf l

Transient Single-Phase Flow. Consider a space element in the form of a


cylindrical capillary of length Dx and cross-sectional area A (Fig. 19). A mass flow
(% vx)x A is injected at the inlet end and the mass flow (% vx)x+Dx A emerges at the
outlet point at a distance x.
Under transient flow conditions, the difference between these two
mass flows corresponds to the mass stored in the volume element DV. The

Fig. 19. Material balance in a cylindrical capillary of length Dx and cross-sectional area A.
90 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

material balance must be maintained because of the principle of conservation


of mass.

Mass inMass out ¼ Mass stored


A t
ð%vx Þx  ð%vx Þxþx Þ

divided by A Dx Dt

ð%vx Þxþx  ð%vx Þ %t


lim ¼
x !0 x t
t !0

qð%vx Þ q%
þ ¼0
qx qt

The differential equation for the continuity law is obtained from the differ-
ence equation assuming small increments of time and space. The nabla operator
and the porosity in the storage volume DV are introduced so that this equation
can be applied to a porous medium and extended to three-dimensional flow
conditions. For horizontal flow of a medium of low compressibility, the extended
continuity equation becomes:

qðf%Þ
rð%vÞ þ ¼0 ð2:4Þ
qt

In one dimension:
k qp
vx ¼ 
m qx
 
q k qp qðf%Þ
 % þ ¼0 ð2:5Þ
qx m qx qt
 
k q% q p q2 p q%
þ% 2 ¼ 
m qx qx qx qt
where k/m is the transmissibility of the pore fluid for a given porous medium.
As an initial condition it was assumed that the transmissibility k/f and the
porosity f are constant throughout the pore space system. The compressibility
equation
1 q%
c¼
% qp
for a fluid of low compressibility is introduced and used to replace the change in
density % by the change in pressure p.
"   
k% q p 2 q2 p qp
c þ 2 ¼ %c
m qx qx qt
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 91

Both sides are multiplied by the product m/k%. The flow rate, at some dis-
tance from the well, is only in the range of a few centimeters per hour so the pres-
sure gradient p/x and the square of the pressure gradient are very small in
comparison with the other variables in Equation (2.5). This term can therefore
be neglected.
This then gives Equation (2.6) which is similar in form to the diffusivity
equation, and is referred to as the pressure propagation or hydraulic conductiv-
ity equation for the one-dimensional case.

q2 p fmc q p
¼ ð2:6Þ
qx2 k qt

Equation (2.5) can also be solved for other initial conditions. If the density of
the fluid phase is defined by a mass law, eg, the gas law for ideal gases, then the
following form of the hydraulic conductivity equation is obtained

q2 ð p 2 Þ fm q p
2
¼2 ð2:7Þ
qx k qt

These equations can also be represented in three-dimensional form. To


solve Equations (2.6) and (2.7) it is necessary to adapt the flow geometry and
boundary conditions to match the situation in the deposit. For a cylindrical coor-
dinate system Equation (2.8) is obtained.

q2 p 1 q p fmc q p
þ ¼ ð2:8Þ
qr2 r qr k qt

With a linear pressure sink as the well, and for infinite extension of the reservoir,
ie, re ¼ 1, the Boltzmann transformation can be used for solving Equation (2.8).
The following boundary conditions apply:

t¼0 r ¼ rw p ¼ pi
r ¼ re p ¼ pi
t ¼ t1 r ¼ rw p ¼ pwf
r = re p ¼ pi

The result of the Boltzmann transformation is the exponential integral


equation:
 
qm fmct r2
pwf ¼ pi þ Ei ð2:9Þ
4phk 4kt

where the symbol Ei denotes exponential integral and pwf the flowing pressure at
the bottom of the well.
In an infinitely large homogeneous reservoir system with a line source
as the well, the flowing pressure at any point can be calculated from
Equation (2.9) as a function of the radius r and the time t. The natural
92 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

logarithm can be used in place of the exponential integral provided the argu-
ment of the exponential integral is significantly smaller than the limiting value
of 0.01 (83).

qm gfmct rw 2
pwf ¼ pi þ ln ð2:10Þ
4phk 4kt

where g denotes the Euler constant.


To reconstruct the situation in a deposit, the deposit’s hydraulic system
must be transformed back to its original state. This can be done by, for example,
closing off an operating well. The pressure in the well will then gradually build
up again until it reaches the original reservoir pressure, assuming that the
deposit is comparable to an infinitely large reservoir system (influence of aqui-
fers). The shut-in of the well is simulated mathematically by superimposing a
negative flow rate q on the radial flow system. Combination of the two equations
for the influx into the well and the out flow from the well gives an equation for
the shut-in pressure pws of the following form:

qm tp þ t
pws ¼ pi  ln ð2:11Þ
4phk t

where tf ¼ tp denotes the flow duration and Dt the shut-in time (84). This equation
has been used to extrapolate the initial reservoir pressure pi from the graphical
presentation of the shut-in pressure plotted against the logarithm of the time
t þt
ratio ð p t ¼ 1Þ (Fig. 20).
For an infinitely large reservoir the initial pressure in the well will
t þt
be reconstituted after an infinite shut-in time pt ¼ 1: The gradient of the

tp þt
Fig. 20. Shut-in pressure pWS versus logarithmic time ratio t for extrapolation of
initial reservoir pressure pi (83).
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 93

straight line m corresponds to the coefficient in Equation (2.11) in front of


the logarithmic function. As the flow rate is normally determined in a mea-
suring tank above ground, it has to be transferred to the state in the deposit.
This is done by incorporating the formation volume factor B in the defining
equation for m.

qmB

4phk

From the gradient m the permeability capacity k h can be calculated for


the reservoir system.

qmB
kh ¼
4pm

In addition to this simple radial model with the selected initial and bound-
ary conditions many methods exist for finding solutions for specific reservoir con-
ditions, which can be applied to individual well and reservoir problems.
Individual solutions have therefore been found for the differential equations to
the following conditions; finite reservoir extension, near wellbore effects, frac-
tured reservoirs, stratified reservoirs, and partial penetration and particular
direction of the well in the producing formation.
Multiphase Flow. Capillary Pressure and Saturation. If there is
more than one phase in the pore space system of a reservoir rock then the distri-
bution of the phases is determined by the interfacial tensions between the fluids
and the solid phase. The equilibrium of forces between two liquid phases (oil and
water) and a solid phase is shown in Figure 21.
The wetting angle y is a measure of the wettability of the solid phase
by the oil and water phases. A value of 08 for y represents an ideally water
wet solid, whereas 1808 defines a completely oil wet surface. The distri-
bution of the phases for a pore cross section therefore can be idealized as
follows:

– Water, as the most strongly wetting phase, is positioned directly against


the surface of the rock

Fig. 21. Equilibrium of forces between two liquid phases (water and oil) and the solid
phase (77). sso = interfacial tension solid – oil; ssw = interfacial tension solid – water; swo
interfacial tension water – oil.
94 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 22. Idealized distribution of oil, water, and gas phase in a pore cross section.
a) Water; b) Gas; c) Oil; d) Rock.

– Gas, which is the least wetting, is positioned in the center of the pore duct
– The oil phase is distributed between them (Fig. 22)
Due to the adhesive forces the pressure in the wetting phase (water) is
lower than in the nonwetting phase (oil). The difference between these two pres-
sures is the capillary pressure pc where

pc ¼ po  pw

Capillary pressure curves (see Fig. 23) are of fundamental importance for
the mobility of the phases in multiphase flow. They determine the irreducible,
and, therefore, fixed saturation of the wetting phase at infinite capillary pressure
and characterize the capillary ascent of the water table at the boundary of the
oil – water zones. The capillary pressure can be directly transferred into eleva-
tion by use of the capillary/gravity force equilibrium.

2sow cosy
¼ ð%w  do Þgh
r

where sow denotes the interfacial tension between oil and water phase, r the
equivalent hydraulic radius of the porous system, %w the density of water, do
the density of oil, g the gravitational constant, and h the elevation above the
water table.
For the same capillary pressure, a higher water saturation is found in a
pore space system of lower permeability (a) than in a highly permeable
one (b). If it is assumed that water represents the wetting phase this means a
reinforced capillary imbibition of water from zones of high permeability into
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 95

Fig. 23. Capillary pressure versus wetting fluid saturation (85). a) Low permeability;
b) High permeability.

low permeability and therefore an improved displacement of oil from the pores of
lower permeability.
Relative Permeabilities. The mobilities of the individual phases in mul-
tiphase flow in a porous medium depend on the wettability of the rock, on the
interfacial tension between the fluids, and on their viscosities. A new flow coeffi-
cient, which is introduced into the phase flow equation as the relative permeabil-
ity, is needed to describe the different mobilities of the phases in the medium as a
function of their fractional saturation. Relative permeability kr is defined as the
ratio of the effective permeability of a phase (kw, ko, kg) in the presence of one or
more adjacent phases to a reference permeability. The absolute rock permeabil-
ity according to Darcy, or the permeability to water at 100 % water saturation,
or the permeability to oil for irreducible water saturation can be used as the
reference permeability. The relative permeabilities for a binary oil – water sys-
tem for a more strongly water-wet and a more strongly oil-wet rock are shown
in Figure 24. If the saturation of the wetting phase Sw is plotted on the abscissa
then the curves are displaced to the right with increasing water wettability. On
the other hand, the final value for the relative permeability of the water phase
increases if the water wettability decreases.
This type of representation can only be used for two-phase permeabilities;
for a three-phase system STONE (87) has proposed a paired representation of the
water – oil and gas – oil relative permeabilities.
96 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 24. Relative permeabilities for a binary oil – water system (86). A) Water-wet rock;
B) Oil-wet rock.

Apart from this, a ternary diagram is the only possible way of representing
the relative permeabilities of a three-phase system.
The flow equations for a two-phase system can therefore be represented
while taking account of the functions of the relative permeabilities and the capil-
lary pressure.

Oil:
   
kro k%o qSw o qw
r  ro ¼  q 
mo q pc qt qt

Water:
   
krw k%w qSw qo qw
r  rw ¼ f 
mw q pc qt qt

where F is the flow potential:

1
 ð p þ %gzÞ
%

and kro and krw denote the relative permeabilities of oil and water, respectively.
Depending on the boundary conditions, this system of partial differential
equations, can, in general, only be solved numerically.
Drive Mechanisms and Recovery Efficiency. The energy reservoir for
the flow of oil, gas, and water in a porous medium consists of the compression
energy of the aquifer volume surrounding the deposit; the mechanical energy
of the gas liberation process; the influence of gravity; and the capillary pressure
effect.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 97

Fig. 25. Binary diagram for a gas – oil mixture (each phase consisting of one component)
(77). A = bubble point pressure (pressure depletion at room temperature); A0 = initial reser-
voir pressure; C = critical point.

As shown in Section 2.2, the initial pressure in a deposit is mainly hydro-


static. The ratio of the initial pressure to the saturation pressure of the oil – gas
mixture characterizes the individual steps of the oil recovery process and the
drive mechanism. Figure 25 shows a simplified binary diagram for a mixture
of gas and oil, each phase consisting of one component.
In this p – T diagram the oil recovery process is considered isothermal.
If the initial reservoir pressure pi lies above the bubble point pressure pb
the deposit is termed a gas undersaturated oil deposit. When the pressure
is released the volume expansion drive alone is initially effective. When the
pressure falls below the bubble point pressure, this is supplemented by
the mechanical effect of the gas liberation process. If the bubble point pres-
sure and initial reservoir pressure are equal then the deposit is termed a
gas-saturated oil deposit. If the initial reservoir pressure lies below the bubble
point pressure then the expansion effect of a gas cap can provide an additional
drive mechanism. From Figure 25 it can also be seen that no oil deposits
exist to the right of the critical point C and that the gas reservoir or
condensate phase type then begins. The influence of the different drive
mechanisms is best calculated from a material balance. The physical and
mathematical treatment of these material balances assumes closed sys-
tems in which removal of fluid results in a reduction of pressure. If, however,
the removal of fluid is replenished by an aquifer feed, or by water or gas injec-
tion then the system pressure remains essentially unchanged. A pressure
98 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

change in the system also changes the properties of the phases in a way
which is disadvantageous to the oil recovery process; measures which
maintain the pressure are, therefore, favorable. Water drive is by far the
most effective drive mechanism; it can be introduced by artificial flooding
measures in locations where water cannot be fed from natural sources (ie,
inadequate aquifer size or inadequate energy replenishment due to low aqui-
fer permeability). The oil recovery process by natural aquifer expansion and
by the other drive mechanisms mentioned rate as primary extraction pro-
cesses. On the other hand, replenishment of the reservoir energy by water
or gas injection is one of the secondary extraction processes.
The recovery efficiency of a deposit is given by the percentage of the
cumulative production of the quantity of hydrocarbon originally present. The
performance of extraction and the recovery efficiency are determined largely
by the geological factors of the deposit and the physics of the oil recovery pro-
cess. The physical approach to oil recovery processes presupposes a micro-
scopic dimension, and the geological approach a macroscopic one; the
recovery factor must, therefore, include microscopic and macroscopic para-
meters. The total recovery factor or recovery efficiency ER is the product of
the microscopic recovery factor ED, the areal recovery factor EA, and the ver-
tical recovery factor (flood invasion efficiency) EI. Figure 26 shows that the
recovery efficiencies lie between 20 % and 40 % of the oil present, depending
on the characteristic drive mechanisms and oil viscosities. In gas reservoirs
the recovery efficiency lies between 80 and 90 % of the gas present. This is

Fig. 26. Oil recovery efficiency as a function of pressure for various drive mechanisms
(88). a) Depletion drive; b) External gas drive (small gas cap); c) External gas drive (large
gas cap); d) Water drive.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 99

attributable to the high mobility of the gas and the favorable form of the
volume expansion drive.
Microscopic Recovery Factor. BUCKLEY, LEVERETT, and WELGE (89–91)
have developed the theory of the displacement of a fluid by a second, immiscible
fluid in a homogeneous medium. This microscopic process determines which fluid
exhibits the stronger wetting properties towards the porous medium and which
has the greater mobility. The displacement process in the pore space is not as
perfect as the displacement of a fluid by a piston in a machine; this is because
a mobile residual phase saturation of the displaced fluid (oil) remains after the
displacement front has passed. Continued flooding can reduce this mobile resi-
dual phase saturation down to an irreducible, fixed percentage. Through a linear
flooding process it is possible to determine the saturation distribution S along the
flooding section, ie, in direction x (Fig. 27).

Fig. 27. A) Saturation distribution along flooding section (direction x); B) Fractional flow
for water along flooding section.
100 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

At the inlet cross section the water saturation is only slightly smaller than
the difference of 1 minus the physical residual oil saturation Sor. Up to the flood-
ing front, the water saturation decreases to a value Swf (front saturation),
depending on the mobilities of the phases. The difference between the ultimate
water saturation (1  Sor) and the actual water saturation at any section is
defined by the profile f, the saturation of mobile oil. The front is characterized
by a step change in the water saturation down to the irreducible water saturation
Swi. The proportions of the flowing phases change along this saturation profile
and can be expressed by the fractional flow f of one of the two flowing phases,
as it is shown for different times t1 and t2 in Figure 27 B.

qw 1
fw ¼ ¼
qo þ qw 1 þ M1

lw krw mo
M¼ ¼
lo kro mw

The fractional flow for water is a function of the mobility ratio M, ie, the
mobility of water kw/mw as the displacing phase to the mobility of the oil ko/mo
as the displaced medium. At the front the fractional flow of water falls to zero,
as only the mobile oil phase can flow ahead of the front. By using a continuity
approach in the porous medium the velocity has been determined at which the
front or the saturation profile moves through a pore continuum (90). If an infi-
nitely small volume unit is considered at the boundary between oil and water
then the difference between input and output water flow volume in the time
unit Dt must be equal to the change in saturation in the space element A Dx
(see Fig. 28).

tðq fw1  q fw2 Þ ¼ A xfðSw1  Sw2 Þ

This finite difference equation can then be transformed into a differ-


ential equation which gives the velocity of propagation of a point on the

Fig. 28. Mass balance for an infinitely small volume unit at the saturation boundary
between oil and water.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 101

saturation profile.

lim q fw ut ¼ AxfSw


x !0
t !0

qx q q fw
¼
qt Af qSw

Integration between the limits x and 0 and the times t and 0 gives the equa-
tion for determining position x which has reached some arbitrary saturation
value at the time t after injection of a certain volume of water Wi.

W i q fw

Af qSw

Zt
Wi ¼ qdt
0

If, instead of some arbitrary saturation, the intention is to characterize the


saturation at the front, and hence the front velocity vf, then Figure 29 illustrates
the graphic method for determining the front saturation.
The saturation at the displacement front Swf is defined by the point where
the tangent to the fractional flow curve from its origin intersects the curve itself.
An extension of this tangent until it intersects the fractional flow value 1, char-
acterizes the average water saturation Swbt as the front breaks through at the
end of the flooding section. The difference 1  Sor  Swbt provides a measure for
the residual oil which can be mobilized. In practice this mobilizable residual oil
can only be extracted after an infinitely long flooding time. The greater the mobi-
lity of the displaced phase in comparison with the displacing phase the more
rapidly the residual oil saturation reaches a quasi steady-state final condition
Sor. This physical residual oil saturation serves as a measure for establishing
the microscopic oil recovery factor ED.

Soi  Sor Sor


ED ¼ ¼1
Soi Soi

Areal Recovery Factor. Consider a homogeneous deposit with no vertical


stratification and a concentric arrangement of the peripheral production wells at
a distance r from an injection well. Here the areal oil recovery factor EA, ie, the
ratio between swept area AS and the total area A, would be equal to 1 by the time
the flooding front has reached the peripheral production wells.
For a five-point well pattern as shown in Figure 30 a considerable propor-
tion of the area A remains unswept when the flooding front reaches the four
peripheral production wells. The areal oil recovery factor which can be achieved
102 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 29. Fractional flow curve for water, definition of the front saturation Swf and break
through saturation Swbt according to Welge (92).

Fig. 30. Concentric arrangement of four peripheral production with respect to one injec-
tion well, 5-point pattern.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 103

before breakthrough of the flooding front depends on the mobility ratio of the
front and the arrangement of the wells.
In a homogeneous producing formation the well spacing initially plays no
part. However, as the capacity of the producing formation to accept the injection
medium (water) depends on permeability, the well spacing cannot be set at any
arbitrary distance. There is little point in optimizing the well spacing on the
basis of a homogeneous deposit model, as in practice more complex deposit for-
mations are often found.
Vertical Recovery Factor or Flood Invasion Efficiency. If the oil pro-
ducing formation were to have no vertical subdivisions then the overall oil recov-
ery factor from this deposit would be the product of the microscopic and areal oil
recovery factors. However, the conditions under which the sediments were
deposited have normally led to a vertical, facially subdivided stratification,
with the result that the vertical recovery factor also has to be taken into account
when determining the overall oil recovery.
If water is invading into a stratified producing formation from an aquifer or
an injection well the strata fill in the order of their permeabilities. The most
permeable layer is flooded first.
DYKSTRA and PARSONS (93) have developed a simple theory for determining
the vertical recovery factor in a stratified producing formation. For this purpose
they have further simplified the microscopic displacement process as described
by BUCKLEY and LEVERETT and have assumed that no mobile residual oil remains
behind the front. However, unlike BUCKLEY and LEVERETT, they have taken into
account the fact that the velocity of the displacement front depends on the mobi-
lity ratio of the phases and on the position of the displacement front along the
flooding section. The higher the mobility of the displacing phase the more rapidly
the flooding front extends toward the end of the flooding section. The flooding
velocity at the inlet corresponds to the effective Darcy flow velocity of the oil
phase (see Fig. 31).

Fig. 31. Relative velocity of flooding front as function of fractional distance for different
mobilities of displacing phase (94). a) M ¼ 0.1; b) M ¼ 0.2; c) M ¼ 0.5; d) M ¼ 1; e) M ¼ 2; f)
M ¼ 5; g) M ¼ 10.
104 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 32. Cross section of a layered reservoir; position of flood front in the individual
layers at a certain time level t (94).

The flood invasion efficiency EI is given by integration of the water satura-


tion profile in Figure 32 using the following formula

P
i P
n
xj
hj þ l hj
1 iþ1
EI ¼
P
n
hj
1

The position of the flooding front at time t in the incompletely swept strata
j ¼ i + 1, relative to the layer i in which the water has just broken through, is
determined by the equation
 0:5
2 kj 2
M þ ki ð1  M Þ
xj M
¼ 
l 1M 1M

where xj/l is the position of the flooding front.


The following formula applies to the time until breakthrough ti

fl2 ðM þ 1Þmw
ti ¼ 
2kwi  p

With the aid of these three equations the vertical oil recovery factor can be
determined at any given time. The product of the three factors gives the overall
recovery efficiency ER:

ED ¼ ED EA EI

2.3. Modeling of Reservoir and Production Performance. An effec-


tive reservoir characterization comprises a description of the reservoir’s geologi-
cal boundaries, the petrophysical properties of the rock, and the material
properties of the contents. Its aim is to calculate the production behavior and
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 105

plan measures to optimize production from the reservoir. The calculation meth-
ods used are the basis for predictions of future production performance and are a
strategic instrument of reservoir management. All methods of prediction are
based on the physical fundamentals of the recovery process and the flow pro-
perties of the fluids in a geologically complex reservoir body. The credibility
and reliability of these prediction methods depend on how closely the calculation
techniques model the reality of the recovery process and the reservoir. This
accuracy of the model must be checked by a model validation before use. Once
the model is chosen it must be adjusted to the production history of the reservoir
(model calibration). The better the model is calibrated, the more reliable the
prediction.
The following models for describing the reservoir behavior, with increas-
ing complexity of process description and/or reservoir characterization, can be
distinguished:

1) Tank models (material balances)


2) Combined tank – flow model (Tarner prediction method)
3) Multiphase flow models (reservoir simulation)

Material Balances Method. The tank model is the simplest form of recon-
struction of the reservoir contents. It is based on the conservation of mass and an
equation of state (88). All forms of material balance use the following approach:

Original contents  Quantity produced ¼ Quantity remaining

In order to set up a material balance the following assumptions must be


made:

1) The reservoir is considered as an energetic unit


2) The material properties and the pressure can be described by mean values
representative of the reservoir
3) The material laws for the pore contents can be determined with sufficient
accuracy

Material Balance for a Gas-Undersaturated Oil Reservoir Without


Aquifer Influence. This material balance is presented as an example.

Initial State:

1) The initial pressure pi exceeds the bubble point pressure pb


2) The reservoir contents consist of the volume of gas-undersaturated oil NBoi
and the volume of initial water Vwi
106 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

N Boi
Vpi ¼
1  Swi
Vwi ¼ Vpi Swi

(for explanation of terms see the abbreviations list at the beginning of this
article).

State Above the Bubble Point Pressure pb:

1) The actual pressure exceeds the bubble point pressure pb but is lower
than pi.
2) The oil volume NpBo is produced. The volume lost is replaced by the expan-
sion of the rock matrix, the residual oil, and the initial water.

Vp ¼ Vpi ½1  cf ð pi  pÞ
Vw ¼ Vwi ½1 þ cw ð pi  pÞ
ðN  Np ÞBo ¼ Residual oil

(for explanation of terms see the abbreviations list at the beginning of this
article).
State Below the Bubble Point Pressure pb:

1) The reservoir pressure p is less than the bubble point pressure pb.
2) Aside from the volume expansion of oil, initial water, and rock matrix, the
release of gas from solution occurs as a further drive process. To simplify
the calculation of the equilibrium behavior as far as possible, the compres-
sibility equations for the oil, water, and rock phases as well as the real gas
equation for the gas phase are used as material laws.

Vp ¼ Vpi ð1  cf ð pi  pÞÞ
Vw ¼ Vwi ½1 þ cw ð pi  pÞ
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 107

From the material balance for gas it follows that:

Free gas ¼ ðNRsi  Gp  ðN  Np ÞRs ÞBg ð2:12Þ

From the displacement process in the pore space it follows that:

Free gas ¼ Vp  ðVw  Wp Bw Þ  ðN  Np ÞBo ð2:13Þ


Vw  Wp Bw ¼ net water expansion
ðN  Np ÞBo ¼ residual oil volume

Equating (2.12) to (2.13) gives:

NRsi Bg  Gp Bg  ðN  Np ÞRs Bg ¼ Vp  Vw þ Wp Bw  ðN  Np ÞBo

Separating reservoir and production terms:

N Boi
N Rsi Bg  N Rs Bg þ N Bo þ ðSwi ð1 þ cw  pÞ  1 þ cf  pÞ
1  Swi

¼ Gp Bg  Np Rs Bg þ Wp Bw þ Np Bo

Combining and rearranging the terms with

G p ¼ N p Rp
gives
 
Boi
N ðRsi  Rs Þ Bg þ Bo þ ððcf þ Swi cw Þ p  ð1  Swi ÞÞ
1  Swi

¼ Np ½ðRp  Rs Þ þ Bo þ Wp Bw

Further rearrangements with

Bo  Boi
¼ co
Boi  p
gives

 
Bg Bo  Boi cf þ swi cw
N Boi ðRsi  Rs Þ þ þ p
Boi Boi 1  Swi
 
Bg co Soi þ cw Swi þ cf
N Boi ðRsi  Rs Þ þ p
Boi 1  Swi

¼ Np ðBo þ Bg ðRp  Rs ÞÞ þ Wp Bw
108 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

with

co Soi þ cw Swi þ cf
ceff ¼
1  Swi

it follows that

Wp Bw
Np Bg ðRsi  Rs Þ þ ceff  pBoi  N
¼ ð2:14Þ
N Bo þ Bg ðRp  Rs Þ

The ratio Np/N in Equation (2.14) indicates the recoverable quantity of


oil — as a fraction of the oil in place — for the dissolved-gas drive mechanism
(first term of the equation) and for the fluid- or rock-expansion drive mechanism
(second term of the equation). The third term is normally negligible for reservoirs
without an aquifer.
Tarner Prediction Method. A tank model is only capable of describing the
energetic aspect of the recovery process. For a complete characterization the ade-
quate inclusion of the flow processes in the porous reservoir rock is absent.
TARNER solved this problem by a combination of a material balance and flow equa-
tions based on the Darcy law (95).
The calculation is as follows:
For a differential pressure change in the closed reservoir, the expected
cumulative oil production is predicted and the cumulative gas – oil ratio
calculated by using a gas material balance. To check the accuracy of the
assumption, the change in saturation in the porous rock caused by fluid
removal and the resulting relative permeabilities for gas and oil are deter-
mined. From the flow equations for the gas and oil phases a gas – oil ratio
results which is compared with that calculated from the material balance.
If the results are identical, the predicted quantity of oil produced during
an incremental pressure change is correct. Otherwise, the calculation
must be repeated with a new assumption until the two gas – oil ratios are
equal.
In the following, the equations for the calculation of the two gas – oil ratios
are derived by way of example for a gas-saturated oil reservoir.
Material Balance for Gas. Original gas content ¼ Quantity of gas pro-
duced þ residual gas
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 109

Material Balance for Gas

N Rsi ¼ ½N Boi  ðN  Np Þ Bo 1=Bg þ ðN  Np Þ Rs þ Gp


Gas cap content Dissolved gas

Tz p0
Bg ¼
pT0

Gp ¼ Np Rp

qg
Rp ¼ when qg and q0 ¼ constant
q0

Separating the variables:

N Rsi Bg  N Boi þ N Bo  N Rs Bg ¼ Np Rp Bg þ Np Bo  Np Rs Bg

N½Bo þ Bg ðRsi  Rs Þ  Boi ¼ Np ðBo þ Rp Bg  Rsi Bg þ Rsi Bg  Rs Bg Þ

with

Bo þ ðRsi  Rs ÞBg ¼ Bt

Boi ¼ Bti

it follows that:

N½Bt  Bti ¼ Np ½Bt þ Bg ðRp  Rsi Þ

Np Bt  Bti
¼ ð2:16Þ
N Bt þ Bg ðRp  Rsi Þ

or solved with respect to Np Rp:

NðBt  Bti Þ  Np ðBt  Rsi Bg Þ


Gp1 ¼ Np Rp1 ¼
Bg

The index 1 refers to the gas – oil ratio derived from the material balance.

Change of saturation

OOIP ¼ NBoi

Residual oil ¼ ðN  Np ÞBo


110 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

N Boi
Vp ¼
1  Swi
ðN  Np ÞBo
So ¼ ð1  Swi Þ ð2:17Þ
N Boi
Np Bo
So ¼ ð1  Þ ð1  Swi Þ
N Boi

Dynamic gas – oil ratio equation

qg qo Rs
  þ
qg Bg Bo
Rp2 ¼ ¼ qo
qo To ; po
Bo

The index 2 refers to the gas – oil ratio derived from flow equation.

qg Bo
Rp2 ¼ þ Rs
qo Bg

Darcy equation (2 immisible phases):

kg A
qg ¼  r pg
mg

when rpc is rather low it follows rpg ’ rpo

ko A
qo ¼  r po
mo
kg mo Bo
Rp ¼ þRs ð2:18Þ
ko mg Bg
Gp2 ¼ Np Rp2

Each pressure stage can be graphically extrapolated to the convergence of the


DGp values or numerically iterated (see Fig. 33).
Figure 34 shows the typical development of the produced gas–oil ratio Rp
plotted against the cumulative oil produced. The gas – oil ratio curve above the
bubble point pressure (1–2) is constant; when the pressure falls below pb then
initially no free gas reaches the well as long as the required critical gas satura-
tion is not reached (2–3) in the surroundings.
Then follows a steep rise of the gas – oil ratio (3–4), resulting from the rapid
transport of the free gas. The quantity of dissolved gas, which is the source of the
gas yield, is limited. Therefore with advancing production time the cumulative
gas – oil ratio reaches a maximum, which is then followed by a rapid drop
(4–5) until the free gas is exhausted. In the last phase of production (5–6) it is
principally the gas dissolved in the residual oil that is produced.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 111

Fig. 33. Graphical method of solution in the Tarner method. a) Material balance; b) Flow
equation.

Reservoir Simulation. The prediction methods discussed in the previous


sections are applicable only to problems in which the reservoir can be described
by uniform mean properties of the rock and the pore fluids. The individual char-
acter of a group of wells or the characteristic drainage volume of their reservoir,
and the occurrence of multiphase flow can be described only by reservoir simu-
lation. This entails modeling the whole body of the reservoir and solving, simul-
taneously or stepwise, the coupled differential equations for pressure and
saturation in each of its volume elements. This section can only give a short
introduction to the basic principles of the numerical simulation of reservoir
behavior. For further details, see (96), (97). Here the mathematical foundations
for an engineering understanding will be treated with reference to two simple
models (98).

Fig. 34. Influence of the critical gas saturation on Rp (88).


112 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 35. Grid model for steady-state flow (98).

Model for Steady-State Flow. A five-point well pattern with an injec-


tion well at the center of the square is taken as the basis. According to symmetry
considerations, in a homogeneous formation, only one-eighth of the grid area
including the injection well and a production well, need be shown. Figure 35
shows the pattern section dividing the reservoir area into square blocks, the
representative portion being bound by two orthogonal sides and the diagonal
between injection well and production well.
At the center of each block there is a node at which all the block’s properties
are concentrated. The flow geometry in all blocks is linear, except for the drai-
nage areas of the two wells. Here, radial flow geometry prevails and, instead
of a node, there is an external radius re and a well radius rw, representing the
equivalent block boundaries. An incompressible medium is assumed to be
injected into well 1 and produced from well 15. Because of the prevailing pres-
sure regime, outflow from the nodes is possible only to the north or east and
inflow to the nodes only from the south or west. For steady-state flow it is
assumed that for each node the sum of inflows and outflows at every instant is
zero. This is illustrated here for node 8:

q78 þ q58 ¼ q812 þ q89 ð2:19Þ

If the permeability is equal in all nodes (ie, in all blocks) and the areas of
flow are likewise identical (since the block width is constant), the general form of
flow equation for the Darcy law (see Section 2.2.1.1) can be written as follows:

Ak
qi ¼   pi ð2:20Þ
mx

Since the coefficients before Dpi are constant in all terms, Equation (2.21)
can be written in place of Equation (2.19)

ðq7  q8 Þ þ ðq5  q8 Þ ¼ ðq8  q12 Þ þ ðq8  q9 Þ ð2:21Þ


Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 113

For node 8 there then follows the pressure balance set up in Equation (2.21)
or rearranged to:

q5 þ q7  4 q8 þ q9 þ q12 ¼ 0

Such a pressure balance can be set up for each of the 13 nodes, in addition
radial flow conditions must be included in the pressure balance for nodes 1 and
15. The 15 equations with 15 unknowns are arranged in matrix form and solved
for the pressure values at the nodes with, eg, the Gauss – Jordan or Gauss – Sei-
del method. The calculation can also be performed with different permeabilities
in the individual blocks.
Model for Unsteady-State Flow. If the same model is used, but instead
of an incompressible fluid a compressible medium is injected into the well, then,
instead of a material balance for the individual volume flows in the nodes, a bal-
ance for the mass flows must be set up. As already derived earlier from the mass
balance in a cylindrical volume element, a variant of the diffusivity equation
must be used to describe unsteady-state flows. For linear flow through the
volume element of edge length Dx, the following differential equation can be writ-
ten for the change of flow along the distance Dx:
   
kA p kA p
t  ð2:22Þ
m x xþx m x x

t ¼ Axfc p

After separating the variables Dx and Dt, the diffusivity equation for
unsteady-state flow is obtained (Eq. 2.23):

p k ð p=xÞ
¼ ð2:23Þ
t fmc x

This differential equation is solved by the finite difference method. In this


procedure the pressure gradient Dp/Dx is regarded as the difference of pressure
in two successive segments, divided by the distance between the centers of the
segments. Figure 36 shows such a linear sequence of segments.
p p j  p j1

¼ ð2:24Þ
x j1; j x
 p p jþ1  p j

¼ ð2:25Þ
x j; jþ1 x

ð p=xÞ ð p jþ1  p j Þ=x  ð p j  p j1 Þ=x


¼ ð2:26Þ
x x

p jþ1  2 p j þ p j1
¼
x2
114 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 36. Linear sequence model.

The pressure gradient between segments j  1 and j is calculated accord-


ing to Equation (2.24), and the pressure gradient between segments j and
j þ 1 according to Equation (2.25). The change of pressure gradient between
segments j  1 and j þ 1 is then shown by Equation (2.26). For unsteady-
state processes, the change of pressure in the time Dt is superimposed on
the change of pressure with position Dx. In Figure 36, the index j designates
the position and the index n the time. In the lower part of the Figure 36 the
pressure variation at times n and n þ 1 is shown. If Equation (2.23) is applied
in general form to define the pressure variation at the beginning and at the
end of the time interval, Equations (2.27) and (2.28) respectively result (see
below).
The pressure variation at the beginning of the time interval is given by:

p j;n  p j;nþ1 k p jþ1;n  2 p j;n þ p j1;n


¼ ð2:27Þ
tnþ1  tn fmc x2

and at the end of the time interval by:

p j;n  p j;nþ1 k p jþ1;nþ1  2 p j;nþ1 þ p j1;nþ1


¼ ð2:28Þ
tnþ1  tn fmc x2

Equations (2.27) and (2.28) can give the same result only when the time
interval Dtn approaches zero. If the pressures at time n in all segments are
known, the pressure in element j at time n þ 1 can be calculated with Equa-
tion (2.27). This method is known as the explicit solution method. The advantage
of the explicit solution method is that it is very easy to program. However, it is
only accurate if very small time steps are chosen.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 115

In contrast to the explicit solution (Eq. 2.27) in which there is only one
unknown, ie, the pressure pj at time n þ 1, three unknown values are present
in Equation (2.28) at the end of the interval (ie, pj1, pj, and pj+1).
If, however, Equation (2.28) is formulated for all three nodes, j  1, j and
j þ 1, the three unknown pressures can be determined from the three equations.
The equations are arranged in matrix form and solved simultaneously either by
the Gauss – Seidel or by the Gauss – Jordan method. This procedure is known
as the implicit solution method and has the advantage of yielding sufficiently
accurate results even at quite large time intervals. For modeling problems in
multiphase flow, the saturation equation for each block of the reservoir model
must be solved in addition to the pressure equation. Hence, the chosen block
size of a reservoir model is influenced not only by the availability of geological
and petrophysical information but also by the accuracy of the numerical compu-
tation method. In regions of the reservoir where relatively rapid changes occur in
the transport characteristics — such as pressure gradient, transmissibilities, and
saturation changes — fine-mesh grids are necessary for numerical accuracy of
the solution method. For complex reservoir studies, simulation models with sev-
eral ten thousand blocks can be produced which require considerable computer
time. There are numerous simulation codes with a dynamicized fine grid size
that follows the movement of flood fronts.
The powerful computer programs currently available to the natural gas and
petroleum industry not only take into account simple processes such as gas dis-
solution, gravitational force, and capillary pressure effects but also can describe
the infuence of compaction in unconsolidated formations and the influence of
the diffusion and sorption of components. They can also treat particular rheologi-
cal properties of the phases, eg, the structural viscosity of polymer solutions in
polymer flooding the phase behavior during mixing processes, temperature-
controlled fractionation of materials (distillation and cracking), thermal conduction,
and thermal convection. A survey of common simulator types is given in Table 3.
Statistical Prediction Methods (Decline Curve Methods). Aside from
tank models or flow models to describe and predict reservoir behavior, statistical

Table 3. Types of reservoir simulators


Description Physical/chemical processes
Multiphase flow simulator multiphase flow in porous/fissured media solution of gas in oil
properties of phases depend on pressure and temperature
Compositional simulator mixing of phases equilibrium of components between the
phases dispersion flow properties depend on interfacial
tension properties of phases depend on composition
Thermal simulator like compositional simulator, in addition : thermal convection,
conduction, and radiation flow properties of phase depend
on temperature chemical reaction of components of a solid
phase (oxidation)
Chemical flood simulator adsorption/desorption of components retention of phases
rheology of aqueous phases properties depend on salt
content of the aqueous phase mass transfer between
aqueous phase and rock
116 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

prediction methods also can be used. These methods are based on observation of
the decline of production or pressure in a reservoir; analysis of the mathematical
trend; and extrapolation of the decline of production or pressure over the period
of prediction. The observed trend curves often show uniform patterns. This is due
to the fact that the change in the flow rate is proportional to the pressure change,
and the pressure change is linked logarithmically, as a result of the compressi-
bility relation, with the volume removed from a closed system. In many cases,
therefore, an exponential decline of the production rate with time can be derived
from the trend analysis.
To characterize the decline curve analysis method, some basic definitions
are given in advance (98). The fractional production decline is defined as the
change of flow rate divided by the initial flow rate

dq
 ¼ fractional production decline
q

The change of the fractional production decline with time is called the pro-
duction decline rate D.

dq
 ¼D
q dt

In a closed or semiclosed reservoir with no infinitely large, fully active aquifer


available for energy replacement, three types of production rate decline can be
distinguished:

1) Constant production decline rate:

D ¼ a0 q0 exponential production decline

2) Linearly proportional production decline rate:

D ¼ a0 q1 harmonic production decline

3) Production decline rate is proportional to a power function of the produc-


tion rate:

D ¼ a0 qn hyperbolic production decline

In order to permit simple prediction methods, the production rate or a func-


tion of the production rate is plotted against a function of time or against cumu-
lative production, choosing scales such that straight lines are obtained. The basic
method is explained with the aid of exponential and harmonic production
declines as examples.
Exponential Production Decline. Figure 37 shows a semilogarithmic
plot of the production rate versus time. According to Equation (2.29) the gradient
of the straight line is a measure of the production decline rate D.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 117

Fig. 37. Exponential production decline, production rate versus time.

dq
D¼ ¼ a0 q0 ¼ a0
q dt

after integration

Zq Zt
dq
¼ a0 dt
q
q0 0

q
ln ¼ a0 t ¼ Dt
q0
q
¼ eDt
q0 ð2:29Þ

By combining the variables q and t a new quantity — the cumulative produc-


tion Np — is obtained.
tN p
Z
Np ¼ qdt
0
tNp
Zq Z
dq
¼ a0 dt
q
q0 0
tNp
Zq Z
0
dq ¼ a qdt
q0 0

q ¼ q0  DNp ð2:30Þ

A plot of production rate q versus Np is shown in Figure 38. From this plot
the production decline rate D is also obtained from the gradient. As soon as
118 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 38. Exponential production decline, production rate versus cumulative production.

the production rate of a well falls below a marginal value determined by the cost,
production must be stopped. The corresponding cumulative amount produced
can be determined graphically by extrapolating the production decline curve to
the known limiting production rate qa. The difference between the maximum cu-
mulative production Npa and the present cumulative production Np gives the re-
serves of the well or reservoir.
Harmonic Production Decline. The defining equation for harmonic
production decline is:

dq
D¼0 ¼ a0 q
qdt

Initial conditions are: when t = 0, D = D0 and q ¼ q0,

D0
D0 ¼ a0 q0 ; a0 ¼
q0

After separating the variables and integrating

Zq Zt
dq 0
 ¼ a dt
q2
q0 0

1 1 D0
 ¼ t
q q0 q0

q0

1 þ D0 t
q
log ¼ logð1 þ D0 tÞ ð2:31Þ
q0
Next Page

Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 119

Fig. 39. Harmonic production decline, production rate plotted against cumulative
production.

For harmonic production decline, the double logarithmic plot of the produc-
tion rate against the time function (1 + D0t ) gives a straight line of gradient  1.
Again, if a new quantity is introduced by combining q and t a semilogarithmic
relation between production rate and cumulative production is obtained after
integration (see Fig. 39).

tNp
Z
Np ¼ qdt
0
tN p
Zq Z
dq
 ¼ a0 qdt
q
q0 0

q D0
ln ¼ Np ð2:32Þ
q0 q0

With the appropriate plot, a straight line of gradient  D0/q0 is obtained,


that again — if the limiting production rate qa is known – can be extrapolated
to give the maximum cumulative production Npa.
Hyperbolic Production Decline. From the equation for the hyperbolic
production decline rate, after rearrangement of the variables and integration,
the following relations are obtained:
Initial condition: t ¼ 0, q ¼ q0, D ¼ D0

dq D0
D¼ ¼ a0 qn a0 ¼
qdt qn0
Previous Page

120 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 40. Hyperbolic production decline, production rate plotted against the time
function.

q0
q¼ ð2:33Þ
ð1 þ nD0 tÞ1=n
 
qn0 1 1
Np ¼  ð2:34Þ
ð1  nÞD0 qn1 0 qn1

A plot of production rate versus log (1 + n D0 t ) is given in Figure 40.


The analysis of the production decline can also be based on the fraction of
oil in the total production rate or on the water cut in the production, instead of
on the oil production rate. The same relations then apply. In order to eliminate
the influence of individual measurement inaccuracies or of unrepresentative
sampling, it is advisable to integrate or smooth the dependent variables q
and the water cut fw respectively over fairly long periods, e.g., over a month’s
or year’s production. These steps enable improved reliability of the prediction
methods.

3. Deep Drilling and Production Engineering

3.1. Deep Drilling Engineering (119–123). Introduction. To produce


raw materials and fuels from the Earth’s crust, two basic problems have to be
solved: Firstly, the deposit must be discovered, its parameters determined, and
then investigated whether it can be developed by considering all the available
data. Secondly, the raw materials present in the deposit have to be produced –
ie, transported to the surface – economically (119). To solve these problems,
particularly for liquid and gaseous materials, deep drilling is indispensable.
Boreholes for locating and exploring deposits are known as exploratory wells,
while boreholes used for producing from the deposits are known as production
wells. Successful exploratory wells are converted into production wells using
suitable equipment.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 121

Exploratory wells must be drilled in such a way that they provide accurate
information on the rock formation drilled through and on the deposit to be
explored. This is done by sampling the rocks, eg, in the form of drill cores, sam-
pling the materials present in the rocks (eg, water, oil, gas), gathering data from
a great variety of geophysical borehole measurements, carrying out productivity
tests, and then measuring the resulting behavior of the reservoir pressure.
Production wells must be drilled and completed in such a way that the cost
of subsequent production is minimal, as much as possible of the reservoir con-
tents can be produced economically, optimum use is made of its productive capa-
city, a minimum number of wells is required to develop the reservoir and trouble-
free production can be achieved, preferably over the reservoir’s lifetime (120).
Boreholes compete with mining methods as a means of prospecting for and
producing raw material and energy, and are in many cases, superior to them
from economic, operational, and safety viewpoints. Deep drilling engineering is
therefore becoming increasingly important with th development of deposits and
reservoirs that lie ever deeper and in inhospitable regions, under the oceans, or
in perpetual ice.
History (124–126). Deep drilling is far from being a new technology
(Fig. 41 ). As early as 600 B.C. CONFUCIUS reports that wells have been sunk in
China to a depth of about 500 m into salt brine with hammer drilling devices.
In Europe, both in ancient times and in the Middle Ages, wells were sunk to
obtain water and later for prospecting shallow ore lodes. Development of
drilling engineering and well depths that could be achieved stagnated for

Fig. 41. History of drilling depth.


122 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

many centuries. The search for raw materials (potash, iron ore, and coal) in the
course of the industrialization in the 19th century forced the development of
deep drilling engineering. At that time, drilling depths already reached
2000 m.
In Europe during the second half of the 19th century and the start of
the 20th century a extensive search for raw materials took place. It was pioneers
like VON OEYNHAUSEN, FAUCK, CRAELIUS and ANTON RAKY who encouraged the pro-
gression of deep drilling by, for example, developing high-speed rotary percus-
sion drilling with circulating drilling mud. Only then could the frozen
boreholes for shaft sinking of mines for coal, potash, and rock salt be produced.
The rotary drilling method, developed in the United States at the end of
the 19th century for the exploration and development of oil deposits, was
an important step forward. Since then, rotary drilling has been continuously
developed and today is still the dominant drilling method by which depths
of more than 9000 m have been reached in commercial boreholes. The
present depth record has been achieved with the Kola SG 3 well in the Com-
monwealth of Independent States, which reached a depth of 12 261 m, using
turbodrilling.
Rotary Drilling (127–131). The circulating viscoplastic drilling mud and
powerful roller cone drill bits are the most important characteristics of rotary
drilling. These, in combination with other typical features, have contributed to
the overriding importance of rotary drilling in the exploration and development
of deep and high-pressure reservoirs. Spectacular blowouts, which were numer-
ous in the past, are now rare.
The principle of mode of action in rotary drilling is as follows (Fig. 42):
By means of a rotary table (e), a torque is applied to the drill string that
consists of individual drill pipes (steel tubes). The drill string is suspended in
the derrick by a hook (b) attached to a traveling block. This hook, which allows
the drill string to rotate, can be moved vertically with a drawwork on the rig
floor. The drill string transfers the torque to the drill bit (n), whose impact on
the rock is controllable via the weight of the drill string and drill collars (m)
(massive thick-walled tubes arranged above the bit), which is not taken up by
the drilling hook.
The rotary table, which is a ‘‘surface drive’’, can be replaced by a hydraulic
drilling motor immediately above the bit. This downhole motor, which can be a
positive-displacement machine on the Moineau principle or a multistage turbine,
is installed in the drill pipe and driven by the circulating mud.
Drilling Tool. The drilling tool, which is adapted to the rock through
which the hole has to be sunk, performs the drilling work on the hole bottom.
This can be done by machine drilling (cutting bit), fragmentation drilling (roller
cone bit, Fig. 43 A – E), or abrasive drilling (diamond bit Fig. 43 G).
The key factor with regard to drilling costs is the rate of penetration (v). The
rate of penetration depends on the rotary speed of the bit n, the weight on bit
P, the drillability of the rock S, and the hole diameter D. In the empirical
relation

nP2
v¼ k
D2 S2
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 123

Fig. 42. Diagram of a rotary drilling system. a) Parked drill pipe; b) Drill hook;
c) Rotary swirrel; d) Flexible hose; e) Rotary table; f) Preventer; g) Vibrating screen
(shaker); h) Pit; i) Cement sheath; j) Casing; k) Hollow drill pipe; l) Circulating drilling
mud; m) Drill collar as part of the drill string; n) Roller cone bit; o) Rock strata;
p) Reservoir rock horizon.
124 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 43. Various types of bit A) Roller bit with coarse teeth (for soft formation); B) Roller
bit (for medium hard rock); C) Roller bit with fine teeth (for tough rock); D) Roller bit with
hard metal inserts (for very hard formations); E) Roller cone bit; F) Jet bit; G) Diamond
full-hole bit.

a ‘‘drillability coefficient’’ k combines factors such as rock properties, mud para-


meters, pressure conditions, bit type, wear of the bit teeth, etc. It is assumed that
the hole bottom is clean.
Rotary speed and bit load cannot be increased without limit. However, the
‘‘penetration factor’’, defined as the product of the bit load and bit rotary speed,
should be as high as possible.
The power required to turn the drill pipe, which depends on the mud lubri-
cation properties, rises with the square of the drill pipe diameter and linearly
with the depth. With increasing depth the torque imparted to the bit by a surface
drive becomes smaller.
The bit’s mechanical power is directly proportional to the product of torque
and rotary speed. In addition, due to the mud stream emerging from the bit,
hydraulic power is available at the bit, which is equal to the pump power
minus the pressure drops in the mud system.
With limited mud pressure due to conventional borehole and pipe sizes, the
optimum hydraulic power at the bit is reached at ca. two thirds of the power
required to pump the mud. The hydraulic power is much greater than the
mechanical power, especially with increasing depth. The bit shape, nozzle
arrangement, and diameter are chosen so that all requirements for the hydraulic
pressure conditions, cleaning of the hole bottom, lubrication, and cooling of the
bit, as well as for securing the cuttings’ removal capacity are met.
The resistance to the flow of mud on its return through the annulus between
drill pipe and borehole wall requires hydraulically optimum compatibility between
borehole and drill pipe diameters to achieve minimum resistance to flow. The mini-
mum flow velocity in the annulus should be 0.5 – 0.75 m/s. The pump power must
be so designed that sufficiently high pressures and large volumes can be handled.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 125

The drill string is loaded by tension (the drill pipes’ own weight), torsion,
bending (resulting from borehole deviations), constant or dynamic internal pres-
sure and buoyancy forces, and periodical vibration. In some cases, the super-
posed stresses in drilling can only be calculated with difficulty.
Drilling Mud. The cuttings are carried in the annulus between drill pipe
and borehole wall by the mud. A thixotropic liquid, containing bentonite and cel-
lulose for example, is used as the drilling mud. Its functions include cooling and
lubricating the drilling tool and cleaning the borehole bottom, discharging the
cuttings, and supporting the borehole wall (hydraulic casing). The mud is
pumped with piston pumps through the drill pipe to the bit, from which it
emerges at high speed to ensure that the cuttings are removed as completely
as possible. The cuttings and finer solids are removed from the mud at the sur-
face and the mud is then recycled by pumping it back again into the hole. Vibrat-
ing screens, centrifuges, and hydrocylones are used to seperate the solids
(Fig. 44).
The mud parameters (eg, the rheological properties, the mechanical, and
thermal state of the colloid system) are adjusted according to the drilling
speed, pump power, bit type, borehole diameter, pore pressure in the rock, tem-
perature, nature of the rock drilled (mud loss zones in the drilled formations),
borehole inflows, and solids content.

Fig. 44. Processing of drilling mud, schematic diagram. a) Flow measurement; b) Mud
pumps; c) Gas traps; d) Sieve; e) Centrifuge tank; f) Degasser tank (vacuum degasser); g)
Intermediate tank; h) Suction tank; i) Pill tank; j) Water tank; k) Reserve tank; l) Mixing
tank; m) Moineau pump; n) Centrifuge; o) Cuttings pit; p) Sludge pit; q) Container, cut-
tings sampling; r) Mixer; s) Separator sampler.
126 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Mud densities between 1.10 and 1.40 g/cm3 are usually sufficient for com-
pensating the reservoir pressure and the ‘‘trip margins’’, ie, the piston effect
and inertia forces during pulling the drill string. In special cases, however,
values below 1 g/cm3 or up to 2.5 g/cm3 are also required.
The carrying power of the mud for the cuttings is given by:

Fy ¼ Vð%G  %mÞg

where F is the surface area of the cutting particles (cm2); V the volume
of the cuttings (cm3); %G, %m are the density of rock, and mud, respectively
(g/cm3), g is the acceleration due to gravity (dm/s2), and y the static shear stress
(dyne/cm2).
The filtration of water from the mud in porous rock strata, with the forma-
tion of a filter cake, proceeds according to Darcy’s law of filtration. The quantity
of filtrate is proportional to the permeability of the filter cake, the applied pres-
sure difference, and the active filter area, and inversely proportional to the fil-
trate viscosity and the filter cake thickness.
Pressure and temperature changes, particularly in deep holes, make it
difficult to assess and calculate the formation of the filter cake. The filter
cake can attain a high mechanical strength and cause drilling difficulties.
Prolonged filtration can also lead to considerable reservoir damage.
The electrical conductivity of the mud and its pH value and contents of
chloride, sulfate, calcium, and hydrogencarbonate ions are continuously moni-
tored. Additives used to adjust the mud system include:

1) Clay materials (eg, bentonite, attapulgite)


2) Weighting materials (eg, barite, hematite)
3) Viscosity-lowering substances, dispersants, protective colloids
4) Emulsifiers, lubricants (eg, oil, to reduce internal friction)
5) Inhibitors

There are water- and oil-based drilling muds. Aside from the technical dril-
ling requirements for the mud, the following requirements should be met: the
formations drilled through remain identifiable (mud-logging see below, indica-
tion of traces of gas and oil); borehole measurements, tests, core drilling, and
other borehole studies can be carried out; the mud causes as little damage as pos-
sible to the formations drilled through (obstruction, swelling, mud losses); and
corrosion of the drilling equipment is limited.
Drilling muds for use in geoscientific research such as in the German Con-
tinental Deep Drilling Program (kontinentales Tiefbohrprogramm, KTB) the
main well must satisfy not only the drilling engineering requirements but
also those of the geoscientists. For example, such drilling mud must have ade-
quate rheological properties, lubrication, and borehole stabilization without
influencing the geoscientific analysis or even disturbing it.
Data on the drilling parameters, mud properties and lithology are col-
lected and stored in a mud logging system by modern data acquisition and
a computer system for drilling rigs. A network of monitors, work stations,
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 127

Fig. 45. Continental Deep Drilling Program (KTB), drilling and casing diagram.

plotters, and printers gives continual access to continuous and discontinuous


data and information.
The hole tapers in sections from top to bottom. Each of these sections is
secured by means of a string of casing, which is bonded by cementing into the
surrounding formation (Fig. 45).
New Developments in Deep Drilling Engineering (131–143). Superdeep
Wells. Commercial wells for hydrocarbons with final depths of at least
5000 m are common nowadays, especially in Europe, where exploration has
moved to even deeper strata – at present the Rotliegendes. The technology,
especially the control of high pressure and temperature, is essentially avail-
able for this depth range. However, considerable development work remains
to be done in certain areas, especially to improve the economics of drilling
and the operating life of the well. Superdeep scientific wells such as that
of the KTB, which will be more than 10 km deep, represent an intensification
of the requirements and hence are of importance for technology development
(Fig. 46).
A few degrees’ change in the inclination and azimuth of the well bore would
cause such large lateral forces, abrasive loads, and wear that the well would have
to be abandoned. The 10 000 m drill string is analogous to a wire 1.5 mm thick
and 100 m long, at whose lower end a drilling tool has to be controlled with
an accuracy in the 0.1 mm range. Active vertical drilling systems have been
developed that continuously measure the inclination and direction of the bore-
hole. They pulse the values through the mud to the surface, so that compensatory
steering can be applied to the tool by extending hydraulic pistons or ribs (Fig. 47).
Straight, superdeep vertical holes are therefore a particularly demanding
128 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 46. Drilling rig UTB 1 for the 10 000 m main well of the Continental Deep Drilling
Program (Germany).

variant of directional drilling engineering. The size and performance data of the
drilling rig for a superdeep hole clearly exceed those for conventional extra-heavy
rigs.
Horizontal Drilling. The second main focus of development in deep dril-
ling engineering is directional drilling where the drilling of holes along a defined
course over long distances and at considerable depth in the underground rock is
now possible and economic (Fig. 48).
Offshore, virtually all wells are directional drilled holes, and the oil and gas
fields are developed with up to 60 wells per platform. Onshore, the classical
applications of horizontal deep drilling include recovery under residential
areas and traffic routes, borehole deviation in the event of damage, and relief
wells in the event of blowouts. A directional drilled hole is produced by using
mud-driven drilling motors. The deviation is created with a bent shaft (bent
sub) combined with the drilling motor, or the motor itself is tilted. In Western
Europe, 85 % of the onshore wells are directional drilled.
Inclination and azimuth (ie, the course of the borehole) are continually
measured and the values transmitted to the surface via pressure pulses through
the mud column. From the surface the azimuth can be corrected accordingly. The
continual checking, known as measuring while drilling (MWD), is crucial for
success. The azimuth is measured by magnetometer or, if in the neighborhood
of ferromagnetic masses, by gyro compass. The inclination is evaluated by accel-
erometers (pendulum). In addition, these instruments indicate the tool face,
ie, the angle between the desired direction and magnetic north (notch and
bead). The availability of powerful, slow-running drilling motors together
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 129

Fig. 47. A) Active steering vertical drilling system; B) Cross section of X; a) Flex sub;
b) Data pulser; c) Sensors; d) Top stabilizer; e) Electronic pack; f ) Generator; g) Control
valves; h) PDM – downhole motor; i) Nonrotating body; j) Hydraulic pistons with movable
ribs; k) Rotating shaft; l) Drill bit.

with the MWD technique are prerequisites for successful directional drilling
engineering.
The inclination build-up of a hole to be deviated into the horizontal can be
abrupt or gradual (see Fig. 49). At an inclination change of 108/10 m, the result-
ing deviated section has a radius of 6 m, whereas at 18/10 m, the radius is ca.
600 m. Holes have been deviated to reach destinations of up to 3000 m, and
holes exist which have been drilled horizontally for almost 1000 m.
3.2. Investigation of the Reservoir Rocks (144,145). With suitable
drilling tools, rock samples (mud samples, drill cores) are obtained from the
reservoir rock zone and can subsequently be examined comprehensively. Special
geophysical measurement (logging) and the first, brief productivity tests (drill-
stem tests) are also carried out.
Core Drilling. During core drilling, the drilling bit (full-hole bit) is
replaced by a core barrel with crown (ring bit). Single- or double-tube core barrels
130 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 48. Selected horizontal wells curvature. a) Texaco, 1981 (horizontal section
>300 m); b) Esso Resources Canada Ltd., Cold Lake 1978 – 1984 (horizontal section
1000 m); c) Esso Resources Canada Ltd., Norman Wells, 1981 (horizontal section 500 m);
d) Elf France, CR 163, 1986 (horizontal section 360 m); e) Elf France, LA 90/LA 91, 1980 –
1981 (horizontal sections 270 and 470 m); f) BP 1985 (horizontal section 324 m); g) Elf
Italy, RSM 6, 1982 (horizontal section 600 m); h) Elf Italy, RSM 211, 1986 (horizontal sec-
tion 600 m); i) Preussag, 1984 (horizontal section 200 m); j) Marsvoskoe, 1968 (horizontal
section 320 m); k) Elf France CLU 110, 1983 (horizontal section 330 m); l) Sohio, Y 20, 1986
(horizontal section 305 m); m) Sohio, B 30, 1985 (horizontal section 452 m); n) Sohio, JX 2,
1986 (horizontal sction 415 m); o) Sohio, BW 1, 1985 (horizontal section 579 m).

are used, some with diamond-type cutter heads. Single-tube core barrels have
the drawback that the core is directly exposed to the mud (core losses); With dou-
bletube core barrels this is prevented by the inner tube of steel or rubber
(Fig. 50). Wedgejaw catchers, flap catchers, or combinations of both prevent the
core from falling out. Orientation devices and core markings allow an accurate
determination of inclination (dip), course (strike), and pattern of the strata
drilled through.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 131

Fig. 49. Horizontal drilling strategies A) Conventional; B) Drainhole, medium radius


technology; C) Drainhole, short hole technology.

Since 1985, remarkable successes have been achieved by a combination of


rotary and mining drilling methods. The mining drilling method involves rapid
rotational drilling with thin-lipped diamond crowns and a double tube core
barrel. Externally and internally smooth drill pipe is used which almost fills
the whole borehole cross section. After boring out the core, the inner core bar-
rel is left on the cable (wireline coring), so that the bit can remain on the bot-
tom until it is worn out (Fig. 51). The thin-lipped crown and thus the distinctly
smaller area of rock which has to be destroyed, makes greater drilling progress
possible in comparison with rotary drilling technology. With the combination
of both methods the drilling progress can be doubled with complete core recov-
ery; by use of the wireline technique, the number of round trips can be
reduced; and by use of a rotary drilling rig, the time for the round trips can
be reduced.
Aside from core drilling, cores can also be shot, drilled, or cut (slicer) from
the borehole wall.
The core investigations provide measurements on pore filling, pore volume,
liquid saturation, capillary pressure, wettability, grain sizes, chemical composi-
tion, petrographic relationships, and strength, etc.
Geophysical Borehole Measurements. Continuous coring is impossi-
ble for economic reasons. Geophysical borehole measurements allow the contin-
uous determination of rock parameters, especially the porosity, the original
saturation conditions, the nature of the pore fluid, the type of rock, the mineral
composition, and the bed thickness.
The laboratory examinations of the core samples taken point-by-point,
however, take longer.
132 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 50. Schematic diagram of double core barrel a) Coupling to drill pipe; b) Overshot
for pulling out with wireline; c) Outer tube, rotates core bit; d) Inner tube, nonrotating;
e) Bearing; f) Mud passage; g) Ball valve; h) Core catching spring; i) Core catching
cone; j) Mud channels.

A variety of instruments is available for borehole measurements. The


methods used can be divided into four main groups

1) Electrical methods: macroresistance, microresistance, conductivity, scan-


ner technique, dip meter
2) Sonic measurements: sonic velocity, travel time, amplitude, reflection,
refraction
3) Radioactivity measurements: natural, induced gamma, and induced neu-
tron radioactivity, spectral analysis
4) Determination of geophysical fields: temperature, pressure, stress, mag-
netics, Earth’s gravitation

The temperature limit of application of geophysical methods is ca. 250 8C,


corresponding to about 8000 m borehole depth.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 133

Fig. 51. Comparison of oil and mining core drilling A) Oil core drilling (double tube core
bit); B) Mining core drilling (cable core bit).

The data from borehole measurements are subjected to comprehensive and


complex methods of interpretation, for which powerful computers and specialists
are required. The results obtained describe the rock, its pore filling, and the
geology in situ. Greater reliability is obtained when more drilled rock material
from cores and cuttings have been used for calibration and correlation of the
measurements.
Tests. In (productivity) tests, qualitative and quantitative conclusions
about the inflow process from the formation to the well are obtained in order
to decide whether a well allows economic production. At the start of a test, the
formation interval to be examined is isolated from the rest of the formation in
order to exclude possible effects from other beds and mud pressure. After pres-
sure relief of the interval, the rate of flow to the well, the pressure, and tempera-
ture are measured and plotted as a function of time. The inflow into the drill
string is then stopped and the pressure build-up which then occurs is recorded.
A second flow cycle is initiated by reopening the drill string.
In principle it is also possible to use a test device on a self-supporting elec-
tric cable. Here, however, the inflow rate is strictly limited for design reasons, so
that only zones near the boreholes can be tested.
In order to save the casing costs of wells which may later, prove to be une-
conomic, tests are carried out in the uncased borehole during or immediately
after the end of the drilling operations. For stability reasons, the test duration
134 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

in the uncased and therefore unbalanced, unstable borehole (drill-stem test)


must be kept short, whereas in later capacity tests in the cased hole the test
duration is theoretically unlimited.
The interpretation of the measurements must provide information on the
following important parameters:

1) Initial reservoir pressure


2) Flow pressure as a function of the amount produced per unit time (produc-
tivity index)
3) Shut-in pressure
4) Reservoir and borehole temperature
5) Permeability and permeability capacity
6) Drainage radius of the well
7) Location of formation damage

During these tests, samples of the reservoir contents are also obtained, so
as to carry out chemical and physical studies. The results serve as a basis or aid
to planning decisions on: the completion of the well, well and reservoir treat-
ments, the production method, the surface installations, the further development
of the field, and exploitation strategy.
3.3. Casing and Cementation. Casing (146). The casing of a borehole
usually consists of several concentric strings of casing pipe, which are installed
according to the geological – engineering requirements at different depths
(Fig. 42). It protects the borehole wall against scouring and blowouts, seals
permeable horizons to prevent undesirable inflow or outflow, and prevents
sloughing from the formation into the hole. The casing is in part already inserted
during the drilling operation, in order to protect problem sections of borehole. It
represents a considerable cost component of the well.
With exploratory wells, the planning of the casing program is rather diffi-
cult, since knowledge of the formation is still imperfect. For production wells a
standard program is usually set up, in which tube dimensions, wall thicknesses,
and landing depths are optimized.
The casing can either be run to the surface and landed in the wellhead, or hung
in the borehole as a so-called liner in a string of casing already installed in the bore-
hole. To ensure a tight bond between casing string and formation, the casing is
cemented, either up to the surface or only over a certain length of the borehole.
The casing is subject to a multiplicity of loads, which generally arise in
combination:

1) Differential pressure between the outside and inside due to hydrostatic


pressure and the pressure of the fluids in the pore space
2) Radial component of the formation stress
3) Tensile stress from the string’s own weight, especially during installation
4) Bending stress in holes with changes of direction, especially in horizontal
holes
5) Thermal stresses
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 135

Since the individual stress contributions superpose one another, precise


assessment of them eg, by landing under tension, is essential. A series of
standardized material grades and casing joints is available to meet the
requirements.
Tubing String. The tubing string transports the fluid produced to the
surface. The tubing is landed in the wellhead at the surface and anchored, if
necessary, at the lower end. The annulus between casing and tubing can be
sealed with a packer if the conditions in the well require it. Packers are usually
elastically deformable sealing elements (of elastomers), which are so combined
with peripheral sliding wedges that during up and down movements the
wedges grip, the seal deforms, and the annulus between tubing and casing is
sealed. Tubing, like casing, is manufactured in a range of standardized mate-
rial grades and dimensions.
Cementing (134). The cementing of the casings serves the following
purposes:

1) Massive bond of casing and formation, in order to divert the forces result-
ing from the loading of the casing into the formation, ie, absorption of the
axial and radial stresses which act on the casing string
2) Isolation of permeable formations, in order to prevent flow between differ-
ent rock strata and, in particular, the contamination of the groundwater
3) Corrosion protection of the casing

The cement is mixed with water to form a cement slurry, and a wide
variety of additives are added to control the rheological properties, the
setting time, the final strength, etc. The slurry is then injected with high
pressure pumps through the borehole into the annulus between casing and
formation.
The cements used in the oil and gas industry are standardized according to
API. Table 4 gives some important characteristics of various API cements. Aside
from the cement quality, the technical execution of the cementing is of crucial
importance for the quality of the cement sheath. For control, the cement bond
log is used: this is an acoustic measurement method for determining the bonding
of cement and casing or of the cement with the formation.

Table 4. Characteristics of some API Cements


Class

A D, E, F G, H

Density, g/cm3 3.14 3.14 3.12


Specific surface, cm2/g 1500 – 1900 1100 – 1500 1600 – 2000
Optimum water content of 46 38 44
the slurry, wt %
Density of the slurry, g/cm3 1.87 1.97 1.89
Temperature range, 8C 75 110 – 160 260 – 370
Sulfate resistance normal medium/high medium
136 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

For cementing at high temperature, Class G cements are used, to which


SiO2 flour is added to avoid a decrease in compressive strength and an increase
in the permeability of the cement. The effect of the added SiO2 depends on the
formation of hydration products, for instance according to

2 CaðOHÞ2 þ SiO2 ! 2 CaOSiO2 1:25 H2 O þ 0:75 H2 O

The elevated temperature lowers the hardening time, so retarders and anti-
friction agents must be added.
3.4. Production Engineering (148,149). Objectives (150–153). The
purpose of the exploitation and production planning of hydrocarbon reservoirs
is to produce optimum amounts of saleable (transportable) products at mini-
mum cost and with close attention to all aspects of safety and ecology. In
initial stages of field development, with few wells and few drill-stem and pro-
duction tests available, the base for planning is not valid enough. However,
the data obtained from continuous production and field development lead to
continual refinement of production planning, in which reservoir and engineer-
ing are regarded as a synergism. Only in this way can an optimum overall
technical and economic result be achieved.
Production engineering has to solve the following problems in oil
production:

1) The time for water injection – when not using it to adjust the pressure
above the bubble point and/or for adjusting the dynamic level to below a
certain material stress, corresponding to the fatigue endurance limit of
the pump rods
2) Dependence of the productivity index (IPR) on viscosity of the oil and water
cut
3) Gas production and availability in the gas lift method
4) Advantages and disadvantages of the artificial lift methods

and for gas production:

1) Occurrence of toxic and/or other problematic substances, eg, hydrogen sul-


fide, elementary sulfur, carbon dioxide, nitrogen, which requires complex
process engineering installations (see also Natural Gas)
2) Heterogeneous multilayer and selective water incursion

The safety and ecological aspects have always been important; nowadays it
is preeminantly so. In the second half of the 19th century, a blowout was proof of
discovery after finding the reservoir by drilling and a reason for celebration.
Thanks to the continual improvement of technical equipment and the regular
training of technical staff, blowouts have become rare, but they can never be
totally avoided due to human error.
Over wide areas, oil and gas production engineering have common princi-
ples, which are treated in the following section whereas specific questions of
production engineering for oil and gas are dealt with.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 137

Fig. 52. Production equipment of a typical gas well. a) Nozzle; b) Block-type Christmas
tree with five shut-off valves; c) Double P-seal; d) Hanger nipple with seat for plug;
e) Tubing seal; f) Flange slide valves; g) Tubing double flange with additional casing
seal; h) Bottom flange; i) Corrosion-inhibited protective liquid; j) API tubings; C 75 or
J 55; k) Intermediate casing; l) Circulation slide valve with seat for plug; m) End casing;
n) Tubing receptacle, overshot type, with seat for wireline plug; o) Tubing anchor; p) Pro-
duction packer; q) Perforation.

General Production Engineering. General production engineering encom-


passes completion, setting up production, perforation, well and reservoir treat-
ment, fracturing technology, workover, the inflow – performance relationship,
one- and two-phase flow through tubing and lines, safety engineering, environ-
mental protection, and the plugging and abandonment of wells.
Completion. If, after borehole measurements and tests, a well is consid-
ered worth producing from, it is completed, ie, it is equipped with all technical
installations necessary for production (Fig. 52). Components necessary for the
completion are the wellhead (including flanging), casing, cementing, tubing
string or strings (single or multiple completion), and the ‘‘bottom hole comple-
tion’’. This includes the special equipment used in the reservoir zone, ie, liners,
liner hangers, sand filters, packers, and all the ‘‘wireline equipment’’ which can
be operated by cable and wire.
138 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 53. Bottom-hole equipment in the reservoir zone. A) Main dolomite, uncased;
B) Carboniferous, casing with liner; a) Tubing, 2 7/8 inch (2.22 cm); b) Cement;
c) Production packer; d) Casing, 7 inch (17.8 cm); e) Liner hanger; f) Liner, 5 inch
(12.7 cm).

Bottom Hole Completion. Essentially two types of well completion


exists in the zone of the producing formation (pay zone), open-hole completion
or casing and later perforation (perforated casing) (Fig. 53).
Open-hole completion, in which the producing formation is not cased, is
only rarely used. Especially with fairly thick, poorly consolidated formations
and at high production rates, difficulties arise with the penetration of sand
from the formation into the well.
For this reason, the casing is set on top of producing formation or below it
and cemented. To allow the production fluid to enter the well, the casing and the
cement sheath are then perforated in the reservoir rock zone.
Perforation (154). By perforating the casing and cement sheath, the well
is hydraulically connected with the reservoir, so that the production fluid can
enter the well with the least possible pressure drop. Currently, shaped charges
are exclusively used for perforation. They are introduced into the borehole on
a self-supporting cable and ignited. The number of perforations, their arrange-
ment, penetration, and size depend on the rock hardness and borehole
conditions.
With the hollow-charge perforators mainly in use, the penetration depth in
the rock is 10 to 40 cm, depending on the compressive strength of the rock
(Fig. 54).
Normal shaped charge perforators can be used up to ca. 170 8C. For higher
temperature, special explosives and techniques have been developed. These
explosives are so designed that they do not detonate by self-ignition when
overheated but become unusable as a result of decomposition to gaseous phases.
Explosives stable to high temperature such as hexanitrostilbene (HNS), can be
used at up to 250 8C and depths down to 8000 m for up to 24 h downhole time.
In the high-temperature range which is sensitive to misfiring, safety measures
prevent the effects of stray and leakage currents on the charges.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 139

Fig. 54. Schematic of a shaped-charge perforation. A); a) Case or container; b) Liner;


c) Main explosive charge; d) Booster; e) Detonating cord groove, point of initiation;
B); a) Detonating cord; b) Shaped charge; c) Electric detonator; d) Port plug; e) Air; f) Steel
carrier; C); a) Shaped charge; b) Slug; c) Jet Pressure on target p1 = 35 kPA (5106 psi),
velocity of foreward jet v1 = 6100 m/s.

The method of tubing-coupled perforating (TCP) (Fig 55), which has been
available since ca. 1980, has captured a significant market, because of the fairly
large number of deep gas wells and the rapid development of horizontal drilling.
By this method the charge carriers are installed on the tubing string below the
packer and landed with the latter. They are then connected (wet) in a separate
run with an ignition mechanism on the cable, and then ignited. The advantages
of this method are

1) A greater shot density — 39 shots per meter with the TCP instead of
13 shots per meter conventionally
2) Considerably longer perforation sections, since the load limit of the cable
used in conventional perforation is no longer limiting
3) No hanging of the charge carriers in the inclined hole (> 708)
4) Greater safety, because the charges are sensitized underground

Well and Reservoir Treatments (155–158). The treatment of the well


and of the reservoir formation is intended to increase the productivity of wells.
The decrease of this productivity can stem from the drilling process or arise in
the course of production. Well treatment is intended to restore the original
state or, in other words, to restore an inflow to the well free from restrictions.
Reservoir bed treatment is aimed at a sustained increase of the flow through
the reservoir rock.
140 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 55. Tubing-conveyed perforation (TCP).

Well Treatment. The type of obstruction of the well must be known in


order to choose the most suitable treatment from the various possibilities. The
obstruction can be caused

1) By solids from the mud (filter cake)


2) By increased water saturation around the well (water block)
3) By swelling of the clay minerals present in the rock’s matrix
4) By chemical precipitation
5) By emulsification

In practice the first two causes occur most frequently. The obstructions are
usually removed by chemical treatments. The most important chemical treat-
ment agents are: aqueous solutions of hydrochloric acid (5 – 15 wt %), hydrofluo-
ric acid (1 – 6 wt %), acetic acid, citric acid, and surfactants (all in concentrations
of 0.1 – 5 wt %).
Corrosion inhibitors are added to the acids. Hydrochloric acid is generally
used with carbonate- containing reservoir rock. Whose carbonate content should
be ca. 20 %. The hydrochloric acid partly dissolves the rock. Hydrofluoric acid,
usually in conjunction with hydrochloric acid, is used for dissolving silicates:
here, the hydrofluoric acid is added to the hydrochloric acid as NH4HF2.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 141

Fig. 56. Schematic of fracturing technology A) Before fracturing; B) After fracturing.

Surfactants are added in almost all well treatments in order to promote


the dispersion of the solids, to improve the wettability of the rock, and to prevent
emulsification.
If it is intended to treat only the borehole wall, it is sufficient to circulate the
treatment liquid into the well zone in question. If the obstruction extends into the
reservoir bed, eg, by clay swelling, the treatment liquid must be uniformly
injected into the reservoir, avoiding pressure which leads to cracking. In any
case, the reaction products should be removed from the reservoir bed or from
the well as quickly as possible.
Reservoir Bed Treatments. Two methods basically exist for extensive
improvement of the inflow conditions in a reservoir bed of low original permeability:

1) Pressure acidizing (usually in carbonate rocks)


2) Hydraulic fracturing (frac treatment)

The best-known method of reservoir-bed treatments is fracturing technol-


ogy (Fig. 56), which is mainly used for producing artifical flow paths in the reser-
voir rock.
Pressure acidizing is also a hydraulic fracturing method by injection of
acid. In pressure acidizing, however an increased permeability is achieved,
by dissolution of a part of the carbonate matrix along the fracture faces. In
frac treatment oil, water, or acid are injected together with gelling or viscos-
ity-enhancing agents. In addition, to hold the fractures open and thereby
ensure a high-permeability flow path, a proppant (eg, fluvial, natural quartz
sand) is added while pumping in the liquids. This proppant is deposited in
the fracture and so produces the high-permeability artificial flow path. The
reservoir treatments are very highly developed and so make it possible to
solve very specific problems. The fracturing process will be treated briefly in
the following:
Fracturing. In fracturing, only a part of the injected liquid is available to
widen the fracture and act as a medium for transporting the proppant, or for the
chemical reaction along the fracture faces. With a widening fracture, the other
part of the liquid increasingly migrates into the pore space of the rock. In
142 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

order to limit this effect, which is unfavorable for the success of the operation,
fluid loss agents are added that raise the viscosity and form a filter cake.
These agents cause the initially high fluid loss to decrease substantially after
the filter cake has built up. Later, these additives become ineffective due to
the influence of temperature and pressure, and the production stream reaches
the well unhindered.
In order to ensure the success of reservoir bed treatments, a sufficient depth
of penetration into the reservoir rock must be achieved. This penetration depth
must be greater with less permeable reservoir rock. Penetration depths of
40 – 70 m are usual with vertical or horizontal fractures.
The development of deep (ca. 5000 m), low-permeability gas reservoirs in
the Rotliegendes and Carboniferous has led to remarkable developments of
this technology. New developments now include a temperature-stable and suffi-
ciently viscous supporting fluid, the availability of pressure-resistant proppants,
and powerful equipment. For example, a treatment volume of 1100 m3 requires a
fracture length of 300 m, maximum pressure of 750 bar at a depth of 4200 m, and
15 pumping units with an output of 9000 hp.
Workover. An important chapter of constant concern for the production
engineer is workover technology. More than 106 production wells worldwide, of
which three quarters are oil wells, require continual upkeep and maintenance
and — repeatedly during the course of their use – repair. Thus the number of
workover hoists is three to four times the number of drilling rigs. In addition,
there are the modern systems, ie, the wireline technique (lowering activation
of tolds by wireline), the coiled tubing technique, and hydraulic lifting and low-
ering of the strings with the snubbing unit. Workover should always follow a
sequence of diagnosis, therapy (work program), operational execution, and mon-
itoring of success. Reliable measurement techniques and supervision of operation
and production are the prerequisites for correct diagnosis. The workover hoist,
equipped with rotary table or power swivel and pump unit, can carry out the
functions of lifting, lowering, rotation, and circulating fluid, likewise the snub-
bing unit. With the wireline technique, only running in and out is possible.
The coiled tubing (Fig. 57) enables lifting, lowering, and circulating. Rotation

Fig. 57. Schematic of a coiled-tube unit. a) Drive unit; b) Operator’s cabin; c) Coiled-
tube drum; d) Coiled tubing; e) Injector crane; f) Injector head; g) Preventer device; h)
Wellhead.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 143

is possible by using suitable downhole motors. The coiled tubing technology has a
high technical standard and a growing market, particularly also with regard to
horizontal drilling. Coiled strings with diameters of 1 inch (2.54 cm), 11/4 inch
(3.18 cm), 13/4 inch (4.44 cm) and 2 inch (5.08 cm) are available, and can be
used at a depth of up to 5500 m and in combination depending on the loading
situation.
Horizontal Wells (132,135). Horizontal drilling, which has been deve-
loped to a state of maturity since the beginning of the 1980s, makes great
demands on production engineering. Drilling engineering is able to pro-
duce, horizontal drilled sections of a few hundred to more than 1000 m in
any rock. In the deviation to the horizontal, the inclination build-up can be
abrupt or gradual (Fig. 46). Whether this value is 108/10 m (final radius
60 m) or 18/10 m (final radius 600 m) is very important for the subsequent
completion.
The essential reason for the rapid growth of horizontal drilling is that a
markedly larger zone of the reservoir is developed for production than with ver-
tical wells, so that a considerable (15 – 20-fold) production increase is achieved at
lower differential pressure; better sweep efficiency; and a higher recovery. In
heterogeneous, fissured rock the effect is even more significant. The potential,
especially if specific enhanced oil recovery (EOR) processes are incorporated in
this technology, is far from exhausted. In gas reservoirs, turbulence around
the borehole is avoided, and in underground storage, a capacity increase of
3- to 8-fold seems possible with horizontal drilling. There is still, however, a
considerable need to develop devices and methods specific to the process,
cf. Schlumberger coiled-tubing-logging (CTL). Completion in deviated and hori-
zontal wells can be carried out as follows:

1) Open hole (ie, uncased): only possible with stable formations


2) Open hole with slotted or ‘‘prepacked’’ liner (ie, plastic-reinforced sand
filter)
3) Slotted liner with external casing packers (ECP) in the open hole
4) Cemented and perforated liner

With the first two methods, no isolation of unwanted sections is possible.


However, the use of correctly placed ECP’s (method 3) and successful cement-
ing (method 4) allows selective production. For liner installation, the clear-
ances and the borehole geometry in the deviation zone and the permissible
bending stress must be taken into account. Cementing in the horizontal sec-
tion is impeded by the fact that the gravitational force favors segregation of
liquid and gas at the top and sedimentation of cuttings and solid at the bottom.
For the perforation, only the TCP technique is possible from a 708 incli-
nation; and, as an alternative to the tubing string, coiled tubing or the drill
string. The wireline technique in horizontal holes is virtually ruled out. For
workover, the coiled tubing technique mainly is applied (Fig. 54). Methods
independent of the borehole inclination are suitable as production facilities
such as the submersible centrifugal pump, the hydraulic piston pump, and
gas lift. The lengths of rigid units must be geared to the maximum possible
144 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

flexure in the deviation zone, and the necessary adaptation of systems must be
carried out.
Oil Production Engineering (159–163). In oil production engineering, not
only the technical instruments and developments determine success. Efficiency
and overall profitability crucially depend on qualified supervision, damage eva-
luation, the logistics of material and distribution, and a sound operating organi-
zation. The use of tertiary methods (eg, steam injection and steam flooding)
signify additional tasks for production engineering: In steam injection wells,
and also in the hot production wells of the steam flooding projects, extreme
thermal stresses (140 bar, 340 8C) and corrosion problems in the underground
equipment must be kept under control.
Production engineering has powerful production processes and production
facilities available to it. Herewith, the particular conditions of each reservoir can
be taken into account. This in turn enables, firstly, the reservoir fluids to be pro-
duced as efficiently as possible with extensive use of the reservoir energy; and
secondly, production difficulties, caused for example by accumulated sand or
water or by corrosion, to be obviated in good time.
With the opening up and development of a reservoir, its most important
qualities are determined: reservoir pressure, permeability of the rocks, viscosity
of the production fluid, and the presence and characteristics of a natural water
drive. These are the foundations for planning production rates and methods, and
equipment.
In principle, the following production processes are available:
1) Flowing production
2) Gas lift
3) Centrifugal pumps
4) Piston pumps, sucker rod, or hydraulic

With these methods, ca. 3.25109 m3 crude oil per year are produced from
depths of up to 6000 m.
Flowing Production (156,164). As a result of production, a pressure
drop occurs which expands, funnel-shaped, around the hole. Under the influence
of this pressure gradient, the reservoir liquid flows to the borehole, enters the
casing through the perforation holes, rises up in the tubing string to the surface,
and flows at the surface by lines to a collection manifold at a central plant. Here
the output from the field is combined, dewatered, treated, and transported to the
refinery. The flow pressure at the depth of the reservoir bed is sufficient to over-
come the hydrostatic pressure of the production column and the frictional loss in
the tubing string and in the surface lines.
The flowing production rate is calculated as a function of bottom flow pres-
sure, tubing diameter, and pressure head according to the law of conservation of
energy, while taking into account the empirical parameters which determine the
two-phase flow.
The relationships become more complicated when the pressure in the
tubing string falls below the degassing pressure (bubble point) and free gas
evolves. Progressive pressure reduction gives rise to larger gas bubbles
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 145

Fig. 58. Flow patterns in two-phase flow. A) Bubble flow; B) Bubble flow (turbulent);
C) Slug flow; d) Froth Flow; E) Annular flow; F) Spray flow.

(see Fig. 58) and the liquid column breaks up (slug flow, Fig. 58 C), or the rising
gas bubbles entrain oil drops which, owing to friction at the tube walls, form an
upward-moving film (annular flow, Fig. 58 E) which finally changes into mist
flow, ie, the liquid film disappears and the liquid separates into droplets which
are mixed with the gas stream (Fig. 58 F).
The duration of this economically advantageous eruptive production phase
can be extended by a good matching of the production rate to the tube cross section.
Gas-Lift Production. In the gas-lift method, gas is injected into the
tubing string at one point or – to minimize the starting pressure – at several
successive points and in this way bubble or slug flow is initiated, as has been
described for eruptive production (see Fig. 59). This method is used continu-
ously or intermittently, depending on rate of flow, production rate, and gas
availability.
The eruptive production phase is often already supported with gas lift or is
switched to gas lift when the pressure gradient between borehole and reservoir
decreases. The applicability of this flexible production method is limited when
there is a shortage of gas or the gas compression costs are too high. Where suffi-
cient gas is available, the scope for application is wide because of the flexibility of
146 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 59. Continuous gas-lift operation A) Schematic; a) Annulus; b) Valves closed; c)


Tubing; d) Valves open; e) Production zone; B) Pressure distribution at different depths;
a) Annulus pressure; b) Pressure curve in tubing for production well; c) Pressure curve in
tubing for inoperative well.

this method with regard to production rate and depth. In strongly deviated bore-
holes and if abrasive media are coproduced, gas lift is usually superior to other
production methods (no moving parts).
Gas lift valves controlled by the injection pressure are fitted in the tubing
string, into which they inject the lift gas. The pressure at the valve is controlled
via a pressure bellow which operates with or without spring support (according
to type). At the surface before installation the bellow is set to a certain opening
and closing pressure.
Since on start-up of a well in gas-lift operation, the back-pressure of the
whole liquid column requires a high injection pressure, several lift valves
are installed and distributed over the tubing string, whose opening pressures
are coordinated with each other.
Production with Subsurface Pumps (165,166). Subsurface pumps are
the most widespread form for the production of the oil, which is usually contami-
nated by considerable amounts of water. Types of pump used include electric
submersible centrifugal pumps and mechanical or hydraulic subsurface piston
pumps. The pumps are installed below the liquid level in the wells and, in diffi-
cult underground conditions, must overcome hydraulic heads of several hundred
to a few thousand meters (Fig. (60).
Electric Submersible Centrifugal Pumps (Fig. 57 B). These are multistage
centrifugal pumps that are submerged in the production liquid. They are
installed in the borehole, directly coupled with the motor, and suspended on
the tubing string. The pump unit comprises from bottom to top: electric motor
(bipolar asynchronous squirrel- cage, oil-immersed, j), centrifugal pump (with
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 147

Fig. 60. Different types of pump. A) Subsurface rod pump with walking beam drive;
B) Electric submersible centrifugal pump; C) Hydraulic subsurface piston pump with
pressure oil supply in the open system; a) Casing; b) Tubing string; c) Wellhead;
d) Pump rod; e) Subsurface pump; f) Drive pump, g) Production pump; h) Pressure oil
pump with i) Pressure raising pump on the suction side; j) Motor; k) Sealing adaptor;
l) Gas separator; m) Nonreturn valve; n) Tubing drain valve; o) Three-core cable;
p) Circulation piece; r) Pressure oil tank; s) High voltage transformer; t) Autotransformer;
u) Switch cabinet; v) Walking beam; w) Gearing; x) Tubing anchor; y) Production packer.

numerous stages e, f ), protector (sealing adaptor between motor and pump, k),
gas separator (for separating free gas from the produced liquid, l), electrical
switching device, and power cable (these limit the installation depth, since the
electric insulation resistance is temperature dependent).
The hydraulic head (hydrostatic pressure), frictional pressure losses,
and wellhead pressure determine the number of pump stages. Because of
possible changes in the hydraulic head, pumps with stable characteristics
are preferred.
At depths between 1000 and 4000 m, submersible centrifugal pumps cope
with minimum delivery rates between 260 and 50 m3/d and maximum delivery
rates between 340 and 150 m3/d.
Mechanical Rod Subsurface Pumps (Sucker Rod Pumping (167)) (Fig. 57 A).
Oil production was for a long time synonymous with the use of rod
pumps, which are installed as tubing pumps with the tubing or are landed at
the pumping depth as inserted pumps on the pump rods in the tubing string.
These pumps are driven from the surface by beam pumping units with motor,
148 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

gearing, and a pumping jack, that includes the typical walking beam with
‘‘horse’s head’’ on which the pump rod string is suspended. The weak point of
the system, the ‘‘pump rod string’’, requires particular attention. Liquid load,
rod string weight, vibrations, and mass acceleration, which occurs particularly
when the direction of motion reverses, determine the efficiency and the limits
of this kind of artificial lift.
The supply of types of subsurface pumps that differ according to material,
valve arrangement, piston area, and stroke length also take account of extreme
operating conditions. Valve function and pumping efficiency are periodically checked
by dynamometer measurements, ie, by recording the load uptake. Corrosion, sand,
or friction in deviated wells reduce the service life of pumps and rod string.
Hydraulic Subsurface Piston Pumps (Fig. 57 C). Like the electric submersi-
ble centrifugal pumps, these pumps are run underground, yet by high-pressure
fluid. The difficulties with the pump rod string are avoided by hydraulic
transmission of the automotive energy with an auxiliary liquid from the surface
to a pump installed in the borehole. A high-pressure pumping installation pro-
vides the ‘‘power oil’’ that is supplied to the individual producing wells and the
pump via the tubing string. The pump combines two identical sections in one
block, the upper of which operates as a motor which is driven by pressure oil
(f) and for its part drives the lower pump (g) which is hydraulically in complete
equilibrium.
The pressure oil is either taken from the produced oil and carefully cleaned
(open system) or is contained in a closed system. Usually the pressure oil is sup-
plied down the tubing and the flow back through the annulus or, where that is
not possible, it is fed through additional tubing strings. For repairs, the liquid
stream is reversed and the pump pumped to the surface (free pumps in contrast
to fixed pumps).
The range of application of hydraulic piston pumps exceeds that of rod
pumps by far. As with gas lift, a number of wells are connected to one power
oil high pressure pump station, respectively, so that an additional system of
lines is required. To account for the particular conditions of each well, a specially
devised flow and pressure control system is required.
With the electric submersible centrifugal pumps and rod pumps, neither
additional lines above ground nor an expensive control technique are used.
These basic differences are significant for economic efficiency (168).
Developments and Limitations. In addition there are jet pumps (mainly in
the United States); Moineau pumps with subsurface drive or driven by pump
rods from the surface; and turbine-driven submersible centrifugal pumps. How-
ever, these pumps are rarely used. The proven methods described above main-
tain their position, thanks to their high standard of technical maturity and
reliability. The correct choice of pump for the pressure and inflow conditions
(inflow performance) and the flow resistances in the production path (outflow
performance), both determined by the reservoir, and the product composition
(gas, oil) ensures a long and optimized service life. An example of this is Shell
Oil’s Reno field in Wyoming in the United States. There, since 1975, 190 wells
at depths ranging from 2000 to 4000 m have been converted from hydraulic
piston pumps to rod pumps for economic reasons alone. By using combined
pump rod strings of high-strength steel, a further developed heavy-duty beam
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 149

a b c

e f g h c

i j c

Well water

Gas
Oil
Water
Bacteria Injection well
Production well
Scale
Internal corrosion

Fig. 61. Flow diagram of an oilfield a) Compressor; b) Drying; c) Line; d) Gas separator;
e) Heater; f) Separator; g) Desalter; h) Tank; i) Secondary sedimentation; j) Filter.

(Mark II) drive, and subsurface pumps suitable for the great depths, 105 m3/d
wet oil with 30 % water are produced from 4420 m lifting depth. The field is mon-
itored by electronic dynamometers.
Collection and Treatment of Crude Oil (146). Production rates, crude
oil properties, and field conditions determine in what way and to what extent the
production of individual wells, groups of wells, parts of the overall field or several
neighboring fields are combined. Treatment at the field includes gas separation
and dewatering (Fig. 61).
Gas Separation. The degassing process starts already in the reservoir
and intensifies with the decrease of pressure and temperature through the tub-
ing string and the pipe system during production. Consequently, a two-phase
mixture enters the gas separators. To obtain a maximum liquid yield and, in par-
ticular, at high gas – oil ratios, a multistage separation should be aimed for.
However, apparatus cost and special requirements, such as, eg, gas delivery at
high pressure to the consumer, lead to the widespread use of two- and three-
stage separation.
Dewatering and Desalting. Since crude oil can still be produced at 95 %
water content, water – together with sand and contaminants – is removed as
early as possible to reduce transportation costs and corrosion. The refinery accep-
tance standard in Germany limits the total contamination (water and sludge) of
the crude oil generally to 1 % (salt content below 0.02 %). The water separates
only partly as ‘‘free water.’’ Water and oil mainly occur as an emulsion, which
must be broken by thermal, mechanical, chemical, or electrical methods, or by
a combination of these. At the same time, the emulsifiers present in the crude
oil (constituents that are concentrated at the interface) must be made ineffective.
The free water droplets must coalesce and, with the aid of gravitational or cen-
trifugal forces, separate.
150 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Since the 1980s, a greatly simplified water separation method has evolved
from the variety of process combinations used. Particular features of separation
plants are:

1) Early feeding of the demulsifier in order to use the pipeline for internal
mixing
2) Moderate heating (‘‘cold treatment’’) to save thermal energy and reduce the
vaporization losses
3) Simple apparatus; the water separation can often be carried out in a single
tank, whereas previously a downstream fine purification in the gas-heated
treater or an electrical fine purification has been required

These improvements have been made possible by the further development


of demulsifying chemicals and by continual constructional improvements. These
include also the improvement of the gas separation, by which the inconvenient
degassing after processing has been eliminated.
Special Problems in Crude Oil Production. Paraffin Precipitation.
When the crude oil is cooled below the pour point, high-boiling hydrocarbons can
precipitate and lead to clogging of the tubing string or the output lines. Counter-
measures are internal plastic lining of pipes, hot oil treatment, additions of che-
mical inhibitors to the oil, and periodic cleaning of tubing and lines with scrapers
or go-devils. Incompletely removed paraffin precipitates are often the reason why
reservoir bed treatments and measures to improve the inflow are unsuccessful.
Chemical Precipitates. Impurities such as carbonates, sulfates, and
sulfur can impair production, treatment, and water injection by obstructions in
the reservoir, sealing of the perforation, and narrowing of lines to the point of
shutdown.
Sand. Fine sand from the reservoir bed can considerably hinder opera-
tion of the pumps and cause damage by blocking and increased wear. Counter-
measures are:

1) Borehole measurements for the early detection of danger and the subse-
quent establishment of a safe production rate which must not be
exceeded.
2) Retention of the sand by filters consisting of slotted tubes which can be
wound with one or more layers of wire or jacketed with a cylinder of plas-
tics-reinforced sand. When the product liquid flows into the filter, the en-
trained sand is filtered off, and then it deposits and/or fills the annulus
between filter and borehole wall. This effect is also achieved by artificially
jetted and injected sand packings (gravel pack, frac pack). The slot and
mesh widths of the filters and the particle size composition of the gravel
packings must be carefully matched to the particle size of the entrained
sand and the flow rates.
3) Artificial sand consolidation by two-component resins that are injected into
the formation. Shrinkage during the consolidation or flushing (flooding
after gelling) ensures the necessary permeability. Failures can force the
abandonment of a well.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 151

Fig. 62. Present development of production depths.

Corrosion. Well production and oilfield installations are exposed to corro-


sion both externally and internally. External corrosion results from aggressive
soils or parasitic electrical currents and internal corrosion from salt water, carbon
dioxide, or hydrogen sulfide contained in the produced fluids. Parts of a plant par-
ticularly at risk are those in which attack by corrosive media takes place in the
presence of mechanical stresses, for example, rod strings and subsurface pumps
in high water cut wells. Salt water dramatically lowers the fatigue strength of
steels under dynamic conditions. Lines and tanks are protected cathodically
and/or by addition of suitable inhibitors to the liquid.
Natural Gas Production Engineering (169 –172). Characteristic
features of natural gas production engineering are reservoir depths of more
than 5000 m (Fig. 62), the completion problems connected with this, and
the technically safe control of the accompanying gases hydrogen sulfide,
carbon dioxide, and nitrogen. To avoid stress corrosion cracking induced by
hydrogen sulfide, sour gas wells are provided with completions that ensure
the circulation of solvents and/or corrosion inhibitors. Selection of appro-
piate materials plays a crucial role for casing and completion. Large treat-
ment plants remove environmentally damaging hydrogen sulfide by gas
scrubbing (Grossenkneten) and the nitrogen by liquefaction (Clenze) (see
also Natural Gas).
High pressure and high pressure differentials, extreme temperature differ-
ences, and aggressive gas constituents require special attention in the produc-
tion equipment of gas wells. The gas is kept away from the actual borehole
casing by the production equipment (Fig. 49) which, in the simplest case, consists
152 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

of tubing and packer. The necessity of complete and durable sealing makes extre-
mely high demands on the material and of machining precision. For gas wells,
special gas-tight tubing joints have been developed. In the selection of tubing
and its joints, the large stress resulting from high pressure and temperature
changes, causing large load reversals, must be taken into account.
The evolvement of reservoir pressure with time, its depth, the inflow area of
the well, the expected production period, and the possible liquid yield during pro-
duction determine the selection of tubing size and consequently the production
rate.
Packers seal off one or several producing zones from each other and from the
annulus. The sealing materials used must be temperature-stable, permanently
elastic, and gas- and H2S-resistant. For the further security of the well, the
annulus is usually filled with a protective liquid (inhibited water).
Essential elements of the tubing string are devices for well-engineering
operations, for the control and monitoring of production, and for remedial
measures, such as the injection of inhibitors or agents against hydrate formation.
These devices are installed, pulled out, and actuated on the wire line. Multizone
completions, borehole repairs, selective test and treatment operations, the killing
of wells, and gas-lift production require the selective, manipulable opening of the
tubing string. For this purpose sliding side door couplings are installed, which
are also actuated with tools on the wire line.
Production is controlled and monitored by control valves and measuring
sections. It is generally controlled as a function of rate or pressure. The rates
are determined by orifice meter measurements. (Fig. 63).

Fig. 63. Flow diagram of a natural gas drying plant with external cooling and pressure
increase. a) Prefractionator; b) Continuous heater; c) Glycol regeneration; d) Glycol pump;
e) Heat exchanger; f) Motor needle valve; g) Chiller (coolant NH3); h) Refrigeration plant;
i) Collector; j) Condenser; k) Piston compressor; l) Low-temperature separator (10 8C); m)
Glycol – oil separator; n) Condensate tank; o) Flare; p) Compressor plant; r) Measuring
section.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 153

Safety is particularly important. At the surface, for example, the valves are
normally kept open via a low pressure gas system. If the pressure drops in the
sensor system, these valves close the production line. With underground safety
valves, a safety valve kept open via a hydraulic control line closes the tubing
string as soon as the pressure in the control line falls. Other arrangements pre-
vent the freezing-up of control devices and lines at the surface by hydrate forma-
tion, and protect the surface installations from excessively high pressure.
Numerous exhausted natural gas fields are converted into natural gas
deep storage reservoirs, where demand and suitability allow (see Natural Gas).
Production engineering questions are also very important for storage reservoirs,
especially since withdrawal rates at up to ten times the production rate of a gas
deposit are aimed at.
Offshore Recovery (168,173). Liquid and gaseous hydrocarbons in
reservoirs under the sea bottom are recovered and produced by equipment
and devices which correspond in principle to those of onshore installations. A
distinction is made between supporting structures designed to incorporate dril-
ling rigs used only temporarily, and those to incorporate production and treat-
ment facilities which must be installed in the sea over the whole period of
exploitation of the deposit, frequently more than 20 years.
Because of the difficult environmental conditions and of the danger to the
staff from the processing of highly inflammable hydrocarbons in a very small
space (especially gases), the safety requirements for offshore units are particu-
larly high.
The positioning of the supporting structures over the defined locations in
the sea no longer presents difficulties in principle, since a very wide range of
navigation systems, such as satellite, Doppler, and radio navigation can be used.
Offshore wells require, for the distance between sea bottom and platform, a
special string of casing known as a riser, which must resist the weather condi-
tions, the motion of the sea, and the current. With exploration in ever deeper
water, platforms are being built and operated in ocean depths down to 600 m
(Fig. 64) and development and planning is already geared to ocean depths ca.
1200 m. The riser technology has also been pushed ahead; the safety engineering
aspects of continuous well monitoring under extreme weather conditions being a
priority. For the Shell Offshore East Coast Project a riser of 4000 feet (2121
meters) is planned in connection with a dynamically positioned drilling vessel.

Fig. 64. Supporting structures for offshore production plants.


154 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

Fig. 65. Mobile offshore drilling rigs. A) Submersible, water depth to 300 m; B) Jack-up
platform, water depth to 100 m; C) Drill ship, water depth to 500 m.

In deeper waters, in areas with shipping lanes, and for smaller fields, subsea
completion is sometimes used, in which wellhead and manifold are directly on
the sea bottom.
Supporting Structures for Offshore Drilling Rigs (Fig. 65). Barges
were the first constructional systems used for installing drilling equipment and
are still in use. These consist of flat hulls which are sunk (flooded) in several stor-
eys above each other in shallow water regions. Such installations must not be
exposed to currents and waves. The barge platforms are preferably used in
coastal swamp and mud flat regions, inland seas, and similar locations. These
installations can be used to water depths of up to ca. 10 m. They are dismantled
by refloating the barges.
Fixed Platforms. These are working platforms erected over the surface
of the sea on piles rammed into the bottom. Sometimes it can be appropriate
that only the derrick is erected on such a platform, while the remaining
equipment, materials, energy plants, crew accommodation, etc., are accommo-
dated on a ship anchored nearby. The field of use of these platform – ship
installations extends to ca. 30 m depth of water. However, this combined construc-
tion method can be used only in regions without difficult currents and weather
conditions.
Jack-Up Platforms (Fig. 62 B). Jack-up platforms are flat, multi-storey
barges that have three, four, six or more legs. As soon as these hulls have
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 155

been towed to their place of use, the legs are generally lowered vertically into the
sea bottom. The platform is raised on the legs so far above the sea floor (which
must be stable) that storm waves of maximum height 32 m (measured in centen-
nial storms) cannot beat under the hull of the barge.
These platforms are equipped for a payload of a few 1000 t of material,
depending on their operating location. The derrick and all necessary auxiliary
equipment, such as a generator (electric drive for the rotary table, hoist equip-
ment, and mud pumps), cementing, and mud installations, measurement ser-
vices, etc., are permanently installed. Crew rooms and a helicopter landing
deck are necessary support facilities.
Material is transferred from the supply ships by permanently installed
rotary cranes. The field of application of such rigs extends to about 100 m
depth of water. Because of good stability, the dependence of operations on the
weather is very small.
Semisubmersibles (Fig. 62 A) are floating offshore platforms that can be self-
propelled. The work platform rests on 3 – 8 slender columns, which themselves
rest on two long, parallel floats. The latter are usually submerged to
6 – 9 m during transportation to the location, to give more favorable sailing
conditions. At the location, sea water ballast is taken on and the draft increased
to about 20 – 25 m to lower the center of gravity. This results in small movements
with the heaving swell which together with the reduced water plane area pre-
vent large amplitudes of motion when in the drilling condition.
In addition, the semisubmersible is anchored to its location. Despite the use
of these measures there is an increased effect of waves and storms on these rigs
compared to other offshore units so that servicing is more restricted than of the
units previously described. The main fields of use are water depths between 50
and 600 m. The variable deck load reaches ca. 5000 t, depending on type. The
engineering installations required are the same as those of the jack-up platform.
Drillships (Fig. 62 C). Drillships were used very early in offshore engineer-
ing. Conventional ships were converted by erecting the derrick amidships and
also accommodating the other engineering installations on the ship. Under the
derrick a square recess (moonpool) was provided to make drilling possible. Gra-
dually special drillships with different shapes have been built for very diverse
applications. However, drillships are used only in small numbers since, owing
to their large motions with the heaving swell and despite dynamic positioning
over the borehole, they can be used only to a limited degree on the open seas
and oceans.
Demand also plays a part. Because of their greater hydrocarbon expecta-
tion, it is mainly the shelf edges which have been explored up to now, and the
deep sea hardly at all, so that drillships have largely been dispensed with. But
they offer considerable advantages in regions with fairly stable weather, because
of their high transit speed and fairly large carrying capacity (and therefore small
dependence on external supply).
The deep sea drillships ‘‘Joides Resolution,’’ and its predecessor the
‘‘Glomar Challenger’’ have undertaken studies of the deep sea bottom and its
underground as part of a broad program of geoscientific research.
Drilling rigs on fixed production platforms are used in fairly deep waters
where a field has already been explored by other drilling rigs and a larger
156 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

number of production wells must now be drilled. These platforms are already
prepared for later production. More than 40 holes can be drilled fanwise, direc-
ted to defined target points. The production installations for the extraction
of the hydrocarbons are installed later as required with the aid of rigging-up
ships.
Supporting Structures for Offshore Production Plants (Fig. 61).
The production platforms used for offshore oil and gas production are of the
same type as those for the drilling rigs of fixed platforms and semisubmersi-
bles. The first platforms were built as wooden structures, but since 1940
they have been constructed only of steel or concrete. All submersed steel
structures require corrosion protection in order to ensure a sufficient service
life. For this, direct protection paint and/or galvanization is used and in addi-
tion cathodic protection devices feeding external current or sacrificial anodes
are applied.
In the Norwegian and British sectors of the North Sea, several large con-
crete structures have been set down on the sea bottom. These consist of three hol-
low conical bodies of concrete, constructed on the sliding molding principle, that
contain, as an upper annex, multistorey production installations and personnel
accommodation; the derrick also is arranged there.
From these large structures, a fairly large number of deflected holes are
sunk. Two legs contain the production casing of the wells. The third leg receives
essentially the supply and line system. The base of the structure is provided with
a large number of concrete tanks in which the crude oil, salt water, and other
products are stored. The first stages of the structures were built in Norwegian
fjords, on a construction deck raised on sheet piles. Then, when the structure
had reached a certain stage, the installations were floated off and towed into dee-
per water, in order to allow immersion during the further weight increase result-
ing from concreting to the final height. The maximum total height of these
installations in the Brent and Statfjord fields is 250 m at a water depth down
to 175 m. A small wave pressure and short installation times at the location
(no ramming work) were reasons for constructing such structures, which, how-
ever, can be set down only on a suitably firm sea bottom. The largest offshore
platform in the world is 271 m high and has been installed in the Norwegian
Statfjord field.
Offshore Well Equipment. An important technical problem in drilling
at sea is the effect of the heaving swell and of the current on the sea bottom.
Movements and vibrations of the drilling rig in relation to the pipe strings
leading to the sea bed (riser and drill string) must be avoided. However,
these undesirable effects cannot completely be removed especially on floating
drilling rigs (semisubmersible and drillship). Devices are used to combat these
difficulties. They are installed in the hoist installations and in the casing
string (riser). For wells in shallow water areas there are no longer any special
problems. Here completion arrangements can be employed which are also in
use on land, with casing held in a ‘‘well protector’’ and led out above the
surface of the sea.
However, in fairly deep water and in the area of shipping lanes, underwater
completion must be used. This was first applied in 1943 in the 11.5 m deep water
of Lake Erie. Meanwhile, such systems, usually remote-controlled, have been
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 157

installed in a great variety of sea areas of the world. Installations that cannot be
reached by divers are not yet in use. Two types of completion exist: wet comple-
tion, in which all the components are in contact with the sea, and dry completion,
in which the components are in a pressure chamber under atmospheric condi-
tions, so that the personnel can carry out assembly and maintenance of the
installation by ‘‘submarine passenger ferry.’’ The criteria for the selection of sui-
table completion include water depth, weather conditions, areas with shipping
lanes, icebergs, mechanical properties of the sea-bottom and the possibility of
sediments wandering at sea bottom.
The use of divers, drilling rigs, or workover ships is required for necessary
repairs, possible replacement of the Christmas tree and the tubing, and workover
operations.
Logistics. As soon as the production of oil and gas in the offshore sector
is taken up, provision must be made for the transportation of these products.
Underwater pipelines and ships are primarily used. An essential prerequisite
for the smooth and secure operation of offshore drilling and production activ-
ities is the existence of a land base. The land base provides the maintenance
service, all supplies of materials, as well as the catering, intelligence, and
rescue services, and usually the exchange of personnel by helicopter, etc.

Acknowledgement

Chapter 1 is based on the article Erdöl und Erdgas, Bildung, Suche und
Vorkommen by Hans-Ulrich Hark { in the 4th edition of Ullmann’s.

Abbreviations

A cross-sectional area, m2
As swept area, m2
B formation volume factor
Bo,w,g formation volume factor for oil, water, gas
Bt two-phase formation volume factor
c compressibility coefficient, Pa1
cf pore volume compressibility, Pa1
co,w,g compressibility for oil, water, gas, Pa1
cpc pseudocritical compressibility
cpr pseudoreduced compressibility
cr reduced compressibility
C critical point
D production decline rate, s1
D hole diameter
EA areal oil recovery factor
ED microscopic oil recovery factor
EI vertical recovery factor
ER recovery efficiency
158 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

f fractional flow
fo,w,g fractional flow for oil, water, gas
F surface area of cutting particles
g acceleration due to gravity, m s2
Gp gas produced, m3
h thickness, elevation, m
k permeability of rock, m2
k drillability coefficient
ko,w,g effective permeability for oil, water, gas, m2
kr relative permeability
kro,w,g relative permeability for oil, water, gas
l length, m
m gradient
Mr molecular mass
M mobility ratio
N initial oil volume, original oil in place, m3
Np oil produced, m3
Npa maximum cummulative production rate
OGIP original gas in place
OOIP original oil in place
p pressure, Pa
pb bubble point pressure, degassing pressure, Pa
pc capillary pressure, Pa
pe pressure at reservoir boundary, Pa
pm mean pressure
ppc pseudocritical pressure
ppr pseudoreduced pressure
pr reduced pressure
pwf flowing pressure in well, Pa
pws shut-in pressure, Pa
P weight on bit
q flow rate, m3 s1
qa limiting production rate
qo,w,g flow rate for oil, water, gas, m3 s1
r radius, m
re radius of area drained by the well
rw,c well radius, reservoir boundary radius, m
Rp produced oil – gas ratio
Rs gas – oil ratio
Rsi initial gas – oil ratio
S saturation
S drillability of the rock
So,w,g saturation oil, water, gas
Sor residual oil saturation
Swf front saturation
Swi initial water saturation, irreducible water saturation
t time, s
tp producing time, s
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 159

T temperature, K
Tc capillary temperature
Tpc pseudocritical temperature
Tr reduced temperature
v velocity, m s1
v apparent velocity
V cuttings volume, m3
Vb bulk volume, m3
VB total rock volume
Vp pore volume, m3
Wi volume of water, m3
Wp water produced, m3
x distance
z real gas factor
g Euler constant
y contact angle
Y static shear stress, dyne/cm2
m dynamic viscosity
mg viscosity of a real gas, Pa s
%o,w,g,f density of oil, water, gas, fluid, kg/m3
%m density of mud
s surface tension, interfacial tension
f porosity
fd dead porosity
feff effective porosity
F flow potential, Pa
lo,w,g mobility of oil, water, gas

Subscripts

i, 0 initial
r residual
b at bubble point
0 standard
j component j
f fluid

REFERENCES

GENERAL REFERENCES

1. A. W. Bally: ‘‘Seismic Expression of Structural Styles,’’ AAPG Stud. Geol. 15


(1983).
2. T. Bandlova, G. Katzung: ‘‘Zum Auftreten von Erdgas in Verbreitungsgebieten koh-
leführender Ablagerungen,’’ Z. Angew. Geol. 21 (1975) 467 – 572.
160 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

3. H. Bartenstein, R. Teichmüller: ‘‘Inkohlungsuntersuchungen, ein Schlüssel zur


Prospektierung von paläozoischen Kohlenwasserstoff-Lagerstätten?’’ Fortschr.
Geol. Rheinl. Westfalen 24 (1974) .
4. R. Beck, P. Lehner: ‘‘Oceans, Now Frontier in Exploration,’’ Am. Assoc. Pet. Geol.
Bull. 58 (1974) 376 – 395.
5. H. Beckmann: Geological Prospecting of Petroleum, Enke Verlag, Stuttgart 1976,
pp. 176.
6. F. Bender: ‘‘Geologie der Kohlenwasserstoffe,’’ Angew. Geowissenschaften, vol. 3,
Enke Verlag, Stuttgart 1984.
7. P. Bitterli: ‘‘On the Origin of Petroleum,’’ Bull. Ver. Schweiz. Pet. Geol. Ing. 30 (1964)
no. 79, 74 – 77.
8. H. Boigk: Erdöl und Erdölgas in der Bundesrepublik Deutschland, Enke Verlag,
Stuttgart 1981, pp. 330.
9. H. Boigk, H. W. Hagemann, W. Stahl, G. Wollanke: ‘‘Zur Herkunft und Migration
des Stickstoffs nordwestdeutscher Erdgase im Oberkarbon und Rotliegend,’’ Erdöl
Kohle Erdgas Petrochem. 75/76 (1986) 112 – 139.
10. H. Boigk, H. U. Hark, H. J. Meyer, H. Wehner: ‘‘Beziehungen zwischen Geochemie,
Migration und Lagerstättengenese im Niedersächsischen Becken (westlich der
Ems),’’ Erdöl Kohle Erdgas Petrochem. 74/75 (1986) no. 1, 98 – 114.
11. C. Bois, P. Bouche, R. Pelet: ‘‘Global Geologic History and Distribution of Hydrocar-
bon Reserves,’’ Am. Assoc. Pet. Geol. Bull. 66 (1982) no. 9, 1248 – 1270.
12. L. F. Brown, Jr., W. L. Fisher: Seismic Stratigraphic Interpretation and Petroleum
Exploration, AAP Continuing Education Course Series, 1980, p. 16.
13. M. S. Burštar: Osnovy teorii formirovanija zeležej nefti i gaza (Grundlagen der
Theorie der Bildung von Erdöl- und Erdgaslagerstätten), Nedra, Moskau 1973,
pp. 255.
14. R. E. Chapman: Petroleum Geology, Elsevier, Amsterdam 1983.
15. G. Dohr: Applied Geophysics in Geology of Petroleum, vol. 1, Enke Verlag, Stuttgart
1981.
16. W. Dow: ‘‘Petroleum Source Beds on Continental Slopes and Rises,’’ Am. Assoc. Pet.
Geol. Bull. 59 (1970) no. 6, 1584 – 1606.
17. H. Grunau: ‘‘Abundance of Source Rocks for Oil and Gas Worldwide,’’ J. Pet. Geol. 6
(1983) no. 1, 39 – 54.
18. H. D. Hedberg: Geologic Aspects of Origin of Petroleum, Am. Pet. Geol. Bull. 48
(1964) 1755 – 1803.
19. D. Heling et al.: Sediments and Environmental Geochemistry, Sedimentological and
Petrophysical Aspects of Primary Petroleum Migration Pathways, Springer Verlag,
Berlin 1990, 152 pp.
20. G. Hosson, W. Pohl: Modern Petroleum Technology, 4th ed., Inst. of Petroleum, Lon-
don 1973.
21. J. M. Hunt: Petroleum Geochemistry and Geology, W. H. Freeman, San Francisco
1979, pp. 617.
22. V. C. Illings: ‘‘Geology Applied to Petroleum,’’ Proc. Geol. Assoc. 53 (1942) 156 – 187.
23. A. Jegorow: ‘‘Paläogeographische Beziehungen zwischen kohle-, erdöl- und erdgas-
führenden Gebieten des Karbons,’’ C. R. Sept. Congre`s Internat. de Stratigraphie et
de Geologie sur Carbonife`re, vol. 3 , Krefeld 1974, pp. 249 – 259.
24. H. Jüntgen, J. Klein: ‘‘Entstehung von Erdgas aus kohligen Sedimenten,’’ Erdöl
Kohle Erdgas Petrochem. 74/75 (1986) no. 1, 52 – 69.
25. R. F. K. Kinghorn: An Introduction to the Physics and Chemistry of Petroleum,
Wiley-Interscience, New York 1983, pp. 432.
26. I. Köster, H. Kulke, H. W. Ladwein: ‘‘Neuentdeckte und potentielle Erdölvorkom-
men,’’ Erdöl Kohle Erdgas Petrochem. (1992) no. 3, 106.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 161

27. D. Leythaeuser, H. W. Ladwein: ‘‘Wechselbeziehungen zwischen Erdölgenese und


Migration,’’ Erdöl Kohle Erdgas Petrochem. (1992) no. 3, 108.
28. T. P. Lowell, A. J. D. Harding: ‘‘Structural Styles, Their Plate Tectonic Habitat, and
Hydrocarbon Traps in Petroleum Provinces,’’ Am. Assoc. Petrol. Geol. Bull. 63 (1979)
1016 – 1058.
29. R. McQuillin, M. Bacon, W. Barclay: An Introduction to Seismic Interpretation—Re-
flection Seismics in Petroleum Exploration, Graham & Trotman, London 1986,
pp. 287.
30. A. Melton, A. C. Giardini: ‘‘A Scientific Explanation for the Origin and Location of
Petroleum Accumulations,’’ J. Pet. Geol. 6 (1983) 117 – 138.
31. G. B. Moody: Petroleum Exploration Handbook, McGraw-Hill, New York 1961.
32. G. J. Moore, G. T. Demaison: ‘‘Anoxic Environments and Oil Source Bed Genesis,’’
Am. Assoc. Pet. Geol. Bull. 64 (1980) no. 8, 1179 – 1209.
33. R. J. Murris: ‘‘Middle East: Stratigraphic Evolution and Oil Habitat,’’ Am. Assoc. Pet.
Geol. Bull. 64 (1980) no. 5, 597 – 618.
34. N. S. Neidell: ‘‘Stratigraphic Modeling and Interpretation: Geophysical Principles
and Techniques,’’ AAPG Education Course Note Series 13 (1981) 141.
35. F. K. North: Petroleum Geology, Allen & Unwin, Boston 1985, pp. 607.
36. M. R. Pegrum, G. Rees, D. Naylor: Geology of the North-West European Continental
Shelf, vol. 2, The North Sea. Graham Trotman Dudlex Publ., London 1975.
37. R. G. Schäfer, D. H. Welte: ‘‘Entstehung von Erdöl und Erdgas,’’ Prax. Naturwiss.
Chem. 41 (1992) no. 2.
38. M. Schoell: ‘‘Kohlenstoff- und Wasserstoffisotopenanalysen in der Erdöl- und
Erdgasforschung,’’ 3eme Cycle Romand des Sciences de la Terre, 1980, pp. 1 – 47.
39. M. Schoell: Genetic Characterisation of Natural Gases, Am. Assoc. Pet. Geol. Bull. 67
(1983) no. 12, 2225 – 2238.
40. R. C. Selley: Elements of Petroleum Geology, Freeman, New York 1985, pp. 449.
41. R. E. Sheriff: ‘‘Seismic Detection of Hydrocarbons: The Underlying Physical
Principles,’’ Offshore Technology Conference 6, Preprint 1, Houston 1974,
pp. 637 – 649.
42. V. Sokolov, A. Geodekian, Z. Buniat-Zade: ‘‘The General Scheme of Petroleum and
Gas Formation, Alteration and Migration in the Earth’s Crust,’’ Adv. Org. Geochem.
Proc. Int. Meet. 4th (1968) 279 – 288.
43. V. Stroganov: ‘‘Principal Phases in Origin of Gaseous and Liquid Hydrocarbons and
Conditions of Formation of Zones of Oil – Gas Accumulations,’’ Int. Geol. Rev. 16
(1974) 769 – 776.
44. M. Teichmüller: ‘‘Anwendung kohlenpetrographischer Methoden bei der Erdöl- und
Erdgasprospektion,’’ Erdöl Kohle Erdgas Petrochem. 24 (1971) 69 – 76.
45. E. N. Tiratsoo: Oilfields of the World, Scientific Press, Beaconsfield 1984, pp. 392.
46. G. D. Tiratsoo, A. E. N. Hobson: Introduction to Petroleum Geology, Scientific Press,
Beaconsfield 1981, pp. 352.
47. B. P. Tissot: ‘‘Recent Advances in Petroleum Geochemistry Applied To Hydrocarbon
Exploration,’’ Am. Assoc. Pet. Geol. Bull. 68 (1984) no. 5, 545 – 563.
48. A. Treibs: ‘‘Entstehung des Erdöls,’’ Erdöl Kohle 1 (1948) 137 – 143, 185 – 199.
49. F. Troost, P. J. P. M. van Beek: ‘‘The Groningen Gas Field: A Case History of
the Development of a Giant Gas Field,’’ JPT J. Pet. Technol. (1979) July, 815 –
820.
50. N. B. Wassojewitsch et al.: ‘‘Die Hauptphase der Erdölbildung,’’ Z. Angew. Geol. 15
(1969) 612 – 615.
51. D. H. Welte: ‘‘Zur Entwicklungsgeschichte von Erdölen auf Grund geochemisch-
geologischer Untersuchungen,’’ Erdöl Kohle Erdgas Petrochem. 20 (1967)
65 – 77.
162 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

52. D. H. Welte: ‘‘Petroleum Exploration and Organic Geochemistry,’’ J. Geochem.


Explor. 1 (1972) 117 – 1939.
53. E. Whitehead: ‘‘Molecular Evidence for the Biogenesis of Petroleum and Natural
Gas,’’ Proc. Symp. Hydrogeochem. Biogeochem. 2 (1973) 158 – 211.
54. P. A. Ziegler: ‘‘Öl- und Gas-Provinzen der Nordsee,’’ Erdöl Erdgas Z. 91 (1975)
207 – 217.
55. P. A. Ziegler: Geological Atlas of Western and Central Europe, Shell International
Petroleum Maatschappij B.V., 1982.

SPECIFIC REFERENCES
56. V. B. Porfirev: ‘‘Inorganic Origin of Petroleum, Am. Assoc. Petrol. Geol. Bull. 58
(1974) 3 – 33.
57. B. Tissot, D. Welte: Petroleum and Occurrence, Springer Verlag, Heidelberg
1978.
58. D. H. Welte: ‘‘Organischer Kohlenstoff und die Entwicklung der Photosynthese auf
der Erde,’’ Naturwissenschaften 57 (1970) 17.
59. W. Schott: ‘‘Aufsuchen von Erdöl- und Erdgaslagerstätten’’ in A. Bentz, H. J. Martini
(eds.) Lehrbuch der Angewandten Geologie, vol. 2, part 1, Enke Verlag, Stuttgart
1968, pp. 563 – 671.
60. C. C. Cornelius: ‘‘Geothermal Aspects of Hydrocarbon Exploration in the North Sea
Area’’, NGU-Nor. Geol. Unters. 1975, no. 316, 29 – 67.
61. A. Hood, J. R. Castano: Organic Metamorphism: Its Relationship to Petroleum Gen-
eration and Application To Studies of Authigenic Minerals, United Nations ESCAP
CCOP, Techn. Bull. 8, 8 – 113.
62. G. Philippi: ‘‘The Influence of Marine and Terrestrial Source Material on The Com-
position of Petroleum,’’ Geochim. Cosmochim. Acta 38 (1974) 947 –966.
63. K. Krejci-Graf: ‘‘Moderne Anschauungen über die Entstehung des Erdöls,’’ Erdöl
Kohle Erdgas Petrochem. 13 (1960) 836 – 845.
64. U. Mann: ‘‘Revealing Hydrocarbon Migration Pathways,’’ Geol. Rundschau 78 (1990)
no. 1, 337 –348.
65. H. Klemme: Petroleum Basins – Classification and Characteristics,‘‘ J. Pet. Geol. 3
(1980) no. 187 – 207.
66. P. A. Ziegler: ‘‘Faulting and Graben Formation in Western and Central Europe,’’
in: P. Kent, M. H. P. Bott, D. P. McKenzie, C. A. Williams (eds.): ‘‘The Evolu-
tion of Sedimentary Basins,’’ Philos. Trans. R. Soc. Ser. London A 305 (1982)
113 –143.
67. W. Stahl, J. Koch: ‘‘13 C/12 C — Verhältnis norddeutscher Erdgase — Reifemerkmal
ihrer Muttersubstanzen,’’ Erdöl Kohle Erdgas Petrochem. 27 (1974) no. 10.
68. W. Stahl: ‘‘Carbon Isotope Ratios of German Natural Gases in Comparison with
Isotope Data of Gaseous Hydrocarbons from Other Parts of the World,’’ Adv. Org.
Geochem. Proc. Int. Meet. 6th (1973) 453 – 462.
69. D. H. Welte, M. A. Jükler: ‘‘Petroleum Origin and Accumulation in Basin
Evolution — A Quantitative Model,’’ Am. Ass. Petrol. Geol. Bull. 65 (1981)
1387 – 1390.
70. W. Rühl: Energiefaktor Erdöl, Edition Interform, Zürich 1989.
71. C. D. Masters, D. H. Root, E. D. Attanasi: ‘‘World Resources of Crude Oil and
Natural Gas,’’ Proceedings of World Petrol. Congress, Publ. J. Wiley and Sons,
Chichester 1991.
72. O. Serra: Fundamentals of Well Log Interpretation, vol. 2, Acquisition of logging
Data; Development in Petroleum Science, Elsevier, Amsterdam 1987.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 163

73. L. C. Graton, H. J. Fraser: ‘‘Systematic Pacing of Spheres – with particular Relation


to Porosity and Permeability,’’ J. Geol. (1935) Nov.– Dec., 785 – 909.
74. H. Pape, L. Riepe, J. R. Schopper: ‘‘Theory of Self-Similar Network Structures in
Sedimentary and Igneous Rocks and their Investigations with Microscopical. and
Physical Methods,’’ J. Microsc. (Oxford) 148 (1987) part 2, 121 – 147.
75. J. Geertsma: ‘‘The Effect of Fluid Pressure Decline Oil Volumetric Changes of Porous
Rocks,’’ Trans. Am. Inst. Min. Metall. Pet. Eng. 210 (1957).
76. N. H. Hall: ‘‘Compressibilities of Reservoir Rocks,’’ Trans. Am. Inst. Min. Metall. Pet.
Eng. 198 (1953) 309.
77. J. W. Amyx, D. M. Bass, R. L. Whiting: Petroleum Reservoir Engineering, Physical
Properties, McGraw-Hill, New York 1960.
78. W. D. McCain: The Properties of Petroleum Fluids, Petroleum Publishing Com-
pany, 1973.
79. E. J. Lynch: Formation Evaluation, Harper Row, New York 1962.
80. A. R. Martinez et al.: ‘‘Classification and Nomenclature Systems for Petroleum and
Petroleum Reserves 1987 Report,’’ Proc. World Pet. Congr. 12th (1987).
81. F. Bender: ‘‘Methoden der angewandten Geophysik und mathematische Verfahren
in den Geowissenschaften,’’ in Angewandte Geowissenschaften, vol. 2, Enke Verlag,
Stuttgart 1985.
82. E. Winnock, Y. Pontalier: ‘‘Lacq Gas Field; Geology of giant Petroleum Fields,’’ Am.
Ass. Pet. Geol. Bull. Nov. (1970) 370 – 387.
83. J. Lee: ‘‘Well Testing,’’ SPE Textbook Series, vol. 1, Texas A & M University,
1982.
84. D. R. Homer: ‘‘Pressure Build-Up in Wells,’’ Proc. Third World Petroleum Congress,
vol. II, Leiden 1951, p. 503.
85. R. E. Collins: Flow of Fluids through Porous Materials, Reinhold Publishing Cor-
poration, New York 1961.
86. M. Honarpour, L. Koederitz, A. H. Harvey: Relative Permeability of Petroleum Reser-
voirs, CRC Press, Boca Raton 1986.
87. H. L. Stone: ‘‘Probability Model for Estimating Three-Phase Relative Permeability,’’
J. Petr. Techn. (1970) Feb., 214 – 219.
88. F. W. Cole: Reservoir Engineering Manual, Gulf Publ. Corp., Houston 1969.
89. M. C. Leverett: ‘‘Capillary Behavior in Porous Solids,’’ Trans. Am. Inst. Min. Metall.
Pet. Eng. 142 (1941) 152 – 169.
90. S. E. Buckley, M. C. Leverett: ‘‘Mechanism of Fluid Disphacements in Sands,’’
Trans. Am. Inst. Metall. Pet. Eng. 146 (1942) 107 – 116.
91. H. J. Welge: ‘‘A Simplified Method for Computing Oil Recovery by Gas or Water
Drive,’’ Trans. Am. Inst. Metall. Pet. Eng. 195 (1952) 91 – 98.
92. F. F. Craig: ‘‘The Reservoir Engineering Aspects of Waterflooding,’’ SPE Monograph,
vol. 3, New York 1971.
93. H. Dykstra, R. L. Parsons: ‘‘The Prediction of Waterflood Performance with Varia-
tion in Permeability Profile,’’ Prod. Monthly 15 (1950) 9 – 12.
94. B. C. Craft, M. F. Hawkins: Applied Petroleum Reservoir Engineering, Prentice Hall,
Englerwood Cliffs 1959.
95. J. Tarner: ‘‘How Different Size Gas Caps and Pressure Maintenance Programs Affect
Amount of Recoverable Oil,’’ Oil Weekly 144 (1944) June 12, 32 – 34.
96. D. W. Peaceman: ‘‘Fundamentals of Numerical Reservoir Simulation,’’ vol. 6, Devel-
opments in Petroleum Science, Elsevier, Amsterdam 1977.
97. H. B. Crichlow: Modern Reservoir Engineering – A Simulation Approach, Prentice-
Hall, Englewood Cliffs, New Jersey 1977.
98. H. C. Slider: Worldwide Practical Petroleum Reservoir Engineering Methods, Penn
Well Publ. Co., Tulsa 1983.
164 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

GENERAL REFERENCES

99. W. Arnold: Eroberung der Tiefe, VEB Dtsch. Vlg. f. Grundstoffind., Leipzig 1965.
100. D. Kett, Gas-Wasserfach 108 (1967) 221 – 229.
101. Oil Gas J. 73 (1975) Jan., 72.
102. Oil Gas J. 72 (1974) July, 33.
103. S. J. Pirson, Oil Reservoir Engineering, McGraw-Hill, New York 1958.
104. P. L. Randolph, World Oil 1974, Aug., 33 – 36; Sept., 65 – 71; Oct., 131 – 134; Nov. ,
81 – 84.
105. H. Rischmüller, Erdöl Erdgas Z. 19 (1966) 91 – 104.
106. H. Rischmüller, Erdöl Erdgas Z. 82 (1966) 90 – 99.
107. H. Rischmüller, H.-M. Tröndle: ‘‘Meerestechnik beim Ozeanbergbau,’’ in G. Dietrich
(ed.): Erforschung des Meeres, Umschau-Vlg., Frankfurt 1970, 267 – 285.
108. H. Rischmüller, Erdöl Erdgas Z. 88 (1972) 110 – 114.
109. K. Trenel, Erdöl Erdgas Z. 83 (1967) 188 – 191.
110. W. Tunn, Erdöl Erdgas Z. 87 (1971) July, 202 – 208.
111. E. Wichert, K. Aziz, Can. J. Chem. Eng. 49 (1971) 267 – 273.
112. R. A. Sumbatov, B. N. Khakhaev, V. N. Gramolin: ‘‘Drilling Rigs and Technical
Facilities for Superdeep Well Drilling,’’ Paper presented at the IGC, Washington
1989.
113. D. Noran: Oil Gas J. 73 (1975) Nov., 77 – 81.
114. Erdölinformationsdienst 28 (1975) no. 41.
115. Erdölinformationsdienst 29 (1975) no. 5.
116. W. Simmler, Gas-Wasserfach 109 (1968) 724 – 730.
117. Mannesmannröhren-Werke: Rohre für Öl- und Gasfelder, 6th ed., vols 1 – 3,
1962.
118. International Petroleum Encyclopedia, The Petroleum Publ. Co., Tulsa 1975.

SPECIFIC REFERENCES

119. R. D. Faniev: Abbau von Erdöl- und Erdgaslagerstätten, VEB Dtsch. Vlg. f. Grund-
stoffind., Leipzig 1963.
120. A. Mayer-Gürr: ‘‘Erdöl- und Erdgasgewinnung in der Bundesrepublik Deutschland,’’
Oel-Zeitschrift für die Mineralölwirtschaft 1 (1973).
121. K. K. Millheim: ‘‘The New Engineering Paradigm and the Emergence of Investiga-
tive Engineering,’’ J. Petrol. Technol. 1989, Aug.
122. Ullmann 4th ed., 11, pp. 19 – 40.
123. H. Rischmüller: ‘‘Tiefbohrtechnik als Schlüsselinstrument in der Energiewirtschaft,’’
Geowissenschaften 10 (1990) 317 – 323.
124. Ö. Aliquander: Das moderne Rotarybohren, VEB Dtsch. Vlg. f. Grundstoffind.,
Leipzig 1965.
125. W. Arnold: Eroberung der Tiefe, VEB Dtsch. Vlg. f. Grundstoffind., Leipzig 1983.
126. H. F. Spörker: ‘‘Entwicklungstendenzen der Tiefbohrtechnik,’’ Erdöl Erdgas Z. 88
(1972) 104 – 110.
127. K.-H. Grodde: Bohrspülungen und Zementschlämme in der Tiefbohrtechnik, Vlg. d.
Erdöl-Z. Otto Vieth, Hamburg 1963.
128. C. Marx, H. Rischmüller: ‘‘Drilling and Coring Techniques for Hard Rock,’’ in
H. J. Behr, F. G. Stehli, H. Vidal (eds.): Observation of the Continental Crust through
Drilling II, Springer, Berlin 1987, 149 – 160.
129. Maurer Engineering Inc.: Deep Drilling Cost Sensitivity Analysis, Study for Gas
Research Institute, Chicago 1989, GRI Contract No. 5088-260-1731.
Vol. 1 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING 165

130. F. P. Springer: ‘‘Bohrtechnische Aspekte bei der Ausrüstung von Tiefbohranlagen,’’


Erdöl Erdgas Z. 88 (1972) 410 – 417.
131. R. S. Andrews: A Survey of Recent Technology Development in International Con-
tinental Scientific Drilling Programs, Joint Oceanographic Institutions, Inc.,
1990.
132. H. Baker: Horizontal Wells, Reprinted from a series of articles appearing Oil & Gas
J. Sept. 24 – 31, 1990.
133. H.-J. Behr, R. Emmermann: Scientific Objectives and Site-Selection Studies of the
Continental Deep Drilling Programme of the Federal Republic of Germany (KTB),
in K.-J. Behr, F. G. Stehli, H. Vidal (eds.): Observation of the Continental Crust
through Drilling II, Springer, Berlin 1987, 85 – 120.
134. H.-J. Behr, P. Kehrer, H. Rischmüller: ‘‘The German Continental Deep Drilling
Program, Objectives and State of Work,’’ in A. Boden, K. G. Erikson (eds.): Deep
Drilling in Crystalline Bedrock, vol. 2, Springer, Berlin 1988, 64 – 81. W. B. Bleakley,
Oil Gas J. 72 (1974) March, 69 – 78.
135. T. Burgess, P. Van de Slijke: ‘‘Horizontal Drilling Comes of Age,’’ Oilfield Review
1980, July.
136. C. Chur, B. Engeser, J. Oppelt: ‘‘Das Vertikalbohrkonzept für die KTB-Hauptbohrung,’’
Erdöl Erdgas Kohle 106 (1990) 486 – 490.
137. M. Ellins, T. Tran Viet: ‘‘Spülungskonzept und Mud Logging-System für die
KTB-Hauptbohrung,’’ Erdöl Erdgas Kohle 106 (1990) 190, 491 – 495.
138. R. Emmermann: ‘‘Das Deutsche Kontinentale Tiefbohrprogramm,’’ Geowissenschaf-
ten in unserer Zeit 1 (1986) 19 – 33.
139. E. P. Ilaysov et al.: ‘‘Ultradeep Drilling for the Investigation of the Earth’s Crust,’’ 12th
World Petroleum Congress, Preprint of Review and Forecast, Topic 12, Houston 1987.
140. H. Rischmüller, C. Chur: ‘‘Technical Concept and Status of Planning for the Wells of
the Continental Deep Drilling Program (KTB),’’ in H.-J. Behr, F. G. Stehli, H.
Vidal (eds.): Observation of the Continental Crust through Drilling II, Springer,
Berlin 1987, 136 – 149.
141. H. Rischmüller: ‘‘Das kontinentale Tiefbohrprogramm der Bundesrepublik
Deutschland – ein Synergismus von Geowissenschaften und Technik,’’ in Bergbau
im Wandel, Akademische Druck- und Verlagsanstalt, Verlag Glückauf, Essen
1988, 143 – 152.
142. A. Sperber et al.: ‘‘Das technische Konzept der KTB-Hauptbohrung,’’ 3rd Collo-
quium DFG-Schwerpunktprogramm, Gießen 1990.
143. H. F. Spörker: ‘‘Thoughts on Problems and Possible Solutions in Ultradeep
Drilling,’’ in H.-J. Behr, F. G. Stehli, H. Vidal (eds.): Observation of the Continental
Crust through Drilling II, Springer, Berlin 1987, 170 – 184.
144. T. C. Frick: Petroleum Production Handbook, McGraw-Hill, New York 1962.
145. J. E. Kastrop, Petroleum Eng. 47 (1975) Feb., 27 –31.
146. G. V. Chilingar, C. M. Beeson: Surface Operations in Petroleum Production, Elsevier,
New York 1969.
147. K. H. Grodde: Bohrspülungen und Zementschlämme in der Tiefbohrtechnik, Vlg. d.
Erdoel-Z. Otto Vieth, Hamburg 1963.
148. H.-G. Graf, H.-G. Haddenhorst, H. Rischmüller, W. De la Sauce: ‘‘Förder- und Spei-
chertechnik,’’ Erdöl, Erdgas 100 (1984) no. 4.
149. A. P. Szilas: Production and Transport of Oil and Gas, Elsevier, Amsterdam
1975.
150. J. W. Amyx, D. H. Bass, R. L. Whitting: Petroleum Reservoir Engineering, McGraw-
Hill, New York 1960.
151. N. J. Clark: Elements of Petroleum Reservoirs. Soc. Petroleum Eng. of Amer. Inst. of
Mining, Metall. Petroleum Eng., Dallas 1960.
166 EXPLORATION, DRILLING, AND PRODUCTION ENGINEERING Vol. 1

152. B. C. Craft, M. F. Hawkins: Applied Petroleum Reservoir Engineering, Prentice-Hall,


Englewood Cliffs 1959.
153. C. R. Smith: Mechanics of Secondary Oil Recovery, Reinhold Publ. Corp., New York
1966.
154. M. K. Robson: ‘‘Introduction to and Benefits of Tubing-Conveyed Perforating,’’ JPT,
Feb. 1990.
155. H.-J. Dietzel, G. v. Hantelmann, H. Rischmüller: ‘‘Artifizielle Fließwege im
flözführenden Karbon, Erdöl Kohle-Erdgas, 37 (1984) no. 12.
156. J. Gulbis et al.: ‘‘Taking the Brakes off Proppant-Pack Conductivity,’’ Oilfield Review
1991 Jan.
157. G. C. Howard, C. R. Fast: ‘‘Hydraulic Fracturing,’’ H.L. Doherty Series Monograph.
vol. 2, Am. Inst. Min. Metall. & Petrol. Eng., 1970.
158. M. L. Slusser, M. Rieckmann: Erdöl Kohle Erdgas Petrochem. Compendium 75/76,
149 – 161.
159. J. D. Clegg: ‘‘High-Rate Artificial Lift,’’ J. Petrol. Technol. 1988, March.
160. B. C. Craft, W. R. Holden, E. D. GravesJr. Well Design, Drilling and Production,
Prentice-Hall, Englewood Cliffs 1962.
161. T. E. W. Nind: Principles of Oil Well Production, Mc Graw-Hill, New York 1964.
162. H. Rischmüller: ‘‘Alternative synergistische Förderhilfsmittelplanung für produzier-
ende Erdölfelder,’’ Erdöl-Erdgas-Z. 88 (1972) April.
163. H. Rischmüller: ‘‘Kurzer Überblick über den Stand bewährter und die Entwicklung
neuer Verfahren in der Erdölfördertechnik,’’ Erdöl, Erdgas, Kohle 102 (1986)
no. 1.
164. H. Muskat: The Flow of Homogeneous Fluids through Porous Media, Edwards
Broth. Inc., Ann Arbor 1946.
165. C. J. Gott: ‘‘Successful Rod Pumping at 14.500 Feet,’’ SPE Production Engineering
1986 Nov.
166. J. Zaba: Modern Oil-Well Pumping, Petroleum Publ. Co., Tulsa 1962.
167. H. Rischmüller: Erdölförderung mit Gestängetiefpumpen, Verlag Vereinigte
Edelstahlwerke AG und Böhler AG, Düsseldorf 1984.
168. K. Horn, R. Putzer, H. Rischmüller: ‘‘Komplettierung und Förderung von Offshore-
Sonden,’’ Erdöl Z. 80 (1964) 379 – 395.
169. D. L. Katz et al.: Handbook of Natural Gas Engineering, McGraw-Hill,
New York 1959.
170. H. Laurien: Taschenbuch Erdgas, 2nd ed., R. Oldenbourg Vlg., München 1970.
171. M. Medici: The Natural Gas Industry, Butterworth, London 1974.
172. H. Rischmüller: ‘‘Die Gaslagerstätten des Westemslandes,’’ Erdöl Erdgas Z. 84
(1968) 322 – 329.
173. K. Hinz, H. Amann, G. Schendl: ‘‘Erdölexploration im tieferen Wasser,’’ Erdöl Kohle
28 (1975) April, 175 – 182.

KLAUS WEGGEN
Hemmingen, Federal Republic of Germany
GÜNTER PUSCH
Technische Universität Clausthal, Clausthal-Zellerfeld
Federal Republic of Germany
HEINRICH RISCHMÜLLER
Technische Universität Clausthal Clausthal-Zellerfeld
Federal Republic of Germany
Vol. 1 ENHANCED OIL RECOVERY 167

ENHANCED OIL RECOVERY


1. Introduction

Excluding oil sands mining, enhanced oil recovery (EOR) accounted for 199,000 m3/
day (1.25 million b/day) of the world’s total 2000 oil production, 3  105m 3/day
(1.9  106 bbl/day) or 1.6% (1). In 2004, U.S. EOR production [105,740 m3/day
(663,451 bbl/day)] represented 12.2% of total production of crude oil plus gas
condensate (2). Most of the oil basins in the United States are mature.
These fields will require extensive enhanced oil programs to maintain current
production rates or, in some cases, merely to minimize rapidly increasing
depletion rates. For example, production in Prudhoe Bay, the largest U.S.
field, is declining 10% annually despite large investments in enhanced oil recov-
ery technology (3).
Primary and secondary oil recovery together recover only 25–50% of the oil
originally in place in a reservoir. After conventional oil recovery operations, the
U.S. will have 60 billion m3 (377  109) barrels of oil remaining after economic
production ceases. The U.S. Department of Energy has estimated as much as
43.3 billion barrels of oil could be produced using EOR technologies (4). The
most likely location of large new U.S. oil fields is Alaska and deep water offshore.
The costs of discovering and developing these reserves make unrecovered oil in
known fields an economically attractive target in part because much of the infra-
structure is also already in place.
Another factor promoting use of EOR technology in existing oilfields is
that countries with the greatest potential of containing undiscovered conven-
tional oil fields either limit Western investment in oil production (5).
The relatively low oil prices of 1981–2003 have resulted in significant
changes in the EOR technology being developed and field tested. Injection of
steam or oil-miscible gases remains of great interest. Micellar polymer flooding
can efficiently recover oil and was the focus of a large research and development
effort from 1970 to 1986, but this process remains relatively expensive and con-
sequently is seldom applied in field operations.
Using relatively low concentration surfactants as CO2 (6) and steam mobi-
lity control agents (foam) can improve recovery process cost effectiveness. Com-
binations of older EOR processes, eg, surfactant-enhanced alkaline
flooding and alkaline–surfactant–polymer flooding show promise of improved
cost effectiveness.
Enhanced oil recovery, the injection of fluids other than water to increase
oil recovery is but one means of recovering more oil from existing fields. Other
methods include better siting of injection and production wells due to improved
reservoir mapping technologies, infill drilling, horizontal wells, hydraulic frac-
turing, and acidizing.
With the increase of oil prices since 2003, major oil companies and national oil
companies have initiated quite large enhanced oil recovery projects usually using
carbon dioxide or steam as the injectants and are planning more such projects.
Using industrially generated CO2 for EOR while simultaneously sequestering

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.0514080102151803.a01
168 ENHANCED OIL RECOVERY Vol. 1

the injected CO2 in the rock formation has the potential to produce more oil while
reducing emissions of this greenhouse gas (7).

2. The Nature of Oil Reservoirs

Oil reservoirs are layers of porous sandstone or carbonate rock, usually sedimen-
tary. Impermeable rock layers, usually shales, and rock faults trap the oil in
the reservoir. The oil exists in microscopic pores in rock. Various gases and water
also occupy rock pores and are often in contact with the oil. These pores are inter-
connected with a complicated network of microscopic flow channels. The weight of
overlaying rock layers places these fluids under pressure. When a well penetrates
the rock formation, this pressure drives the fluids into the wellbore. The flow chan-
nel size, wettability of flow channel rock surfaces, oil viscosity, and other properties
of the crude oil determine the rate of this primary oil production.
As reservoir pressure is reduced by oil production, additional recovery
mechanisms may operate. One such mechanism is natural water drive. Water
from an adjacent more highly pressured formation is forced into the oil-bearing
formation by the pressure differential between the two formations. Another
mechanism is gas drive. Expansion of a gas cap above the oil as oil pressure
declines can drive additional oil to the wellbore. Produced gas may be reinjected
to maintain gas cap pressure as is done on the Alaskan North Slope. Additional
oil may also be produced by compaction of the reservoir rock as oil production
reduces reservoir pressure.
As the natural pressures in the reservoir decrease, oil production declines.
A pump is then used to maintain oil production at economic levels by drawing oil
to the surface and lowering the height of the fluid column in the wellbore. The
pressure of a column of fluid can decrease the rate of fluid entry into the wellbore.
Primary production typically recovers 10–25% of the oil originally in the
reservoir. Efficiency of primary production is related to oil properties, reservoir
properties, geometric placement of oil wells, and the drilling and completion
technology used to drill the wells and prepare them for production. Pumping
the well can maintain production at economic levels for years.
2.1. Waterflooding. Injection wells are used when the natural pres-
sures driving fluids to production wells are depleted and pumping alone is no
longer economical. Fluid injection repressurizes the reservoir, restoring a driving
force and promoting oil production. For economic reasons, water is the usual
injection fluid. Water injection or waterflooding is usually termed secondary oil
recovery. It accounts for 40% of total U.S. oil production. Additional oil recovery
by waterflooding is typically 15–25% of the oil originally in the reservoir.
Determining and using the optimum pattern and arrangement of production
and injection wells for a particular reservoir can have a significant effect on oil
recovery and production rates (8). Infill drilling and horizontal production wells
can drain oil reservoirs more efficiently (9). Horizontal injection wells have also
improved oil recovery by increasing volumetric sweep efficiency and increasing
fluid injection rates (10). One 600–1200-m long horizontal well can replace several
vertical wells decreasing both overall drilling and operational costs (11).
Vol. 1 ENHANCED OIL RECOVERY 169

3. Oil Recovery Mechanisms

There are two principal mechanisms of enhanced oil recovery: increasing volu-
metric sweep efficiency of the injected fluid and increasing oil displacement effi-
ciency by the injected fluid. In both, chemicals are used to modify the properties
of an injected fluid whether water, steam, a miscible gas, eg, CO2 or natural gas,
or an immiscible gas, usually nitrogen. Poor reservoir volumetric sweep effi-
ciency is the greatest obstacle to increasing oil recovery (12).
Wettability is defined as the tendency of one fluid to spread on or adhere to
a solid surface (rock) in the presence of other immiscible fluids (8). As many as
50% of all sandstone reservoirs and 80% of all carbonate reservoirs are oil-wet
(13). Strongly water-wet reservoirs are quite rare (14). Rock wettability can
affect its relative permeability to water and oil and thus influence fluid injection
rates, flow patterns of fluids within the reservoir, and oil displacement efficiency
(8,14,15). Alteration of rock wettability by adsorption of polar materials, eg, sur-
factants and corrosion inhibitors, or by the deposition of polar crude oil compo-
nents (16), can strongly alter the behavior of the rock (15).
When water is injected into a water-wet reservoir, oil is displaced ahead of
the injected fluid. Injection water preferentially invades the small- and medium-
sized flow channels or pores. As the water front passes, unrecovered oil is left in
the form of spherical, unconnected droplets in the center of pores or globules of
oil extending through interconnected rock pores. In both cases, the oil is comple-
tely surrounded by water and is immobile. There is little oil production after
injection water breakthrough at the production well (8).
In an oil-wet rock, water resides in the larger pores, oil exists in the smaller
pores or as a film on flow channel surfaces. Injected water preferentially flows
through the larger pores and only slowly invades the smaller flow channels
resulting in a higher produced water/oil ratio and a lower oil production rate
than in the water-wet case.
3.1. Injection Well Considerations. Fluid injection rate can have a
significant effect on oil recovery economics. Flow is radial from the wellbore
into the reservoir. Thus the region near the injection wellbore acts as a choke
for the entire reservoir.
Addition of surfactant to the injection water (17,18) can displace the oil
remaining near the well. The lower oil saturation results in an increase in the
water relative permeability (8). Therefore, a greater water injection rate may
be maintained at a given injection pressure. Whereas ultimate oil recovery
may not be increased, the higher water injection rate can increase oil production
rates improving oil recovery economics. The concentration of surfactant in the
injection water is relatively high (1–3%). However, the total amount of surfac-
tant used is not great because it is necessary only to displace the oil from a
2–3-m radius around the injection well.
Decreased injection rates resulting from formation damage, ie, reduction of
the rock fluid carrying capacity, near injection wells can reduce oil production
rates at offset (adjacent) production wells. Formation damage may result from
invasion of rock capillaries by solid particles in wellbore fluids during well dril-
ling and completion. Plugging of rock capillaries adjacent to fractures by fine
170 ENHANCED OIL RECOVERY Vol. 1

solid particles in fracturing fluids may also occur. Acidizing the rock immediately
adjacent to the wellbore can dissolve clays (qv), silica particles to improve the
rock permeability. However, treatments must be carefully designed to prevent
formation of precipitates that can plug rock flow channels.
Precipitate formation can occur upon contact of injection water ions and
counterions in formation fluids. Solids initially present in the injection fluid, bac-
terial corrosion products, and corrosion products from metal surfaces in the injec-
tion system can all reduce near-wellbore permeability. Injectivity may also be
reduced by bacterial slime that can grow on polymer deposits left in the wellbore
and adjacent rock. Strong oxidizing agents such as hydrogen peroxide, sodium
perborate, and occasionally sodium hypochlorite can be used to remove these bac-
terial deposits (19–21).
Formation damage can also be caused by chemical and physical interactions
of fluid and rock. Low salinity injection fluids are often preferred to obtain max-
imum viscosity from a given amount of water-soluble polymer. However, low sali-
nity fluids can cause swelling of water-expandable clays. This swelling reduces
the fluid-carrying capacity of rock flow channels. Because clays act as the
cementing medium in many sandstone formations, this swelling weakens cemen-
tation and can result in the release of mineral fine particles which can migrate to
constrictions and plug the flow channels.
Long-lasting stabilization of water-swelling clays may be achieved by using
materials, eg, hydroxyaluminum (22) and certain quaternary ammonium salt
polymers (23). Quaternary ammonium salt polymers have been used in drilling
fluids, completion fluids, acidizing treatments, and hydraulic fracturing as well
as in injection water and injected steam for enhanced oil recovery (23). Adsorp-
tion is long lasting and limited chemical treatment volumes are needed. Treat-
ment concentration is usually 0.1–1.0% by weight. For injection water and
injected steam, it is usually necessary to treat a 2–3-m radius from the injection
wellbore. The addition of potassium hydroxide to injection waters has also been
used to stabilize clays and maintain injection rates (24).
3.2. Injection Fluids. Whereas water is the most commonly used injec-
tion fluid, other fluids can provide higher oil recovery efficiency. Injecting gases
miscible with reservoir crude oil can result in low interfacial tension promoting
high oil displacement efficiency (25). The process of miscible gas flooding using
carbon dioxide (qv) is depicted in Fig. 1. The primary source of CO2 is high pres-
sure natural reservoirs of highly pure CO2. Any CO2 produced with the crude oil
is recovered, purified, and reused. However, CO2 produced in industrial opera-
tions, eg, separation from produced natural gas and coal gasification is increas-
ingly being used while use of flue gases from power plants, refinery units, and
other industrial operations is under consideration (26). For example, Norway’s
Sleipner Field project stores 1 million tonnes/year of CO2 associated with natural
gas production in a saline geologic aquifer. Norway’s carbon emission tax of 315
krone/tonne CO2 equivalent ($45/metric ton) encourages this storage (26). Other
suitable gases include natural gas and flue gas.
Overall, CO2 flooding can increase oil recovery by 7–15% of the oil originally
in place and can be sustained for 10–30 years (25). Improvement of CO2 flood per-
formance has resulted from advances in computer simulation of CO2 flood perfor-
mance and greater experience in handling carbon dioxide–water mixtures. Other
171
Fig. 1. Carbon dioxide flooding. The WAG process, in which a CO2 slug is followed by alternate water and CO2 injections, is usually
employed. The viscosity of the oil is reduced providing more efficient miscible displacement. (Courtesy of the U.S. Dept. of Energy).
172 ENHANCED OIL RECOVERY Vol. 1

recently reduced technology that can increase oil recovery includes horizontal
injection and production wells, four-dimensional (4D) seismic analyses to track
injected CO2 flow, automated field monitoring systems, and injecting larger
volumes of CO2 than were used in earlier EOR projects (27).
Many CO2 miscible EOR projects are located in the west Texas Permian
Basin where as much as two-thirds of the oil remains after waterflooding. An
incremental (10%) recovery is typical for Permian Basin CO2 floods, which
could correspond to as much as 0.5  109 m3 (3–4  109 bbl) (28). Pipelines (qv)
connect these large EOR projects to natural CO2 sources in Colorado and New
Mexico. Industrial point sources of CO2 have also been used for projects in
other areas.
Combining an in situ thermal enhanced oil recovery process (see below)
with CO2 injection and subterranean storage has been reported in the patent
literature (29).
The pressure–composition requirement for miscibility limits the oil reser-
voirs in which CO2 enhanced oil recovery has been applied. The low injected
fluid viscosity often results in poor volumetic sweep efficiency (see below).
Despite these concerns, a Department of Energy study indicates that carbon
dioxide injection has the potential of recovering 43 billion barrels of oil currently
‘‘stranded’’ in six mature U.S. oil-producing regions (30).
Supercritical CO2 (25,31) and various hydrocarbon injectants (25,32)
undergo physical interactions with crude oil that result in stripping out of the
low molecular weight components, which increases oil production (see SUPER
CRITICAL FLUIDS). The rapid or gradual development of miscibility with the remain-
ing crude oil constituents results in oil mobilization. Either partial or complete
miscibility with the oil may be developed depending on the nature of the injec-
tant, crude oil properties, and reservoir conditions, particularly temperature.
However, interaction of the injectant with the crude oil can alter rock wettability
and thus reduce injection rates and decrease oil recovery.
Another method of using CO2 is called cyclic CO2 stimulation or huff ‘n’puff
(33). A limited amount of CO2 is injected into a reservoir over hours or days. The
well is then shut in for a soak period of days to weeks to allow the CO2 to interact
with the crude oil, swelling the oil and reducing its viscosity. The well is then
opened, the CO2 provides a solution gas drive, and oil mobilized by the CO2
soak is produced.
3.3. Reducing Greenhouse Gas Emissions. Carbon dioxide EOR
could also play a role in reducing greenhouse gas emissions. Carbon dioxide cap-
ture and storage (CCS) underground could constitute as much as one-half of the
global emissions reduction by 2050, according to the International Energy
Agency (34). Carbon dioxide sequestration can provide two added economic
incentives for CO2 EOR in addition to the value of the additional oil recovered.
First, by reducing CO2 emissions, field operators can avoid being subject to pay-
ing CO2 emissions penalties. Second, they can receive credits for reducing their
CO2 emissions; credits that can be sold to other firms that have high levels of CO2
emissions. While these systems are being deployed in Norway and other
European countries, they are not yet in place in the U.S. Eight commercial CO2
sequestration projects are underway in the Norwegian sector of the North Sea,
Algeria, Alaska and Saskatchewan, Canada. Governments in Canada, Norway
Vol. 1 ENHANCED OIL RECOVERY 173

and other oil-producing areas are providing financial incentives for the under-
ground sequestration of CO2.
CCS involves three distinct processes: capturing CO2 from the gas streams
emitted during electric power production, industrial processes, or fuel proces-
sing; transporting the captured CO2; and storing it underground in rock forma-
tions where it cannot leak to the surface. Using this CO2 to increase oil
production would offset the costs of CCS and perhaps even render it profitable
(after factoring in governmental financial incentives).
3.4. Injection of Other Gases. Nonmiscible gases, eg, nitrogen, have
been used as EOR injection fluids. Increasingly, the nitrogen used is separated
at the oil field from produced natural gas rather than being produced in indus-
trial gases plants. Oil recovery mechanisms include volatilization of low molecu-
lar weight components of the crude oil and displacement of oil from the top of the
reservoir (27). The latter mechanism occurs as a result of gravity override of the
low density injectant.
Gas injection into a gas cap overlaying an oil reservoir is considered
an EOR method. The resulting repressurization of the reservoir promotes addi-
tional oil production. Reinjection of natural gas is responsible for a significant
fraction of Alaskan North Slope oil production. With the construction of a pipe-
line to transport North Slope natural gas to world markets, CO2 may be used
instead.
High temperature steam is also used for recovery of viscous crude oils
(28). Heat from the steam thins the oil, reducing viscosity and increasing mobi-
lity. The mobilized oil is produced at offset production wells. In heavy oil fields,
water flooding is often omitted and steam injection begun immediately after
primary production. Steam injection temperature is typically 175–2308C in
California oil fields. Injection temperature can reach 3008C in Canadian and
Venezuelan EOR projects.
The injection of large volumes of steam, steam flooding, also called steam-
assisted gravity drainage (SAGD), is used to mobilize oil, which is produced at
offset production wells. By reducing oil viscosity, the oil flows more readily
through the reservoir to production well bores. Smaller volumes of steam are
injected in the cyclic steam stimulation or huff ‘n’puff process (Fig. 2). Many
wells are placed on several cycles of steam stimulation and then used as injection
or production wells in steam flood projects. Large-scale steam projects can be
quite expensive. For example, a steamflood project in Oman’s Mukhaizna
heavy oil field is projected to cost > $2 billion and involve drilling 1800–2200
new wells (35). Production is expected to increase from the current 10,000 to
150,000 bbl/day.
An organic chemical may be added to the steam to provide additional oil
viscosity reduction in addition to the heat-thinning effect provided by injected
steam. For example, after the oil recovery rate in a steam flood reaches 25–75%
of the predicted maximum recovery rate, a hydrocarbon solvent may be injected
to recover additional heavy oil from an oil reservoir (36). Alternatively, steam or
hot water and solvent may be injected simultaneously in a ratio of > 5 : 1 (37). The
C1–5 alcohols have been used as additives to high temperature steam and injected
into heavy oil reservoirs in such a manner that the injectant does not form an
emulsion with the crude oil (38).
174
Fig. 2. Cyclic steam stimulation of an oil well: (a) steam, injected into a well over a period of days or weeks in a heavy oil reservoir, in-
troduces heat (huff) that, coupled with (b), alternate soak periods lasting a few days to allow (c) a production phase of weeks or months
(puff), thins the oil. This process may be repeated until production falls below a profitable level. (Courtesy of the U.S. Dept. of Energy).
Vol. 1 ENHANCED OIL RECOVERY 175

3.5. Heavy Oil Upgrading. Upgrading of heavy oil prior to conventional


refining is increasingly being used to improve SAGD economics. For example,
Iraq’s Qaiyarah Field produces heavy oil that cannot be refined in conventional
refineries and is used mainly for asphalt (39). However, upgrading processes
have been developed by several oil companies that convert heavy crude oils to
light oils that can be processed in conventional refineries. Such an upgrading
plant will be build to process the Qaiyarah heavy oil and several plants are
already operating in Canada and Venezuela to process the bitumen and very
heavy oil recovered in SAGD and oil mining projects. Syngas has been used to
initiate predominantly gas-phase heavy oil upgrading reactions (40). There is
also increased interest in the catalytic reaction of very heavy crude oil and
resid, the bottoms product remaining after vacuum distillation in a conventional
refinery, with steam to produce methane (41).
Bitumen mining in Canada using open-pit mines has reached commercial
status. Upgrading of this often solid-appearing material is required before it
can be processed in conventional refineries.
3.6. Improving Volumetric Sweep Efficiency. Volumetric sweep effi-
ciency is determined by the permeability and wettability distribution in the
reservoir and by the properties of injected fluids. High permeability rock streaks
or layers (thief zones) and natural or induced rock fractures can channel the
injected fluid through a small portion of the reservoir, resulting in a low rock
volumetric sweep efficiency. Low viscosity injection fluids exhibit poor volumetric
sweep efficiency, which leads to low oil production. Thus, proper diagnosis of the
cause of poor volumetric sweep efficiency is critical in designing a successful well
treatment. For example, sealing fractures requires different well treatment
designs than reducing the permeability of thief zones.
Both sodium silicate gelation (42) and in situ cross-linking of organic poly-
mers (43,44) can reduce the permeability of fractures and high permeability
streaks. Polymers are usually injected at concentrations of 1000–5000 ppm.
In situ cross-linking treatments are restricted to fractures and the near-wellbore
region owing to the kinetics of the cross-linking process. The polymer may be
injected into the well with a cross-linker or the cross-linker may be injected
after the polymer. The well is shut in for from 1–7 days and then normal injection
operations are resumed.
The most commonly used polymers are partially hydrolyzed polyacryl-
amides (45). The optimum degree of hydrolysis depends on the application, injec-
tion water composition, and reservoir conditions (46,47). More salt-tolerant
acrylamide copolymers may permit using this technology in higher salinity injec-
tion water (48). Field applications of cross-linked xanthan gum have also been
reported (49).
Chromium(III), Cr(III), compounds have largely replaced Al(III) compounds
as cross-linkers (48–52). The Cr(III) acetate [1066-30-4] cross-links acrylamide
polymers rapidly. The Cr(III) complexes composed of strong ligands, eg, glycolate
or malonate give extended polyacrylamide gelation times compared to salts such
as Cr(III) acetate (53). Delaying cross-linking permits the use of this technology
at higher (up to 1508C) reservoir temperatures. Sodium bisulfite and thiourea
have been used to reduce injected Cr(VI) to the reactive Cr(III) species that
promotes cross-linking (54). Gradual dissolution of colloidal Cr(OH)3 can also
176 ENHANCED OIL RECOVERY Vol. 1

delay cross-linking (55) as can Cr(III) propionate (56). Encapsulation of the cross-
linking agent permits deeper penetration of gel-forming compositions into oil-
bearing formations (57). Injection of unhydrolyzed polyacrylamide followed by
in situ hydrolysis also delays cross-linking (58).
Cross-linked xanthan gums have also been used to reduce the permeability
of thief zones. Trivalent chromium is the preferred cross-linker (59). Cross-linker
effectiveness is less at high salinity. However, Cr(III) has been used in the field
at salinities as great as 166,000 ppm total dissolved solids (60).
Proper placement of the treatment fluid in the reservoir is critical to treat-
ment success. Careful sizing of the treatment and choice of injection rates are
required because overtreatment can cause plugging of the oil-containing rock
and excessive reduction of the injection rate. Even after a well-designed treat-
ment, the fluid injection rate is often significantly less than before well treat-
ment. Many successful applications of this technology in waterfloods and in
surfactant polymer floods have been reported. Wells in CO2 EOR projects have
also been treated using this technology.
Polymerization may also occur in situ. Reactive monomers, eg, acrylamide
in concentrations of 2–5 wt% and various additives including a free-radical poly-
merization initiator may be used (61,62). A difunctional monomer, eg, N,N0 -
methylenebis(acrylamide), can be added to the injected formulation to form a
cross-linked polymer in situ. Low viscosity aqueous monomer solutions can be
injected at higher rates and preferentially enter high permeability zones to a
greater extent than do non-Newtonian polymer solutions (63). If no difunctional
monomer is used, the viscous polymer mass may be slowly dissolved by injection
water increasing water viscosity and providing a second means of increasing oil
recovery.
Lignosulfonates may be cross-linked in situ using Cr(III) (64) or an acidic
gas, eg, CO2 (65). Cross-linked lignosulfonate can be an effective plugging
agent at high temperatures, such as in steam injection wells. Lignosulfonate con-
centration is usually 2–3 wt%. Blends of lignosulfonate and sodium silicate have
also been used (66). Other systems cross-linked in situ for water or stream
injection wells are phenol–formaldehyde (67), urea–formaldehyde (68), furfuryl
alcohol (69), formaldehyde resin plus sulfonated tannin extract (70), and formal-
dehyde resin and alkali kraft lignin (71).
Using foamed gels can reduce the overall amount of chemicals required to
form a given volume of gel within a subterranean formation. For example, a
foamed gel formed from a cross-linkable carboxylate-containing polymer, a reac-
tive transition metal cross-linking agent, polyvinyl alcohol, an aqueous solvent
and an added gas has been described (72).
Surfactant precipitation may be used for in-depth permeability reduction of
thief zones (73). Thief zones have a low oil saturation owing to the preferential
flow of injected fluids through high permeability rock. This process is based on
the sequential injection of a slowly propagating ionic surfactant followed by an
aqueous spacer containing no surfactant. Then a more rapidly propagating
ionic surfactant of the opposite charge type is injected. The oppositely charged
surfactants gradually mix in the high permeability portions of the reservoir
(thief zones) causing precipitation, which plugs flow channels reducing perme-
ability in the most flooded portions of the reservoir thereby diverting injectant
Vol. 1 ENHANCED OIL RECOVERY 177

to rock zones containing higher oil saturations. The economically limiting fac-
tors in the use of this process are cost and low propagation rate of the cationic
surfactant.
Rock surface chemistry can affect volumetric treatment effectiveness and
economics. Metal ion cross-linkers and cationic polymers can adsorb on mineral
surfaces, particularly clays, by ion-exchange (qv) processes. The subsequently
lower concentration of materials in solution decreases treatment effectiveness
by reducing the rate of polymer cross-linking and decreasing gel strength.
Both in situ cross-linking of partially hydrolyzed polyacrylamides (74) and
injection of quaternary ammonium salt polymers having long hydrophilic side
chains (75) have been used to reduce the permeability of water-producing
zones adjacent to production wells. This permeability reduction decreases the
produced water/oil ratio as does injection of polyacrylamide in high hardness
brine to reduce rock permeability (76). The polymers exist in saline solution as
tightly coiled chains and are readily adsorbed owing to relatively low solubility in
hard water. Subsequent injection of soft, low salinity water uncoils the adsorbed
polymer chains increasing water viscosity and reducing rock permeability. This
technology could also be used to reduce the permeability of thief zones adjacent
to injection wells. However, mechanical isolation of these zones may be necessary
for cost-effective treatments.
3.7. Polymer Flooding. Even in the absence of fractures and thief
zones, the volumetric sweep efficiency of injected fluids can be quite low. The
poor volumetric sweep efficiency exhibited in waterfloods is related to the mobi-
lity ratio, M, the mobility of the injected water in the highly flooded (low oil
saturation) rock, mw, divided by the mobility of the oil in oil-bearing portions
of the reservoir, mo (77,78). The mobility ratio is related to the rock permeability
to oil, kro, and injected water, krw, and to the viscosity of these fluids by the
following equation:

M ¼ mw =mo ¼ ðkrw =Zw Þ=ðkro =Zo Þ

The terms hw and ho represent the viscosity of the aqueous and oil phases,
respectively.
The polymer flooding process is depicted in Fig. 3. The displacing or driving
fluid may be steam, supercritical carbon dioxide, hydrocarbon miscible gases,
nitrogen, or solutions of surfactants or polymers instead of water. The volumetric
sweep efficiency increases with lower mobility ratio values (79). A mobility ratio
of 1.0 or lower is considered optimum. The mobility of water is often high relative
to that of oil. Steam and oil-miscible gases, eg, supercritical carbon dioxide exhi-
bit even higher mobility ratios. Consequently, these more expensive injectants
can have low volumetric sweep efficiencies.
Mobility control agents reduce the mobility ratio. Virtually all polymer flood
field projects have used hydrolyzed polyacrylamides having molecular weights
of 1–5  106 or xanthan gum, a biopolymer (80). Variations in polymer molecular
weight and structure have been made to improve performance properties.
Relatively low (100 ppm for fresh water, 1000 ppm or more for saline systems)
polymer concentrations can significantly increase injected water viscosity.
178
Fig. 3. This polymer flooding method requires a preflush to condition the reservoir, the injection of a polymer solution for mobility control
to minimize channeling, and a driving fluid (water) to move the polymer solution and resulting oil bank to production wells. (Courtesy of the
U.S. Dept. of Energy).
Vol. 1 ENHANCED OIL RECOVERY 179

Table 1. Properties of EOR Polymersa


Property Polyacrylamide Xanthan gum
brine tolerance very limited, especially good to both mono- and
to Ca2þ, Mg2þ divalent cations
shear stability undergoes irreversible reversible shear thinning
shear degradation
maximum use 71–82b 71–77
temperature, 8C
hydrolytic stability hydrolysis promoted by hydrolytic depolymeriza-
acid or base; partially tion promoted by acid or
hydrolyzed product base especially at high
more sensitive to temperatures
Ca2þ, Mg2þ
oxidative stability susceptible particularly susceptible
especially at high
temperatures
microbial degradation susceptible very susceptible
a
Ref. 75.
b
In very low salinity reservoirs temperatures can go to 107–1218C.

Adsorption of these polymers on rock can result in a long-term decrease in rock


permeability to aqueous fluids (residual resistance). Some polymer field projects
have exhibited injected water permeability reductions, attributed to residual
resistance effects, that have lasted for more than three years after polymer
injection (81).
Each EOR polymer type has important advantages and significant disad-
vantages (Table 1). When dissolved in more saline waters, xanthan gum pro-
duces a higher apparent viscosity than the same concentration of partially
hydrolyzed polyacrylamide (82). Xanthan gum is more soluble in saline waters
than are polyacrylamides, particularly in injection waters containing divalent
metal ions. Xanthan gum also generally adsorbs less on rock surfaces and
is substantially more resistant to shear degradation than polyacrylamides
(83). However, xanthan gum is also more expensive and the extensional viscos-
ity of the semirigid xanthan molecule is less than that of the flexible polyacry-
lamide (84). Both polymers cross-link easily in the presence of transition
metals.
In addition to the normal problems of completely dissolving particles of
water-thickening polymers, xanthan gum contains insoluble residues that
decrease polymer injectivity. Various methods of reducing insolubles content
and improving xanthan solution injectivity are available (86–92). None appears
economically viable. Oxygen scavengers (93) and bactericides (85,94) are com-
monly used to stabilize injected polyacrylamide and xanthan gum solutions
(95–108).
At low polymer concentrations, xanthan containing the intact pyruvate ring
exhibits higher brine solution viscosity and better filterability than its ring-
opened analogue (109). A xanthan gum containing pyruvate rings in most of the
polymer repeat units has been produced by a proprietary strain of Xanthamonas
campestris (102). Genetic engineering techniques were used to obtain mutant
X. campestris bacterial strains that provided genetic control of the extent of
180 ENHANCED OIL RECOVERY Vol. 1

acetylation and pyruvation of xanthan polymers that provided improved per-


formance in polymer flooding (110). Xanthamonas bacteria have also been used
to produce a polymer, having a glucose and mannose unit in a 2 : 1 ratio, which is
claimed to be a better water viscosifier than xanthan gum (111). Another promis-
ing microbial polysaccharide is scleroglucan (112), which has been evaluated for
North Sea applications. The high temperature behavior of different microbial
polysaccharides has been studied (113). In highly saline media, another biopoly-
mer, succinoglycan, more readily flows through microporous media than does
xanthan gum (113). This glycan forms higher viscosity solutions than does
xanthan gum at equal concentrations.
Most polyacrylamides used as mobility control agents are partially hydro-
lyzed or are acrylamide–acrylic acid (or sodium acrylate) copolymers produced
by emulsion copolymerization (114). Emulsion polymers are used to avoid high
shear degradation and undissolved solid particle problems. Another method of
avoiding these problems is acrylamide solution polymerization at the wellhead.
The polymerization can be designed to proceed at adequate rates and in saline
injection waters to provide polymers of adequate viscosity characteristics (115).
Polyarylamide is usually hydrolyzed in base to produce a random distribution of
acrylate groups (116). Acid hydrolysis results in a more block-like distribution of
acrylate units (117). Electrostatic repulsion of the anionic carboxylate groups
elongates the polymer chain of partially hydrolyzed polyacrylamides, increasing
the hydrodynamic volume and solution viscosity. Cobalt-60, 60Co, irradiation has
been used to initiate polymerization and prepare particularly high molecular
weight polyacrylamides (118).
Maximum freshwater viscosity of polyacrylamide occurs at 35% hydroly-
sis; maximum viscosity in a Ca2þ-containing brine occurs at 10–15% hydrolysis
(119). Metal ions interact with carboxylate groups reducing their mutual repul-
sion and thus decreasing hydrodynamic volume and solution viscosity. Divalent
metal ions reduce viscosity more than monovalent ones (120). Chelating and
sequestering agents have been proposed to reduce the adverse effect of divalent
(119) and multivalent metal ions on polyacrylamide solution viscosity (120,121).
Proper well completion, particularly perforation design, reduces polyacrylamide
shear degradation during injection (122).
Excessive hydrolysis of polyacrylamide in situ can promote undesirable
polymer precipitation in the reservoir. Acrylamide block and random copolymers
designed to reduce undesired amide group hydrolysis, increase thermal stability,
and improve solubility in saline media have been studied for EOR applications
(123–132).
Acrylamide graft copolymers such as those with starch (133), dextran (134),
and lignin (135), have been studied to try to reduce copolymer costs. A general
disadvantage of acrylamide copolymers is greater cost compared to partially
hydrolyzed polyacrylamides.
Among the other types of polymers showing promise for polymer flooding
are hydrophilic N-vinyl lactams (136).
Propagation of enhanced oil recovery chemicals through rock is critical to
the success of an EOR project. Mechanical entrapment of polymers as well as
adsorption can reduce the effective polymer concentration (137,138). Calcium
ions in formation and injection waters increase anionic polymer adsorption
Vol. 1 ENHANCED OIL RECOVERY 181

(139). Sacrificial adsorption agents such as lignosulfonates (141–143) can be used


to reduce the adsorption of more expensive polymers and surfactants. Other che-
micals tested include poly(vinyl alcohol) (144), sulfonated poly(vinyl alcohol)
(145), sulfonated poly(vinylpyrrolidinone) (146), low molecular weight polyacry-
lates (147), and sodium carbonate (148).
3.8. Surfactants for Mobility Control. Water, which can have a mobi-
lity up to 10 times that of oil, has been used to decrease the mobility of gases and
supercritical CO2 (mobility on the order of 50 times that of oil) used in miscible
flooding. Gas/oil mobility ratios, M, can be calculated by the following (26):
  
M¼ kg =ms þ ðkw =mw Þ =½ðko =mo Þ þ ðkw =mw Þ

where k refers to permeability, m to viscosity, and the subscripts g, s, o, and w to


gas, miscible solvent, oil, and water, respectively. The water may be injected
simultaneously with the gas or in alternate slugs with the gas (WAG process).
X-ray computerized tomography of core floods has demonstrated the increased
volumetric sweep efficiency attained in the WAG process (149) compared to
injection of CO2 alone. The design parameters most affecting WAG CO2
flood oil recovery are CO2 and water slug sizes, produced gas:oil ratio as a func-
tion of time, and total volume of injected CO2 (150) (see Fig. 1). Injecting water
saturated with pressurized CO2 and relieving the pressure after injection results
in gas bubble formation in the smallest rock pores forcing oil into more permeable
portions of the formation thereby increasing oil recovery. This strategy can
increase the oil recovery beyond that typically reached in WAG processes (151).
The WAG process has been used extensively in the field, particularly in
supercritical CO2 injection, with considerable success (25,150,152). However, a
method to further increase the viscosity of injected gas or supercritical fluid is
desirable. One means of increasing the viscosity of CO2 is through the use of
supercritical CO2-soluble polymers and other additives (147). The use of surfac-
tants to form low mobility foams or supercritical CO2 dispersions within the for-
mation has received more attention (153–155). Foam has also been used to
reduce mobility of hydrocarbon gases and nitrogen. X-ray computerized tomo-
graphic analysis of core floods indicate that addition of 500 ppm of an alcohol
ethoxyglycerylsulfonate increased volumetric sweep efficiency substantially
over that obtained in a WAG process (150).
One reason for widespread interest in the use of surfactants as gas mobility
control agents is the effectiveness at concentrations of
0.1 wt% (149,156). Some
surfactants are effective below their critical micelle concentration (157). This low
chemical requirement can significantly improve process economics.
Among the classes of surfactants studied for this application are alcohol
ethoxylates and their sulfate and sulfonate (150,158–161) and carboxylate
(162) derivatives, alkylphenol ethoxylates (163), alpha-olefin sulfonates (162),
and alkylated diphenylether disulfonates (164). Increased linear carbon chain
length, decreased branching, and increased ethoxy group chain length increase
foam stability (158). Addition of a water-thickening polymer to the aqueous
phase may stabilize the foam (165). Using a blend of an alkaline agent, a
surfactant and a polymer as the supercritical CO2 foaming agent has been
proposed (166).
182 ENHANCED OIL RECOVERY Vol. 1

In addition to the mobility control characteristics of surfactants, critical


issues in gas mobility control processes are surfactant salinity tolerance, hydro-
lytic stability under reservoir conditions, surfactant propagation through the
reservoir, and foam stability in the presence of crude oil saturations. Lignosulfo-
nate has been reported to increase foam stability and function as a sacrificial
adsorption agent (167). Addition of sodium carbonate or sodium bicarbonate to
the surfactant solution reduces surfactant adsorption by increasing the aqueous-
phase pH (168).
Alcohol ethoxysulfates have been used in field tests as nitrogen (169) and
carbon dioxide (170) foaming agents. Field use of alcohol ethoxysulfates is
restricted to low temperature formations owing to its limited hydrolytic stability
at low pH and elevated temperature (171). It has been reported that some foams
can reduce residual oil saturation, not by oil displacement, but by emulsification
and imbibition of the oil into the foam (172).
Gravity override of low density steam leads to poor volumetric sweep effi-
ciency and low oil recovery in steam floods. Nonchemical methods of improving
steam volumetric sweep efficiency include completing the injection well so steam
is only injected in the lower part of the oil-bearing zone (173), alternating the
injection of water and steam (174), and horizontal steam injection wells
(175,176). Surfactants frequently are used as steam mobility control agents to
reduce gravity override (177). Field-proven surfactants include C16–18 alpha-
olefin sulfonates (AOS), alkyltoluene sulfonates, and neutralized dimerized
alpha-olefin sulfonic acid.
Addition of long-chain (C8–20) alcohols to AOS or alkylaromatic sulfonates
increases foam strength and permits the use of lower surfactant concentra-
tions (178). Increasing the carbon number in alpha-olefin sulfonates to > 25 in-
creases foam strength (179,180). In alkylaromatic sulfonates, longer linear alkyl
groups (181,182) or dialkyl substitution (183) has the same effect. Other alky-
laromatic sulfonates containing benzene, toluene, or xylene rings (182,184),
two fused aromatic groups (185), and the diarylether group (186) have been
evaluated favorably as steam-foaming agents. The neutralized dimer of an
alpha-olefin sulfonate has also been used (187).
To maintain foam strength as the steam cools and eventually condenses, a
noncondensible gas, usually nitrogen or methane, is often added to the injectant
composition (188).
Critical parameters affecting surfactant performance are surfactant propa-
gation rate and surfactant stability at steam temperatures that can reach
> 3168C. Surfactant propagation rate can be reduced by adsorption, precipita-
tion, and partitioning into the oil phase. Anionic surfactant adsorption increases
with increasing salinity and decreases with increasing temperature (189).
Additives can improve surfactant propagation. Both anionic surfactant par-
titioning and precipitation increase with increasing calcium ion concentration
(190) so minimizing divalent metal ion concentration in the surfactant solution
is desirable. Injection of a surfactant preslug containing NaCl converts clays
from the calcium to the sodium form and reduces later ion-exchange processes
that add Ca2þ ions to the surfactant solution (191,192). The use of a hydrotrope
such as sodium xylene sulfonate has been reported to increase oil recovery in
laboratory steam-foam flood tests (193). Hydrotropes are additives that increase
Vol. 1 ENHANCED OIL RECOVERY 183

surfactant solubility. They also may function as sacrificial adsorption agents or


act as foam stabilization agents.
Steam-foaming agents that efficiently mobilize heavy crude oil by heat
transfer can reduce the residual oil saturation. This can increase foam stability
and improve the diversion of subsequently injected steam into oil saturated
zones thereby increasing oil recovery (194).
Thermal stability of the foaming agent in the presence of high temperature
steam is essential. Alkylaromatic sulfonates possess superior chemical stability
at elevated temperatures (195,196). However, alpha-olefin sulfonates have suffi-
cient chemical stability to permit their use at steam temperatures characteristic
of most U.S. steamflood operations.
The addition of high pH agents, eg, sodium hydroxide to the surfactant
solution has been reported to increase foam strength, stability and surfactant
propagation (197–199). These additives can also maintain the pH at a high
enough value to reduce the rate of surfactant decomposition. In addition, the
added base may interact with organic acids naturally found in the crude oil.
The resulting soap generation provides surfactant to more efficiently displace
oil (197). The consequent lower oil saturation can result in a more stable
foam.
Water-soluble polymers can increase the viscosity of the foam external
phase. This improves foam stability and reduces mobility. Gelation of the foam
external phase can reduce chemical requirements to plug thief zones and frac-
tures (200).
3.9. Improving Oil Displacement Efficiency. The use of relatively
large (2–5 wt%) concentrations of surfactants to increase oil displacement effi-
ciency has been studied extensively (35,201,202). This method, called the micel-
lar flooding or surfactant–polymer flooding, usually involves the injection of a
brine preflush to adjust reservoir salinity. The preflush is followed by injection
of a micellar slug comprised of the surfactant, a cosurfactant (usually a C4–6
alcohol), and a hydrocarbon. A polymer solution is then injected to reduce viscous
fingering of the drive fluid into and through the micellar slug. Viscous fingering
causes dilution of the surfactant, reduced contact of the micellar slug with the
crude oil, and trapping of some of the micellar slug in the reservoir. These effects
reduce oil recovery. A freshwater buffer to protect the polymer follows, prior to
injection of the driving fluid, ie, saline water (an oilfield brine), to move the
chemicals and the resulting oil bank to the well.
Process effectiveness depends on maintaining an ultralow [ 10 10 N/m
3
(10 dynes/cm)] interfacial tension between the injected surfactant slug and
the crude oil (203). The effect of petroleum composition on oil solubilization by
surfactants has been the subject of extensive study (204). Other critical surfac-
tant issues are performance in saline injection waters, adsorption on reservoir
rock, partitioning into reservoir crude oil, chemical stability in the reservoir,
interactions with the mobility control polymer, and production problems caused
by resultant emulsions. Reservoir heterogeneity can also greatly reduce process
effectiveness. Micellar processes remain relatively high cost relative to other
EOR technologies.
Besides crude oil petroleum sulfonates, nonionic surfactants, eg, alcohol
ethoxylates, alkylphenol ethoxylates (205) and propoxylates (206), and alcohol
184 ENHANCED OIL RECOVERY Vol. 1

propoxylates (206) have been evaluated for this application. More recently,
anionic surfactants have been used (207–220). Alpha-olefin sulfonates (AOS)
have been found to possess good salt tolerance, chemical stability at elevated
temperatures, and good oil solubilization (210,221), whereas less salt tolerant
alkylaromatic sulfonates exhibit excellent chemical stability. The nature of
the alkyl group, the aryl group, and the aromatic ring isomer distribution can
be adjusted to improve surfactant performance under a given set of reservoir
conditions (222,223).
Surfactant blends can provide improve cost effectiveness compared to use of
a single surfactant. For example, the use of a blend of a synthetic polyisobutylene
with a sulfonate surfactant, alcohols or nonionic surfactants has been described
(224). Cosurfactant requirements can be minimized using a surfactant having a
short-branched hydrophobe or a branched-alkyl substituent on an aromatic
group (222,225) and a long ethoxy group chain (225). Blends of surfactants
optimized for seawater or reservoir brine salinity include linear alkylxylene
sulfonate–alcohol ether sulfate mixtures (226).
High (1–10%) concentrations of lignosulfonate have sufficient interfacial
activity to increase oil recovery from unconsolidated sands (227). Lignosulfonates
and petroleum sulfonates undergo a synergistic interaction resulting in ultralow
interfacial tension and substantially increased oil recovery (228).
The effect of temperature, pressure, and oil composition on oil recovery
efficiency have all been the subjects of intensive study (229). Surfactant propaga-
tion is a critical factor in determining the EOR process economics (230). Surfactant
retention owing to partitioning into residual crude oil can be significant compared
to adsorption and reduce surfactant propagation rate appreciably (231).
Various low cost sacrificial agents decrease surfactant adsorption on reser-
voir rock and increase the surfactant propagation rate. These agents include lig-
nosulfonates and chemically modified lignosulfonates (6,80,143). Alkaline
chemicals (198,199), particularly sodium silicate (232), which precipitate in
the presence of divalent metal ions, can increase the surfactant propagation
rate. Intermixing of polymer mobility control fluid with a previously injected sur-
factant slug can result in undesirable surfactant–polymer interactions affecting
interfacial behavior and reducing oil displacement efficiency (233).
3.10. Alkaline Flooding. Alkaline or caustic flooding involves injection
of high pH agents, eg, sodium hydroxide, sodium carbonate, or sodium silicate
solutions. At equivalent Na2O levels, the three alkaline agents gave equivalent
recovery of each of nine different crude oils in laboratory core floods (234). How-
ever, the use of buffered sodium carbonate rather than strong alkali can result in
reduced interaction with mineral surfaces. The lower reagent consumption can
reduce the amount of sodium carbonate required.
These chemicals generate surfactants in situ by reacting with organic acids
present in crude oil (232,235). Several oil recovery mechanisms may be operative.
Probably the most significant oil recovery mechanism for this process is lowering
of the capillary number (the ratio of viscous to capillary forces) through interfa-
cial tension reduction. Other possible mechanisms are altering rock wettability
(usually from oil-wet to water-wet), oil emulsification and entrapment resulting
in lower injectant mobility (in turn resulting in a greater injectant volumetric
sweep efficiency), oil emulsification and entrainment in the flowing aqueous
Vol. 1 ENHANCED OIL RECOVERY 185

phase, and possibly the solubilization of rigid films that may form at the oil–
water interface.
Caustic flooding chemicals are relatively inexpensive. However, project eco-
nomics are adversely affected by the large quantities that must be injected. The
high pH agents react with reservoir clays (236) and are precipitated by divalent
metal ions present in formation waters. Coinjection of a lignosulfonate (237) or a
polyacrylate (238) reduces precipitation. This precipitation has been used to
advantage to reduce adverse surfactant and polymer interactions with dissolved
divalent metal ions. Injecting a caustic preflush causes divalent metal ion
precipitation before beginning a micellar polymer flood (239). Ion-exchange pro-
cesses promoting solubilization of divalent metal ions limit the effectiveness of
preflushes injected before the caustic solution (240).
Including a surfactant in the caustic formulation (surfactant-enhanced
alkaline flooding) can increase optimal salinity of a saline alkaline formulation.
This can reduce interfacial tension and increase oil recovery (239,241,242). Both
nonionic and anionic surfactants have been evaluated in this application
(243,244). Encouraging field test results have been reported (245). Current
(2005) operating costs in a waterflood in Oklahoma’s Delaware-Childers Field
are $20–22 per barrel of recovered oil (245). An alkaline-surfactant-polymer
waterflood is being initiated that will increase the produced oil from 0.3 to 5%
of total produced fluids. At the estimated oil peak production of 4300 barrels/
day expected in 2007, the higher oil cut and reduced water pumping costs should
reduced operating costs to an estimated $10 per barrel of produced oil.
Surfactants evaluated in surfactant-enhanced alkaline flooding include
internal olefin sulfonates (244), linear alkylxylene sulfonates (246), petroleum
sulfonates (246), alcohol ethoxysulfates (242,247,248), and alcohol ethoxylates/
anionic surfactants (241). Water-thickening polymers, either xanthan or polya-
crylamide, can reduce injected fluid mobility in alkaline flooding (249) and
surfactant-enhanced alkaline flooding (244,248). Cross-linked polymers have
been used to increase volumetric sweep efficiency of surfactant–polymer–alkaline
agent formulations (250). The combined use of alkali, surfactant, and water-
thickening polymer has been termed the alkali–surfactant–polymer (ASP)
process.

4. Other Technologies

Microbial-enhanced oil recovery involves injection of carefully chosen microbes.


Subsequent injection of a nutrient is sometimes employed to promote bacterial
growth. Molasses is the nutrient of choice owing to its relatively low ($100/t)
cost. In its absence, the main nutrient source for the microbes is often the
crude oil in the reservoir. A rapidly growing microbe population can reduce the
permeability of thief zones improving volumetric sweep efficiency. Microbes, par-
ticularly species of Clostridium and Bacillus, have also been used to produce sur-
factants, alcohols, solvents, and gases in situ (251). These chemicals improve
waterflood oil displacement efficiency (see also BIOREMEDIATION).
Microbes adsorb and grow on reservoir rock surfaces fed by injected nutrients
(252) and may have application in plugging thief zones near injection well bores.
186 ENHANCED OIL RECOVERY Vol. 1

However, there is concern that the microbes can also enter lower permeability
zones containing higher oil saturations thereby reducing oil production. Control-
ling the rate and location of bacterial growth and chemical production can be dif-
ficult. Bacterial growth near wellbores has been a common problem causing
reduced injection rates and productivity. However, field test results have been pro-
mising (253). Variations of this technology continue to be patented (254).
The in situ combustion method of enhanced oil recovery through air injec-
tion (32,255,256) is a chemically complex process. There are three types of in situ
combustion: dry, reverse, and wet. In the first, air injection results in ignition of
crude oil and continued air injection moves the combustion front toward produc-
tion wells. Temperatures can reach 300–6508C. Ahead of the combustion front is
a 90–1808C steam zone, the temperature of which depends on pressure in the oil
reservoir. Zones of hot water, hydrocarbon gases, and finally oil propagate ahead
of the steam zone to the production well.
The oil zone is fairly cool, and in a viscous oil reservoir this can result in
little oil movement (liquid blocking). Reverse combustion, in which oil ignition
occurs near the production well, can avoid this problem. The combustion zone
moves countercurrent to the flow of air from the injection well. Oil flows through
heated rock and remains mobile. Reverse combustion requires more air and con-
sumes more oil than forward combustion.
In wet combustion, water is injected concurrently and alternately with air,
extending the steam zone and aiding heat transfer to the crude oil reducing oil
viscosity. This can decrease the injected air/produced oil ratio and improve pro-
ject economics.
Maintenance and propagation of the combustion front are problems. This
has led to a near-wellbore technology in which the same well is used for air injec-
tion and oil production. The combustion front needs to be propagated for a rela-
tively short distance (257).

5. Economic Aspects

World EOR production in 2000 averaged 199,000 m3/day (1.25 million b/day) the
world’s 2000 oil production (258). In 2004, U.S. EOR production was 105,740 m3/
day (663,451 bbl/day) (259). Due to the rapidly fluctuating price of oil it is difficult
to calculate the current value of this level of EOR production.
U.S. 2004 CO2 EOR production of crude oil was 205,877 b/day (260). The
price of crude oil must exceed US$25/bbl before CO2 injection becomes econom-
ical (261). A CO2 price sensitivity analysis indicates CO2 must be priced at less
that US$0.50 per thousand cubic feet before CO2 injection becomes more econom-
ical than conventional production. Incentives for reduction of CO2 emissions if
industrially generated CO2 is used would increase the CO2 price that could be
tolerated while achieving economic viability.
Injection of hydrocarbons accounted for 97,300 b/day of U.S. 2004 oil pro-
duction (260). The amount of crude oil produced by this tehnology has been
declining since 1992 due to the increasing price of natural gas. Notrogen injec-
tion accounted for 14,700 b/day of the 2004 U.S. oil production (260).
Vol. 1 ENHANCED OIL RECOVERY 187

All thermal recovery methods accounted for 345,514 b/day of U.S. 2004 oil
production with steam injection accounting for 98.6% of this amount (260).
Chemical injection: micellar polymer, polymer, alkaline chemicals, and surfac-
tants, accounted for only 60 b/day of U.S. 2004 oil production (260). The contri-
bution of other EOR technologies to U.S. oil production was negligible.
The value of the oil produced by EOR methods varies with crude oil prices.
Given recent oil price volatility, any financial estimates would have limited
value.

BIBLIOGRAPHY

‘‘Enhanced Oil Recovery’’ under ‘‘Petroleum’’ in ECT 3rd ed., Vol. 17, pp. 168–182, by H.
R. Froning, D. D. Fussell, and E. W. Heffern, Amoco Production Co.; in ECT 4th ed., Vol.
18, pp. 405–432, by J. K. Borchardt, Shell Development Co.; ‘‘Enhanced Oil Recovery’’ in
ECT (online), posting date: December 4, 2000, by J. K. Borchardt, Shell Development Co.

CITED PUBLICATIONS

1. B. Williams, Oil Gas J., 18–25 (Aug. 4, 2003).


2. Anonymous, U.S. Energy Information Agency, http://www.eia.doe.gov/emeu/
international/petroleu.html#IntlProduction.
3. Anonymous, U.S. Energy Information Agency, http://www.eia.doc.gov/oiaf/servicerpt/
depletion/chapter_1.html.
4. A. Bailey, Petroleum News (May 1, 2005).
5. Anonymous, The Economist, www.economist.com/surveys (Apr. 30, 2005).
6. D. H. Smith, Surfactant-Based Mobility Control—Progress in Miscible-Flood
Enhanced Oil Recovery, ACS Symposium Series No. 373, American Chemical
Society, Washington, D.C., 1988.
7. G. Miritis, Oil Gas J., 39–44 (Mar. 3, 2003).
8. F. F. Craig Jr., The Reservoir Engineering Aspects of Water Flooding, Society of
Petroleum Engineers, Monograph No. 3, Dallas, Tex., 1971, pp. 19–22.
9. U.S. Pat. 5,320,170 (June 14, 1994), W. S. Huang, J. J. Hsu, and B. N. Wang
(to Texaco, Inc.).
10. R. M. Butler, J. Can. Pet. Tech. 28, 39 (Mar. 1989).
11. S. D. Joshi, J. Pet. Technol. 43, 1302 (1991).
12. Technical Constraints Limiting Application of Enhanced Oil Recovery Techniques to
Petroleum Production in the United States, U.S. Dept. of Energy, DOE/BETC/RI-83/
9 (DEB4003910), Washington, D.C., Jan. 1984.
13. G. V. Chilinger and T. F. Yen, Energy Sources 7, 67 (1983).
14. N. R. Morrow, J. Pet. Tech. 42, 1476 (1990).
15. W. G. Anderson, J. Pet. Technol. 39, 1453 (1987).
16. W. G. Anderson, J. Pet. Technol. 38, 1125 (1986).
17. U.S. Pat. 4,690,217 (Sept. 1, 1987), D. L. Taggart and E. W. Heffern (to Amoco
Corp.).
18. P. F. Dymond and P. R. Spurr, SPE Reservoir Eng. 3, 165 (1988).
19. D. M. Clementz, D. E. Patterson, R. J. Aseltine, and R. E. Young, J. Pet. Technol. 34,
2087 (1982).
188 ENHANCED OIL RECOVERY Vol. 1

20. F. Cusack, H. M. Lappin-Scott, and J. W. Costerton, Oil Gas J. 85, 87 (Nov. 9, 1987).
21. W. N. Hensel Jr., R. L. Sullivan, and R. H. Stallings, Pet. Eng. Int., 155 (May 1981).
22. M. G. Reed, J. Pet. Technol. 24, 860–864 (1972).
23. J. K. Borchardt, in J. K. Borchardt and T. F. Yen, eds., Oilfield Chemistry: Enhanced
Recovery and Production Stimulation, American Chemical Society, Washington,
D. C., 1989, p. 204.
24. R. D. Sydansk, J. Pet. Technol. 36, 1366 (1984).
25. F. Stalkup Jr., Miscible Displacement, Monograph Vol. 8, Society of Petroleum
Engineers, Dallas, Tex., 1983, p. 62.
26. B. M. Friedman, R. J. Wissbaum, S. G. Anderson, Oil Gas J., Aug. 23, 2004.
27. Anonymous, Oil Gas J., April 20, 2005.
28. D. F. Martin and J. J. Tabor, J. Pet. Technol. 44, 396 (1992).
29. U.S. Pat. 6,871,707 (March 29, 2005), J. M. Karanikas and co-workers, (Shell Oil
Company).
30. Anonymous, Oil Gas J., April 20, 2005.
31. L. Minssieux, in Ref. 23, p. 273.
32. R. T. Johansen, in Improved Oil Recovery, Interstate Oil Compact Commission,
Oklahoma City, Okla., 1983, pp. 91–112.
33. C. Bardon and co-workers, SPE Res. Eng. 9, 92 (1994).
34. D. Leblond, Oil Gas J., Jan. 10, 2005.
35. Anonymous, Oil Gas J., May 6, 2005.
36. U.S. Pat. 6,662,872 (Dec. 16, 2003), A. M. Gurtek and co-workers (ExxonMobil
Upstream Research Company).
37. U.S. Pat. 6,591,908 (July 15, 2003), T. N. Nasr (Alberta Science and Research
Authority).
38. U.S. Pat. 6,403,472 (Oct. 23, 2001), R. C. Richardson and co-workers (Texaco, Inc.).
39. Anonymous, Oil Gas J., Oct. 13, 2004.
40. U.S. Pat. 6,852,215 (Feb. 8, 2005), M. Y. Wen (ExxonMobil Upstream Research
Company).
41. N. C. Nahas, Oil Gas J., Oct. 4, 2004.
42. J. K. Borchardt, Coll. Surf. 63, 189 (1992) and references cited therein.
43. R. D. Sydansk and P. E. Moore, Oil Gas J., 405 (Jan. 20, 1992).
44. P. D. Moffitt, J. Pet. Technol. 45, 356 (1993).
45. J. K. Borchardt, in J. I. Kroschwitz, ed., Encyclopedia of Polymer Science and Engi-
neering, Vol. 10, John Wiley & Sons, Inc., New York, 1987, pp. 350–351.
46. A. Moradi-Araghi and P. H. Doe, SPE Reservoir Eng. 2, 189 (1987).
47. S. K. Nanda, R. Kumar, K. L. Sindhwani, and K. L. Goyl, ONGC Bull. 23, 175 (1986).
48. Ref. 47, p. 344.
49. M. Prats, Thermal Recovery, Monograph Vol. 7, Society of Petroleum Engineers,
Dallas, Tex., 1982.
50. R. K. Prud’homme, J. T. Uhl, J. P. Poinsatte, and F. Halverson, Soc. Pet. Eng. J. 23,
804 (1983).
51. U.S. Pat. 4,413,680 (Nov. 8, 1983), B. B. Sandiford, H. T. Dovan, and R. D. Hutchins
(to Union Oil Co., Calif.).
52. B. Sloat, Pet. Eng. Int., 20 (1977).
53. T. P. Lockhart and P. Albonico, SPE Prod. Engr. 9, 273 (1994).
54. D. J. Jordan, D. W. Green, R. E. Terry, and G. P. Willhite, Soc. Pet. Eng. J. 22, 463
(1982).
55. U.S. Pat. 3,687,200 (Aug. 29, 1972), W. G. Routson (to Dow Chemical Co.).
56. N. A. Mumallah, SPE Reservoir Eng. 3, 243 (1988).
57. U.S. Pat. 6,387,986 (May 14, 2002), A. Moradi-Araghi (no assignee).
58. U. S. Pat. 4,744,418 (May 17, 1988), R. D. Sydansk (to Marathon Oil Co.).
Vol. 1 ENHANCED OIL RECOVERY 189

59. S. Hejri and co-workers, SPE Res. Eng. 8, 229 (1993).


60. U.S. Pat. 4,574,887 (Mar. 11, 1986), M. K. Abdo (to Mobil Oil Corp.).
61. U.S. Pat. 4,439,334 (Mar. 27, 1984), J. K. Borchardt (to Halliburton Co.).
62. U.S. Pat. 4,503,909 (Mar. 12, 1985), P. A. Argabright and J. S. Rhudy (to Marathon
Oil Co.).
63. U.S. Pat. 4,428,429 (Jan. 31, 1984), B. J. Felber and C. A. Christopher (to Standard
Oil Co.).
64. U.S. Pat. 3,334,689 (Oct. 8, 1967), H. C. McLaughlin (to Halliburton Co.).
65. U.S. Pat. 4,321,968 (Mar. 30, 1982), E. E. Clear (to Phillips Petroleum Co.).
66. U.S. Pat. 4,275,789 (June 30, 1981), D. D. Lawrence and B. J. Felber (to Standard Oil
Co., Indiana).
67. U.S. Pat. 4,299,690 (Nov. 11, 1980), B. W. Allan (Texaco, Canada, Inc.).
68. Can. Pat. 1,187,404 (May 21, 1985), M. Navritil, J. P. Batycky, M. Sovak, and M. S.
Mitchell (Borden Co., Ltd.).
69. P. M. Hess, J. Pet. Technol. 32, 1834 (1980).
70. U.S. Pat. 4,663,367 (May 5, 1987), M. Navritil, M. S. Mitchell, and M. Sovak
(to Borden Co., Ltd.).
71. Can. Pat. 1,217,932 (Feb. 17, 1987), M. Navritil, M. S. Mitchell, and M. Sovak
(to Borden Co.).
72. U.S. Pat. 6,103,772 (August 15, 2000), R. D. Sydansk (Marathon Oil Company).
73. U.S. Pat. 4,745,976 (May 24, 1988), J. H. Harwell and J. F. Scamehorn (to Board of
Regents for the University of Oklahoma).
74. R. D. Sydansk, SPE Res. Eng. 5, 346 (1992).
75. U.S. Pat. 4,460,627 (July 17, 1984), J. D. Weaver, L. E. Harris, and W. M. Harms
(to Halliburton Co.).
76. A. Zaitoun, N. Kohler, and Y. Guerrini, J. Pet. Technol. 43, 862 (1991).
77. K. S. Sorbie, Polymer-Improved Oil Recovery, Blackie and Son, Ltd., London, 1991.
78. Ref. 5, pp. 45–47.
79. B. C. Crafts and W. F. Hawkins, Applied Petroleum Reservoir Engineering, Prentice-
Hall, New York, 1959.
80. J. K. Borchardt, in Ref. 45, pp. 328–369.
81. P. B. Lorenz, J. C. Trantham, P. R. Zornes, and C. G. Dodd, SPE Res. Eng. 1, 341
(1986).
82. C. S. H. Chen and E. W. Sheppard, J. Macromol. Sci. Chem. Part A, 239 (1979).
83. D. M. Jones, K. Walters, and P. R. Williams, Rheological Acta 26, 20 (1986).
84. U.S. Pat. 4,053,699 (Oct. 11, 1977), P. T. Cahalan, J. A. Peterson, and D. A. Arndt (to
General Mills Chemicals Inc.).
85. J. Chatterji and J. K. Borchardt, J. Pet. Technol. 32, 2042–2056 (1981).
86. U.S. 4,104,193 (Aug. 1, 1978), W. H. Carter, C. A. Christopher, and T. Jefferson (to
Texaco Inc.).
87. P. A. Sandford and A. Laskin, eds., Extracellular Microbial Polysaccharides, ACS
Symposium Series No. 45, American Chemical Society, Washington, D.C., 1977.
88. U.S. Pat. 4,212,748 (July 15, 1980), B. M. Casad, D. Conley, H. H. Ferrell, and O. M.
Stokke (to Conoco, Inc.).
89. Brit. Pat. 2,115,430 (Sept. 7, 1983), J. R. Bragg (to Exxon Production Research Co.).
90. M. Rinaudo and M. Milas, Int. J. Biol. Macromol. 2, 45 (1980).
91. N. Kohler, D. Longchamp, and M. Thery, J. Pet. Technol. 39, 835 (1987).
92. U.S. Pat. 4,165,257 (Aug. 21, 1979), O. M. Stokke (to Conoco, Inc.).
93. S. L. Wellington, Soc. Pet. Eng. J. 23, 901 (1983).
94. M. C. Cadmus and co-workers, Appl. Environ. Microbiol., 5 (Aug. 1982).
95. Ger. Pat. 2,715,026 (Oct. 13, 1977), S. L. Lee (to Shell Internationale Research
Maatschappij BV).
190 ENHANCED OIL RECOVERY Vol. 1

96. Eur. Pat. 106,666 (Apr. 24, 1984), C. J. Philips (to Pfizer Corp.).
97. Can. Pat. 1,070,492 (Jan. 29, 1980), S. L. Wellington (to Shell Canada, Ltd.).
98. U.S. Pat. 4,454,620 (June 12, 1984), F. Dawans, D. Binet, N. Kohler, and D. V.
Quang (to Institut Francais du Petrole).
99. U.S. Pat. 4,481,316 (Nov. 6, 1984), S. Kanda and G. Kawamura (to Nitto Chemical
Industry Co.).
100. Jpn. Kokai Tokkyo Koho 61 275,337 (Dec. 5, 1986), S. Kanda and G. Kawamura
(to Nitto Chemical Industry Co.).
101. Jpn Kokai Tokkyo Koho 62 184,048 (Aug. 12, 1987), A. Niita, T. Arai, R. Funato, and
T. Sato (to Mitsui Cyanamid KK).
102. Jpn Kokai Tokkyo Koho 62 184,047 (Aug. 12, 1987), A. Niita, T. Arai, R. Funato, and
T. Sato (to Mitsui Cyanamid KK).
103. Jpn Kokai Tokkyo Koho 62 177,052 (Aug. 3, 1987), A. Niita, T. Arai, R. Funato, and
T. Sato (to Mitsui Cyanamid KK).
104. Jpn Kokai Tokkyo Koho 62 277,407 (Dec. 2, 1987), Y. Nakanishi (Mitsubishi Chemical
Industries KK).
105. Eur. Pat. Appl. 196,199 (Oct. 1, 1986), C. P. Judson (to Pfizer, Inc.).
106. Jpn Kokai Tokkyo Koho 61 136,545 (June 24, 1986), A. Niita and Y. Ito (to Mitsui
Toatsu Chemicals, Inc.).
107. Eur. Pat. Appl. 241,340 (Oct. 14, 1987), F. Contat and J. Boutin (to Rhône Poulenc
Chemicals).
108. Eur. Pat. 103,483 (Mar. 21, 1984), G. M. Holzwarth (to Exxon Research and
Engineering Co.).
109. J. C. Phillips, J. W. Miller, W. C. Wernau, B. E. Tate, and M. H. Auerbach, Soc. Pet.
Eng. J. 25, 594 (1985).
110. U.S. Pat. 6,316,614 (Nov. 13, 2001), D. H. Doherty (CP Kelco U.S., Inc.).
111. U.S. Pat. 4,457,372 (July 3, 1984), M. S. Doster, A. J. Nute, and C. A. Christopher
(to Texaco, Inc.).
112. Eur. Pat. Appl. 211,288 (Feb. 25, 1987), R. W. Vanderslice and P. Shanon (to Getty
Science and Development Co.).
113. A. Audibert, C. Noik, and J. Lecourtier, J. Can. Pet. Technol. 32(7), 53 (1993).
114. U.S. Pat. 4,439,332 (Mar. 27, 1984), S. Frank, A. T. Coscia, and J. M. Schmitt
(to American Cyanamid Co.).
115. U.S. Pat. 4,439,342 (May 8, 1984), J. K. Borchardt (to Halliburton Co.).
116. J. Klein and R. Heitzman, Makromol. Chem. 179, 1895 (1978).
117. F. H. Halverson and J. E. Lancaster, Macromolecules 18, 1139 (1985).
118. Fr. Pat. 2,495,217 (June 4, 1982), J. Boutin and F. Contat (to Rhône Poulenc
Industries).
119. G. Muller, J. P. Laine, and J. C. Fenyo, J. Polym. Sci. Polym. Chem. Ed. 17, 659
(1979).
120. J. S. Ward and F. D. Martin, Soc. Pet. Eng. J. 21, 623 (1981).
121. M. Szabo, J. Pet. Technol. 31, 553 (1979).
122. M. Szabo, J. Pet. Technol. 31, 561 (1979).
123. J. G. Southwick and C. W. Manke, SPE Reservoir Eng. 3, 1193 (1988).
124. C. L. McCormick, J. Macromol. Sci. Chem. Part A. 22, 955 (1985).
125. U.S. Pat. 4,644,020 (Feb. 17, 1987), G. A. Stahl (to Phillips Petroleum Co.).
126. P. H. Doe, A. Moradi-Araghi, J. E. Shaw, and G. A. Stahl, SPE Res. Eng. 2, 461
(1987).
127. C. L. McCormick and K. P. Blackmon, J. Polym. Sci. A, Polym. Chem. 24, 2635
(1986).
128. Ger. Pat. 3,627,456 (Feb. 26, 1987), T. Y. Ching (to Chevron Research Co.).
129. Eur. Pat. Appl. 233,533 (Aug. 26, 1987), R. G. Ryles (to American Cyanamid Co.).
Vol. 1 ENHANCED OIL RECOVERY 191

130. Ger. Pat. 3,220,503 (Dec. 1, 1983), P. Engelhardt, U. Greiner, H. Schmitz, W. Gulden,
and S. P. von Halasz (to Casella AG).
131. Jpn Kokai Tokkyo Koho 62 015,279 (Jan. 23, 1987), Y. Ito, A. Niita, Y. Sudo, and K.
Hayashi (to Mitsui Toatsu Chemicals, Inc.).
132. U.S. Pat. 4,540,498 (Sept. 10, 1985), M. M. Wu and L. E. Ball (Standard Oil Co., Ohio).
133. R. W. Dexter and R. G. Ryles, in Ref. 20, p. 102.
134. H. Pledger Jr., J. J. Meister, T. E. Hogen-Esch, and G. B. Butler, Polym. Prepr. Am.
Chem. Soc. Div. Polym. Chem. 22 (2), 72 (1981).
135. H. H. Neidlinger and C. L. McCormick, Polym. Prepr. Am. Chem. Soc. Div. Polym.
Chem. 20(1), 901 (1979).
136. J. J. Meister, Polym. Prepr. Am. Chem. Soc. Div. Polym. Chem. 25(1), 266 (1984).
137. U.S. Pat. 6,030,928 (Feb. 29, 2004), G. A. Stahl (Phillips Petroleum Company).
138. F. Friedmann, SPE Reservoir Eng. 1, 261 (1986).
139. Y. Cohen and F. R. Christ, SPE Reservoir Eng. 1, 113 (1986).
140. L. T. Lee, J. Lecourtier, and G. Chaveteau, in Ref. 20, p. 224.
141. S. A. Hong and J. H. Bae, SPE Res. Eng. 5, 467 (1990).
142. U.S. Pat. 4,627,494 (Dec. 9, 1986), G. Kalfoglou (to Texaco, Inc.).
143. U.S. Pat. 4,713,185 (Dec. 15, 1987), J. Howard and M. Stirling (to Canadian Patents
and Development Ltd.).
144. U.S. Pat. 4,284,517 (Aug. 18, 1981), C. S. H. Chen and E. W. Sheppard (to Mobil Oil
Corp.).
145. Brit. Pat. 2,148,356 (Mar. 4, 1987), J. H. Clint, P. K. G. Hodgson, and E. J. Tinley (to
British Petroleum PLC).
146. U.S. Pat. 4,574,885 (Mar. 11, 1986), R. L. Horton (to Phillips Petroleum Co.).
147. J. P. Heller, D. K. Dandge, R. J. Card, and L. G. Donaruma, Soc. Pet. Eng. J. 25, 679
(1985).
148. T. G. Arf and co-workers, SPE Reservoir Eng. 2, 166 (1987).
149. S. L. Wellington and H. J. Vinegar, J. Pet. Technol. 39, 885 (1987).
150. G. J. Parlani, K. A. McColloch, S. L. Warden, and D. R. Edens, J. Pet. Technol. 44,
984 (1992).
151. U.S. Pat. 6,105,672 (August 22, 2000), C. Deruyter (Institut Francais du Petrole).
152. F. Stalkup Jr., J. Pet. Technol. 36, 815–826 (1984).
153. J. P. Heller, in L. L. Schramm, ed., Foams: Fundamentals and Applications in the
Oil Industry, American Chemical Society, Washington, D.C., 1994, p. 201.
154. K. Mannhardt and J. J. Novosad, in Ref. 158, p. 259.
155. J. E. Hanssen and T. Holt, Rogoland Research Report, RF-80/93, Stavanger,
Norway, 1993, and references cited therein.
156. U.S. Pat. 3,185,634 (May 25, 1965), F. F. Craig Jr., and J. L. Lummus (to Pan
American Petroleum Corp.).
157. M. I. Kuhlman and co-workers, SPE Res. Eng. 7, 445 (1992).
158. J. K. Borchardt , in Ref. 20, pp. 181–204.
159. U.S. Pat. 4,706,752 (Nov. 17, 1987), L. W. Holm (to Union Oil Co. of California).
160. U.S. Pat. 4,706,750 (Nov. 17, 1987), J. J. Buckles (to Mobil Oil Corp.).
161. U.S. Pat. 4,502,538 (Mar. 5, 1985), S. L. Wellington, J. Reisberg, E. F. Lutz, and
D. B. Bright (to Shell Oil Co.).
162. U.S. Pat. 4,799,547 (Jan. 24, 1989), J. K. Borchardt (to Shell Oil Co.).
163. U.S. Pat. 4,380,266 (Apr. 19, 1983), S. L. Wellington (to Shell Oil Co.).
164. U.S. Pat. 4,739,831 (Apr. 26, 1988), L. A. Settlemeyer and M. J. McCoy (to Dow
Chemical Co.).
165. U.S. Pat. 4,676,316 (June 30, 1987), T. O. Mitchell (to Mobil Oil Corp.).
166. U.S. Pat. 6,439,308 (August 27, 2002), D. Wang (Da Qing Petroleum Administration
Bureau).
192 ENHANCED OIL RECOVERY Vol. 1

167. Can. Pat. 1,221,305 (May 5, 1987), N. F. Djabbarah (to Cities Service Oil and Gas
Corp.).
168. U.S. Pat. 4,733,727 (Mar. 29, 1988), A. H. Falls (to Shell Oil Co.).
169. L. W. Holm, J. Pet. Technol. 32, 1499 (1970).
170. L. W. Holm and W. H. Garrison, SPE Reservoir Eng. 3, 112 (1988).
171. L. D. Talley, SPE Reservoir Eng. 3, 235 (1988).
172. Brit. Pat. 2,239,278 (June 21, 1991), L. L. Schramm, C. Ayasse, K. Mannhardt, and
J. Novosad (to Alberta Oil Sands Technology and Research Authority).
173. G. P. Small, SPE Prod. Eng. 1, 388 (1986).
174. K. C. Hong and C. E. Stevens, SPE Reservoir Eng. 7, 407 (1992).
175. T. J. C. Fontaine, L. Hayes, and G. Reese, J. Can. Pet. Technol. 32, 44 (Nov. 1993).
176. P. J. Jesperson and T. J. C. Fontaine, J. Can. Pet. Technol. 32, 52 (May 1993).
177. D. K. Olsen and P. S. Sarathi, Proceedings of the Field Application of Foams for Oil
Production Symposium, U.S. Dept. of Energy, N.I.P.E.R., Bartlesville, Okla., 1993.
178. U.S. Pat. 5,333,687 (Aug. 2, 1994), W. T. Osterloh (to Texaco, Inc.).
179. E. E. Isaacs, F. C. McCarthy, and J. D. Maunder, SPE Reservoir Eng. 3, 565 (1988).
180. U.S. Pat. 4,693,311 (Sept. 15, 1987), H. M. Muijs and P. P. M. Keijzer (to Shell Oil
Co.).
181. Ger. Pat. 3,510,765 (Sept. 26, 1985), R. Janssen van Rosmalen, H. M. Muijs, and
P. P. M. Keijzer (to Shell Internationale Research Maatschappij BV).
182. U.S. Pat. 5,273,682 (Dec. 28, 1993), M. Danzik (to Chevron Research and Technology
Co.).
183. Ger. Pat. 3,734,075 (Apr. 4, 1988), H. M. Muijs and P. P. M. Keijzer (to Shell Inter-
nationale Research Maatschappij BV).
184. Ger. Pat. 3,503,532 (Aug. 8, 1985), W. S. Huang, Z. Z. Gassmann, J. T. Hawkins,
V. H. Schievelbein, and W. L. Hall (to Texaco Development Co.).
185. U.S. Pat. 4,699,214 (Oct. 13, 1987), H. P. Angstadt (to Sun Refining and Marketing
Co.).
186. PCT Int. Pat. Appl. 85 05,146 (1985), T. Lim (to Dow Chemical Co.).
187. U.S. Pat. 4,576,232 (Mar. 18, 1986), J. H. Duerksen, R. G. Wall, and J. D. Knight (to
Chevron Research Co.).
188. A. H. Falls, J. B. Lawson, and G. J. Hirasaki, J. Pet. Technol. 40, 95 (1988).
189. J. Novosad, B. B. Maini, and A. Huang, J. Can. Pet. Technol. 25, 42 (1986).
190. H. C. Lau and S. M. O’Brien, SPE Reservoir Eng. 2, 1177 (1988).
191. U.S. Pat. 4,597,442 (July 1, 1986), R. E. Dilgren, H. C. Lau, and G. J. Hirasaki
(to Shell Oil Co.).
192. U.S. Pat. 4,617,995 (Oct. 21, 1986), H. C. Lau (to Shell Oil Co.).
193. Fr. Pat. 2,557,198 (Juen 28, 1985), H. P. Angstadt, D. F. Rugen, and J. L. Cayias
(to Sun Marketing and Refining Co.).
194. E. E. Isaacs, J. D. Maunder, and L. Juan, in Ref. 20, p. 325.
195. B. B. Maini and V. Ma, J. Can. Pet. Technol. 25, 65–69 (1986).
196. H. P. Angstadt and H. Tsao, SPE Reservoir Eng. 2, 613 (1987).
197. U.S. Pat. 4,702,317 (Oct. 27, 1987), C. W. Shen (to Texaco, Inc.).
198. U.S. Pat. 4,727,938 (Mar. 1, 1988), H. C. Lau (to Shell Oil Co.).
199. H. C. Lau and J. K. Borchardt, SPE Reservoir Eng. 6, 470 (1991).
200. U.S. Pat. 5,105,884 (Apr. 21, 1992), R. D. Sydansk (to Marathon Oil Co.).
201. D. O. Shah and R. S. Schecter, eds. Improved Oil Recovery by Surfactant and
Polymer Flooding, Academic Press, Inc., New York, 1977.
202. H. K. van Poolen, Fundamentals of Enhanced Oil Recovery, PennWell Publishing
Co., Tulsa, Okla., 1980, p. 114.
203. J. C. Morgan, R. S. Schechter, and W. H. Wade, in K. L. Mittal, ed., Proc. Sect. 52nd
Colloid Surf. Sci. Symp., Vol. 2, Plenum Press, New York, 1979, pp. 749–775.
Vol. 1 ENHANCED OIL RECOVERY 193

204. M. Bourrel, F. Verzaro, and C. Chambu, SPE Res. Eng. 2, 41 (1987).


205. Ger. Pat. 3,303,894 (Aug. 8, 1983), H. S. Aldrich, T. L. Ashcraft, M. C. Puerto, and
R. L. Reed (to Exxon Production Research Co.).
206. A. Graciaa, L. M. Fortney, R. S. Schechter, W. H. Wade, and S. Yiv, Soc. Pet. Eng. J.
20, 743 (1982).
207. Ger. Pat. 3,542,063 (June 12, 1986), M. C. Puerto (to Exxon Production and Research Co.).
208. U.S. Pat. 4,270,607 (June 2, 1981), R. L. Cardenas, B. G. Harnsberger, and
J. Maddox Jr., (to Texaco, Inc.).
209. Y. Barakat and co-workers, Soc. Pet. Eng. J. 23, 913 (1983).
210. U.S. Pat. 4,555,351 (Nov. 26, 1985), H. Morita, Y. Kawada, J. Yamada, and T. Ukigai
(to Lion Corp.).
211. U.S. Pat. 4,512,404 (Apr. 23, 1985), H. Morita, Y. Kawada, J. Yamada, and T. Ukigai
(to Lion Corp.).
212. Brit. Pat. 2,160,242 (Sept. 9, 1987), R. Thaver (to British Petroleum PLC).
213. Brit. Pat. Appl. 2,184,763 (July 1, 1987), R. Thaver (to British Petroleum PLC).
214. Ger. Pat. 3,422,613 (Dec. 19, 1985), N. Greif, K. Oppenlaender, and K. U. Sewe
(to BASF AG).
215. Eur. Pat. Appl. 264,867 (Apr. 27, 1988), R. Schmidt, W. Rupp, G. Schneider, and
E. M. Kohn (to Hoechst AG).
216. U.S. Pat. 4,426,302 (Jan. 17, 1984), D. R. McCoy, R. M. Gipson, and C. G. Naylor
(to Texaco, Inc.).
217. Y. C. Chiu and H. J. Hwang, Colloids Surf. 28, 53 (1987).
218. J. E. Shaw, J. Am. Oil Chem. Soc. 61, 1395 (1984).
219. U.S. Pat. 4,692,551 (Sept. 8, 1987), K. C. Liu (to GAF Corp.).
220. M. Abe, R. S. Schechter, R. D. Selliah, B. Sheikh, and W. H. Wade, J. Dispersion Sci.
Technol. 8, 157 (1987).
221. M. Baviere, B. Bazin, and C. Noil, SPE Reservoir Eng. 3, 597 (1988).
222. M. C. Puerto and R. L. Reed, Soc. Pet. Eng. J. 23, 669 (1983).
223. Fr. Pat. 2,589,858 (May 15, 1987), H. S. Aldrich, M. C. Puerto, and R. L. Reed
(to Exxon Production Research Co.).
224. U.S. Pat. 6,828,281 (December 7, 2004), W. Hou (Akzo Nobel Surface Chemistry
LLC).
225. C. Lalanne-Cassou and co-workers, J. Dispersion Sci. Technol. 8, 137 (1987).
226. T. A. B. M. Bolsman and G. J. R. Daane, SPE Reservoir Eng. 1, 53 (1986).
227. B. B. Bansal, V. Hornof, and G. Neale, Can. J. Chem. Eng. 57, 203 (1979).
228. K. Manasrah, G. H. Neale, and V. Hornof, Cellul. Chem. Technol. 19, 291 (1985).
229. O. S. Hjelmeland and L. E. Larrondo, SPE Reservoir Eng. 1, 321 (1986).
230. P. Somasundaran and H. H. Shafick, Soc. Pet. Eng. J. 25, 343 (1985) and references
therein.
231. P. B. Lorenz, J. C. Trantham, D. R. Zornes, and C. G. Dodd, SPE Reservoir Eng. 1,
341 (1986).
232. J. K. Burk, SPE Reservoir Eng. 2, 9 (1987).
233. P. H. Krumrine, I. B. Ailin-Pyzik, and J. S. Falcone Jr., Prepr. Am. Chem. Soc., Div.
Pet. Chem. 26(1), 195 (1981).
234. U.S. Pat. 5,311,943 (May 17, 1994), S. L. Wellington (to Shell Oil Co.).
235. Ref. (214), pp. 104–111.
236. J. D. Neil, H. L. Chang, and T. M. Geffern, in Improved Oil Recovery, Interstate Oil
Compact Commission, Oklahoma City, Okla., 1983, pp. 52–66.
237. S. M. Mohnot, J. H. Bae, and W. L. Foley, SPE Reservoir Eng. 2, 653 (1987).
238. U.S. Pat. 4,714,113 (Dec. 12, 1987), S. M. Mohnot and P. M. Chakrabarti (to PPG
Industries, Inc.).
239. L. W. Holm and S. D. Robertson, J. Pet. Technol. 33, 161 (1981).
194 OIL SHALE Vol. 1

240. U.S. Pat. 4,466,892 (Aug. 21, 1984), K. S. Chan and S. J. Majoros (to Standard Oil
Co., Ohio).
241. A. L. Bunge and C. J. Radke, Soc. Pet. Eng. J. 23, 657 (1983).
242. P. J. Shuler, D. L. Kuehne, and R. M. Lerner, J. Pet. Technol. 41, 80 (1989) and re-
ferences cited therein.
243. R. C. Nelson, J. B. Lawson, D. R. Thigpen, and G. L. Stegemeier, Proc. 1984 SPE
Enhanced Oil Recovery Symposium, Society of Petroleum Engineers, Dallas, Tex.,
1984, pp. 417–424.
244. R. C. Nelson and co-workers, SPE Res. Eng. 9, 217 (1994).
245. G. Moritis, Oil Gas J., March 28, 2005.
246. S. M. Saleem and M. J. Faber, Rev. Tec. INTEVEP 6, 133–142 (1986).
247. S. M. Saleem and A. Hernandez, J. Surf. Sci. Technol. 3, 1 (1987).
248. M. J. Pitts, Prepr. Am. Chem. Soc., Div. Pet. Chem. 33 (1), 169 (1988).
249. U.S. Pat. 4,502,541 (Mar. 5, 1985), J. B. Lawson and D. R. Thigpen (to Shell Oil Co.).
250. G. H. Neale, K. C. Khulbe, and V. Hornof, Can. J. Chem. Eng. 65, 700 (1987).
251. G. E. Jenneman, R. M. Knapp, M. J. McInerney, D. E. Menzie, and D. E. Revus, Soc.
Pet. Eng. J. 24, 31 (1984).
252. D. O. Hitzman, Microbial Enhancement of Oil Recovery—Recent Advances, Elsevier
Science Publishing Co., Inc., New York, 1991, pp. 11–20.
253. I. Lazar, Dev. Petr. Sci. 31, 485 (1991).
254. U.S. Pat. 5,758,270 (July 6, 2004), E. Sunde and co-workers (Statoil ASA).
255. P. B. Crawford and C. Cju, Improved Oil Recovery, Interstate Oil Compact Commis-
sion, Oklahoma City, Okla., 1983, pp. 251–309.
256. Ref. 214, pp. 41–55.
257. B. Nzekwu, R. J. Hallam, and G. J. J. Williams, SPE Reservoir Eng. 5, 163 (1990);
Ref. 224, pp. 41–55.
258. B. Williams, Oil Gas J., pp. 18–25 (Aug. 4, 2003).
259. U.S. Energy Information Agency, http://www.eia.doe.gov/emeu/international/
petroleu.html#IntlProduction.
260. Anonymous, Oil Gas J., pp. 45–52 (Apr. 12, 2004).
261. A. Y. Zekri and K. K. Jerbi, Oil & Gas Science and Technology–Rev. IFP 57,
pp. 259–267 (2002).

JOHN K. BORCHARDT
Shell Development Company

OIL SHALE
1. Introduction

Oil shale is a sedimentary mineral that contains kerogen, a mixture of complex,


high molecular weight organic polymers. The solid kerogen is a three-dimensional
polymer that is insoluble in conventional organic solvents. Upon heating,
kerogen decomposes to form gas composed of hydrogen, low molecular weight
hydrocarbons (qv), and carbon monoxide; liquids, composed of water and shale
oil; and a solid char residue.

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.15091216091605.a01
Vol. 1 OIL SHALE 195

Oil shale deposits were formed in ancient lakes and seas by the slow deposi-
tion of organic and inorganic remains. The geology and composition of the inor-
ganic minerals and organic kerogen components of oil shale vary with deposit
locations throughout the world (1) (see also PETROLEUM).

2. Reserves

Estimates of oil shale deposits by continent are given in Table 1 (2). Characteris-
tics of many of the world’s best known oil shales are summarized in Table 2 (3,4).
Oil shale deposits in the United States occur over a wide area (Table 3). The
most extensive deposits, covering ca 647,000 km2 (250,000 mi2), are the Devo-
nian-Mississippian shales of the eastern United States (5). The richest U.S. oil
shales are in the Green River formation of Colorado, Utah, and Wyoming. Typical
mineral and organic analyses for Green River oil shale are given in Table 4.
The Green River formation includes an area of ca 42,720 km2 (16,500 mi2),
and in-place reserves are ca ð0:5 1:1Þ  1012 m3 ðð3 7Þ  1012 bblÞ of which ca
80% are federally owned. The richest portion (85% of the reserve) of the Green
River formation is in the Piceance Basin of Colorado. The deposits in Utah and
Wyoming contain 10% and 5% of the reserves, respectively (6).
The Parachute Creek member contains the majority of the oil shale in
the Piceance Creek Basin and is ca 580 m thick at the depositional center
of the basin. The members of the Green River formation and the thickness of
the various zones are indicated in Figure 1. Organic and saline mineral con-
tents increase toward the depositional center of the basin. The rich Mahogany
zone extends across the Piceance Basin and into the Uinta Basin in eastern
Utah. In addition to its high contents of organic matter, the Parachute
Creek member contains large reserves of nahcolite[15752-47-3],NaHCO3,
and dawsonite [12011-76-6],NaAl(OH)2CO3, which are present in the deepest
parts of the basin.

Table 1. Shale Oil Resourcesa, 109 m3b,d


Marginal or submarginal
Total resourcec resourcesd
Geographic area 21–42 42–104 104–417 21–42 42–104 104–417
Africa 71,500 12,700 636 small small 14
Asia 93,800 17,500 874 2 11
Australia and 15,900 3,200 159 small small
New Zealand
Europe 22,260 4,100 223 1 6
North America 41,400 8,000 477 350 254 99
South America 33,400 6,400 318 119 small
Total 278,260 51,900 2,687 350 376 130
a
Ref. 2.
b
To convert m3 to bbl, divide by 0.159.
c
Includes oil shale in known resources, in extensions of known resources, and in undiscovered
but anticipated resources.
d
Numbers represent shale oil yield range in L/t. To convert L/t to gal/short ton, multiply by 0.2397.
Table 2. Properties of World Wide Oil Shalesa
New
Property Timahdit Irati Nagoorin Kentucky Maoming Colorado Condor Alpha Brunswick Israeli Kunkersite
Fischer assay, %
oil weight 6–9 6–12 14.1 5.3 9.7 16.5 6.3 52.0 6–12 6.2 28.6
water, bound 2.1–2.7 0.2–2.1 6.9 1.9 3.8 1.0 1.9 4.0 0.9–1.4 2.8 2.5
spent shale 85–88 83–90 72.4 90.0 82.0 78.6 87.3 33.0 91.1–84.5 87.4 62.7
gas þ loss 2.8–3.7 2–4 6.6 2.8 4.5 3.8 4.5 11.0 2.0–2.1 3.6 6.2

Other properties
moisture, wt% 6.7–9.8 0.2–6 23.2 2.8 11.3 0.7 7.7 2.8 5.4–6.7 8.1 5.8
specific gravity 1.88–1.99 1.9–2.1 1.47 2.22 1.73 1.94 2.05 1.16 2.32–1.97 1.57 1.60
gross heating value, 5.230–6.904 5.439–6.987 11.950 5.791 8.577 9.113 4.728 30.669 3.766–6.908 4.209 16.07
J/gb
total carbon, wt% 14.78–19.46 12–17 25.67 12.82 18.74 23.45 10.50 70.54 10.75–16.58 15.5 36.8
total hydrogen, % 1.9–2.0 0.9–2.4 3.7 1.5 2.9 2.9 1.7 8.39 1.4–2.2 1.6 4.3
total sulfur, % 2.1–2.7 3.9–5.6 1.0 4.4 1.6 1.1 0.9 1.4 0.9–1.0 2.9 2.0
nitrogen, % 0.46–0.63 0.3–1.9 0.3 1.3 0.6 0.3 1.0 0.1

196
loss on ignition, at 31.4–38.9 20–24 41.9 21.3 32.1 38.0 18.6 91.7 21.1–28.5 56.2
9508C, %
Ash composition, wt%
SiO2 31.6–37.5 5.0–5.6 64.8 57.2 45.2 73.2 53.4 54.8–55.7 33.2
Fe2O3 3.5–5.8 7.6–9.8 10.7 12.2 5.5 8.1 9.9 6.8–5.7 6.6
Al2O3 8.6–13.0 9.8–12.6 12.5 19.5 2.3 12.1 24.3 17.9–15.0 8.9
CaO 15.7–26.7 1.3–3.9 1.9 1.1 18.9 2.0 3.4 8.9–13.8 33.7
MgO 5.6–7.4 2.0–3.7 0.6 0.8 17.4 1.0 4.1 6.1–3.7 9.5
Fischer assay oil
specific gravity, 208C 0.962 0.906 0.918 0.926 0.890 0.902 0.895 0.905 0.880 0.980 0.958
total carbon, wt% 78.73 84.60 83.40 84.95 84.81 84.21 84.72 84.32 85.6 80.8 83.4
hydrogen, % 9.69 12.50 11.37 11.85 11.65 11.29 12.54 11.89 12.3 10.4 10.7
sulfur, % 6.33 1.10 1.16 1.40 0.52 0.92 0.46 1.72 0.6 5.0 0.7
nitrogen, % 1.52 0.90 1.18 1.12 2.60 1.78 1.30 0.69 1.1 1.2 0.1
gross heating value, 40.074 42.547 43.070 41.773 42.447 42.723 42.677 42.539 43.932 39.748 39.790
J/gb
a
Ref. 3.
b
To convert J to cal, divide by 4.184.
Vol. 1 OIL SHALE 197

Table 3. Shale Oil Resources of the United Statesa, 109 m3b


Marginal or submarginal
Total resourcec,d resourcesd
Geographical area 21–42 42–104 104–417 21–42 42–104 104–417
Green River Formation, ie, 636 445 191 318 223 83
Colorado, Utah, and
Wyoming
central and eastern United 318 159 32 32 0
States
Alaskan deposits large 32 40 small small small
other 21,300 3,537 80 small small
Total 22,254 4,173 311 350 254 83
a
Ref. 2.
b
To convert m3 to bbl, divide by 0.159.
c
Includes oil shale in known resources, in extensions of known resources, and in undiscovered
but anticipated resources.
d
Numbers represent shale oil yield range in L/t. To convert L/t to gal/short ton, multiply by 0.2397.

3. Analytical and Test Methods

Sample preparation for the modified Fischer assay technique, a standard method
to determine the liquid yields from pyrolysis of oil shale, is necessary to achieve
reproducible results. A 100-g sample of >230 mm (65 mesh) of oil shale is
heated in a Fischer assay retort through a prescribed temperature range, eg,
ca 25.5–5008C, for 50 min and then soaked for 20 min. The organic liquid
which is collected is the Fischer assay yield (7). The Fischer assay is not an abso-
lute method, but a qualitative assessment of the oil that may be produced from a
given sample of oil shale (8). Retorting yields of greater than 100% of Fischer
assay are possible.

Table 4. Composition of Green River Oil Shalea,b


Composition,
Material wt%
mineral (inorganic, 85 wt% of total)
carbonates 40.8
feldspars 17.8
quartz 12.8
clays 11.0
pyrite and analcite 2.6
kerogen (organic, 15 wt% of total)
carbon 11.7
hydrogen 1.5
nitrogen 0.3
sulfur 0.15
oxygen 1.35
a
Ref. 1.
b
Shale oil yield of 104 L/t (25 gal/short ton).
198 OIL SHALE Vol. 1

Upper aquifer
Uinta Thickness,
formation 0 – 500 m

Lean upper
oil shale
zone
Mahogany zone

member (60 – 500 m)


Parachute Creek
25 – 70 m thick
(relatively
impermeable)

aquifer
Lower
Leach zone

Saline zone
(nahcolite,
dawsonite,
etc)
member
Garden
Gulch

Thickness,
90 – 300 m
Douglas

member
Creek

Thickness,
0 – 140 m

Wasatch
formation Ancient
lake bed

Fig. 1. Green River formation, Colorado.

A total material balance assay is a Fischer assay in which the retort gases
are collected. A complete material balance closure and yields in excess of those
expected from Fischer assay results are achieved. More complete descriptions of
both the Fischer assay and the Tosco material balance assay methods have been
reported (9).

4. General Properties

4.1. Kerogen Decomposition. The thermal decomposition of oil shale,


ie, pyrolysis or retorting, yields liquid, gaseous, and solid products. The amounts
of oil, gas, and coke which ultimately are formed depend on the heating rate of
the oil shale and the temperature–time history of the liberated oil. There is little
effect of shale richness on these relative product yields under fixed pyrolysis
conditions, as is shown in Table 5 (10).
Numerous kinetic mechanisms have been proposed for oil shale pyrolysis
reactions (11–14). It has been generally accepted that the kinetics of the oil
shale pyrolysis could be represented by a simple first-order reaction
(kerogen ! bitumen ! oil), or

sequential A ! B ! C ð1Þ
Vol. 1 OIL SHALE 199

Table 5. Conversion of Kerogen by Fischer Assaya


Grade of shale, L/t
Component 43.8 111.4 151.5 238.3 257.9 312.9
oil, wt% 51 65 69 66 69 71
gas, wt% 14 12 11 12 12 11
organic residue, wt% 35 23 20 22 19 18
a
Ref. 10.

This sequential first-order reaction adequately describes the kinetics of pyrolysis


of the Green River oil shale in western United States. Additional kinetic studies
(15,16) indicate that sequential reactions are inadequate to describe the kinetic
reactions for the thermal decomposition of oil shales worldwide. First, there is no
well-defined chemical induction time as predicted by first-order reactions. Sec-
ondly, kerogen decomposition is a complex array of thermal reactions involving
a variety of organic materials, water, and gases, such as CO and CO2, as well as
hetero-atom reactions involving nitrogen, sulfur, and oxygen. It is impossible to
define the process using simple individual reactions. The kinetic reactions can
best be described using a global approach that encompasses the sequential
first-order reaction (eq. 1) as well as

parallel A C

and

A B
alternate
C

4.2. Temperature and Product Yields. Most oil shale retorting pro-
cesses are carried out at ca 4808C to maximize liquid product yield. The effect
of increasing retort temperature on product type from 480 to 8708C has been stu-
died using an entrained bed retort (17). The oil yield decreased and the retort gas
increased with increased retorting temperature; the oil became more aromatic as
temperature increased, and maximum yields of olefinic gases occurred at about
7608C. Effects of retorting temperatures on a distillate fraction (to 3008C) are
given in Table 6.
4.3. Carbonate Decomposition. The carbonate content of Green River
oil shale is high (see Table 4). In addition, the northern portion of the Piceance
Creek basin contains significant quantities of the carbonate minerals nahcolite
and dawsonite. The decomposition of these minerals is endothermic and occurs
at ca 600–7508C for dolomite, 600–9008C for calcite, 350–4008C for dawsonite,
and 100–1208C for nahcolite. Kinetics of these reactions have been studied (19).
Carbon dioxide, a product of decomposition, dilutes the off-gases produced from
retorting processes at the above decomposition temperatures.
200 OIL SHALE Vol. 1

Table 6. Effect of Retorting Temperature on Product Typea,b


Retorting temperature of Saturates, Olefins, Aromatics,
distillate (reduced to 3008C) vol% vol% vol%
5378C 18 57 25
6498C 7.5 39.5 53
7608C 0 2.5 97.5
8718C 0 0 100
gas combustion 30 50 20
simulated in situ 41 37 22
in situ 59 16 25
median U.S. crude 60–100 <5 0–40
a
Ref. 18.
b
Colorado oil shale.

5. Retorting

Oil shales are solid minerals, impervious to the flow of fluids, and are generally
situated in deposits below the earth’s surface. Therefore, several process steps
must be undertaken to produce crude shale oil. In the case of the commonly
used above-ground retorting (AGR), these steps involve mining, crushing, and
heating. The grade (volume of oil per weight of rock) of most oil shales is low
(see Table 2), and large amounts of the oil shale rock must be processed to pro-
duce crude shale oil. Depending on the grade, 2 to 25 metric tons of oil shale must
be processed to produce one cubic meter of crude shale oil (0.4–4.6 short tons per
barrel of crude shale oil). In order to eliminate the costs of mining and material
handling, direct underground retorting (in situ retorting) has been considered as
an alternative to the conventional AGR.
Historically, direct combustion has been employed in which some of the
organic matter of the kerogen is combusted to provide the heat necessary for
retorting. Although these direct heat (DH) processes do not require a supplemen-
tal source of fuel, some of the kerogen is consumed and the gaseous products of
the kerogen decomposition are diluted with the products of combustion. In order
to obviate these shortcomings, indirect heat (IH) processes were developed in
which the heat required for retorting was supplied by hot gases or solids that
were heated externally. However, the IH processes do not utilize any of the
solid residual carbon or char resulting from kerogen decomposition and they
do require an external source of fuel.
There are numerous means of classifying the many processes that have
been employed to retort oil shale. In addition to the types of retorting, the
retorting process can be classified by the type of feed used and by the flows
within the retort. Types of oil shale feed can be classified as coarse, >5 mm
(>0:25 in:) or fine, <65 mm (<0:25 in:). The flows within the retort can be classi-
fied as concurrent, ie, all materials flowing in the same direction, or countercur-
rent, ie, the solid oil shale flowing in one direction, the air and gases in the
opposite direction. The coarse feed systems usually result in the disposal of
the raw shale fines as waste. The fine feed systems obviate the latter problem
but result in increased crushing costs and greater environmental impacts from
Vol. 1 OIL SHALE 201

Table 7. Retorting Technologies


Technology Country Heating processa,b Feed Flowc
Above-ground retorting
Chevron United States DH fine
FBC Israel DH fine CC
Fuschun China DH coarse CC
Galoter Russia IH coarse CO
Gas combustion United States DH coarse
Kiviter Russia IH and DH coarse CO and CC
LLNL/HRS United States IH (ash) fine CO
Lurgi United States IH (ash) fine CO
Paraho DH United States DH coarse CC
Paraho IH United States IH (gas) coarse CC
Petrosix Brazil IH (gas) coarse CC
Superior United States IH (gas) coarse
Taciuk Australia IH (gas) fine CC
TOSCO II United States IH (solids) fine CC
Unishale A United States DH coarse CC
Unishale B United States IH (gas) coarse CC
In situ retorting
Equity BX United States IH (steam)
IGT United States IH (H2/steam) CO
LOFRECO United States DH
MultiMineral United States DH
RISE United States DH
VMIS United States DH
a
DH ¼ direct heat; IH ¼ indirect heat.
b
Heat-transfer medium is given in parentheses.
c
CC ¼ countercurrent; CO is concurrent.

particulate emissions during the material handling operations. A list of most


of the oil shale retorting processes in use worldwide since the 1940s is provided
in Table 7.
Retorting processes consist of several well-defined steps, or zones, within
the retort, as shown in Figure 2 for the batch-process Nevada-Texas-Utah
(NTU) retort, the forerunner of most of the technologies listed in Table 7. For
DH systems the zones are the oil shale preheating or off-gas oil-mist cooling
zone; the pyrolysis zone, where the solid organic kerogen is converted into
gases, oil mists and vapors, and residual carbon; the combustion zone, where
carbon is burned to provide heat; and the shale cooling zone, where the retorted
shale is cooled and the incoming air is preheated.
5.1. Above-Ground Retorting. AGR processes can be grouped into DH
or IH processes. Numerous design configurations as well as a variety of heat-
transport mediums have been used in the indirect heated processes (Table 7).
5.2. Gas Combustion Retort. The continuous gas combustion retort
(GCR) has been modeled after the earlier batch-operation NTU retort. Although
the term ‘‘gas combustion’’ has been applied to this process, it is a misnomer in
that, in a well-designed and properly operated system, the residual char on the
202 OIL SHALE Vol. 1

Combustion air

Combustion air
preheat Spent shale zone:
spent shale residual carbon burned
cooling

Heat generation Combustion zone:


for pyrolysis residual carbon burning

Oxygen depleted
Pyrolysis zone:
kerogen conversion to
oil, gas, coke

Raw oil shale


preheat, off-gas
and oil mist Preheat zone
cooling

Low energy and


oil mist

Fig. 2. Nevada-Texas-Utah (NTU) retort schematic.

retorted shale supplies much of the fuel for this process. The GCR is the forerun-
ner of most continuous AGR processes (Table 7).
PETROSIX. The PETROSIX technology is operated in the IH mode using
hot recycle gas as the heat-transport medium. The PETROSIX retort has only
one level of heat input, uses countercurrent flows, and uses a circular grate to
control the flow of solids (Fig. 3). The PETROSIX has been operated by Petrobras
(Brazil) since the 1950s and is one of the few retorting processes producing shale
oil in 1995.
Paraho. The Paraho retorting technology is similar to the PETROSIX
technology except that it can be operated in the direct heat (DH) mode. The
unique feature of the Paraho technology is the two levels of heat input (Fig. 4).
In the IH mode, the air blower shown in Figure 4 is replaced by a recycle gas
heater. The Paraho DH operation has been carried out near Rifle, Colorado
since the 1970s; operations to produce asphalt from shale oil are continuing.
Tosco II. The Tosco II retorting technology, developed by The Oil Shale
Corporation, represents IH technology employing concurrent flow using fine-
sized feedstock. It was tested in Colorado from the late 1950s until the early
1980s. The unique feature of the Tosco II process is the use of ceramic balls as
the heat-transport medium (Fig. 5). These 125-mm (0.5-in.) balls, larger than the
Vol. 1 OIL SHALE 203

Oil shale feed


Seal
gas High heating
value gas
Feed hopper

Shale
distributor Electrostatic
precipitator Condenser
Cyclone
Separator Light
Heavy
shale
Pyrolysis shale Heavy Compressor oil
vessel Hot oil shale oil
gas
Discharge Waste
mechanism water

Cool gas

Retorted shale
facilities

Seal systems

Water Retorted shale slurry


to disposal

Fig. 3. PETROSIX process (20).

finely crushed shale feed, are separated from the retorted shale, recycled,
reheated, and reused in the process.
UNISHALE B. The UNISHALE process, like the Paraho process, uses
lump feed and countercurrent flows, and can be operated in either the DH or
IH mode. The UNISHALE B process is an IH process that uses hot recycled
gas as the heat-transport medium (Fig. 6). The unique feature of the UNISHALE
processes is the rock pump. The solids move upward through the retort as
the vapors are moving downward. The rock pump was used in the UNISHALE
technology at Parachute, Colorado to produce more than 0:64  106 m3 (four
million barrels) of crude shale oil. Operations were shut down in 1991.
Lurgi. The Lurgi process, developed by Lurgi-Ruhrgas GmbH for mild
gasification of coal, has been tested as an oil shale retort. The Lurgi process
is similar to the Tosco II process in that it uses finely divided oil shale as a
feedstock and employs indirect heating and concurrent flows. The unique feature
of the Lurgi process is the use of hot combusted retorted shale as the heat-transport
medium (Fig. 7) and retorting is carried out by mixing the hot combusted shale
and raw oil shale in a screw conveyor. Because a portion of the retorted shale is
combusted to supply the process heat, the Lurgi process utilizes more of the
organic matter in the oil shale than other IH processes (see Table 7).
Superior. The Superior retort is different from all the other AGR pro-
cesses in that it consists of a slowly rotating circular grate instead of the vertical
204 OIL SHALE Vol. 1

Raw shale Oil/gas


separators

Net product
Off-gas Gas
gas
Retort
Mist formation zone
Recycle gas
Recycle
blower
gas
Retorting zone Product oil

Top Top dilution gas


Retorted shale distributor
cooling zone
Top air
Mid- Mid-dilution gas
distributor
Mid-
Retorted shale Air
air
cooling zone
Air blower

Bottom cooling gas

Retorted shale

Fig. 4. Paraho DH process (21).

Flue gas
to atmosphere

Minus 1.27 cm Balls


Scrubber
raw shale
Gas

Ball Air
heater Fractionator Naphtha
Surge
Dilute phase fluid bed lift pipe

hopper Preh Gas oil


Ball elevator

ea 648
°
sha ted Ho C
le ball t Accum- Resid
s
lator
Pyrolysis
as

drum Trommel
eg

482°C
t flu

Wa
Ho

rm
bal
ls
Hot
spent
shale

200 mesh
Spent (ca 74 µm)
shale spent shale
cooler to disposal

Fig. 5. Tosco II process (20). The resid is residual oil.


Vol. 1 OIL SHALE 205

Raw Plant
shale in Hot recycled gas fuel gas
Recycled
Retort gas heater
gas
Oil
seal
Sulfer
R recovery
sh etor t plant
ale ed Venturi
ou scrubber Compressor
t Pump
Retorted oil Crude shale oil

Retorted shale

Retorted
shale
Upflow cooler
feeder

Fig. 6. UNISHALE B process (20).

shafts, rotating drums, or screw conveyors used by other technologies


(Fig. 8). The Superior technology is an adaptation of the circular grate sys-
tem used to calcine limestone. Raw oil shale is loaded onto the rotating cir-
cular grate which transports the solids through the same zones as shown in
Figure 2: solids preheat, retorting, char combustion, shale cooling, and air

Gas/solids Waste heat


separation bin

Recovery
Dust
cyclone

Oil Gaseous
Oil shale Waste and liquid
product
Condenser
Dust removal
Mixing screw cyclone
type retort Solids
surge
bin
Lift Solids
pipe to waste

Air + fuel
(if required)

Fig. 7. Lurgi process (20).


206 OIL SHALE Vol. 1

Cool recycle gas

Pre-heated
Hot recycled gas combustion air

Direct Air in
heating
low HV
burners

Carbon Cooling by Cooling


Shale Heating recovery recycle gas by air
feed Spent
in shale
out

Liquid
seal

Oil
removal
Gas
compressor
low heating-
value surplus
retort gas out
Oil and water out

Fig. 8. Superior process (20). HV ¼ heating value.

preheating. Because the gas composition varies within each of these zones, each
zone is separated by a baffle screen as the solids are transported around the
circle. The Superior retort was tested in Colorado in the late 1970s.
5.3. In Situ Retorting. True in situ retorting has been considered as a
means of avoiding the costs of mining, crushing, and surface disposal of spent
shale, and the associated environmental impacts of AGR. However, the imper-
vious nature of the oil shale formation and the overburden pressures have pre-
vented true in situ operations. Shale oil yields, the amount of oil produced
divided by the theoretical amount estimated to be in the oil shale rock, for
in situ retorting are usually half that experienced with AGR retorting. A true
in situ experiment, using drilling and resource fracturing procedures typical of
conventional petroleum development, was tried by the Energy Research Devel-
opment Administration (a forerunner of the U.S. Department of Energy) in 1975
in Rock Springs, Wyoming. No significant yields of shale oil were produced (22).
Other true in situ tests were conducted using the Equity BX superheated steam
process in Colorado, and Dow hot air process in Michigan; neither produced sig-
nificant yields of oil shale. It appears that true in situ retorting is not a practical
Vol. 1 OIL SHALE 207

Combustion

Gas
Air in Oil
Surface uplift

Preblast
surface

Overburden

Co
m

t
ga bu

fron
Oi
Oil Air flow l se stio
s n
shale Fire
bed

Fig. 9. LOFRECO process (20).

approach for the thick strata of oil shale normally situated deep below the
surface.
LOFRECO. The LOFRECO process, developed by Geokinetics, Inc.
(Utah) is a true in situ process. It is limited to relatively thin deposits of oil
shale situated beneath a relatively thin overburden (Fig. 9). The LOFRECO pro-
cess has been successful because the oil shale is rubblized in place by raising the
overburden. Retorting consists of direct combustion horizontally through the
rubblized formation, similar to that shown for the NTU retort (see Fig. 2).
Although the costs of mining and materials handling are obviated, the
LOFRECO process causes significant surface disturbance and results in oil
yields significantly lower than those obtained in the better controlled AGR
processes.
VMIS. The Vertical Modified in situ (VMIS) process consists of construct-
ing an underground retort of rubblized oil shale within the deep, thick deposits
situated in the Piceance Basin in western Colorado. In order to provide space for
the rubblization without upheaval of the overburden, a portion of the oil shale is
mined out and taken to the surface (Fig. 10). Retorting is carried out in the DH
mode exactly as shown in Figure 2. Steam and air are pumped into the top of the
VMIS retort, combustion proceeds down through the rubblized bed, and oil and
gas are pumped out from the bottom. Although yields are significantly lower
than those attained by AGR processes, the VMIS has demonstrated that modified
in situ can produce shale oil from thick deposits situated deep below the surface.
These operations have resulted in approximately one-fifth of the oil shale being
mined out to provide space for the VMIS rubblizing and retorting. A project
involving both VMIS and AGR processing, to utilize the mined-out shale, had
been planned, but was canceled in 1991 (24).

6. Crude Shale Oil

6.1. Properties. The composition of shale oil has depended on the shale
from which it was obtained as well as on the retorting method by which it was
208 OIL SHALE Vol. 1

Fig. 10. VMIS process (23).

produced. Properties of shale oils from various locations are given in Table 8. A
comparison of a Green River shale oil and a Michigan Antrim shale oil, retorted
under similar conditions, is given in Table 9.
Shale oil contains large quantities of olefinic hydrocarbons (see Table 8),
which cause gumming and constitute an increased hydrogen requirement for
upgrading. Properties for crude shale oil are compared with petroleum crude
in Table 10. High pour points prevent pipeline transportation of the crude
shale oil. Arsenic and iron can cause catalyst poisoning.
The primary difference in shale oils produced by different processing meth-
ods is in boiling point distribution. Rate of heating, as well as temperature level
and duration of product exposure to high temperature, affect product type and
yield (28). Gas combustion processes tend to yield slightly heavier liquid products
because of combustion of the lighter, ie, naphtha, fractions.
Carbon-to-hydrogen weight ratios for typical hydrocarbon fuels are natural
gas (methane), 3; gasoline, 6; crude oil, 6–7; shale oil, 7–8; Green River kerogen,
7; diesel and fuel oil, 8; residual oil, 10; and coal and coke, 12. A typical Green
River shale oil contains 40 wt% hydrocarbons and 60 wt% organic compounds
which contain nitrogen, sulfur, and oxygen. The nitrogen occurs in ring com-
pounds such as pyridines and pyrroles, as well as in nitriles, and these materials
comprise 60 wt% of the nonhydrocarbon organic components. Another 10 wt% of
these components is comprised of sulfur compounds which exist as thiophenes
and some sulfides and disulfides. The remaining 30 wt% is oxygen compounds
occurring as phenols and carboxylic acids (1).
6.2. Upgrading Shale Oil. Crude shale oil has a high (2 wt%) content
of organic nitrogen which acts as a catalyst poison, contains a large (20–50 wt%)
atmospheric residuum fraction, and has a high (>5 C) pour point (29,30).
Vol. 1 OIL SHALE 209

Table 8. Properties of Oils Produced from Shales


a
Analysis of distillate, wt%
Country or
company Retort Sp gr 8API N, wt% S, wt% Saturates Olefins Aromatics
Australia, Pumpherston 0.828 27.9 0.52 0.56 42 39 19
Glen Davisb
Brazil, gas combustion 0.919 22.5 1.06 0.68 23 41 36
Tremembeb
Franceb
Autun Pumpherston 0.931 20.5 0.90 0.51 33 36 31
Severac Marcecaux 0.925 21.5 0.53 3.0 30 32 38
Severac Petit 0.959 16.0 0.65 3.40 25 20 55
St. Hilaine Lantz 0.908 24.3 0.54 0.61 31 44 25
Scotlandb Pumpherston 0.874 30.4 0.77 0.35 42 39 19
South Africa, Salermo 0.906 24.7 0.85 0.64 35 44 21
Ermelo
Spain, Pumpherston 0.901 25.6 0.68 0.40 51 27 22
Puertollanob
Sweden, Rockesholm 0.977 13.3 0.68 1.65 12 24 64
Kvarntorpb
United States
Colorado gas combustion 0.943 18.6 2.13 0.69 27 44 29
Colorado Pumpherston 0.900 25.7 1.57 0.77 30 38 32
Superior 0.630 0.93 2.0 0.8 25 25 50
Shale Oilc,d
Rundle 0.636 0.91 0.99 0.41 48 2 50
Shale Oilc,e
Israeli Shale 0.623 0.955 1.2 7.1
Oilf
a
Boiling at 3158C.
b
Ref. 4.
c
Ref. 25.
d
Initial boiling point to 2048C.
e
Whole oil.
f
Ref. 26. Also contains 79.8 C, 9.7 H, and 2.2 wt% O.

Table 9. Comparison of Colorado and Michigan


Antrim Shale Oilsa
Property Colorado Michigan
naphtha, vol% 6.8 3.5
light distillate, vol% 24.9 41.1
heavy distillate, vol% 43.6 38.6
residuum, vol% 23.9 16.3
specific gravity (8API) 0.911 (23.8) 0.934 (20.0)
pour point, 8C 10 15
hydrogen, wt% 12.5 11.1
carbon, wt% 84.7 83.6
nitrogen, wt% 1.6 0.7
sulfur, wt% 0.8 3.5
a
Ref. 27.
210 OIL SHALE Vol. 1

Table 10. Comparison of Green River Crude Shale Oil and Median U.S. Crudea
U.S. Bureau of Mines
Gas Simulated Union Oil Median U.S.
Property Tosco II combustion In situ in situ Co. A crude
distillation boiling
point, 8Cb
ibpc200 18 6 11–15 7 5 30
200–315 24 19 41–48 31 20 22
315–480 34 38 27–35 46 40 28
>480 24 37 9–14 17 35 20
pour point, 8C 1d, 15e 21–28 1 to 5 10–15 32 < 15
specific gravity 0.927 0.934 0.892 0.910 0.940 0.850
8API 21 20 27 24 19 35
nitrogen, wt% 1.9 1.5–2.1 1.4–1.8 1.6 2.0 0.09
sulfur, wt% 0.7 0.8 0.7 0.6–0.9 0.9 0.6
oxygen, wt% 0.8 1.7 0.9
viscosity,
mm2 =s ð¼cStÞ
at 378C 22 59 8–15 21 46 6
1008C 04 07 06
saturates, vol% 30 59 41 60–100
olefins, vol% 50 16 37 <5
aromatics, vol% 20 25 22 0–40
carbon-to-hydro- 7–8 5–7
gen ratio
arsenic, ppm 40 <0:03
a
Refs. (1,27).
b
Values represent % of product.
c
ibp ¼ initial boiling point.
d
After a patented heat treatment which temporarily reduces pour point.
e
No heat treatment.

Prerefining crude shale oil to produce a synthetic crude that is compatible with
typical refineries generally is necessary (31–33). Prerefining to reduce organic
nitrogen content to low levels consists usually of either a delayed coking step
of the crude shale oil or residuum fraction, followed by one or more
hydrogenation steps, or a more severe direct hydrogenation of the crude shale
oil. Conditions for the hydrogenations are ca 4008C, 13.8 MPa (2000 psi) hydro-
gen partial pressure, and up to 356 standard cubic meters of hydrogen uptake
per cubic meter of shale oil (2000 ft3 at STP/bbl) (34). The nitrogen and sulfur
are converted to ammonia and elemental sulfur, and the hydrogen content of
the oil is increased. Upgraded shale oil is a desirable refinery feedstock. It is par-
affinic and characterized by low residuum, nitrogen, and sulfur (see FEEDSTOCKS).
Shale oil has been refined to produce gasoline, kerosene, jet fuel, and diesel
fuel (33). Different procedures have been tested to produce different product
states, eg, hydrotreating followed by hydrocracking for jet fuel production, hydro-
treating followed by fluid catalytic cracking for gasoline production, and coking
followed by hydrotreating for diesel fuel production. Production of military fuels
from the refining of 1590 m3 (10,000 bbl) of Paraho crude shale oil at the Gary
Western refinery in Colorado has been reported (35) and 15,900 m3 (100,000 bbl)
Vol. 1 OIL SHALE 211

Table 11. Properties of Shale Oil during the Unocal Upgrading Processa
Raw shale oil After particulates After arsenic After
from retort removal removal unicracking Syncrude
gravity, 8API 22 22 25 38
pour point, 8C 23.9 23.9 26.6 <3:9
particulates, 300 0 0 0
ppm
arsenic, ppm 25 25 0 0
distillation, 8C
initial 65.5 433
maximum 590 538
a
Ref. 36.

of Paraho shale oil has been processed under a U.S. Navy contract, at Sohio’s
Toledo refinery (33).
At the Parachute Creek Project, Unocal designed and operated an oil shale
upgrading unit to prerefine crude shale oil into syncrude, ie, upgraded shale oil
(36). The unit was designed to handle 1600 m3 (10,000 bbl) of crude shale oil per
stream day. More than 650,000 m3 (four million barrels) of syncrude were pro-
duced. Results of the Unocal shale oil upgrading process are given in Table 11.
The syncrude is compared with Arabian light crude oil in Table 12. Although
Arabian light is considered a premium crude oil among petroleum refiners, the
Unocal shale oil syncrude shows improvements in each of the characteristics
listed. Production of the conventional fuels and lubricants using Unocal shale
oil syncrude as the refinery feedstock is less difficult and less costly than using
Arabian light crude oil (see LUBRICATION AND LUBRICANTS).

7. Alternative Uses

Oil shale is an energy resource that produces a liquid fuel that can be used to
replace conventional crude oil or petroleum. However, the costs associated
with processing oil shale into conventional refined products are significantly
greater than that of processing conventional crude oil. In order to develop the
oil shale resource, other uses have been considered. These functions include

Table 12. Properties of Shale Oil Syncrude and


Arabian Light Crudea
Property Syncrude Arabian Light
gravity, 8API 40 34
sulfur, ppm wt 5 17,000
nitrogen, ppm wt 60 800
carbon residue, wt% 0.05 3.6
heavy metals, ppm wt 20
distillation, vol%
X–5388C 100 85
538 Cþ 15
a
Ref. 36.
212 OIL SHALE Vol. 1

direct combustion to produce process heat for power generation, direct gasifica-
tion of the oil shale geological deposit, and special petrochemical production.
7.1. Direct Combustion. Direct combustion of oil shale has been used
to produce heat for power generation at specific sites and is being carried out
in Estonia and Israel.
In Estonia, most of the rich oil shale, 209 L/t kukersite, is burned as a
solid fuel to produce electric power (37). The kukersite, although technically
an oil shale, is actually similar to a high ash, low grade coal ideally suited
for this use.
In the remote Negev desert region of Israel, oil shale is being burned in a
fluidized-bed combustor to supply process heat and produce electric power.
Unlike the Estonian kukersite, this oil shale is lean, about 63 L/t, but is being
mined to access an underlying phosphate deposit (26).
7.2. Gasification. For significant conversion of shale oil or oil shale to
gaseous products, considerable hydrogen must be used. Hydrogasification is
the main process under consideration for gasification of oil shale. Hydroretorting
of oil shale has been studied extensively (38,39). Gasification of Colorado oil shale
in hydrogen and synthesis gas has been carried out at 524–7608C and at up to
38.3 MPa (5540 psig). Another study involves conventional oil shale retorting fol-
lowed by gasification of the resulting shale oil in a fluidized bed (40). The hydro-
gen is supplied by gasification of the coke on the spent shale and from the
gasifier. Other hydrogasification work includes the processing of Green River
and eastern Devonian oil shales (41,42). Because of their character, the eastern
Devonian shales produce less product oil, ca 35 wt% Fischer assay, than do
Green River shales. The hydrogasification process is claimed to recover about
90 wt% of the kerogen content. Additional oil shale gasification research at the
Laramie Energy Technology Center (U.S. DOE) has been reported (43–45).
7.3. Petrochemicals Production. Early Consideration. In the
1950s, the U.S. Bureau of Mines studied pyrolysis of both oil shale and shale
oil for the production of light olefins. High temperature retorts were used to
determine the effect of continuing the cracking, which is begun when the kerogen
is converted to shale oil (46). Low temperature shale oils are low in aromatic con-
tent, thus one goal has been the production of an aromatic-rich naphtha. High
temperature retorting of this type on Green River oil shales has the disadvantage
of the additional energy requirements of the endothermic carbonate decomposi-
tion, with over 50 wt% decomposition occurring at 8158C. This effect has little
importance for low carbonate oil shale, eg, the eastern U.S. Antrim oil shales
(41). Comparison of thermal cracking of conventional shale oil to high tempera-
ture retorting under the same conditions illustrates that naphtha production is
enhanced considerably by high temperature retorting (46).
Utilization of shale oil products for petrochemical production has been stu-
died (47–51). The effects of prerefining on product yields for steam pyrolysis of
shale oil feed and the suitability of Green River shale oil as a petrochemical feed-
stock were investigated. Pyrolysis was carried out on the whole oil, vacuum
distillate, and mildly, moderately, and severely hydrogenated vacuum distillates.
Specialty Chemicals. Specialty chemicals have been considered an eco-
nomically attractive means of using the oil shale resource. These specialty che-
micals consist of high value, niche market items (52) that utilize the high
Vol. 1 OIL SHALE 213

Table 13. Tensile Strength Retained, %a


Number of freeze–thaw cycles
Asphalt 1 3 5 9
AC-10 85 79 72 48
SOMAT 85 96 87 84
a
Ref. 56.

concentration of heteroatoms (nitrogen, sulfur, and oxygen) found in most crude


shale oils (see Table 8). The use of shale oil with its complex, high molecular
weight, low pour point resid materials, and high concentration of various func-
tional groups could be used to produce waxes, aromatic lubricating oils, sulfonate
feeds, substitutes for coal-tar acids and bases, resins, and special organic inter-
mediates (53). Revenue that may be achieved from shale oil would be greatly
enhanced if used in their production.
Shale Oil Asphalt. The New Paraho Corporation has been producing
asphalt (qv) made from crude shale oil (54,55). This shale oil asphalt, SOMAT,
represents a specialty product that utilizes many of the properties of crude
shale oil that would reduce its value as a refinery feedstock, ie, low pour point,
high boiling point, and large quantities of heteroatoms and organic functional
groups, especially basic nitrogen groups. These properties reduce the value of
crude shale oil as a refinery feedstock, but tend to produce an improved asphalt.
SOMAT is superior to conventional petroleum asphalt (AC-10) in terms of tensile
strength retained after freeze–thaw cycles (Table 13).
Shale oil asphalt also meets or exceeds the performance of the improved,
but more costly, polymer-based asphalt. Since 1989, more than 8 km (5 mi) of
test strips of SOMAT have been placed on various roadways in seven U.S. states.
The SOMAT has demonstrated marked improvement over conventional petro-
leum-based asphalt. These assessments are continuing. As of this writing
(ca 1995), Paraho is preparing preliminary designs for an oil shale facility to
produce about 325 m3/d (2000 bbl/d) of the shale oil modifier used to produce
SOMAT (56).

8. Environmental Issues

The plans to develop a commercial oil shale industry in the three-state region of
Colorado, Utah, and Wyoming in the 1970s raised the possibility of significant
adverse environmental, health, safety, and socioeconomic (EHSS) impacts. Pro-
cessing oil shale to produce oil on a large-scale commercial basis requires a large
amount of mining, crushing, material transport, and disposal operations.
Adverse EHHS impact could result from uncontrolled, or inadequately con-
trolled, large-scale oil shale operations. Without controls, significant amounts of
dust, ie, particulates, would be produced. Because the gas produced from kerogen
breakdown contains significant amounts of hydrogen sulfide and ammonia,
uncontrolled release, or direct combustion with no control technology, could
pose adverse health impacts and air pollution. The liquids produced from retort-
ing operations, ie, process water and crude shale oil, contain significant levels of
214 OIL SHALE Vol. 1

Table 14. Estimate of Colony Oil Shale Project Emissions Production, kg=ha,b
Solid Hydro- Carbon
Source SO2 NOx particulates carbons monoxide
crushing and conveying
primary crusher dust 0 0 27 0 0
collection system
final crusher dust 0 0 136 0 0
collection system
fine-ore storage dust 0 0 27 0 0
collection system
pyrolysis and oil recovery unit
preheat systems 529 2172 51 128 17.4
steam superheater ball 9 56 103 0.2 1.0
circulation systems
processed shale 0 0 116 0 0
moisturizing systems
hydrogen unit
reforming furnaces 146 244 4.9 0.8 4.6
gas–oil hydrogenation unit
reactor heaters 10 54 1.1 0.4 0.45
reboiler heater 2.7 17.7 0.4 0.04 0.3
naphtha hydrogenation unit
reactor heater 2.2 4 0.09 0.01 0.09
sulfur recovery unit
sulfur plants with 29
common tailgas plant
delayed coker unit
heater 21 35 0.7 0.14 0.64
utilities
boilers 51 246 5.4 1.8 1.4
Total 799.9 2828.7 472.59 131.39 25.88
a
Ref. 57.
b
Using no control technology.

toxic metals, suspected or known carcinogens, and other hazardous materials.


Discharge of this water would thus require treatment. Combusting and/or refin-
ing the crude shale oil would also require adequate treatment and environmental
controls. The large quantities of materials involved in oil shale development
means that disposal of the retorted shale poses special problems. Proper controls
are needed to avoid significant air pollution from dust emissions, and surface and
groundwater contamination from leaching and runoff. The amount of water
required for commercial oil shale operations poses water quality impact on the
semiarid region of Colorado, Wyoming, and Utah. Engineering technology was
thus developed for oil shale operations. Most predictions of significantly adverse
EHHS impacts (Table 14) were based on assumptions from earlier foreign opera-
tions and impacts from similar industries (57–61), and were not realized during
the 1975–1990 oil shale boom.
8.1. Air Pollution. Particulates and sulfur dioxide emissions from com-
mercial oil shale operations would require proper control technology. Compliance
monitoring carried out at the Unocal Parachute Creek Project for respirable
particulates, oxides of nitrogen, and sulfur dioxide from 1986 to 1990 indicate
a þ99% reduction in sulfur emissions at the retort and shale oil upgrading
Vol. 1 OIL SHALE 215

1600 100

Density, kg/m3 Optimum water


1520 95

Density, lb/ft3
No added water
1440 90

1360 85
500 1500 2500
Compactive effort, kJ/m3

Fig. 11. Density of retorted shale vs compactive effort (64).

facilities. No violations for unauthorized air emissions were issued by the U.S.
Environmental Protection Agency during this time (62).
8.2. Water Quality. All commercial oil shale operations require substan-
tial quantities of water. All product water is treated for use and operations are
permitted as zero-discharge facilities. In the Unocal operation, no accidental
releases of surface water have occurred during the last four years of sustained
operations from 1986 to 1990. The Unocal Parachute Creek Project compliance
monitoring program of ground water, surface water, and process water
streams have indicated no adverse water quality impacts and no violations of
the Colorado Department of Health standards (62).
8.3. Solids. Proper handling and disposal techniques can obviate poten-
tial problems associated with the solid waste-retorted shale. Retorted shale dis-
posal and revegetation have posed no adverse environmental impacts at the
Unocal Parachute Project (62). Earlier studies carried out using Paraho and
Lurgi retorted shales indicated that these materials behave as low grade
cements (63,64) and can be engineered and compacted into high density materi-
als (Fig. 11) and water impervious structures (Table 15).

Table 15. Permeability of Retorted Shalea


Permeability cm=s  10 6
Loading, No Optimum
Compaction kPab water water
standard, 593 kJ/m3c 345 43.0 6.8
690 29.3 1.4
1380 19.1 0.8
heavy, 2693 kJ/m3c 345 38.1 1.1
690 32.4 0.6
1380 25.2 0.1
a
Ref. 64.
b
To convert kPa to psi, multiply by 0.145.
c
To convert kJ to kcal, divide by 4.184.
216 OIL SHALE Vol. 1

8.4. Health and Safety. Much of the adverse health issue publicity invol-
ving risks of exposures to carcinogens, such as benzo-a-pyrene, have been based on
recorded exposures of Scottish oil shale workers that took place nearly 100 years
ago. It is believed that the increase in cancer was due more to poor personal
hygiene than exposure to shale oil. Industrial hygiene monitoring and health sur-
veys indicate no significantly increased health risks among oil shale workers (62).
8.5. Socioeconomics. Impact from recurring boom-and-bust cycles
typified many of the earlier mining developments in the western United States
(65). However, this and the Oil Shale Trust Fund, established by the legislation
that set up the Federal prototype leases (C-a, C-b, U-a, and U-b), has provided
funds to local towns and counties to assist in the construction and upgrading
of infrastructures. These infrastructures are needed to accommodate the work
force needed to construct and operate the large oil shale processing facilities.
Further, companies such as Unocal have provided socioeconomic grants to
these counties (66).

9. Economic Aspects

As of 1995, there were only a few commercial oil shale facilities operating in the
world. These facilities are located in countries where the economic, political, and
environmental requirements for commercial oil shale development are met.
There are commercial oil shale facilities in Brazil, China, Estonia, and Israel.
No commercial oil shale facilities have existed in the United States because
the costs of shale oil processing exceed those associated with conventional petro-
leum crude processing.
In the United States, estimates of oil shale retorting have ranged from
$113/m3 ($18/bbl) to $567/m3 ($90/bbl). The lower estimate is based on using
the actual costs of constructing a commercial oil shale retorting facility and
using a high grade (>25 L=t ð>30 gal=short tonÞ) of oil shale (3). The higher
estimate is based on conservative estimates utilizing unproven or noncommercial
oil shale retorting technology.
The estimated costs for upgrading crude shale oil range from $38/m3 ($6/bbl)
to $63/m3 ($10/bbl). However, the resulting upgraded shale oil is superior to most
conventional crude petroleum and is more valuable as a refinery feedstock (67).
The costs for upgrading crude shale oil depend on the upgrading techniques, ie,
hydrotreating and coking or hydrotreating and fluid catalytic cracking. However,
the greatest economic factor in oil shale upgrading is the amount of hydrogen
required as reflected in the concentration of heteroatoms such as nitrogen,
sulfur, and oxygen.
The commercial production of shale oil as an alternative energy source has
not been economically feasible. As of 1995, all commercial oil shale operations in
the world (Petrobras, Brazil; PAMA, Israel; The Chinese Petroleum Corporation,
Fushun and Maoming, China; Kivioli Oil Shale Processing Plant, Kohtla-Jarve,
Estonia) receive some sort of economic incentives or assistance from the
countries in which they are operating.
The first stage of the Stuart oil shale project near Gladstone, Australia,
6000 t/d (6600 short tons/d), is scheduled to be constructed by Southern
Vol. 1 OIL SHALE 217

Pacific Petroleum. Financial assistance from the Australian government, con-


sisting of special depreciation incentives and exemption of gasoline taxes
equivalent to about U.S. $1.91/m3 of crude shale oil ($12.00/bbl) has been
assured (68).
As a result of the 1980 Energy Security Act, the United States Synthetic
Fuels Corporation (USSFC) was established to provide financial assistance in
the development of alternative energy sources to reduce the dependence on for-
eign petroleum to meet the needs for liquid fuels. More than $15 billion was
authorized for financial assistance to those projects having the potential for pro-
ducing about 318,000 m3/d of crude shale oil (two million bbl/d) by 1992. Whereas
letters of intent to negotiate for assistance were authorized by USSFC for three
oil shale projects, ie, Cathedral Bluffs, Occidental Oil Shale, Inc. (Rio Blanco,
Colorado); Seep Ridge, Geokinetics, Inc. (Vernal, Utah); and Parachute Creek,
Union Oil Company (Parachute, Colorado) (69), the only oil shale project to
receive financial assistance from the USSFC was the $654 million Parachute
Creek project which received about $114.5 million assistance in the form of a
price guarantee of $56–73/bbl from July 1983 until Union Oil Company ceased
operations in Parachute in June 1991. Less than one-third of the available funds
had been utilized (70).
The Deficit Reduction Act of 1984 significantly curtailed the available
funding for the USSFC, which ceased operations in 1985.

10. Commercial Operations

The number of commercial oil shale operations worldwide has decreased signifi-
cantly since the decade 1975–1985 and are producing only a fraction of the
world’s liquid fuels’ needs. Most commercial oil shale operations have been scaled
back.
10.1. PETROSIX Operations in Brazil. Petroleo Brasilerio (Petrobras)
has a dedicated facility to produce crude shale oil from the Irati formation in
southern Brazil. The facility is called the Oil Shale Industrialization Superinten-
dency (SIX) and uses the PETROSIX retorting technology (see Table 7 and
Fig. 3).
During its 40-year development, three different sizes of PETROSIX retorts
have been operated on a continuous basis: a 1.83-m (6-ft) diameter demonstra-
tion plant; a 5.49-m (18-ft) diameter Prototype Unit (UPI); and a 10.97-m
(36-ft) diameter Industrial Module (MI). Within the SIX facility are numerous
pilot plants available for retorting coarse-sized oil shale, fines utilization, and
oil shale upgrading (3,67).
The UPI and MI retorts are processing 7000 t/d (7700 short tons/d) of Irati
shale to produce 24,381 m3/d (3870 bbl) of shale as well as 80 t (55 short tons) of
LPG, 132 t (145 short tons) of clean fuel gas, and 98 t (108 short tons) of sulfur.
The SIX plant has reached its design rate (Table 16) in an energy efficient
manner with a high on-stream (operating) factor.
Pilot-plant studies have been conducted by Petrobras on many of the differ-
ent oil shales from around the world. Tests indicate that many oil shales could be
processed using the PETROSIX retorting technology (67) (Fig. 12).
218 OIL SHALE Vol. 1

Table 16. MI-Main Operation Dataa


energy consumption/energy produced 0.38
oil yields, % 87–90
gas yields, % 140–150
operation factor, % 88–90
retorting rate, kg=ðhm2 Þ 2300–2900
a
Ref. 3.

10.2. Oil Shale Operations in Israel. Oil shale, the only fossil fuel
resource in Israel, is being used to generate electric power. The oil shale feed
stock, typical of the low grade Israeli oil shale (see Table 2), is situated in a
deposit overlying phosphate ore. The oil shale operations are being carried out
because the oil shale has to be mined to obtain the phosphate ore.
A circulating fluidized-bed boiler, using raw shale oil as a feedstock, is being
used to supply process heat for the phosphate operations and to operate a
100-MW power plant. Scale-up in the 1990s should increase the electric power
generation to 1000 MW (71).
10.3. United States. In 1980, Unocal began constructing the Parachute
Creek Project, designed to produce 1600 m3 (10,000 bbl) of upgraded shale oil per
day. The project included a conventional underground room-and-pillar mine, the
Unishale B (see Table 7) retort, and a special Unocal upgrading facility. Plant
startup occurred in 1986, and daily shale oil production reached 1100 m3/d
(7000 bbl/d). By 1991, total production exceeded 0:6  106 m3 (four million
barrels). However, the Parachute Creek Project was shut down in mid-1991 for
economic reasons.
The New Paraho Corporation has been conducting research on asphalt
derived from shale oil, SOMAT, at its pilot plant (Rifle, Colorado) (54,55). It is
the only active oil shale operation in the United States as of 1995. New Paraho
is continuing its pilot operations while designing a commercial facility to produce
SOMAT. The economics appear promising (56).

Utah (U.S.)
Wyoming (U.S.)
Colorado 1 (U.S.)
Colorado 2 (U.S.)
Indiana (U.S.)
Kentucky (U.S.)
China
Rundle (Australia)
Nagoorin (Australia)
Condor (Australia)
Estonia Oil content, wt % Abrasion index
Israel
Canada
10 20 30 40 50 60 70 80

Fig. 12. Shale oil characteristics compared to Irati oil shale (3).
Vol. 1 OIL SHALE 219

10.4. Other Oil Shale Operations. As of this writing, commercial pro-


duction of shale oil is still being conducted in the People’s Republic of China and
Estonia. However, production rates continue to dwindle owing to the availability
of conventional petroleum and other sources of energy as well as continued
worldwide energy conservation.
Commercial shale oil production in the People’s Republic of China is
still being carried out in Fushin and Maoming. However, production in both
areas is declining because of continued emphasis on conventional petroleum
and coal. Annual shale oil production in China is about 10  106 m3 ð1:6
106 barrelsÞ (72).
Shale oil production in the former Soviet Union is also declining. The only
significant shale oil operations are in Estonia. Most of the rich (208 L/t (50 gal/
short ton)) Estonian oil shale is combusted directly as fuel.
Plans are underway to develop commercial shale oil operations in Austra-
lia. Southern Pacific Petroleum, N.L. is planning a commercial oil shale project
utilizing the Stuart deposit (Brisbane, Australia). Favorable economics
are attained by tax incentives to the Stuart project in the form of increased
depreciation writeoffs and exempting excise tax for gasoline produced from
shale oil. In Stage 1 of the Stuart project, expected to be operational in
1996, surface mining is to be used with AGR Taciuk retorting (see Table 7)
to produce about 675 m3 (4250 barrels) of hydrotreated naphtha and fuel oil
per stream day. In the full-scale, Stage 3 of the Stuart project, daily production
is estimated to be nearly 10,000 m3 (about 60,000 barrels) of upgraded shale oil
syncrude (73).

BIBLIOGRAPHY

‘‘Shale Oil’’ in ECT 1st ed., Vol. 12, pp. 207–221, by B. Guthrie and H. M. Thorne, Bureau
of Mines, U.S. Dept. of the Interior; in ECT 2nd ed., Vol. 18, pp. 1–20, by R. E. Gustafson,
Cameron and Jones, Inc.; ‘‘Oil Shale’’ in ECT 3rd ed., Vol. 16, pp. 333–357, by P. F. Dickson,
Colorado School of Mines.

CITED PUBLICATIONS

1. S. Siggia and P. Uden, eds., Analytical Chemistry Pertaining to Oil Shale and
Shale Oil, National Science Foundation Grant Number GP 43807, June 24–25,
1974, 11–13.
2. D. C. Duncan and V. E. Swanson, U.S. Geol. Surv. Circ. 523(7), 9 (1965).
3. D. L. Bachman and co-workers, 1993 Eastern Oil Shale Symposium, Lexington, Ky.,
Nov. 16–19, 1993.
4. H. M. Thorne and co-workers, U.S. Bur. Mines Inf. Circ. 8216, 6 (1964).
5. G. L. Baughman, ed., Synthetic Fuels Data Book, 2nd ed., Vol. 4, Cameron Engineers
(Division of The Pace Co.), Denver, Colo., 1978, 67–104.
6. Oil Shale and the Environment, EPA-600/9-77-033, U.S. Environmental Protection
Agency, Washington, D.C., Oct. 2, 1977.
7. ASTM D3904, American Society for Testing and Materials, Philadelphia,
Pa., 1980.
220 OIL SHALE Vol. 1

8. R. N. Heistand, Am. Chem. Soc. Div. Fuel Chemis. Preprint 21(6), (1976).
9. L. Goodfellow and M. T. Atwood, Colo. Sch. Mines Q. 69, 205 (Apr. 1974).
10. K. E. Stanfield, U.S. Bur. Mines Rep. Invest. 4825, (1951).
11. V. D. Allred, Colo. Sch. Mines Q. 62, 657 (1967).
12. J. J. Cummins and W. E. Robinson, U.S. Bur. Mines Rep. Invest. 7620, 1 (1972).
13. R. L. Braun and A. J. Rothman, Fuel 54, 129 (Apr. 1975).
14. J. H. Campbell and co-workers, paper presented at The 10th Oil Shale Symposium
Proceedings, Colorado School of Mines Press, July 1977, 148–165.
15. A. K. Burnham and co-workers, Comp. Stud. Various Oil Shales 34(1), 36–42
(Mar. 1989).
16. A. K. Burnham, Relationship Between Hydrous and Ordinary Pyrolysis, UCRL-
JC-114130, Lawrence Livermore National Laboratory, Livermore, Calif., June
1993.
17. S. S. Tihen, Ind. Eng. Chem. 47, 464 (1955).
18. G. U. Dineen, ‘‘Effect of Retorting Temperature on the Composition of Shale Oil,’’
AlChE Meeting, Denver, Colo., Aug. 1962.
19. J. H. Campbell and A. K. Burnham, paper presented at The 11th Oil Shale Sympo-
sium Proceedings, Colorado School of Mines Press, Golden, Colo., Nov. 1978,
242–259.
20. E. M. Piper, 6th IIASA Resource Conference, Luxenborg, Austria, Colorado School of
Mines Press, Golden, Colo., 1981.
21. J. B. Jones, Jr. and R. N. Heistand, Proceedings, 12th Oil Shale Symposium,
Colorado School of Mines Press, Golden, Colo., 1979, 184–195.
22. Energy Development Consultants, ‘‘Oil Shale in the United States, 1981,’’ EDC,
Golden, Colo., 1980.
23. A. Stevens and R. L. Zahradnik, Proceedings, 16th Oil Shale Symposium, Colorado
School of Mines Press, Golden, Colo., 1983, 267–268.
24. L. Dockter, Colo. Sch. Mines Q. 83(4), 120–122 (1988).
25. R. F. Crane, in Ref. 21, 1–16.
26. A. H. Pelofsky and co-workers, in Ref. 21, 32–42.
27. A. Long, Jr., N. W. Merriam, and C. G. Mones, in Ref. 15, 120–135.
28. N. D. South and co-workers, Colo. Sch. Mines Q. 71, 153 (Oct. 1976).
29. R. E. Poulson, C. M. Frost, and H. B. Jensen, ACS Div. Pet. Chem. Prepr. 17(2), 175
(1972).
30. M. T. Atwood, paper presented at The ACS Symposium: Fuels of the Future, Dallas,
Tex., Apr. 4–13, 1973.
31. C. M. Frost, R. E. Poulson, and H. B. Jensen, ACS Div. Pet. Chem. Prepr. 17(2), 156
(1972).
32. D. P. Montgomery, Ind. Eng. Chem. Prod. Res. Dev. 7, 274 (1968).
33. R. F. Sullivan, B. E. Stangeland, and H. A. Frumkin, ACS Div. Pet. Chem. Prepr. 22,
998 (1977).
34. V. F. Yesavage, C. F. Griswold, and P. F. Dickson, paper presented at The 180th
National ACS Meeting, San Francisco, Calif., Aug. 25–29, 1980.
35. H. Batrick and co-workers, Final Report—The Production and Refining of Crude
Shale Oil into Military Fuels, Applied Systems Corp. for Office of Naval Research,
Washington, D.C., Aug. 1975.
36. C. P. Reeg, A. C. Randle, and J. H. Duir, Proceedings, 23rd Oil Shale Symposium,
Colorado School of Mines Press, Golden, Colo., 1990, 68–95.
37. J. D. Baker and C. O. Hook, in Ref. 21, 26–31.
38. H. F. Feldman and co-workers, Inst. Gas Technol. Res. Bull. 36 (Aug. 1966).
39. S. A. Weil and co-workers, paper presented at The 167th National ACS Meeting, Los
Angeles, Calif., Apr. 1974.
Vol. 1 OIL SHALE 221

40. U.S. Pat. 3,703,052 (1972), H. R. Linden (to Institute of Gas Technology).
41. F. C. Schora and co-workers, Hydrocarbon Process. 56(4), 107 (1977).
42. F. C. Schora, ‘‘The Application of the IGT Hydroretorting Process to Eastern Shale,’’
Energy Topics—a Supplement to IGT Highlights, Institute of Gas Technology,
Chicago, Ill., May 9, 1977.
43. E. L. Burwell and I. A. Jacobson, Jr., ‘‘Concurrent Gasification and Retorting of Oil
Shale—a Dual Energy Source,’’ Rocky Mountain Regional Meeting, SPE 5535,
Society of Petroleum Engineers (SPE), AIME, Denver, Colo., Apr. 7–9, 1975.
44. E. L. Burwell and I. A. Jacobson, Jr., U.S. Bur. Mines Tech. Prog. Rep. 85, 1
(Nov. 1974).
45. E. L. Burwell and I. A. Jacobson, Colo. Sch. Mines Q. 71, 139 (Oct. 1976).
46. H. W. Sohns and co-workers, Ind. Eng. Chem. 47, 461 (1955).
47. C. F. Griswold, V. F. Yesavage, and P. F. Dickson, ACS Div. Fuel Chem. Prep. 21, 207
(1976).
48. E. A. Fritzler, V. F. Yesavage, and P. F. Dickson, Proceedings, The Second Pacific
Chemical Engineering Conference, American Institute of Chemical Engineers, New
York, 1977, p. 542.
49. P. D. Smith, P. F. Dickson, and V. F. Yesavage, ACS Div. Pet. Chem. Prepr. 23,
756 (1978).
50. C. F. Griswold, A. Ballut, H. R. Kavianian, P. F. Dickson, and V. F. Yesavage, Chem.
Eng. Prog. 75(9), 78 (1979).
51. C. F. Griswold, A. Ballut, H. R. Kavianian, P. F. Dickson, and V. F. Yesavage, Energy
Commun. 6(2), 153 (1980).
52. J. E. Sinor, ‘‘Niche Market Assessment for a Small Western Oil Shale Project,’’ Final
Report, DE-FC-86MC11076, U.S. Dept. of Energy, Washington, D.C., 1979.
53. J. E. Bunger and A. V. Deveni, Proceedings, 25th Oil Shale Symposium, Colorado
School of Mines Press, Golden, Colo., 1992, 281–294.
54. L. A. Lukens, Proceedings, 22nd Oil Shale Symposium, Colorado School of Mines
Press, Golden, Colo., 1989, 196–206.
55. L. A. Lukens and M. A. Plummer, Colo. Sch. Mines Q., 108–114 (1988).
56. L. A. Lukens, Col. Sch. Mines Q., 115–119 (1988).
57. An Environmental Impact Analysis for a Shale Oil Complex at Parachute Creek,
Colorado, Vol. I, II, III, Colony Development, Denver, Colo., 1974.
58. Detailed Development Plan, Vol. 1 and 2, Oil Shale Lease Tract C-b, C-b Shale Oil
Project, 1976.
59. Modifications to Detailed Development Plan, Oil Shale Tract C-b, C-b Shale Oil
Venture, 1977.
60. Project Independence Oil Shale Task Force Report, FEA, Washington, D.C., 1974,
p. 154.
61. T. L. Thoem and E. F. Harris, in Ref. 19, 1–9.
62. J. B. Benton, FUEL 71, 238–242 (Feb. 1992).
63. J. P. Fox, Proceedings of the 13th Oil Shale Symposium, Colorado School of Mines
Press, Golden, Colo., 1980, 131–139.
64. R. N. Heistand and W. G. Holtz, in Ref. 63, 140–150.
65. A. Gulliford, Boomtown Blues: Colorado Oil Shale, 1885–1985, University Press of
Colorado, Boulder, 1989.
66. J. Evans, Colo. Sch. Mines Q., 103–105 (1988).
67. E. M. Piper and co-workers, Proceedings of the 25th Oil Shale Symposium, Colorado
School of Mines Press, Golden, Colo., 1992, 221–242.
68. 1992 Annual Report, Southern Pacific Petroleum N.L., 1993.
69. 1984 Annual Report, United States Synthetic Fuels Corp., 1985.
70. Oil Gas J., 38 (Apr. 8, 1991).
222 TAR SANDS Vol. 1

71. J. Yerushalmi, Proceedings of the 1992 Eastern Oil Shale Symposium, IMMR Press,
Lexington, Ky., 1993, p. 367.
72. C. Du, Proceedings of the 18th Oil Shale Symposium, Colorado School of Mines
Press, Golden, Colo., 1985, 210–215.
73. B. C. Wright, Alternate Energy ’89 Proceedings, Council on Alternate Fuels,
Washington, D.C., 1989, 175–194.

EDWIN M. PIPER
Piper Designs LLC
ROBERT N. HEISTAND
Consultant

TAR SANDS
In addition to conventional petroleum (qv) and heavy crude oil, there remains
another subclass of petroleum, one that offers to provide some relief to poten-
tial shortfalls in the future supply of liquid fuels and other products. This
subclass is the bitumen found in tar sand deposits (1,2). Tar sands, also
known as oil sands and bituminous sands, are sand deposits impregnated
with dense, viscous petroleum. Tar sands are found throughout the world,
often in the same geographical areas as conventional petroleum.
Petroleum, and the equivalent term crude oil, cover a vast assortment of
materials consisting of gaseous, liquid, and solid hydrocarbon-type chemical com-
pounds that occur in sedimentary deposits throughout the world (3). When pet-
roleum occurs in a reservoir that allows the crude material to be recovered by
pumping operations as a free-flowing dark- to light-colored liquid, it is often
referred to as conventional petroleum.
Heavy oil is another type of petroleum, different from conventional petro-
leum insofar as the flow properties are reduced. A heavy oil is much more diffi-
cult to recover from the subsurface reservoir. These materials have a high
viscosity and low API gravity relative to the viscosity and API gravity of conven-
tional petroleum (Fig. 1) (3, 4), and recovery of heavy oil usually requires thermal
stimulation of the reservoir.
The definition of heavy oil is usually based on API gravity or viscosity, but
the definition is quite arbitrary. Although there have been attempts to rationa-
lize the definition based on viscosity, API gravity, and density (2,3), such defini-
tions, based on physical properties, are inadequate, and a more precise definition
would involve some reference to the recovery method.
In a general sense, however, the term heavy oil is often applied to a petro-
leum that has a gravity <208API. The term heavy oil has also been arbitrarily
used to describe both the heavy oil that requires thermal stimulation for recovery
from the reservoir and the bitumen in bituminous sand (also known as tar sand
or oil sand) formations, from which the heavy bituminous material is recovered

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.20011819160509.a01
Vol. 1 TAR SANDS 223

106

105

Athabasca
4
10
Viscosity, cps

103 Cold lake

102

Heavy oil
(Lloydminster)
10

Light crude oil

1
50 100 150
Temperature, °C

Fig. 1. Relative viscosity data for conventional petroleum, heavy oil, and bitumen.

by a mining operation. Extra heavy oil is the subcategory of petroleum


that occurs in the near-solid state and is incapable of free flow under ambient
conditions. The bitumen from tar sand deposits is often classified as an extra
heavy oil.
Tar sand, also variously called oil sand (in Canada) or bituminous sand, is
the term commonly used to describe a sandstone reservoir that is impregnated
with a heavy, viscous black extra heavy crude oil, referred to as bitumen (or,
incorrectly, as native asphalt). Tar sand is a mixture of sand, water, and bitu-
men, but many of the tar sand deposits in the United States lack the water
layer that is believed to cover the Athabasca sand in Alberta, Canada, thereby
facilitating the hot-water recovery process from the latter deposit. The heavy
asphaltic organic material has a high viscosity under reservoir conditions and
cannot be retrieved through a well by conventional production techniques.
It is incorrect to refer to bitumen as tar or pitch. Although the word tar is
somewhat descriptive of the black bituminous material, it is best to avoid its
use in referring to natural materials. More correctly, the name tar is usually
applied to the heavy product remaining after the destructive distillation of
coal or other organic matter. Pitch is the distillation residue of the various
types of tar.
Physical methods of fractionation of tar sand bitumen usually indicate
high proportions of nonvolatile asphaltenes and resins, even in amounts up to
50% wt/wt (or higher) of the bitumen. In addition, the presence of ash-forming
224 TAR SANDS Vol. 1

metallic constituents, including such organometallic compounds as those of


vanadium and nickel, is also a distinguishing feature of bitumen.
Asphalt is prepared from petroleum and often resembles bitumen. When
asphalt is produced simply by distillation of an asphaltic crude, the product
can be referred to as residual asphalt or straight-run petroleum asphalt. If the
asphalt is prepared by solvent extraction of residua or by light hydrocarbon (pro-
pane) precipitation, or if blown or otherwise treated, the term should be modified
accordingly to qualify the product, eg, propane asphalt.

1. Origin of Bitumen

There are several general theories regarding the origin of the bitumen. One the-
ory is that the oil was formed locally and has neither migrated a great distance
nor been subjected to large overburden pressures. Because under these condi-
tions the oil cannot have been subjected to any thermal effects with the resulting
decomposition or molecular changes, it is geologically young and therefore dense
and viscous.
Another theory promotes the concept of a remote origin for the bitumen, or,
more likely, the bitumen precursor, both geographically and in geological time.
The bitumen precursor, originally resembling a conventional crude oil, is
assumed to have migrated into the sand deposit, which may originally have
been filled with water. After the oil migrated, the overburden pressures were
relieved, and the light portions of the crude evaporated, leaving behind a
dense, viscous residue.
Included in the remote origin theory is the postulate that the light hydro-
carbons were destroyed by bacteria carried into the petroleum reservoirs in oxy-
genated, meteoric waters. The remote origin theory would explain the water
layer surrounding sand grains in the Athabasca deposit. However, because the
metals and porphyrin contents of bitumen are similar to those of some conven-
tional Alberta crude oils of Lower Cretaceous age and because Athabasca bitu-
men has a relatively low coking temperature, the bitumen may be of Lower
Cretaceous age. This is the age of the McMurray formation (Canada), which is
geologically young. This evidence supports the theory that the oil was formed
in situ and is a precursor, rather than a residue of some other oil. The issue
remains unresolved as of this writing (ca 1997).

2. Occurrence

Many of the reserves of bitumen in tar sand formations are available only with
some difficulty, and optional refinery methods are necessary for future conver-
sion of these materials to liquid products, because of the substantial differences
in character between conventional petroleum and bitumen (Table 1).
Because of the diversity of available information and the continuing
attempts to delineate the various world oil sands deposits, it is virtually impos-
sible to reflect the extent of the reserves in terms of barrel units with a great
degree of accuracy. The potential reserves of hydrocarbon liquids that occur
Vol. 1 TAR SANDS 225

Table 1. Bitumen vs Conventional Petroleum Properties


Property Bitumen Conventional
gravity, 8API 8.6 25–37
distillation Vol % IBPa, 8C
5 221
10 293
30 437
50 543
viscosity, suspension
at 388C 35,000 <30
at 998C 513
pour point, 8C 10
0
elemental analysis, wt%
carbon 83.1 86
hydrogen 10.6 13.5
sulfur 4.8 0.1–2.0
nitrogen 0.4 0.2
oxygen 1.1
hydrocarbon type, wt%
asphaltenes 19
5
resins 32
oils 49
metals, ppm
vanadium 250
nickel
iron
copper
100
75
5
}
100
ash, wt % 0.75 0
Conradson carbon, wt% 13.5 1–2
net heating value, kJ/g b 40.68 ca 45.33
a
IBP ¼ initial boiling point.
b
To convert kJ/g to btu/lb, multiply by 430.2.

in tar sand deposits have, however, variously been estimated on a world basis to be
in excess of 477  109 m3 ð3  1012 bblÞ. Reserves that have been estimated for
the United States are believed to be in excess of 795  104 m3 ð50  106 bblÞ,
although estimates vary. Bitumen reserves throughout the world can compare
favorably with reserves of conventional crude oil.
Tar sand deposits are widely distributed throughout the world (Fig. 2)
(5,6) and the various deposits have been described as belonging to two types:
stratigraphic traps and structural traps (Table 2; Fig. 3) (7). However, there
are the inevitable gradations and combinations of these two types of deposits,
and thus a broad pattern of deposit entrapment is believed to exist. In general
terms, the entrapment character of the very large tar sand deposits involves a
combination of both stratigraphic and structural traps.
The largest tar sand deposits are in Alberta, Canada, and in Venezuela.
Smaller tar sand deposits occur in the United States (mainly in Utah), Peru,
Trinidad, Madagascar, the former Soviet Union, Balkan states, and the Philip-
pines. Tar sand deposits in northwestern China (Xinjiang Autonomous Region)
also are large; at some locations, the bitumen appears on the land surface around
226 TAR SANDS Vol. 1

Melville Island
Athabasca
Olenek

Selenizza
Derna
Whiterocks Asphalt Kobystan
Sunnyside Davis-Dismal
Burgan
Asphalt Ridge Creek Melang
Peor springs Kyrock
Guanoco Leyte
Santa Cruz La Brea
Edna Incarte Bemolanga
Sisquoc Bolivar
Santa Rosa

Oficina-Temblador

Fig. 2. Principal tar sand deposits of the world, where . represents > 2; 385; 000 m3
< 15  106 bblÞ bitumen; N, probably >159; 000 m3 ð < 1  106 bblÞ bitumen; and H,
reported occurrence information limited.

Karamay, China. The largest deposits are in the Athabasca area in the province
of Alberta, Canada, and in the Orinoco region of east central Venezuela.
The Athabasca deposit, along with the neighboring Wabasca, Peace River,
and Cold Lake heavy oil deposits, have together been estimated to contain
1:86  1011 m3 ð >1:17  1012 bblÞ of bitumen. The Venezuelan deposits may at
least contain >1:60  1011 m3 ð1:0  1012 bblÞ bitumen (2). Deposits of tar sand,
each containing >3  106 m3 ð20  106 bblÞ of bitumen, have also been located
in the United States, Albania, Italy, Madagascar, Peru, Romania, Trinidad,
Zaire, and the former Soviet Union, comprising a total of ca 450  109 m3
ð2:8  1012 bblÞ.

Table 2. Tar Sand Deposits and Mode of Entrapmenta


Number Deposit Location
1. stratigraphic trap: structure of little Sunnyside, P.R. Springs, Santa Cruz
importance; short-distance
migration assumed
2. structural/stratigraphic trap: Oficina–Temblador tar, Bemolanga,
folding/faulting and unconformity Asphalt Ridge, Melville Island,
equally important Guanoco, Kentucky deposits
3. structural trap: structure important; Whiterocks, La Brea
long-distance migration assumed;
unconformity may be absent
4. intermediate between 1 and 2 Athabasca, Edna, Sisquoc,
Santa Rosa
5. intermediate between 2 and 3 Selenizza, Derna
a
See Fig. 3.
Vol. 1 TAR SANDS 227

Melville Island

Athabasca

Asphalt Ridge Peor Springs Derna


Whiterocks Sunnyside
Selenizza
Asphalt
Kyrock
Santa Cruz Dismal Creek
Edna Guanoco

Sisquoc La Brea
Bemolanga

Santa Rosa

Oficina-Temblador

Fig. 3. Types of traps for tar sand deposits, where  represents a stratigraphic trap, ,
an intermediate between stratigraphic and structural/stratigraphic traps; , a structural/
stratigraphic trap;., an intermediate between structural/stratigraphic and structural
traps; and 4, a structural trap.

The Alberta (Athabasca) tar sand deposits are located in the northeast part of
that Canadian province (Fig. 4). These are the only mineable tar sand deposits
undergoing large-scale commercial exploitation as of this writing (ca 1997).
The Athabasca deposits have been known since the early 1800s. The first
scientific interest in tar sands was taken by the Canadian government in
1890, and in 1897–1898, the sands were first drilled at Pelican Rapids on the
Athabasca River. Up until 1960, many small-scale commercial enterprises
were attempted but not sustained. Between 1957 and 1967, three extensive
pilot-plant operations were conducted in the Athabasca region, each leading to
a proposal for a commercial venture, eg, Suncor and Syncrude.
The Venezuelan tar sands are located in a 50–100-km belt extending east to
west for >700 km, immediately north of the Orinoco River. The precise limits of
the deposit are not well defined because exploration efforts in the past concen-
trated on light and medium crude accumulations.
The geological setting of the Orinoco deposit is complex, having evolved
through three cycles of sedimentation. The oil is contained by both structural
and stratigraphic traps, depending on location, age of sediment, and degree of
faulting. The tar sands are located along the southern flanks of the eastern Vene-
zuelan basin, where three distinct zones are apparent from north to south: a zone
of tertiary sedimentation, a central platform with transgressive overlapping
sediments, and a zone of erosional remnants covered by sediments. The deposit
also contains three systems of faulting. All the faults are normal and many are
concurrent with deposition.
Tar sands in the United States are contained in a variety of separate
deposits in various states (Fig. 5) but because many of these deposits are
228 TAR SANDS Vol. 1

120° District of Mackenzie 110°


60°
Al be rta
N
W
E
S
Athabasca
Plant
Peace River sites
Fort
lumbia Peace McMurray
River
British Co

Saskatchewan
Wabasca
Grand
Prairie

Cold Lake

Edmonton
Jasper

Calgary
0 100 200 km

Medicine
Hat
Canada
49°
United States 110°

Fig. 4. Tar sand and heavy oil deposits in Alberta, Canada.

small, information on most is limited (8). Attempts at development of the depos-


its have occurred primarily in Utah.

3. Properties

Tar sand has been defined as sand saturated with a highly viscous crude hydro-
carbon material not recoverable in its natural state through a well by ordinary
production methods (2–8). Technically the material should perhaps be called
bituminous sand rather than tar sand because the hydrocarbon is bitumen, ie,
a carbon disulfide-soluble oil.
The data available are generally for the Athabasca materials, although
workers at the University of Utah (Salt Lake City) have carried out an intensive
program to determine the processibility of Utah bitumen and considerable
data have become available. Bulk properties of samples from several locations
(Table 3) (9) show that there is a wide range of properties. Substantial
differences exist between the tar sands in Canada and those in the United
Vol. 1 TAR SANDS 229

States; a difference often cited is that the former is water-wet and the latter,
oil-wet (10).

Canada United States


sand is water-wet, thus disengagement sand is oil-wet, thus efficient dis-
of bitumen is efficient using hot-water engagement of bitumen requires
process (caustic ¼ sodium hydroxide; high shear rates (caustic ¼ sodium
bitumen recovery >98%) carbonate; bitumen recovery 95%)
formations usually unconsolidated formations usually consolidated
to semiconsolidated by mineral
cementation
few deposits have been identified numerous deposits identified (33 major
(Alberta contains ca 0.4 m3 bitumen) deposits ¼ 12 m3 bitumen; 20 minor
deposits ¼ 12 m3 bitu-men);
total resource ¼ 6.5 m3 bitumen
(2.6 m3 measured and 3.8 m3
billion speculative)
problems exist in settling and removal little is known about the nature
of clay from tar sand deposits and and effect on processing of clays
process streams
bitumen properties fairly uniform bitumen properties diverse
(sulfur ¼ 4:5 5:5 wt%, (sulfur ¼ 0:5 10 wt%,
nitrogen ¼ 0.1–0.5 wt %; nitrogen 0.1–1.3 wt %;
H/C ratio 1:5; API gravity H=C ratio ¼ 1:3 1.6; API
from 6 to 128) gravity from 2 to 148)
bitumen deposits large with uniform bitumen deposits small and not
quality; recovery and upgrading plants of uniform quality; recovery and
on-stream since 1970s upgrading methods need to be
site-specific

Fig. 5. Tar sand deposits in the United States.


230 TAR SANDS Vol. 1

Table 3. Bulk Properties of Tar Sands


Asphalt P.R. Tar Sand
Property Alberta Ridgea Springsa Sunnysidea Trianglea Texas Alabama
bulk density, 1.75–2.19 1.83–2.50
g/cm3
porosity, vol% 27–56 16–27 6–33 16–28 9–32 32 6–25
permeability, 99–5,900 4,905– 553– 5,265– 2,043– 3158 9.9–
m2  10 16b 5,950 14,902 7,402 7,777 6,316
specific heat, 1.46–2.09
J/(g̊:C)c
thermal 0.0071–
conductivity, 0.0015
J/(s̊CmÞc
a
Deposit in Utah.
b
To convert m2 to millidarcies, multiply by 1:013  1012 .
c
To convert J to cal, divide by 4.184.

The sand component is predominantly quartz in the form of rounded or angular


particles (11), each of which is wet with a film of water. Surrounding the wetted
sand grains and somewhat filling the void among them is a film of bitumen. The
balance of the void volume in the Canadian sands is filled with connate water
plus, sometimes, a small volume of gas. Usually the gas is air but methane
has been reported from some test borings in the Athabasca deposit. Some com-
mercial gas deposits were developed in the late 1980s. The sand grains are
packed to a void volume of ca 35%, corresponding to a mixture of ca 83 wt%
sand; the remainder is bitumen and water which constitute ca 17 wt% of the
tar sands.
3.1. Bitumen. There are wide variations both in the bitumen saturation
of tar sand (0–18 wt% bitumen), even within a particular deposit, and the visc-
osity. Of particular note is the variation of density of Athabasca bitumen with
temperature, and the maximum density difference between bitumen and water
(70–808C (160–1758F)); hence the choice of the operating temperature of the hot-
water bitumen-extraction process.
The API gravity of tar sand bitumen varies from 5 to ca 108API, depending
on the deposit, and the viscosity is very high. Whereas conventional crude oils
may have a high (>100 MPs ¼ cPÞ) viscosity at 408C, tar sand bitumen has a
viscosity on the order of 10 100 kPs (105 –106P) at formation temperature
(ca 0–108C), depending on the season. This offers a formidable obstacle to
bitumen recovery and, as a result of the high viscosity, bitumen is relatively
nonvolatile under conditions of standard distillation (Table 4) (12,13), which
influences choice of the upgrading process.
3.2. Minerals. Usually >99% of the tar sand mineral is composed of
quartz sand and clays. In the remaining 1%, more than 30 minerals have been
identified, mostly calciferous or iron-based (14). Particle sizes range from large
grains (99.9% finer than 1000 mm) to 44 mm (325 mesh), the smallest size that
can be determined by dry screening. The size between 44 and 2 mm is referred
to as silt; sizes <2mm (equivalent spherical diameter) are clay.
Vol. 1 TAR SANDS 231

Table 4. Distillation Data for Various Bitumens


Cut point, Athabasca, NW Asphalt Ridge, P.R. Springs, Tar Sand Triangle,
8C wt% distilleda wt% distilleda wt% distilleda wt% distilleda
200 3.0 2.3 0.7 1.7
225 4.6 3.3 1.4 2.9
250 6.5 4.4 2.4 4.4
275 8.9 5.8 3.8 5.9
300 14.0 7.5 4.9 8.4
325 25.9 8.8 6.8 12.4
350 18.1 11.7 8.0 15.2
375 22.4 13.8 10.1 18.6
400 26.2 16.8 12.5 22.4
425 29.1 19.5 16.0 26.9
450 33.1 23.7 20.0 28.9
475 37.0 28.4 22.5 32.3
500 40.0 34.0 25.0 35.1
525 42.9 40.0 27.3 38.5
538 44.6 44.2 28.0 40.0
538+ 55.4 55.8 72.0 60.9
a
Cumulative.

Clays are aluminosilicate minerals, some of which have definite chemical


compositions. In regard to tar sands, however, clay is only a size classification
and is usually determined by a sedimentation method. According to the previous
definition of fines, the fines fraction equals the sum of the silt and clay fractions.
The clay fraction over a wide range of fines contents is a relatively constant 30%
of the fines.
The Canadian deposits are largely unconsolidated sands having a porosity
ranging up to 45% and good intrinsic permeability. However, the deposits in
Utah range from predominantly low porosity, low permeability consolidated
sand to, in some instances, unconsolidated sands. In addition, the bitumen prop-
erties are not conducive to fluid flow under normal reservoir conditions in either
Canadian or U.S. deposits. Nevertheless, where the general nature of the depos-
its prohibits the application of a mining technique, as in many of the U.S. depos-
its, a nonmining technique may be the only feasible bitumen recovery option (6).

4. Recovery

Oil prices and operating costs are the key to economic development of tar sand
deposits. However, two technical conditions of vital concern for economic devel-
opment are the concentration of the resource (percent bitumen saturation) and
its accessibility, usually measured by the overburden thickness.
The remoteness of the U.S. tar sands is often cited as a deterrent to devel-
opment but topography of the site, overburden-to-ore body ratio, and richness of
the ore body are also important. In the 1990s context of mining tar sand deposits
in the United States, the Utah deposits (Tar Sand Triangle, P.R. Springs, Sunny-
side, and Hill Creek) generally have an overburden-to-net pay zone ratio above
232 TAR SANDS Vol. 1

Heavy oil and bitumen

Mining In situ

Surface Subsurface Thermal Nonthermal

Steam Electrical Combustion Nuclear Bacterial Diluents Emulsification


and hot
water
Reverse Forward Miscible Solvent Chemical
displacement
Stimulation Flood
Wet Dry
Hydrocarbons Inert Carbon Polymer Caustic Surfactant
gases dioxide polymer

Fig. 6. Recovery processes.

the 0.4–1.0 range, with a lean oil content. On the other hand, the Asphalt Ridge
deposit is loosely consolidated and could be mined using a ripper/front-end loader
(without drilling and blasting) at the near-surface location of the deposit.
Recovery methods are based either on mining combined with some further
processing or operation on the oil sands in situ (Fig. 6). The mining methods are
applicable to shallow deposits, characterized by an overburden ratio (ie, overbur-
den depth-to-thickness of tar sand deposit) of ca 1.0. Because Athabasca tar
sands have a maximum thickness of ca 90 m and average ca 45 m, there are indi-
cations that no more than 10% of the in-place deposit is mineable within 1990s
concepts of the economics and technology of open-pit mining.
The bitumen in the Athabasca deposit, which has a gravity on the API scale
of 88, is heavier than water and very viscous. Tar sand is a dense, solid material,
but it can be readily dug in the summer months; during the winter months when
the temperatures plunge to 45 C, tar sand assumes the consistency of concrete.
To maintain acceptable digging rates in winter, mining must proceed faster than
the rate of frost penetration; if not, supplemental measures such as blasting are
required.
4.1. Nonmining Methods. Nonmining (in situ) processes depend on
injecting a heating-and-driver substance into the ground through injection
wells and recovering bitumen through production wells. Such processes need a
relatively thick layer of overburden to contain the driver substance within the
formation between injection and production wells (2).
In principle, the nonmining recovery of bitumen from tar sand deposits is
an enhanced oil recovery technique and requires the injection of a fluid into
the formation through an injection well. This leads to the in situ displacement
of the bitumen from the reservoir and bitumen production at the surface through
an egress (production) well. There are, however, several serious constraints that
are particularly important and relate to the bulk properties of the tar sand and
the bitumen. In fact, both recovery by fluid injection and the serious constraints
on it must be considered in toto in the context of bitumen recovery by nonmining
techniques (see Petroleum, enhanced oil recovery).
Vol. 1 TAR SANDS 233

Another general constraint to bitumen recovery by nonmining methods is


the relatively low injectivity of tar sand formations. It is usually necessary to
inject displacement/recovery fluids at a pressure such that fracturing (parting)
is achieved. Such a technique, therefore, changes the reservoir profile and intro-
duces a series of channels through which fluids can flow from the injection well to
the production well. On the other hand, the technique may be disadvantageous
insofar as the fracture occurs along the path of least resistance, giving undesir-
able or inefficient flow characteristics within the reservoir between the injection
and production wells, which leave a part of the reservoir relatively untouched by
the displacement or recovery fluids.
In steam stimulation, heat and drive energy are supplied in the form of
steam injected through wells into the tar sand formation. In most instances,
the injection pressure must exceed the formation fracture pressure in order
to force the steam into the tar sands and into contact with the oil. When suffi-
cient heating has been achieved, the injection wells are closed for a soak period
of variable length and then allowed to produce, first applying the pressure cre-
ated by the injection and then using pumps as the wells cool and production
declines.
Steam can also be injected into one or more wells, with production coming
from other wells (steam drive). This technique is effective in heavy oil formations
but has found little success during application to tar sand deposits because of the
difficulty in connecting injection and production wells. However, once the flow
path has been heated, the steam pressure is cycled, alternately moving steam
up into the oil zone, then allowing oil to drain down into the heated flow channel
to be swept to the production wells.
If the viscous bitumen in a tar sand formation can be made mobile by an
admixture of either a hydrocarbon diluent or an emulsifying fluid, a relatively
low temperature secondary recovery process is possible (emulsion steam drive).
If the formation is impermeable, communication problems exist between injec-
tion and production wells. However, it is possible to apply a solution or dilution
process along a narrow fracture plane between injection and production wells.
To date (ca 1997), steam methods have been applied almost exclusively
in relatively thick reservoirs containing viscous crude oils. In the case of
heavy oil fields and tar sand deposits, the cyclic steam injection technique
has been employed with some success. The technique involves the injection
of steam at greater than fracturing pressure, usually in the 10.3–11.0 MPa
(1500–1600 psi) range, followed by a soak period, after which production is
commenced (15).
Variations include the use of steam and the means of reducing interfacial
tension by the use of various solvents. The solvent extraction approach has had
some success when applied to bitumen recovery from mined tar sand but when
applied to unmined material, losses of solvent and bitumen are always an
obstacle. This approach should not be rejected out of hand because a novel
concept may arise that guarantees minimal acceptable losses of bitumen and
solvent.
Combustion has also been effective for recovery of viscous oils in moderately
thick reservoirs where reservoir dip and continuity promote effective gravity
drainage, or where several other operational factors permit close well spacing.
234 TAR SANDS Vol. 1

During in situ combustion or fire flooding, energy is generated in the formation


by igniting bitumen in the formation and sustaining it in a state of combustion or
partial combustion. The high temperatures generated decrease the viscosity of
the oil and make it more mobile. Some cracking of the bitumen also occurs,
and the fluid recovered from the production wells is an upgraded product rather
than bitumen itself.
The recovery processes using combustion of the bitumen are termed for-
ward combustion or reverse combustion, depending on whether the combustion
front moves with or counter to the direction of air flow. In either case, burning
occurs at the interface where air contacts hot, unburned oil or, more likely, coke.
Thus, if the flame front is ignited near the injection well, it propagates toward
the production well (forward combustion). However, if the front is ignited near
the production well, it moves in the opposite direction (reverse combustion). In
forward combustion, the hydrocarbon products released from the zone of combus-
tion move into a relatively cold portion of the formation. Thus, there is a definite
upper limit of the viscosity of the liquids that can be recovered by a forward com-
bustion process. On the other hand, because the air passes through the hot for-
mation before reaching the combustion zone, burning is complete; the formation
is left completely cleaned of hydrocarbons. In reverse combustion, some hydro-
carbons are left in the formation. The theoretical advantage of reverse combus-
tion is that the combustion products move into a heated portion of the formation
and therefore are not subject to a strict viscosity limitation. However, most
attempts to implement reverse combustion in field pilot installations have
been unsuccessful. In many cases, the failure resulted from the onset of second-
ary combustion at the production well.
Using combustion to stimulate bitumen production is attractive for deep
reservoirs and in contrast to steam injection usually involves no loss of heat.
The duration of the combustion may be short (days) depending on requirements.
In addition, backflow of oil through the hot zone must be prevented or excessive
coking occurs (15,16). Another variation of the combustion process involves use of
a heat-up phase, then a blow-down (production) phase, followed by a displace-
ment phase using a fire–water flood (COFCAW process).
4.2. Mining Methods. The alternative to in situ processing is to mine
the tar sands, transport them to a processing plant, extract the bitumen value,
and dispose of the waste sand (17,18). Such a procedure is often referred to as oil
mining. This is the term applied to the surface or subsurface excavation of pet-
roleum-bearing formations for subsequent removal of the oil by washing, flota-
tion, or retorting treatments. Oil mining also includes recovery of oil by
drainage from reservoir beds to mine shafts or other openings driven into the
oil rock, or by drainage from the reservoir rock into mine openings driven outside
the oil sand but connected with it by bore holes or mine wells.
On a commercial basis, tar sand is recovered by mining, after which it is
transported to a processing plant, where the bitumen is extracted and the
sand discharged. For tar sands of 10% wt/wt bitumen saturation, 12.5 metric
tons of tar sand must be processed to recover 1 m3 (6.3 bbl) of bitumen. If the
sand contains only 5% wt/wt bitumen, twice the amount of ore must be processed
to recover this amount. Thus, it is clear that below a certain bitumen concentra-
tion, tar sands cannot be processed economically (19).
Vol. 1 TAR SANDS 235

The Athabasca tar sands deposit in Canada is the site of the only commer-
cial tar sands mining operations. The Suncor operation (near Fort McMurray,
Alberta), started production in 1967. The Syncrude Canada project, located
8 km away, started production in 1978. In both projects, about half of the terrain
is covered with muskeg, an organic soil resembling peat moss, which ranges from
a few centimeters to 7 m in depth. The primary part of the overburden, however,
consists of Pleistocene glacial drift and Clearwater Formation sand and shale.
The total overburden varies from 7 to 40 m in thickness, and the underlying
tar sand strata averages about 45 m, although typically 5–10 m must be
discarded because of a bitumen content below the economic cut-off grade of ca
6% wt/wt.
Mining of the Athabasca tar sands presents two principal issues: in-place
tar sand requires very large cutting forces and is extremely abrasive to cutting
edges, and both the equipment and pit layouts must be designed to operate
during the long Canadian winters at temperatures as low as 40 C.
There are two approaches to open-pit mining of tar sands. The first uses a
few mining units of custom design, which are necessarily expensive, eg, bucket-
wheel excavators and large drag lines in conjunction with belt conveyors. In the
second approach, a multiplicity of smaller mining units of conventional design is
employed at relatively much lower unit costs. Scrapers and truck-and-shovel
operations have been considered. Each method has advantages and risks. The
first approach was originally adopted by Suncor and Syncrude Canada, Ltd.,
with Suncor converting to large-scale truck and shovel technology in 1993.
In the Suncor pit design, the ore body is divided into two layers (benches),
each nominally 23 m high. The pit floor and the dividing plane between the upper
and lower bench are roughly horizontal, and 7300-t/h bucket-wheel excavators
are employed as the primary mining equipment (Fig. 7). Tar sands loosened
from the face of each bench by the bucket-wheels are discharged onto a series
of conveyors. The overburden is stripped by an electric shovel that discharges
to trucks for removal of the overburden material. Syncrude utilizes a single-
bench design with four 60-m3 capacity draglines as the primary mining equip-
ment (Fig. 8). The draglines pile tar sands in windrows along the edge of the
pit; four 60,000-t/h bucket-wheels transfer the tar sands to a system of trunk con-
veyor belts that move the material to the extraction plant. The mining operations

Final Overburden
dyke
First mining
elevation
bench

Dyke base Pit floor Bottom bench (second mining face)

90 m
(min)

Fig. 7. Mining with a bucket-wheel excavator, where ( ) represents the ultimate size of
the dyke.
236 TAR SANDS Vol. 1

Overburden

Overburden
Orebody piles
Pit floor Worked-out area

Fig. 8. Mining with dragline.

at the two plants differ by choice of the primary mining equipment; the bucket-
wheel excavators sit on benches, whereas the draglines sit on the surface.
Bucket-wheel excavators use units having a 10 m dia digging wheel on the
end of a long boom. Each wheel has a theoretical capacity of 8700 t/h, but
the average output from digging is about 4500 t/h. At the rate of 122,000 t/d,
tar sand can be transferred from mine to plant by a system of 152-cm wide con-
veyor belts and 183-cm trunk conveyors, operating at 333 m/min. The bucket-
wheel excavators are supplemented by front-end loaders used to dig overburden
and load it through twin chutes onto ca 135 t capacity trucks. Additional equip-
ment is used for maintaining the haul roads and for spreading and compacting
the spoiled material. Overburden may be stripped with 14-m3 hydraulically oper-
ated shovels and a fleet of ca 135 t trucks.
Draglines are equipped with a 71-m3 bucket at the end of a 111-m boom and
can be employed to dig both a portion of the overburden, which is free-cast into
the mining pit, and the tar sand, which is piled in windrows behind the machine.
Bucket-wheel reclaimers, similar to bucket-wheel excavators, load the tar sand
from the windrows onto conveyor belts which transfer it to the plant.

5. Processing

5.1. Hot-Water Process. The hot-water process is the only successful


commercial process to be applied to bitumen recovery from mined tar sands in
North America as of 1997 (2). The process utilizes linear and nonlinear varia-
tions of bitumen density and water density, respectively, with temperature so
that the bitumen that is heavier than water at room temperature becomes
lighter than water at 808C. Surface-active materials in tar sand also contribute
to the process (2). The essentials of the hot-water process involve conditioning,
separation, and scavenging (Fig. 9).
In the conditioning step (mixing or pulping), tar sand feed is heated and
mixed with water to form a pulp of 60–85% solids at 80–908C. First the lumps
of tar sand as mined are reduced in size by ablation, ie, successive layers of lump
are warmed and slough off, revealing cooler layers. The pulp is mechanically
mixed, reacts with any chemicals added, is further heated to the process tem-
perature, and is conditioned by open-steam heating in a horizontal rotating
Naptha

Feed
bin Make up Make up
water water Final
Centrifuges
bitumen

Conditioning Vibr
drum at
scre ing Tailings
en

Settled froth
Oversize
Caustic
Steam
addition Froth
settler

Froth

237
Froth

Middlings
return Froth
Middlings
Scavenger
Middlings cells
Air

Sand
Separation cell

Tailings

Fig. 9. Hot-water recovery process.


238 TAR SANDS Vol. 1

drum. With regard to equipment scale-up, conditioning is essentially a heat-


transfer process. The effluent from the conditioning drum is screened to remove
tramp material or lumps that were not sufficiently reduced in size. The screened
pulp is mixed with any added water, adjusted to the proper consistency for
pumping as described below, and sent to the separation cell.
The separation cell is an open vessel with straight sides and a cone bottom.
Mechanical rakes on the bottom move the sand toward the center for discharge.
Wiper arms rotating on the surface push the froth to the outside of the separa-
tion cell, where it overflows into launders for collection. The cell acts like two set-
tlers, one on top of the other. In the lower settler sand settles down; in the upper
settler, bitumen floats. The bulk of the sand in the feed is removed from the bot-
tom of the separation cell as tailings. A large portion of the feed bitumen floats to
the surface of the separation cell and is removed as froth. A middlings stream
consists mostly of water, with some suspended fine minerals and bitumen parti-
cles. A portion of the middlings may be returned for mixing with the conditioning
drum effluent in order to dilute the separation cell feed for pumping. The remain-
der of the middlings is called the drag stream, which is withdrawn from the
separation cell to be rejected after processing in the scavenger cells.
Tar sand feed contains a certain portion of fine minerals that, if allowed to
build up in concentration in the middlings, increases viscosity and eventually
disrupts settling in the separation cell. The drag stream is required as a purge
in order to control the fines concentration in the middlings. The amounts of water
that can enter with the feed and leave with the separation cell tailings and froth
are relatively fixed. Thus, the size of the drag stream determines the makeup
water requirement for the separation cell.
The third step in the hot-water process is scavenging. Depending on the
drag-stream size and composition, enough bitumen may leave the process in
the drag stream to make another recovery step economical. Froth flotation
with air is usually employed, and the scavenger froth is combined with the
separation cell froth to be further treated and upgraded to synthetic crude oil.
Tailings from the scavenger cell join the separation cell tailings stream and go
to waste. Conventional froth-flotation cells are suitable for this step.
Froth from the hot-water process may be mixed with a hydrocarbon diluent,
eg, coker naphtha, and centrifuged. The Suncor process employs a two-stage cen-
trifuging operation, and each stage consists of multiple centrifuges of conven-
tional design installed in parallel. The bitumen product contains 1–2 wt%
mineral (dry bitumen basis) and 5–15 wt % water (wet diluted basis). Syncrude
also utilizes a centrifuge system with naphtha diluent.
An attempt has been made to develop the hot-water process for the Utah
sands (Fig. 10) (20). With oil-wet Utah sands, this process differs significantly
from that used for the water-wet Canadian sands, necessitating disengagement
by hot-water digestion in a high shear force field under appropriate conditions of
pulp density and alkalinity. The dispersed bitumen droplets can also be recov-
ered by aeration and froth flotation (21).
The hot-water separation process involves extremely complicated surface
chemistry with interfaces among various combinations of solids (including both
silica sand and aluminosilicate clays), water, bitumen, and air. The control of pH
is critical. The preferred range is 8.0–8.5, achievable by use of any of the mono-
Tar sand ore Flowtation
air

Ore surge
re
do
s an
Tar
Crushed tar sand ore
Conditioning

Soda ash
Hot recycle water
Bitumen
Steam concentrate
Settling

Fine sand Wash water


Recycle water
Coarse sand Filtration
Clean sand

239
Air

Solvent
Filtration

Solvent and water

Bitumen– solvent Semi-clean


Solvent recycle

Steam Bitumen
clean up Bitumen– solvent Sand and water
Clean bitumen
Sand and water

Fig. 10. Hot-water recovery for Utah bitumen.


240 TAR SANDS Vol. 1

3m
Tails
Clear water layer

30 m or more 6 m
Transition zone
Dyke
Dyke
Equilibrium
sludge volume
About 15cm

Consolidation
layer

Fig. 11. Aqueous zones in tailings pond.

valent bases. Polyvalent cations must be excluded because they tend to flocculate
clays and thus raise viscosity of the middlings in the separation cell.
One problem resulting from the hot-water process is disposal and control of
the tailings. Each ton of oil sand in place has a volume of ca 0.45 m3, which gen-
erates ca 0.6 m3 of tailings and gives a substantial volume gain. If the mine pro-
duces 200,000 t/d of oil sand, volume expansion represents a considerable solids
disposal problem.
Environmental regulations in Canada and the United States do not allow
the discharge of tailings streams into the river, onto the surface, or onto any
area where contamination of groundwater domains or the river may occur. The
tailings stream is essentially high in clays and contains some bitumen; hence the
need for tailings ponds, where some settling of the clay occurs (Fig. 11). In addi-
tion, an approach to acceptable reclamation of the tailings ponds must be accom-
modated at the time of site abandonment. Problems may be alleviated somewhat
by the development of process options that require considerably less water in the
sand–bitumen separation step. Such an option would allow a more gradual
removal of the tailings ponds.
5.2. Cold-Water Process. The cold-water bitumen separation process
has been developed to the point of small-scale continuous pilot plants. The pro-
cess uses a combination of cold water and solvent. The first step usually involves
disintegration of the tar sand charge, which is mixed with water, diluent, and
reagents. The diluent may be a petroleum distillate fraction such as kerosene
and is added in a ca 1:1 weight ratio to the bitumen in the feed. The pH is main-
tained at 9–9.5 by addition of wetting agents and ca 0.77 kg of soda ash per ton of
tar sand. The effluent is mixed with more water, and in a raked classifier the
sand is settled from the bulk of the remaining mixture. The water and oil over-
flow the classifier and are passed to thickeners, where the oil is concentrated.
Clay in the tar sand feed forms emulsions that are hard to break and are wasted
with the underflow from the thickeners.
The sand reduction process is a cold-water process without solvent. The
objective is removal of sand to provide a feed suitable for a fluid coking process.
In the first step, the tar sand feed is mixed with water at ca 208C in a screw
conveyor at a ratio of 0.75–3 t/t tar sand (lower range preferred). The mixed
pulp from the screw conveyor is discharged into a rotary-drum screen, which
Vol. 1 TAR SANDS 241

is submerged in a water-filled settling vessel. The bitumen forms agglomerates


that are retained by an 840-mm (20-mesh) screen. These agglomerates settle and
are withdrawn as oil product. The sand readily passes through the 840-mm
(20-mesh) screen and is withdrawn as waste stream. Nominal composition of
the oil product is 58 wt% oil (bitumen), 27 wt% mineral, and 15 wt% water.
A process called spherical agglomeration closely resembles the sand-
reduction process. Water is added to tar sands and the mixture is ball-milled.
The bitumen forms dense agglomerates of 75–87 wt% bitumen, 12–25 wt%
sand, and 1–5 wt% water.
5.3. Solvent Extraction. An anhydrous solvent extraction process for
bitumen recovery has been attempted and usually involves the use of a low boil-
ing hydrocarbon. The process generally involves up to four steps. In the mixer
step, fresh tar sand is mixed with recycle solvent that contains some bitumen
and small amounts of water and mineral. Solvent-to-bitumen weight ratio is
adjusted to ca 0.5. The drain step consists of a three-stage countercurrent
wash. Settling and draining time is ca 30 min for each stage. After each extrac-
tion step, a bed of sand is formed and the extract drained through the bed until
the interstitial pore volume of the bed is emptied. The last two steps of the pro-
cess are devoted to solvent recovery from the solids. Although solvent extraction
processes have been attempted and demonstrated for the Athabasca, Utah, and
Kentucky tar sands, solvent losses influence economics of such processes and
they have not yet been reduced to commercial practice.

6. Bitumen Conversion

Bitumen is a hydrogen-deficient oil that is upgraded by carbon removal (coking)


or hydrogen addition (hydrocracking) (2, 4). There are two methods by which
bitumen conversion can be achieved: by direct heating of mined tar sand and
by thermal decomposition of separated bitumen. The latter is the method used
commercially, but the former has potential for commercialization (see FUELS,
SYNTHETIC).
6.1. Direct Heating of Mined Tar Sand. An early process (Fig. 12)
involved a coker for bitumen conversion and a burner to remove carbon from
the sand (22). A later proposal suggested that the Lurgi process might have
applicability to bitumen conversion (23). A more modern approach has also
been developed which also cracks the bitumen constituents on the sand (24).
The processor consists of a large horizontal rotating vessel which is arranged
in a series of compartments, a preheating zone, and a reaction zone.
Direct coking of tar sand using a fluid-bed technique has also been tested.
In this process, tar sand is fed to a coker or still, where the tar sand is heated to
ca 4808C by contact with a fluid bed of clean sand from which the coke has been
removed by burning. Volatile portions of the bitumen are distilled, whereas non-
volatile material is thermally cracked, resulting in the production of more liquid
products and the deposition of a layer of coke around each sand grain. Coked
solids are withdrawn down a standpipe, fluidized with air, and transferred to a
burner or regenerator, operating at ca 8008C where most of the coke is burned off
the sand grains. The clean, hot sand is withdrawn through a standpipe. Part
242 TAR SANDS Vol. 1

Flue gas

Tar sand Product


receiver
Coker Burner
Fuel

Oil product
Sand Spent
recycle sand By-product gas

Air Recycle gas

Fig. 12. Direct heating of tar sand for oil recovery.

(20–40%) is rejected and the remainder is recirculated to the coker to provide the
heat for the coking reaction. The products leave the coker as a vapor, which is
condensed in a receiver. Reaction off-gases from the receiver are recirculated
to fluidize the clean, hot sand, which is returned to the coker.
6.2. Conversion of Separated Bitumen. The overall upgrading pro-
cess by which bitumen is converted to liquid products is accomplished in two
steps (Fig. 13). The first step is the primary upgrading process, which improves
the hydrogen-to-carbon (H/C) ratio by either carbon removal or hydrogen addi-
tion, cracking the bitumen to lighter products which are more easily processed

Primary Hydrotreating
conversion Naptha

Bitumen Light gas oil Synthetic


crude
Heavy gas oil

Pitch
or
coke

Collected gas

Sulfur
H2S Recovery
S Plant
Gas

H2 Plant
Gas

Fig. 13. Conversion of separated bitumen.


Vol. 1 TAR SANDS 243

downstream. The secondary upgrading process involves hydrogenation of the


primary products and is the means by which sulfur and nitrogen are removed
from the primary products. The upgraded or synthetic crude can then be refined
to consumer goods such as gasoline, jet fuel, and home heating oil by conven-
tional means.
Conversion of petroleum feedstocks is accomplished using two basic process
concepts: carbon rejection, of which the coking processes are examples, and
hydrogen addition, of which the hydroprocesses are examples (3). The conversion
of a feedstock such as tar sand bitumen has added another dimension to upgrad-
ing insofar as the feedstock is one of the most complex accepted by any refinery.
Thus, coking has become the process of choice for bitumen conversion and bitu-
men is converted commercially by delayed coking (Suncor) and by fluid coking
(Syncrude). In each case the charge is converted to distillate oils, coke, and
light gases. The coke fraction and product gases can be used for plant fuel.
The coker distillate is a partially upgraded material in itself and is a suitable
feed for hydrodesulfurization to produce a low sulfur synthetic crude oil. In
each case, bitumen conversion to liquids is on the order of 75%þ. Fluid coking
gives a generally higher (þ1 5%) yield of liquids compared to delayed coking.
The remainder appears as coke (ca 15% wt/wt) and gases.
Sulfur is distributed throughout the boiling range of the delayed coker dis-
tillate, as for distillates from direct coking. Nitrogen is more heavily concen-
trated in the higher boiling fractions but is present in most of the distillate
fractions. Raw coker naphtha contains significant quantities of olefins and diole-
fins which must be saturated by downstream hydrotreating. The gas oil has a
high aromatic content typical of coker gas oils.
Finishing and stabilization (hydrodesulfurization and saturation) of the
liquid products are achieved by hydrotreating the liquid streams as two or
three separate streams, because of the variation in conditions and catalysts
necessary for treatment of a naphtha fraction relative to conditions necessary
for treatment of a gas oil (13). It is more efficient to treat the liquid product
streams separately and then to blend the finished liquids to a synthetic crude
oil. In order to take advantage of optimum operating conditions for various dis-
tillate fractions, the Suncor coker distillate is treated as three separate fractions:
naphtha, kerosene, and gas oil. In the operation used by Syncrude, the bitumen
products are separated into two distinct fractions: naphtha and mixed gas oils.
Each plant combines the hydrotreated fractions to form synthetic crude oil
(Table 5), which is then shipped by pipeline to a refinery (see Pipelines). Other
processes which have received attention for bitumen upgrading include partial
upgrading (a form of thermal deasphalting), flexicoking, the Eureka process,
and various hydrocracking processes.
Partial coking or thermal deasphalting processes provide minimal upgrad-
ing of bitumen. In partial coking, the hot-water process froth is distilled at atmo-
spheric pressure, and minerals and water are removed. In flexicoking a gasifier
vessel is added to the system in order to gasify excess coke with a gas–air mix-
ture to a low heating value gas, which can be desulfurized and used as a plant
fuel. The Eureka process is a variant of delayed coking and uses steam stripping
to enhance yield and produce a heavy pitch rather than coke by-product. Hydro-
cracking has also been proposed as a means of bitumen upgrading. The overall
244 TAR SANDS Vol. 1

Table 5. Properties of Synthetic Crude Oil


Property Value
gravity, 8A PI 32
sulfur, wt% 0.15
nitrogen, wt% 0.06–0.10
viscosity, at 378C, mm2/s (¼cSt) <10
components, vol/vol%
C4 4
C5/2208C 24
430/3468C 32
650/5508C 40
550 Cþ 0

liquid yield of direct hydrogenation or hydrocracking of bitumen is substantially


higher than that of coking, and significant amounts of sulfur and nitrogen are
removed. Large quantities of external fuel or hydrogen plant feedstock are
required, however. To prevent coking, the processes operate at high pressure,
with direct contact between bitumen feed and circulating hydrogen. Hydrocrack-
ing processes include the H-Oil process, the LC-Fining process, the Vebe process,
and the Chiyoda process.
The hydrocracker products have higher hydrogen and lower sulfur and
nitrogen contents than those from the coking route and require less secondary
upgrading. However, disadvantages of the hydrogen route include relatively
high hydrogen consumption and high pressure operation. Processes that use con-
ventional, eg, Co–Mo or Ni–Mo, catalysts are susceptible to metals poisoning,
which may limit applicability to, or economics of, operation on feeds high in
metals such as bitumen.

7. Health and Safety Factors

Health and safety factors associated with tar sand processing depend on
the nature of the process and products (Table 6; Fig. 14). Issues arising from
tar sand mining (Table 7) (25) are similar to other large-scale surface mining

Table 6. Emissions from Tar Sand Plant


Source Potential contaminant
process wastewater suspended solids, dissolved solids, phenols, ammonia,
oils, organics, sulfides, metals
sanitary wastewater suspended solids, dissolved solids, biochemical oxygen
demand, organics, nitrates, phosphate, residual
chlorine, coliform organisms, metals
runoff from upgrader area: suspended solids, oils, organics, inorganics, sulfur,
coke storage pile, sulfur metals
storage pile,
solid waste landfills
power plant stacks particulates, SO2, NOx, CO
sulfur plant stacks sulfides, SO2, H2S, particulates
upgrader heaters SO2, NOx, CO2, hydrocarbons, particulates
Cracked products Conventional
Thermal Hydrogen refinery
process treatment feedstock

Mining Sand Metals Nitrogen Sulfur


Tar sand and coke
or
quarrying

Water or Cracked
solvent Bitumen products Hydrogen Conventional

245
Cracking treatment refinery
extraction
process feedstock

Sand Metals Nitrogen Sulfur


and coke

Vis reduced Sulfur and


and nitrogen
metals removed removed

Fig. 14. Products from tar sand processing.


Table 7. Environmental Impact of Surface Mining Operations
Changes in ground
Surface changes water regime
Increased Destruction of
landslide existing Alteration of Topographic Drainage Increased
Operation or source of impact risk vegetation habitats changes diversion noise Physical Chemical
site preparation þ þ þ þ þ
surface cleaning (cleared area) þ þ þ þ þ
stripping (stripped area) þ þ þ þ þ
tar sand extracting (mined area) þ þ þ þ
haul road transportation þ þ þ þ þþ
(construction)

246
tailings disposal þ
bitumen in tailings or low grade þ þ
tar sand waste
fines in tailings
stripped waste
solubles or water transportation þ þ þ
particles in overburden
new surface þ
increases in surface slope from þ
waste disposal
rehandling of materials: back-filling, þ
grading, and recontouring
Vol. 1 TAR SANDS 247

Table 8. Projects for Bitumen and Heavy Oil Recovery/Conversion


Investment, Production,
Project $106 Timing Descriptiona m3/d (bbl/d)b
Amoco 500 immediate SAGD heavy 6,000 (50,0000)
Primrose oil production heavy oil

Alberta Energy
Co., Ltd.
Cold Lake 13 immediate SAGD heavy 120 (1,000)
oil pilot plant heavy oil
Foster Creek 200 1997–1998 commercial 3,600 (30,000)
heavy oil heavy oil
Elan Energy, Ltd. 60 immediate SAGD heavy increased 1,680–
Lindbergh oil plant 2,640 (14,000–
22,000)
Elk Point
Gibson Petroleum, 7 1996–1997 SAGD bitumen increased 240–
Ltd., project 540 (2,000–
Utf 4,500)
Gulf Canada
Resources, Ltd.
Surmont 30 1996 phase 1 SAGD 360 (3,000)
heavy oil
1998–1999 phase 2 increased 2,400
(20,000)
Imperial Oil Limited
Cold Lake 250 immediate phases 9–10 in increased
situ bitumen 14,400;
production 3,000
(120,000;
25; 000þ)
Cold Lake 250 1997 phases 11–12 add 2,400 (20,000)
(proposed) in situ bitumen
production
following
completion
of phases 9–10
Imperial
Cold Lake 40 immediate heavy oil 1,128 (9,400)
development heavy oil
below current
Cold Lake lease
Koch Oil Co., Ltd.
Cold Lake 1997–1998 in situ bitumen 6,000 (50,000)
project
Reita Lake
Shell Canada Ltd.
Peace River 3 immediate SAGD heavy oil increased
project
Solvex
Bitumount 170 1997 bitumen mine/ 168 (14,000)
extraction bitumen, 100,000 t
complex, alumina, 200,000 t
minerals synthetic silica
extraction plant (tailings)
248 TAR SANDS Vol. 1

Table 8. (Continued)
Investment, Production,
Project $106 Timing Descriptiona m3/d (bbl/d)b

Suncor Inc.
Primrose 17 1996–1997 phase 1, 300 (2,500)
SAGD heavy oil
pilot plant
Burnt Lake 122 1998–1999 phase 2, increased 1,500
commercial (12,500)
plant heavy oil
Suncor Inc., Osg
Steepbank Mine 300 2001 new bitumen increased
mine production
10,800 (90,000þ)
light sweet crude,
custom blends
Fort McMurray 300 1998–2001 upgrader increased 9,600
improve- (80,000) light
ments, sweet crude,
plant custom blends
modifications
Fort McMurray 190 immediate environmental reduced SO2
measures emissions
Syncrude
Canada, Ltd.
North Mine 500 1998 new bitumen increased
Mildred Lake mine, 13  106 m3
debottlene-
cking plant
Syncrude Canada,
Ltd.
Aurora Mine 500 2001 new bitumen, first train, 2001;
remote increased 11.3
plant, (94  106 bbl=yr)
debottleneck
Total 3,452 $  109
a
SAGD ¼ steam assisted gravity draining.
b
To convert m3/d to bbl/d, multiply by 6.29.

operations involving large equipment and the movement of huge quantities of


material. The principal environmental consideration relating to the mining pro-
cess is land reclamation following the completion of the mining and, in particu-
lar, those areas affected by the deposition of tailings. Both air and liquid effluents
are subject to controls.
Health and safety factors in in situ operations are associated with high
temperature, high pressure steam, or high pressure air. Environmental consid-
erations relate to air and water quality and surface reclamation. In some envir-
onmentally sensitive areas such as the oil sands deposits in Utah, environmental
considerations may make development unfeasible.
Vol. 1 TAR SANDS 249

8. Future Outlook

The government of the Canadian Province of Alberta has announced a standard


royalty formula for the oil sands industry and embraced the principles and, to a
large degree, put into practice the recommendations of the National Task Force
on Oil Sands Strategies. The Canadian government plans to extend the mining
tax regulation to include in situ operations. Over $3,400,000,000 (Canadian $) in
new projects and expansions have been set aside to allow the industry to move
forward with projects in the initial stages of development (Table 8) which should
encourage further development of Canadian tar sand resources.

BIBLIOGRAPHY

‘‘Tar Sands’’ in ECT 1st ed., Vol. 13, pp. 633–645, by K. A. Clark, Research Council of
Alberta, and G. B. Shea (U.S. Deposits), U.S. Department of the Interior; in ECT 2nd
ed., Vol. 16, pp. 682–732, by F. W. Camp, Sun Oil; in ECT 3rd ed., Vol. 22, pp. 601–
627, by D. Towson, Petro-Canada.

CITED PUBLICATIONS

N. Berkowitz and J. G. Speight, Fuel 54, 138 (1975).


J. G. Speight, in J. G. Speight, ed., Fuel Science and Technology Handbook, Part II,
Marcel Dekker, Inc., New York, 1990.
J. G. Speight, The Chemistry and Technology of Petroleum, 2nd ed., Marcel Dekker, Inc.,
New York, 1991.
M. R. Gray, Upgrading Petroleum Residues and Heavy Oils, Marcel Dekker, Inc.,
New York, 1994.
P. H. Phizackerley and L. O. Scott, in G. V. Chilingarian and T. F. Yen, eds., Bitumens,
Asphalts, and Tar Sands, Elsevier, Amsterdam, the Netherlands, 1978, p. 57.
R. F. Meyer and W. D. Dietzman, in R. F. Meyer and C. T. Steele, eds., The Future of
Heavy Crude Oil and Tar Sands, McGraw-Hill Book Co., Inc., New York, 1981, p. 16.
E. J. Walters, in L. V. Hills, ed., Oil Sands: The Fuel of the Future, Canadian Society of
Petroleum and Geology, Calgary, Alberta, Canada, 1974, p. 240.
L. C. Marchant and C. A. Koch, in R. F. Meyer, J. C. Wynn, and J. C. Olson, eds., Proceed-
ings of the 2nd International Conference on the Future of Heavy Crude and Tar Sands,
McGraw-Hill Book Co., Inc., New York, 1984, p. 1029.
D. Smith-Magowan, A. Skauge, and L. G. Hepler, Can. J. Petrol. Technol. 21(3), 28 (1982).
Heavy Oiler (Jan. 1986).
M. C. Harris and J. C. Sobkowicz, in D. A. Redford and A. G. Winestock, eds., The Oil
Sands of Canada–Venezuela, Special Publication No. 17, Canadian Institute of Mining
and Metallurgy, Ottawa, Canada, 1977, p. 270.
J. W. Bunger, K. P. Thomas, and S. M. Dorrence, Fuel 58, 183 (1979).
J. G. Speight, The Desulfurization of Heavy Oils and residua, Marcel Dekker, Inc.,
New York, 1981.
W. N. Hamilton and G. S. Mellon, in M. A. Carrigy and J. W. Kramers, eds., Guide to
the Athabasca Oil Sands Area, Information Series No. 65, Alberta Research Council,
Alberta, Canada, 1978.
250 DRILLING FLUIDS Vol. 1

J. Burger, in Ref. 5, p. 191.


R. Mungen and J. H. Nicholls, Proc. 9th World Petrol. Cong. 5, 29 (1975).
R. A. Dick and S. P. Wimpfen, Sci. Amer. 243(4), 182 (1980).
R. Houlihan, in Ref. 8, p. 1076.
F. K. Spragins, in Ref. 5, p. 92.
J. C. Miller and M. Misra, Fuel Process. Technol. 6, 27 (1982).
K. E. Hatfield and A. G. Oblad, in Ref. 8, p. 1175.
P. E. Gishler, Can. J. Res. 27, 104 (1949).
R. W. Rammler, Can. J. Chem. Eng. 48, 552 (1970).
W. Taciuk, Energy Proc. Canada 74(4), 27 (1981).
N. A. Frazier and co-workers, Report No. 68-02-1323, United States Environmental
Protection Agency, Washington, D.C., 1976.

JAMES G. SPEIGHT
Western Research Institute

DRILLING FLUIDS
1. Introduction

A range of specialty chemicals is used by the petroleum industry for oil and gas
exploration, drilling and production, and for enhanced oil recovery. Broadly
speaking, drilling fluids are used to lubricate the drill bit, to control formation
pressure and to remove formation cuttings. Workover and completion fluids
are utilized when operating in producing formations. Chemicals are used to
cement steel pipes or casing to the sides of the borehole and to encourage the
flow of crude oil to the well (stimulation). Chemicals find applications at all
stages from oil production at the well bore continuing on to the delivery of
crude oil to the refinery. During Enhanced Oil Recovery (EOR), chemicals
are used for the various techniques (ie, steam, carbon dioxide, polymer and
micellar-polymer flooding) that renew the oil flow from fields that have ceased
to produce by conventional methods.
Drilling fluids or muds are considered an essential component of the rotary
drilling process used in drilling for oil and gas on land and in offshore environ-
ments. This fluid performs a variety of functions that influence the drilling rate,
and the cost, efficiency, and safety of the operation (1–7). Some of the most
important functions of drilling fluids include (1) the transporting of the drilling
cuttings to the surface (2) balancing subsurface and formation pressures
required for the prevention of well blowouts (3) cooling, lubricating and support-
ing part of the weight of the drill and drill pipe. During the drilling of a well, the
drilling fluid mud is pumped from the mud tanks down the hollow drill pipe and
through nozzles in the drill bit. Typical diameter of a drill pipe is 4 in. (10 cm).
The borehole at the bottom of a deep well might be 8 in. (20 cm) in diameter. The
flowing mud sweeps the crushed rock cuttings from under the bit and carries

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.0418091203120118.a01.pub2
Vol. 1 DRILLING FLUIDS 251

them back up the annular space between the drill pipe and the bore hole or cas-
ing to the surface. After reaching the surface, the drilling fluid is passed through
a series of vibrating screens, settling tanks or pits, hydrocyclones and centrifuges
to remove the cuttings brought to the surface. It is then treated with additives to
obtain a set of desired physical and chemical properties. Once treated, the fluid is
pumped back into the well and the cycle repeated.
Drilling fluids generally are composed of liquids: eg, water, petroleum oils,
and other organic liquids; dissolved inorganic and organic additives; and sus-
pended, finely divided solids of various types. The chemistry of the liquid
phase and the level of suspended solids determine the treatment strategy and
the efficiency of the mud-handling equipment. This chemistry also influences
the type and amount of materials needed to maintain or change the density,
viscosity, and other properties (8). Drilling fluids are dynamic systems. These
muds constantly change in response to the changing conditions while the well is
being drilled and downhole temperatures and pressures increase. Occasionally,
the downhole environment requires replacement of one fluid with another of a
different type.
Drilling fluid costs range from several thousand to several million dollars
per well and depend on the nature of the well being drilled. The length of the
drilling time may vary from a few days to more than a year. On the average,
6–8% of the total drilling cost arises directly from the drilling fluid and
additives. Additional fluid and total well costs may arise from improperly formu-
lated or treated fluids that can prolong the drilling time. Total worldwide sales of
drilling fluids in 2000 has been estimated to be $2.28  109 (excluding Eastern
Europe and China) (9). About $1.1  109 or 40% of the total was spent in North
America including the United States. It has been predicted that the worldwide
sales of drilling fluids will rise to $2.72  109 in 2005.
The leading three companies in the North American market that specialize
in the supply of certain aggregates of chemicals, equipment, and technology used
for finding oil and drilling wells are M-I L.L.C. (owned by Schlumberger and
Smith Industries), Haliburton Company (which owns Baroid), and Baker Hughes
Inteq (owned by Baker Hughes). These companies are commonly referred to as
drilling fluid (or drilling mud) companies.
The earliest fluid used for drilling oil wells was water. Various additives
controlled the loss of water to permeable formations exposed in the borehole.
As early as 1901, ground-formation material and surface clays were used to
impart viscosity and density in the Spindletop well drilled near Beaumont,
Texas. High density solids such as hematite and barite were applied in the
early 1920s to increase the drilling fluid density, prevent influx of formation
fluids, and control well blowouts. Drilling fluids were commercialized 1926.
Drilling fluid materials are also used as completion and workover fluids.
When the drilling reaches the oil producing zone, the subsequent operation is
referred to as completion. These fluids are formulated to enhance production
rates of crude oil and minimize permeability damage to the formation while dril-
ling through a production zone. Workover fluids are used on existing oil and gas
wells with the aim of improving or maintaining current production levels and
during remedial work on older wells. Some of the workover operations in oil
well drilling can include sand control measures, casing repair, reperforation,
252 DRILLING FLUIDS Vol. 1

and repair or replacement of subsurface equipment. Historically, water or oil-


based drilling fluids have been employed as workover and completion fluids.
With the increasingly difficult conditions encountered during the drilling of
very deep wells, special water- or oil-based fluids are being developed to maxi-
mize hydrocarbon recovery. Workover fluids also find use in many abandoned
wells that are being reworked and converted back to producing wells.

2. Classification of Drilling Fluids

Drilling fluids are classified as to the nature of the continuous phase: gas, water,
oil, or synthetic. Within each classification, there are divisions based on composi-
tion or chemistry of the fluid or the dispersed phase.
2.1. Gas-Based Muds. Gas-based drilling fluids are used mostly for
hard-rock drilling. These fluids range from compressed dry air or natural gas
to water-based mist and stable foams. Foam is considered gas based since the
gas comprises the bulk of the fluid volume. Bottomhole pressures imposed by a
gas-based fluid are low, and therefore, formation strengths must be relatively
high, having little or no influx or formation fluid. Air, gas, or mist drilling
requires a high annular velocity to remove drill cuttings. On the other hand,
large cuttings can be removed at low annular velocities when using stable foams.
Chemical additives for gas-based drilling fluids are generally biodegradable
mixtures of surfactants, certain polymers, and occasionally salts such as sodium
or potassium chloride. An aqueous solution of the additives is injected into the air
or gas flow to generate a mist or foam. No additives are used in dry air or gas
drilling operations. Gas-based fluids are not recirculated and materials are
added continuously. As the fluid exits the well, air or water vapor escapes to
the atmosphere, gas and oil are burned, and water and formation solids are col-
lected into a pit for later disposal. Stable foams must be destabilized to separate
the air from the liquid phase for disposal.
2.2. Water-Based Muds. The vast majority of all drilling fluids are
water-based systems. The types depend on the composition of the water phase
(pH, ionic content, etc), viscosity builders (clays or polymers), and rheological
control agents (deflocculants or dispersants).
Freshwater fluids can range from clear water having no additives to high
density muds containing clays, barite, and various organic additives. Onshore
wells typically use freshwater muds, as do some offshore wells where highly
weighted muds are needed. Freshwater muds may be operated at pH levels
ranging from 7 to 11. When drilling using clear water, small amounts of poly-
meric flocculants may be added to remove drill solids in a large settling pit in
order to maintain a clean fluid for fast drilling. When a viscous fluid is
required, clays or water-soluble polymers are added. Freshwater is ideal for
formulating stable drilling fluids as many mud additives are most effective
in a system of low ionic strength. Inorganic or organic additives control the
rheological behavior of the clays, particularly at elevated temperatures. An
organic polymer may be used for filtration control. Mud pH is generally alka-
line and, in fact, many viscosity control agents require an environment of
pH >9. Sodium hydroxide is by far the most widely used alkalinity control
Vol. 1 DRILLING FLUIDS 253

agent. Clay-based freshwater muds can be weighted to any desired density


required to control formation pressures.
Many offshore wells are drilled using a seawater system because of ready
availability. Seawater muds generally are formulated and maintained in the
same way that a freshwater mud is used. However, because of the presence of
dissolved salts in seawater, more additives are needed to achieve the desired
flow and filtration properties.
In many drilling areas both onshore and offshore, salt beds or salt domes
are penetrated. Mud saturated with the salt present in the formation is used
to reduce the hole enlargement that would result from salt dissolution by contact
with an undersaturated liquid. In the United States, the salt formations are pri-
marily made up of sodium chloride. In other areas, eg, northern Europe, the salt
may be composed of mixed salts, predominantly magnesium and potassium
chlorides. It has become quite common to use high (20–23 wt% NaCl) salt
muds in wells being drilled in deep (>500-m water depth) water regions of the
Gulf of Mexico. The reasons are twofold: stabilization of water-sensitive shales
(10) and inhibition of the formation of gas hydrates (11).
The high salinity of salt water muds may require different clays and organic
additives than those used in fresh- or seawater muds. Salt water clays and
organic polymers contribute to viscosity. Filtration properties are adjusted
using starch or cellulosic polymers. Alternatively, clays used primarily in fresh
and seawater muds can be prehydrated in freshwater and then added to a salt
mud for viscosity and some filtration control. The pH ranges from that of the
makeup brine, which may be somewhat acidic, to 9–11 through use of sodium
hydroxide or lime. The presence of soluble calcium or magnesium in the mud
determines whether it is feasible to maintain a high pH.
Fresh- or seawater muds may be treated with gypsum or lime to alleviate
drilling problems that may arise from drilling water-sensitive shale or clay-
bearing formations. Gyp muds (gypsum added) are generally maintained at a pH
of 9–10, whereas lime muds (lime added) are in the 12–13 pH range. Calcium-
treated muds generally require more additives to control flow and filtration
properties than those without gypsum or lime.
Potassium treated systems combine one or more polymers and a potassium
ion source, primarily potassium chloride, in order to prevent problems associated
with drilling certain water-sensitive shales (12,13). The flow and filtration prop-
erties may be quite different from those of the other water-base fluids. Potassium
muds have been applied in most active drilling regions around the world. Envir-
onmental regulations in the United States have limited the use of potassium
muds in offshore drilling owing to the apparent toxicity of high potassium levels
in the bioassay test required by discharge permits.
Freshwater, clay, and polymers for viscosity enhancement and filtration
control make up low solid/nondispersed muds. Low solids muds are maintained
using minimal amounts of clay and require removal of all but modest quantities
of drill solids. These are called nondispersed systems because no additives are
used to further disperse or deflocculate the viscosity building clays. Most
water-based muds are considered dispersed because deflocculating additives
are used to control the flow properties. Nondispersed muds can be weighted
to high densities, but are used primarily in the unweighted state. The main
254 DRILLING FLUIDS Vol. 1

advantage of these systems is the high drilling rate that can be achieved because
of the lower colloidal solids content. These are normally applied in hard forma-
tions where increasing the penetration rate can reduce drilling costs significantly
and the tendency for solids buildup is minimal.
2.3. Oil-Based Muds. Oil-based drilling fluids have diesel or mineral oil
(14) as a continuous phase with both internal water and solid phases. Fluids hav-
ing no or very low water content are usually called oil-base muds or all oil muds;
fluids having higher water contents are called invert oil–emulsion muds, or sim-
ply inverts. Most oil muds maintain a fixed oil–water ratio depending on the
desired properties. Oil muds are employed for high angle wells where good lubri-
city is required, for high temperature wells where water-based systems may be
thermally unstable, for drilling water-sensitive shale formations, or where corro-
sive gases such as hydrogen sulfide and carbon dioxide may be encountered.
Environmental restrictions and cost often limit use, although higher drilling
rates achievable using oil muds and polycrystalline diamond compact (PDC)
bits can often offset the high fluid and disposal costs.
Oil-base muds have diesel or mineral oil as the continuous phase and either
are formulated using no internal aqueous phase or have only a minimal water
content (15). Organophilic clay or colloidal asphalt are used to control viscosity
and filtration rates. The internal water phase, either added as part of the formu-
lation or incorporated while drilling, is stabilized using emulsifiers. The desired
density is reached by adding a powdered high specific gravity solid. A wetting
agent ensures oil wetting of added or drilled formation solids.
Oil-based muds that incorporate an internal brine phase as an integral part
of the formulation are called inverts or invert oil–emulsion muds. The water con-
tent can be increased up to 50 vol% of the total liquid phase using the proper
emulsifiers and wetting agents. Suspension properties are achieved by the addi-
tion of organophilic clays. Colloidal solids such as oil-dispersible lignite, resins,
asphalt, or gilsonite impart filtration control. Increased density is achieved using
standard solid materials. The internal water phase is nearly always composed of
a concentrated sodium or calcium chloride brine to provide a sufficiently reduced
water activity to prevent osmotic uptake of drilling fluid water into water-
sensitive formations.
Oil muds are usually characterized by very low filtration rates and at one
time oil muds were not considered stable if other than minimal filtration rates
were observed in standard tests. This feature contributed to the low drilling
rates experienced for oil muds when using rock or natural diamond bits. Relax-
ing the normally tight filtration control specifications on oil muds can result in
higher drilling rates without loss of emulsion or mud stability (16). These relaxed
fluid-loss muds also are termed low colloid oil muds or relaxed filtration oil muds
because the higher filtration rates are achieved by omitting some of the colloidal
solids from the formulation and reducing the concentration of emulsifiers and
surfactants in the fluid. Most relaxed fluid-loss oil muds contain 5–15 vol%
brine, depending on the mud density. In most other respects these fluids are
similar to a conventional invert oil–emulsion mud and can easily be converted
to an invert. Use of relaxed filtration oil muds has dropped considerably owing
to improvements in bit design. These improvements allow high drilling rates
using the more conventional low filtration-rate oil muds.
Vol. 1 DRILLING FLUIDS 255

The synthetic-based muds (drilling fluids) are a relatively new class of dril-
ling muds that are particularly useful for deepwater and deviated hole drilling.
These muds were developed to provide an environmentally superior alternative
to oil-based drilling muds as well as offering an alternative to the high costs asso-
ciated with the disposal of drill cuttings generated when diesel or mineral oil-
based muds are used. The synthetic muds have a continuous phase that consists
of synthetic organic liquid (17). The synthetic-based liquids, because of their
similarity with the oil-based muds, are often called pseudo-oil muds outside
the United States. While the technical and economic benefits of using such fluids
in the deepwater Gulf of Mexico are recognized, the environmental impacts of the
discharged drill cuttings produced are not well understood. While the drill cut-
tings themselves are considered toxicologically inert, there is concern that any
adhering drilling fluid ingredients may be toxic, particularly if the cuttings are
produced during drilling with oil-based or synthetic-based fluids. The high cost
for the synthetic-based fluids can be offset by the enhanced drilling performance
for fast drilling achieved when using them, and the on-site discharge of the SBFs,
when this is permitted. The synthetic-based fluids (SBF) introduced at the start
of the 1990s may be classified into four general categories (1) Synthetic hydrocar-
bons, eg, polymerized olefins that include poly alpha olefins (PAO), linear alpha
olefins (LAO), and internal olefins (IO) (2) ethers (3) esters and (4) acetals. Dur-
ing the second half of the 1990s, the groups of synthetic drilling fluids that began
to achieve enhanced popularity were the LAOs, IOs, and linear paraffins (LPs),
which were often referred to as second generation SBFS (18). These second gen-
eration SBFs were found to have benefits over their predecessors in that they
had lower kinematic viscosities and were less expensive. The current concerns
with toxicity, biodegradation rates, environmental impacts and costs have vir-
tually eliminated the use of synthetic-based fluids except for olefins and esters
in offshore drilling. The environmental aspects of the SBFs are discussed more
fully in Section 7.

3. Properties

The drilling fluids are suspensions or solutions of solids in water or oil, or of


solids and one of these liquids dispersed in the other. The type of drilling fluid
required for a particular well is determined by the geological formation.
The functions of the drilling fluids include:
Removal of formation cuttings from beneath the bit and transporting these
cuttings to the surface.
Prevention of gases in the formation from escaping through the boreholes.
Cooling and lubrication of the bit and drill string at points of contact with the
cased or uncased borehole.
Prevention of the influx of formation fluids into the wellbore.
Sealing of exposed permeable formations.
Maintainenance of the stability of exposed formations.
Helping to suspend the weight of the drill string and casing.
Control of pressure.
256 DRILLING FLUIDS Vol. 1

Furthermore, the fluid should not damage productive formations, it should


not be corrosive to the equipment, and it must be cost effective. The fluid must be
safe for handling and be compatible with the environment or be disposable in an
environmentally sound manner.
How well the drilling fluid fulfills these functions is determined primarily
by the response of the well. The whole drilling operation suffers if the fluid is not
adequate. In extreme cases, drilling may be stopped or a hole may have to be
redrilled. A variety of physical and chemical properties are monitored to ensure
satisfactory performance and guide fluid formulation and treatment (19,20).
3.1. Density. The density of the drilling fluid is adjusted using powdered
high density solids or dissolved salts to provide a hydrostatic pressure against
exposed formations in excess of the pressure of the formation fluids. In addition,
the hydrostatic pressure of the mud/clear brine column prevents collapse of weak
formations into the borehole. Fluid densities may range from that of air to >2500
kg/m3 (20.8 lb/gal). Most drilling fluids have densities >1000 kg/m3 (8.33 lb/gal),
the density of water. The hydrostatic pressure imposed by a column of drilling
fluid is expressed as follows:

P ¼ 0:098 Lrm ð¼ 0:052 Lrm Þ

where P ¼ the hydrostatic pressure in kPa (psi); rm ¼ the drilling fluid density,
kg/m3 (lb/gal); and L ¼ the column length or well depth, m(ft). Wells with bottom-
hole pressures of 100 MPa (14,500 psi) at 5000 m (16,400 ft) are not unusual.
Such a well would require a mud density >2040 kg/m3 (17 lb/gal) to prevent for-
mation-fluid influx. At least 2 MPa (290 psi) above formation pressures normally
is recommended to ensure well control. Bottomhole pressures imposed by the
mud/clear brine column that are higher than necessary can reduce drilling
rates and induce fractures in exposed formations, allowing loss of drilling fluid
to the fractures (lost circulation). Operational safety can be adversely affected
if lost circulation occurs.
3.2. Flow Properties. The fluid viscosity and annular flow velocity must
be high enough to remove cuttings generated by the drill bit and other formation
material that may fall into the wellbore. These solids are carried up the annulus
to the surface where they are separated with varying degrees of efficiency and
then disposed of. In order to accomplish this, low viscosity drilling fluids are cir-
culated at high flow rates or high viscosity fluids at low flow rates. In addition,
for maximum drilling rate, a low effective viscosity is desired at shear rates
generated through the bit nozzles (10,000–100,000/s). The varying demands on
the flow properties are best met by fluids exhibiting non-Newtonian rheological
characteristics. Drilling fluids are normally shear thinning, having apparent
viscosity decreasing with increasing shear rate.
The gel strength (thixotropy) is a measure of the capability of a drilling fluid
to hold particles in suspension after the mud circulation ceases. It results from
the electrical charges on the individual clay platelets. The positively charged
edges of a platelet are attracted to the negatively charged flat surfaces of
adjacent platelets. In a bentonite mud in which the particles are completely
dispersed, the bonds between particles are broken while the mud is flowing.
When the mud pump is shut off and flow ceases, the attraction between the
Vol. 1 DRILLING FLUIDS 257

clay particles causes the platelets to bond to each (referred to as flocculation).


This edge-to-face flocculation results in an open card-house structure capable
of suspending cuttings and sand and gravel particles. This property also sus-
pends finely ground, high specific gravity material such as barite when high-
density muds are required. The ability to keep cuttings in suspension prevents
sandlocking (sticking) of the tools in the borehole while drill rods are added to the
string and minimizes sediment collecting in the bottom of the hole after reaming
and before going back in the hole with a sampler. A drawback to this property is
that cuttings do not readily settle out of the drilling mud in the mud pit and may
be recirculated, thus resulting in grinding of particles by the drill bit, increased
mud density, increased mud pump wear, and lower penetration rate. Thixotropic
properties characterized by a gel strength provide suspension of solids when
mud/clear brine circulation stops. The rheological properties are adjusted
using inorganic and organic additives to achieve the capacity to carry cuttings
and generate adequate bit hydraulic horsepower for fast drilling at moderate
flow rates. The high flow rates required to remove drill cuttings using low viscosity
muds and in high angle wells can increase pump maintenance and induce
fracturing and lost circulation because of excessive bottomhole pressures.
3.3. Filtration Properties. Drilling fluids have a natural tendency to
flow into permeable formations because the borehole pressure is generally higher
than that in the formation. To prevent excessive leak-off, a thin, low permeability
filter cake is formed using additives. Filtration occurs under both dynamic (dur-
ing circulation) and static (no circulation) conditions. Additives may affect each
of these filtration conditions differently. The filtration rate is adjusted using col-
loidal solids and organic polymers to reduce loss of filtrate to the formation and
prevent buildup of a thick filter cake which would restrict the wellbore. Excessive
decrease of the filtration rate can be costly and may result in a viscous fluid that
may affect the drilling rate adversely.
Large solid particles of various sizes and shapes may be added to control
circulation loss where natural or induced fractures, highly porous formations,
or vugular zones, ie, formations containing small cavities (vugs) larger than
the matrix grain size, are encountered. Particle sizes from those large enough
to bridge the opening down to those fine enough to seal small spaces between
the larger bridging particles may be required to prevent drilling fluid losses
into these zones.
3.4. Water Chemistry. Water is present in all but purely gaseous or oil
drilling fluids, both of which comprise only a small percentage of drilling fluid
applications. The water may be present as fine droplets in a mist, emulsified
in an organic continuous phase, or as is most common, comprise the continuous
phase of the drilling fluid. Water added to drilling fluids may be fresh water, sea-
water, or saturated salt solutions. The salt in the last is normally sodium chlor-
ide, but may be another halide or alkali or alkaline-earth salt. The nature of the
dissolved salts affects colloidal clays and other additives and thus must be mon-
itored together with properties such as salinity, total hardness (calcium plus
magnesium), pH, and alkalinity. Drilling fluids are nearly always basic with a
pH in the range of 6–13, depending on the type of system used. Concentrations
of soluble carbonates, sulfide, sulfite, etc, may also be determined. These ions
may be added intentionally or incorporated during drilling.
258 DRILLING FLUIDS Vol. 1

4. Drilling Fluid Materials

The pressure exerted by the column of drilling fluid in the well balances forma-
tion pressures to prevent uncontrolled influx of formation fluids that may result
in a blowout. The mud density must be controlled accurately by suitable weight-
ing materials that do not adversely affect the other properties. Most important is
the specific gravity of the weighting agent as well as its insolubility in water and
its chemical inertness. The weighting material should be ground to the preferred
particle-size distribution and be relatively nonabrasive. Various finely ground,
solid weighting materials that have found application in drilling are listed in
Table 1. As of 1995, little weighting material other than barite was used in the
United States.
Barite, which is predominately composed of BaSO4, meets the overall
requirements for weighting material better than other materials and is used
for increasing the density of drilling fluids throughout the world. Barite in its
pure form has a density of 4500 kg/m3. Commercial barite has a lower specific
gravity than pure barium sulfate owing to the presence of associated minerals,
such as silica. Barite is virtually insoluble in water and does not react with other

Table 1. Solid Materials Used to Increase Density of Drilling Muds


CAS Registry Specific Hardness,
Material number Formula gravity Mohs’ Characteristics Advantages
a
barite [13462-38-2] BaSO4 4.5 , 2.5–3.5 white gray-red readily
4.2b available;
low cost
hematite [1317-60-8] Fe2O3 4.9–5.3 5.5–6.5 iron oxide, low attrition
impurities; rate; API
black to red high
depending on density
particle size
magnetite [1309-38-2] Fe3O4 5.0–5.2 5.5–6.5 iron ore, often high density;
Ti and Mg; HCl
black ore soluble;
scavenges
H2 S
siderite [14476-16-5] FeCO3 3.7–3.9 3.5–4 spathic iron acid soluble
ore; various
colors
dolomite [16389-88-1] CaCO3 2.8–2.9 3.5–4 carbonate of acid soluble
MgCO3 Mg, Ca
calcite [13397-26-7] CaCO3 2.6–2.8 3 limestone, highly acid
occurs in soluble,
sedimentary range of
rock particle
sizes
sodium [7647-14-5] NaCl 2.165 2 cubic structure, water soluble,
chloride used as
bridging
solids
a
Value is for pure material.
b
Value for API grade.
Vol. 1 DRILLING FLUIDS 259

Table 2. API Specifications for Barite and Hematite


Assay Baritea Hematiteb
specific gravityc 4.20 5.05
wet-screen analysis, % residued
>75 mm 3.0 1.5
>45 mm 15
particles, < 6 mm, %d 30 15
soluble alkaline-earth metals as calcium, mg/kgd 250 100
a
Test equipment and procedures from Ref. 21.
b
Test equipment and procedures from Ref. 22.
c
Value given is minimum.
d
Values given are maximum.

mud constituents. Most operators prefer barite that meets API specifications
(Table 2) (21). The barite content in mud depends on the desired density but it
can be as high as 2000 kg/km3 (700 lb/bbl).
Drilling technology favors systems having the lowest possible solids content
because of the resulting higher penetration rate, easier control of mud proper-
ties, and fewer problems experienced during drilling. Alternative weighting
materials having specific gravities higher than that of barite offer this advan-
tage. The only other high specific gravity material that is used to any degree
as of 1995 is hematite (iron oxide).
The specifications for drilling fluid hematite have been set by the API and
are listed in Table 2 (22). Hematite is used most frequently in high density oil-
based muds to minimize the total volume percent solids (23). The abrasivity of
hematite limits its utility in water-based muds.
Calcite and siderite (24) are used occasionally because of their solubility in
hydrochloric acid, which offers a method of removing mud filter cake deposited
on productive formations. The calcite (calcium carbonate) is supplied in the form
of ground marble, with the particle size adjusted to the porosity of the reservoir,
Calcite and siderite are used most frequently in workover or completion fluids
when a nondamaging fluid is required, ie, one that can be removed by acidizing
at a later time.
Solid salt, ground and packaged in several particle size grades, can be used
in saturated salt brines to increase the fluid density (25). However, sized salt is
also as a weighting agent, having specific use in sensitive reservoirs. In such
cases, an oversaturated solution is used to form a filter cake that can easily be
removed with fresh water flushing. This nondamaging has found increasing
use since the early 1990s. At one time the sized salt systems were used primarily
for completion or workover operations, but use has also increased as in drill-in
fluids for horizontal wells (26).
The use of solids-free fluids or clear brine fluids (CBF) is occasionally
resorted to for achieving high drilling rates, and also in workover and completion
operations, for avoiding damage arising from particulate invasion of the produc-
tive formations. Clear brine fluids and their additives are discussed more exten-
sively in the Section 5, Completion and Workover Fluids.
The chemical and mechanical dispersion of the drilled solids tends to
increase the percentage of small-sized solids in a mud as drilling progresses.
260 DRILLING FLUIDS Vol. 1

The incorporation of a limited amount of drilled solids (several volume percent) is


an economical way of increasing the density of low density muds, but it also
reduces penetration rates; hence, drilled solids are usually kept to a minimum.
The common clay and formation solids and their specific gravities encountered in
normal drilling operations are as follows: dolomite (2.8–3.2), limestone (2.8–3.0),
feldspar (2.6–2.7), sandstone (2.3–2.6), clay (2.3–2.6), bentonite (2.3–2.4), salt
(2.16) cement (1.6–2.0), and coal (1.35). The cement is encountered when drilling
a cement plug or out of new casing.
4.1. Viscosity Buildup. The drilling fluid removes cuttings from the
wellbore as drilling progresses. This process is governed by the angle of the hole
and the velocity at which fluid travels up the annulus, as well as by the fluid visc-
osity or flow properties, and fluid density. The cuttings removal efficiency usually
increases with increasing viscosity and density, although at high wellbore angles
a less viscous fluid may be desirable provided high flow rates can be achieved.
Viscosity depends on the concentration, quality, and state of dispersion of sus-
pended colloidal solids.
Although numerous mud additives aid in obtaining the desired drilling fluid
properties, water-based muds have three basic components: water, reactive
solids, and inert solids. The water forming the continuous phase may be fresh-
water, seawater, or saltwater. The reactive solids are composed of commercial
clays, incorporated hydratable clays and shales from drilled formations, and
polymeric materials, which may be suspended or dissolved in the water phase.
Solids, such as barite and hematite, are chemically inactive in most mud sys-
tems. Oil and synthetic muds contain, in addition, an organic liquid as the
continuous phase plus water as the discontinuous phase.
The most important commercial clays used for increasing the viscosity of
drilling fluids are bentonite [1302-78-9], attapulgite [1337-76-4], and sepiolite
[15501-74-3]. For oil-base and synthetic-base muds, organophilic clays are used.
These clays produce viscosity and help suspend weighting materials. The active
ingredient in bentonite is a smectite, montmorillonite [1318-93-0], an expandable
2:1 aluminosilicate mineral, having a three-layer plate-shaped crystalline struc-
ture. The three-layer sheets or platelets consist of a middle octahedral alumina
layer and two outer tetrahedral silica layers. Because of lattice defects in the alu-
mina, and less often in the silica layers, the flat planar surfaces are negatively
charged and have associated cations (primarily sodium and calcium) to achieve
electroneutrality. Bentonite hydrates in the presence of freshwater and disperses
to varying degrees, depending on the nature of the cations that are loosely held
and exchangeable. Hydration and dispersion are enhanced by the presence of
sodium ions in the exchange or planar surface positions. Divalent cations such
as calcium reduce the degree to which the individual platelets disperse. As
hydration of bentonite proceeds in freshwater, the individual platelets separate
and eventually form a stable colloidal dispersion that is stabilized by electrical
interactions between clay platelets (27). This colloidal system has highly non-
Newtonian rheological properties and the suspending and shear-thinning char-
acteristics desired in a drilling fluid. The hydration and dispersion of bentonite
clay, and thus the viscosity of the final system, can be altered considerably by the
presence of electrolytes in the water, exchange reactions that convert the clay
from a high swelling form (sodium) to a low swelling form (such as calcium), or
Vol. 1 DRILLING FLUIDS 261

Table 3. API Specifications for Bentonite Claya


Nontreated OCMA
Assay Bentonite bentonite bentonite
bentonite content in 350 cm3 water, g 22.5 25.0 22.5
suspension properties
viscometer dial reading at 600 rpmb 30 10c 30
yield point/plastic velocity ratiod 3 1.5 6
filtrate volumed, cm3 15.0 12.5 16.0
residue, % >75mmd 4.0 2.5
moisture, %d 10.0 13.0
a
Test equipment 1 and procedures from Ref. 28.
b
Value given is minimum.
c
Value is dispersed plastic viscosity in mPasð¼ cPÞ.
d
Value given is maximum.

adsorption of polyelectrolytes which prevent separation of the clay platelets.


Bentonite concentrations in drilling fluids vary widely, but may range up to
100 kg/m3 (35 lb/bbl). Specifications for the three grades of drilling fluid bento-
nite recognized by the API are listed in Table 3 (28). The higher performance
grades are produced mainly in the Wyoming–Montana–South Dakota area.
This bentonite contains montmorillonite clay in both sodium and calcium
forms. The sodium form predominates. Most such bentonites are processed
using a small amount of a peptizing polymer to enhance the viscosity building
properties of the clay. This is the standard API bentonite. High quality bentonite
that is not treated in any way to enhance its viscosity building characteristics is
sold as API nontreated bentonite.
Fast-yielding bentonite treated with high molecular weight polyacrylamide
and polyacrylate polymers is also available commercially. It produces roughly
the same viscosity as twice the amount of untreated bentonite. Additions of
small amounts of certain polyacrylamides, called bentonite extenders, to a fresh-
water mud containing bentonite give the same effect. Drilling fluids formulated
with bentonite and an extending polymer in this manner are called low solids,
nondispersed muds (29).
Another bentonite extender, which also enhances the viscosity of attapul-
gite slurries, is a mixed metal (magnesium and aluminum) layered hydroxide
[7732-18-5], usually called MMH (30). This material is most frequently used to
increase the viscosity sufficiently so that metal cuttings can be removed from
the well during milling operations. MMH-treated muds are also finding appli-
cation in horizontal wells. A mixed metal (calcium, magnesium, and aluminum)
silicate [1327-39-5] (MMS), is being applied in a similar way.
When the water phase of the drilling fluid contains substantial amounts
of electrolyte, saltwater clays such as attapulgite and sepiolite are added
to raise viscosity. Attapulgite is used solely for its suspending qualities.
It has a fibrous texture and crystalline, needle-like, hydrated magnesium
silicate particles. Attapulgite clays increase viscosity regardless of the compo-
sition of the makeup water. This ability does not depend on hydration,
but rather on the extent to which the bundles of needles are broken up by a
shearing force.
262 DRILLING FLUIDS Vol. 1

Table 4. Properties of Attapulgite and Sepiolite


Assay Value
3
clay concentration in 350 cm saturated salt water, g 20
minimum viscometer dial reading at 600 rpm 30
maximum residue, %>75 mm 8.0
maximum moisture, % 16.0

Sepiolite is a hydrated magnesium silicate that contains less substituted


aluminum than attapulgite, which it closely resembles. Sepiolite occurs in
fibrous and elongated lath-like particles. It is more stable at higher temperatures
than attapulgite (31,32) and therefore is used in geothermal drilling fluids. The
properties of attapulgite and sepiolite specified by the API for drilling fluid use
are given in Table 4 (33).
Oil-dispersible or organophilic clay provides viscosity and suspending qua-
lities in oil-based muds. It is prepared from bentonite, hectorite [12173-47-6], or
attapulgite and aliphatic amine salts. The products obtained from amines having
12 or more straight-chain carbon atoms swell and form gels in hydrocarbon fluids
(34–36). The amino groups replace the sodium and calcium originally present on
the clay surface. Oil-dispersible clays can suspend solids in oil without requiring
additional soaps and emulsifying agents. Addition levels of organophilic clays are
3–11 kg/m3 (1–4 lb/bbl) in diesel oil muds and 8–23 kg/m3 (3–8 lb/bbl) in
mineral oil and synthetic-base muds.
A wide variety of organic polymers serve a number of useful purposes in
drilling fluids, the most important of which are to increase viscosity and control
filtration rates (Table 5). These polymers are either natural polysaccharides, eg,
starch [9005-25-8], guar gum, xanthan gum [11138-66-2], and other biopolymers;
or derivatives of natural polymers, eg, cellulose, lignosulfonate, and lignite and
synthetic polymers, eg, polymers and copolymers of acrylic acid, acrylonitrile,
acrylamide, and 2-acrylamido-2-methylpropanesulfonic acid (AMPS). The most
commonly used polymeric viscosity builders are the cellulosics, xanthan gum,
and polyacrylamides.
Sodium carboxymethyl cellulose [9004-32-4] (CMC) and hydroxyethyl cellu-
lose [9004-62-0] (HEC) are the cellulosics most widely used in drilling fluids (37).
CMC is manufactured by carboxymethylation of cellulose, which changes the
water-insoluble cellulose into the water-soluble CMC (38). Hydroxyethyl cellu-
lose and carboxymethyl hydroxyethyl cellulose (CMHEC) are made by a similar
process. The viscosity grade of the material is determined by the degree of sub-
stitution and the molecular weight of the finished product.
The effectiveness of sodium carboxymethyl cellulose, an anionic polymer, as
a viscosity builder decreases with increasing electrolyte concentration. This poly-
mer can be coprecipitated with calcium and magnesium by raising the pH of the
mud. The polyanionic cellulose (PAC), which has a higher degree of substitution
than CMC, was introduced to overcome some of these limitations (37). CMC and
PAC are available in several viscosity and purity grades. API specifications for
high viscosity, technical-grade CMC for use as a viscosity builder are listed in
Table 6 (39). The primary application of both CMC and PAC is in the control
of filtration rates.
Vol. 1 DRILLING FLUIDS 263

Table 5. Polymers Used in Water-Based Drilling Fluids


Thermal
Chemical degrada-
Polymer description Ionic state tion, 8C application limitations
Fermentable
starch polysaccharide: nonionic 110 filtration fermentable at
amylose, low pH and
amylopectin salinity
xanthan gum microbial anionic 140 viscosity poor yield in
polysaccharide builder calcium
brines
hydroxy- polysaccharide: nonionic 95 viscosity low tempera-
propyl galactose, builder ture stability
guar mannose
Unfermentable
CMC/PAC sodium anionic 140 filtration sensitive to
carboxy-methyl control, salinity,
cellulose viscosity multivalent
builder ions
HEC hydroxyethyl nonionic 110 viscosity primarily for
cellulose builder, acid completion/
degradable workover
fluids
polyacrylates, sodium mixed 200 dispersant, poor perfor-
polyamides polyacrylate, filtration mance in
polyacryl- control, presence of
amide, viscosity multivalent
polyacrylic builder,shale ions
acid-co-
acrylamide stabilizer
lignosulfonate sulfonated lignin, anionic 175 deflocculant, requires
cations vary but filtration alkaline
are often Cr control conditions,
may contain
heavy metals
tannin quebracho sulfo- anionic 200 deflocculant wide pH range
methylated
lignite leonardite anionic 205þ filtration requires
control, alkaline
deflocculant conditions,
sensitive to
salinity and
alkaline
conditions

Hydroxyethyl cellulose (HEC), a nonionic thickening agent, is prepared


from alkali cellulose and ethylene oxide in the presence of isopropyl alcohol
(40). HEC is used in drilling muds, but more commonly in completion fluids
where its acid-degradable nature is advantageous. Magnesium oxide stabilizes
the viscosity-building action of HEC in salt brines up to 1358C (41).
Xanthan gum is a high molecular weight microbial polysaccharide produced
by the bacterium Xanthomonas campestris (42). Commercially, xanthan gum is
produced by a fermentation process and precipitation of the gum in alcohol. It is
264 DRILLING FLUIDS Vol. 1

Table 6. API Specifications for Technical-Grade High Viscosity CMCa


CMC concentration, Minimum viscometer Maximum filtrate
Solvent system g/350 cm3 dial reading at 600 rpm volume, cm3
distilled water 2.20 30
40 g/L salt water 2.70 30
saturated salt water 2.50 30
slurryb 3.15 10
a
Test equipment and procedures from Ref. 39.
b
Slurry contains 35 g of API Standard Evaluation Base Clay per 350-cm3 saturated salt solution.

a viscosity builder and suspending agent and can be used in almost any type of
water (43). Although xanthan gum solutions support bacterial growth, a preser-
vative is usually not needed unless the solution is to be stored. Because of sus-
pending ability at low concentrations and in electrolyte solutions, xanthan gum
is widely used for drilling, workover, and completion fluids. Two other biopoly-
mers, succinoglucan gum [39464-87-4] (44) and welan gum [96949-22-3]
(42,45), are also finding some use in drilling fluids at concentrations similar to
xanthan gum.
Guar gum is a nonionic, branched-chain polysaccharide, a galactomannan
that is usually hydroxypropylated for use in drilling (46). It produces viscous
solutions in fresh or saltwater at concentrations of 3–6 kg/m3 (1–2 lb/bbl). It
is used in solids-free and low solids muds and degrades rapidly above 808C, limit-
ing its use to shallow wells.
High molecular weight polyacrylamides are used as viscosity builders in
freshwater muds (47) or as bentonite extenders. The ionic nature of the polyacryl-
amide may range from nonionic to anionic (30% hydrolyzed) depending on the
situation.
Occasionally polymers are used to increase the viscosity of oil-base and
synthetic-base muds. The polymers for this use are typically sulfonated polystyr-
enes or ethylene–propylene terpolymers (EPDM) (48,49). Such polymers are
usually used in conjunction with an organophilic clay.
4.2. Viscosity Reduction. Proper control of viscosity and gel strengths
is essential for efficient cleaning of the borehole, suspension of weight material
and cuttings when circulation is interrupted, and to minimize circulating
pressure losses and swab/surge pressures owing to axial movement of the
drill string. Viscosity may be increased as previously indicated, but there is
often the necessity of reducing the viscosity. A reduced viscosity can be achieved
by thinning or deflocculating clay–water suspensions. Thinning is measured as
a reduction of plastic viscosity, yield point, or gel strength, or a combination of
these properties. Viscosity is reduced by decreasing the solids content and the
number of particles per unit volume, or by neutralizing the attractive forces
between particles. Although mud thinners or viscosity-reducing chemicals are
added to reduce flow resistance and gel development, reduction of filtration
rates and filter cake thickness and stabilization of mud properties at elevated
temperatures often results. Typical mud-thinning chemicals are polyanionic
materials that are adsorbed on positive edge sites of the clay particles, thereby
reducing the attractive forces between the particles without affecting clay
hydration (27,50).
Vol. 1 DRILLING FLUIDS 265

Table 7. Phosphate Drilling Mud Thinners


CAS Registry Molecular pH of aqueous
Compound number formula solution Observations
sodium acid pyrophos- [7758-16-9] Na2H2P2O7 4.2 for cement
phate (SAPP) determination
sodium tetraphosphate [7727-67-5] Na6P4O13 7.5 generally
preferred
tetrasodium pyrophos- [7722-88-5] Na4P2O7 10
phate (TSPP)
sodium hexametaphos- [10124-56-8] (NaPO3)6 7 glassy material
phate

Thinners or deflocculants for clay–water muds include polyphosphates,


tannins, lignites, lignosulfonates, and low molecular weight polyacrylates and
their derivatives. These materials also can remove chemical contaminants by
precipitation or chelation. Sodium polyphosphates are effective deflocculants
for clays in freshwater and were among the first thinners used in drilling fluids
(51). The reversion of the polyphosphates to orthophosphates occurs rapidly as
the temperature approaches 958C and may cause thickening of the mud. Table 7
gives the four most commonly used phosphates. The application of polyphos-
phates is limited to temperatures <808C, and supplemental control along with
organic thinners.
Tannins occur in many plants and are separated by extraction. Quebracho
extract is an acidic mud thinner and performs best at high pH. It is an excellent
thinner for lime-treated and cement-contaminated muds. However, it is not
effective at high salt concentrations. Sulfomethylated tannin products are func-
tional over a wide range of pH and salinity and have either been treated with
chromium for good thermal stability (52) or are chrome free.
Lignite, generally leonardite, and lignite derivatives are applied in water-
based muds as thinners and filtration control agents. Leonardite is an oxidized
lignite having a high content of humic acids, which may be described as carboxy-
lated phenolic polymers (53,54).
Natural lignite is not a good thinner for low pH water-based muds at mod-
erate temperatures, but can be an excellent thinner or mud conditioner at high
pH and high temperatures where the humic acids are solubilized. It has better
temperature stability than most plant tannins or lignins. The performance of lig-
nite can be greatly improved by solubilizing processes such as caustic treatment,
sulfonation, sulfomethylation, or chelation with heavy metals.
Lignosulfonate thinners are among the most versatile and important che-
micals used in water-based drilling fluids, and are the largest volume additives
used in this category worldwide. Lignosulfonates are by-products of the sulfite
process for separating cellulose pulp from wood. These can be described as poly-
anionic natural polymers. Their mud-thinning quality depends on the source of
the lignin, degree of sulfonation, molecular weight distribution, impurities, and
the amount and type of chelated heavy metals (55). Commercial chrome lignosul-
fonate [9066-50-6] contains up to 3.5–4.5 weight% trivalent chromium. The
chrome-free lignosulfonates may not be as effective overall as those containing
trivalent chromium, but they are useful where chromium is undesrable for
266 DRILLING FLUIDS Vol. 1

environmental reasons. In fact, the chrome-based lignosulfonates have been


banned in many European countries due to their toxicity. Other cations asso-
ciated with the lignosulfonate molecule include sodium, potassium, calcium,
and titanium–zirconium (56).
Chrome and ferrochrome lignosulfonates are effective deflocculants in most
water-based muds over a wide range of salinity, hardness, and pH (8.5–12.5).
Lignosulfonates at high concentrations provide some filtration control and inhi-
bit disintegration and dispersion of shale cuttings. Thinning effectiveness
declines at 120–1508C, although field experience confirms that muds containing
chrome lignosulfonate can be used at bottomhole temperatures of >1758C.
The effectiveness of lignosulfonates at high temperatures depends on the solids
content and the type and concentration of electrolytes. Low molecular weight
(1000–5000) polyacrylates and copolymers of acrylic acid and AMPS are used
as dispersants for weighted water-base muds (57). These materials, 40–50%
of which is the active polymer, are usually provided in a liquid form. They are
particularly useful where high temperatures are encountered or in muds,
which derive most of their viscosity from fine drill solids, and polymers such as
xanthan gum and polyacrylamide. Another high temperature polymer, a sulfonated
styrene maleic–anhydride copolymer, is provided in powdered form (58,59). All
of these materials are used in relatively low concentrations in the mud.
A hydrolyzed cereal solid, predominately a hexasaccharide, is used in high
pH lime muds for reducing the yield point and gel strength (60). This additive
has been used in systems treated with both sodium hydroxide and potassium
hydroxide in addition to other additives common to lime muds (61). A second
viscosity-reducing additive used in lime muds is a graft copolymer of acrylic
acid and calcium lignosulfonate (62).
Filtration control is particularly important in permeable formations where
the mud hydrostatic pressure exceeds the formation pressure. Proper filtration
control reduces drill-string sticking and drag, and rotary torque, as well as mini-
mizing damage to protective formations; in some formations it improves borehole
stability. Several types of materials are available for water-based muds and
application varies according to the type and the chemical environment of the
mud. These include clays, organic polymers, and lignite derivatives. The bento-
nite present in the system often acts as the primary filtration control agent. It
not only develops viscosity, but also lowers the filtration rate, particularly in
freshwater muds. The ability of bentonite clay to control filtration is attributed
to the flat, plate-like particle shape, the capacity to disperse and hydrate, the
ability to form a compressible filter cake, and the colloidal to near-colloidal
particle size.
Hydrating bentonite in freshwater before adding it to the mud greatly
increases its efficiency when the makeup water is contaminated with salt and/
or hardness. Prehydrated bentonite can be protected from dehydration by ligno-
sulfonate or sulfomethylated tannin when used in saturated saltwater. Salt-
water clays, such as sepiolite and attapulgite, provide no filtration control and
are normally used with suitable filtration control agents.
Although a combination of bentonite clay and an organic thinner provides
filtration control in many water-based muds, additional control generally is
needed. Filtration additives for both fresh- and saltwater muds are usually
Vol. 1 DRILLING FLUIDS 267

organic polymers and lignites (see Table 5). Cost and viscosity-building charac-
teristics should be considered, particularly in weighted high solids muds.
Starches, used first in the late 1930s for filtration control, are still in use in
the 1990s. Cornstarch is most commonly used in the United States. Potato starch
is also used in the United States but primarily in Europe and elsewhere. Both
corn and potato starches are pregelatinized before dispersion in water (63).
The API specifications for drilling fluid starch are listed in Ref. 64.
Starch is subject to fermentation by many microorganisms and, unless the
mud is saturated with salt or the pH is 11.5, a preservative or biocide must be
added if the mud is to be used for an extended period of time. The most common
biocide until the mid-1980s was paraformaldehyde [9002-81-7]. This material
has been largely replaced by isothiazolones (at 5–10 ppm conc) (65), carbamates,
and glutaraldehyde [111-30-8]. Alternatively, the biocide may be incorporated
during the processing of the starch and is present in the commercial product.
Starch is degraded by heat and agitation. Under continued circulation at
 958C, starch can breakdown rapidly. Its performance also is affected by pH
changes in the presence of calcium and magnesium ions. Starch coprecipitates
with calcium when caustic soda is added to mud containing dissolved calcium
salts. Starch molecules are hydrolyzed at very low or high pH in the presence
of calcium and magnesium >1108C. An acidic environment also may prevent
hydration and thus limit the effectiveness of starch as a filtration control agent.
Numerous modifications and derivatives of starch have been made for
application in drilling and workover fluids (66). Most modified starches are
cross-linked to some degree to improve thermal stability. Carboxymethyl and
hydroxypropyl starches are finding increasing application in drilling fluids.
There are some applications involving cationic starches as well, although toxicity
must be considered for offshore use.
Carboxymethyl cellulose (CMC) and polyanionic cellulose (PAC) are avail-
able in several viscosity grades and high and low purity grades. All grades can be
effective filtration control agents depending on the well conditions. The effective-
ness of CMC in reducing filtration decreases with increasing salt concentration.
A polyanionic cellulose (PAC) polymer is available that is designed particularly
for application in muds of high salt concentration (37,67). Most CMC and PAC
materials are sodium salts but there are also potassium versions to take advan-
tage of the shale stabilizing properties of this ion (68). While not designed to con-
trol filtration, HEC may be effective as a filtration control agent in combination
with other organic polymers in waters having salinities up to saturation.
Acrylate and acrylamide polymers have several uses in drilling fluids, one
of which is for filtration control. Sodium polyacrylates [9003-04-7] having mole-
cular weights near 250,000 are excellent temperature-stable filtration control
agents for both fresh- and saltwater muds, provided the concentration of
water-soluble calcium is <400 mg/L (69). The calcium ions are precipitated
using a carbonate such as soda ash, before adding the polyacrylate.
Lignite products, mined, ground, and possibly treated with sodium or potas-
sium hydroxide, are economical filtration control additives for some water-based
muds, in addition to improving flow properties (70). A sulfonated lignite
complexed with a sulfonated phenolic resin is an effective high temperature,
filtration control additive, for both fresh- and seawater muds that contain high
268 DRILLING FLUIDS Vol. 1

concentrations of soluble calcium, and does not affect viscosity (71,72). A high
molecular weight polyanionic lignin has also found application for high tempera-
ture muds with a high electrolyte content (73). These and similar products can
provide filtration control for high density, high solids muds above 1208C.
A number of synthetic polymers having the ability to control filtration rates
at high temperature and in the presence of calcium and magnesium have also
been developed (74). Such materials include vinyl sulfonate–vinyl amide copoly-
mers (75,76), a copolymer of AMPS and N,N-dialkyl (meth) acrylamide (77) and a
sulfonated hydroxylated polymer (78).
Filtration control in oil- and synthetic-base fluids is achieved by the emul-
sified aqueous phase, by the emulsifier package, and by additions of powdered
solid materials. The powdered solids used for this purpose consist of asphalt
[8052-42-4], gilsonite [12002-43-6], and amine-treated lignite. The choice of addi-
tive depends on the nature of the base fluid, the emulsifier package, and most
importantly the downhole temperature. Styrene–butadiene copolymers in the
form of aqueous dispersions have also found application as filtration control
agents in oil-base muds.
Cellulosic fibers, powdered limestone, gilsonite, and asphalt are frequently
added to both water and oil muds at levels of 10–25 kg/m3 (4–10 lb/bbl) when
high differential pressures are encountered to control seepage losses to the for-
mation. This treatment also is used to improve the quality of the mud filter cake
to reduce the chance of differential pressure sticking.
4.3. Alkalinity Control. Water-base drilling fluids are generally main-
tained at an alkaline pH. Most mud additives require a basic environment to
function properly and corrosion is reduced at elevated pH. The primary additive
for pH control is sodium hydroxide [1310-73-2].
The second most common alkalinity control agent is lime [1305-78-8], nor-
mally in the form of calcium hydroxide [1305-62-0], used in both water and oil
muds. In the latter, the lime reacts with added emulsifiers and fatty acids to sta-
bilize water-in-oil emulsions. Lime is used in brine systems containing substan-
tial quantities of soluble calcium and in high pH lime muds.
Potassium hydroxide [1310-58-3] is occasionally used for alkalinity control.
This is particularly true for some polymer and lime muds where a low sodium
level is desired. The potassium level of such muds is quite low but has been
attributed by some to provide stability to water-sensitive shale formations (61,79).
A fourth alkalinity control additive is magnesium oxide [1309-48-4], which
is used in clay-free polymer-base fluids (41). Magnesium oxide provides an alka-
line environment and, as it is only slightly soluble, also has a buffering effect. It
enhances the thermal stability of polymer solutions by preventing a pH decrease
to neutral or slightly acidic conditions at elevated temperatures. It is mainly
applied in completion or workover operations where clay-free acid-soluble fluids
are desired. There are occasions where the mud pH must be lowered such as
after drilling fresh cement or overtreatment by one of the alkaline materials dis-
cussed. Organic acids that have been used for this purpose include acetic acid
[64-19-7], citric acid [77-92-9], and oxalic acid [144-62-7], but these materials
have been employed only infrequently. Inorganic additives used to lower pH
levels include sodium bicarbonate [144-55-8] and sodium acid pyrophosphate
[7758-16-9] (SAPP). Of the two, sodium bicarbonate is used the most by far.
Vol. 1 DRILLING FLUIDS 269

4.4. Removal of Contaminants. A drilling fluid contaminant is any


material or condition encountered during drilling operations that adversely
affects the performance of the fluid. Elevated temperatures and drill solids are
encountered in every drilling operation. In most wells these are handled easily,
but in some wells one or both can seriously reduce drilling efficiency. Tempera-
ture problems normally are treated by using viscosity or filtration control addi-
tives, materials having better thermal stability, or possibly by replacement of the
mud system with an oil or synthetic mud. Drill solids are removed mechanically
by various combinations of screens, hydrocyclones, and centrifuges, or chemi-
cally by flocculants. Dilution or replacement of part or all of the mud system
may reduce drill solids to tolerable levels.
Various inorganic chemicals remove soluble contaminants encountered
during drilling. Salt, NaCl, is a common contaminant that can be removed
only by dilution. The adverse effects of salt, primarily clay flocculation, can be
overcome by a deflocculant such as a lignosulfonate or sulfomethylated tannin.
Soluble calcium from cement, gypsum, or anhydrite is frequently encountered
and has a flocculating effect on drilling clays. Calcium is removed using a phos-
phate, sodium carbonate [497-19-8], sodium bicarbonate (particularly in the case
of cement contamination), and occasionally oxalic acid.
If calcium problems persist, the mud may be converted to a calcium-based
system by adding a calcium source, such as gypsum [13397-24-5] or lime.
Gypsum and lime are also used to control bicarbonate/carbonate ions that can
cause mud gelation and rheological problems, particularly in high density
muds in deep, hot wells (80). Sources of carbonate include carbon dioxide bearing
zones and soluble carbonate formations. Carbon dioxide is also generated by
some mud additives and by excessive treatment for removal of calcium.
4.5. Stabilization of Water-Sensitive Formations. Many subsurface
formations encountered during drilling are water-sensitive shales containing
various amounts of clay minerals. The clay mineral components may include a
highly swelling smectite or less water-sensitive illite, mixed-layer smectite–
illite, kaolinite, or chlorite. All shales appear to swell to some extent when con-
tacted by fresh water. Those containing smectite or mixed layer clays are much
more sensitive than illitic shales. The uptake of water by shales has two effects: a
volume change owing to swelling, and a strength reduction as the water content
increases. This may result in flow of plastic shale into the wellbore, softening and
erosion of the exposed shale, or spalling of hard shale, all of which can cause
expensive operational problems. In the latter case, large hard pieces of formation
fall around the drill string, the removal of which may become difficult.
A variety of methods have been devised to stabilize shales. The most suc-
cessful method uses an oil or synthetic mud that avoids direct contact between
the shale and the emulsified water. However, preventing direct contact does not
prevent water uptake by the shale, because the organic phase forms a semi-
permeable membrane on the surface of the wellbore between the emulsified
water in the mud and the water in the shale. Depending on the activity of
the water, it can be drawn into the shale (activity lower in the shale) or into
the mud (activity higher in the shale) (81–83). This osmotic effect is favorable
when water is drawn out of the shale; thus the aqueous phase of the oil or syn-
thetic mud is maintained at a low water activity by adding a salt, either sodium
270 DRILLING FLUIDS Vol. 1

chloride or more commonly, calcium chloride. The salt concentration is carried


somewhat above the concentration required to balance the water activity in
the shale to ensure water movement into the mud.
High initial cost and environmental restrictions prevent use of oil and syn-
thetic muds in many cases where shale problems are expected. It is necessary
then to treat a water-base mud to minimize the destabilizing effect of the drilling
fluid. Salts, polymers, and other organic materials are added to the mud to
reduce the water sensitivity of the shale, shale swelling, and weakening arising
from mud contact, or the rate of water uptake by the shale.
Addition of a salt can transform the shale by cation exchange to a less sen-
sitive form of clay, or reduce the osmotic swelling effect by reducing the water
activity in the mud below that which occurs in the shale. These effects depend
on the salt concentration and the nature of the cation. Salts containing sodium,
potassium, calcium, magnesium, and ammonium ions are used to varying
degrees. Sodium chloride has long been used as a shale stabilizer because of
low cost, wide availability, and its presence in many subsurface formations.
The inhibitive nature of salt muds increases as the salt content increases from
seawater to saturated sodium chloride. In addition to the sodium chloride con-
sumed annually for drilling fluid, considerable quantities are incorporated
while drilling salt zones. This material has been used more for minimizing
washouts in salt zones than for stabilizing shales. High salt levels have found
application in deep water drilling.
Calcium sources, such as gypsum and lime, promote cation exchange from
sodium clay to a less-swelling calcium clay. Calcium concentrations are normally
low and osmotic swelling is only reduced if other salts are present. Calcium chlor-
ide has been used infrequently for this purpose but systems are available that
allow high calcium chloride levels to be carried in the mud system (84).
A variety of shale-protective muds are available that contain high levels of
potassium ions (13). The reaction of potassium ions with clay, well known to soil
scientists, results in potassium fixation and formation of a less water-sensitive
clay. Potassium chloride, potassium hydroxide, potassium carbonate [584-08-7]
(85), tetrapotassium pyrophosphate [7320-34-5] (86), have all been used as the
potassium source. Potassium chloride is generally preferred because of its low
cost and availability. It is generally used in combination with a polymer or
other organic material as discussed below.
Ammonium chloride [12125-02-9], ammonium sulfate [7783-20-2], and di-
ammonium phosphate [7708-28-0] have also been used for shale stabilization
(87,88). Ammonium ions have essentially the same effect on shales as potassium
ions but use of ammonium salts is often objectionable because of the alkaline
nature of the mud. In the North Sea and northern Europe, where magnesium-
bearing salt formations are encountered, magnesium chloride [7786-30-3] is
used, but in the United States it is used only on a small scale.
A number of nonionic and anionic polymers are employed in water-based
muds to stabilize shales. These may be added to a freshwater mud or to a system
containing one of the salts mentioned. Typically, shale-stabilization polymers
include modified starches (89); cellulosic polymers such as CMC and HEC;
gums such as guar, xanthan (90), and flax meal (91); and high molecular weight
polyacrylamides of varying degrees of hydrolysis (12,47,92). A fluid containing a
Vol. 1 DRILLING FLUIDS 271

combination of potassium silicate [1312-76-1] and poly(vinyl alcohol) [9002-89-5]


has also been used (93).
A number of cationic muds have been developed and used. These are formu-
lated around quaternary amines or positively charged polymers (94,95). The
polymer in some instances may be a cationic polyacrylamide. Poly(dimethyl-
amine-co-epichlorohydrin) is another material that has been used successfully
for drilling shale formations (96,97). Some of these additives may require a
salt such as sodium or potassium chloride for achieving best results.
A number of glycol and glycerol-base additives are being used to formulate
shale protective muds usually in conjunction with a salt and/or a polymer (98–
101). The glycols consist of a number of combinations of ethylene oxide and pro-
pylene oxide depending on the purpose of the additive, not all of which are
strictly for shale stabilization, and the desired cloud point. Two polyglycerol
materials have been touted as having shale-protective properties (102–104).
A low molecular weight poly(amino acid) has also been touted for its shale-
stabilizing properties (105,106).
Solid materials, such as gilsonite and asphalt, and partially soluble
sulfonated asphalt may also be added to plug small fractures in exposed
shale surfaces and thereby limit water entry into the formation (91,107).
The asphalts are oxidized or treated to impart partial solubility. These mate-
rials may be softened by the downhole temperature, causing them to deform
and squeeze into small openings exposed to the borehole. Laboratory tests
designed to evaluate shale-stabilizing muds have confirmed the beneficial
action of these materials (108).
4.6. Surfactants. Surfactants perform a variety of functions in a drilling
fluid. Depending on the type of fluid, a surfactant may be added to emulsify oil in
water (o/w) or water in a nonaqueous liquid (w/o), to water-wet mud solids or to
maintain the solids in a nonwater-wet state, to defoam muds, or to act as a foam-
ing agent.
Lignites and lignosulfonates can act as o/w emulsifiers, but generally are
added for other purposes. Various anionic surfactants, including alkylarylsulfo-
nates and alkylaryl sulfates and poly(ethylene oxide) derivatives of fatty acids,
esters, and others, are used. Very little oil is added to water-base muds in use
offshore for environmental reasons. A nonionic poly(ethylene oxide) derivative
of nonylphenol [9016-45-9] is used in calcium-treated muds (109).
Emulsifiers are incorporated in oil and synthetic mud formulations to main-
tain a stable emulsion of the internal brine phase. These materials include cal-
cium and magnesium soaps of fatty acids and polyamines and amides and their
mixtures (110,111). The specific chemistry of these additives depends on the
nature of the continuous phase of the mud, ie, whether it is diesel oil, mineral
oil, or a synthetic liquid. Lime is added along with the fatty acid to form the
soap emulsifier.
Solids present in oil and synthetic muds must be kept wet with the non-
aqueous phase to prevent coagulation and settling and mud instability. Oil-wetting
agents are normally incorporated in the basic mud package. These materials
are typically amines or quaternary ammonium salts having hydrocarbon chains
of 10 or more carbon atoms. They also render clays or lignites oil-wet for use in
viscosity and filtration control (112).
272 DRILLING FLUIDS Vol. 1

Defoamers are frequently needed for salty muds, although in very small
quantities. A common defoamer is aluminum stearate [637-12-7]. Others include
tributyl phosphate [55612-35-6], alkyl aryl sulfonates, silicones, and alcohols
such as octanol [111-87-5].
Foaming agents maintain stable drilling foams in areas where minimal bot-
tomhole pressures are required. A large number of chemicals generate drilling
foams, including anionics, such as alkyl and alkylaryl sulfates and alkylarylsul-
fonates; nonionics, such as ethoxylated fatty alcohols and alkylaryl alcohols; and
cationics, such as imidazolines and tertiary amines. The foaming agent must be
chosen to handle a variety of possible contaminants (salt, crude oils, solids) and
downhole temperatures (113,114).
4.7. Lost Circulation Control. To function properly, a drilling fluid
must be circulated through the well and back to the surface. Occasionally, highly
permeable or cavernous formations and fractured zones, both natural and
induced by the mud pressure, are encountered and circulation is partially or
completely lost. Loss of drilling fluid, owing to openings in the formation, can
result in loss of hydrostatic pressure at the bottom of the hole and allow influx
of formation fluids and possibly loss of well control. It is essential that circulation
be regained for drilling to continue. A wide variety of materials can be added to
the drilling fluid to seal off the lost circulation zones (115,116). The particle sizes
of these materials are typically much larger than the particle sizes of solids nor-
mally suspended in the mud which are generally <150 mm in diameter. Some of
the same materials may be used for controlling filtration rates or for stabilizing
shale formations but in such cases would be much smaller in particle size than a
typical lost circulation material (LCM).
Lost circulation materials are flake, fiber, or granular-shaped particles.
Each type is sold individually, often in two or more size grades, or two or more
materials of different shapes may be sold as a blend (117–120). Materials of dif-
ferent shapes and sizes are often blended into the mud at the well site. Some
common flake-shaped LCMs consist of shredded cellophane and paper, mica,
rice hulls, cottonseed hulls, or laminated plastic. These materials lie flat across
the opening to be sealed or are wedged into an opening such as a fracture. Some
are sufficiently strong to withstand considerable differential pressure, whereas
others are weak and the seal may be broken easily. Weaker flake materials typi-
cally are used near the surface or in combination with fibrous or granular addi-
tives. Fibrous additives include a variety of cellulose fibers, sawdust, sugar-cane
bagasse, paper, straw, leather, and many others of similar size, shape, and avail-
ability. The larger fibers function by forming a brush-heap-type mat over the
opening. The seal so formed may require smaller fibrous particles to stop seepage
of mud through the mat. Fibers generally have little strength and cannot with-
stand high differential pressures. The brush-heap seal may extend far enough
into the wellbore to be dislodged by the drill string, or be rigid enough to interfere
with drill string movement.
Granular LCMs generally are much stronger than the other types and
include ground rubber, nylon, plastics, limestone, gilsonite, asphalt, and ground
nut shells, eg, walnut and pecan. Fine-, medium-, and coarse-size grades are
available. Granular-shaped particles enter the opening, bridge it, and form
a tight seal against further mud losses. Particle size and distribution are
Vol. 1 DRILLING FLUIDS 273

important for this mechanism to be effective. Bridging particles must generally


have a diameter one-half the opening width for a fracture or one-third the dia-
meter of a circular opening (121). An effective seal requires proper gradation of
particle sizes (122). The advantage of this mechanism is that the seal is formed
outside the wellbore and is not subject to drill string action.
4.8. Removal of Solids. Solids incorporated in the mud during drilling
generally are separated mechanically, reduced by dilution, or removed chemi-
cally by flocculation. It is desirable to maintain a low concentration of drill solids
(4–8 vol%) and in some cases total removal is required. In the latter case, the
drilling fluid is clear, consisting of freshwater or brine, and high drilling rates
can be achieved. Polymeric flocculants added in small (0.03–0.06 kg/m3 (0.01–
0.02 lb/bbl)) quantities maintain a clear liquid (123).
Polymers used for flocculating drill solids generally are high molecular
weight polyacrylamides of varying degrees of hydrolysis (124). The polymer
may be cationic, nonionic, or anionic, depending on the chemistry of the drilling
fluid and the nature of the solids. At higher concentrations, some polymers act as
protective colloids and stabilize and enhance the viscosity of bentonite suspen-
sions and protect water-sensitive shales.
4.9. Lubricants and Spotting Fluids. The frictional resistance gener-
ated by the rotating drill string against the formation or casing may require
extra torque if the hole is crooked or being drilled directionally. Considerable
frictional resistance to raising and lowering the drill string may also occur;
this is referred to as drag. Under certain conditions, such as in highly deviated
holes, holes with frequent changes in direction, undergauge holes or poor drill
string dynamics, or increased torque and drag entail a high risk of lost rig
time, expensive pipe recovery operations, and limitation in well development
(125). Excessive torque required to rotate the pipe generates greater strain on
the drill pipe with the danger of twist-offs. Another problem occurs when the
drill pipe sticks to the wall of the borehole. This may be caused by running or
pulling the pipe into an undergauge section of the hole, a key seat, or a bridge
of cavings. Another form of sticking, known as differential pressure sticking,
occurs when the drill pipe becomes motionless against a permeable formation
and a portion of the contact area of the pipe is isolated by the filter cake. The
differential pressure (mud pressure-formation pressure) acts on the isolated
area to press the pipe against the wall.
To overcome these difficulties, drilling fluids are treated with a variety of
mud lubricants available from various suppliers. They are mostly general pur-
pose, low toxicity, nonfluorescent types that are blends of several anionic or non-
ionic surfactants and products such as glycols and glycerols, fatty acid esters,
synthetic hydrocarbons, and vegetable oil derivatives. Extreme pressure lubri-
cants containing sulfurized or sulfonated derivatives of natural fatty acid
products or petroleum-base hydrocarbons can be quite toxic to marine life and
are rarely used for environmental reasons.
Water-dispersible and sulfonated asphalt derivatives have been used as an
aid to improve mud lubricity, although these are primarily for filtration control
and shale protection. Solid lubricants, including graphite, plastic beads (126),
glass beads (127), and ground walnut shells, may be incorporated to impart lubri-
city and reduce the risk of pipe sticking. Gilsonite and air-blown asphalts
274 DRILLING FLUIDS Vol. 1

blended with a glycol-carrier are used to prevent pipe sticking and to act as lubri-
cants for water-base muds (128,129).
If the drill string becomes differentially stuck, mechanical methods or spot-
ting fluids can be applied, or the hydrostatic pressure can be reduced (130). In
general, penetration of water- or oil-based spotting fluids into the interface
between the filter cake and the pipe accompanied by dehydration and cracking
results in reduction of differential pressure across the drill string (130,131).
Spotting fluids are usually positioned in the open hole to completely cover the
problem area.
Recently, a number of water-base spotting fluids have been developed
(132) and consist of a variety of surfactants, glycols (133,134), organic acids
(135), glyceride esters (136), and vegetable oil derivatives. The newer water-
base spotting fluids have the distinct advantage that they can be incorporated
into the drilling fluid following use, without having an adverse effect on the
toxicity of the system (137).
Drill string and casing corrosion can present serious problems in some dril-
ling operations. Corrosion is generally caused by oxygen dissolved or entrained
in the mud as it is circulated through the well (138). Acid gases, such as carbon
dioxide and hydrogen sulfide, contribute to corrosion, particularly hydrogen sul-
fide which can produce catastrophic drill string failures through stress cracking.
Preventing entry of these corrosive gases into the well is the most effective
method of control, but is not always feasible. Therefore additives are employed
to counteract corrosive attack (139,140).
Oxygen corrosion is reduced significantly at pH >11 and is generally
lower at any alkaline pH than in an acidic environment. Maintaining a high
pH is the most common means of corrosion control. Other methods include oxy-
gen removal by scavengers such as sodium sulfite [7757-83-7] or ammonium
bisulfite [17026-44-7] (140,141), or protecting the pipe from attack by coating
with amines.
A high pH also affords protection against hydrogen sulfide and carbon diox-
ide. Zinc and iron compounds are used as sulfide scavengers (142). Zinc com-
pounds include the carbonate, oxide, hydroxide, sulfate, and water-soluble
organic chelates. High surface area iron oxides, particularly a synthetic magne-
tite, also are used (143). Scale inhibitors such as organic phosphonates and low
molecular weight polyacrylates may be added to prevent buildup of carbonate
scale on the pipe. Drilling rates can be severely reduced when drilling formations
that tend to stick to the surface of the drill bit, a situation known as bit balling. If
the bit cannot be kept clean and free of sticky material, drilling performance suf-
fers and the operational cost can increase substantially. The new generation syn-
thetic polycrystalline diamond compact (PDC) bits are capable of achieving high
penetration rates if the cutters are kept clean and the proper hydraulics and bit
rotary speeds can be maintained. Use of an oil- or a synthetic-base mud is nor-
mally preferred when drilling with a PDC bit as there is much less tendency for
bit balling to occur and high drilling rates can be realized (144). Water-base
muds are more prone to bit balling and PDC bit performance can be significantly
poorer than when a nonaqueous continuous-phase mud is used (145). Additives
being used to enhance the drilling performance of PDC bits in water-base muds
include water-soluble glycols (146,147) and various terpenes (144).
Vol. 1 DRILLING FLUIDS 275

5. Completion and Workover Fluids

Completion and workover fluids that are used during the completion phase of the
well or when performing a workover during the life of the well. Completion fluids
should (1) cause minimal formation damage, especially to reservoir permeability,
(2) maintain long-term stability at downhole temperatures, (3) be resistant to
contaminants, and (4) protect down-hole equipment against corrosion. These
fluid types can divided into two broad classifications: solids-free and solids-
laden. The solids-free completion/workover fluids are generally brines ranging
in composition from potassium chloride to zinc bromide, and their densities fall
into the range of 1.02 kg/L (8.5 lb/gal or ppg) to 2.46 kg/L (20.5 ppg). Weighted
solids-free fluids are obtained by the use of solutions of various water-soluble or
acid-soluble salts. In particular, the chlorides and bromides of sodium, potas-
sium, calcium, and zinc, and their mixtures are used to adjust the densities of
these brines. The solids-laden completion and workover fluids are generally com-
posed of a brine as a base fluid and solids, in the form of water or acid-soluble
particles, for density and fluid-loss control as well as polymers for suspension-
carrying capacity and filtrate control. In general, the solids-laden fluids cover
the same density range as the solids-free fluids.
Solids-free fluids are the predominant completion/workover fluids in use.
Use of these fluids isdictated in many completion operations, especially in gravel
pack or frac pack type completions. Use of a solids-laden fluid in such operations
can lead to failure. Solids-free fluids cause a minimum amount of formation
damage and allow tools to be run into and pulled out of the well with ease. Solids
can penetrate into the permeable media of the formation and become lodged,
thus causing reduced permeability and ultimately lower production. Solids-
laden fluids also have typically increased viscosities, which can inhibit tools
being run into and pulled out from the well. In order to keep solids-free fluids
free of solids, filtration, either cartridge or diatomaceous earth, is generally
used while these fluids are being employed.
There are a number of factors that determine the choice of fluid to be used.
If there is no milling or drilling operations to be carried out, then a solids-free
fluid can be used. The particular solids-free fluid/soluble salt to be employed
also depends on a number of factors that include:

(1) Density required: A compilation of some of the soluble salts and their max-
imum brine densities that have been used as completion and workover
fluids is given in Table 8. A specific density may be achieved by mixing to-
gether one or more brines listed in the table. The monovalent brines are
generally mixed with monovalent brines and divalent brines mixed with
divalent brines. Mixing of monovalent and divalent brines can cause the
formation of precipitates and turbidity. Density adjustment can also be
achieved by adding a salt to a specific brine.
(2) True crystallization temperature (TCT) required: It is important to know
expected ambient temperatures and expected sea floor temperatures. The
crystallization temperature can be adjusted by blending brines or water
to raise or lower the crystallization temperature.
276 DRILLING FLUIDS Vol. 1

Table 8. Soluble Salts that Increase the Density of Drilling Mud and Workover
and Completion Fluids
CAS Registry Specific gravity Maximum brine
Salt number Formula of solid density, g/L (ppg)
potassium chloride [7447-40-7] KCl 1.984 1.14 (9.5)
sodium chloride [7647-14-5] NaCl 2.165 1.20 (10.0)
calcium chloride [10043-52-4] CaCl2 2.150 1.39 (11.6)
sodium bromide [7647-15-6] NaBr 3.203 1.50 (12.5)
calcium bromide [7789-41-5] CaBr2 3.353 1.70 (14.2)
zinc bromide [7699-45-8] ZnBr2 4.201 2.46 (20.5)
potassium formate [590-29-4] KCHO2 3.48 1.60 (13.3)
sodium formate [141-53-7] NaCHO2 1.92 1.34 (11.2)
cesium formate [3495-36-1] CsCHO2  H2O 3.372 2.37 (19.7)

(3) Compatibility with the formation: Formation gases may not be compatible
with the completion/workover fluid. For example, calcium-based brines
should not be used in the presence of carbon dioxide as the formation of in-
soluble calcium carbonate can precipitate. Other problems can be that the
formation water may be incompatible with brine, or emulsions may form
with the crude oil.
(4) Cost: Generally speaking, the higher the density of the brine, the higher
its cost. Another rule of thumb is the lower the crystallization temperature
required, the higher will be its cost. Other important factors to weigh are
brine corrosivity (eg, metallurgy of the casing and/or tubing in the well),
brine stability, brine toxicity and brine availability (to remote and generally
international locations).

Brine sources may include seawater, natural brines and manufactured salt
brines. Densities up to 2.3 kg/L (19.2 ppg) can be obtained using mixtures of cal-
cium bromide and zinc bromide. Calcium bromide is commercially available as a
52 wt% solution in water, having a density of 1.704 kg/L (14.2 ppg). This brine
remains clear down to a temperature of 78C. Also available is a calcium bro-
mide/zinc bromide brine having a density of 2.304 kg/L (19.2 ppg), where the
brine contains approximately 52.8 wt% zinc bromide and 22.8 wt% calcium bro-
mide. Evey manufacturer of clear brines provides detailed blending tables show-
ing the relative amount of various brines, salts, and water in order to prepare a
brine with the desired specific gravity and crystallization point. Two examples
are given below. Table 9 provides the volume of water [fraction of a barrel
(bbl)] and weights of 95% calcium bromide solid that are needed to prepare 1 bar-
rel (bbl) of a calcium bromide brine having a specific brine density (ppg). The
crystallization point of the brine is also included in the table. Table 10 presents
the volumes of calcium bromide (1.7 kg/L /14.2 ppg) brine and zinc bromide/
calcium bromide (2.3 kg/L /19.2 ppg) that are needed for preparing calcium
bromide/zinc bromide brines of varying densities, together with their correspond-
ing crystallization points.
Slightly higher densities than the above-mentioned brines can be attained
using nearly saturated solutions of ZnBr2 having a density of 2.46 kg/L (20.5 ppg)
Vol. 1 DRILLING FLUIDS 277

Table 9. Calcium Bromide Solution Requirements Using Calcium Bromide (95%)a


Brine density, 708F
Crystallization
ppg g/L Water, bbl 95% CaBr2, lb point, 8F

11.6 1392 0.860 186.3 32


11.8 1416 0.852 197.5 40
12.0 1440 0.844 208.7 50
12.2 1464 0.836 219.9 61
12.4 1488 0.828 231.1 73
12.6 1512 0.820 242.3 88
12.8 1536 0.811 253.7 96
13.0 1560 0.803 265.1 78
13.2 1584 0.794 276.5 63
13.4 1608 0.786 287.9 43
13.6 1632 0.777 299.4 34
13.8 1656 0.768 311.0 20
14.0 1680 0.758 322.6 7
14.2 1704 0.751 333.4 10
14.4 1728 0.739 346.0 23
a
To prepare 1 barrel (bbl) or 42 gal.

or of cesium formate having a density of 2.367 kg/L (19.7 ppg) (148). The mono-
valent formate brines, which are highly soluble in water, were first introduced
into the oil industry in the early 1990s. The formates are organic and biodegrad-
able. Their chemistry stabilizes and protects the conventional polymers used in
viscosity modifiers and other additives, allowing such additives to be used at
much higher temperatures—often >3508F (1768C). Other relevant information
on formate brines is included in a series of papers that are cited in Refs. 149–154.

Table 10. Calcium Bromide/Zinc Bromide Solution Requirements Using 14.2 ppg
Calcium Bromide and 19.2 ppg Zinc Bromide/Calcium Bromidea

Brine density, 708F


19.2 ppg CaBr2/ Crystallization
ppg g/L 14.2 ppg CaBr2, bbl ZnBr2, bbl point, 8F
15.0 1800 0.840 0.160 22
15.3 1836 0.780 0.220 29
15.6 1872 0.720 0.280 35
15.9 1908 0.660 0.340 37
16.2 1944 0.600 0.400 26
16.5 1980 0.540 0.460 16
16.8 2016 0.480 0.520 6
17.1 2052 0.420 0.580 2
17.4 2088 0.360 0.640 4
17.7 2124 0.300 0.700 6
18.0 2160 0.240 0.760 9
18.3 2196 0.180 0.820 13
18.6 2232 0.120 0.880 19
19.0 2280 0.060 0.940 20
19.3 2316 0.000 1.000 16
a
To prepare 1 barrel (bbl) or 42 gal.
278 DRILLING FLUIDS Vol. 1

Drilling fluids made both with CaBr2/ZnBr2 brines and with formate-based
fluids generally need no solid weighting agents, since the density is an inherent
property of the fluid itself.
One inherent problem with solids-free brines is that since they have no
solids and in general have low viscosities, there is nothing to prevent the liquid
from leaking off into the formation. Therefore, an important part of planning in
the use of the solids-free completion and workover fluids is fluid loss control pills.
These pills are generally a blend of the brine being used and polymers for impart-
ing viscosity. The viscosifiers used in brine fluids should be stable at a tempera-
ture of 1508C and have good acid solubility. The polymers used are
hydroxyethylcellulose, xanthan gum and other specialty polymers. Guar gums
can be included in completion/workover fluids, but their use is more limited
because they are not completely soluble in acid; in addition, guar gum is also
only stable to 668C. For filtrate loss reduction, a blend of polymers and soluble
solids are usually incorporated into the fluid. These polymers are used to plug
the interparticle spaces within the bridging solids, thus reducing the amount
of filtrate entering the formation. The selection of the solids to be used will be
based on the pore throat openings in the formation in order that a bridge can
be rapidly formed on the outside of the formation, preventing solids and liquids
from entering the reservoir.
Some additional details about the additives used in workover/completion
fluids are listed in Table 11. Some solids-laden completion and workover fluids
are now being utilized as reservoir drill-in fluids. These are used once the well

Table 11. Workover/Completion Fluid Additives


Additive Function Region used Comments
commodity salts
calcium bromide, calcium for density increase United States calcium chloride is a
and sodium chloride, of brine fluids major product;
zinc bromide, sodium blends of salts are
bromide, monovalent generally used in
formates workover fluids
viscosifiers
hydroxyethylcellulose build fluid viscosity United States
(HEC), xanthan gum without use of high
polyacrylamides solids system;
additive must
be water or acid
soluble
fluid-loss additives
(filtrate control agents) for reducing the loss of
calcium carbonate, water or fluid by
starches, preventing its
penetration into the
reservoir
formation
other additive types
corrosion inhibitors,
biocides
Vol. 1 DRILLING FLUIDS 279

has reached the depth at which hydrocarbons are expected to be encountered.


These fluids are designed especially to have the properties of a good drilling
fluid, yet deposit a filter cake that is easily removed ie all the components are
soluble in either water or acid, while the filtrate does not cause any damage to
the reservoir.

6. Economic Aspects

Considerable changes have occurred in the makeup of drilling fluids during the
the last two decades and these changes are expected to continue during the
tweny first century. The price of the drilling fluid additives used in these fluids
varies according to company and location so that it is almost impossible to indi-
cate typical prices for the more common additives (used in drilling operations),
which are listed in Table 12. Price information on the full line of products nor-
mally is available from drilling fluids companies. The prices for those materials
listed in Table 12 would also depend on the nature of the contract between the
supplier and operating companies.
It is instructive to consider the price fluctuations that have occurred for bar-
ite, a densifier that constitutes the largest of the chemical expenses incurred in

Table 12. Summary of Drilling Fluid Additives in Use


and Their Function
Function Additive
increase density barite
hematite
increase viscosity bentonite
attapulgite
xanthan gum
hydroxethylcellulose
reduce viscosity sodium acid pyrophosphate
lignite
chrome lignosulfonate
chrome-free lignosulfonate
sodium polyacrylate (liquid)
reduce filtration rate corn starch
modified starch
carboxymethyl cellulose
sodium polyacrylate
modified lignite
shale stability/viscosity polyacrylamide
alkalinity control sodium hydroxide
lime
potassium hydroxide
lost circulation control walnut shells
mica
cellulose fibers
miscellaneous inorganic sodium chloride
compounds potassium chloride
calcium chloride
sodium carbonate
280 DRILLING FLUIDS Vol. 1

drilling muds. Market competition has been significant in the barite segment, so
that its prices have been historically under pressure. The oil price crash and dril-
ling recession in 1998 resulted in the price of barites in Yarmouth (U.K.) falling
from an average of $100–120 per metric ton in 1997 to $75–90 per metric ton in
mid-1998. The price decrease continued in 1999 but started to increase in mid-
2000 to  $80–100 per metric ton at the end of 2000.

7. Analytical and Test Methods

Procedures for determining drilling fluid properties are available (19,20,155).


Tests and test methods are constantly reviewed by API committees to ensure
acceptable accuracy when performed under field conditions as well as in the
laboratory. API publications are republished as new tests are added or existing
tests are modified.

8. Environmental Aspects

At present, the major challenge that must be met in formulating drilling fluids is
need to satisfy the increasingly demanding conditions of high temperature and
pressure that are found in some deep wells and horizontal wells, while avoiding
harm to the environment. Drilling fluid components should be selected in a way
such that the discharge of mud or cuttings will have a minimum possible envir-
onmental impact. In addition to environmental concerns being a major driving
force behind current drilling fluid R&D, the health of rig workers has become
an important concern in the development and use of new products.
While drilling fluids are essential for the success of an well drill, they can
also be one of messier parts of a drilling operation. Cuttings brought up from out
of a borehole have to be disposed of as well as any drilling fluid attached to them.
The degree of impact that drilling fluids can have on the environment depends on
the type of mud used and the existing environmental conditions. Offshore, water-
based muds are generally the least damaging when compared with oil-based
muds. The disposal of waste drilling fluids and drill cuttings in the United States
has long been regulated either by local authorities, the individual states, or by
the federal government. These regulations continue to change. In many
instances, it is environmental issues that have been the main driving force for
product choice in the drilling fluids business. It is no secret that the mud-
contaminated cuttings constitute a major portion of the waste that is generated
during drilling.
The offshore disposal of both diesel and mineral oil drilling fluids and asso-
ciated cuttings has always been prohibited in U.S. waters. However, offshore dis-
charge of mineral oil mud cuttings overboard has been permitted in the North
Sea and elsewhere as long as the oil content of the cuttings was below some reg-
ulatory limit. The regulatory oil-on-cuttings limit in some sectors of the North
Sea has been lowered significantly over the last decade. There is a definite
move toward alternative fluid systems many of which are used in U.S. offshore
areas, eg, synthetic-based drilling muds. Thus, in the 1990s, due to environmental
Vol. 1 DRILLING FLUIDS 281

considerations, oil-based muds were partially replaced by synthetic-based


muds whose synthetic-based fluids were deemed biodegradable. However, recent
biodegradability tests on synthetic-based muds under real sea-bottom conditions
have shown that in the absence of oxygen, biodegradability is poor. As a result,
beginning in 2001, all but one type of synthetic-based muds, ie, the ester-based
mud which indergoes rapid degradation, have been effectively prohibited from
being used in the North Sea. At the end of 2000, it was reported that rig opera-
tors in the U.K. sector have reached zero discharge with respect to synthetic-
based drilling fluid cuttings. A look at synthetic-based mud usage in the North
Sea shows that from 2000 to 2001, the usage was reduced from 10–15% of all
muds used to somewhere 5%.
As of 2001, legislation has been enacted to prohibit the discharge of envir-
onmentally unfavorable cuttings into the North Sea. When oil-based muds are
employed, there also exist strict limitations on how much oil can be discharged.
In the U.K. sector of the North Sea, no discharge of oil-based muds is allowed.
For water-based muds usage offshore, there are controls on the kinds of che-
micals that can be used in the muds. The screening process for the water-based
mud additives may include the toxicity and persistency of the chemical, its poten-
tial to bioaccumulate in marine organisms, and the toxicity of any degradation
process. It is therefore not surprising that the search for environmentally
friendly drilling mud additives and biodegradable alternatives to the conven-
tional mineral oil used continues.
The most significant change in the regulations on discharges of drilling
fluids and cuttings in the Gulf of Mexico occurred in July 1986 when a new
toxicity-based limit was placed on disposal (156). The specific toxicity limit is
based on a 96-h bioassay test in which the drilling fluid is diluted with nine
parts seawater and mixed well, the suspension is allowed to settle for one
hour, and the suspended particulate phase (SPP) is decanted for testing.
Mysid (mysidopsis bahia) shrimp are exposed to a range of SPP concentrations
and the lethal concentration to 50% of the shrimp (LC50) is calculated from the
observed shrimp mortality at the end of 96 h. (LC50 is the lethal concentration,
where 50% of the organisms are killed.) The toxicity limit must be 30,000 ppm or
higher in order to meet EPA standards and allow the overboard discharge of fluid
or drill cuttings. Very few instances of water-based muds failing the mysid bioas-
say test existed during the 1990s.
In summary, there exist a variety of regulations that apply in different off-
shore drilling areas in the United States and around the world. All have had a
profound effect on drilling fluid technology (157,158). Operators and service com-
panies have eliminated use of the more toxic additives, reformulated old mud
systems, while developing new ones to ensure acceptable environmental perfor-
mance based on pertinent regulations.

9. New Directions in Drilling Fluids

Some recent developments and innovations in drilling fluids as proposed in


recent patents are presented in the following section. These describe newly
developed formulations for use in the various stages of oil and gas well drilling.
282 DRILLING FLUIDS Vol. 1

9.1. Water-Based Drilling Fluids. The most significant developments


in mud materials made within the last decade have been achieved with the
aim of improving the performance of the water-based muds, due to increasing
limitations that have been placed on the use of oil-based muds. The main
drawback of the water-based muds is their poor performance in shale stabiliza-
tion, ie, the water fraction can cause swelling of the clay. Performance can be
ameliorated by using inhibiting muds, ie, the addition of salt and special addi-
tives. One of the first additives used for this purpose was partially hydrolyzed
polyacrylamide (PHPA), which was later followed by glycols as shale stabilizers.
A recent development has been the introduction of soluble silicate, sodium
or potassium, that seal up the walls of the hole. While soluble silicates were pro-
posed in the 1930s, they were not successful because they could not be integrated
into the mud system. The new mud systems built up around the silicates have
been successfully used in both the Gulf of Mexico and the North Sea. An example
of such a silicate-based mud was given in a 1997 patent (159), which was
proposed for use in drilling through reactive clays and shales and other soft for-
mations. The drilling fluid contains, in addition to polymeric agents for fluid loss
and rheological control, a stabilizer consisting of an alkali metal silicate (28–35%
by weight potaassium silicate). This drilling fluid does not contain bentonite or
barite.
Another patent (160) describes a brine for use in drilling operations that
has water as the continuous phase together with at least one inorganic salt
and an amount of a low molecular weight water-soluble compound that is effec-
tive in reducing the crystallization point of the brine. The low molecular weight
water-soluble organic compound may be an alcohol, polyol, glycol, polyglycol,
polyalkylene oxide, alkylene glycol ether, carbohydrate, amino acid, etc.
Preferred compounds mentioned are ethylene glycol and tripropylene glycol
bottoms. Data from the patent for two solutions are presented in Table 13 for
20% NaBr/water and 20% NaBr/20% ethylene glycol/water mixtures.
A Great Lakes Chemical patent (161) proposes a composition for either dril-
ling and/or well bore treating operations, using a carboxylate-based brine solu-
tion to inhibit corrosion in fluid-handling equipment. In one embodiment
presented, the carboxylate-based brine consists of a carboxylate salt dissolved
in water for use in the preparation of halide-free drilling fluids. The drilling
fluid can contain a variety of additives such as viscosifiers, weight-up agents,
buffering agents and/or fluid loss control agents. In a second alternative embodi-
ment, a mixed salt brine consisting of a carboxylate salt and a halide salt is used
in drilling and/or well bore treating operations. The carboxylate salts mentioned
included sodium formate, sodium acetate and sodium propionate.

Table 13. Crystallization Point Dataa


20% NaBr/20% Ethylene
20% NaBr/Water,8F glycol/water,8F
FCTA 10 11
LCTD 18 8
a
Ref. 160.
Vol. 1 DRILLING FLUIDS 283

A drilling, drill-in, or completion fluid described in a 1998 patent (162) had


water as its continuous phase along with a nontoxic low molecular weight water-
soluble organic compound, which serves as a gas hydrate inhibitor. Preferred
organic compounds included ethylene glycol alone, or a blend of an ethylene gly-
col derivative and a propylene glycol derivative in an amount effective for inhi-
biting hydration of the shale by the water base.
Another patent invention (163) described clay-free well drilling, biopolymer-
free well drilling, and servicing fluids that were comprised of an aqueous
divalent-cation containing water-soluble salt, a bridging agent, and a pregelati-
nized cross-linked amylopectin starch suspending agent and a fluid-loss control
additive. The water soluble divalent-cation salt is selected from the group con-
sisting of alkaline earth metal halide salts, zinc halide salts and mixtures of
these. The low shear rate viscosity was increased without increasing the high
shear rate viscosity above 70 cP by adding magnesium oxide and/or dipotas-
sium hydrogen phosphate.
A 1998 patent (164) described a potassium formate solution, used as an aque-
ous well servicing fluid, which also contained potassium citrate. The potassium
citrate enhanced the density of the potassium formate solution and also mini-
mized any risk of caking if potassium formate is used alone. The amount of potas-
sium citrate used is in the range of 5–30% by weight, based on the total weight of
potassium formate and potassium citrate in the liquid.
A different route used for increasing the density of fresh or depleted clear
brine drilling fluids that are based on calcium bromide and calcium chloride, is
achieved by the addition of solid hydrates of calcium bromide, in particular cal-
cium bromide tri- and tetrahydrates. These hydrates, which are scarcely men-
tioned in the literature, are highly effective in fixing the brine concentration.
The density of the solid used is about 2200 kg/m3 and consists of 70–78% calcium
bromide. This novel product, known as Calcium Bromide Forte, is currently
being marketed by Dead Sea Bromine Group (DSBG).
A U.S. patent (165) described a water-based drilling fluid which is composed
of 1–8% of an additive comprising a saturated terpene, a mixture of saturated
terpenes, or a mixture of saturated and unsaturated terpenes. The drilling
fluid is considered to be substantially nontoxic and was claimed to have an
LC50 >30,000 ppm and preferably >100,000 ppm. The most economic additives
for this fluid were a mixture of saturated and unsaturated terpenes, having at
least 30% saturated terpenes. The drilling fluid also contains a weighting mate-
rial, a fluid loss additive, and a viscosifier. Examples given of saturated terpenes
were p-menthane, m-menthane, and pinanne.
A Perstorp AB patent (166) described the use of a saturated aqueous salt
solution containing at least one alkali formate salt and at least one alkali bro-
mide salt in a drilling fluid. The fluid was composed of 40–50% by weight
water, 50–60% by weight of the salts mentioned in a weight ratio of formate to
bromide of 30:70 to 60:40. The saturated aqueous salt solution had a density and
a total salts content exceeding that of a saturated aqueous solution of the bro-
mide salt. Example 1 in the patent cited a 26% sodium formate 30.4% sodium
bromide aqueous mixture having a density of 1.561 g/cm3. The individual satu-
rated solutions at 228C had the following densities: sodium formate 1.318 g/cm3
and sodium bromide 1.520 g/cm3.
284 DRILLING FLUIDS Vol. 1

In the drilling of oil wells with water-based fluids, instability problems


arising in shales have to be frequently dealt with. These problems arise
for two reasons: (1) the reactivity of the claystone to water (2) the transmission
of pressure from the mud column to the inside of the formation, thus modifying
the stress within the rock, causing it to break. Another patent (167) presents
a composition for a water-based drilling fluid that is an oil-in-water emulsion.
The newly proposed drilling fluid has been found particularly effective
in preventing the pressure transmission. In the patent’s preferred embodi-
ment, a nonionic emulsifier was used, preferably polyoxy ethylene–sorbitan
monooleate.
A nontoxic drilling fluid composition has been patented (168) and claimed to
be of use as a lubricant or spotting fluid in drilling oil and gas wells. The pro-
posed drilling fluid consists of a water-based continuous phase and a silicone
fluid together with an emulsifier. The silicone fluid is comprised of dimethylsilox-
ane polymers and was claimed to be nontoxic to marine life.
9.2. Organic-Based Drilling Fluids. An invention described in a
patent (169) is an improved glycol-based drilling fluid that is comprised of:
(1) A predominantly glycol–continuous phase (miscible glycol at a concentration
of at best 70% by volume of the continuous phase. (2) An antisticking drilling
fluid additive present at a concentration of at least 0.5% by weight of the drilling
fluid. The antisticking drilling fluid additive is selected from silicone fluids,
silicone-based surfactants, hydrocarbons, glycol-ethers, carboxylic acid esters,
and combinations of these.
An Exxon patent (170) presented an environmentally friendly drilling fluid
or mud composition claimed to be useful in oil and gas production. The biodegrad-
able drilling fluid was comprised of one or more weighting agents, emulsifiers,
wetting agents, viscosifiers, fluid loss control agents, and other particulates
such as used in a gravel pack, emulsified with a paraffin solvent composition
that formed a continuous oil phase, or water-in-oil invert phase. The solvent com-
position (continuous oil phase) was comprised of a mixture of C10 –C24 n-paraffins
and isoparaffins having an isoparaffin: n-paraffin molar ratio ranging from
0.5:1 to 9:1, and the isopaffins of the mixture contain >50% of mono-methyl
species, based on the total weight of the isoparaffins in the mixture.
A 2003 patent (171) describes nonaqueous suspensions suitable for formu-
lation of environmentally friendly drilling fluids that consists of a nonaqueous
base fluid with a concentration of suspension aids, a dispersant, and optionally,
a polymeric viscosifying agent. The base fluid was selected from glycols, polygly-
cols, polyoxyalkylenes, glycol ethers, glycol esters; the suspension aid was
selected from hydroxypropyl cellulose, ethyl cellulose; the dispersant was selected
from calcium stearate, lecithins, stearic acid, oleic acid; and the viscosifiers were
selected from xanthan gum, hydroxyethyl cellulose, CM-cellulose, guar gum,
polyacrylamide and starch. The compositions could be used as additives for
well drilling and treatment fluids.
9.3. Viscosifiers. A 2000 patent (172) proposed the use of novel polymer
compositions as viscosifiers, which can be used in well service fluids (completion
fluids, work-over fluids or drilling fluids). The drilling fluid, for use at high tem-
peratures, is composed of water-soluble copolymers having sulfonate and
carboxylate groups together with alkali metal salts of carboxylic acids. The alkali
Vol. 1 DRILLING FLUIDS 285

metal carboxylate was selected from sodium, potassium, and cesium salts of
acetic and formic acids.
A 1997 patent (173) described a composition that is useful for viscosifying
aqueous drilling fluids that are utilized in subterranean drilling activities such
as drilling, drill-in, completion, hydraulic fracturing, work-over, packer and well
treating fluids. The composition is composed of a prehydrating alcohol having at
least two hydroxyl groups (with a molecular weight of 60 to 600), an inorganic
salt, and a polymer. The prehydrating alcohol reacts physically with the polymer
to cause a partial swelling effect, solvating effect or hydrating effect upon the
polymer.
A Great Lakes Chemical patent (174) presented aqueous mixed salt
systems that were viscosified with water-soluble or water-dispersable polymers.
These systems were comprised of water, a water-soluble or water-dispersable
polymers capable of viscosifying an aqueous media, and mixed salts comprised
of at least two cations and two anions. The cations cited were Li, Na, K, Cs,
Mg, Ca, Zn, or mixtures. The anions cited were Cl, Br, I, formate, nitrate, acet-
ate, cyanate, thiocyanate, or a Zn-complex anion. These compositions were pro-
posed as replacements for conventional drilling, drill-in, completion, hydraulic
fracturing, work-over, packer, well-treating, testing, spacer, or hole abandon-
ment fluids.
A patent (175) disclosed a low solid, high density fluid that was suggested as
being useful in drilling, workover and completion fluids in oil and gas well opera-
tions. The fluids comprised (1) one or more brines having a density of 1500 kg/m3
to 2400 kg/m3 (salt chosen from calcium bromide, zinc bromide or mixtures of
these), (2) a viscosifier that is a high surface area silica, preferably a fumed silica,
(3) a fluid loss control additive that was preferably selected from the group con-
sisting of starch derivatives, cellulose derivatives, lignocellulose derivatives or
mixtures of these; (4) one or more water soluble or acid soluble bridging agents,
and (5) optionally, an alkaline buffer and a polar additive.
A 1998 patent (176) described a drilling fluid comprised of brine and addi-
tives for maintaining rheological properties in a neutral or slightly acidic brine
and a method for using the drilling fluid to drill into a producing formation. The
drilling fluid was comprised of a viscosifying polymer and a fluid loss polymer
that function in the brine without substantially adversely affecting their proper-
ties. The brine was virtually free of insoluble solids.
A recent review (177) was published on the use of complex carbohydrates,
ie, polysaccharides as additives in the petroleum and natural gas industries,
especially as viscosifiers and fluid loss control additives for drilling fluids and
muds. Viscosifiers are typically xanthan gum, hydroxyethyl cellulose and modi-
fied guar gum, while fluid loss control additives are typically pregelatinized
starch, and polyanionic cellulose derivatives, especially CM-cellulose.
9.4. Densifiers. The invention (178) presented in another patent relates
to an additive which increases the density of the well bore fluids used during the
construction or repair of oil, gas, injection, water or thermal wells. The additive
is comprised of solid, colloidal particles of weight average particle diameter (D50)
of < 2 m. The particles have been deflocculated by the action of a dispersant, pre-
ferably incorporated during the grinding or comminution process. The colloidal
particles used include barium sulfate, calcium carbonate,dolomite, illmenite, or
286 DRILLING FLUIDS Vol. 1

hematite. The liquid medium can be an aqueous phase or an organic liquid of


kinematic viscosity <40 cS at 408C and of flash point >608C.

BIBLIOGRAPHY

‘‘Drilling Fluids’’ in ECT 1st ed., Suppl. 1, pp. 246–258, by G. R. Gray and E. E. Huebotter,
Baroid Div., National Lead Co.; in ECT 2nd ed., Vol. 7, pp. 287–307, by E. E. Huebotter
and G. R. Gray, Baroid Div., National Lead Co.; ‘‘Drilling Fluids’’ under ‘‘Petroleum’’ in
ECT 3rd ed., Vol. 17, pp. 143–167, by R. K. Clark and J. J. Nahm, Shell Development
Co; in ECT 4th ed., Vol. 18, pp. 370–405, by R. K. Clark, Shell C&P Technology Company;
‘‘Drilling Fluids’’ in ECT (online), posting date: December 4, 2000, by R. K. Clark, Shell
E&P Technology Company.

CITED PUBLICATIONS

1. N. J. Adams and T. Charrier, Drilling Engineering, A Complete Well Planning


Approach, PennWell Books, Tulsa, Okla., 1985.
2. P. L. Moore, Drilling Practices Manual, PennWell Books, 2nd ed., Tulsa, Okla., 1985.
3. A. T. Bourgoyne, K. K. Millheim, M. E. Chenevert, and F. S. Young, Applied Drilling
Engineering, Society of Petroleum Engineers, Richardson, Tex., 1991.
4. J. K. Borchardt, ACS Symposium Series No. 396, 3 (1989).
5. T. G. J. Jones and T. L. Hughes, Adv. Chem. Ser. 251, 463 (1996).
6. R. D. Dikshit, Chem. Eng. World 26(7), 65 (1991).
7. R. D. Dikshit, Chem. Eng. World 26(8), 41 (1991).
8. G. R. Gray and H. C. H. Darley, Composition and Properties of Oil Well Drilling
Fluids, 5th ed., Gulf Publishing Co., Houston, Tex., 1988.
9. E. Anderson and co-workers, ‘‘Oil Field Chemicals’’, Specialty Chemicals – SRI
International, December 2001.
10. A. H. Hale and F. K. Mody, Society of Petroleum Engineers Symposium on Oilfield
Chemistry, SPE 25180, New Orleans, La., March. 2–5, 1993.
11. A. H. Hale and A. K. Dewan, SPE Drilling Eng. 5, 109 (1990).
12. R. K. Clark, R. F. Scheuerman, H. Rath, and H. van Laar, J. Pet. Tech. 28, 719
(1976).
13. R. P. Steiger, J. Pet. Tech. 34, 1661 (1982).
14. R. B. Bennett, J. Pet. Tech. 36, 975 (1984).
15. L. J. Fraser, SPE Drilling Eng. 7, 20 (1992).
16. J. P. Simpson, J. Pet. Tech. 31, 643 (1979).
17. F. B. Growcock, S. L. Andrews, and T. P. Frederick, Society of Petroleum Engineers
Drilling Conference, IADC/SPE Paper 27450, Dallas, Tex., Feb. 15–18, 1994.
18. J. E. Friedheim, J. Pet. Tech. 49(7), 724 (1997).
19. Standard Procedure for Field Testing Water-Based Drilling Fluids, API RP 13B-1,
1st ed., American Petroleum Institute, Washington, D.C., June 1990.
20. Standard Procedure for Field Testing Oil-Based Drilling Fluids, API RP 13B-2, 2nd
ed., American Petroleum Institute, Washington, D.C., Dec. 1991.
21. Specification for Drilling Fluid Materials, API Spec. 13A, 15th ed., Sect. 2, American
Petroleum Institute, Washington, D.C., May 1, 1993.
22. Ref. 21, Sect. 3.
23. S. W. Golis, Society of Petroleum Engineers Annual Meeting, SPE 13156, Houston,
Tex., Sept. 16–19, 1984.
Vol. 1 DRILLING FLUIDS 287

24. J. P. Sloan, J. P. Brooks, and S. F. Dear, J. Pet. Tech. 27, 15 (1975).


25. T. Mondshine, Oil Gas J., 79 (Jan. 19, 1981).
26. J. M. McLarty, J. W. Dobson, and M. A. Dick, Offshore Technology Conference, OTC
7352, Houston, Tex., May 3–6, 1993.
27. H. van Olphen, An Introduction to Clay Colloid Chemistry, 2nd ed., John Wiley &
Sons, Inc., New York, 1977.
28. Ref. 21, Sect. 4–6.
29. J. L. Lummus and L. J. Field, Pet. Engr. 59 (Mar. 1968).
30. J. L. Burba, W. E. Holman, and C. R. Crabb, Society of Petroleum Engineers Drilling
Conference, IADC/SPE 17198, Dallas, Tex., Feb. 28–March 2, 1988.
31. L. L. Carney and R. L. Meyer, Society of Petroleum Engineers Annual Meeting, SPE
Paper 6025, New Orleans, La., Oct. 1976.
32. L. L. Carney and N. Guven, Soc. Pet. Eng. J. 20, 385 (1980).
33. Ref. 21, Sects. 7 and 8.
34. J. W. Jordan, J. Phys. Chem. 53, 294 (1949).
35. J. W. Jordan, B. J. Hook, and C. M. Finlayson, J. Phys. Chem. 54, 1196 (1950).
36. U.S. Pat. 4,450,095 (May 22, 1984), C. M. Finlayson (to NL Industries).
37. T. L. Hughes, T. G. J. Jones, and O. H. Houwen, SPE Drilling Complet. 8, 157
(1993).
38. K. S. Kang and D. J. Pettitt, in R. L. Whistler and J. N. BeMiller, eds., Industrial
Gums, 3rd ed., Academic Press, Inc., New York, 1993, Chapt. 20, pp. 537–578.
39. Ref. 21, Sect. 10.
40. A. J. Desmarais and R. F. Wint, in Ref. 38, Chapt. 19, pp. 505–535.
41. U.S. Pat. 3,953,335 (Apr. 27, 1976), J. M. Jackson.
42. K. S. Kang and D. J. Pettitt, in Ref. 38, Chapt. 13, pp. 341–397.
43. R. D. Carico, Society of Petroleum Engineers Regional Meeting, SPE Paper 5870,
Long Beach, Calif., Apr. 8–9, 1976.
44. G. P. Sanz, M. C. Gunningham, H. C. Lau, and A. J. Samuel, SPE Drilling Complet.
9, 139 (1994).
45. J. W. Powell, C. F. Parks, and J. M. Seheult, Society of Petroleum Engineers Arctic
Technology Conference, SPE 22066, Anchorage, Alaska, May 29–31, 1991.
46. H. Maier, M. Anderson, C. Karl, K. Magnisuson, and R. L. Whistler, in Ref. 38,
Chapt. 8, pp. 181–226.
47. A. G. Kadaster, G. J. Guild, G. L. Hanni, and D. D. Schmidt, SPE Drilling Eng. 44,
191 (1992).
48. R. C. Portnoy, R. D. Lundberg, and E. R. Werlein, IADC/SPE Drilling Conference,
IADC/SPE 14795, Dallas, Tex., Feb. 10–12, 1986.
49. U.S. Pat. 4,447,338 (May 8, 1984), R. D. Lundberg, T. O. Walker, C. P. O’Farrell, and
H. S. Makowski (to Exxon Research and Engineering Co.).
50. M. R. Rabaioli, F. Miano, T. P. Lockhart, and G. Burrafato, in Ref. 7, SPE 25179.
51. G. R. Gray and H. C. H. Darley, Composition and Properties of Oil Well Drilling
Fluids, 5th ed., Gulf Publishing Co., Houston, Tex., 1988, pp. 582–583.
52. U.S. Pat. 3,479,287 (Nov. 18, 1969), J. C. Floyd and F. J. Shell (to Phillips Petroleum
Co.).
53. C. M. Frost, J. J. Hoepner, and W. W. Fowkes, J. Chem. Eng. Data 4, 173 (1959).
54. W. W. Fowkes and C. M. Frost, U.S. Bur. Mines Rep. Invest., 5611 (1960).
55. L. L. Carney, American Petroleum Institute Southern District Meeting, API Paper
926–15-M, Division of Production, Houston, Tex., March 4–6, 1970.
56. L. S. Park, Society of Petroleum Engineers International Symposium on Oilfield
Chemistry, SPE 16281, San Antonio, Tex., Feb. 4–6, 1987.
57. R. D. Wilcox and M. A. Jarrett, Society of Petroleum Engineers Drilling Conference,
IADC/SPE 17201, Dallas, Tex., Feb. 28–March 2, 1988.
288 DRILLING FLUIDS Vol. 1

58. B. G. Chesser and D. P. Enright, Society of Petroleum Engineers Annual Conference,


SPE 8224, Las Vegas, Nev., Sept. 23–26, 1979.
59. U.S. Pat. 4,518,510 (May 21, 1985), P. A. Gleason and I. E. Brase (to National Starch
and Chemical Corp.).
60. U.S. Pat. 3,998,742 (Dec. 21, 1976), T. O. Walker (to Texaco, Inc.).
61. T. O. Walker, H. L. Dearing, and J. P. Simpson, World Oil, 93 (Nov. 1983).
62. J. A. Neal, J. M. L. Zdybak, and W. G. Bannerman, Oil Gas J., 118 (Mar. 25, 1985).
63. J. N. BeMiller, in Ref. 38, Chapt. 21, pp. 579–600.
64. Ref. 21, Sect. 11.
65. T. K. Haack, D. A. Shaw, and D. E. Greenley, Oil Gas J., 81 (Jan. 6, 1986).
66. D. C. Thomas, Soc. Pet. Engrs. J. 22, 171 (1982).
67. J. R. Pettigrove, Drilling, 120 (May 15, 1964).
68. S. Palumbo, D. Giacca, M. Ferrari, and P. Pirovano, Society of Petroleum Engineers
International Symposium on Oilfield Chemistry, SPE 18477, Houston, Tex., Feb. 8–
10, 1989.
69. C. S. Scanley, World Oil, 122 (July 1959).
70. U.S. Pat. 3,441,504 (Apr. 29, 1969), W. C. Browning (to Milchem, Inc.).
71. U.S. Pat. 3,766,229 (Oct. 16, 1973), F. Turner and J. W. Nahm (to Dresser Industries,
Inc.).
72. U.S. Pat. 3,956,140 (May 11,091 1976), J. W. Nahm and D. A. Rowe (to Dresser
Industries, Inc.).
73. W. R. Clements, M. A. Jarrett, and E. K. Morton, Society of Petroleum Engineers
Annual Conference, SPE 16686, Dallas, Tex., Sept. 27–30, 1987.
74. J. P. Plank, Oil Gas J., 40 (Mar. 2, 1992).
75. M. Hille, Society of Petroleum Engineers International Symposium on Oilfield and
Geothermal Chemistry, SPE 13558, Phoenix, Ariz., Apr. 9–11, 1985.
76. A. J. Son, T. M. Ballard, and R. E. Loftin, SPE Drilling Eng. 2, 209 (1987).
77. U.S. Pat. 4,547,299 (Oct. 15, 1985), J. M. Lucas (to Milchem Inc.).
78. K. H. W. Ujma and J. P. Plank, SPE Drilling Eng. 4, 41 (1989).
79. T. Hemphill, World Oil, 81 (Nov. 1989).
80. L. Altazin, Oil Gas J., 76 (Aug. 4, 1980).
81. T. C. Mondshine, Oil Gas J., 70 (July 14, 1969).
82. M. E. Chenevert, J. Pet. Tech. 22, 1309 (1970).
83. A. H. Hale, F. K. Mody, and D. P. Salisbury, SPE Drilling Complet. 8, 207 (1993).
84. J. Elward-Berry and E. W. Thomas, Society of Petroleum Engineers Drilling Confer-
ence, IADC/SPE 27453, Dallas, Tex., Feb. 15–18, 1994.
85. G. Radenti, S. Palumbo, and G. Zucca, Pet. Eng. Int., 32 (Sept. 1987).
86. P. I. Reid, R. C. Minton, and A. Twynam, Society of Petroleum Engineers European
Petroleum Conference, SPE 24979, Cannes, France, Nov. 16–18, 1992.
87. F. J. Shell, Drilling-DCW, 47 (Nov. 1969).
88. D. P. Vesely, Society of Petroleum Engineers Rocky Mountain Regional Meeting, SPE
17511, Casper, Wyo., May 11–13, 1988.
89. T. C. Mondshine, Oil Gas J., 84 (March 21, 1966).
90. D. E. O’Brien and M. E. Chenevert, J. Pet. Tech. 25, 189 (1973).
91. D. B. Anderson and C. D. Edwards, Pet. Eng., 105 (Sept. 1977).
92. U.S. Pat. 5,208,216 (May 4, 1993), C. D. Williamson, D. W. Fong, and R. K. Gabel
(to Nalco Chemical Co.).
93. S. E. Alford, Society of Petroleum Engineers Drilling Conference, SPE/IADC 21936,
Amsterdam, March 11–14, 1991.
94. T. Hemphill, R. Valenziano, P. Bale, and B. Sketchler, Oil Gas J., 60 (June 8,
1992).
95. O. Welch and L. Lee, Oil Gas J., 53 (July 13, 1992).
Vol. 1 DRILLING FLUIDS 289

96. T. W. Beihoffer, D. S. Dorrough, C. K. Deem, and D. D. Schmidt, Oil Gas J., 47


(Mar. 16, 1992).
97. T. W. Beihoffer and co-workers, Society of Petroleum Engineers Annual Conference,
SPE 24588, Washington, D.C., Oct. 4–7, 1992.
98. R. Bland, Oil Gas J., 54 (June 29, 1992).
99. P. I. Reid, G. P. Elliott, R. C. Minton, and D. A. Burt, Society of Petroleum
Engineers/EPA Exploration & Production Environmental Conference, SPE 25989,
San Antonio, Tex., March 7–10, 1993.
100. J. D. Downs, E. van Oort, D. I. Redman, D. Ripley, and B. Rothmann, Society of Pet-
roleum Engineers Offshore European Conference, SPE 26699, Aberdeen, Scotland,
Sept. 7–10, 1993.
101. A. J. Twynam, P. A. Caldwell, and K. Meads, in Ref. 98, IADC/SPE 27451.
102. U.S. Pat. 4,780,220 (Oct. 25, 1988), T. E. Peterson (to Hydra Fluids, Inc.).
103. M. E. Chenevert, Oil Gas J., 60 (July 17, 1989).
104. U.S. Pat. 5,076,373 (Dec. 31, 1991), A. H. Hale, G. C. Blytas, and A. K. R. Dewan
(to Shell Oil Co.).
105. U.S. Pat. 5,149,690 (Sept. 22, 1992), A. D. Patel and H. C. McLaurine (to M-I Drilling
Fluids Co.).
106. J. R. Bruton and H. C. McLaurine, in Ref. 18, SPE 26327.
107. W. S. Cagle and L. F. Schwertner, Oil Gas J., 61 (March 27, 1972).
108. N. Davis and C. E. Tooman, SPE Drilling Eng. 4, 47 (1989).
109. R. F. Burdyn and L. D. Wiener, World Oil, 101 (Nov. 1957).
110. J. A. Headley, T. O. Walker, and R. W. Jenkins, Society of Petroleum Engineers
Drilling Conference, SPE/IADC 29404, Amsterdam, Feb. 28–March 2, 1995.
111. D. K. Clapper and D. P. Salisbury, World Oil, 125 (Nov. 1984).
112. F. B. Growcock, C. F. Ellis, and D. D. Schmidt, SPE Drilling Complet. 9, 39
(1994).
113. W. J. McDonald, L. J. Remont, W. A. Rehm, and M. E. Chenevert, Development of
Drilling Foams for Geothermal Applications, SAND79–7122, Sandia National
Laboratories, Albuquerque, N.M., Jan. 1980.
114. P. B. Rand and O. J. Montoya, Aqueous Foam Surfactants for Geothermal Drilling
Fluids: I. Screening, SAND79–2036, Sandia Laboratories, Albuquerque, N. M., Jan.
1980.
115. G. R. Gray and H. C. H. Darley, Composition and Properties of Oil Well Drilling
Fluids, 5th ed., Gulf Publishing Co., Houston, Tex., 1988, pp. 595–598.
116. J. U. Messinger, Lost Circulation, PennWell Books, Tulsa, Okla., 1981.
117. G. E. Loeppke, D. A. Glowka, and E. K. Wright, J. Pet. Tech. 42, 328 (1990).
118. T. M. Nayberg, SPE Drilling Eng. 2, 229 (1987).
119. U.S. Pat. 5,207,282 (May 4, 1993), G. F. Fuh, N. Morita, D. L. Whitfill, and D. A.
Strah (to Conoco Inc.).
120. A. Ali, C. L. Kalloo, and B. Singh, SPE Drilling Complet. 9, 32 (1994).
121. C. J. Coberly, Drilling and Production Practices, American Petroleum Institute, New
York, 1937, pp. 189–201.
122. C. Gatlin and C. E. Nemir, J. Pet. Tech. 13, 575 (1961).
123. J. P. Gallus, J. L. Lummus, and J. F. Fox, Trans. AIME 213, 70 (1958).
124. U.S. Pat. 3,637,031 (Jan. 25, 1972), J. D. Hull and R. E. Finch (to Standard Brands
Chemical Industries, Inc.).
125. J. E. Fontenot, Oil Gas J., 58 (Jan. 14, 1974).
126. U.S. Pat. 4,063,603 (Dec. 20, 1977), J. J. Rayborn.
127. U.S. Pat. 4,123,367 (Oct. 31, 1978), A. A. Dodd.
128. U.S. Pat. 5,114,598 (May 19, 1992), J. J. Rayborn and J. P. Dickerson (to Sun Drilling
Products Corp.).
290 DRILLING FLUIDS Vol. 1

129. R. G. Bland, W. T. Micho, and H. J. Howard, Society of Petroleum Engineers Annual


Conference, SPE 24591, Washington, D.C., Oct. 4–7, 1992.
130. N. Adams, Pet. Eng., 40 (Oct. 1977).
131. H. D. Outmans, Trans. AIME 213, 265 (1958).
132. R. K. Clark and S. G. Almquist, SPE Drilling Eng. 7, 121 (1992).
133. U.S. Pat. 4,614,235 (Sept. 30, 1986), I. G. Keener, J. C. Newlove, R. C. Portnoy, D. F.
Fehler, and K. W. Pober (to Exxon Chemical Patents, Inc.).
134. U.S. Pat. 5,057,234 (Oct. 15, 1991), R. G. Bland and D. K. Clapper (to Baker Hughes
Inc.).
135. U.S. Pat. 5,247,992 (Sept. 28, 1993), R. Lockhart.
136. U.S. Pat. 5,002,672 (Mar. 26, 1991), J. R. Hayes and G. J. Campbell (to Turbo-Chem.
International, Inc.).
137. W. S. Halliday and D. K. Clapper, Society of Petroleum Engineers Drilling Confer-
ence, SPE/IADC 18684, New Orleans, La., Feb. 28–March 3, 1989.
138. C. C. Patton, Oil Gas J., 66 (July 22, 1974); Oil Gas J., 163 (July 29, 1974); Oil Gas
J., 76 (Aug. 5, 1974).
139. H. E. Bush, Society of Petroleum Engineers Annual Meeting, SPE Paper 5123,
Houston, Tex., Oct. 6–9, 1974.
140. T. Cox, Drilling Contractor, 37 (May–June 1975).
141. U.S. Pat. 3,899,293 (Aug. 12, 1975), H. E. Bush (to NL Industries, Inc.).
142. R. L. Garrett, R. K. Clark, L. L. Carney, and C. K. Grantham, J. Pet Tech. 31, 787
(1979).
143. U.S. Pat. 4,008,775 (Feb. 22, 1977), I. Fox (to Ironite Products Co.).
144. R. K. Clark and T. Hemphill, SPE Drilling Complet. 9, 176 (1994).
145. J. M. Simon, Society of Petroleum Engineers Annual Conference, SPE 24602,
Washington, D.C., Oct. 4–7, 1992.
146. D. P. Enright, W. M. Dye, and F. M. Smith, SPE Drilling Eng. 7, 15 (1992).
147. U.S. Pat. 5,007,489 (Apr. 16, 1991), D. P. Enright, W. M. Dye, F. M. Smith, and A. C.
Perricone (to Baker Hughes Inc.).
148. J. D. Downs, Society of Petroleum Engineers International Symposium on Oilfield
Chemistry, SPE 25177, New Orleans, La., March 2–6, 1993.
149. M. Byrne, I. Patey, L. George, J. Downs and J. Turner, Society of Petroleum
Engineers International Symposium and Exhibition on Formation Damage Control,
SPE 73766, Lafayette, La., Feb. 20–21, 2002.
150. J. D. Downs, European Petroleum Conference, SPE 24973, Cannes, France, Nov. 16–
18, 1992.
151. J. D. Downs and co-workers, 2nd International Conference on Health, Safety &
Environment in Oil & Gas Exxploration & Production, SPE 27143, Jakarta,
Indonesia, Jan. 25–27, 1994.
152. S. K. Howard, SPE 30498, Society of Petroleum Engineers Annual Technical Confer-
ence, Dallas, Texas, Oct. 22–25, 1995.
153. W. Benton and J. Turner, Drilling Contractor, 38, 41 (May/June 2000).
154. A. Saasen and co-workers, IADC/SPE 74541. Society of Petroleum Engineers Drilling
Conference, Dallas, Texas. Feb. 26–28, 2002.
155. Standard Procedure for Laboratory Testing Drilling Fluids, API RP 13I, 4th ed.,
American Petroleum Institute, Washington, D.C., June 1990.
156. U.S. Fed. Reg. 51(131), 24897 (July 9, 1986).
157. R. Bleier, A. J. J. Leuterman, and C. Stark, J. Pet. Tech. 45, 6 (1993).
158. R. K. Clark, J. Pet. Tech. 46, 804 (1994).
159. PCT Int. Appl. WO 97/05212, (Feb. 13, 1997) J. C. Urquhart.
160. Brit. UK Pat. App. GB 2334279 (Aug.18, 1999), W. S. Halliday and D. K. Clapper
(to Baker Hughes Inc.).
Vol. 1 PETROLEUM REFINERY PROCESSES 291

161. PCT Int. Appl. WO 99/23188 (May 14, 1999), D. P. Vollmer and R. L. Horton
(to Great Lakes Chemical).
162. PCT Int. Appl. WO 98/40446a (Sep. 17, 1998), W. S. Halliday, D. K. Clapper, and
M. R. Smalling (to Baker Hughes Inc.).
163. Eur. Pat. Appl. EP 1074598 (Feb. 7, 2001), J. W. Dobson, K. O. Tresco and J. Lay
(to Texas United Chemical Co.LLC).
164. PCT Int. Appl. WO 98/08914 (March 5, 1998), J. A. C. Van Ooyen (to Verdugt B.V.).
165. U.S. Pat. 5,547,925 (Aug. 20, 1996), W. M. Duncan (to Integrated Industries Inc.).
166. PCT Int. Appl. WO 99/48994 (Sep. 30, 1999), C.-A. Sjogreen and B. Sallmen (to
Perstorp AB).
167. Eur. Pat. Appl. EP 1103589 (Aug. 16, 2001), S. Carminati, A. Guarneri, and
M. Brignoli.
168. Eur. Pat. Appl. EP 0764709 (March 26, 1997), A. D. Patel (to M-I Drilling Fluids).
169. PCT Int. Appl. WO 97/10313 (March 20, 1997), L. Lee and co-workers (to M-I Drilling
Fluids).
170. PCT Int. Appl. WO 97/34963 (Sep. 25, 1997), R. J. Wittenbrink, C. J. Mart, and D. F.
Ryan (to Exxon Research and Engineering Co.).
171. U.S. Pat. 2003/203821 (Oct. 30, 2003), K. B. Fox.
172. PCT Int. Appl. WO 00/71635 (Nov. 30, 2000), E. E. Miller and W. J. Benton (to Cabot
Corp.).
173. PCT Int. Appl. WO 97/26310 (July 24, 1997), D. P. Vollmer, P. H. Javora, and R. L.
Horton (to Great Lakes Chemical).
174. PCT Int. Appl. WO 97/26311 (July 24, 1997), D. P. Vollmer, P. H. Javora and R. L.
Horton (to Great Lakes Chemical).
175. Eur. Pat. Appl. EP 0726302 (Feb. 9, 1996), J. W. Dobson, T. D. Robertson, and K. B.
Mondshine (to Texas United Chemical Co.).
176. PCT Int. Appl. WO 98/21291 (May 22, 1998), J. R. Murphey and co-workers (to Tetra
Technologies Inc.).
177. K. K. Saxena and co-workers, Trends in Carbohydrate Chemistry 5, 139–151 (1999).
178. PCT Int. Appl. WO 98/03609 (Jan. 29, 1998), C. A. Sawdon and A. J. Bradbury
(to Schlumberger Canada Ltd.).

SHMUEL D. UKELES
BARUCH GRINBAUM
IMI (TAMI) Institute for Research and
Development, DSBG (Israel)

PETROLEUM REFINERY
PROCESSES
1. Introduction

Petroleum (also called crude oil), in the unrefined or crude form, like many
industrial feedstocks has little or no direct use and its value as an industrial com-
modity is only realized after the production of salable products. Even then, the
market demand dictates the type of products that are needed. Therefore, the

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.1805060919160509.a01.pub2
292 PETROLEUM REFINERY PROCESSES Vol. 1

Table 1. Distillation Fractions of Petroleum


Fraction Boiling, 8C
light naphtha 1 to 150
gasoline 1 to 180
heavy naphtha 150 to 205
kerosene 205 to 260
stove oil 205 to 290
light gas oil 260 to 315
heavy gas oil 315 to 425
lubricating oil >400
vacuum gas oil 425 to 600
residuum >600

value of petroleum is directly related to the yield of products and is subject to the
call of the market.
Petroleum refining, also called petroleum processing, is the recovery and/or
generation of usable or salable fractions and products from crude oil, either by
distillation or by chemical reaction of the crude oil constituents under the effects
of heat and pressure. Synthetic crude oil, produced from tar sand (oil sand) bitu-
men, is also used as feedstocks in some refineries. Heavy oil conversion (1), as
practiced in many refineries, does not fall into the category of synthetic fuels
(syncrude) production. In terms of liquid fuels from coal and other carbonaceous
feedstocks, such as oil shale, the concept of a synthetic fuels industry has dimin-
ished over the past several years as being uneconomical in light of current pet-
roleum prices.
As the basic elements of crude oil, hydrogen and carbon form the main input
into a refinery, combining into thousands of individual constituents and the eco-
nomic recovery of these constituents varies with the individual petroleum accord-
ing to its particular individual qualities, and the processing facilities of a
particular refinery. In general, crude oil, once refined, yields three basic groupings
of products that are produced when it is separated into a variety of different gen-
eric, but often overlapping fractions (Table 1). The amounts of these fractions pro-
duced by distillation depend on the origin and properties of crude petroleum (2).
The gas and gasoline cuts form the lower boiling products and are usually
more valuable than the higher boiling fractions and provide gas (liquefied petro-
leum gas), naphtha, aviation fuel, motor fuel, and feedstocks, for the petrochem-
ical industry. Naphtha, a precursor to gasoline and solvents, is extracted from
both the light and middle range of distillate cuts and is also used as a feedstock
for the petrochemical industry. The middle distillates refer to products from
the middle boiling range of petroleum and include kerosene, diesel fuel, distillate
fuel oil, and light gas oil; waxy distillate and lower boiling lubricating oils are
sometimes include in the middle distillates. The remainder of the crude oil
includes the higher boiling lubricating oils, gas oil, and residuum (the nonvolatile
fraction of the crude oil). The residuum can also produce heavy lubricating oils
and waxes, but is more often used for asphalt production. The complexity of
petroleum is emphasized insofar as the actual proportions of light, medium,
and heavy fractions vary significantly from one crude oil to another.
Vol. 1 PETROLEUM REFINERY PROCESSES 293

When petroleum occurs in a reservoir that allows the crude material to be


recovered by pumping operations as a free-flowing dark-to-light colored liquid, it
is often referred to as conventional petroleum. In some oil fields, the downhole
pressure is sufficient for recovery without the need for pumping. Heavy oil differs
from conventional petroleum in that its flow properties are reduced and it is
much more difficult to recover from the subsurface reservoir. These materials
have a much higher viscosity and lower API (American Petroleum Institute)
gravity than conventional petroleum, and primary recovery of these petroleum
types usually requires thermal stimulation of the reservoir.
Heavy oil generally has an API gravity of <208 and usually, but not always,
a sulfur content of >2% by weight. Extra heavy oil occurs in the near-solid state
and is virtually incapable of free flow under ambient conditions. Tar sand bitu-
men, often referred to as native asphalt, is a subclass of extra heavy oil and is
frequently found as the organic filling in pores and crevices of sandstones, lime-
stones, or argillaceous sediments, in which case the organic and associated
mineral matrix is known as rock asphalt.
A residuum, often shortened to resid, is the residue obtained from petroleum
after nondestructive distillation has removed all the volatile materials. The tem-
perature of the distillation is usually <3458C because the rate of thermal decom-
position of petroleum constituents is substantial >3508C. Temperatures as high as
4258C can be employed in vacuum distillation. When such temperatures are
employed and thermal decomposition occurs, the residuum is usually referred to
as pitch. By inference, the name is used in the same manner as when it refers to
the nonvolatile residue from the thermal decomposition of coal tar (3).
Asphalt, prepared from petroleum, often resembles native asphalt. When
asphalt is produced by distillation, the product is called residual, or straight-run,
asphalt. However, if the asphalt is prepared by solvent extraction of residua or by
light hydrocarbon (propane) precipitation, or if it is blown or otherwise treated, the
name should be modified accordingly to qualify the product, eg, propane asphalt.
Sour and sweet are terms referring to a crude oil’s approximate sulfur con-
tent, which relates to odor. A crude oil that has a high sulfur content usually con-
tains hydrogen sulfide, H2S, and/or mercaptans, RSH; it is called sour. Without
this disagreeable odor, the crude oil is judged sweet.

2. History

The use of petroleum or derived materials, such as asphalt, and the heavier non-
volatile crude oils is an old art (2). In fact, petroleum utilization has been docu-
mented for >5000 years. The earliest documented uses occurred in Mesopotamia
(ancient Iraq) when it was recognized that the nonvolatile derivatives (bitumen
or natural asphalt and manufactured asphalt) could be used for caulking and as
an adhesive for jewelry or as a mastic for construction purposes. There is also
documented use of bitumen for medicinal use.
Approximately 2000 years ago, Arabian scientists developed methods for the
distillation of petroleum, which were introduced into Europe by way of the Arabian
incursions into Spain. Petroleum, used in China since it was encountered when
drilling for salt, appears in documents of the third century. The Baku region of
294 PETROLEUM REFINERY PROCESSES Vol. 1

northern Persia was also reported by Marco Polo in 1271–1273 as having a com-
mercial petroleum industry.
Interest in naphtha (nafta) began with the discovery that petroleum could be
used as an illuminant and as a supplement to bituminous incendiaries, which were
becoming increasingly common in warfare. Greek fire was a naphtha–bitumen (or
naphtha–asphalt) mix; the naphtha provided the flame and the bitumen (or
asphalt) provided the adhesive properties that prolonged the incendiary effect.
Modern refining began in 1859 with the discovery of petroleum in
Pennsylvania. After completion of the first well, the surrounding areas were
immediately leased and extensive drilling took place. In the post-1945 era, Middle
Eastern countries continued to rise in importance because of new discoveries of
vast reserves. The United States, though continuing to be the biggest producer,
was also the principal consumer and thus was not an exporter of oil. At this time,
oil companies began to roam much farther in the search for oil, which has
resulted in significant discoveries in Europe, Africa, and Canada.
The impetus to develop the petroleum refining industry came from several
changes in life-styles. The increased needs for illuminants, for fuel to drive the fac-
tories of the industrial revolution, for gasoline to power the automobiles, as well as
the demand for aviation fuel, all contributed to the increased use of petroleum.
The product slate has also changed. The increased demand for gasoline and
lubricants brought about an emphasis on refining crude oil. This, in turn,
brought about changes in the way crude oil was refined and led to innovations
and developments in the refining industry, thereby giving birth to the integrated
petroleum refinery (Fig. 1).

3. Feedstock Evaluation

The evaluation of petroleum feedstocks is determined by using a series of staan-


dard test methods that are set by organizations varying from country to country.
For example, in the United States, the American Society for Testing and Materi-
als (ASTM) and the American Petroleum Institute (API) are recognized for estab-
lishing specifications on both products and methods for testing. In the United
Kingdom, it is the Institute of Petroleum (IP); in Germany, it is Deutsche Institut
für Normung (DIN); and in Japan, it is the Ministry of International Trade and
Industry (MITI). Three frequently specified properties are density–specific
gravity–API gravity, characterization factor, and sulfur content (2,4,5).
Many petroleum companies use these standards but many also have their
own individual standards that are proprietary and very rarely published.
The API gravity is a measure of density or specific gravity:

API ¼ ð141:5=specific gravityÞ 131:5

Specific gravity is the ratio of the weight of a given volume of oil to the
weight of the same volume of water at a standard temperature, usually 608F
(15.68C). This method of measuring density and gravity first arose as a result
of the need to define the character of products in more detail; it was natural to
extend the measure to crude oils in general.
Vol. 1 PETROLEUM REFINERY PROCESSES 295

LPG Petrochemical
H
fuel feedstock

Gas C4S
plant H2SO4
Catalytic Gasoline
alkylation
reformer

160°C Naphtha
hydrofiner Catalytic
hydrocracker

160– 230°C Kerosene Jet fuel


Atmospheric hydrofiner diesel oil
Feedstock pipe still 230– 300°C

300– 340°C

Mid-distillate
Kerosene
hydrofiner

Catalytic
cracker

Heating oil
340°C
Fuel oil

Refinery heavy-
ends processing
(a)

340– 470°C Catalytic


Clay
cracker
treater

Atmospheric Vacuum
residuum pipe still 470– 560°C
Vacuum Gasoline
gas– oil
hydrotreater

560°C H2
Coker Hydrocracker in
fuel processing
Coke
Solvent Precipitate
deasphalting Asphalt

Soluble
Solvent Aromatic oils
extraction

Lube, waxes,
greases
(b)

Fig. 1. General refinery operations: (a) light petroleum refining section; (b) heavy
feedstock refining section.
296 PETROLEUM REFINERY PROCESSES Vol. 1

The Watson characterization factor (K) has also been used as a measure of
the chemical character of a crude oil or its fractions:

K ¼ ðTB Þ1=3 =specific gravity

where TB is the absolute boiling point in degrees Rankine (8R ¼ 9/5 K) and speci-
fic gravity is specific gravity compared to water at 608F (15.68C).
For a wide boiling range material such as crude oil, the boiling point is
taken as an average of the five temperatures at which 10, 30, 50, 70, and 90%
of the material is vaporized. A highly paraffinic crude oil can have a character-
ization factor as high as 13, whereas a highly naphthenic crude oil can be as low
as 10.5, and the breakpoint between the two types of crude oil is 12.

4. Refinery Processes

A refinery is a group of manufacturing plants that vary in number according to


the variety of products produced (1,2). Refinery processes must be selected to
convert crude oil into products according to demand. A refinery must also be
flexible and be able to change operations as needed, especially if heavier oils
are the primary feedstocks. This is accomplished through two basic process con-
cepts: carbon rejection (eg, coking processes) and hydrogen addition (eg, hydro-
processes). However, certain downstream processes, such as catalytic reforming,
applied to the product streams do not fit into either of these categories.
The conversion of conventional petroleum to products is well established
but the conversion of high boiling feedstocks (such as heavy oils and residua)
is now in a significant transition period as the demand for transportation
fuels increases. In order to satisfy the changing pattern of product demand, sig-
nificant investments in heavy feedstock conversion processes will be necessary
and technologies are needed that will take conversion of these feedstocks beyond
current limits and, at the same time, reduce the amount of coke and other non-
essential products.
Such a conversion schemes may require the use of two or more technologies in
series rather than an attempt to develop a whole new one-stop conversion technol-
ogy. New processes for the heavy feedstock conversion probably are used perhaps
not in place of but in conjunction with of visbreaking and coking options with
some degree of hydroprocessing as a primary conversion step. In addition, other
processes may replace or, more likely, augment the deasphalting units in many
refineries.
There remains room for improving coking and hydroconversion processes
by reducing hydrocarbon gas formation, by inhibiting the formation of poly-
nuclear aromatic compounds not originally present in the resid, and by separat-
ing an intermediate quality fraction, which produces low amounts of coke, before
or during conversion. In addition, the challenge for hydroconversion is to take
advantage of the nickel and vanadium in the heavy feedstocks to generate an
in situ dispersed catalyst and to eliminate catalyst cost.
4.1. Desalting and Dewatering. Crude oil is recovered from the reser-
voir mixed with a variety of substances: gases, water, and dirt (minerals) (6).
Vol. 1 PETROLEUM REFINERY PROCESSES 297

Thus, refining actually commences with the production of fluids from the well or
reservoir and is followed by pretreatment operations that are applied to the
crude oil either at the refinery or prior to transportation. Pipeline operators,
eg, are insistent upon the quality of the fluids put into the pipelines; therefore,
any crude oil to be shipped by pipeline or, for that matter, by any other form of
transportation must meet rigid specifications in regard to water and salt content.
In some instances, sulfur content, nitrogen content, and viscosity may also be
specified.
Field separation, which occurs at a field site near the recovery operation, is
the first attempt to remove the gases, water, and dirt that accompany crude oil
coming from the ground. The separator may be no more than a large vessel that
gives a quieting zone for gravity separation into three phases: gases, crude oil,
and water containing entrained dirt.
Desalting operations are necessary to remove salt from the brines that
are present with the crude oil after recovery. The salt or brine suspensions
may be removed from crude oil by heating (90–1508C, 200–3008F) under pres-
sure (50–250 psi) that is sufficient to prevent vapor loss and then allowing the
material to settle in a large vessel. Alternatively, coalescence is aided by passage
through a tower packed with sand, gravel, and the like.
Desalting is a water-washing or electrostatic operation performed at the
production field and at the refinery site for additional crude oil cleanup. If the
petroleum from the separators contains water and dirt, water washing can
remove much of the water-soluble minerals and entrained solids. If these crude
oil contaminants are not removed, they can cause operating problems during
refinery processing, such as equipment plugging and corrosion as well as catalyst
deactivation.
The usual practice is to blend crude oils of similar characteristics, although
fluctuations in the properties of the individual crude oils may cause significant
variations in the properties of the blend over a period of time. Blending several
crude oils prior to refining can eliminate the frequent need to change the proces-
sing conditions that may be required to process each of the crude oils individu-
ally. However, incompatibility of different crude oils, which can occur if, eg, a
paraffinic crude oil is blended with a heavy asphaltic oil, can cause sediment for-
mation in the unrefined feedstock or in the products, thereby complicating the
refinery process (7).
Emulsions may also be broken by addition of treating agents, such as soaps,
fatty acids, sulfonates, and long-chain alcohols. When a chemical is used for
emulsion breaking during desalting, it may be added at one or more of three
points in the system. First, it may be added to the crude oil before it is mixed
with fresh water. Second, it may be added to the fresh water before mixing
with the crude oil. Third, it may be added to the mixture of crude oil and water.
A high potential field across the settling vessel also aids coalescence and breaks
emulsions, in which case dissolved salts and impurities are removed with the
water.
4.2. Distillation. The first and most fundamental step in the refining
process (after the crude oil has been cleaned and any remnants of brine removed)
is distillation and which is often referred to as the primary refining process. Dis-
tillation involves the separation of the different hydrocarbon compounds that
298 PETROLEUM REFINERY PROCESSES Vol. 1

occur naturally in a crude oil, into a number of different fractions (a fraction is


often referred to as a cut).
Originally, distillation was a batch operation in which the still was a cast-
iron vessel mounted on brickwork over a fire and the volatile materials were
passed through a pipe or gooseneck that led from the top of the still to a conden-
ser. The latter was a coil of pipe, or a worm (hence the expression worm end
products), immersed in a tank of running water.
Fractions from the atmospheric and vacuum towers are often used as feed-
stocks to these second stage refinery processes that break down the fractions, or
bring about a basic chemical change in the nature of a particular hydrocarbon
compound to produce specific products.
Atmospheric Distillation. The petroleum distillation unit in the 1990s
brings about a fairly efficient degree of fractionation (separation). The feed to a
distillation tower is heated by flow through pipes arranged within a large fur-
nace. The heating unit is known as an atmospheric pipe still heater or pipe still
furnace, and the heating unit and fractional distillation tower make up the
essential parts of a distillation unit or pipe still.
The pipe still furnace heats the feed to a predetermined temperature,
usually a temperature at which a predetermined portion of the feed changes
into vapor. The vapor is held under pressure in the pipe in the furnace until it
discharges as a foaming stream into the fractional distillation tower. Here the
nonvolatile or liquid portion of the feed descends to the bottom of the tower to
be pumped away as a bottom nonvolatile product, while the vapors pass up the
tower to be fractionated into gas oil, kerosene, and naphtha.
Pipe still furnaces vary greatly and, in contrast to the early units where
capacity was usually 31.8–79.5 m3/day (200–500 bbl/day), can now accommodate
3975 m3 (25,000 bbl) or more of crude oil per day. The walls and ceiling are insu-
lated with firebrick and the interior of the furnace is partially divided into two
sections: a smaller convection section where the oil first enters the furnace and a
larger section fitted with heaters where the oil reaches its highest temperature.
The primary fractions from a distillation unit are equilibrium mixtures
and contain some proportion of the lighter constituents characteristic of a
lower boiling fraction. The primary fractions are stripped of these constituents
(stabilized) before storage or further processing.
Vacuum Distillation. Vacuum distillation evolved as the need arose to
separate the less volatile products, such as lubricating oils, from petroleum
without subjecting these higher boiling materials to cracking conditions.
The boiling point of the heaviest cut obtainable at atmospheric pressure
(101.3 kPa ¼ 760 mmHg) is limited by the temperature (3508C, 6628F) at
which the residue starts to decompose or crack. It is at this point that distillation
in a vacuum pipe still is initiated.
Operating conditions for vacuum distillation are usually in the range of
7–13 kPa (50–100 mmHg). In order to minimize large fluctuations in pressure
in the vacuum tower, the units are usually shorter in height and larger in
diameter than the respective atmospheric units. By this means, a heavy gas
oil can be obtained as an overhead product at temperatures of 1508C (3028F)
and lubricating oil cuts can be obtained at 250–3508C (482–6628F). In some
designs, the partial pressure of the hydrocarbons is reduced still further by the
Vol. 1 PETROLEUM REFINERY PROCESSES 299

injection of steam that is added to the column for stripping the nonvolatile con-
stituents in the base of the column.
Fractions from the atmospheric and vacuum towers are often used as feed-
stocks to these second stage refinery processes that break down the fractions, or
bring about a basic chemical change in the nature of a particular hydrocarbon
compound to produce specific products.
Alternatively, the atmospheric residuum or the vacuum residuum may be
sent to a deasphalting unit to remove the very high molecular weight constitu-
ents (collectively called asphalt) and the deasphalted oil is used as feedstock to a
catalytic cracking unit. Solvent deasphalting is a unique separation process in
which the residue is separated by molecular weight (density), instead of by boil-
ing point as in the distillation process. The purpose of a deasphalting unit is to
produce asphalt as final product and to produce a soluble stream (deasphalted oil
that can be used in a catalytic cracking unit and have reduced coke deposition
on the catalyst. The solvents used in a deasphalting unit vary from propane to
pentane and the yields of asphalt and deasphalted oil vary with the hydrocarbon
used in the unit.
Azeotropic Distillation and Extractive Distillations. Effective as they are
for producing various liquid fractions, distillation units generally do not produce
specific fractions. In order to accommodate the demand for such products, refi-
neries have incorporated azeotropic distillation and extractive distillation in
their operations.
The principle of azeotropic distillation depends on the ability of a chemically
dissimilar compound to cause one or both components of a mixture to boil at a
temperature other than the one expected. Thus, the addition of a nonindigenous
component forms an azeotropic mixture with one of the components of the
mixture, thereby lowering the boiling point and facilitating separation by distil-
lation. The separation of components of similar volatility may become economical
if an entrainer can be found that effectively changes the relative volatility. It is
also desirable that the entrainer be reasonably cheap, stable, nontoxic, and read-
ily recoverable from the components. In practice, it is probably the ready recover-
ability that limits the application of extractive and azeotropic distillation.
The majority of successful processes are those in which the entrainer and
one of the components separate into two liquid phases on cooling if direct recov-
ery by distillation is not feasible. A further restriction in the selection of an azeo-
tropic entrainer is that the boiling point of the entrainer is 10–408C (50–1048F)
below that of the components.
4.3. Thermal Cracking. When kerosene (lamp oil) was the principal
desired product, gasoline was the portion of crude petroleum too volatile to be
included in kerosene. The refiners of the 1890s and early 1900s had no use for
gasoline and often discarded an accumulation of it. However, as the demand
for gasoline and aviation fuel increased with the onset of World War I and the
increased production of automobiles during the 1920s, more fuels had to be pro-
duced to meet the demand.
The problem of how to produce more of the lower boiling fractions
from crude oil was solved in 1913 when cracking units were incorporated into
refinery operations and fractions heavier than gasoline were converted into gaso-
line by thermal decomposition. The use of residua as feedstocks for thermal
300 PETROLEUM REFINERY PROCESSES Vol. 1

processes has become economically advantageous because, on the one hand, the
end result is the production of lower boiling salable materials, and, on the
other, the asphaltic materials in the residua are regarded as the unwanted
coke-forming constituents.
The basic processes introduced to bring about thermal decomposition of the
higher boiling streams are known as cracking. In these processes, the higher
boiling fractions are converted to lower boiling products. Catalytic cracking is
the most common cracking process, in which heavy feedstock or cuts are broken
down or changed by being heated, and reacted with catalysts.
The concept behind thermal cracking is the thermal decomposition of
higher molecular weight constituents of petroleum to produce lower molecular
weight, normally more valuable, products. The first commercial process was in
1913 and is known as the Burton process. Even though catalyst cracking gene-
rally replaced thermal cracking in 1940s, noncatalytic cracking processes using
high temperature to achieve the decomposition are still in operation. Catalytic
processes usually produce more gasoline having a higher octane number, but
with lesser yields of gases and heavy fuel oil. The gases produced by catalytic
cracking contain more olefins than those produced by thermal cracking.
Thermal Cracking. In the thermal cracking process, a feedstock (eg, gas
oil) is fed to the fractionator with their thermal reactivity to separate gasoline,
light, and heavy oil. The light oil is then fed to the heater at 540–5958C (1000–
11008F) and a pressure of 350–700 psi, the light oil transforms to the vapor
phase and is sent to the soaker. If the feedstock is heavy oil, temperatures on
the order of 400–4808C (750–9008F) are used and higher pressures (350–
700 psi) are used to maintain the feedstock in the liquid phase, then it is fed to
the soaker. The liquid- and vapor-phase mix in the soaker and sent to the separa-
tor, with the products coming out on the bottom as fuel oil and the light recycle
back to the fractionator. Coking in the reactor is the main problem when heavy
oil is heated at high temperatures.
Visbreaking. Viscosity breaking (reduction) is a mild cracking operation
used to reduce the viscosity of residual fuel oils and residua (8). The process,
evolved from the older and now obsolete thermal cracking processes, is classed
as mild because the thermal reactions are not allowed to proceed to completion.
Residua are sometimes blended with lighter heating oils to produce fuel oils
of acceptable viscosity. By reducing the viscosity of the nonvolatile fraction,
visbreaking reduces the amount of the more valuable light heating oil that is
required for blending to meet the fuel oil specifications. The process is also
used to reduce the pour point of a waxy residue. Visbreaking conditions range
from 455 to 5108C (851–9508F) and 345 to 2070 kPa (50 to 300 psi) at the heating
coil outlet. Liquid-phase cracking takes place under these low severity condi-
tions. In addition to the primary product, fuel oil, material in the gas oil and
gasoline boiling range is produced. Gas oil can be used as additional feed for cat-
alytic cracking units or as heating oil.
In the process (Fig. 2), a crude oil residuum is passed through a furnace
where it is heated to 4808C (896 8F) under an outlet pressure of 690 kPa
(100 psi). The heating coils in the furnace are arranged to provide a soaking
section of low heat density, where the charge remains until the visbreaking
reactions are completed; subsequently, the cracked products are passed into a
Vol. 1 PETROLEUM REFINERY PROCESSES 301

Crude Gasoline
oil
Heavy gas oil

Vacuum
Atmospheric
fractionator
fractionator

Reactor

Flasher

Tar

Light gas oil

Fig. 2. The visbreaking process.

flash-distillation chamber. The overhead material from this chamber is then


fractionated to produce a low quality gasoline as an overhead product and a
light gas oil as bottom. The liquid products from the flash chamber are cooled
with a gas oil flux and then sent to a vacuum fractionator. This yields a heavy
gas oil distillate and a residual tar of reduced viscosity. Quench oil may also
be used to terminate the reactions (2,9).
The main limitation to thermal conversion is that the products can be
unstable. Thermal cracking at low pressure gives olefins, particularly in the
naphtha fraction; such olefins yield an unstable product that tends to form
gum as well as heavier products that form sediments (7).
Coking Processes. Coking is a generic term for a series of thermal pro-
cesses used for the conversion of nonvolatile heavy feedstocks into lighter, distill-
able products (10). The feedstock is typically a residuum and the products are
gas, naphtha, fuel oil, gas oil, and coke. Gas oil can be the primary product of
a coking operation and serves primarily as a feedstock for catalytic cracking
units. The coke obtained is usually used as fuel, but specialty uses, such as elec-
trode manufacture and the production of chemicals and metallurgical coke, are
also possible, thus increasing the value of the coke.
Delayed coking (Fig. 3) is a semicontinuous process in which the heated
charge is transferred to large soaking, or coking, drums, which provide the resi-
dence time needed for the cracking reactions to proceed to completion (11,12).
The feed to these units is normally a vacuum residuum, although residua from
302 PETROLEUM REFINERY PROCESSES Vol. 1

Gas

Gasoline (naphtha)

Gas oil

Fractionator

Crude
Operative Nonoperative
oil
coking drum coking drum
(soaker) (soaker)

Coke

Heater

Fig. 3. The delayed coking process.

other thermal processes are also used. The feedstock is introduced into the pro-
duct fractionator. The fractionator bottoms, including a recycle stream of heavy
product, are heated in a furnace whose outlet temperature varies from 480 to
5158C (896–9598F). The heated feedstock then enters one of a pair of coking
drums where the cracking reactions continue.
The cracked products leave as overhead materials, and coke deposits form
on the inner surface of the drum. To provide continuous operation, two drums are
used; while one drum is on-stream, the one off-stream is being cleaned, steamed,
water-cooled, and decoked in the same time interval. The temperature in the
coke drum is in the range of 415–4508C (779–8428F) with pressures in the
range of 103–621 kPa (15–90 psi). Overhead products go to the fractionator,
where naphtha and heating oil fractions are recovered. The nonvolatile material
is combined with preheated fresh feed and returned to the furnace. The coke
drum is usually on stream for 24 h before becoming filled with porous coke,
after which the coke is removed hydraulically.
Fluid coking (Fig. 4) is a continuous process that uses the fluidized solids
technique to convert atmospheric and vacuum residua to more valuable products
(12,13). The residuum is converted to coke and overhead products by being
Vol. 1 PETROLEUM REFINERY PROCESSES 303

Overhead
products Flue gas

Reactor Burner

Heavy
feedstock Coke

Steam Makeup Air


coke

Excess coke

Fig. 4. Fluid coking.

sprayed into a fluidized bed of hot, fine coke particles, which permits the coking
reactions to be conducted at higher temperatures and shorter contact times than
they can be in delayed coking. Moreover, these conditions result in decreased
yields of coke; greater quantities of more valuable liquid product are recovered
in the fluid coking process.
Fluid coking uses two vessels: a reactor and a burner; coke particles are cir-
culated between the two to transfer heat generated by burning a portion of the
coke to the reactor. The reactor holds a bed of fluidized coke particles, and steam
is introduced at the bottom of the reactor to fluidize the bed.
Flexicoking (Fig. 5), also a continuous process, uses the same configuration
as the fluid coker, but has a gasification section in which excess coke can be gasi-
fied to produce refinery fuel gas. Flexicoking is a process by which excess coke-
make is reduced in view of the gradual incursion of the heavier feedstocks into
refinery operations. Such feedstocks are notorious for producing high yields of
coke (>15% by weight) in thermal and catalytic operations.
4.4. Catalytic Cracking. Fluid catalytic cracking (FCC) (Fig. 6) was
first introduced in 1942 and uses a fluidized bed of catalyst with continuous feed-
stock flow. In the process, which has progressively supplanted thermal cracking,
the thermal decomposition of petroleum constituents occurs in the presence of a
catalyst (14,15). The acid catalysts first used in catalytic cracking were desig-
nated low alumina catalysts; amorphous solids composed of 87% silica, SiO2,
and 13% alumina, Al2O3. Later, high alumina catalysts containing 25% alumina
and 75% silica were used. However, this type of catalyst has largely been
replaced by catalysts containing crystalline aluminosilicates (zeolites) or molecu-
lar sieves (16–18).
The catalyst is employed in bead, pellet, or microspherical form and can be
used as a fixed, moving, or fluid bed. The fixed-bed process was the first process
used commercially and employs a static bed of catalyst in several reactors, which
allows a continuous flow of feedstock to be maintained. The cycle of operations
consists of (1) the flow of feedstock through the catalyst bed; (2) the discontinuance
Reactor products Coke gas to
to fractionator Steam sulfur removal
generation

Fines
Scrubber removal
Cooling

Recycle

Bitumen Coke fines

Heater

Steam Reactor

304
Gasifier

Purge
coke

Air

Steam

Fig. 5. The flexicoking process.


Flue gas
(to final dust
collection)
Regenerator
Fractionator

Cold
water
Reactor
Gas to recovery
Waste Stripping steam
heat
Water
boiler Spent catalyst
Gasoline

305
Regenerator Air Light gas oil
catalyst

Fresh feed Slurry


settler
Air blower

Heavy gas oil


Recyle Wash
oil

Slurry decant oil

Fig. 6. Fluid-bed catalytic cracking with product separation.


306 PETROLEUM REFINERY PROCESSES Vol. 1

of feedstock flow and removal of coke from the catalyst by burning; and (3) the
insertion of the reactor back on-stream. The moving-bed process uses a reaction
vessel, in which cracking takes place, and a kiln, in which the spent catalyst is
regenerated and catalyst movement between the vessels is provided by various
means. The fluid-bed process differs from the fixed-bed and moving-bed
processes, insofar as the powdered catalyst is circulated essentially as a fluid
with the feedstock. The several fluid catalytic cracking processes in use differ pri-
marily in mechanical design. Side-by-side reactor-regenerator construction and
unitary vessel construction (the reactor either above or below the regenerator)
are the two main mechanical variations.
Catalytic cracking has progressively supplanted thermal cracking as the
most advantageous means of converting distillate oils into gasoline and a better
yield of higher octane gasoline can be obtained than by any known thermal
operation. At the same time the gas produced consists mostly of propane and
butane with less methane and ethane. The production of heavy oils and tars,
higher in molecular weight than the charge material, is also minimized, and
both the gasoline and the uncracked cycle oil are more saturated than the pro-
ducts of thermal cracking.
4.5. Hydroprocessing. Hydroprocesses use the principle that the pre-
sence of hydrogen during a thermal reaction of a petroleum feedstock will termi-
nate many of the coke-forming reactions and enhance the yields of the lower
boiling components such as gasoline, kerosene, and jet fuel. Thus, in this
group of refining processes, which includes hydrotreating and hydrocracking,
the feedstock is heated with hydrogen at high temperature and under pressure.
The outcome is the conversion of a variety of feedstocks to a range of products
(Table 2) (1,2,4,9,19).
The purpose of hydroprocessing is (1) to improve existing petroleum pro-
ducts or develop new products or uses; (2) to convert inferior or low grade mate-
rials into valuable products; and (3) to transform near-solid residua to liquid
fuels. Products are as follows: from naphtha, reformed feedstock and liquefied
petroleum gas (LPG); from atmospheric gas oil, diesel and jet fuel, petrochemical
feedstock, and naphtha; from vacuum gas oil, catalytic cracker feedstock, kero-
sene, diesel and jet fuel, naphtha, LPG, and lubricating oil; and from residuum,
catalytic cracker and coker feedstock, diesel fuel, etc.
Hydroprocesses for the conversion of petroleum and petroleum products can
be classified as destructive or nondestructive. The former (hydrogenolysis and

Table 2. General Process Characteristics for Hydroprocessing Various Feedstocks


Atmospheric gas Vacuum gas
Process Naphtha oil oil Residuum
hydrocracking þ þ þ þ
aromatics removal þ þ
sulfur removal þ þ þ
nitrogen removal þ þ þ
metals removal þ þ
coke mitigation þ
n-paraffins removal þ
olefins removal þ
Vol. 1 PETROLEUM REFINERY PROCESSES 307

Table 3. Sulfur Removal from Organic Compounds by Hydrotreating


Name Typical reaction
thiols (mercaptans) R SH þ H2 ! RH þ H2 S
disulfides R S S R0 þ 3 H2 ! RH þ R0 H þ 2 H2 S
sulfides R S R0 þ 2 H2 ! RH þ R0 H þ H2 S

thiophenes + 4 H2 n-C4H10 + H2S


S
CH2CH3
benzothiophenes + 3 H2 + H2S
S

dibenzothiophenes + 2 H2 + H2S
S

hydrocracking) is characterized by the rupture of carbon–carbon bonds and is


accompanied by hydrogen saturation of the fragments to produce lower boiling
products. Such treatment requires rather high temperatures and high hydrogen
pressures, the latter to minimize coke formation.
Hydrogenolysis is analogous to hydrolysis and ammonolysis, which involve
the cleavage of a bond induced by the action of water and ammonia, respectively.
Chemical bonds that are broken by hydrogenolysis reactions include carbon–car-
bon, carbon–oxygen, carbon–sulfur, and carbon–nitrogen bonds. An example of
hydrogenolysis is the hydrodealkylation of toluene to form benzene and methane:

C6 H5 CH3 þ H2 ! C6 H6 þ CH4

On the other hand, nondestructive, or simple hydrogenation is generally


used for the purpose of improving product quality without appreciable alteration
of the boiling range. Examples are the removal of various sulfur compounds
(Table 3), which would otherwise have an adverse effect on product quality.
Treatment under such mild conditions is often referred to as hydrotreating or
hydrofining and is essentially a means of eliminating, in addition to sulfur, nitro-
gen and oxygen as ammonia and water, respectively.
Hydrocracking. Hydrocracking is a catalytic process (>3508C, 6628F) in
which hydrogenation accompanies cracking (20). Relatively high pressures,
6900–21,000 kPa (1000–3000 psi), are employed and the overall result is the con-
version of the feedstock to lower boiling products. Another attractive feature of
hydrocracking is the low yield of gaseous components, such as methane, ethane,
and propane, which are less desirable than the gasoline components. Essentially
all the initial reactions of catalytic cracking occur, but some of the secondary
reactions are inhibited or stopped by the presence of hydrogen.
In the first, pretreating stage of a hydrocracking process, the main pur-
pose is to convert organic nitrogen compounds and organic sulfur in the feed-
stock to hydrocarbons and to ammonia and hydrogen sulfide by hydrogenation
308 PETROLEUM REFINERY PROCESSES Vol. 1

and mild hydrocracking. The purpose is to reduce the organic nitrogen and
sulfur compounds to low levels (<50 ppm). Typical conditions are 340–3908C
(644–7348F), 10,350–17,250 kPa (1500–2500 psi), and a catalyst contact time
of 0.5–1.5 h, up to 1.5 wt% hydrogen is absorbed, partly by conversion of the
nitrogen compounds, but chiefly by aromatic compounds that are hydrogenated.
This first stage is usually carried out with a bifunctional catalyst containing
hydrogenation promoters, eg, nickel and tungsten or molybdenum sulfides, on an
acidic support. The metal sulfides hydrogenate aromatics and nitrogen com-
pounds, and retard deposition of carbonaceous deposits; the acidic support accel-
erates nitrogen removal as ammonia by breaking carbon–nitrogen bonds. The
catalyst is generally used as 0.32  0.32-cm or 0.16  0.32-cm pellets, as well as
spheres or other shapes.
Most of the hydrocracking is accomplished in the second stage. Hydrogen
sulfide, ammonia, and low boiling products are usually removed from the first-
stage product; the remaining oil, which is low in nitrogen–sulfur compounds,
is passed over the second-stage catalyst. Some catalyst systems do not require
the removal of adsorbed compounds between stages. In the second stage, typical
conditions are 300–3708C (572–6988F), 7000–17,250 kPa (1000–2500 psi) hydro-
gen pressure, and 0.5–1.5 h contact time; 1–1.5 wt% hydrogen may be absorbed.
Conversion to gasoline or jet fuel is seldom complete in one contact with the cat-
alyst, so the lighter oils are removed by distillation of the products, and the hea-
vier, high boiling product is combined with fresh feed and recycled over the
catalyst until it is completely converted.
The catalyst for the second stage is also a bifunctional catalyst containing
hydrogenating and acidic components. Metals such as nickel, molybdenum, tung-
sten, or palladium are used in various combinations and dispersed on solid acidic
supports, such as synthetic amorphous or crystalline silica–alumina, eg, zeolites.
These supports contain strongly acidic sites and sometimes are enhanced by
the incorporation of a small amount of fluorine.
A long period of operation (eg, >3 years) between catalyst regeneration is
desirable; this is achieved by keeping a low nitrogen content in the feed and
avoiding high temperatures and high end-point feedstock, which leads to excess
cracking and consequent deposition of coke on the catalyst. Feedstock conversion
is the key insofar as the conversion dictates the temperature employed. When
activity of the catalyst has decreased, it can often be restored by controlled burn-
ing of the coke.
Hydrotreating. On the other hand, hydrotreating is a catalytic process
converts sulfur- and/or nitrogen-containing hydrocarbons into low sulfur low
nitrogen liquids, hydrogen sulfide, and ammonia (21). A wide variety of metals
are active hydrogenation catalysts; those of most interest are nickel, palladium,
platinum, cobalt, and iron. Special preparations of the first three are active at
room temperature and atmospheric pressure. The metallic catalysts are easily
poisoned by sulfur- or arsenic-containing compounds, and even by other metals.
To avoid such poisoning, less effective, but more resistant metal oxides or sulfides
are frequently employed, generally those of tungsten, cobalt, or molybdenum.
Alternatively, catalyst poisoning can be minimized by mild hydrogenation to
remove nitrogen, oxygen, and sulfur from feedstocks in the presence of more
resistant catalysts, such as cobalt–molybdenum–alumina, Co–Mo–Al2O3.
Vol. 1 PETROLEUM REFINERY PROCESSES 309

The process temperature affects the rate and the extent of hydrogenation as
it does any chemical reaction. Practically every hydrogenation reaction can be
reversed by increasing temperature. If a second functional group is present,
high temperatures often lead to the loss of selectivity and, therefore, loss of
desired product yield. As a practical measure, hydrogenation is carried out at
as low a temperature as possible which is still compatible with a satisfactory
reaction rate.
Hydrotreating is carried out by charging the feed to the reactor together
with hydrogen at 300–3458C (572–6538F); the hydrogen pressures are 3450–
6900 kPa (500–1000 psi). The reaction generally takes place in the vapor phase
but, depending on the application, can also be a mixed-phase reaction.
After passing through the reactor, the treated oil is cooled and separated
from the excess hydrogen recycled through the reactor. The treated oil is pumped
to a stripper tower where hydrogen sulfide, formed by the hydrogenation reac-
tion, is removed by steam or by hydrocarbon vapor via reboiling, and the finished
product leaves the bottom of the stripper tower. The catalyst can be regenerated
in situ and ultimately be replaced after several regenerations.
4.6. Reforming. Reforming processes are used to change the inherent
chemical structures of the hydrocarbons that exist in distillation fractions
crude oil into different compounds. Catalytic reforming is one of the most impor-
tant processes in a modern refinery, altering straight run fraction or fractions
from a catalytic cracker into new compounds through a combination of heat
and pressure in the presence of a catalyst. Reforming processes are particularly
important in producing high quality gasoline fuels. Reforming processes are clas-
sified as continuous, cyclic, or semiregenerative, depending upon the frequency
of catalyst regeneration.
When the demand for higher octane gasolines increased during the early
1930s, attention was directed to ways and means of improving the octane number
of fractions within the boiling range of gasoline. Straight-run (distilled) gasolines
frequently had low octane numbers, and any process that could improve the octane
numbers would aid in meeting the demand for higher octane number gasoline.
Such a process, called thermal reforming, was developed and used widely, but to
a much lesser extent than thermal cracking. Thermal reforming was a natural
development from older thermal cracking processes; cracking converts heavier
oils into gasoline whereas reforming converts (reforms) gasolines into higher
octane gasolines. The equipment for thermal reforming is essentially the same
as for thermal cracking, but higher temperatures are used in the former.
In carrying out thermal reforming, a feedstock such as 2058C (4008F)
end-point naphtha or a straight-run gasoline is heated to 510–5958C (950–
11008F) in a furnace, much the same as a cracking furnace, with pressures
from 400 to 1000 psi (27 to 68 atm). As the heated naphtha leaves the furnace,
it is cooled or quenched by the addition of cold naphtha. The material then enters
a fractional distillation tower where any heavy products are separated. The
remainder of the reformed material leaves the top of the tower to be separated
into gases and reformate. The higher octane of the reformate is due primarily to
the cracking of longer-chain paraffins into higher octane olefins. The products of
thermal reforming are gases, gasoline, and residual oil or tar, the latter being
formed in very small amounts (1%). The amount and quality of the gasoline,
310 PETROLEUM REFINERY PROCESSES Vol. 1

known as reformate, is very dependent on the temperature. A general rule is:


The higher the reforming temperature, the higher the octane number, but the
lower the yield of reformate.
Catalytic reforming usually is carried out by feeding a naphtha (after pre-
treating with hydrogen if necessary to remove nitrogen and sulfur compounds)
and hydrogen mixture to a furnace where the mixture is heated to the desired
temperature (450–5208C; 842–9688F) and then passed through fixed-bed cata-
lytic reactors at hydrogen pressures of 350–2700 kPa (50–400 psi). Normally,
several reactors are used in series and heaters are located between adjoining
reactors in order to compensate for the endothermic reactions taking place.
Thermal reforming, less effective and less economical than catalytic pro-
cesses, has been largely supplanted. Like thermal reforming, catalytic reforming
converts low octane gasolines into high octane gasolines, ie, reformate. Whereas
thermal reforming produces reformate having research octane numbers in the
65–80 range, depending on the yield, catalytic reforming produces reformate
having octane numbers on the order of 90–105. Catalytic reforming is conducted
in the presence of hydrogen over hydrogenation–dehydrogenation catalysts, eg,
in the platforming process (22). Catalytic reformer feeds are saturated, ie, not
olefinic, materials. Catalytic cracker naphtha and hydrocracker naphtha that
contains substantial quantities of naphthenes are also suitable reformer feed-
stocks.
Dehydrogenation is a main chemical reaction in catalytic reforming, and
hydrogen gas is consequently produced in large quantities. Hydrogen is recycled
through the reactors where the reforming takes place to provide the atmosphere
necessary for the chemical reactions, and also prevents carbon from being depos-
ited on the catalyst, thus extending its operating life. Because of an excess of
hydrogen above, whatever is consumed in the process is produced, catalytic
reforming processes are unique in that they are the only petroleum refinery pro-
cesses to produce hydrogen as a by-product.
The composition of a reforming catalyst is dictated by the composition of
the feedstock and the desired reformate. The catalysts used are principally
platinum or platinum–rhenium on an alumina base. The purpose of platinum
on the catalyst is to promote dehydrogenation and hydrogenation reactions.
Nonplatinum catalysts are used in regenerative processes for feedstocks contain-
ing sulfur, although pretreatment (hydrodesulfurization) may permit platinum
catalysts to be employed.
4.7. Isomerization. Isomerization is used with the objective of providing
additional feedstock for alkylation units (isobutane) or high octane fractions for
gasoline blending (pentane and hexane) (23). The latter application is useful in
the production of reformulated gasoline by increasing the octane number while
converting or removing benzene (24,25).
Initially, aluminum chloride was the catalyst used to isomerize butane,
pentane, and hexane. Since then, supported metal catalysts have been developed
for use in high temperature processes that operate at 370–4808C (698–8968F)
and 2070–5170 kPa (300–750 psi), whereas aluminum chloride and hydrogen
chloride are universally used for the low temperature processes.
Nonregenerable aluminum chloride catalyst is employed with various
carriers in a fixed-bed or liquid contactor. Platinum or other metal catalyst
Vol. 1 PETROLEUM REFINERY PROCESSES 311

processes that utilize fixed-bed operation can be either regenerable or nonregen-


erable. The reaction conditions vary widely, between 40 and 4808C (104–8968F)
and 1035–6900 kPa (150–1000 psi), depending on the particular process and
feedstock.
4.8. Alkylation. The combination of olefins with paraffins to form higher
isoparaffins is termed alkylation. Alkylate is a desirable blendstock because it
has a relatively high octane number and serves to dilute the total aromatics con-
tent. Reduction of the olefins in gasoline blendstocks by alkylation also reduces
tail pipe emissions. In refinery practice, butylenes are routinely alkylated by
reaction with isobutane to produce isobutane–octane (26). In some plants, propy-
lene and/or pentylenes (amylenes) are also alkylated (27).
Alkylate is composed of a mixture of isoparaffins whose octane numbers
vary with the olefins from which they were made. Butylenes produce the highest
octane numbers, propylene the lowest, and amylenes (pentylenes) the intermedi-
ate values. All alkylates, however, have high (>87) octane numbers that make
them particularly valuable.
Propylene, butylenes, or amylenes are combined with isobutane in the pre-
sence of an acid catalyst, eg, sulfuric acid or hydrofluoric acid, at low tempera-
tures (1–408C; 33.8–104) and pressures, 102–1035 kPa (1 to 10 atm). Sulfuric
acid or hydrogen fluoride are the catalysts used commercially in refineries.
The acid is pumped through the reactor and forms an emulsion with reactants,
and the emulsion is maintained at 50% acid. The rate of deactivation varies with
the feed and isobutane charge rate. Butene feeds cause less acid consumption
than the propylene feeds.
4.9. Polymerization. In the petroleum industry, polymerization is the
process by which olefin gases are converted to higher molecular weight liquid
products that may be suitable for gasoline (polymer gasoline) or other liquid
fuels.
The feedstock, usually consisting of propylene and butylenes (various iso-
mers of C4H8) from cracking processes, may even consist of selective olefins for
dimer, trimer, or tetramer production:
RCH CH2 monomer

[CH(R) CH2)2 dimer

[CH(R) CH2)3 trimer

[CH(R) CH2)4 tetramer

The molecular size of the product is limited insofar as the reaction is termi-
nated at the dimer or trimer stage. Thus the process is more properly termed oli-
gomerization. The 4- to 12-carbon compounds required as the constituents of
liquid fuels are the prime products.
Thermal polymerization is not as effective as catalytic polymerization, but
has the advantage that it can be used to polymerize saturated materials that can-
not be induced to react by catalysts. The process consists of the vapor-phase
cracking of, eg, propane and butane, followed by prolonged periods at high
temperature (510–5958C; 950–11038F) for the reactions to proceed to near com-
pletion. Olefins can also be conveniently polymerized by means of an acid
312 PETROLEUM REFINERY PROCESSES Vol. 1

catalyst. Thus, the treated olefin-rich feed stream is contacted with a catalyst,
such as sulfuric acid, copper pyrophosphate, or phosphoric acid, at 150–2208C
(302–4288F) and 1035–8275 kPa (150–1200 psi), depending on feedstock and
product requirement.
Phosphates are the principal catalysts used in polymerization units; the
commercially used catalysts are liquid phosphoric acid, phosphoric acid on kie-
selguhr, copper pyrophosphate pellets, and phosphoric acid film on quartz. The
last is the least active and has the disadvantage that carbonaceous deposits must
occasionally be burned off the support. Compared to other processes, the one
using liquid phosphoric acid catalyst is far more responsive to attempts to
raise production by increasing temperature.

5. Treating

Since the original crude oils contain some sulfur compounds, the resulting
products and gasolines also contain sulfur compounds, including hydrogen sul-
fide, mercaptans, sulfides, disulfides, and thiophenes. The processes used to
sweeten, ie, desulfurize, the products depend on the type and amount of the sul-
fur compounds present and the specifications of the finished gasoline or other
stocks.
Hydrotreating is the most widely practiced treating process for all types of
petroleum products. However, there are other treating processes suitable for the
removal of mercaptans and hydrogen sulfide; such processes are necessary and
are performed as part of the product improvement and finishing procedures. For
example, mercaptan, RSH, removal is achieved by using regenerative solution
processes, in which the treatment solutions are regenerated rather than dis-
carded. Mercaptan conversion is essentially a process of oxidation to disulfides,
RSSR, by lead sulfide treatment, copper chloride–oxygen treatment, sodium
hypochlorite treatment, or oxygen treatment with a chelated cobalt catalyst in
either a caustic solution or fixed bed.
Hydrogen sulfide, H2S, is removed by a variety of processes, of which one is
a regenerative solution process using aqueous solutions of sodium hydroxide,
NaOH; calcium hydroxide, Ca(OH)2; sodium phosphate, Na3PO4; and sodium
carbonate, Na2CO3.
5.1. Alkali Treatment. Caustic washing is the treatment of materials,
usually products from petroleum refining, with solutions of caustic soda. The
process consists of mixing a water solution of lye (sodium hydroxide or caustic
soda) with a petroleum fraction. The treatment is carried out as soon as possible
after the petroleum fraction is distilled, since contact with air forms free sulfur,
which is corrosive and difficult to remove. The lye reacts either with any hydro-
gen sulfide present to form sodium sulfide, which is soluble in water, or with mer-
captans, followed by oxidation, to form the less nocuous disulfides.

H2 S þ 2 NaOH ! Na2 S þ 2 H2 O
RSH þ NaOH ! NaSR þ H2 O
4 NaSR þ O2 þ 2 H2 O ! 2 RSSR þ 4 NaOH
Vol. 1 PETROLEUM REFINERY PROCESSES 313

Nonregenerative caustic treatment is generally economically applied when


the contaminating materials are low in concentration and waste disposal is not a
problem. However, the use of nonregenerative systems is on the decline because
of the frequently occurring waste disposal problems that arise from environmen-
tal considerations and because of the availability of numerous other processes
that can effect more complete removal of contaminating materials.
Steam-regenerative caustic treatment is directed toward the removal of
mercaptans from such products as gasoline and low boiling solvents (naphtha).
The caustic is regenerated by steam blowing in a stripping tower. The nature and
concentration of the mercaptans to be removed dictate the quantity and tempera-
ture of the process. However, the caustic gradually deteriorates because of the
accumulation of material that cannot be removed by stripping; the caustic qua-
lity must be maintained by either continuous or intermittent discard and/or
replacement of a minimum amount of the operating solution.
5.2. Acid Treatment. The treatment of petroleum products with acids
has been in use for a considerable time in the petroleum industry. Various
acids, such as hydrofluoric acid, hydrochloric acid, nitric acid, and phosphoric
acid, have been used in addition to the most commonly used sulfuric acid, but
in most instances there is little advantage in using any acid other than sulfuric.
Sulfuric acid also has been employed for refining kerosene distillates and
lubricating oil stocks. Although a greater part of the acid-treating processes
has been superseded by other processes, acid treating has continued to some
extent; it is used for desulfurizing high boiling fractions of cracked gasoline dis-
tillates, for refining paraffinic kerosene, for manufacturing low cost lubricating
oils, and for making specialty products such as insecticides, pharmaceutical oils,
and insulating oils.
5.3. Clay Treatment. The original method of clay treating was to perco-
late a petroleum fraction through a tower containing coarse clay pellets. As the
clay adsorbed impurities from the petroleum fraction, the clay became less effec-
tive. The activity of the clay was periodically restored by removing it from the
tower and burning the adsorbed material under carefully controlled conditions
so as not to sinter the clay. The percolation method of clay treating was widely
used for lubricating oils, but has been largely replaced by clay contacting.
However, this use of clay treating has been superseded by other processes;
in particular, by the use of inhibitors, which, when added in small amounts to
gasoline, can prevent gums from forming. Nevertheless, clay treating is still
used as a finishing step in the manufacture of lubricating oils and waxes.
The clay removes traces of asphaltic materials and other compounds that give
oils and waxes unwanted odors and colors.
5.4. Solvent Treatment. Solvent processes can be divided into two
main categories: solvent extraction and solvent dewaxing. The solvent used in
the extraction processes include propane and cresylic acid, 2,20 -dichlorodiethyl
ether, phenol, furfural, sulfur dioxide, benzene, and nitrobenzene. In the
dewaxing process (28), the principal solvents are benzene, methyl ethyl ketone,
methyl isobutyl ketone, propane, petroleum naphtha, ethylene dichloride,
methylene chloride, sulfur dioxide, and N-methylpyrrolidinone.
The early developments of solvent processing were concerned with the
lubricating oil end of the crude. Solvent extraction processes are applied to
314 PETROLEUM REFINERY PROCESSES Vol. 1

many useful separations in the purification of gasoline, kerosene, diesel fuel, and
other oils.
In solvent dewaxing, the oil is diluted with a solvent that has a high affinity
for oil, chilled to precipitate the wax, filtered to remove the wax, stripped of
solvent, and dried. The solvents [principally propane, naphtha, methyl ethyl
ketone (MEK)] act as diluents for the high molecular weight oil fractions to
reduce the viscosity of the mixture and provide sufficient liquid volume to permit
pumping and filtering. Wax, produced by the solvent dewaxing process, is used to
make (1) paraffins for candle wax, (2) microwax for cosmetics, and (3) wax for
petroleum jelly.
Catalytic dewaxing is a process in which the chemical composition of the
feed is changed. The process involved catalytic cracking of long paraffin chains
into shorter chains to remove the wax and produced lower molecular weigh pro-
ducts suitable for other uses. As an example, the feedstock is contacted with
hydrogen at elevated temperature and pressure over a catalyst (such as a zeolite)
that selectively cracks the normal paraffins to methane, ethane, and propane.
This process also knows as hydrodewaxing. There are two types of catalytic
dewaxing: (1) single-catalyst process that is used for pour point reduction and
to improve the oxygen stability of the product, and (2) a two-catalyst process
that uses a fixed-bed reactor and essentially no methane or ethane is formed
in the reaction.
In addition, solvent extraction can replace fractionation in many separation
processes in the refinery. For example, propane deasphalting (Fig. 7) has

Propane
recovery
and recycle

Stripper

Heavy
feedstock Deasphalted
oil

Evaporator
Stripper
Propane
(liquid)
Flasher

Asphalt

Fig. 7. Propane deasphalting.


Vol. 1 PETROLEUM REFINERY PROCESSES 315

replaced, to some extent, vacuum distillation as a means of removing asphalt


from reduced crude oils.

6. Gas Processing

The gas streams produced during petroleum refining usually contain many nox-
ious constituents that have an adverse effect on the use of the gas for other pur-
poses, eg, as a fuel or as a petrochemical feedstock, and some degree of cleaning is
required (29).
Gas purification processes fall into three categories: the removal of gaseous
impurities, the removal of particulate impurities, and ultrafine cleaning. The
extra expense of the last process is only justified by the nature of the subsequent
operations or the need to produce a pure gas stream. Because there are many
variables in gas treating, several factors must be considered (1) the types and
concentrations of contaminants in the gas; (2) the degree of contaminant removal
desired; (3) the selectivity of acid gas removal required; (4) the temperature,
pressure, volume, and composition of the gas to be processed; (5) the carbon
dioxide/hydrogen sulfide ratio in the gas; and (6) the desirability of sulfur recov-
ery on account of process economics or environmental issues.
Process selectivity indicates the preference with which the process removes
one acid gas component relative to or in preference to another. For example,
some processes remove both hydrogen sulfide and carbon dioxide, whereas
other processes are designed to remove hydrogen sulfide only. Thus it is impor-
tant to consider the process selectivity for hydrogen sulfide removal compared to
carbon dioxide removal, ie, the carbon dioxide/hydrogen sulfide ratio in the nat-
ural gas, in order to ensure minimal concentrations of these components in the
product.
One of the principal aspects of refinery gas cleanup is the removal of acid
gas constituents, ie, carbon dioxide, CO2, and hydrogen sulfide, H2S. Treatment
of natural gas to remove the acid gas constituents is most often accomplished by
contacting the natural gas with an alkaline solution. The most commonly used
treating solutions are aqueous solutions of the ethanolamines or alkali carbo-
nates. There are several hydrogen sulfide removal processes, most of which are
followed by a Claus plant that produces elemental sulfur from the hydrogen
sulfide (29).

7. Petroleum Products

The constant demand for products, such as liquid fuels, is the main driving force
behind the petroleum industry (5,30). In fact, it is the changes in product
demand that have been largely responsible for the evolution of the industry.
In general, when the product is a fraction that has been produced from cov-
nentioal petroleum or from heavy oil or resid and includes a large number of indi-
vidual hydrocarbons, the fraction is classified as a refined product. Examples of
refined products are gasoline, diesel fuel, heating oils, lubricants, waxes, asphalt,
and coke. In contrast, when the product is limited to, perhaps, one or two specific
316 PETROLEUM REFINERY PROCESSES Vol. 1

Table 4. Constituents of Light Ends


Carbon Molecular Boiling
Hydrocarbon atoms weight point, 8C Uses
methane 1 16 182 fuel gas
ethane 2 30 89 fuel gas
ethylene 2 28 104 fuel gas, petrochemicals
propane 3 44 42 fuel gas, LPG
propylene 3 42 48 fuel gas, petrochemicals, poly-
mer gasoline
isobutane 4 58 12 alkylate, motor gasoline
n-butane 4 58 1 motor gasoline
isobutylene 4 56 7 synthetic rubber and chemi-
cals, polymer gasoline, alky-
late, motor gasoline
butylene-1a 4 56 6 synthetic rubber and chemi-
cals,
butylene-2a 4 56 1 alkylate, polymer
gasoline, motor
gasoline
isopentane 5 72 28 motor and aviation gasolines
n-pentane 5 72 36 motor and aviation gasolines
pentylenes 5 70 30 motor gasolines
isohexane 6 86 61 motor and aviation gasolines
n-hexane 6 86 69 motor and aviation gasolines
a
Numbers refer to the positions of the double bond; eg, butylene-1 (or butene-1 or but-1-ene)
is CH3CH2CH CH2 and butylene-2 (or butene-2 or but-2-ene) is CH3CH CHCH3.

hydrocarbons of high purity, the fraction is classified as a petrochemical product.


Examples of petrochemicals are ethylene (CH – CH2), propylene (CH3CH – CH2),
benzene (C6H6), toluene (C6H5CH3), and xylene (CH3C6H4CH3).
7.1. Liquefied Petroleum Gas. Certain specific hydrocarbons, such as
propane, butane, pentane, and their mixtures, exist in the gaseous state under
atmospheric ambient conditions, but can be converted to the liquid state under
conditions of moderate pressure at ambient temperature. This is termed LPG.
Liquefied petroleum gas is a refinery product and the individual constituents,
or light ends (Table 4), are produced during a variety of refining operations.
The presence of ethane in LPG must be avoided because of the inability of
this lighter hydrocarbon to liquefy under pressure at ambient temperature and
its tendency to register abnormally high pressures in the LPG containers. On the
other hand, the presence of pentane in liquefied petroleum gas must also be
avoided because this particular hydrocarbon, a liquid at ambient temperatures
and pressures, can condense and separate in the liquid state in the gas lines.
Liquefied petroleum gases precipitate asphaltic and resinous materials
from crude residues while the lubricating oil constituents remain in solution.
Although all liquefied gases possess this property to some extent, propane and
butane are used to deasphalt residual lubricating oils because of their relative
low cost and their ease of separation from lubricating oils.
Fuel gas or refinery gas (that also contains low boiling hydrocarbons) is pro-
duced in considerable quantities during the different refining processes and is
Vol. 1 PETROLEUM REFINERY PROCESSES 317

used as fuel for the refinery itself and as an important feedstock for the petro-
chemical industry.
7.2. Gasoline. Gasoline (also known as petrol in many parts of the world)
is one of the more important refinery products. The final gasoline product as a
transport fuel is a carefully blended mixture having a predetermined composition
that is adequate to power requirements (octane number) and climatic conditions.
Octane number is a measure of a fuel’s ability to avoid knocking. The octane
number of a gasoline is determined in a special single-cylinder engine where various
combustion conditions can be controlled. The test engine is adjusted to give trace
knock from the fuel to be rated. Various mixtures of isooctane (2,2,4-trimethyl
pentane) and normal heptane are then used to find the ratio of the two reference
fuels that produce the same intensity of knock as that by the unknown fuel.
By defining isooctane as having an octane number of 100 and n-heptane
as having an octane number of 0, the volumetric percentage of isooctane in hep-
tane that matches the knock from the unknown fuel can be calculated as the
octane number of the fuel. For example, 90 vol% isooctane and 10 vol% normal
heptane produce a 90-octane-number reference fuel.
Two kinds of octane number ratings are specified, although other methods
are often used for engine and fuel development. Both methods use the same
reference fuels and essentially the same test engine, but engine operating condi-
tions are different. In one test, called the research method (hence the Research
Octane Number, RON), the spark advance is fixed, the air inlet temperature is
1258F (528C), and engine speed is 600 rpm. The other, called the motor method
(hence the Motor Octane Number, MON), uses variable spark timing, a higher
mixture temperature of 3008F (1498C), and a faster engine speed of 900 rpm.
The more severe conditions of the motor method have a greater influence on
commercial blends than they do on the reference fuels. Thus, the motor octane
number of a commercial blend tends to be lower than the research octane num-
ber. Common practice is to label gasoline with an arithmetic average of both rat-
ings, abbreviated as (RþM)/2, and often referred to as road octane number.
The naphtha fraction from crude oil distillation is ultimately used to make
gasoline. The two streams are isolated early in the refining scheme so that each
can be refined separately for optimum blending in order to achieve the required
specifications, including the octane number.
Gasoline is a complex mixture of hydrocarbons that boils <2008C (3928F).
The hydrocarbon constituents in this boiling range are those that have 4–12
carbon atoms in their molecular structure. Gasolines can vary widely in compo-
sition, even those having the same octane number can be quite different (30).
Because of the differences in composition of the various gasolines, gasoline blend-
ing is necessary. The physical process of blending the components is simple, but
determination of how much of each component to include in a blend is much more
difficult. The operation is carried out by simultaneously pumping all the compo-
nents of a gasoline blend into a pipeline that leads to the gasoline storage, and
the pumps must be set to deliver automatically the proper proportion of each
component. Sophisticated instrumentation is employed to achieve the desired
blends.
The objective of gasoline (in fact, product) blending is to allocate the
available blending components in such a way as to meet product demands and
318 PETROLEUM REFINERY PROCESSES Vol. 1

specifications. In the blending process, product streams from other units are col-
lected and blended to produce the desired product. For example, in gasoline
blending, product streams from hydrotreating units, reforming units, polymeri-
zation units and alkylation units are blended to produce specification gasoline.
Aviation gasoline, usually used in light aircraft and older civil aircraft, have
narrower boiling ranges (38–1708C; 100.4–3388F) than automobile gasoline
(0–2008C; 32–3928F). In winter, the extreme temperatures in the northern
climates allow a certain amount of butane to be dissolved in the gasoline to faci-
litate vaporization. The narrower boiling range ensures better distribution of the
vaporized fuel through the more complicated induction systems of aircraft
engines. Aircraft operates at altitudes where the prevailing pressure is less
than the pressure at the surface of the earth, eg, at 5334 m, 52 kPa (7.5 psi) com-
pared to 101.3 kPa (14.7 psi). As a result, the vapor pressure of aviation gaso-
lines, which is a function of the fuel’s boiling range, must be limited to reduce
boiling in the tanks, fuel lines, and carburetors.
Another condition to keep gasoline engines running smoothly is to allow the
fuel–air mixture to start burning at a precise time in the combustion cycle. An elec-
trical spark starts the ignition. The remainder of the fuel–air mixture should be con-
sumed by a flame front moving out from the initial spark. Under certain conditions,
a portion of the fuel–air mixture ignites spontaneously instead of waiting for the
flame front from the carefully timed spark. The extra pressure pulses resulting
from spontaneous combustion are usually audible above the normal sounds of a run-
ning engine and give rise to the phenomenon called knock, of which pinging and
rumbling are special attributes. However, knocks are undesirable because they
waste some of the available power of an otherwise smooth-running engine.
Reformulated gasoline is believed to be the answer to many environmental
issues that arise from the use of automobiles and there has been a serious effort
to produce reformulated gasoline components (Table 5) from a variety of
processes (25).
However, reformulation may increase gasoline consumption, when in fact
the converse is preferable (24).
The production of low sulfur gasoline (and diesel fuel) is a necessary step to
assure clean air and to reduce air pollution from the transportation sector. Sulfur
is a pollutant and prevents the adoption of all major pollution control technolo-
gies. No significant air pollution reduction strategy can work without reducing

Table 5. Refining Technologies for Producing Reformulated Gasoline Constituents


Technology Objective
catalytic reformer prefractionation reduce benzene
reformate fractionation reduce benzene
isomerization increase octane
aromatics saturation reduce total aromatics
catalytic reforming oxygenates for octane enhancement
MTBE synthesis provide oxygenates
isobutane dehydrogenation feedstock for oxygenate synthesis
catalytic cracker naphtha fractionation increase alkylate; increase oxyge-
nates; reduce olefins and sulfur
feedstock hydrotreating reduce sulfur
Vol. 1 PETROLEUM REFINERY PROCESSES 319

sulfur to near-zero levels. When sulfur-containing fuels are burned, sulfur is


emitted as sulfur dioxide, SO2, or sulfate particulate matter. Any reduction in
fuel sulfur immediately reduces these sulfur compounds and, as sulfur levels
decline past a certain point, the benefits increase to include total pollutant emis-
sions. Reduced sulfur fuel (150 ppm sulfur) makes existing vehicles cleaner and
decreases emissions. These benefits increase as vehicles are designed to meet
higher emissions standards and sulfur levels are reduced further. Low sulfur
fuel (50 ppm sulfur) allows for the further benefit of advanced control techno-
logies for diesel vehicles. Diesel particulate filters can be used with low sulfur
fuel, but only achieves 50% control efficiency. Selective catalytic reduction
can be used for >80% control of NOx emissions.
Near-zero sulfur fuel (<10 ppm sulfur) allows for the use of NOx adsorbers,
increasing NOx control to >90% in both diesel and gasoline vehicles. This enables
more fuel-efficient engine designs, designs that are incompatible with current emis-
sions control systems. Particulate filters achieve the maximum efficiency with near-
zero sulfur fuels, approaching 100% control of particulate matter emissions.
The technologies required to reduce sulfur to near-zero levels are in use in
many refineries and the refining industry continues to make progress in devel-
oping more active catalysts and novel processes for removal of sulfur. Refiners
are using hydrotreating technologies in which the feedstock (be it a distillate
fraction or even the residuum) is treated with hydrogen to reduce sulfur levels,
and hence to make processing easier. The result has been reduced sulfur levels
and lower permitted emissions standards for all types of vehicles. Long-term
environmental challenges and the prospect of still stricter requirements in the
future are spurring further research and technology development.
7.3. Solvents. Petroleum naphtha is a generic term applied to refined,
partly refined, or unrefined petroleum products. Naphthas are prepared by
any of several methods, including fractionation of distillates or even crude petro-
leum, solvent extraction, hydrocracking of distillates, polymerization of unsatu-
rated (olefinic) compounds, and alkylation processes. Naphtha can also be a
combination of product streams from more than one of these processes.
The main uses of petroleum naphtha fall into the general areas of solvents
(diluents) for paints, etc, dry-cleaning solvents, solvents for cutback asphalt,
solvents in rubber industry, and solvents for industrial extraction processes.
Turpentine, the older, more conventional solvent for paints, has been almost
completely replaced by the cheaper and more abundant petroleum naphtha.
7.4. Kerosene. Kerosene, also called kerosine, originated as a straight-
run (distilled) petroleum fraction that boiled over the temperature range of
205–2608C (401–5008F). Kerosene was the major refinery product before the
onset of the automobile age, but now kerosene might be termed as one of several
other petroleum products after gasoline. In the early days of petroleum refining,
some crude oils contained kerosene fractions of high quality, but other crude
oils, such as those having a high proportion of asphaltic materials, had to be
thoroughly refined to remove aromatics and sulfur compounds before a satisfac-
tory kerosene fraction could be obtained.
Kerosene is believed to be composed chiefly of hydrocarbons containing
12–15 carbon atoms per molecule. Low proportions of aromatic and unsaturated
hydrocarbons are desirable to maintain the lowest possible level of smoke during
320 PETROLEUM REFINERY PROCESSES Vol. 1

burning. Although some aromatics may occur within the boiling range assigned
to kerosene, excessive amounts can be removed by extraction.
The significance of the total sulfur content of kerosene varies greatly with the
type of oil and the use to which it is put. Sulfur content is of great importance when
the kerosene to be burned produces sulfur oxides, which are of environmental con-
cern. The color of kerosene is of little significance, but a product darker than usual
may have resulted from contamination or aging; in fact, a color darker than speci-
fied may be considered by some users as unsatisfactory. Kerosene, because of its
use as a burning oil, must be free of aromatic and unsaturated hydrocarbons;
the desirable constituents of kerosene are saturated hydrocarbons.
Diesel fuel, jet fuel, kerosene (range oil), No. 1 fuel oil, No. 2 fuel oil, and
diesel fuel are all popular distillate products coming from the kerosene fraction
of petroleum. One grade of jet fuel uses the heavy naphtha fraction, but the ker-
osene fraction supplies the more popular and heavier grade of jet fuel, as well as
smaller amounts that are sold as burner fuel (range oil) or No. 1 heating oil.
Some heating oils (generally No. 2 heating oil) and diesel fuel are similar and
can sometimes substitute for each other. The home heating oil is intended to
be burned with a furnace for space heating. The diesel fuel is intended for
compression–ignition engines.
The cetane number of a diesel fuel is a number that indicates the ability of a
diesel engine fuel to ignite quickly, and burn smoothly, after being injected into the
cylinder. In high speed diesel engines, a fuel with a long ignition delay tends to pro-
duce rough operation. The cetane number should not be confused with the cetene
number, an obsolete designation for the starting and running quality of diesel
fuel that uses cetene, C16H30, as the reference fuel. The cetene number has been
replaced by the cetane number, a scale based on the ignition characteristics of
two well-defined hydrocarbons, cetane, (n-hexadecane), and 2,3,4,5,6,7,8-hepta-
methylnonane.
Cetane has a short delay period during ignition and is assigned a cetane num-
ber of 100; heptamethylnonane has a long delay period and has a cetane number of
15. Just as the octane number is meaningful for automobile fuels, the cetane number
is a means of determining the ignition quality of diesel fuels and is equivalent to the
percentage by volume of cetane, in the blend with heptamethylnonane, that
matches the ignition quality of the test fuel. The cetane number of diesel fuel usually
falls into the 30–60 range; a high cetane number is an indication of the potential for
easy starting and smooth operation of the engine.
Other methods are also available for the estimation of diesel fuel quality.
For example, the diesel index (DI) is defined by the relation:

DI ¼ ðA F  APIÞ=100

where A8F is the aniline point in degrees Fahrenheit and 8API is the American
Petroleum Institute gravity. A high aniline point corresponds to a high pro-
portion of paraffins in the diesel fuel; such a fuel has a high diesel index and,
therefore, a high cetane number.
7.5. Fuel Oil. Fuel oil is classified in several ways, but generally into two
main types: distillate fuel oil and residual fuel oil. Distillate fuel oil is vaporized
and condensed during a distillation process; it has a definite boiling range and
Vol. 1 PETROLEUM REFINERY PROCESSES 321

does not contain high boiling oils or asphaltic components. A fuel oil that con-
tains any amount of the residue from crude distillation hydrocracking is a resi-
dual fuel oil. However, the terms distillate fuel oil and residual fuel oil are losing
their significance because fuel oils are made for specific uses and can be either
distillates, residuals, or mixtures of the two. The terms domestic fuel oil, diesel
fuel oil, and heavy fuel oil are more indicative of the uses of fuel oil.
Domestic fuel oils are those fuel oils used primarily in the home and include
kerosene, stove oil, and furnace fuel oil. Diesel fuel oils are also distillate fuel oils,
but residual oils have been successfully used to power marine diesel engines, and
mixtures of distillates and residuals have been used on locomotive diesels. Heavy
fuel oils include a variety of oils, ranging from distillates to residual oils, that
must be heated to 2608C or higher before they can be used. In general, heavy fuel
oil consists of residual oil blended with distillate to suit specific needs. Heavy fuel
oil includes various industrial oils and, when used to fuel ships, is called bunker oil.
Stove oil is a straight-run (distilled) fraction from crude oil, whereas other
fuel oils are usually blends of two or more fractions. The straight-run fractions
available for blending into fuel oils are heavy naphtha, light and heavy gas oils,
and residua. Cracked fractions such as light and heavy gas oils from catalytic
cracking, cracking coal tar, and fractionator bottoms from catalytic cracking
may also be used as blends to meet the specifications of different fuel oils.
Heavy fuel oil usually contains residuum that is mixed (cut back) to a spe-
cified viscosity with gas oils and fractionator bottoms. For some industrial pur-
poses in which flames or flue gases contact the product (eg, ceramics, glass, heat
treating, and open hearth furnaces), fuel oils must be blended to low sulfur spe-
cifications; low sulfur residues are preferable for these fuels.
The manufacture of fuel oils at one time largely involved using what was
left after removing desired products from crude petroleum. Now fuel oil manufac-
ture is a complex matter of selecting and blending various petroleum fractions to
meet definite specifications.
Fuel oil that is used for heating is graded from No. 1 Fuel Oil, to No. 6 Fuel
Oil, and cover light distillate oils, medium distillate, heavy distillate, a blend
of distillate and residue, and residue oil. For example No. 2 and No. 3 Fuel oils
refer to medium to light distillate grades used in domestic central heating fuel
oil refers to a medium heavy residual oil used for heating large commercial
premises.
7.6. Lubricating Oil. Lubricating oil is distinguished from other petro-
leum fractions by the high (>4008C) boiling point as well as their high viscosity.
Lubricating oil may be divided into many categories according to the types of ser-
vice; however, there are two main groups: oils used in intermittent service, such
as motor and aviation oils, and oils designed for continuous service, such as turbine
oils.
Lubricating oil used in intermittent service must show the least possible
variation in viscosity with respect to temperature and must be changed at fre-
quent intervals to remove the foreign matter collected during service. The stabi-
lity of such oil is therefore of less importance than the stability of oil used in
continuous service for prolonged periods without renewal. Lubricating oil for
continuous service must be extremely stable because the engines in which it is
used operate at fairly constant temperature without frequent shutdown.
322 PETROLEUM REFINERY PROCESSES Vol. 1

Lubricating oil may be divided into many categories according to the types
of service it is intended to perform. However, there are two main groups: (1) oils
used in intermittent service, such as motor and aviation oils and (2) oils designed
for continuous service such as turbine oils.
7.7. Wax. Petroleum waxes are of two general types: paraffin wax in
distillates and microcrystalline wax in residua. The melting point of wax is not
directly related to its boiling point because waxes contain hydrocarbons of dif-
ferent chemical structure. Nevertheless, waxes are graded according to their
melting point and oil content. Paraffin wax is a solid crystalline mixture of
straight-chain (normal) hydrocarbons ranging from mostly C20 –C30 and higher.
Wax constituents are solid at ordinary temperatures (258C; 778F) whereas petro-
latum (petroleum jelly) contains both solid and liquid hydrocarbons.
Wax production by wax sweating was originally used in Scotland to sepa-
rate wax fractions by employing various melting points from the wax obtained
from shale oils. Wax sweating is still used to some extent, but is being replaced
by the more convenient wax recrystallization process. In wax sweating, a cake of
slack wax, also known as crude or raw wax, is slowly warmed to a temperature at
which the oil in the wax and the lower melting waxes become fluid and drip (or
sweat) from the bottom of the cake, leaving a residue of higher melting wax.
Insofar as they are used to purify other products, several processes used in
the refinery fall under the classification of dewaxing processes; however, such
processes must also be classified as wax production processes (1,2). Most com-
mercial dewaxing processes utilize solvent dilution, chilling to crystallize the
wax, and filtration (28). The MEK process (MEK–toluene solvent) is widely
used. Wax crystals are formed by chilling through the walls of scraped surface
chillers, and wax is separated from the resultant wax–oil–solvent slurry by
using fully enclosed rotary vacuum filters.
Solvents used for dewaxing are naphtha, propane, sulfur dioxide, acetone–
benzene, trichloroethylene, ethylenedichloride-benzene (Barisol), MEK–benzene
(benzol), methyl n-butyl ketone, and methyl n-propyl ketone. Other solvents in com-
mercial use for dewaxing include N-methylpyrrolidinone, MEK–MIBK (methyl iso-
butyl ketone), dichloroethane-methylene dichloride, and propylene–acetone.
Solvent dewaxing can be applied to light, intermediate, and heavy lubricat-
ing oil distillates, but each distillate produces a different kind of wax, and each of
these waxes is actually a mixture of a number of waxes. For example, the wax
obtained from light paraffin distillate consists of a series of paraffin waxes that
have melting points in the range of 30–708C (86–1588F) and are characterized
by a tendency to harden into large crystals. However, heavy paraffin distillate
yields a wax composed of a series of waxes that have melting points in the
range of 60–908C (140–1948F) and that harden into small crystals from which
they derive the name microcrystalline waxes or microwaxes.
On the other hand, intermediate paraffin distillates contain paraffin waxes
and waxes intermediate in properties between paraffin and microwaxes. Thus,
the solvent dewaxing process produces three different slack waxes depending
on whether light, intermediate, or heavy paraffin distillate is processed. The
slack wax from heavy paraffin distillate may be sold as dark raw wax, the wax
from intermediate paraffin distillate as pale raw wax. The latter is treated with
lye and clay to remove odor and improve color.
Vol. 1 PETROLEUM REFINERY PROCESSES 323

In the propane process, part of the propane diluent is allowed to evaporate


by reducing pressure so as to chill the slurry to the desired filtration tempera-
ture, and rotary pressure filters are employed. Complex dewaxing requires no
refrigeration, but depends on the formation of a solid urea–n-paraffin complex
that is separated by filtration and then decomposed. This process is used to
make low viscosity lubricants that must remain fluid at low temperatures (refrig-
eration, transformer, and hydraulic oils) (28).
Another method of separating petrolatum from residua is by centrifuge
dewaxing. In this process, the reduced crude oil is dissolved in naphtha
and chilled to 188C or lower, which causes the wax to separate. The mixture
is then fed to a battery of centrifuges where the wax is separated from the liquid.
However, the centrifuge method has been largely displaced by solvent dewaxing
methods, as well as more modern methods of wax removal. Similar use is antici-
pated for catalytic dewaxing processes based on selective hydrocracking of the
normal paraffins; such processes use a molecular sieve-based catalyst in which
the active hydrocracking sites are accessible only to the paraffin molecules.
Catalytic dewaxing (31) is a hydrocracking process operated at elevated
temperatures (280–4008C; 536–7528F) and pressures, 2070–10,350 kPa (300–
1500 psi). However, the conditions for a specific dewaxing operation depend on
the nature of the feedstock and the product pour point required. The catalyst
employed for the process is a mordenite-type catalyst that has the correct pore
structure to be selective for normal paraffin cracking. Platinum on the catalyst
serves to hydrogenate the reactive intermediates so that further paraffin degra-
dation is limited to the initial thermal reactions.
Another catalytic dewaxing process also involves selective cracking of nor-
mal paraffins and those paraffins that may have minor branching in the chain. In
the process (Fig. 8), the catalyst can be reactivated to fresh activity by relatively
mild nonoxidative treatment. The time allowed between reactivations is a func-
tion of the feedstock; after numerous reactivations it is possible that there will
be coke buildup on the catalyst.
A catalytic dewaxing process can be used to dewax a variety of lubricating
base stocks; as such, it has the potential to replace solvent dewaxing, or even be
used in combination with solvent dewaxing (Fig. 9), as a means of relieving the
bottlenecks which can, and often do, occur in solvent dewaxing facilities.
7.8. Asphalt. Asphalt manufacture is, in essence, a matter of distilling
everything possible from crude petroleum until a residue with the desired
properties is obtained (32). This is usually done by stages; crude distillation at
atmospheric pressure removes the lower boiling fractions and yields a reduced
crude that may contain higher boiling (lubricating) oils, asphalt, and even wax.
Distillation of the reduced crude under vacuum removes the oils (and wax) as volatile
overhead products and the asphalt remains as a bottom (or residual) product. At this
stage the asphalt is frequently (and incorrectly) referred to as pitch. In terms of meet-
ing specifications, asphalt can be made softer by blending hard asphalt with the
extract obtained in the solvent treatment of lubricating oils. On the other hand,
soft asphalts can be converted into harder asphalts by oxidation (air blowing).
Road oils are liquid asphalt materials intended for easy application to earth
roads. They provide a strong base or a hard surface and maintain a satisfactory
passage for light traffic. Liquid road oils, cutbacks, and emulsions are of recent
H2 makeup

Recycle gas
Hydrotreater
or fuel gas

Low
temperature
flash

Naphtha
Furnace

324
Distillation

Reactor
Waxy
raffinate
or
deasphalted
residuum

High
temperature
flash

Finished lube
base stock

Fig. 8. Catalytic dewaxing.


Vol. 1 PETROLEUM REFINERY PROCESSES 325

Light
Partial hydrocarbons
Neutrals or Solvent Raffinate Catalytic
solvent
bright stock extract dewaxing Dewaxed lube
dewaxing

Foot oil
Extract
Deoil

Hardwax

Fig. 9. Catalytic dewaxing used in conjunction with solvent dewaxing.

date, but the use of asphaltic solids for paving goes back to the European prac-
tices of the early 1800s.
Cutback asphalts are mixtures in which hard asphalt has been diluted
with a lighter oil to permit application as a liquid without drastic heating.
They are classified as rapid, medium, and slow curing, depending on the vola-
tility of the diluent, which governs the rates of evaporation and consequent
hardening.
Asphalt can be emulsified with water to permit application without heat-
ing. Such emulsions are normally of the oil-in-water type. They reverse or
break on application to a stone or earth surface, so that the oil clings to the
stone and the water disappears. In addition to their usefulness in road and
soil stabilizations, they are useful for paper impregnation and waterproofing.
The emulsions are chiefly either the soap or alkaline type, or the neutral or clay
type. The former breaks readily on contact, but the latter is more stable and
probably loses water mainly by evaporation. Good emulsions must be stable dur-
ing storage or freezing, suitably fluid, and amenable to control for the speed of
breaking.
Recently, asphalt has grown to be a valuable refinery product. In the post-
1980 period, a shortage of good quality asphalt has developed. This is due in no
short measure to the tendency of refineries to produce as much liquid fuels (eg,
gasoline) as possible. Thus, residua that would have once been used for asphalt
manufacture are now being used to produce liquid fuels (and coke).
7.9. Coke. Coke is the residue left by the destructive distillation (coking)
of residua. The composition of coke varies with the source of the crude oil, but
in general, large amounts of high molecular weight complex hydrocarbons
(rich in carbon, but correspondingly poor in hydrogen) make up a high propor-
tion. The solubility of coke in carbon disulfide has been reported to be as high
as 50–80%, but this is, in fact, a misnomer, since the coke is an insoluble,
honeycomb-type material that is the end product of thermal processes.
Petroleum coke is employed for a number of purposes; its principal use is in
the manufacture of carbon electrodes for aluminum refining, which requires a
high purity carbon that is low in ash and free of sulfur. In addition, coke is
employed in the manufacture of carbon brushes, silicon carbide abrasives, struc-
tural carbon (eg, pipes and Rashig rings), as well as calcium carbide manufacture
from which acetylene is produced. Coke produced from low quality crude oil is
326 PETROLEUM REFINERY PROCESSES Vol. 1

mixed with coal and burned as a fuel. Flue gas scrubbing is required. Coke is
used in fluidized-bed combustors or gasifiers for power generation.

8. Petrochemicals

The petrochemical industry began in the 1920s as suitable by-products became


available through improvements in the refining processes. It developed parallel
with the oil industry and has rapidly expanded since the 1940s, with the oil refin-
ing industry providing plentiful cheap raw materials.
A petrochemical is any chemical (as distinct from fuels and petroleum pro-
ducts) manufactured from petroleum (and natural gas) and used for a variety
of commercial purposes. The definition, however, has been broadened to
include the whole range of aliphatic, aromatic, and naphthenic organic chemi-
cals, as well as carbon black and such inorganic materials as sulfur and ammo-
nia. Petroleum and natural gas are made up of hydrocarbon molecules, which
are comprised of one or more carbon atoms, to which hydrogen atoms are
attached. Currently, oil and gas are the main sources of the raw materials
because they are the least expensive, most readily available, and can be pro-
cessed most easily into the primary petrochemicals. Primary petrochemicals
include olefins (ethylene, propylene and butadiene) aromatics (benzene,
toluene, and the isomers of xylene); and methanol. Thus, petrochemical feed-
stocks can be classified into three general groups: olefins, aromatics, and
methanol; a fourth group includes inorganic compounds and synthesis gas
(mixtures of carbon monoxide and hydrogen). In many instances, a specific che-
mical included among the petrochemicals may also be obtained from other
sources, such as coal, coke, or vegetable products. For example, materials such
as benzene and naphthalene can be made from either petroleum or coal, while
ethyl alcohol may be of petrochemical or vegetable origin.
Petrochemicals are generally divided into three groups: (1) aliphatics, such
as butane and butene; (2) cycloaliphatics, such as cyclohexane, cyclohexane
derivatives, and aromatics (eg, benzene, toluene, xylene, and naphthalene);
and (3) inorganics, such as sulfur, ammonia, ammonium sulfate, ammonium
nitrate, and nitric acid.
8.1. Aliphatics. Methane, obtained from crude oil or natural gas, or as a
product from various conversion (cracking) processes, is an important source of
raw materials for aliphatic petrochemicals (Fig. 10). Ethane, also available from
natural gas and cracking processes, is an important source of ethylene, which, in
turn, provides more valuable routes to petrochemical products (Fig. 11).
Ethylene (CH2 – CH2), an important olefin, is usually made by cracking
gases such as ethane, propane, butane, or a mixture of these as might exist in
a refinery’s off-gases. When gas feedstock is scarce or expensive, naphthas and
even whole crude oil have been used in specially designed ethylene crackers.
The heavier feeds also give significant quantities of higher molecular weight ole-
fins and aromatics. Ethylene is consumed in larger amounts than any other
hydrocarbon for the production of aliphatic petrochemicals, but it is by no
means the only source of aliphatic petrochemicals. Propane and butane are
also important aliphatic hydrocarbons (Fig. 12). Propane is usually converted
to propylene by thermal cracking, although some propylene is also available
Air
Nitric acid
Partial combustion Ammonia
Carbon black
Nitrogen
from air
Pyrolysis Ammonium
Hydrogen nitrate

Hydrogen Steam
Steam or oxygen Hydrogen
Methane and carbon and carbon
monoxide Urea
dioxide

Oxygen Air
Hydrogen cyanide Carbon dioxide

327
Acetylene
Air
Methanol Formaldehyde
Acrylonitrile

Hydrogen chloride

Dimer Hydrogen
chloride Chlorine or Methyl chloride
or acetic alkali
Hydrogen acid
chloride
Methylene dichloride

Chloroprene Vinyl chloride Chloroethylenes


Chloroform
or acetate

Chlorine Carbon tetrachloride

Fig. 10. Methane as a source of petrochemicals.


Polyglycols

Ethanolamines Diethylene
(Catalyst) and triethylene
Polyethylene glycols
Ammonia
Water

Oxygen (catalyst) Ethylene oxide Ethylene glycol


Alkali Alcohols or
Hydrogen Dehydrogenation
alkyl phenois
cyanide
Refinery- Ethylene
chlorohydrin Glycol ethers and
cracked Hypochlorous
polyglycol ethers
gas acid Acrylonitrile

Chlorine Ethylene dichloride Vinyl chloride

Oxygen

328
Hydrogen chloride
Bromine
Ethylene Ethylene dibromide
Hydrogen
chloride
Ethyl chloride
Water (catalyst)
Ethyl alcohol Acetaldehyde
Sulfuric
acid Sulfuric Water
Cracking
of ethane, esters
propane, Benzene
heavy Styrene
Ethylbenzene
liquids
Oxygen, acetic acid
(catalyst)
Vinyl acetate

Fig. 11. Ethylene as a source of petrochemicals.


Vol. 1 PETROLEUM REFINERY PROCESSES 329

from refinery gas streams. The various butylenes are more commonly obtained
from refinery gas streams. Butane dehydrogenation to butylene is known, but
is more complex than ethane or propane cracking, and its product distributions
are not always favorable. The production of gasoline and other liquid fuels con-
sumes large amounts of butane.

Hydrogen
Butanols
Carbon monoxide,
Butyraldehydes
hydrogen, catalyst Hydrogen
2-Ethylhexanol
(Catalyst)
Polypropylene

Water Oxidation or dehydrogenation


Isopropyl alcohol Acetone
(sulfuric acid)
Phenol
Cracking of
propane,
Trimer Nonylphenol
heavy liquids Sulfuric
Acid Benzene acid or
Tetramer Dodecylbenzene Detergents sulfur
dioxide
Benzene
Alkylbenzene

Benzene Air Cumene


Propylene Cumene
hydroperoxide
Oxygen, ammonia
Acrylonitrile
Nitric acid, catalyst Epichlorohydrin Glycerol
Chlorine
Allyl chloride
(high temperature)
Refinery-
cracked Hypochlorous Allyl alcohol Isooctyl alcohol
gas Propylene oxide Alkali
acid, alkali
Hydrogen
Acid Carbon monoxide, Aldehydes containing
Heptenes
hydrogen 8-carbon atoms

Water Dehydrogenation
Methyl ethyl ketone
(sulfuric acid)

Dehydrogenation Polymers and copolymers with


Butadiene
styrene and acrylonitrile

Sodium cyanide, Adiponitrile, then


Chlorobutenes hexamethylenediamine
hydrogen

Water
(sulfuric acid)
Copolymer with 2% isoprene
(butyl rubber)
Isobutylene
(Sulfuric acid)
Di- and triisobutylene

Polyisobutylene
Boron trifluoride (low temperature)

Fig. 12. Use of propylene and butylene(s) as sources of petrochemicals.


Polymethylene Mixed alcohols Olefins
Methanation Oxo alcohols and
to substitute Ethanol Ethylene other oxo products
natural gas
Synthesis
Reducing gas
H2
gas
Fischer-Tropsch paraffins
Ethylene and olefins
glycol and
other glycols
Olefins Petrol and Polyolefins
diesel Heavy
water
Fuel (eg, for Ethanol and Carbonyls Ammonia
combined higher alcohols Acrylates
cycle power)
Toluene
diisocyanate
CO
Acetic

330
anhydride Diphenylmethane
Vinyl diisocyanate Urea
acetate Polyurethanes
Ethylene
Cellulose glycol
acetate Formic
Single-cell acid Acetal resins
protein Acetic Urea]
acid formaldehyde
Acrylates resins
Fuel
Methanol Formaldehyde
C4-diols Melamine- and
phenol-formaldehyde
Pentaerythritol resins
Ethylene diamine
Olefins Hexamethylene tetraacetic acid
tetramine
Gasoline Methacrylates

Fig. 13. Chemicals from synthesis gas.


Vol. 1 PETROLEUM REFINERY PROCESSES 331

Air
Xylenes Phthalic anhydride

Air Isophthalic acid


or methyl esters

Air Terephthalic acid


Existing aromatics Benzene,
or methyl esters
Refinery toluene, xylenes,
naphthas ethyl-benzene
Reforming of cyclopentanes,
cyclohexanes, and paraffins

(Minor)
(Minor)
(Hydrogen) Nitric acid
Benzene Toluene Nitrotoluenes

Cyclohexane Propylene Acetylene Reduction


(Minor)
Ethylene
Air Nitric Ditolylethane Toluenediamine
acid

Cyclohexanol, Ethylbenzene Cumene


cyclohexanone
Air
Air Toluene Toluene
Dehydro- diisocyanate diisocyanate
Hydroxylamine, genation
acid
Hydroperoxide

Adipic acid Styrene


Acid Acetone Bisphenol

Sulfonation and alkali fusion


Hexamethylene- or chlorination and hydrolysis
diamine
Polymers
Phenol

Propylene trimer
Caprolactam
Epichlorhydrin
Formaldehyde,
urea

Nylons Dodecylbenzene Detergents Nonylphenol Phenolic Epoxy


resins resins

Fig. 14. Aromatic and cycloaliphatic compounds as sources of petrochemicals.

The gaseous constituents produced in a refinery give rise to a host of che-


mical intermediates that can be used for the manufacture of a wide variety of
products (1,2). Synthesis gas (carbon monoxide, CO, and hydrogen, H2) mixtures
are also used to produce valuable industrial chemicals (Fig. 13).
8.2. Cycloaliphatics and Aromatics. Cyclic compounds (cyclohexane
and benzene) are also important sources of petrochemical products (Fig. 14). Aro-
matics are in high concentration in the product streams from a catalytic refor-
mer. When aromatics are needed for petrochemical manufacture, they are
extracted from the reformer’s product using solvents, such as glycols (eg, the
Udex process) and sulfolane.
The mixed monocyclic aromatics are called BTX as an abbreviation for
benzene, toluene, and xylene. The benzene and toluene are isolated by distilla-
tion, and the isomers of the xylene are separated by superfractionation, frac-
tional crystallization, or adsorption. Benzene is the starting material for
styrene, phenol, and a number of fibers and plastics. Benzene and cyclohexane
are responsible for products such as nylon and polyester fibers, polystyrene,
epoxy resins, phenolic resins, and polyurethanes.
Toluene is used to make a number of chemicals, but most of it is blended
into gasoline. Xylene use depends on the isomer: p-xylene goes into polyester
332 PETROLEUM REFINERY PROCESSES Vol. 1

and o-xylene into phthalic anhydride. Both are involved in a wide variety of con-
sumer products.
Benzene, toluene, and xylene are made mostly from catalytic reforming of
naphthas with units similar to those already discussed. As a gross mixture, these
aromatics are the backbone of gasoline blending for high octane numbers.
However, there are many chemicals derived from these same aromatics; thus
many aromatic petrochemicals have their beginning by selective extraction
from naphtha or gas-oil reformate.
8.3. Inorganics. Of the inorganic chemicals, ammonia is by far the most
common. Ammonia is produced by the direct reaction of hydrogen with nitrogen;
air is the source of nitrogen: N2 þ 3 H2 ! 2 NH3. Refinery gases, steam reforming
of natural gas (methane) and naphtha streams, and partial oxidation of hydrocar-
bons or higher molecular weight refinery residual materials (residua, asphalts)
are the sources of hydrogen. Ammonia is used predominantly for the production
of ammonium nitrate, NH4NO3, as well as other ammonium salts and urea,
H2NCONH2, which are primary constituents of fertilizers.
Carbon black, also classed as an inorganic petrochemical, is made predomi-
nantly by the partial combustion of carbonaceous (organic) material in a limited
supply of air. Carbonaceous sources vary from methane to aromatic petroleum
oils to coal tar by-products. Carbon black is used primarily for the production of syn-
thetic rubber.
Sulfur, another inorganic petrochemical, is obtained by the oxidation of
hydrogen sulfide:
2 H2 S þ O 2 ! 2 H2 O þ 2 S

Hydrogen sulfide is a constituent of natural gas and also of the majority


of refinery gas streams, especially those off-gases from hydrodesulfurization
processes. A majority of the sulfur is converted to sulfuric acid for the manu-
facture of fertilizers and other chemicals. Other uses for sulfur include the pro-
duction of carbon disulfide, refined sulfur, and pulp and paper industry
chemicals.

BIBLIOGRAPHY

‘‘Petroleum Refinery Processes’’ in ECT 1st ed., Vol. 10, pp. 109–161, by W. L. Nelson and
A. P. Buthod, University of Tulsa; in ECT 2nd ed., Vol. 15, pp. 1–77, by W. L. Nelson,
University of Tulsa; ‘‘Petroleum (Refinery Processes, Survey)’’ in ECT 3rd ed., Vol. 17,
pp. 183–256, by C. E. Jahnig, Consultant; in ECT 4th ed., Vol. 18, pp. 433–469, by
J. Speight, Western Research Institute; ‘‘Petroleum Refinery Processes’’ in ECT
(online), posting date: December 4, 2000, by J. Speight, Western Research Institute.

CITED PUBLICATIONS

1. J.G. Speight and B. Ozum. Petroleum Refining Processes, Marcel Dekker, New York,
2002.
Vol. 1 PETROLEUM REFINERY PROCESSES 333

2. J. G. Speight, The Chemistry and Technology of Petroleum, 3rd ed., Marcel Dekker,
New York, 1999.
3. J. G. Speight, The Chemistry and Technology of Coal, 2nd ed, Marcel Dekker,
New York, 1994.
4. J. G. Speight, The Desulfurization of Heavy Oils and Residua, Marcel Dekker,
New York, 1981.
5. M. S. Matar, in H. K. Abdel-Aal, B. A. Bakr, and M. A. Al-Sahlawi, eds., Petroleum
Economics and Engineering, 2nd ed., Marcel Dekker, New York, 1992, p. 33.
6. D. R. Burris and J. D. McKinney, in J. J. McKetta, ed., Petroleum Processing Hand-
book, Marcel Dekker, New York, 1992, p. 666.
7. G. W. Mushrush and J. G. Speight, Petroleum Products: Instability and Incompatibil-
ity, Taylor & Francis, Washington, D.C., 1995.
8. J. R. Salazar, in R. A. Meyers, ed., Handbook of Petroleum Refining Processes,
McGraw-Hill Book Co., New York, 1986, pp. 7–68.
9. J. G. Gary and G. E. Handwerk, Petroleum Refining: Technology and Economics, 3rd
ed, Marcel Dekker, New York, 1994.
10. J. R. Salazar, in Ref. 8, pp. 7–69.
11. H. M. Feintuch, J. A. Bonilla, and R. L. Godino, in Ref. 8, pp. 7–3.
12. J. D. McKinney, in Ref. 4, p. 245.
13. D. E. Blaser, in Ref. 4, p. 253.
14. D. G. Tajbl, in Ref. 8, pp. 2–9.
15. E. C. Luckenbach and co-workers, in Ref. 4, p. 349.
16. S. A. Bradley, M. J. Gattuso, and R. J. Bertolacini, eds., Characterization and Catalyst
Development: An Interactive Approach, American Chemical Society, Washington, D.C.,
1989.
17. R. T. K. Baker and L. L. Murrell, eds., Novel Materials in Heterogenous Catalysis,
American Chemical Society, Washington, D.C., 1990.
18. M. L. Occelli, ed., Fluid Catalytic Cracking II: Concepts in Catalyst Design, American
Chemical Society, Washington, D.C., 1991.
19. G. E. Weismantel, in Ref. 4, p. 592.
20. D. G. Tajbl, in Ref. 8, pp. 2–33.
21. C. N. Cabrera, in Ref. 8, pp. 6–17.
22. J. A. Weiszmann, in Ref. 8, pp. 3–3.
23. P. Greenough and J. R. K. Rolfe, in Ref. 8, pp. 7–69.
24. A. Douaud, Tomorrow’s Engines and Fuels, Report No. 94-2. Institut Francais du
Petrole, Paris, 1994.
25. R. Ragsdale, Oil Gas J., 51 (Mar. 21, 1994).
26. B. R. Shah, in Ref. 8, pp. 1–1.
27. T. Hutson, Jr. and W. C. McCarthy, in Ref. 8, pp. 1–23.
28. G. G. Scholten, in Ref. 4, p. 583.
29. J. G. Speight, Gas Processing: Environmental Aspects and Methods, Butterworth
Heinemann, Oxford, U.K., 1993.
30. H. L. Hoffman, in Ref. 4, p. 2.
31. J. D. Hargrove and J. D. McKinney, in Ref. 4, p. 558.
32. C. P.-C. Chang, J. R. Murphy, and J. D. McKinney, in Ref. 4, p. 527.

JAMES G. SPEIGHT
Consultant, CD & W Inc.
334 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS Vol. 1

OIL REFINING,
ENVIRONMENTAL
CONSIDERATIONS
1. Environmental Protection

1.1. Introduction 1–3. The natural product crude oil and the products
obtained from it are generally easy to handle and with careful treatment —
which is necessary because of their flammability — they cause little pollution
to the environment. (The CO2 emissions associated with all fossil fuels and
the corresponding effects on the earth’s climate cannot be considered in this
context.)
Environmental problems arise, however, in dealing with crude oil and its
products because the petroleum industry — like hardly any other branch of
industry — maintains exploration, transport, and refining installations for
crude oil that are scattered over the entire globe. The distribution and consump-
tion of burning and transportation fuels are even more scattered (Fig. 1). Since
the geographical location of crude oil fields does not coincide with that of the
large consuming and refining regions, enormous transportation distances exist.
The large amounts handled — the annual refining capacity in 1988 was ca.
3109 t in more than 600 refineries worldwide — means that the industry
needs to take precautions in all areas with regard to air emissions, water and
soil pollution and, to a lesser degree, noise. The problems of introducing uniform
standards are caused by legal standards which differ from country to country
depending on the local concentrations of industrial areas with their correspond-
ing environmental pollution and population density.
Because of the high cost of technical measures for environmental conserva-
tion, harmonization of legislation is absolutely necessary. As these costs amount
to more than 25 % of the total processing costs in the refineries equal competition
between different countries must be guaranteed (see Section Cost of Environ-
mental Conservation).
In the following sections, the environmental problems in crude oil refining,
storage and loading, and in the application of the products are discussed sepa-
rately, because the problems in the various fields differ.
1.2. Manufacturing Emissions. Although closed, gas-tight systems are
generally used in refinery units, emissions into air and water cannot be comple-
tely avoided even with careful handling during refining and storage of the crude
oil and its products. This is due to the management and control of the process
and the properties of the products concerned.
Hydrocarbons are discharged into the air because of their high vapor pres-
sure and they appear in refinery wastewater effluents because of their water
solubility, which is, however, small. The carcinogenic aromatic hydrocarbons
are particularly dangerous.
Further attention must be paid to the sulfur and nitrogen compounds
originating from the heteroatomic compounds in the crude oil, both because of
their smell and toxicity and because of the air pollution which arises in the
form of SO2 and NOx emissions during firing in process plants.

ß2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim


10.1002/14356007.a18_051
Vol. 1 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS 335

Fig. 1. Petroleum’s route from the well to the consumer shown as an example for
Germany.

Hydrocarbons (HC) (4). Hydrocarbons in Air. Hydrocarbon emis-


sions can arise in production plants during normal operation (1) from leaking
flanges in the pipework system; (2) at the seals of valves, pumps, and compres-
sors; and (3) in the course of sampling. In the case of an accident, gases are led in
closed systems to flares, collected as far as possible in gas recovery systems,
compressed and returned to the process. The remainder of the gases is burnt
in elevated or ground flares at efficiencies which can exceed 99 %. Liquid pro-
ducts are collected in closed slop systems that are equipped with pressure reser-
voirs and tanks and later returned to the production circuit.
In new plants measures to reduce emissions are taken during their
construction, whereas continuous retrofitting is necessary in existing plants.
Examples are flangeless piping, low-emission stuffing boxes, and seals, such as
duplicated slide ring packings. For intensely odorous, poisonous, and carcinogenic
336 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS Vol. 1

Fig. 2. Emission-free loading and unloading of gasoline hydrocarbons with off-gas


purification (refinery, intermediate storage) a) Storage tank with floating roof; b)
Exhaust gas washes (gasoline); c) Fine purification (adsorption); d) Low-temperature
cooling (to  40 8C).

substances more far-reaching measures are necessary (canned-motor pumps, spe-


cial extraction devices etc.).
Most hydrocarbon emissions in processing occur in the storage areas, ie,
tank farms for crude, feedstocks, intermediate, and final products. Pressure –
vacuum relief valves and floating cover tanks are generally used to reduce
emissions. More recent developments are emission-free tank farms, where
several fixed roof tanks fitted with internal floating roofs breathe into a closed
system at one common gasometer, which normally absorbs all changes in the
tank level. In the rare event of unusually large changes in the system (large
amounts flowing into or out of storage, solar irradiation, heavy rain), the surplus
quantity is burnt harmlessly in an associated flare or, in the case of a pressure
drop, the system is topped up with inert gas.
Considerable amounts of hydrocarbons are also emitted in the loading facil-
ities, especially in the loading of gasolines. Here low-emission or emission-free
loading for transport by ship, rail, and road has become largely accepted. Various
methods are used (Fig. 2):

1. Vapor recovery, where in a closed system the displaced gasoline vapors are
either returned to the product tank or collected in a gasometer and used for
the firing of process plants
2. Regenerative adsorption of the vapors on suitable adsorbents
3. Recovery of the products in liquid form after cooling or washing out the
vapors

Care must be taken to prevent the formation of explosive gasoline – air


mixtures which can occur in the road tanker to be loaded and in the adjacent
piping and equipment.
This can be achieved by (1) keeping the concentrations outside the explosive
range, (2) short transportation paths and exclusion of ignition sources, and
(3) extremely strict control of the oxygen contents.
Vol. 1 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS 337

Dispensing of gasoline from the road tanker to the service station is also
increasingly carried out with vapor recovery between the road tanker and
the installed storage tank of the service station. The possibilities of reducing
emissions in the final step (filling the customer’s motor vehicle) are described
in Section Transportation Fuels.
The measures described reduced the atmospheric emissions from proces-
sing, storage, and distribution of refinery products in the mid 1980s in Western
Europe to < 8 % of the total man-made HC emissions. The portion from the
refineries themselves was in turn only a quarter of this value. More recent inves-
tigations by Concawe show further improvements. Thus, relative to the crude oil
throughput, refinery emission losses of ca. 0.05 wt% are expected, depending on
the structure and size of the plant. For inland refineries in Germany, for exam-
ple, values of 0.02 – 0.03 wt% have been published.
Hydrocarbons in Wastewater. Hydrocarbon-containing wastewater is
unavoidably obtained at various points in the refinery because of the water con-
tent of the crude oil itself and because steam is employed in various processing
steps. The total amount of hydrocarbon-containing wastewater in a normal
refinery is of the order of 60 – 100 m3/h.
This wastewater must be removed from the process units after separation of
the oil phase, and led via a closed system to wastewater purification. Rainwater
from exposed plant areas and from tank yards, and possibly contaminated cool-
ing water from leaks or accidents must be treated in the same way. Considerable
buffer volumes must be made available for the latter amounts of water which are
formed discontinuously and sometimes in large quantities.
Treatment in the wastewater purification system is carried out stepwise by:

1. mechanical separation (sieves, filters, oil – water separators)


2. physicochemical purification (stripping, flocculation, flotation)
3. biological treatment

Biological wastewater treatment of hydrocarbons in refinery wastewaters is


normally problem-free. However, the incoming streams and corresponding buffer
volume must be continuously monitored to detect pollution by sulfur or nitrogen
compounds and by oxygen-containing components such as phenols.
In many countries there is an increasing legal requirement for covered
water treatment plants to avoid odor and for total nitrogen removal for further
protection of surface waters (rivers, lakes etc.). The latter usually requires an
additional purification stage with increased residence time.
After the biological stage, the water is clean and can be discharged into the
receiving water. The average analyses of the wastewater from a modern refinery
(in mg/L) are:

COD 60 – 100
BOD 5 – 15
Oil 0.5 – 2
Settleable solids 0.1 – 0.3
Phenols 0.1 – 0.2
338 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS Vol. 1

Fig. 3. Development of the discharge of hydrocarbons with wastewater in Western


Europe (reports from 87 refineries, ca. 80 %).

As a result of increased wastewater purification, the amount of hydrocar-


bons discharged by the refineries in Western Europe was reduced by ca. 90%
in the period 1970 – 1985 for about the same amount of crude oil processed
(Fig. 3) [55].
Hydrocarbons in Soil and Groundwater. Since hydrocarbons are
water-soluble (even if only to a small degree), their penetration into the soil
with possible contamination of the groundwater must be carefully avoided,
whenever crude oil or any of its products are handled.
Transport from the oil terminal to the refinery is carried out almost exclu-
sively in underground pipelines, which are also the safest means of transport.
The choice of high-grade steels as construction material, good insulation, catho-
dic corrosion protection, and continuous monitoring for leaks, including visual
monitoring from aircraft or ground inspection, ensure a high level of safety.
On difficult terrain and in areas of extreme temperature fluctuations, additional
measures must be taken (pipeline compensation, elevated piles, intermediate
tanks etc.).
The location of the refinery must be carefully selected with regard to possi-
ble dangers to drinking water. According to new legislation, all HC-handling
units must be erected so as to prevent the discharge of spilled product to the
underground, to adjacent streets, or canals; the storage tanks are placed in col-
lection spaces which are made impermeable to oil using clay layers, plastic tilts,
or concrete lining; in case of a leak they must be capable of receiving the entire
tank contents.
If hydrocarbon contamination occurs in the soil, the affected portion of
soil must be removed to prevent subsequent pollution of groundwater. If
small amounts have escaped, the contaminated soil is usually combusted
in incinerating plants. With larger amounts — and particularly if large areas
are polluted with chemical residues and dangerous refuse — the damage
Vol. 1 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS 339

must be treated in situ. Depending on the nature of the soil and the
corresponding migration of the oil, this can be done by pumping off and purify-
ing the contaminated water, possibly by additional injection of fresh water in
adjacent wells.
Degradation of the oil by microorganisms is becoming increasingly
important. This can also be done in situ or by excavating the soil and treating
it externally (6).
Loading and storage of the products outside the refinery are subject to
similar regulations. However, these are still very different in individual
countries. As a result of the large number of external distribution depots for
transport fuels and heating oils and of service stations, and because of the
enormous number of oil-heated households, special care must be taken
against overfilling and escape of products due to leaks. Corrosion-resistant
and nonageing steels, plastic-lined steel tanks, and novel, glass-fiber-
reinforced, plastic tanks are widely used. These tanks must also be con-
structed such that the whole volume can be collected if leakage occurs (double-
walled tanks).
In some countries with a population density and dense housing and
industrial areas, the regulations which apply to pure hydrocarbons (i.e., pro-
ducts) are also applied to the handling of process water with much lower
HC-concentrations.
Sulfur and Nitrogen Compounds. As a natural product, crude oil also
contains heteroatomic compounds containing sulfur, nitrogen, and oxygen in
addition to hydrocarbons. Whereas the nitrogen and oxygen contents are in
the ppm range and play only a secondary role in atmospheric emissions, the
sulfur content of the crudes can be as high as several percent. The distribution
over the individual refinery fractions varies, but the content increases with
increasing molecular size.
Sulfur Compounds. Sulfur and its compounds are catalyst poisons and
adversely affect atmospheric emissions. Hydrogen sulfide, mercaptans, and dis-
ulfides are odor nuisances, and sulfur dioxide is formed during the combustion of
crude oil products. Therefore, sulfur and its compounds must be removed or their
contents reduced.
The light refinery products, liquefied petroleum gas (see LIQUEFIED PETRO-
LEUM GAS) and gasoline, must be almost completely sulfur-free; for diesel fuels
(see AUTOMOTIVE FUELS) and light heating oils (see HEATING OIL), a substantial
sulfur reduction to 0.1–0.5% is required by recent legislation.
Serious problems exist with heavy fuel oil, which is used almost
exclusively as fuel in large industrial furnaces and power stations and leads
to considerable SO2 emissions. Many countries have established a maximum
sulfur content in fuels of 1 – 2 wt%. This value can be reached without
additional treatment only with a few, low-sulfur crude oils, whose supplies
are limited.
Sulfur is generally removed from distillates by hydrodesulfurization,
whereby the chemically bound sulfur is converted to hydrogen sulfide. The
H2S is then removed in a gas scrubber, converted to elemental sulfur in
the downstream Claus process, and supplied to the chemical industry as a raw
material.
340 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS Vol. 1

Fig. 4. Sulfur removal during crude oil processing.

The intense smell of the H2S-containing gases to be processed and the high
toxicity of H2S, even at high dilution, means that precautions must be taken
when handling. All sulfur-processing plants must be completely gastight: in
the hazard zones, instruments and alarm devices are installed which automati-
cally shut down the plants in case of danger.
Sulfur recovery in refineries has increased greatly as the result of the lower
permitted sulfur content of the products. In conventional Claus plants with two
reactors connected in series the conversion of the H2S feed is ca. 95 %, a third
reactor increases the conversion to 96 %. For a further reduction of the H2S con-
tent an aftertreatment step (Scot process, Sulfren, see NATURAL GAS) is required.
Almost complete removal that is required in various countries (eg, Germany
10 ppm H2S in off-gas) can only be achieved with an additional high-tempera-
ture combustion step. Figure 4 compares the sulfur removal from the processed
crude oil with the crude oil capacity in West German refineries for the period
1970–1988. The increase in sulfur removal resulted in a dramatic reduction of
overall SO2 emissions since most of the desulfurized hydrocarbons are used as
fuels.
Sulfur dioxide emissions in the refineries themselves are relatively small.
Depending on the refinery complexity, only 4–6% of the crude oil used is
required for fueling in the process plants. About half of this amount is covered
by the low-sulfur or sulfur-free gases obtained during processing, and the
remainder is heavy fuel oil. With more severe requirements to lower SO2 emis-
sions, the refiner might be forced to use as fuel all of the LPG that is produced in
the refinery.
Legislation differs greatly in the individual countries: the stipulated
limit values can be related to a single process unit, a stack, or the refinery
as a whole and can refer to the total sulfur discharged or the sulfur concentra-
tion in the flue gas. Combinations of several measures are also possible, and so-
called ‘‘bubble concepts’’ are used, particularly in the United States. (Bubble
Vol. 1 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS 341

concept means that for a given amount of emission from the total complex,
different limit values or amounts discharged are allowed for each individual
plant.)
The expensive flue gas desulfurization is normally uneconomic for firing
installations of the size of those employed in a normal refinery. Therefore it is
used only in special cases, eg, if the refinery is connected with a large power
station or if the sulfur content of the fuel is extremely high.
Nitrogen Compounds. The nitrogen content of most crude oils is rela-
tively low, and in the distillates it is reduced during hydroprocessing to a few
ppm. The residual nitrogen content causes no further difficulty in the use of
the products.
Ammonia, which is formed in the hydrogenation steps and added in many
refineries for process control in various process stages (pH adjustment), enters
the wastewater and can be removed in the biological stage of wastewater purifi-
cation (see Section Hydrocarbons in Watewater).
Noise. In the past, noise pollution in the neighborhood of petroleum refi-
neries played only a secondary role. Worldwide, pumps and compressors were
installed not in closed buildings but in the open air, because of the flammability
of petroleum and its products. However, because the plants were smaller at that
time, the noise emission of the units was relatively small. In addition, the sound
radiation from furnaces and their burners produced relatively low sound levels
in the refinery surroundings, because of the compact construction of the main
process plants that were simultaneously screened by lownoise auxiliary installa-
tions (tank farms) at the periphery.
Problems arose, though, in densely populated areas with insufficient dis-
tances between industrial and residential areas. Most noise problems were
caused by flare noise in the event of process disturbances.
Owing to growing environmental awareness in the public and a correspond-
ingly greater strictness of legislation, further measures will have to be taken in
the future to reduce sound emissions. This is also necessary for technical rea-
sons, because the complex conversion refinery maintains more and larger units
with greater furnaces and higher-powered gas compressors with high-speed
steam turbines etc.
Important measures for noise reduction could cover the following items:

1. Low-noise burners and additional noise insulation on process heaters and


piping.
2. Sound hoods on the drive motors and turbines of pumps and compressors.
A complete ‘‘in-housing,’’ however, may also give rise to safety problems,
because fire fighting is more difficult and the danger of explosion is greater
in closed rooms.
3. Sound insulation on control valves.
4. Low-noise flare stack tips for elevated flares or additional ground flares.

A noise reduction by 10 dB(A), ie, a reduction of the observable noise level


by about one-half, is possible according to the states of the art, though only with
considerable expenditure.
342 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS Vol. 1

Table 1. Hydrocarbon emissions in the petroleum field (Western


Europe 1986)
Proportion
3
Source of emission 10 t %

Crude oil production 20 0.4


Refinery, storage 170 4.1
Distribution 310 7.4
Vehicles
Refueling 180 4.3
Exhaust gas 2500 59.7
Evaporation 1010 24.1
Total 4190 100.0

1.3. Consumer Related Emissions. The emissions of SO2 and NOx


from the combustion of refinery products outside the refinery are much greater
than the emissions during refining itself. In addition, the HC emissions in the
transportation field, both in refueling and in running motor vehicles, are also
considerably greater than in the steps from production to the filling station
(see Table 1).
Transportation Fuels. When gasoline is combusted in the spark ignition
(Otto) engine, varying amounts of CO, NOx, unburnt hydrocarbons, and lead
compounds are emitted with the exhaust gas in addition to the combustion
products water and CO2. The amounts differ according to the driving style,
type of fuel, and engine construction. With the diesel engine, SO2 and soot
particles are also emitted.
In addition to exhaust emissions, the emission of hydrocarbon vapor during
refueling and running of motor vehicles is an important environmental pollution
factor.
Motor Gasoline (7). Different factors cause emissions of hydrocarbons
from motor vehicles: firstly, the unburnt hydrocarbons in the motor vehicle
exhaust, and secondly, the hydrocarbons emitted during refueling and running,
because of their high vapor pressure. In Western Europe in 1986, the emissions
(in 103 t/a) were distributed as follows:

Exhaust emissions 2500 %


Evaporative emissions 1010 %
Refueling losses 180 %

Specific measures are provided to reduce these different types of emission


either in the motor vehicle or at the filling station.
For exhaust emissions, the use of catalysts in the exhaust gas line has gen-
erally become accepted.
Efforts to reduce evaporative emissions and refueling losses are less
clear-cut. Their proportion in the emissions from the motor vehicle sector is ca.
Vol. 1 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS 343

Fig. 5. Reduction of hydrocarbon emissions A) Vapor recovery at the service station; B)


Large carbon filter in the motor vehicle a) Gas displacement pipe; b) Vent; c) Gas venting
valve actuated by filling nozzle; d) Gas – liquid separator; e) Gas line; f ) Magnetic valve
and regeneration control orifice; g) Standard gas vent and overturn protection; h) Outlet;
i) Fuel tank; j) Liquid seal in filling tube (reduces escape of gases); k) Activated carbon
filter with 4.5 L capacity (traps gases).

one-third of the total output, corresponding to ca. 12% of the total man-made
emissions of organic substances.
Two solutions are under discussion (Fig. 5) (8):

1. Vapor recovery at the service station, possibly supplemented by a smaller


activated carbon canister in the motor vehicle.
2. A larger activated carbon canister fitted into the car, whose charge can
adsorb all of the displaced hydrocarbons.

Vapor recovery at the service station is only a partial solution, because


most of the emission is produced when the car is running and is not influenced
by this measure. The advantage of vapor recovery is that it can be carried out
at the service station within a rather short period of time. The system
is already in use in the United States. The large activated carbon canister
(‘‘on-board system’’) is considered to be the better solution in the long term
because of its greater effectiveness and lower cost. Technical trials have
344 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS Vol. 1

been carried out with various types of vehicles. Installation in new cars pre-
sents no problems; retrofitting to old vehicles will, however, only be possible in
a few cases.
Owing to the wide use of motor fuels, particular attention must be
paid to benzene emissions. The limit for the benzene content of gasoline
(1 – 5 vol% is under discussion) presents the oil industry with problems,
particularly in the production of unleaded gasoline, because the benzene
content of various blending stocks can be considerably higher, depend-
ing on the quality or origin of the crude, up to 8 vol% for high severity refor-
mate and 18 – 40 vol% for pyrolysis gasoline. Benzene extraction from HC-
stocks does not solve this problem because of the cost and the question of
reuse of the benzene recovered. Therefore, the evaporative losses must be
minimized.
Diesel Fuel. The main exhaust gas problems of diesel vehicles con-
cern the emissions of soot particles and SO2. The NOx and CO components
are less important than in the spark ignition (Otto) engine. Demands are
increasingly being made for a reduction of these emissions, because of the
smell of the exhaust gases, and because the soot particles may have a carcino-
genic effect.
Burning Fuels. The largest proportion of refinery products worldwide is
used as a fuel for domestic or industrial heating as well as for energy produc-
tion. These are the fractions of LPG (see LIQUEFIED PETROLEUM GAS) the middle
distillates, and heavy fuel oil (see HEATING OIL). Emissions of SO2, NOx, and
CO that are associated with combustion, as well as emission of ash and
unburnt carbon (in the form of soot) can be considerably influenced by selection
and pretreatment of the products used and by the burner design. This is par-
ticularly true for the large heating installations operated with sulfur-rich
heavy fuel oils.
Liquefied Petroleum Gas. (see LIQUEFIED PETROLEUM GAS). The liquefied
petroleum gases used as fuels are almost sulfur-free as a result of pretreatment
in the refineries, and they burn without formation of soot. Their use presents no
problem for the user.
Light Heating Oil (No. 2 Fuel). Middle distillates contain up to 1.5 wt%
sulfur, depending on the origin of the crude. Most of the sulfur can be removed
in the refinery by hydrosulfurization.
The permissible sulfur limits for light heating oil in most countries are
< 0.5 wt%, because of its use in domestic heating. In the EU, the sulfur content
of domestic heating oil is now limited to 0.2 wt% (see HEATING OIL – ENVIROMENTAL
PROTECTION). If lower sulfur contents are to be achieved, desulfurization costs will
rise steeply with the current range of crudes, according to studies by Concawe (9)
(Fig. 6).
Heavy Fuel Oil (No. 6 Fuel). The residues from crude oil refining
are used as heavy fuel oil for industrial heating and energy production. Sulfur
dioxide, nitrogen oxides, CO, and particulate emissions must be considered for
heavy fuel oils; for environmental protection measures in this field.
1.4. Cost of Environmental Conservation. Because the understand-
ing of environmental problems, technical solutions, and legislation are still in
Vol. 1 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS 345

Fig. 6. Investment to reduce gas oil sulfur in Western Europe.

rapid development, a view of the expenditure on environmental conservation can


only be predicted. This estimation is necessary for an overall view of the problem,
because the costs of environmental conservation have reached such a high level
that they markedly influence the total processing costs. The great variety of pos-
sible solutions worldwide and the different legislation in the various countries
and regions also have a corresponding impact on costs.
Besides the investment costs required for new plants (15–20% of the total
cost is used for environmental conservation) and the conversion of old plants,
there are continuous operating costs for energy, maintenance, personnel, etc.
It is difficult to give current, generally applicable figures due to the increasing
debate on the environment and the rapid development in this sector. Table 2
shows the growth of environmental conservation costs in Germany from
1977–1987.
A study was set up in 1986 by the Commission of the European Communities
to determine the environmental conservation costs for refineries in the EC member
countries (10). This considered the current (1985) and future (1993) operation of
typical refineries and determined the costs of keeping to environmental regula-
tions — at the refinery sites and with regard to product specifications. The current

Table 2. Investment and operating costs for environmental conservation in the Federal
Republic of Germany in comparison with crude oil processing
1977 1979 1981 1983 1985 1987
6
Crude oil processed, 10 t/a 99.5 109.6 98.9 90.2 87.3 81.4
Investment costs, 106 DM 294 127 130 305 103 221
Operating costs, 106 DM 826 894 1129 1212 1097 900
346 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS Vol. 1

standards and regulations and those expected in future were taken into account.
The most important measures in the processing field are:

1. Gas washing systems and Claus plants for sulfur recovery from H2S
followed by fine purification.
2. Closed systems for the discharge of gaseous (via a flare) and liquid hydro-
carbons.
3. Floating roofs and floating covers in storage tanks for crude oil and
products.
4. Hydrocarbon vapor recovery systems for the storage and loading of volatile
products.
5. Collection spaces for escaping hydrocarbons and precipitation water in
production plants, tank farms and loading plants. Drainage to wastewater
purification plants.
6. Reduction of emissions (SO2, NOx) from the plant’s firing installations.
7. Closed sampling systems, laboratory analysis, and on-line instruments for
pollutant measurement in air and wastewater.

The following measures are for the production of less polluting products:

1. Hydrodesulfurization of gasoline and middle distillates.


2. Reforming and isomerization for the production of high-octane components
as a basis for unleaded gasoline.
3. Synthesis of suitable components for unleaded gasoline (MTBE).
4. Conversion of heavy residues into light, clean products.
5. Separate storage and loading of unleaded types of transportation fuels.

In determining the measures required and the resulting costs, the study
considered the environmental conservation laws applicable in each EC member
country at the time concerned or scheduled up to 1993. The results are given in
Table 3.
They show that: (1) the environmental costs rose in all EC countries;
(2) there are large differences in cost between the individual countries; (3)
for comparable costs arising from the EC directives, the additional national

Table 3. Environmental costs of petroleum processing in the EC member states (in ECU/t)
1993 1993
Country 1985 EC directives only ECþnational regulations

Belgium 1.01 6.25 8.57


France 1.05 3.11 4.54
Germany 3.38 5.87 19.28
Italy 0.34 2.15 2.52
Netherlands 1.01 6.12 17.07
Spain 0.31 3.21 3.58
United Kingdom 0.28 6.64 7.01
Vol. 1 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS 347

legislations lead to different costs; (4) the burden of environmental costs is in


some countries of the same order of magnitude as that of the other processing
costs.
The figures given are more likely to be near the lower limit: the study did
not, e.g., consider noise protection measures, soil clean-up or special regional reg-
ulations, because these can differ greatly according to the location of the refinery.
Moreover, the conservation of groundwater and consideration of climatic changes
are becoming increasingly important, and are continually leading to a tightening
of the limits of the (current) standards.
The investigations carried out for the EC can be applied to all other regions
of the world. Taking into account the high expenditure for minimizing potential
dangers (flammability, explosion risk, transport risks) and for protection of the
plant workers, crude oil refining is burdened with extremely high additional
costs related to product value and technical efforts.

2. Integrated Refinery Models

2.1. Trends of Refinery Structures. The drastic increase in crude oil


prices in 1973–1974 and 1979–1980 as well as changes in the product demand
pattern in major consumer regions, i.e., North America, Europe, and Japan, have
led to new considerations in integrated refinery processing to achieve better
utilization of residual materials, a maximization of distillate production,
an improvement in process economics, and to meet current environmental
standards.
These trends resulted in a worldwide restructuring of oil refining with the
aim of increasing the yields of the high-value products — chiefly transportation
fuels and industrial feedstocks — at the expense of low-value heavy fuel oils.
Great efforts were made to change from the simple ‘‘hydroskimming’’ type
refining scheme to more complex refinery structures with integrated residue
conversion processes. A comparison of three conversion concepts versus hydro-
skimming is given by the following schemes which are based on Arabian light
crude processing (11). Examples of developments in thermal and catalytic
cracking processes for heavy fuel oils and of developments in residue hydropro-
cessing, both in integrated refinery structures, are given in (12) and (13),
respectively.
2.2. Hydroskimming Refinery. The process configuration, known as
hydroskimming or ‘‘fuel oil refinery,’’ is shown in Figure 7. The plant lineup
consists of a crude distillation unit, hydrotreaters for gasoline and gas oil, a
catalytic reformer, and a gas and sulfur workup section. It does not include a
residue conversion unit. Atmospheric residue with a yield of 43.5% of the
crude represents by far the largest product portion, which has to be marketed
as heavy fuel oil.
The naphtha yield is 17.9% of the crude (15.9% gasoline and 2% chemical
feedstock), and the middle distillates amount to 34.9% (9.6% kerosene, 25.3%
gas oil). A comparison of the product yields of the hydroskimming refinery
with the yield structures attainable with different conversion concepts is given
in Table 4.
348 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS Vol. 1

Sulfur

Fuel gas

Amines gas Propane


plant and
Claus plant Butane
Boiling from IBP
up to 170°C Hydro- Light naphtha
treater Catalytic Regular and
reformer premium
Crude oil
feed
170-225°C Sweetening Kerosene
or HDS

Atmospheric
distillation

225-375°C
HDS Gas oil

Boiling 375°C
and higher
Fuel oil

Fig. 7. Hydroskimming.

Table 4. Yield structures of refinery conversion schemes for Arabian light crude
processing (11)
Product Hydroskimming a FCC –VBb HC – SDA –FCC c HC –Coking d
Intake, wt %
Crude oil 100.0 100.0 100.0 100.0
Hydrogen 0.4 0.5
Output, wt %
Fuel gas 1.8 3.4 3.1 4.7
LPG 1.6 3.0 2.9 2.9
Chemical 2.0 1.7
feed
Gasoline 15.9 33.0 32.9 24.4
Kerosene 9.6 9.6 9.6 9.6
Gas oil 25.3 26.3 40.2 50.9
Fuel oil 43.5 22.7 9.2
Coke 1.0 e 1.1 e 5.0 f
Sulfur 0.3 1.0 1.4 1.3
a
Hydroskimming = refinery configuration without residue conversion processes.
b
FCC – VB = conversion scheme with cat cracker and visbreaker.
c
HC – SDA – FCC = conversion scheme with hydrocracker, solvent deasphalting, and cat cracker.
d
HC – coking = conversion scheme with hydrocracker and coker.
e
Coke burnt in FCC unit.
f
Coke produced in coker unit.
Vol. 1 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS 349

Sulfur

Fuel gas

Amines gas Propane


plant and
Butane
Claus plant
Boiling from IBP
up to 170°C Hydro-
treater
Catalytic Regular and
reformer premium
Crude oil
feed
170-225°C Sweetening Kerosene
or HDS
Atmospheric
distillation
225-375°C
HDS Gas oil

FCC
HDS
375-375°C alkylation
Vacuum
Boiling 375°C distillation Coke
and higher
Boiling 550°C Visbreaker Fuel oil
and higher

Fig. 8. Catalytic cracking – visbreaking.

2.3. Conversion Concepts. Cat Cracking –Visbreaking Refinery. A


typical refinery for maximum gasoline production using an FCC unit is shown
in Figure 8. The FCC feedstock, vacuum gas oil, is produced by vacuum distil-
lation and pretreated in a hydrodesulfurizer for sulfur removal and FCC yield
optimization. In a visbreaker unit, the vacuum residue is thermally cracked to
produce additional distillates. As can be seen from Table 4, the residue output
from this refinery type is 22.7% of the crude, i.e., only about half of that of the
hydroskimming type. The gasoline yield amounts to 33% and middle distillates
make up 35.9% of the crude.
Hydrocracking – Cat Cracking Refinery. A high-conversion refinery
which combines fluid catalytic cracking and hydrocracking is shown in
Figure 9. The HC feed is vacuum gas oil, whereas the FCC feed (deasphalted
oil) is produced in a solvent deasphalting unit and desulfurized in a HDS unit.
The asphalt product of the SDA unit can be used as heavy fuel oil; however, it
must be blended with a ‘‘fuel thinner’’ (middle distillate) to meet fuel speci-
fications. The hydrogen produced in the catalytic reformer is not sufficient
for supplying the HC unit, therefore makeup hydrogen is produced in a
steam reformer. As shown in Table 4, the net fuel oil portion is only 9.2% of
the crude. Gasoline and middle distillate yields are 32.9% and 49.8% respec-
tively, ie, an improvement in comparison with the two schemes previously
described.
350 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS Vol. 1

Sulfur

Fuel gas

Amines gas Propane


plant and
Claus plant Butane

Boiling from IBP


up to 170°C Hydro-
treater Catalytic
Regular and
reformer
premium
Crude oil
feed
170-225°C
Sweetening Kerosene
Atmospheric or HDS
distillation 225-375°C
HDS Gas oil
Boiling 375°C
and higher H2

Hydro-
375-550°C FCC and
cracking
alkylation
Steam Vacuum Coke
reformer distillation SDA HDS
Boiling 550°C
and higher Fuel oil

Fig. 9. Hydrocracking – catalytic cracking.

Hydrocracking–Coking Refinery. The hydrocracking–coking refinery is


illustrated in Figure 10. In this configuration the production of middle distillates
is maximized by using a hydrocracker and a coker. Vacuum residue is used as
coker feed while the HC feed is a combination of vacuum gas oils and heavy
coker distillates. As in the previous scheme, extra hydrogen has to be supplied
from a steam reformer.
The only residual product is 5% coke (Table 4) which can be sold to the che-
mical industry or for electrode manufacturing. Since no heavy fuel oil is pro-
duced, this process configuration is often referred to as a ‘‘zero-fuel’’ refinery.
The yield of middle distillates (kerosene and gas oil) is >60% and the naphtha
yield (gasoline and chemical feedstock) is 26%.

3. Corrosion and Materials

Prevention of corrosion in refining equipment is determined by material selection


in plant construction and day-to-day operational control. As various corrosion cri-
teria are dealt with elsewhere, only the main refining-related corrosion problems
and the selection of corrosion-resistant materials are summarized as follows:
Corrosion problems are mainly caused by organic sulfur compounds
and inorganic chlorides in crude oils. The danger of steel embrittlement
by hydrogen in all processes involving hydrogen must also be taken into account.
Vol. 1 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS 351

Corrosion in distillation processes is caused by the following:

1. Formation of iron sulfide and subsequent corrosion by H2S and mercaptans


at elevated temperatures. Suitable materials for corrosion prevention are
Cr and austenitic Cr–Ni alloys, particularly for transfer lines and reboiler
tubes, and Cr–Mo alloys for furnace tubes. The lower parts of the crude
and vacuum columns should be lined with Cr steels.
2. Attack by naphthenic acid to form iron naphthenates and subsequent cor-
rosion. Austenitic Cr–Ni–Mo alloys can be used and naphthenic acid can
be removed upstream of the columns.
3. Hydrogen chloride corrosion in overhead condensation systems. Monel or
other Ni–Cu alloys can be used. Measures which can be taken to prevent
corrosion include desalting of crude oils, neutralization of overhead vapors
with ammonia, and pH-control, and use of corrosion inhibitors in endan-
gered equipment sections.

Corrosion in cracking processes is caused by the following:


1. High-Temperature Sulfur Corrosion. Cr–Mo alloys in the furnace and
heat-exchanger tubes and Cr steel lining in the columns are suitable mate-
rials for corrosion prevention.
2. Erosion by Catalysts. Ceramic and concrete linings in vessels and cyclones
and regenerator internals made of Cr–Mo and austenitic Cr–Ni alloys are
suitable materials.

Corrosion in hydrogen-carrying processes is caused by hydrogen


attack at high temperatures (blistering and embrittlement of steels). Cr or Mo
steels, at higher temperatures austenitic Cr–Ni alloys, can be used to prevent
corrosion. In the presence of hydrogen and H2S, i.e., in hydrotreating and hydro-
cracking processes, steels with high chrome contents must be used.

4. Testing and Analysis

The chemical and physical properties of crude oils and their fractions depend on
the types and concentrations of their consituents. The typical properties of crude
oils and their products must be defined and determined by analytical methods, to
meet quality requirements for production, transport, storage and handling of
crude oils, and to assess specific quality criteria for further processing of products.
This quality testing is performed by regular laboratory analysis or by auto-
matic instruments.

4.1. Crude Oil and Product Properties. Density and API Gravity.
The density ranges of various crudes are as follows:

Paraffin-based crudes 815–835 kg/m3


Mixed-based crudes 835–860 kg/m3
Naphthalene-based crudes 860–955 kg/m3
352 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS Vol. 1

Sulfur

Fuel gas
Amines gas Propane
plant and
Claus plant Butane

Boiling from IBP


up to 170°C Hydro- Light naphtha
treater Reformer
Regular and
Crude oil
premium
feed
170-225°C Sweetening Kerosene
or HDS
225-375°C
Atmospheric
distillation
HDS Diesel

H2
Hydro-
cracking
Steam
reformer Vacuum
distillation Fluid
coking Coke

Fig. 10. Hydrocracking – coking.

The API gravity (introduced by the American Petroleum Institute) is the


internationally used density measure for crude oils. The relation between
degrees API and density at 15 8C is:

 141:5
API ¼  131:5
Density½kg=L

Distillation. The boiling range of crude oils is one of the most important
quality criteria for determining the yield distribution of the different product
fractions. Boiling analyses under atmospheric conditions can be supplemented
by distillation under reduced pressure for oil fractions that boil above 360 8C.
More detailed and precise results are obtained by applying fractional or true
boiling point (TBP) distillation.
Viscosity. The viscosity of crude oils is determined by the ratio of low and
high molecular mass constituents in the oil. The kinematic viscosity of crude oils
(at 38 8C ¼ 100 8F) can vary from 4 mm2/s (0.04 St) for light (North Sea) crudes to
700 mm2/s (7 St) for heavy (Venezuelan) crudes.
Flash Point. The flash point depends on the concentration of light
(low-boiling) constituents. For crude oils, flash points between  20 8C and
þ 80 8C are observed.
Vol. 1 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS 353

Pour Point. The pour point is the temperature at which an oil or oil pro-
duct ceases to flow. It is an important measure of its cold flow properties.
Other Properties. that are analytically tested include specific heat;
calorific value; contents of sulfur, nitrogen, oxygen, and inorganic compounds;
the C/H ratio; the hydrocarbon composition; and product-specific criteria (see
Section OIL PRODUCTIONS).
4.2. Testing Methods and Standards. In the oil industry a frame-
work of standard analytical procedures has been established, e.g., by ASTM,
DIN and the British Institute of Petroleum, (IP) standards. In addition,
various house standards have been introduced by most international oil
companies.

A selection of frequently applied DIN methods is listed below:

Density DIN 51757


Distillation (boiling range) DIN 51751
Vacuum distillation DIN 51356
Fractional distillation DIN 51567
Flash point DIN 51755, 51758, 51376,
51584, 53213
Viscosity DIN 53012, 51550
Cold properties (cloud DIN 51583, 51597, 51421
point, pour point,
freezing point)
Water content DIN 51582
Water and sediment content DIN 51793
Salt content DIN 51576
Sulfur content DIN 51400, 51768, 51409,
51418, 51450

Analytical testing by automatic continous analyzers has come into use


worldwide. The continuous analysis of process streams by automatic ‘‘onstream
analyzers’’ has proven to be an indispensable tool for the quality control of pro-
ducts in integrated refining systems.

5. Storage and Transport

Surface Storage. Crude oils are stored in cylindrical fixed-roof or


floating roof tanks; LPG is stored in spherical or cylindrical pressure vessels.
Suitable storage tanks for gasolines, kerosenes, and gas oils are cylindrial
floating-roof or fixed-roof tanks (optionally with vapor removal). Heavy fuel
oils and bitumen are stored in cylindrical fixed-roof tanks (optionally with
gas blanketing).
Underground Storage. Crude oil and liquid (eg, light heating oil), and
gaseous products (LNG) are stored in underground caverns. The under-
ground storage of crudes and liquid products in caverns is usually intended
354 OIL REFINING, ENVIRONMENTAL CONSIDERATIONS Vol. 1

as long term (emergency) storage; caverns for gas are used as buffer storage
(see NATURAL GAS).
Subsea Storage. Crude oils (from offshore production) and hydrocarbon
condensates (from offshore separation) are stored in undersea tanks.
Transport of Crude Oils and Products. Crude oils and bulk (mass)
products are normally transported by large tankers and pipelines. Finished
products are usually transported by smaller tankers and river barges, product
pipelines, rail tank cars, and road tank trucks.

REFERENCES

1. The Oil Refining Industry and the Environment, CONCAWE, The Hague, Sept.
1987.
2. O. Neuwirth: ‘‘Umweltfragen bei der Nutzung von Mineralöl,’’ Erlanger Forsch.
Reihe B; Naturwiss. Med. 14 (1984) 119–134.
3. Mineralöl und Umweltschutz, 3rd ed., Publ. Mineralölwirtsch. Verband, Hamburg
1989.
4. A. H. Edwards et al.: ‘‘Volatile Organic Compound Emissions in Western Europe,’’
CONCAWE Rep. 1986, no. 2.
5. A. de Roocker et al.: ‘‘Trends in Oil Discharged with Aqueous Effluents from Oil
Refineries in Western Europe,’’ CONCAWE Rep. 1985, no. 61.
6. C. L. F. Webb: ‘‘Biodegradation of Contaminated Soil,’’ Erdoel Erdgas 106 (1990)
no. 11, 453–455.
7. J. S. McArragher et al.: ‘‘Motor Vehicle Emission Regulations and Fuel Specifica-
tions, CONCAWE Rep. 1990, no. 2.
8. International Workshop Berlin: ‘‘Emissionsarme Betankung von Kraftfahrzeugen,’’
Erdoel Kohle Erdgas Petrochem. 42 (1989) 468.
9. R. J. Ellis: ‘‘Cost Estimating Aspects of Environmental Control Technology — HDS
as a Case Example,’’ CONCAWE Rep. 1988, no. 51.
10. Commission of the European Communities: Study on the Costs Borne by the Refining
Industry in the Member States in Order to Comply with Environmental Legislation,
prepared by Chem. Systems Int., London 1986.
11. G. Heinrich, Y. Jacquin, A. Vidal, Proc. World Pet. Congr., 10th, 1979, vol. 4, paper
PD 18–1, Heyden & Sons, London 1979.
12. F. G. Turpin, C. R. Lipuma, A. M. Edelman, in (60), paper PD 18–2.
13. C. T. Douwes, J. van Klinken, J. B. Wyffels, W. C. van Zijll Langhout, in (60), paper
18–3.

WALTHER W. IRION
Deutsche Shell AG, Hamburg,
Federal Republic of Germany
OTTO S. NEUWIRTH
UK Wesseling, Wesseling,
Federal Republic of Germany
Vol. 1 NATURAL GAS 355

NATURAL GAS
1. Introduction

Natural gas is a mixture of gaseous hydrocarbons with varying quanti-


ties of nonhydrocarbons, which are normally considered impurities. Natural
gas is often associated with liquid hydrocarbons (ie, condensate and crude
oil).
Natural gas formed during many epochs in the history of earth and
accumulated after migrating from its place of origin to subsurface deposits
in porous or fractured layers of many geologic formations. Natural gas is recov-
ered with wells that are drilled into these deposits. Worldwide annual pro-
duction amounts to ca. 2.8  1012 m3. Estimated gas reserves are on the order of
150  1012 m3.
1.1. Composition. Usual Composition. Natural gas is composed
mainly of methane, ethane, propane, and butane. In addition, it usually contains
minor quantities of heavier hydrocarbons and varying amounts of gaseous non-
hydrocarbons such as nitrogen, carbon dioxide, and hydrogen sulfide. A typical
composition is given in Table 1.
Main Impurities. In many areas the natural gas recovered is sour,
which means that it contains appreciable quantities of hydrogen sulfide and
carbon dioxide. Although larger quantities of carbon dioxide are undesirable
because they make the transport of natural gas more costly, diminish its com-
bustion qualities, and have a corrosive effect under certain conditions, carbon
dioxide can be tolerated in percentage levels. In contrast, hydrogen sulfide
concentrations in salable gas must be reduced below 3 ppm because of its
high toxicity. Furthermore, it is very corrosive in combination with water
and, when burned, produces sulfur dioxide, which is also toxic and corrosive.
Therefore, sour gas is generally desulfurized or sweetened to reduce the hydro-
gen sulfide content to traces (clean gas).
Higher Hydrocarbons and Water Vapor. Natural gas may contain
larger quantities of higher hydrocarbons than shown in Table 1 (so-called
rich or wet gas) and, in addition, is saturated with water vapor under natural
conditions. These constituents can condense at lower temperature and higher
pressure and thereby impair the transport and use of natural gas. Under
certain conditions, water and gas can form icelike solids, so-called hydrates
(see Section 1.2). Therefore, the proportion of condensable constituents is nor-
mally decreased substantially by refrigeration, adsorption, or absorption pro-
cesses before the natural gas (then called lean or dry gas) is transmitted
through pipelines or liquefied.
1.2. Physical Properties. Molar Mass. Table 2 shows the molar
mass of the commonly occurring components of natural gases. The apparent
molar mass Ma of a gas mixture such as natural gas is equal to the sum of the
molar fraction yi times the molar mass Mi of each component:

Ma ¼ Syi Mi (1)

ß2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim


10.1002/14356007.a17_073.pub2
356 NATURAL GAS Vol. 1

Table 1. Typical Composition Range of Natural Gas


Compound Molar fraction
Hydrocarbons
Methane 0.75–0.99
Ethane 0.01–0.15
Propane 0.01–0.10
n-Butane 0.00–0.02
Isobutane 0.00–0.01
n-Pentane 0.00–0.01
Isopentane 0.00–0.01
Hexane 0.00–0.01
Heptane plus higher hydrocarbons 0.00–0.001

Nonhydrocarbons
Nitrogen 0.00–0.15
Carbon dioxide 0.00–0.30
Hydrogen sulfide 0.00–0.30
Helium 0.00–0.05

Molar Volume and Ideal-Gas Law. The volume occupied by an ideal


gas depends on the temperature, pressure, and number of molecules present.
It does not depend on the type of molecules (ideal-gas law).
The volume of 1 kmol of an ideal gas at STP (273.15 K, 0.101325 MPa) can
be calculated, by using the ideal-gas law, to be 22.414 m3. The effective molar
volumes at STP for common natural gas components are shown in Table 2.
The calculation
P may also be employed for gas mixtures if the apparent molar
mass, Ma ¼ yiMi, is used in computing the number of kilomoles n.
Density and Relative Density. The density of several components of
natural gas is given in Table 2. The density of a gas mixture can be calculated
by the same procedure described in Equation (1). For deviating temperature
and pressure conditions, or when the ideal-gas law cannot be applied, the density
of a gas under any conditions can be calculated from the real-gas law:

pV ¼ znRT (2)

where z is the so-called compressibility factor (see page 5).


With Equation (2) the density of a gas can be calculated from the
relationship

% ¼ pM=zRT (3)

where V and n in Equation (2) are replaced by

% ¼ nM=V (4)

For a gas mixture such as natural gas, the molar mass M in Equation (3) is
replaced by the apparent molar mass, Ma, and z is the compressibility factor of
the gas mixture at the temperature and pressure conditions of interest.
Vol. 1 NATURAL GAS 357

Table 2. Physical Properties of Natural Gas Components at STP (4)


Molar mass M, Molar volume V, Density %, Relative density d
Compound kg/kmol m3/kmol kg/m3 ( air = 1)

Methane 16.043 22.360 0.7175 0.5549


Ethane 30.069 22.191 1.355 1.048
Propane 44.096 21.928 2.011 1.555
n-Butane 58.123 21.461 2.708 2.094
Isobutane 58.123 21.550 2.697 2.086
n-Pentane 72.150 * 20.90 * 3.452 * 2.670 *
Isopentane 72.150 * 21.06 * 3.426 * 2.650 *
n-Hexane 86.177 * 20.10 * 4.29 * 3.315 *
n-Heptane 100.203 * 18.3 * 5.48 * 4.235 *
Nitrogen 28.0134 22.403 1.2504 0.9671
Carbon dioxide 44.0098 22.261 1.9770 1.5290
Hydrogen sulfide 34.076 22.192 1.5355 1.1875
Helium 4.0026 22.426 0.17848 0.1380
*Compound is in liquid state at STP.

The relative density d (see Table 2) (traditionally designated specific


gravity) of a gas is the ratio of its density to the density of dry air at standard
conditions:
%gas
d¼ (5)
%air

For a gas mixture, the relative density can also be calculated from the
apparent molar mass Ma of the gas mixture divided by the molar mass of air
(28.963):

Ma
d¼ (6)
28:963

Critical Temperature and Pressure of Natural Gas Components.


The critical temperature Tc is the temperature above which a gas cannot be
liquefied even if high pressure is applied. At the critical temperature a gas
can be liquefied by application of a certain minimum pressure, the critical
pressure pc. The volume of 1 kg of a gas at this critical point is the critical
volume Vc; the reciprocal value of Vc is the critical density %c.
The critical constants for the primary constituents of natural gases are
given in Table 3. The critical temperature and pressure are constants for each
species.
Principle of Corresponding States. Properties of gases can be corre-
lated if they are compared at ‘‘corresponding’’ values of absolute temperature
and pressure. The critical state of a gas is used as reference point, and so-called
reduced values are obtained. The following terms are employed:

pr ¼ p= pc ; Tr ¼ T=Tc ; and Vr ¼ V=Vc

where Tc, pc, and Vc are the critical point conditions, and T, p, and V are the
values of the system, respectively.
358 NATURAL GAS Vol. 1

Table 3. Critical Constants of Natural Gas Components (5)


Compound pc, MPa Tc, 8C Vc, 103 m3/kg %c, kg/m3
Methane 4.60 82.55 6.1709 162
Ethane 4.88 32.25 4.9218 203
Propane 4.24 96.65 4.6035 217
n-Butane 3.80 152.05 4.3872 228
Isobutane 3.65 134.95 4.5248 221
n-Pentane 3.37 196.45 4.2134 237
Isopentane 3.38 187.25 4.2411 236
n-Hexane 2.97 234.25 4.2934 233
n-Heptane 2.73 267.05 4.3112 232
Nitrogen 3.39 146.95 3.1949 313
Carbon dioxide 7.38 31.05 2.1359 468
Hydrogen sulfide 8.94 100.05 2.8903 346
Helium 0.227 267.96 14.3143 69.9

If the principle of corresponding states is applied to a mixture of gases,


‘‘pseudocritical’’ temperatures and pressures are defined as the molar average
critical temperatures and pressures of the components:
X
ppc ¼ yi pci
X
Tpc ¼ yi Tci

where the subscript pc designates the pseudocritical values of the gas mixture,
ci the critical values of the components, and yi is the molar fraction of compo-
nent i. The pseudocritical pressure and temperature are then used to determine
the pseudoreduced conditions:

ppr ¼ p= ppc
Tpr ¼ T=Tpc

Although they are convenient for correlating purposes the pseudocritical values
have no physical significance.
Viscosity. For an ideal gas, the pressure and temperature dependence of
viscosity is opposite to that of liquids; the viscosity of an ideal gas increases with
increasing temperature and is independent of pressure. Actual hydrocarbon
gases, however, deviate from ideal behavior and approach that of liquids; their
viscosity increases with increasing pressure, and decreases with increasing tem-
perature at intermediate or high pressure.
Figure 1 is employed for estimating the dynamic viscosity uga of natural
gases composed primarily of hydrocarbons at atmospheric pressure (6). The pre-
sence of carbon dioxide, hydrogen sulfide, and nitrogen increases the viscosity.
The viscosity ug at any desired temperature and pressure can be obtained from
Figure 2, where the ratio ug/uga is plotted as a function of the pseudoreduced
temperature and pressure.
Compressibility of Real Gases. At higher pressure the attractive
or repulsive forces between gas molecules become important because the
Vol. 1 NATURAL GAS 359

Fig. 1. Viscosity of paraffin hydrocarbon gases at 101.3 kPa as a function of molar mass
(6). [(8F32)/1.8 ¼ 8C] (Copyright Society of Petroleum Engineers, 1954).

molecules possess finite volumes. The assumptions for ideal-gas behavior


are then no longer valid; consequently, the real-gas law (Eq. 2) was formulated
by introducing a proportionality term (the compressibility factor) to the
ideal-gas law. Compressibility factors for natural gases are determined by labora-
tory measurements, estimated by correlations, or calculated by equations of state.
Figure 3, which was developed by STANDING and KATZ (7) from data collected
on methane and natural gases in the range of up to 56 MPa and 1218C, presents a
correlation between compressibility factors and pertinent pseudoreduced
temperatures and pressures. Correlations are also available for pressures and
temperatures above these ranges (2,3)
The specific p–V–T relationship of fluids is described mathematically by
equations of state. These equations number in the hundreds, ranging from those
for a specific pure compound to generalized forms that claim to relate the proper-
ties of multicomponent mixtures. Well-known examples are the van der Waals
equation (8) with the modifications of REDLICH and KWONG (9), SOAVE (10),
and PENG and ROBINSON (11) and the virial equation. The van der Waals type of

Fig. 2. Viscosity ratio as a function of pseudoreduced temperature (6). (Copyright


Society of Petroleum Engineers, 1942).
360 NATURAL GAS Vol. 1

Fig. 3. Compressibility factor of natural gases as a function of pseudoreduced pressure


and temperature (7). (Copyright Society of Petroleum Engineers, 1942).

equation of state can be solved directly for the compressibility factor z. The most
reliable methods for typical natural gases are those of ROBINSON and JACOBY (12),
and HALL and YARBOROUGH (13).
Well-known virial equations are the Benedict–Webb–Rubin equation (14)
and the Starling–Hon (15) extension of the Benedict–Webb–Rubin equation.
The European Gas Research Group (GERG) and the van der Waals Institute
at Amsterdam carried out an extensive research project (16) to develop a virial
equation of state for the accurate prediction of compressibility factors for natural
gas mixtures. They used the virial equation solved for the compressibility factor z
and truncated after the third term:

z ¼ 1 þ BðTÞ%M þ CðTÞ%2M

where %M is the molar density, %M ¼ 1/VM, and B and C are the second and third
virial coefficients, which depend on temperature and gas composition. With this
Vol. 1 NATURAL GAS 361

Fig. 4. Phase diagram of a hydrocarbon system of fixed composition.

equation, compressibility factors can be predicted within the range 0 to 12 MPa


and 8 to 628C, which deviate <0.1% from experimental data.
Condensation Behavior of Heavier Hydrocarbons. Phase Diagram.
Figure 4 shows the phase diagram of a natural gas system. It represents the
phase changes with temperature and pressure for a particular constant
composition, and designates the pressure and temperature regions in which
only gas or liquids can occur and those in which they coexist at equilibrium.
The two-phase region or ‘‘envelope’’ is enclosed on the left side by the bubble
point curve up to the critical point C and from there on, via the points of maxi-
mum pressure Cp and maximum temperature CT, by the dew point curve. The
bubble point curve designates the points at which the first infinitesimal bubble of
gas is formed from the liquid phase, and the dew point curve indicates the points
at which the first infinitesimal drop of liquid condenses from the gaseous phase.
The critical point C ( juncture of bubble and dew point curves) is the point at
which all intensive properties of gas and liquid phases are identical. The pres-
sure pc and the temperature Tc at this point are the critical values of the gas mix-
ture. The cricondenbar point Cp is defined as the maximum pressure at which
two phases can exist in equilibrium. In natural gas systems this point lies
right of the critical point C. The cricondentherm point CT is defined as the max-
imum temperature at which two phases can exist in equilibrium.
Retrograde Condensation. The dashed lines within the two-phase envel-
ope of Figure 4 are liquid-content lines commonly represented by volume percent
liquid lines. They originate at the critical point C and are intersected by a line
362 NATURAL GAS Vol. 1

connecting the critical point C with the maximum temperature point CT. This
line separates the two-phase envelope in two regions. In the lower region the
gas mixture behaves as expected: At constant or decreasing temperature, the
liquid content in the gas mixture increases with increasing pressure. The beha-
vior of the gas mixture in the two-phase region above the connection line C–CT is
quite different. At constant temperature, the liquid content in the gas mixture
increases with decreasing pressure, and in the range between the critical pres-
sure and the maximum pressure, the liquid content increases at a pressure drop
even when the temperature increases.
This phenomenon, called retrograde condensation, may seriously impair the
transport of natural gas through pipelines because liquid slugs can be formed if
the higher hydrocarbons have not been removed sufficiently. The condensation
behavior of natural gas can be determined by employing equations of state or
by laboratory measurements (17).
Gas Hydrates. Gas hydrates are icelike solids composed of water and
hydrocarbons. Hydrocarbons ranging from methane to cyclopentane are known
to form hydrates. The formation of hydrates depends on pressure, temperature,
molecular size, and concentration of the component. Generally, hydrates form
at high pressure and low temperature, but they can occur as high as 308C and
below 0.7 MPa. Hydrate formation can be predicted from empirical vapor–solid
equilibrium ratios, empirical correlations, and laboratory measurements (18).
To avoid the formation of gas hydrates in pipelines, the water vapor content
of natural gas is commonly reduced by dehydration before transport. Lowering
the water dew point to 5 to 88C in relation to the maximum transmission pres-
sure, is a common stipulation in gas specifications.
1.3. Combustion Properties. Calorific Value. The calorific or heat-
ing value of natural gas is defined as the heat evolved by complete combustion
of a unit quantity of dehydrated gas with pure oxygen at 0.101325 MPa. The
initial temperature of gas and oxygen is 258C. The flue gas resulting from the
combustion process is cooled to the same temperature.
In determining the total calorific value, also called the gross calorific value
(higher heating value), all the water formed in the combustion process is con-
densed to the liquid state. For determination of the net calorific value (lower
heating value), all the water formed by the combustion reaction remains in the
vapor state: ie, the gross and net calorific values differ in the condensation
enthalpy of water. The calorific value is expressed in megajoules per cubic
meter or, occasionally, kilowatt-hours per cubic meter.
Calorific Value of Gas Mixtures. The calorific value of natural gas
can be calculated from the molar composition and the calorific values of the
components:
X
H¼ yi Hi

where yi is the molar fraction of component i, Hi is the gross or net calorific value
of component i. Table 4 shows calorific values for commonly occurring compo-
nents of natural gas.
Wobbe Index. When burned at the same pressure, natural gases with
different compositions generate the same amount of heat per time unit if their
Vol. 1 NATURAL GAS 363

Table 4. Calorific Values of Natural Gas Components at STP (19)


Gross calorific Net calorific
Compound value, MJ/m3 value, MJ/m3

Methane 39.819 35.883


Ethane 70.293 64.345
Propane 101.242 93.215
n-Butane 134.061 123.810
Isobutane 133.119 122.910
n-Pentane 169.19 * 156.56 *
Isopentane 167.53 * 154.99 *
n-Hexane 208.70 * 193.38 *
n-Heptane 265.22 * 245.99 *
Hydrogen sulfide 25.336 23.353
*Compound is in liquid state at STP.

calorific values divided by the square root of their relative densities (Wobbe
index) are equal. The Wobbe index can be referred to the gross or net calorific
value:
pffiffiffi
Wgross;net ¼ Hgross;net = d

Since gas burners tolerate only limited fluctuations in the heating value of fuel
gas, these fluctuations must be restricted. Therefore, natural gas specifications
generally stipulate ranges that are defined by Wobbe index rather than heating
values (20). Consequently, certain gas streams may require blending before
sales, or separate pipeline networks may be operated on different gas streams
with deviating Wobbe index levels.

2. Treating

Natural gas is produced at the well site in compositions of considerable variety.


The objectives of natural gas treatment are correspondingly diverse. The compo-
sition range that must be encompassed is shown in Table 5 for selected examples.
Besides its primary constituent methane, natural gas may contain various
amounts of impurities, which lead to varying problems during production and
transportation:

Water. Condensed water forms solid hydrates with hydrocarbons or hydro-


gen sulfide, or in addition leads to liquid slugs in pipelines and to corrosion (21).
Higher Hydrocarbons. If natural gas contains higher concentrations of
C2þ hydrocarbons, the recovery of liquid petroleum gas (LPG) and gas conden-
sate is economically important. The desired Wobbe index (see Section 1.3.2)
and calorific value could also require a reduction in the concentration of higher
hydrocarbons. Even very low concentrations of higher hydrocarbons can con-
dense in pipelines (retrograde condensation). These condensed hydrocarbons
can attack plastic pipes and control devices.
Hydrogen Sulfide. Hydrogen sulfide together with free water can cause
corrosion, particularly stress corrosion and hydrogen-induced cracking. Removal
Table 5. Composition of Natural Gas
Higher Carbon Hydrogen Elemental Organic
Methane, Ethane, Propane, Butane, hydrocarbons, Nitrogen, dioxide, sulfide, sulfur, sulfur, Helium, Mercury,
Source vol% vol% vol% vol% vol% vol% vol% vol% g/m3 ppm vol% mg/m3
Netherlands, 81 2.8 0.4 0.2 0.1 14.3 0.9 0.03 0.002
Groningen a
Norway, Ekofisk b 85 8.4 2.9 1 0.15 0.4 2 10 ppm 1
Federal Republic of
Germany, 85 1.5 0.07 0.03 2 g/m3C8+ 12.5 0.5 0.03 1–5
Söhlingenc
Indonesia, Arund 75 5.5 2.3 1.1 0.75 0.3 15 100 ppm 0.2
United States, 73.2 6.1 3.2 1.6 0.6 14.3 0.3 0.7
Panhandle, Tx.e
Federal Republic of 43 0.6 0.1 56 0.3 0.04
Germany,
Wustrowf
Federal Republic of 77 0.1 7 8 8 0.5–3 200

364
Germany,
Süd-Oldenburgg
former states of 42 8.5 5.2 3.3 22 0.8 2.6 16 1000 ppm
USSR, Tenguizh
Canada, Bearberryi 4 1 5 90 80
a
Sweet gas with low calorific value, only adjustment of water dew point.
b
Natural gas with high calorific value, H2S conversion to COS at molecular sieve.
c
Mercury removal, C3þ separation by low temperature separation and adsorption.
d
Liquefaction, separation of gas condensate, CO2 – H2S – Hg removal.
e
Helium production.
f
Nitrogen separation, CO2 removal.
g
Sour gas, use of sulfur solvent, H2S, CO2 scrubber, sulfur production.
h
Condensate, LPG, sulfur recovery, removal of organic sulfur compounds.
i
Ultrasour gas, only for sulfur production.
Vol. 1 NATURAL GAS 365

of hydrogen sulfide is performed almost exclusively in centralized treatment


plants.
Carbon Dioxide. Carbon dioxide together with free water causes pitting
corrosion in carbon and low-alloy steels. Some natural gases that contain signif-
icant amounts of CO2 must be treated to increase the methane concentration
prior to sales.
Sulfur. Gases containing hydrogen sulfide can also contain elementary sul-
fur as vapor. Some gas fields, primarily in Canada, Germany, and the United
States, contain such high quantities that the sulfur, depending on pressure, tem-
perature, and gas composition, can precipitate and plug the production pipe, which
then becomes blocked. Furthermore, elemental sulfur and free water are corrosive.
Mercury. Natural gas can contain mercury in concentrations up to
several milligrams per cubic meter, the bulk of which exists in elemental form.
Separated liquid mercury causes mercury-induced corrosion in pipes and fit-
tings, corrosion damage to aluminum heat exchangers in cryogenic plants, and
damage to measuring and control valves containing nonferrous metals by amal-
gam formation. Mercury must also be removed because of its toxicity.
Other Components. In Alabama, for example, cyclic hydrocarbons such
as adamantane and diamantane are found in natural gas; these compounds are
inclined to sublime and must be removed by injection of oil.
Some gases contain high amounts of nitrogen, which must be separated to
obtain a sales gas with the required combustion properties.
Recovery of Components with Economic Value. Another important
factor is whether additional components must be removed to meet product
specifications, or whether the recovery of these products is of economic interest.
Examples of the latter include the recovery of helium, the production of sulfur
from sour gas, and the recovery of LPG and hydrocarbon condensates. A special
case is so-called ultrasour gas (see line 9 of Table 5). In this case the main inter-
est is not to produce a fuel gas but rather to provide a feedstock in the production
of sulfur. In Canada, production of gas with a hydrogen sulfide content of 90%
began. In addition, the removal of carbon dioxide for enhanced oil recovery in
crude oil production should be mentioned.
By contrast, if these components are present only in trace amounts they are
not removed because of economic reasons, but rather to avoid corrosion, environ-
mental, or production problems.
The individual process by which undesired gas contaminants can be elimi-
nated depends on the composition of the gas, the amount and type of contami-
nant, and ultimately whether a salable pure gas must be delivered at the well
or whether the gas is piped to a processing plant.
2.1. Objectives of Natural Gas Treating. Two objectives must be met
in treating natural gas:

–Adjustment to the required quality standards


–Recovery of byproducts

The main objective worldwide is production of a sales gas of pipeline stan-


dard. In this respect, both the extent of removal of undesirable components and
the range of conditioning measures are predetermined.
366 NATURAL GAS Vol. 1

Natural gas treating can take place both directly at the well and in
centralized plants. Treating at the well is always necessary when the gas cannot
be transported to a processing plant by pipeline without hazard. To avoid corro-
sion, sour gas is often dehydrated at the well so that free water can no longer
condense in the pipeline. The recovery of sulfur and the separation of nitrogen,
carbon dioxide, or helium are carried out in centralized plants, to which streams
of different wells are fed. The same holds true for liquefaction of natural gas,
recovery of LPG, and removal of higher hydrocarbons.
2.2. Natural Gas Wells and Field Installations. Installations at indi-
vidual gas wells must fulfill the following objectives:

to guarantee trouble-free gas flow in the production pipes and surface


equipment;
to separate or, if necessary, treat coproduced liquids;
to condition the gas for transportation to the processing plant;
to meet the required feed quality for the pipeline distribution network; and
to protect the production pipes and field installations from corrosion.

The type and design of plants are determined by:

type and concentration of accompanying substances in the gas;


behavior of the gas during pipeline transportation; and
applicable sales specifications.

Basic Flow Diagram of Natural Gas Treating. A flow diagram for


the most important treatment steps is shown in Figure 5. To prevent deposition
of elemental sulfur, which can occur, for example, in some sour gas wells, sol-
vents are pumped into the production pipe. Corrosion inhibitors can also be
introduced in the same way. On the surface, liquids condensed under the
well-head flow conditions are separated. After being depressurized and cooled
to collection pipe conditions, further condensate is precipitated. If necessary,
separated condensate must be treated further ( phase separation, stabilizing,
stripping). If large amounts of formation water occur or considerable amounts
of liquids (inhibitors, sulfur-dissolving agents) are injected during sour gas
production, hydrogen sulfide may be released during depressurization and trans-
portation of these liquids. Emissions can be minimized by high-pressure liquid
stripping or by recompression of the flashed gas. When no further condensable
hydrocarbons are present, the gas is dehydrated to a dew point equal to or
below the dew point specified for pipeline transportation.
Whereas sweet gas can be fed directly from the well site to the distribution
network, sour gas is usually treated further in a centralized plant serving several
gas fields. If pressure at the production well is low, compressors are installed
upstream of the pipeline network or treatment plant.
For wet gases (see Section 1.1), both the dew point of hydrocarbons and that
of water must be lowered. After the injection of hydrate inhibitors (eg, glycols,
methanol), water vapor and higher hydrocarbons are condensed by reducing
the temperature (22).
Vol. 1 NATURAL GAS 367

Fig. 5. Treating of natural gas. —— Flow sheet for treatment of sweet gas.– – –
Additional steps required for treatment of sour gas.

If reduced pressure in the downstream pipeline system causes retrograde


hydrocarbon condensation, adsorption processes are used for further removal
of higher hydrocarbons (23).
Various adsorbents are available to remove traces of mercury; activated
charcoal impregnated with sulfur is particularly useful.
2.3. Processing Steps During Production. Use of Sulfur Solvents.
If elemental sulfur is present in the reservoir, sour natural gases are enriched
with sulfur to a greater or lesser extent. The ability of the gas to absorb sulfur
depends strongly on pressure, temperature, and gas composition (24). Cooling
and pressure reduction during production minimize the solubility of sulfur and
lead to its precipitation; at low temperature, solidification and blockage of the
pipe may occur. These problems are generally overcome by injection of suitable
solvents into the production equipment which permits sulfur to be removed
with the injected liquid, either chemically bound or physically dissolved. Two dif-
ferent reliable processes are applied:
368 NATURAL GAS Vol. 1

Polysulfide Process (25). By the addition of aqueous ammonia or mono-


ethylamine solutions, sulfur is dissolved, in the presence of hydrogen sulfide, as
polysulfide and separated at the surface in a liquid separator. The sulfur-laden
liquid can be regenerated at ca. 1508C. Ammonia or amine, water, hydrogen
sulfide, and carbon dioxide are dissolved in the condensed water and reinjected
into the well. Sulfur is precipitated as a liquid and removed from the bottom of
the regenerating column.
With this process, only a single liquid phase occurs above ground. Traces of
ammonia or amine cause no problems in the downstream treatment. For gases
with a high carbon dioxide content the process can be employed only under cer-
tain conditions. In this case the alkaline media cannot dissolve sufficient hydro-
gen sulfide for the formation of polysulfide. This solvent also cannot be employed
in the presence of formation water containing alkaline-earth metal chlorides
because the calcium carbonate and magnesium hydroxide formed cause
blockages that can be dissolved only by acid treatment. Specific measures are
necessary to prevent down-hole tubing corrosion.
Physical Solvents (26). In the production of sour gas, sulfur may also be
physically dissolved by means of suitable hydrocarbons. A suitable solvent must
have the following properties: no reaction with formation water;

stability under reservoir conditions;


no reaction with sulfur;
high solvent power for sulfur;
very low vapor pressure; and
noncorrosiveness.
Spindle oil and mixtures of spindle oil with special aromatic coal tar fractions,
which have a very high solvent power for sulfur, are the most common solvents for
this purpose. These oils simultaneously serve as carriers for corrosion inhibitors.
Sulfur-laden oils can be regenerated, for example, by extraction with amine
solutions that also contain hydrogen sulfide. After removal of sulfur by formation
of polysulfides these solutions can be regenerated as already mentioned. Phase
separation may be hindered by formation of emulsions of the oil with the simul-
taneously separated water, making specific measures necessary.
Typical concentric piping for sour gas production is shown in Figure 6, in
which the solvent is pumped into the well through the outer concentric pipe
and returns to the well head with the gas production stream. A sour gas well
with solvent injection and the necessary regeneration plant is depicted schema-
tically in Figure 7.
Removal of Mercury. Natural gas from predominantly ‘‘Rotliegenden’’
reservoirs often contains mercury in vapor and/or aerosol form (27). Natural
gas fields containing mercury are known in Algeria, Indonesia, The Netherlands,
and Germany. Mercury has not resulted in any problems in production from
sour gas fields. To avoid health hazards during treatment and use, the
mercury content is lowered in separate mercury-removal plants from 5 mg/m3
to <10 mg/m3.
A considerable portion of the mercury has often already been removed by
low-temperature separation (LTS) in the dehydration unit at the well. Mercury
Vol. 1 NATURAL GAS 369

Fig. 6. Schematic of a typical sour gas well completion. a) Well head; b) Anchor casing;
c) Cement; d) Intermediate casing; e) Injection annulus; f ) Tubing; g) Production packer;
h) Liner; i) Perforations

levels of <5 mg/m3 are attained in downstream chemisorption fixed-bed reactors


with, for example, sulfur-impregnated activated carbon. Condensable compo-
nents, such as water or hydrocarbons, reduce the loading ability for mercury
or even deactivate the adsorbent. The reactors are not regenerated; rather, the
laden adsorbent is fed to an external treatment plant. Regenerative amalgama-
tion processes are currently being tested.
Mercury is also collected from the liquid hydrocarbon stream by phase
separation. Further treatment with solid adsorbents reduces the mercury con-
tent from 300 to 20 ng/g.
Dehydration. In most cases, natural gas is dehydrated close to the gas
wells for safety reasons. The formation of gas hydrates in high-pressure gas pipe-
lines can thus be avoided and problems arising from liquids in the transportation
system minimized. In general, a water dew point of 88C at the pipeline
pressure is adequate. In particular, in the presence of carbon dioxide, hydrogen
sulfide, and chlorides, corrosion can be minimized by decreasing the water con-
tent. In some gas fields, dehydration is not employed; production is maintained
370 NATURAL GAS Vol. 1

Fig. 7. Flow diagram of a sulfur solvent inhibitor system. a) Lean oil storage; b) Sour gas
well; c) Free water knockout; d) Liquid control valves; e) Flow control valve; f ) Cooler; g)
Separator; h) Stripper; i) Liquid storage; j) Phase separation; k) Extraction unit; l) Phase
separator; m) Ethylamine polysulfide; n) Ethylamine regeneration; o) Ethylamine storage.

by the use of specific corrosion-resistant materials together with corrosion inhi-


bitors and thermally insulated equipment.
Cryogenic Dehydration. To decrease the higher hydrocarbon content in
addition to maintaining a specific water dew point, cryogenic dehydration of nat-
ural gas is often employed (see Fig. 8). The wet gas is cooled until the components
to be removed precipitate by condensation or formation of hydrates. Methanol, gly-
col, or a paraffin solvent is often also injected. The reduction in temperature can be
achieved by Joule–Thomson expansion. This route is employed particularly in the
first phase of production in high-pressure wells. When the production pressure is
close to that of the transportation system, external cryogenic units are used.
Dehydration by Absorption Processes (28). Standardized dehydra-
tion plants using glycol absorption are employed most widely. Triethylene glycol
(TEG) is used in preference to other glycols (mono- and diethylene glycols)
because of its high absorption capacity for water vapor, its low vapor pressure
(small losses from evaporation), and its high thermal stability. This type of dehy-
dration plant is shown in Figure 9.
In the absorber (h), glycol and gas are brought in contact countercurrently.
Plant pressures of 2–10 MPa are typical.
Vol. 1 NATURAL GAS 371

Fig. 8. Dehydration with a hydrate inhibitor and cooling at a high-pressure well.


a) High pressure separator (free water knockout); b) Heater; c) Air cooler; d) Separator;
e) Joule–Thomson valve; f ) Gas–gas heat exchanger; g) External cooling system for
low-pressure wells.

In the approximately pressureless regeneration step, water-laden glycol is


heated to 2008C. The energy requirement is reduced by extensive heat exchange
between laden and regenerated glycol in a coiled tube in the storage vessel and in
the top of the regeneration column. Both direct gas-fired regenerators and those
with steam or electric heating are used. The residual water content (ca. 1%) in
the glycol after thermal regeneration can be reduced still further by stripping
with dry gas or applying a vacuum. The stripping medium, often sweet gas,
is passed through the storage vessel to be heated in the spiral tube before the

Fig. 9. Glycol dehydration. a) Well head; b) Corrosion inhibitor–sulfur solvent injection


pump; c) High-pressure separator (free water knockout); d) Heater; e) Flow control valve;
f ) Air cooler; g ) Separator; h) Glycol absorber; i) Demister; j) Glycol–gas heat exchanger; k)
Glycol pump; l) Glycol reboiler; m) High-concentration stripper; n) Glycol storage tank; o) Gly-
col stripper; p) Water–solvent stripper; q ) Glycol filter; r) Produced liquids storage tank; s)
Flare.
372 NATURAL GAS Vol. 1

Fig. 10. Attainable dew points of natural gas as a function of the (gas) contact tempera-
ture with triethylene glycol–water solutions.

desorption of the water vapor in the distillation section is increased. A residual


water content in the glycol of <0.3% can be achieved (29). The dew points attain-
able for natural gas in equilibrium with a triethylene glycol–water mixture are
given in Figure 10 as a function of contact temperature. The plants are self- con-
trolled; the dehydration process is maintained by indirect control of the glycol cir-
culation rate, the regeneration temperature, and the stripping gas rate (30).
Enrichment of hydrocarbons in the glycol can be limited by the use of oil
skimmers. Degradation products resulting from entry of oxygen, thermal decom-
position, or influx of acids from bore hole treatment render glycol corrosive. Cor-
rosion control is usually carried out by the addition of inhibitors that increase the
pH. The glycol is usually filtered through activated charcoal. The service life of
glycol can be several years (31).
Further measures are necessary for dehydration of sour gas, as shown by
the dashed lines in Figure 5. At the usual plant pressure of 10.0 MPa, glycol
absorbs considerable amounts of hydrogen sulfide, in addition to water vapor.
The former is released during regeneration of the glycol, thereby leading to
corrosion and emission problems. High-pressure strippers are therefore installed.
At a pressure slightly above that of the glycol absorber, hydrogen sulfide is
removed from glycol by stripping with heated sweet gas. The sweet gas often ori-
ginates from neighboring fields or plants. The strip gas, enriched with hydrogen
sulfide (representing < 2% of the dehydration unit capacity), is recycled to the
absorber. Glycol dehydration units are constructed of normal carbon steel.
In the final phase of gas production from a well, considerable amounts of
salt-containing formation water are coproduced. During flashing and transporta-
tion of these laden liquids, gases are released, including hydrogen sulfide and
benzene–toluene–xylene (BTX). Emissions can be minimized and risks reduced
by additional stripping or by recompression of these flash gases.
Vol. 1 NATURAL GAS 373

Fig. 11. Hydrocarbon removal with adsorption and regeneration under pressure.
a)–c) Adsorbers; d) Heater; e) Cooler; f ) Chiller; g) Separator; h) Phase separator.

Adsorptive Dehydration. In adsorptive dehydration the gas is brought


in contact with molecular sieves, silica gel, or SORBEAD (ie, Na2O- containing
SiO2). Plants with only two adsorbers, working alternately, are possible. Dew
points < 708C are attainable with adsorption plants. This is particularly neces-
sary for cryogenic plants and liquid natural gas (LNG) plants, where traces of
water and carbon dioxide can lead to blockage by ice formation.
For regeneration, ca. 5% purge gas is drawn from the main gas flow and fed
through the laden adsorber at plant pressure and 200–3308C (Fig. 11). Removed
water vapor from purging of the fixed bed is condensed so that the purge gas,
which is at plant pressure, can be recycled. Regeneration by depressurization
is seldom employed in natural gas treating.
Under the normal requirements of natural gas treating dehydration units
of this type are economically less attractive than glycol and cryogenic dehydra-
tion plants described above. However, they are nevertheless employed when, in
addition to simply removing water, adjustment to a specific hydrocarbon dew
point is required, removal of hydrogen sulfide or carbon dioxide is necessary,
or an extremely low dew point is required.
The adsorbent lifetime depends mainly on the amount of impurities (eg,
salt, inhibitors, sulfur compounds, and waxy hydrocarbons) that originate from
well production. Conversion of sulfur compounds to hydrogen sulfide or carbonyl
sulfide in the adsorber has been observed. Adsorption plants are often used in
offshore fields to minimize the need for scrubbing solvents.
Removal of Hydrocarbons. Recovery of hydrocarbons is mainly done for
economic reasons. However, removal may also be necessary to avoid uncontrolled
appearance of hydrocarbon condensates in long-distance pipelines and low-
pressure distribution systems in cities (32,33). Such condensates in pipelines
may produce liquid slugs, especially during varying loads, which lead to
errors in downstream measuring and controlling devices. In addition, plastic
374 NATURAL GAS Vol. 1

Fig. 12. Phase diagram for natural gas containing higher hydrocarbons and for pure
methane. a) Dew point curve; b) Boiling point curve; c) Critical point; d) Cricondentherm
point.

pipelines or plastic components in meters are adversely affected. Furthermore,


condensates impair leak detecting by use of odorants. Therefore, in addition to
a water dew point, a hydrocarbon dew point is often specified. Thus, condensa-
tion of hydrocarbons is avoided at ground temperature and typical pressure
ranges of the transportation system (10–0 MPa) (34).
Higher hydrocarbons strongly influence the dew point curves of natural gas
and also determine the conditions under which retrograde condensation can
occur (see Section 1.2). As an example, the dew point lines of an untreated and
a treated natural gas are shown in Figure 12. The vapor pressure curve of pure
methane is also shown.
The required hydrocarbon dew point cannot be attained with cryogenic
processes. Industrially, adsorption plants with activated charcoal or SORBEAD
are used, in which the higher hydrocarbons are removed and recovered prefer-
entially at elevated regeneration temperatures and short cycle times
(see Fig. 11). Presence of water causes problems and is usually removed first
in upstream dehydration plants (32).
2.4. Removal of Carbon Dioxide and Sulfur Components. Removal
of the sour gas components hydrogen sulfide and carbon dioxide is one of the
most common aims of natural gas treating. Furthermore, other sulfur com-
pounds (eg, COS or mercaptans) are frequently present and must also be
removed. The quality demands on purified gas and the anticipated type of proces-
sing or further use of the separated components are quite varied. As a result, a
choice of many different processes is available (35).
Absorption Processes. The method with the broadest application is gas
scrubbing. Here, the sour gas components are reversibly bound to a solvent by
chemical or physical absorption. In the regeneration step the components are
desorbed unchanged and the solvent is recycled to the scrubber.
Physical Absorption Process (35–37). If selective removal of sulfur
compounds is required for natural gases with a high content of sour components,
Vol. 1 NATURAL GAS 375

Fig. 13. Physical scrubbing with regeneration by depressurization and stripping.


a) Absorber; b) Solvent recovery; c) Stripper; d) Recompressor.

physical absorption processes are preferred. In physical absorption, carbon


dioxide and hydrogen sulfide are only physically dissolved in the solvent.
The process is reversible: during regeneration under strongly reduced pres-
sure, absorbed gaseous components are released unchanged from the solvent
(Fig. 13).
Important processes include the following: (1) the Purisol process, which
operates with N-methylpyrrolidone (NMP); (2) the Selexol process, in which
the solvent is a mixture of poly(ethylene glycol dimethyl ether)s; (3) the Sepasolv
process, with poly(ethylene glycol methyl isopropyl ether); (4) the Fluor solvent
process, with propylene carbonate; and (5) the cold methanol process (Rectisol).
With the exception of the Rectisol process (which operates at 10 to 308C),
all operate at ambient temperature and employ high-boiling organic solvents
(Table 6). The absorption properties for carbon dioxide are similar; the selectivity
of NMP for hydrogen sulfide is higher, which is often desirable.
Physical scrubbing agents are usually anhydrous; thus both dehydration
and removal of carbon dioxide and/or hydrogen sulfide can be carried out in a sin-
gle step; the separate dehydration step usually required in chemical absorption

Table 6. Characteristic Properties of Physical Solvents


Propylene
Property Purisol Selexol carbonate Sepasolv Rectisol

Vapor pressure (258C), Pa 53.2 0.097 11.3 0.049 13 000


Solubility of CO2 (258C), m3/m3 3.54 3.6 3.38 3.38 13.43 *
Solubility relative to CO2
CH4 0.072 0.067 0.038 0.066 0.051
C3þ 1.07 1.02 0.51 2.35
COS 2.72 2.33 1.88 2.54 3.92
H2S 10.2 8.53 3.29 6.86 7.06
*at 258C.
376 NATURAL GAS Vol. 1

process is superfluous. At a water content of ca. 3–5% in the scrubbing agent (eg,
Selexol), it is also possible to regenerate without a strip gas compressor; the
stripping agent is then the water vapor present.
To reduce losses and prevent condensation of the scrubbing agent in the
pipeline, small posttreatment units are often installed (eg, with glycol). Washers,
activated carbon adsorbers, or refrigeration systems, are also possible. Because
of the good solubility of higher hydrocarbons in these physical solvents, these
processes also allow the recovery of LPG.
Amount of Scrubbing Agent. The circulating amount of scrubbing
agent required is almost independent of the concentration of the components
to be removed but does depend on pressure and raw gas throughput. Loading
of the physical scrubbing agent increases with the partial pressure of the compo-
nents to be removed: Thus, physical absorption is used advantageously at higher
concentration of these components and higher process pressure. A large amount
of coabsorbed methane can be released in the first depressurization step by flash-
ing (usually multistage flashing) and can be fed profitably into internal fuel-gas
systems.
A high selectivity for hydrogen sulfide compared to carbon dioxide leads
to process streams with high hydrogen sulfide concentration, which can be
used in the Claus process. A certain residual amount of carbon dioxide in the
treated gas may be desirable to adjust its calorific value to required specifica-
tions. To match quality specifications, physical scrubbing processes can be
operated with variable selectivity by adjusting circulation of the solvent and
temperature.
Physical scrubbing processes are noncorrosive, and regular carbon steel is
generally used as construction material. Complete removal of carbonyl sulfide
from the crude gas requires sophisticated equipment.
Chemical Absorption Processes (35,38–40). Alkanolamine scrubbing,
known since 1930, is currently one of the most frequently used processes
(! Gas Production). Aqueous solutions of monoethanolamine (MEA), diethanola-
mine (DEA), diisopropylamine (DIPA), diglycolamine (DGA), and methyldietha-
nolamine (MDEA), in particular, are used as absorbents. The choice of solvent is
determined by different factors.
Monoethanolamine allows a high final purity but requires a considerable
amount of energy for regeneration, and absorption of carbonyl sulfide leads to
degradation products that must be eliminated in a reclaimer (see Section 2.8).
Selective removal of hydrogen sulfide is not possible. Diglycolamine, like MEA
a primary amine, is suitable for the complete absorption of both hydrogen
sulfide and carbon dioxide and can also be used for scrubbing of carbonyl sul-
fide and mercaptans. The secondary amines DEA and DIPA show a lower heat
of reaction for absorption–desorption of sour gas components; as a result,
they require less regeneration steam than MEA. At higher carbon dioxide
levels, various degradation compounds are formed that cannot be regenerated,
and require use of a reclaimer (see Section 2.8). A certain selectivity for hydro-
gen sulfide is possible, particularly with DIPA, because its reaction with
carbon dioxide is slower. Tertiary amines such as MDEA have considerably
better selectivity for hydrogen sulfide. The heat of reaction with MDEA is
lower, degradation products are not formed with carbonyl sulfide and carbon
Vol. 1 NATURAL GAS 377

Table 7. Comparison of chemical solvents


Heat of reaction, Relative CO2
kJ/kg absorption, Capacity *
Absorbent H2S CO2 % H2S CO2

MEA 1905 1920 100 0.07 0.50


DEA 1190 1510 40 0.09 0.32
DIPA 1140 2180 20
MDEA 1050 1420 10 0.10 0.12
*Moles of sour gas per mole of amine for solutions with 1.3 mol of amine per kilogram water.

dioxide, and only a few corrosion problems arise. If gases at low pressure are
purified, the normal pipeline specification for hydrogen sulfide is not attained.
Comparative data for characterization of various alkanolamines are presented
in Table 7.
Addition of corrosion inhibitors and special chemically active components
leads to advanced processes that allow higher alkanolamine concentration and
higher loading, for example, or better selectivity for hydrogen sulfide and, there-
fore, lower circulation rates. A lower corrosion tendency and a reduction of
energy required for regeneration can thereby be obtained. Examples are the
Amine Guard, Econamine, activated MDEA (BASF), Gas/Spec, and Ucarsol
processes (41,42). Selectivity can be increased, and further savings in energy
can be obtained by the use of sterically hindered amines (Flexsorb, Ucarsol
Innovator Solvent 111). The energy consumption for the hydrogen sulfide/carbon
dioxide or carbon dioxide scrubbing system is also lower with these amines
(43,44).
Activated and inhibited potassium carbonate solution (Benfield, Catacarb)
is widely applied for purification of natural gas (45). The process can in principle
be used to remove carbon dioxide, hydrogen sulfide, carbonyl sulfide, and carbon
disulfide. However, it is used mostly in treating gases with high carbon dioxide
content. The steam requirement for regeneration can also be reduced by addition
of hindered amines.
Physical–Chemical Absorption Processes. The advantages of chemical
absorption (eg, removal of COS, low coabsorption of methane) are combined with
those of physically effective organic solvents (eg, high loading at high partial
pressure). As amines, DIPA and MDEA are preferentially employed. The physi-
cal absorbents are high-boiling solvents (eg, Sulfolan or Selexol). In addition,
5–30% water is present. The Sulfinol process has found widespread use (46,47)
and has proved useful in the United States, Canada, and the Federal Republic of
Germany for simultaneous removal of hydrogen sulfide, carbon dioxide and
organic sulfur compounds. However, the coabsorption of hydrocarbons is also
increased by a physical solvent, so that an intermediate flash step with reabsor-
ber for the recovery of sulfur-free fuel gas (shown as a dashed line in Fig. 14) is
indispensable. Selective hydrogen sulfide scrubbing can be attained with this
process by proper choice of the amine.
Other physical–chemical absorption processes include Selefining, Flexsorb
PS (44), and Amisol. With the Amisol process a mixture of methanol and
378 NATURAL GAS Vol. 1

Fig. 14. Schematic of physical-chemical scrubbing with hot regeneration. a) Absorber;


b) Air cooler; c) Regenerator; d) Air condenser; e) Intermediate flash stage with reabsorber;
f ) Reboiler; g) Phase separator; h) Expansion turbine.

alkanolamines or aliphatic alkylamines is used (48). Like in all chemical scrub-


bing processes, the gas is saturated with water vapor in the absorption step,
making a downstream dehydration unit necessary.
Selection Criteria for Absorption Processes (35,38,40,49). Selection
of the best suited process is essentially determined by the pressure and com-
position of the raw natural gas, the trace components, and the desired quality
of marketable natural gas and separated sour gas components. Investment
and operating costs are also important. The design of the absorber and the
amount of solvent required are determined by the absorption behavior of
the least soluble compound to be removed. Of the investment cost, 50 – 70%
are directly related to the required amount of scrubbing agent in circulation.
In contrast, operating costs are determined to about 70% by the energy
requirements for regeneration and recirculation of the scrubbing agent. In
view of the many parameters to be considered, a simple selection of a process
by means of a diagram or a table is not possible. Only a few factors can be men-
tioned here.
Raw Gas Composition. Physical absorbents are generally not used for
the purification of natural gas that contains higher hydrocarbons. The solubility
of butane, for example, in most solvents is similar to that of hydrogen sulfide.
Selective removal of the latter, therefore, is not possible. Hydrocarbon solubility
in aqueous chemical scrubbing agents is low. Unfortunately, some chemical
absorbents react with gas components (eg, COS, CO2, and CS2) to form corrosive
products, whereas physical solvents remain unchanged.
Gas Throughput. Physical scrubbing systems are in general more com-
plex and, therefore, require higher investment. For this reason, chemical scrub-
bers are often more economical for small gas streams if no side reactions with the
chemical absorbent occur. For large plants, the low operating costs of physical
absorption processes outweigh the higher investment costs.
Vol. 1 NATURAL GAS 379

Pressure. In chemical absorption processes the capacity of the solvent is


determined by reaction of the sour gas components with the active solvent com-
ponent; ie, saturation is reached at relatively low pressure. The circulation rate
is then proportional to the quantity of the components to be removed and thus
also to the total pressure. For physical scrubbing agents, loading increases
with pressure, and the circulation rate is inversely proportional to the total
pressure.
Claus Gas. In most cases, the sour gas removed is fed to a Claus plant for
recovery of elemental sulfur. For this purpose, the gas should have as high a
content of hydrogen sulfide and as low a content of hydrocarbons as possible.
If the gas to be purified contains a high level of carbon dioxide in addition to
hydrogen sulfide, a selective absorption process is preferable. This requirement
is better met by physical than chemical processes. If strip gas is used for regen-
eration, the hydrocarbon content of the Claus gas should be lowered by inter-
mediate flashing stages and recompression.
Sales Gas. When hydrogen sulfide and carbon dioxide must be simulta-
neously and almost entirely removed, chemical or physical–chemical scrubbing
processes can be attractive. The presence of organic sulfur compounds excludes
some solvents. If only partial removal of carbon dioxide is required, selective
amine absorption processes are used at low pressure, and physical processes
are preferred at high pressure.
Special solvent properties such as vapor pressure, viscosity, toxicity, heat of
absorption–desorption, and corrosion behavior must be considered, along with
foaming tendency, water-absorbing capacity, pricing, and environmental impact.
The decomposition products that must be disposed of can often be reduced by a
reclaim and recovery unit.
Liquid Oxidation Processes. The conventional absorptive removal and
separate processing of hydrogen sulfide is often too expensive, particularly
when the gas stream to be treated is relatively small, the hydrogen sulfide con-
tent is low, or the pressure is low. In this case selective liquid oxidation processes
can be used (50).
The following steps are generally involved in this process:

Absorption of hydrogen sulfide in alkaline solution


Oxidation of dissolved hydrogen sulfide ions to elemental sulfur with simulta-
neous reduction of a chemical oxygen carrier or microoganisms
Reoxidation of the active component with air, in special processes with the aid
of bacteria
Separation of elemental sulfur by filtration, centrifugation, flotation, or sedi-
mentation and, if necessary, further purification.
The processes are differentiated from each other only by the specific redox
system employed. A flow sheet is shown in Figure 15. The sour gas to be purified
is brought in contact with the absorption solution in a venturi scrubber or an
absorption column. The liquid, which may now contain elemental sulfur in colloi-
dal form, is then treated with air to complete the formation of elemental sulfur
and to reoxidize the reduced oxygen carrier which is then recycled to the absorp-
tion step. The sulfur sludge is collected and treated further.
380 NATURAL GAS Vol. 1

Fig. 15. Schematic of a liquid oxidation process. a) Absorber; b) Oxidizer; c) Settling


tank; d) Centrifuge; e) Recirculating pump; f ) Air blower; g) Chemical additive pump;
h) Sulfur slurry pump; i) Sulfur melter.

The following processes are used:

1) Stretford, Sulfolin, Unisulf


Vanadates are employed for oxidation of the absorbed hydrogen sulfide.
Organic oxygen carriers (eg, anthraquinonedisulfonic acids) are required
to reoxidize the vanadium compounds. Optimized formulations suppress
side reactions (eg, formation of sulfate and thiosulfate) and lower the
consumption of chemicals.
2) Takahax, Hiperion
For these processes, naphthoquinonesulfonates or chelates are used as
oxidizing agents. Some environmental problems associated with the use
of vanadates can thus be avoided. In both groups of processes, hydrogen
sulfide is absorbed very selectively; they are also suitable for gases with
a high carbon dioxide content.
3) Lo-Cat, SulFerox, Sulfint, Bio-SR
After absorption in a solution of NaHCO3 –Na2CO3, hydrogen sulfide is oxi-
dized with iron(III) aminopolycarboxylic acid chelates. The resulting
iron(II) chelates are reoxidized with air. Differences in composition of the
solutions employed in individual processes influence the absorption capa-
city for hydrogen sulfide, the side reactions, the selectivity of absorption
in the presence of carbon dioxide, and the ease of removal of sulfur parti-
cles. The Bio-SR process works at pH 2 and without chelates. Iron(II)
sulfate is oxidized by Thiobacillus Ferroxidans (81).
4) Sulfa-Check
In this batchwise process, hydrogen sulfide is oxidized to elemental sulfur
with a nitrite solution. The solution used and the sulfur-containing sludge
must be disposed of.
Vol. 1 NATURAL GAS 381

Fig. 16. Loading of molecular sieve 5A.

5) Thiopaq
Thiopaq is a biological process for SO2 reduction, employing absorption
in caustic solution and a sulfur formation step by biological oxidation. It
requires a sulfate bleed stream (82,83).

Adsorption Processes. Adsorption processes for the removal of hydrogen


sulfide and carbon dioxide are used in natural gas treatment predominantly
when only small amounts of acidic components must be removed (51,52).
Activated charcoal and zeolitic molecular sieves are used as adsorb-
ents. Zeolites, in particular, are suitable for selective separations because
of their ordered crystal lattice with extremely uniform pore size and their
high adsorptive capacity. Typical equilibrium loading curves are shown in
Figure 16.
Adsorption processes are generally run continuously in such a way that
the raw gas to be purified passes through at least one adsorber, while the
other adsorber(s) undergo a regeneration cycle. Regeneration can be carried
out by temperature swing, pressure swing, inert gas stripping, or a combina-
tion of these.
Adsorption of Carbon Dioxide. Molecular sieves are used widely for the
purification of natural gas when it is used as feedstock for a cryogenic plant for
production of liquefied gas. In these plants, even low levels of carbon dioxide
cause problems because the gas freezes in the low-temperature unit and can
lead to blockage. The simultaneous removal of traces of water by molecular
sieves is also an advantage here. A gas quality of <50 ppm of carbon dioxide
and 1 ppm of water can be attained in this way.
The main elements of an adsorption plant with three towers based on
the temperature swing process are shown in Figure 11. The first tower (a) is
in the adsorption stage for purification of the raw gas; the second (b) is cooled
with the treated gas which, after being heated, regenerates the adsorbent
in the third tower (c). After completion of this adsorption – regeneration
cycle the function of the three identical towers is changed by automati-
cally controlled valves. The regeneration gas, which is enriched in carbon
dioxide, can be used for heating. In large plants, however, the volume of
382 NATURAL GAS Vol. 1

regeneration gas may become so large that absorptive workup and recircula-
tion should be considered.
Adsorption of Hydrogen Sulfide. Various adsorptive processes are
available for the removal of hydrogen sulfide; these are differentiated by the
type of binding to the adsorbent and the method of regeneration. Activated
charcoal, iron oxide, and zinc oxide are generally used only for the removal of
smaller amounts of hydrogen sulfide because of the limited regeneration capabil-
ity of these adsorbents (53).
Molecular sieves display a high affinity for sulfur compounds. These
materials can be selected to permit simultaneous dehydration and desulfuriza-
tion without removal of carbon dioxide from the gas. Plants are in service which
can purify gas containing 10–6000 ppm of hydrogen sulfide and 0.1–50%
carbon dioxide. The valuable but hydrogen sulfide-rich regeneration gas is
usually treated further (54) and reused.
With special molecular sieves, hydrogen sulfide can be removed and, in the
presence of carbon dioxide, converted to carbonyl sulfide:
H2 S þ CO2 ! COS þ H2 O

For gas with low hydrogen sulfide content the permitted level of carbonyl
sulfide in marketable natural gas can be exploited by this means. Regeneration
and further treatment of the hydrogen sulfide-containing desorption gas are then
no longer necessary. A survey on adsorption processes is given in Table 8.
2.5. Membrane Processes. Gas separation with membranes has devel-
oped into an economically interesting process in the last decade. Membranes can
be used for almost all aspects of natural gas treating. The following are of parti-
cular interest:

Separation of carbon dioxide and methane


Removal of carbon dioxide and hydrogen sulfide
Dehydration
Recovery of higher hydrocarbons (55–57)

Poreless, dense polymer films are used for the separation of gas mixtures.
Individual components of the gas dissolve in the polymer and move through it by

Table 8. Adsorption Processes


Basic process Examples Product
Adsorption adsorption from oxygen-free gases H2S rich gas
on active carbon or
molecular sieves
Adsorption and chemical reaction with ZnO ZnS
chemical reaction
Adsorption and oxidation regeneration by extraction or desorption
Chemical reaction reaction of H2S with iron to Fe2S3 S/Fe2S3
and oxidation regeneration by oxidation to Fe2O3
and elemental sulfur
Adsorption, catalysis oxidation of H2S with O2 to elemental sulfur S
under active carbon catalysis
Vol. 1 NATURAL GAS 383

Fig. 17. Two-stage membrane process.

diffusion. Material transport can be described by the following steps: absorption


from the gas phase into the membrane matrix, diffusion through the membrane,
and desorption out of the membrane into the gas phase. The general rule is that a
larger available membrane surface area, thinner effective membrane separating
layer, greater difference in partial pressures, higher membrane-specific flow rate
of the components to be separated ( permeation coefficient), and better selectivity
for individual gases result in more complete separation of a mixture. A multitude
of polymer membranes are connected in a module to reach as high a packing den-
sity as possible. Hollow fiber, spiral-wound, and envelope-type modules have
become generally accepted. The principle of a two-step module plant for the
treatment of sour natural gas is shown in Figure 17.
The most widespread membrane processes are those for the separation of
carbon dioxide from natural gas. An overview of the membrane and module
types used in this area is given in Table 9.
A large number of membrane plants are in service in the United States.
They are used to reduce the carbon dioxide, hydrogen sulfide, and water content
of natural gas directly at the well. Membrane separators are simple to operate
and require no electrical energy, coolant, or chemicals. They function without
moving parts and can be operated with a minimum of supervision and mainte-
nance. The permeate containing the separated components can, depending on its
composition, be used as low- calorific fuel gas employed in the recovery of sulfur,
flared, or vented directly to the atmosphere.
The service life of membranes, the pressure difference required, and
the loss of gas to the permeate must all be taken into consideration for process
selection.
2.6. Cryogenic Natural Gas Separation. Cryogenic processes are
always employed when very low-boiling components are to be separated because
none of the processes thus far described is suitable. Examples are reduction of
the nitrogen content in natural gas and recovery of helium. In these processes,
distillation steps are combined in which the considerably different boiling ranges
of the primary components methane, nitrogen, and helium can be employed for
separation.
384 NATURAL GAS Vol. 1

Table 9. Types of Membranes and Modules for Carbon Dioxide Separation and
Dehydration (76–80)
Selectivity Permeability
aaCO =CH , of COa2 , H2Ob
Type of membrane 2 4
m3 (STP)
and polymer Type of module Company abH
2 O=CH4 m –2 h –1 bar –1
CO2 Separation
Asymmetric spiral-wound Separex 25–30a 0.25–0.4a
cellulose acetate
Asymmetric hollow fiber Cynara, Dow 25–30a 0.25–0.4a
cellulose
triacetate
Asymmetric spiral-wound Grace 25–30a 0.25–0.4a
cellulose acetate
Composite hollow fiber Permea, 17–24a & 0.1a
polysulfone/ Monsanto
silicone (PRISM)
Polyaramide hollow fiber Medal
(Du Pont/Air
Liquide)
Asymmetric flat sheet Grace 19a 0.4a
polyimide (spiral-wound)
(6 FDA/DMB (79))
H2O Separation
Polysulfone, hollow fiber Permea
polyphenylene oxide
(PRISM CACTUS)
Composite flat sheet GKSS 500b 35b
polyetherimide– (envelope type)
ethylcellulose ether
Asymmetric cellulose spiral-wound Grace 300b 3.2b
acetate

2.7. Recovery of Sulfur. Claus Process. Sulfur is recovered directly


in elemental form from treated natural gas only by liquid oxidation processes
(see Section 2.4.2). With all other absorption and adsorption processes, the regen-
eration gas stream, containing mainly carbon dioxide and hydrogen sulfide is
processed further, predominantly by the Claus process. This process route to
liquid sulfur is used widely in the natural gas industry (58).
High conversion rates of hydrogen sulfide are mandated worldwide by clean
air regulations. Furthermore, the recovery of sulfur is also of economic interest.
Liquid sulfur is stored and transported easily, and solid sulfur can be marketed
in the form of slates, prills, and pills.
In the Claus process, hydrogen sulfide conversion occurs in a thermal
and a catalytic step. The Claus reaction is based on the exothermic overall
reaction

2 H2 S þ O2 ! 2 S þ 2 H2 O (1)
Vol. 1 NATURAL GAS 385

The first exothermic reaction step (Eq. 2) takes place in the combustion
chamber at ca. 900–11008C:

2 H2 S þ 3 O2 ! 2 SO2 þ 2 H2 O (2)

Subsequently, sulfur dioxide reacts with additional hydrogen sulfide to


form elemental sulfur (Eq. 3). This reaction occurs in the presence of catalyst
in several consecutive steps at a stepwise reduction of temperature from 350 to
1908C.

2 H2 S þ 2 SO2 ! 4 S þ 2 H2 O (3)

Cooling of the process gas from the combustion chamber occurs in the
waste-heat boiler, which is combined with a brick-lined combustion chamber.
For each tonne of sulfur produced, more than 2 t of steam is coproduced. Conden-
sers are arranged between individual stages, in which the evaporation of water is
used to cool the reaction gases as much as possible; the aim is to attain as high a
degree of separation of elemental sulfur as possible.
Prior to the catalytic reaction, reheating of the gas occurs to the activation
temperature of the catalyst (ca. 180–2208C). Inactivation caused by condensa-
tion of sulfur must be excluded. In larger plants, both gas–gas heat exchangers
and in-line burners are used. The latter must be run exactly and substoichiome-
trically because the Claus reaction is very sensitive to disturbances and catalysts
are deactivated by reaction with oxygen. Attainable yields depend on the hydro-
gen sulfide content of the gas and the number of catalytic steps. The choice of
catalyst and the heater–separator concept are also important. The pressure is
predetermined: it is approximately the atmospheric regeneration pressure of
the upstream gas scrubbing units. The pressure drop in sulfur recovery plants
is 20–50 kPa.
In the combustion chamber and the waste-heat boiler, 50–70% of the hydro-
gen sulfide is converted to elemental sulfur. The combustion chamber conditions
and the construction of the burner—both multijet burners and single burners
have been built for plants generating sulfur at up to 1500 t/d—influence the
course of combustion and the formation of compounds that influence the process.
Significant amounts of carbonyl sulfide, carbon disulfide, carbon monoxide,
hydrogen, and sulfur trioxide are formed (59,60).
Catalysts for the Claus process are generally based on aluminum oxide,
which may also contain other metal oxides. Traces of sulfur trioxide, which
may be formed by reaction of sulfur dioxide and traces of oxygen during start-
up–shut-down, lead to rapid deactivation of the catalyst by formation of alumi-
num sulfate. Catalysts based on titanium dioxide are also available that are not
sensitive to oxygen and hydrolyze more than 90% of the COS–CS2 formed in the
combustion chamber, which would otherwise require additional measures in
downstream off-gas purification plants (61,62).
To achieve a high degree of sulfur recovery in a Claus plant, an exact
amount of combustion air is required. This amount cannot be calculated precisely
in advance because side reactions, such as the formation of hydrogen, carbon
monoxide, carbon disulfide, and carbonyl sulfide, also have some effect on the
optimal air quantity.
386 NATURAL GAS Vol. 1

Fig. 18. Sulfur recovery requirements for Alberta sour gas plants (August 1988).

Continuously operating analytical devices for determination of hydrogen


sulfide and sulfur dioxide concentration have been established. A sulfur recovery
degree of 94–96% can be attained with two catalytic reaction stages; with three
reactors, >97% recovery can be achieved.
Tail-Gas Cleanup Processes. The sulfur recoveries attainable in Claus
plants often do not meet the emission standards in force (exceptions are very
small plants). Usually, various standards are set, depending on plant capacity.
Mandated sulfur recoveries for plants in the Canadian province of Alberta are
shown in Figure 18. Specific emission limits based on both the plant capacity
and the hydrogen sulfide content of the sour gas are applied in the United States.
In the Federal Republic of Germany a recovery of at least 99.5% is demanded for
all plants with a capacity of >50 t/d (63–65).
Posttreatment of Claus tail gases is therefore generally required. A series of
tail-gas purification processes is available which also produce elemental sulfur.
The most important industrially applied tail-gas treating processes are described
below (41).
Processes Utilizing the Claus Reaction. In the first group the forma-
tion of sulfur according to the Claus reaction from hydrogen sulfide and sulfur
dioxide is performed in an additional catalytic step. This step is operated at a
temperature below the sulfur dew point, thereby shifting the equilibrium in
the direction of sulfur formation. In the so-called dry-bed processes the reaction
and precipitation of sulfur occur over aluminum oxide catalysts. When the max-
imum sulfur load is reached the catalyst is regenerated with hot hydrogen sul-
fide-containing recycle gas, and then tail gas is passed to an already desorbed
and cooled reactor. Sulfur recoveries of 98–99% can be reached in this
way. Established processes include the following: Sulfreen, CBA, MCRC, and
Maxisulf.
In the IFP Clauspool process the reaction is carried out in a high-boiling
solvent that contains a catalyst; the sulfur formed is drawn off as a liquid. Yields
correspond to those of the dry processes.
Sulfur recoveries of 99% can also be attained with the Superclaus 99
process, first introduced commercially in 1989. A conventional Claus plant is
run with a substoichiometric amount of air, so that practically no sulfur dioxide
appears in the tail gas. Unreacted hydrogen sulfide is converted to sulfur by reac-
tion with air in a downstream reactor using a special iron–chromium catalyst
with a small specific surface area (66,67).
Vol. 1 NATURAL GAS 387

Processes with Catalytic Conversion to Hydrogen Sulfide. In the


second group, all sulfur-containing components in the Claus tail gas are cataly-
tically converted to hydrogen sulfide. Further treatment takes place after cooling
and condensation of a large part of the water vapor (68). The following treating
processes exist:

1. Absorption processes with regeneration. The hydrogen sulfide removed


is returned to the Claus plant. The Scot, Sulften, and BSR– MDEA
processes are representatives of this group. To reduce energy con-
sumption, absorbents are being developed which show better selectivity
for hydrogen sulfide and lower energy requirements for regeneration
(Flexsorb, Ucarsol).
2. Liquid oxidation processes by which absorbed hydrogen sulfide is con-
verted in solution directly to elemental sulfur (Stretford, Sulfolin, and
Unisulf process). Sulfur of lower quality is produced than with the other
methods.
3. Catalytic processes in which hydrogen sulfide is converted to sulfur in the
gas phase by addition of air, oxygen, or sulfur dioxide. The sulfur is recov-
ered by condensation. Representatives of this group are MODOP (69), Se-
lectox, and Ultra.

All processes in this group can be designed so that sulfur recoveries of 99.5%
are attainable. With absorption processes, a recovery of > 99.7% is possible.
Processes with Catalytic Oxidation to Sulfur Dioxide. This group
encompasses processes in which all sulfur-containing components are oxidized
to sulfur dioxide. After cooling, sulfur dioxide is removed by absorption. The sul-
fur dioxide-rich gas arising from regeneration of the laden solvent is recycled to
the Claus plant. The Wellman–Lord and Solinox processes, in which a sulfur
recovery of >99.5% can be achieved, is commercially important.
Incineration. With only a few exceptions, the off-gas from Claus plants or
tail-gas cleanup plants must be fed to an incinerator in which the residual
sulfur-containing components are oxidized to sulfur dioxide. For thermal
incinerations 600–8008C is necessary, whereas catalytic plants require only
300–3508C. All these processes lead to an off-gas with <10 mg/m3 of hydrogen
sulfide. If the off-gas to be treated contains other sulfur compounds such as car-
bonyl sulfide or carbon disulfide, then a value of 10 mg/m3 can also be obtained
in a thermal incineration plant, provided the temperature is sufficiently high.
For the catalytic process in this case, however, up to 80 mg/m3 of sulfur must be
expected in the off-gas. In principle, energy recovery by waste-heat boilers or
gas–gas heat exchangers is possible for both processes. Nevertheless, traces
of sulfur trioxide can lead to condensation of sulfuric acid in the heat exchan-
ger. To avoid corrosion, heat exchange must be limited which means that less
energy can be saved (70).
2.8. Example of an Integrated Plant. A flow diagram of an operating
plant for the purification, dehydration, and conditioning of two different raw nat-
ural gases is shown in Figure 19. The design capacity amounts to a maximum of
14  106 m3/d for raw gas 1 and 5  106 m3/d for raw gas 2. Characteristic data for
the units processing raw gas are presented in Table 10.
388 NATURAL GAS Vol. 1

Table 10. Balance of a Sour Gas Treating Plant (Nonselective Chemical–Physical


Scrubbing)
Sour gas Pure gas Flash gasa* Claus gas Sulfurb**
Quantity 600 000 m3/h 473 200 m3/h 4500 m3/h 127 000 m3/h 86.5 t/h

H2S 10 vol% max. 5 mg/m3 ca. 10 ppm 47.2 vol%


CO2 10 vol% ca. 100 ppm 4 vol% 47.2 vol%
N2 5 vol% 6.3 vol% 3.7 vol% 0.1 vol%
CH4 balance balance balance 1.7 vol%
C2þ 0.2 vol% 0.3 vol% 0.1 vol%
Organic sulfur 220 ppm max. 5 mg/m3 400 ppm
compounds
Elemental sulfur 1 g/m3
H2O dew point, 8C saturated 88C saturated saturated
a
Fuel gas for internal use.
b
Marketable product (highly pure liquid sulfur).

Production of sulfur is set at 2100 t/d. The sour gases are predehydrated in
the field (see Fig. 5) and are low in higher hydrocarbons. Treatment is thereby
reduced to the removal of sulfur components and the adjustment of the inert-
gas content. This special plant design has been chosen because of the following
criteria (71):
The high degree of carbonyl sulfide removal, which is necessary for under-
ground storage of the pure gas, can be achieved with the chemical–physical Sul-
finol scrubbing process. The latter is unselective but, depending on the ratio of
hydrogen sulfide to carbon dioxide in the crude gas, may provide a sufficient
Claus gas quality. The high energy consumption of solvent regeneration is
also taken into account. The high-calorific pure gas can be stored in a nearby
underground storage, but, after addition of nitrogen, can also be fed directly
to the sales pipeline as low-calorific gas. The gas must be dehydrated because
a water-containing scrubbing agent is used. The three-stage Claus plant attains
a sulfur recovery of 97.5%; the total recovery is increased to 99.8% in a down-
stream tail-gas cleanup plant using the Scot process. After degassing, the sulfur
produced is stored as a liquid and transported by rail. Sometimes the sulfur is
solidified to prills or pellets outside the plant. Steam generated in a Claus plant
is used in the reboiler for heating the Sulfinol solvent. Expansion turbines are
employed for depressurization of the rich Sulfinol, which produce surplus elec-
tricity. In many cases the energy recovered from expansion, heat exchangers,
and Claus waste-heat boilers is sufficient for regeneration and running the
pumps. The total energy requirement of the sour gas treatment described
here is ca. 3% of the feed gas. One of the greatest consumers of energy is the
tail gas purification plant.
Degradation of the DIPA content of the solvent to oxazolidone occurs
because of the high carbon dioxide content of crude gas. The oxazolidone is regen-
erated to reusable scrubbing agent in a secondary plant. In another secondary
plant a sulfur-dissolving agent, which prevents sulfur plugging of the wells, is
regenerated for further use.
Lean raw gas 2, which contains a low level of hydrogen sulfide, is trea-
ted in a Purisol scrubber. Carbon dioxide remaining in the raw gas helps in
Vol. 1 NATURAL GAS 389

adjusting the calorific value in the total pure gas stream from the Sulfinol
unit. The flash gases are so low in hydrogen sulfide that they can be burned in
a boiler.
Auxiliary Plants. Natural gas treating plants are connected to the public
energy distribution network. Reliable auxiliary plants are required to guarantee
a high level of availability and independence.
Steam Generation. Waste heat from the Claus process is used in the pro-
cessing plant to drive process pumps and turbine-driven generators, and to heat
solvents. The high-pressure steam generated is first expanded in turbines. The
saturated low-pressure steam obtained can be used reboiling of the solvent at
constant temperature. The reboiler steam pressure is 0.5–0.7 MPa and is
adjusted to match the thermal stability of the solvent. Waste-heat boilers in
the Claus plant operate at steam pressures of 1.6–4.0 MPa; the sulfur conden-
sers at 0.1–0.6 MPa. By proper choice of driving equipment and steam pressure,
energy requirements can be balanced so that only a little external energy is
needed. Small amounts of blow-down streams must be continuously removed
from boilers and condensers to avoid corrosion. These streams must be replen-
ished by water from a treatment plant and a boiler feedwater degassing plant.
Electrical Power. Large compressors, circulating pumps, and blowers are
often driven by steam turbines to achieve a better control of their speed. Fre-
quently, the plant’s own power station is only used for start-up and shutdown,
whereby both steam- and gas turbine-driven generators are employed.
Inert Gas. Occasionally, purified natural gas must be mixed with inert
natural gas to meet specified quality standards. When the admixing of air is
not possible because of limitations of the oxygen content, inert gas plants are
required. Inert gases produced from both catalytically purified flue gases and
air separation are used. The dry gases are generally produced at low pressure
and brought to pipeline pressure (ca. 10 MPa) by multistage compressors. Inert
gases are also employed for safe start-up and shutdown of plants and for regen-
eration of the Claus catalyst (72).
Tank Storage. Both higher hydrocarbons and sulfur are stored in tanks as
liquids. They are transported further by road, rail, and frequently, pipeline. With
special thermal insulation, liquid sulfur transportation through pipes of more
than 8 km in length is possible without additional heating (73).
Wastewater. Natural gas treating plants produce little noxious waste-
water. The boiler- and cooling tower blow-down water is fed to public sewage
systems. Waste waters from piping and equipment cleaning are collected and
treated in a reclaimer. Remaining residues are injected with the salt-containing
formation water in separate disposal wells.
Gas Mixing. Different field qualities are mixed in the so-called raw gas
manifold, so that the absorbers can be charged with gas of constant composition.
Special separators minimize the entrainment of liquids and impurities that
would damage the solvent or cause foaming. Pipelines from the field can be
charged from the manifold with pigs for cleaning purposes. Mixing of different
pure gases to adjust the Wobbe index and the net calorific value is possible in
the pure gas manifold.
Reclaimer, Filter. Through side reactions, chemical scrubbing solvents in
particular become enriched with degradation products, frequently of unknown
390 NATURAL GAS Vol. 1

Fig. 19. Flow diagram of a natural gas treating plant in the Federal Republic of
Germany. a) Sour gas manifold; b) Town gas manifold; c) Disposal well. – – –
Steam;– – – – Condensate.

nature. These substances are corrosive and may cause foam formation in the
absorbers. Solvent quality is stabilized by removing a small side stream and
addition of fresh solvent. This side stream is processed in a reclaimer, which is
usually an auxiliary distillation plant. Purified absorbent can be reused in the
process (74).
With physical scrubbing agents, removal of undesired substances (eg,
hydrocarbons) is frequently achieved by addition of water and phase separation.
Mechanical filters are used in a side stream.
Special Construction Features. High partial pressures of carbon dioxide
and hydrogen sulfide in a wet environment lead to serious corrosion. In particu-
lar, the stress corrosion cracking (SCC) caused by hydrogen sulfide corrodes
Vol. 1 NATURAL GAS 391

high-strength steel from inside: atomic hydrogen can be formed by acid corrosion
in connection with water, and can diffuse into the steel because of its small size of
the atom. The hydrogen atoms recombine in the area of microcracks and local
inhomogenities; the resulting pressure, which can reach 105 MPa, leads to cracks
and blistering (hydrogen-induced cracking).
High-alloy hard steels are particularly susceptible; the necessary construc-
tion and testing regulations must be adhered to. Consequently, the carbon
content of the steel is limited. Equipment and pipeline welding joints must be
annealed. A hardness HRC of <22 must be maintained. Excessive stresses
( particularly at plate edges) and thickness changes which are caused by the fab-
rication techniques must be avoided. Specially designed forged nozzles based on
finite element analysis are customary (75).

3. Liquefaction

The principal reason for liquefying natural gas is a 600-fold reduction in


volume that occurs with the vapor-to-liquid phase change. In its naturally
occurring vapor state, natural gas is a bulky energy source, which is some-
times cumbersome to handle. Storage of the vapor requires huge underground
caverns or large telescoping storage tanks. Transporting natural gas from
production sources to points of consumption necessitates large pipeline
networks. Thus, only overland or somewhat shorter undersea routes can
be considered. Finally, natural gas at atmospheric pressure is too bulky to
be considered as a fuel for transportation purposes; at the very least, it
must be compressed.
Liquefaction of natural gas serves to overcome these obstacles, and permits
transport over larger distances and more diverse application of liquefied natural
gas (LNG) as an energy source. For example, LNG produced in the Middle East
and throughout the Pacific Rim supplies ca. 10% of Japan’s primary energy con-
sumption. Storage of LNG near urban areas permits peak demands for natural
gas to be satisfied without building additional pipelines that would be underuti-
lized most of the time.
Compressed natural gas is a feasible transportation fuel for truck and bus
fleets. However, more recent aircraft and railroad applications (87) require the
higher energy density of LNG to be commercially viable.
3.1. Physical Properties of LNG. In the nomenclature of the energy
industry, LNG is on the light end of the liquefied gas spectrum that also includes
natural gas liquids (NGL, primarily ethane with some propane) and liquefied
petroleum gas (see Liquefied Petroleum Gas, which consists mainly of propane
with a significant proportion of butane). Common to each of these products is the
lack of a standard set of physical properties. Nevertheless, certain characteriza-
tions of LNG are possible and useful.
Because methane is its primary component, the boiling point of LNG at
atmospheric pressure is ca. 1608C. Liquefied natural gas is odorless and color-
less. Trace contaminants, especially mercury, can render LNG potentially corro-
sive if they are not removed to sufficiently low levels during the prepurification
(see Chap. 2) and liquefaction processes. Liquefied natural gas is a commercial
392 NATURAL GAS Vol. 1

product, so the need to avoid solidification of trace contaminants is buttressed by


rigid specifications for an environmentally acceptable fuel. Therefore, contami-
nants such as water vapor, carbon dioxide, and sulfur compounds, which may
be found in vapor-phase natural gas, generally do not exist in LNG. Similarly,
the relatively low boiling point of LNG limits the maximum concentration of pen-
tane and heavier hydrocarbons that can exist in the liquid phase.
3.2. History. Early attempts to commercialize LNG failed. Barge trans-
port between stockyards in Chicago, Illinois and gas fields in Louisiana did not
employ adequately designed storage (88). Similarly, a prototype peak-shaving
storage tank in Cleveland, Ohio failed in 1944 because of poor material selection.
The resultant spill from the tank vaporized, traveled through an adjacent sewer
system, and caused a number of remote casualties upon ignition. These early set-
backs, as well as the subsequent explosion of a poorly designed LNG tank under-
going repair on Staten Island, New York in 1973, did much to hinder the
commercialization of LNG in the United States. Much of the early development
centered around Europe and North Africa.
The first baseload LNG trade began in 1964 from Algeria to Canvey Island
in England and to Le Havre in France. Subsequent projects were built in Libya
and Algeria to supply LNG to Europe and its expanding pipeline grid. Japan
became a purchaser of LNG in 1969, with the receipt of its first cargo from
Kenai, Alaska. Since that time, Japan has become the preeminent buyer of
LNG in the world. Major production facilities now also exist in Indonesia,
Qatar, Abu Dhabi (United Arab Emirates), Brunei, Malaysia, Australia, Oman,
Nigeria, Trinidad, and Egypt. As of 2005, more than 120 million metric tons per
year of LNG is being produced, and facilities under construction will produce an
additional 60 million metric tons per year.
Peak-shaving applications of LNG expanded rapidly in the late 1960s and
throughout the 1970s in both Europe and North America. This growth mirrored
the increased use of natural gas in urban areas and, in particular, the preponder-
ant use of natural gas for heating. Virtually no growth of the peak-shaving busi-
ness occurred in the 1980s because of a number of diverse factors, ranging from
energy conservation and environmental siting restrictions to changes in govern-
mental regulations.
Experiments in LNG fueling of transportation vehicles have been carried
out for more than 20 years. Several driving forces exist for this work, including
the environmental advantages of natural gas combustion compared to conven-
tional fuels. However, no major infrastructure for LNG fueling exists to support
the conversion of automobiles and trucks. Current development for railroad and
marine vessels is, therefore, focusing on closed-loop applications in which the
vehicle returns to the fueling station as part of its normal routing. Fishing ves-
sels may also profit from the low temperature of LNG, which can be used to
refrigerate the fish that are caught. Similarly, to best utilize LNG as a fuel for
high-performance aircraft, the leading edges of the wings may have to be cooled
with LNG to reduce skin-friction drag, while the LNG is being vaporized for use
in the aircraft’s engines.
3.3. Production. Because of the high critical pressure of methane, LNG
must be produced commercially by refrigeration rather than pressurization. A
number of processes have been commercialized for LNG production.
Vol. 1 NATURAL GAS 393

Common to all natural gas liquefaction processes is the need to pretreat the
gas feed for removal of components that would freeze out (thus restricting the
flow of the process), cause corrosion, or lead to pollution upon combustion of
the revaporized LNG (see Chap. 2).
Baseload Liquefaction Processes. Large-capacity or baseload liquefac-
tion of natural gas is performed with emphasis on process efficiency. The scale
of operations means that production with lower installed power capacity and
lower fuel consumption is economically favored. Because the heat that must be
removed from natural gas to cool it to 1608C is eventually rejected to ambient
air or water, rather elaborate systems have resulted.
Cascade Refrigeration Processes. The earliest liquefaction processes
employed cascades of single-component refrigerants in series (see Fig. 20).
Each refrigerant is established as a separate closed-loop refrigerator that
supplies refrigeration at discrete temperature levels. Typically, propane, ethy-
lene, and methane are used to provide a wide, balanced range of refrigeration.
After compression (a) three temperature levels for each of the three refrige-
rants form a nine-stage cascade. Each of these temperature levels corresponds
to a preset pressure letdown (d) for evaporating the refrigerant in heat
exchange with the natural gas feed and a separate refrigerant stream that
requires cooling (e). In this fashion, heat is removed from the natural gas at
successively lower temperatures; ie, the refrigerant is boiled at successively
lower pressure. Heat is rejected to ambient air or water via the warmest
refrigerant, generally propane, and compressor aftercoolers (b). The methane

Fig. 20. Nine-stage cascade liquefaction process. a) Compressor; b) Condenser;


c) Accumulator; d) Phase separator; e) Heat exchanger.
394 NATURAL GAS Vol. 1

refrigerant loop is open in that it is combined with the natural gas feed and after
the final pressure letdown the liquid methane forms part of the LNG product.
Cascade processes allowed use of single-component refrigerants at a time
when thermodynamic correlations and thermophysical data bases were not as
well developed as they are at present. In addition, the processes could be made
very efficient (ie, the amount of irreversibility could be reduced) by increasing
both the number of refrigerants employed and the number of boiling stages
for each refrigerant. This improvement in efficiency, however, had significant
cost penalties. Each refrigerant requires its own compressor, compressor
driver, vessels, and heat exchangers, together with attendant piping, insulation,
and instrumentation. Each additional stage of boiling adds to the number
of heat exchangers and vessels and to the number of sidestreams entering the
compressor.
All-Mixed-Refrigerant Processes. The problems of complex layout,
relatively high capital cost, and limited train capacities inherent in cascade pro-
cesses were addressed by the mixed-refrigerant processes that succeeded them.
As equipment and analytical capabilities advanced, the refrigerants could be
combined in a single refrigeration loop (see Fig. 21). In such processes, a single
refrigerant mixture of pentane, butane, propane, ethane, methane, and nitrogen,
for example, is constituted to match the cooling curve of the natural gas to be
liquefied; ie, the temperature–enthalpy warming curve of the mixed refrige-
rant closely tracks the natural gas cooling curve in (g). This reduces the irrever-
sibility of the process. The same refrigerant is then condensed, a part against air

Fig. 21. All-mixed-refrigerant (MR) liquefaction process. a) First stage MR compressor;


b) Cooler; c) Second stage MR compressor; d) Cooler; e) Phase separator; f ), g) Heat
exchangers; h) Joule–Thomson valve; i) Product letdown valve; j) Driver.
Vol. 1 NATURAL GAS 395

or water at (d) and the balance at lower temperature in heat exchange against
itself in (f). In the latter case, incoming gaseous refrigerant is cooled and con-
densed at elevated pressure in countercurrent flow against evaporating liquid
refrigerant fractions, each flashed (h) to a much lower pressure and, hence,
temperature. The all-mixed-refrigerant process has performed well in a number
of plants and represents a simplification compared to the cascade process. How-
ever, it is not thermodynamically efficient enough to be economical in the face of
rising energy prices. To match the cooling curve of the natural gas feed over the
wide range from the temperature of cooling water or ambient air down to the
liquefaction temperature, compromises in the mixed-refrigerant composition
are necessary. The wide range of boiling points for the refrigerant components
also means that some of the heavier components are compressed to higher pres-
sures than actually required for their condensation to ensure condensation of the
lighter, lower boiling components such as nitrogen and methane. Such a recom-
pression penalty cannot be avoided without a certain degree of separation of the
refrigerant components such as occurs in precooled mixed refrigerant processes.
Precooled Mixed-Refrigerant Processes. In the early 1970s, a third
generation of processes, precooled mixed-refrigerant processes, developed from
direct combination of the other two. The most widely used process employs two
separate refrigeration systems, a propane cascade refrigerant loop in series with
a mixed-refrigerant cycle that incorporates propane, ethane, methane, and nitro-
gen as components (89) (Fig. 22 shows the cooling curves for this process). The
propane loop precools the natural gas and serves as an intermediate refrigerant

Fig. 22. Cooling curve for a propane precooled mixed-refrigerant liquefaction process.
a) Cooling curve of natural gas; b) Propane cascade; c) Mixed refrigerant (stage 1);
d) Mixed refrigerant (stage 2).
396 NATURAL GAS Vol. 1

for heat rejection from the mixed-refrigerant section to cooling air or water. The
propane section can be fabricated from lower cost carbon steel, whereas the lower
temperature sections require aluminum or nickel steels. Thus breaking the cool-
ing process at the propane stage is compatible with economical material selec-
tion. Finally, by reducing the range of cooling that must be achieved by the
mixed refrigerant, its composition can be optimized and energy losses due to
recompression significantly reduced.
The increase in worldwide demand for LNG has prompted an increase in
the capacity of liquefaction trains in order to take advantage of the economies
of scale. This is being achieved by improvements in the liquefaction process as
well as increases in the equipment capacity. For the most widely used processes,
capacity increases of refrigerant compressors, gas turbines, and cryogenic heat
exchangers have led to train capacities of nominally 5 million metric tons
per year.
For larger train capacities, a hybrid process cycle (known commercially as
AP-XTM) has been developed (90,91). In this process, a cycle similar to the
precooled mixed-refrigerant process is used to precool and liquefy the LNG
(see Fig. 23A). However, the liquefied LNG is subcooled in a nitrogen refrigera-
tion closed-loop process cycle (Fig. 23B). Nitrogen gas is compressed, cooled to
near ambient conditions with cooling water or ambient air, and then further
cooled to cryogenic conditions by expansion to lower pressure. The gaseous nitro-
gen is then used to subcool LNG, after which it is returned to be recompressed,
completing the refrigeration cycle. By employing the nitrogen refrigeration cycle
to subcool the LNG, the mixed-refrigerant only has to cool the LNG to about
1158C (versus 1508C to 1608C for the typical precooled mixed-refrigerant
process). This allows for a larger production capacity without a substantial
increase in equipment size.
Additional process schemes have been developed that fall in the category of
precooled mixed-refrigerant processes. If ethane is added to the precooling refrig-
erant to form a dual mixed-refrigerant process, then the precooling temperature
can be lowered in accordance with the operating conditions of the loop and the
proportion of ethane added (92). With this process the refrigeration duty can
be shifted between the two refrigeration loops – a capability that may be useful
in the face of changing feed gas conditions or changing power availability. The
latter can occur because of ambient temperature changes that affect the power
output of gas turbine drivers. A binary precooling refrigerant is somewhat
more complicated to operate because simple pressure control is no longer ade-
quate, but it may be preferred in certain scenarios.
Another precooled process has been proposed for plants that employ gas
turbine drivers in the mixed-refrigerant section. In this configuration, waste
heat from the gas turbines is used to separate ammonia and water in an
ammonia absorption refrigerator (93). This system could provide precooling
duty while eliminating the need for costly compressors and drivers. This system
has not been employed in an operating facility.
Peak-Shaving Liquefaction Processes. Peaking-shaving LNG plants
differ from baseload facilities in several significant aspects that affect plant
design. Peak-shaving plants are much smaller, operate only seasonally, and
are often located near the point at which the pressure in gas trunk lines is
Vol. 1 NATURAL GAS 397

Fig. 23. AP-XTM process cycle. A) Propane precooled mixed-refrigerant process;


B) Nitrogen refrigeration closed-loop process.

decreased and the lines branch into lower pressure local gas distribution
systems. The design emphasis for peak shavers is thus on capital cost minimiza-
tion rather than high thermodynamic efficiency. All-mixed-refrigerant liquefaction
cycles have therefore been employed in a large proportion of peak-shaving faci-
lities. If a local distribution pipeline is available at a pressure substantially
below that of the main gas trunk line, expander processes can be utilized to take
advantage of the pressure difference as shown in Figure 24. Several variations
exist, but the principle of these plants is to expand the inlet gas nearly isentropi-
cally through a turbine, thus rapidly lowering the temperature and partially lique-
fying the gas. The liquid is sent to storage, and the residual gas is compressed in a
compressor mechanically linked to, and driven by, the expander. This gas is then
sent to the lower pressure pipeline for distribution outside the plant.
3.4. Storage. Storage tanks for LNG are an important part of both base-
load and peak-shaving LNG facilities. In addition, they constitute the major por-
tion of the investment in import receiving terminals. Because of the high cost of
these units and their importance in the overall safety of LNG facilities, a great
deal of attention has focused on LNG tank design.
One of the earliest efforts in tank design was directed at proper material
selection. The Cleveland, Ohio tank failure in 1944 was attributable to the use
of 3.5% nickel steel which became embrittled at LNG temperature. Subsequent
to this, large-scale programs have proved the suitability of 9% nickel steel, stain-
less steels, and certain aluminum alloys (5000 series) for LNG tank fabrication.
Designs with the latter have been somewhat limited in size, because the coeffi-
cient of thermal expansion for aluminum is approximately double that for steel;
398 NATURAL GAS Vol. 1

Fig. 24. Schematic of an expander liquefaction process. a) Turbine expander; b) Separa-


tor; c) Compressor.

in large tanks, such thermal movement during cooldown could result in tank
failure.
Tank designs have also evolved as more sophisticated safety analysis has
been applied to LNG installations (94). Early designs featured an internal cryo-
genic liquid tank enclosed by an outer tank that contained the insulation system
for the inner one. In some designs the outer tank contained gaseous nitrogen and
was, in turn, connected to a variable-volume or bladder tank, which compensated
for changes in the nitrogen volume due to changes in ambient temperature while
avoiding pressurization or depressurization of the outer tank. In other designs,
the roof of the inner tank was not gastight but merely supported insulation, and
the outer tank served as a natural gas holder (see Fig. 25). In both designs, the
outer tanks were fabricated of carbon steel and surrounded by a low dike to con-
tain any LNG spills.
Analytical studies indicated that the prime safety risk with an LNG spill is
the formation of a large product vapor cloud that can drift, ignite, and cause
widespread damage. Subsequent designs incorporated outer tank materials
less prone to failure at cryogenic temperatures and taller dikes built closer to
the tanks. These measures lead to less free surface for any LNG spill from a
tank failure and therefore a lower feed rate to the ensuing vapor cloud.
Additional studies reviewed the consequence of an external impact, such
as a crashing aircraft, leading to tank failure and the effect of a catastrophic
inner tank failure on outer tank containment. These studies all focused on the

Fig. 25. Simplified cross-sectional drawing of single-integrity LNG storage tank.


a) Suspended deck; b) Outer tank; c) Sidewall insulation; d) Inner tank; e) Load-bearing
insulation; f ) Foundations; g) Dike.
Vol. 1 NATURAL GAS 399

need for much more secure outer containment. Resultant tank designs have
double-integrity tankage; ie, a liquid spill from a failure of the inner tank is con-
tained by a second concentric tank that is structurally independent of the first
(95). The outer tanks have been built of reinforced concrete with a stainless
steel liner. Finally, full-height earthen berms have been built to protect the tank-
age from external damage and to contain the postulated asymmetric loading on
the outer tank resulting from catastrophic failure of the inner tank. Where the
site is too small to permit full development of earthen berms, inground storage or
additional reinforced concrete outer containment is employed.
3.5. Transportation. Peak Shaving. Peak-shaving LNG facilities typi-
cally vaporize LNG from storage and send it out via a local distribution network.
In some installations, however, LNG is merely stored and vaporized for send-out.
These satellite facilities receive LNG from other production sites, which is trans-
ported in over-the-road trailers. The trailers are similar to those employed in
transporting liquid nitrogen and oxygen, and consist of an inner cryogenic vessel
and an outer protective vessel of noncryogenic material. The space between
vessels is typically filled with expanded perlite insulation and evacuated to sub-
atmospheric pressure to further decrease heat in-leak.
Baseload. Baseload LNG facilities are linked to their corresponding
receiving terminals by fleets of large oceangoing tankers. Modern tankers typi-
cally can hold 125 000 m3 of LNG. Each tanker employs one of several contain-
ment systems, which can be divided into three principal types. One design
employs free-standing spherical tanks supported on cylindrical skirts (see
Fig. 26). Each tank is insulated and located between bulkheads in a separate
section of the ship. These compartments are blanketed with inert nitrogen gas,
which is sampled periodically to detect any LNG leakage.
A second family of designs is the membrane tank. In ships using this type,
the insulation and a secondary barrier are applied to the hull of the ship (Fig. 27).
The insulation also transmits the load from the LNG cargo to the hull and its

Fig. 26. Spherical ship tank. a) Weather cover; b) Tank; c) Insulation; d) Support skirt;
e) Thermal break; f ) Inner hull; g) Ballast tank; h) Outer hull.
400 NATURAL GAS Vol. 1

Fig. 27. Membrane ship tank. a) Membrane liner; b) Insulation; c) Inner hull; d) Ballast
tank; e) Outer hull.

structure. A metallic membrane containment barrier is then installed over the


surface of the insulation. The membrane may consist of Invar (highly alloyed
steel with an extremely low coefficient of thermal expansion) or of a ‘‘waffle’’
structure (made of nickel steel), where the cross section of the waffle allows
two-dimensional thermal movement while maintaining intimate contact with
the load-bearing insulation. Membrane tanks are also compartmentalized and
enable more efficient storage use of the hull volume. This manifests itself in a
flattop deck and a lower center of gravity, whereas spherical tanks extend up
through the deck and require domed covers.
A third type of LNG tanker employs free-standing prismatic tanks. This
design seeks to combine efficient use of hull space with the structural indepen-
dence of a self-supporting structure. A number of earlier ships were built with
this design type, but it is relatively expensive and, with the exception of two
smaller ships for the Alaska–Japan trade, is not employed in newer tankers.
3.6. Safety Aspects. Liquefied natural gas may be viewed as a creation
of the modern era, and as such, it has been subjected to more thorough scrutiny
than traditional fuels such as gasoline and fuel oil. The hazards associated with
LNG are that it is a low-temperature fluid, is flammable, and can create vapor
clouds dense enough to cause asphyxiation. Because none of these is a problem
when LNG is contained in properly designed storage tanks or piping and equip-
ment, safety studies have generally focused on LNG spills. In LNG facilities,
extensive safety systems exist to detect gas leaks, to detect and counter fires,
and to detect smoke. Spill prevention and containment are implicit in the design
of plant facilities, particularly of storage tanks because they contain the greatest
quantity of LNG.
If an LNG spill is ignited soon after it occurs, a pool fire results. This fire is
certainly a problem, but it is contained in a facility staffed and equipped to deal
with such emergencies. A greater concern is that the same spill may be given suf-
ficient time to evaporate and form a vapor cloud that can travel for some distance
before ignition. A number of variables such as the nature of the surface under-
neath the spill, the wind velocity, and the presence of obstructions to the cloud
act to shape and direct it (96). Mechanically, the cloud is formed as LNG boils
from the surface of the spill (97). The evolving vapors are much colder than air
Vol. 1 NATURAL GAS 401

and initially form a dense, low-lying cloud. As the cloud mixes with air and is
warmed by its surroundings, it begins to rise. Wind serves to add a horizontal
component to the cloud’s motion. As air mixes with the natural gas, the mixture
becomes flammable (ie, local compositions are between the lower and upper
flammability limits of LNG). The flammability limits vary with LNG composi-
tion, particularly the proportion of propane. This cloud is then potentially hazar-
dous to areas beyond the battery limits of the LNG facility.
A number of large-scale tests have demonstrated certain characteristics of
LNG vapor cloud fires. These fires are deflagrations-rapid combustions in which
the flame front that moves through the cloud is preceded by a weak, decoupled
shock wave (98). An accompanying radiation hazard exists, but not the pressure
damage of a shock wave that occurs when mixtures of other hydrocarbons and air
are detonated.

4. Transmission, Storage, and Distribution

More than 50% of world gas reserves are located in remote areas. For example,
part of the gas used in Western Europe is produced in the harsh permafrost
environment of Siberia or in the North Sea.
In most cases, producers ship the gas from the fields to the borders of the
countries in which it is used. Importers buy the gas at these points under long-
term contracts and resell the fuel to local utilities as well as to industrial users
and power stations connected directly to the transmission system. Residential
and commercial consumers are normally served by the local distribution
companies. The European natural gas transmission system is shown in Figure 28.
Natural gas is used chiefly for heating. Therefore, gas demand varies sub-
stantially between winter and summer, weekdays and week-ends, or day and
night. The ratio between summer and winter loads in Europe is between 1 : 5
and 1 : 10. Production, transmission, storage, and distribution facilities must be
designed and sized to handle these load variations.
Of the gas traded across international borders, 75% is shipped by pipelines,
and 25% is carried by LNG tankers.
Gas field development and the construction of transportation systems from
distant areas of production to gas-importing nations are highly capital-intensive
projects. Thus, the load factors at which gas is imported are nearly always very
high. For load equalization, gas is stored in underground storage facilities during
the off-peak season and produced from storage during periods of peak demand in
winter. Pipeline pressurization (line packing) in the transmission and distribu-
tion systems themselves, as well as peak-shaving installations, also help to han-
dle load variations.
4.1. Transportation. Pipeline Transmission. Transportation of large
gas volumes is best achieved by large-diameter pipelines operating at high pres-
sure. Currently, the pipeline diameter may be up to 1400 mm and working pres-
sure up to 8 MPa. Such pipelines carry gas at a flow rate of 2  106 m3/h (STP)
over a distance of ca. 1000 km. This capacity is insufficient, however, to ship
gas from distant fields to markets. Compressor stations must therefore be built
to increase the pressure of the gas. Gas compressors are normally driven by
402 NATURAL GAS Vol. 1

Fig. 28. European natural gas transmission system.

turbines or engines fueled by their own pipeline gas for reliability and lower cost.
The number of compressors required for a given carrying capacity depends on the
distance over which the gas must be shipped and on the p compression ratio,
which is given by
po
p¼ (1)
pi

where po is the maximum compressor outlet pressure (bar) and pi the minimum
compressor inlet pressure (bar).
As the compression ratio increases, the number of compressors needed to
operate a gas transmission line decreases. Hence, transmission engineers must
seek to optimize capital outlay and follow-up operating expenses for pipelines
and compressor stations.
For large-diameter pipelines, the optimum compression ratio is generally
relatively low; it is between p ¼ 1.25 and p ¼1.4 (99). Offshore pipelines are
currently built with diameters up to 1000 mm. Pipelines from the North Sea to
continental Europe are 1000-mm-diameter lines that are laid at a water depth of
Vol. 1 NATURAL GAS 403

150 m; Italy and North Africa are connected by 500-mm lines that cross 600-m-
deep water. The very high working pressures at which offshore lines can be oper-
ated partly make up for the loss in capacity due to the smaller diameters. If
necessary, pipeline bundles are laid (100).
Offshore compressor stations are extremely costly structures because they
must be built on platforms. For this reason, the optimum compression ratio of
offshore pipelines is higher than that of onshore pipelines. Depending on flow,
it may be as high as p ¼ 4.
LNG Transportation. Although the energy input required for liquefac-
tion of natural gas is substantial, the volume advantage makes liquefaction
economically viable. Cryogenic LNG is shipped by LNG tankers at atmospheric
pressure. The cargo tanks are insulated to minimize evaporation. Neverthe-
less, the boil-off rate is between 0.2 and 0.25% of the cargo per day. The gas
that evaporates is used to fuel the ships engine. Reliquefaction is presently
not viable.
The transportation of LNG is often the only way of bringing gas from dis-
tant gas-producing countries to market outlets. Any cost comparison between
LNG transportation and pipeline transmission must, of course, be geared to
the circumstances of the project. Generally speaking, an LNG chain is the only
answer if pipeline transmission is not feasible for technical or other reasons or if
the distance is sufficiently long. The cost of LNG transportation is lower than
that of offshore pipeline transmission even for distances of several hundred
kilometers, whereas onshore transmission is nearly always less expensive
than LNG transportation unless the distance is extremely long (99,101–103). A
comparison of costs of LNG transportation and pipeline transmission is given in
Figure 29.
Gas Transmission Pipeline Design. Pipelines for gas transmission must
be designed so that, under peak load conditions, gas is allowed to flow from the
point of delivery at the inlet of the system to the outlet of the system at a pres-
sure between the maximum and minimum allowable operating pressure. Gener-
ally, pipeline capacity is somewhat lower in summer than in winter because of
the following reason: The atmospheric temperature increases soil temperature

Fig. 29. Cost of LNG transportation and cost of pipeline transmission as a function of
transportation distance. a) Offshore pipeline (900-mm diameter); b) LNG transportation
with liquefaction onshore (125 800 m3); c) Onshore pipeline (1000-mm diameter).
404 NATURAL GAS Vol. 1

and hence gas temperatures, and power output from the compressor engine
decreases because of the higher temperature of the intake air.
According to DARCY and WEISBACH, the pressure drop in a gas transmission
pipeline is given by the following equation:

16 l T
p21  p22 ¼ f % q2 pn K (2)
p2 d5 n n Tn

where p1 and p2 are absolute pressures (Pa); pn is the normal pressure,


1.01325  105 Pa; p21  p22 represents the pressure drop (Pa2); T is the absolute
temperature (K); Tn the normal temperature, 273.15 K; %n the gas density
under normal conditions (kg/m3 ); f the pipe friction factor; qn the flow rate
under normal conditions (m3/s); l the length of the pipeline (m); d the diameter
of the pipeline (m); K the gas flow deviation coefficient; and subscript 1 denotes
the pipeline inlet and subscript 2 the pipeline outlet.
Equation (2) is usually transformed into the following equation in which the
pressure is expressed in bar and the flow rate in cubic meters per hour:

l
p21  p22 ¼ 4:6401 1015 f Q2n %n T K (3)
d5
where Qn is the hourly flow rate under normal conditions (m3/h); p1 the absolute
pressure at the pipeline inlet (bar); and p2 the absolute pressure at pipeline
outlet (bar).
Equation (3) may be rewritten using the following terms

p21  p22 ¼ K c Q2n (4)

where

l
c ¼ 4:6401 1015 f % T
d5 n (5)
cðbar2 h2 m6 Þ

According to NIKURADSE, the pipe friction factor f for turbulent flow and
rough-surface pipes is given by (104).

1
f ¼ (6)
ð2 log dk þ 1:14Þ2

According to the Colebrook equation, the following applies to the friction


factor in the transition regime between a hydraulically smooth and a hydrauli-
cally rough pipe surface:
!
1 2:52 k
pffiffiffiffi ¼ 2 log pffiffiffiffi þ (7)
f Re f 3:71 d

where k is the surface roughness factor.


Vol. 1 NATURAL GAS 405

The limit between hydraulic roughness and the transition regime is given
by
 2
200 d
fl ¼ (8)
Re k

where the Reynolds number Re can be calculated by using the following


Equation:

Q2n %n
Re ¼ 353:678 (9)
d Z

where Z is the dynamic viscosity of the fluid (106 Pa s).


If the pipe friction factor is fl, then the flow is in the transition regime.
For the design of new lined large-diameter pipelines, the pipe roughness
factor k (in mm) is used in flow calculations:

Hydraulically rough regime (Nikuradse) 0.012 mm


Transition regime (Colebrook) 0.006–0.008 mm

The gas flow deviation factor K describing the real-gas behavior of natural
gases is another important term in flow equations. For most natural gases, K
may be approximated by

K ¼ Kð pm ; Tm Þ

2 p31  p32 (10)


pm ¼
3 p21  p22

Increase in Capacity. If the capacity of an existing pipeline must be


increased, (1) new compressor stations can be installed or (2) pipeline sections
can be looped.
Installation of New Compressor Stations. The increase in capacity
that can be achieved by the installation of compressor stations is given by

Qn pffiffiffiffiffiffiffiffiffiffiffiffi
¼ n þ 1 100% (11)
Q0

where Qn is the capacity of the system with compressor stations (m3/h); Q0 is


the carrying capacity without compressor stations (m3/h) and n is the number
of compressor stations.
To maximize the increase in carrying capacity achieved by the installation
of compressor stations, the compressor station sites must be optimized. The opti-
mum site depends on whether flow is constant throughout the pipeline system
or varies because of offtakes between the inlet and the outlet of the system
(see Table 11) (105).
Looping of Pipeline Sections. The increase in capacity due to the
installation of compressor stations increases the system cost in steps (106). For
406 NATURAL GAS Vol. 1

Table 11. Optimum Compressor Station Sites


Site, % of total length
Number of Pipeline Pipeline Maximal
compressor without with transport
stations offtake offtake capacity, %

0 100
1 50 21 141
2 33/67 13/31 173
3 25/50/75 9/21/38 200
4 20/40/60/80 7/16/26/42 224

this reason, additional compressor horsepower is not a cost-efficient method of


increasing capacity for all additional capacity requirements. Alternatively, pipe-
line sections may be looped to obtain exactly the additional capacity that is
needed. The increase in capacity achieved by a loop line of the same diameter
as the main line is given by the following formula (if no gas is taken off between
the inlet and the outlet of the pipeline system):

QL 1
¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 100% (12)
Q0 13 L 4 L1

where QL is the capacity after addition of a loop line of length L1 (m3/h); Q0 is the
capacity without looping (m3/h); L is the length of pipeline between inlet and
outlet (m); and L1 the length of the loop line (m). The length of the loop line
required to provide a specified additional capacity depends critically on whether
or not gas is tapped from the pipeline system between the inlet and the outlet
points (105).
The increase in capacity as a function of loop line length is plotted in
Figure 30. If a loop line of the same diameter is assumed, 75% of the main line
must be looped to increase capacity by 50% if all gas flows from the inlet to the
outlet of the transmission system. If gas is taken off regularly between the inlet
and the outlet (as in the case of a regional line), only 40% of the original pipeline
has to be looped.
Transient Flow in Gas Transmission Systems. Flow rate and pressure
in pipelines are a function not only of distance but also of time. Therefore, line
packing effects play an important role. Simulation of transient flow, of course,
produce different results than steady-state simulation. In a typical gas system,
for instance, an hour of peak demand is usually preceded by a phase of low
demand. As pressure increases when demand is low, a large gas volume is
packed in the line when the demand peaks. During the period of peak load off-
takes, the pressure in the line decreases because of the pressure drop resulting
from the higher gas flow velocity through the line. Part of the gas packed in the
line is therefore discharged from the system, covering peak demand.
Figure 31 shows the flows through a pipeline system into which gas is deliv-
ered at location 1 and from which gas is taken off at location 2 at different load
factors. If the flow rate at the pipeline inlet (location 1) is constant (Fig. 31 A),
Vol. 1 NATURAL GAS 407

Fig. 30. Increase in carrying capacity as a function of loop length. a) Pipeline without
offtake; b) Pipeline with offtake.

inlet and outlet pressures will vary considerably and line packing will contribute
substantially to covering peak load demand. As gas is normally delivered into a
pipeline system at a continuous rate, the line will be repacked during periods of
low demand, for example, during the night. If delivery into the pipeline system is
at constant pressure (Fig. 31 B), the delivery rate will vary. However, the
response at the point where gas is delivered into a pipeline system is always
delayed with respect to the system behavior at the point where gas is taken off
from the system; for this reason, delivery into the system and delivery from the
system never coincide exactly.
4.2. Storage. Gas storage facilities are integrated into the pipeline sys-
tem to match the high load factor at which gas is produced and delivered by sup-
pliers to the pattern of end-user demand, which varies between winter and
summer, over the week, and from hour to hour.
Determination of Storage Requirements. Gas demand is influenced by
many different physical and economic parameters. Because substantial quanti-
ties of gas are used for space heating, ambient temperature is the most critical
one. Typical average daily deliveries Qdi in Northwestern Europe as a function of
temperature are shown in Figure 32. Gas demand increases as a linear function
of temperature decrease and is constant above a certain limit temperature. The
correlation is represented by the following equation:

Qdi ¼ Qdg þ mðtl  ti Þ (13)

where Qdi is the daily gas demand at the daily average temperature ti (kW h/d);
Qdg the daily baseload gas demand (kW h/d); tl the limit temperature for space
heating (8C); ti the average temperature on a given day (8C); and m the increase
in deliveries per 8C decrease in temperature (kW h d1 8C1). This equation
allows sufficiently accurate predictions of daily gas offtakes (100).
Transmission, storage, and distribution facilities must, of course, be sized
for the design day peak load on the system, determined by Equation (13). The
ratio between system capacity specified for peak load conditions and capacity uti-
lization is a crucial commercial and economic factor. Utilization time or system
load is expressed differently in various countries.
408 NATURAL GAS Vol. 1

Fig. 31. Transient flow in gas transmission systems. A) Constant flow rate at pipe-
line inlet; B) Constant pressure at pipeline inlet

In Central Europe, the concept of utilization time expressed in annual uti-


lization hours or annual utilization days is frequently employed. Utilization time
is given by Equations (14) and (15):

Annual gas volume


Utilization hours per year ¼ (14)
Peak hourly volume
Annual gas volume
Utilization days per year ¼ (15)
Peak daily volume

In the United States and the United Kingdom, the load factor expresses the same
relationship. It correlates with the quantities given in Equations (14) and (15) as
follows:

Annual volume
Load factor ¼
8760  Peak hourly flow rate
Vol. 1 NATURAL GAS 409

Fig. 32. Daily average delivery as a function of temperature. 10 m (slope) ¼ increase in


delivery per degree temperature decrease.

Annual utilization hours


Load factor ¼ (16)
8760
Irrespective of the actual load of the system, transmission, storage, and dis-
tribution facilities must be designed for adequate gas supplies regardless of
ambient temperature or other conditions. With regard to storage facilities, gas
deliveries from storage must at all times be sufficient to satisfy any demand in
excess of gas supplies from gas fields into the transmission system. The volume of
gas that is stored must be adequate to maintain this delivery throughout the per-
iod for which demand exceeds field supply. The necessary storage capacity is
thus determined by the difference between the field supply load pattern and
the demand load pattern, as well as by ambient temperature during the period
in which supply is lower than demand. The volume of gas needed to satisfy peak
demand must be stored from excess field supplies during off-peak periods. A typi-
cal daily load curve for a period of one year (for Northwestern Europe) is shown
in Figure 33.
Storage Systems. Different systems are used by the gas industry to store
gas during off-peak periods and to deliver gas from storage for peaking service.
Various storage systems serve different peaking needs (see Table 12). The
storage of gas in low-pressure containers has decreased substantially in impor-
tance because the quantity of gas stored in such a facility is insufficient.
Storage pipe arrays or spherical gas tanks in which gas is stored at high pres-
sure provide peaking service for diurnal swings and short peaks within a week.
Spherical gas tanks operate between 0.6 and 1.6 MPa and can have a diameter
as large as 30 m. Pipes used for storage may hold gas at a pressure up to 8 MPa.
410 NATURAL GAS Vol. 1

Fig. 33. Daily load curve showing gas delivered from storage and gas added to storage.

Line packing in high-pressure gas transmission pipelines is a method of


handling diurnal peaks. For line packing, the pipeline must have a large diameter
and must be designed for high working pressure. Line packing is feasible only if at
least temporary spare capacity is available in the system. Normally, the transmis-
sion system design engineer plans a line packing capacity approximately eight
times the difference between the hourly delivery rate and the peak hourly sendout
into the system. This capacity corresponds to ca. 6–7% of daily sendout.
Any demand peaks not provided for at the regional or the local level must be
handled at the supraregional level. The quantities of gas that must be stored for
seasonal peak shaving are substantial, and large underground storage facilities
are practically the only system that will hold sufficiently large volumes of gas.
Figure 34 shows the storage capacity that must be created to handle seasonal
space heating load peaks as a function of supply load factor.
Gas is stored below ground in appropriate rock formations (see Fig. 35). Sui-
table formations for underground storage of gas include porous or fissured rock
as well as caverns mined in salt rock, for example. These formations must be
covered with impermeable caprock to prevent gas from escaping.
The following types of underground storage facility exist:

1) depleted gas fields, oil fields, and aquifer storage facilities (porous rock storage);
2) salt cavern storage facilities; and
3) mined cavern storage facilities (only of minor importance).

Table 12. Suitability of Various Storage Systems for Different Peak Needs
Peaks

Daily Weekly Seasonal System operator

Low-pressure container  local distribution companies


High-pressure container  
Line packing  limited regional transmission companies
Cavern, fields limited   supraregional transmission
companies
Vol. 1 NATURAL GAS 411

Fig. 34. Storage capacity required for seasonal peak shaving as a function of the supply
pattern.

Underground storage facilities hold both working gas and cushion gas. Working
gas is the gas that may be produced from storage for peak-shaving operations.
Cushion gas, sometimes referred to as base gas, is the gas required to
ensure sufficient reservoir pressure for recovery of the working gas. It provides
the pressure needed to cycle the normal working storage volume.
Porous Rock Storage. Gas can be stored below ground in rock pores.
The rock should have sufficient porosity and permeability, and the storage hor-
izon (which must be at sufficient depth) should be covered by tight caprock and
be a structure that entraps gas injected into the rock. Structures used for porous
rock storage include depleted oil and gas fields as well as aquifer structures.
Development of a gas storage facility in a depleted gas field does not pose a
major risk because data on rock properties and information on the tightness of
the caprock are available. The working gas capacity, the maximum reservoir
pressure, and the delivery rates as a function of reservoir pressure are known
from production history. For the storage of gas in depleted oil fields, gas–oil
separators are needed; however, secondary oil production by gas injection often
makes a depleted oil field a commercially attractive storage site (107).
In aquifer storage reservoirs, interstitial water in the pores is displaced by
the gas injected at high pressure. Considerable geological and reservoir engi-
neering effort is required to decide whether an aquifer structure is suitable for
underground storage of gas: ie, the structure of the potential storage horizon
must be evaluated; the thickness, porosity, and permeability of the storage
rock must be determined; and the thickness of the caprock must be verified.
Salt Cavern Storage. Gas may also be stored in caverns leached into
salt rock. Unlike the volume of a natural storage rock reservoir, the volume of
these cylinder-shaped caverns can be controlled within certain limitations. The
diameter of such a cavern may be as large as 80 m and its height 400 m. The
cavern volume can reach 500 000 m3.
The capacity of cavern storage facilities can be adapted to increased storage
needs by adding further caverns in a given formation as required. The cushion
gas volume required for operation of a storage cavern is much lower than the base
gas volume in a porous rock reservoir and may amount to only 20% of total capa-
city. Storage caverns allow the production and injection of gas at very high rates
that are limited chiefly by the capacity of the above-ground installations.
412 NATURAL GAS Vol. 1

In salt storage caverns, convergence can occur because cavern leaching


operations and pressure cycling after commissioning produce strain in the
rock. As a result, salt flows into the cavern, reducing its volume. Within limits,
convergence may be controlled by retaining a minimum quantity of gas in the
cavern (104,107–109).
Optimized Storage Planning. As the load curve in Figure 33 shows, sto-
rage facilities produce gas for only relatively short periods of time. Depending on
supply and demand load patterns, gas withdrawal is required for peak shaving
on 80–150 days per year. If the ambient temperature rises, the withdrawal
phase may be interrupted on days in which demand is lower than supply thereby
allowing more gas to be injected into a partially depleted storage facility.
With porous rock storage facilities, production rates are limited by the resis-
tance of the rock to gas flow to the wells. Depending on depth, reservoir pressure,
and permeability, wells may produce gas at rates between 1000 and 70 000 m3/h
(STP). At salt cavern storage sites, the withdrawal rate is limited only by pres-
sure loss in the tubing and the design of above-ground facilities. Storage caverns
today are normally designed for production rates of up to 250 000 m3/h (STP).
However, because gas cools rapidly during depressurization, production at
such a high rate is feasible only for a relatively short duration up to 60 h in a
limited number of production phases during the year. Similar restrictions do
not exist for porous rock storage facilities, which normally hold substantially lar-
ger quantities of working gas that is produced at lower rates. The working gas
volume at a salt cavity storage site, on the other hand, may be increased by
leaching an appropriate number of caverns.
Transmission system operators optimize their peak-shaving operations by
employing both porous rock and salt cavern storage facilities. The peaking ser-
vice baseload is provided by the production of gas from porous rock facilities at
relatively constant load, whereas caverns are used in periods of maximum
demand. The optimum employment of storage caverns and storage reservoirs
depends, of course, on supply and demand load patterns (see Fig. 36).
4.3. Distribution. Distribution systems take gas from regional supply
centers and transport it to residential, commercial, and industrial users. These
systems consist of high-pressure, medium-pressure, and low-pressure grids. Dis-
tribution systems tend to be operated at low load factors because they serve a
market in which demand varies considerably. If most gas is used for heating,
the grid is designed for peak load on a winter day.
Distribution grids have grown in most areas over the years because gas-
works were built in many cities long before natural gas became available. For
this reason, distribution grids consist of mains of different diameters and differ-
ent materials laid in different years.
Old distribution lines often operate at 2–8 kPa, whereas new distribution
mains are frequently designed for a working pressure of 100 kPa. Low-pressure
distribution grids are normally tied into medium-pressure or high-pressure dis-
tribution systems from which they receive gas at special supply stations.
Since local distribution networks consist of mains cross-connected at numer-
ous points, computers are required to predict flow in distribution lines. Early mod-
els and methods for solving these nonlinear patterns were proposed by CROSS in
1936 and MAC ILLROY in 1949 (110). Substantial computing time is required for
Vol. 1 NATURAL GAS 413

Fig. 35. Underground storage facilities for natural gas.

such a nonlinear model, which originally consisted of very few nodes in view of the
limited performance of early computers. As data-processing technology progressed
during the 1950s, network analysis techniques become more widely used. They are
now state of the art and are employed as a matter of course in distribution system
design. Modern software models all the details of distribution systems.
Many of the input data required for network analysis can be determined
only by statistical methods and field tests. For example, statistics are necessary
to predict system loads controlled by the demand of numerous gas users whose
meters are usually read only once each year. Pipe roughness, which is crucial for
the hydraulic behavior of the grid, must be measured in many cases because pipe
properties depend on the year in which the distribution line was constructed.
Distribution networks must also be designed to handle peak loads that can
be 10 or 15 times higher than the baseload. To optimize the economics of gas dis-
tribution, distribution system capacity is regularly adapted to growing demand.
Methods of increasing capacity include the following:

1) replacement of mains or laying of new lines;


2) increasing distribution system pressure often to a maximum of 10 kPa in
old systems and installation of service regulators to lower the pressure to
2 kPa at which most appliances are operated; or
3) installation of new medium-pressure or high-pressure system stations to
feed gas into the low-pressure grid.
414 NATURAL GAS Vol. 1

Fig. 36. Optimum production of gas from porous rock storage and salt cavern storage for
peak shaving.

New governor stations to interconnect ring mains and distribution lines


constitute a particularly effective method of increasing distribution grid capacity
(111,112). Local distribution companies normally combine all three methods to
optimize the economics of gas distribution. The new distribution lines that are
increasingly being built to develop service in rural regions are designed mainly
for a pressure of 100 kPa, because the higher distribution pressure substantially
reduces pipeline diameter for the same pipeline capacity (see Fig. 37). In fact,
high distribution pressures are often necessary to serve rural areas because
100-kPa distribution lines can transport gas over much longer distances.

5. Economic Aspects

Natural gas is the world’s third most important energy source, behind oil and
coal, covering ca. 23% of the 1995 world energy demand.
Reserves. Proven world gas reserves at the end of 1995 totaled about
150  1012 m3(Vst ie, standard cubic meter of natural gas at 158C and 101.3 kPa
with a net calorific value of 35.2 MJ or 9.776 kW h) (equivalent to 5171  1018 J
or 134  109 t of oil), corresponding to ca. 95% of proven oil reserves (113). A geo-
graphical breakdown is shown in Table 13. These reserves equal about 55 years of
current annual gross production.
In addition, ca. 222  1012 m3(Vst ) (equivalent to ca. 7770  1018 J or
200  109 t of oil (114)) has been identified as probable gas reserves. Including
this figure, the total 1995 gas inventory equals approximately 135 years of cur-
rent annual gross production.
Production. The world’s 1995 gross gas production amounted to
2.75  1012 m3(Vst) (95.15  1018 J or 2.4  109 t of oil), of which about 80% was
marketed. The rest covers reinjection, flaring, losses, etc. The total 1995 mar-
keted volume of 2.21  1012 m3(Vst), broken down into major geographic zones,
is shown in Table 14.
Vol. 1 NATURAL GAS 415

Fig. 37. Pipeline diameter as a function of working pressure for constant pipeline
capacity (in relation to pipeline diameter for 2.5-kPa working pressure).

Consumption. Consumption patterns in various regions of the world


generally follow production profiles, as shown in Table 14. The Former Soviet
Union and Eastern Europe as well as North America both produce and consume
ca. 30% of the world’s annual natural gas availability.
International Trade. Although marginal compared to total consump-
tion, a sizable trade over national borders has developed since 1960. Initial inter-
national deliveries were mainly to neighboring countries through pipelines,
whereas shipments of LNG developed in the 1970s. Table 15 summarizes the
natural gas volume traded internationally in 1995. Of the total of
395  109 m3(Vst) (equivalent to 13  1018 J or 352  106 t of oil) traded, approxi-
mately 92  109 m3(Vst) or 23% was received in the form of LNG.
Pricing. Except for a minor amount used as chemical feedstock, natural
gas is primarily employed as clean heating fuel in industry, power plants,

Table 13. World’s Proven Natural Gas Reserves as of January 1, 1996 (113)
Area 1012 m3(Vst) 1018 J %
North America 6.5 222 4.3
Latin America 8.0 277 5.4
Western Europe 6.3 220 4.3
Former Soviet Union, 59.2 2037 39.3
Eastern Europe
Africa 10.0 346 6.7
Middle East 45.3 1558 30.1
Asia, Oceania 14.8 512 9.9
Total 150.2 5171 100.0
416 NATURAL GAS Vol. 1

Table 14. World Natural Gas in 1995 [109 m3(V ST)] (113)
Marketed
Area production Consumption

North America 689 686


Latin America 101 103
Western Europe 241 350
Former Soviet Union, 734 657
Eastern Europe
Africa 84 46
Middle East 145 145
Asia, Oceania 213 1220
Total 2207 2207

commercial installations, and households. To maintain its competitiveness ver-


sus alternative energies, natural gas prices for the final consumer are generally
determined by prices of competitive fuels such as light heating oil, heavy fuel oil,
coal, LPG, or electricity, with due consideration of the inherent environmental,
handling and financial advantages of gas utilization.
Since gas transportation costs significantly exceed those of oil or other fuels,
gas netbacks at the wellhead are generally lower than those for crude oil. To com-
pensate for the heavy front-end investment in production, purification, and
transportation facilities, a large share of internationally traded natural gas is
normally sold by producers under long-term contracts of up to 35 years’ duration,
which oblige sellers to make predetermined annual quantities available, whereas
buyers are obliged to offtake certain minimum quantities per year (take-or-pay
obligation: usually 80% of availability). Prices of such long-term gas deliveries
are generally tied to official international crude or fuel oil quotations.

6. Testing and Analysis

Natural gas analysis can be divided into two areas: chemical analysis of the com-
position, and determination of the physical properties. Chemical analysis meth-
ods for natural gas are summarized in (115,116). Instructions for both the
chemical analysis and determination of physical properties for natural gases
are given in (116).
6.1. Analysis of Natural Gas Composition. The most important
method for chemical analysis of natural gas is gas chromatography (117–123).
Sampling must be carried out before the gas can be analyzed.
Sampling. Glass equipment is mainly used for sampling because glass is
not attacked by any component of natural gas and does not affect the composition
of the gas (124). Vessels with two taps (see Fig. 38) are used, which must be com-
pletely dry. The gas to be analyzed is first passed through the sampling vessel to
completely displace the air from it. Both taps are then closed so that the sample
is contained in the vessel. The sample should always be under slight overpres-
sure, so that no air is drawn into the vessel even if the taps are not closed
tightly.
Table 15. International Gas Trade in 1995 [109 m3 (V ST)]
Imports to
Supplies from Japan Germany United States France Italy United Kingdom Belgium Others Total

Former Soviet Union 32 13 14 58 117


Canada 79 79
Netherlands 25 6 4 5 1 41
Norway 11 7 2 3 5 28
Algeria 1 8 17 4 8 38
Indonesia 23 10 33

417
Malaysia 11 2 13
Brunei 7 1 8
United Arab Emirates 5 5
United States 2 2 4
Others 9 1 19 29
Total 57 69 80 34 35 2 12 106 395
418 NATURAL GAS Vol. 1

Fig. 38. Sampling vessel.

Gas Chromatography. Gas chromatography has developed since 1952


and has replaced the Orsat method, previously used for analysis of natural
gas. Gas chromatography is the preferred method for most gas analyses. The
analysis of natural gas by gas chromatography has been standardized in ASTM
D-1945. In gas chromatography, the gas to be analyzed is transported by a car-
rier gas stream (usually helium or argon) through a separating column. Car-
bon dioxide may also be used as carrier gas according to the Janak (125).
Figure 39 shows the setup for gas chromatography (120) and Figure 40 an
example of a gas chromatogram of natural gas (120). The separating column
is filled with a molecular sieve or with Chromosorb.
Thermal conductivity detectors and flame ionization detectors are most
commonly used, mass spectrometers may also be used for detection (119). In
some cases, two columns must be used–one for the separation of hydrocarbons,
the other (for a new sample) mainly for determination of nitrogen and oxygen.
Standard reference mixtures are used for two gas qualities: lean gas (ie, gas

Fig. 39. Gas chromatograph. a) Carrier gas; b) Regulating valve; c) Manometer; d) Chart
recorder; e) Air thermostat; f ) Separating column; g) Thermal conductivity detector;
h) Flow meter; i) Three-way tap; j) Trap.
Vol. 1 NATURAL GAS 419

Table 16. Standard Reference Mixtures for the Chromatographic


Analysis of Natural Gas
Lean gas, Rich gas,
Component mol% mol%

Methane 85.0 74.0


Ethane 6.0 10.0
Propane 4.0 7.0
n-Butane 2.0 3.0
Isobutane 2.0 3.0
n-Pentane 0.5 1.0
2,2 Dimethylpropane 0.5 1.0
Isopentane 0.5 1.0
Hexanes and higher 0.1 0.2
hydrocarbons
Nitrogen 4.0 0.5
Carbon dioxide 1.0 1.0
Helium 1.0 0.5
Hydrogen 3.0 0.5

containing few or no condensable hydrocarbons and thus having a lower calorific


value) and rich gas (ie, gas containing a higher amount of condensable hydrocar-
bons and thus having a higher calorific value) (see Table 16).
Analysis of Nonhydrocarbons. Natural gases can contain various
gases apart from hydrocarbons: nitrogen, carbon dioxide, and noble gases can
be determined during gas chromatographic analysis of the hydrocarbons. A gas
chromatographic analysis method for noncondensable gases in light hydrocar-
bons is given in (126). The content of water, sulfur compounds, and mercury
may also be determined.
Determination of Water. Water must be removed from natural gas
to prevent its condensation and to avoid the formation of gas hydrates (127)
(see Section 1.2). Low water content of the order of milligrams per cubic meter
can be titrated according to the classical Karl–Fischer method, in which water
reacts with iodine in an alcoholic solution of sulfur dioxide and pyridine (128).
The iodine can also be determined coulometrically. Water can be determined
gravimetrically by absorption using a desiccant such as magnesium perchlorate.
A variety of hygrometers (129) measure, for example, the variation in the
conductivity of glycerol with water content. Dew point measurements indirectly
give the water content.
Determination of Sulfur Compounds. Natural gases can contain
hydrogen sulfide and occasionally mercaptans. The total sulfur content can be
determined by combustion with excess air or oxygen. Either the sulfur dioxide
formed can be determined, eg, by means of a coulometric titration, or the combus-
tion products can be absorbed in an oxidizing solution and the sulfuric acid
formed is determined.
Hydrogen sulfide can be detected qualitatively using lead acetate producing
a black coloration due to the formation of lead sulfide.
In the ASTM standard test method a measured volume of gas is bubbled
through two solutions – first a neutral cadmium sulfate solution and then a
basic cadmium sulfate solution (130). Hydrogen sulfide is found in the neutral
420 NATURAL GAS Vol. 1

Fig. 40. Gas chromatogram of natural gas.

solution and the mercaptans in the basic solution. Hydrogen sulfide and mercap-
tans are then determined iodometrically.
Determination of Trace Components. Heavy metals, especially
mercury, are the most important trace components. There are at present no
standard determination methods; atomic absorption spectroscopy (AAS) is
usually used.
6.2. Determination of Physical Properties. Natural gases are charac-
terized and classified according to their physical properties. The Wobbe index
(W ) is the most important characteristic for the classification of natural gases
and they are divided into two groups: group L (low Wobbe index) and group H
(high Wobbe index). The Wobbe index is calculated according to
pffiffiffi
WS ¼ HS = dðMJ=m3 Þ;

where HS is the gross calorific value and d the relative density (based on air). In
addition to the more commonly used superior Wobbe index WS, the inferior Wobbe
index WI can be calculated, whereby the inferior calorific value HI is used instead
of HS. The Wobbe indices indicate gas quality for combustion.
Usually the Wobbe index is calculated from HS and d. Automated calori-
meters are available for its direct determination (Wobbe recorders) which pffiffiffi use
a gas stream that is adjusted so that the flow rate is proportional to 1= d.
Important physical properties of natural gas include the following:

Density
Calorific value
Molecular mass
Dew point

Density. Density can be measured with a gas density balance. The method
is based on the different buoyancy of a nitrogen-filled glass sphere suspended
from a balance in various test gases (131). The gas to be measured and nitrogen
Vol. 1 NATURAL GAS 421

should have the same temperature and pressure. Instruments of this type
also allow continuous measurement and recording of gas density in a gas
stream.
A particularly reliable method is to weigh a specific volume of gas in a glass
flask (eg, a Chancel flask according to IP 59). Either the density of the gas (mass
per unit volume) or the relative density d (the ratio of the density of the gas to
that of dry air) is determined. The density and relative density can also be calcu-
lated from the composition determined by gas analysis (132).
Calorific Value. The calorific value in MJ/m3 can be determined in three
ways: (1) directly by combustion of the gas at constant pressure in a gas calori-
meter, (2) by stoichiometric combustion and measurement of the gas : air ratio,
and (3) by calculation from the values obtained from gas analysis.
The principle of continuous recording gas calorimeters is combustion of a
controlled stream of gas, followed by heat exchange between the combustion
gases and a liquid, whose temperature increase is measured and recorded
(133). This method is the most widely used. Discontinuous gas calorimeters
involve combustion of a specific amount of gas and the resulting temperature
increase is measured directly.
Molecular Mass. The molecular mass of natural gas is calculated as an
average value from the concentrations determined by gas analysis and the mole-
cular masses of the components.
Dew Point. In order to avoid condensation of water, the gas tempe-
rature should not drop below the dew point of the water contained in the
gas. The dew point can be determined directly by passing a constant stream
of gas over a polished metal mirror that is slowly cooled. The dew point is the
temperature at which the first signs of condensation appear on the mirror.
These methods can be used for gases under pressures of up to 10 MPa.
Automated instruments using photocells are also employed (134). If the
gas stream contains easily condensable hydrocarbons, then these must be
washed out by passing the gas through high-boiling liquid paraffin prior to
determination.

7. Acknowledgement

The entire topic was coordinated by AXEL COMMICHAU.

REFERENCES

GENERAL REFERENCES

1. H. B. Bradley: Petroleum Engineering Handbook, Society of Petroleum Engineers,


1987, Chap. 20.
2. D. L. Katz et al.: Handbook of Natural Gas Engineering, McGraw-Hill, New York
1959.
3. G. G. Brown et al.: Natural Gasoline and the Volatile Hydrocarbons, Natural Gas
Association of America, Tulsa 1948.
422 NATURAL GAS Vol. 1

SPECIFIC REFERENCES
4. DIN-Norm 1871 : ‘‘Gasförmige Brennstoffe und sonstige Gase,’’ Beuth Verlag,
Berlin–Köln 1980.
5. VDI-Wärmeatlas, VDI-Verlag, Düsseldorf 1988.
6. N. L. Carr, R. Kobayashi, D. B. Burrows: ‘‘Viscosity of Hydrocarbon Gases Under
Pressure,’’ Trans. Am. Inst. Min. Metall. Pet. Eng. 201 (1954) 264–278.
7. M. B. Standing, D. L. Katz: ‘‘Density of Natural Gases,’’ Trans. Am. Inst. Min.
Metall. Pet. Eng. 146 (1942) 140–144.
8. J. D. van der Waals, Proc. Acad. Sci. Amsterdam 3 (1901) 515.
9. O. Redlich, J. N. S. Kwong: ‘‘On the Thermodynamics of Solutions. V—An Equation
of State. Fugacities of Gaseous Solutions,’’ Chem. Rev. 44 (1949) 233–244.
10. G. Soave: ‘‘Equilibrium Constants from a Modified Redlich–Kwong Equation of
State,’’ Chem. Eng. Sci. 27 (1972) 1197–1203.
11. O. Y. Peng, D. B. Robinson: ‘‘A New Two-Constant Equation of State,’’ Ind. Eng.
Chem. Fundam. 15 (1976) no. 1, 59–64.
12. R. L. Robinson, Jr., R. H. Jacoby: ‘‘Better Compressibility Factors,’’ Hydrocarbon
Process. 44 (1965) no. 4, 141–145.
13. K. R. Hall, L. Yarborough: ‘‘A New Equation of State for Z-Factor Calculations,’’ Oil
Gas J. 71 (1971) June 18, 82–91.
14. M. Benedict, G. B. Webb, L. C. Rubin: ‘‘An Empirical Equation for Thermodynamic
Properties of Light Hydrocarbons and Their Mixtures,’’ Chem. Eng. Prog. 47 (1951) 419.
15. K. E. Starling, M. S. Hon: ‘‘Thermo Data Refined for LPG,’’ Hydrocarbon Process. 52
(1972) no. 5, 129–133.
16. M. Jaeschke et al.: ‘‘Accurate Prediction of Compressibility Factors by the GERG
Virial Equation,’’ SPE Paper 17 766, SPE Gas Techn. Symp., Dallas, TX, 1988.
17. ISO 6570: ‘‘Natural Gas—Determination of Potential Hydrocarbon Liquid Content,’’
1983/1984.
18. D. L. Katz: ‘‘Prediction of Conditions for Hydrate Formation in Natural Gases,’’
Trans. Am. Inst. Min. Metall. Pet. Eng. 160 (1945) 140.
19. DIN 51 850: ‘‘Brennwerte und Heizwerte gasförmiger Brennstoffe,’’ Beuth Verlag,
Berlin-Köln 1980.
20. Deutscher Verein des Gas- und Wasserfaches: Techn. Regeln ‘‘Gasbeschaffenheit,’’
Arbeitsblatt G 260/I, Eschborn 1983.
21. D. L. Katz: Handbook of Natural Gas Engineering, McGraw-Hill, New York 1959.
22. G. Drewniok, D. Huber: ‘‘Neue Ergebnisse bei der Erdgasbehandlung im Feld,’’
Lecture at the 26th DGMK-Haupttagung, DGMK-Compendium 78/79, Berlin,
Oct. 4–6, 1978.
23. J. H. z. Siederdissen, F. Wundram, GWF Gas Wasserfach: Gas Erdgas 127 (1985)
no. 1, 3–11.
24. F. Brunner, W. Woll, JPT J. Petr. Technol. 10 (1980) 377–384.
25. O. Buchholz, S. Peter, Erdöl Kohle Erdgas Petrochem. 20 (1967) 268–272.
26. G. Wilken, SPE Paper 19 813, 64th SPE Technical Conference, San Antonio, Oct.
8–11, 1989.
27. W. Tunn, Erdöl Kohle Erdgas Petrochem. 26 (1973) no. 9, 498–500.
28. Dow Chemical Comp.: Gas Conditioning Fact Book, Midland, Michigan 1972.
29. NGPSA (Nat. Gas Proc. Suppl. Ass.): Engineering Data Book, Tulsa, Oklahoma, 1972.
30. A. Misra, Erdöl Erdgas Z. 91 (1975) Apr., 99–105.
31. D. Ballard, Hydrocarbon Process. 45 (1966) June, 171–180.
32. J. Kaliner et al.: ‘‘High Pressure Separation of Higher Hydrocarbons From Natural
Gas With a Continuous Regeneration of the Absorbent By Natural Gas,’’ 17th World
Gas Conference, Washington, D.C., June 5–9, 1988.
Vol. 1 NATURAL GAS 423

33. J. McAllister, P. Parsons, The Chemical Engineer 1987, April, 24–26.


34. J. Voogd, GWF Gas Wasserfach: Gas Erdgas 128 (1987) 4: 192–200.
35. A. L. Kohl, F. C. Riesenfeld: Gas Purification, Gulf Publ., Houston 1979.
36. K. Thormann: Absorption, Springer Verlag, Berlin 1959.
37. C. W. Sweeney et al., Chem. Eng. (N.Y.) 95 (1988) no. 9, 119–125.
38. J. Polasek, J. Bullin, Energ. Prog. 4 (1984) no. 3, 146–150.
39. J. A. Lagas, Sulphur 159 (1982) Mar./Apr., 35–39.
40. J. A. Lagas: ‘‘Selection of Gas Sweetening Processes,’’ Comprimo Gas Sweetening
and Sulphur Recovery Seminar, Amsterdam, Nov. 1982.
41. 1989 Gas Process Handbook issued by Hydrocarbon Process. Gulf Publishing Co,
Houston, 1989.
42. R. E. Meissner, U. Wagner, Oil Gas J. 81 (1983) Feb. 7, 55–58.
43. J. C. Thomas: ‘‘Improved Selectivity Achieved With Ucarsol Innovator Solvent 111,’’
38th Annual Laurance Reid Conference, Norman, Oklahoma, March 7, 1988.
44. W. Weinberg et al.: ‘‘New Gas Treating Alternatives For Saving Energy in Refining
and Natural Gas Treating,’’ 11th World Petroleum Congress, London, Aug. 31,
1983.
45. S. A. Newman: ‘‘The Benfield Process For Acid Gas Removal,’’ AIChE Symposium
Acid And Sour Gas Treating, Atlanta, Mar. 12, 1984.
46. J. Frazier, Hydrocarbon Process. 49 (1970) no. 4, 101–102.
47. W. G. J. van Dijk, J. G. Tönis, Erdöl Kohle Erdgas Petrochem. 19 (1966) June,
404–405.
48. M. Kriebel, Energ. Prog. 4 (1984) no. 3, 143–145.
49. S. A. Newman: Sour Gas Treating Processes, Gulf Publ., Houston 1985.
50. ‘‘Sweeting Natural Gas II,’’ Sulphur 193 (1987) Nov./Dec., 26–31.
51. K. Hedden, R. Rao, GWF Gas Wasserfach: Gas Erdgas 123 (1982) no. 1, 1–6.
52. P. J. Parsons et al.: ‘‘Improvement in Adsorption Processes For Testing Natural
Gas,’’ 51st Autumn Meeting Institute of Gas Engineers, London, Nov. 13, 1985.
53. K. D. Henning et al., GWF Gas Wasserfach: Gas Erdgas 126 (1985) no. 1, 19–24.
54. M. R. Cines et al., Chem. Eng. Prog. 72 (1976) Aug., 89–93.
55. R. W. Spillman et al., ‘‘Gas Membrane Process Optimization,’’ AIChE Meeting, New
Orleans, March 9, 1988.
56. E. W. Funk,, N. N. Li: ‘‘Purification of natural gas by membranes,’’ Gas Separation
Technology, Elsevier Sci. Publ., Amsterdam, 1990, 355–372.
57. K. Ohlrogge: ‘‘Membranes for Dewpointing of Gas Streams,’’ 13th Ann. Membr.
Techn./Sep. Plann. Conf., Newton, Massachusetts, Oct. 23–25, 1995.
58. U. Sander et al.: Sulphur, Sulphur Dioxide and Sulphuric Acid, The British Sulphur
Corp. Ltd., London 1984.
59. Sulphur 187 (1986) Nov./Dec., Suppl. 1–20.
60. J. B. Hyne, Sulphur 198 (1988) Sept./Oct., 24–39.
61. M. J. Pearson, Hydrocarbon Process. 94 (1973) Feb., 81–85.
62. J. Nougayrede et al., Oil Gas J. 85 (1987), Aug. 10, 65–70.
63. Energy Resources Conservation Board (ERCB): Sulphur Recovery Guidelines for
Sour Gas Plants in Alberta, Aug. 1988.
64. Environmental Protection Agency 40 CFR Part 60, United States Federal Register,
vol. 50, no. 190, Part IV, Oct. 1985.
65. VDI-Guideline 3454 : ‘‘Emission Control Claus Units,’’ Part 1, Beuth-Verlag, Berlin,
June 1989.
66. J. E. Meuler: ‘‘Claus Tailgas Treatment,’’ Part 1, Comprimo Gas Sweetening and
Sulphur Recovery Seminar, Amsterdam, Nov. 1985.
67. B. G. Goar: ‘‘Emerging New Sulfur Recovery Technologies,’’ Laurence Reid Gas Con-
ditioning Conference, Norman, Oklahoma, wMarch 6–8, 1989.
424 NATURAL GAS Vol. 1

68. T. J. van Pol: ‘‘Claus Tailgas Treatment,’’ Part 2, Comprimo Gas Sweetening and
Sulphur Recovery Seminar, Amsterdam, Nov. 1985.
69. R. Kettner, N. Liermann, Oil Gas J. 86 (1988) Jan. 11, 63–66.
70. R. Kettner, T. Lübcke, VDI Ber. 730 (1989) 255–274.
71. R. Müller, Erdöl Ergas Kohle Petrochem. 28 (1975) 132–137.
72. R. Kettner, N. Liermann, Erdöl Erdgas 102 (1986), Oct., 448–453.
73. S. Lechler, A. Misra, A. Feizlmayr: New Technology Required for West German
Sulfur Line, Pipe Line Industry, June 1978.
74. T. Lübcke et al.: ‘‘The Mobil Alkanolamine Waste Recovery Process,’’ Comprimo
Sulphur Workshop, Amsterdam, Nov. 1987.
75. R. N. Tuttle, R. D. Kane, National Association of Corrosion Engineers Publication,
Houston 1981.
76. W. J. Schell, C. D. Houston, Chem. Eng. Prog. 78 (1982) Oct. 33–37.
77. S. S. Kulkarni, E. W. Funk, N. N. Li, R. L. Riley, AIChE Symp. Ser. No.229, 79
(1983) 172–178.
78. T. E. Sulpizio, B. S. Minhas, H. S. Meyer, AIChE National Meeting, New Orleans,
Louisiana, April 2, 1992.
79. L. S. White, T. A. Blinka, H. A. Kloczewski, I. Fan Wang, J. Membr. Sci. 103 (1995)
Oct., 73–82.
80. W. J. Koros, G. K. Fleming, J. Membr. Sci. 83 (1993) 1–80.
81. T. Shiratori, H. Sonta: ‘‘Application of Iron-Oxidizing Bacteria to Hydrometallurgi-
cal Flue Dust Treatment and Hydrogen Sulfide Desulfurization,’’ FESM Micro Biol.
Rev. 11 (1–3) 1993, 165–174.
82. C. J. N. Buisman, W. L. Prins: ‘‘New Process for Biological (Flue) Gas Desulphurisa-
tion,’’ Biologische Abgasreinigung, VDI-Bericht 1104, VDI-Verlag Düsseldorf, 1994.
83. H. Satoh, J. Yoshizava, S. Kametani: ‘‘Bacteria Help Desulfurise Gas,’’ Hydrocarbon
Process 67 (1988) no. 5, 76D–76F.

SPECIFIC REFERENCES

84. J. L. Birnbaum: ‘‘The Reemergence of LNG,’’ vols. 1 and 2, Energy Research Assoc.,
New York 1989.
85. ‘‘LNG World Overview,’’ Gotaas-Larsen Shipping Corp., New York, updated annually.
86. R. N. Davis, Y.-N. Liu: ‘‘Baseload LNG: An International Energy Source,’’ 1987
Spring National Meeting of the American Institute of Chemical Engineers, Houston,
USA, Mar. 29–Apr. 2, 1987.

SPECIFIC REFERENCES

87. R. N. Davis, L. E. Olson: ‘‘The Future of Natural Gas as a Fuel in Railroad Opera-
tions,’’ 16th Energy Technology Conference (ET ’89), Washington, D.C., USA, Feb.
28–Mar. 2, 1989.
88. D. Rooke: ‘‘LNG Industry—A Retrospective,’’ 9th International Conference on Lique-
fied Natural Gas (LNG-9), Nice, France, Oct. 17–20, 1989.
89. L. S. Gaumer, C. L. Newton, US 3 763 658, 1973.
90. M. J. Roberts, R. Agrawal, US Patent 6,308,531 B1, 2001.
91. M. J. Roberts, Y. Liu, J. C. Bronfenbrenner, J. M. Petrowski, ‘‘Reducing LNG Capital
Cost in Today’s Competitive Environment’’, PS2-6, The 14th International Conference
and Exhibition on Liquefied Natural Gas (LNG14), Doha, Qatar, March 21-24, 2004.
Vol. 1 NATURAL GAS 425

92. Y.-N. Liu, J. W. Pervier, US 4 545 795, 1985.


93. I. Aoki, Y. Kitsukawa, US 3 817 046, 1974.
94. N. J. Cuperus: ‘‘Developments in Cryogenic Storage Tanks,’’ 6th International
Conference on Liquefied Natural Gas (LNG-6), Kyoto, Japan, Apr. 7–10, 1980.
95. A. C. Timmers, Y. Lev, C. R. McClune, M. J. Pikaar: ‘‘Development of Fire Preven-
tion Measures for Large LNG Storage Installations,’’ 6th International Conference
on Liquefied Natural Gas (LNG-6), Kyoto, Japan, Apr. 7–10, 1980.
96. D. R. Jenkins, A. C. Timmers: ‘‘LNG Plant Safety in the Light of Recent Experimen-
tal Trials and Theories,’’ 7th International Conference on Liquefied Natural Gas
(LNG-7), Jakarta, Indonesia, May 15–19, 1983.
97. J. A. Eyre, M. J. Pikaar, J. L. J. Rosenfield: ‘‘The Assessment of Combustion Related
Hazards Associated with the Spillage of LNG,’’ 8th International Conference on
Liquefied Natural Gas (LNG-8), Los Angeles, USA, June 15–19, 1986.
98. M. C. Parnarouskis, C. D. Lind, P. P. K. Raj, J. M. Cece: ‘‘Vapor Cloud Explosion
Study,’’ 6th International Conference on Liquefield Natural Gas (LNG-6), Kyoto,
Japan, Apr. 7–10, 1980.
99. F. Tuppeck: ‘‘Perspektiven für den interkontinentalen Transport von Energieträgern—
dargestellt am Beispiel des Erdgastransportes,’’ Erdöl Erdgas Z. 95 (1979) no. 6.
100. F. Tuppeck: ‘‘Die energiewirtschaftliche Bedeutung natürlicher und wirtschaftlicher
Einflußgrößen für die Ferngaswirtschaft,’’ (Dissertation, TH Karlsruhe), Gesam-
melte Ber. Betr. Forsch. Ruhrgas AG Essen 1959.
101. K. Liesen: Der internationale Erdgashandel und die Bedeutung des Erdgases für ein
ressourcenarmes Land, Jahrbuch für Bergbau, Energie, Mineralöl und Chemie 83/
83, Verlag Glückauf, Essen 1982.
102. K. Schwier: ‘‘Technik der LNG-Kette am Beispiel der deutschen Flüssigerdgas-
Importverträge,’’ GWF Gas- Wasserfach: Gas/Erdgas 121 (1980) no. 10.
103. H. G. Graf (ed.): ‘‘Inline-Kompressorstation in Ferngasleitungen DVGI-Symposion,’’
Erdöl Erdgas Z. 98 (1982) no. 2.
104. F. Herning: Stoffströme in Rohrleitungen, 4th ed., VDI-Verlag, Düsseldorf 1966.
105. R. Eberhard, R. Hüning: Handbuch der Gasversorgungstechnik, R. Oldenbourg Ver-
lag, München–Wien 1984.
106. H. Oermann: ‘‘Der Fernleitungstransport von Erdgas in technischer und wirtschaf-
tlicher Sicht,’’ GWF Gas- Wasserfach: Gas/Erdgas 112 (1971) no. 1.
107. H. Lübbeck, G.-J. Gralla: ‘‘Kriterien und Möglichkeiten für die Speicherung von Gas in
deutschen Erdöl- und Erdgaslagerstätten,’’ Erdöl Erdgas Z. 91 (1975) no. 11, 381–387.
108. W. F. Burke: ‘‘Simultaneous Underground Gas Storage and Secondary Oil Recov-
ery,’’ J. Technol. (1960) no. 22/26.
109. E. Schindewolf: ‘‘Untertagespeicherung–Bindeglied zwischen Erdgasbezug und
Erdgasmarkt,’’ GWF Gas- Wasserfach: Gas/Erdgas 121 (1980) no. 10, 455.
110. H. Cross: Analysis of Flow in Networks of Conduits or Conductors, Engineering Ex-
periment Station Bulletin no. 286, University of Illinois, Nov. 1936. M. S. McIllroy:
‘‘Pipeline Network Flow Analysis Using Ordinary Algebra,’’ J. Water Works Assoc.
41 (1949) no. 5, 422–428.
111. F. Tuppeck, H. Schachelhuber: ‘‘Erfahrungen bei der Berechnung von Ortsvertei-
lungsnetzen,’’ Gasverwendung 18 (1967) no. 8, 331–337.
112. F. Tuppeck: ‘‘Optimaler Aufbau der Transportkapazität von Ortsgasnetzen mit Hilfe
systematischer Rohrnetzberechnungen,’’ GWF Gas- Wasserfach: Gas/Erdgas 105
(1964) no. 19, 505–509.
113. CEDIGAZ (Centre international d’information sur le gaz naturel, Paris) Le gaz nat-
urel dans le monde en 1995, July 1996.
114. BGR (Bundesanstalt für Geowissenschaften und Rohstoffe), Reserven, Ressourcen
und Verfügbarkeit von Energierohstoffen, Hannover 1995, p. 163.
426 NATURAL GAS Vol. 1

115. A. W. Drews (ed.): ASTM Manual on Hydrocarbon Analysis, ASTM Manual Series
MNL 3, 5th ed., Philadelphia, PA, 1992.
116. DVGW-Regelwerk, Arbeitsblatt G 261: Prüfung der Gasbeschaffenheit, ZfGW-
Verlag, Frankfurt 1985.
117. K. H. Altgelt, T. H. Gouw: Chromatography in Petroleum Analysis, Marcel Dekker,
New York 1979.
118. S. H. Kägler: Neue Mineralölanalyse, Hüthig Verlag, Heidelberg 1987.
119. T. A. Norris: ‘‘Gas Analysis: Analysis of C5 and Lighter Hydrocarbons,’’ in (115), pp. 1–5.
120. ASTM D 1945 - 91 in (115), pp. 326–337.
121. C. J. Cowper, A. J. De Rose: The Analysis of Gases by Chromatography, Pergamon
Press, Oxford 1984.
122. DIN 51 405, ed. May 1987.
123. E. Leibnitz, H. J. Struppe: Handbuch der Gaschromatographie, 3rd ed., Verlag
Chemie, Weinheim, Germany 1984.
124. H. Laurien: Taschenbuch Erdgas, R. Oldenburg Verlag, München–Wien 1970.
125. DIN 51 872 Part 3, ed. Feb. 1987.
126. ASTM D 2504–83 in (115), pp. 466–469.
127. DVGW-Regelwerk, Arbeitsblatt G 285: Hydratinhibierung in Erdgasen mit Metha-
nol, ZfGW-Verlag, Frankfurt 1974.
128. ASTM E 700–79, DIN 51 777, ed. March 1983.
129. M. A. Berliner: Feuchtemessung, VEB Verlag Technik, Berlin 1980.
130. ASTM 2385–81 in (115) , pp. 428–431.
131. ASTM D 1070–85.
132. ASTM D 3588–81, DIN 51 858, ed. Nov. 1982, ISO 6976–1983.
133. ASTM D 1826–88.
134. ASTM D 1142–90, ISO 6327–1981, DIN 51 871, ed. Dec. 1985.

GEORG HAMMER
TORSTEN LÜBCKE
ROLAND KETTNER
Mobil Erdgas – Erdöl GmbH, Celle,
Federal Republic of Germany
MARK R. PILLARELLA
Air Products and Chemicals, Inc.,
Allentown, PA, United States
HERTA RECKNAGEL
Ruhrgas AG, Essen, Federal Republic of
Germany
AXEL COMMICHAU
Mobil Europe Gas Inc., The Hague, Netherlands
HANS-JOACHIM NEUMANN
German Petroleum Institute,
Clausthal – Zellerfeld, Federal Republic
of Germany
BARBARA PACZYNSKA-LAHME
Consultant, Osterode, Federal Republic
of Germany
Vol. 1 BTX PROCESSING 427

BTX PROCESSING
Benzene (71-43-2) (B), toluene (108-88-3) (T), and the xylenes (X) are the lowest
molecular weight aromatic hydrocarbon homologues. They are each very large
scale chemical feedstocks. Since they are often produced together in the same
process, they can be considered as a group, ie, BTX. However, BTX as such is
not an article of commerce. It is either an important component of a crude mix-
ture such as reformate or pyrolysis gasoline, or it is separated and purified into
its individual components. This article mainly discusses the processes for making
those crude mixtures. Other articles cover the physical and chemical properties
of the individual BTX compounds themselves (see BENZENE; TOLUENE; and
XYLENES AND ETHYLBENZENE). This article focuses on technology involving the
group as a whole.
Originally BTX was obtained commercially by pyrolysis of coal (see COAL).
Since World War II, the production of BTX has been intimately connected with
the production of gasoline. BTX constitutes part of an important gasoline compo-
nent called reformate which is discussed below (see GASOLINE AND OTHER MOTOR
FUELS). Reformate is highly valued for gasoline because it has a very high octane
rating. This results from the high concentration of aromatic compounds, all of
which have very high octane values.
Any BTX needed for chemical use is separated from the reformate stream
before it is blended into the gasoline pool. Although at a given refinery the total
volume of gasoline production (eg, 16,000 m3/d) usually dwarfs the BTX volume
(eg, 800 m3/d) and may have a higher priority, BTX production is often important
enough to support its own reforming facilities and should not be considered sim-
ply as a gasoline by-product. This independence from gasoline may be even
further emphasized in the future because of restrictions on the allowed level
of BTX in gasoline and because new BTX processes may utilize light feeds or
natural gas.
The need for BTX in gasoline has varied considerably. In the 1970s and
1980s in the United States, more high octane reformate was needed as the use
of lead antiknock compounds was decreased for environmental reasons. More
reforming capacity was put into use. Now there is environmental pressure to
reduce the aromatic content (especially benzene) of gasoline (1). This may drama-
tically reduce production of gasoline reformate. Octane number requirements
would have to be met with other high octane components such as oxygenated
hydrocarbons (eg, methyl t-butyl ether (MTBE)). Therefore, the production of
BTX for gasoline probably will drop. However, despite possible dislocations in
supply, the demand for chemical uses will still be readily satisfied and the avail-
ability and price of BTX for chemical uses probably will not be greatly affected by
the change in gasoline composition.
The principal chemical uses of BTX are illustrated in Figure 1 and listed in
Table 1 (2). A very wide range of consumer products from solvents to fibers, films,
and plastics are based on BTX. The consumption of BTX is approximately in the
proportions of 67:5:28, respectively. However, no BTX process gives BTX in these
proportions. The economic value of benzene and xylenes (especially p-xylene) is
normally higher than that of toluene. Because of this, processes that convert

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.02202419230505.a01
428 BTX PROCESSING Vol. 1

Polystyrene
Ethylbenzene Styrene ABS resins
SBR elastomers

Phenol Phenolic resins


Cumene
Methacrylates
Acetone Solvents

Cyclohexane
Adipic acid

Caprolactam } Nylon-6 and nylon-6,6

Benzene
Toluene
Dinitrotoluene Toluene diisocyanate Polyurethanes

}
0-Xylene Phthalic anhydride
Plasticizers
Polyesters and alkyd resins
Xylenes m-Xylene Isophthalic acid

p-Xylene
Dimethyl terephthalate
Terephthalic acid } Polyester fibers and films

Fig. 1. Principal uses of BTX.

toluene to benzene by hydrodealkylation (3) and disproportionate toluene to ben-


zene and xylenes (4) have been commercialized. In addition, reforming processes
that emphasize production of either benzene or p-xylene (106-42-3) have been
described (5). Since these are not classified as BTX processes they are not
discussed in detail here.

Table 1. Consumption of BTX in the Manufacture of Chemicals, 1989, 103 t


Product Worldwide United States
benzene
cumene [98-82-8] 4,319 1,437
cyclohexane [110-82-7] 3,656 915
ethylbenzene [100-41-4] 10,863 3,248
all others 3,157 568
Total (benzene) 21,995 6,168
toluene
cresol [1319-77-3] 93
phenol [108-95-2] 271 36
toluene diisocyanate [1321-38-6] 602 241
and toluenediamine [26764-44-3]
all others 564 194
Total (toluene) 1,530 471
o-xylene
phthalic anhydride [85-44-9] 1,964 393
p-xylene
dimethyl terephthalate [120-61-6] 2,349 870
terephthalic acid [100-21-0] 4,990 1,363
Total (o- and p-xylene) 9,303 2,626
Vol. 1 BTX PROCESSING 429

1. Reforming

Reforming, as currently practiced, is a platinum-catalyzed high temperature


vapor-phase process which converts a relatively nonaromatic C6 –C12 hydrocar-
bon mixture (naphtha) to an aromatic product called reformate (6). The catalyst
often contains less than 1% of platinum, possibly modified with other metals,
supported on a high surface area support such as alumina, which provides
acidity (see Catalysis). The gasoline octane rating of the reformate is directly
related to its aromatic content (Fig. 2). The aromatic content is higher when
the reformer is operated at high severity (high temperature, low space velocity).
Some cracking to light products also occurs, and this also increases at high sever-
ity. A typical reformate contains BTX in the proportions of 19/49/32, respectively,
although these proportions can be varied by tailoring the feed composition. In
response to the environmental pressure on benzene in motor gasoline mentioned
previously, it is probable that many U.S. refiners will choose to reduce the pro-
portion of benzene in their reformate by raising the cut point on the naphtha feed
to their reformers.
1.1. Feedstock. Feed for reformers is normally petroleum hydrocarbons
that boil roughly in the 70–1908C range. When this feed (straight run naphtha)
is obtained by fractionating a crude oil, its composition varies considerably with
the source of the crude as shown in Table 2 (see PETROLEUM) (7). The aromatic
content of these naphthas is quite low. Because of this and the lack of highly
branched paraffins, the octane numbers are low. Also, some naphthas have
high cycloparaffin (naphthene) contents. Since, in the reforming process, open
chain paraffins first have to cyclize to cycloparaffins before they aromatize,
those naphthas with high cycloparaffin content are generally easier to reform.
In addition to straight run naphthas, 70–1908C cuts obtained by distillation
from streams produced by cracking high boiling petroleum fractions can also be
used as feed to reformers. Naphthas produced by hydrocracking are particularly
suitable.
Impurities containing sulfur, nitrogen, and oxygen are undesirable in a
reformer feed because they harm the catalyst. Therefore, these elements are

100
Aromatics in reformate, liquid vol %

90

80

70

60

50

40

30
80 90 100 110
Octane, research

Fig. 2. Octane rating as a function of aromatics in reformate.


Table 2. Properties of Straight Run Naphthas from Various Crude Oils
Crude Naphtha
properties properties
Gravity, Sulfur, Vol % of Gravity, Octane, Paraffins, Cyclo-paraffins, Aromatics,
Name (source) 8API wt % crude 8API research wt % wt % wt %
Arab Light 33.1 1.9 17.0 59.2 33.2 72 17 11
(Saudi
Arabia)
Arab Heavy 27.4 2.8 13.0 59.9 33.9 70 19 11
(Saudi Arabia)
Iran Light (Iran) 33.5 1.4 17.8 56.2 45.9 59 28 13
Bonny Light 34.3 0.1 16.1 51.9 63.0 37 51 12
(Nigeria)
Sumatra Light 35.9 0.1 10.9 58.0 41.8 58 37 5
(Indonesia)
Ardjuna 36.0 0.1 21.7 50.9 66.5 36 45 19
(Indonesia)

430
Brent (North 38.2 0.4 20.2 54.0 62.3 47 39 14
Sea, UK)
Mayan (Mexico) 21.5 3.4 12.0 56.9 45.3 62 26 12
Isthmus (Mexico) 33.3 1.2 19.2 56.5 65.9 61 26 13
Alaska North Slope 27.6 1.1 13.3 52.1 63.2 42 41 17
(Alaska, U.S.)
Huntington Beach 21.8 1.5 14.1 48.2 70.6 14 80 6
(California, U.S.)
West Texas 39.1 0.3 23.4 54.6 53.8 45 45 10
Intermediate
(Texas, U.S.)
Empire Mix 32.3 0.3 12.7 52.4 61.2 43 44 13
(Gulf Coast,
U.S.)
Vol. 1 BTX PROCESSING 431

largely removed by pretreating the feed with hydrogen (hydrotreating) in the


presence of catalysts containing nickel–molybdenum, cobalt–molybdenum, or
combinations (6). This process converts them into hydrogen sulfide, ammonia,
and water, respectively, which are readily removed by distillation.
The molecular weight distribution of the feed affects the distribution of the
product. If the naphtha is concentrated in the C6 –C8 range, more benzene and
toluene are found in the product. If the feed is weighted to C8 –C10, more xylenes
and higher aromatics are found. Some carbon number ‘‘slippage’’ occurs by deal-
kylation: some C7s form benzene by losing a methyl group, some C8s form
toluene, etc.
1.2. Reforming Conditions. The main process variables are pressure,
450–3550 kPa (50–500 psig), temperature (470–5308C), space velocity, and
the catalyst employed. An excess of hydrogen (2–8 moles per mole of feed) is
usually employed. Depending on feed and processing conditions, net hydrogen
production is usually in the range of 140–210 m3/m3 feed (800–1200 SCF/bbl).
The C1 –C4 products are recovered and normally used as fuels.
A flow diagram for a typical semiregenerative reformer based on the Rheni-
forming Process (8) is shown in Figure 3. The hydrotreated feed is heat exchanged
with the reformer product. Further heating is achieved by passage through a fur-
nace. The feed then enters the first of several reactors which, in the Rheniforming
Process, contain platinum–rhenium on alumina catalyst. To replenish the
endothermic heat of reaction of the aromatization process, several sets (three
are shown in Fig. 3) of alternating furnaces and reactors are used.
The excess hydrogen present actually inhibits aromatics formation but is
necessary to reduce catalyst fouling by coke formation. Catalyst fouling is a
critical issue in reforming. Process conditions and catalyst compositions balance
conversion against fouling rate. Today, by using improved catalysts that foul
more slowly or by using continuous catalyst regeneration, commercial reformers
can get high conversions and yields by operating at relatively low pressures,

Recycle compressor

Booster compressor Overhead to


light ends
Once-through recovery
hydrogen and
desulfurizer
Hydrotreater offgas to fuel Stablizer
reactor Rheniformer reactors

Desulfurizer

Reformate

Straight run
naphtha

Fig. 3. Simplified process flow diagram of a naphtha hydrotreater and rheniformer.


432 BTX PROCESSING Vol. 1

C8 Aromatics, liquid vol %


20
A
18
B
16

14 C

12

10
16
Toluene, liquid vol %

A B
14 C

12

10

8
Benzene, liquid vol %

6 A
B
5 C

2
0 10 20 30 40 50 60 70 80 90 100
C7 Paraffin conversion, liquid vol %

Fig. 4. Higher yields from lower pressures; reactor pressure: A, 791 kPa (100 psig);
B, 1480 kPa (200 psig); C, 2515 kPa (350 psig). Data for Arabian Naphtha, 54–1548C
fraction.

eg, 450–965 kPa (50–125 psig). Figure 4 shows the effects of pressure (mainly
partial pressure of hydrogen) and of conversion on BTX yield.
As coke fouling occurs, the catalyst loses activity. Temperature must be
raised to keep the conversion level constant. Typically only about a 20–308C tem-
perature rise is allowable. Then the very valuable platinum catalyst must be
regenerated. Three approaches are used. If the run length (fouling time) is
about six months or more, it is practical to shut down the unit and regenerate
the catalyst in place. This approach, called semiregenerative reforming,
is used in a number of commercial processes: Houdriforming, Magnaforming,
Platforming, and Rheniforming (6).
If, because of feed type, process severity, or catalyst composition, the run
length is short, the catalyst is regenerated in swing reactors or continuously.
Processes using four to six reactors, one or more of which may be undergoing
regeneration at any given time, have been developed, eg, Powerforming and
Ultraforming (6). The catalyst has a higher average activity in these pro-
cesses than in semiregenerative ones. Continuous regeneration is offered
by UOP in the Continuous Catalyst Regeneration (CCR) Platforming Pro-
cess, and by IFP in their Aromizing Process (9). A portion of the catalyst is
Vol. 1 BTX PROCESSING 433

continuously removed, regenerated in a separate regeneration loop, and


returned to the reformer.
During regeneration the coke is burned off the catalyst. The techniques
employed are fairly sophisticated so as to maintain the platinum and any
other active metals in a well dispersed form and to restore the original catalyst
activity. Regeneration usually takes several days.
The choice of reforming process depends on the product desired, plant size,
and capital availability. If BTX is to be only a coproduct, the refiner might select
a semiregenerative process. Severity is usually lower, and the important factor is
the yield of gasoline. This yield for some combinations of feeds and catalysts can
improve with successive catalyst regeneration cycles (10). For high BTX yields a
swing reactor or continuous regeneration process might be the choice because
BTX yields are highest at high severity and low pressure.
Patents cover a new reforming catalyst based on L-zeolite which gives a sig-
nificantly higher yield of BTX, especially benzene, from light paraffinic feeds
(11). Other new zeolites (12) may also offer advantages over the traditional
reforming catalyst supports.
1.3. Reforming Chemistry. The main reactions occurring in a reformer
are shown in Figure 5 (6,13–15); most are reversible indicating the potential
importance of reaction equilibrium.
In the alkylcyclohexane (ACH) to aromatic equilibrium, ACH Ð AR, aro-
matics are favored by high temperatures and low pressures. Normal reforming
conditions promote rapid ACH dehydrogenation and a high conversion to
aromatics.
The alkylcyclopentane (ACP) to aromatics process (ACP Ð ACH Ð Ar) is
less efficient than ACH dehydrogenation, owing to the slowness of the first
step and to ACP ring opening. Under conditions where cyclohexane is converted
to benzene with close to 100% efficiency, only 50–75% of methylcyclopentane
may be converted to benzene.

Pt, acid
Isoparaffins Alkycyclopentanes
(ACP)
Pt, acid

cracked Acid Acid


products

Pt, acid
Pt
- Paraffins Alkylcyclohexanes Aromatics Lighter
(ACH) (Ar) aromatics

Hydrocracking
Paraffins Cycloparaffins Dealkylation
insomerization insomerization
Paraffins Alkylcyclohexane
dehydrocyclization dehydrocyclization

Fig. 5. Main reactions of catalytic reforming. Pt and acid refer to predominant active
catalytic sites.
434 BTX PROCESSING Vol. 1

Aromatization of isoparaffins and n-paraffins is more difficult. Not only is


the reaction slower, but also the ultimate yield is lower because more cracking
occurs. Under mild reforming conditions a straight run naphtha with a high
cycloparaffin (naphthene) plus aromatic content achieves a much higher octane
rating (and aromatics content) than one with a high paraffin content. This latter
naphtha can be more severely reformed to achieve a high octane rating, but then
the yield of liquid product is lower.

2. Hydrocarbon Pyrolysis

A large amount of BTX is obtained as a by-product of ethylene manufacture (see


Ethylene). The amount produced strongly depends on the feed to the ethylene
plant. This is illustrated in Table 3 for various feeds to a typical large scale
plant producing 450,000 t/yr of ethylene (16). Note that only about 1–2% of
the ethane/propane feeds end up as BTX and it is almost completely benzene
and toluene. As the feed goes up in molecular weight, the yield of BTX increases
from 4% with butane feed to about 10% with gas oils, and the BTX proportions go
from 72:20:8 respectively, to 44:34:22 respectively.
Outside the realm of typical hydrocarbon pyrolysis is the high temperature
pyrolysis of methane. In one variant of this process, which has only been
commercialized to produce acetylene (with some BTX), methane reacts in an
electric arc at about 15008C (17) with very short contact times. At higher tem-
peratures or with a catalyst and added hydrogen, BTX is produced with fairly
high selectivity (18).

3. BTX from Light Hydrocarbons

A completely new approach for BTX production has emerged in recent years. It
converts C2 to C6 paraffins into aromatics using a modified ZSM-5 zeolite catalyst
which contains gallium (19). An example of this approach, the Cyclar process,
has been in commercial operation by British Petroleum at Grangemouth,

Table 3. BTX Yields from Various Pyrolysis Feeds,a,b 103 t/yrc


Gas oil, distilled
Mid-crude C6 –C8
Feed C 2 H6 C3 H8 n  C4 H10 naphtha Raffinate Atm Vac
ethylene 450 450 450 450 450 450 450
product rate
feed rate 583 1080 1135 1349 1535 1750 2213
ethylened 48.2 34.2 35.8 30.0 26.0 23.0 18.0
benzene 5.0 26.6 34.3 90.0 72.7 105.5 82.5
toluene 0.7 5.8 9.4 45.1 41.5 50.8 64.3
C8 aromatics 4.0 23.8 18.4 38.0 41.4
a
Ethane recycled to extinction.
b
Ref. 16.
c
Unless otherwise noted.
d
Once through yield.
Vol. 1 BTX PROCESSING 435

Scotland since August 1990 (20). It uses C3 –C4 feed and employs UOP’s CCR
technology to compensate for rapid catalyst coking.
The mechanistic steps are as follows: paraffins dehydrogenate to olefins; the
olefins oligomerize and cyclize; and the cyclics aromatize. Because the first step
is rate controlling, very little olefin is actually present. The BTX product is rela-
tively free of nonaromatics and therefore is very desirable as a chemical feed. As
in reforming, some C1 –C2 fuel gas is produced along with a valuable hydrogen
stream. From a C3 –C4 feed the BTX product is roughly 35:45:20, respectively.

4. Other BTX Processes

Because of the importance of the petroleum-based processes discussed


previously, only about 1% of the U.S. supply of BTX currently comes from coal
pyrolysis (21).
Outside the United States, coal pyrolysis is more important as a source of
BTX. The proportions are about 70:20:10, but can vary greatly depending on the
coal and on the pyrolysis process used. Product quality is not as good as petro-
leum-derived BTX. This source could become more important again if petroleum
costs escalate. Much higher yields of BTX from coal can be obtained by first
hydrogenating the coal (22).
Another very interesting route to BTX is the Mobil Methanol To Gasoline
(MTG) Process (23). Methanol is converted into gasoline containing about 50%
aromatics by passing it over a ZSM-5 catalyst. The BTX composition is about
10:20:70, respectively. More than half of the aromatics present are C9s and
C10s, and these would have to be dealkylated to yield more BTX. The methanol
can be obtained from any syngas source such as coal, natural gas, or petroleum
(see Methanol). The economic attractiveness of this approach depends on the cost
of methanol relative to gasoline from petroleum. A commercial plant has been
operating successfully since 1985 in New Zealand where natural gas is abundant
and petroleum is scarce (24).

5. BTX Recovery

The complexity of separating and purifying the individual BTX components from
crude BTX products depends on the amount of nonaromatic impurities present.
If the amount is small enough, simple distillation can suffice. If not, it is obvious
from Figure 6 that distillation alone will not be sufficient because the BTX
aromatic compounds are close in boiling point to some of the cycloparaffins of
the same carbon number or to paraffins of the next higher carbon number.
Because of this, extraction or extractive distillation with a polar solvent is
used to separate the slightly polar aromatic hydrocarbons from the nonpolar non-
aromatic hydrocarbons. When olefins are also present, as in pyrolysis gasoline,
these interfere with the extraction and must first be hydrogenated or removed
by adsorbents.
An option for avoiding the cost of extraction is to increase the severity of the
BTX formation step. This reduces the quantity of residual paraffins, and,
436 BTX PROCESSING Vol. 1

200

Temperature,°C 150

100

50

P C A P C A P C A P C A
C6 C7 C8 C9

Fig. 6. Boiling points of C6 –C9 hydrocarbons. P, iso and normal paraffins; C, C5- and
C6-cycloparaffins; and A, aromatics.

depending on the BTX formation process, may leave the BTX clean enough to
purify by distillation. The final impurity concentrations may still be too high
for merchant sale, but may be acceptable in some downstream operations. For
example, xylenes going into an isomerization/separation loop for p-xylene pro-
duction can contain some paraffins if the isomerization catalyst is capable of
decomposing them. Disadvantages of high severity processing are the increased
catalyst fouling rate and the potential increase in undesirable olefin impurities.
5.1. Extraction and Extractive Distillation. The choice of an extrac-
tion or extractive distillation solvent depends upon its boiling point, polarity,
thermal stability, selectivity, aromatics capacity, and upon the feed aromatic
content. Capacity, defined as the quantity of material that is extracted from
the feed by a given quantity of solvent, must be balanced against selectivity,
defined as the degree to which the solvent extracts the aromatics in the feed in
preference to paraffins and other materials. Most high capacity solvents have
low selectivity. The ultimate choice of solvent is determined by economics. The
most important extraction processes use either sulfolane or glycols as the polar
extraction solvent.
Sulfolane [126-33-0], used in UOP and Shell processes (25,26), offers good
thermal and hydrolytic stability, high density and boiling point, and a good bal-
ance of solvent properties. Its high density and boiling point make it easy to sepa-
rate from the hydrocarbon streams. A diagram of a sulfolane extraction unit is
shown in Figure 7. Fresh feed enters the extractor and flows countercurrent to
Vol. 1 BTX PROCESSING 437

Raffinate

Rich Extract to
solvent treating and
fractionation

E ED RC

Feed

Recycle

Lean solvent

Fig. 7. Shell sulfolane extraction process. E, extraction; ED, extractive distillation; RC,
recovery column. (Courtesy of UOP, Inc.)

the down flowing solvent. The raffinate is withdrawn at the top of the extractor
and leaves the system after water washing. The solvent, now rich in aromatics, is
sent to the top of the extractive stripper where the nonaromatic hydrocarbons
are removed. Aromatics and sulfolane are separated in the recovery column.
The lean solvent is recycled to the extractor and the aromatics are washed
with water and removed. The recovery of benzene and toluene is usually
99þ%; of C8 aromatics 97%; and of Cþ 9 aromatics 75–90%.
The widely employed UOP Udex Process uses a glycol solvent (27). Diethy-
lene glycol was used in early versions of the process; however, increased capacity
was obtained by adding dipropylene glycol or, in some cases, a change was made
to triethylene glycol. Further improvement was made by using tetraethylene gly-
col (28). The Union Carbide Tetra Process also employs tetraethylene glycol (29).
Other extraction processes are included in Table 4.
Extractive distillation, using similar solvents to those used in extraction,
may be employed to recover aromatics from reformates which have been prefrac-
tionated to a narrow boiling range. Extractive distillation is also used to recover
a mixed benzene–toluene stream from which high quality benzene can be pro-
duced by postfractionation; in this case, the toluene product is less pure, but is
still acceptable as a feedstock for dealkylation or gasoline blending. Extractive
distillation processes for aromatics recovery include those listed in Table 4.
5.2. Downstream Processing. In addition to extraction, various down-
stream operations are often carried out on the BTX product to produce products
in proportions to fit the market demand. A typical aromatics processing scheme
is shown in Figure 8 in which benzene, p-xylene, and o-xylene are the products.
438 BTX PROCESSING Vol. 1

Table 4. Extractive Processes for BTX Recovery


CAS Registry
Company process Solvent Number Reference
Extraction
Shell Process sulfolane [126-33-0] (25,26)
UOP Udex Process diethylene glycol [111-46-6] 27
triethylene glycol [112-27-6]
tetraethylene glycol [112-60-7] 28
Union Carbide tetraethylene glycol [112-60-7] 29
Tetra Process
Lurgi Arosolvan N-methyl-2-pyrrolidinone [872-50-4] (30,31)
and monoethylene glycol [107-21-1]
Institut Français du dimethyl sulfoxide (DMSO) [67-68-5] 32
Petrôle
SNAM Progetti Formex N-formylomorpholine [4394-85-8] 33
Howe-Baker Aromex diglycolamine [929-06-6] 34
Krupp-Koppers Morphylex N-formylmorpholine [4394-85-8] 35
Extractive distillation
Institut Français du dimethylformamide [68-12-2] 5
Petrôle DMF (DMF)
Krupp-Koppers Octenar N-formylmorpholine [4394-85-8] 5
Lurgi Distapex N-methyl-2-pyrrolidinone [872-50-4] 5
UOP Sulfolane sulfolane [126-33-0] 5

Fig. 8. General BTX processing sequence.


Vol. 1 BTX PROCESSING 439

After the crude BTX is formed, by reforming in this case, a heart cut is
sent to extraction. Actually, the xylenes and heavier components are often
sent to downstream processes without extraction. The toluene produced is con-
verted to benzene, a more valuable petrochemical, by running it through a
hydrodealkylation unit. This catalytic unit operates at 540–8108C with an
excess of hydrogen. Another option is to disproportionate toluene or toluene
plus C9 aromatics to a mixture of benzene and xylenes using a process such
as UOP’s Tatoray or Mobil’s Selective Toluene Disproportionation Process
(STDP) (36).
The o-xylene (95-47-6) in Figure 8 is recovered by a two-stage distillation.
First it is separated (or split) from m-xylene (108-38-3) and the other C8
aromatics in a superfractionating column, the xylene splitter, (Unit H). The
bottoms, a mixture of o-xylene and Cþ 9 aromatics, is redistilled (or rerun) in
Unit I to recover o-xylene of 96 þ% purity.
The distillate (overhead) from Unit H, containing mostly ethylbenzene
(100-41-4), p-xylene, and m-xylene, and some o-xylene becomes the feed for the
p-xylene separation process (Unit J).
p-Xylene (106-42-3) can be purified by crystallization or adsorption.
When a typical reformate-derived C8 aromatic mixture is cooled, p-xylene
crystallizes first. Most plants employing crystallization operate at 60 to
75%, depending on feed composition (37). The process is limited by a eutectic
temperature below which o- or m-xylene also crystallize. The solubility of
p-xylene in the remaining C8 aromatic mixture over the range of 60 to
758C is 9.6 to 6.2%.
UOP’s ParexProcess can be used to purify p-xylene by adsorption (38).
Toray has a similar process. These processes take advantage of the fact that
p-xylene is adsorbed more easily than the other C8 aromatics by a suitable mole-
cular sieve. The p-xylene is desorbed by either a lighter or heavier hydrocarbon
which is subsequently removed by distillation. p-Xylene is recovered in about
97% yield (see Adsorption).
The mother liquor obtained from the crystallization, or the raffinate after
removal by adsorption, is isomerized using an acidic catalyst to convert m-xylene
to the o- and p-isomers (Unit K in Fig. 8).
In the isomerization step, light and heavy impurities are generated,
including saturated hydrocarbons, benzene, toluene, and Cþ 9 aromatics.
These are removed by recycling the isomerate through the xylene splitter
and the light isomerate column (Unit L). If the light isomerate contains
much benzene or toluene, it can be recycled to Unit G. m-Xylene may be recov-
ered between the xylene splitter and the p-xylene separation plant by selective
sulfonation followed by regeneration or via complex formation with a mixture
of HF and BF3 (39).
To this point the presence of ethylbenzene in the mixed xylenes has been
ignored. The amount can vary widely, but normally about 15% is present.
The isomerization process must remove the ethylbenzene in some way to
ensure that it does not build up in the isomerization loop of Figure 8. The
ethylbenzene may be selectively cracked (40) or isomerized to xylenes (41)
using a platinum catalyst. In rare cases the ethylbenzene is recovered in high pur-
ity by superfractionation.
440 BTX PROCESSING Vol. 1

There are many variations of the basic processing loop shown in Figure 8.
Processing to produce only BT is common, often in conjunction with a toluene-
to-benzene dealkylation unit. If benzene and toluene are not to be recovered,
Column B may be used to remove toluene and lighter components. In
that case, Units E, F, and G would be eliminated. To recover p-xylene only,
the xylene splitter is reduced in size and is used to split between o-xylene and
the Cþ
9 aromatics. The o-xylene rerun still is then eliminated.

6. Environmental Considerations

BTX processing has come under steadily increasing pressure to reduce emissions
and workplace exposures (see Industrial hygiene). Reductions in the permissible
levels of both benzene and total aromatics (BTX) in gasoline have been legislated.
Whereas all BTX components are to be controlled, the main focus is on benzene
because it is considerably more toxic than the others and is classified as a known
carcinogen (42).
Workplace exposure limits for benzene have been regulated to levels as low
as 0.5 ppm (43). Industrial emissions affecting the public are now low enough
that the EPA considers that a greater hazard exists from mostly indoor sources
such as smoking, automobile exhausts, and consumer products (44).
The stringent controls over manufacturing and handling benzene may dis-
suade refiners from installing new equipment to extract benzene from gasoline to
satisfy gasoline requirements. Instead they will probably modify their reforming
operations to produce less benzene or to convert the remaining benzene to other
compounds. They will achieve the necessary octane rating for gasoline via
branched paraffins and oxygenates. Overall, this is likely to result in further
segregation of petrochemicals BTX production from gasoline production.

BIBLIOGRAPHY

‘‘BTX Processing’’ in ECT 3rd ed., Vol. 4, pp. 264–277, by D. L. Ransley, Chevron Research
Company.

CITED PUBLICATIONS

1. Oil Gas J. 87, 30 (Sept. 11, 1989).


2. 1989 Petrochemical Report, SRI International, Palo Alto, Calif., 1989; 1989 World
Benzene Analysis, Chemical Marketing Association, Inc., Houston, Tex., 1989;
1990–91 Toluene/Xylenes Annual, DeWitt & Co., Houston, Tex., 1990.
3. A. P. Dossett, in E. G. Hancock, ed., Toluene, the Xylenes, and Their Industrial
Derivatives, Elsevier, Amsterdam, The Netherlands, 1982, ‘‘Chapt. 5’’.
4. P. Grandio and co-workers, Hydrocarbon Process., 85 (Aug. 1972).
5. K. J. Day and T. M. Snow, in Ref. 3, Chapt. 3.
6. D. M. Little, Catalytic Reforming, PennWell, Tulsa, Okla., 1985.
7. Data from Chevron Research and Technology Co. Crude Analysis Library, 1990.
8. T. R. Hughes, R. L. Jacobson, K. R. Gibson, L. G. Schornack, and J. R. McCabe, Oil
Gas J. 74, 121 (May 17, 1976).
Vol. 1 BTX PROCESSING 441

9. P. Bonnifay and co-workers, Oil Gas J. 74, 48 (Jan. 19, 1976).


10. T. R. Hughes and co-workers, Hydrocarbon Process., 75 (May 1976).
11. D. V. Law, P. W. Tamm, and C. M. Detz, Energy Progress 7, 215 (1987).
12. U.S. Pat. 4,347,394 (Aug. 31, 1982), C. M. Detz and L. A. Field (to Chevron
Research Company).
13. J. Henningsen and M. Bundgaard-Nielson, Brit. Chem. Eng. 15, 1433 (Nov. 1970).
14. W. S. Kmak and A. N. Stuckey, American Institute of Chemical Engineers Meeting,
New Orleans, La., Mar. 14, 1973, Paper 56A.
15. A. M. Kueelman, Hydrocarbon Process., 95 (Jan. 1976).
16. J. G. Freiling and A. A. Simone, Oil Gas J. 71, 25 (Jan. 1, 1973).
17. R. N. Shreve and J. A. Brink, Jr., Chemical Process Industries, 4th ed., McGraw-Hill,
New York, 1977, ‘‘Chapt. 7’’.
18. J. Oro and J. Han, Science 153, 1393 (1966); U.S. Pat. 4,814,533 (Mar. 21, 1989),
L. Devries and P. R. Ryason (to Chevron Research Company).
19. N. S. Gnep and co-workers, Appl. Catal. 43, 155 (1988).
20. P. C. Doolan and P. R. Pujado, Hydrocarbon Process., 72 (Sept. 1989).
21. 1989 Petrochemical Report, SRI International, Palo Alto, Calif., 1989.
22. W. R. Ladner, J. O. H. Newman, and P. W. Sage, J. Inst. Energy 53, 76 (1980).
23. C. D. Chang and A. J. Silvestri, J. Catalysis 47, 249 (1977).
24. Oil Gas J. 83, 52 (Oct. 28, 1985).
25. C. E. Beecher, R. E. Mann, and M. L. Renquist, Oil Gas J. 63, 80 (Nov. 22, 1965).
26. Hydrocarbon Process., 194 (Sept. 1982).
27. Hydrocarbon Process., 248 (Sept. 1970).
28. G. S. Somekh and B. I. Friedlander, Hydrocarbon Process., 127 (Dec. 1969).
29. Hydrocarbon Process., 195 (Sept. 1982).
30. E. Muller, Chem. Ind. (London), 518 (June 2, 1973).
31. E. Muller and G. Hochfield, 7th World Petrol. Congr. 4, 13 (1967).
32. B. Choffe and co-workers, Hydrocarbon Process., 188 (May 1966).
33. E. Cinelli and co-workers, Hydrocarbon Process., 141 (Apr. 1971).
34. W. T. Jones and V. Payne, Hydrocarbon Process., 91 (Mar. 1973).
35. M. Stein, Hydrocarbon Process., 139 (Apr. 1973).
36. Hydrocarbon Process., 93 (Nov. 1989).
37. D. L. McKay, G. H. Dale, and D. C. Tabler, Chem. Eng. Progress 62, 104 (1966).
38. M. Seko, T. Miyake, and K. Inada, Hydrocarbon Process., 133 (Jan. 1980).
39. T. Ueno, in T. C. Lo, and co-eds. Handbook of Solvent Extraction, John Wiley & Sons,
Inc., New York, 1983, p. 575.
40. D. B. Broughton, 186th ACS National Meeting, Am. Chem. Soc. Div. Pet. Chem.
Prepr. 28(4), 1072 (Aug. 1983).
41. C. V. Berger, Hydrocarbon Process., 173 (Sept. 1973).
42. K. B. Clansky, ed., A Guide to Industrial Chemicals Covered under Major Federal
Regulatory and Advisory Programs, Roytech, Burlingame, Calif., 1990.
43. California OSHA Regulations, Sec. 5218, 1990.
44. J. Raloff, Sci. News Lett. 136, 245 (Dec. 1989).

W. A. SWEENEY
P. F. BRYAN
Chevron Research and
Technology Company
442 FLUID CATALYTIC CRACKING, UNITS, REGENERATION Vol. 1

FLUID CATALYTIC
CRACKING, UNITS,
REGENERATION
1. Introduction

Fluid catalytic cracking (FCC) is the central process in a modern, gasoline-


oriented refinery. In U.S. refineries, the amount of feed processed by fluid cata-
lytic cracking units (FCCU) is equivalent to 34% of the total crude oil processed
in the United States (1). As of January 2003, installed FCCU capacity in the
United States was 9.0  105m 3/day (5.7  106 barrels/day calendar day basis).
The popularity of the catalytic cracker stems from its ability to produce
large quantities of high octane gasoline and other valuable light products and
to use a wide range of refinery process streams as feed materials. The FCCU
feeds include high molecular-weight vacuum gas oils (VGO), which are tradi-
tional feeds; atmospheric and vacuum residues; coker gas oils; gas oils from
other thermal and hydrocracking operations; lube oil extracts; and deasphalted
oils. A survey of U.S. refiners conducted in the mid-1990s indicated that 39% of
the U.S. gasoline pool was produced by the FCCU (2).
Although the FCCU has long been an important refinery processing tool
in the United States, refinery gasoline demands outside the United States have
been satisfied largely from the gasoline naturally present in crude oil (3). This situa-
tion is changing, and the need for catalytic cracking is now growing steadily every-
where. As of January 2003, the installed worldwide FCCU capacity stood at
22.6  105 m3/day (14.2  106 barrels/day). Worldwide, excluding North America,
and also excluding the Former Soviet Union and mainland China, for which data
did not exist 10 years ago, the installed FCC capacity increased nearly 60% in
the 10-year period from 1981 to 1991, and another 22% in the period from 1992 to
2002 (4,5).
The FCC process is highly complex but self-contained. A typical UOP FCC
unit of 2002 design is shown in Figure 1a. Figure 1a was originally prepared by
UOP for inclusion in the 3rd edition of Meyer’s Handbook of Petroleum Refining
Processes, currently in press. The 2002 design is contrasted to a mid-1970s FCC
unit in Figure 1b (6). In the 2002 design, hot catalyst from the regenerator enters
the base of the riser and is carried up the lower portion of the riser with an accel-
erating media, typically steam. The accelerating media conditions the catalyst
flow pattern so that catalyst is uniformly dispersed across the riser cross-section.
At the end of the acceleration zone, feed is injected into the riser through a series
of feed distributors positioned around the riser circumference. Sufficient pres-
sure drop is taken across the distributors to produce an atomized spray that cov-
ers the riser cross section and provides intimate contact with the upflowing
catalyst. This careful orchestration of feed and catalyst contacting contrasts
with the less elegant procedure used in the 1970s design. At that time feed,
and catalyst mixed at the base of the riser, with no attempt at catalyst dispersion
and little effort made toward feed dispersion. The improved contacting in the
2002 design has resulted in decreased coke deposition on the catalyst, which

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.0612210921161915.a01.pub2
Vol. 1 FLUID CATALYTIC CRACKING, UNITS, REGENERATION 443

Fig. 1. (a) Modern UOP FCC unit with: Short contact time riser VSS disengager
combustor-style regenerator. (b) Mid-1970s UOP FCC unit with: Riser cracking Stacked
reactor bubbling-bed regenerator.

leads to lower regenerator temperatures, higher catalyst circulation rates, and


higher and more selective conversion of the feed (7).
Catalyst and hydrocarbon vapors pass up the riser in a dilute-phase, plug
flow. In the 2002 design, catalyst and hydrocarbon vapors are rapidly separated.
Rapid separation is desired to minimize back mixed, nonselective reactions that
can occur when catalyst and hydrocarbon leave the plug flow riser environment.
These nonselective reactions also result in further unwanted increases in coke on
catalyst. Rapid separation is achieved by discharging the riser components
directly into a form of a vortex (cyclone) separator. A hydrocarbon residence
time in this style separator of 2 s have been reported, compared to a typical
residence time of 20–30 s for the 1970 separator designs (8).
When in contact with the catalyst, the hydrocarbon feed cracks to form
gasoline, middle distillates, liquefied petroleum gas (LPG), and fuel gas. Coke,
which is also formed, deposits on the catalyst surface and reduces the catalyst
activity. To restore its activity, the catalyst passes into the regenerator section
where the coke is burned off. The temperature of the catalyst is increased
150–2008C as a result of the coke burn. Care must be taken on the reactor
side to control coke production, so that the heat released in the regenerator
does not raise the regenerator temperature beyond the metallurgical limitations
of the regenerator. The 2002 UOP design utilises a high efficiency, combustor-
style regenerator, which typically operates in the fast-fluidized, catalyst
444 FLUID CATALYTIC CRACKING, UNITS, REGENERATION Vol. 1

transport regime. By contrast, the typical 1970s style regenerator utilized a bub-
bling fluidized bed for air/catalyst contacting.
The heat released from coke burning is used to provide the heat require-
ments on the reactor side: to vaporize the feed, to compensate for the endother-
mic heat of cracking, to bring the reactor to the desired reactor temperature, and
to compensate for heat losses to the atmosphere. Thus the FCCU always operates
in complete heat balance at any desired hydrocarbon feed rate and reactor tem-
perature. In FCC units, such as the ones shown in Figure 1, riser temperature
control adjusts the catalyst circulation rate to achieve the heat balance. Catalyst
flow is controlled by a slide valve located in the catalyst transfer line from the
regenerator to the reactor and in the catalyst return line from the reactor to
the regenerator. In some older style units, where catalyst flow is controlled by
pressure balance between the reactor and regenerator, the heat-balance control
is more often achieved by changing the temperature of the hydrocarbon feed
entering the riser.
Prior to entering the regenerator, the spent catalyst from the reactor passes
through a steam stripper where the hydrocarbon vapors that are entrained by
the flowing catalyst particles are displaced by steam. The steam also acts to
desorb some of the lighter hydrocarbons that are adsorbed during the reaction
step. These stripped materials, which represent valuable reaction products,
are carried back into the reactor by the upflowing steam from the stripper and
go to the product-recovery section with the other reaction products. The removal
of these materials from the catalyst stream flowing to the regenerator also
reduces the amount of burnable hydrocarbon going to the regenerator, which
helps to decrease the regenerator temperature.

2. FCCU Heat Balance

Because of the thermal coupling of reactor and regenerator, any change on


the reactor side creates a rapid change on the regenerator side, which, in turn,
influences the reactor side, and vice versa. This dynamic interaction rapidly
comes to equilibrium, and the catalytic cracker adjusts to a new steady-state,
heat-balanced condition. The first law of FCCU catalytic cracking is that the
FCCU will always adjust itself to stay in heat balance (9). If an FCCU is not
in heat balance, it is out of control. The operating characteristics of an FCCU
regenerator are thus dictated by the constraints of the heat balance (Fig. 2).
Considering the overall balance, the burning of coke in the regenerator pro-
vides all of the energy required to raise the temperature of the reactor hydrocar-
bon feed and products to the reactor vessel temperature; raise the temperature of
the combustion air from the temperature of the air blower discharge to the tem-
perature of the regenerator flue gas; provide the endothermic heat of reaction;
provide the heat that is removed from the system, including natural heat loss
from the reactor and regenerator to the atmosphere as well as intentional heat
removal from catalyst coolers.

HCoke combustion ¼ HProducts-feed þ HAir þ HSteam þ HCracking þ HLoss ð1Þ


Vol. 1 FLUID CATALYTIC CRACKING, UNITS, REGENERATION 445

Fig. 2. Heat balances of the FCCU. Heat balance around the reactor A, heat balance
around the regenerator B, and the overall heat balance around the entire catalytic cracker
C must be considered.

Thus the amount of heat that must be produced by burning coke in the re-
generator is set by the heat balance requirements and not directly set by the
coke-making tendencies of the catalyst used in the catalytic cracker or by the
coking tendencies of the feed. Indirectly, these tendencies may cause the cracker
operator to change some of the heat-balance elements, such as the amount of
heat removed by a catalyst cooler or the amount put into the system with the
feed, which would then change the amount of heat needed from coke burning.
If FCCU operations are not changed to accommodate changes in feed or cat-
alyst quality, then the amount of heat required to satisfy the heat balance essen-
tially does not change. Thus the amount of coke burned in the regenerator
expressed as a percent of feed does not change. The consistency of the coke
yield, arising from its dependence on the FCCU heat balance, has been classified
as the second law of catalytic cracking (9).

3. Coke Formation

The coke burned in an FCCU regenerator is a poorly defined, hydrocarbonaceous


material that has either not been desorbed from the catalyst surface or has not
been purged from between catalyst particles during the passage of the catalyst
through the steam stripper. This coke originates from four different sources (10):
(1) catalytic coke is produced directly from the acid catalyzed cracking reaction;
(2) contaminant coke arises from the dehydrogenation activity of contaminating
446 FLUID CATALYTIC CRACKING, UNITS, REGENERATION Vol. 1

Table 1. FCCU Coke Distributions,a % of Total Coke


Feed type Catalytic Contaminent Additive Cat/oil
VGO 65 15 5 15
resid 29 29 35 7
a
Ref. 8 and 10.

metals (principally nickel and vanadium) in the feed to the FCCU; (3) additive
coke is directly related to the high boiling, refractory components in the feed
to the FCCU. This coke correlates with the basic nitrogen and molecular weight
of the feed (11) and with the carbon residue content of the feed (10); and
(4) cat-to-oil coke is related to the catalyst circulation rate and derives
from the hydrocarbons leaving the stripper. These hydrocarbons are potentially
strippable but because of incomplete stripping are entrained by the flowing cat-
alyst into the regenerator. FCCU coke distributions for operations with zeolitic
catalysts are given in Table 1 (10,12).
The amount of catalytic coke that is formed depends on the type of catalyst
used in the FCCU, the coking tendency of the feed, the degree of conversion of
the feed, and the length of time the catalyst is exposed to the feed (eq. 2) (13).

Coke on catalyst ¼ kc t1=2 ð2Þ

Coke deposition is essentially independent of space velocity. These observa-


tions, which were developed from the study of amorphous catalysts during the
early days of catalytic cracking (13), still characterize the coking of modern
day zeolite FCC catalysts over a wide range of hydrogen-transfer (H-transfer)
capabilities.
Coke on the catalyst is often referred to as delta coke (D C), the coke content
of the spent catalyst minus the coke content of the regenerated catalyst. Delta
coke directly influences the regenerator temperature and controls the catalyst
circulation rate in the FCCU, thereby controlling the ratio of catalyst: hydrocar-
bon feed (cat/oil ratio, or C/O). The coke yield as a fraction of feed CF is related to
delta coke through the C/O ratio (eq. 3).
 
C
CF ¼ C ð3Þ
O

The catalytic conversion of gas oil is well approximated by a second-order


reaction, where, sv ¼ space velocity (14).

conv k
conversion2nd order ¼ ¼ ð4Þ
100  conv sv

Space velocity is related to the C/O ratio through the catalyst residence time as

C 1
¼ ð5Þ
O ðsvÞt
Vol. 1 FLUID CATALYTIC CRACKING, UNITS, REGENERATION 447

Combining equations 3–5 yields

C
CF ¼ conv2nd Order ð6Þ
kt

Equation 6 relates the catalytic coke yield (as a fraction of the feed) to the delta
coke, to the conversion, and to the catalyst residence time.
It has been shown that coke yield as a fraction of feed does give a linear
relationship with second-order conversion (15) indicating a positive coke yield
at zero conversion. This coke yield at zero conversion is the additive coke contri-
bution to the total coke yield and is related to feed properties, particularly
Conradson carbon content. The amount of this additive coke is significantly
less than the Conradson carbon value of the feed (16), probably in the range of
50% of the Conradson carbon.
Catalytic coke is also a function of the type of feed being processed. The
more aromatic the feed, the higher the coke (on feed) yield at a given conversion
(15). The type of cracking catalyst also influences the coke yield. Increasing
the H-transfer characteristics of a Y-type zeolite, by increasing the amount of
rare earth exchanged on the zeolite, results in an increase in the catalytic coke
yield (17). Modern catalysts for FCC applications frequently contain two dis-
tinctly different, active components; an active zeolite component for producing
high yields of gasoline while yielding minimum catalytic coke and an active,
nonzeolitic component, typically a catalytically active alumina located in the
so-called matrix portion of the catalyst surrounding the zeolite. The active alu-
mina matrix provides improved conversion of the heavy portion of the feed but
typically increases catalytic coke yield at constant conversion (18).
Contaminant coke can vary substantially because of large variations in the
metals content of the feed being processed and from variations in the manner
that cracking catalysts respond to these metals. Nickel is considered the most
active dehydrogenation agent, and hence the most troublesome of the contami-
nant metals. Copper is equally active as a dehydrogenation catalyst, but the
amount of copper present is usually too small to have an effect. Typically, copper
in the feed to the FCCU is only 2–5% of the nickel content. Vanadium is a less
active dehydrogenation agent than nickel, about one-fourth as effective as nickel
(10). The metals content of the equilibrium catalyst in the FCCU is frequently
referred to as the equivalent nickel content, which is defined as

V
equivalent nickel ¼ Ni þ ð7Þ
4

The equivalent nickel content of the feed to the FCCU can vary from
<0.05 ppm for a well-hydrotreated VGO to >20 ppm for a feed containing a high
resid content. The nickel and vanadium deposit essentially quantitatively on the
cracking catalyst and, depending on catalyst addition rates to the FCCU, result
in total metals concentrations on the equilibrium catalyst from 100 to 10,000 ppm.
The coking characteristics of a cracking catalyst increase as the equivalent
nickel level on the catalyst increases. At high equivalent nickel levels, contami-
nant coke can be the major coke component on the catalyst. The effect of the
448 FLUID CATALYTIC CRACKING, UNITS, REGENERATION Vol. 1

contaminant nickel on the catalyst coke content can be mitigated through the use
of a metal passivator added to the feed or through the incorporation of an effec-
tive metals traps into the cracking catalyst. Coking problems due to contaminant
vanadium still remain a problem.
Feed additives were frequently used in the 1980s and early 1990s to reduce
the contaminant coke produced through nickel-catalyzed reactions. These inhibi-
tors were injected into the feed stream going to the catalytic cracker. The addi-
tive forms a nickel complex that deposits the nickel on the catalyst in a less
catalytically active state. The first such additive was an antimony compound
developed and first used in 1976 by Phillips Petroleum. During the 1980s, anti-
mony was widely used in FCCUs that had a problem with contaminant nickel.
The use of the antimony additive reportedly reduced coke yields by 15% in a com-
mercial trial (19). In the late 1980s, other additives were introduced to combat
contaminant nickel, eg, Chevron introduced a bismuth-based additive, which is
claimed to provide performance similar to antimony (20).
In the late 1980s, cracking catalysts were developed with metals traps that
appear to be so effective in containing the adverse effects of contaminant nickel
that additive-type inhibitors are no longer needed (21). These traps consisted of
reactive alumina. It has been reported that these aluminas promote a chemical
reaction with the nickel to form NiAl2O4 which then diffuses into the bulk alu-
mina phase, thereby minimizing any contact between nickel and the potential
coke-forming hydrocarbons (22). By the late 1990s, all FCC catalyst manufac-
turers incorporated such a trap into their metals resistant catalysts.

4. Coke Burning

The burning of coke in the regenerator provides the heat to satisfy the FCCU
heat balance requirements as shown in equation 1. The heat released from the
burning of coke comes from the reaction of carbon and hydrogen to form carbon
monoxide, carbon dioxide, and water. The heat generated from burning coke thus
depends on the hydrogen content of the coke and the relative amounts of carbon
that burn to CO and CO2, respectively.

Reaction DHcomb kJ/kg of C or H2


H2 þ 12 O2 ! H2O (g) 121,000
C þ O2 ! CO2 (g) 32,700
C þ 12 O2 ! CO (g) 9,200

The impact that variations in coke content and burning conditions can have
on the overall heat of coke combustion is shown in Table 2.
Because the heat balance dictates the amount of heat that is required from
burning coke, the heat of combustion then determines the amount of coke that
must be burned.
combustion heat required; kJ=h
coke required; kg=h ¼ ð8Þ
specific heat of combustion; kJ=kg
Vol. 1 FLUID CATALYTIC CRACKING, UNITS, REGENERATION 449

Table 2. Specific Heat of Coke Combustion


H2, wt% in coke Flue gas, CO2/ CO Specific heat kJ/ kg cokea
12 1 43,300
7 1 38,900
7 1.0 28,000
a
To convert kJ to kcal, divide by 4.184.

Decreasing the specific heat of combustion increases the amount of the


coke that must be burned; the coke component that increases is essentially
the cat-to-oil coke, which is increased by increasing the catalyst circulation
rate. Cat-to-oil coke can be expressed as (23):

coke ¼ f ½ðcatalyst circulation rateÞ0:65 ð9Þ

Thus decreasing the specific heat of combustion results in an increase in catalyst


circulation rate. Because of this relationship to coke yield (eq. 9), the increase in
the catalyst circulation rate results in a decrease in regenerator temperature.
The hydrogen content of the coke can vary considerably, depending on the
efficiency of the stripping operation to remove the lighter hydrocarbons from the
catalyst flowing into the regenerator. Hydrogen content in the coke can vary
from 4 to 5 wt% for a highly efficient stripping operation, up to 13–15 wt%
for poor stripping. The hydrogen content in a well run FCCU is in the range of
6–8 wt%.
The oxidation of carbon on the catalyst surface (coke burning) is believed to
proceed through the formation of solid surface oxides, which decompose to give
both CO and CO2 as primary products (24). In an early study on the burning of
graphite, it was shown that the CO2/CO ratio is constant throughout the burning
cycle, independent of oxygen partial pressure, and dependent only on the com-
bustion temperature (25). The burning of carbonaceous deposits from porous oxi-
des produced CO2/CO ratios at the burning site in exact agreement with the
ratios of the earlier study (25) over the temperature range 450–6008C (26). Dif-
fusion effects in the pores of large catalyst particles influence the CO2/CO ratio in
the bulk vapor phase (26). However, for particle sizes typical of FCC operations
(60–70-mm average particle size), diffusion is not a factor, and thus particle size
would not influence the CO2/CO ratio. Consequently, this molar ratio remains
relatively constant at a value of 1.0 for many regenerators operating in the
region below 7008C and in the absence of a CO combustion promoter.
The rate of coke burning for coke deposited on a zeolite-containing catalyst
has been reported to be first order with respect both to coke concentration cC and
oxygen partial pressure (26):

dcC
¼ kcC PO2 ð10Þ
dt

This equation is in agreement with coke-burning studies on amorphous silica–


alumina catalysts (27,28). Other investigators have found that carbon burning
was second order (29) or mixed order (30) with respect to carbon. It has been
450 FLUID CATALYTIC CRACKING, UNITS, REGENERATION Vol. 1

0.9
0.8
0.7
0.6
0.5
0.4

Fraction of initial coke remaining


0.3

0.2

0.1
0.08
0.06
0.05
0.04
0.03

0.02

0.01
0.5 1.0 1.5 2.0 2.5 3.0
Regeneration time, h

Fig. 3. Burning at 5008C in flowing air of coke from H-Y FCC catalyst.

argued that coke burning follows first-order kinetics, but only after an initial
period of several minutes, when burning is not first order with carbon content
(31). The data in Figure 3 on coked FCC catalyst of the H-Y type agree with
this argument.
The activation energy for burning from a coked zeolite has been reported as
109 kJ/mol (32) and 125 kJ/mol (30 kcal/mol) has been found for coke burning
from a H-Y FCC catalyst. Activation energies of 167 kJ/mol (40 kcal/mol) (27)
and 159 kJ/mol (28) have been reported for the burning of carbon from a coked
amorphous silica–alumina catalyst.
The coke-burning rate in the regenerator is thus a function of the oxygen
partial pressure, the carbon content on the catalyst leaving the reactor, the tem-
perature in the regenerator, and the residence time in the regenerator. The last
three variables are all coupled together through the FCCU heat-balance require-
ments. For example, an increase in regenerator temperature normally results in
a decrease in catalyst circulation rate to maintain a constant temperature on the
reactor side. The decrease in catalyst circulation rate increases the residence
time in the regenerator and is accompanied by a change in the carbon content
of the catalyst coming into the regenerator. A typical interrelationship between
the O2 content in the regenerator flue gas, the regenerator temperature, and the
carbon content of the regenerated catalyst (CRC) is shown in Figure 4. At a con-
stant flue gas O2 content, a decrease in regenerator temperature of 208C results
in a CRC increase of 0.15–0.20 wt%.
The hydrocarbon feed rate to the reactor also affects the burning kinetics in
the regenerator. Increasing the reactor feed rate increases the coke production
rate, which in turn requires that the air rate to the regenerator increase.
Vol. 1 FLUID CATALYTIC CRACKING, UNITS, REGENERATION 451

1.0

0.9

0.8

0.7

0.6

0.5
Coke on regenerated catalyst, wt %

0.5%
0.4

1.0%
0.3
0.1% Excess O2

0.2

2.0%

0.1
620 640 660 680 700 720 740
Regenerator temperature, °C

Fig. 4. Coke on regenerated catalyst (CRC) versus regenerator temperature at varying


O2 content (33).

Because the regenerator bed level is generally held constant, the air residence
time in the dense phase decreases. This decrease increases O2 content and con-
sequently CO combustion in the dilute phase, frequently referred to as after burn
(Fig. 5).
The coke-burning rates on silica–alumina catalysts have been found to
increase several orders of magnitude when transition-metal oxides are present
(28). The addition of 0.15 wt % chromium (as Cr2O3) to a silica–alumina catalyst
produces a nearly fivefold increase in carbon-burning rate. The cation exchanged
into a zeolite-containing catalyst affects the burning rate (32). A zeolite
exchanged with copper was found to have a burning-rate constant that was
18 times greater than a zeolite exchanged with chromium.
The hydrogen contained in coke burns at a higher rate than carbon.
Hydrogen-burning rates are four to five times greater than carbon-burning
rates (27,29).
452 FLUID CATALYTIC CRACKING, UNITS, REGENERATION Vol. 1

+20

Regenerator temperature change, °C


+10

–10

Complete combustion
CO = 150 ppm
–20 O2 = 0.2 Vol %

50 60 70 80
Regenerator gas velocity, cm/s

Fig. 5. Effect of gas residence time on afterburning. The temperature change DT is the
difference in temperature between flue and dense phase (33).

4.1. CO Combustion. The combustion products leaving the coke-


burning site consist of both CO2 and CO, typically at a CO2/CO mole ratio of
1.0. The CO formed can then be further oxidized. When a CO combustion
promoter is present, the reaction of CO and O2 to form CO2 occurs readily
in the regenerator dense phase. In the absence of such a promoter, the oxidation
of CO occurs slowly in the dense phase below 7008C. This reaction occurs much
more rapidly in the regenerator dilute (vapor) phase where it is believed to be
catalyzed by the combined presence of fine, decoked metals containing catalyst
particles and water vapor (34).
The oxidation of CO ! CO2 releases twice as much heat as does the burning
of C ! CO. The large heat release coupled with the relatively small amount of
mass in the dilute-phase produces a large DT between the dilute and dense
phases in the regenerator. For FCC operations in which a CO combustion promo-
ter is not used, the dilute-phase temperature is frequently 15–408C higher than
the dense phase. In such operations, the dilute-phase temperature, hence the
dilute-dense DT, becomes the limiting factor in regenerator control. The control
of the regenerator temperature, in such cases, is normally achieved by adjusting
the air rate to the regenerator, to control the oxidation of CO to CO2 in the dilute
phase, and thereby maintaining an acceptable dilute-dense DT.
With a sufficent excess of O2 in the regenerator flue gas (1–2%), complete
CO combustion can be achieved without a CO combustion promoter, if dense-
phase temperatures >7208C are maintained.
4.2. Promoted CO Combustion. In the early 1970s, researchers dis-
covered that certain Groups 8–10 (VIII) metals, particularly platinum, can be
incorporated into an FCC catalyst system at low concentrations (1–3 wppm)
to catalyze the combustion of CO to CO2 effectively (35,36). Of even greater
Vol. 1 FLUID CATALYTIC CRACKING, UNITS, REGENERATION 453

importance was their discovery that at this low incorporation, the Groups 8–10
(VIII) metals did not catalyze undesirable dehydrogenation reactions during the
cracking reaction. The cracking-selectivity characteristics achieved by the crack-
ing catalyst in use in the FCCU are thus not altered when such a CO combustion
promoter is added to the catalyst inventory.
The use of 1–2 ppm platinum in FCCU circulating catalyst inventories is
now widely practiced throughout the world to catalyze the combustion of CO
to CO2 in the FCCU regenerator. The promoter is added to the FCC catalyst
inventory either as an integral part of the fresh FCC catalyst, where it is
included in the fresh catalyst at 1–2 wppm, or as a separate additive. The addi-
tive, which typically contains 500–1000 wt ppm of platinum, is added to the
FCCU by a small metering system that is independent of the fresh cracking-
catalyst addition system. In the United States, the separate additive approach
is generally used; in Europe, the additive is most commonly incorporated with
the fresh cracking catalyst.
Through the use of a combustion promoter, CO combustion occurs readily in
the dense phase at temperatures well <7008C. Guegan reported that promoted
CO combustion occurred as low as 6508C in a commercial FCCU operation (37).
Promoted combustion has a number of important benefits. First, the after-
burn problem, ie, burning in the dilute phase, is greatly reduced. By consuming
CO in the dense phase, the potential heat release from burning in the dilute
phase is significantly decreased. A commercial example of a 508C reduction in
regenerator afterburn by using a CO combustion promoter is as follows (38):

Without promoter With promoter


typical flue gas temperature, 8C 699 660
typical dense phase temperature, 8C 663 674
afterburn DT, 8C þ 36 14

The increased flexibility of FCCU operations resulting from using a CO


combustion promoter is an even greater advantage. When using a promoter,
the air rate to the regenerator can be varied to achieve any degree of CO combus-
tion that is desired. By this mechanism, heat can readily be added to or removed
from the regenerator to produce changes in regenerator temperature, catalyst
circulation rate, coke yield, and reactor feed conversion. An example of how var-
iations in the degree of CO combustion can effect the performance of an FCCU is
seen in Figure 6 (33).
In actual practice, the regenerator temperature is normally controlled at
the maximum temperature allowed by the regenerator metallurgy. Variations
in feed quality in the FCCU can result in significant regenerator temperature
excursions if the degree of CO combustion is unchanged (constant heat of com-
bustion). With a CO combustion promoter present, the degree of CO combustion
can be varied to hold the regenerator temperature constant at the desired max-
imum value even though significant changes in feed quality have occurred. This
flexibility is achieved mainly through regenerator air rate control. To a lesser
degree, additional flexibility is achieved by controlling the addition rate of
454 FLUID CATALYTIC CRACKING, UNITS, REGENERATION Vol. 1

Conversion, wt %
.74

.73

.72

0.4
CRC, wt %

0.3
0.2
0.1
0

7
Cat:oil ratio

6
5
4
3

760
Regen. temp, °C

740
720
700
680
660
1 2 3 4 5 6 7 8
CO content in flue gas, vol %

Fig. 6. Typical FCCU interdependency curves (33).

fresh CO combustion catalyst to the catalytic cracker. When a CO combustion


promoter is used, changes in the air rate affect the amount of CO converted to
CO2 both in the dilute and in the dense phase. Changing the amount of CO
burned in turn affects both the dense- and dilute-phase temperatures. Increasing
the amount of excess O2 in the regenerator causes an increase in CO burning
in the regenerator dense phase as indicated by an increase in dense-phase tempe-
rature. Additional burning in the dilute phase also occurs, causing an increase
in DT (afterburning) between dense and dilute phases. As a typical example, a
decrease in CO content in the regenerator flue gas from 5 to 3 vol% would
result in a dilute-phase temperature increase of 318C, a dense-phase tempera-
ture increase of 168C, and an increase in dilute-dense DT of 158C.
The amount of promoter used can also be a variable in influencing the
degree of CO combustion. A promoted catalyst system can be classified as fully
or partially promoted. A partially promoted system is one in which an increase in
promoter content results in a decrease in the dilute-dense DT. Conversely, a fully
promoted system sees no effect on afterburn DT when the spromoter level is
Vol. 1 FLUID CATALYTIC CRACKING, UNITS, REGENERATION 455

70

60 10% promoted

Dilute – dense ∆T, °C 50

Fully promoted 50% promoted


40

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10 11
CO in flue gas, vol %

Fig. 7. Effect of promoter concentration on regenerator dilute-dense DT (9).

changed. An increase in the promoter content increases the proportion of CO that


is burned in the regenerator dense phase. A reported response of afterburn DT to
promoter change is shown in Figure 7 (9). The change in dilute-dense DT is
mainly because of changes in the dilute-phase temperature. Changes in promo-
ter concentration have only a small effect on dense-phase temperature but
greatly affect dilute-phase temperatures.
The carbon-burning reaction competes with the coke-burning reaction for
the available O2 in the regenerator. When the O2 level is low, the amount of pro-
moter present can influence the degree that carbon is burned from the catalyst.
In such situations, a high level of promoter enhances the combustion of CO and
retards coke burning. A fully promoted catalyst with essentially no excess O2 in
the regenerator equilibrates at a coke content on the regenerated catalyst (CRC)
of 0.1 wt% higher than does a partially promoted system at the same degree of
CO combustion (33).
Recent environmental concerns related to NOx emissions from the FCCU
regenerator have focused on the effect of platinum CO combustion promoters
also promoting the formation of NOx (39,40). Interest, consequently, has arisen
in the use of non-platinum CO combustion promoters. Pladimum has been used
in place of platinum with favorable results (39). Additives for the reduction of
NOx in regenerator flue gas are discussed in the next section.

5. Environmental Aspects

The FCCU regenerator is one of the principal sources of air pollutants from a
refinery. In the mid-1990s, a typical 23,850 m3/day (150,000 barrel per day), mod-
ern U.S. refinery with a 50,000 BPD FCCU processing a low sulfur, low metals
content VGO feed, and operating in complete CO combustion would typically
emit the following into the atmosphere from the FCCU regenerator flue gas
stack: fine catalyst particles, 2–3 t/day; sulfur oxides, 3–4 t/day; nitrogen oxides,
0.5 t/day; and CO, 1.5–2 t/day. Recently, governmental regulations, particularly
456 FLUID CATALYTIC CRACKING, UNITS, REGENERATION Vol. 1

in the United States, have significantly reduced the allowable FCCU emissions
for particulates and sulfur oxides. A new 50 MBPD unit or an existing 50 MBPD
unit being significantly revamped in 2002 would be required to meet the
following contaminant levels in the regenerator flue gas: catalyst particulates
<0.5 t/day, sulfur oxides <0.7 t/day (41). Nitrogen oxide limits are not yet defined
but are expected to be similarly constrained. Carbon monoxide emissions would
be limited to <3 t/day. Much research has been done in the past and is ongoing to
develop technologies to meet these emissions levels and reduce them even
further.
5.1. Catalyst Emissions. High velocity air passing up through the
regenerator at velocities typically around 1 m/s carries large quantities of cata-
lyst into the vapor space above the catalyst bed. To contain these catalyst parti-
cles within the regenerator vessel, a highly efficient, generally two-stage cyclone
system is installed inside the regenerator. The flue gas and the entrained cata-
lyst particles pass through the cyclone system which is installed inside the regen-
erator. The flue gas and the entrained catalyst particles are typically returned by
the cyclones back into the regenerator. The particles that are not retained and
that leave with the exiting flue gas are small, typically <20 mm particle size.
These particles generally represent fines that have been produced as a
result of catalyst attrition. Two attrition mechanisms are generally recognized:
particle shattering and surface abrasion of particles. The abrasion mechanism
generates fines of much smaller average size. The manufacturing techniques
and chemical compositions of the various FCC catalysts influence the type and
severity of the attrition mechanism. A small amount of fines (<20 mm in size)
may also come from fresh catalyst additions.
Current particulate regulations vary from site to site depending on the local
regulatory agency’s standards. The New Source Performance Standards, NSPS,
set by the U.S. federal government state that a new unit must not exceed 1lb of
particulates per 1000 lbs of coke burned (1 kg/ 1000 kg). It is possible that an
existing unit, that undergoes a substantial revamp, can be reclassified as a
new source.
There are several ways to capture the particles remaining in the flue gas as
it exits the regenerator. The most common technique is through the use of elec-
trostatic precipitators (ESP), which can reduce particulate emissions to very low
levels (41). An ESP consists of rows of collector plates and voltage discharge elec-
trodes arranged in one or more gas tight chambers. The electrodes apply electri-
cal charges to particles in the FCCU regenerator flue gas. The charged particles
are attracted to the collector plates and are thus removed from the flue gas.
While this equipment is well known and has a long record of reliable perfor-
mance, it is expensive and can be dangerous. Explosions in ESPs have occured.
An alternative to the ESP is a third stage separator (TSS). The TSS is an
additional cyclonic separator, external to the regenerator, which supplements
the catalyst separation achieved by the two stages of cyclones typically present
in an FCCU regenerator.
In the TSS, catalyst is removed from the flue gas by passing the catalyst-
entrained gas stream through a set of cyclonic elements that separate out the
catalyst particles. Separation efficiency depends upon the particle-size distribu-
tion of the catalyst leaving the regenerator cyclones, being most efficient on large
Vol. 1 FLUID CATALYTIC CRACKING, UNITS, REGENERATION 457

particles, and less efficient on small particles. Essentially 100% removal of


particles >5 mm in diameter have been reported (41).
Fines-depleted flue gas leaves the upper chamber of the TSS. Recovered
catalyst and blow-down gas ( 3% of the flue gas) exit from the bottom of the
separator to an electrostatic precipitator, or to a small, conventional, fourth
stage collection device for further concentration and disposal of the catalyst fines.
Recent improvements in third stage separators have put them in a class
acceptable to most environmental agencies, at a fraction of the cost of an ESP.
The devices are proving to be highly reliable and efficient. Current performance
has been reported to yield a flue gas containing 0.3–0.6 lb of particulates per
1000 lb of coke burned, well below the NSPS standard (41,42). When a flue gas
power recovery unit (described later) is present, the TSS protects the power
recovery unit from erosion by catalyst fines.
Much effort has been made by catalyst manufacturers to improve catalyst
attrition resistance and thus reduce the formation of fines (see CATALYSTS,
SUPPORTED). In the 10-year period from 1980 to 1990, most catalyst manufacturers
improved the attrition resistance of their catalyst by a factor of at least 3–4.
This improvement was achieved even though the catalyst zeolite content during
this period was continually increasing, a factor that makes achieving catalyst
hardness more difficult. As an example of the type of attrition improvement
that has been achieved, the catalyst attrition index, which is directly related to
catalyst loss rate in a laboratory attrition test, decreased from 1.0 to 0.35 for one
constant catalyst grade during 1989–1990 (43). This trend has continued with
the recent introduction of premium catalyst technology for loss sensitive applica-
tions where power recovery equipment may be damaged by excess catalyst
fines (44).
5.2. SOx Emissions—Source and Control Options SOx, which
refers to the combination of SO2 and SO3, is a major emission concern from a
FCCU regenerator flue gas. A 1998 survey of 79 refineries in western Europe
reveal that 13.2% of all refinery SOx emissions originate in the FCCU (45). Of
the amount of organic sulfur in the reactor feed (non-hydrotreated), 10%
(range 5–30%) goes to the regenerator with the coke, where it is converted to
SOx (39). The sulfur content of the coke is highly dependent on feed type (aromatic,
paraffinic, or hydrotreated), the amount of organic sulfur in the feed, and the reac-
tor conversion level. A general rule of thumb to estimate the regenerator flue gas
SOx content (if no attempt is made to reduce it), in parts per million:

Straight run feeds: SOx ¼ 800–1100 x wt% feed sulfur


Hydrotreated feeds: SOx ¼ 2050–2080 x wt% feed sulfur (46)

The refiner does have several options to reduce SOx emissions, and gener-
ally will use a combination of these options to reach regulatory requirements.
These include

Hydrotreating the feed


Flue gas scrubbing with NaOH, CaCO3, or other basic solution.
SOx reduction catalyst additive
Feed stock selection (low sulfur content feeds)
458 FLUID CATALYTIC CRACKING, UNITS, REGENERATION Vol. 1

Hydrotreating In hydrotreating, the FCCU feed is pretreated to convert


the organic sulfur in the feed to H2S by reacting it with hydrogen. The severity
of the operation is a function of the partial pressure of hydrogen used. Mild
hydrotreating at 3550–5600 kPa (500–800 psi) H2 partial pressure, with a mod-
ern cobalt–molybdenum catalyst can reduce the sulfur content of the FCC feed
by 90% or greater (47). In one published study, a FCC feed with 17,500 wppm
sulfur was hydrotreated at two levels of severity. At moderate severity the sulfur
in the feed was reduced to 100 wppm, a decrease of 99.4%. A resultant decrease
in SOx emission from the regenerator of 97.4% was observed. Increasing severity,
by increasing H2 partial pressure 1.4 times, resulted in a decrease in feed sulfur
to 33 wppm or 99.8%. A corresponding decrease in SOx emission of 99.2% was
observed (48).
Sulfur compounds in the FCC feed can be placed into the following classifi-
cations: alkyl sulfides, aromatic sulfides, substituted benzothiophenes, and mul-
tiring benzothiophenes. Hydrotreating can effectively convert most of these
sulfides, but is not very effective in converting the multi-ring benzothiophenes.
These compounds are also the most difficult to crack in the FCC reactor, and
usually accumulate in the heavy FCCU products (cycle oils and coke) (49). There-
fore, the percent of feed sulfur on coke increases significantly with hydrotreating.
Scrubbing. Scrubbing of the flue gas is a technique commonly used
among all combustion processes. In an FCC environment, however, the scrub-
bing technique must be such that it adds very little pressure drop to the system
(on the order of several inches of H2O). The scrubber may see large amounts of
catalyst dumped to it in an upset of the FCC. It must therefore have accommoda-
tion to handle such episodes of high particulate loading without plugging or
damaging circulation equipment. To maintain low pressure drop, most FCC
scrubbing systems either induce flow via venturi type scrubbers or use sprays
to contact the flue gas with the scrubbing liquid. The high temperature of the
FCCU flue gas may require a quench section in the scrubber. Scrubbing efficien-
cies can vary somewhat between technologies, with reported efficiencies ranging
from 85 of 99% SO2 remova1 (41,50,51).
The most common type of scrubbing solution used is sodium hydroxide
(caustic). The acid–base reaction is very fast and SO2 removal efficiencies of
99% have been reported (50). The disadvantages to using caustic are the cost
of the reagent, and the disposal of the resultant sodium sulfate solution, which
requires further treatment to neutralize it.
One unique solution to the disposal of this sodium sulfate waste solution is
a two-step process, which uses the THIOPAQ biological conversion technology.
This technology has seen success in other scrubbing applications. It converts
the SO2 to elemental sulfur with no liquid effluent. The first step is to
scrub the SO2 from the flue gas with a buffer solution of sodium bicarbonate
and caustic. This buffer solution is produced by allowing NaOH to react with
CO2 until the resultant sodium bicarbonate solution is in equilibrium with the
carbon dioxide in the flue gas. The SO2 reacts to form NaHSO3 and Na2SO4.
This buffered solution is then sent to a series of biological reactors. The first reac-
tor operates in an anaerobic environment to convert the sodium sulfite and
sodium sulfate to sodium sulfide, NaHS. The second reactor operates in an aero-
bic mode and oxidizes the sodium sulfide to elemental sulfur. The elemental
Vol. 1 FLUID CATALYTIC CRACKING, UNITS, REGENERATION 459

sulfur is separated from the liquid and the sodium bicarbonate solution returned
to the scrubber. The elemental sulfur is then either dried and sent to landfill as
nonhazardous waste or further purified for resale (50).
Calcium carbonate (limestone) slurry is another solution that is used for
scrubbing the FCCU flue gas. This material is less expensive to purchase, as it
is shipped as a solid. However, it can be difficult to get into solution and plugging
issues can be problematic. Its SO2 removal efficiencies are usually 95–98% (50).
The reaction chemistry produces calcium sulfite, CaSO3 , which for handling pur-
poses is oxidized to calcium sulfate (gypsum), CaSO4 . The gypsum is dried and
recycled as an additive for cement production. One drawback to this calcium car-
bonate scrubbing is the release of CO2 in the reaction. Current concerns over CO2
emissions as a greenhouse gas may make this scrubbing technique less desirable.
Sea water scrubbing to remove SO2 is being practiced by StatOil in Mong-
stat, Norway. The seawater contains natural calcium and sodium buffers that
absorb the SO2. The water is returned to the sea after dilution with sea water
cooling streams (52).
Although often lumped together in analysis results, SO2 and SO3 create
there own unique emission and control issues. The SO3 content, in FCCU flue
gas, can vary from 1 to 10% of the total SOx (53). The SO3 emissions, are espe-
cially troublesome in that SO3 condenses to form sulfuric acid mist that is not
readily removed by wet scrubbers. The light scattering properties of sulfuric
acid mist creates a visible plume that, depending on measurement technology
utilized, is often measured and treated as a particulate emission. SBS Injection
Technology, commercialized in the utility industry in 2002 has been proposed as
a solution to FCCU SO3 problem (53). Sodium bisulfite solution is injected into
the flue gas to react the SO3 to produce sodium sulfate and sodium bisulfate
particle (solids). These then can be readily removed by an electrostatic precipita-
tor or in an SO2 scrubber.
SOx Reduction Catalyst Additive. Currently, many refiners find that the
use of a SOx adsorbing additive is the most effective way to comply with the EPA
regulations. With such an additive, the following steps occur
Oxidation of SO2 to SO3 in the regenerator:

2 SO2 þ O2 ! 2 SO3

Adsorption of SO3 in the regenerator by the additive where MO ¼ metal oxide


and M is most commonly magnesium:

SO3 þ MO ! MSO4

Reduction of the metal sulfate in the reactor:

MSO4 þ CH4 ! MS þ CO2 þ 2 H2 O

Release of H2S and regeneration of the metal oxide adsorbent in the


stripper:

MS þ H2 O ! MO þ H2 S
460 FLUID CATALYTIC CRACKING, UNITS, REGENERATION Vol. 1

The H2S is carried out with the reactor products, goes through the product-
recovery system of the FCCU, and eventually to further processing for sulfur
recovery. The metal oxide adsorbent recirculates with the spent cracking catalyst
back to the regenerator for the next SOx adsorption cycle.
The first commercially effective metal oxide adsorbent consisted of a solid
solution of a pure magnesium aluminate spinel (MgAl2O4) with MgO (54).
Such a solid solution (Mg2AlO5) does not destroy the spinel framework. The
adsorption activity of the dispersed MgO in the spinel is much greater than
that of pure MgO itself (54).
Cerium is effective in oxidizing SO2 to SO3. Consequently, a cerium-impreg-
nated Mg2Al2O5 actively converts the SO2 to SO3, which is then strongly
adsorbed by the dispersed MgO as MgSO4. A completely cyclic SOx removal cat-
alyst contains, in addition to the above component, a fourth metal component
such as vanadium to catalyze the conversion of MgSO4 to MgO (55).
Multifunctional SOx removal catalyst systems have been in commercial use
since 1985 in the United States. Such systems have successfully reduced SOx
emissions in the FCCU regenerator by typically 20–60%, with reductions up to
90% reported (39,56,57). Additive levels in the circulating catalyst inventory
range from 1 to 10%. The additive level and the amount of SOx reduction depend
on conditions, such as feed quality, the presence or absence of a CO combustion
promoter, regenerator temperature, regenerator mixing efficency, and excess O2
content (39).
Hydrotalcite, a MgO/Al2O3 containing compound, has also been found to
have SOx adsorbent capability and is now being used in some SOx adsorbent
additives (39). In 2003, both spinel and hydrotalcite types of SOx reduction tech-
nology have been upgraded to higher performance standards, frequently by the
addition of more magnesium oxide component (39).
Because O2 is necessary to convert SO2 to SO3, decreasing O2 in the regen-
erator has been found to reduce the effectiveness of the SOx removal additive.
The SOx additives used in regenerators operating in a partial CO combustion
mode, where excess O2 is frequently limited to <0.2 vol% in the flue gas, are
less successful in reducing SOx. In such cases, SOx removal is typically 20–
30% less than for a full CO combustion (1þ% excess O2) case (56).
5.3. Nitrogen Oxide Emissions Source and Controls. The (NO)x
(nitrous oxides ) emissions are now recognized as a significant contributor to
photochemical smog and acid rain, and therefore have come under greater reg-
ulatory scrutiny than in the past. In terms of tons emitted per year, the FCCU
regenerator is one of the largest point source of NOx emissions. Only a small por-
tion of the NOx present in the flue gas is produced through the oxidation of N2 in
the regenerator air stream (40,58). The main source of the NOx comes from the
combustion of organic nitrogen, originating in the FCC feed of which 50% ulti-
mately reaches the regenerator section in the coke. However, only 5–20% of the
organic nitrogen compounds entering the regenerator end up as NOx, predomi-
nantly NO. The remainder is converted to N2. Since most of the organic nitrogen
ends up as N2, there is likely a secondary reaction with CO or coke to reduce NO
to N2 (59). Although feed nitrogen is the source of NOx, researchers have found that
final NOx concentrations are due more to the type of N in the feed and to the regen-
erator conditions, than they are to the absolute content of N in the feed (40).
Vol. 1 FLUID CATALYTIC CRACKING, UNITS, REGENERATION 461

The amount of excess oxygen may have a more direct correlation with the final
NOx concentration leaving the regenerator (39,40). Increasing the O2 in the
regenerator flue gas from 0.1 to 1.6% has been reported to double NOx emissions
when no CO combustion promoter is present. Adding a platinum-based combus-
tion promoter, which increases the available atomic O2, can increase the NOx
content much more (60).
Hydrotreating the reactor feed can reduce feed nitrogen by 20% (low sever-
ity), 90% (moderate Hydrotreating), and up to 99% (severe Hydrotreating) (61),
and thus can be a tool in reducing FCCU flue gas NOx emissions.
An emerging picture of the complex NOx chemistry in the FCCU regenera-
tor includes the following reaction networks:

Pyrolysis Oxidation
N (Coke) ! HCN ! N2, NO, N2O
N (Coke) ! NH3 ! N2, NO, N2O

NO Reduction
coke
2 NO þ 2 CO ! N2 þ 2 CO2
NO þ CFAS  ! 1=2 N2 þ CO 
FAS ¼ free active surface

Catalyst additives and regenerator design have both been found to be useful
in reducing flue gas NOx emissions. Uniform distribution of coked catalyst in the
regenerator improves the probability of NOx adsorption onto the carbon surface,
necessary for NOx reduction reactions to occur. Air grid design, spent catalyst
distribution, catalyst withdrawl location, and the fluidization regime can all
play major factors in the effectiveness of mixing and distribution. Poor design
and/or erosion of air grids and catalyst distributors can lead to maldistribution
and therefore higher NOx values. Regenerator designs that operate in a fast flui-
dized regime (combustor style) rather than a turbulent regime (bubbling bed
style) afford better catalyst-air mixing and therefore lower NOx emissions.
Recent research in NOx additives has taken two approaches: the develop-
ment of CO combustion promoters that do not catalyze NOx formation and addi-
tives that directly reduce NOx emissions. New NOx reduction additives have been
designed to catalyze the NO reduction by CO or to block the formation of NO by
oxidation of the reduced intermediates, HCN and NH3. The major FCC catalyst
vendors are now publishing the results of commercial trials of NOx reduction
additives in US FCC units. Variying degrees of successes have been reported
(39,57,62). This suggests that the complex interplay between FCCU regenerator
hardware, operating variables and additives for NOx reduction requires more
detailed study.
Post regenerator NOx Reduction Technology Two technologies have
been commercialized to reduce FCCU NOx emissions by treatment of the flue
gas after it leaves the regenerator.
Selective Catalytic Reduction (SCR), which is the most proven of these pro-
cesses, removes NOx by passing it over a vanadium pentoxide/titanium dioxide
462 FLUID CATALYTIC CRACKING, UNITS, REGENERATION Vol. 1

catalyst bed in the presence of NH3 and O2. The reactions are

4 NH3 þ 4 NO þ O2 ! 4 N2 þ 6 H2 O

4 NH3 þ 2 NO2 þ O2 ! 3 N2 þ 6 H2 O

Reaction is preferably carried out in the temperature range of 315–4008C , where


a NOx removal efficiency of >90% i1s claimed (41). An SCR unit installed at the
Scanraff refinery, Lysekil, Sweden was reported to remove 85% of the 800 mg/
nm3 of NOx present in the FCCU regenerator flue gas (51). For an FCCU revamp,
however, considerable space in the refinery is required for the installation of the
SCR reactor, which frequently makes SCR impractical or not cost effective (51).
The SCR has two additional drawbacks if SOx is not removed from the flue gas:

1. An SCR side reaction typically converts 1–5% of the SO2 to SO3. This sig-
nificant increase in SO3 can result in a visible plume in the flue gas leaving
the regenerator stack (53).
2. Unreacted NH3 can react with SO3 leaving the SCR to produce (NH3)2SO4
and NH3HSO4, which can increase particulate emissions and can also
contribute to plume formation (41).

Another dry process utilizes activated coke to remove both SOx and NOx
from the flue gas in a two stage, moving-bed process (51). The SOx is removed
in the first stage via adsorbtion on the activated carbon. NH3 is injected into
the flue gas, which then passes through the second bed, where the activated car-
bon acts to catalyze the reaction of NOx with NH3 to convert the NOx to N2 and
H2O. The activated carbon flows from the second stage to the first stage and then
to a regenerator to desorb the SOx. The desorption step releases SO2, which must
be recaptured and converted to elemental sulfur. This technology was commer-
cialized on a resid-processing FCCU in Japan in 1987 and achieved a reported
70% removal of both NOx and SOx (63).
Several emerging technologies, yet to be fully commercialized also appear to
have promise for reducing NOx from the flue gas leaving the regenerator. Scrub-
bing of NOx, as is done to remove SOx , is not an effective means of removing NOx,
since NO and NO2 are much less soluble than SO2. A process using ozone injected
into the flue gas stream converts NOx to N2O5, which is highly water soluble and
rapidly converts to nitric acid. Scrubbing of the flue gas with a NaOH solution is
then very effective in removing NOx and further converts the nitric acid to the
neutral salt (NaNO3) (53). A combination chemical–biological process utilizes
an iron chelate aqueous solution to convert NOx to a water soluble nitrosyl com-
plex. The solution is sent to a bioreactor, where the complex is reduced to N2,
CO2, and H2O (50). Finally, a high temperature (8708C) noncatalytic process,
suitable for use with FCCUs having a CO boiler, uses NH3 to convert NOx to
N2 and H2O (59).
5.4. CO2 Emissions Reduction Recently, the concern of CO2 emissions
as a contributor to green house gas levels has drawn attention to the FCCU as a
point source. In a FCCU based refinery, 20% of the CO2 emitted originates in the
FCCU (64). Carbon dioxide is the natural result of complete combustion of the
Vol. 1 FLUID CATALYTIC CRACKING, UNITS, REGENERATION 463

coke in the catalyst regeneration, the heat from which is also essential to heat
balance the unit. As a result of the Kyoto Protocol, the European Union has
already begun CO2 credit trading, and therefore interest among international
refiners is very strong for ways to lower FCCU based CO2 emissions. The United
States, having withdrawn from the Kyoto Protocol, has announced a separate
strategy for addressing climate change, and several climate change bills have
been introduced in the 108th Congress. The primary vehicle for climate change
legislation is comprehensive energy legislation (65).

6. Regenerator Operating Parameters

To maximize the performance of an FCCU, most units run at one or more unit
constraints. Frequently, one of these constraints is the regenerator temperature,
which is set by metallurgical limits for safe operations. Process variables on both
the reactor and the regenerator side are thus manipulated to keep the regenera-
tor temperature as close as possible to this regenerator temperature limit.
Changes in feed quality, which will affect the coking tendency of the feed
and the coke-making characteristics of the circulating catalyst, influence the
regenerator temperature. The refiner can rapidly respond to these changes by
manipulating certain independent process variables, which include reactor
riser temperature, temperature of the hydrocarbon feed entering the riser,
fresh catalyst addition rate, and cooling duty if a catalyst cooler exists. If the
change in feed quality is of prolonged duration, the refiner can also change
catalyst type.
The heavy material in the reactor feed, which remains as a residue after
evaporation and pyrolysis of the oil, is designated as Conradson carbon (ASTM
test method D189-76). This residue is frequently used as an indicator of the
coking tendency of the feed. Incorporating heavy feedstocks such as vacuum
resids into the FCCU feed greatly increases the feed’s Conradson carbon
content. The influence of increasing resid content on regenerator temperature
has been determined and correlated against the feed Conradson carbon content
(66). A 2% increase in Conradson carbon can cause a regenerator temperature
increase of 288C. Other studies indicate that an increase in the resid content
of the feed that results in an increase in Conradson carbon from 1.5 to 2.9%
produces a regenerator temperature increase of 378C at constant riser tem-
perature (13).
The presence of contaminant metals on the equilibrium catalyst can signi-
ficantly increase the catalyst coking tendency, which in turn results in an
increase in regenerator temperature if all other factors remain unchanged. As
one example, if the metals on an FCCU equilibrium catalyst increased from an
equivalent-nickel value of 2000 to 3500 wt ppm, the catalyst coking tendency
would increase 30–50%. If all controllable parameters remained constant, the
regenerator temperature would be expected to increase 35–508C, resulting in a
decrease in catalyst circulation rate, and a subsequent drop in conversion.
To decrease the regenerator temperature, heat must be taken out of the
system either by
464 FLUID CATALYTIC CRACKING, UNITS, REGENERATION Vol. 1

1. Decreasing conversion, which decreases the amount of catalytic coke pro-


duced. Decreasing conversion can be achieved by either decreasing reactor
temperature or by decreasing equilibrium catalyst activity. For a given
decrease in conversion, the latter is more effective in reducing regenerator
temperature. Typically, decreasing reactor riser outlet temperature by 88C
decreases conversion 2–2.5% and decreases regenerator temperature by
6–88C (8,67). Reducing equilibrium catalyst activity by four numbers,
either by decreasing the fresh catalyst addition rate or by using a lower ac-
tivity fresh catalyst, also reduces conversion by 2% and decreases regen-
erator temperature by >428C, depending on the initial temperature (68).
2. Through the physical removal of heat. Heat can be removed by using a
catalyst-cooling heat exchanger or by cooling the feed entering the reactor.
Typically, decreasing the feed temperature by 288C decreases the regenera-
tor temperature by 68C (8).
3. Through a decrease in the relative amount of CO converted to CO2 in the
regenerator. Although burning coke to CO rather than to CO2 releases 70%
less heat, this method is less frequently used. It requires that more feed be
converted to coke to satisfy the heat balance, thus decreasing yield profit-
ability, and it also requires that the unconverted CO be burned to CO2 in
an external CO boiler, to prevent any CO emission to the atmosphere.

7. FCCU Regenerator Configuration and Mechanical Hardware

Since the first fluid-bed catalytic cracking unit was commissioned in 1942, the
process has seen considerable improvement in mechanical construction, reliabil-
ity, and process flow. A modern FCCU typically operates continuously for 3–
5 years between turnarounds, during which time quantities on the order of
1010 kg of feedstock are processed and quantities of catalyst in the range of
7  1010 kg are circulated through the regenerator. Early FCCU designs (69),
were complex compared with the configuration of more recent designs.
In the modern unit design, the main vessel elevations and catalyst transfer
lines are typically set to achieve optimum pressure differentials because the
process favors high regenerator pressure, to enhance coke-burning kinetics
and power recovery from the flue gas, and low reactor pressure to enhance pro-
duct yields and selectivities.
Comparing a modern design with a typically less desirable configuration of
the past (70), the principal improvement factors have been aimed toward:
reduced catalyst inventory and temperature limitations, enhanced coke-burning
kinetics, catalyst and air contacting; and catalyst residence time distribution.
New designs have eliminated horizontal transport, long standpipes, additional
standpipe aeration requirements, and catalyst fines recycle.
On the regenerator side, there have been several significant design changes
from the bed regenerators that were utilized in early FCC unit designs. In the
1970s, with an emphasis on more complete combustion of the coke deposits, a
fast-fluidized, combustor-style regenerator design was introduced. Some of the
characteristics of this new, widely applied design are described later in this section.
Vol. 1 FLUID CATALYTIC CRACKING, UNITS, REGENERATION 465

In the 1980s, to more effectively process high boiling, contaminated, resi-


dual feedstocks, two-stage regenerators and catalyst coolers were commercia-
lized to allow greater unit heat balance flexibility. One version of the two-stage
regenerator designs makes use of two separate regenerator vessels (71), with the
first stage vessel operating in a partial combustion mode, with conventional two-
stage cyclones followed by an external CO boiler. Catalyst from this first stage of
regeneration is then dilute phase transported to the second stage regenerator,
where the rest of the coke is burned. This second regenerator vessel has no inter-
nals in the dilute phase. The flue gas containing excess oxygen passes from the
second stage regenerator to externally located, refractory lined cyclones.
An alternate two-stage regenerator design was independently developed
(70,72) incorporating two distinct bed-regeneration zones within a single regene-
rator vessel. Coked catalyst from the catalyst stripper enters the upper regenera-
tion zone. This first-stage is operated in partial combustion, with CO in the
exiting flue gas, to produce a lower heat of combustion. Catalyst passes by gra-
vity flow from the first stage to the second zone, located below the first stage.
This second stage operates in a complete combustion mode with excess oxygen,
to complete the coke burn and provides regenerated catalyst with low carbon-on-
catalyst. The excess oxygen-containing flue gas from this second zone passes
upward through the first stage. Additional air is also provided to the first
stage through a separate air distributor. Togeteher, these two air streams pro-
vide the oxygen needed to achieve partial combustion in the first stage. A single
flue gas line leaves the regenerator. Since this flue gas stream contains CO, it is
sent to a CO boiler.
An external dense-phase catalyst cooler was developed specifically to
remove heat from catalyst that passes from this upper regeneration zone
to the lower regeneration zone. The amount of heat removed can be adjusted
to provide better control of the process conditions—a key feature of the catalyst
cooler.
7.1. The Effect of Fluidization on Regeneration Several fluidization
regimes have been used in FCCU regenerators depending on the technology
evolution and the objectives of the particular design. Early designs had little
concern for CO emissions, but were very much concerned about catalyst loss
and equipment (primarily cyclone) damage due to high temperatures of uncon-
trolled CO after burning. Construction materials were typically chome alloys,
which limited regenerator temperatures to <7008C (13008F).
Bubbling Bed Regime Early FCCU regenerators operated in the bub-
bling–bed regime (73). The bubbling-bed regenerator design offered limited
solids carryover and transport through the freeboard region. The bubbling-bed
regime ranges from the minimum bubbling velocity, which typically for FCC-
type particles is from 0.006 to 0.03 m/s, up to 0.3 m/s superficial velocity. Dis-
crete gas bubbles flowing through the bed produce abrupt pressure fluctuations
at the bed’s surface. The relative fluctuations are set by the bubble frequency or
superficial gas velocity. Most of the larger particles that are entrained are
returned to the bed through the two-stage cyclone diplegs; the smaller fines
being carried out with the flue gas (74).
The bubbling-bed design had several major problems chief among these
being poor catalyst–air mixing. As a result much of catalyst returned to the
466 FLUID CATALYTIC CRACKING, UNITS, REGENERATION Vol. 1

reactor unregenerated or partially coked, leading to poor activity and low reactor
conversion (52).
Turbulent-Bed Regime. The commercialization of high alumina catalysts
in 1955 was one of the chief reasons for a shift in regenerator design to the tur-
bulent-bed regime. The catalyst was much more active, but highly sensitive to
coke deactivation, so the poor regeneration of the bubbling-bed design was no
longer acceptable. The need to improve mixing of the combustion air and the cat-
alyst, led designs to change to the turbulent fluidization regime. At the higher
gas velocities of the turbulent-bed regime (0.3–1.0 m/s), the distinct bubble
phase disappears, and the bulk of the gas flows in a manner described as: ‘‘In
voids which continually coalesce and split tracing tortuous passages as they
rise through the bed’’ (75). The surface of the upper bed is considerably more dif-
fuse and has reduced pressure fluctuations and substantially higher entrain-
ment of solids into the freeboard region. Because of the higher coke-burning
capacity requirements and improved contacting efficiency, the vast majority of
commercial regenerators converted to the turbulent-bed regime. In this config-
uration, the ultimate regeneration capacity is set by the sharp increase in
solid entrainment that occurs as the linear velocity approaches 1.0 m/s and the
corresponding effect this air velocity has on the cyclone separation efficiency and
dipleg hydraulics (76).
Over the years, considerable efforts have been made to improve regene-
ration efficiency. These efforts focused on improving the radial mixing char-
acteristics and residence-time distributions (reduce short circuiting) of the
turbulent- and bubbling-bed systems. Some of these efforts include: optimizing
fluidized-bed length and diameter ratios, optimizing air distribution through
the proper grid pressure drop and grid plugging pattern, and optimizing catalyst
distribution via cyclone dipleg discharge orientation, and proper location of the
spent catalyst addition and withdrawal points into the regenerator. These addi-
tion and withdrawal devices include tangential inlets, swirl distributors, deflec-
tor plates, and baffles as well as the entry elevation into the fluidized bed (74).
Fast Fluidized Regime. In the late 1970s, the high efficiency combustor
style regenerator was commercialized. This type of regenerator operates in the
fast fluidized regime. The fast fluidized regime (1.0–3.0 m/s superficial velocity)
extends into the transport phase (74), where a sharp increase in the rate of solids
carryover occurs as the transport velocity is approached. In the absence of any
solid recycle, the bed would rapidly disappear. Beyond this velocity, catalyst
fed to the base of the regenerator travels upward in a fully entrained transport
flow. The voidage or density of the resulting suspension is dependent not only on
the velocity of the gas but also on the solid flow rate. If the solid rate is low, a
dilute-phase flow will result. If solids are fed to the regenerator at a sufficiently
high rate, eg, by recirculating the solid carried over back to the regenerator, then
maintaining a relatively large solid concentration is possible. This condition is
referred to as the fast-fluidized bed (73).
The high velocities in the combustor promote catalyst to air contact . As a
result, the catalyst residence time required for catalyst regeneration is greatly
reduced. The highly efficient mixing of the air and the catalyst improves the
overall conversion of CO to CO2, thus minimizing and in many cases eliminating
the need for CO combustion promoter (51).
Vol. 1 FLUID CATALYTIC CRACKING, UNITS, REGENERATION 467

Exit wt fraction/time
A

Time, s

Fig. 8. Theoretical residence time distributions: A, combustor style approach to plug


flow; B, turbulent bed (100% backmixed).

Figure 8 compares theoretical particle residence-time distributions of a


turbulent-bed regenerator with those of a modern high efficiency combustor.
The basic difference in response curves is a result of both the fluidization regime
(fast fluidized) and the solids entry and exit configuration. In the combustor
design, a spent catalyst particle cannot possibly leave the regenerator without
passing through the dilute-phase combustion zone.
The overall benefits of this high efficiency combustor over a conventional
bubbling- or turbulent-bed regenerator are enhanced and controlled carbon-
burn kinetics (carbon on regenerated catalyst at <0.05 wt%); ease of start-up
and routine operability; uniform radial carbon and temperature profiles; limited
afterburn in the upper regenerator section and uniform cyclone temperatures;
and reduced catalyst inventory and air-blower horsepower (73). As of 2003, this
design was well established, with more than 45 units in commercial operation.

8. Mechanical Hardware

From a mechanical point of view, the FCCU regenerator can be divided into two
main sections: the regenerator vessel and its internals, which include the air dis-
tributor, the cyclones, the plenum chamber, and catalyst coolers; and the flue gas
handling section, which includes power, heat, and emission control systems (SOx,
NOx, and particulates). Continuous advances in the mechanical design of both
these areas have led to significant improvements in overall FCCU reliability,
even though processing conditions, at the same time, have become more difficult
(harder and denser catalysts, higher regenerator temperatures).
8.1. Regenerator Vessel and Internals. The FCCU regenerator is one
of the largest vessels in the refinery (up to 18 m in diameter) and operates at tem-
peratures up to 7758C. The regenerator is usually a carbon steel vessel, intern-
ally lined with refractory insulation to maintain the wall temperature in the area
of 3408C, which is suitable for carbon steel.
Air Distributor Grid. The function of the FCCU regenerator air distributor
is to uniformly introduce O2 into the regenerator vessel to achieve even burning
of coke throughout the regenerator cross-section. Every air distributor must be
468 FLUID CATALYTIC CRACKING, UNITS, REGENERATION Vol. 1

Fig. 9. Regenerator pipe grid air distributor.

designed for three conditions: start-up, normal operation, and emergency. The
actual design temperatures and pressures may vary from unit to unit, depending
on the process requirements and configuration. However, these three design
conditions are common to all units and must be incorporated into the
design. The extreme temperature of the emergency case usually dictates the
selection of the distributor metallurgy.
Basically, two types of air distribution systems are used in FCCU regenera-
tors: those that distribute air into the catalyst bed through a perforated plate and
those that distribute air through a series of small pipes, known as a pipe grid
(77). The pipe grid design is the most commonly used. A typical pipe grid design
is shown in Figure 9. The pipe, which is typically constructed of 304 H stainless
steel or 1 14 Cr-12 Mo steel, distributes the air through three or four large laterals
into a number of small branches. The grid is designed for a total pressure drop in
the range of 3.5–7.0 kPa (0.5–1 psi). This pressure drop is set by regulating the
air leaving the branch arms through a series of nozzles that point downward.
Erosion patterns at the nozzle tip are minimized with the dual-diameter jets
that allow the air to develop a full-flow velocity profile across the entire nozzle
cross-sectional area prior to exiting. Frequently, some of these nozzles are
plugged to prevent jet impingement on adjacent pipes; increase pressure drop;
and enhance carbon and air distribution within the bubbling bed. The upper
operations limit of the grid is based on jet velocity limitations so as not to
cause catalyst attrition. Conversely, the downturn capacity is set by low pressure
drop, which can lead to aspiration of catalyst through the peripheral branches.
Such aspiration results in severe and accelerated erosion.
8.2. Cyclones. The high gas velocities inside the regenerator can result
in considerable catalyst entrainment into the vapor space above the catalyst bed.
Cyclones are used to separate these catalyst particles from the exiting regenera-
tor flue gas and return essentially all of these particles to the regenerator bed. In
a typical regenerator, the cyclones are generally installed in sets of two (Fig. 10).
In this two-stage arrangement, the gas outlet of the first cyclone is channeled
directly to the inlet of the second cyclone.
The cyclones are typically designed with diameters of 100–150 cm for ease
of maintenance. Cyclone inlet velocities are usually restricted to 18–22 m/s
in the first stage and to 21–26 m/s in the second stage to achieve satisfactory
Vol. 1 FLUID CATALYTIC CRACKING, UNITS, REGENERATION 469

Vapor outlet

Vapor/catalyst
inlet
P (Hopper)

Catalyst head

Dip legs

P (Dip leg
outlet)

Fig. 10. Cyclone system.

efficiency, pressure drop, and erosion characteristics (57). The number of sets of
two-stage cyclones thus depends on the total gas flow. Finding room to house all
the necessary cyclones within the regenerator frequently requires considerable
ingenuity (78).
Catalyst leaving the cyclone flows down the cyclone discharge pipe, known
as the dipleg, and builds a layer of catalyst at the bottom of this pipe. Sufficient
catalyst head is developed in the dipleg to overcome the cyclone inlet and cyclone
cone outlet pressure drop and thus allow the entrained solids to return to the
regenerator bed. The second stage sets the required overall dip leg length, and
the catalyst seal is maintained with either the dipleg being submerged in the
bubbling bed or with a trickle-flapper valve arrangement at the catalyst dis-
charge end. The open bottom, submerged dipleg with only a splash baffle, is
used primarily on regenerator first-stage cyclones. When the diplegs are unsub-
merged in a bubbling bed, a flapper valve arrangement is utilized at the catalyst
discharge end of the dipleg.
470 FLUID CATALYTIC CRACKING, UNITS, REGENERATION Vol. 1

The metallurgy of the cyclone equipment has in recent years focused


primarily on type 304 H stainless steel. The 304 H material is durable and
easy to fabricate and repair, withstands the high regenerator temperatures,
and is oxidation- and corrosion-resistant. Essentially all internal surfaces of
the cyclone that are subject to erosion are protected with a 2 cm layer of
erosion-resistant lining. When installed and cured, most refractory linings are
highly resistant to erosion.
8.3. FCC Catalyst Coolers. Heat-removal systems have been used in
commercial FCCUs since the early 1940s. The earliest coolers utilized tubular
cooling coils located inside the regenerator. The tube banks were normally
arranged horizontally in rows of three or four, but because of their location
in a continuously bubbling, or turbulent catalyst bed, they offered limited duty
flexibility, with no shutdown or start-up potential.
Because of the drawbacks to the internal cooling coil, cooler design has
evolved to an external exchanger type. The exchanger is either designed to have
catalyst flowing dense phase down and through the exchanger to the lower portion
of the regenerator, or to have catalyst flowing down into the cooler, backmixing,
and returning back up into the regenerator. External cooler design improvements
introduced in the middle 1980s have resulted in excellent mechanical reliability,
and the same, or better, on-stream efficiency as many of the other mechanical com-
ponents of the FCCU system. The design also has a wide range of easily controlled
heat removal duties. This all-vertical bundle design allows either a backmixed or
flow-through configuration (Fig. 11). The bayonet tube design uses forced water
circulation and allows for catalyst to flow on the shellside of the bundle. Both
the catalyst flux and aeration velocities are controlled at low levels and at
the same time controllably produce a wide range of heat-transfer coefficients.
Concerns regarding erosion or differential are minimal. The independently con-
trolled catalyst flow produces a countercurrent gas (bubble-phase) and solid

Fig. 11. Catalyst cooler configurations where (a) is a backmix cooler and (b) is a flow-
through catalyst cooler (79).
Vol. 1 FLUID CATALYTIC CRACKING, UNITS, REGENERATION 471

flow. Typically, a single cooler can be operated and controlled anywhere within a
6–50  106 watt saturated-steam production range.
8.4. Flue Gas Handling. Waste Heat Boilers. In a conventional
FCCU flue gas system, the regenerator combustion gases generally pass through
two stages of cyclonic separators, a slide valve, orifice chamber, waste heat boi-
ler, and electrostatic precipitator or third stage separator. The slide valve and
orifice chamber act in combination to reduce the flue gas to essentially atmo-
spheric pressure.
If the CO is not completely combusted to CO2 in the regenerator, a
CO boiler is used to complete the combustion. The resulting heat of combus-
tion and the sensible heat of the flue gas along with any heat from auxiliary
fired fuel are recovered in the form of high pressure steam. When the regenerator
is operated in total CO burn, the CO boiler is replaced with either a shell and
tube exchanger or a box-type waste heat boiler.
The shell and tube exchanger, which is a single-purpose piece of equip-
ment, has the standard parts of a normal heat exchanger, but certain parts
have been modified to withstand the erosive service of the particulate-
laden flue gas stream. This design has the limitation of being able to produce
only saturated steam. The flue gas leaves the regenerator at temperatures up
to 7608C and exits the exchanger at 2908C. The flue gas then proceeds through
the pressure-control slide valve and orifice chamber, which reduces its pressure
to essentially atmospheric prior to entering the emission control system or stack.
The box cooler receives regenerator flue gas after it has been reduced to
essentially atmospheric pressure. This arrangement is not limited to the produc-
tion of saturated steam. Any number of coils can be installed in the box, and nor-
mally both steam-generating and superheater coils are present. The tube
temperatures within the box must be maintained above the SO3 dew point of
150–1758C to prevent sulfuric acid corrosion (80).
The principal differences between the shell and tube and the box designs
may be summarized as follows:

Shell and tube Box


flue gas pressure 138–276 kPa atmospheric
type of steam saturated saturated plus superheated
steam pressure limit 3500 kPa none

8.5. Power Recovery. All FCC Units emit a form of ‘‘zero cost’’ energy
as part of the flue gas discharge from the FCC regenerator. Typically, these
gasses are in the 7008C range at anywhere from 240 to 380 kPa. In many
cases, this hot, pressurized gas can economically be utilized by a Power Recov-
ery System to either drive the FCC main air blower, or produce electricity for
the refinery grid.
Types of Systems. There are several possible arrangements of Power
Recovery Systems available for incorporation into new or existing FCC Unit
operations. These systems can include 2, 3, 4, and 5 component systems (expan-
der, main air blower, steam turbine, gear, motor/generator), specifically designed
to meet a multitude of processing and economic requirements (81).
472 FLUID CATALYTIC CRACKING, UNITS, REGENERATION Vol. 1

For new FCC unit applications, where plot space is generally not a primary
consideration, the power recovery expander will most often be directly connected
to the main air blower to drive the compressor in place of a steam turbine or
motor drive. This eliminates the need for either a large electric motor, or a
steam turbine to drive the main air blower requiring a surface condenser, utility
support system, and condensate pumps.
For revamp applications, where plot space is most often a primary consid-
eration, and the main air blower and driver are already in place, it is generally
unlikely that a new expander can economically be positioned, and directly con-
nected to the existing main air blower. In these situations, a stand-alone power
recovery expander is typically used to drive an electrical generator to produce
electricity for the refinery, rather than drive the main air blower. This installa-
tion can be a three or four component system. Generally, the three component
power recovery system (expander, gear, generator), protected by a third stage
separator, is the easiest configuration to add to an existing FCC unit. On
many occasions, the addition of a let-down steam turbine (four body system)
can be added to this train to generate additional power. In all cases, the installa-
tion of a power recovery system requires a large capital investment.
Process Flow. In a typical power recovery system, hot flue gas leaving
the regenerator goes to a third stage separator, where roughly 98% of the 10 m
and larger particulate material is removed to protect the expander blades in the
power recovery unit, as described earlier. Other particulate removal systems
(ESP or scrubber) are not appropriate for power recovery use, since they operate
at essentially atmospheric pressure.

Fig. 12. Regenerator Flue Gas System, Before and After Installation of a Power
Recovery System.
Vol. 1 FLUID CATALYTIC CRACKING, UNITS, REGENERATION 473

This ‘‘clean’’ flue gas leaving the separator is then sent through two butter-
fly control valves used to direct and throttle the flue gas, etc, and through the
Power Recovery Expander where the pressure in the flue gas is converted to
direct mechanical energy to drive a generator and produce electricity. The expan-
der acts as a restriction orifice and together with the inlet line butterfly valves,
maintains the regenerator operating pressure. A bypass valve is provided
upstream of the inlet butterfly valves to divert a portion of the flue gas around
the expander during periods when the potential recoverable energy exceeds the
capacity of the Power Recovery Train, or when the Power Recovery Train needs
to be bypassed for maintenance. Finally, the expander exhaust gas, generally
110–120 kPa at this point, is directed back to the flue gas system (downstream
of the orifice chamber) for further processing: steam production, particulate
removal, and atmospheric discharge, as seen in Figure 12.
Power Production. The power recovered from the flue gas is approxi-
mately equal to the main air blower requirements. On a BPSD basis, using a
50,000 BPSD FCC ( 2.5 MM t/year) operating at 310 kPa (30 PSIG), producing
approximately 200,000 kg/h (450,000 lb/h) of flue gas at 7208C (13258F), one
could expect this system to produce 11 MW of electrical power.

BIBLIOGRAPHY

‘‘Catalyst, Regeneration Fluid’’ in ECT 4th ed., Vol. 5, pp. 419–448, by L. L. Upson and
D. A. Lomas, UOP Research Center; ‘‘Catalyst Regeneration, FCC Units’’ in ECT (online),
posting date: December 4, 2000, by L. L. Upson, D. A. Lomas, UOP Research Center.

CITED PUBLICATIONS

1. D. Nakimura, Oil Gas J., 62 (Dec. 23, 2002).


2. 1996 American Petroleum Institute/ National Petroleum Refiners Association Survey
of Refining Operations and Product Quality, Final Report, July 1997.
3. L. G. Southard, 1971 International Petroleum Annual, Bureau of Mines, U.S. Depart-
ment of the Interior, Washington, D.C., Mar. 1973.
4. ‘‘World Wide Report,’’ Oil Gas J. (Dec. 31, 1990).
5. ‘‘World Wide Report,’’ Oil Gas J. (Dec. 29, 1989).
6. C. A. Cabrera and D. Knepper, ‘‘Advanced Reactor Design for FCC Units,’’ paper AM
90–39, presented at 1990 NPRA Annual Meeting, San Antonio, Tex., Mar. 1990.
7. UOP 2002 Technical Data Sheet, ‘‘Optimix FCC Feed Distribution System.’’
8. L. L. Upson and D. A. Wegerer, ‘‘Rapid Disengager Techniques in Riser Design,’’
presented at the ACS 3rd International Symposium on Advances in Fluid Catalytic
Cracking, Chicago, Il., Aug. 22–27, 1993.
9. L. L. Upson, I. Dalin, and W. R. Wichers, ‘‘Heat Balance: The Key to Catalytic Crack-
ing,’’ presented at Katalistiks 3rd FCC Symposium, Amsterdam, The Netherlands,
May 1982.
10. R. N. Cimbalo, R. L. Foster, and S. J. Wachtel, ‘‘Deposited Metals Poison FCC
Catalyst,’’ Oil Gas J (May 15, 1972).
11. U.S. Pat. 3,791,962 (1974), E. J. Demmel and H. Owen.
12. J. L. Mauleon and J. C. Courcelle, ‘‘FCC Heat Balance Considerations with Heavy
Feeds,’’ presented at Katalistiks 6th Annual FCC Symposium, Munich, Germany,
May 1985.
474 FLUID CATALYTIC CRACKING, UNITS, REGENERATION Vol. 1

13. A. Voorhies, Jr., Ind. Eng. Chem. 37, 318 (Apr. 1945).
14. V. W. Weekman, Jr., and D. M. Nace, AIChE J. 16, 397 (May 1970).
15. R. W. Mott, Oil Gas J., 73 (Jan. 26, 1987).
16. R. R. Dean, J. L. Mauleon, and W. S. Letzsch, Oil Gas J., 75 (Oct. 4, 1982).
17. J. S. Magee, W. E. Cormier, and G. M. Woltermann, Oil Gas J., 59 (May 27, 1985).
18. L. L. Upson, ‘‘How to Design Your Own FCC Catalyst,’’ presented at Katalistiks 8th
Annual FCC Symposium, Budapest, Hungary, June 1987.
19. G. H. Dale and D. L. McKay, Hydrocarbon Process. 56, 97 (1977).
20. P. Ramamoorthy, A. R. English, J. V. Kennedy, L. W. Jossens, and A. S. Krishna,
‘‘A New Metals Passivator in Fluid Catalytic Cracking,’’ presented at 1988 Annual
NPRA Meeting, San Antonio, Tex., Mar. 1988.
21. W. S. Letzsch, A. G. Ashton, and L. L. Upson, Hydrocarbon Proc. 70, 89 (June 1991).
22. C. C. Wear ‘‘Orion-LC Catalysts Set New Standards for Nickel Tolerance,’’ paper pre-
sented at Grace Davison FCC Technology Conference, Athens, Greece, September, 1994.
23. E. G. Wollastin, W. J. Haflin, W. D. Ford, and G. J. D’Souza, Hydrocarbon Proc., 93
(Sept. 1975).
24. P. L. Walker, F. Rusinko, and L. G. Austin, Advances in Catalysts XI, Academic Press,
New York, 1959, pp. 133–221.
25. J. R. Arthur, Trans. Faraday Soc. 47, 164 (1951).
26. P. B. Weisz, J. Catal. 6, 425 (1966).
27. F. E. Massoth, I. E.C. Proc. Des. Devel. 6, 200 (Apr. 1967).
28. P. B. Weisz and R. B. Goodwin, J. Catal. 6, 227 (1966).
29. J. C. Dart, R. T. Savage, and C. G. Kirkbride, Chem. Eng. Prog. 45, 102 (1949).
30. R. G. Haldeman and M. C. Botty, J. Phys. Chem. 63, 489 (1959).
31. T. B. Metcalfe, Brit. Chem. Eng. 12, 388 (Mar. 1967).
32. T. Hano, F. Nakashio, and K. Kusanoki, J. Chem. Eng. (Jpn) 8, 127 (1975).
33. L. L. Upson and H. van der Zwan, Oil Gas J., 65 (Nov. 23, 1987).
34. B. Lewis and G. Von Elbe, Combustions, Flames, and Explosions of Gases, 2nd ed.,
Academic Press, New York, 1961, p. 71.
35. U.S. Pat. 4,064,039 (Jan. 1978), J. E. Pennick.
36. U.S. Pats. 4,072,600 (Feb. 7, 1978) and 4,093,535 (June 6, 1978), A. B. Schwartz.
37. Y. Guegan, ‘‘Use of Combustion Promoter,’’ presented at Katalistiks 1st FCCU
Symposium, Bordeaux, France, Oct. 1980.
38. L. L. Upson, ‘‘Catalytically Promoted Combustion Improves FCC Operations,’’ paper No.
AM 79–39 presented at the 1979 NPRA Annual Meeting, San Antonio, Tex., Mar. 1979.
39. A. Vierhelig, R. Becker, and M. Evans, ‘‘The Role of Additives in Reducing FCC Emis-
sions to Meet Legislation,’’ paper No. AM-03-97 presented at the 2003 NPRA Annual
Meeting, San Antonio, Tex., Mar. 2003.
40. A. W. Peters, X. Zhao, and G. Weatherbee, ‘‘The Origin of NOx in the FCCU Regenerator,’’
Paper No. AM-95-59, presented at the 1995 NPRA Annual Meeting, San Francisco, Calif.
41. Phillip K. Niccum, Eusebius Gbordzoe, Stephen Lang, ‘‘FCC Flue Gas Emission
Control Options,’’ Paper No. AM-02-27 presented at the 2002 NPRA Annual Meeting,
San Antonio, Tex., Mar. 2002.
42. V. J. Memmott and B. Dodds ‘‘Innovative Technology Meets Processing and Environ-
mental Goals: Flying J Commissions New MSCC and TSS,’’ Paper AM-03-13
presented at the 2003 NPRA National Meeting, San Antonio, Tex., March, 2003.
43. L. L. Upson, R. J. Lawson, W. E. Cormier, and F. J. Baars, Oil Gas J., 64–74 (Oct. 1, 1990).
44. M. T. Smith and T. F. Petti, High Unit Retention Critical to Efficient Operation of
Power Recovery Trains, Grace Davison Catalagram, No. 88, 2001.
45. B. Poot, F. Fabiani, L. Reuvers, A. Selvaggi, L. White, Sulfur Dioxide Emissions from
Oil Refineries and Combustion of Oil Products in Western Europe and Hungary,
CONCAWE REPORT No. 10/02.
Vol. 1 FLUID CATALYTIC CRACKING, UNITS, REGENERATION 475

46. J. L. Mauleon and J. Marvillet, ‘‘Control and Monitoring of FCC Flue Gas SOx Emis-
sions,’’ presented at 83rd Annual Meeting Air and Waste Management Association,
Pittsburgh, Pa., June 1990.
47. R. H. Gilman, M. Y. Asim, and T. A. Reid, ‘‘Optimizing FCC Operations using
Pretreatment to Meet Future Market Challenges,’’ paper No. AM-91-37, presented
at the 1991 NPRA Annual meeting, San Antonio, Tex., March 1991.
48. L. D. Krenzke, K. Baron, ‘‘FCC Pretreating to meet New Environmental Regulations
on Gasoline,’’ Paper No. AM-95-67 presented at 1995 NPRA Annual Meeting, San
Francisco, Calif., March 1995.
49. T. A. Nguyen, and M. Skripek, ‘‘Reducing Sulfur in FCC Gasoline via Hydrotreating’’
Paper #54C, Presented at AICHE 1994 Spring meeting April 1994.
50. B. J. Arena, M. W. Schnaith, W. R. Abma, H. Dijkman, S. F. Meyer, and D. C. Nokes,
‘‘State of the Art Processes for Handling FCC Flue Gas Emissions,’’ Paper No. ENV-99-
186 presented at 1999 NPRA Environmental Conference, Dallas, Tex., September 1999.
51. CONCAWE Air and Water Quality Management Group, Special Task Force AQ/STF-
55 and WQ/STF-28, ‘‘Best Available Techniques to reduce emissions from Refineries,’’
Document No. 99/01, Brussels, May 1999.
52. J. W. Wilson, Fluid Catalytic Cracking; Technology and Operation, PennWell
Publishing Co. 1997.
53. E. H. Weaver, M. J. Barrasso, and J. B. Jarvis, ‘‘An Update of Wet Scrubbing Tech-
nology for FCCUS–Multiple Pollutant Control,’’ Paper No. AM-03-120 presented at
2003 NPRA Annual Meeting, San Antonio, Tex., March 2003.
54. A. A. Bhattacharyya, G. M. Woltermann, J. S. Yoo, J. A. Karch, and W. E. Cormier,
‘‘Catalytic SOx Abatement of FCC Flue Gases,’’ presented at 194th National Meeting
of the ACS, New Orleans, La., Aug. 31–Sept. 4, 1987.
55. U. S. Pat. 4,790,982 (Dec. 13, 1988), J. S. Yoo, C. A. Radlowski, J. A. Karch, and A.
Bhattacharyya.
56. J. W. Powell, W. S. Letzsch, R. M. Benslay, K. C. Chuang, and R. Bartek, ‘‘Advanced
FCC Flue Gas Desulfurization Technology,’’ paper No. AN-88-49 presented at the
1988 NPRA Annual Meeting, San Antonio, Tex., Mar. 1988.
57. T. J. Dougan, S. K. Purnell, and J. R. Riley, ‘‘FCC Catalyst Solutions for Gasoline
Sulfur, SOx and NOx,’’ paper presented at the 2003 KBR-Exxon Mobil FCC Licensee
Symposium.
58. A. W. Peters, K. L. Dishman, P. C. Doolin, and L. D. Turlock, Ind. Eng. Chem. Res. 37,
4631 (1998).
59. R. Sadeghbeigi, ‘‘NOx Control in FCC,’’ Paper No. AM-02-55 presented at the 2003
NPRA Annual Meeting, San Antonio, Tex., March 2002.
60. M. L. Batchelder, ed., 1989 NPRA Question and Answer Session on Refinery and
Petrochemical Technology, Gerald L. Farrar & Assoc., Tulsa, Okla., pp. 76–77.
61. R. Miller and M. Skripek, ‘‘Upgrading FCC Feeds with UNOCAL Hydroprocessing
Technology,’’ paper presented at UNOCAL Symposium, Vienna, Austria, June
1990.
62. C. Kuehler, ‘‘Meeting Clean Fuels and Clean Air Objectives with the FCC,’’ paper
presented at the 2003 Akzo Nobel Catalyst Technical Seminar, Banff, Canada,
July, 2003.
63. D. G. Olson, K. Tsuji, I. Shiraishi, and M. R. Miller ‘‘The Reduction of Gas Phase Air
Toxics from Combustion and Incineration Sources Using the MET-Mitsui-BF
Activated Coke Process,’’ paper presented at 1999 International Ash Utilization
Symposium, Lexington, N.Y., Oct. 1999.
64. G. Phillips, ‘‘CO2 Management in Refineries,’’ I. Chem. E. 5th International Gasifica-
tion Conference, Noordwijk, the Netherlands, April, 2002.
65. NPRA Web Site, www.NPRAdc.org, ‘‘Global Climate Change,’’ NPRA Issues.
476 FLUID CATALYTIC CRACKING, UNITS, REGENERATION Vol. 1

66. L. C. Yen, R. E. Wrench, and C. M. Kuo, ‘‘The Important Role of Regenerator Tem-
perature in Catalytic Cracking,’’ presented at Katalistiks 6th Annual FCC Sympo-
sium, Munich, Germany, May 1985.
67. J. R. Murphy and Y. L. Cheng, ‘‘The Interaction of Heat Balance and Operating
Variables in Zeolitic Catalyst Operations,’’ presented at Katalistiks 5th Annual
FCC Symposium, Vienna, Austria, May 1984.
68. G. Wilson, in G. L. Farrar, ed., transcript of 1977 NPRA Q and A Symposium, 1977, p. 54.
69. D. Kunii and O. Levenspiel, Fluidization Engineering, Krieger Publishing, New York,
1977, p. 18.
70. D. A. Lomas, C. A. Cabrera, D. M. Cepla, C. L. Hemler, and L. L. Upson, ‘‘Controlled
Catalytic Cracking,’’ presented at 1990 UOP Technology Conference, Chicago, Ill.,
Apr. 1990.
71. J. L. Mauleon, J. B. Sigaud, and G. Heinrich, ‘‘FCC Heat Balance Management with
Heavy Feeds, MTC Approach,’’ presented at JPI Petroleum Refinery Conference,
Tokyo, Japan, Oct. 1986.
72. D. A. Kauff and M. W. Schnaith, ‘‘FCC Regenerator Options to Process Residue
Feedstocks,’’ paper presented at the Rationalizing Capacity and Profitability in Asian
Oil Refining Symposium, Phuket, Thailand, February, 1997.
73. D. H. Turner, ‘‘Turbulent and Fast Fluidization (High Velocity Fluidization)’’ City
University of N.Y., Ph.D. dessertation, 1979.
74. L. L. Upson, C. L. Hemler, and D. A. Lomas, ‘‘Unit Design and Operational Control:
Impact on Product Yields and Product Quality, Chapter 11 of Fluid Catalytic Crack-
ing Science and Technology,’’ Elsevier, 1993.
75. J. Yerushalmi and N. T. Cankurt, Powder Technol. 24, 187 (1979).
76. Cyclone Separators, API Publication No. 931, May 1975.
77. P. E. Glascow and A. A. Murcia, ‘‘Process and Mechanical Design Considerations
for FCC Regeneration Air Distributors,’’ presented at Katalistiks 5th Annual FCC
Symposium, Vienna, Austria, May 1984.
78. N. L. Giuricich and B. Kalen, ‘‘Dominant Criteria in FCC Cyclone Design,’’ presented
at Katalistiks 3rd Annual FCC Symposium, Amsterdam, The Netherlands, May 1982.
79. C. A. Cabrera, C. L. Hemler, and S. P. Davis, ‘‘Improve Refinery Economics via
Enhanced FCC Operations,’’ presented at Katalistiks 8th Annual FCC Symposium,
Budapest, Hungary, June 1987.
80. R. R. Pierce, Chem. Eng. 84, 125 (Apr. 11, 1977).
81. L. A. Hissink and A. W. Drake ‘‘Power Recovery Expander Experience,’’ paper
presented at Katalistiks 5th Annual FCC Symposium, Vienna, Austria, May. 1984.

GENERAL REFERENCE

D. Selberg and J. R. Riley, Super DeSOx Provides Industry Leading Effectiveness, Grace
Davison Catalagram, No. 92, 2003.

L. L. UPSON
F. S. ROSSER
C. L. HEMLER
P. PALMAS
L. E. BELL
W. J. REAGAN
B. W. HEDRICK
UOP LLC
Vol. 1 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES 477

FLUID CATALYTIC
CRACKING, CATALYSTS
AND ADDITIVES
1. Introduction

Fluid catalytic cracking (FCC) is a central technology in modern refining (1,2)


and is used to upgrade a wide variety of gas, oil, and residual feedstocks to
gasoline, diesel fuel, and light gases. The first FCC unit, PCLA-1, came on
stream in 1942 and though considered a mature technology, both hardware
and catalyst developments continue to be made. A brief history of the impor-
tant events in the development of FCC catalysts is shown in Table 1. In the
1960s, the introduction of zeolite Y revolutionized the process by increasing
gasoline selectivity, adding almost $2 of value per barrel of feed processed
(3). Before this breakthrough, amorphous silica-alumina catalysts were
used that had poor gasoline selectivity and made much coke and gas com-
pared with zeolite-based catalysts. More than 40 years later, zeolite Y, or
more precisely, modified versions of zeolite Y, continues to be at the heart
of the modern FCC catalyst.
In the 1980s, the introduction of a second zeolite, ZSM-5, dramatically
improved the ability of refiners to increase both gasoline octane number and
the yields of light olefins (propylene and butylenes). Originally employed to
improve gasoline octane, the focus of ZSM-5 additive use has switched more
toward the production of propylene as the demand for FCC propylene has
increased significantly to supply the growing petrochemicals industry.
In the early 1990s, the introduction of new alumina technologies with a
higher tolerance to the contaminant metals Ni and V further increased the flex-
ibility of the FCC unit to process heavier crude sources (4).
In this article, we focus on developments in FCC catalyst technology and
the driving forces behind them. Although the FCC process has been used for
more than one-half of a century, new and important developments continue to
be made in several areas. Some of these new developments are a direct response
to environmental regulation. Another important driving force is the goal to pro-
cess heavier crude sources with higher levels of contaminants. Changes in hard-
ware, such as the increasing use of advanced riser termination technology to
further reduce catalyst/oil contact time have also resulted in significant changes
in FCC catalyst technology. With the increasing integration of petroleum refin-
ing and chemicals production, the value of specific molecules in the product spec-
trum is driving FCC catalysts toward specific selectivities. Each of these trends
has long-term implications for the development of future catalysts and hardware
for the FCC unit.

2. FCC Catalyst Design Fundamentals

FCC catalysts are generally comprised of zeolite, clay, active-matrix, and binder.
The primary source of cracking activity comes from the zeolite component.

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.fluidnee.a01
478 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES Vol. 1

Table 1. Important Dates in FCC Catalyst Development


Development
year
aluminum chloride catalyst 1915
activated clay catalysts (Houdry) 1928
silica/alumina catalyst (Houdry) 1940
first commercial production of FCC catalyst (Davison) 1942
commercial production of microspheroidal catalysts (Davison) 1948
synthetic high alumina catalyst (Davison) 1955
commercial production of zeolite Y (Union Carbide) 1959
introduction of zeolitic FCC catalyst (Mobil) 1962
introduction of USY and REUSY zeolites (Davison) 1964
silica-sol bound catalysts (Davison) 1973
introduction of Pt combustion promoter (Mobil) 1974
feed-added Ni passivation technology (Phillips) 1975
alumina-sol bound catalysts (Davison) 1981
commercialization of spinel SOx additive (ARCO) 1982
introduction of ZSM-5 (Mobil) 1984
coke-selective, deep bottoms cracking matrices (Davison) 1986
introduction of Ni-tolerant matrix technologies (Davison) 1990
commercialization of V-trapping technology (Davison) 1992
introduction of additives for gasoline sulfur reduction (Davison) 1995
introduction of CSSN/Z-17 high activity zeolites (Davison) 1996
commercialization of NOx control additive technologies (Davison) 1997
introduction of catalyst for sulfur reduction (Davison) 2001
introduction of Z-28 and Z-30 high activity zeolites (Davison) 2003
introduction of Tunable Reactive Matrices (Davison) 2004

Zeolite Y (faujasite) has been used in FCC catalysts since the early 1960s,
however, a tremendous amount of R&D effort over the years has led to the con-
tinuous introduction of zeolite Y derivatives with a wide variety of performance-
enhancing properties.
The principle modifications involve zeolite ultrastabilization and rare-earth
ion exchange. In the ultrastabilization process, steam calcination is used to
increase the framework Si/Al ratio of the zeolite Y. Not only does this process
result in an increase in the intrinsic stability of the zeolite, it also results in
the development of a secondary mesoporous network of pores that enhance the
bottoms upgrading activity of the zeolite.
Rare earth ion exchange increases the activity of the zeolite and retards
crystal destruction and dealumination in the hydrothermal environment of the
regenerator. The higher the degree of rare-earth exchange of the zeolite, the
higher the unit-cell size of the equilibrated zeolite in the equilibrium catalyst.
Increasing the degree of rare earth exchange increases the gasoline selectivity
of the FCC catalyst.
Most fluid cracking catalysts, especially those used in resid applications,
are manufactured with an active-matrix component usually based on specialty
aluminas. Active-matrix contributes a catalytically active surface that derives
its activity (and selectivity) predominantly from porous silica-alumina that
Vol. 1 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES 479

may have undergone chemical or physical modification to enhance activity as


well as other aspects of its performance. Of importance is the maximization of
pores in the range 100–600 Å, since these pores are known to be important for
coke-selective bottoms upgrading (5). Recently, new FCC catalyst technology
was introduced based on the ability to tune pore size distribution and matrix
acidity (6).
Clay is an important component that fulfills the remaining catalyst particle
performance requirements. Clay serves as a heat sink and transfer medium. It
can also serve as a sink for sodium, improving the catalyst’s resistance to sodium
poisoning. Clay provides little or no activity to the catalyst, but it does provide
mechanical strength and density to the particle for optimum fluidization
properties.
The binder is the ‘‘glue’’ that holds all catalyst components together and
provides particle physical integrity. Some binders, such as alumina-sol polymers,
have intrinsic cracking activity and thus aid in bottoms cracking.
Figure 1 shows the principle steps in the manufacture of FCC catalysts.

ZEOLITE
MATRIX CLAY
SYNTHESIS
SYNTHESIS PROCESSING
& PROCESSING
ACTIVE
MATRIX
ZEOLITE INERT

BINDER BINDER
SYNTHESIS

MIXING /
DISPERSION

SPRAY DRYING

FINISHING

Major Advantages:

Careful control over quality


of ingredients

Flexibility in formulation

Consistent product properties possible


because of continuous processing

Fig. 1. Typical FCC catalyst manufacturing flow scheme.


480 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES Vol. 1

3. Residual Oil Processing

In today’s highly competitive market, refiners are under increasing pressure to


maximize profits in an environment where they, not only must operate their FCC
units to meet increasingly stringent environmental regulations, but also have to
produce products that meet tighter quality specifications. Nevertheless, with
declining demand for fuel oil many refiners have found that processing larger
amounts of lower cost resid feedstocks in the FCCU or RFCCU can substantially
increase profitability (7).
Resid feedstocks have a higher boiling range compared with vacuum gas
oil’s and are characterized by high concentrations of polynuclear naphthenes,
aromatics, and asphaltenes as well as high levels of contaminant metals, notably
nickel and vanadium and high Conradson Carbon residue. Not surprisingly,
proper design of the FCC catalyst is crucial for units operating with such
feedstocks.
For units processing resid feeds where performance is impacted by high
levels of contaminant metals and high feed Conradson Carbon residue, catalyst
coke selectivity is the most important performance characteristic. Catalyst stabi-
lity and metals tolerance are also very important features of a catalyst designed
for resid applications. To facilitate the cracking of the large hydrocarbon mole-
cules found in resid feeds, the catalyst must be designed to maximize diffusion
of these bulky species to the active acid sites (5,8). Zeolitic and matrix activity
must be optimized for each application and is dependent on feedstock properties
along with the dominant bottoms cracking mechanisms characteristic of the feed.
Resistance to less common catalyst poisons such as iron is critical for an increas-
ing number of refiners and can be addressed by the correct choice of matrix
technology. Finally, catalyst physical properties must be optimized to ensure
maximum unit retention (8). The influential effects of metals tolerance is
described below.
3.1. Contaminant Metals (Ni, V, and Fe). In the 1990s, a wide range of
technologies were developed to improve the tolerance of FCC catalysts to con-
taminant metals in the feed, primarily vanadium, and nickel (4). The dearoma-
tization function of nickel is responsible for an increase in coke yield as the Ni
level increases on the equilibrium catalyst. Moreover, the dehydrogenation
activity of the Ni leads to the formation of molecular hydrogen. Since hydrogen
is the most difficult gas to compress, an increase in the FCC unit can quickly
limit the wet gas compressor. In order to increase the tolerance of FCC catalysts
to Ni contamination, special matrix alumina technologies were developed. These
react with contaminant Ni, removing it from the surface and forming stable Ni–
aluminate phases in the bulk of the alumina (Fig. 2). Oxygen TPD measurements
provide evidence for the solid-state diffusion of Ni into the alumina away from
the surface and therefore the reactants.
Vanadium is particularly harmful to catalyst performance because it deac-
tivates the zeolite component and causes yields to deteriorate (9). Vanadic acid is
believed to form from vanadium oxides in FCCU regenerators in the presence of
steam:

V2 O5 ðsÞ þ 3 H2 OðvÞ ! 2 H3 VO4 ðvÞ


Vol. 1 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES 481

Ni
(a) Highly Dispersed - Poor Ni Tolerance
Ni Ni Ni Ni Ni Ni Ni Ni
Ni
Good Ni Support Ni Ni Ni Ni
High Ni Dehydrogenation Activity Ni

Ni Ni
Ni Ni Ni Ni Ni
(b) Nickel Agglomeration - Better Ni Tolerance Ni Ni Ni
Chemical Reaction
Poor Ni Support 100 Å
Low Ni Dehydrogenation Activity

(c) Solid State Diffusion - Best Ni Tolerance


Chemical Reaction Al
Strong Metal Support Interaction Al
Ni
Low Ni Dehydrogenation Activity
Al

Al NiAl2O4
Al

Fig. 2. Matrix technology passivates Ni by strong metal–support interactions.

Vanadium mobility varies from unit to unit, depending on, among other
things, the level of excess oxygen in the regenerator (10). Catalyst manufacturers
have developed more stable zeolites (eg, Z-17, Z-28, Z-30, CSSN, and CSX) and a
series of V traps to increase the ability of the zeolite to handle vanadium. These
traps are based on Ba, Ti, rare earth, and other elements. Some are more effec-
tive than others, but the basic idea is the same, ie, to keep the V away from the
zeolite by binding to the surface of an inactive particle (Fig. 3).

Reactor Heat (V3+, V4+) in Regenerator


Feed Porphyrins (V3+, V4+) + FCC Catalyst Catalyst Surface
and Reducing Oxidation
Coke
Environment
V2O3(V4+, V5+ )
on Catalyst Surface

c
Destruction of Zeolite Particle to Particle Mobile H3VO4, V5+ + Steam
and Formation of Unstable
Migration of H3VO4, V5+ on Catalyst
Zeolite as in A and B

b a

Destruction of Zeolite by + Rare Earth (RE)


Hydrolysis of SiO2/Al2O3 in Zeolite
Framework

RE + H3VO4 Forms
RE Vanadates Making
Zeolite Unstable

Fig. 3. Vanadium deactivation mechanism.


482 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES Vol. 1

9.0
DCR Testing on Resid Feedstock
8.5 Max Coke Selectivity CPS Deactivation at 1450˚F:
Benchmark 4000 ppm Ni, 4000 ppm V

Bottoms, wt% 8.0

7.5

7.0 Max Bottoms Cracking


Benchmark
6.5

6.0

5.5

5.0
2 3 4 5 6 7 8
Coke, wt%

Fig. 4. Recent developments have resulted in the introduction of step-out coke-selective


bottoms upgrading FCC catalyst technologies such as IMPACT.

Catalyst manufacturers are continuously striving to improve the stability of


their catalysts to metals poisoning as resid processing becomes more important
in the future. For example, AURORA-LC contains high stability Z-17 zeolite
and metals trapping matrix technology that in one commercial example,
allowed a refiner to increase profitability by $4.30/mton of feed in the presence
of 18,000 ppm Ni þ V (11).
More recently, a breakthrough in resid cracking catalyst technology was
made with the introduction of IMPACT (6). Utilizing new high stability Z-28 zeo-
lite technology in combination with a proprietary tunable reactive matrix system
that incorporates an integral vanadium trap, IMPACT has demonstrated step-
out coke selectivity and bottoms upgrading in the presence of very high levels
of contaminant metals (Fig. 4). In one U.S. Gulf Coast application, IMPACT
increased profitability by an estimated $3.10/mton at only 20% concentration
in the circulating catalyst inventory (Table 2) (6).
Understanding of the effect of Fe contaminants on the FCC catalyst is rela-
tively new and comes from direct observations in the field: dramatic losses in
activity sometimes coupled with a loss in bottoms cracking selectivity that
appears to correlate with decreased ABD (Average Bulk Density measured in
g/cm3) due to catalyst particles sticking to each other (12). The catalyst sticking
appeared in turn to be correlated with increasing Fe content on the equilibrium
catalyst. A detailed account of Fe deactivation is given in Ref. 12. A pictorial
representation of the mechanisms leading to catalyst deactivation by Fe for
SiO2 bound and Al2O3 bound catalysts is presented in Figure 5.
To minimize catalyst deactivation by contaminant Fe, refiners are recom-
mended to focus on the following where possible:

Feedstocks: Use low Fe, Na, and Ca feeds and reduce its acid content to lower
equipment corrosion.
FCC unit: Minimize the regeneration temperature.
FCC catalyst: Use Al2O3 bound catalysts instead of SiO2 bound ones.
Vol. 1 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES 483

Table 2. Performance of IMPACT Resid Cracking Catalyst


Technology at 20% Turnover
Resid catalyst benchmark 20% IMPACT
product yields
total C2-, wt% 3.1 3.1
total C3, vol% 11.5 11.2
total C4, vol% 18.0 17.4
naphtha, vol% 55.4 58.3
LCO, vol% 16.9 16.1
slurry, vol% 6.0 5.4
coke, wt% 5.6 5.6
conversion, vol% 77.1 78.5
profit margin
$/mton of fresh feed þ3.10

4. Short Contact Time (SCT) Cracking

In order to minimize nonselective post-riser cracking, many refiners are revamp-


ing their units to operate in the so-called ‘‘short contact time’’ mode. In the FCC
operation, catalyst and feed are contacted at the bottom of the riser with cracking

Fig. 5. Mechanism of catalyst deactivation by Fe.


484 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES Vol. 1

reactions proceeding along the length of the riser. Contact time over the length of
the riser is 5–8 s. At the end of the riser, the catalyst and products are sepa-
rated in the disengager. It is here that product vapor can undergo nonselective
(essentially thermal) cracking as hydrocarbon vapors come into contact with hot,
coked catalyst.
In the early 1990s, a closed cyclone system was introduced that featured
cyclones directly coupled to the end of the riser (13), which allowed rapid separa-
tion of the catalyst and products, reducing post riser cracking in the disengager.
Today, all major process licensors offer advanced riser termination technologies
of one sort or another with one objective in common—reduction of nonselective
post-riser cracking.
New processes are emerging where the catalyst/feed contact time is reduced
to <2 s. The so-called millisecond catalytic cracking (MSCC) process was intro-
duced in the early 1990s in the United States (14).
4.1. FCC Catalyst Design Considerations for SCT Operations.
Catalysts for SCT operations must be designed taking into account the following
considerations (15):

 Catalyst/oil contact time is shorter—leading to lower intrinsic conversion


and bottoms cracking.
 Reactor temperature is generally higher—leading to increased conversion
and light olefins yields (including increased gasoline olefinicity).
 Catalyst circulation rate is generally increased—leading to an overall
increase in selective conversion.
 Regenerator/regenerated catalyst temperatures are generally lower—
leading to cooler catalyst/oil mix zone temperature, hence the need for im-
proved feed nozzles to ensure efficient feed vaporisation.

To compensate for the reduced catalyst/feed contact time and consequently


loss in conversion, high activity catalysts are needed. The zeolite component
within the catalyst has the greatest effect on activity and therefore, proper zeo-
lite selection is very important. In addition, it is very important to choose a
matrix with the right level of activity for a particular application to address
the reduction in bottoms upgrading that occurs with shorter contact time. Excess
matrix activity is undesirable because it leads to poorer coke and gas selectiv-
ities. In addition to providing bottoms cracking activity, the matrix should be
selected to provide low coke and gas selectivities, particularly in the presence
of contaminant metals.

5. Clean Fuels Production

5.1. Gasoline Sulfur Reduction. Gasoline vapors and tailpipe emis-


sions contain NOx and VOCs, which can react in the atmosphere to produce
ozone, a major component of smog. Toxic compounds (eg, benzene, a known car-
cinogen) are also emitted by vehicles. In the United States, Federal Reformu-
lated Gasoline (RFG) Phase 2 specifications went into effect in 2000, requiring
Vol. 1 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES 485

reductions in tailpipe emissions of these pollutants as calculated by the


Environmental Protection Agency’s (EPA) Complex Model. Gasoline RVP, sulfur,
olefins, and aromatics content all play important roles in meeting RFG Phase 2
requirements.
Sulfur recently has come under even further scrutiny. Sulfur in gasoline
degrades the performance of catalytic converters by poisoning the active sites
(16). In order to improve catalytic converter performance and meet lower tail-
pipe emissions standards, the EPA has determined that gasoline sulfur levels
must be reduced. The EPAs Tier 2 specifications require that gasoline sulfur
average 30 ppm at the refinery level by 2006 with a per-gallon cap of
80 ppm. The 30 ppm requirement represents a >90% reduction in gasoline
sulfur from baseline levels. The California Air Resources Board (CARB) has
gone even further. Current CaRFG-2 specifications already require an average
gasoline sulfur level of 30 ppm and proposals for CaRFG-3 would lower average
gasoline sulfur to 15 ppm.
The rest of the world is not far behind. European specifications will
require 50 ppm sulfur in gasoline and diesel in 2005 with a likely move to
10 ppm by 2008. Many Asian countries including, but not limited to, Japan,
Korea, and India have also recently lowered sulfur specifications in transporta-
tion fuels.
In order to achieve these low levels, refiners have a number of choices
including gas oil hydrotreating, FCC feed hydrotreating, gasoline sulfur adsorp-
tion processes, and gasoline hydrofinishing (17). Each of these techniques has
debits associated with it. Some, like gasoline hydrofinishing, lead to a reduction
in gasoline olefins and thereby reduce gasoline octane. Some, like feed hydro-
treating, have a high demand for molecular hydrogen, which is often in short
supply in the refinery. Lowering gasoline end point can significantly lower gaso-
line volume. Furthermore, it may be difficult to find a home for the high sulfur
heavy gasoline, without further processing. Thus, there is great incentive for
refiners and catalyst manufacturers to find ways to reduce gasoline sulfur
directly in the FCC unit.
Gasoline Sulfur Speciation. FCC gasoline contributes to >90% of the
total gasoline pool sulfur with the remaining 10% originating from straight
run naphtha. The sulfur in the gasoline is concentrated (up to 50%) in the higher
boiling region (>1938C). Moreover, only 2–5% of the sulfur in the FCC feed-
stock ends up in the gasoline as summarized in Figure 6.
The sulfur compounds in the FCC gasoline typically comprise mercaptans,
thiophene, C1–4 substituted thiophenes, thiophenol, C1–2 substituted thiophe-
nols, tetrahydrothiophene, and benzothiophene (Fig. 7).
Reaction Mechanism. In an FCC unit, typically 35–45% of the feed
sulfur is converted to H2S, 2–5% is found in coke, 5% in gasoline and the
rest ends up in the LCO and bottoms (18,19). The reactivity of feed sulfur com-
pounds and the mechanism by which they end up in gasoline is a subject of
ongoing study. Mercaptans and sulfides are converted to H2S and do not signif-
icantly increase the amount of sulfur in gasoline. Hydrogen sulfide production
from FCC has been found to correlate well with the amount of non-thiophenic
sulfur in the feedstock (19). Huling and co-workers (20) and Nguyen and
486 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES Vol. 1

Fuel Gas, H2S 20–60%

F Gasoline 2–5%
Feed Sulfur
C Light Cycle Oil 10–25%

C Heavy Cycle Oil 5–35%

2–30%

Fig. 6. Sulfur distribution in the FCC products.

Skripek (21) correlated sulfur in FCC gasoline with the sulfur species in FCC
feed. Alkylthiophenes and aromatic sulfides are believed to be the key contribu-
tors to gasoline range sulfur.
The concentration of sulfur in gasoline is only weakly dependent on con-
version as a result of opposing effects (18,21). Mercaptans and alkylthiophenes
are unstable with respect to desulfurization. The concentration of these species
decrease with increasing contact time or as gasoline overcracking occurs. How-
ever, benzothiophene, formed by dealkylation of higher molecular weight sul-
fur compounds, is relatively stable with respect to desulfurization. In general,
the concentration of benzothiophene increases with conversion. At constant
conversion, the concentration of sulfur in gasoline increases with riser tem-
perature due to increased dealkylation reactions of aromatic sulfur species.

Benzothiophene
2.8e5
2.7e5
Methylthiophenol
2.6e5
2.5e5 Methylthiophene
2.4e5 Thiophene
Benzenethiol

Me-THT
2.3e5
C3-thiophene

C2-Thiophene
2.2e5
C2-THT

2.1e5 THT C2-thiophenol


2.0e5
1.9e5
1.8e5
1.7e5
20 40 60 80 100
Time (min)

Fig. 7. GC-AED Analysis of a typical FCC gasoline.


Vol. 1 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES 487

The sulfur content of gasoline increases as gasoline cut point temperature is


increased, because of the inclusion of benzothiophenes.
Gasoline Sulfur Reduction Catalyst/Additive Technologies. Approaches
for developing GSR catalyst/additive technologies may include: (1) direct desul-
furization of thiophenes and thiophenols from the product gasoline, (2) adsorp-
tion of gasoline sulfur precursors; or (3) alkylation of gasoline sulfur
precursors into a higher boiling range fraction. The best alternatives are
obviously (1) and (2) since they do not transfer the sulfur to the other liquid pro-
ducts. In addition, it is desirable that the additive converts the sulfur compounds
into H2S and hydrocarbon products rather than to coke.
Sulfur reduction additives such as GSR-1, commercialized in 1995, and its
successor, D-PriSM, commercialized in 2001, have been shown to be most effec-
tive at reducing sulfur species in the light FCC naphtha and have been used in
>20 refineries worldwide (17). Based on solid Lewis acid (ZnO-based) technology,
D-PriSM has provided up to 35% sulfur reduction on a reduced endpoint FCC
gasoline. The availability of D-PriSM as an additive allows the inventory to be
base loaded for rapid sulfur reduction response and it can be used in conjunction
with any base FCC catalyst.
As an alternative to sulfur reduction additives, catalyst technologies such
as SuRCA and SATURN have been successfully commercialized (17). The
SuRCA catalysts are designed to completely replace the conventional FCC cata-
lyst in the circulating inventory, while maintaining or even enhancing existing
yields and selectivities. In addition to inherent cracking activity and selectivity,
SuRCA catalysts contain sulfur reduction functionality, which has provided
full-range gasoline sulfur reductions as high as 30% commercially. In contrast
to sulfur reduction additives, SuRCA is effective across the entire gasoline
boiling range. Additionally, reductions of 10–20% in LCO sulfur have been
observed in some applications.
SATURN catalyst technology incorporates a gasoline sulfur reduction func-
tionality that is effective over the entire gasoline boiling range. SATURN has
exhibited gasoline sulfur reductions as high as 50% in pilot plant testing. Reduc-
tions in LCO sulfur from 10 to 30% have also been observed. As with SuRCA, the
SATURN catalyst is custom-formulated with both cracking activity and sulfur
reduction functionality so that it can replace the conventional FCC catalyst in
the circulating inventory with minimal impact on yields and selectivities. The
magnitude of the sulfur reduction observed with this product is such that select
refiners may find that use of SATURN would allow them to comply with upcom-
ing clean fuels regulations while avoiding the significant capital expense of an
FCC feed or naphtha hydrotreater (17).
5.2. Gasoline Olefins Reduction. It is well known that the hydrogen-
transfer activity of the FCC catalyst can significantly affect gasoline olefins
levels. Typically, increasing the rare-earth content of the catalyst increases
hydrogen-transfer activity, thereby decreasing gasoline olefins. However, when
using conventional FCC catalysts, the decrease in gasoline olefins is often accom-
panied by large reductions in LPG olefins and gasoline research octane number
(RON), and poorer coke selectivity (22). The loss of valuable propylene,
butylenes, and octane barrels is often not economical, even for the gasoline
olefin- constrained refiner.
488 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES Vol. 1

RFG catalysts have been introduced as a solution to this problem. While


RFG catalysts are specifically designed with maximum hydrogen-transfer
activity to reduce FCC gasoline olefins, this is accomplished without the eco-
nomic penalties mentioned above. Gasoline olefins can be significantly reduced
[reductions up to 40% rel. (15% abs.) have been observed commercially] while
keeping propylene, butylenes, and gasoline RON constant. Coke selectivity is
also maintained, even at very high metals levels (22).
Following the early success in the Murphy-Meraux, LA refinery (23), RFG
catalysts are now in use in >10 FCC units around the world and continue to
show outstanding activity and stability in addition to olefins reduction in some
of the most severe resid applications. Equilibrium catalyst (Ecat) zeolite unit
cell sizes of 24.40 Å and higher have been maintained without sacrificing coke
selectivity in FCC units with Ecat Ni þ V levels of 8,000–12,000 ppm. A recent
example of the commercial performance of an RFG catalyst show a reduction in
gasoline olefins from 38 to 28 vol%, while maintaining LPG olefins, gasoline
RON, and coke selectivity, and increasing conversion all at a higher average
metals level (17).

6. Environmental Issues: Controlling Emissions from the FCC Unit

6.1. Reducing Particulate Emissions. During normal operation,


an average FCC unit may replace 1% of its catalyst inventory to make up
for particle attrition. The physical properties of FCC catalysts are designed
for optimum fluidization and low attrition. A trade-off exists between the
mechanical strength of the catalyst and the desire to limit erosion of FCC
hardware (24–26).
The nature of the binder is an important consideration. Commercial experi-
ence shows that alumina-sol bound catalysts have outstanding unit retention
characteristics compared with other catalyst systems as indicated by the data
in Figure 8 (8).
6.2. Reducing SOx Emissions. Sulfur oxides (generally referred to as
SOx) have been shown to be a leading cause of environmental damage, espe-
cially due to their contribution to the formation of acid rain, and their role in
the formation of aerosols. In the United States, the EPA New Source Perfor-
mance Standards (NSPS) enforcement initiative has led to voluntary agree-
ments between the EPA and refiners to reduce emissions (27). Known as
Consent Decrees, these agreements currently cover 45% of the industry and
typically require refiners to reduce FCC unit SOx emissions to <25 vppm. The
SOx is formed in the FCC regenerator from the combustion of sulfur containing
molecules bound in coke. Although only a small percentage of the feed sulfur
ends up in coke, typically <10%, all of the sulfur in coke is oxidized to SOx
(18). The concentration of SOx in the FCC flue gas can vary from <100 ppm
to several thousand ppm, depending on feed sulfur levels and speciation.
Although hydrotreated feeds typically produce lower SOx emissions, the per-
centage of the feed sulfur going to coke is usually higher than for virgin
feeds. Up to 30% of feed sulfur has been reported in the coke depending on the
nature of the feedstock (28).
Vol. 1 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES 489

Refinery B
60

55

50
Stack Opacity North, %

45
Si-Sol
Si-Sol Al-Sol
Al-Sol
40

35

30

25

20

15

10
–1 month 0 +1 month +3 months +5 months

Fig. 8. Al-Sol catalysts show outstanding unit retention in commercial operations (8).

FCC unit SOx control options include (1) feed hydrodesulfurization, (2)
‘end-of-pipe’ solutions such as flue gas scrubbing, and (3) the use of SOx-transfer
catalysts. Options (1) and (2) are capital intensive but do provide additional
benefits. The SOx-transfer catalysts or additives require little or no capital and
can be implemented quickly.
SOx Additives. SOx additives have been commercially available since
the early 1980s. These early additives were marginally effective at reducing
SOx. However, this changed with the introduction of DESOX. Its patented
spinel structure has proven to be the most effective material available for
SOx reduction. DESOX reduces (or more correctly transfers) SOx through a
three-step mechanism: (1) oxidation of SO2 to SO3, (2) capture of SO3 by pro-
prietary metal spinel, and (3) release of H2S in the FCC reactor (Fig. 9) (29).
DESOX technology has been used extensively throughout the refining
industry (>100 units worldwide) (27). Refiners have achieved SOx reductions
in excess of 90%, which has usually been accomplished at a cost of $500–
900/ton of SOx removed. There is no thermodynamic limit to the amount of
SOx that can be reduced using DESOX. Indeed commercial experience at several
locations shows that emissions can be <10 vppm.
Figure 10 shows the results of a test a refiner conducted to determine if
reaching 25 vppm SOx was feasible using a catalytic additive (30). Since the
only goal of the test was to determine if 25 vppm could be reached, the base
case was considered relatively unimportant, and was reduced to one day. Upon
base loading to 1% of inventory, the refiner saw SOx emissions drop to the 25 ppm
range, and was able to hold emissions at that level for an extended period with
DESOX additions of 3 wt% of fresh catalyst additions.
No yield debits have been reported with DESOX (30), even at addition rates
>10% of catalyst additions, and DESOX has not had any negative effects on equi-
librium catalyst properties (30).
490 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES Vol. 1

Fig. 9. Catalytic SOx reduction.

6.3. Reducing NOx Emissions. Environmental legislation aimed at


reducing the formation of ground-level ozone is forcing refiners to reduce their
emissions of NOx, a recognized ozone precursor. The FCCU regenerator is typi-
cally the single largest point source of NOx within a refinery and poses a very
challenging environment for controlling NOx. The NOx levels in the FCC regen-
erator flue gas are typically in the range of 100–500 ppm and may need to be
reduced to <20 vppm.

200

180

160

140

120

100

80

60

40

20

0
9/17/2001 9/19/2001 9/21/2001 9/23/2001 9/25/2001 9/27/2001 9/29/2001 10/1/2001

Fig. 10. DESOX is able to sustain SOx emissions below 25 vppm at refinery A . Average
SOx reduction is 85% with 71% of the SOx emissions below 25 ppm.
Vol. 1 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES 491

(Coke) N NO + CO, Coke N2

Pt Promoter
Pt Promoter

N2 HCN, NH 3, ...
CO2

Fig. 11. Mechanism of NOx formation.

A nitrogen balance around an FCC unit shows that  50% of the feed nitro-
gen ends up in the coke deposited on the catalyst (31). The burning of the nitro-
gen containing coke during the regeneration, generates primarily N2, but also
NO and occasionally small amounts of NO2 and N2O.
The chemistry of NOx formation has been studied in detail and a proposed
mechanism is summarized in Figure 11. According to this scheme, NO can be
reduced to N2 with the help of reductants such as coke, CO, and hydrocarbons.
Conventional combustion promoters (Pt/Al2O3) used to oxidize CO to CO2 have
been shown to increase NOx emissions. One likely mechanism is that they also
oxidize nitrogen intermediates (eg, NH3 and HCN). In 1996, a low-NOx combus-
tion promoter, XNOx, was introduced and shown to reduce NO emissions by
>50%. This NO reduction was achieved while maintaining good CO combustion
promotion. The results from the commercial use of a low-NOx promoter are
shown in Figure 12 (30). In this case, replacing conventional CO promoter
with XNOx resulted in a 60–70% NOx reduction while maintaining afterburn
and CO emissions control.
NOx reduction can also be achieved with NOx reducing additives that do not
have CO oxidation functionality. DENOX, was introduced commercially in 1997
and has been shown to decrease NOx by >50%.
In the United States, Consent Decree protocols typically call for replacing
conventional combustion promoters with low NOx promoters and conducting
optimization of NOx reducing additives at various levels. Figure 13 summarizes
Consent Decree test results with DENOX (30). In these trials, uncontrolled or
‘‘Base NOx’’ has been estimated using six sigma modeling techniques. Drivers
for NOx formation were found to be unit specific, with models containing between
4 and 17 input variables, including nonlinear and interaction terms. ‘‘Reduced
NOx’’ is the average NOx measured during the trial. In some instances, XNOx
additions started coincidentally with DENOX and it is not possible to separate
the relative contribution of each additive.
The amount of additive used in each trial varied greatly, as did the
results, with observed NOx reductions as high as 80%. The effectiveness
of an additive for controlling NOx emissions depends on unit design, mechan-
ical condition, operating parameters, feedstock, etc. The chemistry of NOx
formation and reduction is very complex. As a result the performance of NOx
additives is unit specific.
492 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES Vol. 1

120 50

Promoter Addition Rate (kg/D)


Pt-based Promoter 45
100
40
Usage Rate
80 35
FCC Design: UOP Riser, 30
565 ton inventory 60 25
20
40
15
62,000 BBL/D of HDT feed 20 10
5
0 0
Base emissions were
90-120 ppm with conventional 31-May 7-Jun 14-Jun 21-Jun 28-Jun 5-Jul 12-Jul
Pt-based promoter Date

160
>50% Reduction in Flue
140 Promoted Base

120
100
80
60
40
20
0
0 500 1000 1500 2000 2500 3000
CO Emissions (ppm)

Fig. 12. Commercial performance of XNOx.

300

50%
250

47%
ppm NO in flue gas

200

8%
150
17%
80%
67%
100
19%
0%
50

0
0.12 0.6 0.6 0.6 0.6 1.0 2.0 2.5
DENOX Addition Rate (% of fresh catalyst addition)

Fig. 13. Summary of consent decree test results with DENOX.


Vol. 1 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES 493

7. Production of Light Olefins in the FCCU

The next 2–3 years will see a significant growth in the demand for propylene
derivatives, particularly polypropylene, that will lead to structural tightness in
the global propylene market. Since steam cracker capacity expansion, which is
tied to a lower growth rate in the demand for ethylene, is forecast to lag pro-
pylene demand and much of the new planned capacity is in the Middle East
where light feedstocks will be processed, the petrochemicals industry is looking
to operators of FCC processes to fill the propylene supply gap (32).
In the short term, this will lead to opportunities for refiners to further
increase margins by using ZSM-5 based additive technologies to produce incre-
mental propylene for sale into the petrochemicals market. In the long term, a
number of refiners are considering either revamping their FCC units to operate
in ‘‘petrochemicals mode’’ or are considering investing in grass routes ‘‘on pur-
pose’’ propylene production facilities that utilize ZSM-5 based FCC catalyst
technologies.
7.1. ZSM-5 Technologies. The addition of ZSM-5 additives is the sin-
gle most important method to increase the yields of light olefins, particularly pro-
pylene, at the expense of gasoline (Fig. 14). Davison Catalysts was the first
catalyst manufacturer to supply ZSM-5 based additives for use in FCC units
back in 1984, where their primary role was to enhance gasoline octane. Since
then, the role of these additives has changed to that of optimizing the yields of
high value C3¼ and C4¼ olefins, while at the same time capturing the boost in
gasoline octane (33).
ZSM-5 additives utilize unique matrix technologies that are designed not
only to stabilize the ZSM-5 zeolite and provide the maximum activity per
pound of zeolite, but also to minimize unselective cracking that reduce the
yield of valuable light olefins. In addition, additives are designed with excellent

Fig. 14. ZSM-5 converts C6þ olefins to C5 olefins.


494 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES Vol. 1

(a) (b)

Propylene Selectivity (C3=/Conversion)


OlefinsMax OlefinsUltra OlefinsMax

Fresh Catalyst Addition Rate


Propylene Selectivity, (%)

ca.10wt% Reduction in
Fresh Catalyst Addition Rate

Propylene yield >14vol% (8wt%)


OlefinsUltra

70 ECat Microactivity 75
Time

Fig. 15. OlefinsUltra enabled a european refiner to reduce fresh catalyst addition rate
while maintaining high propylene yield.

attrition resistance to minimize losses (33). OlefinsUltra, for example, is formu-


lated with the highest crystal content allowable under a new license agreement.
This ultrahigh activity additive enables refiners to achieve high propylene yields
that would otherwise require high loadings of less active additives that lead to
dilution of the circulating catalyst inventory. In one commercial example, a
refiner was able to reduce fresh catalyst additions by 10% on switching to
OlefinsUltra (Fig. 15).
Most recently, Davison Catalysts’ APEX project focused on developing
catalysts that give yields of propylene in excess of 15 wt% from the FCC unit
(33). A new family of Propylene Maximizing Catalysts (PMC) has subsequently
been developed based on the IMPACT technology platform that will allow re-
finers to take propylene production in the FCC unit to a new level. By using
proprietary zeolite and matrix technologies, catalysts have been designed
that not only produce exceptional yields of propylene (Table 3) but also demon-
strate low coke make and good bottoms cracking activity in the presence of con-
taminant metals.
Propylene yields in excess of 15 wt% are of particular interest to refiners
that are considering revamping their FCC units to operate in a ‘‘petrochemicals
mode’’. This technology is currently in the commercialization stage, and a first
application in an FCC unit is expected to take place later this year.
7.2. Understanding Performance Differences: Phosphorous
Stabilization. One of the important variables that accounts for much of the per-
formance differences observed among ZSM-5 additives manufactured by different
suppliers is the stabilization of the zeolite with phosphorus (34). Figure 16 illus-
trates the effect of phosphorus stabilization for additives formulated with 25 wt%
ZSM-5 crystal. The phosphorus-stabilized sample retains much of its initial activ-
ity (indicated by comparable delta propylene yield) on steam deactivation com-
pared to the nonstabilized additive. Clearly, this represents the extremes in
terms of the effect of phosphorus stabilization on additive activity. The activity
Vol. 1 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES 495

Table 3. APEX Project: Propylene Maximization Catalyst Technology


Typical yields (wt%)
Conventional
Conventional FCC þ ZSM-5 APPMC APPMC APPMC
FCC Additive 140 150 200
feed type Heavy Heavy Heavy Heavy Light
Ni þ V/ppm 5000 5000 5000 5000 <100
reactor temperature (8C) 535 535 550 565 565
conversion (wt%) 80 80 80 85 85
propylene (wt%) 5 9 15 18 22
butylenes (wt%) 8 10 14 12 17
C5 þ gasoline (wt%) 52 46 36 32 28
LCO þ HCO (wt%) 20 20 20 15 15

level to which the additive equilibrates will depend on how the ZSM-5 zeolite was
stabilized during manufacture (34).
Nuclear magnetic resonance (nmr) spectroscopy analysis indicates that pho-
sphorus-stabilized ZSM-5 zeolite is less dealuminated after steam deactivation
than standard nonstabilized additives (34). This observation implies that the
higher activity/stability of the ZSM-5 in these additives is due to a retardation
of the dealumination rate caused by the phosphorus stabilization process.
Two independent experimental results substantiate this conclusion. Figure 17
shows the 27Al nmr spectra of (a) a fresh (heat-treated nonstabilized) ZSM-5
and two hydrothermally deactivated ZSM-5 samples; (b) nonstabilized ZSM-5;

3.5
Delta Propylene Yield, wt%

2.5

1.5

0.5

0
1200 °F 1500 F/4 hrs/100% Stm

Unstabilized Stabilized

Fig. 16. Activity of stabilized and unstabilized ZSM-5 crystal.


496 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES Vol. 1

27
Fig. 17. Al nmr spectra of stabilized and unstabilized ZSM-5 crystal.

and (c) P-stabilized. As can be seen, the phosphorus stabilized ZSM-5 retains a
significant amount on aluminum in the framework compared to the unstabilized
version as evidenced by the large tetrahedral Al peak. Again, the activity reten-
tion of the additive in the actual unit will depend on the stabilization process
carried out during manufacture.
Differences in phosphorus stabilization of the ZSM-5 zeolite account for
much of the performance differences between additives from different suppli-
ers. Surface area and percent ZSM-5 crystal content are not good indicators
of catalytic performance especially between additives from different suppliers.
ZSM-5 additive performance can vary substantially as a result of differences
in ZSM-5 zeolite crystallization and stabilization. This has been seen commer-
cially (34).

8. Conclusions

The FCC unit remains the primary hydrocarbon conversion unit in the modern
petroleum refinery. To further improve profitability, refiners are increasing the
amount of lower cost, residual feedstocks processed. These feeds are not only
heavier, but also contain high levels of contaminant metals compared with con-
ventional vacuum gas oils. Zeolite and matrix technologies in the FCC catalyst
have evolved to maximize conversion of these difficult feeds to valuable transpor-
tation fuels and light olefins.
Increasingly stringent environmental regulations have also changed the
objective function of the modern refinery. Special gasoline sulfur reduction addi-
tive and catalyst technologies are now being widely applied in FCC units to help
Vol. 1 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES 497

refiners comply with tighter gasoline specifications. Developments in additives


technologies have been made in response to environmental regulations on
emissions (eg, NOx, SOx). In addition, breakthroughs in the stabilization of
ZSM-5-based additives and technologies capable of double-digit propylene yields
open up opportunities for refiners to increase margins by participating in the
petrochemicals market.
Future catalyst developments in the area in FCC are likely to continue to
be driven by response to environmental regulation, changes in feed quality, and
advances in reactor hardware.

BIBLIOGRAPHY

CITED PUBLICATIONS

1. J. W. Wilson, Fluid Catalytic Cracking Technology and Operations, Penwell Publish-


ing, Oklohoma, 1977.
2. P. B. Venuto and E. T. Habib, Jr., Fluid Catalytic Cracking with Zeolite Catalysts,
Marcel Dekker, New York, 1979.
3. S. Gambicki, Stud. Surf. Sci. Catal. 130, 147 (2000).
4. T. J. Dougan, U. Alkemade, L. T. Boock, and B. Lakhanpal, Oil Gas J. 92(39) (1994).
5. X. Zhao, W. C. Cheng, J. A. Rudesill, ‘‘FCC Bottoms Cracking Mechanisms and
Implications for Catalyst Design for Resid Applications’’, NPRA Annual Meeting,
Paper AM-02-53, San Antonio, Tex., March 2002.
6. S. K. Purnell, ‘‘IMPACT: A Breakthrough Technology for Resid Processing’’, Catala-
gram No. 93, Davison Catalysts Publication, 2003.
7. J. T. Haley, T. F. Lim, C. Yap, S. K. Purnell, J. R. D. Nee, M. Lesemann, and T. G.
Roberie, ‘‘FCC Profit Maximization: New Catalyst Technologies for the Asia Pacific
Market, Petrochemical Feedstock Maximization and Capital Savings Options for
Clean Fuels’’, ARTC 7th Annual Conference, Singapore, April 2004.
8. S. K. Purnell, ‘‘A Comprehensive Approach to Catalyst Design for Resid Applica-
tions’’, NPRA Annual Meeting, Paper AM-03-126, San Antonio, Tex., March 2003.
9. R. F. Wormsbecher, W. C. Cheng, G. Kim, and R. H. Harding, Vanadium Mobility in,
Fluid Catalytic Cracking, in Proceedings of the 210th ACS National Meeting,
Chicago, Ill., 1995.
10. J. A. Rudesill and A. W. Peters, in M. Occelli and P. O’Connor, eds., The Effect of
Oxidation State of Vanadium on the Selectivity of Fluid Catalytic Cracking Cata-
lysts, Fluid Catalytic Cracking Technology, Marcel Dekker, New York, 1998.
11. T. F. Lim, C. Yap, K-H Chew, S. K. Purnell, and J. R. D. Nee, ‘‘Maximizing FCC Unit
Profitability Through Premium Catalyst Technologies’’, in Hydrocarbon Asia, Mar/
April 2004, p. 42.
12. G. Yaluris, W. C. Cheng, M. Peters, L. J. Hunt, and L. T. Boock, ‘‘The Effects of
Fe Poisoning on FCC Catalysts’’, NPRA Annual Meeting, Paper AM-01-59,
New Orleans, March 2001.
13. U.S. Pat. 4,909,993 (1990) J. H. Haddad, H. Owen, and K. W. Schatz.
14. M. W. Schnaith, A. Sexson, D. True, D. B. Bartholic, Y. K. Lee, I. S. Yoo, and H. S.
Kang, Oil Gas J. 96(25), 53 (1998).
15. P. A. Diddams and H. Haave, Grace Davison FCC Technology Conference, Lisbon,
1998.
498 FLUID CATALYTIC CRACKING, CATALYSTS AND ADDITIVES Vol. 1

16. Auto/Oil Air Quality Improvement Research Program, Phase I Final Report, May,
1993.
17. S. K. Purnell, D. A. Hunt, and D. Leach, ‘‘Catalytic Reduction of Sulfur and Olefins
in the FCCU’’, NPRA Annual Meeting, Paper AM-02-37, San Antonio, Tex., March
2002.
18. R. F. Wormsbecher, G. D. Weatherbee, G. Kim, and T. J. Dougan, ‘‘Emerging Tech-
nology for the Reduction of Sulfur in FCC Fuels’’, NPRA Annual Meeting, Paper
AM-93-55, San Antonio, March 1993.
19. E. G. Wollaston, W. L. Forsythe, and I. A. Vasalos, Oil Gas J. 64 (Aug. 2, 1971).
20. G. P. Huling, J. P. McKinney, and T. C. Readal, Oil Gas J. 73 (May 19, 1975).
21. T. A. Nguyen and M. Skripek, AIChE Spring National Meeting, Atlanta, April,
1994.
22. R. W. Mott, T. G. Roberie, and X. Zhao, ‘‘Suppressing FCC Gasoline Olefinicity while
Managing Light Olefins Production’’, NPRA Annual Meeting, Paper AM-98-11, San
Francisco, March 1998.
23. J. W. Balko, D. Podratz, and J. A. Olesen, ‘‘Grace Davison FCC Catalytic Technolo-
gies for Managing New Fuels Regulations’’, NPRA Annual Meeting, Paper AM-00-14,
San Antonio, Tex., March 2000.
24. Petroleum Technol. Q. 3 (1) 39 (1998).
25. F. J. A. Geiger, H. W. A. Dries, and E. Bakker, Erosion, Fouling and Cracking of FCC
Cyclone Systems Resolved, VDI Berichte No. 1511, 1999.
26. H. von Uetz and K.-J. Grob, Stromungsmechanische Grundlagen der Erosion,
Literatur S. 235, pp. 228–235.
27. Davison KBR conference paper.
28. J. G. McDaniel and D. J. Neuman, NPRA Annual Meeting, Paper AM-92-60,
New Orleans, March 1992.
29. S. W. Davey, Environmental Fluid Catalytic Cracking Technology, in Proceedings of
the European Refining Technology Conference, Paris, France, Nov. 1999.
30. T. J. Dougan and J. R. Riley, ‘‘SOx, NOx and Consent Decrees; The FCCU Additive
Perspective’’, NPRA Environmental Conference, Paper ENV-02-198, New Orleans,
September 2002.
31. A. W. Peters, G. D. Weatherbee, and X. Zhao, Fuel Ref. 5(3), 45 (1995).
32. S. J. Zinger, ‘‘Propylene Supply for the Global Petrochemical Markets’’, Catalagram
No. 94, Special Edition Propylene, Davison Catalysts Publication, 2004.
33. J. R. D. Nee and M. Lesemann, ‘‘Propylene Production in the FCC Unit’’, Catalagram
No. 94, Special Edition Propylene, Davison Catalysts Publication, 2004.
34. M. Lesemann, J. R. D. Nee, and S. J. Zinger, ‘‘The Evolving Role of Fluid Catalytic
Cracking in the Petrochemicals Market’’, NPRA Annual Meeting, Paper AM-04-58,
San Antonio, Tex., March 2004.

J. R. D. NEE
R. H. HARDING
G. YALURIS
W. C. CHENG
X. ZHAO
T. J. DOUGAN
J. R. RILEY
W. R. Grace & Co.
Vol. 1 CATALYSIS 499

CATALYSIS
1. Introduction

Homogeneous catalysis, 324 Heterogeneous catalysis, 340 Catalysis, is the key


to efficient chemical processing. Most industrial reactions and almost all biologi-
cal reactions are catalytic. The value of the products made in the United States
in processes that at some stage involve catalysis is approaching several trillion
dollars annually, which is more than the gross national products of all but a few
nations of the world. Products made with catalysis include food, clothing, drugs,
plastics, detergents, and fuels. Catalysis is central to technologies for environ-
mental protection by conversion of emissions.
This article is an introduction and survey that states the fundamental prin-
ciples and definitions of catalysis, demonstrates the unity of the subject, and
places it in an applied perspective. The selection of industrial catalytic processes
discussed has been made for the sake of illustrating principles and representa-
tive characteristics of catalysis and catalytic processes.
A catalyst is a substance that increases the rate of approach to equilibrium
of a chemical reaction without being substantially consumed itself. A catalyst
changes the rate but not the equilibrium of the reaction. This definition is almost
the same as that given by Ostwald in 1895. The term catalysis was coined in
1835 by Berzelius, who recognized that many seemingly disparate phenomena
could be described by a single concept. For example, ferments added in small
amounts were known to make possible the conversion of plant materials into
alcohol; and there were numerous examples of both decomposition and synthesis
reactions that were apparently caused by addition of various liquids or by contact
with various solids.
Berzelius attributed catalytic action to ill-defined forces, and the value of
Ostwald’s more lasting definition is that it identified catalysis as a phenomenon
that was consistent with the emerging principles of physical chemistry. Now it is
well recognized that catalysts function by forming chemical bonds with one or
more reactants, thereby opening up pathways to their conversion into products
with regeneration of the catalyst. Catalysis is thus cyclic; reactants bond to one
form of the catalyst, products are decoupled from another form, and the initial
form is regenerated. The simplest imaginable catalytic cycle is therefore depicted
as follows:
R

C RC

where R is the reactant, P the product, C the catalyst, and RC an intermediate


complex. The intermediate complexes in catalysis are often highly reactive and
not observable.

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.0301200107012005.a01.pub2
500 CATALYSIS Vol. 1

Ideally, the catalyst would cycle forever between C and RC without


being consumed. But in reality there are competing reactions, and catalysts
are converted into species that are no longer catalysts. In practice, catalysts
must be regenerated and replaced. Catalyst manufacture is a large industry;
catalysts worth some several billion dollars are sold annually in the United
States.
Catalysts may be gases, liquids, or solids. Most catalysts used in techno-
logy are either liquids or surfaces of solids. Catalysis occurring in a single gas
or liquid phase is referred to as homogeneous catalysis (or molecular catalysis)
because of the uniformity of the phase in which it occurs. Catalysis occurring in
a multiphase mixture such as a gas–solid mixture is referred to as hetero-
geneous catalysis; usually this is surface catalysis. Biological catalysts are
proteins, ie, poly(amino acids), called enzymes, and metalloenzymes, which
are proteins incorporating inorganic components, eg, iron sulfide clusters.
Some enzymes are present in cytoplasmic solution in cells and others are
anchored in cell membranes or on surfaces. Traditionally, homogeneous cataly-
sis, heterogeneous catalysis, and enzymic catalysis have been regarded as
almost separate disciplines, and the language and literature have developed
without much coherence and overlap.
The performance of a catalyst is measured largely by criteria of chemical
kinetics, as a catalyst influences the rate and not the equilibrium of a reaction.
The catalytic activity is a property of a catalyst that measures how fast a cata-
lytic reaction takes place and may be defined as the rate of the catalytic reaction,
a rate constant, or a conversion (or temperature required for a particular conver-
sion) under specified conditions. The selectivity is a measure of the property of a
catalyst to direct a reaction to particular products. There is no single definition of
selectivity, but it is sometimes defined as a ratio of activities, such as the ratio of
the rate of a desired reaction to the sum of the rates of all the reactions that
deplete the reactants. Selectivity may also be represented simply as a product
distribution. Because catalysts typically lose activity and/or selectivity during
operation, they are also evaluated in terms of stability and lifetime. The stability
of a catalyst is a measure of the rate of loss of activity or selectivity. In practical
terms the stability might be measured as a rate of deactivation, such as the rate
of change of the rate of the desired catalytic reaction or as the rate at which the
temperature of the catalyst would have to be raised to compensate for the activity
loss. Catalysts that have lost activity are often treated to bring back the activity,
ie, reactivated; the regenerability is a measure, often not precisely defined, of
how well the activity can be brought back. Technological catalysts are also eval-
uated in terms of cost; in a typical process, the cost of the catalyst is a small frac-
tion of the total processing cost, often only a few cents or less per kilogram of
product.
Liquid-phase catalysts are used in batch and flow reactors. Batch reactors
are predominant in small-scale operations and those designed for flexibility such
as pharmaceutical and fine chemicals manufacture. Flow reactors predominate
in large-scale processes such as commodity chemicals manufacture and petro-
leum refining. Corrosion, separation, and catalyst recovery and recycle are
usually important issues in these processes. Solid catalysts are used in batch
and flow reactors; in most large-scale operations, solid particles of catalyst are
Vol. 1 CATALYSIS 501

present in fixed- or fluidized-bed reactors with gaseous or liquid reactants flow-


ing through them.
The solids used as catalysts are typically robust porous materials with high
internal surface areas, typically, hundreds of square meters per gram. Reaction
occurs on the internal catalyst surface. The typical solid catalyst used in industry
is a composite material with numerous components and a complex structure.
Catalytic processes are classified roughly according to the nature of the pro-
duct and the industry of application, and to some degree separate literatures
have developed. In chemicals manufacture, catalysis is used to make heavy che-
micals, eg, ammonia and sulfuric acid; commodity chemicals, eg, acetic acid and
n-butyraldehyde; and fine chemicals, eg, fragrances and flavorings. Catalysis is
used extensively in the manufacture of pharmaceuticals. In fuels processing,
catalysis is used in almost all the processes of petroleum refining and in coal
conversion and related synthesis gas (CO and H2) conversion. Most of the recent
large-scale developments in industrial catalysis have been motivated by the
need for environmental protection. Many processes for abatement of emissions
are catalytic; automobile exhaust conversion catalysts are now the most impor-
tant of all in terms of sales volume, and catalysts for conversion of nitrogen
oxides in stationary power plant effluents are used on a massive scale in
Japan and parts of Europe. Recent innovations in pharmaceutical manufacture
involve catalysis (usually homogeneous) for asymmetric synthesis. Most of the
applications of catalysis in biotechnology are fermentations, often carried out
in stirred reactors with gases, liquids, and solids present; the catalysts are
enzymes present in living organisms such as yeasts. There are applications of
whole biological cells and of individual enzymes mounted on supports, ie, car-
riers, and used in fixed-bed reactors.

2. History

The science of catalysis is driven by technology, as it has been from the begin-
ning. Some of the earliest known examples of controlled chemical transforma-
tions are catalytic. Fermentation was used in ancient times to make alcoholic
beverages, and a number of the earliest examples of chemical technology were
also exploitations of catalysis (1). For example, before the sixteenth century
ether was made by distilling spirits in the presence of sulfuric acid. In 1746,
nitric oxide was used as a catalyst in the lead chamber process for oxidation of
sulfur dioxide to give sulfur trioxide in the manufacture of sulfuric acid. In 1781,
acids were used to catalyze the conversion of starch into sugar. In 1817, H. Davy
discovered that in the presence of platinum, mine gases were oxidized at low
temperatures; he designed a safety lamp for miners in which the platinum
glowed if the flame was extinguished (2). Many more practical discoveries
were made in the 1800s (1,2).
These examples existed prior to Ostwald’s 1895 definition (1), ie, before the
nature of catalysis was well understood. But in 1850 Wilhelmy (3) had already
made the first measurements of kinetics of catalytic reactions in an investigation
of sugar inversion catalyzed by mineral acids. In the years following Ostwald’s
definition, just as the principles of chemical equilibrium and kinetics were
502 CATALYSIS Vol. 1

becoming known, the field of catalysis became more quantitative and developed
rapidly. Kinetics of surface-catalyzed reactions were measured by Bodenstein
(4) just after the turn of the century. The defining work that set the stage for
modern catalytic technology was the development of the ammonia synthesis
process by Haber, Bosch, Mittasch, and co-workers, beginning 1908 (1,5).
These pioneers worked out methods of catalyst testing and process develop-
ment that are essentially the methods of choice today (6,7). They understood the
interplay between chemical equilibrium and reaction kinetics; indeed, Haber’s
research, motivated by the development of a commercial process, helped to
spur the development of the principles of physical chemistry that account for
the effects of temperature and pressure on chemical equilibrium and kinetics.
The ammonia synthesis reaction is strongly equilibrium limited. The equilibrium
conversion to ammonia is favored by high pressure and low temperature. Haber
therefore recognized that the key to a successful process for making ammonia
from hydrogen and nitrogen was a catalyst with a high activity to allow operation
at temperatures low enough for a relatively favorable equilibrium.
Bosch and co-workers (6) devised laboratory reactors to operate at high
pressure and temperature in a recycle mode. These test reactors had the essen-
tial characteristics of potential industrial reactors and were used by Mittasch
and co-workers (6) at BASF in Germany to screen some 20,000 samples as can-
didate catalysts. The results led to the identification of an iron-containing
mineral that is similar to catalysts used in industry today. The researchers
recognized the need for porous catalytic materials and materials with more
than one component, today identified as the support, the catalytically active com-
ponent, and the promoter. Today’s technology for catalyst testing has become
more efficient because much of the test equipment is automated and some is
miniaturized. The analysis of products and catalysts is also much faster and
more accurate today.

3. Homogeneous Catalysis

3.1. Characterization of Solution Processes. There are many impor-


tant examples of catalysis in the liquid phase, but catalysis in the gas phase is
unusual. From an engineering viewpoint, most of the liquid-phase processes
have the following characteristics in common.
Pressure and Temperature. The pressure and temperature are rela-
tively low, typically less than 2 MPa (20 atm) and 1508C. High pressures
are used in some applications, but they require expensive thick-walled
reactors. High temperatures are largely avoided because they result in high
autogenous pressures and because many of the catalysts are unstable at high
temperatures.
Corrosiveness. The catalyst solutions are corrosive, and the reactors,
separation devices, etc, that come in contact with them must be made of expen-
sive corrosion-resistant materials.
Separation Processes. Separation of the catalyst from the products is
expensive; the process flow diagram and the processing cost are often domi-
nated by the separations. Many soluble catalysts are expensive, eg, rhodium
Vol. 1 CATALYSIS 503

complexes, and must be recovered and recycled with high efficiency. The most
common separation devices are distillation columns; extraction is also applied.
Gas Handling. The reactants are often gaseous under ambient condi-
tions. To maximize the rate of the catalytic reaction, it is often necessary to mini-
mize the resistance to gas–liquid mass transfer, and the gases are therefore
introduced into the liquid containing the catalyst as swarms of bubbles into a
well-stirred mixture or into devices such as packed columns that facilitate
gas–liquid mixing and gas absorption.
Exothermicity. The catalytic reactions are often exothermic bond-forming
reactions of small molecules that give larger molecules. Consequently, the reac-
tors are designed for efficient heat removal. They may be jacketed or contain coils
for heat-transfer media, or the heat of reaction may be used to vaporize the pro-
ducts and aid in the downstream separation by distillation.
There are also a number of generalizations about the chemistry of these
processes. Often the reactants are small building blocks, many formed from
organic raw materials, namely, petroleum, natural gas, and coal. The reactants
include O2, low molecular weight olefins, and synthesis gas (CO and H2). In older
technology, acetylene was a common building block. Many reactions are cata-
lyzed by acids and bases, usually in aqueous solution. Many reactions are cata-
lyzed by transition-metal complexes, usually in nonaqueous organic solvents, but
sometimes in water and in ionic liquids with low melting points such as alkylam-
monium salts. The transition metal complex catalysts used in technology are
often highly selective.
3.2. Influence of Mass Transport on Reaction Rates. When a rela-
tively slow catalytic reaction takes place in a stirred solution, the reactants are
supplied to the catalyst from the immediately neighboring solution so readily
that virtually no concentration gradients exist. The intrinsic chemical kinetics
determines the rate of the reaction. However, when the intrinsic rate of the reac-
tion is relatively high and/or the transport of the reactant is relatively slow, as in
a viscous polymer solution, the concentration gradients become significant, and
the transport of reactants to the catalyst cannot keep the catalyst supplied suffi-
ciently for the rate of the reaction to be that corresponding to the intrinsic che-
mical kinetics. Assume that the transport of the reactant in solution is described
by Fick’s law of diffusion with a diffusion coefficient D, and the intrinsic chemical
kinetics is of the following form

r ¼ k½C½A ð1Þ

where C is catalyst and A the reactant. Then in the general case in which the
diffusion influences the rate (8),

1
Z ¼ r=rmax ¼ ð2Þ
1 þ RAkþRC 
4p 2 D

where Z, the ratio of the reaction rate to the maximum value that could occur, ie,
the rate in the absence of a transport influence, is called the effectiveness factor,
and RC and RA are the radii of the catalyst and reactant molecules, respectively.
504 CATALYSIS Vol. 1

There are two limiting cases; if k  4pðRA þ2 RB ÞD, then there is no diffusion influ-
ence and the rate is described by equation 1, where the concentrations are the
bulk average concentrations in the solution. On the other hand, when the
inequality is inverted, the rate becomes
 
RA þ R B
r ¼ 4p D½C½A ð3Þ
2

In the former case, the rate is independent of the diffusion coefficient and is
determined by the intrinsic chemical kinetics; in the latter case, the rate is inde-
pendent of the rate constant k and depends on the diffusion coefficient; the reac-
tion is then diffusion controlled. This kind of mass transport influence is different
from that characteristic of dissolution of a gas into a liquid phase.
3.3. Examples of Solution Catalysis. Acid-Base Catalysis. Inex-
pensive mineral acids, eg, H2SO4, and bases, eg, KOH, in aqueous solution
are widely applied as catalysts in industrial organic synthesis. Catalytic reac-
tions include esterifications, hydrations, dehydrations, and condensations.
Much of the technology is old and well established, and the chemistry is well
understood. Reactions that are catalyzed by acids are also typically catalyzed
by bases. In some instances, the kinetics of the reaction has a form such as the
following (9):

r ¼ ðk0 ½H2 O þ kHþ ½Hþ  þ kHA ½HA þ kOH ½OH  þ kB ½BÞ½R ð4Þ

where r is the reaction rate, [R] the concentration of the reactant, and the terms
on the right-hand side represent catalysis by water, hydrogen ions, undisso-
ciated acid HA, hydroxide ions, and undissociated base. Often all the terms
on the right-hand side but one are negligible.
If the second term on the right-hand side in equation 4 is the only signifi-
cant one, then catalysis is by hydrogen ions (hydrated protons), and the catalysis
is called specific acid catalysis; the rate of the reaction is proportional to the con-
centration of the hydrogen ions and to the concentration of the reactant. Many
examples conform to this pattern. Similarly, if the third term on the right-hand
side of equation 4 is the only significant one, then the catalysis is by HA, and
general acid catalysis occurs. If the fourth term is the significant one, hydroxide
ions are the catalyst and specific base catalysis occurs. If the fifth term is the
significant one, general base catalysis occurs.
The chemistry of acid–base catalysis involves the transfer of hydrogen ions,
Hþ and sometimes H . In general or specific acid catalysis, the undissociated
acid molecules or the hydrated protons formed by dissociation of an acid, respec-
tively, donate protons to the reactant. The proton acceptor in an aqueous solution
is a good base, often an organic compound including O, N, or S. In catalysis by
bases, the reactant donates a proton to the catalyst. The proton donor in general
acid catalysis may be a weak or a strong acid, and often the activity of the cata-
lyst increases with its strength as an acid, measured by the acid dissociation
constant Ka.
Donation of a proton to the reactant often forms a carbenium ion or an oxo-
nium ion, which then reacts in the catalytic cycle. For example, a catalytic cycle
Vol. 1 CATALYSIS 505

suggested for the conversion of phenol and acetone into bisphenol A, which is an
important monomer used to manufacture epoxy resins and polycarbonates, in an
aqueous mineral acid solution is shown in Figure 1 (10).
The kinetics of the bisphenol A synthesis has been reported to be of the fol-
lowing form (11)
 
r ¼ k Hþ ½A½P ð5Þ

CH3
O
HO C OH
H3C C CH3
CH3

H OH OH
H3C C CH3 H3C C CH3

2 1
OH

CH3
OH HO C OH
CH3 H

CH3
H + HO C OH
CH3
CH3
C OH
CH3

H CH3
O C OH
H CH3

H2O

Fig. 1. Catalytic cycle for synthesis of bisphenol A from phenol and acetone in the
presence of a dissociated mineral acid (10).
506 CATALYSIS Vol. 1

where A is acetone and P is phenol. When the reaction is carried out in the pre-
sence of low concentrations of an added compound, thioglycolic acid,
HSCH2COOH, the rate is increased, with the kinetics reported to be of the fol-
lowing form (12)

 
r ¼ k0 Hþ ½A½Pðk1 þ k2 ½TÞ ð6Þ

where T is thioglycolic acid, a promoter. A catalyst promoter is a substance


that, although it lacks significant catalytic activity itself, increases the activity,
or possibly the selectivity, of a catalyst. Promoters are common in catalysis.
In some processes, the reactant bases are too weak to be protonated signif-
icantly except in the presence of very strong acids such as fuming sulfuric acid or
a mixture of concentrated sulfuric and nitric acids, ie, mixed acid.
The kinetics of reactions catalyzed by very strong acids are often compli-
cated. The exact nature of the proton donor species is often not known, and typi-
cally the rate of the catalytic reaction does not have a simple dependence on the
total concentration of the acid. However, sometimes there is a simple dependence
of the catalytic reaction rate on some empirical measure of the acid strength of
the solution, such as the Hammett acidity function H0, which is a measure of the
tendency of the solution to donate a proton to a neutral base. Sometimes the
rate is proportional to h0 (H0 ¼ log h0). Such a dependence may be expected
when the slow step in the catalytic cycle is the donation of a proton by the
solution to a neutral reactant, ie, base; but it is not easy to predict when such
a dependence may be found.
An important petroleum refining process, alkylation of propylene or
butenes with isobutane to give high octane-number branched products, is cata-
lyzed by concentrated sulfuric acid or hydrofluoric acid. The reactions are carried
out at subambient temperatures, at which the equilibrium is favorable, in
stirred, refrigerated reactors containing two-phase liquid mixtures. Most of the
hydrocarbon reactants are in the organic phase, and, when the catalyst is H2SO4,
most of the catalyst is in the aqueous phase; the reaction takes place near the
liquid–liquid interface.
In the alkylation process, the reactants are weak bases. The stronger
bases in the mixture, the olefins, are protonated, forming carbenium ions.
The carbenium ions are good Lewis acids, and they react with isobutane, a
good hydride ion donor, to form paraffins and the relatively stable tert-butyl
cation, ie, trimethyl carbenium ion. This is a hydride transfer reaction; such
reactions are important in many acid-catalyzed hydrocarbon conversions.
The tert-butyl cation becomes the predominant carbenium ion in the organic
solution and reacts with propylene or butenes in Lewis acid– Lewis base reac-
tions to form carbon –carbon bonds, giving C7 or C8 products among others.
There are many C7 and C8 isomers formed because isomerizations are rapid.
A partial cycle that illustrates the important characteristics of alkylation
is shown in Figure 2 for the ethylene– isobutane alkylation. The cycle is
shown for ethylene because it gives a simpler product distribution than propy-
lene or butenes; ethylene is much less reactive than propylene or butenes
because on protonation it forms a primary carbenium ion, which is much
Vol. 1 CATALYSIS 507

C C
C
C C
C

C
C C C C
C

C
C C C C
C C
C C
C

Fig. 2. Catalytic cycle for the ethylene–isobutane alkylation (10).

less stable than the secondary carbenium ion formed by protonation of propy-
lene or straight-chain butenes. Ethylene is not used in the commercial alkyla-
tion processes.
Acid-catalyzed reactions of paraffins require much stronger acids than
those of olefins or aromatics, because the paraffins are much weaker
bases. The only acids that catalyze reactions of paraffins alone at relatively
low temperatures are combinations of Brønsted and Lewis acids; examples
include HCl þ AlCl3, HF þ SbF5, and FSO3H þ SbF5 (magic acid). Some of
these are called superacids (13), eg, magic acid. Superacids are stronger acids
than concentrated sulfuric acid and some have values of H0 of 20 and less,
whereas that of 100% sulfuric acid is only about 12. The Hammett acidity
function is logarithmic; thus some superacids are more than eight orders of
magnitude stronger proton donors than sulfuric acid. Hence, these superacids
can protonate paraffins, even methane. They catalyze isomerization and
carbon–carbon bond-forming and bond-breaking reactions at low tempera-
tures but are not found in commercial applications, in part because of their
extreme corrosiveness.
Metal Complex Catalysis. Most of the recent innovations in industrial
homogeneous catalysis have resulted from discoveries of transition-metal com-
plex catalysts. Thousands of transition-metal complexes (including organometal-
lics, ie, those with metal–carbon bonds), are known, and the rapid development
of organotransition metal chemistry in recent decades has been motivated
largely by the successes and opportunities in catalysis. The chemistry of metal
complex catalysis is explained by the bonding and reactivity of organic groups
(ligands) bonded to the metals. The important reactions in catalytic cycles are
those of ligands bonded in the coordination sphere of the same metal atom. Bond-
ing of ligands such as CO or olefin to a transition metal activates them and facil-
itates the catalysis. The following reactions of transition-metal complexes
account for almost all the steps in the catalytic cycles.
508 CATALYSIS Vol. 1

Ligand exchange:

MLx þ L0 Ð MLx 1 0 þ L ð7Þ

where M is metal and L and L0 are ligands. This reaction may be thermally or
photochemically initiated; it may proceed by dissociation of a ligand to give a
coordinatively unsaturated complex, to which another ligand is then bonded,
but it may also proceed associatively, without prior dissociation of a ligand.
Ligand exchange is one way in which reactants become bonded to catalysts.
Oxidative addition and reductive elimination:

oxidative addition
MLx–2 + AB MLx–2 AB ð8Þ
reductive elimination

The forward reaction requires an electron-rich metal in a low oxidation state that
is coordinatively unsaturated. Oxidative addition provides another way for bond-
ing of reactant ligands to a metal, and reductive elminination provides a way for
products to leave a metal.
Migratory Insertion:
O
C O ð9Þ
Lx– 2M CH3 Lx– 2MC
CH3

In this reaction, one ligand migrates to a neighboring ligand that then becomes
formally inserted between the metal and the ligand that has migrated, creating a
site of coordinative unsaturation so that another reactant ligand can be asso-
ciated with the metal. The migratory insertion reaction accounts for chain-
growth steps of olefin polymerization reactions.
Wilkinson Hydrogenation. One of the best understood catalytic cycles is
that for olefin hydrogenation in the presence of phosphine complexes of rhodium,
the hydrogenation named after the Nobel-laureate Geoffroy Wilkinson (14,15).
The reactions of a number of olefins, eg, cyclohexene and styrene, are rapid, tak-
ing place even at room temperature and atmospheric pressure; but the reaction
of ethylene is extremely slow. Complexes of a number of transition metals in
addition to rhodium are active for the reaction.
The Wilkinson hydrogenation cycle shown in Figure 3 (16) was worked out
in experiments that included isolation and identification of individual rhodium
complexes, measurements of equilibria of individual steps, determination of
rates of individual steps under conditions of stoichiometric reaction with certain
reactants missing so that the catalytic cycle could not occur, and determination
of rates of the overall catalytic reaction. The cycle demonstrates some generally
important points about catalysis: the predominant species present in the react-
ing solution and the only ones that are easily observable by spectroscopic
methods, eg, RhCl[P(C6H5)3]3, RhCl[P(C6H5)3]2 (olefin), and RhCl2[P(C6H5)3]4,
are outside the cycle, possibly in virtual equilibrium with species in the cycle,
and not involved in the kinetically significant steps of the cycle. Often the only
C PPh3, C C H
Ph3P Ph3P PPh3 H2, k2 Ph3P H
C
Rh Rh Rh
H2, k– 2
Cl PPh3 C C PPh3 Cl PPh3 Cl PPh3
, PPh3
K5 C C PPh3
,
PPh3 PPh3 PPh3
PPh3
C C C C k1 k– 1 K4

Ph3P Cl PPh3
H C C H H
Rh Rh Ph3P H2 , k3 Ph3P H Ph3P
Ph3P PPh3 Rh Rh Rh
Cl H2 , k– 3
Cl PPh3 Cl PPh3 C C Cl PPh3

509
K7
H2 H2
C C
Fast
PPh3 K6
Ph3P Cl H C C

Rh Rh H H
H C C H
Ph3P
Ph3P H
Cl Rh
PPh3
Cl PPh3

Fig. 3. Catalytic cycle (within dashed lines) for the Wilkinson hydrogenation of olefin (14,16). Ph represents phenyl (C6H5). Values of rate
constants and equilibrium constants are as follows: k1 ¼ 0.68 s 1; k 1 7 104 L/(mols); K1  10 5 mol/L; k 1/k4 ffi 1; K5 ¼ 3 10 4; k2 ¼
4.8 L/(mols); k 2 ¼ 2.8 10 4 s 1; K2 ¼ 1.7 104 L/mol; k3 7 104 L/(mols); k6 ¼ 0.22 s 1.
510 CATALYSIS Vol. 1

species that can be observed during catalysis are those that are not involved in
the cycle. Consequently, elucidation of catalytic cycles is difficult, and inferences
about cycles that are based on easily measured compositions of reacting solutions
are usually less than well founded. The exceptions to this rule are usually the
cycles of least interest, those of slow catalytic reactions.
The Wilkinson hydrogenation illustrates the roles of the oxidative addition,
reductive elimination, and insertion reactions, all of which occur in the cycle
(Fig. 3). Some of the properties of the rhodium phosphine complexes that explain
their success as catalysts include the following: Rhodium exists in two stable oxi-
dation states, Rh(I) and Rh(III), separated by two units. Oxidative addition and
reductive elimination reactions formally require changes of two units in metal
oxidation state. There are no intermediates that are so stable that they form bot-
tlenecks in the cycle. There is an exquisite dynamic balance, and the cycle turns
over rapidly with only low concentrations of the intermediates being present in a
steady state. The phosphines are electron-donor ligands that are important in
controlling the reactivities of the intermediates and regulating the dynamic bal-
ance. Replacement of triphenylphosphine by other phosphines leads to signifi-
cant changes in the rate of the catalytic hydrogenation. The phosphines can
influence the reactivities of the intermediates, and thus the rate of catalysis,
by virtue of both electronic and steric effects.
Asymmetric Hydrogenation. Biological reactions are stereoselective, and
numerous drugs must be pure optical isomers. Metal complex catalysts have
been found that give very high yields of enantiomerically pure products, and
some have industrial application (17,18). The hydrogenation of the methyl
ester of acetamidocinnamic acid has been carried out to give a precusor of
L-dopa, ie, 3,4-dihydroxyphenylalanine, a drug used in the treatment of
Parkinson’s disease.

H COOCH3 H2
C C CH2CHCOOCH3
NHCOCH3 NHCOCH3

The strategy of the catalyst development was to use a rhodium complex similar
to those of the Wilkinson hydrogenation but containing bulky chiral ligands in an
attempt to direct the stereochemistry of the catalytic reaction to favor the desired
L isomer of the product (17). Active and stereoselective catalysts have been found
and used in commercial practice (although there is now a more economical route
to L-dopa than through hydrogenation of the prochiral precursor). The 2001
Nobel Prize in Chemistry was awarded to W. S. Knowles and R. Noyori, recogniz-
ing their work on these catalysts (and to K. B. Sharpless for work on asymmetric
oxidation catalysis).
Two pathways were found for the chiral hydrogenation, and they give pro-
ducts with different stereochemistries (19). One pathway involves the preferred
mode of initial binding of the reactant to the catalyst. The other pathway
involves an isomer of the reactant–catalyst complex that is formed in only
small amounts, but its conversion is energetically favorable and constitutes the
kinetically predominant pathway to products (9) (Fig. 4). Thus the chirality of
Vol. 1 CATALYSIS 511


CO2CH3
HN ‡ ‡
G°major
G°minor
P Minor
Rh
CH3 OPh *
P

Energy

CH3OOC NH
P Ph CH3OOC
* Rh CH3 NH
P O Major H P
*
Rh Ph
CH3
O
P H

CO2CH3
HN P H
*
Ph Rh
CH3
O
H P

Reaction coordinate

Fig. 4. Schematic representation of energy profiles for the pathways for the hydrogena-
tion of a prochiral precursor to make L-dopa (19). The chiral disphosphine ligand is shown
schematically as P * P. Ph represents phenyl (C6H5).

the product is determined not by the preferred mode of the initial binding, but
instead by the more favorable energetics of the pathway involving the minor
isomer of the reactant–catalyst complex.
Methanol Carbonylation. An important industrial process catalyzed by
rhodium (and more recently iridium) complexes in solution is methanol carbony-
lation to give acetic acid.

CH3 OH þ CO ! CH3 COOH ð10Þ

The catalytic cycle (Fig. 5) (20,21) is well established. The CH3I works as
a cocatalyst or promoter because it undergoes an oxidative addition with
[Rh(CO)2I2] , and the resulting product has the CO ligand bonded cis to the
CH3 ligand; these two ligands are then poised for a migratory insertion reac-
tion. When the catalyst is iridium instead of rhodium, a similar cycle pertains,
but the reaction of [H3CIr(CO)2I3] with CO to give [H3CCOIr(CO)2I3] is rate
determining (21). The migratory insertion in the iridium complex can be accel-
erated by the addition of either methanol or a Lewis acid (SnI2); each appears to
facilitate substitution of an iodide ligand by CO, resulting in faster methyl
migration. The greater stability of [H3CIr(CO)2I3] than of [H3CRh(CO)2I3]
accounts for the different characters of the reactions catalyzed by the two
metals (21).
Methanol carbonylation is one of only a few industrially important catalytic
reactions for which the quantitative reaction kinetics is known (22).

r ¼ k½CH3 I½Rh complex ð11Þ


512 CATALYSIS Vol. 1

I CO
Rh
I CO

MeCOI MeI
H2O
COMe Me
OC I I CO
Rh HI Rh
I CO I CO
I I
MeCO2H MeOH

CO

COMe
I I
Rh
I CO

Fig. 5. Catalytic cycle for methanol carbonylation (21).

Here, for the rhodium-complex-catalyzed reaction, k ¼ 3.5 106exp( Eact/RT)


and R is the gas constant, T is the absolute temperature, and Eact is 61.5 kJ/
mol (14.7 kcal/mol). Because the reaction is zero order in the reactants, a stirred
reactor has no disadvantage in comparison with a tubular flow reactor with
respect to the efficiency of use of the reactor volume. A stirred reactor might
therefore be preferred to facilitate rapid heat transfer and good gas–liquid
contacting for rapid gas–liquid mass transfer, ie, dissolution of the CO into
the reactant solution.
The reaction is carried out in a polar solvent, as the intermediates are
anionic. Water is present from the following reaction, which achieves virtual
equilibrium:

CH3 OH þ HI Ð CH3 I þ H2 O ð12Þ

Water is also formed in the acid-catalyzed dehydration of methanol to give


dimethyl ether. The solution is acidic because of the presence of the HI.
The process flow diagram (23) (Fig. 6) is characteristic of those for homoge-
neous catalytic processes in being dominated by the separations devices. The pro-
ducts are purified by distillation, and several columns are required because the
boiling point of the product acetic acid is near the middle of the boiling range of
the product mixture; both the light products, including unconverted CO, CH3I,
and dimethyl ether, and the heavy rhodium complex must be recovered and
recycled efficiently for the process to be economical. Corrosion is also a concern,
and the reactor, separations devices, and lines are made of expensive stainless
steel alloys.
For years, researchers have attempted to minimize the problems of sep-
aration and corrosion by using catalysts that are solid analogues of the metal
complexes used in solution. The early attempts were not successful, in large mea-
sure because the rhodium complex was leached from the support (typically a
cross-linked polymer with pendant phosphine groups) and the polymer was not
Vol. 1 CATALYSIS 513

Scrubber To flare
system

Recycle

Product
to storage

Carbon
monoxide

Methanol Waste
acid
Reactor Light ends Drying Heavy ends Waste acid
column column column stripper

Fig. 6. Process flow diagram for methanol carbonylation catalyzed by a rhodium complex
to make acetic acid (23).

sufficiently stable. Recent work with more stable vinyl pyridine resins as sup-
ports has led to stable catalysts and apparently to a viable industrial process
(24); this is important as perhaps the first successful application of a supported
catalyst that is nearly analogous to a molecular metal complex catalyst.
Olefin Hydroformylation (The Oxo Process). One of the most important
industrial applications of transition-metal complex catalysis is the hydroformyla-
tion of olefins (25), illustrated for propylene:

CHO
CH3CH CH2 + CO + H2 CH3CHCH3 + CH3CH2CH2CHO
ð13Þ

The linear isomer is more valuable than the branched isomer (see BUTYRALDE-
HYDE). The product aldehydes are hydrogenated to give so-called oxo alcohols;
long-chain products are converted into sulfonates and used as detergents.
The oxo process was discovered in 1938 by Roelen and co-workers of
Ruhrchemie. The first catalysts were cobalt carbonyl complexes formed from
[HCo(CO)4]. They are still used on a large scale. The process is carried out at
temperatures of 1508C and pressures of >30 MPa (several hundred atm). The
high pressures are required to maintain the cobalt in the form of the soluble
metal carbonyl complex; at lower pressures, cobalt metal can form and deposit
514 CATALYSIS Vol. 1

on the reactor walls, possibly in a porous form that adds a significant resistance
to heat transfer. Use of cobalt complexes with phosphine ligands led to processes
with improved selectivities. The process flow diagram is complicated because it is
expensive to separate the catalyst from the products and recycle it in the high
pressure process. In the Kuhlmann process (25), the catalyst is extracted into
an aqueous stream and then regenerated with acid to give [HCo(CO)4], which
is absorbed into the reactant solution.
A significant advance in hydroformylation technology was made with the
discovery that phosphine complexes of rhodium, similar to those used in the
Wilkinson hydrogenation but incorporating H, CO, and olefin ligands as well
as triphenylphosphine, have catalytic activities several orders of magnitude
greater than those of cobalt; furthermore, the rhodium complexes are stable
enough to be used at low pressures, typically 1.5 MPa (15 atm) at 1508C, and
separated by distillation (25). Rhodium complex catalysts now are used in indus-
trial processes for hydroformylation of propylene. When a large excess of triphe-
nylphosphine is used as the solvent, the selectivity for the desired straight-chain
product is high, with >95% of the aldehyde being n-butyraldehyde.
The processing costs associated with separation and corrosion are still sig-
nificant in the low pressure process; for the process to be economical, the effi-
ciency of recovery and recycle of the rhodium must be very high.
Consequently, researchers have continued to seek new ways to facilitate the
separation and confine the corrosion. Extensive research was done with rhodium
phosphine complexes bonded to solid supports, but the resulting catalysts were
not sufficiently stable, as rhodium was leached into the product solution. A suc-
cessful solution to the engineering problem resulted from the application of
a two-phase liquid–liquid process (26) . The catalyst is synthesized with polar
–SO3Na groups on the phenyl rings of the triphenylphosphine

SO3Na

NaO3S SO3Na

P H

Rh P

SO3Na
NaO3S

P CO
NaO3S
SO3Na
SO3Na

SO3Na
Vol. 1 CATALYSIS 515

Olefin
Off-gas

CO + H2
M
2
4

3
Raw aldehyde
1

Water 5

Steam

Fig. 7. Process flow diagram for the two-phase hydroformylation of propylene where
1 ¼ reactor; 2 ¼ separator; 3 ¼ phase separator; 4 ¼ stripping column; and 5 ¼ heat exchan-
ger (26).

so that it is water soluble. In the reactor, the catalyst is largely confined to the
aqueous phase and the reactants are largely confined to the organic phase; reac-
tion likely takes place near the interface. A process flow diagram is shown in
Figure 7 (26). The two liquid phases are separated downstream of the stirred
reactor and the catalyst is recycled with a high efficiency. The process is applied
commercially and is competitive with the single-phase process. It is indicative of
a trend toward application of more homogeneous catalytic processes with envir-
onmentally friendly solvents such as water.
Attempts to anchor transition-metal complex catalysts to supports have
been extended to reactions far beyond just methanol carbonylation and hydro-
formylation, but almost all the resulting catalysts have failed to find practical
application. The principal difficulty is that the catalyst at each step of the cata-
lytic cycle must be stably bonded to the support, and this is often too stringent a
design criterion. Perhaps a more generally useful way to confine the catalyst in a
phase separate from that containing most of the reactants is to use a second
liquid phase. Water is one appealing choice, and promising work has also been
done with fluorous phases (27,28).
Olefin Metathesis. The olefin metathesis (dismutation) reaction (29,30)
converts olefins to lower and higher molecular weight olefins. For example, pro-
pylene is converted into ethylene and butene

¼CHR0 !RCH¼
2 RCH ¼ ¼CHR þ R0 CH¼
¼CHR0 ð14Þ

This reaction is catalyzed in solution by complexes of tungsten, molybdenum, or


rhenium in high oxidation states, eg, Re7þ. Examples of catalyst precursors are
[W(CH-t-C4H9)(OR)2X2] (31a), where X is halide. A number of catalysts require
cocatalysts such as CH3Al2Cl3.
516 CATALYSIS Vol. 1

The olefins that undergo metathesis include most simple and substituted
olefins; cyclic olefins give linear high molecular weight polymers. The mechanism
of the reaction is believed to involve formation of carbene complexes that react
via cyclic intermediates, ie, metallacycles. An extension of the chemistry has
led to efficient ring-opening metathesis polymerization (32).
Industrial olefin metathesis processes are carried out with solid catalysts (29).
The Wacker Oxidation of Ethylene to Acetaldehyde. One of the early
industrial examples of organometallic catalysis in solution was described by
Smidt and co-workers in 1959 (33). This is the Wacker oxidation of ethylene to
give acetaldehyde. It had long been known that Pd(II) complexes in solution
would stoichiometrically oxidize ethylene to acetaldehyde. Smidt’s insight was
to make this reaction part of a catalytic cycle. The researchers created a process
by closing the cycle, using an excess of an oxidizing agent, Cu2þ, to rapidly reox-
idize the palladium that oxidized the ethylene and thereby keep palladium from
plating out on the reactor walls as metal. The cycle is completed as O2 reoxidizes
the Cuþ formed from Cu2þ. The sequence of reactions is the following:

CH2 ¼
¼CH2 þ H2 O þ PdCl2 ! CH3 CHO þ Pd þ 2 HCl ð15Þ
Pd þ 2 CuCl2 ! PdCl2 þ 2 CuCl ð16Þ
1
2 CuCl þ O2 þ 2 HCl ! 2 CuCl2 þ H2 O ð17Þ
2

These reactions constitute the cycle; their sum gives the stoichiometry of the
Wacker oxidation:

1
CH2 ¼
¼CH2 þ O2 ! CH3 CHO ð18Þ
2

This process is unusual among those involving organometallic chemistry in that


it takes place in aqueous solution. In one process design, there is a single reactor
in which all three reactions in the sequence take place, and the feed contains
pure O2. In another process, there are two reactors, with the ethylene oxidation
(eq. 15) and the palladium reoxidation (eq. 16) taking place in one reactor and the
reoxidation of Cuþ with air taking place in the other. An advantage of the single-
reactor process is the confinement of the corrosive solutions in one vessel, but
this is compensated by the need for purified oxygen . The feed gas contains ethy-
lene and oxygen, and the composition must be kept outside the explosive limits.
Consequently, the feed gas is not a stoichiometric mixture, and gas recycle
is necessary for economic operation. In contrast, the two-reactor process allows
the use of air as a feed to the reactor where Cuþ is reoxidized; the air flows
once through the reactor and is almost all converted. Ethylene flows into the
other reactor. A disadvantage of the two-reactor process is the need for corro-
sion-resistant materials in both reactors and the lines between them. Both
designs have been successfully applied.
A process similar to the Wacker process has been applied for the oxidation
of ethylene with acetic acid to give vinyl acetate, but now the principal applica-
tions are with a solid catalyst.
Vol. 1 CATALYSIS 517

Free-Radical Oxidation of Hydrocarbons. The chemistry of the Wacker


reaction is atypical for oxidation. A more common pattern in catalytic oxidation
is a chain reaction proceeding through free-radical intermediates; hydroper-
oxides are common products, and they are often converted themselves into
other products (34,35). Free-radical reactions are characterized by complex pro-
duct distributions because O2 has a high reactivity with organic reactants, with
metal centers, and with many ancillary ligands. Metals typically play a role in an
initiation process, helping to start a free-radical chain reaction; free radicals are
often generated by metal-catalyzed decomposition of organic hydroperoxides.
Some of the important applications of such chemistry are oxidations of
petroleum-derived hydrocarbons. Catalysts include iron, manganese, cobalt,
and copper in high oxidation states; a typical solvent is acetic acid. The processes
include oxidation of n-butane to give acetic acid, formation of fatty acids from
paraffin wax, and oxidation of cyclohexane to give adipic acid precursors in
nylon manufacture.
A simplified mechanism of the cyclohexane oxidation is shown in Figure 8
(36). A free-radical chain is set up, with R and ROO being the chain carriers

O2

R• RH

C6H12 C6H11• C6H11OO•

C6H11OOH C6H12

H2O

Co⫹3

OH⫺ H⫹
Co⫹2

C6H11O• C6H11OO•

OH OOH

Fig. 8. Simplified mechanism for the free-radical oxidation of cyclohexane (36).


518 CATALYSIS Vol. 1

(R is C6H11), as shown in the upper right of the figure. One role of Co3þ is
suggested to be that of an electron-transfer agent undergoing a one-electron
change; the role of the metal is that of a redox initiator, as cobalt cycles between
the þ2 and þ3 oxidation states in a process that generates free radicals. As the
cobalt also reacts in a cycle, its role is catalytic, and only small amounts are
needed.
One of the most important oxidation processes is the conversion of p-xylene
into terephthalic acid or dimethyl terephthalate. There are numerous commer-
cial processes, generally using air as the oxidant and cobalt and manganese
salts as the catalysts (37). In one process, the oxidation takes place at 2258C
and 15 atm, and bromides are involved in the catalysis (37).
CO and H2 Reactions. Synthesis gas (CO and H2) can be converted
into numerous organic chemicals, including acetic anhydride formed in a com-
mercial process with a rhodium complex catalyst. Soluble rhodium catalysts
are also active for synthesis of ethylene glycol along with methanol and other
products (38). The process is not applied commercially because extremely
high pressures are required. But the process is conceptually important because
it is believed that the catalysts may be compounds with metal–metal
bonds, called metal clusters. A simple metal cluster is [Rh13(CO)24H3]2 ,
which has the following structure:

There are many related compounds, including rhodium carbonyl cluster


anions that are present in the solutions catalyzing ethylene glycol formation
and which may be the catalytically active species or in equilibrium with
them (38).
There are only a few well-documented examples of catalysis by metal clus-
ters, and not many are to be expected as most metal clusters are fragile and frag-
ment to give metal complexes or aggregate to give metal under reaction
conditions (39). However, the metal carbonyl clusters are conceptually important
because they form a bridge between catalysts commonly used in solution, ie,
transition-metal complexes with single metal atoms, and catalysts commonly
used on surfaces, ie, small metal particles or clusters.
3.4. Phase-Transfer Catalysis. When two reactants in a catalytic
process have such different solubility properties that they can hardly both be
present in a single liquid phase, the reaction is confined to a liquid– liquid
interface and is usually slow. However, the rate can be increased by orders of
magnitude by application of a phase-transfer catalyst (39,40), and these are
Vol. 1 CATALYSIS 519

used on a large scale in industrial processing. Phase-transfer catalysts func-


tion by facilitating mass transport of reactants between the liquid phases.
Often most of the reaction takes place close to the interface.
Industrial examples of phase-transfer catalysis are numerous and growing
rapidly; they include polymerization, substitution, condensation, and oxidation
reactions. The processing advantages, besides the acceleration of the reaction,
include mild reaction conditions, relatively simple process flow diagrams, and
flexibility in the choice of solvents.

4. Heterogeneous Catalysis

4.1. Characterization of Surface Processes. Most of the largest


scale catalytic processes take place with gaseous reactants in the presence of
solid catalysts. From an engineering viewpoint, these processes offer the fol-
lowing advantages, in contrast to those involving liquid catalysts: (1) wide
ranges of temperature and pressure are economically applied. The most direct
way to increase the activity of a catalyst is to increase its temperature, and
high temperatures can be used economically with solid catalysts as the reac-
tants are gases and there is no liquid phase to require a high pressure. The
catalysts must be robust to withstand the high temperatures; solid catalysts
are commonly used at temperatures of 5008C, and some endure tempera-
tures several hundred degrees higher; (2) solid catalysts are only rarely cor-
rosive; (3) the separation of gaseous or liquid products from solid catalysts is
simple and costs little. It is simplest when the catalyst particles are confined in
a fixed-bed reactor; when very small particles are used in a fluidized bed, they
are entrained in the product stream, and a gas–solid separator is required;
(4) the mixing and mass transport in a fixed- or fluidized-bed reactor are faci-
litated by the solid catalyst particles through which the reactants and pro-
ducts flow. If no liquid phase is present, a potentially significant mass-
transfer resistance is eliminated; mass-transfer resistance in the gas phase
is usually small. However, reactions take place on the internal surfaces of
porous catalyst particles, and the resistance to transport through the pores
is often significant; (5) strongly exothermic and strongly endothermic reactions
are routinely carried out with solid catalysts. Fluidized-bed reactors offer the
advantage of excellent mixing and high rates of heat transfer. They are com-
monly used when heat effects are large, eg, for oxidation reactions; alterna-
tively, small tubes are used in bundles to minimize the distance through
which the heat is transferred in a fixed bed. The catalyst particles
themselves constitute a significant heat-transfer resistance, and the temp-
erature gradients in particles are sometimes large.
4.2. Properties of Solid Catalysts. Most solid catalysts used on a large
scale are porous inorganic materials. A number of these and the reactions they
catalyze are summarized in Table 1 (10); a list of new processes was compiled by
Armor (41).
Catalysis takes place as one or more of the reactants is chemisorbed (che-
mically adsorbed) on the surface and reacts there. The activity and selectivity of
a catalyst depend strongly on the surface composition and structure.
520 CATALYSIS Vol. 1

Table 1. Some Large-Scale Industrial Processes Catalyzed by Surfaces of


Inorganic Solidsa
Catalyst Reaction
metals (eg, Ni, Pd, Pt, as C¼
¼C bond hydrogenation (eg, olefin þ H2 ! paraffin)
powders or on supports)
or metal oxides (eg,
Cr2O3)
metals (eg, Cu, Ni, Pt) C¼¼O bond hydrogenation (eg, acetone þ H2 !
2-propanol)
metal (eg, Pd, Pt) Fe, Ru complete oxidation of hydrocarbons, oxidation of CO3
(supported and pro- H2 þ N2 ! 2NH3
moted with alkali
metals)
Ni CO þ 3H2 ! CH4 þ H2O (methanation)
CH4 þ H2O ! 3 H2 þ CO (steam reforming)
Fe or Co (supported and CO þ H2 ! paraffins þ olefins þ H2O þ CO2 þ oxygen-
promoted with alkali containing organic compounds) (Fischer-Tropsch
metals) reaction)
Cu (supported on ZnO, CO þ 2H2 ! CH3OH
with other components,
eg, Al2O3)
Re þ Pt (supported on g- paraffin dehydrogenation, isomerization and
Al2O3 and promoted dehydrocyclization (eg, n-heptane ! toluene þ 4 H2)
with chloride) (naphtha reforming)
solid acids (eg, SiO2- paraffin cracking and isomerization; aromatic alkylation;
Al2O3, zeolites) polymerization of olefins
g-Al2O3 alcohol ! olefin þ H2O
Pd supported on zeolite paraffin hydrocracking
metal-oxide-supported olefin polymerization (eg, ethylene ! polyethylene)
complexes of Cr, Ti, or
Zr
metal-oxide-supported olefin metathesis (eg, 2-propylene ! ethylene þ butene)
complexes of W or Re
V2O5 or Pt 2SO2 þ O2 ! 2SO3
Ag (on inert support, ethylene þ 12 O2 ! ethylene oxide (with CO2 þ H2O)
promoted by alkali
metals)
V2O5 (on metal oxide naphthalene þ 92 O2 ! phthalic anhydride þ
support) 2 CO2 þ 2 H2O
o-xylene þ 3 O2 ! phthalic anhydride þ 3H2O
bismuth molybdate, propylene þ 12 O2 ! acrolein
uranium antimonate, propylene þ 32 O2 þ NH3 ! acrylonitrile þ 3 H2O
other mixed metal
oxides
mixed oxides of Fe and Mo CH3OH þ O2 ! formaldehyde (with CO2 þ H2O)
Fe3O4 or metal sulfides H2O þ CO ! H2 þ CO2 (water gas shift reaction)
½Co Mo=g olefin hydrogenation aromatic hydrogenation
Al2 O3 ðsulfidedÞ hydrodesulfurization hydrodenitrogenation
fNi Mo=g
Al2 O3 ðsulfidedÞ
Ni W=g Al2 O3
ðsulfidedÞ
a
Ref. 10.
Vol. 1 CATALYSIS 521

The catalysts with the simplest compositions are pure metals, and the
metals that have the simplest and most uniform surface structures are single
crystals. Researchers have done many experiments with metal single crystals
in ultrahigh vacuum chambers so that unimpeded beams of particles and radia-
tion can be used to probe them. These surface science experiments have led to
fundamental understanding of the structures of simple adsorbed species, such
as CO, H, and small hydrocarbons, and the mechanisms of their reactions;
they indicate that catalytic activity is often sensitive to small changes in surface
structure. For example, paraffin hydrogenolysis reactions take place rapidly on
steps and kinks of platinum surfaces but only very slowly on flat planes; how-
ever, hydrogenation of olefins takes place at approximately the same rate on
each kind of surface site.
A few industrial catalysts have simple compositions, but the typical catalyst
is a complex composite made up of several components, illustrated schematically
in Figure 9 by a catalyst for ethylene oxidation. Often it consists largely of a por-
ous support or carrier, with the catalytically active components dispersed on the
support surface. For example, petroleum refining catalysts used for reforming of
naphtha have 1 wt% Pt and Re on the surface of a transition alumina such as
g-Al2O3 that has a surface area >100 square meters per gram. The expensive
metal is dispersed as minute particles or clusters so that a large fraction of the
atoms are exposed at the surface and accessible to reactants.
The typical industrial catalyst has both microscopic and macroscopic
regions with different compositions and structures; the surfaces of industrial cat-
alysts are much more complex than those of the single crystals of metal investi-
gated in ultrahigh vacuum experiments. Because surfaces of industrial catalysts
are very difficult to characterize precisely and catalytic properties are sensitive

Microparticle of a -Al2O3

Pore

Ag Particle on Al2O3 surface

Alkali metal promoter


present on surface

Porous support particle

Fig. 9. Schematic representation of a catalyst for ethylene oxide synthesis (not to scale).
The porous support particle consists of microparticles held together by a binder.
522 CATALYSIS Vol. 1

to small structural details, it is usually not possible to identify the specific com-
binations of atoms on a surface, called catalytic sites or active sites, that are
responsible for catalysis. Experiments with catalyst poisons, substances that
bond strongly with catalyst surfaces and deactivate them, have shown that the
catalytic sites are usually a small fraction of the catalyst surface. Although most
models of catalytic sites rest on rather weak foundations, some are well estab-
lished. For example, NO decomposition on ruthenium takes place at steps, as
shown by scanning tunneling microscopy images of individual reacting atoms
on the surface (42).
Important physical properties of catalysts include the particle size and
shape, surface area, pore volume, pore size distribution, and strength to resist
crushing and abrasion. Measurements of catalyst physical properties (43a)
are routine and often automated. Pores with diameters <2.0 nm are called
micropores; those with diameters between 2.0 and 5.0 nm are called meso-
pores; and those with diameters >5.0 nm are called macropores. Pore volumes
and pore size distributions are measured by mercury penetration and by N2
adsorption. Mercury is forced into the pores under pressure; entry into a
pore is opposed by surface tension. For example, a pressure of 71 MPa
(700 atm) is required to fill a pore with a diameter of 10 nm. The amount of
uptake as a function of pressure determines the pore size distribution of the
larger pores (44). In complementary experiments, the sizes of the smallest
pores (those 1 – 20 nm in diameter) are determined by measurements char-
acterizing desorption of N2 from the catalyst. The basis for the measure-
ment is the capillary condensation that occurs in small pores at pressures
less than the vapor pressure of the adsorbed nitrogen. The smaller the dia-
meter of the pore, the greater the lowering of the vapor pressure of the liquid
in it.
Surface areas are determined routinely and exactly from measurements of
the amount of physically adsorbed (physisorbed) nitrogen. Physical adsorption is
a process akin to condensation; the adsorbed molecules interact weakly with the
surface and multilayers form. The standard interpretation of nitrogen adsorption
data is based on the BET model (45), which accounts for multilayer adsorption.
From a measured adsorption isotherm and the known area of an adsorbed N2
molecule, taken to be 0.162 nm2, the surface area of the solid is calculated (see
ADSORPTION).
4.3. Influence of Mass Transport on Catalyst Performance. Reac-
tants must diffuse through the network of pores of a catalyst particle to reach
the internal area, and the products must diffuse back. The optimum porosity
of a catalyst particle is determined by tradeoffs: making the pores smaller
increases the surface area and thereby increases the activity of the catalyst,
but this gain is offset by the increased resistance to transport in the smaller
pores; increasing the pore volume to create larger pores for faster transport is
compensated by a loss of physical strength. A simple quantitative development
(46–48) follows for a first-order, isothermal, irreversible catalytic reaction in a
spherical, porous catalyst particle.
If there is a significant resistance to transport of the reactant in the
pores, a concentration gradient will exist at steady state, whereby the concen-
tration of the reactant is a maximum at the particle periphery and a minimum
Vol. 1 CATALYSIS 523

at the particle center. The product concentration will be higher at the particle
center than at the periphery. The concentration gradients provide the driving
force for the transport.
As a reactant molecule from the fluid phase surrounding the particle
enters the pore structure, it can either react on the surface or continue diffus-
ing toward the center of the particle. A quantitative model of the process is
developed by writing a differential equation for the conservation of mass of
the reactant diffusing into the particle. At steady state, the rate of diffusion
of the reactant into a shell of infinitesimal thickness minus the rate of diffusion
out of the shell is equal to the rate of consumption of the reactant in the shell
by chemical reaction. Solving the equation leads to a result that shows how the
rate of the catalytic reaction is influenced by the interplay of the transport,
which is characterized by the effective diffusion coefficient of the reactant in
the pores, Deff, and the reaction, which is characterized by the first-order reac-
tion rate constant.
The result is shown in Figure 10, which is a plot of the dimensionless effec-
tiveness factor as a function of the dimensionless Thiele modulus f, which is R(k/
Deff)1/2, where R is the radius of the catalyst particle and k is the reaction rate
constant. The effectiveness factor is defined as the ratio of the rate of the reaction
divided by the rate that would be observed in the absence of a mass transport
influence. The effectiveness factor would be unity if the catalyst were nonporous.
Therefore, the reaction rate is

r ¼ Zk½As ð19Þ

1.0

First order
0.8

0.6

Zero order
Effectiveness factor η

0.4

Second order

0.2

0.1
1 2 4 6 8 10 20 40
Thiele modulus φ = R k
D

Fig. 10. The Thiele plot accounting for the influence of intraparticle mass transport on
rates of catalytic reaction. The dimensionless terms Z and f are the effectiveness factor
and the Thiele modulus, respectively, and are defined in the text.
524 CATALYSIS Vol. 1

where Z is the effectiveness factor and [A]s the concentration of the reactant A at
the peripheral surface of the catalyst particle. The observed reaction rate con-
stant is not the intrinsic reaction rate constant k but the product Zk.
Figure 10 shows that Z is a unique function of the Thiele modulus. When
the modulus f is small (91), the effectiveness factor is unity, which means that
there is no effect of mass transport on the rate of the catalytic reaction. When f is
greater than 1, the effectiveness factor is less than unity and the reaction rate
is influenced by mass transport in the pores. When the modulus is large (010),
the effectiveness factor is inversely proportional to the modulus, and the reaction
rate (eq. 19) is proportional to k/f, which, from the definition of f, implies that
the rate and the observed reaction rate constant are proportional to (1/R)(Deffk)1/
2
[A]s. This result shows that both the rate constant, ie, a measure of the intrinsic
activity of the catalyst, and the effective diffusion coefficient, ie, a measure of the
resistance to transport of the reactant offered by the pore structure, influence the
rate. It is not appropriate to say that the reaction is diffusion controlled; it
depends on both the diffusion and the chemical kinetics. In contrast, as shown
by equation 3, a reaction in solution can be diffusion controlled, depending on
D but not on k.
In the case of a significant intraparticle diffusion influence, the tempera-
ture dependence of the observed rate constant, the apparent activation energy,
is not that of the intrinsic rate constant, ie, the true activation energy; rather, as
the Thiele modulus becomes large (>10), the temperature dependence
approaches that of (Deffk)1/2. The temperature dependence of a diffusion coeffi-
cient is usually small, and the measured activation energy is thus about half
the intrinsic activation energy. This is an important and general result: The
kinetics is disguised by the transport influence. When the effectiveness factor
is unity, there is no disguise, and the true activation energy is measured.
This development has been generalized. Results for zero- and second-
order irreversible reactions are shown in Figure 10. Results are given else-
where (48) for more complex kinetics, nonisothermal reactions, and particle
shapes other than spheres. For nonspherical particles, the equivalent spherical
radius, three times the particle volume/surface area, can be used for R to a
good approximation.
Even when there is a transport disguise, the reaction order remains one
for a first-order reaction. But for reactions that are not intrinsically first order,
the transport disguise changes the observed reaction order; for an intrinsically
zero-order reaction, the observed order becomes 1/2 and for an intrinsically
second-order reaction it becomes 3/2 when f010. For all reaction orders
the apparent activation energy is approximately one-half of the intrinsic
value in this limit.
The mass transport influence is easy to diagnose experimentally. One mea-
sures the rate at various values of the Thiele modulus; the modulus is easily
changed by variation of R, the particle size. Crushing and sieving the particles
provide catalyst samples for the experiments. If the rate is independent of the
particle size, the effectiveness factor is unity for all of them. If the rate is inver-
sely proportional to particle size, the effectiveness factor is less than unity and
f010. If the dependence is between these limits, then several experimental
points allow triangulation on the curve of Figure 10 and estimation of Z and
Vol. 1 CATALYSIS 525

f. It is also possible to estimate the effective diffusion coefficient and thereby to


estimate Z and f from a single measurement of the rate (48).
If the effectiveness factor is less than unity, the catalyst is not being used
efficiently, ie, the central region is starved of reactant. The results of Figure 10
show how to increase the effectiveness factor: decrease the Thiele modulus. This
can be done by some combination of the following: (1) decreasing the particle size
R. However, if particles are too small, they may cause too great a pressure drop
in a flow reactor or be entrained in the product stream; (2) changing particle
shape to reduce the transport length. Particles with cross sections resembling
clover leaves and wagon wheels have been used, for example; (3) increasing
the effective diffusion coefficient. Larger catalyst pores accomplish this, but
with a sacrifice in physical strength; and (4) decreasing the activity of the cata-
lyst measured by k. This option is unappealing as researchers strive to make
more active catalysts, but it may be economical to reduce the activity by reducing
the loading of the catalytically active component on a support. It may also be
appropriate to prepare the catalyst with the active component concentrated
near the particle periphery and not in the particle interior.
Intraparticle mass transport resistance can lead to disguises in selectivity.
If a series reaction A!B!C takes place in a porous catalyst particle with a small
effectiveness factor, the observed conversion to the intermediate B is less than
what would be observed in the absence of a significant mass transport influence.
This happens because as the resistance to transport of B in the pores increases, B
is more likely to be converted to C rather than to be transported from the catalyst
interior to the external surface. This result has important consequences in pro-
cesses such as selective oxidations, in which the desired product is an intermedi-
ate and not the total oxidation product CO2.
Rates and selectivities of solid-catalyzed reactions can also be influenced by
mass-transport resistance in the external fluid phase. Most reactions are not
influenced by external-phase transport, but the rates of some very fast reactions,
eg, ammonia oxidation, are determined solely by the resistance to this transport.
As the resistance to mass transport within the catalyst pores is larger than that
in the external fluid phase, the effectiveness factor of a porous catalyst is
expected to be less than unity whenever the external-phase mass transport resis-
tance is significant. A practical catalyst that is used under such circumstances is
the ammonia oxidation catalyst. It is a nonporous metal and consists of layers of
wire mesh.
4.4. Catalyst Components. Industrial catalysts are typically complex
in composition and structure, consisting of catalytically active phases, supports,
binders, and promoters.
Catalytically Active Species. The most common catalytically active mate-
rials are metals, metal oxides, and metal sulfides. Occasionally, these are used in
pure form; examples are Raney nickel, used for fat hydrogenation, and g-Al2O3,
used for ethanol dehydration. More often the catalytically active component is
highly dispersed on the surface of a support and may constitute no more than
1% of the total catalyst. The main reason for dispersing the catalytic species
is the expense. The expensive material must be accessible to reactants, and
this requires that most of the catalytic material be present at a surface. This
is possible only if the material is dispersed as minute particles, as small as
526 CATALYSIS Vol. 1

1 nm in diameter and even less. It is not practical to use minute particles by


themselves, as they would be entrained in products and clog lines and pumps,
and their use in a fixed-bed reactor would cause large pressure drops. Dispersion
on a support may also help stabilize the catalytically active species.
Supports. The principal component of a typical catalyst is the porous
support (49,50). Most supports are robust solids that can be made with wide
ranges of surface areas and pore size distributions. The most widely applied
supports are metal oxides; others are carbon, kieselguhr, organic polymers,
and zeolites.
The most commonly applied catalyst support is g-Al2O3, one of the transi-
tion aluminas (33,51). These are defective metastable solids formed from nonpor-
ous Al(OH)3 by heating it to 5008C or more; continued heating gives the more
stable d-Al2O3. As the Al(OH)3 is heated in air, it decomposes into an oxide with a
micropore system and a surface area of hundreds of square meters per gram. The
solid consists of small, crystalline primary particles; the spaces between these
primary particles are micropores and mesopores. As the solid is heated to
11008C, there is a series of phase changes and ultimately a collapse of the
pore structure and loss of almost all the internal surface area as finally the
extremely hard, crystalline a-Al2O3 (corundum) is formed. It has a melting
point of 21008C.
Transition aluminas are good catalyst supports because they are inexpen-
sive and have good physical properties. They are mechanically stable, stable at
relatively high temperatures even under hydrothermal conditions, ie, in the
presence of steam, and easily formed in processes such as extrusion into shapes
that have good physical strength such as cylinders. Transition aluminas can
be prepared with a wide range of surface areas, pore volumes, and pore size
distributions.
Macropores can be introduced into g-Al2O3 by including particles of
an organic material such as carbon or sawdust with the Al(OH)3 (51). When
the g-Al2O3 is formed, the organic particles are surrounded by the alumina.
The organic material is removed by burning, leaving macropores; the macropore
dimensions are determined by the particle size of the organic material.
Most catalyst supports are simply nearly inert platforms that help stabilize
the dispersion of the catalytically active phase. Sometimes, however, the sup-
ports play a direct catalytic role, as exemplified by the alumina used in supported
Pt and RePt catalysts for naphtha reforming.
The surfaces of g-Al2O3 and most other metal oxides are covered with polar
functional groups including OH groups and O2 ions (52). A structural model of
g-Al2O3 is shown in Figure 11 (53). When the solid is heated, it gives off water in
a reversible process called dehydroxylation (54). As a result, surface OH groups
are lost and Al3þions are exposed at the surface. The microparticles of the oxide
interact strongly with each other through the surface functional groups; eg,
hydrogen bonding can occur. As a result, a microporous solid consisting of
such particles is a strong material. In contrast, some support materials, exempli-
fied by a-Al2O3, have few surface functional groups, and a solid consisting of
microparticles of such a material lacks physical strength.
Binders. To create needed physical strength in catalysts, materials called
binders are added (51); they bond the catalyst. A common binder material is a
Vol. 1 CATALYSIS 527

Fig. 11. Structural model of the (111) face of g-alumina (53). The small solid circles
represent Al3þ, the large open circles OH groups, and the hatched circles oxygen. The
surface is 50% dehydroxylated.

clay mineral such as kaolinite. The clay is added to the mixture of microparticles
as they are formed into the desired particle shape, eg, by extrusion. Then the
support is heated to remove water and possibly burnout material and then sub-
jected to a high temperature, possibly 15008C, to cause vitrification of the clay;
this is a conversion of the clay into a glasslike form that spreads over the micro-
particles of the support and binds them together.
Promoters. Many industrial catalysts contain promoters, commonly che-
mical promoters. A chemical promoter is used in a small amount and influences
the surface chemistry. Alkali metal ions are often used as chemical promoters,
eg, in ammonia synthesis catalysts, ethylene oxide catalysts, and Fischer-
Tropsch catalysts (55). They may be used in as little as parts per million quan-
tities. The mechanisms of their action are often not well understood. In contrast,
seldom-used textural promoters, also called structural promoters, are used in
massive amounts in some catalysts and affect their physical properties. These
are used in ammonia synthesis catalysts.
4.5. Catalyst Treatments. Catalysts often require activation or regen-
eration and their disposal also requires special consideration (56).
Activation. Some catalysts, eg, Ziegler olefin polymerization catalysts
(57), are highly reactive in the presence of air, and some, eg, iron catalysts,
are even pyrophoric; these must be handled under a blanket of inert gas. The
surfaces of most catalysts are reactive and not easily maintained in an active
state in the form in which they are conveniently supplied. Thus, many catalysts
must be activated prior to use. The activation may be as simple as exposure to
reactants under processing conditions, but some catalysts require specialized
treatments. For example, catalysts used for hydroprocessing of fossil fuels are
usually delivered as supported metal oxides, but in the operating state they
are supported metal sulfides. Catalyst suppliers specify detailed procedures
for treatment, eg, with a mixture of hydrogen and oil containing organosulfur
528 CATALYSIS Vol. 1

compounds, to carry out the sulfiding properly. The details are often critical, and
the catalyst manufacturer’s guarantees may be void if the procedure is not
carried out properly.
Deactivation. Catalysts lose activity and selectivity in many ways, and
much effort in process development goes into measuring the deactivation and
finding means to minimize it. Some catalysts undergo physical changes during
normal operation; eg, a catalytically active phase may be transformed into an
inactive phase. Catalysts also undergo sintering, which is a coalescence of parti-
cles to give larger particles, accompanied by loss of surface area. Iron sinters
under conditions of ammonia synthesis, but when a textural promoter such as
alumina is present, the sintering is greatly reduced. Catalyst components may
be volatile and gradually vaporized during use; examples are silica used as a sup-
port and molybdenum oxides present in selective oxidation catalysts.
Catalysts commonly lose activity in operation as a result of accumulation of
materials from the reactant stream. Catalyst poisoning is a chemical phenom-
enon. A catalyst poison is a component such as a feed impurity that as a result
of chemisorption, even in small amounts, causes the catalyst to lose a substantial
fraction of its activity. For example, sulfur compounds in trace amounts poison
metal catalysts. Arsenic and phosphorus compounds are also poisons for a num-
ber of catalysts. Sometimes the catalyst surface has such a strong affinity for a
poison that it scavenges it with a high efficiency. The poison may then adsorb
strongly on the catalyst at the upstream end of a fixed-bed reactor at the begin-
ning of operation, with a wave then moving downstream through the reactor as
the upstream surface becomes saturated.
A selective poison is one that binds to the catalyst surface in such a way
that it blocks the catalytic sites for one kind of reaction but not those for another.
Selective poisons are used to control the selectivity of a catalyst. For example,
nickel catalysts supported on alumina are used for selective removal of acetylene
impurities in olefin streams (58). The catalyst is treated with a continuous feed
stream containing sulfur to poison it to an exactly controlled degree that does not
affect the activity for conversion of acetylene to ethylene but does poison the
activity for ethylene hydrogenation to ethane. Thus the acetylene is removed
and the valuable olefin is barely converted.
Because catalyst surfaces are reactive and often sensitive to their environ-
ments, they may be irreversibly changed by exposure to undesired reactants.
Upsets in plant operations can lead to catastrophic losses of whole catalyst
charges. A large catalyst charge that is ruined can cost hundreds of thousands
of dollars as well as the cost of lost operation.
Catalysts are also deactivated or fouled by physical deposition of materials
present in or formed from feeds. Sometimes massive deposits form on surfaces
and block access of reactants to the catalytic sites. Coke is carbonaceous material
of various compositions, often aromatic and with a high molecular weight and a
typical composition of approximately CH. Coke forms on every hydrocarbon pro-
cessing catalyst and on most catalysts used for organic chemical conversions.
Inorganic materials are also deposited on catalysts. For example, the organova-
nadium and organonickel compounds in petroleum residua react to form vana-
dium and nickel sulfides on the surfaces of hydroprocessing catalysts. The
solid deposits reduce activity by covering catalytic sites and by filling pores
Vol. 1 CATALYSIS 529

and restricting the entry of reactants. When the effectiveness factor for the
deposition reaction is small, the pore mouths can become blocked and catalysts
can suffer near catastrophic failure. Small particles of solid such as dust can also
foul catalysts. This may be a problem in processes for cleanup of NOx emissions
from coal-fired power plants, and catalysts are designed in the form of monoliths
(honeycombs) to minimize the effect.
Regeneration. Deactivated catalysts are treated to bring back the cataly-
tic activity in processes called regenerations. Coke deposits are removed by
controlled combustion. Often low partial pressures of oxygen are used to keep
the rate of combustion and the temperature rise from becoming too large and
leading to damage of the catalyst, such as by sintering. Periodic coke burn-off
can be carried out many times with little damage to many catalysts. Most cata-
lysts last for months or years between regenerations, but catalysts used for
cracking of petroleum are in contact with reactants for only a few seconds;
then they are separated and cycled to another reactor where they are regener-
ated (and then they are cycled back to the cracking reactor) (see CATALYSTS,
REGENERATION).
Redispersion. Expensive catalyst components such as precious metals
are used in high dispersions on supports. During operation, the small metal par-
ticles tend to sinter, ie, migrate and agglomerate, into larger particles with a loss
of metal surface area and thereby a loss of catalytic activity. The metals in such
catalysts may be redispersed as part of the catalyst regeneration (59). For exam-
ple, after the coke is burned off a supported platinum catalyst, the catalyst may
be treated with a reactive atmosphere containing chlorine and oxygen to form
volatile platinum oxychloride species that are transported through the gas
phase and deposited on the pore walls, where they are then treated in H2 and
reduced. The result is an increased dispersion of the platinum and a reactivated
catalyst.
Reclamation, Disposal, and Toxicity. Removal of poisons and inorganic
deposits from used catalysts is typically difficult and usually uneconomical.
Thus, some catalysts are used without regeneration, although they may be pro-
cessed to reclaim expensive metal components. Used precious metal catalysts,
including automobile exhaust conversion catalysts, are treated (often by the sup-
pliers) to extract the metals, and recovery efficiencies are high. Some spent
hydroprocessing catalysts may be used as sources of molybdenum and other
valuable metals.
Some catalysts are hazardous materials, or they react to form hazardous
materials. For example, catalysts used for hydrogenation of carbon monoxide
form volatile metal carbonyl compounds such as nickel carbonyl, which are
highly toxic. Many catalysts contain heavy metals and other hazardous compo-
nents, and environmentally safe disposal has become an increasing concern and
expense.
4.6. Catalyst Preparation. Catalyst preparation is more an art than a
science (60). Many reported catalyst preparations omit important details and
are difficult to reproduce exactly, and this has hindered the development of cat-
alysis as a quantitative science. However, the art is developing into a science and
there are now many examples of catalysts synthesized in various laboratories
that have nearly the same physical and catalytic properties.
530 CATALYSIS Vol. 1

Supports are often prepared first and the catalyst and promoter components
added later. Metal oxide supports are usually prepared by precipitation from aqu-
eous solutions. Nitrates are commonly used anions; alkalies and ammonium are
commonly used cations. Metal oxide supports, eg, silica and alumina, are prepared
in the form of hydrogels. Mixed oxides such as silica–alumina are made by cogela-
tion. Careful control of conditions such as pH is important to give uniform products.
Sol–gel processes are useful for the preparation of some high-area oxides (61).
Supports are washed at controlled values of pH to remove impurities. Ions
and impurities in the preparative solution are easily occluded in the solid and
difficult to remove by washing. Therefore, ions that might poison the catalyst,
eg, Cl , SO42 , or alkali metal ions, are avoided. Many of the materials are cation
exchangers, and washing does not remove cations from them. Metal ions can be
removed by exchange with ammonium ions, which on heating give off ammonia
and leave hydrogen ions in surface OH groups. Drying of precipitates and
hydrated gels leads to evolution of gases and may generate microporosity, as
described above for transition aluminas. Porosity can also be created by reduc-
tion of a nonporous oxide; porous iron can be made this way.
Catalyst components are usually added in the form of precursor metal salts
in aqueous solutions. In impregnation, the support may be dried, evacuated, and
brought in contact with an excess of an impregnating solution containing metal
salts. The processes are complex, possibly involving some dissolution of the sup-
port and reprecipitation of structures including mixed metal species. The solid is
then dried and calcined, ie, brought to a high temperature, usually in air. Alter-
natively, in the incipient wetness method, just enough of the impregnating
solution is used to fill the pores of the support. The chemistry of the interactions
of catalyst precursors with metal oxide supports is beginning to be understood.
Important parameters that control the adsorption of metal complex precursors
from aqueous solution are the isoelectric point of the metal oxide, the pH of
the solution, and the nature of the metal complex (62). Depending on the condi-
tions of the contact, cationic or anionic species may be adsorbed. Sometimes these
are simple mononuclear (single-metal atom) species, but sometimes they are
complicated polynuclear ions. The support may dissolve to some degree in the
preparation solution and be redeposited in some form (perhaps with the catalyst
precursor) (63). The nature of the initially adsorbed species may significantly
affect the structure of the catalytic species in the resultant catalyst. After
impregnation, the catalyst may be activated, for example, by drying and calcin-
ing. Promoters may be added at various stages, eg, as a final step in the prepara-
tion or just prior to operation.
Supported metal catalysts are reduced, eg, by treatment in hydrogen at
temperatures in the range of 300–5008C. The reduction temperature may influ-
ence the stability of the metal dispersion.
4.7. Examples of Surface Catalysis. Molecular Catalysis on
Supports. The term molecular catalysis is commonly applied only to reactions
in uniform fluid phases, but it applies nearly as well to some reactions taking
place on supports. Straightforward examples are reactions catalyzed by polymers
functionalized with groups that closely resemble catalytic groups in solution.
Industrial examples include reactions catalyzed by ion-exchange resins, usually
sulfonated poly(styrene-divinylbenzene) (64). This polymer is an industrial
Vol. 1 CATALYSIS 531

catalyst for synthesis of methyl tert-butyl ether (MTBE) from methanol and iso-
butylene and synthesis of bisphenol A from phenol and acetone, among others.
The former application grew rapidly as MTBE became a component of high-
octane-number gasoline (see ETHERS).
The polymer has the following structure, which is shown schematically.

CH CH2 CH CH2

SO3H
SO3H CH CH2

The sulfonated resin is a close analogue of p-toluenesulfonic acid in terms of


structure and catalyst performance. In the presence of excess water, the SO3H
groups are dissociated, and specific acid catalysis takes place in the swelled
resin just as it takes place in an aqueous solution. When the catalyst is used
with weakly polar reactants or with concentrations of polar reactants that are
too low to cause dissociation of the acid groups, general acid catalysis prevails
and water is a strong reaction inhibitor (65).
The polymer-supported catalysts are thus important conceptually in linking
catalysis in solutions and catalysis on supports. The acid–base chemistry is fun-
damentally the same whether the catalytic groups are present in a solution or
anchored to the support. The polymer-supported catalysts have replaced acid
solutions in numerous processes because they minimize the corrosion, separa-
tion, and disposal problems posed by mineral acids.
Polymer-supported methanol carbonylation catalysts incorporating metal
complexes, mentioned above, also behave in much the same way as their
soluble analogues.
Surfaces of inorganic solids can be functionalized with catalytic groups just
as organic polymers can. For example, the hydroxyl groups on the surface of
silica can be used for synthesis of the following structure:

SO3H

Cl
Si
O O

This is an ion-exchanger like the sulfonated polymer. The silica surface can
also be functionalized with phosphine complexes; when combined with rhodium,
these give anchored complexes that behave like their soluble and polymer-
supported analogues as catalysts for olefin hydrogenation and other reactions:

CO
P Rh P
Cl
CH2 CH2
532 CATALYSIS Vol. 1

These silica-supported catalysts demonstrate the close connections between


catalysis in solutions and catalysis on surfaces, but they are not industrial cata-
lysts. However, silica is used as a support for chromium complexes, formed either
from chromocene or chromium salts, that are industrial catalysts for polymeriza-
tion of a-olefins (66,67). Supported chromium complex catalysts are used on an
enormous scale in the manufacture of linear polyethylene in the Unipol and
Phillips processes. The exact nature of the catalytic sites is still not known,
but it is evident that there is a close analogy linking soluble and supported
metal complex catalysts for olefin polymerization.
The newest industrial catalysts for olefin polymerization are supported metal
complexes, metallocenes (eg, zirconocenes) promoted by aluminum-containing
components such as methylalumoxane (68). These are used for polymerization of
ethylene and for the stereospecific polymerization of propylene giving high yields
of isotactic polypropylene; they are also used for synthesis of copolymers. The dis-
covery and rapid development of the large new class of supported catalysts is one
of the major successes of industrial catalysis in recent years.
The conceptual link between catalysis in solution and catalysis on surfaces
extends to surfaces that are not obviously similar in structure to molecular spe-
cies. For example, the early Ziegler catalysts for polymerization of propylene
were a-TiCl3. (Today, supported metal complexes are used instead). These
catalysts are selective for stereospecific propylene polymerization. The catalytic
sites are believed to be located at the edges of TiCl3 crystals. The surface struc-
tures have been inferred to incorporate anion vacancies, that is, sites where Cl
ions are not present and where Ti3þ ions are exposed (69). These cations exist in
octahedral surroundings. The polymerization has been explained by a mechan-
ism whereby the growing polymer chain and an adsorbed propylene bonded cis
to it on the surface undergo a migratory insertion reaction (70). In this respect,
there is no essential difference between the explanation of the surface catalyzed
polymerization and that catalyzed in solution.
Stereospecific polymerization in solution and on surfaces incorporating ana-
logous metallocenes has been explained in part by the steric restrictions of ligands
bonded to the metal center. For example, the following structure, among numer-
ous others, has been postulated as an intermediate in solution catalysis (71):

R Cz
CH2
M
P θP

The steric constraints imposed by the bulky ligands cause the propylene to bond
almost entirely with a single orientation with respect to the growing polymer
chain, CH2  P , which leads to the stereoregular product.
The explanation for the stereospecificity of the surface catalysis, which
preceded that for the solution catalysis, is based on the structure inferred for
Vol. 1 CATALYSIS 533

the a-TiCl3 crystal edges; the locations of the Cl ions at the anion vacancies cre-
ate an unsymmetrical environment whereby the growing polymer chain and the
adsorbed propylene are oriented predominantly in a single, energetically favored
way that leads to the stereoregular polymer as a result of a series of insertion
reactions. The explanation is simplified, but again there is a strong conceptual
link between molecular and surface catalysis. It is difficult to forge many such
links because of the lack of detailed understanding of most surface-catalyzed
reactions, which is a consequence of the complexity of the surface compositions
and structures of solid catalysts.
4.8. Catalysis by Metals. Metals are among the most important and
widely used industrial catalysts (72). They offer activities for a wide variety of
reactions (Table 1). Atoms at the surfaces of bulk metals have reactivities and
catalytic properties different from those of metals in metal complexes because
they have different ligand surroundings. The surrounding bulk stabilizes surface
metal atoms in a coordinatively unsaturated state that allows bonding of reac-
tants. Thus metal surfaces offer an advantage over metal complexes, in which
there is only restricted stabilization of coordinative unsaturation. Furthermore,
metal surfaces provide catalytically active sites that are stable at high
temperatures. For example, supported palladium catalysts have replaced soluble
palladium for vinyl acetate synthesis; the advantages of the solid include reduced
corrosion and reduced formation of by-products.
CO Oxidation Catalyzed by Palladium. One of the best understood cata-
lytic reactions occurring on a metal surface is the oxidation of carbon monoxide
on palladium:

2 CO þ O2 ! 2 CO2 ð20Þ

This reaction takes place similarly in automobile exhaust converters.


Carbon monoxide oxidation catalysis is understood in depth because poten-
tial surface contaminants such as carbon or sulfur are burned off under reaction
conditions and because the rate of CO oxidation is almost independent of pres-
sure over a wide range. Thus ultrahigh vacuum surface science experiments
could be done in conjunction with measurements of reaction kinetics (73). The
results show that at very low surface coverages, both reactants are adsorbed ran-
domly on the surface; CO is adsorbed intact and O2 is dissociated and adsorbed
atomically. When the coverage by CO is >1/3 of a monolayer, chemisorption of
oxygen is blocked. When CO is adsorbed at somewhat less than a monolayer, oxy-
gen is adsorbed, and the two are present in separate domains. The reaction that
forms CO2 on the surface then takes place at the domain boundaries.
The available results are consistent with the following sequence of steps on
the surface (73):

O2 þ 2 S ! O2;ads ! 2 Oads ð21Þ


CO þ S ! COads ð22Þ
COads þ Oads ! CO2 þ 2 S ð23Þ
CO þ Oads ! CO2 þ S ð24Þ
534 CATALYSIS Vol. 1

CO + 12_ O2

∆ H = 283
Energy, kJ/mol
260 Transition
state

Ea = 105

COads + Oads CO2


~20
CO2, ads

Reaction coordinate

Fig. 12. Schematic potential energy diagram illustrating the changes associated with
the individual reaction steps in CO oxidation on Pd (73). Eact ¼ 105 kJ/mol; DH ¼
283 kJ/mol. To convert kJ to kcal, divide by 4.184.

This depiction is vague because the exact nature of the sites S and their bonding
with reactants are not known. The experimental results have led to an approxi-
mate potential energy diagram characterizing these elementary steps on the
surface (Fig. 12) (73). This shows the role of the surface in providing an efficient
pathway for the reaction. Most of the energy is liberated as the reactants are
adsorbed; the activation energy for reaction of the adsorbed CO with the
adsorbed O is relatively small, and this step is only slightly exothermic.
Ammonia Synthesis. Another well-understood reaction is the ammonia
synthesis:
N2 þ 3 H2 ! 2 NH3 ð25Þ

This reaction is catalyzed by iron, and extensive research, including surface


science experiments, has led to an understanding of many of the details
(74,75). The adsorption of H2 on iron is fast, and the adsorption of N2 is slow
and characterized by a substantial activation energy. N2 and H2 are both
dissociatively adsorbed. Adsorption of N2 leads to reconstruction of the iron sur-
face and formation of structures called iron nitrides that have depths of several
atomic layers with compositions of approximately Fe4N. There is a bulk com-
pound Fe4N, but it is thermodynamically unstable when the surface structure
is stable. Adsorbed species such as the intermediates NH and NH2 have been
identified spectroscopically.
The following sequence of steps explains the observations (73):

N2 þ 2 S ! 2 Nads ð26Þ
H2 þ 2 S ! 2 Hads ð27Þ
2Nads þ Hads ! NHads þ S ð28Þ
NHads þ Hads ! NH2;ads þ S ð29Þ
NH2;ads þ Hads ! NH3 þ 2 S ð30Þ
Vol. 1 CATALYSIS 535

N + 3H

315

NH + 2 H

390
1130

NH2 + H
1400

960

N2Hads 450
540
~30 ~20 125

1
_
+ 3
_ H2
50 ∆ H = 45
N
2 2 2

260 42
1
_ 3
_
30 NH3, ads NH3, g
N
2 2, ads
+ 2
H2 120
NH2, ads + Hads
Nads + 3 Hads NHads + 2 Hads

Fig. 13. Schematic potential energy diagram for the catalytic synthesis and decomposi-
tion of ammonia on iron. The energies are given in kJ/mol; to convert kJ to kcal, divide by
4.184 (73).

where S refers to surface sites, the exact nature of which is unknown. An


approximate potential energy diagram for this sequence of steps is shown in
Figure 13, which shows how the catalyst facilitates the bond breaking reactions.
The energy gain resulting from the formation of the strong metal–nitrogen and
metal–hydrogen bonds makes the first steps endothermic. The dissociative
adsorption of N2 is rate determining, not because of a high activation energy
barrier but because the frequency factor, (preexponential factor) in the rate
constant is small. The ammonia synthesis mechanism is so well understood
that rates of the reaction under practical conditions have been predicted
from the rate of adsorption of N2 measured under low-pressure conditions, far
from those of practical catalysis, combined with the equilibria of the other
steps (75). The prediction was within a factor of two of the observed rate of the
industrial reaction. This result is a satisfying consolidation of decades of funda-
mental research motivated by an important catalytic process.
The industrial catalysts for ammonia synthesis consist of far more than the
catalytically active iron (76). There are textural promoters, alumina and calcium
oxide, that minimize sintering of the iron and a chemical promoter, potassium
(1 wt% of the catalyst), and possibly present as K2O; the potassium is believed
to be present on the iron surface and to donate electrons to the iron, increasing
its activity for the dissociative adsorption of N2. The primary iron particles are
30 nm in size, and the surface area is 15 m2/g. These catalysts last for years.
Some newer catalysts incorporate ruthenium instead of iron.
536 CATALYSIS Vol. 1

4.9. Catalysis by Metal Oxides and Zeolites. Metal oxides are com-
mon catalyst supports and catalysts. Some metal oxides alone are industrial cat-
alysts; an example is the g-Al2O3 used for ethanol dehydration to give ethylene.
But these simple oxides are the exception; mixed-metal oxides are more common.
For example, silica–alumina was used earlier as a catalyst for cracking of petro-
leum and is still a component of such catalysts, and bismuth molybdates were
used for ammoxidation of propylene to give acrylonitrile. Metal oxides supported
on metal oxides are also commonly applied. For example, rhenium oxide, Re2O7,
supported on alumina is used for olefin metathesis, and complicated supported
oxides related to bismuth molybdates are used for ammoxidation.
Metal oxide surfaces are more complex in structure and composition than
metal surfaces, and they are not so easy to characterize with some ultrahigh
vacuum techniques, eg, electron spectroscopies, because they are poor electrical
conductors and build up electrical charge when subjected to streams of charged
particles. Consequently, understanding of catalysis on metal oxide surfaces is
less advanced than understanding of catalysis on metal surfaces, although sig-
nificant progress has been made (77,78).
Acid– base chemistry of metal oxide surfaces is important in catalysis and
is characterized by measurements such as infrared spectroscopy with adsor-
bed probe molecules, eg, the base pyridine. The surfaces have both basic and
acidic character (79). Both OH and O groups have base strengths ranging from
weak, eg, in silica gel, to moderate, eg, in g-Al2O3, to strong, eg, in highly dehy-
droxylated MgO. Surface OH groups are acids with proton donor strengths
ranging from weak, eg, in Al2O3, to strong, eg, in SiO2-Al2O3. Metal ions
exposed at surfaces are Lewis acids. Redox properties are also important in
some catalytic applications, as cations in some oxides, eg, V2O5, can change
oxidation state. Furthermore the principles of organometallic chemistry are
useful in describing the interactions of organic ligands with metal ions exposed
at surfaces.
Zeolites and Catalytic Cracking. Some of the best-understood catalysts
are zeolites, ie, crystalline aluminosilicates (80–82). They are well understood
because they have much more nearly uniform compositions and structures
than amorphous metal oxides such as silica and alumina. Here the usage of
amorphous refers to results of X-ray diffraction experiments; the crystallites of
a metal oxide such as g-Al2O3 that constitute the microparticles are usually so
small that sharp X-ray diffraction patterns are not measured; consequently the
solids are said to be X-ray amorphous or simply amorphous.
Zeolites contain Si, Al, and O ions and various other cations. The structures
are built up of linked SiO4 and AlO4 tetrahedra that share O ions. These tetra-
hedra are arranged in a number of ways to give the different zeolites. The struc-
tures are unique in that they incorporate pores as part of the regular crystalline
structures. The pores have dimensions of the order of molecular dimensions so
that some molecules fit into the pores and some do not. Hence, the zeolites are
molecular sieves, and they are applied in industrial separations processes to
take advantage of this property. Some zeolites and their pore dimensions are
listed in Table 2.
A catalytically important family of zeolites called faujasites (zeolites X
and Y) is represented in Figure 14. Here the points of intersection of the lines
Vol. 1 CATALYSIS 537

Table 2. Zeolites and Their Pore (Aperture) Dimensionsa


CAS Registry Number of oxy- 10 Aperture
Zeolite Number gens in the ring dimensions, nm
chabazite [12251-32-0] 8 3.6 3.7
erionite [12510-42-8] 8 3.6 5.2
zeolite A 8 4.1
ZSM-5 (or silica- [58339-99-4] 10 5.1 5.5; 5.4 5.6
lite)
ZSM-11 10 5.1 5.5
heulandite 10 4.4 7.2
ferrieriteb 10 4.3 5.5
faujasite [12173-28-2] 12 7.4
zeolite LTL 12 7.1
mordenite [12173-98-7] 12 6.7 7.0
offretite 12 6.4
a
The framework oxygen is assumed to have a diameter of 0.275 nm.
b
There are also apertures with eight-membered oxygen rings in this zeolite.

represent Si or Al ions; oxygen is present at the center of each line. This depiction
emphasizes the framework structure of the zeolite and shows the presence of the
intracrystalline pore structure in which there are spaces called supercages, each
with a diameter of 1.2 nm. The pore structure is three dimensional; the super-
cages are connected by apertures with diameters of 0.74 nm. Some rather large
molecules can fit through these apertures (Fig. 15) and undergo catalytic reac-
tion in the cages.
The zeolite frame is made up of SiO4 tetrahedra, which are neutral, and
AlO4 tetrahedra, which have a charge of 1. The charge of the AlO4 tetrahedra
is balanced by the charges of additional cations that exist at various crystallogra-
phically defined positions in the zeolite, many exposed at the internal surface.
Zeolites are thus ion exchangers. The cations may be catalytically active.
When the cations are Hþ, the zeolites are acidic. Acidic zeolite Y finds enormous
industrial application as a component of petroleum cracking catalysts (81). In

Fig. 14. Schematic representation of the structure of a faujasitic zeolite. I and II indicate
cation positions.
538 CATALYSIS Vol. 1

0.6
0.4 0.4 Linear paraffins
Zeolite A

0.7 Erionite

0.8

Critical molecular dimension, nm


Average channel
Ferrierite
0.5 0.5 Benzene

size, nm
0.9 2-Methyl-substituted
olefins
ZSM-5 Isoparaffins
1.0
o - Xylene

1.1

aromatics
0.6 0.6

Higher
1.2
Offretite
Cavity size, nm

1.3
Mordenite
1.4 0.7 0.7

1.5 Faujasite Naphthalene

Fig. 15. Pore dimensions of zeolites and critical dimensions of hydrocarbons (82).

the following simplified structure, the OH groups located near AlO4 tetrahedra
are moderately strong Brønsted acids and responsible for the catalytic activity
for many reactions:

H
H
O O O O O
O O
Si Al Si Si Al Si
O OO OO OO OO OO O

Zeolites are named to represent the exchangeable cations in them; eg, zeo-
lite NaY is zeolite Y with sodium ions in the cation exchange positions.
Another catalytically important zeolite is ZSM-5. There is a three-dimen-
sional network of pores in this zeolite, represented in Figure 16. A set of straight
parallel pores is intersected by a set of perpendicular zigzag pores. These pores
are smaller than those of the faujasites (Fig. 15). ZSM-5 is classified as a medium
pore zeolite, the faujasites are large pore zeolites, and zeolite A (Table 2) is a
small pore zeolite.
Both the faujasites and ZSM-5 in the acidic form catalyze many reactions
that are catalyzed by other soluble and solid acids. The zeolites are not very
strong acids at low temperatures, but at 5008C they are able to protonate paraf-
fins and initiate their cracking. Almost all the catalytic applications of zeolites
take advantage of their acidic properties. Activities of a family of HZSM-5
samples with different Si/Al ratios have been studied (84). When the Al
contents are low, the catalytic activity is proportional to the Al content of the
zeolite over a wide range of compositions. These results identify the proton-
donor sites associated with the Al cations as the catalytic sites for the cracking
Vol. 1 CATALYSIS 539

Chain

Pentasil
5-1 unit
Pair

a
b

(a) (b)

Fig. 16. Structure of the zeolite ZSM-5 (83): (a) framework of the zeolite; (b) schematic
representation of the pore structure.

reaction. At higher concentrations of Al in the zeolite, the dependence is no


longer linear.
ZSM-5 is a component of some catalysts for cracking of petroleum (85–87),
but the larger pored zeolite Y in an acidic form is the principal catalytic compo-
nent. Zeolite Y is sometimes used in a form containing hydrogen ions to provide
acidity and rare-earth ions such as La3þ, which make the structure more stable.
The stability is valuable because the zeolites are deactivated in a matter of a few
seconds of contact with reactant vapors in the catalytic reactor and are subjected
to an atmosphere of oxygen and steam at temperatures as high as 8008C in the
regenerator where coke is burned off.
The feedstocks to crackers are petroleum fractions ranging from gas oil to
residuum. They undergo a complicated set of reactions, including cracking, iso-
merization, disproportionation, and coke formation, that proceed through carbe-
nium ion intermediates and give predominantly lower molecular weight
products, including, eg, many in the gasoline boiling range. The patterns in
the chemistry of catalytic cracking are consistent with the chemistry of hydrocar-
bons and carbenium ions in solution at much lower temperatures. For example,
cracking of a paraffin takes place predominantly in a cycle that is initiated by
formation of a carbenium ion by protonation of an olefin by the catalyst. The
chain reaction is the following:

þ
Rþ þ R0 H ! RH þ R0 ð31Þ
þ
R0 ! olefin þ Rþ ð32Þ

The second step is a b-scission, the breaking of a carbon–carbon bond b to the


charged carbon. The sum of the two reactions is the stoichiometry of the overall
cracking reaction R0 H ! RH þ olefin. Rþ, a relatively stable carbenium ion such
as the tert-butyl cation, is a chain carrier. The role of the catalyst is to donate the
proton to start the chain. This representation is greatly simplified.
540 CATALYSIS Vol. 1

Cracking catalysts are complex composites consisting of a support, eg,


silica–alumina, and the catalytically active zeolite, present as crystallites
roughly 1 mm in size dispersed in a matrix of the amorphous support (85). The
catalyst particles are small, roughly 50 mm in diameter on average so that they
can be fluidized by the vaporized oil entering the cracking reactor. The reactant
vapors carry the catalyst particles with them; such a design is necessary because
the catalyst is largely deactivated by coke after only several seconds of operation
and must therefore be efficiently transported out of the reactor and into the
regenerator where the coke is burned off.
The catalyst may contain 20 wt% zeolite, and more or less is used, depend-
ing on the feedstock and operating goals. The principal component of the catalyst
is the matrix, which has a relatively low catalytic activity, but it is active for
cracking the molecules that are too large to fit into the zeolite pores. The matrix
also plays the role of a heat-transfer medium; the cracking reactions are fast and
endothermic, and the temperature of the catalyst and the rates of the reactions
fall as the catalyst is carried by the oil vapors from the inlet toward the outlet of
the reactor. The thermal mass of the matrix keeps the temperature drop from
being too large and causing the cracking rate to fall off too quickly. After leaving
the reactor, the catalyst is heated up in the regenerator where the coke is burned
off and reenters the reactor at a suitably high temperature.
In addition to the matrix and the catalytically active zeolite, there are small
amounts of a supported metal such as platinum on alumina, which catalyzes CO
oxidation in the regenerator and minimizes the emissions of CO. There are also
metal oxide components that minimize the emission of SOx formed in the regen-
erator from combustion of organosulfur compounds from the oil. The metal
oxides react with SOx in the regenerator to make stable metal sulfates. Cycled
with the regenerated catalyst to the reducing atmosphere of the cracking reactor,
the sulfates are converted into H2S, which is removed by scrubbing the effluent
gas stream.
The discovery of new zeolites continues apace, and the newer ones tend to
have larger pores than those discovered earlier (88). Zeolites can be modified by
incorporation of cations in the crystalline lattice which are not exchangeable
ions, but can play catalytic roles. For example, silicalite, which has the structure
of ZSM-5 but without Al, incorporating Ti in the lattice, is a commercial catalyst
for oxidation of phenol with H2O2 to give diphenols; the catalytic sites are
isolated Ti cations (89).
There are numerous structures that are similar to zeolites, such as alumi-
nophosphate molecular sieves, AlPOs, and these are finding increasing catalytic
applications. A new process for removal of wax by isomerization to remove
straight-chain paraffins reportedly uses a nonzeolitic molecular sieve. Uni-
dimensional pores are selective in producing highly branched products and
few cracking products; the products has a low pour point and a high viscosity
index (90).
Shape-Selective Catalysis. The zeolites are unique in their molecular-
sieving character, which is a consequence of their narrow, uniform pores
(81,91). The transport of molecules in such small pores is different from that
in the larger pores of typical catalysts. Figure 17 is a schematic representation
of the diffusion coefficients of molecules in pores (92). When the pore diameters
Vol. 1 CATALYSIS 541

10 – 2 Molecular

Knudsen
10 – 4

10 – 6
D, cm2/s

10 – 8

Configurational
10–10

10–12

10–14
0.1 1 10 100 1000
Pore size, nm

Fig. 17. Schematic representation of the regimes of diffusion in pores (92).

are large in comparison with the dimensions of the diffusing molecules, then
molecular diffusion occurs, as it does in a fluid phase. When the pores become
smaller, the interactions of the molecules with the pore walls become dominant,
and Knudsen diffusion occurs. When the pores become so small that the mole-
cules barely fit through them, then configurational diffusion occurs. This diffu-
sion may be characterized by a substantial activation energy; the rate of
diffusion is a strong function of the pore and molecule sizes in this regime. In
the limit, a pore is too small for a molecule to fit in, and the diffusion coefficient
becomes zero.
Catalytic processes have been developed to take advantage of the unique
transport and molecular sieving properties of zeolites. The zeolite that has
found the most applications is the medium-pored HZSM-5 (81,91). The term
shape-selective catalysis is applied to describe the unique effects. There are
different kinds of shape selectivity. Mass transport shape selectivity is a conse-
quence of transport restrictions whereby some species diffuse more rapidly than
others in the zeolite pores. In the simplest kind of shape-selective catalysis, small
molecules in a mixture enter the pores and are catalytically converted whereas
large molecules pass through the reactor unconverted because they do not fit into
the pores where the catalytic sites are located. This statement is slightly over-
simplified as there are a few catalytic sites on the outer surfaces of zeolite crys-
tallites. Similarly, product molecules formed inside a zeolite may be so large that
their transport out of the zeolite may be very slow, and they may be converted
largely into other products that diffuse more rapidly into the product stream.
A different kind of shape selectivity is called restricted transition state selectivity
542 CATALYSIS Vol. 1

(81,91). It is not related to transport restrictions; rather, it is related to the size


restriction of the catalyst pore that suppresses the formation of the transition
state for a certain reaction, whereas it may not suppress the formation of a smal-
ler transition state for another reaction. A way to diagnose the nature of the
shape selectivity is to use zeolite crystallites of various sizes as the catalyst.
Mass transport selectivity is influenced by the particle size; restricted transition
state selectivity is not.
For example, HZSM-5 is a component of some cracking catalysts. It selec-
tively cracks the straight-chain paraffins rather than the branched paraffins
because of restricted transition state selectivity. This is a desired processing
goal as the straight-chain isomers have lower octane numbers than the others
and are less desirable gasoline components. In this example, the catalyst favors
the reaction of the straight-chain compounds, even though they are intrinsically
less reactive because they initially form secondary rather than the more stable
and easily formed tertiary carbenium ions. Thus the pore size restriction
reverses the pattern of selectivity that would be expected on the basis of the
intrinsic chemistry alone.
Mass transport shape selectivity is illustrated by a process for disproportio-
nation of toluene catalyzed by HZSM-5 (81,91). The desired product is p-xylene;
the other isomers are less valuable. The ortho and meta isomers are bulkier than
the para isomer and diffuse less readily in the zeolite pores. This transport
restriction favors their conversion to the desired product in the catalyst pores;
the desired para isomer is formed in excess of the equilibrium concentration.
Xylene isomerization is another reaction catalyzed by HZSM-5, and the catalyst
is preferred because of restricted transition state shape selectivity (91). An unde-
sired side reaction, the xylene disproportionation to give toluene and trimethyl-
benzenes, is suppressed because it is bimolecular and the bulky transition state
cannot readily form.
Mixed-Metal Oxides and Propylene Ammoxidation. The best catalysts
for partial oxidation are metal oxides, usually mixed-metal oxides. For example,
phosphorus–vanadium oxides are used commercially for oxidation of n-butane to
give maleic anhydride, and oxides of bismuth and molybdenum with other com-
ponents are used commercially for oxidation of propylene to give acrolein or acry-
lonitrile. A key to the success of a hydrocarbon oxidation catalyst is its ability to
convert the hydrocarbon selectively to a partially oxidized product, rather than
CO2; this is difficult because the partially oxidized product is intrinsically more
reactive than the hydrocarbon, especially when it is a paraffin.
The surface of a mixed-metal oxide exposes two kinds of metal ions in addi-
tion to O2 ions and OH groups. Selective oxidation of hydrocarbons, represented
schematically in Figure 18 (10), takes place on surface sites having oxygen atoms
of limited reactivity, associated with the metal M1 in the figure. These react with
a hydrocarbon to give water and a partially oxidized organic compound rather
than CO2. The surface sites are reoxidized by other components of the solid cat-
alyst rather than by O2 directly. A second metal plays the role of an intermediary
and oxygen is transported as ions through the bulk of the mixed metal oxide
catalyst. A compensating transport of electrons and reaction of O2 with the sur-
face at sites different from those where the hydrocarbon is adsorbed make the
process cyclic. Essentially this same pattern has already been illustrated by
Vol. 1 CATALYSIS 543

Hydrocarbon

M1n + O2 M2(m– y)+

O2–

M1(n– x)+ M2m+

H2O

e–
Partially
oxidized
hydrocarbon

Fig. 18. Schematic representation of the catalytic cycle for ammoxidation of propy-
lene and related reactions. M1 and M2 represent the two metals in a mixed-metal oxide
catalyst (10).

the Wacker oxidation, whereby the hydrocarbon reacts with an oxide, H2O, in a
step mediated by palladium, and a second metal, eg, copper, reacts with O2 and
then reoxidizes the palladium.
An important industrial partial oxidation process is the conversion of
propylene to acrylonitrile:

3
CH2 ¼
¼CHCH3 þ O2 þ NH3 ! CH2 ¼
¼CHCN þ 3 H2 O ð33Þ
2

The first catalysts used commercially to convert the propylene with high
selectivity were mixed oxides of bismuth and molybdenum, referred to as
bismuth molybdates. Improved catalysts consisting of a number of solid phases
have been developed, with each generation becoming more complicated
than its predecessor. Among the catalysts cited in a patent is the following:
Co2þ 2þ 3þ 3þ
6 Ni2 Fe3 Bi ðMoO4 Þ12 on 50 wt% SiO2 with some P and K (93). Silica is
the support. The other components provide new functions, perhaps making the
catalyst more stable. The catalysts are mechanically stable and used in fluidized-
bed reactors at low pressures 100 kPa (1 atm ), and temperatures of 400–
5008C. Years of development and improvement of catalysts for this process have
led to the recognition that lattice oxygen of certain multivalent transition-metal
oxides serves as a versatile, selective oxidizing agent, and these sites must be site
isolated on the catalyst surface (94).
544 CATALYSIS Vol. 1

4.10. Catalysis by Supported Metals. Metals used in industrial cata-


lysis are often expensive, and they are predominantly used in a highly dispersed
form. Metal species dispersed on supports may be as small as the mononuclear
chromium and zirconium complexes used for olefin polymerization, or they may
be clusters containing as few as 10 metal atoms (95), or they may be larger par-
ticles that have three-dimensional structures and resemble small chunks of
metal. The surfaces of the latter expose a number of different crystal faces,
and if the particles are larger than 5 nm the distribution of crystal faces may
be almost independent of particle size and shape. The smaller clusters, however,
are less like bulk metals and are more likely to have unique structures and cat-
alytic properties. The interactions between the metals and the support may be
thought of as effects comparable to the ligand effects in molecular catalysis;
the catalytic properties are sensitive to the structure and size of the metal
cluster (95).
Ethylene Oxidation to Ethylene Oxide. A thoroughly investigated reac-
tion catalyzed by a supported metal is the commercially applied partial oxidation
of ethylene to give ethylene oxide (96). The desired reaction is the formation of
ethylene oxide, ie, epoxidation; the following reaction scheme is a good approxi-
mation:

O2 O
CH2 CH2 CH2 CH2

O2 O2

CO2, H2O

The selective oxidation is catalyzed by silver, which is the only good cat-
alyst. Other olefins are not converted selectively to the epoxides in the pre-
sence of silver. However, propylene epoxidation is applied commercially; the
catalysts are either molybdenum complexes in solution or solids containing
titanium.
The ethylene epoxidation catalysts (Fig. 9) are multicomponent mixtures
consisting of a support (a-Al2O3), the catalytically active component (silver par-
ticles), and chemical promoters (alkali metal ions such as Csþand perhaps anions
such as sulfate), and a binder; the older literature also describes textural promo-
ters. Furthermore, trace amounts of chlorine-containing compounds such as
ethylene dichloride are continuously added with the feed; these compounds,
like the alkali metal promoter, increase the selectivity of the catalyst for ethylene
oxide. The data of Figure 19 illustrate the role of an alkali metal promoter (97).
Extremely small amounts of the promoter markedly improve the selectivity of
the catalyst; selectivities as high as 80% are reported. Under conditions of the
catalytic oxidation, the silver surface is covered with a layer of oxygen and
may be more properly described as an oxide than as a metal. There is only an
incomplete understanding of the nature of the catalytic sites, the role of promo-
ters, and the bonding of the reactants to the catalyst surface.
The support needs to be inert, which explains the choice of a-Al2O3; most
metal oxides, including transition aluminas, catalyze unselective oxidation.
Vol. 1 CATALYSIS 545

80

78

76
Selectivity, % ethylene oxide

74

72

70

68
0 20 40 60 80 100 120 140 160
Potassium content, ppm

Fig. 19. Promotion of ethylene oxidation by potassium. The selectivity is the percentage
of ethylene converted to ethylene oxide (97).

The catalyst has a low surface area, 1 m2/g, and large pores to minimize the
influence of intraparticle diffusion, which would reduce the selectivity.
A new process for epoxidation of butadiene with a supported silver catalyst
has recently been announced (98).
Naphtha Reforming and Bifunctional Catalysis. In some supported
metal catalysts the support is not just an inert platform but plays an active
catalytic role. This point is illustrated by catalysts for reforming of naphtha
to make high octane number gasoline, a process that is a classic example of
bifunctional surface catalysis (99). The catalysts consist of metal, originally
platinum (100), but now largely rhenium – platinum or tin – platinum, on a
transition alumina, g-Al2O3 or Z-Al2O3. Platinum is chosen because it is
more active than any metal except iridium in a number of reactions that
increase the octane number of paraffins without substantially changing
their molecular weights. These reactions include dehydrogenation, eg, conver-
sion of methylcyclohexane to toluene, and dehydrocyclization, eg, conversion
of n-heptane to toluene. Skeletal isomerization is also desired, but platinum
has only a low activity for this reaction. Rather, the reaction is acid catalyzed.
Consequently, platinum supported on an amorphous solid acid is a good
catalyst.
The support must not be too strongly or too weakly acidic; chlorided
Al2O3 is optimal. The catalyst works well for reforming although the individual
functions, ie, the metal and the acid, alone do not. The metal alone does not
catalyze the branching reactions; they require an acidic function to gene-
rate carbenium ions, which undergo the desired isomerization. The acidic
function alone is not sufficient to generate the carbenium ions; if it were so
strongly acidic, the catalyst would be deactivated rapidly by carbonaceous
deposits. The metal in the catalyst catalyzes dehydrogenation of paraffins to
546 CATALYSIS Vol. 1

give olefins, which are easily protonated and thereby converted by carbenium
ion routes.
Some reforming processes operate with platinum on alumina catalysts at
5008C and high pressures, eg, 5 MPa (50 atm), with the reactant stream con-
taining predominantly hydrogen (33,99). It is unusual for a process in which
hydrogen is produced to be carried out in the presence of excess hydrogen, but
this is done to minimize catalyst deactivation. Coke causes deactivation, but
the high partial pressure of hydrogen retards coke formation as aromatic coke
precursors and coke are hydrogenated. Catalyst regeneration by controlled
coke burnoff may be needed only about twice a year.
However, there is a disadvantage to the high hydrogen partial pressures
that became especially apparent with the removal of tetraethyllead from gasoline
and an increased need for reforming processes as a source of high octane number
hydrocarbons. Aromatic compounds have high octane numbers, and the dehydro-
genation of naphthenes, ie, cyclic paraffins, to give aromatics is favored thermo-
dynamically at high temperatures and low hydrogen partial pressures.
Therefore, there was an incentive to find a way to operate reforming processes
economically at much lower pressures.
Low pressure operation became routine with the application of new cata-
lysts that are resistant to deactivation and withstand the low pressures. The cat-
alysts are bimetallic, eg, incorporating tin or rhenium as well as platinum
(101,102). The structures of these catalysts are still not well understood, but
they likely involve cationic forms of tin or rhenium, and the latter resists deac-
tivation better than the monometallic catalyst, especially in the presence of
sulfur-containing compounds.
There are now a number of applications of supported bimetallic catalysts. In
some instances, the role of the added metal is to isolate the atoms of the other
metal at the surface, thereby reducing the rates of reactions that require
ensembles of the atoms as the catalytic sites. These reactions are called struc-
ture-sensitive reactions (103). Some reactions, such as olefin hydrogenation,
are structure-insensitive reactions; they seem to be catalyzed by isolated metal
centers on surfaces, much as the Wilkinson hydrogenation is catalyzed by mono-
nuclear metal complexes. Iridium–platinum catalysts have been used for
naphtha reforming, and one of the roles of the platinum may be to combine
with the iridium in alloylike bimetallic clusters to suppress an undesired struc-
ture-sensitive reaction, viz, paraffin hydrogenolysis (103). This reaction takes
place in reforming and is largely undesired because it reduces the molecular
weight range.
4.11. Catalysis by Metal Sulfides. Metal sulfides such as MoS2, WS2,
and many others catalyze numerous reactions that are catalyzed by metals (104).
The metal sulfides are typically several orders of magnitude less active than the
metals, but they have the unique advantage of not being poisoned by sulfur com-
pounds. They are thus good catalysts for applications with sulfur-containing
feeds, including many fossil fuels.
Metal sulfide catalysts are widely applied in hydroprocessing of petroleum.
The reactions include hydrogenation, hydrodesulfurization, and hydrodenitro-
genation (105). Hydrodesulfurization is the reaction of organosulfur compounds
Vol. 1 CATALYSIS 547

with hydrogen to give hydrocarbons and H2S; hydrodenitrogenation gives


hydrocarbons and ammonia. Hydrodesulfurization is carried out to remove sul-
fur from feeds to naphtha reformers because the metals in the reforming cata-
lysts are poisoned by sulfur. Hydrodesulfurization of heavy petroleum fractions
is carried out to minimize formation of SOx resulting from combustion of the
fuels. Hydrodenitrogenation is carried out to remove basic nitrogen-containing
compounds from fuels that are later hydrocracked, as hydrocracking catalysts
have acidic components that are poisoned by the basic nitrogen-containing
compounds.
MoS2 is one of the most active hydroprocessing catalysts, but it is expen-
sive, and the economical way to apply it is as highly dispersed material on
a support, g-Al2O3. The activity of the supported catalyst is increased by the
presence of promoter ions, Co2þ or Ni2þ. The structures of the catalysts are
fairly well understood; the MoS2 is present in extremely small layer structures
only a few atoms thick on the support surface, and the promoter ions are pre-
sent at the edges of the MoS2 layers, where the catalytic sites are located
(106,107).
The catalysts are prepared by impregnating the support with aqueous
salts of molybdenum and the promoter. In acidic solutions, molybdate ions
are present largely in the form of heptamers, [Mo7O24]6 , and the resulting
surface species are believed to be present in islands, perhaps containing only
seven Mo ions. Before use, the catalyst is treated with H2 and some sulfur-
containing compounds, and the surface oxides are converted into the sulfides
that are the catalytically active species.
The applications of supported metal sulfides are unique with respect to
catalyst deactivation phenomena. The catalysts used for processing of petro-
leum residua accumulate massive amounts of deposits consisting of sulfides
formed from the organometallic constituents of the oil, principally nickel and
vanadium (108). These, with coke, cover the catalyst surface and plug the
pores. The catalysts are unusual in that they can function with masses of
these deposits that are sometimes even more than the mass of the original
fresh catalyst. Mass transport is important, as the deposits are typically
formed with effectiveness factors less than unity, and in the extreme case
the deposits block the pore mouths. Modeling of the transport–reaction phe-
nomena has guided the preparation of catalysts with tailored pore structures
to minimize the detriment of the deposits. These have been some of the most
fruitful applications of the principles of chemical engineering in catalyst
design and preparation.

5. Catalyst Development, Testing, and Production

Catalysts are discovered to meet processing needs and opportunities, but the dis-
covery of a catalytic application to take advantage of some newly discovered
material almost never occurs. Catalyst development is largely a matter of trial
and error testing. The methodology was defined by Mittasch in the development
of the ammonia synthesis process. Catalyst developers benefit from an extensive
548 CATALYSIS Vol. 1

and diverse literature and often can formulate good starting points in a search
for candidate catalysts by learning what has been used successfully for similar
reactions. Deeper insights, such as would arise from understanding of the
mechanistic details of a catalytic cycle, are usually not attained; the exceptions
to this rule largely pertain to molecular catalysis, usually reactions occurring in
solution. Fundamental insights were valuable in guiding the development of the
process for chiral hydrogenation and that for methanol carbonylation, among
others, but it would be inappropriate to infer that understanding of the funda-
mental chemistry led to straightforward design of the catalysts. Indeed, the
initial working hypothesis about the chiral hydrogenation turned out to be incor-
rect. The more complicated processes of surface catalysis are for the most part
only partially understood even when the processes are established and extensive
after-the-fact research has been done. Creative research in catalyst discovery
and development is usually the result of intuition and partial understanding
combined with efficient testing and serendipity. Researchers who are repeatedly
successful in finding new and improved catalysts seem to recognize needs and
opportunities and notice significant exceptions to expected patterns and reason
inductively by imperfect analogies.
Catalyst testing and evaluation have been revolutionized by computers,
automated and miniaturized test reactors, and analytical methods. Combinator-
ial methods are playing an increasing role in the preparation and testing with
modern equipment. Researchers can systematically prepare and screen many
catalysts in a short time and efficiently determine, not only the initial catalytic
activity and selectivity, but also the stability and the appearance of trace
products that may indicate some new catalytic properties worthy of further
development.
Catalyst design is in a primitive stage. There are hardly any examples of
true design of catalysts. However, development of improved catalysts has been
guided successfully in instances when the central issues were the interplay of
mass transport and reaction. An example is catalysts used for hydroprocessing
of heavy fossil fuels.
Almost all industrial catalysts are developed by researchers who are moti-
vated to improve processes or create new ones. Thus the organization that first
uses a new catalyst is usually the one that has discovered it. This organization,
however, only rarely becomes the manufacturer of the catalyst used on a large
scale. Catalysts are for the most part highly complex specialty chemicals, and
catalyst manufacturers tend to be more efficient than others in producing
them. Catalyst manufacturing is a competitive industry. Catalyst users often
develop close relations with catalyst manufacturers, and the two may work
together to develop and improve proprietary catalysts.

6. The Catalysis Literature

Catalysis is a broad, complex subject that is documented in many and varied


sources. The field is rich in opportunity, in part because there is so much infor-
mation that it is possible to find nuggets that competitors miss. Industrial cata-
lysis is a competitive field, and much practical knowledge is proprietary.
Vol. 1 CATALYSIS 549

The literature consists of patents, books, journals, and trade literature. The
examples in patents may be especially valuable. The primary literature provides
much catalyst performance data, but there is a lack of quantitative results char-
acterizing the performance of industrial catalysts under industrially realistic
conditions. Characterizations of industrial catalysts are often restricted to phy-
sical characterizations and perhaps activity measurements with pure component
feeds, but it is rare to find data characterizing long-term catalyst performance
with impure, multicomponent industrial feedstocks. Catalyst regeneration proce-
dures are scarcely reported. Those who have proprietary technology are normally
reluctant to make it known. Readers should be critical in assessing published
work that claims a relevance to technology.
Often the catalysts described in the literature are not quite the same as
those used in industrial processes, and often the reported performance is for
pure single-component feeds. Sometimes the best quantitative approximations
that can be made from the available literature are those based on reported
kinetics of reactions with pure feeds and catalysts that are similar to but not
the same as those used in practice. As a first approximation, one may use the
published results and scale the activity on the basis of a few laboratory results
obtained with realistic feeds and commercially available catalysts.
Catalyst suppliers are valuable sources of information. They often have
extensive experience testing catalysts in long-term operation with real industrial
feedstocks and may be willing to share information to improve their chances of
selling a catalyst. Also they may work with a potential customer to develop
catalysts that they could then supply. There is an extensive literature produced
by catalyst manufacturers and organizations that license and market catalytic
technology. This trade literature should be read critically as its purpose is to gen-
erate sales, but it often contains valuable information. Catalyst manufacturers
and those who license and sell technology are motivated to demonstrate their
technical knowledge. Their success in marketing depends on their technical
reputations and their reliability in supplying catalysts and technology that con-
sistently meet specifications.

BIBLIOGRAPHY

‘‘Catalysis’’ in ECT 1st ed., Vol. 3, pp. 245–272 and Suppl. 1, pp. 144–150, by V. I.
Komarewsky, Illinois Institute of Technology, and J. B. Coley, Standard Oil Co. of Indi-
ana; in ECT 2nd ed., Vol. 4, pp. 534–586, by G. A. Mills, Houdry Process and Chemical
Co.; in ECT 3rd ed., Vol. 5, pp. 16–61, by G. A. Mills, U.S. Department of Energy, and
J. A. Cusumano, Catalytica Associates, Inc.; in ECT 4th ed,. Vol. 5 pp. 320–373, by Bruce
C. Gates, University of Delaware; ‘‘Catalysis’’ in ECT (online), posting date: December 4,
2000, by Bruce C. Gates, University of Delaware.

CITED PUBLICATIONS

1. A. Mittasch, Kurze Geschichte der Katalyse in Praxis und Theorie, J. Springer-


Verlag, Berlin, Germany, 1939.
550 CATALYSIS Vol. 1

2. A. Mittasch and E. Thies, Von Davy und Do€bereiner bis Deacon, ein Halbes
Jahrhundert Grenzflächenkatalyse, Verlag Chemie, Berlin, Germany, 1932.
3. L. Wilhelmy, Ann. Physik. Chem. (Poggendorf) 81, 413, 419 (1850).
4. M. Bodenstein, Bericht des V. Internationalen Kongresses für angewandte Chemie zu
Berlin 1903, Sektion X, Band IV, 1904, p. 561.
5. A. Mittasch, Adv. Catal. 2, 81 (1950).
6. S. A. Topham, in J. R. Anderson and M. Boudart, eds., Catalysis–Science and
Technology, Vol. 5, Springer-Verlag, Berlin, Germany, 1984, p. 119.
7. B. Timm, Proceeding of the 8th International Congress on Catalysis (Berlin) 1, 7
(1984).
8. M. Boudart, Kinetics of Chemical Processes, Prentice-Hall, New York, 1968.
9. R. P. Bell, The Proton in Chemistry, Cornell University Press, Ithaca, New York,
1973.
10. B. C. Gates, Catalytic Chemistry, John Wiley & Sons, Inc., New York, 1992.
11. P. K. Ghosh, T. Guha, and A. N. Saha, J. Appl. Chem. 17, 239 (1967).
12. J. I. de Jong and F. H. D. Dethmers, Rec. Trav. Chim. 84, 460 (1965).
13. G. Olah, G. K. S. Prakash, and J. Sommer, Superacids, John Wiley & Sons, Inc.,
New York, 1985.
14. J. P. Collman, L. S. Hegedus, J. R. Norton, and R. G. Finke, Principles and Applica-
tions of Organotransition Metal Chemistry, 2nd ed., University Science Books, Mill
Valley, Calif., 1987.
15. B. R. James, Homogeneous Hydrogenation, John Wiley & Sons, Inc., New York,
1974.
16. J. Halpern, Trans. Am. Crystallogr. Assoc. 14, 59 (1978); J. Halpern, T. Okamoto,
and A. Zakhariev, J. Mol. Catal. 2, 65 (1976); J. Halpern and C. S. Wong,
J. Chem. Soc., Chem. Commun., 629 (1973).
17. W. S. Knowles, M. J. Sabacky, and B. D. Vineyard, Adv. Chem. Ser. 132, 274 (1974).
18. R. Noyori, Asymmetric Catalysis in Organic Synthesis, John Wiley & Sons, Inc.,
New York, 1994.
19. J. Halpern, Science 217, 401 (1982).
20. D. Forster, J. Am. Chem. Soc. 98, 846 (1976).
21. P. M. Maitlis, A. Haynes, G. J. Sunley, and M. J. Howard, J. Chem. Soc. Dalton
Trans. 1996, 2187.
22. J. Hjorkjaer and V. W. Jensen, Ind. Eng. Chem. Prod. Res. Dev. 15, 46 (1976).
23. R. T. Eby and T. C. Singleton, in B. E. Leach, ed., Applied Industrial Catalysis,
Vol. 1, Academic Press, New York, 1983, p. 275.
24. N. Yoneda, S. Kusano, M. Yasui, P. Pujado, and S. Wilcher, Appl. Catal. A Gen. 221,
253 (2001).
25. H.-W. Bohnen and B. Cornils, Adv. Catal., in press.
26. H. Bach, W. Gick, W. Konkol, and E. Wiebus, Proceeding of the 9th International
Congress on Catalysis (Calgary) Vol. 1, 1988 p. 254.
27. I. T. Horvath, Acc. Chem. Res. 31, 641 (1998).
28. J. A. Gladaysz, Pure Appl. Chem. 73, 1319 (2001).
29. R. L. Banks, in Ref. 23, Vol. 3, p. 215.
30. H. S. Eleuterio, J. Mol. Catal. 65, 55 (1991).
31. (a) J. Kress, M. Wesolek, and J. A. Osborn, J. Chem. Soc., Chem. Commun., 514
(1982); (b) N. Calderon, E. A. Ofstead, J. P. Ward, W. A. Judy, and K. W. Scott,
J. Am. Chem. Soc. 90, 4133 (1968).
32. M. Scholl, S. Ding, C. W. Lee, and R. H. Grubbs, Org. Lett. 1, 953 (2000); C. W.
Bielawski and R. H. Grubbs, Angew. Chem. Int. Ed. Engl. 39, 2903 (2000).
33. B. C. Gates, J. R. Katzer, and G. C. A. Schuit, Chemistry of Catalytic Processes,
McGraw-Hill Book Co., Inc., New York, 1979.
Vol. 1 CATALYSIS 551

34. J. E. Lyons, in Ref. 23, Vol. 3, p. 131.


35. R. A. Sheldon and J. K. Kochi, Adv. Catal. 25, 272 (1976).
36. G. W. Parshall, J. Mol. Catal. 4, 243, (1978).
37. G. W. Parshall and S. D. Ittel, Homogeneous Catalysis, 2nd ed., John Wiley & Sons,
Inc., New York, 1992.
38. B. D. Dombek, Adv. Catal. 32, 325 (1983).
39. E. V. Dehmlow and S. S. Dehmlow, Phase-Transfer Catalysis, VCH, Weinheim,
Germany, 1993.
40. C. M. Starks, C. L. Liotta, and M. Halpern, Phase-Transfer Catalysis: Funda-
mentals, Applications, and Industrial Perspectives, Chapman and Hall, New York,
1994.
41. J. N. Armor, Appl. Catal. A: General 222, 407 (2001).
42. J. Wintterlin, Adv. Catal. 45, 131 (2000).
43. (a) S. J. Gregg and K. S. W. Sing, Adsorption, Surface Area, and Porosity, 2nd ed.,
Academic Press, London, UK, 1967. (b) F. Schüth, K. Sing, and J. Weitkamp, eds.,
Handbook of Porous Solids, Wiley-VCH, Weinheim in press.
44. A. J. Lecloux, in Ref. 6, Vol. 2, p. 171.
45. P. H. Emmett, Adv. Catal. 1, 65 (1948).
46. E. W. Thiele, Am. Sci. 55(2), 176 (1967).
47. P. B. Weisz, Science 179, 433 (1973).
48. C. N. Satterfield, Mass Transfer in Heterogeneous Catalysis, MIT Press, Cambridge,
Mass., 1970.
49. B. G. Linsen, J. M. H. Fortuin, C. Okkerse, and J. J. Steggerola, eds., Physical and
Chemical Aspects of Adsorbents and Catalysts, Academic Press, London, 1967.
50. A. B. Stiles, ed., Catalyst Supports and Supported Catalysts, Butterworths, Boston,
Mass., 1987.
51. R. K. Oberlander, in Ref. 23, Vol. 3, p. 64.
52. H.-P. Boehm and H. Knözinger, in Ref. 6, Vol. 4, p. 39.
53. H. Knözinger and P. Ratnasamy, Catal. Rev.-Sci. Eng. 17, 31 (1978).
54. H. H. Lamb, B. C. Gates, and H. Knözinger, Angew. Chem., Int. Ed. Engl. 27, 1127
(1988).
55. W.-D. Mross, Catal. Rev. Sci. Eng. 25, 591 (1983).
56. J. B. Butt and E. E. Petersen, Activation, Deactivation, and Poisoning of Catalysts,
Academic Press, San Diego, Calif., 1988.
57. K. B. Tripplett, in Ref. 23, Vol. 1, p. 178.
58. A. B. Stiles, in Ref. 23, Vol. 2, p. 110.
59. E. Ruckenstein, in S. A. Stevenson, J. A. Dumesic, R. T. K. Baker, and E. Rucken-
stein, eds., Metal-Support Interactions in Catalysis, Sintering, and Redispersion,
Van Nostrand Reinhold Co., Inc., New York, 1987, p. 141.
60. G. Ertl, H. Knözinger, and J. Weitkamp, eds., Preparation of Solid Catalysts, Wiley-
VCH, Weinheim, Germany, 1999.
61. C. G. Brinker and G. W. Scherer, Sol–Gel Science, Academic Press, New York, 1989.
62. J. P. Brunelle, Pure Appl. Chem. 50, 1211 (1978).
63. M. Che, Stud. Surf. Sci. Catal. 130, 115 (2000).
64. M. A. Harmer and Q. Sun, Appl. Catal. A Gen. 221, 45 (2001).
65. R. Thornton and B. C. Gates, J. Catal. 34, 275 (1974).
66. M. P. McDaniel, Adv. Catal. 33, 47 (1985).
67. J. P. Hogan, in Ref. 23, Vol. 1, p. 149.
68. G. G. Hlatky, Chem. Rev. 100, 1347 (2000).
69. P. Cossee, J. Catal. 3, 80 (1964).
70. E. J. Arlman and P. Cossee, J. Catal. 3, 99 (1964).
71. R. Waymouth and P. Pino, J. Am. Chem. Soc. 112, 4911 (1990).
552 CATALYSIS Vol. 1

72. V. Ponec and G. C. Bond, Catalysis by Metals and Alloys, Elsevier, Amsterdam, The
Netherlands, 1995.
73. G. Ertl, Proceedings of the 7th International Congress on Catalysis, Part A, p. 21,
1981.
74. R. Jennings, ed., Ammonia Synthesis—Theory and Practice, Pergamon Press,
1991.
75. P. Stoltze and J. K. Nørskov, Phys. Rev. Lett. 55, 2502 (1985); P. Stoltze, Phys. Scr.
36, 824 (1987); B. Hammer and J. K. Nørskov, Adv. Catal. 45, 71 (2000).
76. J. S. Merriam and K. Atwood, in Ref. 22, Vol. 3, p. 113.
77. V. E. Henrich and P. A. Cox, The Surface Science of Metal Oxides, Cambridge
University Press, Cambridge, 1994.
78. H. Idriss and M. A. Barteau, Adv. Catal. 45, 261 (2000).
79. B. C. Gates, in Handbook of Heterogeneous Catalysis, John Wiley & Sons, Inc., in
press.
80. D. W. Breck, Zeolite Molecular Sieves, Wiley, New York, 1974.
81. N. Y. Chen, W. E. Garwood, and F. G. Dwyer, Shape Selective Catalysis in Industrial
Applications, Dekker, New York, 1996.
82. E. G. Derouane, in M. S. Whittington and A. J. Jacobson, eds., Intercalation
Chemistry, Academic Press, New York, 1982, p. 101.
83. E. M. Flanigen and co-workers, Nature (London) 271, 572 (1978).
84. W. O. Haag, R. M. Lago, and P. B. Weisz, Nature (London) 309, 590 (1984).
85. P. B. Venuto and E. T. Habib, Catal. Rev. Sci. Eng. 18, 1 (1978).
86. B. W. Wojciechowski and A. Corma, Catalytic Cracking: Catalysts, Chemistry, and
Kinetics, Dekker, New York, 1986.
87. I. E. Maxwell, CATTECH 1, 5 (1997).
88. M. E. Davis and R. F. Lobo, Chem. Mater. 4, 756 (1992).
89. B. Notari, Adv. Catal. 41, 253 (1996).
90. S. J. Miller, Stud. Surf. Sci. Catal. 84, 2319 (1994).
91. W. O. Haag and N. Y. Chen, in L. L. Hegedus, ed., Catalyst Design, Progress and
Perspectives, John Wiley & Sons, Inc., New York, 1987, p. 163.
92. P. B. Weisz, CHEMTECH 3., 498 (1973).
93. U.S. Pat. 3,414,631 (Dec. 3,1968), R. K. Grasselli, G. Heights, and J. L. Callahan
(to Standard Oil Co.); U.S. Pat. 3,642,930 (Feb. 15, 1972), R. K. Grasselli, G. Heights,
and H. F. Hardman (to Standard Oil Co.).
94. R. K. Grasselli, Top. Catal. 15, 93 (2001).
95. B. C. Gates, Adv. Chem. Eng. 27, 49 (2001).
96. J. M. Berty, in Ref. 23, Vol. 1, p. 207.
97. U.S. Pat. 4,356,312 (Oct. 26, 1982), R. P. Nielson and J. H. La Rochelle (to Shell Oil Co.).
98. J. R. Monnier, Appl. Catal. A Gen. 221, 73 (2001).
99. G. A. Mills, H. Heinemann, T. H. Milliken, and A. G. Oblad, Ind. Eng. Chem. 45, 134
(1953).
100. M. J. Sterba and V. Haensel, Ind. Eng. Chem. Prod. Res. Dev. 15, 2 (1976).
101. G. Antos, A. M. Aitani, and J. M. Parera, eds., Catalytic Naphtha Reforming: Science
and Technology, Dekker, New York, 1995.
102. U.S. Pat. 3,415,737 (Dec. 10, 1968), H. E. Klucksdahl (to Chevron Research Co.).
103. J. H. Sinfelt, Bimetallic Catalysts, Discoveries, Concepts, and Applications, John
Wiley & Sons, Inc., New York, 1983.
104. O. Weisser and S. Landa, Sulphide Catalysts, Their Properties and Applications,
Pergamon Press, Oxford, UK, 1973.
105. M. J. Girgis and B. C. Gates, Ind. Eng. Chem. Res. 30, 2021 (1991).
106. H. Topsøe, B. S. Clausen, and F. E. Massoth, Hydrotreating Catalysis, Science and
Technology, Springer, Berlin, 1996.
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 553

107. J. V. Lauritsen, S. Helveg, E. Laegsgaard, I. Stensgaard, B. S. Clausen, H. Topsøe,


and E. Besenbacher, J. Catal. 197, 1 (2001).
108. R. J. Quann, R. A. Ware, C.-W. Hung, and J. Wei, Adv. Chem. Eng. 14, 95 (1988).

GENERAL REFERENCES

References 10, 14, 18, 23, 33, 37, 39, 40, 48–50, 56, 72, 77, and 80 are general references.
Also see B. Cornils, and W. A. Herrmann, eds., Applied Homogeneous Catalysis
with Organometallic Compounds, Wiley-VCH, Weinheim, Germany, 2002; G. Ertl,
H. Knözinger, and J. Weitkamp, eds., Handbook of Heterogeneous Catalysis, Wiley-
VCH, Weinheim, Germany, 1997. Advances in Catalysis, Academic Press, New York,
continuing series. Nomenclature used in catalysis is summarized in B. Cornils,
W. A. Herrmann, R. Schlögl, and C. H. Wong, eds., Catalysis from A to Z, Wiley-VCH,
Weinheim, Germany, 2000.

BRUCE C. GATES
University of California, Davis

CATALYST DEACTIVATION
AND REGENERATION
1. Introduction

Catalyst deactivation, the loss over time of catalytic activity and/or selectivity, is
a problem of great and continuing concern in the practice of industrial catalytic
processes. Costs to industry for catalyst replacement and process shutdown total
billions of dollars per year. Time scales for catalyst deactivation vary consider-
ably; for example, in the case of cracking catalysts, catalyst mortality may be
on the order of seconds, while in ammonia synthesis the iron catalyst may last
for 5–10 years. It is inevitable, however, that all catalysts will decay.
Typically, the loss of activity in a well-controlled process occurs slowly.
However, process upsets or poorly designed hardware can bring about cata-
strophic failure. For example, in steam reforming of methane or naphtha great
care must be taken to avoid reactor operation at excessively high temperatures
or at steam-to-hydrocarbon ratios below a critical value. Indeed, these conditions
can cause formation of large quantities of carbon filaments that plug catalyst
pores and voids, pulverize catalyst pellets, and bring about process shutdown,
all within a few hours.
While catalyst deactivation is inevitable for most processes, some of its imme-
diate, drastic consequences may be avoided, postponed, or even reversed. Thus, de-
activation issues (ie, extent, rate, and reactivation) greatly impact research,
development, design, and operation of commercial processes. Accordingly, there

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.1415021218150209.a01.pub2
554 CATALYST DEACTIVATION AND REGENERATION Vol. 1

is considerable motivation to understand and treat catalyst decay. Indeed, over


the past three decades, the science of catalyst deactivation has been steadily
developing, while literature addressing this topic has expanded considerably to
include books (1–4); comprehensive reviews (5–8); proceedings of international
symposia (9–14); topical journal issues (eg, Ref. 15); and more than 7000 patents
for the period of 1976–2001. (In a patent search conducted in April 2001 for the
keywords catalyst and deactivation, catalyst and life, and catalyst and regenera-
tion, 1781, 3134, and 5068 patents were found respectively.) This area of
research provides a critical understanding that is the foundation for modeling
deactivation processes, designing stable catalysts, and optimizing processes to
prevent or slow catalyst deactivation.

2. Mechanisms of Deactivation of Heterogeneous Catalysts

There are many paths for heterogeneous catalyst decay. For example, a catalyst
solid may be poisoned by any one of a dozen contaminants present in the feed. Its
surface, pores, and voids may be fouled by carbon or coke produced by cracking/
condensation reactions of hydrocarbon reactants, intermediates, and/or pro-
ducts. In the treatment of a power plant flue gas, the catalyst can be dusted or
eroded by and/or plugged with fly ash. Catalytic converters used to reduce emis-
sions from gasoline or diesel engines may be poisoned or fouled by fuel or lubri-
cant additives and/or engine corrosion products. If the catalytic reaction is
conducted at high temperatures, thermal degradation may occur in the form of
active phase crystallite growth, collapse of the carrier (support) pore structure,
and/or solid-state reactions of the active phase with the carrier or promoters. In
addition, the presence of oxygen or chlorine in the feed gas can lead to formation
of volatile oxides or chlorides of the active phase, followed by gas-phase transport
from the reactor. Similarly, changes in the oxidation state of the active catalytic
phase can be induced by the presence of reactive gases in the feed.
Thus, the mechanisms of solid catalyst deactivation are many; neverthe-
less, they can be grouped into six intrinsic mechanisms of catalyst decay: (1) poi-
soning, (2) fouling, (3) thermal degradation, (4) vapor compound formation and/
or leaching accompanied by transport from the catalyst surface or particle,
(5) vapor–solid and/or solid–solid reactions, and (6) attrition/crushing. As
mechanisms 1, 4, and 5 are chemical in nature while 2 and 5 are mechanical,
the causes of deactivation are basically threefold: chemical, mechanical, and
thermal. Each of the six basic mechanisms is defined briefly in Table 1. Mechan-
isms 4 and 5 are treated together, since 4 is a subset of 5.
2.1. Poisoning. Poisoning (3,16–22) is the strong chemisorption of reac-
tants, products, or impurities on sites otherwise available for catalysis. Thus,
poisoning has operational meaning; that is, whether a species acts as a poison
depends upon its adsorption strength relative to the other species competing
for catalytic sites. For example, oxygen can be a reactant in partial oxidation
of ethylene to ethylene oxide on a silver catalyst and a poison in hydrogenation
of ethylene on nickel. In addition to physically blocking adsorption sites,
adsorbed poisons may induce changes in the electronic or geometric structure
of the surface (17,21).
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 555

Table 1. Mechanisms of Catalyst Deactivation


Mechanism Type Brief definition/description
poisoning chemical strong chemisorption of species on catalytic
sites which block sites for catalytic reaction
fouling mechanical physical deposition of species from fluid phase
onto the catalytic surface and in catalyst
pores
thermal degradation thermal thermally induced loss of catalytic surface
area, support area, and active phase-
support reactions
vapor formation chemical reaction of gas with catalyst phase to produce
volatile compounds
vapor–solid and chemical reaction of vapor, support, or promoter with
solid–solid catalytic phase to produce inactive phase
reactions
attrition/crushing mechanical loss of catalytic material due to abrasion loss
of internal surface area due to mechanical-
induced crushing of the catalyst particle

Mechanisms by which a poison may affect catalytic activity are multifold as


illustrated by a conceptual two-dimensional model of sulfur poisoning of ethylene
hydrogenation on a metal surface shown in Fig. 1. To begin with, a strongly
adsorbed atom of sulfur physically blocks at least one three- or fourfold
adsorption/reaction site (projecting into three dimensions) and three or four
topside sites on the metal surface. Second, by virtue of its strong chemical
bond, it electronically modifies its nearest neighbor metal atoms and possibly
its next-nearest neighbor atoms, thereby modifying their abilities to adsorb
and/or dissociate reactant molecules (in this case H2 and ethylene molecules),
although these effects do not extend beyond about 5 atomic units (21). A third
effect may be the restructuring of the surface by the strongly adsorbed poison,
possibly causing dramatic changes in catalytic properties, especially for reactions
sensitive to surface structure. In addition, the adsorbed poison blocks access of
adsorbed reactants to each other (a fourth effect) and finally prevents or slows
the surface diffusion of adsorbed reactants (effect number five).
Catalyst poisons can be classified according to their chemical makeup,
selectivity for active sites, and the types of reactions poisoned. Table 2 lists

H H
S
H C C H H H

M M M M M M

M M M M M

Fig. 1. Conceptual model of poisoning by sulfur atoms of a metal surface during ethylene
hydrogenation.
556 CATALYST DEACTIVATION AND REGENERATION Vol. 1

Table 2. Common Poisons Classified According to Chemical Structure


Chemical type Examples Type of interaction with metals
Groups VA and VIA N, P, As, Sb, O, S, Se, Te through s and p orbitals;
shielded structures are less
toxic
Group VII A F, Cl, Br, I through s and p orbitals;
formation of volatile halides
toxic heavy metals As, Pb, Hg, Bi, Sn, Zn, occupy d orbitals; may
and ions Cd, Cu, Fe form alloys
molecules that adsorb CO, NO, HCN, benzene, chemisorption through multiple
with multiple bonds acetylene, other unsa- bonds and back bonding
turated hydrocarbons

four groups of catalyst poisons classified according to chemical origin and their
type of interaction with metals. It should be emphasized that interactions of
Group VA–VIIIA elements with catalytic metal phases depend on the oxidation
state of the former, ie, how many electron pairs are available for bonding and the
degree of shielding of the sulfur ion by ligands (16). Thus, the order of decreasing
toxicity for poisoning of a given metal by different sulfur species is H2S, SO2,
SO42 , ie, in the order of increased shielding by oxygen. Toxicity also increases
with increasing atomic or molecular size and electronegativity, but decreases if
the poison can be gasified by O2, H2O, or H2 present in the reactant stream (21);
for example, adsorbed carbon can be gasified by O2 to CO or CO2 or by H2 to CH4.
Table 3 lists a number of common poisons for selected catalysts in impor-
tant representative reactions. It is apparent that organic bases (eg, amines)
and ammonia are common poisons for acidic solids such as silica–aluminas
and zeolites in cracking and hydrocracking reactions, while sulfur- and

Table 3. Poisons for Selected Catalysts in Important Representative Reactions


Catalyst Reaction Poisons
silica–alumina, cracking organic bases, hydrocarbons
zeolites heavy metals
nickel, platinum, hydrogenation/ compounds of S, P, As, Zn, Hg,
palladium dehydrogenation halides, Pb, NH3, C2H2
nickel steam reforming of methane, H2S, As
naphtha
iron, ruthenium ammonia synthesis O2, H2O, CO, S, C2H2, H2O
cobalt, iron Fischer–Tropsch synthesis H2S, COS, As, NH3, metal
carbonyls
noble metals on hydrocracking NH3, S, Se, Te, P
zeolites
silver ethylene oxidation to ethylene C2H2
oxide
vanadium oxide oxidation/selective catalytic As/Fe, K, Na from fly ash
reduction
platinum, palladium oxidation of CO and Pb, P, Zn, SO2, Fe
hydrocarbons
cobalt and hydrotreating of residues asphaltenes; N, Ni, V
molybdenum sulfides compounds
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 557

arsenic-containing compounds are typical poisons for metals in hydrogenation,


dehydrogenation, and steam reforming reactions. Metal compounds (eg, of Ni,
Pb, V, and Zn) are poisons in automotive emissions control, catalytic cracking,
and hydrotreating. Acetylene is a poison for ethylene oxidation, while asphal-
tenes are poisons in hydrotreating of petroleum residues.
‘‘Selective’’ poisoning involves preferential adsorption of the poison on the
most active sites at low concentrations. If sites of lesser activity are blocked initi-
ally, the poisoning is ‘‘antiselective.’’ If the activity loss is proportional to the con-
centration of adsorbed poison, the poisoning is ‘‘nonselective.’’ An example of
selective poisoning is the deactivation of platinum by CO for the para-H2 conver-
sion (23) while Pb poisoning of CO oxidation on platinum is apparently antiselec-
tive (24), and arsenic poisoning of cyclopropane hydrogenation on Pt is
nonselective (25). For nonselective poisoning the linear decrease in activity
with poison concentration or susceptibility (s) is defined by the slope of the activ-
ity versus poison concentration curve. Several other important terms associated
with poisoning are defined in Table 4. Poison tolerance, the activity at saturation
coverage of the poison, and resistance (the inverse of deactivation rate) are
important concepts that are often encountered in discussions of poisoning includ-
ing those defined in Table 4.
Activity versus poison concentration patterns are based on the assumption
of uniform poisoning of the catalyst surface and surface reaction rate controlling,
ie, negligible pore-diffusional resistance. These assumptions, however, are rarely
met in typical industrial processes because the severe reaction conditions of high
temperature and high pressure bring about a high pore-diffusional resistance for
either the main or poisoning reaction or both. In physical terms, this means that
the reaction may occur preferentially in the outer shell of the catalyst particle, or
that poison is preferentially adsorbed in the outer shell of the catalyst particle, or
both. The nonuniformly distributed reaction and/or poison leads to nonlinear
activity versus poison concentration curves, but do not represent truly selective
or antiselective poisoning. For example, if the main reaction is limited to an outer
shell in a pellet where poison is concentrated, the drop in activity with concen-
tration will be precipitous.
As sulfur poisoning is a difficult problem in many important catalytic pro-
cesses (eg, hydrogenation, methanation, Fischer–Tropsch synthesis, steam
reforming, and fuel cell power production), it merits separate discussion as an

Table 4. Important Poisoning Parameters


Parameter Definition
activity (a) reaction rate at time t relative to that at t ¼ 0
susceptibility negative slope of the activity versus poison concentration curve
(s) [s ¼ (a 1)/C (t)]. Measure of a catalyst’s sensitivity to a given
poison
toxicity susceptibility of a given catalyst for a poison relative to that for
another poison
resistance inverse of the deactivation rate, property that determines how rapidly a
catalyst deactivates
tolerance activity of the catalyst at saturation coverage (some catalysts
(a(Csat)) may have negligible activity at saturation coverage)
558 CATALYST DEACTIVATION AND REGENERATION Vol. 1

example of catalyst poisoning phenomena. Studies of sulfur poisoning in hydro-


genation and CO hydrogenation reactions have been thoroughly reviewed
(8,21,26–30). Much of the previous work focused on poisoning of nickel metal cat-
alysts by H2S, the primary sulfur poison in many important catalytic processes,
and thus provides some useful case studies of poisoning.
Previous adsorption studies (27–29) indicate that H2S adsorbs strongly and
dissociatively on nickel metal surfaces. Extrapolation of high temperature data
to zero coverage using a Tempkin isotherm (28) yields an enthalpy of adsorption
of 250 kJ/mol; in other words, at low sulfur coverages, surface nickel–sulfur
bonds are a factor of 3 more stable than bulk nickel–sulfur bonds. The absolute
heat of adsorption increases with decreasing coverage and the equilibrium
partial pressure of H2S increases with increasing temperature and increasing
coverage. It is expected that H2S (and other sulfur impurities) will adsorb essen-
tially irreversibly to high coverage in most catalytic processes involving metal
catalysts.
Two important keys to reaching a deeper understanding of poisoning phe-
nomena include (1) determining surface structures of poisons adsorbed on metal
surfaces and (2) understanding how surface structure and hence adsorption stoi-
chiometry change with increasing coverage of the poison. Studies of structures of
adsorbed sulfur on single crystal metals (especially Ni) (3,27,31–34) provide such
information. They reveal, for example, that sulfur adsorbs on Ni(100) in an
ordered P(2 2) overlayer, bonded to four Ni atoms at S/Nis < 0.25 and in a
C(2 2) overlayer to two Ni atoms for S/Nis ¼ 0.25–0.50 (see Fig. 2; Nis denotes
a surface atom of Ni); saturation coverage of sulfur on Ni(100) occurs at S/
Nis ¼ 0.5. Adsorption of sulfur on Ni(110), Ni(111), and higher index planes of
Ni is more complicated; while the same P(2 2) structure is observed at low cov-
erage, complex overlayers appear at higher coverages—for example on Ni(111) in
two additional stages (structures) up to saturation at S/Nis ¼ 0.5. In more open
surface structures such as Ni(110) and Ni(210), saturation coverage occurs at
S/Nis ¼ 0.74 and 1.09 respectively; indeed, there is a trend of increasing S/Nis
with decreasing planar density for Ni while the saturation sulfur concentration
remains constant at 44 ng/cm2 Ni (see Table 5).
Reported saturation stoichiometries for sulfur adsorption on polycrystalline
and supported Ni catalysts (S/Nis) vary from 0.25 to 1.3 (27). The values of
saturation coverage greater than S/Nis ¼ 0.5 may be explained by (1) a higher
fractional coverage of sites of lower coordination number, ie, involving more
open planes or intersections of planes (Table 5); (2) enhanced adsorption capacity
at higher gas phase concentrations of H2S in line with the observed trend of
increasing saturation coverage with increasing H2S concentration; and/or
(3) reconstruction of the surface by adsorbed sulfur at higher adsorption
temperatures.
The first effect would be favored, and in fact is observed, for supported cat-
alysts of higher dispersion (27). The second effect may explain the typically lower
observed values of S/Nis for single crystal Ni, which are measured at extremely
low pressures (high vacuum) relative to the higher values of S/Nis for polycrystal-
line and supported Ni, typically measured at orders of magnitude higher pres-
sure; in the case of the single crystal work the surface is not in equilibrium
with gas phase H2S/H2. The third effect, reconstruction of nickel surfaces by
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 559

Ni

(a) p(2×2), S/Nis = 0.25

(b) c(2×2), S/Nis = 0.5

Fig. 2. Schematic view of sulfur adsorbed on a Ni(100) surface at a (a) S/Nis ¼ 0.25 in a
p(2 2) structure and (b) S/Nis ¼ 0.50 in a c(2 2) structure.

adsorbed sulfur, has been reported by a number of workers (27); for example,
McCarroll and co-workers (33,34) found that sulfur adsorbed at near saturation
coverage on a Ni(111) face was initially in a hexagonal pattern but upon heating
above 700 K reoriented to a distorted C(2 2) structure on a Ni(100) layer. In
another study (32), sulfur adsorbed on a Ni(810) caused decomposition to (100)

Table 5. Sulfur Adsorption Densities on Various Crystal Faces of Nickel a


Sulfur conc. at Number of S Number of S atoms per
saturation, atoms/cm2 Ni atoms/cm2 surface Ni
Crystal face ngS/cm2 ( 1015) ( 1015) atoms
(111) 47  1 0.86 1.8 0.48
(100) 43  1 0.80 1.6 0.50
(110) 44.5  1 0.82 1.1 0.74
(210) 42  1 0.78 0.72 1.09
polycrystalline 44.5  1 0.82 — —
a
Data from Ref. 31.
560 CATALYST DEACTIVATION AND REGENERATION Vol. 1

and (410) facets. On the basis of their review of the reconstruction studies,
Bartholomew and co-workers (27) concluded that at high temperatures and
near saturation coverages, restructuring by sulfur of different facets of Ni to
the more stable Ni(100) is probably a general phenomenon. If so, the S/Nis
ratio at saturation would in principle be 0.5 for the reconstructed surface. In
the first example above, restructuring would not affect the S/Nis ratio at satura-
tion, since it is 0.5 for both (100) and (111) planes; however, in the second exam-
ple, the S/Nis ratio at saturation would probably decrease, as rough planes
transform to smoother ones. Nevertheless, the possibility of increases in the S/
Nis ratio at saturation due to reconstruction cannot be ruled out.
The nature of reconstruction of a surface by a poison may depend on its pre-
treatment. For example, in a scanning tunneling microscopy (STM) study of
room temperature H2S adsorption on Ni(110), Ruan and co-workers (35) found
that the S/Ni structure at saturation varied with the initial state of the surface,
ie, whether clean or oxygen covered (see Fig. 3). This study showed that no recon-
struction occurs by direct exposure to H2S at room temperature, rather only in

Fig. 3. A series of in situ STM images recorded after exposure of Ni(110) to oxygen and
then progressively higher exposures of H2S: (a) (2 1)O overlayer; (b) white islands and
black troughs with a C(2 2)S structure after exposure to 3 and 8 L of H2S; (c) 25 L,
islands transform to low-coordinated rows in the [001] direction; and (d) 50 L, stable,
well-ordered (4 1)S (35).
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 561

the presence of O2 (or air). This emphasizes the complexities inherent in predict-
ing the structure and stability of a given poison adsorbed on a given catalyst dur-
ing a specified reaction as a function of different pretreatments or process
disruptions, eg, exposure to air.
It is evident that structure and stoichiometry of sulfur adsorbed on nickel
are complex functions of temperature, H2S concentration, sulfur coverage, and
pretreatment, phenomena that account at least in part for the complex nature
of nickel poisoning by sulfur (27,36). Could one expect similar complexities in
the poisoning of other metals? Probably, since poisoning of nickel is prototypical,
ie, similar principles operate and similar poisoning behaviors are observed in
other poison/metal systems, although none have been studied to the same
depth as sulfur/nickel.
Since one of the necessary steps in a catalytic reaction is the adsorption of
one or more reactants, investigation of the effects of adsorbed sulfur on the
adsorption of other molecules can provide useful insights into the poisoning pro-
cess (21,27). Previous investigations (27,37–43) indicate that both H2 and CO
adsorptions on nickel are poisoned by adsorbed sulfur. Sulfur poisoning can
affect reaction selectivity as well as activity (27).
Because sulfur adsorbs so strongly on metals and prevents or modifies the
further adsorption of reactant molecules, its presence on a catalyst surface
usually effects substantial or complete loss of activity in many important reac-
tions. The steady-state methanation activities of Ni, Co, Fe, and Ru are relative
to the fresh, unpoisoned surface activity as a function of gas phase H2S concen-
tration. Data indicate that Ni, Co, Fe, and Ru all suffer 3–4 orders of magnitude
loss in activity at 15–100 ppb of H2S, ie, their sulfur tolerances are extremely
low. Moreover, the sharp drop in activity with increasing H2S concentration sug-
gests highly selective poisoning. Nevertheless, the rate of sulfur poisoning and
hence sulfur resistance varies from catalyst to catalyst and is apparently a func-
tion of catalyst composition (27) and reaction conditions (44). Indeed, it is possi-
ble to significantly improve sulfur resistance of Ni, Co, and Fe with catalyst
additives such as Mo and B that selectively adsorb sulfur. Because the adsorp-
tion of sulfur compounds is generally rapid and irreversible, surface sulfur con-
centrations in catalyst particles and beds are nonuniform, eg, H2S adsorbs
selectively at the entrance to a packed bed and on the outer surface of catalyst
particles, making the experimental study and modeling of sulfur poisoning extre-
mely difficult.
There are other complications in the study of sulfur poisoning. For example,
the adsorption stoichiometry of sulfur in CO hydrogenation on Ni is apparently a
function of the temperature, H2/CO ratio, and water partial pressure (44). More-
over, at high CO partial pressures sulfur may be removed from the surface as
COS, which is not as strongly adsorbed as H2S. At low temperature conditions,
eg, those representative of Fischer–Tropsch synthesis or liquid phase hydro-
genations, the gas phase concentration of H2S in poisoning studies must be kept
very low, ie, below 0.1–5 ppm, to avoid formation of bulk metal sulfides —
a phenomenon that seriously compromises the validity of the results. Thus, the
importance of studying poisoning phenomena in situ under realistic reaction
conditions, at low process-relevant poison concentrations, and over a process-
representative range of temperature and concentration conditions is emphasized.
562 CATALYST DEACTIVATION AND REGENERATION Vol. 1

There are a number of industrial processes in which one intentionally poi-


sons the catalyst in order to improve its selectivity. For example, to minimize
unwanted cracking reactions, to improve isomerization selectivity, to minimize
coking, etc.
2.2. Fouling, Coking, and Carbon Deposition. Fouling is the physi-
cal (mechanical) deposition of species from the fluid phase onto the catalyst sur-
face, which results in activity loss due to blockage of sites and/or pores. In its
advanced stages it may result in disintegration of catalyst particles and plugging
of the reactor voids. Important examples include mechanical deposits of carbon
and coke in porous catalysts, although carbon- and coke-forming processes also
involve chemisorption of different kinds of carbons or condensed hydrocarbons
that may act as catalyst poisons. The definitions of carbon and coke are some-
what arbitrary and by convention related to their origin. Carbon is typically a
product of CO disproportionation while coke is produced by decomposition or con-
densation of hydrocarbons on catalyst surfaces and typically consists of polymer-
ized heavy hydrocarbons. Nevertheless, coke forms may vary from high
molecular weight hydrocarbons to primarily carbons such as graphite, depending
upon the conditions under which the coke was formed and aged. A number of
books and reviews treat the formation of carbons and coke on catalysts and
the attendant deactivation of the catalysts (1,4,45–50).
The chemical structures of cokes or carbons formed in catalytic processes
vary with reaction type, catalyst type, and reaction conditions. Menon (50) sug-
gested that catalytic reactions accompanied by carbon or coke formation can be
broadly classified as either coke-sensitive or coke-insensitive, analogous to
Boudart’s more general classification of structure-sensitive and structure-
insensitive catalytic reactions. In coke-sensitive reactions, unreactive coke is
deposited on active sites, leading to activity decline, while in coke-insensitive re-
actions, relatively reactive coke precursors formed on active sites are readily
removed by hydrogen (or other gasifying agents). Examples of coke-sensitive
reactions include catalytic cracking and hydrogenolysis; on the other hand,
Fischer–Tropsch synthesis, catalytic reforming, and methanol synthesis are
examples of coke-insensitive reactions. On the basis of this classification
Menon (50) reasoned that the structure and location of a coke are more impor-
tant than its quantity in affecting catalytic activity.
Consistent with Menon’s classification, it is also generally observed that not
only structure and location of coke vary but also its mechanism of formation var-
ies with catalyst type, eg, whether it is a metal or metal oxide (or sulfide, sulfides
being similar to oxides).
Carbon and Coke Formation on Supported Metal Catalysts. Possible
effects of fouling by carbon (or coke) on the functioning of a supported metal cat-
alyst are as follows. Carbon may (1) chemisorb strongly as a monolayer or phy-
sically adsorb in multilayers and in either case block access of reactants to metal
surface sites, (2) totally encapsulate a metal particle and thereby completely
deactivate that particle, and (3) plug micro- and mesopores such that access of
reactants is denied to many crystallites inside these pores. Finally, in extreme
cases, strong carbon filaments may build up in pores to the extent that they
stress and fracture the support material, ultimately causing the disintegration
of catalyst pellets and plugging of reactor voids.
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 563

CO (a) Cα (a) + O (a)

C in Ni Cv

Ni3C (s)

Cα (a) Cβ (s) Cc (s)

CH4 (g)
4 H (a)

Fig. 4. Formation, transformation, and gasification of carbon on nickel (a, g, s refer to


adsorbed, gaseous, and solid states respectively) (48).

Mechanisms of carbon deposition and coke formation on metal catalysts


from carbon monoxide and hydrocarbons (4,45–49) are illustrated in Figs. 4
and 5. Different kinds of carbon and coke that vary in morphology and reactivity
are formed in these reactions. For example, CO dissociates on metals to form Ca,
an adsorbed atomic carbon; Ca can react to Cb, a polymeric carbon film. The more
reactive, amorphous forms of carbon formed at low temperatures (eg, Ca and Cb)
are converted at high temperatures over a period of time to less reactive, graphi-
tic forms (48).
It should also be emphasized that some forms of carbon result in loss of cat-
alytic activity and some do not. For example, at low temperatures (<300–3758C)
condensed polymer or b-carbon films and at high temperatures (>6508C) graphi-
tic carbon films encapsulate the metal surfaces of methanation and steam
reforming catalysts (48). Deactivation of steam reforming catalysts at high

Fig. 5. Formation and transformation of coke on metal surfaces (a, g, s refer to adsorbed,
gaseous, and solid states respectively); gas phase reactions are not considered (48).
564 CATALYST DEACTIVATION AND REGENERATION Vol. 1

reaction temperatures (500–9008C) may be caused by precipitation of atomic


(carbidic) carbon dissolved in the Ni surface layers to a depth of more than
50–70 nm (50,51). If it accumulates on the metal surface (at high or low tempera-
tures), adsorbed atomic carbon can deactivate metal sites for adsorption and/or
reaction. For example, Durer and co-workers (52) demonstrated that carbon
atoms residing in the fourfold hollow sites of Rh(100) block the adsorption of
hydrogen (and hence could block sites for hydrogenation). In the intermediate
temperature range of 375–6508C, carbon filaments are formed by precipitation
of dissolved carbon at the rear side of metal crystallites, causing the metal par-
ticles to grow away from the support (45). Filament growth ceases when suffi-
cient carbon accumulates on the free surface to cause encapsulation by a
carbon layer; however, encapsulation of the metal particles does not occur if
H2/CO or H2O/hydrocarbon ratios are sufficiently high. Thus, carbon filaments
sometimes formed in CO hydrogenation or steam reforming of hydrocarbons
would not necessarily cause a loss of intrinsic catalyst activity unless they are
formed in sufficient quantities to cause plugging of the pores (48) or loss of
metal occurs as the carbon fibers are removed during regeneration (53,54). How-
ever, in practice, regions of carbon forming potential in steam reforming must be
carefully avoided, since once initiated, the rates of filamentous carbon formation
are sufficiently high to cause catastrophic pore plugging and catalyst failure
within a few hours to days.
The rate at which deactivation occurs for a given catalyst and reaction
depends greatly on reaction conditions—especially temperature and reactant
composition. A fundamental principle for coke-insensitive reactions on metals
(eg, methanation, Fischer–Tropsch synthesis, steam reforming, catalytic reform-
ing, and methanol synthesis) is that deactivation rate depends greatly on the dif-
ference in rates of formation and gasification of carbon/coke precursors, ie, rd ¼
rf rg. If the rate of gasification rg is equal to or greater than that of formation rf,
carbon/coke is not deposited. Rates of carbon/coke precursor formation and gasi-
fication both increase exponentially with temperature, although the difference
between them varies a great deal with temperature because of differences in pre-
exponential factors and activation energies. Thus, carbon/coke formation is
avoided in regions of temperature in which precursor gasification rate exceeds
deposition rate. A similar principle operates in steam reforming, ie, at a suffi-
ciently low reaction temperature, the rate of hydrocarbon adsorption exceeds
the rate of hydrocracking and a deactivating polymer film is formed (55); accord-
ingly, it is necessary to operate above this temperature to avoid deactivation.
In steam reforming filamentous carbon formation rate is a strong function
of hydrocarbon structure; for example, it decreases in the order acetylenes, ole-
fins, paraffins, ie, in order of decreasing reactivity, although activation energies
for nickel are in the same range (125–139 kJ) independent of hydrocarbon struc-
ture and about the same as those observed for formation of filamentous carbon
from decomposition of CO (48). This latter observation suggests that the reac-
tions of CO and different hydrocarbons to filamentous carbon proceed by a com-
mon mechanism and rate-determining step—probably the diffusion of carbon
through the metal crystallites (48).
The rate at which a carbon or coke is accumulated in a given reaction under
given conditions can vary significantly with catalyst structure, including metal
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 565

type, metal crystallite size, promoter, and catalyst support. For example, sup-
ported Co, Fe, and Ni are active above 350–4008C for filamentous carbon forma-
tion from CO and hydrocarbons; the order of decreasing activity is reportedly Fe,
Co, Ni (48). Pt, Ru, and Rh catalysts, on the other hand, while equally or more
active than Ni, Co, or Fe in steam reforming produce little or no coke or carbon.
This is attributed to reduced mobility and/or solubility of carbon in the noble
metals, thus retarding the nucleation process. Thus, it is not surprising that
addition of noble metals to base metals retards carbon formation; for example,
addition of Pt in Ni lowers carbon deposition rate during methanation,
while addition of Cu or Au to Ni substantially lowers carbon formation in
steam reforming (48,56). In contrast to the moderating effects of noble metal
additives, addition of 0.5% Sn to cobalt substantially increases the rate of carbon
filament formation from ethylene (57), an effect desirable in the commercial
production of carbon filament fibers.
Since carbon formation and gasification rates are influenced differently by
modifications in metal crystallite surface chemistry, which are in turn a function
of catalyst structure, oxide additives or oxide supports may be used to moderate
the rate of undesirable carbon or coke accumulation. For example, Bartholomew
and Strasburg (58) found the specific rate (turnover frequency) of filamentous
carbon deposition on nickel during methanation at 3508C to decrease in the
order Ni/TiO2, NiAl2O3, Ni/SiO2, while Vance and Bartholomew (59) observed
Ca hydrogenation rates at 1708C to decrease in this same order (the same as
for methanation at 2258C). This behavior was explained in terms of promotional
or inhibiting effects due to decoration of metal crystallites by the support, for
example silica, inhibiting both CO dissociation and carbon hydrogenation. This
hypothesis is consistent with observations (60,61) that silica evaporated on metal
surfaces and supported metals inhibits formation of filamentous carbon. Simi-
larly Bitter and co-workers (62) observed rates of carbon formation in CO2/CH4
reforming to decrease in the order Pt/g-Al2O3 ! Pt/TiO2 > Pt/ZrO2; while 90% of
the carbon deposited on the support, the authors linked deactivation to carbon
accumulated on the metal owing to an imbalance between carbon formed by
methane dissociation and oxidation by chemisorbed CO2. The rate of formation
of coke in steam reforming is delayed and occurs at lower rates in nickel catalysts
promoted with alkali or supported on basic MgO (63).
Since formation of coke, graphite, or filamentous carbon involves the forma-
tion of C C bonds on multiple atoms sites, one might expect that coke or carbon
formation on metals is structure-sensitive, ie, sensitive to surface structure and
metal crystallite size. Indeed, Bitter and co-workers (62) found that catalysts
containing larger Pt crystallites deactivate more rapidly than those containing
small crystallites. Moreover, a crystallite size effect, observed in steam reforming
of methane on nickel (48,63), appears to operate in the same direction, ie, forma-
tion of filamentous carbon occurs at lower rates in catalysts containing smaller
metal crystallites.
In summary, deactivation of supported metals by carbon or coke may occur
chemically owing to chemisorption or carbide formation or physically and
mechanically owing to blocking of surface sites, metal crystallite encapsulation,
plugging of pores, and destruction of catalyst pellets by carbon filaments. Block-
ing of catalytic sites by chemisorbed hydrocarbons, surface carbides, or relatively
566 CATALYST DEACTIVATION AND REGENERATION Vol. 1

reactive films is generally reversible in hydrogen, steam, CO2, or oxygen.


Further details of the thermodynamics, kinetics, and mechanisms of carbon
and coke formation in methanation and steam reforming reactions are available
in reviews by Bartholomew (48) and Rostrup-Nielsen (55,63).
Coke Formation on Metal Oxide and Sulfide Catalysts. In reactions
involving hydrocarbons, coke may be formed in the gas phase and on both non-
catalytic and catalytic surfaces. Nevertheless, formation of coke on oxides and
sulfides is principally a result of cracking reactions involving coke precursors
(typically olefins or aromatics) catalyzed by acid sites (64,65). Dehydrogenation
and cyclization reactions of carbocation intermediates formed on acid sites lead
to aromatics, which react further to higher molecular weight polynuclear aro-
matics and condense as coke.
Olefins, benzene and benzene derivatives, and polynuclear aromatics are
precursors to coke formation. However, the order of reactivity for coke formation
is clearly structure dependent, ie, decreases in the order polynuclear aromatics >
aromatics > olefins > branched alkanes > normal alkanes. For example, the
weight percent coke formed on silica–alumina at 5008C is 0.06, 3.8, 12.5,
and 23% for benzene, naphthalene, fluoranthene, and anthracene respectively
(66).
Coking reactions in processes involving heavy hydrocarbons are very com-
plex; different kinds of coke may be formed and they may range in composition
from CH to C and have a wide range of reactivities with oxygen and hydrogen
depending upon the time on stream and temperature to which they are exposed.
For example, coke deposits occurring in hydrodesulfurization of residues have
been classified into three types (67):

1. Type I deposits are reversibly adsorbed normal aromatics deposited during


the first part of the cycle at low temperature.
2. Type II deposits are reversibly adsorbed asphaltenes deposited early in the
coking process.
3. Type III deposits result from condensation of aromatic concentrates into
clusters and then crystals that constitute a ‘‘mesophase.’’ This crystalline
phase is formed after long reaction times at high temperature. This har-
dened coke causes severe deactivation of the catalyst (67).

In addition to hydrocarbon structure and reaction conditions, extent and


rate of coke formation are also a function of the acidity and pore structure of
the catalyst. Generally, the rate and extent of coke formation increase with
increasing acid strength and concentration. Coke yield decreases with decreasing
pore size (for a fixed acid strength and concentration); this is especially true in
zeolites where shape selectivity plays an important role in coke formation. How-
ever, in pores of molecular diameter, a relatively small quantity of coke can cause
substantial loss of activity. It should be emphasized that coke yield can vary con-
siderably into the interior pores of a catalyst particle or along a catalyst bed,
depending upon the extent to which the main and deactivation reactions are
affected by film mass transport and pore diffusional resistance.
The mechanisms by which coke deactivates oxide and sulfide catalysts are,
as in the case of supported metals, both chemical and physical. However, some
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 567

aspects of the chemistry are quite different. The principal chemical loss of activ-
ity in oxides and sulfides is due to the strong adsorption of coke molecules on
acidic sites. But as discussed earlier, strong acid sites also play an important
role in the formation of coke precursors, which subsequently undergo condensa-
tion reactions to produce large polynuclear aromatic molecules that physically
coat catalytic surfaces. Physical loss of activity also occurs as coke accumulates,
ultimately partially or completely blocking catalyst pores as in supported
metal catalysts. For example, in isomerization of cis-butene on SiO2/Al2O3 (68)
catalyst deactivation occurs by rapid, selective poisoning of strong acid sites;
coke evolved early in the reaction is soluble in dichloromethane and pyridine
and is slightly aromatic. Apparently, the blocking of active sites does not signifi-
cantly affect porosity or catalyst surface area, as SiO2/Al2O3 contains relatively
large mesopores.
In the case of supported bifunctional metal/metal oxide catalysts, different
kinds of coke are formed on the metal and the acidic oxide support, eg, soft coke
(high H/C ratio) on Pt or Pt–Re metals and hard coke (low H/C ratio) on the alu-
mina support in catalytic reforming (69). In this case coke precursors may be
formed on the metal via hydrogenolysis, following which they migrate to the sup-
port and undergo polymerization and cyclization reactions, after which the larger
molecules are dehydrogenated on the metal and finally accumulate on the sup-
port, causing loss of isomerization activity. Mild sulfiding of these catalysts
(especially Pt–Re/alumina) substantially reduces the rate of hydrogenolysis
and the overall formation of coke on both metal and support; it especially reduces
the hard coke, which is mainly responsible for deactivation.
Several studies (65,70–80) have focused on coke formation during hydro-
carbon reactions in zeolites including (1) the detailed chemistry of coke precur-
sors and coke molecules formed in zeolite pores and pore intersections (or
supercages) and (2) the relative importance of adsorption on acid sites versus
pore blockage. The principal conclusions from these studies can be summarized
as follows: (1) the formation of coke and the manner in which it deactivates a zeo-
lite catalyst are shape-selective processes, (2) deactivation is mainly due to the
formation and retention of heavy aromatic clusters in pores and pore intersec-
tions, and (3) while both acid-site poisoning and pore blockage participate in the
deactivation, the former dominates at low coking rates, low coke coverages (eg,
in Y-zeolite below 2 wt%), and high temperatures, while the latter process dom-
inates at high reaction rates, low temperatures, and high coke coverages. Thus,
pore size and pore structure are probably more important than acid strength and
density under typical commercial process conditions. Indeed, deactivation is typi-
cally more rapid in zeolites having small pores or apertures and/or a monodimen-
sional structure (78). Fig. 6 illustrates four possible modes of deactivation of
HZSM-5 by carbonaceous deposits with increasing severity of coking (78).
These conclusions (in the previous paragraph) are borne out, for example,
in the study by Cerqueira and co-workers (80) of USHY zeolite deactivation dur-
ing methylcyclohexane transformation at 4508C, showing the following:

1. Coke is probably mainly formed by rapid transformation of styrenic C7 car-


benium ions with lesser contributions from reactions of cyclopentadiene,
C3 –C6 olefins, and aromatics.
568 CATALYST DEACTIVATION AND REGENERATION Vol. 1

Mode 1 Mode 2

Mode 3 Mode 4

Fig. 6. Schematic of the four possible modes of deactivation by carbonaceous deposits in


HZSM-5: (1) reversible adsorption on acid sites, (2) irreversible adsorption on sites with
partial blocking of pore intersections, (3) partial steric blocking of pores, and (4) extensive
steric blocking of pores by exterior deposits (78).

2. Soluble coke consists of polynuclear aromatic clusters containing three to


seven five- and six-membered rings having typical compositions of C30H40
to C40H44 and having dimensions of 0.9 1.1 nm to 1.1 1.5 nm, ie, sizes
that would cause them to be trapped in the supercages of Y-zeolite.
3. At short contact times, coking is relatively slow and deactivation is mainly
due to acid-site poisoning, while at long contact times, coking is much
faster because of the high concentrations of coke precursors; under these
latter conditions coke is preferentially deposited at the outer pore open-
ings of zeolite crystallites and deactivation is dominated by pore-mouth
blockage.

That coke formed at large contact times not only blocks pores and/or pore
intersections inside the zeolite but also migrates to the outside of zeolite crystal-
lites where it blocks pore entrances has been observed in several studies
(74,76,77,80). However, the amount, structure, and location of coke in ZSM-5
depends strongly on the coke precursor, eg, coke formed from mesitylene is
deposited on the external zeolite surface whereas coking with isobutene leads
to largely paraffinic deposits inside pores; coke from toluene, on the other
hand, is polyaromatic and is deposited both on external and internal zeolite
surfaces (74).
2.3. Thermal Degradation and Sintering. Background. Thermally
induced deactivation of catalysts results from (1) loss of catalytic surface
area due to crystallite growth of the catalytic phase, (2) loss of support area
due to support collapse and of catalytic surface area due to pore collapse on
crystallites of the active phase, and/or (3) chemical transformations of catalytic
phases to noncatalytic phases. The first two processes are typically referred
to as ‘‘sintering.’’ Sintering processes generally take place at high reaction
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 569

temperatures (eg, >5008C) and are generally accelerated by the presence of


water vapor.
Most of the previous sintering and redispersion work has focused on sup-
ported metals. Experimental and theoretical studies of sintering and redisper-
sion of supported metals published before 1997 have been reviewed fairly
extensively (8,81–90). Three principal mechanisms of metal crystallite growth
have been advanced: (1) crystallite migration, (2) atomic migration, and (3) (at
very high temperatures) vapor transport. Crystallite migration involves the
migration of entire crystallites over the support surface, followed by collision
and coalescence. Atomic migration involves detachment of metal atoms or mole-
cular metal clusters from crystallites, migration of these atoms over the support
surface, and ultimately, capture by larger crystallites. Redispersion, the reverse
of crystallite growth in the presence of O2 and/or Cl2, may involve (1) formation of
volatile metal oxide or metal chloride complexes that attach to the support and
are subsequently decomposed to small crystallites upon reduction and/or (2) for-
mation of oxide particles or films that break into small crystallites during subse-
quent reduction.
There is controversy in the literature regarding which mechanism of sinter-
ing (or redispersion) operates at a given set of conditions. Logically, atomic
migration would be favored at lower temperatures than crystallite migration,
since the higher diffusivities of atoms or small clusters would facilitate their
migration, whereas the thermal energy necessary to induce motion of larger
crystallites would only be available at higher temperatures. Moreover, migration
of small crystallites might be favorable early in the sintering process but
unfavorable as crystallites become larger. However, fixing on only one of the
three sintering mechanisms (and two dispersion mechanisms) is a simplification
that ignores the possibility that all mechanisms may occur simultaneously and
may be coupled with each other through complex physicochemical processes
including the following: (1) dissociation and emission of metal atoms or metal-
containing molecules from metal crystallites, (2) adsorption and trapping of
metal atoms or metal-containing molecules on the support surface, (3) diffusion
of metal atoms, metal-containing molecules and/or metal crystallites across sup-
port surfaces, (4) metal or metal oxide particle spreading, (5) support surface
wetting by metal particles, (6) metal particle nucleation, (7) coalescence of, or
bridging between, two metal particles, (8) capture of atoms or molecules by
metal particles, (9) liquid formation, (10) metal volatilization through volatile
compound formation, (11) splitting of crystallites in O2 atmosphere owing to for-
mation of oxides of a different specific volume, and (12) metal atom vaporization.
Depending upon reaction or redispersion conditions, a few or all of these
processes may be important; thus, the complexity of sintering/redispersion pro-
cesses is emphasized.
In general, sintering processes are kinetically slow (at moderate reaction
temperatures) and irreversible or difficult to reverse. Thus, sintering is more
easily prevented than cured.
Factors Affecting Metal Particle Growth and Redispersion in Supported
Metals. Temperature, atmosphere, metal type, metal dispersion, promoters/
impurities and support surface area, texture, and porosity are the principal para-
meters affecting rates of sintering and redispersion (see Table 6) (8,86–90).
570 CATALYST DEACTIVATION AND REGENERATION Vol. 1

Table 6. Effects of Important Reaction and Catalyst Variables on Sintering Rates of


Supported Metals Based on GPLE Dataa
Variable Effect
tempera- sintering rates are exponentially dependent on T; Eact varies from 30 to
ture 150 kJ/mol. Eact decreases with increasing metal loading; it increases in
the following order with atmosphere: NO, O2, H2, N2
atmosphere sintering rates are much higher for noble metals in O2 than in H2 and
higher for noble and base metals in H2 relative to N2; sintering rate
decreases for supported Pt in atmospheres in the following order: NO,
O2, H2, N2
metal observed order of decreasing thermal stability in H2 is Ru > Ir ffi Rh > Pt;
thermal stability in O2 is a function of (1) volatility of metal oxide and
(2) strength of metal oxide–support interaction
support metal–support interactions are weak (bond strengths of 5–15 kJ/mol);
with a few exceptions, thermal stability for a given metal decreases
with
support in the following order: Al2O3 > SiO2 > carbon
promoters some additives decrease atom mobility, eg, C, O, CaO, BaO, CeO2, GeO2;
others increase atom mobility, eg, Pb, Bi, Cl, F, or S; oxides of Ba, Ca, or
Sr are ‘‘trapping agents’’ that decrease sintering rate
pore size sintering rates are lower for porous versus nonporous supports; they
decrease as crystallite diameters approach those of the pores
a
Refs. 8 and 86–90.

Sintering rates increase exponentially with temperature. Metals sinter relatively


rapidly in oxygen and relatively slowly in hydrogen, although depending upon
the support, metal redispersion can be facilitated by exposure at high tempera-
ture (eg, 500–5508C for Pt/Al2O3) to oxygen and chlorine, followed by reduction.
Water vapor also increases the sintering rate of supported metals.
Normalized dispersion (percentage of metal exposed at any time divided by
the initial percentage exposed) versus time data show that at temperatures of
6508C or higher, rates of metal surface area loss (measured by hydrogen chemi-
sorption) due to sintering of Ni/silica in hydrogen atmosphere are significant,
causing 70% loss of the original metal surface area within 50 h at 7508C. In redu-
cing atmosphere, metal crystallite stability generally decreases with decreasing
metal melting temperature, ie, in the order Ru > Ir > Rh > Pt > Pd > Ni > Cu >
Ag, although this order may be affected by relatively stronger metal–support
interactions, eg, the observed order of decreasing stability of supported platinum
in vacuum is Pt/Al2O3 > Pt/SiO2 > Pt/C. In oxidizing atmospheres, metal crystal-
lite stability depends on the volatility of metal oxides and the strength of the
metal–oxide–support interaction. For noble metals, metal stability in air
decreases in the order Rh > Pt > Ir > Ru; formation of volatile RuO4 accounts
for the relative instability of ruthenium (91).
The effect of temperature on sintering of metals and oxides can be under-
stood physically in terms of the driving forces for dissociation and diffusion
of surface atoms, which are both proportional to the fractional approach to the
absolute melting point temperature (Tmp). Thus, as temperature increases, the
mean lattice vibration of surface atoms increases; when the Hüttig tempera-
ture (0.3Tmp) is reached less strongly bound surface atoms at defect sites (eg,
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 571

edges and corner sites) dissociate and diffuse readily over the surface, while at
the Tamman temperature (0.5Tmp) atoms in the bulk become mobile. Accord-
ingly, sintering rates of a metal or metal oxide are significant above the Hüttig
temperature and very high near the Tamman temperature; thus, the relative
thermal stability of metals or metal oxides can be correlated in terms of the Hüt-
tig or Tamman temperatures (92). For example, sintering of copper catalysts for
methanol synthesis is promoted by traces of chlorine in the feed, which react at
about 2258C (500 K) with the active metal/metal oxide surface to produce a
highly mobile copper chloride phase having a Tamman temperature of only
79–1748C (352–447 K) relative to 405–5278C (678–800 K) for copper metal or
metal oxides (93).
Promoters or impurities affect sintering and redispersion by either increas-
ing (eg, chlorine and sulfur) or decreasing (eg, oxygen, calcium, cesium) metal
atom mobility on the support; in the latter case this is due to their high
resistance to dissociation and migration due to high melting points as well
as their hindering dissociation and surface diffusion of other atoms. Similarly,
support surface defects or pores impede surface migration of metal particles —
especially micropores and mesopores with pore diameters about the same size
as the metal crystallites.
Historically, sintering rate data were fitted to a simple power-law expres-
sion (SPLE) of the form

dðD=D0 Þ=dt ¼ ks ðD=D0 Þn ð1Þ

where ks is the sintering rate constant, D0 the initial dispersion, and n the sin-
tering order, which for typical catalyst systems may vary from 3 to 15; unfortu-
nately, the SPLE is in general not valid for sintering processes because it
assumes that surface area or dispersion ultimately reaches zero, given sufficient
time, when in fact, for a given temperature and atmosphere, a nonzero or limit-
ing dispersion is observed after long sintering times. Moreover, the use of the
SPLE is further questionable because variations in sintering order are observed
as a function of time and temperature for a given catalyst in a fixed atmosphere
(88–90); thus, data obtained for different samples and different reaction condi-
tions cannot be quantitatively compared. Nevertheless, it has been shown by
Fuentes (94) and Bartholomew (87–89) that the effects of temperature, atmo-
sphere, metal, promoter, and support can be quantitatively determined by fitting
sintering kinetic data to the general power-law expression (GPLE)

dðD=D0 Þ=dt ¼ ks ðD=D0 Deq =D0 Þm ð2Þ

which adds a term Deq/D0 to account for the observed asymptotic approach of
the typical dispersion versus time curve to a limiting dispersion Deq at infinite
time; m, the order of sintering, is found to be either 1 or 2. A recently compiled,
comprehensive quantitative treatment of previous sintering rate data based on
the GPLE with an order m of 2 (87–89) quantitatively addresses the effects of
catalyst properties and reaction conditions on sintering rate (91,95–97).
Sintering studies of supported metals are generally of two types: (1) studies
of commercially relevant supported metal catalysts and (2) studies of model
572 CATALYST DEACTIVATION AND REGENERATION Vol. 1

metal–support systems. The former type provides useful rate data that can be
used to predict sintering rates, while the latter type provides insights into the
mechanisms of metal particle migration and sintering, although the results can-
not be quantitatively extrapolated to predict behavior of commercial catalysts.
There is direct evidence from the previous studies of model-supported catalysts
(87,90) for the occurrence of crystallite migration (mainly in well-dispersed sys-
tems early in the sintering process), atomic migration (mainly at longer sintering
times), and spreading of metal crystallites (mainly in oxygen atmosphere). There
is also evidence that under reaction conditions, the surface is dynamic, ie, adsor-
bates and other adatoms rapidly restructure the surface and slowly bring about
faceting; moreover, thermal treatments cause gradual changes in the distribu-
tion of coordination sites to minimize surface energy. There is a trend in increas-
ing sophistication of spectroscopic tools used to study sintering and redispersion.
Additional insights into atomic and molecular processes during reaction at the
atomic scale using STM, analytical HRTEM, and other such powerful surface
science tools are expected during the next decade.
Sintering of Catalyst Carriers. Sintering of carriers has been reviewed by
Baker and co-workers (86) and Trimm (98). Single-phase oxide carriers sinter by
one or more of the following processes: (1) surface diffusion, (2) solid-state diffu-
sion, (3) evaporation/condensation of volatile atoms or molecules, (4) grain
boundary diffusion, and (5) phase transformations. In oxidizing atmospheres,
g-alumina and silica are the most thermally stable carriers; in reducing atmo-
spheres, carbons are the most thermally stable carriers. Additives and impurities
affect the thermal properties of carriers by occupying defect sites or forming new
phases. Alkali metals, for example, accelerate sintering, while calcium, barium,
nickel, and lanthanum oxides form thermally stable spinel phases with alumina.
Steam accelerates support sintering by forming mobile surface hydroxyl groups
that are subsequently volatilized at higher temperatures. Chlorine also promotes
sintering and grain growth in magnesia and titania during high temperature cal-
cination (99). By contrast, sulfuric acid treatment of hydrated alumina (gibbsite)
followed by two-step calcination results in a very stable transitional alumina
with needle-like particle morphology (98). Dispersed metals in supported metal
catalysts can also accelerate support sintering; for example, dispersed nickel
accelerates the loss of Al2O3 surface area in Ni/Al2O3 catalysts.
Effects of Sintering on Catalyst Activity. Baker and co-workers (86) have
reviewed the effects of sintering on catalytic activity. Specific activity (based on
catalytic surface area) can either increase or decrease with increasing metal
crystallite size during sintering if the reaction is structure-sensitive, or it can
be independent of changes in metal crystallite size if the reaction is structure-
insensitive. Thus, for a structure-sensitive reaction, the impact of sintering
may be either magnified or moderated; while for a structure insensitive-reaction,
sintering has in principle no effect on specific activity (per unit surface area). In
the latter case, the decrease in mass-based activity is proportional to the
decrease in metal surface area. Ethane hydrogenolysis and ethane steam reform-
ing are examples of structure-sensitive reactions, while CO hydrogenation on
supported cobalt, nickel, iron, and ruthenium is structure-insensitive.
2.4. Gas/Vapor–Solid and Solid-State Reactions. In addition to poi-
soning, there are a number of chemical routes leading to catalyst deactivation:
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 573

(1) reactions of the vapor phase with the catalyst surface to produce (a) inactive
bulk and surface phases (rather than strongly adsorbed species) or (b) volatile
compounds that exit the catalyst and reactor in the vapor phase; (2) catalytic
solid-support or catalytic solid-promoter reactions, and (3) solid-state transfor-
mations of the catalytic phases during reaction.
Gas/Vapor–Solid Reactions. Reactions of Gas/Vapor with Solid to
Produce Inactive Phases. Dispersed metals, metal oxides, metal sulfides,
and metal carbides are typical catalytic phases, the surfaces of which are similar
in composition to the bulk phases. For a given reaction, one of these catalyst
types is generally substantially more active than the others, eg, only Fe and
Ru metals are active for ammonia synthesis, while the oxides, sulfides, and car-
bides are inactive. If, therefore, one of these metal catalysts is oxidized, sulfided,
or carbided, it will lose essentially all of its activity. While these chemical modi-
fications are closely related to poisoning, the distinction here is that rather than
losing activity owing to the presence of an adsorbed species, the loss of activity is
due to the formation of a new phase altogether.
Examples of vapor-induced chemical transformations of catalysts to inac-
tive phases are listed in Table 7.
Reactions of Gas/Vapor with Solid to Produce Volatile Compounds.
Metal loss through direct vaporization is generally an insignificant route to
catalyst deactivation. By contrast, metal loss through formation of volatile
compounds, eg, metal carbonyls, oxides, sulfides, and halides in CO, O2, H2S,
and halogen-containing environments, can be significant over a wide range of
conditions, including relatively mild conditions. Classes and examples of volatile
compounds are listed in Table 8.
While the chemical properties of volatile metal carbonyls, oxides, and
halides are well known, there is surprisingly little information available on
their rates of formation during catalytic reactions. There have been no reviews
on this subject and relatively few reported studies to define the effects of metal
loss on catalytic activity (27,108–121). Most of the previous work has focused on
volatilization of Ru in automotive converters (108–111); nickel carbonyl forma-
tion in nickel catalysts during methanation of CO (113,119) or during CO chemi-
sorption at 258C (27,115); formation of Ru carbonyls during Fischer–Tropsch
synthesis (116,117); volatilization of Pt during ammonia oxidation on Pt–Rh
gauze catalysts (120,121); and volatilization of Cu from methanol synthesis
and diesel soot oxidation catalysts, leading to sintering in the former and better
catalyst–soot contact but also metal loss in the latter case (92).
Results of selected studies are summarized in Table 9.
Loss of nickel metal during CO chemisorption on nickel catalysts at tem-
peratures above 08C is also a serious problem; moreover, this loss is catalyzed
by sulfur poisoning (27). In view of the toxicity of nickel tetracarbonyl, the
rapid loss of nickel metal, and the ill-defined adsorption stoichiometries,
researchers are advised to avoid using CO chemisorption for measuring nickel
surface areas; instead, hydrogen chemisorption, an accepted ASTM method
with a well-defined adsorption stoichiometry, is recommended (124).
Decomposition of volatile platinum oxide species formed during high tem-
perature reaction may (125–127) lead to formation of large Pt crystallites and/
or substantial restructuring of the metal surface. For example, Wu and Phillips
Table 7. Examples of Reactions of Gases/Vapors with Catalytic Solids to Produce Inactive Phases
Catalytic process Gas/vapor composition Catalytic solid Deactivating chemical reaction Ref.
auto emissions control N2, O2, HCs, CO, NO, Pt–Rh/Al2O3 2 Rh2O3 þ g-Al2O3 ! RhAl2O4 þ 0.5 100,101
H2O, SO2 O2
ammonia synthesis and H2, N2 Fe/K/Al2O3 Fe ! FeO at >50 ppm O2 8
regeneration
Traces O2, H2O Fe ! FeO at >0.16 ppm H2O/H2
catalytic cracking HCs, H2, H2O La-Y-zeolite H2O induced Al migration from 8
zeolite framework causing zeolite
destruction
CO oxidation, gas tur- N2, O2, 400 ppm CO, Pt/Al2O3 2 SO3 þ g-Al2O3 ! Al2(SO4)3 which 8
bine exhaust 100–400 ppm SO2 blocks catalyst pores
diesel HC/soot emissions N2, O2, HCs (gas and Pt/Al2O3 and b-zeolite; formation of Al2(SO4)3 or sulfates of 102–104

574
control liquid), CO, NO, oxides of CaCuFeVK Ca, Cu, Fe, or V which block cat-
H2O, soot, SO2 on TiO2 alysts pores and lower activity for
oxidation; Al2O3 stabilized by
BaO
Fischer–Tropsch CO, H2, H2O, CO2, HCs Fe/K/Cu/SiO2 Fe5C2 ! Fe3O4 due to oxidation at 105
high XCO by-product H2O, CO2
Fischer–Tropsch CO, H2, H2O, HCs Co/SiO2 Co þ SiO2 ! CoO  SiO2 and collapse 106
of SiO2 by-product H2O
selective catalytic N2, O2, NO, PM,a H2O, V2O5/WO3/TiO2 formation of Al2(SO4)3 if Al2O3 is 107
reduction (SCR), SO2 used
stationary
steam reforming and CH4, H2O, CO, H2, CO2 Ni/Al2O3 Ni þ Al2O3 ! Ni2Al2O4 8
regeneration in H2O
a
Particulate matter.
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 575

Table 8. Types and Examples of Volatile Compounds Formed in Catalytic Reactions


Gaseous
environment Compound type Example of compound
CO, NO carbonyls and nitrosyl carbonyls Ni(CO)4, Fe(CO)5 (0–3008C)a
O2 oxides RuO3 (258C), PbO (> 8508C),
PtO2 (>7008C)
H2S sulfides MoS2 (>5508C)
halogens halides PdBr2, PtCl4, PtF6, CuCl2,
Cu2Cl2
a
Temperatures of vapor formation are listed in parentheses.

Table 9. Documented Examples of Reactions of Vapor with Solid to Produce


Volatile Compounds
Catalytic Catalytic Vapor Comments on
process solid formed deactivation process Ref.
automotive Pd–Ru/Al2O3 RuO4 50% loss of Ru during 100-h 111
converter test in reducing automotive
exhaust
methanation Ni/Al2O3 Ni(CO)4 PCO > 20 kPa and T < 4258C 113
of CO due to Ni(CO)4 formation,
diffusion and decomposition
on the support as large
crystallites
CO chemi- Ni catalysts Ni(CO)4 PCO > 0.4 kPa and T > 08C 114
sorption due to Ni(CO)4 formation;
catalyzed by sulfur
compounds
Fischer– Ru/NaY zeolite Ru(CO)5, loss of Ru during FTS (H2/ 116,117
Tropsch Ru/Al2O3, Ru3(CO)12 CO ¼ 1, 200–2508C, 1 atm)
synthesis Ru/TiO2 on Ru/NaY zeolite and Ru/
(FTS) Al2O3; up to 40% loss while
flowing CO at 175–2758C
over Ru/Al2O3; for 24 h, rate
of Ru loss less on
titania-supported Ru and
for catalysts containing large
metal crystallites
(3 nm) relative to small
metal crystallites (1.3 nm);
surface carbon lowers loss
ammonia Pt–Rh gauze PtO2 loss: 0.05–0.3 g Pt/ton HNO3; 8,120,122
oxidation recovered with Pd gauze;
loss of Pt leads to surface
enrichment with inactive Rh
HCN Pt–Rh gauze PtO2 extensive restructuring and 8,123
synthesis loss of mechanical strength
methanol CuZnO CuCl2, mobile copper chloride phase 92
synthesis Cu2Cl2 leads to sintering at reaction
temperature (2258C)
diesel soot oxides of K, Cu, CuCl2, mobile copper chloride improves 92
oxidation Mo, and Cu2Cl2 catalyst–soot contact; catalyst
trace Cl evaporation observed
576 CATALYST DEACTIVATION AND REGENERATION Vol. 1

(125–127) observed surface etching, enhanced sintering, and dramatic surface


restructuring of Pt thin films to faceted particles during ethylene oxidation
over a relatively narrow temperature range (500–7008C). The substantially
higher rate of sintering and restructuring in O2/C2H4 relative to that in nonreac-
tive atmospheres was attributed to the interaction of free radicals such as HO2,
formed homogeneously in the gas phase, with the metal surface to form meta-
stable mobile intermediates. Etching of Pt–Rh gauze in a H2/O2 mixture under
the same conditions as Pt surfaces (6008C, N2/O2/H2 ¼ 90/7.5/2.5) has also been
reported (123). A significant weight loss was observed in a laminar flow reactor
with little change in surface roughness, while in an impinging jet reactor, there
was little weight loss, but substantial restructuring of the surface to particle-
like structures, 1–10 mm in diameter; these particles were found to have the
same Pt–Rh composition as the original gauze. The nodular structures of
about 10-mm diameter formed in these experiments are strikingly similar to
those observed on Pt–Rh gauze after use in production of HCN at 11008C in
15% NH3, 13% CH4, and 72% air. Moreover, because of the high space velocities
during HCN production, turbulent rather than laminar flow would be expected
as in the impinging jet reactor. While little Pt is volatilized from the Pt–Rh
gauze catalyst during HCN synthesis, the extensive restructuring leads to
mechanical weakening of the gauze (8).
Other examples of catalyst deactivation due to volatile compound formation
include (1) loss of the phosphorus promoter from the VPO catalyst used in the
fluidized-bed production of maleic anhydride with an attendant loss of catalyst
selectivity (8), (2) vapor-phase loss of the potassium promoter from steam-
reforming catalysts in the high temperature, steam-containing environment (8),
and (3) loss of Mo from a 12-Mo-V-heteropolyacid due to formation of a volatile
Mo species during oxydehydrogenation of isobutyric acid to methacrylic acid (118).
While relatively few definitive studies of deactivation by volatile compound
formation have been reported, the previous work does provide the basis for enu-
merating some general principles. A generalized mechanism of deactivation by
formation of volatile metal compounds can be postulated (see Fig. 7). In addition,
the roles of kinetics and thermodynamics can be stated in general terms:

Transport
Metal compound vapor Lost vapor

Decomposition
Vaporization of vapor
Formation
Metal + Volatile agent Volatile compound Metal
Decomposition

(a) rate of volatile compound formation = rate of formation – rate of decomposition

(b) rate of metal loss = rate of vaporizaion – rate of vapor decomposition

Fig. 7. Generalized mechanisms and kinetics for deactivation by metal loss (8).
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 577

1. At low temperatures and partial pressures of the volatilization agent (VA),


the overall rate of the process is limited by the rate of volatile compound
formation.
2. At intermediate temperatures and partial pressures of the VA, the rate of
formation of the volatile compound exceeds the rate of decomposition.
Thus, the rate of vaporization is high, the vapor is stable, and metal loss
is high.
3. At high temperatures and partial pressures of the VA, the rate of formation
equals the rate of decomposition, ie, equilibrium is achieved. However, the
volatile compound may be too unstable to form or may decompose before
there is an opportunity to be transported from the system. From the pre-
vious work, it is also evident that besides temperature and gas phase com-
position, catalyst properties (crystallite size and support) can play an
important role in determining the rate of metal loss.

Solid-State Reactions. Catalyst deactivation by solid-state diffusion


and reaction appears to be an important mechanism for degradation of
complex multicomponent catalysts in dehydrogenation, synthesis, partial oxi-
dation, and total oxidation reactions (8,128–139). However, it is difficult in
most of these reactions to know the extent to which the solid-state processes
such as diffusion and solid-state reaction are affected by surface reactions.
For example, the rate of diffusion of Al2O3 to the surface to form an aluminate
may be enhanced by the presence of gas-phase oxygen or water or the nuclea-
tion of a different phase may be induced by either reducing or oxidizing con-
ditions. Recognizing this inherent limitation, the focus here is nevertheless
on processes in which formation of a new bulk phase (and presumably the
attendant surface phase) leads to substantially lower activity. There is prob-
ably some overlap with some of the examples given under Gas/Vapor–Solid
Reactions involving reactions of gas/vapor with solid to produce inactive
phases.
Examples from the literature of solid-state transformations leading to cata-
lyst deactivation are summarized in Table 10.
There are basic principles underlying most solid-state reactions in working
catalysts that have been enumerated by Delmon (135): (1) the active catalytic
phase is generally a high-surface-area defect structure of high surface energy
and as such a precursor to more stable, but less active phases and (2) the basic
reaction processes may themselves trigger the solid-state conversion of the active
phase to an inactive phase; for example, it may involve a redox process, part of
which nucleates the inactive phase.
A well-documented example of these principles occurs in the partial oxi-
dation of propene to acrolein on a Fe2(MoO4)3 catalyst (132,135). This oxidation
occurs by the ‘‘Mars van Krevelen’’ mechanism, ie, a redox mechanism in which
lattice oxygen reacts with the adsorbed hydrocarbon to produce the partially
oxygenated product; the reduced catalyst is restored to its oxidized state
through reaction with gaseous oxygen. In propene oxidation, two atoms of oxy-
gen from the catalyst are used, one for removing two hydrogen atoms from the
olefin and the other one in forming the unsaturated aldehyde. The fresh, calcined
578 CATALYST DEACTIVATION AND REGENERATION Vol. 1

Table 10. Examples of Solid-State Transformations Leading to Catalyst Deactivation


Deactivating
Catalytic process Catalytic solid chemical reaction Ref.
ammonia synthesis Fe/K/Al2O3 formation of KAlO2 at 138
catalyst surface
catalytic combustion PdO/Al2O3, PdO/ZrO2 PdO ! Pd at T > 8008C 131
catalytic combustion Co/K on MgO, CeO2, formation of CoO–MgO 139
or La2O3 solid soln., LaCoO3, or
K2O film on CeO2
dehydrogenation of Fe2O3/Cr2O3/K2O K migration to center of 8
styrene to ethyl pellet caused by
benzene thermal gradient
Fischer–Tropsch Fe/K, Fe/K/CuO transformation of active 136,137
carbides to inactive
carbides
oxidation of SO2 to SO3 V2O5/K2O/Na2O/ formation of inactive 134
kieselguhr V(IV) compounds at
T < 420–4308C
partial oxidation of V2O5 –MoO3 decreased selectivity due 128
benzene to maleic to loss of MoO3 and
anhydride formation of inactive
vanadium compounds
partial oxidation of Fe2(MoO4)3 plus MoO3 structural reorganization 129,135
methanol to to b-FeMoO4; reduction
formaldehyde of MoO3
partial oxidation of Fe2(MoO4)3 reductive transformation 132,135
propene to acrolein of Mo18O52 to Mo4O11
partial oxidation of Fe2(MoO4)3 reduction to FeMoO4 and 130,133
isobutene to MoO3 x
methacrolein

catalyst MoO3 consists of corner-sharing MoO6 octahedra (with Mo at the center


and six oxygen atoms at the corners), but upon reduction to MoO2, octahedra
share edges. However, it has been reported (132,135) that only slightly reduced
(relative to MoO3), open structures such as Mo18O52 and Mo8O23 are the most
active, selective phases; more complete reduction of either of these structures
leads to formation of Mo4O11 having substantially lower selectivity. Delmon
and co-workers (133,135) have shown that addition of an oxygen donor such as
Sb2O4 facilitates spillover of oxygen and thereby prevents overreduction and
deactivation of the catalyst.
2.5. Mechanical Failure of Catalysts. Forms and Mechanisms of
Failure. Mechanical failure of catalysts is observed in several different forms,
including (1) crushing of granular, pellet, or monolithic catalyst forms due to a
load; (2) attrition, the size reduction, and/or breakup of catalyst granules or pel-
lets to produce fines, especially in fluid or slurry beds; and (3) erosion of catalyst
particles or monolith coatings at high fluid velocities. Attrition is evident by a
reduction in the particle size or a rounding or smoothing of the catalyst particle
easily observed under an optical or electron microscope. Washcoat loss is
observed by scanning a cross section of the honeycomb channel with either an
optical or an electron microscope. Large increases in pressure drop in a catalytic
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 579

process are often indicative of fouling, masking, or the fracturing and accumula-
tion of attritted catalyst in the reactor bed.
Commercial catalysts are vulnerable to mechanical failure in large part
because of the manner in which they are formed; that is, catalyst granules,
spheres, extrudates, and pellets ranging in diameter from 50 mm to several milli-
meters are in general prepared by agglomeration of 0.02–2 mm aggregates of
much smaller primary particles having diameters of 10–100 nm by means of pre-
cipitation or gel formation followed by spray drying, extrusion, or compaction.
These agglomerates have in general considerably lower strengths than the pri-
mary particles and aggregates of particles from which they are formed.
Two principal mechanisms are involved in mechanical failure of catalyst
agglomerates: (1) fracture of agglomerates into smaller agglomerates of approxi-
mately 0.2d0 –0.8d0 and (2) erosion (or abrasion) of aggregates of primary parti-
cles having diameters ranging from 0.1 to 10 mm from the surface of the agglomerate
(140). While erosion is caused by mechanical stresses, fracture may be due to
mechanical, thermal, and/or chemical stresses. Mechanical stresses leading to
fracture or erosion in fluidized or slurry beds may result from (1) collisions of par-
ticles with each other or with reactor walls or (2) shear forces created by turbu-
lent eddies or collapsing bubbles (cavitation) at high fluid velocities. Thermal
stresses occur as catalyst particles are heated and/or cooled rapidly; they are
magnified by temperature gradients across particles and by differences in ther-
mal expansion coefficients at the interface of two different materials, eg, catalyst
coating/monolith interfaces; in the latter case the heating or cooling process can
lead to fracture and separation of the catalyst coating. Chemical stresses occur as
phases of different density are formed within a catalyst particle via chemical
reaction; for example, carbiding of primary iron oxide particles increases their
specific volume and micromorphology leading to stresses that break up these
particles (141).
Role of Properties of Ceramic Agglomerates in Determining Strength
and Attrition Resistance. Factors Affecting the Magnitude of Stress
Required for Agglomerate Breakage and the Mechanisms by Which it
Occurs. The extent to which a mechanism, ie, fracture or erosion, participates
in agglomerate size reduction depends upon several factors: (1) the magnitude of
a stress, (2) the strength and fracture toughness of the agglomerate, (3) agglom-
erate size and surface area, and (4) crack size and radius. Erosion (abrasion)
occurs when the stress (eg, force per area due to collision or cavitation pressure)
exceeds the agglomerate strength, ie, the strength of bonding between primary
particles. Erosion rate is reportedly (140) proportional to the external surface
area of the catalyst; thus, erosion rate increases with decreasing agglomerate
size.
Most heterogeneous catalysts are complex, multiphase materials that
consist in large part of porous ceramic materials, ie, are typically oxides, sulfides,
or metals on an oxide carrier or support. When a tensile stress of a magnitude
close to the yield point is applied, ceramics almost always undergo brittle frac-
ture before plastic deformation can occur. Brittle fracture occurs through for-
mation and propagation of cracks through the cross section of a material in a
direction perpendicular to the applied stress. Agglomerate fracture due to a
tensile stress occurs by propagation of internal and surface flaws; these flaws
580 CATALYST DEACTIVATION AND REGENERATION Vol. 1

created by external stresses or inherent defects are stress multipliers, ie, the
stress is multiplied by 2(a/r)0.5, where a is the crack length and r is the radius
of curvature of the crack tip; since a/r can vary from 2 to 1000, the effective stress
at the tip of a crack can be 4–60 times the applied stress. Tensile stress multi-
pliers may be microcracks, internal pores, and grain corners.
The ability of a material to resist fracture is termed fracture toughness. The
plain strain fracture toughness KIc is defined as

KIc ¼ YðpaÞ0:5 ð3Þ

where Y is a dimensionless parameter (often close to 1.0–2.0), the magnitude of


which depends upon both specimen and crack geometries, s is the applied stress,
and a is the length of a surface crack or half the length of an internal crack.
Crack propagation and fracture are likely if the right hand side of equation
3 exceeds the experimental value of plain strain fracture toughness (left-hand
side of eq. 3). Plane strain fracture toughness values for ceramic materials are
significantly smaller than for metals and typically below 10 MPa(m)0.5; repor-
ted values for nonporous, crystalline alumina (99.9%), fused silica, and zirconia
(3 mol% Y2O3) are 4–6, 0.8, and 7–12 MPa(m)0.5 respectively; flexural strengths
(analogous to yield strengths for metals) for the same materials are 280–550,
100, and 800–1500 MPa (142). Thus, on the basis of both fracture toughness
and flexural strength, nonporous, crystalline zirconia is much stronger toward
fracture than alumina, which in turn is much stronger than fused silica.
The introduction of porosity to crystalline or polycrystalline ceramic mate-
rials will, on the basis of stress amplification, significantly decrease elastic mod-
ulus and flexural strength for materials in tension.
Thus far the discussion has focused mainly on tensile strength, the extent of
which is greatly reduced by the presence of cracks or pores. However, for ceramic
materials in compression, there is no stress amplification due to flaws or pores;
thus ceramic materials (including catalytic materials) in compression are much
stronger (approximately a factor of 10) than in tension. In addition, the strength
of ceramic materials can be dramatically enhanced by imposing a residual com-
pressive stress at the surface through thermal or chemical tempering. Moreover,
introduction of binders such as graphite enables agglomerates of ceramic pow-
ders to undergo significant plastic deformation before fracture.
Tensile Strengths and Attrition Resistance of Catalyst Supports and
Catalysts. The strengths cited above for nonporous, annealed crystalline or
polycrystalline materials do not necessarily apply to porous catalyst agglomer-
ates even under compression; rather, agglomerate strength is dependent upon
the strengths of chemical and physical bonds including the cohesive energy
between primary particles. Agglomerate strength would depend greatly on the
preparation of the compact. Representative data for catalyst agglomerates (see
Table 11) suggest they are generally substantially weaker than polycrystalline
ceramic materials prepared by high temperature sintering, such as alumina
(140,142,144–148).
From the data in Table 11 it is evident that even subtle differences in pre-
paration and pretreatment also affect agglomerate strength. For example,
spheres of g-Al2O3 prepared by sol–gel granulation are substantially (17 times)
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 581

Table 11. Mechanical Strengths and Attrition Rates of Catalyst Supports Compared
to Those of Sintered Ceramic Agglomerates
Catalyst Attrition
support or Preparation/pretreatment/ Strength, index,
ceramic properties MPa wt%/h Ref.
High surface area catalyst supports
g-Al2O3, 1.2– sol–gel granulation/dried 10 h at 11.6  1.9 0.033 144
4.25-mm spheres 408C, calcined 3 h at 4508C/
389 m2/g, dpore ¼ 3.5 nm
g-Al2O3, 4.25-mm Alcoa LD-350 0.7 0.177 144
spheres
g-Al2O3, 100 mm VISTA-B-965-500C 6.2  1.3 140
TiO2 (anatase), thermal hydrolysis/dried 1108C, 28a 145
2
30 mm calcined 2 h at 5008C/92 m /g,
<10-nm primary crystallites
TiO2 (anatase), basic precipitation/dried 1108C, 15a 145
90 mm calcined 2 h at 5008C/81 m2/g,
10–14-nm primary crystallites
TiO2 (75% anatase, Degussa P25, fumed/4-mm 0.9 146
25% rutile) extrudates/48 m2/g,
Vpore ¼ 0.34 cm3/g,
dpore ¼ 21 nm
TiO2 (anatase) Rhone-Poulenc DT51, ppt./4 mm 0.9 146
extrudates/92 m2/g,
Vpore ¼ 0.40 cm3/g,
dpore ¼ 8, 65 nm
Low surface area ceramics
Al2O3 spray dried with organic binder; 2.3 147
plastic deformation observed
Al2O3 heat treated (sintered), 99.9% 282–551 142
TiO2 (Rutile) partially sintered 194 147
ZrO2 (yttria commercial samples from three 0.035–0.43 148
additive) companies, spray-dried
ZrO2 (3% Y2O3) heat treated (sintered) 800–1500 142
a
Rough estimates from break points on relative density versus log[applied pressure] curves; data are
consistent with mass distribution versus pressure curves from ultrasonic tests.

stronger than commercial g-Al2O3 spheres (143). Moreover, 30- and 90-mm dia-
meter particles of TiO2 prepared by thermal hydrolysis or basic precipitation are
30 and 15 times stronger than commercially available 4-mm extrudates (146).
Catalyst attrition is a difficult problem in the operation of moving-bed,
slurry-bed, or fluidized-bed reactors. Generally, stronger materials have greater
attrition resistance; this conclusion is supported by representative data in
Table 11 for g-Al2O3, showing that the strength of the alumina prepared by
sol–gel granulation is 17 times higher, while its attrition rate is 5 times lower.
The mechanism by which attrition occurs (erosion or fracture) can vary
with catalyst or support preparation, crush strength, and with reactor environ-
ment; it can also vary with the mechanical test method. There is some evidence
in the attrition literature supporting the hypothesis that in the presence of a
large stress, weaker oxide materials are prone to failure by fracture, while
stronger materials tend to erode (149). However, there is also contrary evidence
582 CATALYST DEACTIVATION AND REGENERATION Vol. 1

(145), showing that fracture may be the preferred mechanism for strong TiO2
agglomerates, while abrasion is favored for weaker agglomerates. Supporting a
third trend, data (140) show that attrition mechanism and rate are independent
of agglomerate strength but depend instead on the type of material. 100-
mm-diameter agglomerates of precipitated Fe/Cu/K Fischer–Tropsch catalyst
[prepared by United Catalyst (UCI)] and having nearly the same strength
shown in Table 11 for Vista-B Al2O3 (6.3 vs. 6.2 MPa), were found to undergo
substantial fracture to 5–30-mm fragments (an increase from 45 to 85%) as
well as substantial erosion to 1 mm or less fragments (increase from 2 to 50%).
Under the same treatment conditions, 90-mm-diameter agglomerates of Vista-B
Al2O3 underwent by comparison much less attrition, mainly by erosion (20%
increase in 0.1–5-mm fragments). The very low attrition resistance of the Fe/
Cu/K UCI catalyst is further emphasized by the unsatisfactory outcome of a
test of this catalyst by the U.S. Department of Energy (DOE) in a pilot-scale
slurry-phase bubble-column reactor in LaPorte, Tex.; following one day of opera-
tion, the filter system was plugged with catalyst fines, preventing catalyst–wax
separation and forcing shutdown of the plant (150).
Thus, based on these three representative examples, it follows that which of
the two attrition mechanisms predominates depends much more on material
composition and type than on agglomerate strength. However, irrespective of
mechanism the rate of attrition is usually greater for the weaker material.
The catalyst preparation method can have a large effect on the attrition
resistance of an Fe/Cu Fischer–Tropsch catalyst (151). This catalyst, prepared
by precipitation, undergoes severe attrition during a 25-min treatment with
ultrasonic radiation; indeed the mass fraction finer than 0.1–5 mm increases
from 0 to 65%. However, after a spray drying treatment of the same catalyst,
less than a 10% increase in the same fractions is evident.
In their review of attrition and attrition test methods, Bemrose and Bridge-
water (152) discuss how attrition varies with reactor type, eg, involves mainly
particle–wall impacts in moving pellet bed reactors and particle–particle
impacts in fluidized-bed reactors of high fluid velocity. In fact, jet attrition of cat-
alyst particles in a gas fluidized bed involving principally abrasion due to colli-
sion of high-velocity particles has been modeled in some detail (149,153). Thus,
given such important differences in attrition mechanism, realistic attrition test
methods should attempt to model reactor operation as closely as possible. In
addition, the ideal test would require only a small catalyst sample, a simple,
inexpensive apparatus, and a few minutes to complete the test. Relatively
quick, inexpensive single-particle crushing tests have been devised (152); how-
ever, properties of a single particle are rarely representative of those for the
bed; moreover, it is difficult to relate the results of this crushing test to the actual
abrasion process. Realistic tests have been devised for two reactor types involv-
ing a moving catalyst, ie, an air-jet test for fluidized-bed catalysts (154,155), and
a rotating drum apparatus for moving-bed catalysts (156); however, the air-jet
test requires a large quantity (eg, 50 g) of catalyst, an expensive apparatus,
and about 20 h to run. In the past decade a new jet-cup test has been developed
for testing of fluidized-bed catalysts (154,155), which requires only a 5-g sample
and about 1 h to complete; comparisons of results for the jet-cup and air-jet tests
indicate that the two tests give comparable results (154,155). Nevertheless, the
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 583

mechanisms for the two tests are different, ie, the air-jet (fluid-bed) test is
abrasion- (erosion-) dominant, while the jet-cup test includes both abrasion
and fracture mechanisms (155). A 30-min, 10-g ultrasonic attrition test based
on cavitation has also been developed in the past decade (145,151,157); while
it likewise involves both abrasion and fracture mechanisms, the results appear
to correlate with other methods. For example, particle size distributions for the
same Co/silica catalyst after ultrasonic, jet-cup, and laboratory-scale, slurry-bed
column reactor (SBCR) tests are very similar, indicating that both fracture and
abrasion mechanisms operate in the small-scale SBCR. Moreover, the good
agreement among the three methods suggests that both the jet-cup and ultraso-
nic tests may provide data representative of the attrition process in laboratory-
scale SBCR reactors. It is evident that these two small-scale methods are espe-
cially useful for screening of a series of catalysts to determine relative strength.
Nevertheless, the more realistic large-scale tests are probably needed for
accurately determining design attrition rates of a commercial catalyst to be
used in a full-scale process. The observation that attrition of a fluid catalytic
cracking (FCC) catalyst initially involves fracture of weak agglomerates followed
by abrasion of strong agglomerates emphasizes the need to collect and analyze
the particle size distribution of attrited fines as a function of time in order to
define which mechanism (or mechanisms) operates at startup as well as in the
steady-state process. Because the mechanism may be time dependent, rapid,
small-scale tests may produce misleading results.
While realistic laboratory-scale tests have been developed for simulating
attrition in large moving-bed and fluidized-bed reactors, no such laboratory
test has been developed and demonstrated yet for simulation of large-scale
SBCR reactors, although recent research has focused on the development of
such tests. For example, in laboratory-scale, SBCR tests of supported cobalt cat-
alysts over several days (157), it was observed that the attrition resistance
decreases in the order Co/Al2O3, Co/SiO2, Co/TiO2 (especially the anatase form
underwent attrition at a high rate); attrition resistance was observed to increase
with increasing cobalt loading from 10 to 40 wt%.
Implications of Mechanistic Knowledge of Attrition for Catalyst Design.
The understanding of mechanisms important in attrition of catalyst supports
and catalysts, the relationship between strength and attrition rate for a given
material, and test data can be used to great advantage in the design of attrition
resistant catalysts. Several alternatives follow from the previous discussion for
increasing attrition resistance: (1) increasing aggregate/agglomerate strength
by means of advanced preparation methods, eg, sol–gel granulation, spray dry-
ing, and carefully controlled precipitation methods (see Table 11 for examples),
(2) adding binders to improve strength and toughness, eg, the addition of a poly-
vinylpyrrolidone binder to agglomerates of quartz sand increases agglomerate
strength from 0.1 to 3 MPa (158), (3) coating aggregates with a porous but
very strong material such as ZrO2, eg, embedding a fluidized-bed catalyst for par-
tial oxidation of n-butane to maleic anhydride in a strong, amorphous matrix of
zirconium hydrogen phosphate significantly improves its attrition resistance
(159), and (4) chemical or thermal tempering of agglomerates to introduce
compressive stresses that increase strength and attrition resistance, eg, heating
and cooling particles rapidly by passing them through a low-residence-time,
584 CATALYST DEACTIVATION AND REGENERATION Vol. 1

high-temperature furnace to harden the agglomerate exterior, while preventing


significant sintering of or phase changes in the porous interior. The subject of
preventing mechanical degradation and other forms of catalyst deactivation is
addressed in greater detail under Prevention of Catalyst Decay.
2.6. Summary of Deactivation Mechanisms for Solid Catalysts.
Causes of solid (heterogeneous) catalyst deactivation are basically threefold:
(1) chemical, (2) mechanical, and (3) thermal. Mechanisms of heterogeneous
catalyst deactivation can be classified into five general areas: (1) chemical de-
gradation including volatilization and leaching, (2) fouling, (3) mechanical
degradation, (4) poisoning, and (5) thermal degradation. Poisoning and thermal
degradation are generally slow processes, while fouling and some forms of chemi-
cal and mechanical degradation can lead to rapid, catastrophic catalyst failure.
Some forms of poisoning and many forms of fouling are reversible; hence, rever-
sibly poisoned or fouled catalysts are relatively easily regenerated. On the other
hand, chemical, mechanical, and thermal forms of catalyst degradation are
rarely reversible.
It is often easier to prevent rather than cure catalyst deactivation. Many
poisons and foulants can be removed from feeds using guard beds, scrubbers,
and/or filters. Fouling, thermal degradation, and chemical degradation can be
minimized through careful control of process conditions, eg, lowering tempera-
ture to lower sintering rate or adding steam, oxygen, or hydrogen to the feed
to gasify carbon or coke-forming precursors. Mechanical degradation can be mini-
mized by careful choice of carrier materials, coatings, and/or catalyst particle
forming methods.
While treating or preventing catalyst deactivation is facilitated by an
understanding of the mechanisms, additional perspectives are provided by exam-
ining the route by which each of the mechanisms causes loss of catalytic activity,
ie, how it influences reaction rate (92). Thus, catalytic activity can be defined in
terms of the observed site-based rate constant kobs, which is equal to the product
of the active site density s (number of sites per area of surface), the site-based
intrinsic rate constant kintr, and the effectiveness factor Z, ie,

kobs ¼ kintr Z ð4Þ

Loss of catalytic activity may be due to a decrease in any of the three factors in
equation 4, whose product leads to kobs. Thus, catalyst deactivation can be
caused by (1) a decrease in the site density s, (2) a decrease in intrinsic activity
(ie, decrease in kintr), and/or (3) lowered access of reactants to active sites
(decrease in Z). Poisoning, for example, leads to a loss of active sites, ie,
s ¼ s0(1 a), where a is the fraction of sites poisoned; sintering causes loss of
active sites through crystallite growth and reduction of active surface area.
Fouling can cause both loss of active sites due to blocking of surface sites as
well as plugging of pores causing a decrease in the effectiveness Z. Moreover,
poisoning, as discussed earlier, can also lead to a decrease in intrinsic activity
by influencing the electronic structure of neighboring atoms. Thus, each of the
deactivation mechanisms affects one or more of the factors comprising
observed activity (see Table 12); all of the mechanisms, however, can effect a
decrease in the number of catalytic sites.
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 585

Table 12. How Deactivation Mechanisms Affect the Rate of a Catalyzed Reaction and the
Rapidity and Reversibility of Deactivation Process
Effects on reaction rate Deactivation process

Decrease Decrease in Decrease in


Deactivation in number intrinisic effectiveness Fast or
mechanism of active sites activity (kintr) factor (Z) slowa Reversible
b,c
chemical varies no
degradation
fouling fast yes
mechanical varies no
degradation
poisoning slow usually
sintering b,d b,e slow sometimes
vaporization/ b,f fast sometimes
leaching
a
Generally.
b
In some cases.
c
Chemical degradation can cause breakdown of support, pore plugging, and loss of porosity.
d
If the reaction is structure-sensitive, sintering could either increase or decrease intrinsic activity.
e
Sintering of the support may cause support collapse and loss of porosity; it may also increase average
pore diameter.
f
Leaching of aluminum or other cations from zeolites can cause buildup of aluminum or other oxides
in zeolite pores.

3. Homogeneous Catalysts and Enzymes

3.1. Homogeneous Catalysts. The discussion of the deactivation of


homogeneous catalysts has received less attention relative to that of hetero-
geneous catalysts (160,161). Indeed, the first comprehensive review of homo-
geneous catalyst deactivation appeared just recently (160). Nevertheless, the
vast literature of homogeneous catalysis provides numerous anecdotal accounts
of problems with catalyst decomposition and references to homogeneous
catalysts having a limited number of turnovers, all testifying to the importance
of these phenomena.
Homogeneous catalysts may undergo degradation by routes similar to those
of heterogeneous catalysts, eg, by chemical modification, poisoning, and thermal
degradation. However, the specific details of these mechanistic routes are gener-
ally somewhat different, since the catalyst is a molecule rather than a solid; that
is, an organometallic complex is quite different from a metal surface in terms of
structure and scale. For example, reaction of impurities with homogeneous cat-
alytic complexes is analogous to poisoning of a heterogeneous catalyst by impu-
rities, although the former is essentially a chemical reaction of two species of
similar dimensions while the latter involves adsorption of a molecule on the sur-
face of a crystallite containing hundreds to thousands of atoms.
Homogeneous catalysts are generally metal–ligand complexes. The metal
center functions as the active site, while the ligands serve to influence site chem-
istry through electronic modifications of the metal that influence activity/selec-
tivity and through geometric constraints that enhance selectivity. Hence
586 CATALYST DEACTIVATION AND REGENERATION Vol. 1

activity and selectivity properties of homogeneous catalysts can be significantly


influenced by processes that change the chemistry either of the metal center or
the ligands or both.
Mechanisms (or causes) of homogeneous catalyst degradation can be classi-
fied as (1) metal deposition reactions, eg, decarbonylation of carbonyl complexes,
loss of protons from cationic species, or reductive elimination of C-, N-, or O-
donor fragments; (2) decomposition of ligands attached to a catalytic complex;
(3) reactions of metal–carbon and metal–hydride bonds with polar species (eg,
water, oxygen, acids, alcohols, olefins, and halides); and (4) poisoning of active
sites by impurities, reactants, or products or by dimerization of the catalyst.
Principal features of these mechanisms and examples thereof are summar-
ized in Table 13. It is noteworthy that mechanisms 1 and 2 lead mainly to deac-
tivation by either loss or modification of ligands, while mechanisms 3 and 4 cause
deactivation largely by either modifying or poisoning the metal, although ligands
are also clearly modified by type 3 mechanisms. Of the four mechanisms, deacti-
vation by metal formation and deposition is the most common, although all are
important.
Mechanisms 1 and 4 are reversible to some extent. Mechanisms 2 and 3,
involving breaking of active site bonds and formation of stable products, are
largely irreversible. Products of ligand oxidation are generally more stable
than the complexes from which they were formed.
3.2. Enzymes. Structural and Catalytic Properties of Enzymes.
Enzymes are globular macromolecular polypeptide proteins (molecular weights
of 104 –106) synthesized by living organisms (8,162). Each enzyme has a unique
three-dimensional structure with a binding site or pocket that is chemically and
geometrically compatible with a single reactant molecule (substrate) or group of
chemically related reactants; in other words enzymes have molecular-recognition
capability. Enzymes are unique in their ability to catalyze biochemical reactions
with high selectivity (essentially 100%) at extraordinarily high rates, ie,
10–10,000 molecules/(enzymes) compared to typical values of 1–10 or less for
conventional catalysts. These activities enable enzymes to be effective catalysts
at extremely low concentrations, eg, 10 5 –10 10 mol/L, at substrate (reactant)
concentrations of greater than 10 6 mol/L.
The high activity of enzymes has been illustrated for urease and catalase
(163,164). The stereochemical specificity of enzymes is unmatched and absolute,
ie, their sites can distinguish between optical and geometrical isomers, almost
always catalyzing only the reaction of one isomer of an enantiomeric pair
(162). Nevertheless, some enzymes catalyze reactions of chemically unrelated
species; for example, nitrogenase reduces N2 to NH3 as well as hydrogenating
acetylene to ethylene (165).
In 1976 there were 1800 known enzymes, and new enzymes were being dis-
covered at the rate of about 60 per year (166); accordingly, there were an esti-
mated 3000 known enzymes in 1996. It is estimated that an average cell
contains 3000 different enzymes (165), and it is speculated that as many as
25,000 different enzymes exist (167).
While they are synthesized in vitro and are only active within a limited
range of pH and temperature, enzymes otherwise have properties similar to syn-
thetic homogeneous and polymer-supported catalysts. Moreover, they can be
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 587

Table 13. Deactivation Mechanisms for Homogeneous Catalystsa


Deactivation
mechanism Comments Examples
1. Metals deposition most common decomposition
reactions mechanism
ligand loss decarbonylation is most 2 HCo(CO)4 ! CO2(CO)8 !
common CO4(CO)12 ! Co metal
loss of protons from reductive elimination as HX; in Wacker reaction elimination
cationic species pH dependent; basic media of HCl from PdHCl leads to
lead to zero-valent metals Pd(0); in Heck
reaction Pd(0) is an
intermediate
reductive elimination occurs in cross-coupling oxidative addition of aryl
of C-, N-, O-donor reactions to form C-C, C-N, halides
fragments C-O bonds
2. Ligand ligands greatly influence activity
decompositions and selectivity of homogeneous
catalysts
oxidation phosphorus or sulfur-based PR3 þ H2O ! H2 þ O ‘‘PR3;
ligands are readily oxidized by PR3 þ 1/2 O2 ! O ‘‘PR3
O2, H2O, CO2, peroxides;
nitrogen-based ligands are
more stable
oxidative addition breaking of C-P bond with decomposition of Rh and Co
insertion of a metal hydroformylation catalysts
(161)
nucleophilic attack internal or external attack decomposition of triphenyl-
causing insertion of metal in phosphines (TPPs) in Pd
C-P bond or displacement of catalyst by acetate ion
metal with Ar
thermal depends on temperature and decomposition of RhH(CO)
decomposition gas composition (PPh3) hydroformylation
catalysts to stable cluster
containing m2-PPh2
fragments in absence
of H2 and CO
reactions with water hydrolysis, alcoholysis, and hydrolysis of diphosphites
alcohols; rearrange- transesterification of in Rh-catalyzed hydro-
ments phosphites, imines, and formylation of alkenes
pyridines
3. Reactions of metal–
carbon and metal–
hydride bonds
with water, oxygen, decomposition of reactive metal deactivation of Ziegler catalysts
acids, and alcohols alkyls with water or oxygen containing alkyl complexes of
Ti, Zr, and V
with olefins, halides, formation of metallated reaction of propene with
and aluminoxanes transition metal ion zirconium alkyl catalyst
complex deactivates forms an alkane and a
polymerization catalysts p-allyl zirconium species
4. Poisoning of active
sites
by dienes forms a stable p-allylic methoxycarbonylation of
and alkynes complex; these poisons propyne using Pd-2-pyridyl-
must be removed in DPP is poisoned by butadiene
polyolefin manufacture and 1,2-propadiene
588 CATALYST DEACTIVATION AND REGENERATION Vol. 1

Table 13. ð Continued Þ


Deactivation
mechanism Comments Examples
by polar impurities such as basic amines enantioselective isomerization
of the allylamine to the
asymmetric enamine (in
menthol synthesis) is
poisoned by a stronger
basic amine isomer
by dimer formation active monomeric catalyst Rh–TPP hydroformylation
species form dimers catalyst dimerizes; Pd(I)
dimers in carbonylation
catalysts
a
Data from 160.

extracted from their biological source, purified, crystallized, and used in labora-
tory studies or industrial processes. Further, they can be attached to glass or
ceramic supports and used as heterogeneous catalysts. And their application in
industrial processes is rapidly increasing.
Enzymes are formed in living systems by condensation/dehydration of
amino acids to produce peptide (C–N) bonds that constitute the backbone of
long protein chains. The active conformation of an enzyme is produced by folding
of the protein chain into secondary (helical), tertiary (folded), and quaternary
(combined tertiary) structures. The folded layers are held in place by hydrogen
bonding and disulfide linkages. There are 20 naturally occurring amino acids,
each having the composition H2N–CHR–COOH, the R group (side chain) having
different molecular functions, eg, proton donation, proton removal, and bridge
formation; for example, amino acids include glycine, alanine, and serine with
side chains of H, CH3, and CH2OH respectively. Upon folding, side chains
become the functional groups of the active site or ligands for binding of metals
ions, which then become functional groups at the active site. Naturally occurring
metal ions in enzymes include Mg2þ, Zn2þ, Ca2þ, Ni2þ, Fe2þ, Fe3þ, Co3þ, and
Mo2þ.
Distinctive catalytic characteristics of enzymes (162) include (1) their flex-
ible structure, which facilitates an ‘‘induced fit’’ of the substrate, the making and
breaking of bonds, and the departure of products, and (2) their sensitivity to reac-
tion effectors (inhibitors or activators), which function similarly to promoters of
heterogeneous catalysts. Some enzymes require a cofactor that combines with
the enzyme to form a catalytic site; metal ions are examples of cofactors. Enzy-
matic reactions may also require a coenzyme that reacts with the reactant to pro-
duce an enzyme-compatible substrate. Living organisms control and optimize
biological processes using a variety of tools: (1) enzyme effectors, (2) regulation
of enzyme growth or activation rates, (3) compartmentalization of enzymes
within organs or organelles, and (4) destruction (editing) of undesired intermedi-
ates or products (162).
Deactivation of Enzymes. Enzymes generally function only under mild
conditions of temperature and pH observed in living organisms. Under typical
commercial reaction conditions (40–608C, 1 atm) enzymes otherwise stable in
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 589

solution may lose activity rapidly as a result of only slight changes in their
environment such as temperature, pressure, pH, and ionic strength that induce
small free energy changes from native to denatured states (168); moreover, their
separation from the product is generally difficult and may cause further dena-
turation and loss of catalytic activity. The modest, largely reversible losses of
activity resulting from small changes in reaction environment are largely due
to modest changes in conformation of the active site. More severe changes in
reaction conditions (eg, a 108C increase in temperature) typically bring about
the dissociation and unfolding of the quaternary and tertiary structures, respec-
tively, into primary chains that subsequently order into fibrous protein bundles;
in the process active sites are irreversibly destroyed. If further exposed to severe
conditions of temperature and pH, the principal chain structure of the protein
may undergo loss or modification of functional groups or amino acid residues.
The activity of a typical enzyme increases exponentially with temperature
in accordance with the Arrhenius law up to about 50–608C, passes through a
maximum and declines precipitously above about 60–708C. Thus, catalyst life
may be on the order of days to weeks at around 508C; however, the deactivation
rate is extremely high at only slightly higher temperatures, eg, 50% loss of activ-
ity in 5 min at 65–708C is typical. Nevertheless, a few enzymes are active
and stable at temperatures exceeding 1008C; for example, a-amylase catalyzes
starch liquifaction at 105–1158C. Because their deactivation rates are highly
temperature-dependent, enzymes are generally shipped and stored under refrig-
eration (0–48C); at these low temperatures they are generally stable for months.
Causes of deactivation can be classified (as in the case of heterogeneous cat-
alysts) as chemical, mechanical, and thermal. However, for enzymes these causes
are closely linked, since mechanically and thermally induced routes almost
always effect chemical changes. Thermally induced chemical change (at elevated
temperature) is the most likely scenario for enzyme deactivation.
Chemical deactivation mechanisms include (1) changes in stereo configura-
tion by protons or hydroxyl ions at or near active sites (169,170), (2) structural
modifications in aqueous or nonaqueous solvents (171–173), (3) poisoning of
active sites by inhibitors (162,174), including ‘‘Trojan-horse inhibitors’’ that are
activated by the target enzyme (162), (4) aggregation (175), (5) unfolding (6) frag-
mentation due to solvolysis, hydrolysis in water, or self-hydrolysis (autolysis) of
proteases, eg, trypsin (176), and (7) oxidation in air (177,178). Mechanisms 1–5
may be reversible, while mechanisms 6 and 7 are generally irreversible. Mechan-
ical deactivation may be caused by hydrodynamic shear forces, eg, by stirring or
gas sparging, sometimes leading to fragmentation and/or aggregation (175,178).
Thermal inactivation of enzymes is a well-studied phenomenon (179–185);
it may be either reversible or irreversible (180). Potentially reversible changes
(due to small, short excursions in temperature near the characteristic unfolding
temperature) include light aggregation, conformational changes, folding without
further chemical change, disulfide exchange, and/or breaking of hydrogen bonds.
Irreversible denaturation (due to prolonged, severe thermal treatment) may be
caused by cleavage of disulfide bonds and/or cystinyl cross-links; unfolding fol-
lowed by chemical change; chemical changes of the primary structure and/or
active site, eg, cleavage of the polypeptide chain by hydrolysis or destruction of
individual amino acid residues; strong aggregation of inactive unfolded forms;
590 CATALYST DEACTIVATION AND REGENERATION Vol. 1

and formation of rubbery, tough fibrous structures due to alignment and bund-
ling of unfolded primary chains (similar to that observed during the boiling of an
egg). Chemical bonding of unfolded primary chains to form fibers is thermodyna-
mically favorable because chemical bonding of hydrophobic functions exposed by
unfolding lowers the entropy and hence free energy of the system.
Table 14 summarizes representative examples of enzyme deactivation by
the various mechanisms.
Methods of enhancing enzyme stability have received considerable atten-
tion (168,173,178,180,183,185–187). Strategies to improve both chemical and
thermal stability include (1) use of soluble additives, (2) immobilization, (3) pro-
tein engineering, and (4) chemical modification. Chemical modification (183,
185–187) and immobilization (164,165,172,188–191) are probably the most suc-
cessful and widely used methods. As examples of the first kind, modification of
protein surfaces by chemical binding with polysaccharides can improve thermo-
stability, while polyol binding increases enzyme solubility in organic solvents
with little loss of activity (171,176,183). Enzyme stability can be greatly
enhanced and recovery problems obviated by immobilizing (heterogenizing)
enzymes (164,165,172,188–191) through (1) covalent binding to a support,
(2) cross-linking of enzymes using a bifunctional agent, (3) adsorption on a
solid surface, (4) entrapment in a gel, or (5) containment in a membrane. More-
over, immobilization enables the catalytic process to be run continuously using a
reactor of substantially lower volume, thereby substantially reducing capital and
operating costs. These important advantages have stimulated the development
of a significant number of commercial immobilized enzyme systems.

4. Prevention of Catalyst Decay

4.1. General Principles of Prevention. The age-old adage that says


‘‘an ounce of prevention is worth a pound of cure’’ applies well to the deactivation
of catalysts in many industrial processes. The catalyst inventory for a large plant
may entail a capital investment of tens of millions of dollars. In such large-scale
processes, the economic return on this investment may depend on the catalyst
remaining effective over a period of up to 3–5 years. This is particularly true
of those processes involving irreversible or only partially reversible deactivation
(eg, sulfur poisoning or sintering). Some typical industrial catalysts, approxi-
mate catalyst lifetimes, and factors that determine their life are listed as
examples in Table 15. It is evident that in many processes more than one
mechanism limits catalyst life. Moreover, there is a wide variation in catalyst
lifetimes among different processes, ie, from 10 6 to 15 years. While there is
clearly greater interest in extending catalyst lifetimes in processes where life
is short, it should be emphasized that great care must be exercised in protecting
the catalyst in any process from process upsets (eg, temperature runaway, short-
term exposure to impure feeds, or changes in reactant composition) that might
reduce typical catalyst life by orders of magnitude, eg, from years to hours.
While complete elimination of catalyst deactivation is not possible, the rate
of damage can be minimized in many cases through understanding of the
mechanisms, thereby enabling control of the deactivation process, ie, prevention
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 591

Table 14. Representative Examples of Deactivation Mechanisms for Enzymes


Deactivation
mechanism Cause(s)/reversibility Examples Ref.
1. Chemical generally involve
formation or breaking
of bonds in enzyme
structure
modest changes caused by (a) introduction model of effects of pH on 184
in active site of Hþ or OH near phytases: enzyme is in
configuration active site, (b) small equilibrium with
changes in pH or protonated and
solvent environment/ hydroxylated forms
largely reversible which are less active or
inactive
poisoning of adsorption of inhibitor mechanistic study of the 174
active site on active site/ inhibition of crotonase by
sometimes reversible (methylenecyclopropyl)-
formyl-CoA; MCP ring
trapping of an active
site nucleophile is
suggested
aggregation caused by changes in pH dimers and trimers of 175
or solvent environment lysozyme are formed and
with partial unfolding/ activity is lost in a stirred
sometimes reversible reactor; mechanism may
involve collision-induced
conversion of enzyme to
inactive state, followed
by formation of disulfide
bridges
unfolding, cleavage of enzyme bonds deactivation of peroxidase 168,172,173
fragmentation, due to interaction with in organic solvents
bundling of solvent, Hþ, or OH due including DMSO;
primary chains to medium to large solvent may strip
into fibers changes in pH/ water from enzyme,
irreversible leading to reduced
conformational mobility
and unfolding
2. Mechanical caused by hydrodynamic lysozyme is aggregated and 175
shear forces, eg, irreversibly inactivated in
stirring or gas a stirred reactor; the
sparging, which can break deactivation rate
bonds and cause constant is
aggregation of proportional to the
enzymes/usually impeller power
irreversible
3. Thermal
modest changes caused by small, short equilibrium measure- 179,180
in active site excursions in tempera- ments of the tempera-
configuration ture near the transition ture-induced unfolding
and reversible temperaturea/ of bovine ribonuclease;
unfolding reversible repeated measurements
after cooling fall on the
same plot of fraction
unfolded versus T
592 CATALYST DEACTIVATION AND REGENERATION Vol. 1

Table 14. ðContinuedÞ


Deactivation
mechanism Cause(s)/reversibility Examples Ref.
irreversible cleavage of enzyme bonds irreversible thermo- 180
unfolding, due to interaction with inactivation of hen egg
fragmentation, solvent, Hþ, or OH white lysozyme at 1008C
bundling of due to medium to large and pH 4, 6, 8; inactivation
primary chains changes in pH/ is due to monomolecular
into fibers irreversible changes in coordination,
eg, hydrolysis of the
Asp-X peptide bonds,
deamidation of Asn residues,
destruction of cystine
residues, and formation of
incorrect structures
a
Characteristic temperature for a specific enzyme above which unfolding occurs and below which
refolding occurs.

is possible through control of catalyst properties, process conditions (ie, tempera-


tures, pressures), feedstock impurities, methods of contacting, and process
design. Figure 8 illustrates general approaches to eliminating or moderating
deactivation through modifications in catalyst and/or process. Examples of how
deactivation can be prevented are discussed below in connection with the most
important causes of deactivation: chemical degradation, fouling by coke and car-
bon, poisoning, sintering, and mechanical degradation. Principles for preventing
deactivation by these mechanisms are summarized in Table 16. Representative
results from studies focusing on prevention or minimization of catalyst deactiva-
tion are found in Refs. 18,48,55,56,192–245.
4.2. Prevention of Chemical Degradation (by Vapor–Solid and
Solid–Solid Reactions). The most serious problems-oxidation of metal
catalysts, overreduction of oxide catalysts, and reaction of the active catalytic
phase with carrier or promoter-can be minimized or prevented by careful cat-
alyst and process design (as enumerated in Table 16). For example, the loss of
Rh due to solid-state reaction with alumina in the automotive three-way cata-
lyst can be prevented by supporting Rh on ZrO2 in a separate layer from Pt
and/or Pd on alumina. In Fischer –Tropsch synthesis, the oxidation of the
active cobalt phase in supported cobalt catalysts to inactive oxides, alumi-
nates, and silicates can be minimized by employing a two- or three-stage pro-
cess in which product steam is moderated in the first stage by limiting
conversion and in subsequent stages by interstage removal of water. It can
also be moderated by addition of noble metal promoters that facilitate and
maintain high reducibility of the cobalt and by coating the alumina or silica
support with materials such as ZrO2 that are less likely to react with cobalt
to form inactive phases.
4.3. Prevention of Fouling by Coke and Carbon. Rostrup-Nielsen
and Trimm (45), Trimm (47), and Bartholomew (48) have discussed principles
and methods for avoiding coke and carbon formation. General methods of pre-
venting coke or carbon formation are summarized in Table 16. Most of these
Table 15. Typical Lifetimes and Factors Determining the Life of Some Important Industrial Catalystsa
Operating Typical life Process affecting life Catalyst property
Reaction conditions Catalyst (years) of catalyst charge affected
Ammonia synthesis 450–4708C, Fe with promoters (K2O) 10–15 slow sintering activity
N2 þ 3 H2 ! 2 NH3 200–300 atm and stabilizer (Al2O3)
methanation (ammonia and 250–3508C, supported nickel 5–10 slow poisoning by S, activity and pore
hydrogen plants) 30 atm As, K2CO3 from blockage
CO/CO2 þ H2 ! CH4 þ H2O plant upsets
acetylene hydrogenation 30–1508C, supported palladium 5–10 slow sintering activity/selectivity
(‘‘front end’’) 20–30 atm and temperature
C2H2 þ H2 ! C2H4
sulfuric acid manufacturing 420–6008C, vanadium and potassium 5–10 inactive compound activity, pressure
2 SO2 þ O2 ! 2 SO3 1 atm sulfates on silica formation; pellet drop, and mass
fracture; plugging transfer
by dust
methanol synthesis 200–3008C, copper on zinc and 2–5 slow sintering; activity
CO þ 2 H2 ! CH3OH 50–100 atm aluminum oxides poisoning by S, Cl,
and carbonyls

593
low temperature CO shift 200–2508C, copper on zinc and 2–4 slow poisoning and activity
CO þ H2O ! CO2 þ H2 10–30 atm aluminum oxides accelerated sintering
by poisons
hydrocarbon hydrode 300–4008C, cobalt and molybdenum 1–10 slow coking, poisoning activity, mass
sulfurization 30 atm sulfides on aluminum by metal deposits in transfer, and
R2S þ 2 H2 ! H2S þ R2 oxide residues pressure drop
high temperature CO shift 350–5008C, Fe3O4 on chromia 1–4 slow sintering, pellet activity and
CO þ H2O ! H2 þ CO2 20–30 atm breakage due to pressure drop
steam
steam reforming, natural gas 500–8508C, nickel on calcium aluminate 1–3 sintering, sulfur- activity and
CH4 þ H2O ! CO þ 3 H2 30 atm or a-alumina poisoning, carbon pressure drop
formation, and
pellet breakage due
to plant upsets
ethylene partial oxidation 200–2708C, silver on a-alumina with 1–3 slow sintering, activity and
2 C2H4 þ O2 ! 2 C2H4O 10–20 atm alkali metal promoters poisoning by Cl, S selectivity
Table 15. ðContinued Þ
Operating Typical life Process affecting life Catalyst property
Reaction conditions Catalyst (years) of catalyst charge affected

butane oxidation to maleic 400–5208C, vanadium phosphorus 1–2 loss of P; attrition or activity and
anhydride C4H10 þ 1–3 atm oxide with transition pellet breakage; S, selectivity
3.5 O2 ! C4H2O3 þ 4 H2O metal additives Cl poisoning
reduction of aldehydes 220–2708C, copper on zinc oxide 0.5–1 slow sintering, pellet activity or
to alcohols 100–300 atm breakage (depends pressure drop
RCHO þ H2 ! RCH2OH on feedstock)
ammonia oxidation 800–9008C, Pt–Rh alloy gauze 0.1–0.5 surface roughness, selectivity
2 NH3 þ 5/2 O2 ! 1–10 atm loss of platinum,
2 NO þ 3 H2O fouling by Fe
oxychlorination of ethylene 230–2708C, copper chlorides on 0.2–0.5 loss by attrition and fluidized state
to ethylene dichloride 1–10 atm alumina (fluidized other causes and activity
2 C2H4 þ 4 HCl þ O2 ! bed) resulting from
2 C2H4Cl2 þ 2 H2O plant upsets
catalytic hydrocarbon 460–5258C, platinum alloys on 0.01–0.5 coking, frequent activity and

594
reforming 8–50 atm treated alumina regeneration mass transfer
catalytic cracking of oils 500–5608C, synthetic zeolites 0.000002 very rapid coking activity and
2–3 atm (continuous mass transfer
(fluidized bed) regeneration)
a
Adapted from Ref. 9.
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 595

Fig. 8. Approaches to eliminating catalyst deactivation.

are based on one important fundamental principle, namely that carbon or coke
results from a balance between the reactions that produce atomic carbon or
coke precursors and the reactions of these species with H2, H2O, or O2 that remove
them from the surface. If the conditions favor formation over gasification, these
species accumulate on the surface and react further to less active forms of carbon
or coke, which either coat the surface with an inactive film or plug the pores,
causing loss of catalyst effectiveness, pore plugging, or even destruction of the
carrier matrix.
Methods to lower rates of formation of carbon or coke precursors relative to
their rates of gasification vary with the mechanism of formation (ie, gas, surface,
or bulk phase) and the nature of the active catalytic phase (eg, metal or oxide).
For example, gas phase formation can be minimized by choosing reaction condi-
tions that minimize the formation of free radicals, by using free-radical traps, by
introducing gasifying agents (eg, H2, H2O) or gas diluents, and by minimizing
the void space available for homogeneous reaction. Similarly, the formation
and growth of carbon or coke species on metal surfaces is minimized by choosing
reaction conditions that minimize the formation of atomic carbon or coke precur-
sors and by introducing gasifying agents. Selective membranes or supercritical
conditions can also be used to lower the gas-phase and surface concentrations
of coke precursors. Since carbon or coke formation on metals apparently requires
a critical ensemble of surface metal atoms and/or dissolution of carbon into the
bulk metal, introduction of modifiers that change ensemble sizes (eg, Cu or S in
Ni or Ru) or that lower the solubility of carbon (eg, Pt in Ni) can be effective in
minimizing these forms of deactivation.
596 CATALYST DEACTIVATION AND REGENERATION Vol. 1

Table 16. Methods for Preventing Catalyst Decay


Basic
mechanism Problem Cause Methods of minimization
chemical oxidation of oxidation of (1) purify feed of oxidants; (2) mini-
degradation metal metal by mize reactant/product water by
catalysts to contaminant recycle/separation, staged reactors,
inactive O2 or and otherwise limiting conversion;
oxides reactant/ (3) incorporate additives that
product water facilitate resistance to oxidation
transformation solid-state reac- (1) avoid conditions (eg, oxidizing
of active tion of active con-dition, high steam pressures,
phase to phase with and high temperatures) that favor
stable, inac- support or solid-state reactions and (2) select
tive phase promoters combinations of active phase and
promoters/supports that are
noninteracting
overreduction (1) stabilize oxidation state using
of active oxide promoters that induce resistance
phases to reduction or that serve as oxy-
gen donors and (2) add steam to the
reactants to prevent overreduction
fouling by coke loss of catalytic free-radical (1) avoid formation of free radicals,
or carbon surface sites reactions in lower temp.; (2) minimize free
due to for- gas phase space; (3) free radical traps,
mation of diluents; (4) add gasifying agents
carbon or (eg, H2, H2O)
coke films
free-radical (1) coat reactor with inert material
reactions at
reactor walls
formation and (1) avoid accumulation of coke pre-
growth on cursors (eg, atomic carbon, olefins)
metal through careful choice of reactant
surfaces conditions or membranes; (2) add
gasifying agents (eg, H2, H2O),
diluents; (3) incorporate catalyst
additives to increase rate of gasifi-
cation or to change ensemble size;
(4) passivate metal surfaces with
sulfur; (5) decrease dispersion; and
(6) recycle inerts to flush surface of
heavy oligomers and to moderate
temperature
formation and (1) utilize measures 1, 2, 3, and 6 for
growth on metal surfaces; (2) design catalyst
metal oxides, for optimum pore structure and
sulfides acidity; and (3) use shape-selective,
coke-resistant molecular sieves
loss of catalyst formation of gas (1) minimize formation of free
effective- phase coke, radicals or coke precursors as
ness; plugg- vermicular above; (2) use gasifying agents;
ing of pores; carbons, and (3) incorporate catalyst additives
destruction liquid or solid that lower solubility of carbon in
of catalyst cokes in metal or that change ensemble
massive size; (4) use supports with large
quantities pores, large pellets
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 597

Table 16. ðContinuedÞ


Basic
mechanism Problem Cause Methods of minimization

hot spots in (1) use wash coat or small pellets;


pellet or bed (2) use slurry- or fluid-bed
reactor, gas diluents
mechanical crushing of brittle fracture (1) minimize porosity of pellets or
failure granules, due to a monoliths; (2) improve bonding of
pellets, or mechanical primary particles in agglomerates
monoliths in load that make up pellets or monoliths
a fixed bed using advanced forming methods,
eg, spray drying and controlled
thermal treatments; (3) add bin-
ders such as carbon to the support
material, which facilitate plastic
deformation and thus protect
against brittle fracture; and
(4) chemically or thermally
temper agglomerates
attrition and/ abrasion of (1) avoid highly turbulent shear flows
or erosion catalyst and/or cavitation, leading to high
in fixed or coatings or erosion rates; (2) avoid thermal
moving beds particles due stresses in the preparation and use
to mechanical, of catalysts that lead to fracture or
thermal, or separation of coatings; and (3) avoid
chemical formation of chemical phases of
stresses substantially different densities or
growth of carbon filaments that
cause fracture of primary particles
and agglomerates. Choose supports,
support additives, and coating
materials such as titanates, zir-
conia, and zirconates, having high
fracture toughness
poisoning loss of blockage of sites (1) purify feed and/or use guard bed
catalytic by strong to adsorb poison; (2) employ addi-
surface adsorption of tives that selectively adsorb poison;
sites impurity (3) choose reaction conditions that
lower adsorption strength;
(4) optimize pore structure and
choose mass transfer regimes that
minimize adsorption of poison on
active sites; and (5) apply coating
that serves as diffusion barrier to
poison
thermal loss of metal metal particle (1) lower or limit reaction
degradation, area or subparticle temperature while facilitating
sintering migration at heat transfer; (2) add thermal
high stabilizers to catalyst; and
temperatures (3) avoid water
loss of support crystallization same as for avoiding loss of
area and/or struc- metal area
tural modifica-
tion or collapse
598 CATALYST DEACTIVATION AND REGENERATION Vol. 1

Coke deposition on oxide or sulfide catalysts occurs mainly on strongly


acidic sites; accordingly the rate of coking can be lowered by decreasing the acid-
ity of the support. For example, silanation of HY and HZSM-5 zeolites decreases
their activities but improves catalyst life (245). In steam reforming, certain
catalyst additives, eg, MgO, K2O, or U3O8, facilitate H2O or CO2 adsorption
and dissociation to oxygen atoms, which in turn gasify coke precursors
(8,48,55).
As in the case of poisoning (see below), there are certain reactor bed or cat-
alyst geometries that minimize the effects of coking on the reaction. For example,
specific film-mass transport or pore diffusion regimes favor coke or carbon
deposition on either the outside or inside of the catalyst pellet (246,247). Choos-
ing supports with relatively large pores minimizes pore plugging; choice of large-
diameter, mechanically-strong pellets avoids or delays reactor plugging. But in
view of the rapidity at which coke and carbon can deposit on, plug, and even
destroy catalyst particles, the importance of preventing the onset of such forma-
tion cannot be overemphasized.
Reforming of naphtha provides an interesting case study of catalyst and
process designs to avoid deactivation by coking (8,206,208,248). The classical
Pt/Al2O3 catalyst is bifunctional; that is, the metal catalyzes dehydrogenation
while the acid sites of the Al2O3 catalyze isomerization and hydrocracking.
Together the two functions catalyze dehydrocylization and aromatization. Addi-
tion of Re, Sn, or Ge to Pt and sulfiding of the Pt–Re catalyst substantially
reduce coke formation by diluting large Pt ensembles that would otherwise pro-
duce large amounts of coke, while addition of Sn and Ir improves selectivity for
dehydrogenation relative to hydrogenolysis, the latter of which leads to coke
formation. Naphtha reforming processes are designed for (1) high enough H2
pressure to favor gasification of coke precursors while minimizing hydrocrack-
ing, (2) maintenance of Cl and S contents throughout the bed to ensure optimum
acidity and coke levels, and (3) low enough overall pressure to thermodynami-
cally and kinetically favor dehydrogenation and dehydrocylization. Accordingly,
optimal process conditions are a compromise between case 1 and case 3. The
above-mentioned improvements in catalyst technologies, especially resistance
to coking, have enabled important process improvements such as optimal opera-
tion at lower pressure; thus, processes have evolved over the past two to three
decades from conventional fixed-bed reactors at high pressure (35 bar) using non-
regenerative Pt catalysts to low pressure (3.5 bar), slowly moving-bed, continu-
ously regenerated units with highly selective Pt/Sn catalysts, resulting in
substantial economic benefits (248).
4.4. Prevention of Poisoning. Since poisoning is generally due to
strong adsorption of feed impurities and since poisoned catalysts are generally
difficult or impossible to regenerate, it is best prevented by removal of impurities
from the feed to levels that will enable the catalyst to operate at its optimal life-
time. For example, it is necessary to lower the feed concentration of sulfur com-
pounds in conventional methanation and Fischer–Tropsch processes involving
base metal catalysts to less than 0.1 ppm in order to ensure a catalyst lifetime
of 1–2 years. This is typically accomplished using a guard bed of porous ZnO
at about 2008C. In cracking or hydrocracking reactions on oxide catalysts, it is
important to remove strongly basic compounds such as ammonia, amines, and
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 599

Mo/alumina
[metals removal]
large pore

[finish HDM, CoMo/alumina


initiate HDS] medium pore

CoMo/alumina
[finish HDS]
small pore

Fig. 9. Staged reactor system with decreasing pore size strategy for HDM/HDS of
residues (224).

pyridines from the feed; ammonia in some feedstocks, for example, can be
removed by aqueous scrubbing. The poisoning of catalysts by metal impurities
can be moderated by selective poisoning of the unwanted metal. For example,
in catalytic cracking of nickel-containing petroleum feedstocks, nickel sites,
which would otherwise produce copious amounts of coke, are selectively poisoned
by antimony (249). The poisoning of hydrotreating catalysts by nickel and vana-
dium metals can be minimized by (1) using a guard bed of inexpensive Mo cata-
lyst or a graded catalyst bed with inexpensive, low-activity Mo at the top (bed
entrance) and expensive, high-activity catalyst at the bottom (see Fig. 9) and
(2) depositing coke prior to the metals since these metal deposits can be physi-
cally removed from the catalyst during regeneration (250).
It may be possible to lower the rate of poisoning through careful choice of
reaction conditions that lower the strength of poison adsorption (48) or by choos-
ing mass-transfer-limiting regimes that limit deposits to the outer shell of the
catalyst pellet, while the main reaction occurs uninterrupted on the interior of
the pellet (246). The manner in which the active catalytic material is deposited
on a pellet (eg, uniformly or in an eggshell or egg yolk pattern) can significantly
influence the life of the catalyst (17,251).
An example of reducing catalyst poisoning (and oxidation) through process
design has been reported in a process patent for staged hydrocarbon synthesis
via the Fischer–Tropsch reaction (252). While cobalt catalysts are favored
because of their high activities and while it is desirable to achieve high conver-
sions of CO in the process, the one-pass conversion for cobalt is limited by (1) its
tendency to be oxidized at high partial pressures of product water observed at
high CO conversions and (2) its tendency to form under these conditions the
oxygenated products (eg, alcohols and aldehydes) that poison or suppress its
synthesis activity. One alternative is to separate products and recycle the unused
CO and H2, but this requires costly recompression and separation of the oxyge-
nates. Costly separation and/or poisoning can be prevented by operating a first-
stage reactor containing a cobalt catalyst to a moderately high conversion
followed by reacting the remaining CO and H2 in a second stage to above 95%
600 CATALYST DEACTIVATION AND REGENERATION Vol. 1

conversion on an iron catalyst, which is not sensitive to the oxygenates and


which shifts some of the product water to H2 and CO2, thus minimizing its
hydrothermal degradation.
An example of reducing catalyst poisoning through catalyst design occurs in
abatement of emissions for automotive and motorcycle engines (18,222). Applica-
tion of an alumina or zeolite coating or alternatively preparing the active phase
in a sublayer provides a diffusion barrier that prevents or slows the access of poi-
sons from the fuel or oil (eg, phosphorus and/or zinc from lubricating oil or corro-
sion products) to the catalyst surface. The principle is to optimize the pore size
distribution of the diffusion barrier to provide access to the catalytic phase of
relatively small hydrocarbon, CO, NO, and O2 molecules while preventing access
of larger molecules such as from lubricating oil and/or particulates.
4.5. Prevention of Sintering. Since most sintering processes are irre-
versible or are reversed only with great difficulty, it is important to choose reac-
tion conditions and catalyst properties that avoid such problems. Metal growth is
a highly activated process; thus by choosing reaction temperatures lower than
0.3–0.5 times the melting point of the metal, rates of metal sintering can be
greatly minimized. The same principle holds true in avoiding recrystallization
of metal oxides, sulfides, and supports. Of course, one approach to lowering reac-
tion temperature is to maximize activity and surface area of the active catalytic
phase.
Although temperature is the most important variable in the sintering pro-
cess, differences in reaction atmosphere can also influence the rate of sintering.
Water vapor in particular accelerates the crystallization and structural modifica-
tion of oxide supports. Accordingly, it is vital to minimize the concentration of
water vapor in high temperature reactions on catalysts containing high surface
area supports.
Besides lowering temperature and minimizing water vapor, it is possible to
lower sintering rates through addition of thermal stabilizers to the catalyst. For
example, the addition of higher melting noble metals (such as rhodium or ruthe-
nium) to a base metal (such as nickel) increases the thermal stability of the base
metal (253). Addition of Ba, Zn, La, Si, and Mn oxide promoters improves the
thermal stability of alumina (254).
Designing thermally stable catalysts is a particular challenge in high tem-
perature reactions such as automotive emissions control, ammonia oxidation,
steam reforming, and catalytic combustion. The development of thermally stable
automotive catalysts has received considerable attention, thus providing a
wealth of scientific and technological information on catalyst design (eg, Refs. 8
and 225–232). The basic design principles are relatively simple: (1) utilize
thermally and hydrothermally stable supports, eg, high-temperature d- or y-alu-
minas or alkaline-earth or rare-earth oxides that form ultrastable spinels with g-
alumina; (2) use PdO rather than Pt or Pt–Rh for high temperature converters,
since PdO is considerably more thermally stable in an oxidizing atmosphere
because of its strong interaction with oxide supports; and (3) use multilayer stra-
tegies and/or diffusion barriers to prevent thermally induced solid-state reac-
tions (eg, formation of Rh aluminate) and to moderate the rate of highly
exothermic CO and hydrocarbon oxidations. For example, a typical three-way
automotive catalyst may contain alkaline-earth metal oxides (eg, BaO) and
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 601

rare-earth oxides (eg, La2O3 and CeO2) for stabilizing Pt and/or PdO on alumina
and ZrO2 as a thermal stabilizer for the CeO2, an oxygen storage material, and as
a noninteracting support for Rh in a separate layer or in a separate phase in a
composite layer.
4.6. Prevention of Mechanical Degradation. While relatively few
studies have focused on this topic, there are nevertheless principles that guide
the design of processes and catalysts in preventing or minimizing mechanical
degradation. In terms of catalyst design it is important to (1) choose supports,
support additives, and coatings that have high fracture toughness, (2) use pre-
paration methods that favor strong bonding of primary particles and agglo-
merates in pellets and monolith coatings, (3) minimize (or rather optimize)
porosity (thus maximizing density), and (4) use binders such as carbon to facil-
itate plastic deformation and thus protect against brittle fracture. Processes (and
to some extent preparation procedures) should be designed to minimize (1) highly
turbulent shear flows or cavitation that lead to fracture of particles or separation
of coatings, (2) large thermal gradients or thermal cycling leading to thermal
stresses, and (3) formation of chemical phases of substantially different densities
or formation of carbon filaments leading to fracture of primary particles and
agglomerates. Nevertheless, thermal or chemical tempering can be used in a con-
trolled fashion to strengthen catalyst particles or agglomerates.
Examples of catalyst design to minimize attrition can be found in the recent
scientific (239,240) and patent (241–244) literature focusing on the Fischer–
Tropsch synthesis in slurry reactors. These studies indicate that (1) spray drying
of particles improves their density and attrition resistance; (2) addition of silica
and/or alumina into titania improves its attrition resistance, while addition of
only 2000–3000 ppm of titania to g-alumina improves alumina’s attrition resis-
tance; and (3) preformed alumina spheres promoted with La2O3 provide greater
attrition resistance relative to silica. Increasing attrition resistance is apparently
correlated with increasing density (239,240,244). According to Singleton and co-
workers (244), attrition resistance of Co/Al2O3 is improved when the g-alumina
support is (1) formed from synthetic boehmite having a crystallite diameter of
4–5 nm and (2) pretreated in acidic solution having a pH of 1–3; moreover,
attrition resistance decreases in the order Co/Al2O3, Co/SiO2, Co/TiO2 and is
greater for catalyst prepared by aqueous versus nonaqueous impregnation.

5. Regeneration of Deactivated Catalysts

Despite our best efforts to prevent it, the loss of catalytic activity in most pro-
cesses is inevitable. When the activity has declined to a critical level, a choice
must be made among four alternatives: (1) restore the activity of the catalyst,
(2) use it for another application, (3) reclaim and recycle the important and/or
expensive catalytic components, or (4) discard the catalyst. The first alternative
(regeneration and reuse) is almost always preferred; catalyst disposal is usually
the last resort especially in view of environmental considerations.
The ability to reactivate a catalyst depends upon the reversibility of the
deactivation process. For example, carbon and coke formation is relatively easily
reversed through gasification with hydrogen, water, or oxygen. Sintering on the
602 CATALYST DEACTIVATION AND REGENERATION Vol. 1

other hand is generally irreversible, although metal redispersion is possible


under certain conditions in selected noble metal systems. Some poisons or fou-
lants can be selectively removed by chemical washing, mechanical treatments,
heat treatments, or oxidation (255,256); others cannot be removed without
further deactivating or destroying the catalyst.
The decision to regenerate/recycle or discard the entire catalyst depends
largely on the rate of deactivation. If deactivation is very rapid, as in the coking
of cracking catalysts, repeated or continuous regeneration becomes an economic
necessity. Precious metals are almost always reclaimed where regeneration is
not possible. Disposal of catalysts containing nonnoble heavy metals (eg, Cr,
Pb, or Sn) is environmentally problematic and should be a last resort; if disposal
is necessary, it must be done with great care, probably at great cost. Accordingly,
a choice to discard depends upon a combination of economic and legal factors
(256). Indeed, because of the scarcity of landfill space and an explosion of envir-
onmental legislation, both of which combine to make waste-disposal prohibitively
expensive, there is a growing trend to regenerate or recycle spent catalysts
(257,258). A sizeable catalyst regeneration industry benefits petroleum refiners
by helping to control catalyst costs and limiting liabilities (259,260); it provides
for ex situ regeneration of catalyst and recovery/recycling of metals, eg, of cobalt,
molybdenum, nickel, and vanadium from hydroprocessing catalysts (257).
Consistent with its importance the scientific literature treating catalyst
regeneration is significant and growing (includes several hundred journal arti-
cles since 1990), (eg. refs. 256,261–263).
The patent literature treating catalyst regeneration/reactivation is enor-
mous (more than 2000 patents); the largest fraction of this literature describes
processes for regeneration of catalysts in three important petroleum refining
processes, FCC, catalytic hydrocarbon reforming, and alkylation. However, a
significant number of patents also claim methods for regenerating absorbents
and catalysts used in aromatization, oligomerization, catalytic combustion,
SCR of NO, hydrocracking, hydrotreating, halogenation, hydrogenation,
isomerization, partial oxidation of hydrocarbons, carbonylations, hydroformy-
lation, dehydrogenation, dewaxing, Fisher –Tropsch synthesis, steam reform-
ing, and polymerization.
Conventional methods for regenerating (largely in situ) coked, fouled, poi-
soned, and/or sintered catalysts in some of these processes and representative
examples thereof (264–296) are summarized in Table 17, while the basic princi-
ples and limitations involved in regeneration of coked, poisoned, and sintered
catalysts are briefly treated in the subsections that follow.
5.1. Regeneration of Catalyst Deactivated by Coke or Carbon.
Carbonaceous deposits can be removed by gasification with O2, H2O, CO2, and
H2. The temperature required to gasify these deposits at a reasonable rate varies
with the type of gas, the structure and reactivity of the carbon or coke, and the
activity of the catalyst. Walker and co-workers (305) reported the following order
for rates of uncatalyzed gasification at 10 kN/m3 and 8008C (relative rates in
parenthesis): O2 (105) > H2O (3) > CO2 (1) > H2 (3 10 3). However, this activity
pattern does not apply in general for other conditions and for catalyzed reactions
(1). Nevertheless, the order of decreasing reaction rate of O2 > H2O > H2 can be
generalized.
Table 17. Conventional Methods for and Representative Examples of Catalyst Regeneration from Scientific and Patent Literatures
Deactivation mechanism/
reaction/catalyst Problem/cause Method(s) of regeneration/phenomena studied/conclusions Refs.
Deactivation by coke, carbon
alkene aromatization catalyst fouling by conden- (1) ZSM-5 catalyst for light olefin oligomerization containing 264,265
oligomerization/zeolites, sation of heavy oligomers 2–3% coke is treated in 8–10% steam/air mixture (1300 kPa,
esp. ZSM-5, -22, -23, to coke 938C inlet) in a fluidized bed (2) a coked crystalline alumogal-
beta-zeolite, ferrierite losilicate is contacted with oxygen at a concentration of 0.05–10
vol%, 420–5808C, and 300–4000 h 1
alkylation of isoparaffins on rapid catalyst deactivation (1) coked zeolite is regenerated in liquid phase (P > 3500 kPa) 266,267
solid catalysts/sulfated due to coke formation; fluid bed with H2 in two steps: (a)at reaction temperature (20–
zirconia, USYa, Nafion, unacceptable product 508C) and (b)at 258C above reaction temperature (2) coked Pd-
silicalite, ZSM-5 quality, and thermal and Pt/Y-zeolite catalysts containing 10–13% coke are regen-
degradation of catalyst erated in either air or H2; H2 treatment enables removal of most
during regeneration of the coke at low-moderate temperatures; higher tempera-
tures are required for air
catalytic reforming of poisoning and fouling by (1) coke on Pt bimetallic reforming catalyst is removed off-stream 268–273

603
naphtha/Pt/Al2O3 pro- coke produced by conden- in a moving bed at 300–6008C, followed by oxychlorination
moted with Re, Sn, Ge, or Ir sation of aromatics and (350–5508C) (2) coke on Pt/zeolite is removed in halogen-free
olefins oxygen-containing gas at T < 4158C (3) sintering during
oxidation of coke on Pt–Ir/Al2O3 catalyst can be minimized at
low regeneration temperature (4) study of influence of heating
rate, temperature, and time on structural properties of regen-
erated Pt–Sn/Al2O3 (5) study of effects of Cl, Sn content, and
regeneration sequence on dispersion and selectivity of
Pt–Sn/Al2O3 (6) regenerated Pt–Re/Al2O3 is more stable that
the fresh catalyst in n-heptane conversion and more selective
for toluene
dehydrogenation of propane catalyst activity is low owing (1) temperatures gradients were measured during burn off of coke 275,275
and butane/Cr2O3/Al2O3, to equilibrium limitations formed on a chromia–alumina catalyst during butene dehy-
Cr2O3/ZrO2, FeO/K/MgO, and build-up of product drogenation; data were used in developing a mathematical
Pt/Al2O3, Pt–Sn/Al2O3, H2; rapid loss of activity model for predicting temperatures and coke profiles (2) coked
Pt–Sn/KL-zeolite occurs owing to coke for- supported palladium catalyst used in the dehydrogenation of
mation dimethyltertrahydronaphthalenes to dimethylnaphthalenes is
reactivated with an organic polar solvent at a temperature
below 2008C
Table 17. ðContinuedÞ
Deactivation mechanism/
reaction/catalyst Problem/cause Method(s) of regeneration/phenomena studied/conclusions Refs.
Fischer–Tropsch synthesis/ loss of activity due to block- (1) carbidic surface carbon deposited on cobalt can be largely 276,277
Co/Al2O3 ing of sites by carbon removed in hydrogen at 170–2008C and in steam at 300–4008C
overlayers and heavy (2) slurry-phase cobalt catalysts may lose 50% activity during
hydrocarbons synthesis over a period of a few days; the activity can be reju-
venated in situ by injecting H2 gas into vertical draft tubes
inside the reactor
fluid catalytic cracking (FCC) rapid loss of activity due to (1) process and apparatus for increasing the coke burning capa- 278,279
of heavy hydrocarbons/ poisoning of acid sites and city of FCC regenerators; auxiliary regenerator partially burns
USY or REO-Yb in silica blocking of small zeolite off the coke at turbulent or fast fluidized-bed conditions (2)
matrix pores by coke multistage fluidized-bed regeneration of spent FCC catalyst in
a single vessel by incorporating two relatively dense phase
fluidized beds beneath a common dilute phase region
hydrocracking of heavy loss of activity due to poi- (1) regeneration of noble metal/zeolite via progressive partial 280,281
naphtha/CoMo, NiW, MoW soning of acid sites and removal of carbonaceous deposits under controlled oxidizing

604
on Al2O3 or SiO2 –Al2O3; Pt blocking of small zeolite conditions to maximize sorption of a probe molecule while
or Pd on Y-zeolite, morde- pores by coke minimizing metal sintering (2) regeneration of noble metal/
nite, or ZSM-5 zeolite in air at about 6008C, followed by a mild treatment in
aqueous ammonia to improve catalytic activity
hydrotreating of gas oil loss of activity due to forma- (1) TPO studies of oxidative regeneration of CoMo and NiW HDS 282,283,
tion of types I, II, and III catalysts; sulfur is removed at 225–3258C, carbon at 375– 297–300
coke on metal sulfide and 5758C. Redispersion of NiW was observed by EXAFS (2) phy-
alumina surfaces and in sicochemical changes in CoMo and NiCoMo HDS catalysts
pores during oxidative regeneration, including redispersion of Co, Ni,
and Mo oxides and surface area loss, were examined (3) changes
in NiMo catalyst structure and coke composition during reac-
tion and regeneration were examined and correlated (4) prop-
erties of NiMo catalyst deactivated during shale oil
hydrogenation and regenerated in O2 or H2 were examined.
regeneration in 1.6% O2 was more effective than that in 5% H2.
Ni aluminate spinel was observed after burn off (5) hard and
soft cokes formed on CoMo catalysts during HDS of gas oil were
characterized. At low coke levels, hard coke was more easily
removed in H2 than in O2 (6) spent catalysts are washed with
solvent and contacted with steam at about 6008C
methanol to olefins or gaso- severe coking and deactiva- (1) kinetics of coke burnoff from a SAPO-34 used in converting 284,285
line/silica–alumina, Y-zeo- tion of silica–alumina and methanol to olefins were studied; kinetics are strongly depen-
lite, ZSM-5, other zeolites, Y-zeolite catalysts dent on the nature of the coke. Kinetics are slowed by strong
and aluminophosphate observed during high con- binding of coke to acid sites (2) ZSM-34 catalyst used in con-
molecular sieves versions of methanol, also version of methanol to light olefins is effectively regenerated in
substantial coking of H2-containing gas; this approach avoids the formation of cata-
ZSM-5, other lyst-damaging products such as steam that would be formed
zeolites, and aluminopho- during burn off in air
sphate molecular sieves
Poisoning
FCC of residues/USY or (1) poisoning of acid sites by (1) organometallic solutions of Sb and Bi are added to process 284,285,
REO-Y in silica matrix N- containing compounds. steam to passivate Ni by forming inactive Ni–Sb and Ni–Bi 301–304
(2) deposition of Ni and V species (2) V metal deposits are trapped by reaction with
metals on acid sites which magnesium orthosilicate to form an unreactive magnesium
change selectivity and vanadium silicate (3) spent metal-contaminated catalyst is
decrease activity demetallized by chlorinating and washing followed by con-
tacting with NH4F and one antimony compound (4) metal-
contaminated catalyst is contacted with an aqueous solution
of a carboxylic acid (eg, formic, acetic, citric, or lactic acid) (5)

605
metal-contaminated catalyst is contacted with HCl, HNO3, or
H2SO4 (6) metal contaminated catalyst is contacted with
reducing CO gas to form gaseous metal carbonyls that sepa-
rated from the catalyst
hydrogenation or poisoning of metal sites by (1) regeneration of Ni/SiO2 catalyst poisoned by thiophene using a 288,289
dechlorination arsenic, sulfur, and other sequence of oxidation–reduction treatments at low PO2 and 1
poisons atm H2 respectively (2) regeneration in dilute hypochlorite
solution of a Pd/Al2O3 catalyst deactivated during the aqueous-
phase dechlorination of trichloroethylene in the presence of
sulfite or HS ions present in ground water
hydrotreating of residues/ pore-mouth poisoning and (1) regeneration of catalysts containing V, Ni, or Fe by contacting 290,291
Al2O3-supported Mo and blockage by Ni, V, and Fe with H2O2 solution and organic acid (2) following removal of
CoMo sulfides present in feed as coke by air or solvent wash, catalyst is acid leached to remove
organometallics undesired metals
Table 17. ðContinuedÞ
Deactivation mechanism/
reaction/catalyst Problem/cause Method(s) of regeneration/phenomena studied/conclusions Refs.

Thermal degradation
Catalytic reforming of sintering of Pt causing (1) redispersion of Pt–Ir bimetallic catalysts using a wet HCl/air 270,273,
naphtha/Pt/Al2O3 pro- formation of large metal treatment, since the conventional oxychlorination is not effec- 292,293
moted with Re, Sn, Ge, or crystallites crystals and tive (2) redispersion of Pt/KL-zeolite using wet HCl/air treat-
Ir; Pt/KL-zeolite loss of active surface areament followed by brief calcination and reduction (3)
redispersion of Pt–Re/Al2O3 in Cl2 and O2 (4) redispersion of
supported Pt, other noble metals, and Ni in Cl2 and O2
hydrocracking of heavy sintering of noble metal redispersion of noble metals on molecular sieves including 294
naphtha/CoMo, NiW, MoW causing formation of large silica-aluminates, ALPOS, SAPOS
on Al2O3 or SiO2 –Al2O3; metal crystallites crystals
Pt or Pd on Y-zeolite, and loss of active surface
mordenite, or ZSM-5 area
hydrotreating of gas oil and sintering of Mo and Co sul- (1) oxidative regeneration of hydroprocessing catalyst at 6008C 295,296
residues/Al2O3-supported fides causing formation of optimizes surface area and Mo dispersion (2) oxidative regen-

606
Mo and CoMo large sulfide crystals and eration in several steps with a final oxidation at 500–6008C to
loss of active surface area restore residual catalyst activity

a
USY: ultrastable Y-zeolite.bREO-Y: rare-earth exchanged Y-zeolite.
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 607

Rates of gasification of coke or carbon are greatly accelerated by the same


metal or metal oxide catalysts upon which carbon or coke deposits.
Because catalyzed removal of carbon with oxygen is generally very rapid at
moderate temperatures (eg, 400–6008C), industrial processes typically regener-
ate catalysts deactivated by carbon or coke in air. Indeed, air regeneration is
used to remove coke from catalysts in catalytic cracking (64), hydrotreating pro-
cesses (261), and catalytic reforming (262).
One of the key problems in air regeneration is avoiding hot spots or over-
temperatures which could further deactivate the catalyst. The combustion pro-
cess is typically controlled by initially feeding low concentrations of air and by
increasing oxygen concentration with increasing carbon conversion (261,306);
nitrogen gas can be used as a diluent in laboratory-scale tests while steam is
used as a diluent in full-scale plant operations (306). For example, in the regen-
eration of hydrotreating catalysts McCulloch (261) recommends keeping the tem-
perature at less than 4508C to avoid the g- to a-alumina conversion, MoO3
sublimation, and cobalt or nickel aluminate formation, which occur at 815,
700, and 500–6008C respectively.
Because coke burn-off is a rapid, exothermic process, the reaction rate
is controlled to a large extent by film heat and mass transfer. Accordingly, burn-
off occurs initially at the exterior surface and then progresses inward with the
reaction occurring mainly in a shrinking shell consistent with a ‘‘shell-progres-
sive’’ or ‘‘shrinking-core’’ model (307); as part of this same work, Richardson (307)
showed how experimental burn-off rate data can be fitted to various coking
transport models, eg, parallel or series fouling. Burn-off rates for coke deposited
on SiO2/Al2O3 catalysts were reported by Weisz and Goodwin (308); burning
rate was found to be independent of initial coke level, coke type, and source of
catalyst.
5.2. Regeneration of Poisoned Catalysts. Much of the previous lit-
erature has focused on regeneration of sulfur-poisoned catalysts used in hydro-
genations and steam reforming. Studies of regeneration of sulfur-poisoned Ni,
Cu, Pt, and Mo with oxygen/air, steam, hydrogen, and inorganic oxidizing agents
have been reported (27). Rostrup-Nielsen (309) indicates that up to 80% removal
of surface sulfur from Mg- and Ca-promoted Ni, steam reforming catalysts occurs
at 7008C in steam. The presence of both SO2 and H2S in the gaseous effluent sug-
gests that the following reactions occur:

Ni S þ H2 O ! NiO þ H2 S ð5Þ

H2 S þ 2 H2 O ! SO2 þ 3 H2 ð6Þ

Although this treatment is partially successful in the case of low-surface-


area steam reforming catalysts, the high temperatures required for these reac-
tions would cause sintering of most high-surface-area nickel catalysts.
Regeneration of sulfur-poisoned catalysts, particularly base metal cata-
lysts, in air or oxygen has been largely unsuccessful. For example, the treatment
of nickel steam-reforming catalysts in steam and air results in the formation of
sulfates, which are subsequently reduced back to nickel sulfide upon contact with
hydrogen. Nevertheless, sulfur can be removed as SO2 at very low oxygen partial
608 CATALYST DEACTIVATION AND REGENERATION Vol. 1

pressures, suggesting that regeneration is possible under carefully controlled


oxygen or species such as CO2 or NO that dissociate to oxygen. Apparently,
at low oxygen pressures the oxidation of sulfur to SO2 occurs more rapidly
than the formation of nickel oxide while at atmospheric pressure the converse
is true, ie, the sulfur or sulfate layer is rapidly buried in a nickel oxide layer. In
the latter circumstance, the sulfur atoms diffuse to the nickel surface during
reduction, thereby restoring the poisoned surface. Regeneration of sulfur-
poisoned noble metals in air is more easily accomplished than with steam,
although it is frequently attended by sintering. Regeneration of sulfur-poi-
soned nickel catalysts using hydrogen is impractical because (1) adsorption
of sulfur is reversible only at high temperatures at which sintering rates are
also high, and (2) rates of removal of sulfur in H2 as H2S are slow even at
high temperature.
Inorganic oxidizing agents such as KMnO4 can be used to oxidize liquid
phase or adsorbed sulfur to sulfites or sulfates (16). These electronically
shielded structures are less toxic than the unshielded sulfides. This approach
has somewhat limited application, ie, in partial regeneration of metal catalysts
used in low temperature liquid-phase hydrogenation reactions or in liquid-
phase destruction of chlorinated organic compounds. For example, Lowrey
and Reinhard (289) reported successful regeneration in dilute hypochlorite
solution of a Pd/Al2O3 catalyst deactivated during the aqueous-phase dechlor-
ination of trichloroethylene (TCE) in the presence of sulfite or HS ions. These
poisons are formed by sulfate-reducing bacteria present in natural ground-
water and are apparently adsorbed on the alumina or Pd surfaces more
strongly than sulfate ions.
5.3. Redispersion of Sintered Catalysts. During catalytic reforming of
hydrocarbons on platinum-containing catalysts, growth of 1-nm platinum metal
clusters to 5–20-nm crystallites occurs. An important part of the catalyst regenera-
tion procedure is the redispersion of the platinum phase by a high temperature
treatment in oxygen and chlorine, generally referred to as ‘‘oxychlorination.’’ A typi-
cal oxychlorination treatment involves exposure of the catalyst to HCl or CCl4 at
450–5508C in 2–10% oxygen for a period of 1–4 h (see details in Table 18). During

Table 18. Typical Regeneration Procedure for Reforming Catalystsa


1. Preliminary operations:
cool the catalyst to about 2008C and strip hydrocarbons and H2 with N2
2. Elimination of coke by combustion:
inject dilute air (0.5% O2) at 3808C and gradually increase oxygen content to about
2% by volume while maintaining temperature below 450–5008C to prevent further
sintering of the catalyst. To prevent excessive leaching of Cl2, HCl or CCl4 may be
injected during the combustion step
3. Restoration of catalyst acidity:
Restoration of acidity occurs at 5008C by injection of a chlorinated
compound in the presence of 100–200 ppm water in air
4. Redispersion of the metallic phase:
expose the catalyst to a few Torr of HCl or CCl4 in 2–10% O2 in N2 at 510–5308C for a
period of about 4 h. After redispersion, O2 is purged from the unit and the catalyst is
reduced in H2
a
Ref. 261 and 262.
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 609

coke burning some redispersion occurs, eg, D increases from 0.25 to 0.51, while dur-
ing oxychlorination the dispersion is further increased, eg, from 0.51 to 0.81 (262).
Some guidelines and principles regarding the redispersion process are
worth enumerating:

1. In cases involving a high degree of Pt sintering or poisoning, special regen-


eration procedures may be required. If large crystallites have been formed,
several successive oxychlorinations are performed (262).
2. Introducing oxygen into reactors in parallel rather than in series results in
a significant decrease in regeneration time (84).
3. Introduction of hydrocarbons present in the reactor recycle after regenera-
tion is said to stabilize the catalyst; solvents such as ammonium acetate,
dilute nitric acid containing lead nitrate, EDTA and its diammonium salt
are reported to dissolve out metal aggregates without leaching out the dis-
persed metal (84).
4. The procedures for redispersion of Pt/alumina are not necessarily applic-
able to Pt on other supports or to other metals. For example, Pt/silica is re-
dispersed at lower temperature and higher Cl2 concentration (150–2008C
and 25% Cl2). Pd/alumina can be redispersed in pure O2 at 5008C. While
Pt–Re/alumina is readily redispersed by oxychlorination at 5008C, Pt–Ir/
alumina is not redispersed in the presence of O2 unless the catalyst is pre-
treated with HCl (270).

An extensive scientific and patent literature of redisperson describes the


use of chlorine, oxygen, nitric oxide, and hydrogen as agents for redispersion
of sintered catalysts. Most of the early literature shows positive effects for chlo-
rine compounds in the presence of oxygen in redispersing alumina-supported
platinum and other noble metals. Recent literature demonstrates the need for
understanding the detailed surface chemistry in order to successfully develop
and improve redispersion processes, especially in more complex catalyst sys-
tems such as alumina-supported bimetallics. For example, on the basis of a
fundamental study of the redispersion surface chemistry, Fung (270) devel-
oped a redispersion procedure for Pt–Ir bimetallic catalysts using a wet HCl/
air treatment, since the conventional oxychlorination is not effective for this
catalyst.
Redispersion of alumina-supported platinum and iridium crystallites is also
possible in a chlorine-free oxygen atmosphere if chlorine is present on the cata-
lyst. The extent of redispersion depends on the properties of the Pt/Al2O3 catalyst
and temperature. The question whether redispersion of platinum occurs only in
oxygen without chlorine present on the catalyst remains controversial.
Two models, ‘‘the thermodynamic redispersion model’’ and ‘‘the crystallite
splitting model,’’ have been advanced to explain the redispersion in oxygen
(84,85,310). The ‘‘thermodynamic’’ redispersion model hypothesizes the forma-
tion of metal oxide molecules that detach from the crystallite, migrate to active
sites on the support, and form surface complexes with the support. Upon subse-
quent reduction, the metal oxide complexes form monodisperse metal clusters. In
the ‘‘crystallite splitting’’ model, exposure of a platinum crystallite to oxygen at
5008C leads to formation of a platinum oxide scale on the outer surface of the
610 CATALYST DEACTIVATION AND REGENERATION Vol. 1

crystallite, which stresses and ultimately leads to splitting of the particle (310).
Dadyburjor hypothesizes that the crystallite splitting model is most applicable to
the behavior of large crystallites and to all particles at relatively small regenera-
tion times while the thermodynamic migration model is useful for small particles
and most particles after longer regeneration times.

6. Summary and Perspective

6.1. Summary

1. The causes of deactivation are basically of three kinds: chemical, mechan-


ical, and thermal. The five intrinsic mechanisms of catalyst decay, (a) poi-
soning, (b) fouling, (c) thermal degradation, (d) chemical degradation, and
(e) mechanical failure, vary in their reversibility and rates of occurrence.
Poisoning and thermal degradation are generally slow, irreversible pro-
cesses while fouling with coke and carbon is generally rapid and reversible
by regeneration with O2 or H2.
2. Catalyst deactivation is more easily prevented than cured. Poisoning by im-
purities can be prevented through careful purification of reactants. Carbon
deposition and coking can be prevented by minimizing the formation of car-
bon or coke precursors through gasification, careful design of catalysts and
process conditions, and by controlling reaction rate regimes, eg, mass
transfer regimes, to minimize effects of carbon and coke formation on activ-
ity. Sintering is best avoided by minimizing and controlling the tempera-
ture of reaction.
3. Prevention and monitoring are important engineering principles in ‘‘stan-
dard of care’’ practice. The prevention of catalyst decay is important in
every aspect of a process including design, construction, operation, and re-
generation. Careful monitoring of process variables is a necessity in under-
standing and preventing catalyst decay problems of either a slow or a
catastrophic nature.
4. The optimization of a catalytic process considers optimum operation and re-
generation policies subject to constraints of catalyst cost, operation cost, re-
generation cost, and product value. The optimum operating policy
maximizes the rate of formation of product during the operating period.
5. Catalyst deactivation kinetics for reactions involving relatively slow deac-
tivation can be experimentally determined using a laboratory fixed-bed,
mixed-fluid (CSTR) reactor. Reactors and processes involving a slowly de-
activating catalyst can be designed using relatively simple numerical ana-
lysis of the design equations and a pseudo-steady-state approximation for
the main reaction.
6. Modeling and experimental assessment of deactivation processes are useful
in providing (a) accelerated simulations of industrial processes, (b) predic-
tive insights into effects of changing process variables on activity, selectiv-
ity, and life, (c) estimates of kinetic parameters needed for design and
modeling, (d) estimates of size and cost for scale-up of a process, and
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 611

(e) a better understanding of the basic decay mechanisms. It is now possible


to develop realistic mathematical models of most catalytic processes, which
can be used in conjunction with short-term experimental tests to accurately
predict catalyst life in a commercial unit. Proper application of this ap-
proach could save companies millions of dollars by alleviating the need
for long-term deactivation tests and/or premature shutdown. For details
on this aspect of assessment, refer to the expanded version of this article
published in the Encyclopedia of Catalysis (78,311).

6.2. Perspectives and Trends. Research and development activities


in the area of catalyst deactivation have grown steadily in the past three
decades. Catalyst deactivation symposia are held annually as part of national
meetings of chemical engineering and chemical societies in the United States
and Europe. The rising quality of work presented at the international sympo-
sium on catalyst deactivation, held every four years, is evident. In view of the
importance of deactivation problems in industrial processes, this trend will
most probably continue.
Several other trends are evident:

1. The increasing use of more sophisticated analytical tools to investigate the


chemistry and mechanisms of deactivation. Surface science tools such as
AES, quantitative HRTEM, XPS, and STM are now routinely applied to in-
vestigate deactivation mechanisms at very fundamental levels.
2. The increasing development of more sophisticated models of deactivation
processes.

These trends are also likely to continue. Moreover, the combination of more
sophisticated methods and models will hasten the practical application of models
for predicting catalyst/process life. This is already happening in selected compa-
nies. For example, for more than a decade now operators at Phillips Petroleum
Co. have been using deactivation models (developed at their corporate research)
in their refineries to predict when shutdown will be necessary. One of these
models enables them to predict accurately the lifetime of hydrotreating catalysts
on the basis of catalyst and feedstock properties.
6.3. Future Needs. Collection of Data. It is evident from careful
examination of the literature that few deactivation rate data are available for
even the most important large-scale catalytic systems. Accordingly, there is a cri-
tical need for collection of such data at the laboratory, bench, and plant scale.
There is much that could be done with good data. Sophisticated analytical
tools and well-designed reactors are available at most companies for collecting
and analyzing such data. The field is ripe and ready to harvest. The wise will
seize these opportunities.
Data Analysis and Model Development. Much of the previously col-
lected data were analyzed using outdated methods. There is much that could be
learned by reanalyzing some of these data using new approaches such as the
GPLE and microkinetic modeling. Critical reviews that include collections of
carefully selected rate and kinetic data would constitute important contributions
612 CATALYST DEACTIVATION AND REGENERATION Vol. 1

to scientific knowledge and technological development. The incorporation of


these data into models would enable more sophisticated design of catalysts, reac-
tors, and processes.

BIBLIOGRAPHY

‘‘Catalyst Regeneration, Metal Catalysts’’ in ECT (online), posting date: December 4, 2000,
by D. W. Robinson, UOP.

1. J. L. Figuerido, in J. L. Figuerido, ed., Carbon Formation and Gasification on Nickel,


M. Nijhoff Publishers, Boston, 1982.
2. R. Hughes, Deactivation of Catalysts, Academic Press, London, 1984.
3. J. Oudar and H. Wise, eds., Deactivation and Poisoning of Catalysts, Marcel Dekker,
New York, 1985.
4. J. B. Butt and E. E. Petersen, Activation, Deactivation, and Poisoning of Catalysts,
Academic Press, San Diego, 1988.
5. P. J. Denny and M. V. Twigg, in Ref. 9, p. 577.
6. C. H. Bartholomew, Chem. Eng. 91, 96 (1984).
7. J. B. Butt, in J. R. Anderson and M. Boudart, eds., Catalysis—Science and Technol-
ogy, Vol. 6, Springer-Verlag, New York, 1984, p. 1.
8. R. J. Farrauto, C. H. Bartholomew, Fundamentals of Industrial Catalytic Processes,
Kluwer Academic Publishers, London, 1997.
9. B. Delmon and G. F. Froment, eds., Catalyst Deactivation 1980 (Studies in Surface
Science and Catalysis, Vol. 6), Elsevier, Amsterdam, 1980.
10. B. Delmon and G. F. Froment, eds., Catalyst Deactivation 1987 (Studies in Surface
Science and Catalysis, Vol. 34), Elsevier, Amsterdam, 1987.
11. C. H. Bartholomew and J. B. Butt, eds., Catalyst Deactivation 1991 (Studies in Sur-
face Science and Catalysis, Vol. 68), Elsevier, Amsterdam, 1991.
12. B. Delmon and G. F. Froment, eds., Catalyst Deactivation 1994 (Studies in Surface
Science and Catalysis, Vol. 88), Elsevier, Amsterdam, 1994.
13. C. H. Bartholomew and G. A. Fuentes, eds., Catalyst Deactivation 1997 (Studies in
Surface Science and Catalysis, Vol. 111), Elsevier, Amsterdam, 1997.
14. B. Delmon and G. F. Froment, eds., Catalyst Deactivation 1999 (Studies in Surface
Science and Catalysis, Vol. 126), Elsevier, Amsterdam, 1999.
15. J. A. Moulijn, ed., A series of papers on Catalyst Deactivation, Appl. Catal., A: Gen.
212, 1–255 (2001).
16. E. B. Maxted, Adv. Catal. 3, 129 (1951).
17. L. L. Hegedus and R. W. McCabe, in Ref. 9, p. 47.
18. L. L. Hegedus and R. W. McCabe, Catalyst Poisoning, Marcel Dekker, New York,
1984.
19. J. B. Butt, in J. L. Figuerido, ed., Progress in Catalyst Deactivation (NATO Advanced
Study Institute Series E, No. 54), M. Nijhoff Publishers, Boston, 1982, p. 153.
20. J. Barbier, in Ref. 3, p. 109.
21. C. H. Bartholomew, in Ref. 10, p. 81.
22. J. R. Rostrup-Nielsen, in Ref. 11, p. 85.
23. V. J. Volter and M. Hermann, Z. Anorg. Allg. Chem. 405, 315 (1974).
24. K. Baron, Thin Solid Films 55, 449 (1978).
25. R. D. Clay and E. E. Petersen, J. Catal. 16, 32 (1970).
26. R. J. Madon and H. Shaw, Catal. Rev. Sci. Eng. 15, 69 (1977).
27. C. H. Bartholomew, P. K. Agrawal, and J. R. Katzer, Adv. Catal. 31, 135 (1982).
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 613

28. J. R. Rostrup-Nielsen, in J. L. Figuerido, ed., Progress in Catalyst Deactivation


(NATO Advanced Study Institute Series E, No. 54), M. Nijhoff Publishers, Boston,
1982, p. 209.
29. H. Wise, J. McCarty, and J. Oudar, in Ref. 3, p. 1.
30. J. R. Rostrup-Nielsen and P. E. Nielsen, in Ref. 3, p. 259.
31. M. Perdereau and J. Oudar, Surf. Sci. 20, 80 (1970).
32. J. Oudar, Catal. Rev. Sci. Eng. 22, 171 (1980).
33. J. J. McCarroll, T. Edmonds, and R. C. Pitkethly, Nature 223, 1260 (1969).
34. T. Edmonds, J. J. McCarroll, and R. C. Pitkethly, J. Cat. Sci. Technol. 8, 68 (1971).
35. L. Ruan, F. Besenbacher, I. Stensgaard, and E. Laegsgaard, Phys. Rev. 69, 3523 (1992).
36. J. Hepola, J. McCarty, G. Krishnan, and V. Wong, Appl. Catal. B 20, 191 (1999).
37. W. Erley and H. Wagner, J. Catal. 53, 287 (1978).
38. K. D. Rendulic and A. Winkler, Surf. Sci. 74, 318 (1978).
39. D. W. Goodman and M. Kiskinova, Surf. Sci. 105, L265 (1981).
40. M. Kiskinova and D. W. Goodman, Surf. Sci. 108, 64 (1981).
41. S. Johnson and R. S. Madix, Surf. Sci. 108, 77 (1981).
42. R. J. Madix, M. Thornberg, and S. B. Lee, Surf. Sci. 133, L447 (1983).
43. E. L. Hardegree, P. Ho, and J. M. White, Surf. Sci. 165, 488 (1986).
44. E. J. Erekson and C. H. Bartholomew, Appl. Catal. 5, 323 (1983).
45. J. R. Rostrup-Nielsen and D. L. Trimm, J. Catal. 48, 155 (1977).
46. D. L. Trimm, Catal. Rev. Sci. Eng. 16, 155 (1977).
47. D. L. Trimm, Appl. Catal. 5, 263 (1983).
48. C. H. Bartholomew, Catal. Rev. Sci. Eng. 24, 67 (1982).
49. L. F. Albright and R. T. K. Baker, eds., Coke Formation on Metal Surfaces (ACS
Symposium Series 202), American Chemical Society, Washington, D.C., 1982.
50. P. G. Menon, J. Mol. Catal. 59, 207 (1990).
51. J. D. Deken, P. G. Menon, G. F. Froment, and G. Haemers, J. Catal. 70, 225 (1981).
52. W. G. Durer, J. H. Craig, Jr., and J. Lozano, Appl. Surf. Sci. 45, 275 (1990).
53. A. D. Moeller and C. H. Bartholomew, Prepr.—Am. Chem. Soc., Div. Fuel Chem. 25,
54 (1980).
54. K. J. Marschall and L. Mleczko, Ind. Eng. Chem. Res. 38, 1813 (1999).
55. J. R. Rostrup-Nielsen, in J. R. Anderson and M. Boudart, eds., Catalysis—Science
and Technology, Vol. 5, Springer-Verlag, New York, 1984, p. 1.
56. F. Besenbacher, I. Chorkendorff, B. S. Clausen, B. Hammer, A. M. Molenbroek, J. K.
Norscov, and I. Stensgaard, Science 279, 1913 (1998).
57. T. Nemes, A. Chambers, and R. T. K. Baker, J. Phys. Chem. 102, 6323 (1998).
58. C. H. Bartholomew, M. V. Strasburg, and H. Hsieh, Appl. Catal. 36, 147 (1988).
59. C. K. Vance and C. H. Bartholomew, Appl. Catal. 7, 169 (1983).
60. R. T. K. Baker and J. J. Chludzinski, J. Catal. 64, 464 (1980).
61. D. E. Brown, J. T. K. Clark, A. I. Foster, J. J. McCarroll, and M. L. Sims, in Ref. 49,
p. 23.
62. J. H. Bitter, K. Seshan, and J. A. Lercher, J. Catal. 183, 336 (1999).
63. J. R. Rostrup-Nielsen, J. Catal. 33, 184 (1974).
64. B. C. Gates, J. R. Katzer, and G. C. A. Schuit, Chemistry of Catalytic Processes,
McGraw-Hill, New York, 1979.
65. C. Naccache, in C. Naccache, ed., Deactivation of Acid Catalysts, Marcel Dekker,
New York, 1985.
66. W. G. Appleby, J. W. Gibson, and G. M. Good, Ind. Eng. Chem. Process Des. Dev. 1,
102 (1962).
67. H. Beuther, O. H. Larson, and A. J. Perrotta, in Ref. 9, p. 271.
68. A. G. Gayubo, J. M. Arandes, A. T. Aguayo, M. Olazar, and J. Bilbao, Ind. Eng.
Chem. Res. 32, 588 (1993).
614 CATALYST DEACTIVATION AND REGENERATION Vol. 1

69. S. M. Augustine, G. N. Alameddin, and W. M. H. Sachtler, J. Catal. 155, 217 (1989).


70. M. Guisnet and P. Magnoux, Appl. Catal. 54, 1 (1989).
71. F. Bauer, V. Karazirev, C. Vlaev, R. Hanisch, and W. Weiss, Chem. Techn. 41, 297
(1989).
72. W. A. Grotten, B. W. Wojciechowski, and B. K. Hunter, J. Catal. 138, 343 (1992).
73. A. Bellare and D. B. Dadyburjor, J. Catal. 140, 510 (1993).
74. M. A. Uguina, D. P. Serrano, R. V. Grieken, and S. Venes, Appl. Catal. 99, 97 (1993).
75. C. Li, Y. Chen, S. Yang, and R. Yen, Appl. Surf. Sci. 81, 465 (1994).
76. J. G. Buglass, K. P. d. Jong, and H. H. Mooiweer, in Proc. 120th National Meeting of
the American Chemical Society, Aug. 20–24, 1995, p. 631.
77. D. Chen, H. P. Rebo, K. Moljord, and A. Holmen, in Proc. 14th International Sympo-
sium on Chemical Reaction Engineering, Part B, May 5–9, 1996, p. 2687.
78. M. Gusinet, P. Magnoux, and D. Martin, in Ref. 13, p. 1.
79. T. Masuda, P. Tomita, Y. Fujikata, and K. Hashimoto, in Ref. 14, p. 89.
80. H. S. Cerqueira, P. Magnoux, D. Martin, and M. Gusinet, in Ref. 14, p. 105.
81. S. E. Wanke and P. C. Flynn, Catal. Rev. Sci. Eng. 12, 93 (1975).
82. P. Wynblatt and N. A. Gjostein, Prog. Solid State Chem. 9, 21 (1975).
83. E. Ruckenstein and B. Pulvermacher, AIChE J. 19, 356 (1973).
84. E. Ruckenstein and D. B. Dadyburjor, Rev. Chem. Eng. 1, 251 (1983).
85. S. E. Wanke, in J. L. Figueiredo, ed., Progress in Catalyst Deactivation (NATO
Advanced Study Institute Series E, No. 54), M. Nijhoff Publishers, Boston, 1982, p. 315.
86. R. T. Baker, C. H. Bartholomew, and D. B. Dadyburjor, Stability of Supported Cat-
alysts: Sintering and Redispersion, Catalytic Studies Division, 1991.
87. C. H. Bartholomew, Catalysis (Spec. Period. Rept.). 10 (1992).
88. C. H. Bartholomew, Appl. Catal., A: Gen. 107, 1 (1993).
89. C. H. Bartholomew, in Ref. 12, p. 1.
90. C. H. Bartholomew, in Ref. 13, p. 585.
91. C. H. Bartholomew and W. Sorenson, J. Catal. 81, 131 (1983).
92. J. A. Moulijn, A. E. van Diepen, and F. Kapteijn, Appl. Catal., A: Gen. 212, 13–16 (2001).
93. G. W. Bridger, and M. S. Spencer, in M. V. Twigg ed., Catalyst Handbook, 2nd ed.,
Manson Publishing, London, 1996, p. 441.
94. G. A. Fuentes, Appl. Catal. 15, 33 (1985); G. A. Fuentes and F. A. Ruiz-Trevino, in
Ref. 11, pp. 637–644.
95. J. P. Bournonville and G. Martino, in Ref. 9, p. 159.
96. G. A. Somorjai, X-ray and Electron Methods of Analysis, Plenum Press, New York,
1968.
97. S. R. Seyedmonir, D. E. Strohmayer, G. J. Guskey, G. L. Geoffroy, and M. A.
Vannice, J. Catal. 93, 288 (1985).
98. D. L. Trimm, in Ref. 11, p. 29.
99. A. G. Shastri, A. K. Datye, and J. Schwank, Appl. Catal. 14, 119 (1985).
100. L. L. Hegedus and K. Baron, J. Catal. 54, 115 (1978).
101. J. Summers and L. L. Hegedus, Ind. Eng. Chem. Prod. Res. Dev. 18, 318 (1979).
102. U.S. Pat. 6,003,303 (Dec. 21, 1999), J. D. Peter-Hoblyn, J. M. Valentine, B. N.
Sprague, and W. R. Epperly (to Clean Diesel Technologies, Inc.).
103. U.S. Pat. 6,013,599 (Jan. 11, 2000), I. Manson (to Redem Corp.).
104. U.S. Pat. 6,093,378 (July 25, 2000), M. Deeba, Y. K. Lui, and J. C. Dettling
(to Engelhard Corp.).
105. M. E. Dry, in J. Anderson and M. Boudart, eds., Catalysis—Science and Technology,
Springer-Verlag, New York, 1981, p. 159.
106. G. W. Huber, C. G. Guymon, B. C. Stephenson, and C. H. Bartholomew, Catalyst
Deactivation 2001 (Studies in Surface Science and Catalysis, Vol. 139), Elsevier,
Amsterdam, 2001, p. 423.
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 615

107. G. Busca, L. Lietti, G. Ramis, and F. Berti, Appl. Catal., B: Environ. 18, 1–36 (1998).
108. T. P. Kobylinski, B. W. Taylor, and J. E. Yong, in Proc. SAE, Detroit, 1974.
109. M. Shelef and H. S. Gandhi, Platinum Met. Rev. 18, 1 (1974).
110. H. S. Gandhi, H. K. Stepien, and M. Shelef, Mat. Res. Bull. 10, 837 (1975).
111. C. H. Bartholomew, Ind. Eng. Chem. Prod. Res. Dev. 14, 29 (1975).
112. R. W. Clark, J. K. Tien, and P. Wynblatt, J. Catal. 61, 15 (1980).
113. W. M. Shen, J. A. Dumesic, and C. G. Hill, J. Catal. 68, 152 (1981).
114. R. B. Pannell, K. S. Chung, and C. H. Bartholomew, J. Catal. 46, 340 (1977).
115. G. Lohrengel and M. Baerns, Appl. Catal. 1, 3 (1981).
116. I. Qamar and J. G. Goodwin, in Proc. 8th Am. Meeting Catal. Soc., Philadelphia,
1983 (Paper C-22).
117. J. G. Goodwin, D. O. Goa, S. Erdal, and F. H. Rogan, Appl. Catal. 24, 199 (1986).
118. O. Watzenberger, T. Haeberle, D. T. Lynch, and G. Emig, in Ref. 11, p. 441.
119. M. Agnelli, M. Kolb, and C. Mirodatos, J. Catal. 148, 9 (1994).
120. H. C. Lee and R. J. Farrauto, Ind. Eng. Chem. Res. 18, 1 (1989).
121. R. J. Farrauto and H. C. Lee, Ind. Eng. Chem. Res. 29, 1125 (1990).
122. F. Sperner and W. Hohmann, Platinum Met. Rev. 20, 12 (1976).
123. J. M. Hess and J. Phillips, J. Catal. 136, 149 (1992).
124. C. H. Bartholomew, Catalysis Lett. 7 (1990).
125. N. L. Wu and J. Phillips, J. Phys. Chem. 89, 591 (1985).
126. N. L. Wu and J. Phillips, Appl. Phys. 59, 769 (1986).
127. N. L. Wu and J. Phillips, J. Catal. 113, 129 (1988).
128. A. Bielanski and M. Najbar, in Ref. 9, p. 127.
129. N. Burriesci, F. Garbassi, M. Petrera, and G. Petrini, in Ref. 9, p. 115.
130. Y. L. Xiong, R. Castillo, C. Papadopoulou, L. Dada, J. Ladriere, P. Ruiz, and
B. Delmon, in Ref. 11, p. 425.
131. R. J. Farrauto, M. Hobson, T. Kennelly, and E. Waterman, Appl. Catal. 81, 227
(1992).
132. P. L. Gai-Boyes, Catal. Rev. Sci. Eng. 34, 1 (1992).
133. B. Delmon, in Ref. 12, p. 113.
134. K. M. Erickson, D. A. Karydis, S. Boghosian, and R. Fehrmann, J. Catal. 155, 32
(1995).
135. B. Delmon, in Ref. 13, p. 39.
136. N. B. Jackson, A. K. Datye, L. Mansker, R. J. O’Brien, and B. H. Davis, in Ref. 13,
p. 501.
137. S. A. Eliason and C. H. Bartholomew, in Ref. 13, p. 517.
138. A. Baranski, R. Dziembaj, A. Kotarba, A. Golebiowski, Z. Janecki, and J. B. C.
Pettersson, in Ref. 14, p. 229.
139. C. A. Querini, F. Ravelli, M. Ulla, L. Cornaglia, and E. Miro, in Ref. 14, p. 257.
140. H. N. Pham, J. Reardon, and A. K. Datye, Powder Technol. 103, 95 (1999).
141. D. S. Kalakkad, M. D. Shroff, S. Kohler, N. Jackson, and A. K. Datye, Appl. Catal.
133, 335 (1995).
142. W. D. Callister, Materials Science and Engineering: An Introduction, John Wiley &
Sons, Inc., New York, 2000.
143. R. L. Coble and W. D. Kingery, J. Am. Ceram. Soc. 39, 381 (1956).
144. S. G. Deng and Y. S. Lin, AIChE J. 43, 505 (1997).
145. S. G. Thoma, M. Ciftcioglu, and D. M. Smith, Powder Technol. 68, 53 (1991).
146. M. Bankmann, R. Brand, B. H. Engler, and J. Ohmer, Catal. Today 14, 225
(1992).
147. V. M. Kenkre and M. R. Endicott, J. Am. Ceram. Soc. 79, 3045 (1996).
148. H. Song and J. R. G. Evans, J. Am. Ceram. Soc. 77, 806 (1994).
149. J. Werther and W. Xi, Powder Technol. 76, 39 (1993).
616 CATALYST DEACTIVATION AND REGENERATION Vol. 1

150. B. L. Bhatt, E. S. Schaub, E. C. Hedorn, D. M. Herron, D. W. Studer, and D. M.


Brown, in G. J. Stiegel and R. D. Srivastava, eds. Proc. of Liquefaction Contractors
Review Conference, U.S. Department of Energy, Pittsburgh, Pa., 1992, p. 403.
151. H. N. Pham and A. K. Datye, Catal. Today 58, 233 (2000).
152. C. R. Bemrose and J. Bridgewater, Powder Technol. 49, 97 (1987).
153. M. Ghadiri, J. A. S. Cleaver, V. G. Tuponogov, and J. Werther, Powder Technol. 80,
175 (1994).
154. S. A. Weeks, P. Dumbill, Oil Gas J. 88, 38 (Apr. 16, 1990).
155. R. Zhao, J. G. Goodwin, K. Jothimurugesan, J. J. Spivey, and S. K. Gangwal, Ind.
Eng. Chem. Res. 39, 1155 (2000).
156. P. K. Doolin, D. M. Gainer, and J. F. Hoffman, J. Testing Evaluation 21, 481 (1993).
157. R. Oukaci, A. H. Singleton, D. Wei, and J. G. Goodwin, Preprints, 217th National
Meeting, ACS Division of Petroleum Chemistry, Anaheim, Calif., 1999, p. 91.
158. M. J. Adams, M. A. Mullier, and J. P. K. Seville, Powder Technol. 78, 5 (1994).
159. G. Emig and F. G. Martin, Ind. Eng. Chem. Res. 30, 1110 (1991).
160. P. W. N. M. van Leeuwen, Appl. Catal., A: Gen. 212, 61 (2001).
161. P. E. Garrou, Chem. Rev. 85, 171 (1985).
162. B. C. Gates, Catalytic Chemistry, John Wiley & Sons, Inc., New York, 1992.
163. J. L. Latham and A. E. Burgess, Elementary Reaction Kinetics, 3rd ed.,
Butterworths, London, 1977.
164. W. Hartmeier (translated by J. Wiesner), Immobilized Biocatalysts, Springer-Ver-
lag, Berlin, 1988; J. A. Hurlbut and co-workers, J. Chem. Educ. 50, 149 (1973).
165. O. R. Zaborsky, in J. Burton and L. Garten, eds., Advanced Materials in Catalysis,
Academic press, New York, 1977.
166. S. L. Neidleman, Catalysis of Organic Reactions, Marcel Dekker, New York, 1984.
167. S. Kindel, Technology 1, 62 (1981).
168. M. N. Gupta, Biotechnol. Appl. Biochem. 14, 1 (1991).
169. A. W. H. Fersht, Enzyme Structure and Mechanism, W. H. Freeman & Co.,
New York, 1984.
170. J. R. Whitaker, Principles of Enzymology for the Food Sciences, 2nd ed., Marcel
Dekker, New York, 1994.
171. M. Matsumoto, K. Kida, and K. Kondo, J. Chem. Technol. Biotechnol. 70, 188 (1997).
172. A. M. Azevedo, D. M. F. Prazeres, J. M. S. Cabral, and L. P. Fonseca, J. Mol. Catal.,
B: Enzymatic 15, 147 (2001).
173. A. M. Klibanov, Tibtech 15, 97 (1997).
174. D. Li, Z. Guo, and H. Liu, J. Am. Chem. Soc. 188, 275 (1996).
175. S. Colombie, A. Gaunand, and B. Lindet, J. Mol. Catal., B: Enzymatic 11, 559 (2001).
176. Z. Zhang, Z. He, and M. He, J. Mol. Catal., B: Enzymatic 14, 85 (2001).
177. G. Toscano, D. Pirozzi, M. Maremonti, and G. Greco, Jr., Biotechnol. Bioeng. 44, 682
(1994).
178. M. Mohanty, R. S. Ghadge, N. S. Patil, S. B. Sawant, J. B. Joshi, and A. V.
Deshpande, Chem. Eng. Sci. 56, 3401 (2001).
179. A. Ginsberg and W. R. Carroll, Biochemistry 4, 2159 (1965).
180. T. J. Ahern and A. M. Klibanov, Meth. Biochem. Anal. 33, 91 (1988).
181. K. A. Dill, D. O. V. Alonso, and K. Hutchinson, Biochemistry 28, 5439 (1989).
182. T. J. Hancock and J. T. Hsu, Biotechnol. Progr. 12, 494 (1996).
183. M. Longo and D. Combes, J. Chem. Technol. Biotechnol. 74, 25 (1999).
184. L. M. M. Tijskens, R. Greiner, E. S. A. Biekman, and U. Konietzny, Biotechnol.
Bioeng. 72, 323 (2001).
185. V. V. Mozhaev and K. Martinek, Enzyme Microb. Technol. 6, 50 (1984).
186. V. V. Mozhaev, N. S. Melik-Nubarov, V. Siksnis, and K. Martinek, Biocatalysis 3,
189 (1990).
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 617

187. M. G. Roig and J. F. Kennedy, Crit. Rev. Biotechnol. 12, 391 (1992).
188. A. M. Klibanov, Anal. Biochem. 93, 1 (1979).
189. P. Monsan and D. Combes, Methods Enzymol. 137, 584 (1988).
190. J. M. Guisan, P. Sabuquillo, R. Fernandez-Lafuent, G. Fernandez-Lorente, C. Ma-
teo, P. J. Halling, D. Kennedy, E. Miyata, and D. Re., J. Mol. Catal., B: Enzymatic
11, 817 (2001).
191. P. Wang, S. Dai, S. D. Waezsada, A. Y. Tsao, and B. H. Davison, Biotechnol. Bioeng.
74, 249 (2001).
192. R. G. Silver, J. C. Summers, and W. B. Williamson, Catalysis and Automotive Pollu-
tion Control II, Elsevier, Amsterdam, 1991, p. 167.
193. G. B. Fisher, M. G. Zammit, and J. LaBarge, SAE Report 920846, 1992.
194. R. J. Farrauto and R. M. Heck, Catal. Today 51, 351 (1999).
195. G. W. Huber, C. G. Guymon, B. C. Stephenson, and C. H. Bartholomew, Catalyst
Deactivation 2001 (Studies in Surface Science and Catalysis, Vol. 139), Elsevier,
Amsterdam, 2001, p. 423.
196. U.S. Pat. 6,169,120 (Jan. 2, 2001), G. L. Beer (to Syntroleum Corp.).
197. C. H. Bartholomew, M. W. Stoker, L. Mansker, and A. Datye, in Ref. 14, p. 265.
198. U.S. Pat. 5,728,894 (Mar. 17, 1998), O. Nagano and T. Watanabe (to Ashahi Kasei
Kogyo Kabushiki Kaisha).
199. T. Maillet, J. Barbier, and D. Duprez, Appl. Catal., B 9, 251 (1996).
200. WIPO Pat. 93/16020A3 (Sept. 16, 1993), G. Mathys, L. Martens, M. Baes,
J. Verduijn, D. Huybrechts, and C. Renata (to Exxon Chem.).
201. U.S. Pat. 5,672,800 (Sept. 30, 1997), G. Mathys, L. Martens, M. Baes, J. Verduijn,
and D. Huybrechts (to Exxon Chem.).
202. U.S. Pat. 6,080,903 (June 26, 2000), L. Stine, B. Muldoon, S. Gimre, and R. Frame (to
UOP).
203. B. Subramaniam, V. Arunajatesan, and C. J. Lyon, in Ref. 14, p. 63.
204. WIPO Pat. 99/33769 (July 8, 1999), D. Ginosar, R. Fox, and P. Kong (to Lockheed
Martin).
205. F. H. Ribeiro, A. L. Bonivardi, and C. Kim, J. Catal. 150, 186 (1994).
206. D. Ginosar and B. Subramaniam, in Ref. 12, p. 327.
207. E. E. Petersen, in Ref. 13, p. 87.
208. J. W. Gosselink and J. A. R. V. Veen, in Ref. 14, p. 3.
209. L. Lin, T. Zao, J. Zang, and Z. Xu, Appl. Catal. 67, 11 (1990).
210. D. E. Resasco and G. L. Haller, Catalysis (Spec. Period. Rept.) 11, 379 (1994).
211. R. D. Cortright and J. A. Dumesic, J. Catal. 148, 771 (1994).
212. H. Weyten, K. Keizer, A. Kinoo, J. Luyten, and R. Leysen, AIChE J. 43, 1819
(1997).
213. P. Praserthdam, T. Mongkhonsi, S. Kunatippapong, B. Jaikaew, and N. Lim, in
Ref. 13, p. 153.
214. WIPO Pat. 00/69993 (May 12, 2000), B. Rose and T. Kiliany (to Mobil Oil Corp.).
215. A. Guerrero-Ruiz, A. Sepulveda-Escribano, and I. Rodriguez-Ramos, Catal. Today
21, 545 (1994).
216. D. Qin and J. Lapszewicz, Catal. Today 21, 551 (1994).
217. S. Stagg and D. Resasco, in Ref. 13, p. 543.
218. K. Fujimoto, K. Tomishige, O. Yamazaki, Y. Chen, and X. Li, Res. Chem. Intermed.
24, 259 (1998).
219. U.S. Pat. 5,191,142 (Mar. 2, 1993), C. Marshall and J. Miller (to Amoco Corp.).
220. A. G. Gayubo, A. T. Aguayo, A. E. S. D. Campo, P. L. Benito, and J. Bilbao, in Ref. 14,
p. 129.
221. U.S. Pat. 5,248,647 (Sept. 28, 1993), P. T. Barger (to UOP).
222. WIPO Pat. 99/42202 (Feb. 20, 1998), J. Cox and J. Evans (to Johnson Matthey).
618 CATALYST DEACTIVATION AND REGENERATION Vol. 1

223. WIPO Pat. 98/50487 (May 2, 1997), S. Leviness, C. Mart, W. Behrmann, S. Hsia, and
D. Neskora (to Exxon Research and Engineering Co.).
224. C. H. Bartholomew, in M. Oballa and S. Shih, eds., Catalytic Hydroprocessing of
Petroleum and Distillates, Marcel Dekker, New York, 1993, p. 1.
225. J. Summers and W. B. Williamson, in J. Armor, ed., Environmental Catalysis 1993,
Vol. 552, American Chemical Society, Washington, D.C., 1993, p. 94.
226. J. Dettling, Z. Hu, Y. K. Lui, R. Smaling, C. Z. Wan, and A. Punke, in Studies in
Surface Science and Catalysis, Vol. 96, Elsevier, Amsterdam, 1995, p. 461.
227. U.S. Pat. 4,910,180 (Mar. 20, 1990), M. Berndt and D. Ksinsik (to Doduco).
228. U.S. Pat. 4,985,387 (Jan. 15, 1991), M. Prigent, G. Blanchard, and P. Phillippe (to
ProCatalyse).
229. U.S. Pat. 5,041,407 (Aug. 20, 1991), W. Williamson, D. Linden, and J. Summers (to
Allied-Signal Inc.).
230. U.S. Pat. 5,116,800 (May 26, 1992), W. Williamson, D. Linden, and J. Summers (to
Allied-Signal Inc.).
231. U.S. Pat. 5,234,881 (Aug. 10, 1993), C. Narula, W. Watkins, and M. Chattha (to Ford
Motor Co.).
232. U.S. Pat. 5,254,519 (Oct. 19, 1993), C. Wan, S. Tauster, and H. Rabinowitz (to
Engelhard Corp.).
233. T. Furuya, S. Yamanaka, T. Hayata, J. Koezuka, T. Yoshine, and A. Ohkoshi, in
Proc. Gas Turbine Conference and Exhibition, Anaheim, Calif., 1987.
234. T. Kawakami, T. Furuya, Y. Sasaki, T. Yoshine, Y. Furuse, and M. Hoshino, in Proc.
Gas Turbine and Aeroengine Congress and Exposition, Toronto, Ont., June 4–8,
1989.
235. U.S. Pat. 5,250,489 (1993), R. D. Betta, F. Ribeiro, T. Shoji, K. Tsurumi, N. Ezawa,
and S. Nickolas (to Catalytica, Inc.).
236. T. Fujii, Y. Ozawa, and S. Kikumoto, J. Eng. Gas Turbines Power 120, 509 (1998).
237. D. O. Borio and N. S. Schbib, Comput. Chem. Eng. 19, S345 (1995).
238. U.S. Pat. 5,028,634 (July 2, 1991), R. Fiato (to Exxon Research and Engineering
Co.).
239. R. Zhao and J. G. Goodwin, Jr., K. Jothimurugesan, S. K. Gangwal, and J. J. Spivey,
Ind. Eng. Chem. Res. 40, 1065–1075 (2001).
240. R. Zhao, J. G. Goodwin, Jr., K. Jothimurugesan, S. K. Gangwal, and J. J. Spivey,
Ind. Eng. Chem. Res. 40, 1076 (2001).
241. U.S. Pat. 5,939,350 (Aug. 17, 1999), A. H. Singleton, R. Oukaci, and J. G. Goodwin
(to Energy International Corp.).
242. U.S. Pat. 6,087,405 (July 11, 2000), S. Plecha, C. H. Mauldin, and L. E. Pedrick (to
Exxon Research and Engineering Co.).
243. U.S. Pat. 6,124,367 (Sept. 26, 2000), S. Plecha, C. H. Mauldin, and L. E. Pedrick (to
Exxon Research and Engineering Co.).
244. W.O. Pat. 00/71253 (Nov. 30, 2000), A. H. Singleton, R. Oukaci, and J. G. Goodwin
(to Energy International Corp.).
245. M. Seitz, E. Klemm and G. Emig, in Ref. 14, p. 211.
246. S. Masamune and J. M. Smith, AIChE J. 12, 384 (1966).
247. Y. Murakami, T. Kobayashi, T. Hattori, and M. Masuda, Ind. Eng. Chem. Fundam.
7, 599–605 (1968).
248. W. H. J. Stork, in G. F. Froment, B. Delmon, and P. Grange, eds., Hydrotreatment
and Hydrocracking of Oil Fractions, (Studies in Surface Science and Catalysis,
Vol. 106), Elsevier, New York, 1997, pp. 41–67.
249. G. D. Parks, A. M. Schaffer, M. J. Dreiling, and C. B. Shiblom, Prepr.—Am. Chem.
Soc., Div. Petr. Chem. 25, 335 (1980).
Vol. 1 CATALYST DEACTIVATION AND REGENERATION 619

250. D. L. Trimm, in J. L. Figueiredo, ed., Progress in Catalyst Deactivation (NATO


Advanced Study Institute Series E, No. 54), M. Nijhoff Publishers, Boston, 1982,
pp. 3–18.
251. E. R. Becker and J. J. Wei, J. Catal. 46, 365–381 (1977).
252. U.S. Pat. 5,498,638 (Mar. 12, 1996), D. C. Long (to Exxon Research and Engineering
Co.).
253. C. H. Bartholomew, Appl. Catal., A: Gen. 107, 1–57 (1993).
254. B. R. Powell, Presented at the Materials Research Society Annual Meeting, Boston,
Nov. 16–21, 1980 (Paper H9).
255. R. Heck and R. Farrauto, Catalytic Air Pollution Control: Commercial Technology,
Van Nostrand Reinhold, New York, 1995.
256. D. L. Trimm, Appl. Catal., A: Gen. 212, 153 (2001).
257. G. Berrebi, P. Dufresne, and Y. Jacquier, Environ. Prog. 12, 97 (1993).
258. R. L. D’Aquino, Chem. Eng. 107, 32 (2000).
259. T. Chang, Oil Gas J. 96(41), 49 (1998).
260. S. R. Blashka and W. Duhon, Int. J. Hydrocarbon Eng. 4(1), 60 (1998).
261. D. C. McCulloch, in B. E. Leach, ed., Applied Industrial Catalysis, Academic Press,
New York, 1983, pp. 103–110.
262. J. P. Franck and G. Martino, in J. L. Figueiredo, ed., Progress in Catalyst Deactiva-
tion (NATO Advanced Study Institute Series E., No. 54), M. Nijhoff Publishers,
Boston, 1982, pp. 355–398.
263. J. J. Spivey, G. W. Roberts, and B. H. Davis, eds., Catalyst Deactivation 2001
(Studies in Surface Science and Catalysis, Vol. 139), Elsevier, Amsterdam, 2001.
264. U.S. Pat. 5,043,517 (Aug. 27, 1991), J. H. Haddad, N. Mohsen, and H. Owen (to Mo-
bil Oil Corp.).
265. U.S. Pat. 5,306,682 (Apr. 26, 1994), M. Ueda, T. Murakami, S. Shibata, K. Hirabaya-
shi, T. Kondoh, K. Adachi, N. Hoshino, and S. Inoue (to Research Association for the
Utilization of Light Oil JP).
266. U.S. Pat. 5,675,048 (Oct. 7, 1997), S. Y.-F. Zhang, C. D. Gosling, P. A. Sechrist, and
G. A. Funk (to UOP).
267. G. Panattoni and C. A. Querini, in Ref. 263, p. 181.
268. U.S. Pat. 5,854,162 (Dec. 29, 1998), P. Dufresne and N. Brahma (to Eurecat).
269. U.S. Pat. 5,883,031 (Mar. 16, 1999), R. A. Innes, D. L. Holtermann, and B. F.
Mulaskey (to Chevron).
270. S. C. Fung, Chemtech 24, 40 (1994).
271. J. C. Alfonso, D. A. G. Aranda, M. Schmal, and R. Frety, Fuel Proc. Technol. 50, 35
(1997).
272. G. J. Arteaga, J. A. Anderson, and C. H. Rochester, J. Catal. 187, 219 (1999).
273. C. L. Pieck, C. R. Vera, and J. M. Parera, in Ref. 263, p. 279.
274. D. R. Acharya, R. Hughes, M. A. Kennard, and Y. P. Liu, Chem. Eng. Sci. 47, 1687
(1992).
275. U.S. Pat. 4,999,326 (Mar. 12, 1991), D. L. Sikkenga, I. C. Zaenger, and G. S. Wil-
liams (to Amoco).
276. A. Ekstrom and J. Lapszewicz, J. Phys. Chem. 88, 4577 (1984); J. Phys. Chem. 91,
4514 (1987).
277. U.S. Pat. 5,268,344 (Dec. 7, 1993), L. E. Pedrick, C. H. Mauldin, and W. C. Behr-
mann (to Exxon Research and Engineering Co.).
278. U.S. Pat. 5,338,439 (Aug. 16, 1994), H. Owen and P. H. Schipper (to Mobil Oil Corp.).
279. U.S. Pat. 5,198,397 (Mar. 30, 1993), M. F. Raterman (to Mobil Oil Corp.).
280. U.S. Pat. 5,393,717 (Feb. 28, 1995), M. R. Apelian, A. S. Fung, G. H. Hatzikos, C. R.
Kennedy, C.-H. Lee, T. R. Kiliany, P. K. Ng, and D. A. Pappal (to Mobil Oil Corp.).
281. U.S. Pat. 5,340,957 (Aug. 23, 1994), D. E. Clark (to Union Oil Co.).
620 CATALYST DEACTIVATION AND REGENERATION Vol. 1

282. Y. Yoshimura, T. Sato, H. Shimada, N. Matsubayashi, M. Imamura, A. Nishijima,


S. Yoshitomi, T. Kameoka, and H. Yanase, Energy Fuels 8, 435 (1994).
283. E. S. Oh, Y. C. Park, and I. C. Lee, J. Catal. 172, 314 (1997).
284. A. T. Aquavo, A. G. Gayubo, A. Atutxa, M. Olazar, and J. Bilbao, J. Chem. Tech.
Biotech. 74, 1082 (1999).
285. U.S. Pat. 4,777,156 (Oct. 11, 1988), N. P. Forbus, M. May-Som Wu (to Mobil Oil
Corp.).
286. A. Krishna, C. Hsieh, A. E. English, T. Pecoraro, and C. Kuehler, Hydrocarbon
Process. 59–66 (Nov. 1991).
287. C. Altomare, G. Koermer, P. Schubert, S. Suib, and W. Willis, Chem. Mater. 1, 459–
463 (1989).
288. A. Aguinaga and M. Montes, Appl. Catal., A: Gen. 90, 131 (1992).
289. G. V. Lowry and M. Reinhard, Environ. Sci. Technol. 34, 3217 (2000).
290. U.S. Pat. 4,830,997 (May 16, 1989), D. C. Trinh and A. Desvard (to Institut Français
du Petrole).
291. U.S. Pat. 5,230,791 (July 27, 1993), D. E. Sherwood (to Texaco Inc.).
292. S. C. Fung, in Ref. 263, p. 399.
293. U.S. Pat. 5,672,801 (Sept. 30, 1997), B. Didillon (to Institut Français Du Petrole).
294. U.S. Pat. 4,929,576 (May 29k 1990), Y.-Y. P. Tsao and R. von Ballmoos (to Mobil Oil
Corp.).
295. P. Dufresne, N. Brahma, and F. Girardier, Revue de I’Institut Français due Petrole
50, 283 (1995).
296. U.S. Pat. 5,275,990 (Jan. 4, 1994), F. T. Clark and A. L. Hensley, Jr. (to Amoco
Corp.).
297. A. Brito, R. Arvelo, and A. R. Gonzalez, Ind. Eng. Chem. Res. 37, 374 (1998).
298. V. L. S. Teixeira-da-Silva, F. P. Lima, and L. C. Dieguez, Ind. Eng. Chem. Res. 37,
882 (1998).
299. C. E. Snape, M. C. Diaz, Y. R. Tyagi, S. C. Martin, and R. Hughes, in Ref. 263, p. 359.
300. U.S. Pat. 5,445,728 (Aug. 29, 1995), D. E. Sherwood, Jr., and J. R. Hardee, Jr. (to
Texaco Inc.).
301. U.S. Pat. 5,021,377 (June 4, 1991), M. K. Maholland, C.-M. Fu, R. E. Lowery, D. H.
Kubicek, and B. J. Bertus (to Phillips Petroleum Co.).
302. U.S. Pat. 5,141,904 (Aug. 25, 1992), D. H. Kubicek, C.-M. Fu, R. E. Lowery, and M.
K. Maholland (to Phillips Petroleum Co.).
303. U.S. Pat. 5,151,391 (Sept. 29, 1992), C.-M. Fu, M. Maholland, and R. E. Lowery (to
Phillips Petroleum Co.).
304. U.S. Pat. 6,063,721 (May 16, 2000), Y. Hu, B. Luo, K. Sun, Q. Yang, M. Gong, J. Hu,
G. Fang, and Y. Li (to China Petro-Chemical Corp.).
305. P. L. Walker, Jr., F. Rusinko, Jr., and L. G. Austin, Adv. Catal. 11, 133–221 (1959).
306. J. W. Fulton, Chem. Eng. 96, 111–114 (1988).
307. J. T. Richardson, Ind. Eng. Chem. Process Des. Dev. 11, 8 (1972).
308. P. B. Weisz and R. B. Goodwin, J. Catal. 6, 227 (1966).
309. J. R. Rostrup-Nielsen, J. Catal. 21, 171–178 (1971).
310. D. B. Dadyburjor, in B. Delmon and G. F. Froment, eds., Catalyst Deactivation,
Elsevier, Amsterdam, 1980, pp. 341–351.
311. C. Bartholomew in I. I. Harvath, eds., Encyclopedia of Catalysis, Vol. 2, John Wiley
& Sons, New York, 2003, pp. 182–315.

CALVIN BARTHOLOMEW
Brigham Young University
Vol. 1 DISTILLATION 621

DISTILLATION
1. Introduction

Distillation is a method of separation that is based on the difference in com-


position between a liquid mixture and the vapor formed from it. This compo-
sition difference arises from the dissimilar effective vapor pressures, or
volatilities, of the components of the liquid mixture. When such dissimilarity
does not exist, as at an azeotropic point, separation by simple distillation is
not possible. Distillation as normally practiced involves condensation of the
vaporized material, usually in multiple vaporization/condensation operations,
and thus differs from evaporation, which is usually applied to separation of a
liquid from a solid but which can be applied to simple liquid concentration
operations.
Distillation is the most widely used industrial method of separating
liquid mixtures and is at the heart of the separation processes in many
chemical and petroleum plants (see SEPARATION PROCESS SYNTHESIS). The
most elementary form of the method is simple distillation, in which the
liquid is brought to boiling and the vapor formed is separated and con-
densed to form a product. If the process is continuous with respect to feed
and product flows, it is called flash distillation. If the feed mixture is
available as an isolated batch of material, the process is a form of batch
distillation and the compositions of the collected vapor and residual liquid
are thus time dependent. The term fractional distillation, which may be
contracted to fractionation, was originally applied to the collection of sepa-
rate fractions of condensed vapor, each fraction being segregated. In modern
practice the term is applied to distillation processes in general, where
an effort is made to separate an original mixture into several components
by means of distillation. When the vapors are enriched by contact with
counterflowing liquid reflux, the process is often called rectification. When
fractional distillation is accomplished with a continuous feed of material
and continuous removal of product fractions, the process is called conti-
nuous distillation. When steam is added to the vapors to reduce the partial
pressures of the components to be separated, the term steam distillation is
used.
Most distillations conducted commercially operate continuously, with a
more volatile fraction recovered as distillate and a less volatile fraction recovered
as bottoms or residue. If a portion of the distillate is condensed and returned to
the process to enrich the vapors, the liquid is called reflux. The apparatus in
which the enrichment occurs is usually a vertical, cylindrical vessel called a
still or distillation column. This apparatus normally contains internal devices
for effecting vapor–liquid contact; the devices may be categorized as plates or
packings.
Distillation has been practiced in one form or another for centuries. It was
of fundamental importance to the alchemists and was in use well before the time
of Christ. The historical development of distillation has been published (1) as has
the history of vapor–liquid contacting devices (2).

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.0409192006010918.a01.pub2
622 DISTILLATION Vol. 1

2. Vapor–Liquid Equilibria

The equilibrium distributions of mixture component compositions in the


vapor and liquid phases must be different if separation is to be made by distil-
lation. It is important, therefore, that these distributions be known. The com-
positions at thermodynamic equilibrium are termed vapor –liquid equilibria
(VLE) and may be correlated or predicted with the aid of thermodynamic
relationships. The driving force for any distillation is a favorable vapor– liquid
equilibrium, which provides the needed composition differences. Reliable VLE
are essential for distillation column design and for most other operations invol-
ving liquid– vapor phase contacting. Many VLE have been measured and
reported in the literature, and compilations of such data are available (3,4).
Also, bibliographic guides have been published, providing source references
for thousands of publications presenting VLE (5 –7). If data are not to be
found, they may be measured, or estimated by generalized methods (8 –10),
with some sacrifice in reliability. Even if carefully measured data are avail-
able, thermodynamic models are usually required to extrapolate or interpolate
the data for conditions not represented by the experiments. Whatever the
source and extent of the VLE, some evaluation should be made with regard
to accuracy.
The VLE for the system at hand may be simple and easily represented by an
equation or, in some systems, may be so complex that they cannot be adequately
measured or represented. Excellent treatises are available for selection and
implementation of vapor–liquid equilibrium studies (11–14). Typical VLE for
binary systems are shown graphically in Figure 1. Figure 1a is a representative
boiling-point diagram showing equilibrium compositions as functions of tempera-
ture at a constant pressure. The lower line is the liquid bubble point line, the
locus of points at which a liquid on heating forms the first bubble of vapor.

1.0
Equilibrium line
Vapor dew yLA
point line 0.8 Point A
TE
e
lin

xLE yLE
y
=
Temperature

°x

0.6
45

yL
xLB xLF yLD
0.4
TA

Liquid bubble xLA


xLA yLA
point line 0.2

0
0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
x L or y L xL

(a) (b)

Fig. 1. Isobaric VLE diagrams: (a) dew and bubble point; (b) vapor–liquid (y–x)
equilibrium.
Vol. 1 DISTILLATION 623

Vapor P

Liquid P

Fig. 2. Equilibrium between vapor and liquid. The conditions for equilibrium are
TV ¼ TL and PV ¼ PL . For a given T and P, phase fugacities are equal, ie, f V ¼ f L and
fi V ¼ fi L .

The upper line is the vapor dew-point line, representing points at which a vapor
on cooling forms the first drop of condensed liquid. The liquid and vapor compo-
sitions are conventionally plotted in terms of the low-boiling (more volatile) sub-
stance, L, in the mixture. The system point A has a vapor composition of yLA in
equilibrium with a liquid composition of xLA at a temperature of TA. Figure 1b is
a typical isobaric phase or y–x diagram. For further discussion, several textbooks
are available (15,16).
2.1. Thermodynamic Relationships. A closed container with vapor
and liquid phases at thermodynamic equilibrium may be depicted as in
Figure 2, where at least two mixture components are present in each phase.
The components distribute themselves between the phases according to their
relative volatilities. A distribution ratio for mixture component i may be defined
using mole fractions:

Ki ¼ yi  =xi ð1Þ

where the asterisk is used to denote an equilibrium condition. This K term,


known as the vapor–liquid equilibrium ratio, or often the K value, is widely
used, especially in the petroleum (qv) and petrochemical industries. For any
two mixture components i and j, their relative volatility, often called the alpha
value, is defined as

K i y i x j y i ð 1 xi Þ
aij ¼ ¼ ¼ ð2Þ
K j x i y j xi ð 1 yi Þ
624 DISTILLATION Vol. 1

Equation 2 may be rearranged to form an expression for the equilibrium curve in


Figure 1b.
a x
yi ¼  ij i  ð2aÞ
1 þ aij 1 xi

The relative volatility, a, is a direct measure of the ease of separation by distilla-


tion. If a ¼ 1, then component separation is impossible, because the liquid- and
vapor-phase compositions are identical. Separation by distillation becomes easier
as the value of the relative volatility becomes increasingly greater than unity.
Distillation separations having a values less than 1.2 are relatively difficult;
those that have values above 2 are relatively easy.
When both phases form ideal thermodynamic solutions, ie, no heat of mix-
ing, no volume change on mixing, etc, Raoult’s law applies:

pi V ¼ xi Pi 0 ð3Þ

where Pi0 is the vapor pressure of i at the equilibrium temperature. Combining


this expression with Dalton’s law of partial pressures, K values and relative vola-
tilities may be obtained:
Ki ¼ Pi 0 =P ð4Þ
0 0
aij ¼ Pi =Pj ð5Þ

Examples of ideal binary systems are benzene–toluene and ethylbenzene–


styrene; the molecules are similar and within the same chemical families.
Thermodynamics texts should be consulted before making the assumption that
a chosen binary or multicomponent system is ideal. When pressures are low
and temperatures are at ambient or above, but the solutions are not ideal, ie,
there are dissimilar molecules, corrections to equations 4 and 5 may be made:

Ki ¼ gi L Pi 0 =P ð6Þ

aij ¼ gi L Pi 0 gj L Pj 0 ð7Þ

where the Raoult’s law correction factor, gL, is a thermodynamically important


liquid-phase activity coefficient.
The development and thermodynamic significance of activity coefficients is
discussed in most chemical engineering thermodynamics texts. The liquid-phase
coefficients are strong functions of liquid composition and temperature and, to a
lesser degree, of pressure. A system with positive deviation, ie, the two compo-
nents having activity coefficients greater than one such that the logarithm of
the coefficient is positive, is shown in Figure 3a; a system with negative devia-
tion, the coefficients are less than unity and logarithms negative, is shown
in Figure 3b. In a few cases one component of a binary mixture has a positive
deviation and the other a negative deviation. Most commonly, however, both
coefficients have positive deviations.
Vol. 1 DISTILLATION 625

10
γ1∝

γ ∝2
Activity coefficient (log scale), γ

1.0

Activity coefficient (log scale), γ


γ 1 γ 2
γ 2

γ ∝2

γ ∝

1
γ 1

1.0 0.2
0 1.0 0 1.0
x1 x1

(a) (b)

Fig. 3. Binary activity coefficients for two component systems having (a) positive and
(b) negative deviations from Raoult’s law. Conditions are either constant pressure or
constant temperature and terminal coefficients, g1
i , are noted.

Terminal activity coefficients, g1i , are noted in Figure 3. These are often
called infinite dilution coefficients and for some systems are given in Table 1.
The hexane–heptane mixture is included as an example of an ideal system. As
the molecular species become more dissimilar, they are prone to repel each other,
tend toward liquid immiscibility, and have large positive activity coefficients, as
in the case of hexane–water.
If the molecular species in the liquid tend to form complexes, the system
will have negative deviations and activity coefficients less than unity, eg, the
system chloroform–ethyl acetate. In azeotropic and extractive distillation (see
DISTILLATION, AZEOTROPIC AND EXTRACTIVE) and in liquid–liquid extraction, nonideal
liquid behavior is used to enhance component separation. An extensive discus-
sion on the selection of nonideal addition agents is available (17).

Table 1. Terminal Activity Coefficients at Atmospheric


Pressurea
Component 1 Component 2 g1
1 g1
2

chloroform ethyl acetate 0.3 0.3


chloroform benzene 0.9 0.7
n-hexane n-heptane 1.0 1.0
ethyl acetate ethanol 2.5 2.5
ethanol toluene 6.0 6.0
benzene methanol 9.0 9.0
ethanol isooctane 11.0 8.0
methyl acetate water 20.0 7.0
ethyl acetate water 100.0 15.0
hexane water >100:0 >100:0
a
Values are approximate.
626 DISTILLATION Vol. 1

A great deal of study and research has gone into the development of work-
ing equations that can represent the curves of Figure 3. These equations are
based on solutions of the Gibbs-Duhem equation:
" # " #
a ln gi a ln gn
xi þ    xn ¼0 ð8Þ
axi axi
T;P T;P

One of the simplest and often used equations, or models, is that of Van Laar
(18). For a binary system of components 1 and 2, these equations are
A12
lng ¼ " #2 ð9Þ
A12 x1
1
A21 x2
A21
lng2 ¼ " #2 ð10Þ
A21 x2
1
A12 x1

It should be noted that only two parameters are involved. They are directly
related to the terminal activity coefficients:

ln g1
1 ¼ A12 ð11Þ
ln g1
2 ¼ A21 ð12Þ

A useful and quite popular model is given (19):


 
12 21
ln g1 ¼ lnðx1 þ 12 x2 Þ þ ð13Þ
x1 þ 12 x2 21 x1 þ x2
 
12 21
ln g2 ¼ lnðx2 þ 21 x1 Þ ð14Þ
x1 þ 12 x2 21 x1 þ x2

This, the Wilson model, is more complex than the Van Laar model, but it
does retain the two-parameter feature. The terminal activity coefficients are
related to the parameters:

ln g1
1 ¼ 1 ln 12 21 ð15Þ
ln g1
1 ¼ 1 ln 21 12 ð16Þ

Whereas the Wilson model has been found to represent a wide variety of nonideal
VLE, it cannot handle the case of partial immiscibility of the liquid phase; for
this purpose a three-parameter relationship, the nonrandom, two-liquid
(NRTL) model was developed (20).
The most recently developed model is called UNIQUAC (21). Comparisons
of measured VLE and predicted values from the Van Laar (22), Wilson, NRTL,
Vol. 1 DISTILLATION 627

and UNIQUAC models, as well as an older model, are available (3). Thousands
of comparisons have been made, and Reference 3, which covers the Dortmund
Data Base, available for purchase and use with standard computers, should be
consulted by anyone considering the measurement or prediction of VLE. The pre-
dictive VLE models can be accommodated to multicomponent systems through
the use of certain combining rules. These rules require the determination of
parameters for all possible binary pairs in the multicomponent mixture. It is
possible to use more than one model in determining binary pair data for a
given mixture (23).
To estimate VLE when no experimental data or model parameters are
available and the cost of special measurements cannot be justified, a group con-
tribution method based on the molecular structures involved called UNIFAC (8)
has been developed. Not all possible groups have been evaluated, but regular
progress reports are published. The UNIFAC method, as well as the other mod-
els, are critically important in extending limited data to conditions in distillation
columns that can cover wide ranges of temperatures, pressures, and composi-
tions. Handling of all these models by computer solution has been described in
some detail (24).
The vapor–liquid equilibria of dilute solutions are frequently expressed in
terms of Henry’s law:

p i V ¼ H i  xi ð17Þ

where, From equation 6, the Henry’s law coefficient is

Hi  ¼ gi L Pi 0 ð18Þ

Henry’s law is useful for handling equilibria associated with gas absorption (qv)
and stripping problems. Henry’s law coefficients are useful for estimating term-
inal activity coefficients and have been tabulated for many compounds in dilute
aqueous solutions (25,26).
The foregoing discussion has dealt with nonidealities in the liquid phase
under conditions where the vapor phase mixes ideally and where pressure–
temperature effects do not result in deviations from the ideal gas law. Such con-
ditions are by far the most common in commercial distillation practice. However,
it is appropriate here to set forth the completely rigorous thermodynamic expres-
sion for the K value:

Z !
p
1
gLi f0i p0i exp vLi dP
R0 T p0i
Ki ¼ ð19Þ
^ iP
f
For nonideal vapor-phase behavior, the fugacity coefficient for component i in the
mixture must be determined:
Z P 
^1 ¼ 1 v R0 T
ln f v i dP ð20Þ
R0 T 0 P
628 DISTILLATION Vol. 1

If the vapor forms an ideal solution,


vvi ¼ vv ¼ zR0 T=P ð21Þ
where z is the compressibility factor for the mixture. The right side of the nu-
merator in Equation 19 is called the Poynting correction, PC:
Z p !
1 L
PC ¼ exp v dP ð22Þ
R0 T p0i i

when the liquid is incompressible,


 L 
vi P P0i
PC ¼ exp ð23Þ
R0 T

At pressures less than 2 MPa (20 bar) and temperatures greater than 273 K,
PC  1:0. When the vapor obeys the ideal gas law, z ¼ 1:0; then for ideal vapor
solutions and for conditions such that PC ¼ 1:0, Equation 19 reduces to
Equation 6.
The fugacity coefficient departure from nonideality in the vapor phase can
be evaluated from equations of state or, for approximate work, from fugacity/
compressibility estimation charts. References 11,14, and 25 provide valuable
insights into this matter.
Journals for the publication of VLE data are available as are comprehensive
tabulations of azeotropic data (27,28); if the composition and temperature of the
azeotrope are known (at a given pressure), then such information may be used to
calculate activity coefficients. At the azeotropic point, by definition, yi ¼ xi ; from
Equation 6,

gi L ¼ P=Pi 0 ð24Þ

The vapor pressure Pi0 can be obtained from any of many sources such as hand-
books.
The measurement of VLE can be carried out in several ways. A common
procedure is to use a recycle still, which is designed to ensure equilibrium
between the phases. Samples are then taken and analyzed by suitable methods.
It is possible in some cases to extract equilibrium data from chromatographic
procedures. Discussions of experimental methods are available (5,11). For the
more challenging measurements, eg, conditions where one or more components
in the mixture can decompose or polymerize, commercial laboratories can be
used.
2.2. Azeotropic Systems. An azeotropic mixture is one that vaporizes
without any change in composition. Figures 4 and 5 represent homogeneous
azeotropic systems. Figure 4 depicts a minimum boiling azeotropic system
such as ethanol–water; Figure 5 describes a maximum boiling azeotropic system
such as acetone–chloroform. The point Z defines the azeotropic composition; this
azeotropic mixture is also called the constant boiling mixture (CBM). Positive
activity coefficients tend to produce minimum boiling azeotropes, and negative
coefficients tend to produce maximum boiling azeotropes.
Vol. 1 DISTILLATION 629

1.0

0.8
C, D

0.6
yL
Temperature

Z
0.4
C D
A, B
A B

0.2

0
0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
xL or yL xL

(a) (b)

Fig. 4. Boiling point (a) and phase diagram (b) for a minimum boiling binary azeotropic
system at constant pressure. A,B and C,D are representative equilibrium points; Z is the
azeotropic point.

Heterogeneous azeotropes are formed when the positive activity coefficients


are sufficiently large to produce two liquid phases that exist at the boiling point,
and a constant boiling mixture that is formed at some composition, generally
within the liquid immiscibility composition range. An example of a heteroge-
neous azeotropic system is the water/1-butanol system shown in Figure 6. Within

Z 1.0
C, D
D
C
A B 0.8
Temperature

0.6
yL

0.4

0.2
A, B

0
0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
xL or yL xL

(a) (b)

Fig. 5. Boiling point (a) and phase diagram (b) for a maximum boiling binary azeotropic
system at constant pressure. A,B and C,D are representative equilibrium points; Z is the
azeotropic point.
630 DISTILLATION Vol. 1

120

1.0

110 C, D
0.8 M
Temperature, °C

Mol fraction water in vapor


Z N
A B
0.6
y vs T

100 0.4 A, B

Z
D
0.2
C
x vs T M N

90 0
0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
Mol fraction water in liquid or vapor Mol fraction water in liquid

(a) (b)

Fig. 6. Boiling point (a) and phase diagram (b) for the heterogeneous azeotropic system,
water/1-butanol at atmospheric pressure. A,B and C,D are representative equilibrium
points; Z is the azeotropic point; M and N are liquid miscibility limits.

the immiscible range, M–N, the equilibrium vapor is the heterogeneous azeo-
trope, Z, of constant composition and the equilibrium temperature is constant.
At liquid compositions lower in water than in the azeotrope, the relative volati-
lity of water/1-butanol is greater than one; at liquid compositions higher in water
than in the azeotrope, the relative volatility of water/1-butanol is less than one
(see DISTILLATION, AZEOTROPIC AND EXTRACTIVE).

3. Distillation Processes

Basic distillation involves application of heat to a liquid mixture, vaporization of


part of the mixture, and removal of the heat from the vaporized portion. The
resultant condensed liquid, the distillate, is richer in the more volatile compo-
nents and the residual unvaporized bottoms are richer in the less volatile com-
ponents. Most commercial distillations involve some form of multiple staging in
order to obtain a greater enrichment than is possible by a single vaporization and
condensation.
For ease of presentation and understanding, the initial discussion of
distillation processes involves binary systems. Examining the binary boiling
point (Fig. 1a) and phase (Fig. 1b) diagrams, the enrichment from liquid com-
position xL to vapor composition yL represents a theoretical step, or equili-
brium stage.
3.1. Simple Distillations. Simple distillations utilize a single equili-
brium stage to obtain separation. Simple distillation, also called differential
Vol. 1 DISTILLATION 631

distillation, may be either batch or continuous, and may be represented on boil-


ing point or phase diagrams. In Figure 1a, if the batch distillation begins with a
liquid of composition xLA the initial distillate vapor composition is yLA. As the
distillate is removed, the remaining liquid becomes less rich in the low boiler,
L, and the boiling liquid composition moves to the left along the bubble point
line. If the distillation is continued until the liquid has a composition of xLE,
the last vapor distillate has a composition of yLE. Simple batch distillation is
not widely used in industry, except for the processing of high-valued chemicals
in small production quantities, or for distillations requiring regular sanitization.
Calculation methods are found in most standard distillation texts, and as parts of
commercial process simulation computer software packages.
Simple continuous distillation, also called flash distillation, has a continu-
ous feed to a single equilibrium stage; the liquid and vapor leaving the stage are
considered to be in phase equilibrium. On the boiling point diagram (Fig. 1a), the
feed is represented by xLF, the bottoms liquid by xLB, and the equilibrium vapor
distillate by yLD. The overall mass balance is

F ¼DþB ð25Þ
the component L balance is

x L F F ¼ yL D D þ xL B B ð26Þ

Flash distillations are widely used where a crude separation is adequate.


Examples of flash multicomponent calculations are given in standard distillation
texts (29).
3.2. Multiple Equilibrium Staging. The component separation in sim-
ple distillation is limited to the composition difference between liquid and vapor
in phase equilibrium. To overcome this limitation, multiple equilibrium staging
is used to increase the component separation. Figure 7 schematically represents
a continuous distillation that employs multiple equilibrium stages stacked one
upon another. The feed, F, enters the column at equilibrium stage f. The heat
qs required for vaporization is added at the base of the column in a reboiler or
calandria. The vapors VT from the top of the column flow to a condenser from
which heat qc is removed. The liquid condensate from the condenser is divided
into two streams: the first, a distillate D, which is the overhead product (some-
times called heads or make), is withdrawn from the system, and the second, a
reflux R, which is returned to the top of the column. A bottoms stream B is with-
drawn from the reboiler. The overall separation is represented by feed F separ-
ating into a distillate D and a bottoms B.
Above the feed a typical equilibrium stage is designated as n; the stage
above n is n þ 1 and the stage below n is n 1. The section of column above
the feed is called the rectification section and the section below the feed is
referred to as the stripping section.
The mass balance across stage n is (1) vapor (V n 1 ) from the stage below
(n 1) flows up to stage n; (2) liquid (Lnþ1 ) from the stage above (n þ 1) flows
down to stage n; (3) on stage n the vapors leaving V n are in equilibrium with the
liquid leaving Ln. The vapors moving up the column from equilibrium stage to
632 DISTILLATION Vol. 1


qc

Cond
enser
T
V
Condensate

Vn+1 Distillate D
n+1 Reflux R
Ln + 1 Balance loop
Vn
n
V n−1 L n

n−1
f
V
Feed F Ln−1 −
f
Lf
V m+1
m+1
Vm
Lm + 1
m
V m−1
Lm
m−1
Vs
Lm−1

q− s
Ls
Calandria

Bottoms B

Fig. 7. Distillation column with stacked multiple equilibrium stages. Terms are defined
in text.

equilibrium stage are increasingly enriched in the more volatile components.


Similarly, the liquid streams moving down the column are increasingly dimin-
ished in the more volatile components.
The overall column mass balances are

F ¼DþB ð27Þ
and for any component i,

Fxi F ¼ Dxi D þ Bxi B ð28Þ


The overall enthalpy balance is

HF F þ HS ¼ HD D þ HB B þ HC ð29Þ
A mass balance around plate n and the top of the column gives:

V n 1 ¼ Ln þ D ð30Þ
Vol. 1 DISTILLATION 633

And for any component:

V n 1 yi n 1 ¼ Ln xi n þ Dxi D ð31Þ
 n   
L D
yn 1
i ¼ x n
þ xD ð32Þ
V n 1 i V n 1 i

Below the feed, a similar balance around plate m and the bottom of the column
results in:
 m   
L B
ym 1
i ¼ x m
þ xB ð33Þ
V m 1 i V m 1 i
Equation 32 represents the upper (or rectifying) operating line equation, and
Equation 33 represents the lower (or stripping) operating line equation. The
slopes Ln =V n 1 and Lm =V m 1 can vary, depending on heat effects.
Graphical Method. The graphical McCabe–Thiele (30) design method
facilitates a visualization of distillation principles while providing a solution to
the material balance and equilibrium relationships. Here, the subscripts L and
H are not used, and x and y refer to the lower boiler, ie, more volatile component,
in the binary system. A McCabe–Thiele diagram is given in Figure 8 where P, Q,
and S are the xB, xF, and xD compositions on the y ¼ x, 458 construction line,
respectively. Line OP is the stripping operating line and line OS is the rectifying
operating line.

1.0
S

y n, x n
y n+1, x n+1
y n−1, x n
Mol fraction low boiler in vapor, y

y n, x n+1

O q line plate n
y m +1, x m +1
Q

y m, x m
plate m
y m, x m+1

y m−1, x m

P
0
xB xF x D 1.0
Mol fraction low boiler in liquid, x

Fig. 8. McCabe–Thiele diagram. Terms are defined in text.


634 DISTILLATION Vol. 1

The McCabe–Thiele method employs the simplifying assumption that the


molal overflows in the stripping and the rectification sections are constant.
This assumption reduces the rectifying and stripping operating line equations to:
" #  
n 1 L n D
y ¼ x þ xD ð34Þ
V R VR
" #  
m 1 L B
y ¼ xm þ xB ð35Þ
V S VS

The constant molal flows in each section are designated by L and V. The
McCabe–Thiele assumption of constant molal overflow implies that the molal
latent heats of the two components are identical, the sensible heat effects are
negligible, and the heat of mixing and the heat losses are zero. This simplified
situation is closely approximated for many distillations. Equation 34 now
represents the straight upper operating line OS, and Equation 35 represents
the straight lower operating line OP. The upper operating line has the slope
(L/V)R and the intercept at xD (¼ yD ) on the x ¼ y line. Note that this operating
line slope is less than one. Similarly, the lower operating line has a slope of
(L/V)S and the intercept is at xB on the y ¼ x line. This operating line has a
slope greater than one. The line QO from the feed intercept Q to the intersection
of the operating lines at O is called the q line.
The equilibrium curve gives the vapor–liquid relationships of yn and xn
above the feed and of ym and xm below the feed. The upper operating line gives
the relationship between yn 1 and xn and the lower operating line gives the rela-
tionship ym 1 and xm, ie, the streams passing each other. The graphical represen-
tation of theoretical equilibrium stages n and m is shown. The ym 1 , xm to ym,
xLþ1 represent the mass balance and phase equilibrium for theoretical stage m.
Similarly, yn 1 , xn to yn, xnþ1 represent theoretical stage n. The total number of
theoretical stages in the column can now be stepped off starting either at the
composition xB and stepping upward or starting at xD and stepping downward.
Condition of Feed (q Line). The q line, which marks the transition from
rectifying to stripping operating lines, is determined by mass and enthalpy bal-
ances around the feed plate. These balances are detailed in distillation texts (15).
The slope of the q line is q/(q 1) where:

heat needed to vaporize one mole of feed


q¼ ð36Þ
molal latent heat of feed
The q line, therefore, depends on the enthalpy condition of the feed. Types
of q lines are shown in Figure 9 and are listed below.

Feed enthalpy condition q Slope of q line q Line coordinates


cold liquid >1 þ Q–E
saturated liquid 1 1 Q–D
partially vaporized 0–1 Q–C
saturated vapor 0 0 Q–B
superheated vapor <0 þ Q–A
Vol. 1 DISTILLATION 635

e r
lin pe ine
riu
m Up ng l
ib i
uil era
t
Eq op e
E lin
y
D =
x
5 °,
C 4

B
Q
A

Lower
operating
lines
xF

Fig. 9. McCabe–Thiele q lines for various feed enthalpy conditions. Terms are defined in
text.

Reflux and Reflux Ratio. The liquid returned to the top of the column is
called reflux. The molar ratio R/D is the external reflux ratio. The ratio (L/V)R,
which is the slope of the rectifying operating line, is the rectifying internal reflux
ratio. Similarly, the ratio (L/V)S, which is the slope of the stripping operating
line, is the stripping internal reflux ratio. As the ratio R/D increases, the rectify-
ing internal reflux ratio increases and numerically approaches unity; similarly,
the stripping internal reflux ratio decreases and numerically approaches unity.
In the McCabe–Thiele plot the two operating lines move away from the equili-
brium line toward the y ¼ x diagonal as the reflux ratio increases, and the indi-
vidual theoretical stage steps become larger; accordingly, fewer theoretical
stages are required to make a given separation.
McCabe–Thiele Example. Assume a binary system L–H that has ideal
vapor–liquid equilibria and a relative volatility of 2.0. The feed is 100 mol of
xF ¼ 0:6; the required distillate is xD ¼ 0:95, and the bottoms xB ¼ 0:05, with
the compositions identified and the lighter component L. The feed is at the boil-
ing point. To calculate the minimum reflux ratio, the minimum number of theo-
retical stages, the operating reflux ratio, and the number of theoretical stages,
assume the operating reflux ratio is 1.5 times the minimum reflux ratio and
there is no subcooling of the reflux stream, then:

(1) Calculate the vapor composition in equilibrium with the liquid feed. From
Equation 2a and for x ¼ 0:60 mol fraction;

2ð0:6Þ
y ¼ ¼ 0:75 mol fraction ð37Þ
1 þ ð2:0 1Þ0:6
636 DISTILLATION Vol. 1

1.0 15
14
13
0.9 xD
12
11
0.8 10
9
8
0.7
7
Mol fraction L in vapor, y

0.6
6
xF
0.5
5

0.4
4

0.3
3

0.2
2

0.1 1

xB
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Mol fraction L in liquids, x

Fig. 10. McCabe–Thiele example. See text.

(2) Similarly, the entire equilibrium curve is calculated and is plotted in


Figure 10. The feed is at the boiling point so the q line is drawn vertically
with an infinite slope.
(3) Calculate mass balances on the basis of 100 mol of feed:
F ¼ D þ B ðoverall column balanceÞ
0:60F ¼ 0:95D þ 0:05B ¼ 60 ðcomponent L balanceÞ
D ¼ 61:11 mol distillate
B ¼ 38:89 mol bottoms

(4) Calculate reflux ratios. The minimum internal reflux ratio is a line
from the intercept of the q line with the equilibrium curve to the xD point
on the 458 line:

slope ¼ ðL=VÞR ¼ ð0:95 0:75Þ=ð0:95 0:60Þ


¼ 0:5714 minimum internal
reflux ratio
V ¼ L þ D ¼ 0:5714V þ 61:11
V ¼ 142:58 mol ðat minimum refluxÞ
L ¼ 81:47 mol ¼ R ðat minimum refluxÞ
ðR=DÞmin ¼ 81:47=61:11 ¼ 1:333 ðminimum reflux ratioÞ
Vol. 1 DISTILLATION 637

(5) Operating reflux ratio ¼ ðR=DÞoperating ¼ 1:5 1:333 ¼ 2:0.


(6) Reflux flow ¼ R ¼ 2:0ð61:11Þ ¼ 122:22 mol ¼ L (at operating reflux ratio).
(7) Rectifying section vapor flow ¼ V ¼ L þ D ¼ 122:22 þ 61:11 ¼ 183:33 mol.
(8) Upper operating line (eq. 34):
yn 1 ¼ ð122:22=183:33Þxn þ ð61:11=183:33Þ0:95 ¼ 0:667xn þ 0:317

(9) Stripping section liquid and vapor flows, because the feed is at the boiling
point,

LS ¼ LR þ F ¼ 122:22 þ 100 ¼ 222:22 mol


VS ¼ VR ¼ 183:33 mol

(10) Lower operating line (eq. 35):


ym 1 ¼ ð222:22=183:33Þxm þ ð38:89=183:33Þ ð0:05Þ ¼ 1:212xm 0:0106

(11) Theoretical stages: the complete construction is shown in Figure 10. Stages
were stepped off starting at the base. Approximately 14.2 theoretical stages
are required. This includes the reboiler, which normally functions as an
equilibrium stage. Therefore, a capability of 13.2 theoretical stages in the
column is needed. If the condenser were to condense only reflux, with the
distillate product leaving the process as a vapor, it could be counted also as
an equilibrium stage, making 12.2 stages needed for the column.

Unequal Molal Overflow. The McCabe–Thiele method is based on the


simplifying assumption that the molal overflow is constant in both the rectifying
and stripping sections. For many problems this assumption is not valid and more
precise calculations are necessary. For the more general case, detailed enthalpy
balances are made around individual stages or groups of stages. Standard distil-
lation texts discuss the internal enthalpy calculations by algebraic balances or by
graphical procedures; eg, Reference 15 details the stage-to-stage mass and
enthalpy balances with equilibrium calculations and also by means of the graphi-
cal Ponchon–Savarit procedure (31,32). Hand algebraic and graphical methods
requiring internal enthalpy calculations have been largely superseded by simu-
lations performed on modern computing devices, including personal computers
(see COMPUTER TECHNOLOGY).
Minimum Number of Theoretical Stages and Minimum Reflux Ratio.
There are infinite combinations of reflux ratios and numbers of theoretical
stages for any given distillation separation. The larger the reflux ratio, the
fewer the theoretical stages required. For any distillation system with its
given feed and its required distillate and bottoms compositions, there are two
constraints within which the variables of reflux ratio and number of theoretical
stages must lie: the minimum number of theoretical stages and the minimum
reflux ratio. The minimum reflux ratio occurs when the reflux ratio is reduced
so that the upper and lower operating lines and the q line are coincident at
a single point on the equilibrium line as shown in Figure 11a. When this con-
dition exists, an infinite number of theoretical stages would be required to make
638 DISTILLATION Vol. 1

xD
xD
Mol fraction L in vapor, y

Mol fraction L in vapor, y


xF

xB xB

Mol fraction L in liquids, x Mol fraction L in liquids, x

(a) (b)

Fig. 11. Limiting conditions in binary distillation. (a) Minimum reflux and infinite
number of theoretical stages; (b) total reflux and minimum number of theoretical
stages.

the separation. The minimum number of theoretical stages occurs when the
system is at total reflux: no feed, distillate, or bottoms. This is illustrated in
Figure 11b, where the operating lines are coincident with the 458, y ¼ x line.
For the McCabe– Thiele example presented above, the graphical proce-
dure would give slightly less than nine minimum theoretical stages including
the reboiler.
Simple analytical methods are available for determining minimum
stages and minimum reflux ratio. Although developed for binary mixtures,
they can often be applied to multicomponent mixtures if the two key com-
ponents are used. These are the components between which the specifica-
tion separation must be made; frequently the heavy key is the component
with a maximum allowable composition in the distillate and the light key
is the component with a maximum allowable specification in the bottoms.
On this basis, minimum stages may be calculated by means of the Fenske
relationship (33):

ln½ðyi =yj ÞDðxj =xi ÞB 


Nmin ¼ ð38Þ
ln aij;avg

where i and j are the light and heavy components of a binary mixture, or the light
key and heavy key in a multicomponent mixture. The average relative volatility
is often taken as the geometric average of the relative volatilities at the top and
bottom of the column. For the McCabe–Thiele example,

ln½ð0:95=0:05Þ ð0:95=0:05Þ
Nmin ¼ ¼ 8:50 stages
ln 2:0
Vol. 1 DISTILLATION 639

For minimum reflux ratio, the following equations (35) may be used:
X ai xif
¼1 q ð39Þ
i
ai f
X ai ðxid Þ
¼ Rmin þ 1 ð40Þ
i
ai f

where the value of q is determined as in the McCabe–Thiele procedure. Equa-


tion 39 is solved for root f, the value of which must lie between 1.0 and the
light key volatility. The root value so determined is then used in equation 40
to obtain the value of Rmin. Although a trivial example, the McCabe–Thiele
problem would yield

2:0ð0:6Þ 1:0ð0:4Þ
þ ¼ 1 q ¼ 0 ðbecause q ¼ 1Þ
2:0 f 1:0 f

solving f ¼ 1:25. Substituting in equation 40, for the given distillate composi-
tions,

2:0ð0:95Þ 1:0ð0:05Þ
þ ¼ Rmin þ 1
2:0 1:25 1:0 1:25

from which Rmin ¼ 1:333.


Both of these limits, the minimum number of stages and the minimum
reflux ratio, are impractical for useful operation, but they are valuable guidelines
within which the practical distillation must lie. As the reflux ratio decreases
toward the minimum reflux, the required number of stages increases rapidly.
Similarly, as the minimum number of stages is approached, the required reflux
ratio increases rapidly. A representative plot of the number of theoretical stages
vs reflux ratio for some distillation separation is shown in Figure 12. Both mini-
mum limits may be calculated for any distillation, thereby bracketing the prac-
tical design. Actual operating reflux ratios for most commercial columns are in
the range of 1.1 to 1.5 times the minimum reflux ratio.
The operating, fixed, and total costs of a distillation system are functions of
the relation of operating reflux ratio to minimum reflux ratio. Figure 13 shows
a typical plot of costs; as the operating to minimum reflux ratio increases, the
operating cost (principally energy cost for the boil-up) increases almost linearly.
Similarly, the fixed costs at first decrease from the infinite number of stages, pass
through a minimum, and then increase again as the diameter of column
increases with increased reflux ratio. These costs for typical distillations have
been calculated (36); the ratio of the economic optimum reflux to the minimum
reflux is often 1.2 or less.
Minimum Reflux with Pinch Zone. There are some distillations where the
minimum reflux does not occur at the intersection of the upper and lower oper-
ating lines and the q line. These cases arise when the equilibrium is skewed from
positive activity coefficients and when the operating line intersects the equili-
brium line in a zone of constant composition, a pinch zone, which is not at the
640 DISTILLATION Vol. 1

Theoretical stages

Minimum stages

Minimum reflux ratio

Reflux ratio

Fig. 12. Representative plot of theoretical stages vs reflux ratio for a given separation.
Each curve is the locus of points for a given separation. Note the limiting conditions of
minimum reflux and minimum stages.

q line intersection. Figure 14 illustrates such a case. An example of such a pinch


zone in an ethanol–water column is available (36).
Multicomponent Calculations. The calculations that determine the
reflux and stage requirements are more difficult to make for multicompo-
nent systems than for binary systems. When the concentration of a component

Total cost
Annual cost

Operating cost

Fixed cost

Economic optimum, R/D

(R/D)min
Reflux ratio, R/D

Fig. 13. Fixed, operating, and total costs of a typical distillation, as a function of reflux
ratio.
Vol. 1 DISTILLATION 641

1.0

xD
P
Mol fraction of L in vapor, y

Q
q line

xF

xB
0 1.0
Mol fraction of L in liquids, x

Fig. 14. False minimum reflux for system of skewed equilibria. Minimum reflux occurs
at intersection P of operating line and equilibrium line, not at intersection of q line and
equilibrium line. Terms are defined in text.

in the distillate and in the bottoms is specified for the overall solution of a
binary distillation, the component balance around the column also is comple-
tely specified. In the multicomponent case, only a single high-boiling key
component can be specified in the distillate and a single low-boiling key com-
ponent in the bottoms; the split of other components can be determined
only by detailed calculations. These require a series of trial and error com-
putations to obtain the solution at any given reflux ratio and number of
stages. As the number of components and number of stages become large,
the mathematical problem becomes formidable. Two approaches may be fol-
lowed: use of approximate, ie, shortcut, methods, or use of a suitable computer
program that provides rigorous solutions. The former are used when approxi-
mate solutions are adequate or when a computer is not available. For the
latter, numerous commercial programs are available and may be used with
personal computers.
Most shortcut methods involve: (1) calculating the minimum number of
stages; (2) calculating the minimum reflux ratio; and (3) estimating, from empiri-
cal correlations, the actual number of stages at an operating reflux. For mini-
mum stages the Fenske relationship (eq. 38) is used, whereas for minimum
reflux ratio the Underwood relationships (eqs. 39 and 40) are used. The relation-
ship of operating to minimum reflux ratio and of operating to minimum number
of plates is then estimated from the Gilliland correlation (37), or from a more
recent correlation such as that of Reference 38.
642 DISTILLATION Vol. 1

The Gilliland correlation is in graphical form and the curve has been fitted
by several workers, an example of which is (39):
" #0:5668
Nt Nmin R Rmin
¼ 0:75 0:75 ð41Þ
Nt þ 1 Rþ1

For the McCabe–Thiele example, and using eq. 41


" #0:5668
Nt 8:50 2:0 1:333
¼ 0:75 0:75
Nt þ 1 2:0 þ 1

from which Nt ¼ 15:7 stages. The original plot of Gilliland would give Nt ¼ 14:8,
closer to the McCabe–Thiele value of 14.2 stages.
Discussions of shortcut methods have appeared many times in the litera-
ture (16,35), accompanied by the usual admonition to use such methods only
for approximate designs or analyses. For multicomponent systems having signif-
icant nonidealities, the shortcut methods can be grossly in error.
Rigorous computer solutions are used for complex distillations involving
multiple stages, multiple components, nonideal phase equilibria, multiple feeds
and drawoffs, and heat addition or removal at intermediate stages. Most
calculations are made by computer and the algorithms are generally based on
the Thiele–Geddes model (40), which rates a given number of stages and reflux
ratio for separation capability. A detailed discussion of computer solutions,
including the handling of convergence problems, is available (41).
Computer solutions entail setting up component equilibrium and compo-
nent mass and enthalpy balances around each theoretical stage and specifying
the required design variables as well as solving the large number of simulta-
neous equations required. The explicit solution to these equations remains too
complex for present methods. Studies to solve the mathematical problem by algo-
rithm or iterational methods have been successful and, with a few exceptions, the
most complex distillation problems can be solved.
3.3. Multiple Products. If each component of a multicomponent distilla-
tion is to be essentially pure when recovered, the number of columns required for
the distillation system is N* 1, where N* is the number of components. Thus, in
a five-component system, recovery of all five components as essentially pure pro-
ducts requires four separate columns. However, those four columns can be
arranged in 14 different ways (42).
The number of columns in a multicomponent train can be reduced from the
N* 1 relationship if side-stream draw-offs are used for some of the component
cuts. The feasibility of multicomponent separation by such draw-offs depends
on side-stream purity requirements, feed compositions, and equilibrium relation-
ships. In most cases, side-stream draw-off distillations are economically feasible
only if component specifications for the side-stream are not tight. If a single com-
ponent is to be recovered in an essentially pure state from a mixture containing
both lower- and higher-boiling components, a minimum of two columns is
required, one column to separate the lower boilers from the desired component
and another column to separate the component from the higher boilers.
Vol. 1 DISTILLATION 643

The economics of the various methods that are employed to sequence multi-
component columns have been studied. For example, the separation of three-,
four-, and five-component mixtures has been considered (43) where the heuris-
tics (rules of thumb) developed by earlier investigators were examined and an
economic analysis of various methods of sequencing the columns was made.
The study of sequencing of multicomponent columns is part of a broader field,
process synthesis, which attempts to formalize and develop strategies for the
optimum overall process (44).

4. Distillation Columns

Distillation columns are vertical, cylindrical vessels containing devices that


provide intimate contacting of the rising vapor with the descending liquid.
This contacting provides the opportunity for the two streams to achieve some
approach to thermodynamic equilibrium. Depending on the type of internal
devices used, the contacting may occur in discrete steps, called plates or trays,
or in a continuous differential manner on the surface of a packing material.
The fundamental requirement of the column is to provide efficient and economic
contacting at a required mass-transfer rate. Individual column requirements
vary from high vacuum to high pressure, from low to high liquid rates, from
clean to dirty systems, and so on. As a result, a large variety of internal devices
has been developed to fill these needs. The column devices discussed herein are
used for absorption (qv) and stripping as well as distillation. The principal opera-
tional difference is that in absorption or stripping, the gas flowing up the column
is primarily a noncondensable phase at column conditions, whereas in distilla-
tion the gas phase is a condensable vapor.
4.1. Plate Columns. There are two general types of plates in use: cross-
flow and counterflow. These names refer to the direction of the liquid flow rela-
tive to the rising vapor flow. On the cross-flow plate the liquid flows across the
plate and from plate to plate via downcomers. On the counterflow plate liquid
flows downward through the same orifices used by the rising vapor.
Crossflow Plates. As indicated in Figure 15, liquid enters a crossflow plate
from the bottom of the downcomer of the plate above and flows across the active or
bubbling area where it is aerated by the vapors flowing through orifices from the
plate below. It is in this aerated zone where most of the vapor–liquid mass trans-
fer occurs. The aerated mixture flows over the exit weir into a downcomer. A
vapor–liquid disengagement takes place in the downcomer and most of the
trapped vapor escapes from the liquid and flows back to the interplate vapor
space. The liquid, essentially free of entrapped vapor, leaves the plate by flowing
under the downcomer to the inlet side of the next lower plate. The vapor, disenga-
ging from the aerated mass on the plate, rises to the next plate above.
The pressure drop incurred by the vapor as it passes through the orifices of
the plate is fundamental to plate operation. In most plate designs, the pressure
drop prevents the crossflowing liquid from falling through the plate. The pres-
sure drop also results from the energy consumed to disperse the vapor–liquid
mixture, eg, to atomize a portion of the liquid to provide increased interfacial
area for mass transfer. Diameters of commercial crossflow plate columns range
644 DISTILLATION Vol. 1

Downcomer
Liquid
deentraining
zone
Bubbling
area
Exit weir

Downcomer
Vapor
clearance
disengaging
from liquid

Degassed
liquid

Fig. 15. Flow pattern in a crossflow plate distillation column.

from 0.3 to 15 m and plate spacings range from 0.15 to 1.2 m. The total pressure
drop per plate is often in the range of 0.25–1.6 kPa (2–12 mm Hg).
Three principal vapor–liquid contacting devices are used in current cross-
flow plate design: the sieve plate, the valve plate, and a hybrid of the tow in
which the ‘‘valves’’ are in a fixed, open position. In past years another type of
crossflow plate containing bubble caps was specified, and there are some existing
columns containing such vapor dispering devices. Details of the bubble cap tray
and its design features were covered by Bolles (45). All the devices mentioned
function to provide the needed intimate contacting of vapor and liquid, requisite
to maximizing transfer of mass across the vapor–liquid interfacial boundary.
Sieve Plates. The conventional sieve or perforated plate is inexpensive
and the simplest of the devices normally used. The contacting orifices in the con-
ventional sieve plate are holes that measure 1 to 12 mm diameter and exhibit
ratios of open area to active area ranging from 1:20 to 1:7. If the open area is
too small, the pressure drop across the plate is excessive; if the open area is
too large, the liquid weeps or dumps through the holes.
Valve Plates. Valve plates are categorized as proprietary and details of
design vary from one vendor to another. These represent a variation of the sieve
plate in which the holes are large and are fitted with liftable valve units such as
those shown in Figure 16. The principal advantage over sieve plates is the ability
to maintain efficient operation over a wider operating range through the use of
variable orifices (valves), which open or close depending on vapor rate. The most
common valve units consist of flat disks having attached legs that allow the valve
Vol. 1 DISTILLATION 645

w
flo
uid
Liq

(a) (b) (c)

Fig. 16. Examples of individual valve units used in valve plates: (a) Flexitray valve,
courtesy of Koch-Glitsch Inc.; (b) Float Valve, and (c) Fixed Valve (V-Grid), both courtesy
of Sulzer Chem-Tech, Inc.

to move upward or downward. Sometimes two weights of valves are used on a sin-
gle plate to extend operating range and improve vapor distribution. The valve
units usually have a tab or indentation that provides a minimum open area of
vapor flow, even when the valve is closed, and also prevents the valve from stick-
ing under corrosive or fouling conditions. Details on valve plate geometry, along
with methods for valve plate design, are available from valve plate vendors.
Fixed Valve Plates. For several years plates with a combination of sieve
holes and movable valves were marketed by several vendors, and can still be
obtained on special order. Of more current interest is the crossflow tray con-
taining small ‘‘valves’’ that are fixed in the full open position. Some designers
are specifing that the area under the downcomers be perforated with these fixed
valves, requiring the downcomers to be suspended as shown in Figure 17. This

Fig. 17. Flow pattern in a high-capacity crossflow plate with suspended downcomer.
646 DISTILLATION Vol. 1

arrangement allows an increase in vapor flow but requires careful design of the
downcomer to prevent vapor by-passing up the downcomer.
Multiple Liquid-Path Plates. As the liquid flow rate increases in large
diameter crossflow plates (ca 4 m or larger), the crest heads on the overflow weirs
and the hydraulic gradient of the liquid flowing across the plate become exces-
sive. To obtain improved overall plate performance, multiple liquid-flow-path
plates may be used, with multiple downcomers. These designs are illustrated
and discussed in detail in the literature (45).
Counterflow Plates. Counterflow plates are used less frequently than
crossflow plates. The liquid flows downward and the vapor upward through the
same orifices in a counterflow plate and the plate does not have downcomers.
The openings are round holes (dualflow tray) or slots (Turbogrid tray). A variation
of the dualflow tray is the Ripple tray in which the tray floor is shaped in a corru-
gated fashion (46). Counterflow plates are used advantageously in fouling services
because for each hole vapor and liquid flow alternately, providing a self-cleaning
action that is quite effective. The dualflow and Turbogrid plates have similar oper-
ating characteristics, and typical operating data have been published (47).
Another important plate which has characteristics similar to a counterflow
plate is the Multiple Downcomer (MD) plate (48). This is a plate where the active
area occupies the full column cross section but with a plurality of small downco-
mers interspersed among the perforations. The downcomers are specially sealed
to prevent upflow of vapor through them, as in the case of high-capacity sus-
pended down comer plates (Fig. 17). The plate has been used successfully in
many high-liquid-flow cases.
Vapor Capacity Parameters. The diameter of a distillation column is
determined by the capacity of the column to handle the required flows of vapor
and liquid. The vapor capacity parameter is
" #0:5

rg
Csb ¼ V ð42Þ
rL rg

and its simplification


 0:5
F  ¼ V  rg ð43Þ

The term Csb in equation 42 is called a Souders–Brown capacity parameter and


is based on the tendency of the upflowing vapor to entrain liquid with it to the
plate above. The term F* in equation 43 is called an F-factor. For Csb and F* to
be meaningful the cross-sectional area to which they apply must be specified. The
capacity parameter is usually based on the total column cross section minus the
area blocked for vapor flow by the downcomer(s). For the F factor, typical
operating ranges for sieve plate columns are

Area basis (kg/(m  s2 ))0.5 (lb/(ft  s2 ))0.5


F S* total cross section 0.6–3.0 0.5–2.5
F A* active area 0.85–4.3 0.7–3.5
F H* hole area 8.5–30 7–25
Vol. 1 DISTILLATION 647

, m/s
0.20
0.2

( ( ( 20
σ 0.10
900
600
0.08
450
0.5

0.06
300
ρL − ρg

225
ρg

0.04 150
( Csb = V ∗

0.02

Mixed froth Emulsion flow


0.01
0.01 0.02 0.05 0.1 0.2 0.5 1.0 2.0
Flow parameter

Fig. 18. Flooding correlation for crossflow trays (sieve, valve, bubble-cap) where the
numbers represent tray spacing in mm. Also shown are approximate boundaries of the
spray zone, and mixed froth and emulsion flow regimes.

Entrainment Flooding. The vapor capacity of a column is limited by


excessive entrainment, usually called flooding. A flooding condition can be
observed when the holdup of liquid becomes excessive, the pressure drop
increases dramatically, and the mass-transfer efficiency falls precipitously. Esti-
mates of the vapor velocity for a flooding condition may be made from the chart in
Figure 18 (49). The abscissa term L/G(rg/rL)0.5 is called a flow parameter and its
value can indicate several things about the character of the aerated mass on the
plate. For example, a very low value can indicate a phase inversion in which the
vapor flow is continuous (spray flow), whereas a high value can indicate a bubbly
mass (emulsion flow). The value of the flow parameter is easily determined from
the stage calculations (reflux and boilup ratios) and densities of the phases. The
ordinate value in Figure 18 leads to a value of the flooding velocity, and prudent
design calls for limiting actual flows to 70–80% of this velocity.
Downcomer Flooding. For cases of very high liquid-to-vapor flow ratios
the limiting capacity of the column is based on the ability of the downcomers
to move the de-aerated liquid from a plate to the next plate below. It is clear
that there can be constrictions in the downcomer design or that even with no con-
strictions there is simply not enough flow area to accommodate the high volume
of liquid. Thus, the downcomer can flood, or choke, when it becomes completely
filled with liquid or aerated mass. Typical design heuristics include limiting the
downcomer velocity (clear liquid basis) to no more than 0.12 m/s. Also, to allow
for complete disengagement of vapor from liquid in the downcomer, a minimum
residence time of 4 s is often used. The actual limiting values of these parameters
varies somewhat with the properties of the fluids and the exact dimensions of the
plate components.
Stable Operating Range. All plates have a stable operating envelope
bound by a range of liquid and vapor flow rates as shown in Figure 19. The size
and shape of the stable area depends on the plate design and on the system prop-
erties. The line AD represents the minimum operable vapor flow rate at various
648 DISTILLATION Vol. 1

B
Floo
ding
Blowing

Vapor rate

r
t

come
inmen

ing
entra h
Stable operation

Hig

flood
Down
D
p ing
g, dum
Weepin
A

Liquid rate

Fig. 19. Stable operating range for crossflow plates.

liquid flow rates. Below AD, the vapor rate is too low to maintain the liquid on
the plate and, as a result, the liquid weeps excessively or dumps through the
plate orifices. Above line BC the column floods by entrainment. To the right of
CD the high liquid rate causes downcomer flooding. The area to the left of AB
represents high entrainment at low liquid flow rates, with vapor jets at the ori-
fices. Design procedures for sieve plates have been published (eg 49,50). Vendors
of valve trays make available their design methods.
Plate Efficiencies. Column requirements are calculated in terms of theo-
retical stages or plates. Actual plates must, however, be specified in the design.
Thus the effectiveness of the plate in approaching the equilibrium condition
must be predicted. This approach is called the plate efficiency, which is a mea-
sure of the rate of mass transfer on the actual plate. This efficiency, expressed
either as a fraction or as a percentage, depends on three principal factors: the
geometry of the plate (hole arrangement, valve design, etc); the loading of
vapor and liquid traffic on the plate; and the diffusional properties of the fluids.
The simplest efficiency is the overall column efficiency, which is the number
of theoretical plates in a column divided by the number of actual plates:
Eo ¼ Nt =Na ð44Þ
Thus the overall efficiency is an averaged efficiency of all the individual plates.
A more useful plate efficiency for theoretical prediction is the Murphree
plate efficiency:
yn yn 1
Emv ¼ ð45Þ
yn yn 1
where yn and yn 1 are the vapor compositions from plate n and n 1 (the plate
below n), and yn* is the vapor composition that would be in equilibrium with
Vol. 1 DISTILLATION 649

the liquid composition leaving plate n. Thus, for a given plate, Emv is a ratio of
the actual vapor composition change to the change that would occur if the plate
were effective enough to bring the vapor and liquid to thermodynamic equili-
brium. This definition is based on the outlet liquid composition, and says nothing
about the average liquid composition on the plate. In cases where a significant
concentration gradient exists in the liquid composition across the plate, it is pos-
sible for Emv to have a value greater than 1.0 (100%). Equation 45 is written in
terms of vapor composition. A similar equation can be written in terms of the
liquid compositions and is denoted as EmL.
Of still more theoretical importance is the efficiency at some point on the
plate:
" #
yn yn 1
Eog ¼ ð46Þ
yn yn 1
point

This parameter is called the point efficiency (or local efficiency). It cannot have a
value greater than 1.0, and it has a counterpart term for liquid compositions.
Prediction of Plate Efficiency. As of this writing, the most comprehen-
sive study of plate efficiency known was made in the mid-1950s, based on the
then-still-popular bubble cap plates (51). Unfortunately, the predictive model
developed has been shown to be inadequate for many industrial distillations.
An improvement of the model, more oriented toward the sieve plate, was pub-
lished in 1984 (52). There has been continuing research effort directed toward
a better understanding of the mechanisms that occur in the rather complex aer-
ated mass on the typical plate (53–55). A complicating factor is the lack of uni-
form liquid flow across the plate, and situations have been found where the liquid
actually stagnates in certain zones of larger-diameter plates. For larger columns
it is possible for the observed Murphree efficiency to exceed 100%. A satisfactory
method for predicting plate efficiency does not exist. Most recently there have
been studies of the various types of flow regimes that occur on operating plates
and of the effect of these regimes on tray performance, including plate efficiency.
Pursuit of the flow regime studies (56–59) may lead to improved plate efficiency
prediction methods. For example, a newer model (60) takes into account the
regime as well as the vapor bubble (froth flow) or liquid drop (spray flow) char-
acteristics in determining mass-transfer coefficients in the aerated zone on the
plate.
Empirical methods for predicting plate efficiency have been proposed. Prob-
ably the most widely used method correlates overall column efficiency as a func-
tion of feed viscosity and relative volatility (61).
General Comments on Plate Efficiency. The plate efficiencies of well-
designed commercial bubble cap, sieve, and valve plates are approximately the
same when the plates are operated within their normal design range. The
plate efficiency decreases both at the low end of the plate’s operating range,
where the liquid tends to leak through the plate, and at the high end of the oper-
ating range, where liquid entrainment becomes substantial.
Most distillation systems in commercial columns have Murphree plate effi-
ciencies of 70% or higher. Lower efficiencies are found under system conditions of
650 DISTILLATION Vol. 1

a high slope of the equilibrium curve (Fig. 1b), of high liquid viscosity, and of
large molecules having characteristically low diffusion coefficients. Finally,
most experimental efficiencies have been for binary systems where by definition
the efficiency of one component is equal to that of the other component. For mul-
ticomponent systems it is possible for each component to have a different effi-
ciency. Practice has been to use a pseudo-binary approach involving the two
key components. However, a theory for multicomponent efficiency prediction
have been developed (62,63) and are amenable to computational analysis.
4.2. Packed Columns. In packed columns, the vapor–liquid contacting
takes place in continuous beds of solid packing elements rather than in discrete
individual plates. The contacting can be visualized as occurring in differential
increments across the height of the packing; thus packings are known as coun-
terflow devices rather than stagewise devices. Mechanically, the packed column
is a relatively simple structure. In its simplest form the packed column comprises
a vertical shell having dumped or carefully arranged packing elements on an
open-type support, together with a suitable liquid distribution device above the
packed bed. A packed column having two packed beds and a midcolumn feed is
shown in Figure 20. The vapor enters the column below the bottom bed and flows

Reflux Vapor Reflux


liquid distributor

Packing
Packing support
Feed
Feed
distributor

Packing
support
Packing

Vapor Liquid

Fig. 20. Packed column shell and internals. Column shown has single packed beds above
and below the feed. For separations requiring a large number of stages, additional beds,
separated by redistribution devices, are likely to be needed.
Vol. 1 DISTILLATION 651

upward through the column. The liquid (reflux or other liquid stream) enters at
the top through the liquid distributor and flows downward through the packing
counter-currently to the rising vapor. The height of the individual packed beds is
limited to 2–9 m by the mechanical strength of the packing or by the need to
redistribute the liquid so that good mass-transfer efficiency can be maintained.
Packings. For many years packed columns consisted of randomly dum-
ped packings almost exclusively, with occasional applications of regularly
stacked packings or pads of woven or knitted wire. In the late 1960s a partial
trend away from random packings began when a special structured packing
made of wire gauze was introduced by Sulzer Brothers in Switzerland (64).
The indicated advantages of the structured packings were high mass-transfer
efficiency and very low pressure drop. These devices appeared to be ideal for
high-vacuum distillations. However, cost of fabrication was very high and they
were considered mainly for the vacuum distillation of specialty chemicals. In
1977, a lower-cost sheet metal version was introduced (65), and since that time
a large business in structured sheet metal packings has arisen. At the same time,
improved random packings have been developed and a comprehensive discussion
of their characteristics has been published (66). Some of the common random
packings are shown in Figure 21. The Raschig ring, one of the oldest of packings,
is an open cylinder of equal height and diameter. The Berl saddle and the cera-
mic Intalox saddle (Norton Co.) have a higher capacity and efficiency than the
Raschig ring. The Pall ring is a modification of the Raschig ring which allows
through-flow of liquid and vapor, with consequent lower pressure drop and better
efficiency. The newer Intalox metal saddle (IMTP) is an example of a random
packing having a very high void fraction and low resistance to the flowing
phases. Other newer random packings, not shown in Figure 21, include the
CMR ring (Koch-Glitsch, Inc.) and the Nutter ring (Sulzer Chemtech Inc.). The
random-type packings can generally be made from metal, plastic, or ceramic
materials; the approximate nominal size range for the individual elements is
12–75 mm.
Common structured packing geometries are shown in Figure 22. Flat plates
of gauze or sheet metal are perforated, or embossed or lanced, and corrugated.
Corrugated sheets are then stacked together such that adjacent sheets have
opposite corrugation directions. The corrugations have angles with the hori-
zontal of 45 to 60 degrees. Vapor and liquid contact each other in wetted-wall

Raschig

(a) (b) (c) (d) (e)

Fig. 21. Random packing elements for distillation columns: (a) Raschig ring (metal);
(b) Berl saddle (ceramic); (c) Intalox saddle (ceramic); (d) Pall ring (metal); and (e) Intalox
saddle (metal).
652 DISTILLATION Vol. 1

Flow channel

(a)

s
h

(b)

Fig. 22. (a) Flow channel arrangement; (b) flow channel triangular cross section where
for angles of 908, DEQ ¼ 4RH ¼ 4ð S2 Þ 2S
1
¼ S.

fashion, and the perforations plus other surface enhancements, eg, texturing,
serve to promote liquid spreading into thin films. Dimensions, performance char-
acteristics, and design procedures for the structured packings are summarized in
Reference 67.
Packed Column Internals. In order to ensure good packed column mass-
transfer efficiency, the liquid must be distributed uniformly over the surface of
the packing. As a general rule there should be at least 100 pour points per square
meter (10 points/ft2), although fewer points may be used for random packings of
the bluff-body type such as Raschig rings and Berl saddles. Although they have
capacity and pressure drop limitations, the bluff-body packing elements are able
to divide the downflowing liquid and thus improve on an initially marginal
distribution. On the other hand, the through-flow-type random packings, eg,
Pall rings and Intalox metal tower packings, as well as the structured packings,
are not able to correct the initial distribution and in fact may allow some dete-
rioration of distribution if the bed heights are greater than about 5 m.
Vol. 1 DISTILLATION 653

2.0
2.0

Pressure drop per unit height of packing


1.0
1.0
in. of water/ft

e
rat
0.5

rat

kPa/m
g
uid
0.5

uid

kin
iq

ac
liq
hl

yp
Low
Hig

Dr
0.1
0.1
Gas rate, mass per unit cross-sectional area per unit time

Fig. 23. Log–log plot of pressure drop per unit height of typical packing as a function of
gas rate at two liquid rates and for the unirrigated packing.

Considerable research is in progress on methods for ensuring good liquid


distribution in large-diameter columns, and the packing manufacturers main-
tain large test stands where a particular design of distributor can be tested
using water before being installed in the column. The distributor design problem
becomes more severe at low, ie, <700 cm3 =ðsm2 Þ ð1 gal=ðmin ft2 ÞÞ liquid rates or
in large (>3 m) diameter towers. An example of a more fundamental study of
liquid distribution is available (68), as are typical liquid distributor designs
and typical packing supports (50).
Packed Column Operation. In the packed column, liquid flows down-
ward in opposition to the upward flow of vapor; both phases flow through the
same open space or interstices between the packing elements. At low liquid
and gas flow rates, the descending liquid occupies only a small fraction of the
interstices and, therefore, offers little hindrance to the rising vapor flow.
Figure 23 shows a schematic plot of pressure drop per unit of height as a function
of the gas rate at low and high liquid flow rates. At a low rate of gas flow, the log
slope of each curve is approximately 2. As the gas flow rate increases, there is an
increasing tendency for the liquid to be held up in the void space, thereby
decreasing the space available for the gas flow. As the gas flow rate increases
further, more liquid is held up until at some high gas rate the packing floods.
At this point, the liquid is essentially filling the interstices and can no longer
flow downward. At flooding, the log slope is practically infinite. The pressure
drop at the inception of flooding ranges from 1.6 to 3.3 kPa/m (2 to 4 in. water/ft)
of packing. More comprehensive discussions of packed column hydraulics may
be found in distillation texts (15,17,69), monographs (66,70,71), or handbooks
(72,73).
Capacity of Packed Columns. Packed columns are usually designed to
operate at some percentage approach to flooding, eg, 60–70%, or at some speci-
fied pressure drop per unit height of packing, eg, 0.8 kPa/m (1 in. water/ft) of
packing. Flooding correlations have been proposed (74), one revision introducing
654 DISTILLATION Vol. 1

0.2
1.2
0.82
0.1
0.41
0.06
,(mm2 /s) 0.1

0.04 0.20

0.02
0.08
gρ g ( ρL − ρG )
G 2Fp v 0.1

0.01
∆ P = 0.04
0.006

0.004

0.002

0.001

0.0006

0.0004
0.004 0.04 0.4 4.0 8.0
0.006 0.02 0.06 0.2 0.6 2.0 6.0
0.01 0.1 1.0

ρg
( (
0.5
L
G ρL

Fig. 24. Generalized method using log scales for estimating packed column flooding and
pressure drop, DP, in kPa/m; g ¼ gravitational constant, 9.81 m/s2; v ¼ kinematic
viscosity in mm2 =s (¼ cSt); L, G have units of kg/(m2 s); rL, rG are in kg/m3; and the pack-
ing factor, Fp, in m 1 can be found in Table 2. To convert kPa/m to mm Hg/m, multiply
by 7.5 (66).

constant-pressure drop lines (75). The most recent revision in these correlations
is shown in Figure 24 (66). The idea of flooding has been eliminated from the
chart with the stipulation that the topmost curve represents the maximum capa-
city. Experimentally determined packing factors Fp, presented in Table 2, should
be used in the ordinate group. These factors distinguish between the various
shapes and sizes of the available packings. The curves are for constant pressure
drop and thus the chart enables estimation of both capacity and pressure drop.
Packing Mass-Transfer Characteristics. The contacting for mass trans-
fer in a packed column occurs differentially along the length of the column. The
separation calculations can thus be made on a differential basis along this
length, using mass-transfer coefficients or heights of transfer units. The calcula-
tions are somewhat imprecise because of the uncertainty in the fundamental
mass-transfer mechanisms in larger-scale columns. Useful models for predicting
the mass-transfer efficiency of randomly packed columns have been published
(76,77), using the same database of commercial-scale performance data. These
models cover the better known packings, eg, metal and ceramic Raschig rings,
ceramic Berl saddles, and metal Pall rings, in nominal sizes in the range of
12 to 50 mm. It has been found that to avoid excessive maldistribution of liquid
Vol. 1 DISTILLATION 655

Table 2. Characteristics of Packinga


Nominal size, Surface area, Packing factor,
Packing mmb m2/m3 Void fraction Fp, m 1
Dumped (random) packing
Intalox saddles
ceramic
25 255 0.77 197
50 118 0.79 98
metal (IMTP) 25 0.97 135
40 0.97 79
50 0.98 59
plastic 25 206 0.91 130
50 108 0.93 92
75 88 0.94 59
Berl saddles, 13 465 0.62 790
ceramic
25 250 0.68 360
50 105 0.72 150
Pall rings
metal 16 0.92 265
25 205 0.94 183
50 115 0.96 88
plastic 16 341 0.87 310
25 207 0.90 180
50 100 0.92 85
Raschig rings, ceramic 13 370 0.64 1902
25 190 0.74 587
50 92 0.74 215
Structured packing
Flexipac
1 6 558 0.91 108
2 12 246 0.93 72
3 37 134 0.96 52
4 50 69 0.98 30
Sulzer-BX 6 490 >0:90 66
a
Ref. 66.
b
For structured packings, values correspond to crimp height.

near the wall, a ratio of column diameter to packing element size of at least eight
should be maintained. Thus if one wishes to conduct pilot-scale packed column
tests, a minimum column diameter of about 100 mm would be used together
with 12-mm packing elements. The models would then permit scale-up to large
columns containing 50-mm size elements of the same type, eg, Pall rings.
These models provide values of the height of a transfer unit for the liquid
phase HL and the vapor phase HV. These values are combined to form the height
of an overall transfer unit, Hov:

Hov ¼ Hv þ ðm0 V=LÞHL ð47Þ

where V and L are molar flow rates of vapor and liquid and m0 is the slope of the
y–x equilibrium curve (Fig. 1b) in the concentration range of interest. The
656 DISTILLATION Vol. 1

required total height of the packed section is then obtained from the simple rela-
tionship,
Zp ¼ ðNov Þ ðHov Þ ð48Þ

In order to determine the packed height Zp it is necessary to obtain a value


of the overall number of transfer units Nov; methods for doing this are available
for binary systems in any standard text covering distillation (69) and, in a more
complex way, for multicomponent systems (63,78). However, it is simpler to cal-
culate the number of required theoretical stages and make the conversion:

Nov ¼ Nt ðln m0 V=LÞ ðm0 V=L 1Þ ð49Þ

An alternative to determining packed height is through the use of an empirical


term, height equivalent to a theoretical plate (HETP). This term can be measured in
a fashion similar to that used for the overall plate efficiency of a column (eq. 44):
total packed height Zp
HETP ¼ ¼ ð50Þ
no: of theoretical plates Nt

Typical experimental values of HETP for a random packing such as 50-mm


Pall rings, and a structured packing, such as Intalox 2T of Norton Co., under
the same system conditions, are shown in Figure 25. Many designers of packed

FS, ft/s (lbs/ft3)0.5


0 1.0 2.0 3.0
0.75 30

0.50 20
HETP, in.
HETP, m

0.25 10

0 0
0 0.5 1.0 1.5 2.0
FS, m/s·(kg/m3)0.5

Fig. 25. Values of HETP as a function of throughput for ( ) 50-mm metal Pall rings
and ( ) No. 2 structured packing at 12-mm crimp height. Conditions are cyclohexane/
n-heptane system, 165 kPa (24 psia) operating pressure, total reflux, 0.43-m diameter
column. Courtesy of The University of Texas at Austin.
Vol. 1 DISTILLATION 657

columns prefer the use of HETP instead of Hov, but the latter is more fundamen-
tal and discriminates between liquid- and vapor-phase resistances. It should be
noted that terms such as Hov and Nov are based on vapor-phase concentrations;
equivalent terms based on liquid concentrations could be used.
For structured packings, methods for predicting Hv and HL are some-
what more reliable, partly because the fluid mechanics are less complicated
than those for random packings (they can be modeled by analogy to wetted
wall columns) and also because they have been extensively tested under distil-
lation conditions and for a variety of test mixture properties. The earliest
mechanistic model for structured packings appeared in 1985 (79), and has
since been expanded significantly for predicting hydraulics (80) and mass
transfer efficiency (81). An alternate model of Olujic (82) was developed
separately, and the two models have been compared (83). For these models
the key design parameters are: corrugation height, nature of metal surface,
angle of corrugation from the horizontal (45 vs 60 deg) in addition to the
usual physical property and flow rate considerations. Methods for predicting
pressure drop and flooding in beds of structured and random packings have
been reviewed (84).
4.3. Packed vs Plate Columns. Relative to plate towers, packed
towers are more useful for multipurpose distillations, usually in small
(under 0.5 m) towers or for the following specific applications: severe corro-
sion environment where some corrosion-resistant materials, such as plas-
tics, ceramics, and certain metallics, can easily be fabricated into packing
but may be difficult to fabricate into plates; vacuum operation where a low
pressure drop per theoretical plate is a critical requirement; high (eg, above
49,000 kg/(h m2 ) [10; 000 lb/(h ft2 )] liquid rates; foaming systems; or
debottlenecking plate towers having plate spacings that are relatively close,
under 0.3 m.
Plate columns have the advantage of lower fabrication cost, less depen-
dence on good liquid and gas distribution, and protection against vapor bypass-
ing the liquid in critical zones, eg, regions of extremely low impurities.
Further, methods for the design on plate columns are somewhat more reliable
than those for many of the packings, especially those packings of a proprietary
nature.
There are notable cases where plate columns have been converted to
packed columns to gain advantage of the low pressure drop exacted from the
vapor stream. More recently the packings have been largely of the structured
type. Illustrative of this is the trend toward the use of structured packing in
ethylbenzene –styrene fractionators, some of which have diameters of 10 m
or higher.

5. Steam Distillation

Steam distillation is used to lower the distillation temperatures of high-boiling


organic compounds that are essentially immiscible with water. If an organic com-
pound is immiscible with water, both liquids exert full vapor pressure upon
vaporization from the immiscible two-component liquid. At a system pressure
658 DISTILLATION Vol. 1

of P, the partial pressures would be:

P ¼ pwater þ porganic ð51Þ

and because the water and organic compound are immiscible:

P ¼ P0 water þ P0 organic ð52Þ

The steam distillation of N-ethylaniline at atmospheric pressure (73) gives


the following: the vapor pressures at 99.158C of water and N-ethylaniline
are 98.27 and 3.04 kPa (737 and 22.8 mm Hg), respectively. Thus, according to
equation 52,

P ¼ 98:27 þ 3:04 ¼ 101:3 kPa ð53Þ

and the concentration of the N-ethylaniline in the vapor is

y ¼ 3:04=101:3 ¼ 0:030 mol fraction ð54Þ

The normal boiling point of N-ethylaniline is 2048C. Therefore, steam


distillation makes possible the distillation of N-ethylaniline at atmosphe-
ric pressure at a temperature of 99.158C instead of its normal boiling point
of 2048C. Commercial applications of steam distillation include the frac-
tionation of crude tall oil (85) the distilling of turpentine, and certain essential
oils.

6. Molecular Distillation

Molecular distillation occurs where the vapor path is unobstructed and the
condenser is separated from the evaporator by a distance less than the
mean-free path of the evaporating molecules (86). This specialized branch
of distillation is carried out at extremely low pressures ranging from 13–
130 mPa (0.1 – 1.0 mm Hg). Molecular distillation is confined to applications
where it is necessary to minimize component degradation by distilling at
the lowest possible temperatures. Commercial usage includes the distillation
ofvitamins (qv) and fatty acid dimers (see DIMER ACIDS).
6.1. Distillation as a Separation Method. Distillation is the most
important industrial method of separation and purification of liquid compo-
nents. Liquid separation methods in less common use include liquid–liquid
extraction, membrane diffusion, ion exchange, and adsorption. However, distilla-
tion does not require a mass-separating agent such as a solvent, adsorbent, or
membrane, and distillation utilizes energy in a convenient heating medium
(often steam). Also, a wealth of experience with design and operations makes dis-
tillation column performance prediction more reliable than equivalent predic-
tions for other methods. At times distillation also competes indirectly with
methods involving solid–liquid separations such as crystallization. An
Vol. 1 DISTILLATION 659

extensive discussion of the selection of alternative separation methods is avail-


able (29).
The suitability and economics of a distillation separation depend on such
factors as favorable vapor–liquid equilibria, feed composition, number of compo-
nents to be separated, product purity requirements, the absolute pressure of the
distillation, heat sensitivity, corrosivity, and continuous vs batch requirements.
Distillation is somewhat energy inefficient because in the usual case heat added
at the base of the column is largely rejected overhead to an ambient sink. How-
ever, the source of energy for distillations is often low-pressure steam, which
characteristically is in long supply and thus relatively inexpensive. Also,
schemes have been devised for lowering the energy requirements of distillation
and are described in many publications (87).
6.2. Favorable Vapor–Liquid Equilibria. The suitability of distilla-
tion as a separation method is strongly dependent on favorable vapor–liquid
equilibria. The absolute value of the key relative volatilities directly determines
the ease and economics of a distillation. The energy requirements and the num-
ber of plates required for any given separation increase rapidly as the relative
volatility becomes lower and approaches unity. For example: given an ideal bin-
ary mixture having a 50 mol% feed and a distillate and bottoms requirement of
99.8% purity each, the minimum reflux and minimum number of theoretical
plates for assumed relative volatilities of 1.1, 1.5, and 4 are

Minimum no. of
Relative volatility Minimum reflux ratio theoretical plates
1.1 20 130
1.5 4.0 31
4 0.66 9

In the example, the minimum reflux ratio and minimum number of theore-
tical plates decreased 14- to 33-fold, respectively, when the relative volatility
increased from 1.1 to 4. Other distillation systems would have different specific
reflux ratios and numbers of theoretical plates, but the trend would be the same.
As the relative volatility approaches unity, distillation separations rapidly
become more costly in terms of both capital and operating costs. The relative
volatility can sometimes be improved through the use of an extraneous solvent
that modifies the VLE. Binary azeotropic systems are impossible to separate into
pure components in a single column, but the azeotrope can often be broken by an
extraneous entrainer (see DISTILLATION, AZEOTROPIC AND EXTRACTIVE).
6.3. Feed Composition. Feed composition has a substantial effect
on the economics of a distillation. Distillations tend to become uneconomical
as the feed becomes dilute. There are two types of dilute feed cases, one in
which the valuable recovered component is a low boiler and the second when
it is a high boiler. When the recovered component is the low boiler, the absolute
distillate rate is low but the reflux ratio and the number of plates is high.
An example is the recovery of methanol from a dilute solution in water.
When the valuable recovered component is a high boiler, the distillate rate,
the reflux relative to the high boiler, and the number of plates all are high.
660 DISTILLATION Vol. 1

An example for this case is the recovery of acetic acid from a dilute solution in
water. For the general case of dilute feeds, alternative recovery methods are
usually more economical than distillation.
6.4. Product Purity. Product purity requirements influence choice of
separation methods. For favorable equilibria, distillation energy requirements
do not increase significantly as purity specifications become tighter. For example,
in an ideal binary distillation of 60 mol % of A in the feed, the minimum and oper-
ating reflux ratios would be essentially the same whether the required purity of A
was 99 or 99.9999%. The number of plates would increase substantially, however,
as the purity requirements became more stringent. The shortcut methods of calcu-
lating minimum reflux ratio, minimum number of plates, operating reflux ratio,
and number of operating plates allow a rapid evaluation of the effect of changes
in purity requirements on the key economic factors in distillation.
6.5. Operating Pressure. The absolute pressure of the distillation may
have substantial economic impact. The temperature at which heat is supplied to
the reboiler and removed from the condenser determines the unit cost of the
energy. The cost of removing heat in the condenser increases rapidly as the con-
densing temperature drops below the range of air or water cooling capability; eg,
the cost of removing a unit quantity of heat at 25 C may be one hundred times
as high as removing it at 1008C. Similarly, the cost of the energy required for the
reboiler increases rapidly as the boiler temperature increases above some level
determined by local conditions. For example, at a particular site low-pressure
waste steam at 1108C may be essentially without cost, but if a temperature level
of 2008C is required, the unit cost of the heat is much higher. The relative
cost of the heat being removed and supplied is the controlling factor determining
the design of some distillations. The use of multiple interstage reboilers and con-
densers at different energy levels, as well as the use of other operational modes
used to optimize the overall economics, has been discussed (88).
The absolute pressure may have a significant effect on the vapor–liquid
equilibrium. Generally, the lower the absolute pressure, the more favorable
the equilibrium. This effect has been discussed for the styrene–ethylbenzene
system (29). In a given column, increasing the pressure can increase the column
capacity by increasing the capacity parameter (see eqs. 42 and 43). Selection of
the economic pressure can be facilitated by guidelines (89) that take into consid-
eration the pressure effects on capacity and relative volatility. Low pressures are
required for distillation involving heat-sensitive material.
6.6. Heat Sensitivity. The heat sensitivity or polymerization tendencies
of the materials being distilled influence the economics of distillation. Many
materials cannot be distilled at their atmospheric boiling points because of
high thermal degradation, polymerization, or other unfavorable reaction effects
that are functions of temperature. These systems are distilled under vacuum in
order to lower operating temperatures. For such systems, the pressure drop per
theoretical stage is frequently the controlling factor in contactor selection. An
excellent discussion of equipment requirements and characteristics of vacuum
distillation may be found in Ref. 90.
6.7. Corrosivity. Corrosivity is an important factor in the economics of
distillation. Corrosion rates increase rapidly with temperature, and in distilla-
tion the separation is made at boiling temperatures. The boiling temperatures
Vol. 1 DISTILLATION 661

may require distillation equipment of expensive materials of construction;


however, some of these corrosion-resistant materials are difficult to fabricate.
For some materials, eg, ceramics, random packings may be specified, and this
has been a classical application of packings for highly corrosive services. On
the other hand, the extensive surface areas of metal packings may make
these more susceptible to corrosion than plates. Again, cost may be the final
arbiter.
6.8. Batch vs Continuous Distillation. The mode of operation also
influences the economics of distillation. Batch distillation is generally limited to
small-scale operations where the equipment serves several different distillations.
6.9. Research. Much of the research on commercial-size distillation
equipment is being done by Fractionation Research, Inc. (FRI), a nonprofit,
industry-sponsored, research corporation located in Stillwater, Oklahoma. The
industrial sponsors of FRI are fabricators, designers and constructors, or users
of distillation equipment. The general policy of FRI has been to publish very little
of the research results in the open literature. However, a number of key papers
have been delivered and several of the motion pictures of operating distillation
trays are available for purchase. Importantly, any proprietary reports of FRI,
older than 30 years, can now be obtained from the library of Oklahoma State
University at Stillwater. The released data cover literally thousands of experi-
mental runs on large-diameter trays such as sieve, bubble-cap, and dualflow. A
recent paper, co-published with FRI, provides an excellent summary of the state-
of-the-art of commercial-scale distillation technology (90).
The published literature dealing with distillation continues at a moderate
pace. Regular reviews are published by Ray (91,92) spanning the period 1967
through 1998.

7. Equipment Costs

A compilation of costs of distillation and related equipment is available (93) but


must be adjusted for the effect of inflation. Some of the commercial computer-
aided process design packages contain equipment cost information. For specialized
internals, such as distributors, support plates, packings, crossflow plates, and so on,
it is usually necessary to obtain cost information directly from the equipment
vendors. It is important to recognize that the cost of a distillation system includes
many components in addition to the column itself. For example, an expensive pack-
ing may be justified on the basis that it can reduce the cost of the column shell, foun-
dations, piping, and so on. Discussions of economics of distillation systems are
available (70,94).

8. Column Control

Distillation columns are controlled by hand or automatically. The parameters


that must be controlled are (1) the overall mass balance, (2) the overall enthalpy
balance, and (3) the column operating pressure. Modern control systems are
designed to control both the static and dynamic column and system variables.
For an in-depth discussion, see Refs. (95–98).
662 DISTILLATION Vol. 1

8.1. Nomenclature
Symbol Definition Units
A12, A21 constants in the Van Laar activity coefficient equation
B bottoms from column mol/s
Csb vapor capacity parameter m/s
D distillate from column mol/s
Eo overall column plate efficiency (eq. 44) fractional
Emv Murphree plate efficiency (eq. 45) fractional
Eog local, or point, efficiency based on vapor concentrations fractional
f fugacity kPa
F feed mol/s
F* F-factor (eq. 43) m=s  ðkg=m3 Þ0:5
FA* F-factor based on active (bubbling) area m=s  ðkg=m3 Þ0:5
FH* F-factor based on hole area m=s  ðkg=m3 Þ0:5
Fp packing factor from Table 2 1/m
G gas mass rate kg/s
H; enthalpy per mole
H enthalpy per unit time
H* Henry’s law constant (eq. 17) kPa/mol fraction
HETP height equivalent of theoretical plate m
HL height of a liquid-phase transfer unit m
Hov height of an overall transfer unit, vapor concentrations m
Hv height of a vapor-phase transfer unit m
K y*/x, vapor–liquid equilibrium ratio (eq. 1)
L liquid rate mol/s
L average liquid rate for section mol/s
L; liquid mass rate kg/s
m an equilibrium stage below the feed
m0 slope of equilibrium line
n an equilibrium stage above the feed
N number of stages
N* number of components
Na number of actual stages
Nov number of transfer units, vapor concentration basis
Nt number of theoretical stages
P total pressure of system kPa
p partial pressure kPa
P0 vapor pressure kPa
q heat to vaporize 1 mol feed divided by molal latent
heat of feed (eq. 36)
q; heat removed or added at column auxiliaries
R reflux mol/s
R0 gas law constant
T temperature K
v vapor molar volume m3/mol
V vapor molar rate mol/s
V; average molar vapor rate for section mol/s
V* vapor velocity m/s
x mole fraction in liquid
y mole fraction in vapor
y* mole fraction vapor in equilibrium with x
z compressibility factor in gas law
Zp height of packed bed m
a relative volatility (eq. 2)
Vol. 1 DISTILLATION 663

Symbol Definition Units


L
g liquid-phase activity coefficient (eq. 6)
g1 terminal activity coefficient, at infinite dilution
L12, L21 constant in Wilson activity coefficient model (eq. 13)
r fluid-phase density kg/m3
f fugacity coefficient (eq. 20)
Superscripts
B bottoms
C condenser
D distillate
E end
F feed
f feed stage
L liquid
m stage number m
m 1 stage below m
mþ1 stage above m
n stage number n
n 1 stage below n
nþ1 stage above n
N Nth component of components i to n
P pressure
S reboiler
T top column
V vapor
Subscripts
1,2,3 . . . n component numbers
B bottoms
D distillate
F feed
g gas
H component H of binary system L–H, H is the high boiler
i,j components of mixture 1 . . . i,j, . . . n
L component L of binary system L–H, L is the low boiler
L liquid
min minimum
P pressure
R rectifying section
S stripping section
T temperature

BIBLIOGRAPHY

‘‘Distillation’’ in ECT 1st ed., Vol. 5, pp. 156–187, by E. G. Scheibel, Hoffmann-LaRoche,


Inc.; in ECT 2nd ed., Vol. 7, pp. 204–248, by C. D. Holland and J. D. Lindsey, Texas A & M
University; in ECT 3rd ed., Vol. 7, pp. 849–891, by E. R. Hafslund, E. I. du Pont de
Nemours & Co., Inc.; in ECT 4th ed., Vol. 8, pp. 311–358, by James R. Fair, The Univer-
sity of Texas at Austin; ‘‘Distillation’’ in ECT (online), posting date: December 4, 2000, by
James R. Fair, The University of Texas at Austin.
664 DISTILLATION Vol. 1

CITED PUBLICATIONS

1. A. J. V. Underwood, Trans. I. Chem. E. 13, 34 (1935).


2. J. R. Fair, AIChE Symp. Ser. No. 235 79, 1 (1984).
3. J. Gmehling, U. Onken, and W. Arlt, Vapor–Liquid Equilibrium Collection (continu-
ing series), DECHEMA, Frankfurt, Germany, 1979.
4. M. Hirata, S. Ohe, and K. Nagahama, Computer Aided Data Book of Vapor–Liquid
Equilibria, Elsevier, Amsterdam, The Netherlands, 1975.
5. E. Hala, J. Pick, V. Fried, and O. Vilim, Vapor–Liquid Equilibrium, 2nd ed.,
Pergamon Press, Oxford, UK, 1967.
6. E. Hala, I. Wichterle, J. Polak, and T. Boublik, Vapor–Liquid Equilibrium at Normal
Pressures, Pergamon Press, Oxford, UK, 1968.
7. I. Wichterle, J. Linek, and E. Hala, Vapor–Liquid Equilibrium Data Bibliography,
Elsevier, Amsterdam, The Netherlands, 1975.
8. A. Fredenslund, J. Gmehling, and P. Rasmussen, Vapor–Liquid Equilibria Using
UNIFAC, Elsevier, Amsterdam, The Netherlands, 1977.
9. J. H. Hildebrand, J. M. Prausnitz, and R. L. Scott, Regular and Related Solutions,
Van Nostrand Reinhold Co., Inc., New York, 1970.
10. E. L. Derr and C. H. Deal, I. Chem. E. Symp. Ser. No. 32 3(40), (1969).
11. D. A. Palmer, Handbook of Applied Thermodynamics, CRC Press, Inc., Boca Raton,
Fla., 1987.
12. J. M. Prausnitz, R. N. Lichtenthaler, and E. G. Azeredo, Molecular Thermodynamics
of Fluid-Phase Equilibria, 3rd ed., Prentice-Hall, Inc., Englewood Cliffs, N.J., 1999.
13. R. C. Reid, J. M. Prausnitz, and B. Pohling, The Properties of Gases and Liquids, 4th
ed., McGraw-Hill Book Co., Inc., New York, 1987.
14. S. M. Walas, Phase Equilibria in Chemical Engineering, Butterworths, Reading,
Mass., 1998.
15. J. D. Seader, Separation Process Principles, John Wiley & Sons, Inc., New York, 1981.
16. P. Wankat, Equilibrium-Staged Separations, Elsevier Science Publishing Co., Inc.,
New York, 1988.
17. M. Van Winkle, Distillation, McGraw-Hill Book Co., Inc., New York, 1967.
18. J. J. Van Laar, Z. Physik. Chem. 72, 723 (1910); 83, 599 (1913).
19. G. M. Wilson, J. Am. Chem. Soc. 86, 127 (1964).
20. H. Renon and J. M. Prausnitz, AIChE J. 14, 135 (1968).
21. D. S. Abrams and J. M. Prausnitz, AIChE J. 21, 116 (1975).
22. Margules, Sitzber. Math.-Naturw. Kl. Kaiserlichen Akad. Wiss. (Vienna) 104, 1243
(1895).
23. H. H. Chien and H. R. Null, AIChE J. 18, 1177 (1972).
24. J. M. Prausnitz and co-workers, Computer Calculations for Multicomponent Vapor–
Liquid and Liquid–Liquid Equilibria, Prentice-Hall, Inc., Englewood Cliffs, N.J.,
1980.
25. Technical Data Book, Petroleum Refining, 3rd ed., Vols. I and II, American Petroleum
Institute, New York, 1976.
26. Y.-L. Huang, J. D. Olson, and G. E. Keller, Ind. Eng. Chem. Research 31, 1759 (1992).
27. J. Gmehling et al. Azeotropic Data, Wiley/VCH, New York, 1994. (2 Vols.)
28. L. Horsley, Azeotropic Data—III, Advances in Chemistry Series No. 116, American
Chemical Society, Washington, D.C., 1973.
29. C. J. King, Separation Processes, 2nd ed., McGraw-Hill Book Co., Inc., New York,
1980.
30. W. L. McCabe and E. W. Thiele, Ind. Eng. Chem. 17, 605 (1925).
31. M. Ponchon, Tech. Mod. 13, 20, 55 (1921).
Vol. 1 DISTILLATION 665

32. R. Savarit, Arts Metiers 65, 145, 178, 266, 307 (1922).
33. M. R. Fenske, Ind. Eng. Chem. 24, 482 (1932).
34. A. J. V. Underwood, Chem. Eng. Progr. 44, 603 (1948).
35. J. R. Fair and W. L. Bolles, Chem. Eng. 75(9), 156 (Apr. 22, 1968).
36. G. G. Brown and co-workers, Unit Operations, John Wiley & Sons, Inc., New York,
1950.
37. E. R. Gilliland, Ind. Eng. Chem. 32, 918 (1940).
38. J. H. Erbar and R. N. Maddox, Petrol. Ref. 40(5), 183 (1961).
39. H. E. Eduljee, Hydrocarbon Proc. 54(9), 120 (1975).
40. E. W. Thiele and R. L. Geddes, Ind. Eng. Chem. 25, 290 (1933).
41. C. D. Holland, Fundamentals of Multicomponent Distillation, McGraw-Hill Book Co.,
Inc., New York, 1981.
42. R. N. S. Rathore, K. A. Van Wormer, and G. J. Powers, AIChE J. 20, 491 (1974).
43. D. C. Freshwater and B. D. Henry, Chem. Eng. (London) (301), 533 (1975).
44. J. E. Hendry, D. F. Rudd, and J. D. Seader, AIChE J. 19, 1 (1973).
45. W. L. Bolles, in B. D. Smith, ed., Design of Equilibrium Stage Processes, McGraw-Hill
Book Co., Inc., New York, 1963, Chapt. 14.
46. M. H. Hutchinson and R. F. Baddour, Chem. Eng. Progr. 52(12), 503 (1956).
47. F. Kastanek, M. V. Huml, and V. Braun, I. Chem. E. Symp. Ser. No. 32, 5(100),
(1969).
48. W. V. Delnicki and J. L. Wagner, Chem. Eng. Progr. 52(1), 28 (1956).
49. J. R. Fair, Chapt. 5 in Handbook of Separation Process Technology, R. W. Rousseau,
ed., John Wiley, New York, 1987.
50. J. R. Fair, in R. H. Perry and D. Green, eds., Perry’s Chemical Engineers’ Handbook,
7th ed., McGraw-Hill Book Co., Inc., New York, 1997, section 14.
51. Bubble-Tray Design Manual, American Institute of Chemical Engineers (AIChE),
New York, 1958.
52. H. Chan and J. R. Fair, Ind. Eng. Chem. Proc. Des. Devel. 23, 814, 820 (1984).
53. M. J. Lockett, Distillation Tray Fundamentals, Cambridge University Press,
Cambridge, Mass., 1986.
54. M. M. Dribika and M. W. Biddulph, Trans. I. Chem. E. 70, Part A, 142 (1992).
55. M. Prado and J. R. Fair, Ind. Eng. Chem. Res. 29, 1031 (1990).
56. K. E. Porter, M. J. Lockett and C. T. LimTrans. I. Chem. E. 50, 91 (1972).
57. W. V. Pincezewski, N. D. Benke, and C. J. D. Fell, AIChE J. 21, 1210 (1975).
58. K. E. Porter, A. Safekouri, and M. J. Lockett, Trans. I. Chem. E. 51, 265 (1973).
59. M. Prado, K. L. Johnson, and J. R. Fair, Chem. Eng. Progr. 83(3), 32 (1987).
60. J. A. Garcia and J. R. Fair, Ind. Eng. Chem. Res. 39, 1809, 1818 (2000).
61. H. E. O’Connell, Trans. AIChE 42, 741 (1946).
62. R. Krishna, H. F. MartinezR. Sreedhar, and G. L. Standart, Trans. I. Chem. E. 55,
178 (1977).
63. R. Taylor and R. Krishna Multicomponent Mass Transfer, John Wiley, New York,
1993.
64. A. Sperandio, M. Richard, and M. Huber, Chem.-Ing.-Tech. 37, 22 (1965).
65. W. D. Stoecker and B. Weinstein, Chem. Eng. Progr. 73(11), 71 (1977).
66. R. F. Strigle, Packed Tower Design and Applications, Gulf Publishing, Houston, Tex.,
1994.
67. J. R. Fair and J. L. Bravo, Chem. Eng. Progr. 86(1), 19 (1990).
68. P. J. Hoek, J. A. Wesselingh, and F. J. Zuiderweg, Chem. Eng. Res. Des. 64, 431
(1986).
69. R. E. Treybal, Mass Transfer Operations, 3rd ed., McGraw-Hill., Inc., New York,
1980.
70. H. Z. Kister, Distilation—Design, McGraw-Hill, Inc., New York, 1992.
666 DISTILLATION Vol. 1

71. J. Stichlmair and J. R. Fair Distilation—Principles and Practices, Wiley-VCH,


New York, 1998.
72. P. A. Schweitzer, ed., Handbook of Separation Techniques for Chemical Engineers,
3rd ed., McGraw-Hill Book Co., Inc., New York, 1997,
73. R. W. Rousseau, ed., Handbook of Separation Process Technology, John Wiley & Sons,
Inc., New York, 1987, Chapt. 5.
74. T. K. Sherwood, G. H. Shipley, and F. A. L. Holloway, Ind. Eng. Chem. 30, 765 (1938).
75. M. Leva, Chem. Eng. Progr. Symp. Ser. No. 10 50, 51 (1954).
76. W. L. Bolles and J. R. Fair, Chem. Eng. 89(14), 109 (July 12, 1982).
77. J. L. Bravo and J. R. Fair, Ind. Eng. Chem. Proc. Des. Dev. 21, 162 (1982).
78. R. Krishnamurthy and R. Taylor, AIChE J. 31, 449, 456 (1985).
79. J. L. Bravo, J. A. Rocha, and J. R. Fair, Hydrocarbon Proc. 64(1), 91 (1985).
80. J. A. Rocha, J. L. Bravo, and J. R. Fair, Ind. Eng. Chem. Res. 32, 641 (1993).
81. J. A. Rocha, J. L. Bravo, and J. R. Fair, Ind. Eng. Chem. Res. 35, 1660 (1996).
82. Z. Olujic, Chem. Biochem. Eng. 11, 31 (1997).
83. J. R. Fair, A. F. Seibert, M. Behrens, P. P. Saraber, and Z. Olujic, Ind. Eng. Chem.
Res. 39, 1788 (2000).
84. J. Stichlmair, J. L. Bravo and J. R. Fair, Gas Sepn. Purif. 3, 19 (1989).
85. J. Drew and M. Propst, eds., Tall Oil, Pulp Chemicals Association, New York, 1981.
86. K. C. D. Hickman, in R. H. Perry and C. H. Chilton, eds., Chemical Engineers’ Hand-
book, 5th ed., McGraw-Hill Book Co., Inc., New York, 1973, section 13.
87. J. R. Fair, in Y. A. Liu, H. A. McGee, and W. R. Epperly, eds., Recent Developments
in Chemical Process and Plant Design, John Wiley & Sons, Inc., New York, 1987,
Chapt. 3.
88. W. C. Petterson and T. A. Wells, Chem. Eng. 84(20), 79 (1977).
89. H. Z. Kister and I. D. Doig, Hydrocarbon Proc. 56(7), 132 (1977).
90. J. G. Kunesh, H. Z. Kister, M. J. Lockett and J. R. Fair, Chem. Eng. Prog. 91(10), 43
(1995).
91. M. S. Ray, Chemical Engineering Bibliography, 1967–1988, Noyes Publications, Park
Ridge, N.J., 1990.
92. M. S. Ray, Sepn. Sci. Technol. 34, 3305 (1999). Also, Sepn. Sci. Technol. 34, 139
(1999); 32, 3067 (1997); 32, 1163 (1997).
93. M. Peters and K. D. Timmerhaus, Plant Design and Economics for Chemical
Engineers, 4th ed., McGraw-Hill Book Co., Inc., New York, 1991.
94. H. Z. Kister, Distillation–Operation, McGraw-Hill., Inc., New York, 1990.
95. A. E. Nisenfeld and R. C. Seeman, Distillation Columns, Instrument Society of
America, Research Triangle Park, N.C., 1981.
96. F. G. Shinskey, Distillation Control, 2nd ed., McGraw-Hill Book Co., Inc., New York,
1984.
97. P. B. Deshpande, Distillation Dynamics and Control, Instrument Society of America,
Research Triangle Park, N.C., 1984.
98. P. S. Buckley, W. L. Luyben, and J. P. Shunta, Design of Distillation Control Systems,
Instrument Society of America, Research Triangle Park, N.C., 1985.

JAMES R. FAIR
The University of Texas at Austin
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 667

DISTILLATION,
AZEOTROPIC AND
EXTRACTIVE
1. Introduction

Distillation (qv) is the most widely used separation technique in the chemical
and petroleum industries. Not all liquid mixtures are amenable to ordinary frac-
tional distillation, however. Close-boiling and low relative volatility mixtures are
difficult and often uneconomical to distill, and azeotropic mixtures are impossible
to separate by ordinary distillation. Yet such mixtures are quite common (1,2)
and many industrial processes depend on efficient methods for their separation
(see also SEPARATIONS PROCESS SYNTHESIS). This article describes several special
distillation techniques for economically separating low relative volatility and
azeotropic mixtures.
Vapor–liquid phase equilibrium (VLE) in a c-component mixture can be
represented by

Z P
ViL dP
yi fVi P ¼ xi gi Psat
i fi
sat
exp for i ¼ 1; :::; c ð1Þ
Psat
i
RT

where yi is the mole fraction of component i in the vapor phase; fiV is the vapor-
phase fugacity coefficient of component i; P is the total system pressure; xi is the
mole fraction of component i in the liquid phase; gi is the liquid-phase activity
coefficient of component i; Pisat is the vapor pressure of component i; fisat is
the saturation fugacity coefficient of component i; and the exponential term is
the Poynting correction factor (Poyi). At low to moderate pressures, the value
of the term fisat/fi*Poyi is typically close enough to unity that it can be ignored
and equation 1 reduces to

yi P ¼ xi gi Psat
i for i ¼ 1; :::; c ð2Þ

For simplicity, this is the VLE equation that will be used throughout the rest of
this article.
The activity coefficient (gi) is a measure of the liquid-phase nonideality of a
mixture and its value varies with both temperature and composition. When gi ¼ 1
the liquid phase is said to form an ideal solution and equation 2 reduces to
Raoult’s law. Nonideal mixtures (gi 6¼ 1) can exhibit either positive (gi > 1) or
negative (gi < 1) deviations from Raoult’s law. Positive deviations are more com-
mon and occur when the molecules of the different compounds in the solution are
dissimilar and have no preferential interactions between them. Negative devia-
tions occur when there are preferential attractive forces (hydrogen bonds, etc.)
between the molecules of the different species that do not occur in the absence
of the other species (3). If these deviations are large enough, the pressure-
composition (P-x,y) and temperature-composition (T-x,y) phase diagrams exhibit
a minimum or maximum point (Fig. 1). At these minima and maxima the liquid

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.0409192004150805.a01.pub2
668 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

Fig. 1. P-x,y and T-x,y phase diagrams showing binary, homogeneous, maximum, and
minimum azeotropes. (a) Chloroform (1)-tetrahydrofuran (2) at 308C; (b) chloroform (1)-
tetrahydrofuran (2) at 101 kPa; (c) ethanol (1)-toluene (2) at 658C; and (d) ethanol (1)-to-
luene (2) at 101 kPa (from Ref. 6). To convert kPa to atm, multiply by 9.87 10 3.

phase and its equilibrium vapor phase have the same composition, ie,

y i ¼ xi for i ¼ 1; :::; c ð3Þ

the mixture boils at constant temperature, and the dew-point (vapor) and bubble-
point (liquid) curves are tangent with zero slope. These are the defining condi-
tions for a homogeneous azeotrope where a single liquid phase is in equilibrium
with a vapor phase. Note that a maximum boiling azeotrope (Fig. 1b) is equi-
valent to a minimum pressure azeotrope (Fig. 1a) and a minimum boiling azeo-
trope (Fig. 1d) is also a maximum pressure azeotrope (Fig. 1c). Minimum boiling
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 669

azeotropes occur in mixtures with positive deviations from Raoult’s law and max-
imum boiling azeotropes occur in mixtures with negative deviations. The vast
majority of the known azeotropes are minimum boiling (see 1,2). Combining
equations 2 and 3 gives an alternative criterion for a homogeneous azeotrope,
in this case, a binary azeotrope:

g2 Psat
¼ 1 ð4Þ
g1 Psat
2

Equation 4 illustrates that, (1) only small deviations from Raoult’s law are neces-
sary for an azeotrope to exist in a close boiling mixture (ie, compounds with simi-
lar vapor pressures) and (2) the larger the difference in boiling points of the
compounds, the more nonideal the mixture must be to form an azeotrope and,
hence, the less likely that an azeotrope will exist. Generally, mixtures of com-
pounds with boiling points more than 308C apart do not exhibit azeotropes
(4). Exceptions to this heuristic do exist, however. For example, hydrogen chlor-
ide and water form a maximum boiling azeotrope and their normal boiling points
differ by 1858C. Martin (5) showed that a binary mixture forms a minimum boil-
ing azeotrope when the infinite dilution activity coefficient of the less volatile
component is greater than the ratio of the pure component vapor pressures, ie,

Psat
g1
2 >
1
ð5Þ
Psat
2

Partially miscible mixtures are more nonideal than completely miscible mixtures
and thus are more likely to form azeotropes. Usually, these will be heterogeneous
azeotropes where two (or more) liquid phases are in equilibrium with a vapor
phase, the mixture boils at constant temperature and at constant composition,
but the composition of the equilibrium phases are all different (see Fig. 17b).
The overall liquid-phase composition is, however, identical to the vapor-phase
composition. Since positive deviations from Raoult’s law are necessary for liquid
phase immiscibility, only minimum boiling heterogeneous azeotropes can exist.
When the immiscibility exists only over a limited composition range, the azeo-
trope can lie outside of the two-liquid-phase regions, giving rise to a homoge-
neous azeotrope. The methyl acetate–water and methyl ethyl ketone–water
mixtures are two such examples.
Separation by distillation depends on the vapor and liquid phases having
different compositions when a liquid (vapor) mixture is partially vaporized (con-
densed). The vapor phase becomes enriched in the more volatile components and
depleted in the less volatile components while the opposite occurs in the equili-
brium liquid phase. By successively repeating these partial vaporizations and
condensations in a countercurrent cascade, it is often possible to achieve the
desired degree of separation. A common measure of the degree of enrichment
or the ease of separation is the relative volatility defined as:

yi xj gi Psat
i
aij ¼ ¼ ð6Þ
xi yj gj Psat
j
670 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

The relative volatility of most mixtures changes with temperature, pressure, and
composition. The larger the value of aij, the easier it is to separate component i
from component j. From equation 3, at a c-component (ie, binary, ternary, etc.)
homogeneous azeotrope, yi ¼ xi for all c components in the mixture. Therefore,
aij ¼ 1 for all components i and j and no enrichment of the vapor takes place dur-
ing a partial vaporization. Thus, homogeneous azeotropes cannot be separated by
ordinary fractional distillation. Similarly, any mixture, be it ideal, nonideal,
close-boiling, etc. where the relative volatilities are close to unity will be difficult
to separate by ordinary distillation because little enrichment occurs with each
partial vaporization step. Although the actual minimum relative volatility neces-
sary for an economical separation is a function of a number of variables (most
notably, the required product purities, but also the value of the products and
the cost of the materials of construction and the heating and cooling mediums,
etc.), alternatives to ordinary fractional distillation should generally start to be
investigated when aij is less than 1.15–1.2.
Most methods for distilling azeotropic and low relative volatility mixtures
rely on the addition of specially chosen chemicals or ‘‘mass separating agents’’
to facilitate the separation. These separating agents can be divided into distinct
classes that define the principal distillation techniques used to separate such
mixtures. The five categories of distillation-based methods for separating low
relative volatility and/or azeotropic mixtures are (1) extractive distillation and
homogeneous azeotropic distillation where the liquid separating agent is comple-
tely miscible. In extractive distillation, the separating agents, variously known
as solvents, extractive agents, extractants, and sometimes entrainers, are high
boiling, non-azeotrope-forming, liquids that alter the relative volatility of the
mixture to be separated. (2) Heterogeneous azeotropic distillation or, more com-
monly, azeotropic distillation, where the liquid separating agent, called the
entrainer, forms one or more azeotropes with the other components in the mix-
ture and causes two liquid phases to exist over a broad range of compositions.
This immiscibility is the key to making the distillation sequence work. (3) Salt-
effect distillation, which is a variation of extractive distillation, where the separ-
ating agent is an ionic salt. The salt dissociates in the liquid mixture and alters
the relative volatilities to make the separation possible. (4) Pressure-swing or
pressure-sensitive distillation where a series of distillation columns operating
at different pressures is used to separate mixtures containing azeotropes
whose composition changes appreciably over a moderate pressure range.
(5) Reactive distillation where the separating agent reacts preferentially and
reversibly with one of the components in the mixture. The reaction product is
then distilled from the nonreacting components and the reaction is reversed to
recover the initial component. An example is the separation of isobutene from
butanes by etherification (7). This last method is not discussed in this article. All
five of these techniques can be used in batch as well as continuous distillations.
Distillation can also be combined with other separation methods such
as liquid – liquid extraction, absorption, adsorption/molecular sieves, melt
crystallization (qv), or membrane-permeation-based methods like pervapora-
tion (8 – 10) to separate azeotropic mixtures. Ordinary fractional distillation
is used for the separation over the composition range where the relative
volatility is appreciable and the separation is easy. One of the other
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 671

separation techniques, whose efficacy does not depend on relative volatility,


is then used to achieve the separation over the ‘‘pinched’’ (low relative
volatility) or azeotropic composition range where distillation would be diffi-
cult or impossible.

2. Residue Curve Maps

The most basic form of distillation, called simple distillation, is a process in


which a multicomponent liquid mixture is slowly boiled in an open pot and the
vapors are continuously removed as they form. At any instant in time the vapor
is in equilibrium with the liquid remaining in the still. Because the vapor is
always richer in the more volatile components than the liquid, the liquid compo-
sition changes continuously with time, becoming more and more concentrated in
the less volatile species. A simple distillation residue curve is a graph showing
how the composition of the liquid residue in the pot changes over time. A residue
curve map is a collection of residue curves originating from different initial com-
positions. Residue curve maps contain the same information as phase diagrams,
but present it in a way that is more useful for understanding how to create a dis-
tillation sequence to separate a mixture. Simple distillation residue curves repre-
sent the liquid composition profiles of packed distillation columns operating at
total (infinite) reflux. Some authors (11,12) prefer to use distillation lines,
which represent liquid composition profiles in trayed distillation columns operat-
ing at total reflux, instead of residue curves. For practical purposes, there is little
difference between the two types of curves and either can be used for distillation
sequence synthesis. (Residue curves and distillation lines have identical fixed
points, ie, the pure components and azeotropes in a mixture, and, except for
regions with high curvature, the two lines typically lie very close to each
other.) Residue curves are used here.
Residue curves can only originate from, terminate at, or be deflected by the
pure components and azeotropes in a mixture. Pure components and azeotropes
that residue curves move away from are called unstable nodes (UN), those where
residue curves terminate are called stable nodes (SN), and those that deflect
residue curves are called saddles (S).
The simplest residue curve map for a ternary mixture is shown in Fig. 2. All
ternary nonazeotropic mixtures, including ideal and constant relative volatility
mixtures, are qualitatively represented by this map. All of the residue curves ori-
ginate at the light (lowest boiling) pure component, move toward the intermedi-
ate boiling component, and end at the heavy (highest boiling) pure component.
Residue curves point in the direction of increasing temperature and must always
move in such a way that the mixture boiling temperature continuously increases
along every curve. (Distillation lines are typically given the opposite orientation.
That is, the arrows on a distillation line point in the direction of decreasing
boiling temperature. Consequently, the discussion in the text has to be reversed
if distillation lines are used.) From this property and the direction of the arrows
in Fig. 2, the light component is an unstable node; the intermediate component,
which deflects the residue curves, is a saddle; and the heavy component is a
stable node.
672 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

Intermediate

x2

Heavy x1 Light

Fig. 2. Residue curve map for a ternary nonazeotropic mixture.

Many different residue curve maps are possible when azeotropes are pre-
sent (see 13–15). For example, there are six possible residue curve maps for tern-
ary mixtures containing a single azeotrope. These six maps differ by the binary
pair forming the azeotrope and by whether the azeotrope is minimum or maxi-
mum boiling. Figure 3 represents the case where the intermediate and heaviest
components (A and B) form a minimum boiling binary azeotrope. Pure compo-
nent D is an unstable node, pure components A and B are stable nodes, the mini-
mum boiling binary azeotrope C is a saddle, and the boiling point order from low

x2

B C A
x1

Fig. 3. Residue curve map for a ternary mixture with a distillation boundary running
from pure component D to the binary azeotrope C.
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 673

to high is D!C!A or B. The residue curve connecting component D to the


azeotrope C has the special property that it divides the composition triangle
into two separate distillation regions. Any initial mixture charged to the
still-pot of the simple distillation apparatus with a composition lying to
the left of the curve D C will result in the last drop of liquid being pure
B; whereas any initial mixture with a composition lying to the right of the
curve D C will yield pure A as the last drop of liquid. Residue curves like
D C that divide the composition space into different distillation regions are
called simple distillation boundaries, infinite reflux boundaries, separatrices,
or, most commonly, distillation boundaries. The residue curve map for a binary
mixture is the one-dimensional line connecting the two pure components and
an azeotropic point is a boundary in the composition space that limits how far
a distillation can proceed. (Consider, for example, the binary mixture of A and
B, which forms the minimum-boiling azeotrope C, represented by the bottom
edge of the triangle in Fig. 3). In a ternary mixture, the composition space is a
two-dimensional triangle and a distillation boundary is a one-dimensional
curve that similarly restricts how far a distillation can proceed. There must
be a saddle on at least one end of every distillation boundary and each distilla-
tion region must contain a stable node, an unstable node, and at least one
saddle. As seen in Fig. 3, different distillation regions can have some saddles
and nodes in common.
Residue curve maps would be of limited usefulness if they could only be gen-
erated experimentally. Fortunately that is not the case. The simple distillation
process can be described (16) by the set of equations:

dxi
¼ xi y i for i ¼ 1; :::; c ð7Þ
dx

where xi and yi are the liquid and vapor mole fractions, respectively, of compo-
nent i, x is a nonlinear time scale that is related to the fraction of liquid remain-
ing in the pot at any instant in time, and c is the number of components in the
mixture. Note that the pure components and azeotropes in the mixture corre-
spond to the steady-states or singular points of equation 7 (where the derivative
goes to zero) and they are all either nodes or saddles. A detailed mathematical
treatment of simple distillation residue curve maps and their properties can be
found in the literature (16,17). Given a method for calculating the VLE for the
mixture of interest, equation 7 can be numerically integrated forward and back-
ward in time from a number of different initial compositions to generate a resi-
due curve map. An alternative method for sketching residue curve maps that
only requires knowledge of the boiling points of the pure components and azeo-
tropes in the mixture has also been published (18–21). This latter method is well
suited to mixtures lacking detailed VLE data and thus can be useful in prelimin-
ary separation feasibility studies.
Even though the simple distillation process is not a practical method
for separating mixtures, simple distillation residue curve maps are extremely
useful. They can be used to test the consistency of experimental azeotropic
data (18,19,22); to predict the sometimes nonintuitive order and content of the
cuts from batch distillations of azeotropic mixtures (23–26); to determine
674 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

whether a given mixture is separable by distillation; to help identify feasible


separating agents; to predict the attainable product compositions; and to
synthesize the corresponding distillation sequences (4,13,18,27–36). By helping
to identify the limits of distillation, residue curve maps can also be useful in
synthesizing separation sequences where distillation is combined with other
methods.
Residue curve maps exist for mixtures having any number of components,
though they cannot be visualized when there are more than four components.
Fortunately, many mixtures of industrial importance contain only three or
four key components and can thus be treated as pseudo-ternary or pseudo-
quaternary mixtures. Quaternary residue curve maps (three-dimensional tetra-
hedrons with two-dimensional surfaces as distillation boundaries) are more
complicated than their ternary counterparts, but can still be understood using
just the boiling point temperatures of the pure components and azeotropes
(37). Mathematical (ie, non-graphical) methods for calculating all of the azeo-
tropes, distillation boundaries, and distillation regions in mixtures containing
any number of components are available (38–43).

3. Homogeneous Azeotropic Distillation

The most general definition of homogeneous azeotropic distillation is the distilla-


tion of any mixture containing one or more azeotropes into the desired pure com-
ponent or azeotropic products without exploiting any liquid-phase immiscibilities
that might be present. Thus, homogeneous azeotropic distillation includes the
distillation of azeotropic mixtures in which the desired separation can be
achieved without the addition of a separating agent, azeotropic mixtures to
which a separating agent that may or may not form new azeotropes is added
to facilitate the separation, and non-azeotropic mixtures to which an azeo-
trope-forming separating agent is added to more easily distill one or more of
the compounds away from the others in the original mixture. (When liquid-
phase immiscibility is used to make the distillation equence work, the separation
is classified as a heterogeneous azeotropic distillation. These systems are dis-
cussed later in the article).
The first step in synthesizing a homogeneous azeotropic distillation
sequence is to determine the desired separation objective. For example, some-
times it is desirable to recover all of the constituents in the mixture as pure com-
ponents; other times it is sufficient to recover only some of the components as
pure products. In still other cases, an azeotrope may be the desired product.
Not every objective is attainable and those that are feasible may require different
distillation sequences.
The next step is to sketch or calculate the residue curve map for the mixture
to be separated. This will allow one to determine whether the desired separation
can be achieved and, if so, how to achieve it or, if not, to determine that the
separation objective needs to be changed. (See pp. 223–227 in Ref. 21 for an
instructive example). Note that the decision to add a separating agent to a mix-
ture in order to achieve a separation objective also requires that an effective
method be found for recovering the agent for reuse.
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 675

As mentioned above, simple distillation boundaries are the same as the dis-
tillation boundaries of a distillation column operated at total reflux. While the
two are not equivalent at finite reflux ratios, the simple distillation boundaries
remain a very good approximation to the finite-reflux distillation boundaries,
except in regions where the boundaries are extremely curved. Although the
composition profile of a distillation column operating at finite reflux can cross
a curved simple distillation (infinite reflux) boundary to a limited degree (28–
33,44,45) the resulting distillation sequences typically require large reflux(re-
boil) ratios and/or large recycle flowrates and, thus, are not normally economical.
Mixtures such as nitric acid–water–sulfuric acid, that have extremely curved
boundaries, can be exceptions. A good rule-of-thumb, therefore, is that the com-
positions of the distillate and bottom streams from a distillation column should
lie in the same simple distillation region in order for the separation to be feasible.
Crossing of simple distillation boundaries should only be attempted as a last
resort and only with caution, because the feasibility of such column designs
can be sensitive to small inaccuracies in the VLE model and, once built, the
column may be sensitive to disturbances.
An overall material balance for a continuously operated distillation column
requires that the feed, distillate, and bottoms compositions lie on a straight line
in the composition space (the composition triangle for ternary mixtures). Thus, a
feasible distillation sequence for separating a homogeneous azeotropic mixture
can be identified by determining whether or not the desired products lie in the
same distillation region and then can be synthesized by superimposing material
balance lines onto the distillation residue curve map. When setting the composi-
tions of the distillate and bottom streams leaving each column in the sequence:
(1) the distillate composition must have a lower boiling temperature than the
bottoms composition, however the component with the lowest (highest) boiling
point is not necessarily removed as the distillate (bottoms); (2) pure components
and azeotropes that are nodes on the residue curve map are easier to obtain
as pure products than saddles; and (3) a double-feed column is almost always
required in order for a saddle on the residue curve map to be the product from
a distillation column, eg, extractive distillations. For simplicity, only ‘‘sharp’’
splits, where a pure component or azeotrope is the product from at least one
end of the distillation column, are considered here. (This excludes situations
where a nonsharp or sloppy split is acceptable, such as when a small amount
of impurities are removed in a small distillate or bottoms stream. The complete
range of product compositions that can be reached in a single column for a given
feed can be determined by calculating the ‘‘bow-tie’’-shaped separation region
(4,21,29,31–34,44)).
As an example, consider the residue curve map for the nonazeotropic mix-
ture shown in Fig. 2. It has no distillation boundary so the mixture can be sepa-
rated into pure components by either the direct or indirect sequence (Fig. 4). In
the direct sequence, the unstable node (light component, L) is taken overhead in
the first column. The bottom stream composition is determined by following the
material balance line (the straight line through the distillate and feed composi-
tion) until a distillation boundary or the edge of the composition space (the com-
position triangle here) is reached. Thus, the bottom stream from the first column
is a binary mixture of the intermediate, I, and heavy, H, components (plus a trace
676 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

Fig. 4. Column sequences and material balance lines for the (a) direct and (b) indirect
sequences for separating nonazeotropic mixtures.

amount of the light component). In the binary I–H mixture, I has the lowest boil-
ing temperature (unstable node) so it is recovered as the distillate in the second
column and the stable node, H, becomes the corresponding bottoms stream. The
indirect sequence removes the stable node (heavy component, H) from the bottom
of the first column. The composition of the distillate stream is determined by fol-
lowing the material balance line through the bottom and feed compositions until
a distillation boundary or the edge of the composition space is reached. Thus, the
overhead stream from the first column is a binary mixture of L and I (plus a trace
amount of H). In the second column, the unstable node, L, is taken overhead
and component I, the stable node in the binary mixture, is recovered as the
bottoms. The relative flowrate of any of the streams can be determined by the
well-known lever-arm rule. For example, in Fig. 4b, the flowrate of stream B1
is represented by the length of the line segment from F1 to D1 and the flowrate
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 677

Fig. 5. The acetone–2-propanol–water system where & represents the 2-propanol–


water azeotrope. (a) Residue curve map (46); (b) material balance line and distillation
boundary (---); and (c) the column sequence.

of stream D1 is represented by the length of the line segment from F1 to B1.


In this example the distillate and bottoms flows are roughly equal, but that is
frequently not true.
A second example is the azeotropic mixture of acetone, water, and 2-propa-
nol encountered in the production of acetone by 2-propanol dehydrogenation. The
goal is to recover the acetone product, recycle the unreacted 2-propanol to the
reactor, and discard the wash water. From the residue curve map (Fig. 5a), it
is clear that, unlike the previous example, the three pure components cannot
be recovered because of the distillation boundary running from the acetone ver-
tex to the 2-propanol–water azeotrope, which divides the triangle into two distil-
lation regions. However, by modifying the separation objective to accept
recycling the 2-propanol–water azeotrope to the reactor instead of pure 2-propa-
nol, a feasible separation sequence is possible (Fig. 5b and c). The acetone
(unstable node) is taken overhead in the first column, leaving a 2-propanol–
water mixture to be distilled in the second column. In this binary mixture, the
678 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

azeotrope is an unstable node so it becomes the distillate and the stable node,
water, is removed from the bottom of the second column. If some up-stream mod-
ification were made which reduced the amount of water in the feed to the first
distillation column enough to move the feed composition into the other distilla-
tion region, the desired separation would no longer be possible. Acetone could
still be recovered as the distillate in the first column (assuming there are suffi-
cient stages and reflux), but it would no longer be possible to get pure water from
the base of the second column. Instead, 2-propanol would be come out of the bot-
tom of the second column. The binary azeotrope would continue to be distilled
overhead in the second column. This means there would no longer be a way to
remove pure water from the process.
The overwhelming majority of all ternary mixtures that can potentially
exist are represented by only 125 different residue curve maps (15, see also 13,
14). (The rare mixtures in which a pair of compounds form two binary azeotropes
(47–51) are exceptions.) For each type of separation objective, these 125 maps
can be subdivided into those that can potentially meet the objective (ie, the resi-
due curve maps where the desired pure component and/or azeotropic products lie
in the same distillation region) and those that cannot. Knowing the structure of
the residue curve map for a mixture is thus sufficient to determine if a given
separation objective is feasible. (There is, however, no guarantee that a feasible
separation will be economical).
The seven most favorable residue curve maps for the common task of dis-
tilling a mixture containing a binary minimum-boiling azeotrope into its two
constituent pure components by adding a suitable separating agent are
shown in Fig. 6. (Favorable residue curve maps for separating binary maxi-
mum-boiling azeotropes are given in Ref. 4). Other maps that exploit distilla-
tion boundary curvature, namely those with a low boiling separating agent
that forms a highly curved boundary ending at the azeotrope, have the poten-
tial to meet the stated objective (28 –33). However, as described earlier, distil-
lation sequences based on such maps have a number of potentially serious
drawbacks. Thus, for initial screening purposes, only compounds that yield
one of these seven maps need to be considered as possible separating agents
for separating binary minimum-boiling azeotropes. Five of the candidate
maps require the separating agent to form a maximum-boiling azeotrope
with one of the other components. Since maximum-boiling azeotropes are
much less common than minimum-boiling azeotropes, these five maps will
also be relatively rare. Two examples are adding formic acid to valeric acid–
water mixtures (4) and adding hydrazine to water–aniline mixtures (52).
The map shown in Fig. 6a requires the separating agent to have a boiling
point between the other two compounds, but not form any azeotropes. Such
separating agents will also be fairly uncommon. Because the likelihood that
two compounds will form an azeotrope increases the closer their boiling points
are and because the compounds of the original mixture already boil close
enough to form an azeotrope, many potential intermediate-boiling separating
agents will also form azeotropes with one or both of the original compounds.
Benzene is an example of a feasible intermediate- boiling separating agent for
separating acetone–heptane mixtures (29,30,32). Thus, of the seven favorable
residue curve maps for separating binary minimum-boiling azeotropes into
A A

A/B
azeo

B E
E B
(a)

A A B

A/B
azeo

E
B E
(b)

A A A

E B E B E B
(c)

E E A

A/B
azeo

B
A B A B
A/E azeo
(d)

Fig. 6. The seven most favorable residue curve maps and corresponding column se-
quences for the homogeneous azeotropic distillation of a minimum-boiling binary azeo-
trope using a mass separating agent E to recover the two constituent pure components
A and B (18). The symbol & represents an azeotrope. (a) Case I, where the separating
agent is intermediate boiling and does not introduce any new azeotropes: (b) Case II, ex-
tractive distillation with a heavy solvent that does not introduce any new azeotropes. In
some cases, B will come off the top of the first column instead of A; (c) Case III, where the
separating agent is intermediate boiling and forms a maximum boiling azeotrope with the
lighter of the two pure components (i.e., A). The agent may or may not form a minimum
boiling azeotrope with B, and there may or may not be a minimum boiling ternary azeo-
trope lean in A; and (d) the same column configuration as case III, but the separating
agent is lower boiling.

679
680 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

their two constituent pure components, the map representing extractive


distillation (Fig. 6b) is by far the most common and the most important. Notice
that, in this map, the azeotrope is an unstable node (lowest boiling), the two
desired pure components are saddles, and the high boiling separating agent is
a stable node.
For other separation objectives or when homogeneous azeotropic distilla-
tion is combined with other non-distillation separation methods, additional
residue curve maps can lead to successful separations and need to be consid-
ered. For example, the minimum-boiling homogeneous azeotrope formed by
cyclohexane and benzene can be separated by adding acetone (53), despite
the fact that acetone and cyclohexane form a binary minimum-boiling homoge-
neous azeotrope that introduces a distillation boundary running from the
acetone–cyclohexane azeotrope to the benzene–cyclohexane azeotrope,
which divides the desired pure components into different distillation regions
(Fig. 7a). Acetone (F2) is fed to the first distillation column along with the
benzene–cyclohexane azeotrope (F1) so that the overall feed to the column
(F1 þ F2) lies within the upper distillation region on the residue curve
map. Benzene, the stable node in this distillation region, is recovered from
the bottom of the first distillation column (B1) and a mixture near the
acetone–cyclohexane azeotrope is distilled overhead (D1). (See the material
balance lines on Fig. 7a). Next liquid –liquid extraction with water is used to
separate the acetone–cyclohexane azeotrope (Fig. 7b). Acetone is extracted
into the aqueous phase and nearly pure cyclohexane is recovered as the raffi-
nate. Finally, acetone and water are separated by distillation with the acetone
recycled to the first column and the water recycled to the extractor.
When neither or only one of the components needs to be recovered as a
pure product, separating agents that form azeotropes and introduce distilla-
tion boundaries that divide the pure components into different distillation
regions can sometimes be used without resorting to a nondistillation step.
For example, even though methyl acetate and water form a minimum-boiling
homogeneous azeotrope, water is commonly used in polyvinyl alcohol plants to
separate the minimum-boiling homogeneous azeotrope formed by methyl acet-
ate and methanol (54). Methyl acetate and water are only partially miscible
but, unlike most acetate–water mixtures, the methyl acetate–water azeotrope
lies outside of the immiscible region. The methyl acetate–methanol –water
mixture contains a distillation boundary running from the methyl acetate –
methanol azeotrope to the methyl acetate–water azeotrope, placing pure
methyl acetate and methanol into different distillation regions and making
it impossible to recover both components as pure products (Fig. 8). Notice
that the larger distillation region is bounded by the methyl acetate –methanol
azeotrope (unstable node), pure methanol (saddle), the methyl acetate –water
azeotrope (saddle) and the high boiling separating agent, water (stable node).
In other words, this four-sided distillation region has the same structure
as the extractive distillation residue curve map (Fig. 6b), except that one of
the saddles is an azeotrope. As long as the overall feed to the column lies in
this distillation region, the column sequence will also be identical to that
shown in Fig. 6b. In the first column, a double-feed column with water as
the upper feed, the methyl acetate–water azeotrope is the overhead product.
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 681

Benzene
(80.1°C)
B1

F1
(77.3°C)

F2
(53.1°C)
(80.8°C) D1 (56.1°C)
Cyclohexane Acetone

(a)

H2O
CH H2O makeup
CH = cyclohexane solvent

Liquid-liquid
extraction

D1
Homogeneous
azeotropic CH-acetone
distillation azeotrope

F2
Feed F1
CH Acetone Recycle
Benzene Acetone

Makeup Distillation
Acetone

Recycle
B1 H2O
Benzene

(b)

Fig. 7. Hybrid process for separating the benzene-cyclohexane azeotrope using acetone
as the separating agent. (a) The residue curve map for the acetone–benzene–cyclohexane
mixture at 101 kPa with superimposed feed (----) and column ( ) material balance lines
for the homogeneous azeotropic distillation step. Azeotropes are denoted by &. (b) The
complete separation sequence (53).

(A wet methyl acetate stream is acceptable here because the methyl acetate is
sent to a hydrolysis reactor where more water is added). A mixture of metha-
nol and water leaves the bottom of the first column and becomes the feed to the
second column. Methanol is recovered as the distillate in the second column.
682 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

Methanol
(64.5°C)

(53.5°C)

(100.0°C) (56.4°C) (56.9°C)


Water Methyl acetate

Fig. 8. Residue curve map for the methyl acetate–methanol–water mixture at 101 kPa,
where & represents an azeotrope. The immiscible region is not shown.

Water leaves the bottom of the second column and is recycled to the first
column, completing the sequence. This process and others with separating
agents that give rise to similarly structured four-sided distillation regions
are thus a variation of extractive distillation where only one of the azeotro-
pic-constituents is recovered as a pure product. Another example is using
methyl ethyl ketone (MEK) to separate the acetone–methanol azeotrope
(21,55). Methanol and MEK form a minimum boiling azeotrope and the tern-
ary mixture contains a distillation boundary running between the two azeo-
tropes, which puts acetone and methanol into different distillation regions,
ie, the residue curve map is analogous to Fig. 8. This extractive distillation
sequence distills the methanol–MEK azeotrope overhead in the first (double-
feed) column while acetone is recovered from the top of the second column.
Successful separating agents with residue curve maps similar to the one
shown in Fig. 8 will have the same attributes as those described later for
extractive distillation solvents, except for the prohibition against azeotrope
formation.

4. Extractive Distillation

Extractive distillation is defined as distillation in the presence of a miscible,


higher boiling, liquid separating agent, commonly called the solvent, which
does not form azeotropes with the other components in the mixture (27). Extrac-
tive distillation is widely used in the chemical and petrochemical industries for
separating azeotropic, close-boiling, and other low relative volatility mixtures,
including those whose phase diagram shows the presence of a severely ‘‘pinched’’
(low relative volatility) region over a limited composition range (ie, a tangent
pinch).
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 683

Extractive distillation works because the solvent is specially chosen to


interact differently with the components of the original mixture, thereby altering
their relative volatilities. Because these interactions occur predominantly in the
liquid phase, the solvent is continuously added near the top of the extractive dis-
tillation column so that it is present at an appreciable concentration in the liquid
phase on all of the trays below. The mixture to be separated is added through a
second feed point further down the column (see Fig. 6b). In the extractive col-
umn, the component with the higher volatility in the presence of the solvent
(not necessarily the component with the lowest boiling point) is distilled over-
head as a relatively pure distillate. The other component leaves with the solvent
via the column bottoms. The solvent is then separated from the remaining com-
ponents, typically by distillation in a second column, and recycled back to the
first column. The product leaving the top of the solvent recovery column is rela-
tively free of both the compound distilled overhead in the extractive column and
the solvent.
Many examples of extractive distillation can be found in the literature.
A selected few are included here to give the reader an idea of the breadth of
possible applications. Extractive distillation has been widely used since the
1940s to separate butanes, butenes, and butadiene. (Butadiene and n-butane
form an azeotrope while the various butanes and butenes have low rela-
tive volatilities). Industrially used solvents have included furfural (56,57),
acetone, acetonitrile (58), dimethylformamide (59), dimethylacetamide (60),
N-methylpyrrolidone (61), and dimethylsulfoxide (62). Other extractive distil-
lation examples include: distilling nonaromatics away from aromatics using
phenol (63), sulfones (64), N-methylpyrrolidone (65), or dimethylformamide
(66); separating toluene from methylcyclohexane using phenol (67); remov-
ing close-boiling heptane isomers from cyclohexane, an important raw mate-
rial for manufacturing nylon intermediates (68); separating the phenol–
cyclohexanone azeotrope using adipic acid diester (69); separating the
cumene –phenol azeotrope using trisubstituted phosphates (70); separating
the low relative volatility mixture of propylene and propane using acrylonitrile
(71); the Ryan-Holmes process for separating the ethane– carbon dioxide azeo-
trope that arises when carbon dioxide is used for enhanced oil-field recovery
(72); separating the azeotropes formed by alcohols and their esters using aro-
matic hydrocarbons (73); separating the methanol–methylene bromide azeo-
trope using ethylene bromide (74); dehydrating ethanol using ethylene glycol
(75 –77); separating acetone and methanol using water (78); separating the
pyridine–water azeotrope using bisphenol (79); dehydrating tetrahydrofuran
using monopropylene glycol (80) or 1,4-butanediol (81); separating close-
boiling vinyl acetate and ethyl acetate using phenol (82); separating mixtures
of dimethyl ether and chloromethane, obtained by reacting methanol and
HCl, using water (83) or methanol (84) as the extractant; purifying crude
propylene oxide reaction mixtures using propylene glycol (85); and purifying
a variety of fluorocarbons used as refrigerants, blowing agents, and aerosol
propellants (86 –91).
Extractive distillations can be divided into three general categories: (1) the
separation of minimum-boiling azeotropes, (2) the separation of maximum-
boiling azeotropes, and (3) the separation of low relative volatility nonazeotropic
684 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

mixtures. For separating binary mixtures into pure components, each category
is represented by a single, unique residue curve map.
4.1. Minimum Boiling Azeotropes. Every extractive distillation that
separates a binary minimum boiling azeotropic mixture into its pure components
has a residue curve map and column sequence like the one shown in Fig. 6b.
Typical tray-by-tray composition profiles are shown in Fig. 9. Notice that the

1.00
EtOH-H2O Ethylene
0.90 Feed point glycol
feed pt.
0.80
EtOH
0.70
Liquid mole fraction

H2O
EG
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0 2 4 6 8 10 12 14 16 18 20 22
Reboiler Theoretical stage number Condenser
(a)

Water
1.0

0.8

0.6
x2
0.4

FL
0.2
B FU F D

0
0 0.2 0.4 0.6 0.8 1.0
Ethylene glycol x1 Ethanol

(b)

Fig. 9. Extractive distillation column profiles for separating ethanol (EtOH) and water
(H2O) using ethylene glycol (EG) as the extractive agent. (a) Liquid compositions versus
theoretical stage number where (-- -^-- -) represents ethanol, (-- -.-- -) water, and (-- ~ --)
ethylene glycol; and (b) liquid composition profiles in mole fraction coordinates where (*)
represents the feeds, (.) is the distillate, (X) is the bottoms; and (.) and (*) are the tray
compositions. (From Example 5.4 in Ref. 21.)
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 685

concentration of the solvent remains high throughout the extractive section of


the column, ie, the section between the two feed points (Fig. 9a). This is a char-
acteristic of all extractive distillations and is due to the need for the solvent to be
present in sufficient quantity on the trays to alter the relative volatilities of the
components being separated. Most of the separation of the components that form
the azeotrope takes place here in the extractive section. The purpose of the strip-
ping section (below the lower feed) is to strip out the remaining ‘‘light’’ compo-
nent(s) to complete the separation of the key components begun in the
extractive section. The rectifying section (above the upper or solvent feed
point) separates the high-boiling solvent from the ‘‘light’’ component(s) so the sol-
vent won’t contaminate the distillate.
When distilling ideal mixtures, the component with the lowest boiling point
is always the one recovered as the distillate. That is not always the case in
extractive distillation. Neither can the design engineer freely pick which azeotro-
pic component to recover in the distillate, despite the apparent symmetry of the
residue curve map (Fig. 6b). For a given solvent, one and only one component can
be recovered overhead in the extractive column and it need not be the pure com-
ponent with the lowest boiling point. For example, the extractive distillation of
ethanol and water using gasoline (76), some phenols (92), cyclic ketones, or cyclic
alcohols (93) causes water to distill overhead in the extractive column while the
lower boiling ethanol leaves in the bottom stream with the solvent. Similarly,
ketones distill the higher boiling methanol overhead from methanol– acetone
mixtures (55,94) and furfural reverses the natural volatility difference of
butane–butadiene mixtures (56). This phenomenon where the intermediate
boiling pure component distills overhead in the extractive column is entirely
due to the way in which the relative volatilities of the components in the mixture
are altered by the liquid-phase interactions between the solvent and the other
components.
One way to identify which component will distill overhead with a given
solvent is to draw pseudo-binary y– x phase diagrams for the mixture to be
separated. A pseudo-binary phase diagram is one in which the VLE data for
the azeotropic constituents (components 1 and 2) are plotted on a solvent-
free basis. When no solvent is present, the pseudo-binary y– x diagram is the
true binary y–x diagram (Fig. 10a). At the azeotrope, where the VLE curve
crosses the 458 line, a12 ¼ 1.0. To determine which component will be the dis-
tillate, a series of pseudo-binary y –x plots are drawn at increasing solvent
concentrations until the pseudo-azeotrope, the point where the solvent-free
VLE curve crosses the 458 line, ie, where a12 ¼ 1.0, disappears into one of the
pure component corners. The resulting pseudo-binary phase diagram will
either resemble Fig. 10b where the solvent increases the volatility of
component 1 relative to component 2, making component 1 the distillate, or
Fig. 10c, where the solvent has the opposite effect, making component 2 the
distillate. For example, aniline increases the volatility of cyclohexane (nbp ¼
80.88C) relative to benzene (nbp ¼ 80.18C) (Fig. 10d). Thus, with aniline as
the extractive agent, the higher boiling cyclohexane will distill overhead in
an extractive distillation. Pseudo- binary y –x phase diagrams can also be
used to determine which component will be the distillate in extractive distilla-
tions of non-azeotropic mixtures.
686 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

Fig. 10. Pseudo-binary (solvent-free) y–x phase diagrams for determining which compo-
nent will be the distillate. (a) No solvent present; (b) and (c) with sufficient solvent pre-
sent to eliminate the pseudo-azeotrope with component 1 and component 2, respectively,
becoming the distillate (96); and (d) experimental VLE data for cyclohexane–benzene
where A, B, C, and D represent 0, 30, 50, and 90 mol% aniline, respectively (97).

Some authors (95) prefer to use ‘‘isovolatility’’ curves to determine which


component will become the distillate in an extractive distillation. For a mixture
of components 1 and 2 and an extractive distillation solvent S, an isovolatility
curve is simply the locus of points where a12 ¼ 1.0 in the ternary composition
space, ie, with different amounts of solvent present. The isovolatility curve starts
at the binary azeotrope and ends on one of the faces of the composition triangle.
If it ends on the 1-S face, then component 1 will distill overhead. Likewise, com-
ponent 2 will be the distillate if the isovolatility curve ends on the 2-S face.
Pseudo-binary y–x phase diagrams and isovolatility curves are simply different
ways of visualizing the same information. When plotted on a solvent-free basis,
points along an isovolatility curve become the pseudo-azeotropic points on a
series of pseudo-binary phase diagrams.
Extractive distillations exhibit unexpected behavior at large reflux ratios.
In ordinary distillations, the number of stages required to achieve a given
separation decreases monotonically as the reflux ratio increases. Equivalently,
in an operating distillation column where the number of stages is fixed, the
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 687

distillate purity monotonically increases as the reflux ratio increases. Conse-


quently, it is very common to increase the reflux when higher purity products
are desired. In extractive distillations, however, the behavior is more complex.
As the reflux ratio increases from its minimum value, the number of theoretical
stages necessary to achieve a given separation initially decreases—the same
behavior observed with ordinary distillations. Eventually, however, a minimum
in the number of stages required is reached and further increases in the reflux
ratio require an increasing number of stages (essentially all in the extractive sec-
tion) in order to continue to achieve the same separation. Fig. 11a illustrates this

45

40
Number of stages

35

30

25

20

15
0 5 10 15 20 25 30
Reflux ratio
(a)

1.00

0.99

0.98

0.97

0.96

0.95
0 5 10 15 20 25 30
Reflux ratio
(b)

Fig. 11. Ethanol–water separation using ethylene glycol as the solvent (from Example
5.4 in Ref. 21). (a) Number of theoretical stages in the extractive column as a function of
reflux ratio. (b) Distillate mole fraction of ethanol as a function of reflux ratio.
688 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

behavior in the ethanol–water–ethylene glycol separation from Fig. 9. For azeo-


tropic mixtures, beyond a certain large reflux ratio, called the maximum reflux
ratio, an infinite number of stages would be required to achieve the same separa-
tion. During the operation of an extractive distillation column, two different
types of behavior can also be observed, depending on where the column is oper-
ating along its distillate purity vs. reflux ratio curve (Fig. 11b). At lower reflux
ratios, the column behaves normally. Increasing the reflux increases the product
purity. At larger reflux ratios, however, the inverse behavior occurs. Increases in
reflux ratio result in decreased product purities. As might be imagined, standard
distillation control systems will have trouble controlling an extractive distillation
column operating in this inverse response region or, worse still, in the vicinity of
the inflection point where the response to reflux ratio changes switches from the
expected to the inverse. Thus, an important step in the design of an extractive
distillation sequence is to locate where the inverse behavior begins for the chosen
solvent and to design the column to operate well away from this point. The range
of reflux ratios over which the normal response occurs increases with the solvent-
to-feed ratio and, in most cases, the inverse response region is easily avoided.
This unexpected increasing stage requirement or decreasing product purity
with increasing reflux ratios is caused by the dilution of the solvent concentra-
tion in the column (which makes the separation more difficult) as more and more
of the solventless reflux is fed back into the column.
4.2. Design Optimization. Optimization is an important yet often neg-
lected step in the design of extractive distillation sequences. The cost of the solvent
recovery step affects the optimization and thus must also be included. Optimization
not only yields the most efficient extractive distillation design, but is also necessary
for meaningful comparisons with other separation sequences and methods.
When several simple heuristics are used, the optimization procedure
usually reduces to a single-variable optimization of the feed ratio, ie, the ratio
of the solvent to process feed flow rates, which has the greatest effect on the
sequence cost (77,98). The simple heuristics are for (1) setting the purity of the
solvent recycle stream midway between unity and the minimum required purity,
calculated by assuming that all of the intermediate volatility component in the
bottom of the solvent recovery column (water in the ethanol–water–ethylene
glycol system) ends up in the distillate of the extractive column (see Eqs. 3–6
of Ref. 77); (2) setting the temperature of the solvent feed 5–158C below the boil-
ing temperature of the distillate leaving the extractive column; (3) setting the
fractional recovery of the component recovered at the top of the extractive col-
umn to between 99.5% and 99.99%; and (4) setting the reflux ratios in all of
the columns to 1.2–1.5 times the minimum. Minimum reflux ratios for extractive
distillations can be calculated using the methods presented in Refs. 99 and 100 to
solve the equations given by Refs. 101 and 102.
Because homogeneous azeotropic mixtures cannot be separated into pure
component products by isobaric distillation without a solvent, but can be sepa-
rated in the presence of a sufficient amount of a suitable solvent, there is clearly
some minimum amount of solvent that just makes the separation possible. This
minimum solvent flow, or minimum solvent-to-feed ratio, depends on the solvent
used and the effectiveness with which it alters the relative volatility of the
components being separated. A solvent yielding a smaller minimum feed ratio
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 689

45

40

Number of stages
35

30

25

20

15
0 5 10 15 20 25 30
Feed ratio

Fig. 12. Number of theoretical stages in the extractive column as a function of the
solvent-to-feed ratio for the extractive distillation of ethanol and water using ethylene
glycol as the solvent (29).

is a better solvent and should result in a lower cost design than a solvent with a
larger minimum feed ratio. The solvent composition required to just make the
pseudo-azeotrope disappear on a pseudo-binary y–x phase diagram, ie, to
‘‘break’’ the azeotrope, can be taken as a qualitative measure of the minimum sol-
vent flow. A simple method for estimating minimum solvent flows has been pub-
lished (95) as has an exact quantitative method (96,98). As a rough rule-of-
thumb, the economically optimal feed ratio is often between two and four
times the minimum feed ratio (96,98). Many industrial extractive distillations
operate with feed ratios between 1 and 5, though we know of commercial units
operating with feed ratios as low as 0.15 and as high as 20. Figure 12 shows how
the number of stages in the extractive column varies with feed ratio for the etha-
nol–water separation shown in Fig. 9. The relationship between stages required
and feed ratio is the same for all extractive distillations of azeotropic mixtures.
The number of stages required is infinite at the minimum feed ratio, rapidly
drops at feed ratios slightly above the minimum, and then continues to decrease
at a slowing rate at higher feed ratios until, eventually, an asymptotic region is
reached where further increases in solvent flow has little or no effect on the num-
ber of stages required. Sometimes the number of stages required in the extrac-
tive column will start to increase again at sufficiently high feed ratios. Fig. 13
shows how the feed ratio influences the total annualized cost (TAC) [defined in
Ref. 103] of an extractive distillation sequence for dehydrating ethanol using
ethylene glycol. Cost diagrams for all extractive distillations have this same dis-
tinctive shape. The cost is very high near the minimum feed ratio, but then
rapidly decreases to a minimum before starting to increase again at higher
feed ratios, as the costs associated with the increasing solvent recirculation
rate (increased heating and cooling loads and larger diameters in both
columns), begin to exceed the cost savings from the reduction in the number of
stages in the extractive column.
690 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

3.90

3.80

A
3.70

3.60
TAC, $ × 106/yr
3.50

1.70

1.60

1.50

1.40 Change in reboiler


steam temperature
1.30

1.20
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Feed ratio

Fig. 13. Extractive distillation sequence cost as a function of the solvent-to-feed ratio for
producing anhydrous ethanol from azeotropic ethanol using ethylene glycol at reflux
ratios of D ¼ 1.15 rmin, * ¼ 1.2 rmin, and & ¼ 1.3 rmin (77). Point A represents another pub-
lished design for the same mixture (75).

4.3. Maximum Boiling Azeotropes. Maximum boiling azeotropes are


far less common than minimum boiling azeotropes and so are successful extrac-
tive distillations of maximum boiling azeotropes using high boiling solvents.
Adding a high boiling solvent to a mixture containing a maximum boiling azeo-
trope causes a distillation boundary running from the maximum boiling
azeotrope to the heavy solvent (Fig. 14), which divides the desired pure compo-
nents into different distillation regions. The only way a high boiling solvent can
yield an economically viable means for separating a maximum boiling azeotrope
is if the resulting distillation boundary is extremely curved. Classic examples
include using sulfuric acid to concentrate aqueous nitric acid (104–106) or hydro-
chloric acid (107) mixtures. N-formylmorpholine has also been proposed for
breaking the maximum-boiling formic acid–water azeotrope (108). Figure 15
shows that the distillation boundary in the nitric acid–water–sulfuric acid mix-
ture is indeed highly curved. Applying the lever arm rule to the material balance
lines superimposed on the residue curve map in Fig. 15a shows that the internal
recycle flow rates (B1 and B2) in this process are relatively high compared with
the product flow rates (D1 and D2).
4.4. Nonazeotropic Mixtures. Extractive distillation is widely used in
the petrochemical industry to separate close-boiling but non-azeotropic mixtures.
Common examples include the purification of aromatics, the separation of olefins
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 691

Maximum-boiling
azeotrope

A
Heavy
Solvent

Fig. 14. Residue curve map for separating a maximum boiling azeotrope using a high
boiling solvent where (- - - -) represents the distillation boundary and & the azeotrope.

and paraffins, diolefins and olefins, paraffins and napthalenes, styrenes and
aromatics, and a number of other hydrocarbon mixtures (109).
The extractive distillation of any close-boiling or other low relative volati-
lity, but non-azeotropic, mixture using a high boiling solvent is represented
by the residue curve map shown in Figure 2. Although this map is different

HNO3
D1 D1

F M Air

HNO3
F

Steam
D2
M Distillation
Air H 2O
boundary
B 1—
D2
B2
H2SO4 B2 B1 H 2O H2SO4–rich

(a) (b)

Fig. 15. Extractive distillation of nitric acid, HNO3, and water, H2O, using sulfuric acid,
H2SO4, as the solvent. (a) Residue curve map and material balance lines where & repre-
sents the maximum boiling azeotrope and (b) the column sequence (11).
692 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

than the one for the extractive distillation of minimum boiling azeotropes, the
distillation sequence is identical (see Fig. 6b) and the process works for the
same reason. The solvent alters the relative volatilities of the components to
be separated via liquid-phase interactions. Depending on the nature of these
interactions, either the low boiling or the intermediate boiling pure component
will be distilled overhead in the extractive column. Pseudo-binary y–x phase dia-
grams can again be used to determine which component will distill overhead in
the extractive column. When the composition of the lower boiling component is
plotted and increasing solvent concentrations cause the pseudo-binary VLE
curve to lie above the true binary VLE curve, then this solvent accentuates the
natural volatility difference and will cause the lower boiling component to distill
overhead. If the pseudo-binary VLE curve lies below the true binary VLE curve,
then this solvent reverses the natural volatility difference of the mixture and, at
sufficiently high solvent concentrations, the intermediate boiling component will
distill overhead.
Because there is no azeotrope, these mixtures could in theory be separated
without adding a solvent, though it would be a difficult and expensive separation.
Consequently, there is no minimum feed ratio (minimum solvent flow). Curves
relating the number of stages in the extractive column to the feed ratio (eg,
Fig. 12) end at a large but finite number of stages when the solvent flow rate
goes to zero. By using the heuristics provided earlier, optimization of extractive
distillation sequences for separating non-azeotropic mixtures usually reduces to
a one-variable optimization of the feed ratio. These extractive distillations can
also exhibit inverse response at sufficiently large reflux ratios, due to dilution
of the solvent concentration in the column. (See p. 447 of Ref. 110 for additional
information.)
References 109 and 112 provide additional information on the extractive
distillation of close-boiling, non-azeotropic mixtures and the use of mixed-solvent
systems.
4.5. Solvent Selection. The most important step in developing a
successful (ie, economical) extractive distillation sequence is identifying an
effective solvent for the separation. The solvent selection procedure is typi-
cally carried out in several stages. In the first step, simple qualitative methods
are used to identify general classes of compounds or functional groups that
may make effective solvents for a given separation. In the next step, individual
compounds are evaluated, typically by experimental methods, and then
ranked to identify the most promising candidates. In the final step, detailed
vapor– liquid equilibrium measurements are made for several of the top
candidates, the separation is simulated using chemical process simulation
software and/or tested in a lab-scale or mini-works column, and then the
final solvent selection is made. The final selection process should include an
economic comparison of the optimal separation sequence with each of the
final solvent candidates, because differences in the solvent recovery costs
can influence the decision. The remaining discussion focuses on the first two
stages.
As described earlier, extractive distillation works because the presence
of the solvent alters the relative volatilities of the components being separa-
ted. Consequently, an important measure of the effectiveness of an extractive
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 693

distillation solvent is its selectivity, Sij,

sat sat
ðaij Þs ðgi Pi =gj Pj Þs ðgi =gj Þs
Sij ¼ ¼ ffi ð8Þ
aij gi Psat
i =gj Pj
sat gi =gj

which is simply the ratio of the relative volatilities of the key components with
and without the solvent present. Solvents with selectivities farthest from unity
are generally preferred. Since interactions with the solvent predominantly occur
in the liquid phase—affecting the ratio of the liquid-phase activity coefficients
(gi/gj)—and the ratio of the pure component vapor pressures is normally not
significantly affected by changes in the boiling temperature caused by the pre-
sence of the solvent, equation 8 is often expressed as the ratio of the key compo-
nent activity coefficients with and without the solvent present. Alternatively,
since the denominator in equation 8 is the same for all solvents, the solvent
selectivity is sometimes written as just the ratio of the key component activity
coefficients in the presence of the solvent. While it is possible to find solvents
that increase or decrease the ratio of activity coefficients, it is usually preferable
to select a solvent that accentuates the natural difference in vapor pressures
between the components to be separated. In other words, a solvent that increa-
ses gi relative to gj when Psat i > Pj
sat
(Sij > 1) is favored over one that increases
gj relative to gi (Sij < 1). In the latter case, adding small amounts of the solvent
actually makes the separation more difficult, and relatively large quantities
may be required to completely overcome the natural volatility difference and
enhance the separability of the original mixture (105). Such solvents will cause
the intermediate boiling pure component to distill overhead in the extractive col-
umn. A heuristic, therefore, is to favor solvents that cause the more volatile
(lower boiling) component to distill overhead (ie, solvents with selectivities
greater than 1.0).
To force the naturally more volatile component i overhead, the solvent
should either behave essentially ideally with component j and cause positive
deviations from Raoult’s law for component i (gj1 and gi > 1), or behave essen-
tially ideally with component i and cause negative deviations from Raoult’s law
for component j (gi  1 and gj < 1). Compounds of similar type and size, eg,
pentane–hexane or methanol–ethanol, tend to behave ideally in the liquid
phase and thus have activity coefficients close to unity. Dissimilar molecules
tend to repel each other, causing positive deviations from Raoult’s law and, in
the extreme, resulting in liquid-phase immiscibilities. Compounds that tend to
associate in the liquid phase exhibit negative deviations from Raoult’s law.
Because systems showing positive deviations are more common, the usual
approach is to force the lower boiling component overhead by selecting a solvent
which is chemically similar to the higher boiling species and dissimilar to the
lower boiling species.
Homologues, polarity, and hydrogen-bonding tendencies can be used for a
crude screening of candidate solvents by qualitatively predicting the types of
deviations from ideality that can be expected. When the compounds being sepa-
rated are chemically dissimilar, suitable solvents may be found among the higher
694 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

boiling homologues of either compound, though homologues of the less volatile


compound are generally favored, as they enhance the natural volatility differ-
ence. The molecular similarity of two members of the same chemical family
causes them to form relatively ideal solutions, while the molecular dissimilarity
of the homologue and the other compound tend to cause positive deviations. For
example, higher boiling alcohols could be used to force acetone overhead from the
acetone–methanol azeotrope while the higher boiling ketones would force
methanol overhead (94). Choosing the lowest boiling homologue that does not
azeotrope with the other components (94) or the homologue that just barely
remains miscible (111) have both been recommended.
Mixtures of liquids with similar polarity typically form nearly ideal solu-
tions while liquids with different polarities exhibit positive deviations from
ideality. The degree of deviation from ideality is roughly proportional to the dif-
ference in polarity. Table 1 gives a qualitative ranking of the polarity of various
functional groups. Note that steric hindrance diminishes the polarity of a com-
pound. Other more quantitative measures of a molecule’s polarity include the
dipole moment, the dielectric constant (112), and the solubility parameter
(3,4,113). Dipole moment interactions in hydrocarbon systems are discussed in
Ref. 113. In general, adding a more polar solvent increase the volatility of the
least polar compound in the original mixture. The reverse is true when a less
polar solvent is added. As an example, consider the azeotropic mixture of acetone
and methanol. Methanol is more polar than acetone (Table 1) so, by adding an
even more polar compound like water, the relative volatility of acetone to metha-
nol can be increased enough to ‘‘break’’ the azeotrope. If, on the other hand, a
nonpolar compound like an alkane were added, the volatility of methanol relative
to acetone would increase (105). Since acetone has a lower boiling point than
methanol, the heuristic to favor solvents that enhance the natural volatility dif-
ference tells us that a polar solvent like water is preferred.

Table 1. Relative Polarities of Functional Groupsa


most polar water
organic acids
amines
polyols
alcohols
esters
ketones
aldehydes
ethers
aromatics
olefins
least polar paraffins
Effect of branching
most polar normal
secondary
least polar tertiary
a
From Ref. 3.
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 695

Carlson and Stewart’s (114) guidelines for choosing potential extractive dis-
tillation solvents based on polarity arguments are given below. Be aware, how-
ever, that exceptions do occur. (1) If the mixture to be separated is highly polar
(like acetic acid and water), choose a polar solvent or a nonpolar solvent (like a
hydrocarbon) that is more soluble with the higher boiling component. (2) For
nonpolar mixtures (like cyclohexane–benzene) choose a polar solvent (like ani-
line) that is more soluble with the higher boiling component. (3) For a mixture
where the more volatile component is polar and the less volatile component is
nonpolar, choose a nonpolar or a polar solvent. Be aware that polar solvents
may reverse the relative volatility. (4) For mixtures where the more volatile com-
ponent is nonpolar and the other component is polar, choose a polar solvent.
(5) For mixtures of compounds that are moderately polar, choose a polar solvent
or a nonpolar solvent that is more soluble with the higher boiling component.
For cases where either a polar or a nonpolar solvent is recommended, polar com-
pounds are often preferred because they typically cause the solution to be more
nonideal.
Strong deviations from ideality are often associated with hydrogen bonding
between molecules and many successful extractive distillation solvents are com-
pounds capable of forming strong hydrogen bonds such as phenols, aromatic
amines like aniline, higher alcohols, glycols, etc. (115). Several authors have
published tables based on the hydrogen-bonding tendency of different classes
of compounds and functional groups for qualitatively predicting the type of
deviation likely to occur in a mixture (34,115–118). Table 2 is representative.
To distill the more volatile component overhead, use Table 2 to select classes
of compounds that cause positive deviations with the more volatile component
or negative deviations with the less volatile component.
Once several potentially promising families of compounds have been iden-
tified using the above qualitative methods, individual solvent candidates need to
be identified and ranked. As a starting point, elminate all compounds with boil-
ing points lower than the compounds to be separated. This ensures that,
throughout the column, the solvent will be predominantly present in the liquid
phase where it can alter the activity coefficients of the mixture being separated.
The likelihood that a potential solvent will form azeotropes with any of the mix-
ture components can be minimized by selecting compounds that boil 30–508C or
more above the mixture. On the other hand, the solvent should not boil so high
that excessive temperatures or high vacuum are required in the solvent recovery
column.
Group contribution methods such as modified UNIFAC (Dortmund)
(119,120) can be used to predict activity coefficients in mixtures and thus provide
a nonexperimental method for estimating solvent selectivities. Bear in mind,
however, that all group contribution methods have weaknesses and they some-
times give, not only quantitatively inaccurate, but also qualitatively incorrect
results. COSMO–RS (121,122) is a promising, though relatively new, method
for a priori predicting activity coefficients and other thermophysical properties
based on unimolecular quantum chemical calculations. It has the advantage of
often being applicable to those problems where group contribution methods
fail. Currently, however, on mixtures for which they are well-suited, group con-
tribution methods give more accurate predictions than COSMO-RS (123).
696 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

Table 2. Predicted Deviations from Raoult’s Law for Different Functional Groups Based
on Hydrogen-Bonding Interactionsa
Solvent classb
Solute
Class 1 2 3 4 5 6 7 8 9 10 11 12
H-Donor Groups
1 phenol 0 0 0 þ þ
2 acid, thiol 0 0 0 0 0 0 0 þ þ
3 alcohol, water 0 þ þ 0 þ þ þ þ
4 active H on multihalogen paraffin 0 0 þ 0 0 þ

H-Acceptor Groups
5 ketone, amide with no H on N, þ 0 þ þ þ þ þ þ þ
sulfone, phosphine oxide
6 tertiary amine 0 þ 0 þ þ 0 þ 0 0
7 secondary amine 0 þ þ 0 0 0 0 0 þ
8 primary amine, ammonia, amide with 2H 0 þ þ 0 0 þ þ þ þ
on N
9 ether, oxide, sulfoxide 0 þ þ 0 0 þ 0 þ 0 þ
10 ester, aldehyde, carbonate, phosphate, 0 þ þ þ 0 þ þ 0 þ þ
nitrate, nitrite, nitrile, intramolecular
bonding, e.g.,
o-nitrophenol
11 aromatic, olefin, halogen aromatic, multi- þ þ þ 0 þ 0 0 þ 0 þ 0 0
halogen paraffin without active H,
monohalogen paraffin
Non-H-Bonding Groups
12 paraffin, carbon disulfide þ þ þ þ þ 0 þ þ þ þ 0 0
a
From Reference 115.
b
Negative deviation. þ Positive deviation. 0 No deviation.

The infinite-dilution selectivity provides a convenient quantitative method


for comparing the effectiveness of different solvents. Since activity coefficients
are composition dependent, a fair comparison of the candidate solvents requires
that their selectivities be determined at a consistent solvent concentration and
key component ratio. Most commonly, the activity coefficient of each of the key
components is determined at infinite dilution in each of the potential solvents.
The infinite-dilution selectivity, Sij1, is then simply the ratio of these infinite-
dilution activity coefficients.

g1
S1
ij ¼
i
ð9Þ
g1
j

In most systems, the selectivity steadily, and often almost linearly, increases
with solvent concentration so that the solvent with the largest infinite-dilution
selectivity usually also has the largest selectivity at the lower solvent concen-
trations used in extractive distillation. In other words, solvents with the largest
infinite-dilution selectivities are generally preferred. Be aware, however, that
some systems exhibit a maximum in selectivity at finite solvent concentrations
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 697

(eg, n-butane and butene in furfural) while in other systems, typically those that
approach the point of immiscibility at high solvent concentrations, the selectivity
increases rapidly (ie, nonlinearly) with solvent concentration (eg, 2,4-dimethyl
pentane–benzene–aniline) (110,124). For such mixtures, the infinite-dilution
selectivity is not an accurate indicator of the solvent’s effectiveness in an extrac-
tive distillation. Fortunately, these systems are relatively rare.
Gas–liquid chromatography (GLC) is one of the most common experimental
methods used for screening potential extractive distillation solvents and/or for
determining infinite-dilution activity coefficients (125–127). The selectivity is
determined by measuring the retention time of each solute on an inert chromato-
graphic column that has been preloaded with enough of the candidate solvent to
approximate infinite dilution conditions. Other methods for measuring infinite-
dilution activity coefficients include ebulliometry (127,128), the dilutor technique
(129), and the dew point method (130). The dilutor technique is the only one that
can determine infinite-dilution activity coefficients in solvent mixtures.
Potential extractive distillation solvents can also be evaluated by experi-
mentally measuring the selectivity or relative volatility of a fixed composition
mixture of the components to be separated (often 50% each) in the presence of
a constant amount of each candidate solvent (114). Solvent-to-feed ratios of 1:1
to 3:1 are typically used. The preferred experimental apparatus is a modified
Othmer still operated to yield one theoretical stage of separation (114,131). As
before, the objective is to find the candidate solvents that cause the largest
increase in relative volatility (ie, the highest selectivities).
While good selectivity at reasonable solvent concentrations is essential, it is
not the only important property when selecting an extractive distillation solvent.
A solvent’s capacity (ie, its ability to solubilize the components in the mixture
being separated) is also important. If a solvent causes a second liquid phase to
form during the distillation, much of the selectivity enhancement can be lost.
Since higher selectivities are by achieved by increasing the nonideality between
the solvent and one of the key components, and mixtures with large nonidealities
often have limited miscibilities, solvents with the highest selectivities can have
lower capacities. Thus, a compromise between high selectivity and high capacity
sometimes has to be made. Another alternative is to use a mixture of solvents
where one of the solvents is chosen for its high selectivity and the other is chosen
to maintain high capacity (109,112). The solvent which provides the high selec-
tivity is usually the predominant component in the solvent mixture. Because of
its limited solubility with many compounds, the reverse is true when water is
used as the highly selective solvent. For example, aromatics are separated
from aliphatics using N-methyl-2-pyrrolidone (NMP) with a small amount of
water added to increase the selectivity.
In addition to having a high selectivity at reasonable concentrations and
a high capacity, desirable extractive distillation solvents should: (1) not form
azeotropes with the components in the mixture to be separated—this can almost
be guaranteed by picking solvents with boiling points 30–508C or more above
the other compounds, (2) be easily separated from the other components at
reasonable temperatures and pressures, (3) be thermally stable at the tempe-
ratures encountered in the distillation columns, (4) be nonreactive with the
other compounds, (5) be nontoxic, (6) inexpensive, (7) readily available (perhaps
698 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

a compound already in the process), (8) noncorrosive to commonly used materials


of construction, and (9) have a low latent heat of vaporization (114). In practice,
some compromise in solvent properties is almost always required.

5. Salt-Effect Distillation

Salt-effect distillation (also called salt extractive distillation or salt recti-


fication) is another form of extractive distillation, where, rather than using
high boiling liquids to alter the relative volatility of the mixture being sepa-
rated, the separating agent is a soluble, nonvolatile ionic salt or mixture of
salts. Because the added salt is completely nonvolatile, no stages are needed
above the salt feed point to keep it out of the overhead product. The salt is nor-
mally dissolved in the reflux stream and added at the top of the column so that
it is present on all of the trays of the extractive column. The salt is removed
from the bottom of the column along with the less volatile component(s) in
the feed mixture and sent to the salt recovery step where evaporation and/or
drying is used to recover the salt so it can be recycled to the top of the extrac-
tive column.
The principal advantage of an ionic salt over a liquid separating agent is
that the salt ions typically cause much larger changes in relative volatility
than do liquid separating agents (132). Consequently, much less salt is typically
needed—sometimes as little as a few percent as compared with 50–90% of the
liquid phase, as can be the case with some liquid separating agents (133).
This, in turn, leads to shorter, smaller diameter columns and/or higher purity
products, and noticeably smaller energy requirements (132,133).
Salt-effect distillation also has several significant disadvantages compared
with traditional extractive distillation. Foremost is the relatively limited number
of systems for which there is a known salt that is both soluble in the components
being separated and effective at altering their relative volatility. The solubility
constraints are the most limiting, as it can be difficult to find a salt that is ade-
quately soluble in both components throughout the range of compositions
encountered in the extractive column. The search for suitable salts is also hin-
dered by the complexities of the liquid-phase interactions which make it difficult
to predict or model vapor–liquid equilibrium in such systems without experi-
mental data. It is also more difficult to transport and accurately meter solids
than liquids and it can be difficult to achieve rapid dissolution of the salt after
it has been added. For these reasons, there have been only a limited number
of industrial applications of salt-effect distillation (133–140). Reference 135
demonstrates that, as with liquid separating agents, certain salts will reverse
the natural volatility difference of a mixture, causing the higher boiling com-
pound to distill overhead.
Mixed-agent extractive distillations in which one separating agent is a
liquid and one a salt has found some interest and shown some promise in
China (141). Such processes combine the advantages of extractive distillation
(ie, the ease of recovering, transporting and mixing the separating agent)
with those of salt-effect distillation (ie, the more pronounced effect on relative
volatility so less separating agent, less energy, and smaller equipment are
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 699

needed). At least two industrial plants are mentioned in which ethylene glycol–
potassium acetate mixtures are used to dehydrate ethanol or tert-butanol.
See the articles by Furter (132,133,142–144) for additional information on
salt-effect distillation.

6. Pressure-Swing Distillation

It is well known that varying the system pressure can affect the composition of
an azeotrope. When the change in azeotropic composition is appreciable over a
moderate pressure range, this effect can be exploited to separate both minimum-
and maximum-boiling binary azeotropes using a two-column sequence with the
columns operated at different pressures. This process, called pressure-swing or
pressure-sensitive distillation, is illustrated by Fig. 16. For a binary mixture
forming a pressure-sensitive, minimum-boiling azeotrope, the fresh feed, F, is
mixed with the recycled stream from the second column to form the feed stream,
F1, to the first column, which operates at pressure P1. (Alternatively, these two
streams can be fed to separate points in the first column). Because F1 lies to the
right of the azeotrope at pressure P1 (Fig. 16a), pure component A is removed as
the bottom product, B1, and a mixture near the azeotropic composition at pres-
sure P1 is the distillate, D1. Stream D1 is changed to pressure P2 and fed to the
next column as stream F2. Because F2 now lies to the left of the azeotropic com-
position at pressure P2 (Fig. 16a), the other pure component, B, can be recovered
in the bottom stream, B2, and a near azeotropic mixture becomes the distillate,
D2, which is recycled to the first column. The greater the shift in azeotropic com-
position, the smaller the recycle flow rate will be relative to the product flow
rates, and thus, the smaller the column diameters and the more economical
the process will be. An analogous procedure is used for binary maximum boiling
azeotropes (110), except the pure component products are recovered as distillates
and the near azeotropic streams come out of the bottom of each column.

B2
Pressure P1 Pressure P2
P2
D1
F2 D2 D2

T
B1 F F1 F2
P1 1 2
D1

B1 B2
Pure B F F1 Pure A
Composition Pure A Pure B

(a) (b)

Fig. 16. Pressure-swing distillation of a minimum boiling binary azeotrope. (a) Tem-
perature–composition phase diagram showing the effect of pressure on the azeotropic
composition; (b) column sequence (145).
700 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

Systems amenable to pressure-swing distillation include: tetrahydrofuran


(THF) and water (146,147), acetonitrile and water (148), a variety of alcohol–
ketone azeotropes, including the methanol–acetone azeotrope, which is known
to disappear at both low and high pressure (149), methanol and methyl ethyl
ketone (150), the alcohol–acetate azeotropes arising from transesterification
reactions (151), ethanol and water (76,152), and the maximum-boiling hydrogen
chloride–water azeotrope (153,154). Additional pressure-sensitive azeotropes
can be found in Refs. 1,2,145 and 150. Only a fraction of the known azeotropes,
however, are sufficiently pressure-sensitive for pressure-swing distillation to be
economical. References 96 and 150 provide methods for estimating the effect of
pressure on azeotropic composition.
Pressure-swing distillation can also be applied to multicomponent mixtures
containing distillation boundaries as long as at least one end of the boundary ter-
minates at an azeotrope whose composition changes with pressure. Now when
the pressure is varied, the position of the distillation boundary moves along
with the pressure-sensitive azeotrope to which it is attached. This enables the
distillation boundary to be circumvented by using a multi-pressure distillation
sequence analogous to the one in Fig. 16b [see Fig. 4 in Ref. 145]. Reference
155 includes a four-component example. Similarly, an azeotrope whose composi-
tion is insensitive to pressure can be separated into pure components by adding a
separating agent that forms a pressure-sensitive azeotrope with one of the con-
stituents of the original mixture and introduces a pressure-sensitive distillation
boundary (145). Like their binary counterparts, these ternary pressure-swing
distillation sequences are prone to having large recycle flowrates, requiring lar-
ger column diameters, unless the movement of the distillation boundary is appre-
ciable over a moderate pressure range.

7. Heterogeneous Azeotropic Distillation

Heterogeneous azeotropic distillation, or simply azeotropic distillation, is widely


used for separating nonideal mixtures. The technique uses minimum boiling
azeotropes and liquid–liquid immiscibilities in combination to overcome the
effect of other azeotropes or tangent pinches in the mixture that would otherwise
prevent the desired separation. The azeotropes and liquid heterogeneities that
are used to make the desired separation feasible may either be induced by the
addition of a separating agent, usually called the entrainer, or they may be
intrinsically present, in which case the mixture is sometimes called self-
entrained. The most common case is the former; it includes such classic
separations as ethanol dehydration using either benzene, cyclohexane, toluene,
heptane, ethyl ether, etc, as the entrainer, and acetic acid recovery from water
using either 1-propyl acetate, isopropyl acetate, 1-butyl acetate, or isobutyl acet-
ate as the entrainer. In ethanol dehydration the entrainer is used to break the
homogeneous minimum boiling azeotrope between ethanol and water; in the
acetic acid recovery process the entrainer is used to overcome the tangent
pinch between acetic acid and water.
The first successful application of heterogeneous azeotropic distillation
was in 1902 (156) and involved using benzene to produce absolute alcohol from
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 701

a binary mixture of ethanol and water. This batch process was patented in 1903
(157) and later converted to a continuous process (158). Good reviews of the early
development and widespread application of continuous azeotropic distillation in
the pre-World War II chemical industry are available (159).
Historically azeotropic distillation processes were developed on an indivi-
dual basis using experimentation to guide the design. The use of residue curve
maps as a vehicle to explain the behavior of entire sequences of heterogeneous
azeotropic distillation columns, as well as the individual columns that make
up the sequence, provides a unifying framework for design. This process can
be applied rapidly, and produces an excellent starting point for detailed simula-
tions and experiments.
7.1. Phase Diagrams. For binary mixtures, it is well known that when
a liquid–liquid envelope merges with a minimum boiling vapor–liquid phase
envelope the resulting azeotropic phase diagram has the form shown in
Fig. 17. When the liquid composition (x1) is equal to x1AZ, as in Fig. 17a, then
the vapor composition, y1, is also equal to x1AZ and the mixture boils at constant
temperature and at constant (and equal) composition in each phase. Thus a
homogeneous azeotrope is formed. When the overall liquid composition
x10 ¼ (x10)AZ, as in Fig. 17b, then y1 is also equal to (x10)AZ and again the mixture
boils at constant temperature and at constant composition in each phase. How-
ever, the liquid of composition (x10)AZ splits into two liquid phases, L1 and L2, so
that there are three coexisting equilibrium phases (two liquids and one vapor)
that have different compositions. That is, a heterogeneous azeotrope is formed.
Homogeneous and heterogeneous azeotropes share the common property that
the overall liquid composition is equal to the vapor composition, which provides
a means for identifying azeotropes experimentally and computationally.
The properties of ternary heterogeneous vapor–liquid–liquid equilibrium
(VLLE) phase diagrams are important for understanding azeotropic distillation.
The simplest VLLE phase diagram, where a liquid–liquid envelope merges with

T T

V
V
T AZ — L
L
AZ
T —
L1 L2

L–L L–L

0 x1AZ 1 0 (x01)AZ 1

(a) (b)

Fig. 17. Schematic isobaric phase diagrams for binary azeotropic mixtures (AZ). (a)
Homogeneous azeotrope; (b) heterogeneous azeotrope.
702 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

T T

Vapor line

Heterogeneous
B C B C liquid boiling
surface

Critical line Critical line

Projection
A A of vapor
Heterogeneous line
liquid boiling
envelope

(a) (b)

Fig. 18. Schematic isobaric phase diagrams for ternary (A,B,C) azeotropic mixtures. (a)
Homogeneous liquid phase at all boiling points; (b) heterogeneous liquid phase for some
boiling points.

a VLE surface containing a single minimum boiling binary azeotrope, is shown in


Fig. 18. The region, where the LLE and VLE surfaces merge is called the hetero-
geneous liquid boiling surface, Fig. 18b. Note that the boiling temperature varies
over this surface, so the surface is not flat. When the overall liquid composition
lies inside the heterogeneous boiling envelope, and the temperature lies on the
heterogeneous boiling surface, then the liquid boils and splits into two equili-
brium liquid phases (connected by a tie-line) and one coexisting vapor phase.
The Gibbs phase rule requires that the locus of all the equilibrium vapor compo-
sitions forms a curve in T-y space (called the vapor-line) and not a surface, as
happens in the homogeneous region.
A convenient way of representing the T-x-y phase diagram (Fig. 18b) is by
projection onto the composition triangle at the base of the figure. It is understood
that the temperature varies from point to point on the projected vapor line and
on the projected boiling envelope. The latter looks like an isothermal liquid–
liquid binodal envelope, but is not. Each tie line across the boiling envelope is
associated with a different boiling temperature (Fig. 19).
Little experimental VLLE phase data for ternary mixtures has been pub-
lished (see Table 8.1 in 21) and thus good model parameters are not available
for most VLLE mixtures. Fig. 19a shows a model predicted phase diagram for
the ethanol–water–benzene mixture at 101.3-kPa (1 atm) pressure (160). In
this diagram, the liquids and vapors in equilibrium with each other are signified
by a common number. For example, the coexisting liquids on tie-line number 2
are in equilibrium with vapor number 2 at a boiling temperature of 68.558C. The
ethanol–water–benzene mixture forms a minimum boiling heterogeneous
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 703

Ethanol
1.00

0.80

0.60
x2
A
0.40 8
6
5 7
6
4
0.20
3 2
2
1
1
0 0.20 0.40 0.60 0.80 1.00
x1
Water Benzene
(a)

Ethanol
(78.3°C)
1.00

78.17 C

0.80

0.60
x2
6
68.10°C
0.40

5
4
0.20 3
2
1

0 0.20 0.40 0.60 0.80 1.00


x1
Water Benzene
(100°C) (80.3°C)
(b)

Fig. 19. Isobaric VLLE phase diagrams for the ethanol–water–benzene system at 101.3
kPa; (&-&) represent liquid–liquid tie-lines; (~—~), the vapor line; &, homogeneous azeo-
tropes; ~, heterogeneous azeotropes; * ~ , Horsley’s azeotropes. (a) Calculated, where A is

the end point of the vapor line and the numbers correspond to boiling temperatures in 8C
of 1, 70.50; 2, 68.55; 3, 67.46; 4, 66.88; 5, 66.59; 6, 66.46; 7, 66.47, and 8, the critical point,
66.48. (b) Experimental, where the numbers correspond to boiling temperatures in 8C of 1,
67.0; 2, 65.5; 3, 65.0; 4, 64.85; 5, 64.9; and 6, the critical point, 69.9.
704 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

ternary azeotrope. Experimental VLLE data for this mixture are shown in
Fig. 19b (160). Ternary heterogeneous saddle azeotropes are also possible, eg,
formic acid–water–m-xylene (161), water–acetone–chloroform (162), however,
maximum boiling heterogeneous azeotropes cannot exist (163). This differs
from homogeneous azeotropes where all three types are found in Nature.
The typical phase equilibrium problem encountered in distillation is to cal-
culate the boiling temperature and the vapor composition in equilibrium with a
liquid phase of specified composition at a given pressure. If the liquid separates
into two liquid phases, then the problem is to calculate the boiling temperature
and the compositions of the two equilibrium liquid phases plus the coexisting
vapor phase at the specified overall liquid composition. Robust and practical
numerical methods have been devised for solving this problem (164–171).
Thus, using these techniques and a nonideal solution model that is capable
of predicting multiple liquid phases, it is possible to produce phase diagrams
comparable to those of Fig. 19. These predictions are not, however, always quan-
titatively accurate (160,172–175).
7.2. Residue Curve Maps. Residue curve maps are useful for repre-
senting the infinite reflux behavior of continuous distillation columns and for
getting quick estimates of the feasibility of carrying out a desired separation.
In a heterogeneous simple distillation process, a multicomponent partially mis-
cible liquid mixture is vaporized in a still and the vapor that is boiled off is trea-
ted as being in phase equilibrium with the coexisting liquid phases. The vapor is
withdrawn from the still as distillate. The changing liquid composition is most
conveniently described by following the trajectory (or residue curve) of the over-
all composition of all the coexisting liquid phases (176). An extensive amount of
valuable experimental data for the water–acetone–chloroform mixture, includ-
ing binary and ternary LLE, VLE, and VLLE data, and both simple distillation
and batch distillation residue curves are available (162,177). Experimentally
determined simple distillation residue curves have also been reported for the het-
erogeneous system water–formic acid–1,2-dichloroethane (178).
Using the new generation of VLLE computation techniques, it is possible to
calculate residue curve maps for heterogeneous liquid systems (Fig. 20). In this
map, the heterogeneous liquid boiling envelope has been superimposed on the
residue curves in order to distinguish the homogeneous and heterogeneous
regions of the triangular diagram.
Systems Having a Ternary Heterogeneous Azeotrope. Binary or tern-
ary heterogeneous azeotropes are restricted to being either unstable nodes or
saddles in the residue curve map (163). An example is the computed residue
curve map for the ethanol–water–benzene mixture that exhibits a ternary
minimum boiling hetero-azeotrope (see Fig. 20). This map exhibits three dis-
tillation regions. The three distillation boundaries all begin at the minimum
boiling ternary heteroazeotrope and end at each of the binary azeotropes. If
ethanol is the desired product of the separation, the initial condition for the
simple distillation must lie in the upper region of Fig. 20, which is bounded
by the ternary azeotrope, the binary azeotropes between benzene þ ethanol
and water þ ethanol, and the pure ethanol vertex. The computed phase dia-
gram is shown in Fig. 19a. Other mixtures with a minimum-boiling ternary
heteroazeotrope include, ethanol–water–toluene, ethanol–water–cyclohexane,
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 705

Ethanol
(78.30°C)
1.0

0.8
78.06°C

0.6
67.45°C
x2 A

0.4

0.2
70.50°C

0
0 0.2 0.4 0.6 0.8 1.0
x1
Water Benzene
(100°C) (80.15°C)

Fig. 20. Residue curve map calculated for the ethanol–water–benzene mixture where A
is the end point of the vapor line; & represents a homogeneous azeotrope; * ~ , heteroge-

neous azeotropes; (~—~), the vapor line; (- . -), the heterogeneous liquid boiling envelope;
*, the critical point; and &, the end points of tie-lines (179).

2-propanol–water–benzene, and related mixtures. The water–acetone–


chloroform system exhibits a ternary heterogeneous saddle azeotrope (162),
but such systems are rare and will not be discussed futher in this article.
7.3. Column Sequences. The analysis of residue curve maps and dis-
tillation boundaries for homogeneous azeotropic mixtures provides a simple and
useful technique for distinguishing between feasible and infeasible sequences of
distillation columns, and many of the same insights apply to heterogeneous mix-
tures. Residue curves cross continuously through the liquid boiling envelope
from one side to the other. Thus a simple distillation boundary inside the hetero-
geneous region does not stop abruptly or exhibit a discontinuity at the liquid boil-
ing envelope but passes continuously through it, becoming a homogeneous
distillation boundary thereafter (see Figs. 20 and 22b). As in homogeneous sys-
tems, residue curves cannot cross heterogeneous distillation boundaries. How-
ever, if a point x8, on a residue curve inside of the heterogeneous region phase
splits into two equilibrium liquid phases that lie in two different distillation
regions, then a liquid–liquid phase separation can be exploited to jump across
heterogeneous distillation boundaries in a way that is not possible for homoge-
neous systems. This is the key to devising feasible sequences of columns for
separating heterogeneous mixtures.
Binary Mixtures. A binary mixture containing a homogeneous azeotrope
can be distilled up to, but not beyond, the azeotropic composition. If the binary
706 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

Fig. 21. Column sequence for separating a binary heterogeneous azeotropic mixture, A
and B, where xF represents the process feed mole fraction. (a) Phase diagram; (b) column
sequence.

azeotrope is heterogeneous however, the situation is more favorable and a simple


sequence of columns, shown in Fig. 21, is capable of isolating each pure compo-
nent. The process feed, xF, is fed as a saturated or sub-cooled liquid to a decanter
where it phase-separates into an A-rich phase and a B-rich phase. The A-rich
phase, xF1, is then fed to column 1; the B-rich phase, xF2, is fed to column 2.
As can be seen in Fig. 21a, because of the liquid–liquid phase split the composi-
tions of these two feed streams lie on either side of the azeotrope. Therefore, col-
umn 1 produces pure A as a bottoms product and the azeotrope as distillate,
whereas column 2 produces pure B as a bottoms product and the azeotrope as
distillate. The two distillate streams are fed to the decanter along with the
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 707

process feed to give an overall decanter composition partway between the


azeotropic composition and the process feed composition according to the
lever rule. This arrangement is well suited to purifying water–hydrocarbon
mixtures, such as a C4 –C10 hydrocarbon, benzene, toluene, xylene, etc;
water–alcohol mixtures, such as butanol, pentanol, etc; as well as other immis-
cible systems.
If the process feed does not lie in the liquid–liquid region it can be made to
do so by deliberately feeding either pure A or pure B to the decanter, as required.
This may only be necessary during start-up or for control purposes because the
recycled azeotrope has the beneficial effect of dragging the decanter composition
further into the liquid–liquid region.
Ternary Mixtures. When the binary mixture containing the minimum
boiling azeotrope is completely homogeneous, ie, the liquid is homogeneous for
all compositions, the method given in Fig. 21 requires modification. In this
case, a third component, called the entrainer, is added which induces a liquid–
liquid phase separation over a limited portion of the ternary composition dia-
gram. Many options for sequencing ternary heterogeneous azeotropic distillation
systems exist (159). These sequences generally consist of two, three, or four
columns using various techniques for handling the entrainer recycle stream.
The feasibility of such sequences rests on the use of liquid–liquid phase splits
to provide each column with a feed composition in a different distillation region.
In this regard, the sequences for ternary mixtures resemble the sequences for
binary mixtures. In all cases the heart of the process is the azeotropic column
and its decanter.
The classical example is the separation of ethanol from water using benzene
as the entrainer. Many other separations fit this mold, eg, ethanol–water–car-
bon tetrachloride, 2-propanol–water–benzene, and 2-propanol–water–cyclohex-
ane (180). The task is to separate a homogeneous binary mixture of ethanol and
water, which contains a minimum boiling binary azeotrope, into its pure compo-
nents. An entrainer, which has limited miscibility with one of the components, in
this case water, is used. In addition, benzene (and a number of other entrainers
such as cyclohexane and carbon tetrachloride) causes two more minimum boiling
binary azeotropes (one homogeneous, the other heterogeneous) to form together
with a ternary minimum boiling heterogeneous azeotrope. The resulting residue
curve map is similar to that shown in Fig. 20, in which there are three distilla-
tion regions. Figure 22 shows a schematic of the azeo-column and the material
balance lines for this system (note that the points xB, xF, and xD lie on a straight
line by the lever rule). The map indicates that the components to be separated
(ethanol and water) lie in two different distillation regions, I and II.
The azeo-column must be designed to meet the following target composi-
tions: (1) the bottoms composition is specified to be almost pure ethanol, ie, xB
is specified to lie close to the vertex of distillation region II; and (2) the overhead
vapor composition leaving the last vapor–liquid equilibrium stage of the column,
yN, must lie in the wedge-shaped portion of region II inside the heterogeneous
region near the ternary azeotrope. Typical target values for xB and yN are
shown in Fig. 22. The final values of xB and yN are subject to optimization and
may differ slightly from those shown. All values of yN inside the wedge-shaped
region, except those special values which lie on the vapor line, are in equilibrium
708 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

V, yN

xRich
FRecycle
L
xRecycle xN x 0R
N
D,x0D
xLean
V LT

F1
xF1

V LB

B,xB

(a)

Ethanol
B,xB

F1
F, x F
II

Distillation
boundary
I xN
xRich
xLean
D, xD V, yN

L, x0R
III

Water Benzene

(b)

Fig. 22. Separation of ethanol from an ethanol–water–benzene mixture using benzene


as the entrainer. (a) Schematic representation of the azeo-column; (b) material balance
lines where & denotes the homogeneous and * ~ the heterogeneous azeotropes; &, the
end points of the liquid tie-line and ~, the overhead vapor leaving the top of the column.
The distillation regions, I, II, and III, and the boundaries are marked. Other terms are
defined in the text.
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 709

with a homogeneous liquid. Therefore, the liquid composition leaving the top
tray, xN, lies in the homogeneous portion of region II as shown (Fig. 22). The
azeo-column is therefore designed so that the steady-state liquid composition
profile runs from xB to xN. This can normally be done in such a way that every
stage inside the column is homogeneous, ie, has only one liquid phase. A method
for doing this is given in Ref. 181. Not all azeo-columns, however, are designed to
have a single liquid phase on all the trays. Neither do columns that have been
designed in this manner necessarily remain homogeneous under the action of
disturbances (168,172–174,182–188). For some azeotropic distillations there
can be a significant savings in energy and/or stages by designing the column
so that the entrainer penetrates down the rectifying section causing some of
the stages in the rectifying section to exhibit multiple liquid phases
(180,189,190). One such case is the acetic acid þ water þ n-butyl acetate separa-
tion discussed at the end of this article.
The overhead vapor of composition yN is totally condensed and sent to a
decanter where it splits into two equilibrium liquid phases, an entrainer-rich
phase of composition xrich and an entrainer-lean phase of composition xlean.
The relative proportion of these two liquid phases in the decanter, f, is given
by the lever rule, where f represents the molar ratio of the entrainer-rich
phase to the entrainer-lean phase in the decanter. The overall composition of
the liquids in the decanter is the same as yN. The temperature of the liquid
in the decanter is normally held constant at a value below the VLLE boiling
temperature.
The two decanter liquid phases are used to provide the reflux and distillate
streams. Normally, the reflux ratio, r, is chosen so that r ¼ L/D f. This requires
that the reflux rate be greater than the condensation rate of entrainer-rich phase
and that the distillate rate be correspondingly less than the condensation rate of
entrainer-lean phase. This means that the reflux stream consists of all the
entrainer-rich phase plus some of the entrainer-lean phase, and the distillate
stream consists of the balance of the entrainer-lean phase, ie, xD ¼ xlean. Thus,
the overall composition of the reflux stream, xR0, lies on the decanter tie-line
between the entrainer-rich phase, xrich, and the overall decanter composition,
yN, as shown in Fig. 22.
Completing the Separation Sequence. To close the material balance
around the separation sequence, the distillate stream leaving the azeotropic
column (column 2 in Fig. 23a) is separated (in column 3) into a product stream,
B3, and an entrainer recycle stream, D3. The amount of benzene needed to move
the composition of stream F1 to F is equal to the amount of benzene lost in the
aqueous distillate D2 and recovered in the stream D3.
Kubierschky Three-Column Sequence. If only simple columns are used,
ie, no side-streams, side-rectifiers/strippers etc, then the separation sequence can
be completed by adding an entrainer recovery column, column 3 in Fig. 23a, to
recycle the entrainer, and a preconcentrator column (column 1) to bring the feed
to the azeotropic column up to the composition of the binary azeotrope.
The entrainer recovery column takes the distillate stream, D2, from the
azeo-column and separates it into a bottoms stream of pure water, B3, and a
ternary distillate stream, D3, for recycle to column 2. The overall material bal-
ance line for column 3 is shown in Fig. 23b. This sequence was one of two original
710 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

Entrainer make-up

V 2, y N

D3 , xD3
D3 L2
xRich
x0R2 xLean
D1
D2

2 xD2

F1 1 3

B1 B2 B3

(a)

Ethanol
1.0
B2
D1

0.8

D 1+ D 3

0.6
D3
Column 2 xN
x Lean

0.4
D2, xD2
x Rich

Column 3
0.2 yN
0
L2, xR2
B3

0.0
0.0 0.2 0.4 0.6 0.8 1.0
Water Benzene

(b)

Fig. 23. Separation of ethanol and water from an ethanol–water–benzene mixture. Bot-
toms B1 and B3 are water, B2 is ethanol. (a) Kubierschky three-column sequence where
columns 1, 2, and 3 represent the preconcentration, azeotropic, and entrainer recovery col-
umns, respectively. (b) Material balance lines for the azeotropic and entrainer recovery
columns where & represents the overall vapor composition from the azeo-column, V2,
yN; & the liquid in equilibrium with overhead vapor composition from the azeo-column,
xN; ~, distillate composition from entrainer recovery column xD3; and ^, overall feed com-
position to the azeo-column, D1 þ D3.
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 711

continuous processes disclosed in 1915 (191). More recently, it has been applied
to other azeotropic separations (76,192,193) .
Extensive design and optimization studies have been carried out for this
sequence (193). The principal optimization variables, ie, the design variables
that have the largest impact on the economics of the process, are the reflux
ratio in the azeo-column; the position of the tie-line for the mixture in the decan-
ter, determined by the temperature and overall composition of the mixture in the
decanter; the position of the decanter composition on the decanter tie-line [see
Ref. 181 for a discussion of the importance of these variables]; and the distillate
composition from the entrainer recovery column.
Figure 24 shows material balance lines for three different decanter tie-
lines. The process feed to the preconcentrator in each sequence is a binary
mixture of 4.2 mol% ethanol and 95.8 mol% water. The product purity from
the azeo-column (column 2) is set at 99.9 mol% ethanol, and the water purity
leaving the entrainer recovery column is set at 99.5 mol% water. These specifica-
tions are essentially identical to those used for studying the optimal extractive
distillation sequences (77,78). For design 1, the decanter tie-line is set at the bub-
ble-point of the mixture leaving the top of the azeo-column, having composition
yN. This temperature is 64.48C. For design 2, the decanter composition is the
same as for design 1 but it is subcooled to 258C. For design 3 the decanter com-
position is placed closer to the ternary azeotrope than for designs 1 and 2, and
the decanter temperature is set at the bubble-point of the mixture. In each

Ethanol
1.0

D1
Design 1
0.8 Design 2
Design 3

0.6

0.4

0.2

F1
0.0
0.0 0.2 0.4 0.6 0.8 1.0
Water Benzene

Fig. 24. Three sets of material balance lines for the Kubierschky three-column sequence
where design 1 corresponds to the upper tie-line having rmin ¼ 8.78; design 2, to the sub-
cooled upper tie-line having rmin ¼ 12.23; and design 3, to the lower tie-line having
rmin ¼ 17.31; & represents the overall decanter composition; ^, the overall feed composi-
tion to the azeo-column; }, the distillate composition from the entrainer recovery column;
and *, the end points of the liquid–liquid tie-lines.
712 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

design, the distillate composition from the entrainer recovery column is placed
close to the distillation boundary. The position of the tie-line has a significant
influence on the distillate composition from the azeo-column, which in turn
influences the position of the material balance line for the entrainer recovery
column. The position of the distillate composition from the entrainer recovery
column also influences the overall feed composition to the azeo-column (D1 þ
D3 in Fig. 23).
Optimization studies indicate that the distillate composition from the
entrainer recovery column has a strong influence on the process economics and
the optimum position is always close to the distillation boundary. By the lever
rule, this minimizes the amount of water in the entrainer recycle stream, or
equivalently, makes the overall feed to the azeo-column richer in ethanol. For
each tie-line, the optimal position of the decanter liquid composition is found
by the method proposed in Ref. 181. The minimum reflux ratio for a given tie-
line can be reduced by as much as 50% by making small changes in the decanter
composition. Calculations indicate that the optimal reflux ratio for the azeo-col-
umn is normally in the range 1.1–1.5 rmin and that the cost of the sequence is
insensitive to this factor. This leaves the position of the decanter tie-line as
the sole remaining optimization variable.
The intra-sequence flows, compositions, and reflux ratios are quite sensi-
tive to relatively small changes in the position of the decanter tie-line as evi-
denced by the rmin value for the azeo-column: 8.78 for design 1, 12.23 for
design 2, and 17.3 for design 3. Based on knowledge of homogeneous distilla-
tions, the vapor rate and total annualized cost for sequence 1 would appear
to be the lowest. For homogeneous distillations the feed and product flow
rates, as well as their compositions, can be held constant as the reflux ratio
is changed from one design to another. Thus there is a direct relationship
between increased minimum reflux ratio and increased costs. However, such
a relationship does not occur for heterogeneous distillations so we can’t con-
clude that design 1 will have the lowest cost.
A good approximation for the vapor rate leaving the reboiler, V, for any type
of distillation is

V ¼ ðr þ 1ÞD ð10Þ

where r is the reflux ratio and D is the distillate flow rate. For homogeneous
distillations, D is constant so that V increases as r increases. For azeotropic dis-
tillation, however, both r and D change from one tie-line to another. These effects
may tend either to reinforce or to cancel each other depending on the mixture.
There is no general rule and each mixture must be treated separately. In the
ethanol-water-benzene system, the reflux ratio increases from one design to
another, but the distillate flow rate, D2, decreases, as can be seen by application
of the lever rule to the material balance lines for the azeo-column in Fig. 24.
The net effect is that the vapor rate in the azeo-column hardly changes from
one design to another. All the sequences shown in Fig. 24 have approximately
the same cost. This fortuitous cancellation of effects does not occur in general.
It is always worthwhile exploring the economic impact of variations in the
position of the decanter tie-line.
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 713

Other Sequences. The Kubierschky sequence is not the only way to per-
form the separation. Alternatives include: (1) If the process feed already has a
composition at or near the composition of the binary azeotrope then the precon-
centrator is not needed. (2) The distillate stream from the entrainer recovery col-
umn can be recycled directly to the decanter, which is analogous to the binary
process shown in Fig. 21. This recycle alternative causes the reflux ratio in the
azeotropic column to be much larger than necessary (194) and should be avoided,
even though it has been studied extensively in the literature (172,182,195–199).
(3) Use of the Kubierschky two-column sequence (191), in which the preconcen-
tration and entrainer recovery steps are combined into a single column. For the
ethanol–water–benzene system, this alternative has lower capital costs but
higher operating costs than the Kubierschky three-column sequence so that
the total annualized cost is about the same for both sequences (193). (4) Use of
the Steffen three-column sequence, the basic layout of which is the same as the
Kubierschky three-column sequence. The essential new feature is to replace the
single decanter in the Kubierschky sequence by multiple decanters, adding fresh
or recycled water to each. The entrainer-rich phase from each decanter is
returned to the azeo-column as reflux and the aqueous-rich phase from each
decanter is fed to the next decanter in the train. This effectively replaces the
decanter by a liquid–liquid extraction step which pushes the ‘‘effective decanter
tie-line’’ deeper into the liquid–liquid region than it could possibly go by distilla-
tion alone. (5) Use of the Ricard-Allenet four-column sequence (3,159,191,
193,194,200). The first two columns in this sequence are the same as the first
two columns in the Kubierschky three-column sequence. The third column in
the Ricard-Allenet sequence takes the water-rich phase from the decanter and
splits it into a distillate stream containing the ternary azeotrope, and by material
balance, a bottoms stream consisting of water and ethanol. The ternary azeo-
trope is recycled to the azeo-column decanter, and the ethanol-water stream is
fed to a fourth column, which recovers pure water as bottoms and the binary
azeotrope as distillate. This azeotropic distillate stream is recycled to the feed
into the azeo-column. Ricard-Allenet proposed a variation on this sequence
(191) in which the overhead vapors from the entrainer recovery column (third
column in the sequence) are provided with a separate condenser–decanter
system, similar to the one provided for the overhead vapors from the azeo-col-
umn. (6) Use of the Ricard-Allenet three-column sequence in which the fourth
column is eliminated and the bottom stream from the third column (containing
only ethanol and water) is recycled to the preconcentrator column. This alter-
native has good economic possibilities, but there are no available studies of
this alternative.
In summary, for systems of the ethanol–water–benzene type, the three
most attractive sequences for carrying out azeotropic distillation are the
Kubierschky three-column sequence, the Kubierschky two-column sequence,
and the Ricard-Allenet three-column sequence. For each of these there is the
added possibility of replacing the decanter with a liquid–liquid extraction step.
Other Classes of Entrainers. Not all heterogeneous azeotropic mixtures
are of the ethanol–water–benzene type. The number of azeotropes in the mix-
ture may vary from system to system as may their character, ie, maximum or
minimum boiling, hetereogeneous or homogeneous. In addition, the size and
714 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

A A

B Entrainer B Entrainer
(a) (b)

A A

B Entrainer B Entrainer
(c) (d)

Fig. 25. A selection of feasible residue curve maps for ternary heterogeneous mixtures
where & represents homogeneous and * ~ heterogeneous azeotropes. See text.

shape of the liquid–liquid region varies greatly from system to system. The
feasibility and sequencing strategy for each new system is most conveniently
established using residue curve maps such as those shown in Fig. 25.
Any entrainer that does not divide the components to be separated into dif-
ferent distillation regions is normally a feasible entrainer, and remains feasible
even if it induces a region of partial miscibility. Examples are shown in Fig. 25a
and b and it is possible to construct many more such maps by strategically pla-
cing heterogeneous regions on the feasible residue curve maps given in Fig. 6.
Sequences based on these maps normally have multiple liquid phases on some
of the stages in the column, which may lower the mass-transfer efficiency on
those stages. In practice, mass-transfer efficiency and hydrodynamic perfor-
mance of three-phase columns does not present a problem. As early as 1938 it
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 715

was stated that ‘‘The efficiency of the plates was apparently undiminished by the
heterogeneity of the boiling liquid and that was undoubtedly due to the violent
agitation produced on the plates by the rapid bubbling of the vapours through
the liquid . . .’’ (159). A more recent study and assessment of the literature
from a practitioner’s viewpoint is available (201). However, at the conceptual
design stage of process development, mass-transfer efficiency is of secondary
importance to feasibility.
The map shown in Fig. 25b is relevant to the separation of acetic acid and
water which is of commercial significance. Although this binary mixture does not
form an azeotrope, it does have a severe tangent pinch at high water composi-
tions, which makes it difficult to produce an acid-free water stream. It is not eco-
nomical to separate this mixture into pure product streams without the aid of an
entrainer. All known commercial entrainers for this separation are heteroge-
neous and most produce residue curve maps similar to the one shown in Fig. 25b.
Clarke-Othmer Process for Acetic Acid–Water Separation. Large
amounts of dilute acetic acid are purified industrially. Feasible entrainers
include ethylene dichloride (nbp 83.78C), chloroform (npb 61.88C), ethyl ether
(nbp 34.58C), also called ether or diethyl ether), 1-propyl acetate (nbp 101.68C),
isopropyl acetate (nbp 88.48C), and 1-butyl acetate (nbp 125.98C). All except -
butyl acetate have residue curve maps similar to Fig. 25b. The azeo-column
has a process feed consisting of acetic acid (component A in Fig. 25b) and
water (component B in Fig. 25b). The entrainer is charged to the column, typi-
cally via the reflux stream, and remains captive within the column. The bottom
stream from the column is pure acetic acid, and the overhead vapor is close to the
composition of the minimum boiling heterogeneous binary azeotrope formed by
the entrainer and water (see Fig. 25b). The azeotropic vapors are condensed and
decanted into a water stream that leaves as distillate, and an entrainer-rich
stream that is returned to the column as reflux. In some process alternatives,
part of this stream is also recycled to the column feed (see p. 67 of Ref. 202). Addi-
tional reflux, if needed, is achieved by returning some of the water phase. It is
typical for these systems to have multiple liquid phases present on some stages
in the rectifying section. This process was invented by Clarke and Othmer (203);
more detailed operating information and entrainer comparison for this separa-
tion is available (204). Ethyl acetate (nbp 77.18C) is sometimes mentioned as a
feasible entrainer for separating acetic acid–water mixtures and, although it
does give a residue curve map similar to Fig. 25b, the liquid–liquid region is
so small that it is not nearly as attractive an entrainer as those cited above. It
is for this reason that many users of this system employed multicomponent
entrainers, adding for example cyclohexane and/or methylene chloride to ethyl
acetate so as to enlarge the liquid-liquid region while retaining as much as pos-
sible the favorable distribution coefficients (tie-line slopes) of the ethyl acetate
system. Another entrainer that is sometimes used industrially is n-butyl acetate
that has a normal boiling point of 1268C, which is 88C higher than acetic acid.
This is a feasible entrainer but the residue curve map is different than the one
shown in Fig. 25b (see the section below, Self-Entrained Systems, for more
details).
Wentworth Process for Ethanol–Water Separation. In the Wentworth
process ethyl ether is used as the entrainer for ethanol– water separation,
716 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

producing a residue curve map similar to the one shown in Fig. 25c (A and B
represent ethanol and water, respectively). Ethyl ether and water form a mini-
mum boiling heterogeneous azeotrope at 34.158C containing 98.75 wt% ether and
1.25 wt% water at atmospheric pressure. There is no azeotrope between ethyl
ether and ethanol and no ternary azeotrope. The dilute ethanol process feed is
first preconcentrated up to the composition of the ethanol–water azeotrope in a
preconcentrator column. This stream is fed to the azeo-column that produces
pure ethanol as bottoms product and an overhead vapor close to the composition
of the ethyl ether–water azeotrope. The overhead vapors are condensed and
decanted into an ether-rich layer that is returned to the column as reflux, and a
water-rich layer that leaves as distillate. The distillate contains no alcohol and
very little ether. Pure water may be obtained from this stream by sending it to
a stripping column where water is recovered as the bottom product and the over-
head vapor has a composition near the ethyl ether–water azeotrope. These vapors
are condensed and recycled to the decanter. The azeo-column is normally operated
at  700 kPa (7 atm) pressure because, (1) this increases the amount of water in
the ether–water azeotrope, thereby reducing the amount of ether needed in the
system, and (2) it allows for cooling water to be used in the condenser.
This process is described in the literature in more detail (205–207). Its
main advantage was a lower energy consumption ( 6000 kJ/L (20,000 Btu/
gal) ethanol product) relative to the benzene process (12,000 kJ/L (43,000 Btu/
gal) ethanol) (206). Since the 1940s the gap has been narrowed by better designs
for the benzene process that is now capable of producing 99.8 mol% ethanol at an
energy consumption of 8,400 kJ/L (30,000 Btu/gal) ethanol product (78,193), or
using thermally integrated columns for 5000 kJ/L (18,000 Btu/gal) ethanol pro-
duct (78,208). In recent years, homogeneous separating agents have shown great
promise for the ethanol–water separation, and extractive distillation processes
using ethylene glycol as the solvent have been designed with energy consump-
tions of  6000 kJ/L (22,000 Btu/gal) ethanol product (78), or using thermally
integrated columns for 2200–3300 kJ/L (8,000–12,000 Btu/gal) ethanol product
(78,209,210).
Rodebush Sequence for Ethanol–Water Separation. When ethyl acet-
ate is used as the entrainer to break the ethanol–water azeotrope the residue
curve map is similar to the one shown in Fig. 25d, ie, the ternary azeotrope is
homogeneous. Otherwise the map is the same as for ethanol–water–benzene.
In such cases the liquid leaving the azeo-column condenser does not separate
into two liquid phases, and the sequence is infeasible unless special tricks
are employed. In the Rodebush sequence, water is continuously added to the
decanter in order to shift the overall composition into the two-liquid phase
region. The water-rich phase from the decanter is fed to a distillation column
where pure water is recovered in the bottom stream (some of which is recycled
to the decanter) and the distillate, which lies near the ternary azeotrope, is
recycled to the azeo-column. The ethyl acetate-rich phase from the decanter
is returned to the azeo–column as reflux (191). Pure ethanol is recovered in
the bottom stream of the azeo-column. More recently, a clever variation on
this sequence was patented (211) for separating a ternary feed consisting of
ethanol, water, and diethoxymethane. This variation also has a residue curve
map similar to the one shown in Fig. 25d.
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 717

More Complex Mixtures. All the sequences discussed are type I liquid
systems, ie, mixtures in which only one of the binary pairs shows liquid–liquid
behavior. Many mixtures of commercial interest display liquid–liquid behavior
in two of the binary pairs (type II systems), eg, secondary butyl alcohol–
water–di-secondary butyl ether (SBA–water–DSBE), and water–formic acid-
meta-xylene (162). Sequences for these separations can be devised on the basis
of residue curve maps. The SBA–water–DSBE separation is practiced by
ARCO and is considered in detail in the literature (183,184,188,212).
Self-Entrained Systems. It is quite common to find that the feed mixture
to an azeotropic distillation column is not a binary azeotropic mixture but a tern-
ary mixture that already contains a third component that can act as an entrai-
ner. Such systems are called self-entrained. The presence of the third component
may be unavoidable on account of the reaction chemistry (eg, acetic acid–water–
vinyl acetate (213–215); ethanol–water–diethoxy methane, 211), or may be pre-
sent on purpose due to the choice of process technology. For example, in some
process alternatives for separating acetic acid–water mixtures by azeotropic dis-
tillation part of the entrainer-rich phase from the decanter is recycled back to the
column feed (202, p. 67), thereby creating a self-entrained feed to the column. In
other process alternatives for separating the same mixture, solvent extraction is
used to remove almost all the acetic acid from the water, creating an extract
stream that contains almost all of the acetic acid that entered with the process
feed, some water, and all of the solvent used in the extractor (202, p. 68; 216,217).
This extract stream is then fed (as a self-entrained ternary mixture) to an azeo-
tropic distillation column where glacial acetic acid (>99.7 wt%) is recovered as
the bottom product. In these kinds of separation sequences, the third component
is chosen to be both a good solvent in the extractor and a good entrainer in the
azeotropic distillation. Minotti et al. (218) give a detailed description of the pro-
cess alternatives and process economics of using diethyl ether to separate acetic
acid–water mixtures using a combination of extraction and azeotropic distilla-
tion flow sheets.
Two typical self-entrained azeotropic distillation systems are described
below.
Separation of Acetic–Water–Isopropyl Acetate Mixtures. At 1 atm
pressure, isopropyl acetate boils at 88.48C and forms a binary heterogeneous
azeotrope with water that boils at 77.28C and contains 40 mol% water. Iso-
propyl acetate is the lightest boiling component in the mixture and the residue
curve map, calculated using the NRTL activity coefficeint model, is shown in
Fig. 26. (Note that several other light-boiling entrainers, such as vinyl acetate,
diethyl ether, chloroform, etc, generate similar residue curve maps and have
qualitatively similar azeotropic distillations. Thus, the isopropyl acetate
entrainer is representative of other entrainers of the light-boiling type). For
illustration, a saturated liquid (q ¼ 1) feed composition of 20 mol% isopropyl
acetate, 40 mol% water and 40 mol% acetic acid is assumed. The desired
purity of the bottoms stream is glacial acetic acid containing 99.9 mol% acetic
acid. Two design alternatives exist. With the same feed composition, the
column can be designed to operate with either an entrainer-rich or an
entrainer-lean composition profile. In either case the column configuration is
the same as shown in Fig. 27, and the distillate composition (determined by
718 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

Acetic Acid (2)


(118.2°C)
1.0

0.8

0.6

x2

0.4

0.2

xxxxxx
0.0
0.0 0.2 0.4 77.2 0.8 1.0
Water (3) x1 Isopropyl Acetate (1)
(100.0°C) (88.4°C)

Fig. 26. Residue curve map and liquid–liquid boiling envelope for acetic acid þ
water þ isopropyl acetate at 1-atm pressure.

material balance with the feed and bottoms compositions) is essentially the
same in both cases. (Note that the distillate composition of acetic acid deter-
mines which tie-line in the liquid – liquid region corresponds to the decanter
tie-line, but the relative proportion of phases in the decanter is determined
by the position of the overhead vapor composition on the decanter tie-line,

Fig. 27. Schematic column configuration for self-entrained azeotropic distillation.


Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 719

Acetic acid (2) Components Feed Distillate Bottoms


1.0 Bottoms 1 0.2000 0.3244 0.0010
2 0.4000 0.0256 0.9989
3 0.4000 0.6500 0.0001

0.8 P (atm) = 1 Stages:


r= 1.90 Top stage 17
s= 4.65 Feed stage 16
q= 1.0 Reboiler 1
0.6 φ= 3
D/F = 0.62
x2 B/F = 0.38
V/F = 1.79

0.4 Liquid leaving top


Feed stage of column, xN
Overhead vapor composition
Decanter (decanter composition) yN
tie-line
0.2 Feed
Liquid boiling stage x rich = x reflux
envelope
L2
L1
xxxxxx
0.0
0.0 0.2 0.4 0.6 0.8 1.0
Water (3) Distillate x1 Isopropyl acetate (1)
Vapor line Binary azeotrope

Fig. 28. Azeo-column liquid composition profile .—. (solid line connecting filled circles)
and design data for the acetic acid þ water þ isopropyl acetate separation. Entrainer-rich
design with low recovery of acetic acid (2.56 mol % acetic acid in the distillate).

see Fig. 28). The decanter composition, which is the same as the overhead
vapor composition, is, however, quite different in the two designs, as is the
reflux composition (ie, the proportion of the two decanter liquid phases
returned to the column is different), and the reflux ratios.
The first entrainer-rich design is shown in Fig. 28. The distillate is selected
so it contains some acetic acid in order to make the decanter tie-line visible on the
diagram and the relationship between the various points on it more obvious. The
bottom stream, the acetic acid product, contains more isopropyl acetate (0.001
mole fraction) than water (0.0001 mole fraction). This distribution of the trace
components forces the stripping section composition profile to move along the
acetic acid–isopropyl acetate edge of the diagram and intersect with the rectify-
ing section composition profile in the entrainer-rich region of the diagram. The
result is a column design where the liquid on all of the stages throughout the col-
umn contains more entrainer than water. The top of the column is designed so
that the overhead vapor composition (which becomes the feed to the decanter)
lies to the entainer-rich side of the azeotrope on the decanter tie-line. This causes
the liquid on the top tray of the column to be rich in entrainer (see Fig. 28). The
ratio, f, of entrainer-rich phase (L2) to entrainer-lean phase (L1) in the decanter
is 3:1 (this is determined by the design value of the decanter composition). Part of
the entrainer-rich decanter liquid is refluxed to the column, the balance is mixed
with the entrainer-lean decanter liquid to produce a distillate composition at the
required point along the decanter tie-line. There is a simple algebraic material
balance relationship between the reflux ratio and the reflux composition
720 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

(which also contains the distillate composition and the decanter composition as
known parameters). Therefore, a chosen value of r implies that the reflux com-
position is determined, and vice-versa. When we select the reflux composition to
be equal to the composition of the entrainer-rich phase the corresponding value
of the reflux ratio is r ¼ 1.9. This entrainer-rich design requires 17 theoretical
stages to perform the separation at a reflux ratio of 1.9. Sixteen stages are in
the stripping section and only 1 is in the rectifying section. A key measure of eco-
nomic performance (energy use) is the internal vapor rate relative to the feed
rate (V/F), which is 1.79 for this design. Generally in distillation, values of
V/F < 1 are considered very attractive, while values greater than 3 are unattrac-
tive. Values in between are normal.
A more realistic entrainer-rich design is shown in Fig. 29, where the distil-
late is selected to have only a small amount of acetic acid present (mole fraction
of 0.0001). The decanter tie-line lies almost on top of the isopropyl acetate-water
edge of the diagram. In this design, the feed and bottom composition are the
same as before, so is f ¼ L2/L1 ¼ 3. We again choose to reflux only the entrai-
ner-rich phase (which determines the reflux ratio to be r ¼ 1.88), giving a column
with a total of 19 theoretical stages (15 in the stripping section and 4 in the rec-
tifying section), and V/F ¼ 1.72. This design, therefore, produces high purity
glacial acetic acid at high recovery (ie, with very little acetic acid lost in the dis-
tillate) in a column with both a reasonable number of stages and a reasonable
boilup.

Components Feed Distillate Bottoms


Acetic acid (2)
1 0.2000 0.3329 0.0010
1.0 Bottoms
2 0.4000 0.0001 0.9989
3 0.4000 0.6670 0.0001

0.8 P (atm) = 1 Stages:


r= 1.88 Top stage 19
s= 4.31 Feed stage 15
q= 1 Reboiler 1
φ= 3
0.6
D/F = 0.60
B/F = 0.40
x2 V/F = 1.72

0.4
Feed

Feed stage
0.2

xxxxxx
0.0
0.0 0.2 0.4 0.6 yN 0.8 xN 1.0
Water (3) Distillate x1 Isopropyl acetate (1)

Fig. 29. Azeo-column liquid composition profile and design data for the acetic acid þ
water þ isopropyl acetate separation. Entrainer-rich design with high recovery of acetic
acid (100 ppm (molar) acetic acid in the distillate).
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 721

Components Feed Distillate Bottoms


Acetic acid (2)
1.0 Bottoms 1 0.2000 0.3253 0.00001
2 0.4000 0.0247 0.9990
3 0.4000 0.6500 0.00099

0.8 P (atm) =1 Stages:


r =3 Top stage 22
s = 6.38 Feed stage 14
Feed stage
q =1 Reboiler 1
φ =1
0.6
D/F = 0.61
x2 B/F = 0.39
V/F = 2.46

0.4
Feed

0.2

xN
xxxxxx
0.0
0.0 0.2 0.4 0.6 0.8 1.0
Water(3) Distillate yN xreflux Isopropyl acetate (1)
x1

Fig. 30. Azeo-column liquid composition profile and design data for the acetic acid þ
water þ isopropyl acetate separation. Entrainer-lean design with low recovery of acetic
acid (2.47 mol% acetic acid in the distillate).

By contrast, entrainer-lean designs require more stages and more boilup.


The entrainer-lean designs described below have exactly the same feed composi-
tion, and almost the same acetic acid purity in the bottoms stream and acetic acid
composition in the distillate as the entrainer-rich designs. The first case is shown
in Fig. 30. It is the entrainer-lean analog of the design in Fig. 28. The overhead
vapor composition is selected to lie on the entrainer-lean side of the azeotrope so
that the composition of the liquid on the top stage of the column is also on the
lean side of the diagram. (A phase ratio of 1:1 in the decanter achieves this).
The minimum reflux for this separation is 1.8 and the corresponding value of
(V/F)min is 1.7. Choosing a reflux ratio, r ¼ 3, gives V/F ¼ 2.5, and also deter-
mines that the reflux stream contains roughly equal amounts (on a molar
basis) of the two decanter phases (see Fig. 30). This design requires a total of
22 theoretical stages for the separation. The distribution of stages is very differ-
ent than the design in Fig. 28, with 14 in the stripping section and 8 in the rec-
tifying section. Therefore, this design requires more stages and more boilup to
deliver the same product specifications as the entrainer-rich design. Note that
the trace components in the bottom stream for this design consist mainly of
water (0.001 mole fraction versus 0.00001 mole fraction of isopropyl acetate).
This trace component distribution forces the stripping section composition
profile to move along the acetic acid-water edge of the diagram and intersect
with the rectifying composition profile in the entrainer-lean region of the
diagram. The result is a column design where the liquid on all of the stages
contains less entrainer than water.
722 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

Acetic acid (2) Components Feed Distillate Bottoms


1.0 Bottoms 1 0.2000 0.3333 0.00001
2 0.4000 0.0007 0.9990
3 0.4000 0.6660 0.0010

0.8 P (atm) = 1 Stages:


r = 3 Top stage 30
Feed stage s = 6 Feed stage 14
q = 1 Reboiler 1
φ = 1
0.6
D/F = 0.6
B/F = 0.4
x2
V/F = 2.4

0.4
Feed

0.2

xN
xxxxxx
0.0
0.0 0.2 0.4 0.6 0.8 1.0
Water (3) Distillate yN xreflux Isopropyl acetate (1)
x1

Fig. 31. Azeo-column liquid composition profile and design data for the acetic acid þ
water þ isopropyl acetate separation. Entrainer-lean design with high recovery of acetic
acid (700 ppm (molar) acetic acid in the distillate).

An entrainer-lean design that delivers high purity acetic acid at high


recovery is shown in Fig. 31. Thirty stages (14 in the stripping section and 16
in the rectifying section) are required at a reflux ratio of r ¼ 3 (V/F ¼ 2.4).
Again, the minimum reflux ratio for this separation is 1.8. It is possible to
reduce the number of stages by increasing the reflux ratio, eg, when r ¼ 5, V/
F ¼ 3.6, and the total number of stages is reduced to 25. However, compared to
the corresponding entrainer-rich design in Fig. 29, the entrainer-lean design
requires more stages and more boilup to deliver the same acetic acid product
purity and recovery.
As more entrainer is added to the feed (keeping the relative proportions of
acetic acid and water roughly equal), it remains possible to generate both entrai-
ner-rich and entrainer-lean designs, although the entrainer-rich design is always
the better of the two. However, as the composition of water in the feed increases,
it eventually becomes impossible to generate an entrainer-rich design and, as the
composition of isopropyl acetate in the feed increases, it eventually becomes
impossible to generate an entrainer-lean design.
One final possibility for the design of the azeo-column is to put the overhead
vapor composition exactly on the vapor line, causing the liquid on the top stage of
the column (and possibly the next several stages below it) to be heterogeneous.
The rectifying composition profile then moves through the liquid–liquid region
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 723

Acetic acid (2)


(118.2°C)
1.0

0.8

0.6

x2

0.4

0.2

xxxxxxxxxxx
0.0
0.0 0.2 0.4 0.6 91.0 0.8 1.0
N-Butyl acetate (3) x1 Water (1)
(125.9°C) (100.0°C)

Fig. 32. Residue curve map and liquid–liquid boiling envelope for acetic acid þ
water þ n-butyl acetate at 1-atm pressure.

rather than around the outside of it. This may be advantageous for some systems
such as the n-butyl acetate case that follows.
Separation of Acetic Acid–Water–n–Butyl Acetate Mixtures. At 1-atm
pressure, n-butyl acetate boils at 1268C and forms a binary heterogeneous azeo-
trope with water that boils at 918C and contains 72 mol% water. N-Butyl acet-
ate is the highest boiling component in the mixture and the residue curve map,
calculated using the NRTL activity coefficient model, is shown in Fig. 32. This
map is quite different than the map shown in Fig. 26 for light-boiling entrainers,
and the feasible splits and azeotropic column designs are also different. Only
entrainer-lean designs are possible. This is due to the fact that acetic acid is
the stable node (high boiler) and the entrainer is a saddle (intermediate boiler)
in Fig. 26, while acetic acid is a saddle and the entrainer is a stable node in
Fig. 32.
Consider a saturated liquid feed containing 20 mol% n-butyl acetate,
40 mol% water and 40 mol% acetic acid. The desired purity of the bottoms stream
is again glacial acetic acid containing 99.9 mol% acetic acid. The column config-
uration is the same as in Fig. 27, and the distillate composition is again chosen to
have some acetic acid present in order to make the decanter tie-line visible on the
triangular diagram. A typical design is shown in Fig. 33. The distillate composi-
tion determines the position of the decanter tie-line, and the overhead vapor com-
position is selected to lie on the entrainer-lean side of the azeotrope (the ratio of
entrainer-rich phase (L2) to entrainer-lean phase (L1) in the decanter is 1:3),
which gives a liquid on the top tray of the column that is also lean in entrainer.
The rectifying profile starts at this point and moves toward the stable node (high
724 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

Acetic acid (2) Components Feed Distillate Bottoms


1.0 Bottoms 1 0.4000 0.6500 0.0010
2 0.4000 0.0253 0.9980
3 0.2000 0.3247 0.0010

0.8 P (atm) = 1 Stages:


r= 2 Top stage 30
s= 4.79 Feed stage 23
q= 1 Reboiler 1
φ= 0.333
0.6
D/F = 0.61
B/F = 0.39
x2
V/F = 1.84

0.4
Feed

Feed stage

0.2

xN

xxxxxxxxxxx
xrich xlean
0.0
0.0 0.2 0.4 0.6 yN 0.8 1.0
N-Butyl acetate (3) x1 Distillate xreflux Water (1)

Fig. 33. Azeo-column liquid composition profile and design data for the acetic acid þ
water þ n-butyl acetate separation. Low recovery of acetic acid (2.53 mol % acetic acid
in the distillate).

boiler) at n-butyl acetate (ie, in the direction of increasing temperature down the
column). It intersects the stripping profile which moves from the acetic acid
corner of the diagram along the acetic acid–water edge before turning into the
diagram towards the azeotrope (ie, in the direction of decreasing temperature
up the column). The minimum reflux at which these profiles just intersect is
1. A reasonable trade-off between stages and reflux occurs at r ¼ 2 (V/
F ¼ 1.84), giving 30 theoretical stages in the column (23 in the stripping section
and 7 in the rectifying section), see Fig. 33. As can be seen in this Fig., the reflux
stream and the liquid composition on all of the stages are entrainer-lean. It is not
possible to design a column with an entrainer-rich profile because rectifying pro-
files that start near the entrainer-rich end of the decanter tie-line will move
directly to pure n-butyl acetate (in the direction of increasing temperature
down the column). The rectifying profile can never intersect the stripping profile,
which starts at acetic acid and moves toward the azeotrope (in the direction of
decreasing temperature up the column). Another way of saying this is that
heavy entrainers cannot be forced up the column to generate entrainer-rich
designs.
With n-butyl acetate as the entrainer, high recovery of acetic acid is possi-
ble only with larger reflux ratios (in the range of r ¼ 3–5, giving V/F in the range
of 2.4–3.6) and larger numbers of theoretical stages (40–34, respectively), eg, see
Fig. 34. Reflux ratios < 3 increase the number of stages dramatically (eg, r ¼ 2
requires 50 theoretical stages). This effect is due to the proximity of the column
composition profile to the tangent pinch between acetic acid and water located in
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 725

Acetic acid (2) Components Feed Distillate Bottoms


1.0 Bottoms 1 0.4000 0.6660 0.0010
2 0.4000 0.0013 0.9980
3 0.2000 0.3327 0.0010

0.8 P (atm) = 1 Stages:


r = 3 Top stage 40
s = 6 Feed stage 21
q = 1 Reboiler 1
0.6 φ = 0.333
D/F = 0.6
x2 B/F = 0.4
V/F = 2.4
0.4
Feed

Feed Stage

0.2

xxxxxxxxxxx

0.0
0.0 0.2 0.4 0.6 yN 0.8 xreflux1.0
N-Butyl acetate (3) x1 Distillate
Water (1)

Fig. 34. Azeo-column liquid composition profile and design data for the acetic acid þ
water þ n-butyl acetate separation. Higher recovery of acetic acid (0.13 mol % acetic
acid in the distillate).

the vicinity of pure water. When the distillate contains very little acetic acid, the
decanter tie-line lies on the n-butyl acetate-water edge of the triangle, and this
puts the composition of the liquid on the top stage of the column right on top of
the tangent pinch where the separation per stage is very small, see Fig. 34.
For this system there is a clear advantage to having multiple liquid phases
on the top several trays of the rectifying section. Such designs put the top portion
of the rectifying profile inside of the liquid–liquid region of the diagram, well
away from the tangent pinch. This has the effect of simultaneously reducing
both the reflux ratio and the total number of theoretical stages, see Fig. 35.
Parten and Macpherson Ure (189) disclose a process for achieving this by only
refluxing the organic phase. Urdaneta and co-workers (180) and Wasylkiewicz
and co-workers (190) propose methodologies for systematically designing such
azeotropic distillations with multiple liquid phases on some of the stages inside
the column. Designs for the acetic acid-water-n-butyl acetate system are consid-
ered in detail in (190), and one is reproduced in Fig. 35.
Whereas there is extensive literature on design methods for azeotropic and
extractive distillation, much less has been published on operability and control.
It is, however, widely recognized that azeotropic distillation columns can be dif-
ficult to operate and control because some of these columns can exhibit complex
dynamic behavior, multiple steady-states, and parametric sensitivity (172–
174,182–187,219). In contrast, extractive distillations typically do not exhibit
such complex behavior and even highly optimized columns are typically no
726 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

Acetic acid (2)


1.0

0.8

0.6

0.4

0.2

xxxxxxxxxxxxxx
0.0
0.0 0.2 0.4 0.6 0.8 1.0
N-Butyl acetate (3) Water (1)

Fig. 35. Azeo-column liquid composition profile for the acetic acid þ water þ n-butyl acet-
ate separation with two liquid phases on the top stages in the rectifying section. Taken
from reference (190). The heterogeneous binary azeotrope and vapor-line are also shown.

more difficult to control than ordinary distillation columns producing high purity
products (220). Exceptions do exist, however. Multiple steady-states have been
observed in at least one industrial extractive distillation (221).

8. Nomenclature

B bottom stream or bottoms flow rate


c number of components in a mixture
D distillate stream or distillate flow rate
F feed stream or feed flow rate
L liquid flow rate
P pressure
Psat
i vapor pressure of component i
R gas constant
r reflux ratio
Sij selectivity of a solvent for solutes i and j (defined by eq. 8)
S1ij selectivity of a solvent with the solutes present at infinite dilution
(defined by eq. 9)
T temperature
TAC total annualized cost (see Ref. 103)
V vapor flow rate
ViL liquid-phase molar volume of component i
xi liquid-phase composition of component i (mol fraction)
yi vapor-phase composition of component i (mol fraction)
aij volatility of component i relative to component j (defined by eq.6)
gi liquid-phase activity coefficient of component i
gi1 infinite dilution liquid-phase activity coefficient of component i
x nonlinear time scale (see eq.7)
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 727
f the ratio of the two phases in a decanter
fiv vapor-phase fugacity coefficient of component i
fisat liquid-phase saturation fugacity coefficient of component i
Subscripts
B bottom
D distillate
F feed
lean entrainer-lean phase
min minimum
N stage number N
R reflux
rich entrainer-rich phase
S in the presence of solvent
Superscripts
AZ denotes an azeotrope or azeotropic condition
o overall (composition)

BIBLIOGRAPHY

‘‘Azeotropes’’ in ECT 1st ed., under ‘‘Distillation,’’ Vol. 5, pp. 176–179, by E. G. Scheibel,
Hoffmann-LaRoche Inc.; ‘‘Azeotropy and Azeotropic Distillation’’ in ECT 2nd ed., Vol. 2,
pp. 839–858, by D. F. Othmer, Polytechnic Institute of Brooklyn; ‘‘Azeotropic and Extrac-
tive Distillation’’ in ECT 3rd ed., Vol. 3, pp. 352–377, by D. F. Othmer, Polytechnic Insti-
tute of New York; in ECT 3rd ed., Supl. Vol., pp. 145–158, by G. Prokopakis, Columbia
University.

CITED PUBLICATIONS

1. L. H. Horsley, Azeotropic Data III, Advances in Chemistry Series No. 116, American
Chemical Society, Washington, D.C., 1973.
2. J. Gmehling, J. Menke, J. Krafczyk, and K. Fischer, Azeotropic Data, Parts I & II,
VCH Publishers, Inc., New York (1994).
3. C. J. King, Separation Processes, 2nd ed., McGraw-Hill Publishing Co., New York,
1980.
4. J. D. Seader, J. J. Siirola, and S. D. Barnicki, in R. H. Perry, D. W. Green and J. O.
Maloney eds., Perry’s Chemial Engineer’s Handbook, 7th ed., McGraw-Hill, New
York (1997), Section 13, pp. 54–81.
5. G. Q. Martin, Hydroc. Proc., 241 (Nov. 1975).
6. J. M. Smith and H. C. Van Ness, Introduction to Chemical Engineering Thermody-
namics, 3rd ed., McGraw-Hill, New York, 1975.
7. U.S. Pat. 4,287,379 (Sept. 1, 1981), E. Brunner, E. Schubert, A. Lindner, F. Merger,
K. Volkamer, M. Strohmeyer, and G. Sandrock (to BASF AG).
8. A. Fahmy, D. Mewes, and K. Ebert, Separation Sci. and Technol., 36(15), 3287
(2001).
9. F. Lipnizki, R. W. Field, and P.-K. Ten, J. Membrane Sci., 153, 183 (1999).
10. G. Guerreri, Chem. Eng. Res. Des., 70(A5), 501 (1992).
11. J. G. Stichlmair, J. R. Fair, and J. L. Bravo, Chem. Eng. Prog. 85(1), 63 (1989).
12. S. Widagdo and W. D. Seider, AIChE J., 42(1) 96 (1996).
13. M. F. Doherty and G. A. Caldarola, Ind. Eng. Chem. Fund., 24, 474 (1985).
728 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

14. H. Matsuyama and H. Nishimura, J. Chem. Eng. Jpn., 10, 181 (1977).
15. E. J. Peterson and L. R. Partin, Ind. Eng. Chem. Res., 36, 1799 (1997).
16. M. F. Doherty and J. D. Perkins, Chem. Eng. Sci., 33, 281 (1978).
17. M. F. Doherty and J. D. Perkins, Chem. Eng. Sci., 33, 569 (1978).
18. E. R. Foucher, M. F. Doherty, and M. F. Malone, Ind. Eng. Chem. Res., 30, 760
(1991).
19. M. F. Doherty, Chem. Eng. Sci., 40, 1885 (1985).
20. M. F. Doherty and J. D. Perkins, Chem. Eng. Sci., 34, 1401 (1979).
21. M. F. Doherty and M. F. Malone, Conceptual Design of Distillation Systems,
McGraw-Hill, New York (2001).
22. Y. Yamakita, J. Shiozaki, and H. Matsuyama, J. Chem. Eng. Jpn., 16, 145 (1983).
23. D. B. Van Dongen and M. F. Doherty, Chem. Eng. Sci., 40, 2087 (1985).
24. C. Bernot, M. F. Doherty, and M. F. Malone, Chem. Eng. Sci., 45, 1207 (1990).
25. C. Bernot, M. F. Doherty, and M. F. Malone, Chem. Eng. Sci., 46, 1311 (1991).
26. B. S. Ahmad, Y. Zhang, and P. I. Barton, AIChE J., 44(5), 1051 (1998).
27. M. Benedict and L. C. Rubin, Trans. AIChE, 41, 353 (1945).
28. E. Rev, Ind. Eng. Chem. Res., 31, 893 (1992).
29. L. Laroche, N. Bekiaris, H. W. Andersen, and M. Morari, Ind. Eng. Chem. Res., 31,
2190 (1992).
30. L. Laroche, N. Bekiaris, H. W. Andersen, and M. Morari, AIChE J., 38, 1309 (1992).
31. O. M. Wahnschafft, J. W. Koehler, E. Blass, and A. W. Westerberg, Ind. Eng. Chem.
Res., 31, 2345 (1992).
32. J. G. Stichlmair and J.-R. Herguijuela, AIChE J., 38, 1523 (1992).
33. Z. T. Fidkowski, M. F. Doherty, and M. F. Malone, AIChE J., 39(8), 1303 (1993).
34. G.-J. A. F. Fien and Y. A. Liu, Ind. Eng. Chem. Res., 33, 2505 (1994).
35. S. K. Wasylkiewicz, L. C. Kobylka, and M. A. Satyro, Chem. Eng., 80 (Aug. 1999).
36. D. Y.-C. Thong and M. Jobson, Chem. Eng. Sci., 56, 4369 (2001).
37. V. Julka, A Geometric Theory of Multicomponent Distillation, Ph.D. dissertation,
University of Massachusetts, Amherst, 1992.
38. Z. T. Fidkowski, M. F. Malone, and M. F. Doherty, Comput. Chem. Engng., 17(12),
1141 (1993).
39. S. T. Harding, C. D. Maranas, C. D. McDonald, and C. M. Floudas, Ind. Eng. Chem.
Res., 36, 160 (1997).
40. S. K. Wasylkiewicz, M. F. Doherty, and M. F. Malone, Ind. Eng. Chem. Res., 38, 4901
(1999).
41. B. T. Safrit and A. W. Westerberg, Ind. Eng. Chem. Res., 36, 1827 (1997).
42. R. E. Rooks, V. Julka, M. F. Doherty, and M. F. Malone, AIChE J., 44(6), 1382
(1998).
43. B. S. Ahmad and P. I. Barton, AIChE J., 42(12), 3419 (1996).
44. D. B. Van Dongen, Distillation of Azeotropic Mixtures: The Application of Simple-
Distillation Theory to Design of Continuous Processes, Ph.D. dissertation, University
of Massachusetts, Amherst, 1983.
45. S. G. Levy, Design of Homogeneous Azeotropic Distillations, Ph.D. dissertation,
University of Massachusetts, Amherst, 1985.
46. S. G. Levy, D. B. Van Dongen, and M. F. Doherty, Ind. Eng. Chem. Fund., 24, 463
(1985).
47. J. D. Olson, Fluid Phase Equil., 185(1–2), 209 (2001).
48. C.-P. Kao, R. N. Miller and J. F. Sturgis, J. Chem. Eng. Data, 46(2), 229 (2001).
49. C.-P. Kao, M. E. Paulaitus, and A. Yokozeki, Fluid Phase Equil., 127(1–2), 191 (1997).
50. S. P. Christensen and J. D. Olson, Fluid Phase Equil., 79, 187 (1992).
51. W. J. Gaw and F. L. Swinton, Nature, 212, 283 (1966).
52. R. Q. Wilson, W. H. Mink, H. P. Munger, and J. W. Clegg, AIChE J., 1, 220 (1955).
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 729

53. J. D. Seader and Z. M. Kurtyka, in R. H. Perry, D. W. Green and J. O. Maloney eds.,


Perry’s Chemical Engineer’s Handbook, 6th ed., McGraw-Hill, New York (1984),
p. 13–58.
54. C. A. Finch, ed., Polyvinyl Alcohol, Properties and Applications, John Wiley & Sons,
New York, 1973.
55. An.-I. Yeh, L. Berg, and K. J. Warren, Chem. Eng. Comm., 68, 69 (1988); U.S. Pat.
4,501,645 (Feb. 26, 1985), L. Berg, and An.-I. Yeh.
56. C. K. Buell and R. G. Boatright, Ind. Eng. Chem., 39(6), 695 (1947).
57. J. Happel, P. W. Cornell, D. Eastman, M. J. Fowle, C. A. Porter, and A. H. Schutte,
Trans. AIChE, 42, 189 (1946).
58. C. Chilton, Chem. Eng., 146 (Feb. 1957); Shell Development Co., Petrol. Refiner,
36(11), 297 (1957).
59. S. Takao, Hydrocarbon Process. Petrol. Refiner, 45(11), 151 (Nov. 1966).
60. W. W. Coogler, Jr., Hydrocarbon Process. Petrol. Refiner, 46(5), 166 (May 1967).
61. H. Kroeper, Hydrocarbon Process. Petrol. Refiner, 41(11), 191 (Nov. 1962).
62. West Ger. Pat. 1,215,688 (May 5, 1966), to Union Rheinische Braunkohlen Kraftstoff.
63. Shell Development Co., Petrol. Refiner, 36(11), 299 (1957).
64. F.-M. Lee, Ind. Eng. Chem. Process Des. Dev., 25, 949 (1986).
65. Brit. Pat. 1,302,325 (1973), E. Mueller, K. H. Eisenlohr, G. Hohfeld, and P. A. John.
66. French Pat. 2,176,488 (Nov. 1973), P. Mikitenko and L. Asselineau.
67. H. G. Drickamer and H. H. Hummel, Trans. AIChE, 41, 631 (1945).
68. R. E. Brown and F.-M. Lee, Hydroc. Proc., 70(5) 83 (1991).
69. U.S. Pat. 4,016,049 (Apr. 5, 1977), G. B. Fozzard and R. A. Paul (to Phillips
Petroleum Co.).
70. U.S. Pat. 4,166,772 (Sept. 4, 1979), T. P. Murta (to Phillips Petroleum Co.).
71. E. R. Hafslund, Chem. Eng. Prog., 65(9) 58 (1969).
72. R. L. Schendel, Chem. Eng., Prog., 80(5), 39 (1984).
73. U.S. Pat. 4,473,444 (Sept. 25, 1984), J. Feldman and J. M. Hoyt (to National Distil-
lers and Chemical Corp.).
74. U.S. Pat. 3,794,568 (Feb. 26, 1974), G. A. Daniels and J. A. Wingate (to Ethyl Corp.).
75. C. Black and D. E. Ditsler, in R. F. Gould, ed., Azeotropic and Extractive Distillation,
Advances in Chemistry Series No. 115, American Chemical Society, Washington,
D.C. 1972, pp. 1–15.
76. C. Black, Chem. Eng. Prog., 76(9), 78 (1980).
77. J. R. Knight and M. F. Doherty, Ind. Eng. Chem. Fund., 28, 564 (1989).
78. J. P. Knapp and M. F. Doherty, AIChE J., 36, 969 (1990).
79. U.S. Pat. 3,804,722 (Apr. 16, 1974), E. D. Oliver (to Montecatini Edison S.p.A.).
80. U.S. Pat. 4,918,204 (Apr. 17, 1990), T. T. Shih and T. Chang (to Arco Chemical
Technology, Inc.).
81. U.S. Pat. 4,665,205 (May 12, 1987), T. Yamada and K. Yamaguchi (to Idemitsu
Petrochemical Co., Ltd.).
82. Brit. Pat. 1,305,866 (Feb. 2, 1973), J. Feldman, F. Lerman, and F. D. Miller (to
National Distillers and Chemical Corp.).
83. U.S. Pat. 5,843,286 (Dec. 1, 1998), P. Roth, E. Leistner and H. Haverkamp (to
Hoechst AG).
84. U.S. Pat. 5,763,712 (June 9, 1998), P. Roth, E. Leistner and W. Wendel (to Hoechst
AG).
85. U.S. Pat. 5,849,938 (Dec. 15, 1998), J. C. Jubin and M. A. Rueter (to Arco Chem
Technology LP).
86. WO Pat. 9,910,302 (Mar. 4, 1999), S. Kohno and T. Shibanuma (to Daikin Ind. Ltd.).
87. WO Pat. 9,907,660 (Feb. 18, 1999), S. W. Boehmer, B. A. Mahler, and R. N. Miller (to
E. I. DuPont de Nemours & Co.).
730 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

88. U.S. Pat. 6,175,045 (Jan. 16, 2001), S. Kohno and T. Shibanuma (to Daikin Ind. Ltd.).
89. U.S. Pat. 5,922,175 (July 13, 1999), R. Bertocchio, E. Deslandes, and E. Lacroix (to
Elf Atochem SA).
90. U.S. Pat. 6,039,845 (Mar. 21, 2000), R. Bertocchio, E. Lacroix, and S. Perdrieux (to
Elf Atochem SA).
91. U.S. Pat. 5,874,657 (Feb. 23, 1999), B. A. Mahler and R. N. Miller (to E. I. DuPont de
Nemours & Co.).
92. U.S. Pat. 4,428,798 (Jan. 31, 1984), D. Zudkevitch, S. E. Belsky, and P. D.
Krautheim (to Allied Corp.).
93. U.S. Pat. 4,455,198 (June 19, 1984), D. Zudkevitch, D. K. Preston, and S. E. Belsky
(to Allied Corp.).
94. E. G. Scheibel, Chem. Eng. Prog., 44(12), 927 (1948).
95. L. Laroche, N. Bekiaris, H. W. Andersen, and M. Morari, Can. J. Chem. Eng., 69,
1309 (1991).
96. J. P. Knapp, Exploiting Pressure Effects in the Distillation of Homogeneous Azeotro-
pic Mixtures, Ph.D. dissertation, University of Massachusetts, Amherst, 1991.
97. B. Kolbe, J. Gmehling, and U. Onken, I. Chem. E. Symp. Ser., (56), 1.3/23 (1979).
98. J. P. Knapp and M. F. Doherty, AIChE J., 40(2), 243 (1994).
99. V. Julka and M. F. Doherty, Chem. Eng. Sci., 48, 1367 (1993).
100. Z. T. Fidkowski, M. F. Malone, and M. F. Doherty, AIChE J., 37, 1761 (1991).
101. S. G. Levy and M. F. Doherty, Ind. Eng. Chem. Fund., 25, 269 (1986).
102. J. R. Knight, Synthesis and Design of Homogeneous Azeotropic Distillation
Sequences, Ph.D. dissertation, University of Massachusetts, Amherst, 1986.
103. J. M. Douglas, Conceptual Design of Chemical Processes, McGraw-Hill, New York,
1988.
104. U.S. Pat. 1,074,287 (Sept. 30, 1913), H. Pauling.
105. C. S. Robinson and E. R. Gilliland, Elements of Fractional Distillation, 4th ed.,
McGraw-Hill, New York, 1950.
106. U.S. Pat. 4,966,276 (Oct. 30, 1990), A. Guenkel.
107. T. Grewer, Chemie. Ing. Techn., 43, 655 (1971).
108. U.S. Pat. 4,076,594 (Feb. 28, 1978), H. Buelow, H. Hohenschutz, J. E. Schmidt, and
W. Sachaze (to BASF Aktiengesellschaft).
109. F.-M. Lee and J. C. Gentry, Chem. Eng. Prog., 93(10), 56 (1997).
110. M. Van Winkle, Distillation, McGraw-Hill, New York, 1967.
111. M. Benedict, C. A. Johnson, E. Solomon, and L. C. Rubin, Trans. AIChE, 41, 371
(1945).
112. F.-M. Lee, Chem. Eng., 105(12), 112 (1998).
113. J. M. Prausnitz and R. Anderson, AIChE J., 7(1), 96 (1961).
114. C. S. Carlson and J. Stewart, in E. S. Perry and A. Weissberger, eds., Techniques of
Organic Chemitry, Vol. IV, Distillation, Wiley-Interscience, New York, 1965.
115. L. Berg, Chem. Eng. Prog., 65(9), 52 (1969).
116. R. H. Ewell, J. M. Harrison, L. Berg., Ind. Eng. Chem., 36(10), 871 (1944).
117. L. A. Robbins, Chem. Eng. Prog., 76(10), 58 (1980).
118. L. A. Robbins, in R. H. Perry, D. W. Green and J. O. Maloney, eds., Perry’s Chemical
Engineer’s Handbook, 6th ed., McGraw-Hill, New York (1984), pp. 15–6 to 15–8.
119. J. Gmehling, R. Wittig, J. Lohmann, and R. Joh, Ind. Eng. Chem. Res., 41, 1678 (2002).
120. J. Gmehling, Fluid Phase Equilib., 144, 37 (1998).
121. F. Eckert and A. Klamt, Ind. Eng. Chem. Res., 40, 2371 (2001).
122. F. Eckert and A. Klamt, AIChE J., 48(2), 369 (2002).
123. M. Schiller and O. Muthu, ‘‘Is There Life after UNIFAC? Past, Present, and Future
of Mixture Phase Equilibria Prediction Methods,’’ Paper 4b, AIChE Spring National
Meeting, New Orleans, March, 2002.
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 731

124. D. Tassios, Chem. Eng., 118 (Feb. 10, 1969).


125. D. P. Tassios, in R. F. Gould, ed., Azeotropic and Extractive Distribution, Advances
in Chemistry Series No. 115, American Chemical Society, Washinton, D.C. 1972,
pp. 46–63; D. P. Tassios, Hydroc. Proc. 49(7), 114 (1970).
126. D. P. Tassios, Ind. Eng. Chem. Proc. Des. Dev., 11(1), 43 (1972).
127. L. Dallinga, M. Schiller, and J. Gmehling, J. Chem. Eng. Data, 38, 147 (1993).
128. J. D. Olson, Fluid Phase Equilib., 52, 209 (1989).
129. M. Krummen, D. Gruber, and J. Gmehling, Ind. Eng. Chem. Res., 39, 2114 (2000).
130. D. B. Trampe and C. A. Eckert, AIChE J. 39, 1045 (1993).
131. D. Zudkevitch, Chem. Eng. Commun., 116, 41 (1992).
132. W. F. Furter, Separation & Purification Methods, 22(1), 1 (1993).
133. W. F. Furter, in R. F. Gould, ed., Azeotropic and Extractive Distillation, Advances
in Chemistry Series No. 115, American Chemical Society, Washington, D.C. 1972.
134. D. Barba, V. Brandini, and G. DiGiacomo, Chem. Eng. Sci., 40, 2287 (1985).
135. U.S. Pat. 4,269,666 (May 26, 1981), C. G. Wysocki (to Standard Oil Co.).
136. D. Schmitt and A. Vogelpohl, Separation Science and Technology, 18(6), 547 (1983).
137. U.S. Pat. 5,135,620 (Aug. 4, 1992), R. E. Brown (to Phillips Petroleum Company).
138. R. T. P. Pinto, M. R. Wolf-Maciel, and L. Lintomen, Comput. Chem. Eng., 24, 1689
(2000).
139. U.S. Pat, 4,874,475 (Oct. 17, 1989), D. F. McLaughlin and R. A. Stoltz (to Westing-
house Electric Corp.).
140. U.S. Pat. 4,021,531 (May 3, 1977), P. Besson, J. Guerin, P. Brun, and M. Bakes
(to Ugine Aciers).
141. J. Fu, AIChE J., 42(12), 3364 (1996).
142. W. F. Furter, Can. J. Chem. Eng., 55(6), 229 (1977).
143. W. F. Furter and R. A. Cook, Int. J. Heat Mass Trans., 10, 23 (1967).
144. W. F. Furter, Chem. Eng. Commun., 116, 35 (1992).
145. J. P. Knapp and M. F. Doherty, Ind. Eng. Chem. Res., 31, 346 (1992).
146. S. I. Abu-Eishah and W. L. Luyben, Ind. Eng. Chem. Proc. Des. Dev., 24, 132 (1985).
147. U.S. Pat. 4,093,633 (June 6, 1978), Y. Tanabe, J. Toriya, M. Sato, and K. Shiraga (to
Mitsubishi Chemical Industries Co., Ltd.).
148. U.S. Pat. 6,323,362 (Nov. 27, 2001), P. Lalanne-Magne, B. Grenouiller, and D. Brun-
Buisson (to BP Chemicals Limited).
149. U.S. Pat. 2,324,255 (July 13, 1943), E. C. Britton, H. S. Nutting, and L. H. Horsley
(to The Dow Chemical Company).
150. T. C. Frank, Chem. Eng. Prog., 93(4), 52 (1997).
151. U.S. Pat. 4,362,603 (Dec. 7, 1982), R. D. Presson, H.-C. Wu, and E. J. Sockell (to the
Standard Oil Company).
152. U.S. Pat. 1,676,700 (July 10, 1928), W. K. Lewis.
153. U.S. Pat. 3,394,056 (July 23, 1968), M. Nadler and R. P. Cahn (to Esso Research and
Engineering).
154. Eur. Pat. App. 1,099,666 A1 (May 16, 2001), T. Suzuta and K. Iwanaga (to Sumitomo
Chemical Company, Ltd.).
155. S. K. Wasylkiewicz, L. C. Kobylka, and F. J. L. Castillo, Hung. J. Ind. Chem., 28(1),
41 (2000).
156. S. Young, J. Chem. Soc., 81, 707 (1902).
157. Ger. Pat. 142,502 (June 25, 1903), S. Young.
158. Ger. Pat. 287,897 (Oct. 11, 1915), Kubierschky.
159. H. M. Guinot and F. W. Clark, Trans. Inst. Chem. Eng., 16, 189 (1938).
160. H. N. Pham and M. F. Doherty, Chem. Eng. Sci., 45, 1823 (1990).
161. W. Reinders and C. H. De Minjer, Recl. Trav. Chim., 66, 564 (1947).
162. W. Reinders and C. H. De Minjer, Recl. Trav. Chim., 66, 573 (1947).
732 DISTILLATION, AZEOTROPIC AND EXTRACTIVE Vol. 1

163. H. Matsuyama, J. Chem. Eng. Jpn., 11, 427 (1978).


164. L. E. Baker, A. C. Pierce, and K. D. Luks, Soc. Petrol. Engrs. J., 22, 731 (1982).
165. M. L. Michelsen, Fluid Phase Equil., 9, 1 (1982).
166. M. L. Michelsen, Fluid Phase Equil., 9, 21 (1982).
167. D. J. Swank and J. C. Mullins, Fluid Phase Equil., 30, 101 (1986).
168. B. P. Cairns and I. A. Furzer, Ind. Eng. Chem. Res., 29, 1364 (1990).
169. C. M. McDonald and C. A. Floudas, AIChE J., 41, 1798 (1995).
170. C. M. McDonald and C. A. Floudas, Computers Chem. Engng., 21, 1 (1997).
171. S. K. Wasylkiewicz, L. N. Sridhar, M. F. Doherty, and M. F. Malone, Ind. Eng.
Chem. Res., 35, 1395 (1996).
172. G. J. Prokopakis and W. D. Seider, AIChE J., 29, 49 (1983).
173. M. Rovaglio and M. F. Doherty, AIChE J., 36, 39 (1990).
174. B. P. Cairns and I. A. Furzer, Ind. Eng. Chem. Res., 29, 1383 (1990).
175. J. M. Prausnitz and co-workers, Computer Calculations for Multicomponent Vapor–
Liquid and Liquid–Liquid Equilibria, Prentice-Hall, Englewood Cliffs, NJ, 1980,
Chaptrt 4.
176. H. N. Pham and M. F. Doherty, Chem. Eng. Sci., 45, 1837 (1990).
177. W. Reinders and C. H. De Minjer, Recl. Trav. Chim., 59, 207 (1940).
178. I. N. Bushmakin and P. Ya. Molodenko, Russ. J. Phys. Chem., 37, 2618 (1964).
179. H. N. Pham and M. F. Doherty, Chem. Eng. Sci., 45, 1837 (1990).
180. R. Y. Urdaneta, J. Bausa, S. Bruggemann, and W. Marquardt, Ind. Eng. Chem. Res.,
41, 3849 (2002).
181. H. N. Pham, P. J. Ryan, and M. F. Doherty, AIChE J., 35, 1585 (1989).
182. G. J. Prokopakis and W. D. Seider, AIChE J., 29, 1017 (1983).
183. J. W. Kovach III and W. D. Seider, AIChE J., 33, 1300 (1987).
184. J. W. Kovach III and W. D. Seider, Comput. Chem. Eng., 11, 593 (1987).
185. B. P. Cairns and I. A. Furzer, Ind. Eng. Chem. Res., 29, 1349 (1990).
186. D. S. H. Wong, S. S. Jang, and C. F. Chang, Comput. Chem. Eng., 15, 325 (1991).
187. S. Widagdo, W. D. Seider, and D. H. Sebastian, AIChE J., 38, 1229 (1992).
188. S. Widagdo, W. D. Seider, and D. H. Sebastian, AIChE J., 35, 1457 (1989).
189. U.S. Pat. 5,980,696 (Nov. 9, 1999), W. D. Parten and A. Macpherson Ure (to E. I.
DuPont de Nemours & Co).
190. S. K. Wasylkiewicz, L. C. Kobylka and F. J. L. Castillo, Chem. Eng. J., 79, 219
(2000).
191. D. B. Keyes, Ind. Eng. Chem., 21, 998 (1929).
192. D. W. Townsend, private communication, 1982.
193. P. J. Ryan and M. F. Doherty, AIChE J., 35, 1592 (1989).
194. H. N. Pham and M. F. Doherty, Chem. Eng. Sci., 45, 1845 (1990).
195. W. S. Norman, Trans. Inst. Chem. Eng., 23, 66 (1945).
196. C. S. Robinson and E. R. Gilliland, Elements of Fractional Distillation, 4th ed.,
McGraw-Hill, New York, 1950, p. 312.
197. Zh. A. Bril’, A. S. Mozzhukhin, F. B. Petlyuk, and L. A. Serafimov, Theor. Found.
Chem. Eng., 9, 761 (1975).
198. Zh. A. Bril’, A. S. Mozzhukhin, F. B. Petlyuk, and L. A. Serafimov, Russ. J. Phys.
Chem., 11, 675 (1977).
199. G. J. Prokopakis, W. D. Seider, and B. A. Ross, in R. S. Mah and W. D. Seider, eds.,
Foundations of Computer-Aided Chemical Process Design, Engineering Foundation,
New York, 1981.
200. C. D. Holland, S. E. Gallun, and M. J. Lockett, Chem. Eng., 88, 185 (1981).
201. M. E. Harrison, Chem. Eng. Prog., 86(11), 80 (1990).
202. V. H. Agreda and J. R. Zoeller (eds.), Acetic Acid and its Derivatives, Marcel Dekker,
N.Y. (1993).
Vol. 1 DISTILLATION, AZEOTROPIC AND EXTRACTIVE 733

203. U.S. Pat. 1,804,745 (1931), H. T. Clarke and D. F. Othmer.


204. D. F. Othmer, Chem. Metall. Eng., 40, 91 (1941).
205. D. F. Othmer and T. O. Wentworth, Ind. Eng. Chem., 32, 1588 (1940).
206. T. O. Wentworth and D. F. Othmer, Trans. Am. Inst. Chem. Eng., 36, 785 (1940).
207. T. O. Wentworth, D. F. Othmer, and G. M. Pohler, Trans. Am. Inst. Chem. Eng., 39,
565 (1943).
208. U.S. Pat. 4,217,178 (Aug. 12, 1980), R. Katzen, G. D. Moon, Jr., and J. D. Kumans
(to Raphael Katzen Associates).
209. S. Lynn and D. N. Hanson, Ind. Eng. Chem. Proc. Des. Dev., 25, 936 (1986).
210. U.S. Pat. 5,035,776 (July 30, 1991), J. P. Knapp and M. F. Doherty.
211. U.S. Pat. 4,740,273 (Apr. 26, 1988), D. L. Martin and P. W. Raynolds (to Eastman
Kodak Co.).
212. J. W. Kovach III and W. D. Seider, J. Chem. Eng. Data, 32, 16 (1988).
213. M. L. Luyben and B. D. Tyreus, Computers Chem. Eng., 22, 867 (1998).
214. U.S. Pat. 3,404,177 (Oct. 1, 1968), T. B. Baba and I. L. Mador (to National Distillers
and Chemical Corp.).
215. U.S. Pat. 3,905,875 (Sept. 16, 1975), W. Kronig and W. Schwerdtel (to Bayer Aktien-
gesellschaft).
216. J. J. Siirola, pp. 222–233 in L. T. Biegler and M. F. Doherty eds., Foundations of
Computer-Aided Process Design, AIChE Symp. Series, Vol. 91, American Institute
of Chemical Engineers, N.Y. (1995).
217. J. J. Siirola, Adv. Chem. Eng., 23, 1 (1996).
218. M. Minotti, M. F. Doherty, and M. F. Malone, Trans. IChemE., 76A, 361 (1998).
219. M. Rovaglio, T. Faravelli, G. Biardi, P. Gaffuri, and S. Soccol, Comput. Chem. Eng.,
17,(5–6), 535 (1993).
220. E. W. Jacobsen, L. Laroche, M. Morari, S. Skogestad, and H. W. Andersen, AIChE J.,
37(12), 1810 (1991).
221. A. Zheng, V. Grassi, and G. Meski, Ind. Eng. Chem. Res., 37, 1836 (1998).

GENERAL REFERENCES

J. D. Seader, J. J. Siirola, and S. D. Barnicki, in R. H. Perry, D. W. Green and J. O.


Maloney eds., Perry’s Chemical Engineer’s Handbook, 7th ed., McGraw-Hill,
New York, 1997, Section 13, pp. 54–81.
M. F. Doherty and M. F. Malone, Conceptual Design of Distillation Systems, McGraw-Hill,
New York, 2001.
E. J. Hoffman, Azeotropic and Extractive Distillation, Interscience Publishers, New York,
1964.

MICHAEL F. DOHERTY
University of Massachusetts, Amherst
JEFFREY P. KNAPP
E. I. du Pont de Nemours & Co., Inc.
734 SEPARATIONS PROCESS SYNTHESIS Vol. 1

SEPARATIONS
PROCESS SYNTHESIS
1. Introduction

Chemical process and plant design involves a hierarchy of complex and creative
activities including both routine and innovative design. Routine design is largely
analytical and is primarily concerned with determining values for the specifica-
tion and operation of specific units, eg, reactors, extractors, and distillation col-
umns. By contrast, process synthesis, the generation of conceptual flow sheets
comprising such units, is a more open-ended activity characterized by a com-
binatorially large number of feasible alternatives. Finding better flow sheet
alternatives has a significant impact on overall process competitiveness. This
is particularly true for separations process synthesis, the selection of separation
methods, their interconnection, and their operating parameters. Virtually every
chemical process involves the recovery, isolation, or purification of products,
by-products, intermediates, wastes, or raw materials. The separation systems
to accomplish these tasks often dominate the capital and operating costs of che-
mical manufacturing processes.
Several approaches to the separations process synthesis problem have been
formulated including superstructure optimization, evolutionary modification, and
systematic generation. From a known feed composition, desired product composi-
tions, and a well-defined set of separation methods, superstructure optimization
approaches construct a hypothetical flow sheet that includes all applicable separa-
tion methods interconnected in every possible order so as to include all possible
separations scheme alternatives. Separations synthesis then becomes a problem
of systematically stripping away the less desirable parts of this superstructure
while simultaneously optimizing the design and operating parameters of the
remaining separation methods using mixed-integer nonlinear programming (1).
This purely mathematical approach is computationally intensive, but is becoming
well-established for certain classes of distillation synthesis problems.
Evolutionary modification starts with an existing flow sheet for a similar
separation to which adaptations are made as necessary to meet the objectives
of the specific case at hand. This approach is exemplified by standard flow
sheet patterns, eg, the schemes for breaking heterogeneous minimum-boiling
binary azeotropes or the sequences for extractive distillation (see DISTILLATION,
AZEOTROPIC AND EXTRACTIVE) (2–4). Although rarely resulting in novel designs, evo-
lutionary modification is a frequently used separations synthesis technique
because of the extensive existing repertoire of design heuristics, standard pat-
terns, and encyclopedias of complete flow sheets.
In the systematic generation approach, the separations flow sheet is con-
structed from a portfolio of basic components in a directed fashion so that a given
feed stream is progressively transformed into one or more target compositions.
This process can be viewed as the solution of integrated equipment selection, sequ-
encing, and specification problems. Resource constraints generally limit the num-
ber of separation process alternatives that may be generated and evaluated to a
small fraction of the total number of alternatives that are theoretically possible.

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.1905160102011814.a01.pub2
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 735

As a compromise between thoroughness, efficiency of evaluation, and guar-


antee of optimality, selection and sequencing methods often are reduced to
design heuristics and simple ranked lists of physical and chemical properties
characteristic of specific separation methods (5,6). This article will illustrate sys-
tematic generation separations synthesis strategies for nearly ideal liquid mix-
tures, as well as quite nonideal liquid mixtures. Strategies involving thermally
coupled distillations and reactive separations are also presented.

2. Heuristic Distillation Sequencing for Nonazeotropic Mixtures

2.1. Synthesis Methods for Simple Distillation Systems. One prac-


tical systematic generation problem is the separation of nearly ideal liquid
mixtures by simple distillation. By assuming that distillation is to be used for
all separations, the separation method selection issue is eliminated. Moreover,
if the liquid-phase nonidealities are not severe and the relative volatilities of
the components of a liquid mixture are reasonably constant, then knowledge of
the boiling points of the components provides sufficient thermodynamic infor-
mation. The components are ranked in order of increasing boiling point and a
split is required when two components adjacent in the ranked list, ie, keys of
the separation, are desired to be in different products. Given the additional
assumption of sharp separations, ie, high purity and high recovery, the distillate
will consist of the light key and lower boiling components, whereas the bottoms
will consist of the heavy key and higher boiling components.
Once the components are ranked by boiling point, it is a relatively straight-
forward task to enumerate exhaustively all possible distillation sequences. Each
sequence can then be simulated in detail and the most cost-effective sequence
selected by comparison. However, this search approach to separation flow
sheet generation makes no use of past process knowledge nor gives any indica-
tion of the relative value of any of the sequences without significant computa-
tional effort. The use of detailed distillation simulation is a higher level of
analysis than is required for screening many of the alternative sequences.
Some screening can be accomplished more efficiently with heuristic-based argu-
ments, but with the sacrifice of guaranteed optimality.
Heuristics are reliable, well-established rules for reducing the number of
potential alternative sequences with minimum effort, and often lead to near opti-
mal separation system designs. Most of the heuristics for distillation sequencing
were originally formulated from parametric studies. A number of heuristics have
been suggested, some of which contradict each other (7–10). A representative list
follows:

Two Product Heuristic (No. 1). If there are only two products in a mixture, and
all of the components in one product are more volatile than all of the com-
ponents in the other product, then the next separation should divide the
components into two pure products.
Process Hazard Heuristic (No. 2). If a component is corrosive, unstable, reac-
tive, or otherwise hazardous, then the next separation should work toward
removing the component from the bulk process stream.
736 SEPARATIONS PROCESS SYNTHESIS Vol. 1

Direct Sequence Heuristic (No. 3). If the most volatile component (product) in
the mixture is greater than 20 mol% of the feed, and the most volatile
component (product) has the largest mole fraction in the feed, and the com-
ponent (product) with the second largest mole fraction is present in a ratio
of less than 90% of the most volatile component (product), and the proposed
separation is one of the easiest separations remaining, then the next separa-
tion should split off the most volatile component (product) as distillate.
Indirect Sequence Heuristic (No. 4). If the least volatile component (product)
in the mixture is greater than 20 mol% of the feed, and the least volatile
component (product) has the largest mole fraction in the feed, and the com-
ponent (product) with the second largest mole fraction is present in a ratio
of less than 90% of the least volatile component (product), and the pro-
posed separation is one of the easiest separations remaining, then the
next separation should split off the least volatile component (product) as
bottoms.
Distillate to Bottoms Ratio Heuristic (No. 5). If a separation has a distillate to
bottoms ratio (molar basis) of 40:60–60:40, and the proposed separation
is one of the easiest separations remaining, then the next separation should
be the one closest to a 50:50 distillate to bottoms ratio.
Easiest Next Heuristic (No. 6). If none of the other heuristics apply, then the
next separation to be done should be the easiest, that is the one with the
highest separation coefficient, S, defined as

S ¼ ða  1ÞðD=BÞ for D=B < 1

S ¼ ða  1ÞðB=DÞ for D=B > 1

where a is the ratio of volatilities of adjacent components, D is the molar


flow rate of the distillate, and B is the molar flow rate of the bottoms.
Rules for Determining Difficulty of Distillation. If a  1.5, then the given se-
paration is one of the easiest separations remaining.
If 1.1  a  1.5 and there are no separations with a > 1.5, then the given
separation is one of the easiest separations remaining.
If a < 1.1 and there are no separations with a  1.1 then the given separa-
tion is one of the easiest separations remaining.

Heuristic methods have also been extended to sequencing nonsharp distil-


lation separations and to combinations of distillation, mixing, and stream bypass
operations (11–13).
A heuristic approach is an example of an opportunistic systematic genera-
tion strategy. At any point in the flow-sheet synthesis procedure, the partial
design generated is a feasible consequence of the initial feed composition and
the separation methods so far specified. The resolution of composition differ-
ences between the intermediate streams and remaining product goals are
addressed by specifying additional separation methods. No attempt is made
to look ahead to anticipate or accommodate any potential difficulties in separa-
tion problem resolution. Because distillation is assumed to be feasible for all
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 737

necessary separations, this opportunistic strategy will not lead to interme-


diate streams from which it is impossible to resolve the remaining composition
differences.
2.2. Thermally Coupled Sequences. Simple heuristic methods gene-
rate conventional distillation sequences, comprising rectifying–stripping col-
umns each with a reboiler and a condenser, for each of n  1 separations. Such
sequences, even when optimized can have high capital and energy-related
operating costs. One method of reducing vapor traffic, potentially column dia-
meters, number of pieces of equipment, and energy costs is to thermally couple
sections of columns. A practical and widely practiced application of thermally
coupled distillation sequences is the separation of air into oxygen, nitrogen,
argon, and sometimes helium.
In its simplest form, thermal coupling involves using vapor from a higher
pressure rectifying section to boil the underflow from a lower pressure stripping
section. By using thermally coupled side stripper or rectifier sections, the vapor
and liquid flows can be shared with another column section (rectifying section
with a side stripper, stripping section with a side rectifier). Thus one condenser
can provide the condensing duty for both the rectifying section and the
side stripper. Similarly, one reboiler can provide boil-up for coupled stripping
and side rectifier sections. Generalizing, only one condenser or one reboiler is
required for any two column sections associated with a component of inter-
mediate volatility.
However, there is a limit to the number of condensers and reboilers that can
be exchanged for additional column sections. A condenser is always required
for condensation of the most volatile component of a mixture and a reboiler for
boil-up of the least volatile component. With only one condenser and reboiler, the
separation of an n-component mixture into n pure products requires a minimum
of 4n  6 column sections (14).
Figure 1 illustrates three of many possible flow sheets for separating a four-
component mixture, ranging from a conventional direct distillation sequence
to a fully integrated one-condenser/one-reboiler configuration. The conventional

Fig. 1. Four Component Separation Sequence: (a) noncoupled direct sequence, (b) ther-
mally coupled with 8 sections and 3 reboilers and condensers, and (c) thermally coupled
with 10 sections and 2 reboilers and condensers.
738 SEPARATIONS PROCESS SYNTHESIS Vol. 1

sequence (Fig. 1a) requires 3 columns with a total of 6 column sections and 6
reboilers and condensers. The partially coupled sequence (Fig. 1b) has halved
the number of reboilers and condensers, but increased the number of column sec-
tions to 8. The fully coupled sequence (Fig. 1c) requires 10 column sections with
the minimum 1 reboiler and 1 condenser. Each of these configurations and the
many other possible variations involve capital, energy usage, and operability
trade-offs that are difficult to know a priori. Often highly coupled sequences suf-
fer from difficult control issues and have seen limited application in industry.
Heuristics have been proposed to help generate practical distillation sequence
superstructures (14–16). The synthesis of thermally coupled distillation column
sequences is a significant step up in design complexity over conventional distilla-
tion sequencing. However, sophisticated heuristic/logic and MINLP techniques are
now able to solve robustly and reliably conventional to fully thermally coupled dis-
tillation sequencing problems in acceptable computation times (17,18).

3. Separations Synthesis for Nonideal Liquid Mixtures

3.1. Problem Representation. Typical liquid mixtures encountered


in organic chemicals manufacturing often exhibit a wide range of melting and
boiling points, reactivity, temperature sensitivity, strong thermodynamic non-
idealities resulting in azeotropism, and liquid–liquid-phase formation: in general,
a diverse range of behaviors that tend to complicate separation operations. A
portfolio of separation methods has been developed over the years to deal with
these behaviors, including, eg, simple distillation, azeotropic distillation, dual-
feed extractive distillation, decantation, liquid–liquid extraction, various forms
of crystallization (qv), adsorption (qv), and membrane permeation (see EXTRACTION,
LIQUID–LIQUID; MEMBRANE TECHNOLOGY). Simple single-feed distillation is most widely
used because of predictable, reliable, flexible, robust, and efficient operation, and
because of mature equilibrium-based design techniques that do not require exten-
sive piloting. Furthermore, simple distillation is one of the few methods that
requires only the input of energy to effect the separation. Other common liquid
separation methods including extraction, azeotropic distillation, extractive distil-
lation, and solution crystallization, require the introduction of an additional mass
separating agent (MSA). The MSA must then be recovered and recycled for eco-
nomical operation, adding further complexity to the separations system design.
The generation of separations schemes for liquid mixtures can be thought of as
a problem of finding applications for distillation and the identification and resolu-
tion of situations where distillation cannot be used.
An important aspect of separations process synthesis is effective problem
representation and visualization. The most widely used separation methods for
liquid mixtures, including variants of distillation and extraction, are equilibrium-
based processes for which the pertinent thermodynamics can be represented con-
veniently with graphical methods. Triangular phase diagrams (three-component
systems) and tetrahedral diagrams (four components) are useful tools for visua-
lizing separation method behavior including material balances. Useful thermo-
dynamic representations are residue curve maps (RCM) and distillation region
diagrams (DRD) for vapor–liquid equilibria (VLE), miscibility diagrams for
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 739

liquid–liquid equilibria (LLE) (see DISTILLATION, AZEOTROPIC AND EXTRACTIVE;


EXTRACTION, LIQUID–LIQUID), and solubility diagrams for visualizing solid–liquid
equilibria (SLE) (see CRYSTALLIZATION). While these representations are useful
for predicting the behavior of equilibrium-based separations, they do not predict
kinetic separation methods, eg, adsorption and membrane permeation processes,
although the material balances for these separation methods still can be repre-
sented graphically.
The dominance of distillation-based methods for the separation of liquid
mixtures makes a number of points about RCM and DRD significant. Residue
curves trace the liquid-phase composition of a simple single-stage batch stillpot
as a function of time. Residue curves also approximate the liquid composition
profiles in continuous staged or packed distillation columns operating at infinite
reflux and reboil ratios, and are also indicative of many aspects of the behavior of
continuous columns operating at practical reflux ratios (19).
The family of all residue curves that originate at one composition and ter-
minate at another composition defines a RCM region. All systems with no azeo-
tropes and even some systems with azeotropes have only one region, the entire
composition space. All residue curves originate at the lowest boiling composition
of the system and terminate at the highest boiling composition. However, other
systems in which not all residue curves originate or terminate at the same com-
position have more that one region. The demarcation between regions in which
adjacent residue curves originate from different compositions or terminate at
different compositions is called a separatrix. Separatrices are related to the exis-
tence of azeotropes. In the composition space for a binary system, the separatrix
is a point (azeotropic composition); for three components, the separatrix gener-
ally becomes a curved line; for four components a surface, etc.
All pure components and azeotropes in a system lie on region boundaries.
Within each region, the most volatile composition (either a pure component or a
minimum-boiling azeotrope and the origin of all residue curves) is the low boiling
node (also called the unstable node). The least volatile composition (either a pure
component or a maximum-boiling azeotrope and the terminus of all residue
curves) is the high boiling (or stable) node. All other pure components and azeo-
tropes are called intermediate-boiling saddles because no residue curves origi-
nate, terminate, or quite pass through these compositions. Adjacent regions
may share nodes and saddles. Pure components and azeotropes are labeled as
nodes and saddles as a result of the boiling points of all of the components and
azeotropes in the system. If one component is removed from a mixture, the label-
ing of all remaining pure components and azeotropes, specifically those that were
saddles, may change. Region-defining separatrices always originate or terminate
at nodes or saddle azeotropes, but never at saddle pure components (20).
To a first approximation, the composition of the distillate and bottoms of a
single-feed continuous distillation column lie on the same residue curve. There-
fore, for systems with separatrices and multiple regions, distillation composition
profiles are also constrained to lie in specific regions. The precise boundaries of
these distillation regions are a function of reflux ratio, but they are closely
approximated by the RCM separatrices. If a separatrix exists in a system, a cor-
responding distillation boundary will also exist. Also, mass balance constraints
require that the distillate composition, the bottoms composition, and the net feed
740 SEPARATIONS PROCESS SYNTHESIS Vol. 1

Fig. 2. Fourteen most common DRD.

composition plotted on an RCM for any feasible distillation be collinear and


spaced in relation to distillate and bottoms flows according to the well-known
lever rule.
The pattern of boundaries, nodes, and saddles of a given multicomponent
system is related to the boiling points of the pure components and azeotropes
and is readily definable mathematically. A thorough review of the thermody-
namic principles behind residue curve maps is presented by Kiva and co-workers
(21). Although 125 distinct RCM or DRD are possible for three-component
systems, the 14 shown in Fig. 2 are the only common maps (22). A RCM can
be constructed from experimental data (for many common systems) or calcu-
lated if an equation of state or an activity coefficient expression is available
(eg, Wilson parameters or UNIFAC groups). However, considerable information
on system behavior can still be deduced from semiqualitative sketches based
only on pure component and azeotrope boiling point data and if available
approximate azeotrope compositions. This data can be used to construct DRD
by the method presented in Perry’s Chemical Engineering Handbook (2).
For a given multicomponent mixture, a single-feed distillation column can
be designed with sufficient stages, reflux, and material balance control to pro-
duce a variety of different separations ranging from the direct mode of operation
(pure low boiling node taken as distillate) to the indirect mode (pure high boiling
node taken as bottoms). This range of operability results in a bow-tie shaped set
of reachable compositions that may be opportunistically achieved for single-feed
distillation roughly bounded by the material balance lines corresponding to
the sharpest direct separation and the sharpest indirect separation possible.
The exact shape of the reachable composition space is further limited by the
requirement that the distillate and bottoms lie on the same residue curve, and some-
times further by peculiarities in the shape of the residue curves (23). Since
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 741

MIPK MIPK
95.3°C 95.3°C

B2 B2

F2 F2
B1
A2 A2
80.9°C 80.9°C D2

B1

D2 F1 F1
D3
D3
B3 B3 F3
F3
D1

100°C D1 79.6°C 100°C 79.6°C

Water A1 MEK Water A1 MEK


73.6°C 73.6°C
(a) (b)

Fig. 3. Methyl ethyl ketone–MIPK–water system, where A1 and A2 represent two dif-
ferent azeotropes; F1, F2, and F3, different feed compositions; Bn and Dn the correspond-
ing bottoms and distillates, respectively; (—), the distillation boundary; and (X), the
reachable compositions for the various feeds: (a) approximate bow-tie and (b) exact reach-
able compositions.

residue curves are deflected by saddles, it is generally not possible to obtain a


saddle product (pure component or azeotrope) from a simple single-feed column.
For preliminary design purposes, the lowest and highest boiling nodes and the
compositions on the distillation region boundary directly opposite the feed com-
position from these two nodes are of particular interest.
Figure 3 illustrates the situation for the system methylethylketone (MEK),
methylisopropylketone (MIPK), and water, and the problem of recovering a pure
MIPK product from such mixtures. The bow-tie approximation of reachable com-
positions for several feeds is shown in Fig. 3a; the exact reachable compositions
are shown in Fig. 3b.
For feed F1, the upper edge of the reachable composition region is along the
MEK–MIPK (water-free) face of the composition triangle and part of the lower
edge is along the MEK–water (MIPK-free) face. There exist conditions under
which both the water in the bottoms MIPK product can be driven to low levels
(high product purity) while the MIPK in the distillate is also driven to low levels
(high product recovery), although achieving such operation depends on having
adequate stages and reflux ratio.
The reachable composition region for feed F2 is significantly different with
the upper edge along the water–MIPK (MEK-free) face of the triangle and the
lower edge along the distillation boundary. From this feed, it is not possible to
achieve a high purity MIPK specification while simultaneously obtaining high
MIPK recovery. If the column is operated to get high purity MIPK, the material
balance line is constrained by the distillation boundary. Alternatively, if the
column is operated to obtain a high recovery of MIPK (by removing the MEK–
water azeotrope as the distillate), the material balance requires the bottoms to
lie on the water–MIPK face of the triangle (low purity). From feed F3, which
742 SEPARATIONS PROCESS SYNTHESIS Vol. 1

is situated in a different distillation region than the desired product, pure MIPK
cannot be obtained by simple single-feed distillation at all.
Phase diagrams are particularly effective tools for separation synthesis
work because of their ability to combine various types of thermodynamic infor-
mation onto one representation including boiling points of pure components
and azeotropes, location of azeotropes and any VLE-based distillation bound-
aries, location of LLE binodal curves and tie lines, melting points, eutectic
points, and SLE phase compositions. Many of these thermodynamic features
have a significant impact on the separation system flowsheet. Figure 4 shows
examples of combined vapor–liquid and liquid–liquid phase diagrams for several

MIPK IPA
95.3°C 82.2°C
Node Node Distillation
Distillation region II
region I
Distillation
Distillation A2 region I
A2 region II 80.4°C
80.9°C Saddle
Saddle

100.0°C 79.6°C 100.0°C 39.7°C


Node Saddle Node Saddle

Water A1 MEK Water A1 Methylene


73.6°C 37.9°C chloride
Node Node

(a) (b)

IPA

82.3°C Distillation
Node region II
Distillation
region I
A2
78.7°C
Saddle

A4
61.6°C
Saddle

100.0°C A3 68.8°C
Node 58.7°C Node Saddle

Water A1 Hexane
61.5°C
Saddle

(c)

Fig. 4. Combined residue curve and phase equilibria diagrams for the systems where
A1-4 represent azeotropes, (—) is the distillation boundary, the shaded areas represent
the region of two liquid phases, and (– – –) are tie lines: (a) MEK–MIPK–water;
(b) methylene chloride–IPA–water; and (c) hexane–IPA–water.
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 743

ternary systems. In each of these diagrams, a number of special points, lines, and
regions have been identified. Figure 4a shows the two-phase liquid region for the
MEK–MIPK–water system previously mentioned, while Fig. 4b illustrates the
methylene chloride–2-propanol–water system. Both systems have a distillation
boundary spanned by liquid–liquid tie lines. The n-hexane–isopropyl alcohol
(IPA)–water system is more complicated (Fig. 4c), exhibiting three distillation
regions as well as a two-phase liquid region (24).
Relatively few points or composition regions in the phase diagram are of
particular significance in separations process synthesis. These compositions of
interest include the feed and desired product compositions, as well as azeotropes,
eutectics, and selected points on liquid–liquid binodal curves. The choice of
composition for a mass separation agent, if required, is critical and usually is a
composition that can be conveniently regenerated in the process. Binary and tern-
ary azeotropes that are also high or low boiling nodes (as well as the corresponding
two liquid-phase compositions, if heterogeneous) are of particular interest in this
regard as these points are reachable compositions by distillation and decantation.
Pure components and azeotropes that are saddles are by contrast poor choices for
MSA recycle as they are not included in the set of reachable compositions, unless
they become nodes when another component in the system is deleted.
Although distillation is a favored separation method, quite a number of
situations prevent or interfere with the use of simple direct or indirect mode
distillation schemes. In general, the designer will be faced with avoiding, over-
coming, or exploiting a limited set of critical features in order to accomplish
the overall objective of the separation system. These critical features include
(1) distillation boundaries, ie, if product (or MSA) composition is in a different
distillation region from the feed mixture, the product cannot be obtained directly
by simple single-feed distillation; (2) saddle products, ie, if a product (or MSA) is
a saddle in a particular distillation region, that product cannot be obtained at
high purity directly by simple single-feed distillation; (3) pinched or close-boiling
regions, ie, if a feed and product are separated by a region of low relative vola-
tility, simple single-feed distillation is not precluded, but tends to require a
large number of stages and high reflux ratio; (4) overlapping melting–boiling
points, ie, some solutions may contain components with melting points that
are higher that the boiling points of other components and distillation of
such mixtures may result in solidification within the column; (5) temperature-
sensitive components, ie, some mixtures may contain components that degrade,
decompose, polymerize, or otherwise react in an undesirable manner at the tem-
perature conditions of distillation, thus some milder separation method must be
used (25).
3.2. Flowsheet Construction. A purely opportunistic approach to
separation synthesis for nonideal mixtures may again be considered. Guided
by the general sequencing heuristics given, any separation method may be
picked that is feasible and applicable to the state under consideration, and build-
ing the solution toward the desired products may proceed. However, the exis-
tence of the critical features mentioned previously can preclude the application
of simple distillation. In systems exhibiting critical features, it is quite possible
to reach an intermediate composition from which it is impossible to reduce
remaining composition differences by distillation or any other separation
744 SEPARATIONS PROCESS SYNTHESIS Vol. 1

method. The only solution is either to backtrack to an earlier intermediate com-


position, discarding part or all of the current flow sheet, or to follow an alterna-
tive synthesis strategy.
The purely opportunistic approach may not make use of all thermody-
namic or physical property information known for the whole system. An alter-
native approach to separation flowsheet generation is to look ahead to potential
difficulties and develop contingencies early in the synthesis process to deal with
them before running into dead ends. Once a critical feature has been identified,
it is useful to examine strategic methods for crossing, breaking, by-passing,
reaching, or exploiting the critical feature. The strategies and resulting separa-
tion methods for handling a given critical feature may differ considerably and
the same strategy can often be implemented in several different fashions. For
example, both decantation and extraction are methods for implementing the
strategy of exploiting LLE tie-lines to cross a VLE distillation boundary.
Table 1 lists strategies and implementations associated with several different
critical features.
Early in the synthesis of separation schemes for nonideal liquid mixtures, it
may not be known exactly where in the flow sheet a strategic separation will end
up, only that it will be required someplace in some form in order to overcome or
exploit a particular critical feature. Thus, strategic separations often initially do
not have well-defined feeds and products. The region where the feed must be
located may be known, as well as a general idea of the types of products expected,
but no definite compositions for either. For example, for a strategic decant opera-
tion, all that may be known initially is that the feed must be somewhere in the
two-phase liquid region and the products will be on the binodal curve at opposite
ends of a tie line through the feed possibly specified to be in two different distil-
lation regions.
Along with the forward-looking strategic approach, several opportunistic
separations may also be possible. These include (1) distill overhead the low boil-
ing node of a mixture, with the bottoms composition constrained by mass balance
and region boundary, ie, follow a direct distillation sequence; (2) distill the
high boiling node of a mixture as underflow, ie, follow an indirect distillation
sequence; (3) decant a two-phase liquid mixture. When critical features are pre-
sent, opportunistic operations can be thought of as links between feeds, strategic
separations, and products. An opportunistic operation often is used to reach a
composition where a strategic separation is applicable (eg, opportunistically dis-
till into a two-phase liquid region, then strategically decant the mixture to cross
a VLE boundary). Sometimes a opportunistic separation will be equivalent to a
strategic separation. Alternatively, if no critical features are present, the entire
separations flow sheet can be synthesized heuristically by a series of opportunis-
tic separations alone as discussed previously.
3.3. Separation Synthesis Algorithm. The procedure for the systema-
tic generation of conceptual separation system flow sheets consists of the follow-
ing eight steps

Step No. 1: DEFINE PROBLEM


Determine feed specifications - construct a list of all sources for process (source
list will consist initially of all feeds).
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 745

Table 1. Strategic Separations


Separation
Strategy implementation Comments
Pinched Regions
use existing liquid–liquid decanter By-passes pinched VLE
region or add component liquid–liquid
that causes new or altered extraction
liquid–liquid region
exploit solid–liquid melt crystallization driving force for melt cryst. is
equilibrium to adsorption a melting point difference
circumvent– of impurities and desired
cross-pinch product. Feed must meet
certain melting point and
crystal characteristics for
feasibility (see also
solution crystallization)
change system pressure distillation feasibility highly dependent
to alter–eliminate on extent that pinch is
pinch altered by moderate
change in pressure
exploit solid–liquid solution based on differences in
solubility crystallization melting points and liquid-
differences phase solubilities. Four
modes of operation
possible: drown out,
isothermal evaporation,
adiabatic evaporation,
cooling. Choice depends
on stream characteristics
exploit kinetic phenomena molecular sieve requires the selection
to circumvent–cross-pinch adsorption of solid-phase mass
membrane separation agent
permeation
pervaporation
use VLE to circumvent– distillation choose third component
cross-pinch. Approach azeotropic which does not form
pinched region from distillation pinch with one of the
nonpinched direction key components
distill through pinched simple distillation process may require a
region large number of stages
and a high reflux ratio
use existing component extractive extractive agent modifies
or add new MSA to distillation liquid-phase behavior
by-pass pinch in a (activity coefficients) of key
two-feed column components. Residue curve
map must be of appropriate
form for extractive
distillation to work
Saddle Products
use existing liquid–liquid decanter a binodal composition is
region or add component liquid–liquid often not pure enough for
that causes new or altered extraction the final product. It may
liquid–liquid region still be difficult to
perform the additional
purification
746 SEPARATIONS PROCESS SYNTHESIS Vol. 1

Table 1. (Continued )
Separation
Strategy implementation Comments
exploit solid–liquid melt crystallization driving force for melt cryst. is
equilibrium to reach adsorption a melting point difference
saddle of impurities and desired
product. Feed must meet
certain melting point and
crystal characteristics for
feasibility (see also solu-
tion crystallization)
use existing component or extractive extractive agent modifies
add new MSA to reach distillation liquid-phase behavior
saddle in a single two-feed (activity coefficients) of key
column components. Residue curve
maps must be of appropriate
form for extractive
distillation to work
exploit solid–liquid solubility solution based on differences in
differences to reach saddle crystallization melting points and liquid
phase solubilities. Four
modes of operation
possible: drown out,
isothermal evaporation,
adiabatic evaporation,
cooling. Choice depends
on stream characteristics
reduce or expand the azeotropic a saddle in a three-component
dimensionality of problem distillation system becomes a node in a
to turn saddle into a node; distillation two-component
move to a face of the RCM system. Add-
ing the appropriate
component can alter the
dist. region boundaries
and turn a saddle into
a node
exploit kinetic phenomena adsorption requires the selection
to reach saddle membrane of solid-phase agent
permeation
pervaporation
react to turn saddle into node reaction reactions involve components
reactive already in mixture
distillation
Overcoming Distillation Boundaries
use existing liquid–liquid decanter uses LLE to avoid
region or add component liquid–liquid boundary in VLE
that causes new or altered extraction
liquid–liquid region
exploit solid–liquid melt crystallization driving force for melt cryst. is
equilibrium adsorption a melting point difference
of impurities and desired
product. Feed must meet
certain melting point and
crystal characteristics for
feasibility (see also
solution crystallization)
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 747

Table 1. (Continued )
Separation
Strategy implementation Comments
mix streams to mixer may generate infeasible
cross-boundary solutions due to mass
balance violations. Only
works if there is a way to
get a composition that is
purer than the desired
mixture product (farther
than the middle point of
the balance)
exploit existing convex distillation curvature is difficult to
curvature or add predict accurately
component that causes without experimental
favorable convex curvature data. Boundary must be
to cross boundary highly curved to avoid
large recycles
change system pressure to distillation feasibility highly dependent
alter–eliminate boundary on extent that azeotrope
composition is altered
by moderate change in
pressure
exploit solid–liquid solubility solution based on differences in
differences crystallization melting points and liquid
phase solubilities. Four
modes of operation
possible: drown out,
isothermal evaporation,
adiabatic evaporation,
cooling. Choice depends
on stream characteristics
exploit kinetic phenomena to molecular sieve requires the selection of
cross-boundary adsorption solid-phase agent
membrane
permeation
pervaporation
Vapor–Solid–Liquid Equilibria
use intermediate reflux of sidedraw-reflux occurs when boiling and
low melting point column melting points of
component to prevent components overlap
freezing of high melter
exploit solid–liquid melt crystallization driving force for melt cryst.
equilibrium to separate is a melting point
overlapping difference of impurities
melters–boilers and desired product. Feed
must meet certain
melting point and crystal
characteristics for
feasibility (see also
solution crystallization)
add a component to strip stripping
out lighter components
to separate overlapping
melters–boilers
748 SEPARATIONS PROCESS SYNTHESIS Vol. 1

Table 1. (Continued )
Separation
Strategy implementation Comments
exploit solid–liquid solubility solution based on differences in
differences to separate crystallization melting points and liquid
overlapping melters– phase solubilities. Four
boilers modes of operation
possible: drown out,
isothermal evaporation,
adiabatic evaporation,
cooling. Choice depends
on stream characteristics
Temperature Sensitive Mixtures
use existing liquid–liquid decanter perform separation at
region or add component liquid–liquid conditions below
that causes new or altered extraction reaction–degradation
liquid–liquid region temperature: usually
can be accomplished
at temperatures
between 20 and 458C
exploit solid–liquid melt crystallization driving force for melt cryst.
equilibrium adsorption is a melting point
difference of impurities
and desired product.
Feed must meet certain
melting point and crystal
characteristics for
feasibility (see also
solution crystallization)
alter system pressure vacuum distillation feasibility highly dependent
to decrease reboiler wiped film on pressure required to
temperature evaporation prevent degradation and
short-path stage requirements of
evaporation desired separation
exploit solid–liquid solution based on differences in
solubility differences crystallization melting points and liquid
phase solubilities. Four
modes of operation
possible: drown out,
isothermal evaporation,
adiabatic evaporation,
cooling. Choice depends
on stream characteristics
exploit kinetic phenomena molecular sieve requires the selection
adsorption mem- of solid-phase agent
brane permeation

Determine product specifications - construct a list of all destinations of pro-


cess components (list will consist initially of compositions and amounts
of all desired products and by-products to be produced by the separa-
tions process).
Determine identity of any additional species to be used as a MSA.
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 749

DECISION POINTS:
Product specifications
Identity of MSA species

Step No. 2: EXAMINE EQUILIBRIUM DATA


Examine VLE, LLE, and SLE as needed.
Determine how each varies with temperature, pressure, composition, etc.
Construct a residue curve map (with LLE, SLE overlayed).
List compositions of interest. (potential MSA compositions).
Select a composition specification for the MSA (if there is an MSA). Add
MSA composition to list of destinations.
DECISION POINTS:
MSA composition

Step No. 3: CHOOSE A STREAM TO PROCESS


Choose a stream from the source list to process. If source list contains more
than one stream, this selection may be arbitrary. All streams will even-
tually be processed, but order of processing may have some effect on the
structure of the flowsheet synthesized.
Prepare list of known critical features which must be dealt with to reach all
destinations (products as well as MSA compositions).
DECISION POINTS:
Selection of current stream to process

Step No. 4: PROCESS STREAM


Consult Table 3 Recycle Heuristics, and Table 2 Rules for Selecting Opera-
tions. Do one of the following three things with the current stream:
1. If the composition matches that of a destination, label the stream as a
product. Make note of the fact that this destination has been reached
at least once.
OR
2. Recycle the stream. Consult Table 3 for guidance.
OR
3. Specify an operation (from the list of applicable opportunistic and stra-
tegic operations).
(a) Determine if any of the previously identified critical features pertain
to the current stream.
(b) Examine Table 1 (corresponding to the critical features identified
above), and list all strategic operations which are applicable to the
current stream.
(c) List all opportunistic separations that are applicable to the current
stream (see Table 1).
(d) Consult Table 2 to determine which strategic or opportunistic
operation should be tried first.
750 SEPARATIONS PROCESS SYNTHESIS Vol. 1

Table 2. Rules for Selecting Among Potential Operations


Tier 1 Once all the potential opportunistic and strategic operations have been
identified for the current stream, then the next step is to determined the
order that these operations should be tried. Use the following heuristics
to order alternatives at a decision point:

1. Select a strategic separation that reaches a product (or MSA) composition.


2. Select an opportunistic separation that reaches a product (or MSA)
composition.
3. Select a strategic operation that does not reach a product.
4. Select an opportunistic operation that reaches the feed composition of a
strategic operation.
5. Select any remaining opportunistic separation.

Tier 2 When more than one separation is feasible:

1. Perform hardest separations last.


2. Favor separations that remove corrosive, hazardous components as a
product directly (ie, no further separation required for that stream).
3. Favor separations that remove the largest fraction of mixture as a product
directly.
4. Favor separations that give approximately equal-sized output streams.

DECISION POINTS:
Selection of destination for current stream
Selection of recycle point for current stream
Selection of operation to be applied to current stream

Table 3. Recycle Heuristics


1. Consider recycling a stream when the separation(s) to be performed on that stream
would duplicate a section of the process that has been previously specified. This will
generally occur when the composition of a stream is very similar to the composition
of a previous stream in the process.
2. Avoid recycle to a point that would allow infinite build-up of any component, ie, if there is
no outlet for each component downstream of a recycle, then do not recycle to that point.
3. Similarly to item No. 2, if a stream is the last stream on the source list, and all
products have not been obtained, recycle is not an option.
4. Match stream compositions as much as possible.
5. Avoid recycle of streams that cause major changes in the operation of upstream units,
eg, do not recycle a stream at a point that would cause an upstream composition to
move into another distillation region.
6. If compositions do not match, and item No. 5 is not violated, then it is permissible to
recycle a stream that is small compared to an upstream flow rate.
7. If all product specs have been met and all critical features dealt with, and there are still
streams on the source list, then consider recycling remaining streams.
8. Do not add a new MSA if all products and MSA compositions of interest have been reached
and there are still streams on the source list: consider recycling remaining streams.
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 751

Step No. 5: UPDATE FLOWSHEET


Connect streams.
Update flowsheet.
Add any newly created streams to source list (ie, streams resulting from
separations just added to flowsheet).
Add required feed stream resulting from strategic operations just specified
to destination list.
Update worksheet (stream, possible moves, chosen move, decision point).

DECISION POINTS:
NONE

Step No. 6: OTHER STREAMS TO PROCESS?


If the source list is not empty, then the flowsheet is not complete. Addi-
tional streams must be processed.
RETURN TO STEP No. 3

Step No. 7: ALL PRODUCT (INCLUDING MSA AND STRATEGIC


OPERATION FEED) SPECS MET?
The solution is complete if all product specifications have been met.
If the flowsheet cannot meet the product specs as originally stated. Deter-
mine if they can be ‘‘relaxed’’ - are the products specs that can be
obtained with the current flowsheet actually satisfactory? If yes, then
the solution is complete.

DECISION POINTS:
All problem specs met?

Step No. 8: GENERATE ALTERNATIVE SOLUTIONS


Once an acceptable flowsheet has been generated, consider evolutionary
modification.
If another solution is desired, then return to any/all of the decision points
and choose different alternatives.

The synthesis scheme emphasizes the use of the appropriate knowledge,


identification of critical features and strategies to deal with them, and provides
a means for selecting and sequencing both opportunistic and strategic separa-
tions. The process begins by the construction of lists of sources and destinations
for the process. A source may be the original feed mixture or a stream created by
a strategic or opportunistic separation. A source list is maintained as the algo-
rithm progresses. It contains streams that have not been identified as destina-
tions, have not been recycled, nor fed into another unit operation. Destinations
are the final and sometimes intermediate objectives of the separation flow sheet.
They may be products, by-products, MSA compositions that must be regenerated
and recycled, or the feed to a strategic separation. The destination list also
changes as the design proceeds.
752 SEPARATIONS PROCESS SYNTHESIS Vol. 1

The thermodynamics and physical properties of the mixture to be separated


are examined. VLE nodes and saddles, LLE binodal curves, etc, are labeled. Cri-
tical features and compositions of interest are identified. A stream is selected
from the source list. This stream is either identified as meeting all the composi-
tion objectives of a destination, or else as in need of further processing. Once an
opportunistic or strategic operation is selected and incorporated into the flow-
sheet, any new sources or destinations are added to the respective lists. If a stra-
tegic separation for dealing with a particular critical feature has been
implemented, then that critical feature is no longer of concern. Alternatively,
additional critical features may arise through the addition of new components
such as a MSA. The process is repeated until the source list is empty and all des-
tination specifications have been satisfied.
Separation method selection is governed by a two-tier set of rules in Table 2.
The first tier involves the selection between strategic and opportunistic opera-
tions. Strategic separations are favored, as these are known to be required in
the flowsheet at some point, as are separations that directly reach a desired pro-
duct composition. The second tier uses the same general sequencing heuristics
outlined previously to help guide the process toward lower cost sequences
when more than one alternative is feasible.
The recycling of a stream to a point upstream in the process is often a
powerful alternative to performing additional processing. In particular, regen-
eration and recycle of an MSA composition is essential to the economic operation
of such separation methods as extraction, extractive distillation, and azeotropic
distillation. However, recycling cannot be done indiscriminately and material
balances must be carefully considered. Unit operations that worked before clos-
ing the recycle may no longer function in the same manner, or may become infea-
sible. For example, the recycle may cause an upstream composition to move into
another distillation region making a previously specified distillation now impos-
sible. Alternatively, the recycle may result in the infinite build-up of a particular
component if there is no outlet for that component from the recycle loop. Table 3
presents recycle heuristics for guidance.
3.4. Separation Selection Issues. Separations process synthesis is an
open-ended design activity. Often very distinct flow sheets can be generated if
the choice of MSA is changed, a different separation method is applied to a par-
ticular stream, or even by changing the order in which streams on the source list
are examined. Such decision points are found throughout the algorithm and are
the mechanism for generating alternative flow sheets. Each decision point can be
revisited, another alternative selected, and the section of the flow sheet resulting
from this decision point can be revised. Potential decision points for each step of
the algorithm are noted in the Section 3.3.
It is particularly difficult to give rules for finding the ‘‘best’’ separation
method among several feasible alternatives for nonideal mixtures. Ranked
lists of physical properties may not have enough discriminating power without
resorting to more in-depth calculations. However, knowledge of a few proper-
ties characteristic of each separation often can help eliminate feasible, but poor
choices, as well as indicate better alternatives clearly. Table 4 lists pure
component and mixture properties that characterize several liquid separa-
tion methods. Section in 3.5 present a list of general separation heuristics
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 753

Table 4. Characteristic Properties for Liquid Separation Methods


Separation method Properties
Distillation-Based Methods
distillation relative volatility
D/B ratio of mixture
residue curve map or distillation region diagram
extractive distillation data, model for liquid-phase behavior
residue curve map or distillation region diagram
azeotropic distillation residue curve map or distillation region diagram
pressure-swing distillation effect of pressure on azeotrope composition
type of azeotrope (max or min boiling)
distillation into curved boundaries detailed, accurate residue curve map
high vacuum distillation effect of pressure on boiling points
relative volatility
Crystallization Methods
adiabatic evaporation effect of temperature on solubility
boiling point of solvent
heat of vaporization of solvent
isothermal evaporation effect of temperature on solubility
boiling point of solvent
heat of vaporization of solvent
cooling effect of temperature on solubility
eutectic diagram
drown-out or salt-out solubility in solvent
melt crystallization melting points
eutectic diagram
metastable zone width
Other MSA Methods
steam stripping miscibility with water
azeotrope formation with water
reactivity with water
extraction data, model for liquid-phase behavior
selectivity in solvent
adsorption kinetic diameter
hydrogen bonding characteristics
pervaporation chemical families

for distillation-based separation methods, crystallization methods, and other


methods requiring an MSA. Sources of information on MSA selection is included
in the list of references in the tables.

I. General Separation Heuristics: Distillation-Based Methods


1. Distillation (2,26,27)
(a) Usually cannot be beaten when a (relative volatility)  1.5.
Competitive for 1.1  a  1.5. Other methods usually better for a
 1.05 to 1.1.
(b) Becomes difficult (likely to require large number of stages or reflux)
when temperatures of high boiling and low boiling nodes of region
differ by < 108C.
754 SEPARATIONS PROCESS SYNTHESIS Vol. 1

(c) Very difficult when temperatures of high and low boiling nodes of
region differ by < 58C.
(d) Simple one-column distillation is generally unfavorable when
the bottoms to distillate ratio is either very high or very low
(eg, 95/5  D/B ratio  5/95). For low D/B ratio high reflux
required. For high D/B ratio large amount of feed must be
vaporized.
(e) Consider special multiple-effect distillation schemes if the bottoms
to distillate ratio is either very high or very low, or if the relative
volatility is low.
(f) Avoid attempts to recover simultaneously both high and low boiling
nodes in high purity from mixtures of three or more components,
particularly in columns that reflux compositions different from
the distillate composition (ie, reflux of one phase from a decanter),
as such operation may be difficult to control.
2. Azeotropic Distillation (2) (see DISTILLATION, AZEOTROPIC AND EXTRACTIVE)
(a) The more structurally, chemically similar components are, the
less likely that the separation will be improved by azeotropic dis-
tillation (if an MSA-key component azeotrope is being used to al-
ter the RCM). Any azeotropes formed between one component and
another similar component will tend to have similar boiling points,
compositions.
(b) MSA selection is critical.
(c) See 8a1-6 above.
3. Extractive Distillation (2) (see DISTILLATION, AZEOTROPIC AND EXTRACTIVE)
(a) Usually not favorable for separation of components that show
very similar liquid-phase behavior: steroisomers, homologous series,
isonormal–neoisomers. Components to be separated must have
some different functional group for MSA to affect liquid-phase beha-
vior differentially.
(b) Only certain residue curve maps are favorable for extractive
distillation.
(c) MSA selection is critical.
4. Pressure Swing Distillation (2,26) (see DISTILLATION, AZEOTROPIC AND
EXTRACTIVE)

(a) Azeotrope composition must change at least 5–10 mol% over mod-
erate pressure change for process to be economical. Generally lar-
ger change required for minimum boiling azeotrope than maximum
boiling azeotrope. Extent of pressure change somewhat dictated by
reboiler temperature in high pressure column, condensing tem-
perature in low pressure column (try to avoid refrigeration, special
heat exchange fluids).
(b) For minimum boiling azeotrope, generally less favorable be-
cause recycle composition between columns taken overhead,
pure products are bottoms: results in high energy usage, larger
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 755

column size, products will contain any high boiling contami-


nants.
(c) For maximum boiling azeotrope, generally more favorable because
recycle composition between columns are bottoms streams, pure
products are distillates: recycle not as energy-intensive, products
are distilled once.
5. Distillation into Curved Boundary (2) (see DISTILLATION, AZEOTROPIC AND
EXTRACTIVE)

(a) More curved the boundary, the less recycle required. Generally less
favorable for minimum boiling azeotropes due to large overhead
recycles.
(b) Consider other methods if recycle rates are high (ie, three to five
times feed rate or more).
(c) Boundaries involving maximum azeotropes often highly curved.
(d) Not applicable to binary systems; azeotrope is point in phase
diagram.
6. High Vacuum Distillation (28,29)
(a) Consider high vacuum distillation when material is temperature
sensitive at normal pressure ranges.
(b) Consider wiped-film evaporator (external condenser) when pres-
sure must be in 5–0.1-Torr range.
(c) Consider short-path evaporator (internal condenser) when pressure
must be in 0.1–0.001-Torr range. For larger scale production, 0.01–
0.005 Torr is practical lower pressure limit.
(d) At best, only one equilibrium stage achievable for each WFE or SPE
unit. Separation efficiency decreases as pressure decreases.
(e) Operate at a vacuum no lower than required. Short-path unit typi-
cally 25–35% higher cost than wiped-film unit.
(f) Separate all noncondensables before feeding to WFE or SPE unit.
Even a small amount of noncondensables will overload vacuum sys-
tem, especially at ultrahigh vacuum ranges. Remember that most
low molecular weight compounds will not condense at cooling water
temperatures under high vacuum.
II. General Separation Heuristics: Crystallization Methods (30–32)
1. Solution Crystallization: General
(a) Cost increases greatly with decreasing temperature. Try to keep at
or above 0–208C.
(b) Order of preference: adiabatic evaporation > evaporation > cooling
> drown-out.
(c) If system forms solid solution then multistage crystallization re-
quired. Likely to be expensive option.
2. Solution Crystallization: Adiabatic Evaporation (Vacuum Cooling)
(a) Favored for systems where temperature change has moderate
effect on solubility.
756 SEPARATIONS PROCESS SYNTHESIS Vol. 1

(b) Generally best when crystallizing component is 20–75% of feed


mixture.
(c) Solvent must have sufficient volatility to be easily removed at tar-
get temperature and vacuum levels.
(d) May be inferior for systems with high boiling point elevations (high
vacuum levels required).
(e) Components must be stable at flash point of mixture. Will probably
run under vacuum. Vacuum < 50–100 Torr may require specialized
(scraped-film) crystallizer, more complicated vacuum system.
(f) If running under vacuum, separate noncondensables before feeding
to crystallizer. Excessive noncondensables will overload vacuum
system or lead to oversized design.
(g) Complete recovery possible in theory. In practice recovery dictated
by solubility limit of impurities and viscosity of solution.
3. Solution Crystallization: Isothermal Evaporation
(a) Favored for systems where solubility has little temperature depen-
dence. Can be used in other cases.
(b) Solvent to be evaporated must have high volatility compared to
remainder of solution.
(c) Theoretical yield is 100%, but usually limited by cocrystallization of
impurities.
(d) Use single effect if crystallizing component is greater than 75% of
feed.
(e) Consider multiple effects if crystallizing component is <75% of feed.
(f) Can be very energy intensive. Energy usage per pound of product
highly dependent on heat of vaporization of solvent and number
of effects used.
(g) Components must be stable at boiling point of mixture. May have to
run under vacuum. Vacuum <50–100 Torr may require specialized
(scraped- film) crystallizer, more complicated vacuum system.
(h) If running under vacuum, separate noncondensables before feeding
to crystallizer. Excessive noncondensables will overload vacuum
system or lead to oversized design.
(i) Complete recovery possible in theory. In practice, recovery dictated
by solubility limit of impurities and viscosity of solution.
4. Solution Crystallization: Cooling
(a) Generally best when crystallizing component is 20–75% of feed
mixture.
(b) Solubility should be strong function of temperature. Solubility
should decrease with decreasing temperature.
(c) Feed should be close to saturation limit before cooling to maximize
potential recovery. May consider preconcentration step to remove
excess solvent.
(d) Recovery limited by eutectics.
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 757

5. Solution Crystallization: Drown-Out or Salt-Out


(a) Consider when solvent cannot be removed by evaporation (either
temperature sensitivity or low solvent volatility).
(b) For drown-out, look for MSA in which the component to be
crystallized has low solubility, while other components have
high solubility. MSA should be completely miscible with feed
mixture.
(c) For salt-out, add solute that reduces effective solubility of component
to be crystallized. Salt with common ion usual choice. Salting-out
compound should be cheap, as may be difficult to recover.
(d) Generally best when crystallizing component is large percentage of
feed. Consider preconcentration step if dilute.
6. Melt Crystallization
(a) Mixture must be stable at temperatures 208C above melting point
of crystallizing component.
(b) Crystallizing component should be at least 50% of feed mixture,
preferably 75% or more.
(c) Melting points must be between 30 and 1508C for practical imple-
mentation.
(d) Crystallizing component should show sharp melting point.
(e) Metastable zone width should be <258C for efficient crystallization.
(f) Large difference in melting point and eutectic temperature im-
proves controllability.
(g) Difference of 20–308C between melting points of pure components
more favorable.
(h) Recovery limited by eutectics.
(i) Cost increases greatly with decreasing temperature. Try to keep at
or > 0–208C.
III. General Separation Heuristics: Other Methods Requiring MSA
1. Stripping (steam) (33)
(a) Particularly good for components that form minimum boiling
azeotropes with water and are immiscible with water: tend to
have extremely high volatilities.
(b) Not good for low boilers that are completely miscible with water
(eg, methanol, acetone).
(c) Generally uneconomical if component to be stripped is >10–15% of
feed mixture (too much steam required). Should use other methods
to remove bulk of component.
(d) Be prepared to expect some product contamination if feed compo-
nents can react with water (eg, ester will be partially hydrolyzed
to acid and alcohol). Fate of reaction product species will depend
on above rules (eg, methanol from methyl ester hydrolysis probably
will not be stripped out of bottoms stream).
758 SEPARATIONS PROCESS SYNTHESIS Vol. 1

2. Extraction (26,34,35) (see EXTRACTION, LIQUID–LIQUID)


(a) Usually not favorable for separation of components that show very
similar liquid-phase behavior.
(b) MSA selection is critical.
(c) Consider component already in mixture for MSA if system exhibits
appropriate liquid–liquid behavior.
(d) For given binary pair two-phase behavior likely when infinite dilu-
tion activity coefficient of either component in other component is
greater than or equal to 7.5.
(e) For trace removals (eg, catalyst recovery) usually can get to <100 ppm,
but not much lower. Often difficult to regenerate solvent to such
low levels.
3. Adsorption (liquid phase) (26)
(a) Species can only be separated by molecular sieving effects when
their kinetic diameters fall into different zeolite aperture size
categories. Standard molecular sieve diameters are nominally 3,
4, 5, 8, 10, 13 Å.
(b) Dehydration of organics (removal of <1% water) generally feasible
by molecular sieving, if kinetic diameter of organic >3 Å.
(c) Selectivity generally unfavorable when system components are
miscible in all proportions (may not be a problem when removal
is the objective).
(d) Sparingly soluble minor components are more favorably adsorbed.
(e) Strong hydrogen bonding of component with like or unlike mole-
cules tends to retard adsorption.
(f) Only for dilute solutions (<1–2% of feed), unless molecular sieving
effect also present.
(g) Activated carbon adsorption generally uneconomical for removal of
>1000 ppm contaminant from large stream unless willing and able
to regenerate bed. Steaming often easiest regeneration method, but
creates new waste water problem. Usually 3–5-lb steam required
per pound of carbon for regeneration.
(h) Do not regenerate molecular sieves by steaming. Water typically is
very strongly adsorbed and may not be easily displaced by adsor-
bent in next adsorption cycle.
(i) Resin adsorbents (macroreticular polymer resins) generally good for
removal of up to 1–2% of stream. Often regenerable.
4. Membrane Permeation and Pervaporation (26)
(a) Membranes becoming more widely available for aqueous-organic
separations. Some successful industrial applications reported for
dehydrations of alcohols (ethanol and above), removal of alcohols
from water.
(b) Not good for services where solids, tars are expected. Stream
should be distilled once before coming in contact with
membrane.
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 759

3.5. Solution Evolution. It is often beneficial to reexamine a completed


flow sheet and look for opportunities for simplification and consolidation of unit
operations. A complicated series of unit operations can sometimes be replaced
by a simpler structure that has equivalent material balance lines. When two
or more sections of the flow sheet perform similar functions, ie, both produce
the same product using the same or similar unit operations, one section often
can be eliminated by recycling the stream to the input of the remaining section.
An MSA contaminated by other components in the mixture will often function
as effectively as a pure MSA, without the need for additional purification
operations.
Binary System. Example: Separation of 2-Propanol–Water. Consider
the separation of a binary mixture of 60 mol% water and 40 mol% IPA into two
products consisting of IPA of 99.5 mol% purity and water with <100 ppm impurity
of IPA.
The first task is to examine the characteristics of the IPA–water phase
equilibria (VLE, LLE, SLE) to determine the compositions of interest and any
critical features. Isopropyl alcohol forms a minimum boiling azeotrope with
water (80.48C at 760 Torr, 68 mol% IPA). The azeotrope is between the feed
and the IPA product and is a distillation boundary. Thus, it is impossible to
obtain both desired products from any single-feed distillation column.
Table 1 contains strategic separations to be considered for crossing distilla-
tion boundaries. Many of these can be eliminated after examining the pertinent
physical properties and equilibrium behavior (Table 4) and referring to the gen-
eral separation heuristics in Sections 3.5.1–3.5.3. The results are summarized in
Table 5.
For a binary system the two possible opportunistic distillations are essentially
identical and can be combined to concentrate the feed up to 68% IPA and pro-
duce pure water (Table 6). The feed and products of the two possible strategic sepa-
rations are still ill-defined, while those of the opportunistic separation are known.
Because a preconcentration step will probably be required to make the
final sequence more economical, it is logical to start with the opportunistic
separation. This separation produces one of the products, pure water, as the
underflow and a concentrated distillate appropriate for feed into either strategic
separation. Arbitrarily choosing pervaporation first, the retentate has a composi-
tion on the IPA-rich side of the azeotrope, while the permeate is pure water. No
further strategic separations are required.
Addition of another opportunistic distillation to the flow sheet results in
the production of the IPA product (bottoms) and approximately the azeotrope
composition out the top. The distillate is not a desired product and must be
dealt with in some fashion. It is similar in composition to several other streams
in the process, so recycle (rather than further processing in additional equip-
ment) is considered. Referring to Table 3, recycling directly back into the
same column is unwise as water builds up. Recycle to the front of the process
entails revaporization of this whole fraction with additional water and
increases the membrane load. Recycle to the membrane inlet is probably
best. Although the membrane load is increased, the fraction does not need to
be distilled overhead again. The final dual-distillation, membrane permea-
tion sequence is shown in Fig. 5. A similar sequence can be constructed
760 SEPARATIONS PROCESS SYNTHESIS Vol. 1

Table 5. Strategic Separations for Crossing Distillation Boundaries: IPA–Water


Strategy Heuristic Status Comments
use existing III.2.(c) extraction, water and 2-propanol
liquid–liquid region decantation miscible in all
or add component infeasible with proportions
that causes new or current system no liquid–liquid region
altered liquid–liquid in binary system
region

III.2.(a) may be feasible components show


with addition of different liquid phase
suitable MSA behavior. Exploitation
of LLE possible with
appropriate MSA
exploit solid– II.6.(i) melt crystallization possibly feasible, but
liquid crystallization freezing points of water
equilibrium probably inferior and 2-propanol (273
and
184.7 K, respectively)
relatively low
III.3.(c) adsorption strongly hydrogen-
infeasible bonded,
miscible system.
Adsorption probably
unfavorable
mix streams mixing infeasible binary azeotropes cannot
to cross- be broken in practice
boundary by simple mixing.
Would either require
infinite recycle or
mixing of IPA that is
purer than the desired
product
exploit existing I.5.(d) infeasible no convex curvature to
curvature alternative exploit since azeotrope
or add component that is point in phase
causes favorable diagram
curvature to
cross boundary
may be feasible
with addition of
suitable third
component
change system I.4.(a) pressure-swing azeotropic composition
pressure to distillation varies only slightly
alter–eliminate infeasible with pressure: 13 vol%
boundary water at 95 Torr;
12 vol% water at
760 Torr, 11.7 vol%
water at 3087 Torr
exploit solid–liquid II.1.(a) solution low temperatures
solubility differences crystallization required (well < 08C)
probably inferior
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 761

Table 5. (Continued )
Strategy Heuristic Status Comments
exploit kinetic III.3.(a) molecular sieve difference in kinetic
phenomena to adsorption diameters (water 2.65
cross-boundary feasible with Å; IPA 4.2 Å) acceptable
preconcentration for 3-A mol sieves.
Large amount of water
to adsorb; preconcen-
tration step probably
necessary to keep bed
size reasonable
III.4.(a) membrane dehydration of IPA
permeation known application
feasible with of pervaporation
preconcentration (eg, cellulose acetate
membranes). Large
amount of water to pass
through membrane,
preconcentration step
probably necessary to
keep membrane area
reasonable

using molecular sieve adsorption as the strategic separation in place of per-


vaporation. Both sequences involve trade-offs between water removal in
the membrane or adsorption bed, recycle rate, and isopropanol production
rate. Such an analysis is best carried out at the next level of detail via process
simulation.
Strategies Requiring Introduction of a Third Component. Although the
original binary system does not exhibit liquid–liquid behavior, Strategy (Table 5)
suggests the addition of a mass separating agent to the system in order to cause
the formation of a liquid–liquid region. A number of desirable characteristics for
the MSA can be identified. The solvent must be partially miscible with either IPA
or water, or both. Hydrocarbons, ethers, higher ketones, and halogenated com-
pounds are usually partially miscible with water. Moreover, the liquid–liquid region
must be situated to allow for crossing the azeotrope or any distillation boundaries
formed. Because the MSA must be recovered at some point in the process, it should

Table 6. Opportunistic Distillations, Binary System


Operation Unit OPS Feed(s) Products
opportunistic fractionator feed mixture: D - IPA/water azeotrope
separations from feed 40% IPA B - water (target product)
strategic operations molecular sieve undefined IPA/ adsorbed - water
from feed adsorption water mix raffinate - mixture enriched
in IPA
membrane permeate - water
permeation raffinate - mix enriched
in IPA
762 SEPARATIONS PROCESS SYNTHESIS Vol. 1

IPA–water IPA–water IPA–water


A A A
IPA–water
A
Feed
Feed Water permeate 60%
Mixture IPA
60%
IPA on IPA-rich
side of azeotrope Water Water IPA
Water product desorbed product
IPA
product Mixture on IPA-rich
product
side of azeotrope
(a) (b)

Fig. 5. Distillation of aqueous 2-propanol (IPA), where A is the azeotrope, combined with
(a) pervaporation and (b) adsorption.

have an appreciable boiling point difference from both water and IPA. Further infor-
mation on solvent selection for liquid–liquid extraction is available (34,35).
Different MSAs may lead to completely different separation systems
designs. The systematic generation procedure is demonstrated for two potential
solvents, hexane and methylene chloride.
Water–IPA–Hexane System. The residue curve map for the ternary
system exhibits three distillation regions and a two-liquid-phase region (Fig. 6).
Each pure component is the high boiling node in a different distillation region.
Thus, from the feed mixture a distillation boundary must be crossed in order
to obtain pure IPA. The compositions of interest in this diagram include the
feed mixture (60% water, 40% IPA), the two products (in different distillation
regions), and potential MSA compositions, eg, pure hexane (high boiling node
in region III), the ternary azeotrope (low boiling node in regions I, II, III), and
points on the binodal curve, particularly tie-lines passing through azeotropes.
Since the binary azeotropes are saddle points (and thus hard to regenerate by
distillation), they are probably not good compositions to pick for an MSA. With
the limited information available at this point, pure hexane is a possible MSA.
This is a decision point. Another MSA composition may lead to a different flow
sheet. Moreover, if may be found that pure hexane is not required.
Without revisiting the kinetic-based methods already examined for the
binary case, several strategic operations can be used to cross the distillation
boundary between region I and II including Mixer 1, Extractor 2, and Decanter
3. The decantation strategy could be implemented either with a single feed,
which must be in the liquid–liquid region, or by adding an MSA, which puts
the overall mixture in the liquid–liquid region.
The feed compositions and products of each of these strategic separations
remain ill-defined. The unspecified IPA–water mixture, the input to each stra-
tegic separation, could be, but is not necessarily the original feed composition.
The MSA composition (pure hexane in this case) is such that one of the products
of the strategic separation is in region II (ie, the strategic separation crosses
the distillation boundary). Two opportunistic distillations from the original
feed mixture, Fractionators 4 and 5, are also possible. Since the feed is binary,
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 763

IPA
B9
Region
II

Azeo D13 Hexane


D4,D5
B8

D11 D5 D13 1 D9 O7 D11


Feed 7
Azeo M1 Feed
5 13 M1 9 A7 11
B12 B10 Region
A7 D9 Azeo O7 III
B4,
Region I D8,D10,D12 B11 B5 B13 B9 B11
B5,B13
Water Water IPA
Water Azeo Hexane
(a) (b)

IPA
B9
Region
II

Azeo D5

D5 1 D9 O7
Feed 7
M1 Azeo Feed M1
D9 5 9 A7
Region
A7 Region I Azeo O7
III
B5 B5 B9

Water Azeo Hexane Water IPA

(c) (d)

Fig. 6. Isopropyl alcohol dehydration where Azeo is azeotrope, (– – –) are tie lines,
and the shaded areas represent the region of two liquid phases. Initial dehydration
(a) residue curve map and (b) flow sheet, and evolved IPA dehydration (c) residue curve
map and (d) flow sheet. The  in b and d represent a two-phase decanter. (See text; see
Table 7.)

these separations are identical and are the same as the preconcentrator pre-
viously discussed. (See Table 7).
Following step 4 of the algorithm and consulting Table 2, the first opera-
tion should be Fractionator 5, an opportunistic separation that reaches a
product. None of the strategic separations is able to reach a product composi-
tion. Application of Fractionator 5 to the Feed produces the desired water
product as underflow and a water–IPA mixture (D5) as distillate. Stream D5
requires further processing to obtain the IPA product. At this point no other
opportunistic separations can be applied, but stream D5 is a possible input to
any of the three previously identified strategic operations. Arbitrarily picking
the mixing strategy, sufficient hexane must be added to D5 to produce a compo-
sition in region II.
764 SEPARATIONS PROCESS SYNTHESIS Vol. 1

Table 7. Strategic Separations for Crossing Distillation Boundaries: Water–IPA–Hexane


Operation Unit OPS Feed(s) Products
strategic operations mixer 1 undefined IPA/ M1 - mixture in
for reaching water mix MSA region II
region II
extractor 2 undefined IPA/ E2 - extract in
water mix MSA region II
R2 - raffinate
in region I
decanter 3 undefined IPA/ O3 - organic phase
water/MSA mix in region II
A3 - aqueous phase
in region I
opportunistic fractionator 4 feed mixture: D4 - IPA/water
separations 40% IPA azeotrope
from feed
B4 - &water
fractionator 5 feed mixture: D5 - IPA þ water
40% IPA
B5 - water (target
product)
strategic mixer 6 undefined mixture M6 - mixture in
operations for in region II region III
reaching
region III
decanter 7 undefined mixture O7 - organic phase
in region II in region III
A7 - aqueous phase
in region I
opportunistic fractionator 8 M1 D8 - ternary
separations azeotrope
from mixture
M1
B8 - IPA þ hexane, or
IPA þ water
fractionator 9 M1 D9 - hexane þ IPA þ
water
B9 - IPA (target
product)
opportunistic fractionator 10 O7 D10 - ternary
separations azeotrope
from decant
phase O7
B10 - hexane þ IPA
fractionator 11 O7 D11 - hexane/IPA
azeo þ water
B11 - hexane
(target MSA)
opportunistic fractionator 12 A7 D12 - ternary
separations from azeotrope
decant phase A7
B12 - water þ IPA
fractionator 13 A7 D13 - IPA/water
azeo þ hexane
B13 - water
(target product)
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 765

The strategy for boundary crossing has now been implemented. However,
by the addition of the hexane another critical feature has been created. Hexane
must be regenerated, but it is in a different distillation region than the only
remaining unprocessed stream (M1). In this case, the possible boundary crossing
strategic operations are Mixer 6 and Decanter 7. Two opportunistic distillations,
Fractionators 8 and 9, can also be applied to M1 (note decantation is also a pos-
sible opportunistic separation).
Again referring to Table 2, the next separation should be Fractionator 9,
which produces the IPA product directly. If the distillation is driven far enough
stream D9 will be in the two-phase region and will be an appropriate input into
the strategic separation Decant 7. Decanter 7 yields phases in regions I and III.
Alternatively, D9 could be mixed with a sufficient amount of some stream
(unknown at this time) in region III to bring the overall mixture into region
III. Either strategy could be selected, but the decant is probably better as it
can be accomplished without the addition of another stream. The organic
phase of the decanter, O7, is in region III and needs to be further purified
to obtain pure hexane. Two opportunistic separations, Fractionators 10 and 11,
are possible.
Selection of Fractionator 11 gives pure hexane, which can be recycled
to Mixer 1. The distillate D11, however, is a problem. It cannot be distilled
because of its location next to a distillation boundary. It is outside of the two
phase region, so it cannot be decanted. In essence, no further separations are
possible. However, using the rules of Table 3, it can be mixed into the MSA
recycle stream without changing the operation of Mixer 1 appreciably. However,
as both outlet streams are mixed together, Fractionator 11 is not really needed.
The mixture of hexane and IPA, O7, could have been used as the MSA composi-
tion in the first place.
The final loose end in the process is the aqueous decanter product, A7. The
hexane must be removed before the mixture can be sent to waste water
treatment (ie, accepted as a water by-product). Two opportunistic separations,
Fractionators 12 and 13, are possible. Selection of Fractionator 13 gives pure
water underflow, and a distillate similar to D5. Distillate D13 can be recycled
back and mixed with D5 without affecting the operation of Mixer 1. All streams
have now been processed and the flowsheet produces both desired products
(Fig. 6b).
Although the flowsheet is complete, it does contain some redundant separa-
tion that adds to the capital cost. With the basic structure defined, the next step
is to consider some evolutionary improvements. First, Fractionator 13 performs
essentially the same job as Fractionator 5, ie, concentrating a mixture of water
and IPA. Fractionator 13 can be taken out and A7 can be recycled directly to the
feed of Fractionator 5. Although the flowrate to Fractionator 5 increases, its basic
separation function does not change. As the bubble point of A7 is somewhat
higher than that of Feed F, A7 is probably best introduced as a separate feed
lower in the column. Second, using pure hexane as the MSA was unnecessary.
Fractionator 11 can be eliminated and O7 used as the MSA composition. Finally,
considering the bubble point and amount of O7, operations Mixer 1, Fractionator
9, and Decanter 7 can be rearranged as a distillation-decanter combination with
identical overall material balance lines and function. The evolved flowsheet is
766 SEPARATIONS PROCESS SYNTHESIS Vol. 1

shown in Fig. 6c and d. Other sequences can be generated by choosing different


alternatives at the various decision points.
Methylene Chloride Alternatives. When the solvent is changed to
methylene chloride the residue curve map contains two distillation regions
and a liquid–liquid region (Fig. 4b). Pure methylene chloride is a saddle in dis-
tillation region II and therefore is probably not a good choice for the MSA
composition. Rather, the compositions of interest are the water–methylene
chloride azeotrope (high boiling node), points on the binodal curve, and IPA-
free mixtures of methylene chloride and water with compositions between
those of the binodal curve and the methylene chloride–water azeotrope. The
strategic operations are the same as with the hexane case (decant, extract, or
mix) and again, two opportunistic fractionations are possible (preconcentration
distillations).
In one possible sequence, the MSA composition is chosen as water saturated
methylene chloride expected to be regenerated by decantation. The boundary-
crossing strategic operation is to mix the feed with the MSA. The resulting
two-phase mixture is opportunistically fractionated to produce the IPA product
as bottoms, a mixture of water–methylene chloride as distillate. This distillate is
opportunisticly decanted to recover water-saturated methylene chloride MSA for
recycle. The aqueous decanter phase is the water product, which optionally may
be further purified by stripping (Fig. 7).
If an opportunistic preconcentration of the feed is used instead, an entirely
different flowsheet results. In this case the MSA composition is a two-phase
mixture of methylene chloride and water. Detailed simulations are required to
determine which of these (or other) IPA dehydration flowsheet alternatives is
the economically advantaged process.

4. Flowsheet Generation for Reactive Systems

4.1. Problem Representation. The integration of chemical reaction


with separation offers the potential of significant capital and operating cost
reductions. Consequently, many different types of reactive separations have
been propounded including reactive distillation, extraction, absorption, crys-
tallization, and adsorption (36,37). As with nonreactive systems, effective
graphical techniques have been developed to help visualize the superposition
of complex chemical and phase equilibrium. When the number of components
of a reactive system is >3, it becomes more difficult to visualize composition-
based phase diagrams, as tetrahedral or higher order constructs are required.
However, by the Gibb’s phase rule, the number of degrees of freedom of a sys-
tem in simultaneous phase and chemical equilibrium is reduced by the
number of independent equilibrium reactions. Thus, the reactive phase dia-
gram is constrained to lie on a subset of surfaces embedded in the overall com-
position space. These surfaces can be transformed into lower dimensional
composition coordinates, often referred to as transformed or reaction invariant
composition space (4). The transformed compositions, although dependent on
the number and type of reaction, are independent of the extent of reaction.
For a system of n components with r reactions, the transformed liquid mole
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 767

IPA
B2

D2 O3
Region 3
Azeo II
Feed M1 A3
2

Water
Feed Region I B2

IPA

M1 (b)
A3
Water Azeo D2, 03
Methylene chloride
(a)

IPA
B4
M2
D1 O3 D4
Region 3
Azeo D1 II
Feed M5 A3
1 4
A3
Region I
Feed M5 B1 B4
M2 O3 Water IPA

(d)
B1
Water Azeo D4
Methylene chloride
(c)

Fig. 7. Isopropyl alcohol dehydration using methylene chloride where Azeo = azeotrope,
the shaded areas represent the region of two liquid phases, (– – –) are tie lines, and
, two-phase decanters: (a) residue curve map and (b) flow sheet, and alternative methy-
lene chloride (c) residue curve map and (d) flow sheet. (See text.)

fraction, Xi, is given by


!
xi  vTi ðvref Þ1 xref
Xi ¼ for i ¼ 1 to n  r
1  vTtot ðvref Þ1 xref

where vTi ¼ row vector of stoichiometric coefficients of component i for each


reaction r
768 SEPARATIONS PROCESS SYNTHESIS Vol. 1

vref ¼ square matrix of stoichiometric coefficients of r reference compo-


nents for each reaction r
xref ¼ column vector of mole fractions of reference components
vTtot ¼ row vector of the sum of stoichiometric coefficients for each of r
reactions

A similar equation can be defined for the transformed vapor fractions, Yi.
The reference components are a subset of the reacting species and are used to
indicate the extents of reaction in terms of the mole numbers of the reference
components. The number of reference components is equal to the number of inde-
pendent reactions. Selection of the proper reference component is system depen-
dent. Selection rules have been elucidated (38) and the transformed coordinate
approach has been applied to a variety of reactive separations and phase equili-
bria including reactive distillation, adsorption, and solid–liquid, liquid–liquid
equilibria (39–41).
Transformed compositions are particularly useful for visualization of
reactive distillation systems. It has become well established that phase and
distillation region diagrams (DRD) in transformed compositions are entirely
analogous to those for nonreactive systems, exhibiting the same material balance
properties. Thus, reactive DRD can be used to assist in design feasibility and
operability studies in much the same fashion as outlined above for non-reactive,
nonideal systems (3,4,42).
When an equilibrium reaction occurs in a vapor–liquid equilibrium system,
the phase compositions depend not only on the relative volatilities, but also on
the consumption or production of species. The condition for azeotrope formation
(vapor and liquid-phase compositions are equal for two or more components) no
longer holds true in a reactive system and must be modified to include reaction
stoichiometry:

y1  x1 yi  xi X
n
¼ for all i ¼ 2 to n  1 and vT ¼ vi
v1  vT x1 vi  vT xi i¼1

where xi, yi ¼ mole fractions in the liquid and vapor, respectively, and vi ¼
stoichiometric reaction coefficient (negative for reactants, positive for products).
A reactive azeotrope occurs at a temperature where the rate of mass
exchange between phases and the rate of reaction for each component are such
that no net change in composition occurs in either phase. Whether one assumes
equilibrium reaction behavior or kinetically controlled reaction behavior (very
much dependent on column holdup, catalysis, and other device implementation
parameters) affects the existence and location of reactive azeotropes. Reactive
azeotropes always lie on the chemical equilibrium line. Since the boiling points
of the mixture are affected by the column pressure, and the equilibrium constant
is influenced by temperature (and thus column pressure), the location of the reac-
tive azeotrope can be influenced by changing the column pressure.
A residue curve map in transformed coordinates for the reactive system
methanol–water–acetic acid–methyl acetate is shown in Fig. 8. The water–
methyl acetate azeotrope has disappeared (water and methyl acetate can back
react to methanol and acetic acid), while the methyl acetate–methanol azeotrope
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 769

Fig. 8. Reactive residue curve map for a methyl acetate system. (Reprinted from Ref. 43,
with permission from Elsevier.)

remains. Only those azeotropes containing either all the required reactants or
products will be altered by the reaction. The system exhibits only one distillation
region in which the methanol–methyl acetate azeotrope is the low boiling node
and acetic acid is the high boiling node.
4.2. Flow Sheet Construction. Because of the analogous properties of
reactive and nonreactive residue curve maps, the opportunistic-strategic syn-
thesis methods described earlier can be extended to reactive system synthesis.
There has been limited progress in this direction (44,45). The elucidation of an
elegant reactive separation solution to a complex synthesis problem is often an
appealing prospect. However, there must be good reasons for combining reac-
tion and separation. Reactive separations introduce added design and operability
difficulties that must be weighed against any perceived benefits. In practice, the
operability space of a reactive separation is often quite limited compared to
sequential reaction and separation. For example, reactive distillation is only
applicable when the mixture boils under temperature and pressure conditions
favorable for the reaction. Other minimum compatibility criteria must also be
met. Other unfavorable situations for reactive distillation are the presence of
appreciable solids, operation at supercritical conditions, ie, where no separate
vapor and liquid phases exist, operation at very high or low temperature and
pressure conditions, ie, vapor and liquid phases both must be present and appro-
priate heating and cooling media available.
Once the minimum compatibility criteria have been met, then it is neces-
sary to evaluate whether simultaneous separation of components from the reac-
tion mixture is beneficial. Typical favorable situations include (1) when the
reaction must be carried out with a large excess of one or more of the reactants;
(2) when a reaction can be driven to completion by removal of one of the products
as formed, ie, equilibrium limited reactions; (3) when the desired product
770 SEPARATIONS PROCESS SYNTHESIS Vol. 1

is formed as an intermediate of consecutive reaction steps and can be removed as


formed (improvement of selectivity); (4) when a parallel reaction pathway can be
shutdown by manipulating local concentration gradients; (5) when exothermic
heat of reaction can be used to boil the mixture; (6) when narrow temperature
control is required and the mixture can be made to boil at the appropriate reac-
tions conditions.
A parametric study has been completed on component relative volatilities
and magnitude of reaction equilibrium constants for the generic equilibrium
reaction systems: A $ C, A $ C þ D, A þ B $ C, A þ B $ C þ D (46). In
this work the liquid-phase behavior was assumed to be ideal. Reactive azeotropes
could exist, but no nonreactive azeotropes were allowed. The following observa-
tions about the four types of reaction systems with ideal liquid-phase behavior
were made: (1) every A $ C system has one and only one reactive azeotrope,
independent of relative volatilities, but dependent on the magnitude of the reac-
tion equilibrium constant, Keq; (2) for the A $ C þ D system, reactive azeotropes
are only present for Keq < 1, and if present do no limit the economic viability;
(3) for the A þ B $ C system, reactive azeotropes can affect economic viability
when both reactants are lower or higher boiling than the product; (4) for the
A þ B $ C þ D system, all reactive azeotropes are intermediate boiling and if
present, can affect economic viability when either C and D are higher boiling
than A and B and relative volatilities follow aAB > 1, aAD > aAC, aAB < aAD, or
C and D are lower boiling than A and B and relative volatilities follow aAB > 1,
aAC < 1, aAD < 1, aAC > aAD.
A number of configuration and economic viability heuristics derived
from this work are summarized in Table 8. These heuristics also serve
as a starting point for consideration of reactive systems with nonideal
liquid-phase behavior. Subawalla and Fair also presented some basic
reactive distillation heuristics (47). A number of examples of reactive dis-
tillation synthesis using various synthesis methods have been published
(3,4,48–50).
Example: Methyl Acetate Production. Consider the process for the pro-
duction of methyl acetate via the acid-catalyzed equilibrium-limited esterifica-
tion of methyl with acetic acid, with by-production of water. The reaction is
nearly athermic, with an equilibrium constant on the order of unity.
Conventional Flowsheet. Assuming that a conventional reactor (eg,
CSTR) Using the synthesis approach outlined above, it is possible to develop
a The effluent from the reactor comprises the esterification products of methyl
acetate and water, as well as unreacted methanol and acetic acid. The four
component system exhibits azeotropic behavior, with binary low boiling azeo-
tropes between and methyl acetate– water, and methanol –methyl acetate
(homogenous and lowest boiling point of mixture). The methyl acetate– water
azeotrope is heterogeneous, but the azeotropic composition is outside of
the liquid – liquid region. There are no other binary or ternary azeotropes,
although the acetic acid –water system exhibits a tangent pinch at the
water-rich end.
The first opportunistic separation takes all of the methyl acetate, and
methanol overhead with as much water as azeotropes with the methyl acetate.
The overhead of the first column comprises a ternary system in which both
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 771

Table 8. Reactive Distillation Heuristics


Configuration Heuristics
1. Systems with low boiling reactive azeotropes and low boiling products must have a
nonreactive rectifying section to produce a pure product
2. Systems with high boiling reactive azeotropes and high boiling products must have a
nonreactive stripping section to produce a pure product
3. A double feed column is usually best when the reactant boiling points are far apart
(a AB  4)
4. A single feed column is usually best when the reactant boiling points are close
(1.2  a AB  1)
5. A column comprising a center reactive zone, nonreactive stripping and rectifying
sections is generally favorable for A $ C þ D and A þ B $ C þ D systems
6. The ability to separate C from D dictates the economic viability of reactive distillation
for A $ C þ D and A þ B $ C þ D systems, assuming Keq > 1. For a reactive distillation
to be economically viable and:

(a) To have reflux ratio  100, then should have aCD > 1.1
(b) To have reflux ratio  10, then should have aCD > 1.25
(c) To have reflux ratio  5, then should have aCD > 1.45
(d) To have reflux ratio  2.5, then should have aCD > 2.1

7. For systems in which the reactants are all intermediate boiling and the products are
the highest and lowest boiling components, then a Keq of at least 105 is required for
potential economic viability
8. For systems in which the products are all intermediate boiling and the reactants are
the highest and lowest boiling components, then a Keq of at least 100 is required for
potential economic viability
9. For systems in which the desired product is the highest boiling component, then a Keq of
at least 0.1 is required for potential economic viability
10. For systems in which the desired product is the lowest boiling component, then a Keq of
at least 103 is required for potential economic viability
11. For systems in which the by-product is the lowest boiling component and the desired
product is intermediate boiling, then a Keq of at least 103 is required for potential
economic viability
12. For systems in which the by-product is the highest boiling component and the desired
product is intermediate boiling, then a Keq of at least 102 is required for potential
economic viability

methyl acetate and methanol are saddles. Examining the residue curve map for
this system, the strategic separation of extractive distillation can be used to draw
methanol down the column and reject methyl acetate out the top. Solvents, eg,
ethylene glycol, water, and acetic acid, are potential extractive distillation sol-
vents. This methyl acetate may be wet with water, so if desired an additional
opportunistic separation can be used to take the methyl acetate–water azeotrope
overhead (recycled to the extractive distillation column) and pure methyl acetate
out the bottom. Methanol for recycle is recovered by distilling the underflow of
the extractive distillation column.
The bottoms from the effluent splitter column containing the acetic acid can
be recovered by using a combination of strategic separation separations for
pinched systems. Adding an extraction agent, eg, ethyl acetate, to concentrate
the acetic acid, followed by heterogeneous azeotropic distillation to recover the
acetic acid and solvent finishes the flow sheet (Fig. 9).
772 SEPARATIONS PROCESS SYNTHESIS Vol. 1

Acetic Acid
Methanol
Catalyst

Methyl
Acetate
Solvent

Azeo Water

Water Heavies
Solvent/Entrainer

Water

Fig. 9. Conventional flowsheet for methyl acetate process.

Evolution to Reactive Distillation. The conventional flowsheet thus


far synthesized could be summarized as consisting of a number of sections
or tasks as in Fig. 10. It turns out that the economically expensive task in
the process is the separation of acetic acid from coproduct water. However, if
reactive distillation were exploited, acetic acid might be reacted to completion

Fig. 10. Methyl acetate process tasks.


Vol. 1 SEPARATIONS PROCESS SYNTHESIS 773

Fig. 11. Methyl acetate process exploiting reactive distillation.

thereby eliminating the acid–water separation requirement. For this system


for which Keq is on the order of 1, the order and magnitude of the relative
volatilities and the heuristics in Table 8 suggest that reactive distillation
might be economically advantageous. The residue curve map in transformed
coordinates for this system shown in Fig. 8 shows no distillation boundaries.
However, as noted previously, the methyl acetate –methanol azeotrope
remains, and both desired products from the process, methyl acetate and
water, are saddles that cannot be reached in high purity and recovery by
reactive distillation alone.

Methyl Acetate
Distillation
Acetic Acid Task
Extractive
Distillation
Task
Catalyst
Reactive
Distillation
Task
Reaction
Task
Reactive
Distillation
Methanol Task

Combined
Distillation
Task

Water

Fig. 12. Integrated one-column methyl acetate process.


774 SEPARATIONS PROCESS SYNTHESIS Vol. 1

The process tasks might be developed as follows. The reaction task is


followed by reactive distillation to react all acetic acid remaining in the
water from the reactor effluent. The water stream may contain methyl acetate
and methanol, which may be opportunistically removed by distillation,
taking some water as the methyl acetate–water azeotrope, and recycled. Reactive
distillation can also be employed to react all methanol remaining in the methyl
acetate from the reactor effluent. The methyl acetate stream still contains water
as an azeotrope, and this azeotrope may be broken by extractive distillation with
acetic acid and any remaining acetic acid also removed by opportunistic distillation
from the methyl acetate product and recycled. The resulting process tasks and
recycles are arranged as in Fig. 11.
Six of the seven tasks in the process of Fig. 11 are distillative in nature and
each can be implemented in its own piece of equipment with optimized column
design parameters, hydraulics, reflux ratio, etc. For this particular example,
considering the fortuitous placement of recycles from each task, it is possible
to integrate all seven tasks and their energy requirements into a single column
shell with a single reboiler and condenser as in Fig. 12. The resulting distilla-
tion–reactive distillation–extractive distillation column requires 80% less capi-
tal and energy that the conventional flow sheet in Fig. 9.

BIBLIOGRAPHY

‘‘Separations Process Synthesis’’ in ECT 4th ed., Vol. 21, pp. 923–962, by S. D. Barnicki
and J. J. Siirola, Eastman Chemical Company; ‘‘Separations Process Synthesis’’ in ECT
(online), posting date: December 4, 2000, by S. D. Barnicki and J. J. Siirola, Eastman
Chemical Company.

CITED PUBLICATIONS

1. I. E. Grossmann, in J. J. Siirola, I. E. Grossmann, and G. Stephanopoulos, eds.,


Foundations of Computer-Aided Process Design, Elsevier, New York, 1989, p. 105.
2. S. D. Barnicki and J. J. Siirola, in J. D. Seader, ed., Perry’s Chemical Engineering
Handbook, 7th ed., McGraw-Hill, New York, 1997, Section 13.
3. J. G. Stichlmair and J. R. Fair, Distillation, Wiley-VCH, New York, 1998, Chapt. 5.
4. M. F. Doherty and M. F. Malone, Conceptual Design of Distillation Systems,
McGraw-Hill, New York, 2001, Chapts. 5, 7, 8.
5. J. M. Douglas, Conceptual Design of Chemical Processes, McGraw-Hill, New York, 1988.
6. D. F. Rudd, G. J. Powers, and J. J. Siirola, Process Synthesis, Prentice-Hall,
Englewood Cliffs, N. J., 1973.
7. Y. A. Liu, in Recent Developments in Chemical Process and Plant Design; Y. A. Liu, H. A.
McGee, Jr., and W. R. Epperly, eds., John Wiley & Sons, Inc., New York, 1987, Chapt. 6.
8. N. Nishida, G. Stephanopoulos, and A. W. Westerberg, AIChE J. 27, 321 (1981).
9. R. M. Kelly, in Handbook of Separation Process Technology; R. W. Rousseau, ed.,
Wiley-Interscience, New York, 1987, Chapt. 4.
10. V. M. Nadgir and Y. A. Liu, AIChE J. 29, 926 (1983).
11. Y. W. Huang and L. T. Fan, in Artificial Intelligence in Process Engineering; M. L.
Mavrovouniotis, ed., Academic Press, Boston, 1990, Chapt. 10.
12. Y. A. Liu, T. E. Quantrille, and S. H. Cheng, Ind. Eng. Chem. Res. 29, 2227 (1990).
Vol. 1 SEPARATIONS PROCESS SYNTHESIS 775

13. R. R. Wehe and A. W. Westerberg, Chem. Eng. Sci. 45, 1 (1990).


14. R. Agrawal, Ind. Eng. Chem. Res. 35, 1059 (1996).
15. R. Agrawal, Trans. Inst. Chem. Eng. Part A 77, 543 (1999).
16. R. Agrawal, Ind. Eng. Chem. Res. 40, 4258 (2001).
17. J. A. Caballero and I. E. Grossmann, AIChE J. 49, 2864 (2003).
18. J. A. Caballero and I. E. Grossmann, Comput. Chem. Eng. 28, 2307 (2004).
19. D. B. Van Dongen and M. F. Doherty, Ind. Eng. Chem. Fund. 24, 454 (1985).
20. E. R. Foucher, M. F. Doherty, and M. F. Malone, Ind. Eng. Chem. Res. 30, 760 (1991).
21. V. N. Kiva, E. K. Hilmen, and S. Skogestad, Chem. Eng. Sci. 58, 1903 (2003).
22. S. D. Barnicki, Chem. Proc., 2004.
23. O. M. Wahnschafft, J. Koehler, E. Blass, and A. W. Westerberg, Ind. Eng. Chem.
Res. 31, 2345 (1992).
24. J. Gmehling, J. Menke, K. Fischer, and J. Krafczyk, Azeotropic Data, Parts I–II,
VCH, New York, 1994.
25. S. D. Barnicki and J. J. Siirola, in E. K. Antonsson and J. Cagan, eds., Formal
Engineering Design Synthesis, Cambridge University Press, MA, 2001, Chapt. 10,
pp. 362–390.
26. S. D. Barnicki and J. R. Fair, Ind. Eng. Chem. Res. 29, 421 (1991).
27. P. C. Wankat, Ind. Eng. Chem. Res. 32, 894 (1993).
28. A. Eckles, P. Benz, and S. Fine, Chem. Eng. 98, 201 (1991).
29. A. Eckles and P. Benz, Chem. Eng. 99, 78 (1992).
30. J. W. Mullin, Crystallization, Butterworth-Heinemann, Boston, 1993.
31. C. G. Moyers and R. W. Rousseau, in Ref. 7, Section 11.
32. N. S. Tavare, Industrial Crystallization, Plenum Press, New York, 1995.
33. Y. L. Hwang, G. E. Keller, and J. D. Olson, Ind. Eng. Chem. Res. 31, 1753 (1992).
34. T. C. Lo, M. H. I. Baird, and C. Hanson, Handbook of Solvent Extraction, Krieger
Publishing Co., Malabar, Flor., 1991.
35. L. A. Robbins, Chem. Eng. Prog. 76(10), 58 (1980).
36. A. Stankiewicz, Chem. Eng. Proc. 42, 137 (2003).
37. C. Noeres, E. Y. Kenig, and A. Górak, Chem. Eng. Proc. 42, 157 (2003).
38. S. B. Gadewar, G. Schembecker, M. F. Doherty, Chem. Eng. Sci. 60, 7168 (2005).
39. S. Gruner and A. Kienle, Chem. Eng. Sci. 59, 901 (2004).
40. B. Harjo, K. Hg, and C. Wibowo, Ind. Eng. Chem. Res. 43, 3566 (2004).
41. D. W. Slaughter and M. F. Doherty, Chem. Eng. Sci. 50, 1679 (1995).
42. H. G. Schoenmakers and B. Bessling, Chem. Eng. Proc. 42, 145 (2003).
43. D. Barbosa and M. F. Doherty, Chem. Eng. Sci. 43(7), 1523 (1988).
44. C. P. Almeida-Rivera, P. L. J. Swinkels, and J. Grievink, Comput. Chem. Eng. 28,
1997 (2004).
45. G. Schembecker and S. Tlatlik, Chem. Eng. Proc. 42, 179 (2003).
46. C. A. Hoyme, A parametric Reactive Distillation Study: Economic Feasibility and
Design Heuristics, Ph.D. Dissertation, The University of Tennessee, Knoxville, 2004.
47. H. Subawalla and J. R. Fair, Ind. Eng. Chem. Res. 38, 3696 (1999).
48. J. W. Lee, S. Hauan, K. M. Lien, and A. W. Westerberg, Proc. R. Soc. London Ser. A,
Math. Phys. Eng. Sci. 456, 1953 and 1965 (2000).
49. J. W. Lee, S. Hauan, K. M. Lien, and A. W. Westerberg, Chem. Eng. Sci. 55, 3161 (2000).
50. J. J. Siirola, L. T. Biegler, and M. F. Doherty, eds., Fourth International Conference
on Foundations of Computer-Aided Process Design, Elsevier, New York, 1994, p. 222.

SCOTT D. BARNICKI
CRAIG A. HOYME
JEFFREY J. SIIROLA
Eastman Chemical Company
776 LIQUID–LIQUID EXTRACTION Vol. 1

LIQUID–LIQUID
EXTRACTION
1. Fundamentals and Fields of Application

Liquid – liquid extraction or solvent extraction is a separation process which is


based on the different distribution of the components to be separated between
two liquid phases. It depends on the mass transfer of the component to be
extracted from a first liquid phase to a second one.
Definitions. The phases form definite contact areas and cannot or can
only to a limited extent mix with one another on account of their properties.
The phases are composed of individual substances, so-called components, and
flow as liquid streams through the extractor (see Fig. 1). Phases are liquid
streams that are in equilibrium with each other.
Generally, more than three components are involved in an extraction. To
make the description of the problem simpler, key components are defined. The
feed to a liquid – liquid extraction is the solution that contains the components
to be separated. The key component of this liquid phase (the phase from which
the solute is extracted) is called A, the transferred substance (the solute) C. The
(extraction) solvent is the liquid added to the process to extract the substance C
from the feed. The solvent can be the pure component, B, but usually contains
small quantities of C and A, because the solvent is usually recycled from a recov-
ery system. The solvent phase leaving the extractor is the extract. The extract
contains mainly B and the extracted component C but also small quantities of
A. The raffinate is the liquid phase left from the feed after being contacted by
the extraction solvent; it is composed mainly of A, but generally small quantities
of the extracted component C and solvent B.
Comparison with Distillation. Liquid – liquid extraction has similari-
ties with distillation. In both cases the substances to be separated have concen-
tration ratios 6¼ 1 in the two phases. In the case of distillation the second phase
(the vapor phase) is formed exclusively from the components of the initial (liquid)
phase, with extraction a new substance, the (extraction) solvent, must be added
to form the second (liquid) phase. The solvent enters into the material balance of
the process and its behavior towards the substances to be separated is decisive
for the separation process.
Generally, the extracted substance is separated from the solvent by dis-
tillation and the solvent B is recycled to the extractor. Sometimes the selective
action of the solvent is used in the distillation which generates the reflux for
the extraction, eg, extractive distillation (extraction of aromatic compounds,
see Section 5.2) or azeotropic distillation (extraction of acetic acid). In the extrac-
tion of metal salts (see Section 5.3) the purity is often achieved by adjusting the
pH. Extraction is an isothermal process, normally carried out at ambient tem-
perature and pressure.
Comparison with Absorption. In both absorption and extraction sol-
vents are used which are recycled and thereby take up the required components
selectively and then release them again. Selection of the most suitable solvents
therefore plays an important role in the design of the two separation processes.

ß2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim


10.1002/14356007.b03_06
Vol. 1 LIQUID–LIQUID EXTRACTION 777

Fig. 1. Principles of extraction – streams, phases and components.

Applications of Liquid – Liquid Extraction. Liquid – liquid extraction


is primarily applied where direct separation methods such as distillation and
crystallization cannot be used or are too costly. Liquid – liquid extraction is
also employed when the components to be separated are heat-sensitive (eg, anti-
biotics) or relatively nonvolatile (eg, mineral salts).
Liquid – liquid extraction is used in industry for the following purposes:

1. Separation of systems with similar boiling points (eg, separation of


aromatics from aliphatic hydrocarbons);
2. separation of highboilers and low concentrated solutes from aqueous
solutions (eg, phenol);
3. separation of mixtures with high boiling points (eg, vitamins);
4. separation of temperature-sensitive compounds (eg, acrylates, biotechnol-
ogy);
5. separation of azeotropic mixtures (extraction of acetic or formic acid from
aqueous media using, eg, MTBE as solvent);
6. extraction of organic compounds from salt solutions (eg, caprolactam);
7. extraction of salts from polymer solutions (eg, ketone resins, polyols);
8. extraction of metal salts from low-grade ores (eg, copper);
9. extraction of metal salts from wastewater (eg, copper);
10. recovery of nuclear fuels (Purex process).

Only a general outline is given in this article. For further details, the reader
is referred to books (1–7) and the Proceedings of the International Solvent
Extraction Conference (ISEC) (8–17). Data of liquid – liquid equilibria can be
found in (18,19) and (20).

2. Thermodynamic Fundamentals

A comprehensive treatment of the thermodynamic fundamentals of phase equi-


libria is given elsewhere (! Absorption; ! Distillation and Rectification).
2.1. Graphic Representation of Phase Equilibria. Triangular Diagrams.
Triangular diagrams (Fig. 2) are used for representing three-component systems.
778 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 2. Triangular diagram for representation of a three-component system.

Every possible composition of the ternary mixture corresponds to a point in the


diagram. Each corner of the triangle represents a pure component and its designa-
tion is marked at this corner. On the side opposite to this corner the mass fraction
of this component is zero. The sides of the triangle represent the corresponding
two-component systems. The proportions of the individual components in a
mixture are obtained as follows: The mass fraction of each component is given
by lines parallel to the side opposite to the corner which represents the pure com-
ponent, from 0 kg/kg at this opposite side to 1 kg/kg at the corner. The numbering
can be placed at the edges of the triangle, in which case it is advisable to extend the
lines. The numbers can also be inserted in the middle of the lines. This makes the
diagram easier to use but some of the numbers are always skew and inverted. The
point P has the composition 0.5 kg/kg C, 0.2 kg/kg A, and 0.3 kg/kg B.
The following rules apply in the triangular diagram: If component C is
added to a mixture of 0.8 kg/kg A and 0.2 kg/kg B, represented by the point D,
then the composition of the new mixture follows the path of the straight line
DC. On this straight line, and on all straight lines which pass through a corner,
the ratio between the two other components which do not form the corner is
constant.
If a mixture of composition U and one of composition V are combined then
the resulting mixture lies on the straight connecting line UV. The composition of
the final mixture is obtained by dividing the connecting line between the two ori-
ginal mixtures in inverse ratio to the quantities mixed (lever rule). The mixture
M, for example, comprises three parts U and one part V, and the mixture G com-
prises three parts C and two parts D.
The concentration data obtained from the triangular diagram (eg, point
G: 0.6 kg/kg C, 0.08 kg/kg B, 0.32 kg/kg A; B : A = 1 : 4) apply to the total quantity
of resulting mixture regardless of whether it is homogeneous or separated into
several phases.
As every point in a triangular diagram corresponds to a composition there is
no coordinate free for another reference variable. It is only possible to plot para-
meters such as lines of equal density, boiling points or other physical data, lines
of equal activity coefficients or lines of phase boundaries.
Vol. 1 LIQUID–LIQUID EXTRACTION 779

Fig. 3. Definitions of phase equilibria and mass fractions in a triangular diagram.


A) Phase equilibria; B) Mass fractions.

Only ternary systems with miscibility gaps are suitable for extractions,
and the boundary line between the liquid single-phase region(s) and the
two-phase region is particularly important. This boundary line is called the
binodal curve (Fig. 3). Every point on the binodal curve (except the plait
point in systems with a complete binodal curve, see below) is in equilibrium
with another binodal point. The lines which connect points in equilibrium
with one another are called tie lines. All mixtures which correspond to a
point on the tie line separate into two phases. The composition of these
phases is given by the end points of the tie lines and their quantities are
given by the lever rule. The equilibrium data have to be determined experi-
mentally in each individual case. Lines parallel to the triangle sides CB and
AC respectively, drawn through related points of equilibrium on both
branches of the binodal curve, produce an auxiliary line, formed by the inter-
sections of the parallels, and enable to construct any tie line as demonstrated
in Figure 3 A.
Several types of triangular diagrams are shown in Figure 4. In every sys-
tem with a complete binodal curve (closed diagram, Type II in Fig. 4) the end
points of the tie lines draw closer and closer together until they finally coincide
at the plait point. The plait point on the binodal curve represents a single phase
which does not split into two phases.
Diagram I is called an open diagram in which B is not fully miscible with
the substance to be dissolved. The capacity of such a solvent is correspondingly
low and its selectivity is usually high (see Section 2.4).
780 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 4. Various types of phase equilibria in ternary systems. Type I: Open system,
miscibility gap between the key component of the solvent (B) and solute C. Type II: Closed
system, C is completely miscible with the key component of the feed (A) and the key
component of the extraction solvent (B). Type III: Open system, miscibility gap between
A and C Type IV: System with formation of three phases.

In diagram II the substance to be extracted (C) is totally miscible with both


A and B. This form of extraction diagram is often found in extractions where
solute C and A are present in approximately the same quantities. If in this
case the concentration of C in the AB-mixture is too high (higher than the con-
centration at the plait point) the system forms only one phase and extraction is
not possible.
Diagram III shows a system where C has a miscibility gap with A. Its con-
centration in the B-rich phase is then substantially higher than in the feed. The
diagram is typical for wastewater extractions.
Diagram IV is an open triangular diagram in which three liquid phases are
formed in the middle concentration range. This behavior is often found in sys-
tems consisting of inorganic salts in aqueous solutions and organic solvents.
The appearance of a third liquid phase can cause considerable difficulties during
metal-salt extraction (see Section 5.3). The influence of pressure on the binodal
curve and equilibria is negligible below values of several bar. The influence of
temperature is strong. The lower the temperature the more will the triangular
diagram normally be filled by the miscibility gap.
Other Graphic Representations. Liquid two-phase systems comprising
three components can also be plotted on rectangular coordinates. In many
cases these diagrams are preferred because the measured values can be adjusted
better and the configuration of the calculations is simpler. The most important
forms of representation are shown for three mixtures in Figures 5–7. A compre-
hensive review of forms of representation is given in (27).
In the equilibrium diagram (Figs. 5–7 II) the mass fraction of C in the
B-rich phase yC (ordinate) is plotted as a function of this mass fraction in
the A-rich phase xC (abscissa). This diagram is analogous to the McCabe – Thiele
Vol. 1 LIQUID–LIQUID EXTRACTION 781

Fig. 5. Representation of phase relationship of the ternary system heptane (A), diethy-
lene glycol (B), benzene (C), at 20 8C. All data refers to mass fractions. I) Triangular
diagram; II) Equilibrium diagram; III) Distribution diagram; IV) Selectivity diagram.
782 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 6. Representation of phase relationship of the ternary system trichloroethylene (A),


water (B), caprolactam (C), at 20 8C. All data refers to mass fractions. I) Triangular
diagram; II) Equilibrium diagram; III) Distribution diagram; IV) Selectivity diagram.
Vol. 1 LIQUID–LIQUID EXTRACTION 783

Fig. 7. Representation of phase relationship of the ternary system water (A), butyl
acetate (B), phenol (C), at 20 8C. All data refers to mass fractions. I) Triangular diagram;
II) Equilibrium diagram; III) Distribution diagram; IV) Selectivity diagram.
784 LIQUID–LIQUID EXTRACTION Vol. 1

diagram in distillation (see Distillation). It shows the relationship between equi-


librium and operating conditions.
In the example shown in Figures 5–7 III the distribution ratios of the
extracted components (benzene, caprolactam or phenol) are plotted as functions
of their mass fractions in the A-rich phase (curve a) and in the B-rich phase
(curve b). The distribution ratio k indicates how much better or worse the
extracted component is soluble in the B-rich than in the A-rich phase; the mini-
mal solvent ratio can be evaluated directly from Figures 5–7 III. The distribution
ratio therefore is an important criterion for the selection of a solvent. The mass
fraction in the A-rich phase is easier to visualize as a reference variable. At the
feed stage it corresponds approximately to the initial feed concentration. It gov-
erns the yield at the raffinate outlet. On the other hand, showing k as a function
of yC (curve b) has the advantage that it permits comparisons with respect to sol-
vent loading. It is always advisable to plot both curves because their suitability
for extrapolation and interpolation varies considerably in the different concen-
tration ranges.
In the selectivity diagram (Figs. 5–7 IV) the quotient of two distribution
ratios is plotted as a function of the mass fractions of component C in the A-
rich (curve a) and in the B-rich phase (curve b). The quotient of the distribution
ratios for the extracted component C (yC/xC) and for A (yA/xA) corresponds to the
relative volatility in distillation (see Section 5.1). It is a measure of the selectivity
of a solvent.
2.2. Measuring Methods of Phase Equilibria. Whenever possible a
ternary system is used as the starting point for measurement of liquid – liquid
equilibria. It is first necessary to establish which of the three possible binary sys-
tems have a miscibility gap and which are completely miscible. For this purpose
it is sufficient in each case to mix the two substances in ratios of 3 : 1, 1 : 1 and
1 : 3. If no formation of two liquid phases occurs in any of the cases then the sys-
tem is fully miscible and not suitable for extraction.
If, for example, A and B are not miscible while the third, C, is completely
miscible with the two others (closed triangular diagram) then the binodal
curve can be determined as follows. Equal amounts of A and B are placed in a
controlled-temperature vessel which can be shaken (eg, an Erlenmeyer flask in
a thermostat) and substance C is added until the mixture becomes clear. In two
further experiments pure A or B is placed in the vessel and in each case the sec-
ond substance is then added until turbidity occurs. Three points have now been
obtained in the triangular diagram and the approximate shape of the binodal
curve is known.
For precise determination of the binodal curve, mixtures of A and C
containing 0.9, 0.8, 0.7 (etc., for as far as appears necessary) kg/kg A are
prepared and titrated with B at constant temperature. When turbidity is
reached titration is ceased and a small quantity of the original AC mixture is
added until the solution is clear again. Then a characteristic variable such as
refractive index, density, dielectric constant, freezing point, boiling point, etc.
of this mixture can be determined. In the same way mixtures of C and B are
titrated with A. This measurement gives the precise shape of the binodal
curve and all the compositions lying on this curve can be defined by a physical
parameter.
Vol. 1 LIQUID–LIQUID EXTRACTION 785

To determine phase equilibrium and therewith the tie lines, ie, the points
which are in equilibrium with one another, a mixture from the two-phase region
is placed in a controlled-temperature, shaking cylinder. Equilibrium is estab-
lished by shaking. The phases are allowed to settle and are separated by draining
off into different vessels. Then their quantities are determined and the physical
variable measured during determination of the binodal curve is then measured
for both phases. This immediately gives the two tie line points. Furthermore,
both phases can also be analyzed by well known chemical methods, such as
gas or liquid chromatography and titration. The correctness of the measurement
can be checked by connecting the two tie line points in the triangular diagram.
The connecting line must pass through the point of the original mixture, and this
point must divide the length of the tie line in inverse ratio to the quantities of the
two phases.
If a system of more than three components has to be measured then several
components are combined into one group in order to arrive at a ternary system
again. Determination of a binodal curve would still be possible in such a case but
the two phases into which the test mixture separates are no longer identical with
the mixtures which were obtained during the determination of the binodal curve.
It is therefore better to allow the mixture to reach equilibrium conditions in a
controlled-temperature, shaking cylinder, separate it and then analyze the
phases using some preplanned method.
The time consuming step in the determination of equilibria is the prepara-
tion of the experiment, such as purification of the substances, setting up the cali-
bration curves, etc. Therefore it is senseless to restrict the number of measured
points and use interpolation methods. To increase the accuracy a material bal-
ance is set up for each component and the measured points are adjusted by draw-
ing the curves from Figures 5–7.
2.3. Empirical Correlations of Phase Equilibria. In ternary systems,
the use of empirical correlations to represent the binodal curves and distribution
equilibria of component C is of advantage (28). The mutual solubilities of
the liquid phases are usually determined by a laboratory turbidity titration.
The binodal curve is then determined by extrapolating the measured values
(see Section 2.2). In Type I systems with an open miscibility gap, the binodal
curve can be simply described by parabolic equations:

x ¼ D0 þ E0 x0A þ F 0 x02
A

y ¼ D00 þ E00 x00B þ F 00 x002


B

where x0 A is the mass fraction of A in the A-rich phase, x00 B the mass fraction of
B in the B-rich phase and D0 , D00 , E 0 , E 00 , F 0 and F 00 are coefficients.
In systems with a closed miscibility gap (Type II), the parabolic equations
are only valid for the lower parts of the curve; they cannot be used to represent
the entire binodal curve. The best equation for the closed binodal curve is that
proposed by HLAVATY (29). The correlation is derived from a polynomial equation
and has the advantage that the binodal curve ends exactly at the binary solu-
bility points. Its disadvantage is, however, that the lower, in practice virtually
linear, regions of the binodal curves are represented in wave form as a result
786 LIQUID–LIQUID EXTRACTION Vol. 1

of the polynomial equation. The Hlavaty equation should, therefore, be used with
caution.
The equilibrium distribution of solute C in the A- and B-rich phases is gen-
erally determined independently from the binodal curve by means of laboratory
experiments. The mass fraction xC and yC of the two phases in thermodynamic
equilibrium are thus obtained and determine the positions of the end points of
the tie lines on the binodal curves (see Section 2.2). Equations for correlating
and interpolating the end point of the tie lines are described in the literature,
they are primarily exponential equations such as those proposed by OTHMER
and TOBIAS, and HAND (1). However, all exponential equations generally fail
when the concentration of C is very low. To allow a more accurate estimation
of distribution equilibria and extraction processes at the low raffinate concentra-
tions that are of industrial importance, BENDER and BLOCK (28) proposed the fol-
lowing correlation for the distribution ratio K:

K ¼ G1 þ G2 x0C þ G3 x02 03
C þ G 4 xC

The coefficients Gi must be fitted to the measured values of the tie lines x
and y. Although the equation does not generally permit extrapolation of the
curve above the measured values to the plait point, the distribution coefficient
K can be obtained accurately up to x and y ! 0.
The empirical correlations are simple and clear. They can be calculated
quickly with appropriate computer programs and give an accurate representa-
tion of the measured liquid – liquid equilibria. However, these advantages are
accompanied by several disadvantages. The correlations can only be used with
ternary systems and do not allow extrapolation to higher multicomponent sys-
tems. Extensive laboratory work is required to determine the mutual solubilities
of components A and B as a function of the concentration of C as well as the equi-
librium distribution of component C. Systematic measurement errors cannot be
detected because the experimental values are empirically correlated.
2.4. Selection of Solvent. The selection of the most suitable solvent is
one of the hardest problems of the extraction, and the decision made has a great
influence on the economy of the process. The solvent must fulfill two basic
requirements:

1. it must have a miscibility gap with the A-rich phase, and


2. it must be a better solvent for component C than for A.

A great number of methods have been recommended for discovering suita-


ble solvents (30,31). The more empirical strategy of former times is being
replaced more and more by a systematic research with the aid of data banks
(26). The concept of polarity is used for characterizing suitable parameters (32).
The greater the difference in polarity between two liquids, the greater is the
probability that they have a miscibility gap. The substance to be extracted will
have a higher concentration in the liquid to which it is closer in polarity.
However, the relationships are not so simple that only a scale of polarities
has to be set up and the difference in polarity at which miscibility gaps occur
read off the scale. For example, aniline, phenol, and furfural have miscibility
Vol. 1 LIQUID–LIQUID EXTRACTION 787

gaps with both water and hexane. Acetone, ethanol, and dioxan which have a
similar polarity are miscible with both water and hexane. The molecular form
and the size of the molecule also contribute to solvent behavior. The behavior
of solvents towards hydrocarbons has been investigated particularly intensively
(33,34). It is also possible to obtain effects from mixing two solvents which could
not be predicted from the properties of the mix components (35,36).
The most important property of a solvent is its selectivity. This indicates
how much better the solute is dissolved than the A-rich phase. As can be seen
from Figures 5–7 III the selectivity is strongly dependent on concentration.
In addition to selectivity the capacity of a solvent is also important. This is
defined by the distribution coefficient of C between the B-rich and A-rich phases
and determines the quantity of solvent which has to be recycled. When testing
different solvents for a specific extraction task it is often found that solvents
with higher selectivities have lower capacities and vice versa. The greater the
throughput of a plant the more important the capacity of the solvent is for the
economy of the process. If there are several possible solvents for a process then
their behavior with respect to selectivity and capacity can be tested with a gas
chromatograph without starting with time-consuming equilibrium measure-
ments (37).
In addition to these basic requirements there are also other criteria which
must be taken into consideration when selecting a solvent and which sharply
reduce the number of possible compounds. Most important is the availability of
the solvent in commercial quantities at an economically acceptable price. It is
then necessary to find data concerning boiling points, density, viscosity and —
where available — information about solubility (38,39).
The boiling point of the solvent is important because the extracted
component is usually separated from the solvent by distillation and the
boiling point difference determines the reflux ratio in this distillation. If
the solvent boils at a higher temperature than the solute, this solute is dis-
tilled overhead. If it is necessary to recycle a great deal of solvent then the
solvent has to be purified by distillation in a subsequent step; the specific
heat which is required for heating the solvent to its boiling temperature
must also be taken into account. Too great a temperature difference is then
a disadvantage.
The density of the solvent is important for obtaining a sufficiently large dif-
ference in densities between the two liquid phases for the separation process.
Chlorinated compounds are suitable if a solvent with a high density is to be
used, and particularly low-density compounds are found among the saturated
hydrocarbons and ethers.
The thermal stability of the solvent can be highly significant. The
researcher hardly ever finds the answer to this question in the literature and
is directed to carry out his own measurement. It is necessary to know (1) how
much solvent make-up will be needed in the process; (2) whether a regeneration
process for continuous removal of decomposition products needs to be installed.
Very careful tests must be carried out to find out whether decomposition pro-
ducts can contaminate the extract or have a corrosive effect on the construction
materials of the plant. Sometimes the toxicological properties and biological
degradation of a solvent have to be considered (40).
788 LIQUID–LIQUID EXTRACTION Vol. 1

The viscosity should be as low as possible as this favors mass transfer and
phase separation. However, this property is less significant than those men-
tioned above. If the possible solvents have been reduced to a manageable number
after testing all the aspects then the individual properties can be evaluated with
a point system to discover the best solvent.
The surface tension is another important parameter because it influences
the energy demand necessary to produce a suitable drop dispersion. High surface
tension gives rise to formation of stable drops and needs a high energy input. A
solvent with a high surface tension decreases the danger of producing an emul-
sion. The surface tension is dependent on the concentration of the transfer com-
ponent. Further properties such as suitability for regeneration, disposability,
costs, combustibility, toxicity and corrosiveness also have to be taken into
account to select a suitable solvent.
Bound up with solvent selection is the choice of the dispersed and the con-
tinuous phase, respectively. Direction of mass transfer and phase ratio have to be
taken into account. The direction of mass transfer influences the surface tension
and in this way can promote or hamper the coalescence of drops. The phase to be
dispersed should be the one which impedes the drop coalescence to produce
stable drops, and the one with the higher flow rate to obtain a mass transfer
area as large as possible.
2.5. Calculation of Theoretical Stages (! Absorption, Chap. 4.; !
Distillation and Rectification, Chap. 4.). Two possibilities exist to calculate
the mass transfer performance of extractors. The first one uses the concept of
theoretical stages and is based on the assumption that concentration equilibrium
is established in every section of an extractor. This assumption can be fulfilled
only in the mixer – settler with long contact times of the phases involved. In
this concept the total height of an extraction column is given by the product of
the number of theoretical stages (NTS) and the height equivalent of a theoretical
stage (HETS). NTS follows from a material balance in combination with the dis-
tribution equilibrium; HETS depends on fluid dynamics and mass transfer in the
columns and comprises all real conditions as for instance the conditions of phase
mixing. Therefore this concept is preferred only for rough estimations in practice
because it is easy to handle and sufficient for preparation of laboratory experi-
ments. In the following sections, the method will be presented by aid of equili-
brium diagrams.
The second possibility is a stream concept. The column height required for a
given throughput and concentration gradient of phases along the column can be
calculated from height integration of a set of differential equations, derived from
a differential material balance and the local mass transfer between the both
phases. The stream concept supplies the course of concentrations along the col-
umn and is suitable for analysis of the dynamic behavior of a column and for tak-
ing into account the drop population inside the column. It is preferred until now
for scientific investigations, eg, for stationarily operated pulsed sieve tray col-
umns (45,47–49,51,62) and for instationarily operated rotating extractors
(65,68).
Selection of Calculation Methods. The conditions for an applicable
method of calculation are set by the problem to be solved, by whether simpli-
fied assumptions are permissible and by the main influencing variables of an
Vol. 1 LIQUID–LIQUID EXTRACTION 789

extraction problem. They are, therefore, different. The following points should be
considered when selecting a method of calculation:

1. Do changes in mass flows have to be considered?


2. Is the distribution ratio constant?
3. Is the concentration gradient large or small?
4. Does only one component have to be extracted, or are there several with
different distribution coefficients?

Specific Methods. Method 1. If the following conditions are satisfied:

1. the component A and solvent B are virtually insoluble in each other,


2. the changes in mass flow rates of the extract and raffinate phase are negli-
gible (eg, for a low content of C in the feed)

then the compositions of the phases can be expressed as loads. If a material bal-
ance is made as shown in Figure 8, the two loads Yc and Xc in a given column
cross section can be related as follows

m
_F
YC  YC; S ¼  ðXC  XC; R Þ
m_s

When plotted on the equilibrium diagram, this relation gives a straight


line, the balance or operating line. The number of theoretical stages required
to obtain a desired decrease in the concentration of C from XC, F to XC, R
can be determined graphically by constructing a series of steps as shown in

Fig. 8. A) Material balance of a column cross section; B) Corresponding operating


diagram.
790 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 9. Construction of steps in the operating diagram to obtain the number of theoreti-
cal stages (NTS).

Figure 9. The result of the use of the steps often means that the load obtained
from C in the raffinate (XC, R) does not correspond to the required load (XCR, req).
Figure 8 also shows the position of the balance line for the minimum extrac-
tant ratio Vmin

Vmin ¼ m
_ s; min =m
_F

The desired enrichment of the feed with C using the minimum amount of extrac-
tant ṁs, min can only be achieved with an infinite number of stages.
Method 2. Figure 10 shows the equilibrium diagram for the system
toluene – ethylene glycol – water as a double logarithmic plot. This type of plot

Fig. 10. Determination of the number of stages for a variable distribution ratio and a
high degree of depletion.
Vol. 1 LIQUID–LIQUID EXTRACTION 791

is useful if the distribution ratio varies considerably, despite low loading, and the
concentration range considerably exceeds the desired raffinate purity. In this
case the operating line is at an angle of 458.
 
m
_F
logðYC  YC; S Þ ¼ log þ logðXC  XC; R Þ
m
_S

The position of the straight lines is determined by the extractant ratio V. The
number of required theoretical stages can be determined by constructing a series
of steps.
Method 3. If the distribution ratio K is a constant, further simplification
is possible; in analogy with the Colburn equation (! Absorption, Chap. 4.2 ), the
number of theoretical stages NTS can be calculated:
h  i
X Y
log 1 þ ðE  1Þ XC;C; RF þ XC;C; RS
NTS ¼ 1
logE

where the extraction factor E is given by

E ¼ Km
_ s =m
_F

If NTS is known and the degree of depletion XC, R/XC, F is desired, the follow-
ing equation is obtained:
 
YC; S
XC; R ðE  1Þ 1 þ XC; F
¼
XC; F EðNTSþ1Þ  1
This equation can only be solved iteratively; a graphical solution is shown in
Figure 11.
Method 4. If none of the above-mentioned conditions is satisfied, and the
extraction problem can be limited to consideration of the three key components,
the calculation can be carried out with the aid of the triangular mass fraction
diagram. This method is based on the mixing rules described in Section 2.1.
The specified compositions of the feed F and solvent S streams are plotted
on the phase equilibrium diagram for the ternary mixture (Fig. 12). For material
balance of the feed and product streams, the column can be assumed to be a
simple mixing vessel. The composition M of the mixture lies on the tie line FS;
the position of M on the tie line is determined by the ratio ṁS : ṁF = MF : MS. The
compositions of the raffinate R and extract E streams leaving the column are
obtained from the intercepts of the tie line with the binodal curve. The raffi-
nate composition is determined by the desired mass fraction of C in the raffinate
(xC, R). The composition of the extract E is given by the point at which the
extended line RM intersects the binodal curve. The ratio ṁR : ṁE is obtained
by applying the lever rule. The quantities and the compositions of the incoming
and outflowing streams are thus determined.
The required number of stages in a countercurrent column are obtained by
using the pole point procedure. The procedure can be explained by imaging a
792 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 11. Dependence of the theoretical number of stages, the degree of depletion, and the
yield on the extraction factor E.

Fig. 12. Pole point construction – balance diagram.


Vol. 1 LIQUID–LIQUID EXTRACTION 793

Fig. 13. Pole point construction – operating diagram.

stream P to maintain the material balance. Mixing of this stream with the raffi-
nate stream R gives the composition of the solvent stream S. To maintain the
material balance the extract on the other hand must be composed of the feed
stream F and the imaginary stream P. The imaginary stream P can thus be
regarded as a common mixing partner for F and R to maintain the compositions
of E and S. The composition of the imaginary stream P is represented by a point
outside the triangular diagram, the pole point. The imaginary mass fractions of
the three components in the pole point can be > 1 or negativ. If the material bal-
ance boundaries for the individual stages are taken as shown in Figure 13, the
above procedure gives the stage extracts E2, E3, etc. from the mixture of the ima-
ginary stream P and the individual stage raffinates in the proceeding stages R1,
R2, etc. The composition of the raffinates R1, R2, etc. are linked to the correspond-
ing extracts E1, E2, etc. via the distribution ratios (tie lines).
Evaluation of Stage-Number Calculation for Process Design. In order
to determine the operating conditions and the dimensions of the equipment
used for extraction, the results of several stage-number calculations must be
plotted in a diagram showing the relationship between the solvent circulation,
the number of stages, and the residual content of solute in the raffinate. Such
a diagram is shown in Figure 14 for mixture II (see also Fig. 6). It represents
the extraction of a 10 % solution of caprolactam in trichloroethylene with
water as the extraction solvent. The favorable operating region is usually located
in the region which shows the most pronounced curvature. Other variables
(eg, the effect of temperature and the residual loading of the solvent with solute)
may have to be taken into consideration.

3. Apparatus

3.1. Survey. Extraction apparatus can be classified into countercurrent


columns, centrifugal extractors, and mixer – settlers. All industrial equipment
794 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 14. Extraction of caprolactam from a 10 wt % solution in trichloroethylene with


water as solvent.

designs use the principle to disperse one of the two liquids into the other in order
to enlarge the contact area for mass transfer and to influence the drop generation
by energy input such as mixing or pulsing. After the resulting droplets have been
contacted with the continuous phase long enough to allow the extractable com-
ponent(s) to migrate to the dispersed phase, both liquids have to be separated
again. Densities will define the flow directions of the phases: The heavier liquid
flows from top to bottom, the lighter one from bottom to top. Accordingly, the
fluids have to be transported to the apparatus and removed by pumps.
In the following chapter, a few typical designs have been selected from the
large number of equipment designs actually used in industry. A more detailed
treatment can be found in literature, especially in (5) and (6).
Columns without Energy Input. The simplest extractor is the spray col-
umn (Fig. 15 A). The spray column achieves only very small numbers of separa-
tion stages and is rarely used nowadays. If the column of Figure 15 A is filled
Vol. 1 LIQUID–LIQUID EXTRACTION 795

Fig. 15. Columns without energy input (LIC = level indicating controller). A) Spray
column (light phase dispersed); B) Packed column (heavy phase dispersed); C) Sieve plate
column (light phase dispersed).

mainly with the heavier liquid, the lighter liquid breaks at the feeding point into
drops that rise in the continuous liquid. Drops reaching the interface are kept at
this position by controlling the bottom discharge (LIC). There, they will then gra-
dually coalesce and form a continuous layer floating on top of the continuous
liquid.
The reverse procedure is possible as well, as shown in Figure 15 B. Here, the
distributor is located at the top and the water interface, called the principle
interface, is controlled at the bottom of the column by filling it mainly with the
lighter liquid. In this case, the heavier liquid is the dispersed one, broken into
droplets descending in the lighter liquid. The principle interface can be adjusted
at any height (between top and bottom) of the column as needed, thus the heavier
liquid being dispersed above and the lighter liquid below the principle interface.
Countercurrent columns are equipped with settling zones at the top and the
bottom, with coalescing aids provided if necessary. The liquids should never be
fed into the column in these settling zones because this would cause hydrody-
namic instability of the column thus reducing the operational performance.
The dispersed liquid often cannot be selected from theoretical considerations,
but only by aid of experiments in a pilot column using the real material system.
Packed columns. (structured and unstructured) and sieve plate columns
have been adopted from distillation technology (see Distillation). In packed
columns drop formation occurs on entry of the dispersed phase in the packed
bed (Fig. 15 B). The drops rise through the packed bed without significant
further dispersion. However, mass transfer takes place mainly during the
formation of a new interfacial area. This explains the low efficiency of packed
columns.
In a sieve plate column with downcomers, as shown in Figure 15 C, the
lighter dispersed phase coalesces beneath each plate to give a continuous
layer. Its buoyancy presses the liquid through the holes of the sieve plates, pro-
ducing jets which disintegrate into drops. The continuous liquid flows through
796 LIQUID–LIQUID EXTRACTION Vol. 1

the downcomers from stage to stage and then across the plates. In the case of low
interfacial tension the layers of coalesced phase are too small for a symmetrical
operation of the plates. Then it is advantageous to use dual flow plates without
downcomers (57).
Sieve plate columns have the advantage over packed columns that new
drops are formed at each tray. However, the drop load of the plate is often irre-
gularly distributed, especially if the column diameter is large (more than 1 m).
This is the reason why sieve plate columns can reach efficiencies of 10 – 15 %
only in rare favored cases. Moreover, since columns are generally built with a
plate spacing of 0.3 m, even simple separations require considerable overall
heights.
Mass transfer between the two liquids can be substantially improved by
application of additional mechanical energy in form of superimposed agitation
or pulsation, resulting in higher turbulence near the drops and enlarged mass
transfer area due to smaller drop sizes.
Pulsed Columns. The simplest design of pulsed column is the pulsed
sieve plate column (Fig. 16A). A piston pump with an electrical or an air-pressure
drive causes the entire liquid content of the column to vibrate. Both phases are
alternately forced through the holes in the rhythm of the pulsation. The down-
comers used in the conventional sieve plate column are thus superfluous. The
dispersed phase does not coalesce above or below the plates to give a continuous
layer. The pulsation shock flings the drops off the plate faster than would be
possible as a result of only the density difference. This leads to a fine, uniform
drop distribution and hence enlarges the mass transfer area. Applications are
limited by the physical properties of the substances concerned and high through-
puts. Systems which emulsify easily cannot be processed in pulsed columns just
as sticky and greasy liquids. For very large throughputs (column diameter

Fig. 16. Pulsed and reciprocating-plate columns (Light phase dispersed). A) Pulsed
sieve-plate column; B) Reciprocating-plate column (Karr).
Vol. 1 LIQUID–LIQUID EXTRACTION 797

> 0.6 m), the pulsed volume and pulsing power become very high; however, pul-
sing units for large column diameters have been developed.
The separation effect of the column decreases with increasing diameter,
this has not yet been fully investigated. The plate spacing is 50 – 100 mm. Larger
plate spacings favor backmixing and hence reduce the efficiency of the column.
Smaller plate spacings lead to a strong decrease of the throughput. Sieve plates
have cylindrical holes and a triangular pitch. The relationship between the frac-
tional free area j, the hole diameter di and the pitch p is given by
pffiffiffiffiffiffiffiffiffiffiffiffi
j ¼ pd2i = 12 p2

For usual liquid – liquid systems sieve plates with hole diameters from 2
to 4 mm resulting in free areas of 23 to 40 % are recommended (6, Chap. 10),
the lower value of 23 % is used for high and the higher value of 40 % for low
interfacial tensions. Free areas less than 20 % are used for liquids with higher
viscosities (58).
The plate material has a pronounced influence on drop formation and hence
on the separation efficiency. The plate material and phase selection must be
matched in such a way that the plates are wetted by the continuous phase but
not by the dispersed phase, because this would lead to strong coalescence effects.
If, for example, an aqueous phase is to be dispersed into drops, metal trays are
unsuitable, but plastic plates are satisfactory. The sharp edged side of the sieve
plate, which is due to manufacturing, should exactly face the flow direction of the
dispersed liquid so that drops will shear at the edges of the holes. Gaps between
the sieve plate edge and the column wall should be avoided by precise construc-
tion or by applying edge-sealing elements made of materials which are wetted by
the continuous liquid.
In most cases, piston pumps or diaphragm pumps are used as the pulsators.
The pulsing frequencies are between 30 and 150 min1, with pulsing amplitudes
from 5 – 15 mm (relative to the column cross section). (Pulsing amplitude is
defined as the distance between upper and lower center of the pulsing motion
in the active part of the column.) The dispersed phase has to be fed into the col-
umn via a distributor. The mean drop diameter of the predispersed swarm
should be larger than the mean drop size in the active part of the column, other-
wise the hydrodynamics of the column would mainly be affected by the distribu-
tor. The holes in the distributor should be very sharp-edged to avoid very fine
drops. Even in packed columns, the separation effect can be substantially
improved by pulsing of the continuous phase, at the expense of a reduced
throughput. The increased performance results from greater shear forces caus-
ing a reduction in drop size and an increase in interfacial area. As in distillation
columns, the phenomenon of maldistribution in randomly packed columns of lar-
ger diameter can be overcome by use of structured packings. These packings lead
to a good cross-sectional mixing and have a higher capacity, typically a total flow
rate of both phases of approximately 100 m3m2h1 compared to 30 m3m2h1 for
a randomly packed column. Structured packings are particularly suitable for low
interfacial tension systems or for small density differences; separation efficiency
is nearly independent of scale-up. Hydrodynamics and mass transfer perfor-
mance are well investigated (59).
798 LIQUID–LIQUID EXTRACTION Vol. 1

Pulsing Units. The use of pulsed columns is currently limited to small and
medium throughputs. The column diameter is limited by the pulsing unit. For
very large throughputs, the pulsed volume becomes so large that the use of a
piston pump as the pulsing unit causes difficulties. Consequently, pulsing
units have been developed which operate without a piston pump. The pulsator
developed by MISEK (Fig. 17) is based on a centrifugal pump. The pump (e) is con-
nected on both the suction side and the delivery side to pressure vessels which
are alternately connected to the column by means of a rotating valve (c). With

Fig. 17. Pulsator with rotating valve. A) Principle diagram of compression stroke;
B) Principle diagram of suction stroke; C) Technical design (Bronswerk) a) Suction
drum; b) Discharge drum; c) Rotating valve; d) Column; e) Centrifugal pump.
Vol. 1 LIQUID–LIQUID EXTRACTION 799

Fig. 18. Hydraulic pulsator. a) Pulsing piston; b) Working piston; c) Rinsing; d) Hydrau-
lic aggregate with radial piston pump.

the valve setting shown in Figure 17 A, the pump delivers the liquid into the
column via the discharge drum (b), taking the rate delivered from
the suction drum (a). After changeover of the valve (Fig. 17 B), the pump
delivers from the column via vessel (a) into drum (b). The pump thus always
delivers continuously in one direction. Reversal of the direction of flow in the
pulse line is effected by the change-over valve (c) in accordance with the
pulse frequency. A pulsator based on this principle has been developed by
Bronswerk/DSM (Fig. 17 C), and is now in operation for columns with a maxi-
mum diameter of 2.7 m.
For large pulsed volumes, the hydraulic pulsator from Wepuko (60) is also
suitable (Fig. 18). It is flanged directly to the lower bottom of the column end and
acts in the direction of the column axis. The pulsing piston is connected via a
common piston rod to a working piston, the hydraulic drive of the piston is a
rapidly reversible radial piston pump.
Reciprocating-Plate Columns. In another approach for solving the
problem of large throughputs, the plates, and not the liquid, are made to recipro-
cate, so that the pulsing pump is eliminated. This principle led to the develop-
ment of the reciprocating-plate column which can be regarded as the converse
of the pulsed sieve plate column. In the pulsed sieve plate column, the trays
are fixed and the entire liquid content of the column is caused to vibrate; in
the reciprocating-plate column, the plates are moved. In both cases, the liquid
undergoes an oscillating motion relative to the plate.
In the Karr column (Fig. 16 B), sieve plates with a free area of about 60 %, a
10 – 15 mm hole diameter, and a 25 –50 mm plate spacing, are mounted on a cen-
tral shaft. The complete pack is moved up and down, driven by an eccenter.
800 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 19. Rotating disk contactor (RDC). a) Rotating disk; b) Stator ring.

The relatively large free area of reciprocating plate columns allows much
greater throughputs than conventional pulsed columns. However, recent inves-
tigations showed that pulsed sieve plate columns can be equipped with larger
free areas, too. Both types of pulsed columns show very similar performances
and operational behavior if they are equipped with the same free areas (61).
Research on the established types of pulsed columns has been very intensive
since the 1970s.
Columns with Rotating Internals. Rotating Disk Contactor (RDC). In
the RDC (Fig. 19 A), stator rings are mounted at close intervals. Disks are fitted
between the rings to a rotating shaft. The rotating disks fling the drops of the
dispersed phase off centrifugally, onto the stator rings where they are braked
and deflected. The drop distribution is excellent and can reach the emulsification
limit as the rotational speed increases.
The RDC has a relatively simple construction with a large free cross-
sectional area. On the one hand, it reaches large throughputs, on the other
hand it favors backmixing of phases causing a decrease of efficiency. A large
range of throughputs is possible without remarkable decrease of efficiency.
The loading limit is generally lower than in pulsed columns. The RDC is unsui-
table for mixtures which tend to form emulsions and for products containing dirt.
The separation efficiency decreases sharply with increasing diameter. Neverthe-
less, successful RDCs of up to 3500 mm in diameter have been constructed. They
Vol. 1 LIQUID–LIQUID EXTRACTION 801

Fig. 20. Columns with rotating internals. A) Kühni extractor; B) QVF extractor.

are thus suitable for greater throughputs than pulsed columns and are prefer-
ably used to operate with large product flows when the demands on separation
performance are only low as for instance in the oil industry or in wastewater
treatment (42).
As in pulsed columns, the wetting behavior also influences drop formation
in the RDC, and should therefore be taken into account when selecting a mate-
rial for the column internals.
Kühni Extractor. The Kühni extractor (Fig. 20 A) uses a turbine as the
mixing element. Perforated partition disks separate the extraction stages and
reduce backmixing.
This apparatus is also used for performing reactions between two liquid
phases; the residence time can be adjusted via the free area of the perforated
disk.
QVF-Extractor. The QVF extractor RZE (Fig. 20 B) uses blade stirrer as
mixing elements. The partition disks have meander-formed metal stripes at the
inner circumference to reduce the phase dispersion by channeling of both phases.
The operational behavior is similar to that of the Kühni extractor (see Fig. 20 A).
Graesser Contactor. The Graesser contactor (Fig. 21) is another type
of rotating equipment. Here, not only the internals but the entire apparatus
rotate. Internal half-tubes are fitted as scooping elements in a horizontal
rotating cylinder. Due to the rotation, the half-tubes transport part of
the heavy phase into the overlying light phase, and vice versa. To reduce back-
mixing, the extractor length is divided into chambers by partition disks.
Because of its low dispersing effect, this equipment is particularly suitable
for systems which tend to emulsify. The throughput is low compared with
other extractors.
Mixer – Settlers. Mixer – settlers consist of single or several stages in
series (Fig. 22 A), each stage consisting of a stirred vessel and a settling vessel
(Fig. 22 B). Various arrangements are possible such as cocurrent, counter-
current, or crosswise flow of phases. The mixer – settler is the only extractor
802 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 21. Graesser contactor.

Fig. 22. Principle of a mixer – settler. A) Flow Sheet; B) Arrangement.


Vol. 1 LIQUID–LIQUID EXTRACTION 803

that reaches stage efficiencies of nearly 100 % and is almost insensitive for load
fluctuations.
This extractor has the advantage that backmixing can be completely pre-
cluded due to the complete phase separation in each stage. As a result, there
are no transfer problems in scale-up. The favorable efficiency is almost comple-
tely retained even for large throughputs. Any number of separation stages can be
built, and, moreover, throughput does not appear to be limited. These advan-
tages account for the widespread use of mixer – settlers in industrial processes
for many decades. However, early mixer – settler equipment was very complex.
Each separation stage consisted of a stirred vessel and a settling vessel. The two
phases had to be pumped between the individual stages. Each settler needed an
interface level control. The horizontal arrangement resulted in a large area
requirement.
In an effort to overcome these disadvantages of the mixer – settler batteries
without foregoing any of their advantages, a number of different designs were
developed. Mixer – settlers are used in two different areas. Industrial-scale
equipment is used to obtain extremely high throughputs when a small number
of stages is needed, eg, in the extraction of metal salts. Laboratory-scale devices
are employed in process development to check the number of calculated stages as
a function of the ratio solvent.
Large-scale mixer – settlers were originally developed to extract copper
from dilute aqueous solutions. The basic design is a box-type mixer – settler
according to Denver (Fig. 23). The mixing and separating zones are combined
in a single unit and are separated by an overflow weir for the light phase and
an underflow slit for the heavy phase. The individual stages are arranged so
that the outlet of the settler also acts as the inlet of the mixer. The agitator

Light Light
phase out phase in

Heavy Heavy
phase in phase out
Section A-B

Mixing zone Setting zone


Light Heavy
phase out phase out

Heavy Light
phase in phase in
A B

Fig. 23. Box-type mixer – settler (Denver). A) Side view; B) Ground plain.
804 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 24. Industrial mixer – settlers. A) Davy McKee mixer – settler; B) IMI turbine
pump-mix unit (cross section A – B – C – D shown in the lower part); C) Lurgi tower
extractor.

has two functions, it disperses and conveys the liquid phases. These functions
may conflict with one another. The rate of rotation necessary to overcome the
conveying height may be too high when unstable systems must be dispersed,
causing problems in phase separation, eg, a high degree of entrainment. Mixing
pumps with low shearing forces, and thus a low mixing intensity, have been
developed to avoid these difficulties. Examples are the Pumpmix impeller used
with the Davy McKee mixer – settler (Fig. 24 A) and the IMI turbine pump-mix
unit (Fig. 24 B).
A further development of the Davy McKee mixer – settler has been
described (43). Each stage consists solely of a vessel in which mixing and separa-
tion occur. The product inlet and the agitator are located in the middle and create
a dispersion zone. The light phase separates out at the top and the heavy phase
at the bottom.
In the Lurgi tower extractor, the battery of adjacent settlers is replaced by a
space-saving tower design in which the settlers are arranged on top of each
other. External centrifugal pumps are used for mixing (Fig. 24 C). Like all
mixer –settlers, this device is suitable for very large throughputs provided that
the liquid system can withstand the high energy input of the mixing pump. Back-
mixing to a certain degree is unavoidable since the stages have hydrodynamic
connections.
Multistage mixer – settlers are a valuable aid in process development on a
laboratory scale. They can be integrated in ‘‘mini plants’’; at this stage of devel-
opment no decision should be made which type of equipment is suitable for the
Vol. 1 LIQUID–LIQUID EXTRACTION 805

Fig. 25. Laboratory mixer – settlers. A) Normag mixer – settler; B) Schott/QVF mixer–
settler–>Schott/QVF mixer – settler.

large-scale plant. Several examples of laboratory mixer – settlers are shown in


Figure 25.
Centrifugal Extractors. Centrifugal extractors are used when the density
difference between the two phases is insufficient to produce a countercurrent
under gravity. The best known of these devices is the Podbielniak centrifugal
extractor (Fig. 26 A), in which cylindrical perforated sheets are concentrically
arranged as a compact rotor on a high-speed shaft. The heavy phase is fed in
via the shaft and moves outwards due to the centrifugal force. The light phase
moves in the opposite direction. Similar to columns, either the light phase or
the heavy phase can be dispersed into drops, as desired. If the light phase is to
be dispersed, the main interface must be located near the shaft. Depending on
the system the number of stages which can be combined in one unit is limited
to 3 to 4 theoretical stages.
The Robatel centrifugal extractor is based on a totally different principle
and allows greater numbers of stages (Fig. 26 B). Disks serving as the mixing ele-
ment are centered on a vertical stator. The housing is divided by annular disks
into a number of stages and rotates around the central shaft. Each stage consists
of a mixing chamber and a settling chamber. The disks each dip into the mixing
chamber.
Centrifuges are more complex than columns, their investment and mainte-
nance costs are correspondingly high. They can therefore only stand comparison
with columns in cases where they have advantages which no column can provide.
Phase separation is possible even in systems having a density difference as low
as 15 kg/m3. Systems with large differences in viscosity can be processed, since
the high shear forces effect intensive mixing.
As a result of its compact design, the centrifugal extractor is a space-saving
apparatus with an extremely small operating volume and a short residence time.
These advantages can be desirable if expensive solvents or substances of limited
stability are used. The steady state is reached soon after starting. This extractor
806 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 26. Centrifugal extractors. A) Podbielniak extractor; B) Robatel four-stage counter-


current extractor.

is therefore suitable for batchwise operation and frequent changes of products.


Extraction under pressures of up to 5 MPa is possible.
3.2. Fluid-Dynamic Fundamentals. A comprehensive treatment of
fluid-dynamic fundamentals is given elsewhere.
Problems and Process Strategy. In Section 2.5, it was shown how the
number of theoretical stages and the flow diagram of the unit operation ‘‘extrac-
tion’’ are calculated on the basis of equilibria. The selection and design of appa-
ratus are based on fluid dynamics and are not carried out until the
thermodynamic design is completed. At present, model experiments are still
Vol. 1 LIQUID–LIQUID EXTRACTION 807

necessary for the design of extraction apparatus. For this reason, a knowledge of
the operating characteristics of the various types of apparatus is of great interest
so that the number of experiments can be substantially reduced, and preselection
and preliminary sizing can be carried out at a relatively early stage of process
development.
The fluid-dynamic design of an apparatus is divided into the following
areas:

1. Selection of suitable internals with respect to material of construction,


shape, and size.
2. Determination of the operating conditions with respect to phase selection,
pulsing, and loading.
3. Sizing of the column diameter in accordance with the required throughputs
of feed and solvent.
4. Sizing of the column height in accordance with the actual separation problem.

However, these four problems cannot be considered in isolation. Thus, both


the internals and the operating conditions affect (1) the loading limit and hence
the column diameter and (2) the separation efficiency and thus the column
height. Measures for increasing the loading limit have an adverse effect on the
separation efficiency, and vice versa.
The sizing of the column diameter depends on the flooding point, ie, the
maximum loading limit at which a column can be operated. On flooding, accumu-
lation of the dispersed phase occurs at individual points within the column and
blocks countercurrent flow. Consequently, the light phase appears in the dis-
charge of the heavy phase, and vice versa. This phenomenon is detectable by
holdup measurements. On flooding, the holdup increases continuously and indi-
cates an unstable operating state.
The flooding limit depends on the drop size. Small drops rise less rapidly
than large drops and therefore cause a greater holdup. The more energy is sup-
plied to the system — either by pulsing or by rotating internals — the smaller are
the drops, so that the flooding limit decreases.
Internals also effect dispersing. The smaller their open area, the more the
flooding limit decreases.
The sizing of the column height is determined by the mass transfer, which
is described by the well-known mass-transfer equation: the rate of mass trans-
ferred is equal to the product of the mass-transfer coefficient, the mass-transfer
area and the concentration difference between the two phases. A great progress
in scientific research has been achieved as regards measurement and calculation
of the quantities of influence on mass transfer (see for instance (6,45–49,51,62–
65,68). But in practice, the various influences on mass transfer are best summar-
ized in the expression ‘‘height equivalent of a theoretical stage’’ (HETS), as a
measure for the separation efficiency of the column, as mentioned in Section 2.5.
Measurements in pilot-plant columns will be used below to demonstrate the
way in which various parameters affect the loading limit and separation effi-
ciency. The test systems recommended by the European Federation of Chemical
Engineering (EFCE) are used for this demonstration (44).
808 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 27. Loading limit as a function of the pulsing frequency at different free area for
toluene – water (50). a ¼ 6 mm; vd/vc¼ 1; ~ j= 0.60; d ¼ 6.5 mm; &j = 0.40; d ¼ 4.0 mm;
*j ¼ 0.23; d ¼ 2.5 mm.

Operating Characteristics of Pulsed Columns and Columns with Rotat-


ing Internals. Loading Limit. Figure 27 shows the loading limit vC þ vD as
a function of the pulsing frequency for plates of small free area j for the
toluene – water system, eg, for high interfacial tension. vc and vd are the volume
flows of continuous (c) and dispersed (d) phase related to the total cross-sectional
area of the column, the so-called superficial velocities. This flooding point
curve starts at the origin of the coordinates; in other words, without pulsing,
no throughput is possible. With increasing pulsing, the throughput also rises
up to a maximum. This part of the curve is called the lower flooding limit;
here, pulsing has mainly a transport function. With increasing frequency, how-
ever, the dispersing function predominates, the drops become smaller and the
loading limit falls. The interesting region in practice is that in which dispersing
predominates.
As is to be expected, the loading limit rises with increasing free area over
the entire pulsing range, but the position of the maximum remains unchanged.
At the largest free area of j = 0.6, a modest throughput is possible even without
pulsing.
The effect of the geometry of the internals can also be shown for pulsed
packed columns. There is no lower loading limit. At frequency f ¼ 0, eg, the
unpulsed column, the loading limit has its maximum value, and decreases
with increasing frequency (53).
Of all the physical properties, the interfacial tension has the greatest effect
on the fluid dynamics of an extraction column. Figure 28 shows a comparison of
three test systems with different interfacial tensions. As can be seen, the
throughput optimum is shifted to higher pulsing frequencies with increased sur-
face tension, ie, high surface tensions need a high energy supply.
In addition to the physical properties, the phase ratio is another parameter
determined by the process. Decisions can only be made as to which of the two
phases is to be dispersed (the organic or the aqueous phase) and whether in
the greater or smaller quantity. Figure 29 shows the effect of the parameter
vd/vc (52). As the phase ratio increases, the throughput decreases.
Vol. 1 LIQUID–LIQUID EXTRACTION 809

Fig. 28. Loading limit as a function of pulsation frequency at different interfacial


tension (50): j ¼ 0.23; vd/vc¼ 1. a) Butanol – water: s ¼ 1.8 mN/m; b) Butyl acetate – water:
s ¼ 13.5 mN/m; c) Toluene – water: s ¼ 36.1 mN/m.

A further parameter affecting the loading limit and separation efficiency is


the direction of the mass transfer. Curve b in Figure 30 A shows the flooding
point curve of the pure two-component mixture toluene – water, where no
mass transfer takes place (54). Curve a for mass transfer from the dispersed
organic phase to the continuous aqueous phase lies substantially higher by com-
parison. The reason for this is an increased tendency to coalescence and hence
formation of larger drops. In the converse case (mass transfer from the continu-
ous to the dispersed phase), the interfacial tension and hence the drop size are
reduced by the component being transferred, leading to a lowering of the flooding
point (53,55). The effect of the direction of mass transfer on the drop size also
causes differences in the separation efficiency. In the case of mass transfer

Fig. 29. Relative loading of a pulsed sieve plate column as a function of flow ratio (52)
* Toluene – water; & MIBK – water; ~ Butyl acetate – water; 3 n-Butanol – water .
810 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 30. Influence of mass transfer on loading limit and separation efficiency for the
toluene – acetone – water system (j = 0.60). ~ Mass transfer from the disperse to the con-
tinuous phase; & System without mass transfer; * Mass transfer from the continuous to
the disperse phase.

from the continuous phase to the dispersed phase (into the drop), smaller HETS
values result, ie, better separation efficiency is obtained (Fig. 30 B).
PILHOFER (56) has shown that the loading limit of extraction columns
depends only on the drop size. Accordingly, high loading limits can be obtained
only with relatively large drops.
Separation Efficiency. However to obtain a high separation efficiency a
large mass transfer area, eg, the greatest possible holdup of dispersed phase, is
required. The holdup rises with decreasing drop size. Any input of energy has the
purpose of producing the optimal drop dispersion. The interrelation of separation
efficiency and pulsing can be demonstrated by measurements on the toluene –
water – acetone system, which has a high interfacial tension (41). Figure 31
shows the measured height of a theoretical stage (HETS) for this system as a
function of the pulsation frequency, for the three different plate types (curves
a – c). In all cases, the HETS decreases as expected with increasing pulsing
intensity, eg, pulsing enhances the separation efficiency. The curves show an
Vol. 1 LIQUID–LIQUID EXTRACTION 811

Fig. 31. Separation efficiency measured as a function of interfacial tension (41) a) – c)


Toluene – water – acetone; d) Butanol – water – succinic acid þ j = 0.60; * j ¼ 0.40;
! j ¼ 0.23.

important influence of the plate geometry: with increasing free area, the separa-
tion efficiency decreases sharply.
The butanol – succinic acid – water system shows very favorable HETS
values because of its extremely low interfacial tension (curve d). The effect of
free area is negligible. This means that, for systems of low interfacial tension,
the free area can be increased in favor of higher throughputs, without a signifi-
cant loss in separation efficiency.
A comparison of these two systems clearly shows the pronounced influence
of the interfacial tension on separation efficiency. This fluid-dynamic aspect
should also be taken into account when selecting solvents for extraction pro-
cesses. Hitherto, this selection has been made almost exclusively under thermo-
dynamic criteria.
The relationship between plate geometry, interfacial tension, and sepa-
ration efficiency, as described here for a pulsed sieve plate column, qualitatively
applies also to reciprocating-plate columns and pulsed packed columns.
Generally speaking, in systems with a high interfacial tension, particular atten-
tion must be paid to the choice of the internals, a higher energy input is also
necessary.
Fluid Dynamic Calculation Methods. The calculation methods for deter-
mining the loading limit can be classified in two categories. The first category
consists of empirical correlations which are based on physical and geometrical
data, and energy input. The second category is based on fluid-dynamic models
and presupposes knowledge of the drop size. Both method classes are applicable
only to two-component systems and thus neglect the effect of mass transfer, as
shown in Figure 30. The effects of wetting and of drop coalescence are likewise
disregarded. Moreover, the empirical correlations in particular are frequently
812 LIQUID–LIQUID EXTRACTION Vol. 1

restricted to certain geometries and cannot be extrapolated to other designs,


they can therefore only be used to obtain an approximate estimate. For final
equipment sizing, an experimental determination of the loading limit in the
particular type of equipment with the original process streams is absolutely
necessary.
Packed Column. Unpulsed packed columns have no industrial significance
any longer. A review on empirical equations to calculate holdup, drop size, flood-
ing, mass transfer, and axial dispersion is given in (6, Chap. 8). Performance can
be improved by mechanical pulsation of the continuous phase at the expense of a
reduced throughput. For a rough estimate, pulsing within reasonable ranges
may be assumed to cause the load limits to fall by 20 – 30 % as compared with
the unpulsed column. Still, no equations for calculation are known.
Sieve Plate Column. The conventional sieve plate column has been
investigated by PILHOFER (66) for toluene – water systems. KIRSCHMANN (69)
expanded the investigations of PILHOFER to liquid systems with lower surface
tensions and density differences than toluene – water and to dual flow plates.
For calculation see (69).
Sieve Plate and Rotating Columns. Special empirical equations are
given in the literature for each type of extraction columns in order to calculate
the flooding condition, so, eg, for sieve plate columns in (66) and (67), for pulsed
sieve plate columns in various chapters of (6), for rotating columns in (6) and (5).
However, since important effects are disregarded as mentioned before, here it
will be sufficient to introduce a rough method based on a fluid dynamic model
(66). The model, originally developed for unpulsed sieve tray columns, proved
to be applicable to all types of columns. It is assumed that the liquids occupy a
share of the cross-section according to their holdup e and flow toward each other
as layers. If their superficial velocity v, ie, their volumetric flow rate V related to
the complete column cross section S, is

Vc Vd
vc ¼ ; vd ¼ ð1Þ
S S

they have an effective velocity vc/(1  ed) or vd/ed. That yields the relative velocity
vr, also called slip velocity, between the two liquids
vd vc
vr ¼ þ ð2Þ
ed 1  ed

Either vr or ed can be calculated from Equation (2) if one or more relation-


ships between the two terms are known. In (45) a simple empirical equation is
given to calculate vr
vr
¼ ð1  ed Þk ð3aÞ
vp
k ¼ 4:6  0:13 ln Ar ð3bÞ

when 1 < Ar < 107 and 2.5 < k < 4.6


vP is the individual drop velocity given by (66)
Vol. 1 LIQUID–LIQUID EXTRACTION 813

for circulating drops

Re p ¼ KL0:15 ðAr0:523 KL0:1435  0:75Þ ð4aÞ

if

1:83 KL0:275 < Ar < 3:94 KL0:275

for oscillating drops

Re p ¼ KL0:15 ð4:18 Ar0:218 KL0:00773  0:75Þ ð4bÞ

if

Ar > 372:9 KL0:275 ð4cÞ

The dimensionless numbers are defined as follows:


particle Reynolds number

v p dp
Rep ¼ ð5Þ
nd

Archimedes number

dp 3 g %c  %
Ar ¼ ð6Þ
Zc 2

characteristic number of liquid

%c 2 s3
KL ¼ ð7Þ
 % gZc 4

density difference

 % ¼ j%c  %d j ð8Þ

Mathematical formulation implies the fact that an increase of the flow rate
at flooding cannot increase the drop holdup:

dvd
¼0 ð9aÞ
ded

when vc = constant,

dvc
¼0 ð9bÞ
ded

when vd = constant.
814 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 32. General loading diagram (52,56).

With Equation (2) for the slip velocity vr the equations for calculating the
superficial velocities vc, vd at the flooding point are obtained in a dimensionless
form:
 1=3  
%c ed dRer
vc ¼ Ar1=3 ð1  ed Þ2 Rer þ ð10aÞ
 % vc g ded
 1=3  
%c 1=3 2 dRer
vd ¼ Ar ed Rer  ð1  ed Þ ð10bÞ
% vc g ded

Equations (10a) and (10b) together with Equation (3) for vr and Equation (4)
are the basis for flooding point diagrams according to Figure 32. Ordinate and
abscissa contain the superficial velocities vd and vc, ie, the phase flow rates, mod-
ified by physical data groups to a dimensionless form. The diagram shows the
calculated flooding curves of circulating drops as solid lines with the parameter
Ar. Dashed lines of constant phase ratio vd/vc and the total throughput as the
sum of vd and vc are also shown.
It can be seen from Equation (4b) that ReP is approximately proportional to
Ar0.3 and therefore to dP. That means that the drop velocity vP is nearly indepen-
dent of drop diameter dP in the region of oscillating drops. Therefore Figure 32
can be used also for oscillating drops if Ar is replaced by KL using Equation 4c.
Otherwise the diagram is valid for a certain KL value in the region of circulating
drops, in case of Figure 32, KL = 1011.
The dependence of drop diameter dP upon parameters of influence must be
known in order to use Figure 32 and to calculate flooding curves. Many scientific
investigations on drop disintegration in various extractors are found in the lite-
rature, but often without considering the influence of mass transfer and the
interaction of dispersing and coalescence. Furthermore, particularly for higher
Vol. 1 LIQUID–LIQUID EXTRACTION 815

interfacial tension systems a considerable height of column is required to reach a


steady-state drop size, and this height varies with liquid system and inlet
arrangement for the dispersed phase. Since population balances are not used
in this introductory article, the concept of mean drop sizes is applied – normally
reported as the Sauter mean or surface percentage mean drop size – defined by
P
npi d3pi
i
d32 ¼ P ð11Þ
npi d2pi
i

where npi represent the numbers of drops of the class i of diameter dpi.
Finally, discussion will be restricted to empirical equations by KUMAR and
HARTLAND (6, Chap. 17) which give the best fit to experimental data and have
the broadest range of validity:
Pulsed sieve tray and reciprocating extractor
 0:10    
d32 0:74 h a f %1=4 a f %1=4
rffiffiffiffiffiffiffiffiffiffi ¼ C1 e  exp 3:00 þ exp 28:65
s 0:05 m gs1=4 1=4 gs1=4 1=4
%g
ð12Þ

The best values of the parameter C1 are 1.51, 1.36, and 2.01 for no mass
transfer, c ! d, and d ! c direction of transfer, respectively. The product a  f is
considered as the agitation variable in the equation, since the fit could not be
improved if a and f were treated separately. The average absolute value of
the relative deviation in the predicted values of d32 from the experimental points
is 16.3 %.
Rotating disk extractors

  0:12  0:16 !0:59    


d32 C1 Zc %d DR 2 %c g h 0:25 D 0:46
¼ pffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffi 
DR 0:07 þ FrR s%c DR %c s DC DR
ð13Þ

The optimized values of C1 are 0.63, 0.53, and 0.74 for no mass transfer,
c ! d and d ! c directions of transfer, respectively. The effect of holdup is ignored
due to lack of data. Equation (13) predicts the drop size with an average absolute
value of the relative deviation of 23 %.
Kühni columns

 0:20 !0:24
d32 0:37 0:11 Zc D R 2 %c g
¼ C 1 e nS ½0:14 þ exp ð18:73FrR Þ  pffiffiffiffiffiffiffiffi
DR s%c DR s
ð14Þ

The values of the constant C1 are 9.81  102 for no mass transfer and c ! d
transfer and 0.31 for d ! c transfer. The stage number nS, which varies from 2 to
816 LIQUID–LIQUID EXTRACTION Vol. 1

17 in the present set of data, shows a rather weak effect on drop size. Equa-
tion (14) predicts the drop diameter with an average absolute value and relative
deviation of 17.6 %.
3.3. Apparatus Design. Extraction apparatus can be approximatively
designed purely by calculation using fluid-dynamic models (49,70). For reliable
design, however, pilot-scale tests are indispensable. If possible, the test should
be carried out with the original operating mixture in the type of apparatus
which is also envisaged for the large-scale unit. Desirably, the geometry of the
internals should be the same in the pilot-scale and industrial apparatus. How-
ever, this condition cannot always be satisfied. Generally, a column diameter
of at least 60 – 80 mm is necessary for design tests. Tests in smaller columns can-
not be used as the basis for equipment sizing (for criteria on equipment selection,
see Section 3.4). The number of tests can, however, be substantially reduced if
knowledge of the operating characteristics is utilized.
The physical data and the phase ratio are determined by the process used,
the following parameters are available for optimization: pulsing or rotation, geo-
metry of the internals, material of the internals, and phase selection, eg, the
direction of mass transfer. Energy input and the geometry of the internals are
selected on the basis of the physical data.
Internals and Operating Conditions. The following guidelines apply for
the selection of the free area in pulsed columns. An increase in the free area is
recommended in systems with a low interfacial tension where it leads to a great
reduction in the column volume. At moderate interfacial tension, the column dia-
meter can be reduced by a greater free area, but at the expense of the column
height. In the case of extremely high interfacial tension, there is no point in
increasing the free area.
The material of the internals should be selected so that the continuous
phase, but not the dispersed phase, wets the plates. As a rule of thumb, metal
plates can be used when the organic phase is dispersed; if the aqueous phase
is dispersed, the internals should be of plastic, both in pulsed columns and in
rotary-agitated columns.
In deciding which phase should be dispersed, the direction of mass transfer
must be taken into account. In accordance with Figure 30, mass transfer should
be from the continuous phase to the dispersed phase, if possible. In the final
phase selection, however, fluid-dynamic considerations frequently conflict with
operating and safety requirements. The position of the interface in the column
determines the phase which is dispersed. Figure 33 summarizes the possible
combinations.
The separation efficiency determines the column height and the load limit
determines the column diameter. For optimization, the two parameters are
related by the ratio y ¼ HETS=ðvc þ vdÞ which can be used to optimize the column
internals and energy input. A minimum value of this expression is equivalent to
a minimum column volume. Figure 34 shows that optimum pulsing is obtained
between the maximum of the flooding point curve and the minimum of the HETS
curve. An analogous procedure is used for rotary agitated columns. The design
point for industrial-scale columns is chosen to be 75 – 80 % of the load limit.
The operating point with respect to load and pulsing (rotation) is fixed by this
procedure.
Vol. 1 LIQUID–LIQUID EXTRACTION 817

Fig. 33. Selection of phases and plate materials.

Column Diameter. Assuming the same column geometry, the specific


throughput of the test column is adopted for the large-scale column.

ðvc þ vd Þ2 ¼ ðvc þ vd Þ1

with
ðvc þ vd Þ1 ¼ 0:75  ðvc þ vd Þ1 max
Index 1¼ pilot scale
Index 2¼ large scale
This gives the column diameter:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u m _F m _S

u
u4 þ
t %F %S 2
D2 ¼
p  ðvc þ vd Þ2

Column Height. The separation efficiency of all columns decreases with


increasing column diameter, because the residence time distributions of the
phases increase. Therefore values of the height of a theoretical stage, measured
for a laboratory column, need to be corrected by empirical scale-up equations to
be applicable for industrial-scale columns. However, scientific investigations of
this problem are rather poor and mainly empirical knowledge of industrial
engineers exists. The different chapters of (5) list formulae from experience
with countercurrent extractors as described here, including
818 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 34. Optimization of pulsation.

Pulsed sieve tray columns (71)


HETS2
¼ exp ½1:64 ðD2  D1 Þ ð15Þ
HETS1
(in Equation 15 the factor 1.64, given by THORNTON, is only valid for small col-
umns. Figure 35 shows an empirical dependence of the Thornton factor from
the column diameter published by REISSINGER (49))

Fig. 35. Dependence of Thornton factor on column diameter.


Vol. 1 LIQUID–LIQUID EXTRACTION 819

Reciprocating columns (72)


 0:38
HETS2 D2
¼ ð16Þ
HETS1 D1

When the plate geometry and amplitude of the vibration are left unchanged
in the large-scale column, the frequency should be reduced, according to:
 0:14
f2 D1
¼
f1 D2
The following applies to stirred columns (73):
 0:33
HETS2 D2
¼
HETS1 D1

3.4. Criteria for Equipment Selection. For a given separation pro-


blem, equipment is selected essentially according to three criteria: physical prop-
erties, required number of separation stages, and throughput. Further operating
aspects are frequently important, for example, flexibility regarding load varia-
tions or product quality, sensitivity to small amounts of solids, suitability for
batch operation.
The physical properties of interest in extraction are the density difference
between the two phases, their viscosity, and interfacial tension. These three
parameters affect dispersibility and phase separation behavior.
High density difference, viscosity, and interfacial tension make dispersing
more difficult; systems with such physical properties require a higher dissipation
of energy than in the converse case. Phase separation is favored by a high density
difference and interfacial tension, whereas a high viscosity makes phase separa-
tion more difficult.
The lower limit for the density difference can be taken to be 50 kg/m3. Mix-
tures with higher density differences can generally be processed in columns and
mixer – settlers. If the density difference is smaller, a centrifugal extractor must
be used.
A low interfacial tension leads to phase separation problems. In industrial
units, approximately s = 1 – 2 mN/m can be taken as the lower limit. Therefore,
mixer – settlers can certainly not be used for mixtures of relatively low interfa-
cial tension; because of the shear field generated, centrifugal extractors also have
a dispersing effect which should not be underestimated. In such cases, the energy
input must be very carefully controlled. Rotary-agitated columns and pulsed col-
umns can be used, in which the energy input can be adapted to the physical prop-
erties by the speed of rotation or the pulse frequency.
The selection of equipment is also affected by the number of theoretical
stages (NTS) required for a given separation problem. Only small numbers of
theoretical stages can be achieved in columns without energy input, because of
their poor separation efficiency. Pulsed columns and rotary agitated columns
give higher numbers of separation stages. They are limited by the pulsing
units or by the lengths of the continuous shaft. In centrifugal extractors, the
820 LIQUID–LIQUID EXTRACTION Vol. 1

Robatel design provides more separation stages than, for example, the Podbiel-
niak extractor. Mixer – settlers are constructed in modules so that there is vir-
tually no limit to the number of separation stages.
A first estimate of the maximum possible throughputs in the individual
extractor types can be obtained by considering the largest diameters (Dmax)
which have hitherto been installed:

Dmax, m
Reciprocating-plate column 1.5
Pulsed column 2.7
Kühni extractor 2.5
QVF extractor 2.8
Rotating Disk Extractor 3.0
Mixer – settler tower 8.0

Accordingly, the highest throughputs can clearly be achieved with the


mixer – settler. In pulsed columns, the diameter is limited by the pulsing unit.
In rotation columns, the separation efficiency at large diameters is no longer
satisfactory. However, informations about maximal cross-sectional loads which
are necessary to obtain definitive throughput data of a type of extractor are
not given in the list.
The literature contains some publications which recommend criteria for
the selection of the best apparatus. However, most of these recommendations
are only qualitative, and some are even contradictory; no studies of apparatus
have so far allowed an objective comparison. Only the standardization of
test systems for extraction has allowed fluid-dynamic studies to be carried
out under comparable conditions (44). The following systems have been recom-
mended:

Toluene – water – acetone


Butyl acetate – water – acetone
Butanol – water – succinic acid

These systems provide an objective basis for comparison of equipment. Neverthe-


less, investigations by different authors are not always mutually comparable
with respect to the absolute HETS values. Evidently, the column length and
the phase ratio have an unexpectedly large effect.
For the toluene system, PILHOFER compared the different operating charac-
teristics of various groups of equipment. Analogous studies on the liquid system
butyl acetate – water – acetone were carried out by GOLDMANN (64). Figure 36
shows a diagram of GOLDMANN with the number of theoretical stages NTS per
meter of column height, as a function of the load. The figure contains data for
QVF and Kühni columns – the latter with five different free areas of the perfo-
rated partition disks – and for pulsed sieve plate extractors (PSE) with five dif-
ferent free areas j and one reciprocating column (Karr). The rotary agitated
columns of QVC and Kühni with the smallest free areas j clearly achieve the
highest NTS values, but allow only low loads. Also, the other Kühni columns
Vol. 1 LIQUID–LIQUID EXTRACTION 821

Fig. 36. Comparison of the number of theoretical stages NTS in dependence of total
load for Kühni, QVF, Karr and pulsed sieve plate extractors (PSE) (64) (system butyl
acetate – water – acetone, phase ratio vd/vc & 1.12, measurements of various authors cited
in (64)).

show higher NTS values than the pulsed columns but smaller ranges of load.
Generally speaking, all measures which improve the separation efficiency reduce
the load capacity.
A comparison between pulsed sieve trays and structured packings with the
test system toluene – water – acetone was published (74). Figure 37 shows the
diagram with the extractor efficiency in dependence on the load. It is obvious
that structured packings have the broadest range of load capacity and reach
the same or higher efficiencies as pulsed sieve trays. From the results for two
structured packings without pulsation the positive influence of pulsing can
clearly be seen: The pulsed unstructured packing, Pall rings 15  15 mm, has a
clearly smaller load range.
In Figure 38, the range of loads is compared for sieve trays and structured
packings with pulsation (75). According to that, packed columns have significant
advantages. As shown in (75), the investment costs of pulsed structured packed
columns are 30 % lower than the ones of pulsed sieve tray columns and 40 to 50 %
lower than pulsed unstructured packed columns.
In (76) a systematic study on extractor selection is given.
Adaptation of the Column Geometry to the Separation Problem. In
the early days of extraction, various types of apparatus were restricted to clearly
defined geometries and were thus sharply distinguished from one another.
Recent apparatus developments allow adaptation of the column geometry to phy-
sical data, both for pulsed columns and rotary-agitated columns. The boundaries
are thus no longer distinct and a general fluid-dynamic description of related
types of apparatus is feasible. The optimization of operating conditions and the
geometry of the internals is as equally important as the selection of a suitable
type of apparatus.
822 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 37. Comparison of efficiency versus load of packed columns with and without
pulsing and pulsed sieve tray extractors. a  f ¼ pulsation intensity, Bf ¼ flooding load,
j ¼ relative free area.

In practice the final selection of equipment is often based on product-


specific conditions or plant requirements which include sensitivity to solids,
tendency of the product to polymerize, and safety aspects, for example with
regard to solvent holdup.

Fig. 38. Range of load of pulsed structured packed columns in comparison to pulsed
sieve tray columns in dependence of the mass density difference for aqueous systems
and systems similar to water.
Vol. 1 LIQUID–LIQUID EXTRACTION 823

Untypical Operating Conditions. Usually extractors operate with


nearly the same volume flow for both liquids and for liquids with viscosities
similar to that of water. Therefore, the equations given before are only valid
for such conditions. Some approaches for untypical operating conditions are
given below.
RAUSCHER (62) investigated pulsed sieve tray extractors with phase ratios
between 0.06 and 10 and the liquid system n-butyl acetate (o) – water (d) –
methyl isopropyl ketone (MIPK), which enriches MIPK in equilibrium preferable
in the organic phase. He confirmed the experience that the separation efficiency
declines for phase ratios unequal 1 and found the following causes: By dispersion
of the smaller volume flow (phase ratio  1), stronger backmixing of droplets by
the continuous phase will occur, leading to a broader drop size distribution. By
dispersion of the larger volume flow (phase ratio >> 1), the wake behind the
drops causes remarkable backmixing of continuous phase. These phenomena
have to be taken into account by an adequate calculation via drop population
balances.
G. WAGNER (58) investigated the fluid dynamics of pulsed sieve tray and
Kühni rotating extractors using liquid systems with viscosities up to 50 mPa  s,
mixed from paraffin oil and n-hexane as organic phase and of glycerol and water
as aqueous phase. I. WAGNER (77) continued the work with mass transfer inves-
tigations using the system tridecanol as drop phase (Z = 42 mPa  s), water, and
propanol. The experiments showed that the mean Sauter drop diameter
increases with increasing drop viscosity; holdup is decreased, while the flooding
curve is shifted to higher throughputs. In case the continuous phase is the more
viscous one, the drop swarm contains a larger fraction of small droplets causing
increased droplet entrainment by the continuous phase. In stirred columns, the
drop size can be influenced in a broader range than in pulsed columns; yet, more
energy will be required for a certain drop dispersion. Intensive stirrers as in
Kühni and QVF extractors should be preferred to stirrers in RDC extractors.
In the case of a more viscous continuous phase, the drop dispersion is nearly
independent of drop viscosity and is hardly influenced by the rotating or pulsat-
ing internals. Here, lower intensity dispersers are preferable, such as RDC
extractors.
The mass transfer to drops is distinctly decreased by increasing drop visc-
osity, developing towards the state of transfer to rigid spheres, and depends on
the contact time. It can be improved by suitable internals and by supply of
energy.

4. Phase-Separation Equipment

4.1. Gravity Settlers without Inserts. Each extraction stage consists of


two steps: (1) dispersing and (2) phase separation by coalescence of drops, thus
generating a homogeneous layer of drop phase. Figure 39 A shows a vertical set-
tler at the top of a column for the case that the drop phase has a lower density
than the continuous phase. The drops rise against the sinking continuous liquid
and are blocked by the upper level of the continuous phase, which is the principal
interface. There, coalescence of the dispersed phase will occur both between
824 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 39. Vertical (A) and horizontal (B) gravity settlers for drop separation from a
continuous liquid.

adjacent drops and at the principal interface. The coalesced drop phase leaves
the extractor at the top of the column.
In a horizontal settler, as shown in Figure 39 B, both phases flow in the
same direction. Drops of lighter dispersed phase rise to the interface where
they coalesce to form a homogeneous layer. In the lower part of the settler a
layer of the continuous phase remains, growing in flow direction due to decreas-
ing drop numbers. The wedge-shaped zone of dispersion should end before the
placement of phase outlets.
As the accumulation of impurities in technical liquid systems changes
important properties such as viscosity and surface tension, an exclusively theo-
retical modeling, as developed by many authors (see for instance (80)), is insuffi-
cient. However, the length of a horizontal settler can be estimated by a combined
procedure of calculations and simple and cheap batch settling experiments (79).
The calculation is based on a model of the physical processes in a settler, while
experiments supply one empirical parameter to include surface phenomena that
could not be modeled so far (see Figs. 40 and 41).

Fig. 40. Experiments on batch settling.


Vol. 1 LIQUID–LIQUID EXTRACTION 825

Fig. 41. Dimensioning a continuous settler (79). A) Evaluation of settling experiments;


B) Fitting of measured and calculated curve.

A glass cylinder as shown in Figure 40 A, filled with about 1 L of the original


liquid – liquid system, is vigorously shaken for 1/2 to 1 min. After placing it in an
upright position, the heights of sedimentation and coalescence fronts are mea-
sured in dependence of time as shown in Figure 40 B for drops being lighter
than the continuous phase. The drops are banked up in a dense layer if sedimen-
tation is faster than the drop coalescence. The curves of sedimentation and coa-
lescence meet each other at time te when all drops are coalesced. For a sufficient
826 LIQUID–LIQUID EXTRACTION Vol. 1

reproducibility of measurements it should be made sure that the glass flask is


washed with deionized water, that the distribution of drop sizes is narrow and
the temperatures of the liquid – liquid system in the glass cylinder and the con-
tinuous settler are equal (78). Figure 37 C shows the height H of the wedge-
shaped zone between the curves of sedimentation and coalescence.
The concept of settler dimensioning is shown in Figure 41. All equations
mentioned in the following are given in (79). Settler dimensioning starts with
evaluation of settling experiments according to Figure 41 A. The linear curve of
sedimentation supplies the velocity of sedimentation; a model of drop swarm
sedimentation supplies the diameter of the starting drop swarm (66). The
thickness of the dense layer of drops follows from a balance of sedimenting
and coalescing drops in time steps. The next stage is calculation of drop defor-
mation in vertical steps (80). This serves to calculate coalescence times
between drops and between drops and their homophase, the former to deter-
mine the drop growth, the latter to calculate the drop homophase formed by
coalescence (80). The coalescence model contains a fitting parameter Fp to fit
the calculated and measured coalescence curve in Figure 41 B. This parameter
is the link between the settling experiment and the settler dimensioning; Fp
can be used for the settler calculation, given in Figure 41 B.
The settler needs an inlet length Lin which can be calculated by a model
equation in dependence on settler diameter, operation parameters, and the
mean drop size at the settler inlet, coming from the extractor outlet. For calcu-
lation of coalescence in the dense drop layer the same equations that were mod-
eled for the settling experiment can be applied using the fitting parameter Fp
derived from these experiments. The result is the geometry of the wedge of dis-
persion, eg, its height and length, which has to be taken into account for
dimensioning the settler.
This combined experimental and theoretical procedure yields good results
although the drop distributions in the settling experiment and the real settler
are not the same. The extensive equation system needs computer aided solving.
A computer program is offered by (81). A simplified method is given in (79).
4.2. Settlers with Coalescing Aids. Gravity settlers are often equip-
ped with coalescing aids to fit a settler to special conditions of the liquid system
or of the operational behavior. Industrially proven coalescing aids are inclined or
corrugated plates, fixed beds of small packings, fiber beds, porous sintered mate-
rials, and membranes. In some cases, when the drop sizes to be settled are smal-
ler than 1 mm or if surfactants render the settling more difficult, centrifuges (6,
Chap. 8) or electrical fields must be applied (82). The most important coalescing
aids are inclined packages of plain or corrugated plates and fiber bed filters.
Inclined Plates. Figure 42 shows one possible arrangement of inclined
plates in a horizontal settler. Other plate configurations are possible. In any
case, the feed has to be distributed by a sieve plate at the settler entry over
the cross-sectional area, filled with inclined plates. The plates shorten the sedi-
mentation distance of drops and provide a larger interfacial area for the coales-
cence. The drops coalescing at the plates form trickling films which flow off to the
horizontal, principle interface due to a hydrostatic pressure difference. A plate
material wettable by the drop phase is a prerequisite. Several scientific investi-
gations and experiments with liquid systems of a wide range of properties were
Vol. 1 LIQUID–LIQUID EXTRACTION 827

Fig. 42. Horizontal gravity settler with two inclined packages of plane, parallel plates.
The height of principal interface is best situated, when both phases have the same exit
velocity.

made during the 1970s and 1980s on the hydrodynamics of inclined films and the
coalescence of drops at the film surface, these are summarized in (83–86). The
main results are:

1) Thin laminar films should flow on the plates, because the coalescence
of drops is promoted by the shear stress of film flow near the plate;
the larger the volume flow of cocurrent continuous phase, the smaller
the film thickness and the better the coalescence.
2) The plate should be wettable by the drop phase (advancing contact angle
< 708, receding contact angle > 158).
3) The drop swarm has to be distributed equally over the width of plates.
4) Constructions that can thin the trickling film increase the settler
performance.
5) Short lengths of plates of about 400 mm are sufficient for coalescence of
drops riding on the film.
6) The width bp of a plate can be calculated by the empirical equation of the
Reynolds number of trickling film
0:19
Ref  80 KL;d

Vd  % g sin a
bp ¼ 0:03
ðRef k vd Þ3=2 Zd

with the liquid number of trickling film

s3 %d %d
KL;d ¼
g Z4d  %
where a is the plate inclination, Vd,1 = Vd/n the volume flow of dispersed phase on
a plate, and n the number of plates.
7) The optimum of plate inclination is within the range 108 < a < 158.
828 LIQUID–LIQUID EXTRACTION Vol. 1

8) The distance of adjacent plates should not be smaller than 30 to 40 mm in


order not to disturb film flow and drop coalescence.
9) The trickling film continuously carries surfactants to the outflow of a
plate and thus prevents their accumulation. Therefore surfactants hardly in-
fluence coalescence and settling performance up to medium flow rates of drop
phase.
10) Under certain conditions, the drop coalescence leaves very small droplets,
so-called secondary drops, due to partial coalescence, which generate a
droplet entrainment at the settler exit. Its occurrence is primarily a ques-
tion of liquids involved, ie, their viscosity ratio (86).

The coalescence time of secondary drops is significantly higher than that of


primary drops, due to interfacial effects. Settling of secondary drops requires
other coalescers, for instance membrane coalescers or centrifuges.
Fiber bed filters. are built from short fibers stuffed to rectangular mats
or to disks of 15 to 25 mm thickness, held in a suitable support, or from fiber
fleeces as shown in Figure 43.
They have a definite, large voidage of about 95 %, nevertheless they are
sensitive to solid dirt particles in the liquid. The droplets couple to the fibers
while flowing through the fiber bed, succeeding drops coalesce with them and
enlarge the size of adhering drops until the liquid flow pulls them out of the
bed due to increasing flow resistance. A fiber bed is able to enlarge the sizes of
incoming drops at least by one order of magnitude so that they can coalesce in a
subsequent settler with their homophase. Fiber bed filters are restricted to drop
concentrations < 10 vol % to avoid redispersion.
REBELEIN (87,88) and MAGIERA (89,90) defined heuristic rules for the design
of fiber beds on the basis of careful experiments with a broad variation of para-
meters of influence:

Fig. 43. Combined fiber bed coalescer and gravity settler.


Vol. 1 LIQUID–LIQUID EXTRACTION 829

1) Fiber materials as glass or stainless steel with high energetic surfaces


provide different interaction sites (eg, hydrogen bonds and ion – dipole
interaction) for the liquid and improve the fiber bed performance
compared to materials with low energetic surfaces as for instance
teflon. High energetic surfaces improve the wetting properties of the
fibers.
2) The drop enlargement in a fiber bed decreases with increasing flow rate,
ie, with superficial velocities > 1 cm/s.
3) Small fiber diameters of about 15 mm are favorable.
4) The bed depth required for coalescence increases with the fiber diameter.
5) The concentration of droplets in the incoming liquid hardly influences the
bed performance.
6) Best efficiency means a separation of about 95 % of incoming drops.
7) The larger the surface tension of the liquid system, the better the separa-
tion performance.
8) Smallest drop sizes that can be separated are about 10 mm.
9) Presence of surfactants and mud and a higher viscosity of the continuous
phase reduce the separation performance, the viscosity of the drop phase
has no influence within 50 and 5000 mPa  s.

All attempts to calculate the bed performance in advance via filter coeffi-
cients were fruitless (87–90). MAGIERA developed a mathematical model of physi-
cal mechanisms inside a fiber bed and was able to simulate the bed behavior
and drop enlargement by introduction of two experimental fitting parameters.

Fig. 44. Membrane coalescer (91).


830 LIQUID–LIQUID EXTRACTION Vol. 1

Therefore, a reliable dimensioning of fiber beds needs experiments with the


liquid dispersion which is to be separated.
Membrane Coalescence. A new and very promising method to separate
fine droplets down to 1 mm from emulsions, stabilized by surfactants, was discov-
ered and investigated by HOFFMANN (91) (Fig. 44).
Membrane coalescers are applicable in a broad range of volume concentra-
tions down to a lower concentration limit of the organic phase (0.1 vol %). A thin,
hydrophobic, stretched membrane, for instance polytetrafluoroethylene (PTFE),
held in place by a support, is permeated by two-phase flow of emulsion. The
emulsion leaves the membrane as a liquid foam with large oil domains which
will immediately coalesce at the principle interface in a succeeding settler. A
superficial velocity through the membrane is possible within the range of 0.1
to 19 cm/s, the pressure drop is comparable with that in fiber beds. The main
advantages are huge separation performances of about 98 %, in combination
with a large range in throughput; the ability to separate droplets one order of
magnitude smaller than in fiber beds; and the insensitivity against hydrophobic
surfactants. Hydrophilic surfactants are without influence, too, if their amount is
clearly lower than the critical micelle concentration.
Selection of Appropriate Settler. Most important and the first step in
selection of a settler is a careful analysis of separation problems (78,92). The flow
sheet of the complete process can give hints on origin and kind of the dispersion
to be separated. Sometimes the preceding steps could be altered in order to sim-
plify the phase separation problem. Measurements of properties of the liquid sys-
tem and settling experiments in a glass cylinder show, if and why a phase
separation can be problematic. Drop sizes smaller than 100 mm need suitable coa-
lescing aids which are supposed to be well wettable by the drop phase. When the
drop settling in a glass cylinder is fast, but the liquids appear cloudy, this indi-
cates partial coalescence, resulting in very fine drops, so that a fiber bed or a
membrane coalescer will be necessary.
The presence of solid particles and mud requires filters before the settler.
The wetting behavior of coalescing aids can significantly change the operation
of the process. This should be accounted for in the procces. The phase ratio
can be changed by recycling of one phase to the settler. By this, the type of dis-
persion can be inverted. In case the organic phase has a higher viscosity than the
water phase, an o/w dispersion separates much better than a w/o dispersion, and
vice versa.

5. Liquid – Liquid Extraction Processes

5.1. General. The quality of the raffinate in liquid – liquid extraction can
be controlled easily by changing the ratio of solvent to feed and the extraction
temperature. The quality of the extract is governed only by the conditions of
feed location and no direct control is possible.
This is especially true because most extractions are carried out in a closed
system to minimize solvent requirements. In such a system, a maximum extract
purity is obtainable and some components that should end up in the raffinate
leave with the extract.
Vol. 1 LIQUID–LIQUID EXTRACTION 831

Fig. 45. Use of cooling stage A) Process flow scheme: a) Extractor; b) Settler B)
Triangular diagram: a) Binodal curve at higher temperature; b) Binodal curve at lower
temperature.

Additional steps are, therefore, required to enhance extract purity and thus
raffinate yield. If the extract is the product desired, improving extract purity is
the most important — in fact, the only — objective. Different solutions to this pro-
blem are applied in commercial operation.
Change in Temperature. Extract purity can be increased by cooling the
extract from the main system and adding one more stage (see Fig. 45 A). In the
triangular diagram, this means that the binodal curve is changed from a to b
(Fig. 45 B).
Because the composition of the first extract is now inside binodal curve b, it
separates in a settler b in a purer final extract phase and a small amount of inter-
mediate raffinate phase. The latter is returned to the main extraction system.
The selectivity of separation in the additional stage is not very great
because the loading of the first extract phase is rather high and separation
takes place near the plait point. This system can be used when the main system
is operated at high temperature for other reasons or when an increase in extract
purity and the resultant increase in raffinate yield are desirable.
Addition of Water. The addition of water at or below the feed point gen-
erally changes the system from a closed to an open one (see Section 2.1). In this
way, extract purity can be increased to the required level. Water is soluble, at
least to some extent, in most polar solvents and therefore leaves the system
with the final extract phase. In this way, no unwanted influence is exerted on
the main extraction system.
However, because water must be separated from the solvent by distillation
before its reintroduction in the main extractor, the heat requirements are high.
Addition of water is, therefore, not regarded as a generally attractive solution.
Also, water can only be added when its boiling point is much lower than those
of the solvent and the feedstock.
832 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 46. Extractor for separation using a countersolvent. (a) Extractor.

Use of a Countersolvent. As shown in Figure 46, a second solvent can


be fed into one end of the extraction system, on the opposite side of the solvent
entry point. The feed in this case is introduced somewhere in the middle of the
system.
The countersolvent should be only slightly soluble in the solvent and,
therefore is generally similar in chemical structure to the key component of the
raffinate phase. The boiling point of the countersolvent should be such that it
can easily be separated by distillation from the solvent as well as from the
raffinate and the extract (ie, either a much lower or a much higher boiling
point). Because the countersolvent does not contain any raffinate components
when it enters, it can wash out the raffinate components in the solvent at the
feed point.
Additionally, some extract components are washed back by the countersol-
vent, thus increasing the load in the top section. In this way, a very pure final
extract can be produced, depending on the ratio of countersolvent to feed used.
Vol. 1 LIQUID–LIQUID EXTRACTION 833

Fig. 47. Use of countersolvent with intermediate recovery. a) First extractor; b) Distilla-
tion column; c) Second extractor.

A disadvantage of this system is that the countersolvent enters the top sec-
tion. Consequently, the phase ratio of B-rich to A-rich phase is reduced, resulting
in a less pure raffinate (this also means a lower yield of extract). To restore raf-
finate purity, more B-rich phase must be applied. Such a system is thus costly
and sensitive to changes in feed composition, solvent-to-feed ratio, and ratio of
countersolvent to feed.
Some improvement can be obtained by carrying out the extraction in two
parts (Fig. 47).
In the first extractor (a), the feed is extracted with the primary solvent. The
intermediate extract phase is fed to a second extractor (c) and brought in contact
with the countersolvent. The mixture of countersolvent and intermediate raffi-
nate is fed to a column (b) in which the intermediate raffinate and the counter-
solvent are separated by distillation. The pure intermediate raffinate is then fed
to the first extractor, while the countersolvent is charged to the second extractor.
In this way, some of the disadvantages can be overcome.
5.2. Combined Processes of Extraction and Distillation. Azeotropic
Distillation. Usually it is necessary to combine extraction and distillation pro-
cesses for recovering the solvent from the extract phase and also from the raffinate
phase, if the solvent is soluble in the raffinate.
Figure 48 shows as an example the recovery of acetic acid from an aqueous
solution with methyl tert-butyl ether (MTBE) as solvent. The system of water
and acetic acid is a highly nonideal azeotropic mixture that could be separated
by simple distillation. However, for dilute feed compositions solvent extraction
834 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 48. Triangular diagram (A) and flow sheet (B) of a hybrid extraction – distillation
process for separation of water and acetic acid with MTBE as solvent (93) E = extraction;
C-1, C-2 = distillation; S = settling; H = condenser.

is more advantageous because of the very high energy demand of a water-rich


feed distillation. The triangular diagram of Figure 48 shows that the binary sys-
tem of solvent MTBE and water has a heterogeneous azeotrope. As a conse-
quence, extraction has to be combined with azeotropic distillation as shown in
the flow sheet of Figure 48 B. By extraction in the extractor E extraction feed F
is changed to raffinate R and the solvent L1 to extract phase F1, which on the
other hand is the feed of distillation column C-1. In C-1 the extract is split into
the bottoms acetic acid B1 and an acetic acid-free distillate D1 connected by
the main balance line of distillation in the triangle diagram. The acetic acid-
Vol. 1 LIQUID–LIQUID EXTRACTION 835

free raffinate R is fed into column C-2 where it is separated into the bottoms,
pure water B2 and water – MTBE azeotrope D2. Both distillates D1 and D2 lie
in the immiscible region between MTBE and water. This makes simple decanta-
tion (S) of the condensed phases possible. The solvent-rich fraction L1 is not
further purified since MTBE as solvent needs not to be pure. So L1 can be
recycled directly to the extractor E, a part of it is used as reflux for C-1. In column
C-2 a mixture of water and MTBE has been separated into pure water and
MTBE – water azeotrope. Therefore, L2, the water-rich phase from the settler,
can also be fed into column C-2.
The economy of the hybrid process depends on a careful optimization of
operational and constructional parameters, besides the careful selection of the
solvent. Recently, mixed integer nonlinear programming (MINLP) has been
successfully introduced in optimization of such processes (93,94). MINLP can
be used as a process synthesis tool. It is necessary to generate a so-called
superstructure of the hybrid process which contains a variety of alternative
flow sheets. Then MINLP is able to choose the best solution for given para-
meters and also from competitive unit operations like extractive distillation,
processes with more than one solvent or processes with reextraction for solvent
recovery.
Extractive Distillation. Another possibility to separate azeotropic
mixtures is extractive distillation. A solvent with a much higher boiling point
than the feed is fed at the top of a distillation column to entrain one component
of the distillation feed that enters the column at the bottom (see Distillation).
The combined system of liquid – liquid extraction and extractive distillation is
applied in nearly all processes used to manufacture very pure light aromatics.
N-Formylmorpholine is a suitable solvent to separate the light aromatics
benzene and toluene used as feedstock for the petrochemical industry

Fig. 49. Morpholine solvent process a) Extractor; b) Washing; c) Extractive distillation


column; d) Distillation column; e) Settler.
836 LIQUID–LIQUID EXTRACTION Vol. 1

(95–100). Due to the selectivity of N-formylmorpholine (NFM), extremely high


yields of very pure products are obtained.
A simplified process flow scheme for N-formylmorpholine extraction is given
in Figure 49. Feed enters the extractor at an appropriate point in the bottom half
of the column. The exact location depends on the composition of the feed and can
be optimized for each feedstock. Solvent enters at the top of the extractor at the
selected temperature. Because NFM has a high solvent power, this must often be
reduced by addition of water. In general, 4 – 10 % water is sufficient to ensure a
low solubility of saturated hydrocarbons while still maintaining a high solubility
of aromatics. The raffinate phase leaving the top of the extractor is saturated
with solvent. Washing with water has proved to be a simple and effective
means of recovering dissolved solvent. The water – solvent mixture from this
washing operation is charged to the bottom of the main extractor to increase
the water content of the solvent phase as discussed above.
The solvent – aromatics mixture from the bottom of the extractor still con-
tains some alkanes. This stream is routed to an extractive distillation column. In
the lower part of this column, saturated hydrocarbons are stripped off, which
results in high-purity aromatics at the bottom. Because the volatility of alkanes
in the presence of NFM is much higher than that of aromatics, the need for sharp
separation is not very high. At the top of the extractive distillation column, some
solvent is added. This solvent stream serves to wash back any aromatics evapo-
rated at the feed point. The resulting top product, therefore, has a very low aro-
matic content and is very effective as a reflux in the bottom part of the extractor,
washing out the heavier alkanes.
The bottom product of the extractive distillation column, consisting of sol-
vent and very pure aromatics, is fed to a distillation column where the aromatics
are evaporated along with water. The top product of this column is condensed
and fed to a settler. The extracted aromatics are available as product or can be
routed to a distillation train to separate the individual compounds. Some of the
aromatics are returned to the top of the distillation column as reflux. In this way,
any ascending solvent vapor is condensed and the top products are solvent-free.
The water layer from the settler is used for raffinate washing. Sometimes, part
of the water is used as reflux instead of extract. In this combination, ultimate use
is made of the solvent characteristics in both the extraction and the extractive
distillation steps.
5.3. Reactive Extraction. Introduction. If the solvent reacts with the
extracted substance during the extraction, the whole process is called reactive
extraction. This reaction is normally used to alter the properties of inorganic
cations that are to be extracted from an aqueous solution to make them soluble
in the nonpolar organic solvent. The reaction mechanism can best be compared
with a cation exchange process. The solvent acts like a liquid ion exchanging the
ion to be extracted for protons.
The use of this process for extraction of anions is far less known but is of the
same importance. The mechanism of this type of reaction proceeds most of the
time through ion pair formation and can best be compared with an anion
exchange process.
A third possible mechanism is the solvating and/or chelating reaction. In a
reaction of this type, one or more molecules of water in the hydrate layer of the
Vol. 1 LIQUID–LIQUID EXTRACTION 837

Table 1. Commercial Chelating Extraction Reagents


Name Producer Chemical composition
LIX 63 Henkel 5,8-diethyl-7-hydroxy-6-dodecanone oxime
LIX 64 Henkel 2-hydroxy-5-dodecylbenzophenone oxime
LIX 64N Henkel LIX 65N with added LIX 63 (promoter)
LIX 65N Henkel 2-hydroxy-5-nonylbenzophenone oxime
LIX 70 Henkel chlorinated LIX 65 N with added LIX 63 (promoter)
LIX 71 Henkel mixture of LIX 64 N and LIX 70
LIX 73 Henkel mixture of LIX 64 N and LIX 70
LIX 34 Henkel 8-(p-dodecylphenyl sulfonamido)quinoline
LIX 54 Henkel acetyl-p-dodecyl acetophenone
LIX 622 Henkel
LIX 6022 Henkel
SME 529 Shell 2-hydroxy-5-nonylacetophenone oxime
SME 530 Shell SME 529 with kinetic modifier
P 17 Acorga phenyl benzyl ketone oxime
P 50 Acorga substituted salicyl aldoxime
P 5100 Acorga one part P 50 with equal part nonylphenol
P 5200 Acorga one part P 50 with two parts nonylphenol
P 5300 Acorga one part P 50 with three parts nonylphenol
P 5050 Acorga two parts P 50 with one part tridecanol
Kelex 100 Sherex substituted 8-hydroxyquinoline with 10 % isodecanol
Kelex 120 Sherex similar to Kelex 100 but with 20 % isodecanol

ions are replaced by the active solvent. The new complex will be insoluble in
water but soluble in the solvent.
Generally, the formation of a complex is a slow reaction. Therefore, in many
cases, it is not possible to use columns for this type of extraction. In many appli-
cations mixer – settler are used to make the extraction possible.
A reactive extraction will need a re-extraction to isolate the extracted sub-
stance from the solvent. The active solvents used in reactive extractions are
water insoluble compounds. They are used undiluted or in solution in an inert
solvent.
Extraction Mechanism of Different Types of Solvent. Classical
Complexing and/or Chelating Agents. The active solvents are usually
large, polyfunctional molecules. Most of them are designed to extract specific
cations. Dioximes, hydroxyoximes, quinolines, b-diketones, etc., are rather
expensive but widely used solvents to extract metal ions such as copper and
nickel from the aqueous phase.
Metal complexes are formed in the water phase in a substance-specific pH
range and extracted into an inert solvent. After extraction the metal ion is liber-
ated from the complex by changing the pH, and re-extracted into the water
phase. The complexing/chelating agents are designed such that they are chemi-
cally stable, nontoxic, and insoluble in water to minimize losses. The liquid ion
exchange (LIX) (5) agents used in the production of copper are typical for this
category.
Henkel, Hoechst, Shell, Sherex, and Acorga are the largest producers of
complexing/chelating extractants (see Table 1).
838 LIQUID–LIQUID EXTRACTION Vol. 1

Phosphoric, Sulfonic, and Carboxylic Acids. This class of products


acts typically as cation exchangers in solution. The products have one or more
protons that can be exchanged for metal ions.

Naphthenic acids Versatic acids

Sulfonic acids Phosphonic acids

Thiophosphoric acids Phosphoric acids

The metal salts of these acids are soluble in most nonpolar solvents and,
depending on the pH of the extraction and the pKa of the acid, the cation will
migrate to the organic phase or stay in the water phase. The dependence of
the distributions coefficients on pH is shown in Figure 50.
The different curves give the variation of the distribution coefficient D in
dependence of pH for different cations. These solvents are not specific for certain

Fig. 50. Plots of log D vs. pH for extraction of metals.


Vol. 1 LIQUID–LIQUID EXTRACTION 839

cations, but are very suitable to concentrate and recuperate trace elements from
effluent streams (109,110).
The major producers of this type of compounds are Bayer, Hoechst, Daiha-
chi, Shell, and Albright & Wilson.
Ethers and Thioethers

R2 O þ HX Ð R2 OHþ þ X

R2 S þ HX Ð R2 SHþ þ X

where HX stands for a strong acid, eg, H2SO4 or HCl, and X is the anionic
rest of the acid. Most metal ions that are capable of forming anionic complexes
are extractable with ethers and thioethers. Fe, Au, In, Ga, etc. can be extracted
with these solvents from an aqueous HCl solution (HCl concentration > 6 M)
and re-extracted from the organic phase with water. Most of these anionic com-
plexes form strong acids which protonate with the ether group to form oxonium
(thionium) salts. These salts, however, are soluble in ether or thioether. It is also
possible to extract Cl from water with (R2O)2  H2SO4, the sulfate ion being
replaced by the chloride. This type of solvent is normally used in undiluted
form (106).
Ketones. extract the same anionic complexes as ethers, but in most cases
at lower acid concentration. Complexes with weaker acids can also be extracted.
Methyl isobutyl ketone and cyclohexanone are industrially used in the separa-
tion of Nb – Ta as F complexes (112) (! Niobium and Niobium Compounds,
Chap. 4.3.1) and in the separation of Zr – Hf as CNS complexes (! Zirconium
and Zirconium Compounds, Chap. 1.4.2) (116). The solvents are used in undi-
luted form. A great disadvantage of this and of the former type of solvents is
that they are not completely insoluble in water. A distillation is needed to remove
the dissolved solvent from the water phase.
Trialkylphosphine oxides, Trialkylphosphinates, -phosphonates,
and -phosphates

R3 PO þ HX Ð R3 POHþ þ X
ROR2 PO þ HX Ð ROR2 POHþ þ X
ðROÞ2 RPO þ HX Ð ðROÞ2 RPOHþ þ X
ðROÞ3 PO þ HX Ð ðROÞ3 POHþ þ X 

With this type of extractants most metals from group 3 to group 8 can be
extracted, even if only weak complexes are formed with the solvent. It is for
example possible to extract TiOSO4 (114) from a sulfuric acid solution with
dibutyl butylphosphonate (DBBP) as the solvated complex 2 DBBP  TiOSO4 
H2SO4  5 H2O. These organophosphorus solvents are, due to their outstanding
chemical stability and good physical properties, most widely used in reactive
extraction (107,108).
Free acids can also be extracted, the basicity of the solvents decreases from
phosphine oxide to trialkylphosphate. The distribution coefficients are rather
small so that large volumes of solvent are needed in an extraction process
840 LIQUID–LIQUID EXTRACTION Vol. 1

(104,105,111). These extractants are used as such or diluted in an inert solvent


(petroleum fraction) with or without modifier. The distribution coefficients in
metal extraction are usually very high and depend on the acid concentration.
The products are rather expensive.
Bayer, Hoechst, Daihachi, Shell, and Albright & Wilson are the major producers.
Trialkylsulfonium (Selenium, Tellurium), Tetraalkylammonium,
and Tetraalkylphosphonium (Arsonium, Stibonium) Salts
 
½R4 Nþ A Solvent þ ½B Raffinate Ð ½R4 Nþ B Solvent þ ½A Raffinate

These onium compounds form ions both in acidic and basic medium. The
mechanism of extraction is practically always anion exchange (106). The
onium compounds are regenerated in the same way as anion exchanger resins
by re-extraction with a base. These solvents are used in the production of Cr,
V, Re, Tc, Mo, and W as a solution in a petroleum fraction.
Producers are Henkel, Sherex, Hoechst, and BASF.
Amine and Phosphine Salts

R3 N þ HX ! R3 NHþ þ X
R3 P þ HX ! R3 PHþ þ X

Primary, secondary, and tertiary amines must be protonated, thus con-


verted to the ionic form before they can be used as an ion exchanger. Thus
acids can be extracted with free amines, metals have to be transformed to the
corresponding acids: vanadium can be extracted as HVO3, chromium as
H2Cr2O7, and manganese as HMnO4. Most used are the tertiary amines because
they are chemically inert to most reagents.
In the system trilaurylamine – citric acid a reversible extraction can be
set up, due to the difference in distribution coefficient at different temperatures.
At low temperature (20 8C), the distribution coefficient solvent/raffinate is > 10.
At high temperature (130 8C), the coefficient for the same system is < 1. Thus,
the acid can be extracted from water at low temperature and re-extracted with
water at high temperature. Much less accounted for is the fact that these amines
have very different distribution coefficients for different anions.

 
½R3 NHþ A Solvent þ ½B Raffinate Ð ½R3 NHþ B Solvent þ ½A Raffinate

For trioctylamine in an aromatic solvent the order of extractability is

I  Br > NO   2 3 2


3 > Cl  F > S2 O3 > PO4 > SO4

This series is dependent on the type of amine, the diluent, and the tempera-
ture and must be determined for any given system (5). This series can be
enlarged for organic compounds, capable of forming ion pairs with the amine.
Phenols and substituted phenols can be extracted with better distribution coeffi-
cients than chlorides. In this series thiophenols and some mercaptans are situ-
ated between bromide and nitrate and all sulfonic acids are better extracted than
Vol. 1 LIQUID–LIQUID EXTRACTION 841

phosphate. This makes the sulfate salts of amines a very good choice for new
extraction systems.
(R3N)2  H2SO4 (R = octyl) is capable of extracting Cl and/or NO3 (113) from
a mixture of both anions with SO2 4 . The value of the distribution coefficient of
the anions is surprisingly high.

DCl =DSO4  100


DNO3 =DSO4  400

A disadvantage of these solvents is the fact that per mole amine one mole of
salt is formed after regeneration. However, most of the time, the anions to be
removed are present in the solution only as trace impurities, so that this draw-
back is not significant. Another disadvantage of amines is the poor solubility of
ammonium salts in aliphatic hydrocarbons. To avoid third phase formation, the
use of aromatic solvents (smell) or aliphatics with modifiers is necessary. Amines
can also act as solvating agents in an acid medium. Iron, zinc, and uranium salts
can be extracted in this way (106).
Uses
1) A bleed from a (NH4)2SO4 crystallization unit in a caprolactam plant con-
tains 40 % (NH4)2SO4, 10 % NO 
3 , 50 ppm Cl and 3 – 4 % organic sulfonic
acids. The solution is extracted by trioctylammonium sulfate, the added
amount is stoichiometric to [NO 
3 ] + [Cl ] + [sulfonic acid]. In three stages
> 99 % of the nitrate, 90 % of the chloride, and 85 % of the sulfonic acids
are extracted by the trioctylamine in solution (40 %) in Shellsol AB (aro-
matic hydrocarbon). The raffinate is returned to the crystallization plant
for (NH4)2SO4 recovery. The re-extract contains 35 % of nitrate, 0.016 %
of chloride, the sulfonic acids, and only 0.1 % of sulfate and is treated
separately.
Compared with a conventional stripping, the use of chemicals in this
system is strongly reduced. 80 % less NaOH is consumed because the
base is only used to liberate the amine and not to neutralize the NH3
from (NH4)2SO4 (113).
2) Traces of phenol (20 – 200 ppm) and bisphenol (20 – 100 ppm) can be ex-
tracted from the wastewater of a polycarbonate plant to values < 1 ppm,
thus avoiding biological treatment of the water. The re-extract containing
sodium phenolate and bisphenolate (300 – 500 times more concentrated)
can be returned to the polycarbonate plant. The solvent used is 5 % triocty-
lamine in Shellsol AB (117).
3) From a plant for the production of vulcanization accelerators 2-mercapto-
benzothiazol (100 – 300 ppm) and its oxidation products are completely ex-
tracted in two stages and re-extracted as a 20 – 30 % solution. This
concentrated waste can be easily removed. The solvent is 20 % trioctyla-
mine in Shellsol AB (113).
4) Bromide ions can be extracted selectively from sea water (concentration
in the Dead Sea & 4 %) with amine hydrochlorides and concentrated in a
re-extraction step, thus avoiding the energetically unfavorable stripping
842 LIQUID–LIQUID EXTRACTION Vol. 1

of sea water. The extraction can be carried out with various types of amines
as HCl or H2SO4 salt (106).
5) Mg2+ can be removed from a saturated solution of (NH4)2SO4 by extraction
with the ammonium salt of diethylhexyl phosphonic acid (DEHPA) at pH
> 5. It can be recovered by lowering the pH to < 3 (115).
6) Traces of uranium are extracted from phosphate ores. The uranium content
in these ores is only 30 – 300 ppm. The reserves of those ores are so huge
that the quantity of uranium from phosphates many times exceeds that
of uranium found in other uranium ores.
7) Lanthanides and actinides have only recently become available in higher
quantity and high purity. This was enabled by the development of the ex-
traction technology.
8) Worldwide about 40 % of the copper is produced by reactive extraction. The
plants are an order of magnitude larger then any other type of solvent
extraction presently in operation. With an aqueous feed of 200 – 3000 m3/h
a whole new technology had to be developed. Copper can be extracted after
acid or ammoniacal leaching. The aqueous stripping phase is used directly
in the electrowinning of copper. To avoid the build-up of impurities in the
electrolysis of copper, Fe, Zn, and other metals are removed from this phase
by extraction with other solvents (5,102,111).
9) More than 90 % of the world production of cobalt is performed by reactive
extraction. Separation of Co and Ni can be achieved from alkaline, neutral,
or from acidic media. Most of the impurities (Sn, Bi, Fe, Pb, Cu, Cd, and Zn)
have to be removed from the solution by precipitation or extraction. From a
medium concentrated aqueous chloride solution (100 – 200 g/L), Co can be
extracted by amines together with only traces of Ni. After scrubbing with
4 – 5 M HCl, most of the Ni is removed from the solvent. The cobalt is
stripped by water.
From neutral or weakly acidic media (pH 4.8 – 5.5) Co can be extracted
by the Naþ or NHþ 4 salt of di-2-ethylhexylphosphoric acid (D2EHPA)
(to avoid acidification of the aqueous solution). Some Ni is co-extracted
and is removed by scrubbing with a Co solution thus replacing any ex-
tracted Ni by Co. Co is finally stripped by HCl, H2SO4, or HNO3 depending
on the use.
The use of monooctyl octylphosphonic acid (MOOP) to replace
D2EHPA (101) permits the same separation, but at lower pH and with en-
hanced distribution constants for Co.
In alkaline solution Co and Ni can be extracted by LIX agents from
ammoniacal solution. Prior to the extraction, cobalt is oxidized to the Co3þ
state. Nickel is stripped first with dilute acid, Co remaining in the solvent.
Cobalt is then reduced to the Co2þ state and stripped with diluted acid
(5,102,104,105).

All these examples are used in industry or in pilot plants. Elaborate pro-
cess and flow control are essential to use this technology in a economical and
meaningful way.
Vol. 1 LIQUID–LIQUID EXTRACTION 843

Fig. 51. Distribution coefficient in dependence on acid concentration.

Setting up an Extraction System. A solution R contains three compo-


nents: acid, metal A and metal B. Both metals and the acid are extractable
with a solvent S and metal A is the desired component that is to be obtained
in acid solution. The procedure to set up a new extraction system with these
data is typical for any reactive extraction system (114,118).
The distribution coefficients are to be measured at different acid concentra-
tions at a temperature suitable for the extraction. A typical measurement curve
is shown in Figure 51.
The diagram shows a poor, practically constant distribution coefficient D
for the acid and an increasing one for both metals A and B at increasing acid
concentration.
This means that the acid extraction is in fact a normal physical extraction
and that both metals form solvation complexes. At any acid concentration, all
three components are extracted together. The extraction coefficients E are
related in the same way as are the distribution coefficients,

E ¼ D  S=R:

To extract metal A completely and in a reasonable number of steps extraction


has to be carried out in zone 4 of Figure 51, the extraction zone. At the same time,
metal B and the acid are only partially extracted into the solvent. The number of
steps can be calculated with the Kremser formula. The Kremser formula calcu-
lates the efficiency Z in the extraction

Z ¼ ðEnþ1  EÞ=ðEnþ1  1Þ

over n stages. For this equation the number of stages can easily be calculated

n ¼ log f½1=ð1  ZÞ  ð1  1=EÞ þ 1=Eg=log E

To remove metal B from the solvent the solvent has to be washed with
water or acid in such a way that most of metal A stays in the solvent, while
metal B is washed out as completely as possible. This is done in zone 3 of Fig-
ure 51, the scrubbing zone. The amount of water to be used is given again by
844 LIQUID–LIQUID EXTRACTION Vol. 1

Fig. 52. General flow sheet of a reactive extraction system.

the distribution coefficients of both acid and metal B. If too much acid is removed,
metal A will also be washed out. Therefore it is more adequate to use an
acid solution. The concentration of the dilute acid should be such that the acid
solution can only take up the amount of acid to be removed from the solvent.
By scrubbing with pure water only a limited amount of scrubbing fluid can be
used because of the low distribution coefficient of the acid. By adding a calculated
amount of acid the amount of scrubbing fluid can be increased and more compo-
nent B can be removed without removing component A. The raffinate, leaving
the scrubbing zone contains too much of metal A to be discarded and is sent
back to one of the steps in the extraction. The number of steps can again be cal-
culated with the Kremser formula.
In zone 2 of Figure 51, the stripping zone, metal A is washed out of the sol-
vent with water or a dilute acid solution. The amount of water is chosen such as
to obtain the desired concentration of metal A in the re-extract. With the given
distribution coefficient, it is not possible to remove metal A completely form the
solvent. At any acid concentration, traces of metal A will remain in the solvent.
In zone 1 of Figure 51, the washing zone, the acid and the rest of metal A are
washed out of the solvent. After this step, the solvent is completely regenerated
and can be reused.
General Flow Sheet. Figure 52 shows a general flow sheet of a reactive
extraction system.
Solvent Make-Up. Due to the poor distribution coefficient of the acid seve-
ral extraction steps are needed to load the solvent with sufficient acid to make
extraction of the metals possible. In the given example the acid comes from
the initial solution. To avoid losses of metal A, the solvent must be completely
free of A. At the low acid concentration in the first steps of the solvent make-
up any metal A present in the solvent will be re-extracted in the raffinate.
Reactive Extraction. If sufficient acid is present in the solvent, both metals
will be extracted from the feed solution. To limit the number of steps in the
extraction, concentrated acid is added. The high concentration is necessary for
the metal extraction and in the solvent make-up.
Vol. 1 LIQUID–LIQUID EXTRACTION 845

Solvent Scrubbing. To make sure that most of metal B is washed out from
the solvent, the acid solution containing metal A and acid from the solvent wash-
ing step is used, so that we stay in the extraction zone for metal A. The second
component, with a lower distribution coefficient, is already in its stripping zone
and will be washed out of the solvent. To avoid losses, this stream is sent back to
one of the steps in the extraction.
Solvent Stripping. After scrubbing, the solvent contains only metal A
and acid. Using water or a dilute acid solution, the purified metal A can now
be re-extracted into a water phase.
Solvent Washing. The solvent must be completely regenerated before it
can be recycled to the solvent make-up. This is done by adding some supplemen-
tary washing steps to the system. The amount of water is chosen according to the
requirements in the scrubber and the number of steps is calculated as a function
of that amount.
Diluents and Modifiers. Diluents are used to reduce the solvent viscosity,
improve phase separation, and most of all, when properly chosen, to reduce the
loss of solvent. The solubility of tri-n-butyl phosphate (TBP, one of the most fre-
quently used extractants) in pure water is 420 mg/L at room temperature; a 50 %
dilution of TBP in n-hexane lowers this solubility to 175 mg/L. The choice of
diluent is very important. Higher amines are completely soluble in aliphatic dilu-
ents. Their salts, however, are only slightly soluble in the same solvents. There-
fore it is essential to test the solutions in all steps of the process. If third-phase
formation occurs in one of the steps, the diluent must be changed or a modifier
is to be added. These modifiers are normally long chain alcohols, monoesters of
polyhydric alcohols, or phosphate esters. These modifiers may considerably influ-
ence the kinetics of the extraction. Especially in the extraction of copper the
choice of modifier is very important. The different types of solvent are sold
mixed with a suitable modifier which greatly enhances the extraction perfor-
mance of the solvent.

6. Symbols

2a Pulsing amplitude, m
A key component of the phase from which the solute is ex-
tracted
B key component of the phase into which the solute is ex-
tracted
ci constants
C transferred component
di hole or packing diameter, m
dp particle diameter, m
d32 Sauter mean drop diameter, m
D column diameter, m
DR rotor diameter, m
D distribution coefficient
E = K ṁs/ṁF extraction factor
f pulsing frequency, s1
846 LIQUID–LIQUID EXTRACTION Vol. 1

g acceleration due to gravity, m/s2


Gi coefficients in empirical correlations of the distribution
coefficient
h compartment height, m
H height of column, m
HETS height equivalent of a theoretical stage, m
k = yc/xc distribution ratio of transferred component
(in terms of mass fraction)
K = YC/XC distribution ratio of transferred component
(in terms of loading)
Kp plait point
mj k mass of component j in phase or flow k
( j = A, B, C; k = F, R, S, E), kg
mk = mAk + mBk + mCk total mass of phase or flow k (k = F, R, S, E), kg
ṁ mass flow, kg/h
ns revolutions of stirrer, s1
n number of stages
NTS number of theoretical stages
s column cross section, m
v superficial velocity, m3 m2 h1 or m/s
v = ṁS/ṁF extractant ratio
vp individual drop velocity, m/s
vr slip velocity (Eq. 9), m/s
xC mass fraction of C in A-rich phase
XC = x/(1  x) loading of A-rich phase with component C, kg/kg
yC mass fraction of C in B-rich phase
YC = y/(1  y) loading of B-rich phase with C, kg/kg

Greek symbles

Z dynamic viscosity, Pa  s  103


density, kg/m3
s interfacial tension, N/m  103
j free area

Dimensionless Groups

Ar ¼ d3P g%c %=Z2c Archimedes number


FrR ¼ N 2 DR =g Froude number
Rep ¼ p dp =d Particle Reynolds number
KL ¼ %2c s3 =ð%gZ4c Þ Characteristic number of liquid

Subscripts

A key component of the phase from which the solute


is extracted
Vol. 1 LIQUID–LIQUID EXTRACTION 847

B key component of the phase into which the solute is


extracted
C transferred substance (solute)
c continuous phase
d dispersed phase
E extract
f flooding
F feed
i internals (characteristic size)
o organic
p drop (particle)
R raffinate
S solvent
1 pilot scale
2 large scale

REFERENCES

1. R. E. Treybal: Liquid Extraction, McGraw-Hill, New York – Toronto – London 1963.


2. S. Hartland: Counter Current Extraction, Pergamon Press, Oxford 1970.
3. J. Rydberg, C. Musikas, G. R. Choppin (eds.): Principles and Practises of Solvent
Extraction, M. Dekker, Inc., New York 1992.
4. C. Hanson: Recent Advances in Liquid – Liquid Extraction, Pergamon Press, Oxford
1971.
5. T. C. Lo, M. H. I. Baird, C. Hanson (eds.): Handbook of Solvent Extraction, Wiley-
Interscience, New York 1983.
6. J. D. Godfrey, M. J. Slater (eds.): Liquid – Liquid Extraction Equipment, J. Wiley
& Sons, New York 1994.
7. J. Wisniak, A. Tamir: Liquid – Liquid Equilibrium and Extraction. A Literature
source Book. Physical Sciences Data, 7, Part A 1980, Part B 1981, Physical Sciences
Data, 23, Supplement 1 1985, Elsevier, Amsterdam.
8. Proceedings ISEC ’88, Vernadsky Institute of Geochemistry and Analytical Chemis-
try of the USSR Academy of Sciences, 4 vols., NAUKA, Moscow, 1988.
9. Proceedings ISEC ’90, Chemical Society of Japan, Elsevier, New York 1992.
10. Proceedings ISEC 71, The Hague, 2 vols., Society of Chemical Industry, London 1971.
11. Proceedings ISEC 74, Lyon, 3 vols., Society of Chemical Industry, London 1974.
12. Proceedings ISEC 77, Toronto, 2 vols., CIMM Special Publication, no. 21, Can. Inst.
Min. Met., 1979.
13. Proceedings ISEC 80, Liège, 3 vols., Assocn. des Ingenieurs sortis de l’Université de
Liège, Liège 1980.
14. Proceedings ISEC 83, Denver AIChE 1983.
15. Proceedings ISEC 86, München, 3 vols., DECHEMA, Frankfurt 1986.
16. Proceedings ISEC ’93, Solvent Extraction in the Process Industries, 2 vols., Elsevier,
New York 1993.
17. Proceedings ISEC ’96, Value Adding Through Solvent Extraction, 2 vols., The
University of Melbourne, Parkville, Victoria, Australia 1996.
18. J. M. Sorensen, W. Arlt: ‘‘Liquid – Liquid Equilibrium Data Collection,’’ vol. V, part
1 – 3, DECHEMA Chemistry Data Ser., Frankfurt 1979.
19. Landolt-Börnstein: II 2 c, Springer Verlag, Berlin 1980.
848 LIQUID–LIQUID EXTRACTION Vol. 1

20. O. Madelung: Landolt-Börnstein directory disk 1.1, Springer-Verlag, Berlin 1994.


21. A. Seidel, F. W. Linke: ‘‘Solubilities of Inorganic and Organic Compounds,’’ in A. W.
Francis (ed.): Ternary Systems Separating into Two Liquid Layers, van Norstrand,
London 1952.
22. A. W. Francis: Liquid – Liquid Equilibria, Wiley, New York 1963.
23. H. Stephen, T. Stephen: Solubilities of Inorganic and Organic Compounds, 4 vols.,
Pergamon Press, Oxford 1964.
24. F. Ruiz, D. Prats, A. F. Marcilla: ‘‘Liquid – Liquid Extraction: a Graphical Method
for Equilibrium Stage Calculations for Quaternary Systems,’’ Fluid Phase Equilib.
15 (1984) no. 3, 257 – 265.
25. W. Rückl, R. Marr: ‘‘Multicomponent Liquid – Liquid Extraction,’’ Ger. Chem. Eng.
(Engl. Transl.) 8 (1985) no. 1, 27 – 31.
26. M. J. Hampe: ‘‘Selection of Solvents in Liquid – Liquid Extraction According to
Physico-Chemical Aspects,’’ Ger. Chem. Eng. (Engl. Transl.) 9 (1986) no. 4, 251 – 263.
27. Ullmann, 4th., vol. 2, 550 – 551.
28. E. Bender, U. Block, ‘‘Thermodynamische Berechnung der Flüssig-Flüssig-Extrak-
tion,’’ Verfahrenstechnik (Mainz) 9 (1975) no. 3, 106 – 111.
29. K. Hlavaty: ‘‘Correlation of the Binodal Curve in a Ternary Liquid Mixture with One
Pair of Immiscible Liquids,’’ Collect. Czch. Chem. Commun. 37 (1972) 4005 – 4007.
30. E. Hecker: ‘‘Auswahl von Lösungsmittelsystemen zur multiplikativen Verteilung,’’
Chimia 8 (1954) 229 – 236.
31. C. M. Hanson: ‘‘The Universality of Solubility Parameter,’’ Ind. Eng. Chem. Prod.
Res. Dev. 8 (1969) no. 1, 2 – 11.
32. C. Reichardt, K. Dimroth: ‘‘Lösungsmittel und empirische Parameter zur Charakter-
isierung ihrer Polarität,’’ Fortschr. Chem. Forsch. 11 (1968) no. 1, 1 – 73.
33. R. Anderson, R. Cambio, J. M. Prausnitz: ‘‘Physical and Chemical Forces in Solvent
Selectivity for Hydrocarbons,’’ AIChE J. 8 (1962) 66 – 69.
34. C. H. Deal, E. L. Derr: ‘‘Selectivity and Solvency in Aromatics Recovery,’’ Ind. Eng.
Chem. Prod. Res. Dev. 3 (1964) 394 – 399.
35. R. A. Grieger, C. A. Eckert: ‘‘Solvent Mixtures for Separation Processes,’’ Ind. Eng.
Chem. Prod. Res. Dev. 6 (1967) 250 – 255.
36. E. Müller, G. Höhfeld: ‘‘Aromatenextraktion mit Lösungsmittelgemischen,’’ Erdöl
Kohle Erdgas Petrochem. 24 (1971) 573 – 578.
37. D. Tassios: ‘‘GLC Screens Extraction Solvents,’’ Hydrocarbon Process 49 (1970) no. 7,
114 – 118.
38. I. Mellan: Industrial Solvents Handbook, Noyes Data Corp., New Jersey 1985.
39. H. Gnamm, W. Sommer: Lösungsmittel und Weichmachungsmittel, Wissenschaf-
tliche Verlags-anstalt, Stuttgart 1980.
40. G. Dave, M. Lidman: ‘‘Biological and Toxicological Effects of Solvent Extraction Che-
micals,’’ Hydrometallurgy 3 (1978) no. 3, 201 – 216.
41. R. Berger, W. Leuckel, D. Wolf, Chem. Ing. Tech. 50 (1978) no. 7, 544 – 545.
42. R. Marr, G. Husung, F. Moser, Chem. Ing. Tech. 47 (1975) no. 5, 203.
43. G. A. Rowden, M. Dilley, C. F. Bonney, G. A. Gillet, Lecture at Hydrometallurgy 81,
Manchester 1981.
44. T. Misek, R. Berger, J. Schröter: Standard Test Systems for Liquid Extraction, The
Institution of Chemical Engineers 1985.
45. D. Niebuhr, Ph. D. Thesis, Technische Universität Clausthal (Fed. Rep. of Germany)
1982.
46. E. Aufderheide, Ph. D. Thesis, Technische Universität Clausthal (Fed. Rep. of
Germany) 1985.
47. W. Pietzsch, Ph. D. Thesis, Technische Universität München (Fed. Rep. Germany)
1984.
Vol. 1 LIQUID–LIQUID EXTRACTION 849

48. H. Haverland, Ph. D. Thesis, Technische Universität Clausthal (Fed. Rep. of


Germany) 1988.
49. K. H. Reissinger, Ph. D. Thesis, Technische Universität Graz (Austria) 1985.
50. R. Berger, K. Walter, Chem. Eng. Sci. 40 (1985) no. 12, 2175 – 2184.
51. H. Zimmermann, Ph. D. Thesis, Technische Universität München (Fed. Rep. of
Germany) 1981.
52. Th. Pilhofer, Chem. Ing. Tech. 48 (1976) no. 3, 237.
53. E. Bender, R. Berger, W. Leuckel, D. Wolf, Chem. Ing. Tech. 51 (1979) no. 3,
192 – 199.
54. R. Berger, Lecture GVC annual meeting, Düsseldorf 1981.
55. J. Komasawa, J. Ingham, Chem. Eng. Sci. 33 (1978) no. 3, 341 – 347, no. 4, 479 –
485.
56. Th. Pilhofer, Chem. Ing. Tech. 51 (1979) no. 3, 231.
57. K. Hirschmann, E. Blass, Ger. Chem. Eng. 7 (1984) 280 – 287.
58. G. Wagner, Ph. D. Thesis, Technische Universität München, 1995.
59. B. Hoting, A. Vogelpohl, Chem. Ing. Tech. 68 (1981) 105 – 112.
60. H. W. Brandt, K. H. Reissinger, J. Schröter, Verfahrenstechnik 9 (1975) no. 8,
383 – 387.
61. G. Dichtl, T. Pilhofer, Verf. Tech. 15 (1981) 615 – 617.
62. H. Rauscher, Ph. D. Thesis, Technische Universität München, 1992.
63. M. Lorenz, Ph. D. Thesis, Technische Universität Clausthal, 1990.
64. G. Goldmann, Ph. D. Thesis, Technische Universität München, 1986.
65. H. Hufnagl, Ph. D. Thesis, Technische Universität München, 1992.
66. Th. Pilhofer, D. Mewes: Siebboden-Extraktionskolonnen, Verlag Chemie, Weinheim –
New York 1979.
67. Th. Pilhofer, Ger. Chem. Eng. (Engl. Transl.) 2 (1979) 69 – 76.
68. G. Zamponi, Ph. D. Thesis, Technische Universität München, 1996.
69. K. Hirschmann, Ph. D. Thesis, Technische Universität München, 1984.
70. K. H. Reissinger R. Marr, Chem. Ing. Tech. 58 (1986), 7, 540 – 547.
71. J. D. Thornton, Brit. Chem. Eng. 32 (1958) 247 – 251.
72. A. E. Karr, T. C. Lo, Proc. ISEC 71.
73. R. Bauer, Ph. D. Thesis, ETH Zürich (Switzerland) 1976.
74. Th. Pilhofer, J. Schröter, Chem. Ing. Tech. 56 (1984) 883 – 890.
75. W. Bäcker, J.-P. Schäfer, J. Schröter, Chem. Ing. Tech. 63 (1991) 1008 – 1011.
76. E. Blass, W. Goettert, M. J. Hampe: ‘‘Selection of Extractors and Solvents,’’ in
(6, Chap. 18].
77. I. Wagner, Ph. D. Thesis, Technische Universität München, 1999.
78. R. Berger, Chem. Ing. Tech. 58 (1986) no. 6, 499 – 456.
79. M. Henschke, Ph. D. Thesis, RWTH Aachen, 1995.
80. S. Hartland, S.A.K. Jeelani in (6 , Chap. 13).
81. Computer Aided Settler Design ‘‘CASEDY’’ Rauschert Verfahrenstechnik GmbH,
Postfach 20, D-96347 Steinwiesen.
82. A. Kriechbaumer, R. Marr, Chem. Ing. Tech. 55 (1983) 700 – 707.
83. W. Rommel, W. Meon, E. Blass, Separat. Sci. Technol. 27 (1991) 129 – 159.
84. W. Rommel, E. Blass, W. Meon, Chem. Engng. Sci. 47 (1992) 555 – 564.
85. W. Meon, W. Rommel, E. Blass, Chem. Engng. Sci. 48 (1993) 159 – 168.
86. W. Rommel, E. Blass, W. Meon, Chem. Engng. Sci. 48 (1993) 1735 – 1743.
87. F. Rebelein, Ph. D. Thesis, Technische Universität München, 1989.
88. F. Rebelein, E. Blass, Filtration & Separation 27 (1990) 360 – 363.
89. R. Magiera, Ph. D. Thesis, Technische Universität München, 1995.
90. R. Magiera, E. Blass, Filtration & Separation 34 (1997) 369 – 376.
91. S. Hoffmann, Ph. D. Thesis, Technische Universität München, 1999.
850 LIQUID–LIQUID EXTRACTION Vol. 1

92. Abwassertechnischer Verein e.V. (ed.): Industrieabwässer, Verlag Ernst & Sohn,
Berlin 1999.
93. S. Glanz, Ph. D. Thesis, Technische Universität München, 1998.
94. S. Glanz, J. Stichlmair in R. Darton (ed.): ‘‘Distillation and Absorption ’97,’’ Ichemt,
Symp. Ser. No. 142, 1997.
95. F. Trefny, Erdöl Kohle Erdgas Petrochem. 23 (1970) 337 – 340.
96. E. Cinelli et al., Hydrocarbon Process. 51 (1972) no. 4, 337 – 340.
97. Hydrocarbon Process. 51 (1972) no. 4, 141 – 144.
98. G. Precesser et al., Oil Gas J. July 16 (1973) 114 – 118.
99. Krupp Koppers, Hydrocarbon Process. 52 (1973) no. 4, 139 – 141.
100. Hydrocarbon Process. (1982) 182.
101. G. M. Ritcey, A. W. Ashbrook: Solvent Extraction, Part 1, Elsevier, Amsterdam 1984.
102. G. M. Ritcey, A. W. Ashbrook: Solvent Extraction, Part 2, Elsevier, Amsterdam 1979.
103. Solvent Extraction in Metallurgical Processes, Internat. Symp. Antwerpen 1972.
104. W. W. Schulz, J. D. Navratil: Science and Technology of Tributyl Phosphate, vol. I,
CRC Press, Boca Raton, FL 1984.
105. W. W. Schulz, J. D. Navratil, A. S. Kertes: Science and Technology of Tributyl Phos-
phate, vol. VI, CRC Press, Boca Raton, FL 1991.
106. A. K. De, S. M. Khopar, R. A. Chalmers: Solvent Extraction of Metals, Van Nostrand
Reinhold, London 1970.
107. Bayer AG: Geschäftsbereich Anorganische Chemikalien, Baysolvex – TBP für die
Flüssig-Flüssig Extraktion 1993.
108. Bayer AG: Geschäftsbereich Anorganische Chemikalien, Baysolvex – Dialkan-phos-
phate für die Flüssig-Flüssig Extraktion 1993.
109. Bayer AG: Geschäftsbereich Anorganische Chemikalien, Baysolvex – D2EHPA für
die Flüssig-Flüssig Extraktion 1993.
110. Bayer AG: Geschäftsbereich Anorganische Chemikallien, Baysolvex – D2EHTPA für
die Flüssig-Flüssig Extraktion 1993
111. Bayer AG: Geschäftsbereich Anorganische Chemikalien, Baysolvent for your
Solvent-extraction 1993.
112. F. Fairbrother: The Chemistry of Niobium and Tantalum, Elsevier Amstrad, London –
New York 1967, pp. 8 – 10.
113. Bayer Antwerpen N.V., EP 0 396 790 A1, 1989.
114. Bayer Antwerpen N.V., DE 3 643 711 A1, 1986.
115. Bayer Antwerpen N.V., EP 638 515-A2, 1993.
116. W. D. Jamrack: Rare Metal Extraction by Chemical Engineering Technology, Perga-
mon Press, Oxford 1963.
117. M. L. Wang, B. L. Liu: ‘‘Extraction Equilibrium of Phenol by Sulphuric Acid Salts of
Triisooctylamine,’’ Chem. Eng. Comm. 156 (1996) 131 – 146.
118. Bayer Antwerpen N.V., DE 3 524 053 A1, 1985.

ECKART MÜLLER
Frankfurt a. Main Germany
ROSEMARIE BERGER
Ludwigshafen Germany
ECKHART BLASS
Technische Universität München
München Germany
DOMIEN SLUYTS
Bayer AG Antwerpen Belgium
Vol. 1 PIPELINES 851

PIPELINES
Pipelines or pipe lines are continuous large-diameter piping systems, usually
buried underground where feasible, through which gases, liquids, or solids sus-
pended in fluids are transported over considerable distances. They are used to
move water, wastes, minerals, chemicals, and industrial gases, but primarily
crude oil, petroleum products, and natural gas. In the oil and gas business, a
pipeline system consists of a trunkline, ie, the large-diameter, high pressure,
long-distance portion of the piping system through which crude oil is shipped
to refineries, or natural gas and oil products, respectively, are transported to dis-
tribution points, and smaller low pressure gathering lines that transport oil or
gas from wells to the trunkline. Smaller lines used by natural gas distributors
are not considered part of a gas pipeline system (see NATURAL GAS).
Pipeline transport involves the application of force to the material being
moved, either through the use of pumps to transport liquids, compressors to
move gases, or flowing water to move solids. In some applications, vacuum
may create the pressure differential.

1. Pipeline Transport of Gases

Essentially all substances that are gases at standard conditions of temperature


and pressure are transported commercially by pipeline, this includes ammonia,
carbon dioxide, carbon monoxide, chlorine, ethane, ethylene, helium, hydrogen,
methane (natural gas), nitrogen, oxygen, and others. Gases with moderate boil-
ing points can be pipelined in either gaseous or liquid form; liquefied petroleum
gases (LPG), carbon dioxide, ammonia, and chlorine are usually shipped as
liquids because of the smaller pipeline volume for liquids.
The largest pipeline transport of gas, by far, is the movement of methane
(natural gas). Natural gas can be liquefied, but it is not pipelined in liquid form
because of cost and safety considerations. For overseas transport, it is shipped
as liquefied natural gas (LNG) in insulated tankers, unloaded at special unloading
facilities, vaporized, and then transported over land in pipelines as a gas.
1.1. Gas Pipeline Industry. In early gas industry days, the U.S. Con-
gress mandated that any pipeline must have an ensured 20-year supply of gas
before it could receive a permit to operate. The pipeline companies signed
long-term purchase contracts with producers to satisfy this mandate and trans-
ported the gas to the cities where it was sold and title taken by the gas distribu-
tion companies. In the mid-1980s, however, the Federal Energy Regulatory
Commission (FERC) began a restructuring of the industry that was to remove
the merchant role from the gas pipelines and ultimately establish them as com-
mon carriers, comparable to railroads and liquid pipelines. Distribution compa-
nies and large gas users, such as corporations, may buy their own gas directly
from producers, aggregators, or marketers and contract with pipelines to trans-
port it for a fee.
This industry restructuring, coupled with removal of price controls and a
large deliverability–demand inbalance, the so-called gas bubble, led to low gas

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.1609160523080112.a01
852 PIPELINES Vol. 1

prices during the late 1980s and early 1990s and resulted in the formation of a
spot market for natural gas, a gas futures market, and a need for more sophisti-
cated gas technology. The common carrier status of natural gas pipelines allows
many small producers to enter the pipelines, if pipeline capacity is available, and
make deliveries on a short-term (spot) basis. Flow computers at entry and exit
points along the pipeline provide an instantaneous and accurate account of the
gas that enters or leaves the pipeline. The flow computers tie into a large com-
puter at pipeline headquarters to record an accurate account of gas from produ-
cers entering the pipeline as well as gas exiting to a gas user or distributor. In
addition, FERC Order 636 mandated the establishment of an electronic bulletin
board (EBB) for the gas industry, by which anyone with a personal computer can
obtain the latest information needed to purchase gas and transport it by pipeline,
such as suppliers’ prices, available pipeline capacity, etc (1). The EBB will also
carry information about capacity release in pipeline operations and permit
trading of capacity release options so that elements of pipeline capacity will
become a commodity, just as the natural gas itself. The objective is to optimize
the efficiency of the national gas pipeline grid (2).
1.2. Methane (Natural Gas). Although the first natural gas pipeline
was constructed in 1870, most of the gas consumed at that time was manufac-
tured from coal (qv) and used locally rather than transported by long-distance
pipeline. In the 1990s, natural gas is conveyed in strong, thin-walled, long-dis-
tance pipelines in virtually all principal countries of the world. In the United
States, the total length of gas pipelines is ca 5:82  105 km, which includes
ca 4:24  105 km of transmission pipelines, ca 1:49  105 km of field (gathering)
lines, and ca 0:09  105 km of lines in storage areas (3,4). This represents ca 73%
transmission pipelines, 25.5% gathering lines, and ca 1.5% storage lines. Collec-
tively, long-distance gas pipelines account for only about one-third the natural
gas transport system in the United States; the other two-thirds are gas distribu-
tion mains to consumers.
Polyethylene (PE), poly(vinyl chloride) (PVC), or polypropylene plastic pipe
is being used in increasingly greater amounts throughout the world in both gas
gathering systems and gas distribution systems. U.S. gas distribution companies
have installed 644,000 km of polyethylene piping since its introduction in 1968
and are adding over 40,000 km each year (5). Plastic pipe is used in sizes greater
than 400-mm dia for certain nonpressurized nongas applications, but is not used
in sizes of greater than 400-mm dia for natural gas and not at all for truly long-
distance natural gas pipelines. When plastic pipe is installed under roads, cas-
ings are often required. Railroads do not allow polyethylene pipe to be used at
crossings; the additional installation costs for crossings can reduce the economic
advantages of using plastic pipe for long pipelines (6).
Transport of natural gas starts with small-diameter gathering lines that
convey the raw gas from individual wells to a collection point and from there
to a gas treating plant, where heavier hydrocarbons (qv), particulates, and
water are separated from the gas. The hydrocarbons known as natural gas
liquids (NGL) are transported as raw mix in liquid pipelines to chemical plants,
where they are fractionated into lighter and heavier fractions that are trans-
ported to market areas in products pipelines. The methane-rich gas, which
usually contains some ethane, is processed to remove particulates, water, and
Vol. 1 PIPELINES 853

gaseous impurities such as hydrogen sulfide, carbon dioxide, etc, and then com-
pressed to the appropriate pressure for transmission by large-diameter pipeline.
Pressure is maintained by large compressors at compressor stations along the
length of the pipeline, usually spaced 80–160 km apart (7). Maximum operating
pressures are 3.45 MPa (500 psi) for older lines and 9.93 MPa (1440 psi) for
newer lines.
Of the 125 U.S. gas pipelines listed in 1991, 47 were rated as principal
pipelines, ie, 1:42  109 m3 of gas transported or stored for three consecutive
years (4). The longest gas pipelines originate near the largest U.S. gas producing
areas, primarily the Gulf of Mexico and the mid-continent area of Oklahoma,
Texas, Kansas, New Mexico, etc, and move gas to the more populated areas of
the country, ie, Los Angeles, Chicago, New York, or New Jersey. Maps showing
the locations of natural gas pipelines in the United States (8) and Europe (9) are
available.
Some U.S. natural gas pipeline companies are subsidiaries of gas holding
companies. The largest U.S. natural gas pipeline companies, in terms of overall
length of transmission systems are Northern Natural Gas Company, 26,539 km;
Tennessee Gas Pipeline Corporation, 23,567 km; Columbia Gas Transmission
Company, 18,481 km; Natural Gas Pipeline Company of America, 17,200 km;
and Transcontinental Gas Pipe Line Corporation, 17,071 km. For gas moved in
1994, the four largest pipelines were ANR Pipeline Company, 95; 278  106 m3
(3,363,275 MMcf), of which 40.8% was gas moved for others; Transcontinental
Gas Pipe Line Corporation, 87; 050  106 m3 (3,073,801 MMcf), of which 99.7%
was moved for others; Natural Gas Pipeline Company of America,
83; 089  106 m3 (2,933,940 MMcf), of which 87.1% was moved for others; and
Northern Natural Gas Company, 56; 523  106 m3 (1,995,861 MMcf), with
100% moved for others.
In Europe, Russia’s huge natural gas resources are being moved to Western
Europe by pipelines from Siberia to the consuming countries. Gas from Algeria’s
large resources is moved by 2500-km, 1200-mm dia pipeline from the Hassi R’Mel
gas field across the Mediterranean Sea and the Strait of Messina to Italy in water
depths reaching to 610 m. A 1448-km pipeline will carry Algerian gas to Spain
and Portugal via Morocco. Norway’s Statoil will build a 1095-km pipeline system
(Zeepipe) from the North Sea to Europe. In South America, a 2252-km gas pipe-
line will transport natural gas from producing fields in Bolivia to Sao Paulo, Bra-
zil, and hook up with a line between Sao Paulo and Rio de Janiero. A gas pipeline
that would cover 27,000 km has been proposed by the National Pipeline Research
Society of Japan to transport gas from Siberia and Sakhalin Island to Australia
by way of Japan, China, Thailand, Malaysia, and Indonesia (9).
Ammonia Pipelines. Ammonia [7664-41-7] is a commodity produced from
natural gas and may be produced in plants located near its market areas or
plants may be located near gas-producing regions and the ammonia transported
to the market areas by tank car, barge, or pipeline. The primary ammonia pipe-
lines in the United States are the Mid-America Pipeline Company (MAPCO) and
Gulf Central, both connecting ammonia-producing areas in Texas, Oklahoma,
and/or the Gulf Coast with farming districts throughout the Midwest. The pipe-
lines are made of high strength pipeline steel, 200–250-mm dia, and operate
with an ammonia pressure of 2.07–10 MPa (300–1450 psi) to maintain the
854 PIPELINES Vol. 1

ammonia in the liquid state. A limit of 0.2% H2O and 0% CO2 in the ammonia is
specified to minimize stress–corrosion problems. Stress–corrosion cracking of
the high strength pipeline steel can occur if the ammonia is contaminated by
atmospheric carbon dioxide. Copper-base alloys are corroded by ammonia in
the presence of air and water, and equipment containing these metals can be cor-
roded by even small ammonia leaks. The MAPCO ammonia pipeline has resolved
these problems by using a special steel containing 0.25% C, 1.0–1.7% Mn, and
0.15% Cu; eliminating copper alloys from valves, fittings, and pumps; stress-
relieving fabricated valves or using cast valves and stress-relieving pipe fabrica-
tions containing branch connections; and using neoprene for valve trim (10).
A Russian ammonia pipeline of nearly 2400 km extends from Togliatti
on the Volga River to the Port of Odessa on the Black Sea, and a 2200-km,
250-mm dia branch line extends from Gorlovka in the Ukraine to Panioutino.
The pipeline is constructed of electric-resistance welded steel pipe with 7.9-mm
thick walls but uses seamless pipe with 12.7-mm thick walls for river crossings.
The pipeline is primed and taped with two layers of polyethylene tape and
supplied with a cathodic protection system for the entire pipeline. Mainline
operating pressure is 8.15 MPa (1182 psi) and branch-line operating pressure
is 9.7 MPa (1406 psi) (11).
Hydrogen Pipelines. The manufactured gas distributed by early gas dis-
tribution systems contained up to 50% hydrogen, and at least two large-scale
operations in the 1990s have evaluated 10–20% hydrogen mixtures with natural
gas, but actual experience with long-distance hydrogen pipelines is rather lim-
ited. The oldest hydrogen pipeline (started in 1938) is the Chemische Werke
Hüls AG 220-km, 150–300-mm dia system in the German Ruhr Valley that
transports 100  106 m3 of hydrogen annually to multiple users at a nominal
pressure of 1.55 MPa (225 psi). Following its expansion after 1954, some fires
have occurred but no hydrogen embrittlement or explosions (12,13). Other
shorter H2 pipelines include a 340-km network in France and Belgium, an
80-km pipeline in South Africa, and two short pipelines in Texas that supply
hydrogen to industrial users (14). NASA has piped H2 through short pipelines
at their space centers for several years.
Natural gas is roughly eight times more dense than hydrogen (0.72 g/L vs
0.09 g/L). Because the pipeline capacity of a pipeline depends on the square root
of the gas density, the pipeline capacity of hydrogen is nearly three times greater
than for natural gas. The heating value of hydrogen, however, is only about one-
third of the 37.2 MJ/m3 heating value for an equivalent volume of natural gas.
Thus, the energy carrying capacity of a given size pipeline is approximately the
same when it is carrying either hydrogen or natural gas, provided that it is oper-
ating in turbulent flow with the same pressure drop along its length and at the
same operating pressure. At higher pressures, the ratio of heating values of nat-
ural gas and hydrogen increases and at 5.17 MPa (750 psi); the ratio is 3.8:1,
compared with about 3:1 at atmospheric pressure. This also means that hydro-
gen compressors must handle 3.8 times more gas than natural gas compressors
for the same energy throughput, thus indicating higher pipeline transmission
costs for hydrogen than for natural gas. One study suggests that regional trans-
port costs for hydrogen may be as much as five times greater than for natural gas
at the higher pressures but that long-distance pipeline transport should not show
Vol. 1 PIPELINES 855

such a large difference (15). Some components currently used for natural gas
may be adequate for hydrogen service but not compressors and meters. In
view of the different compressor requirements for hydrogen and natural gas,
the design of rotary compressors used for natural gas is probably not satisfactory
for hydrogen use (12,16).
The question of whether hydrogen embrittlement of pipeline steel is a pro-
blem has not been completely resolved. No hydrogen embrittlement problems
were reported for hydrogen pipeline transmission at 5.17 MPa (750 psi), and
an American Petroleum Institute (API) study of steels operating in hydrogen
service indicates no problems at temperatures below 2048C and pressures
of 69 MPa (10,000 psi). Other tests of pipeline steels at hydrogen pressures of
3.5–6.9 MPa (500–1000 psi) have shown a loss of tensile ductility and acceler-
ated fatigue-crack growth. Embrittlement is a potential problem in high pressure
pipeline transmission of hydrogen (13).
1.3. Industrial Gases. Industrial gas (oxygen, nitrogen, etc) pipelines
are short compared with long-distance pipelines that transport crude oil, natural
gas, or petroleum products; however, more than 80% of the oxygen and more
than 60% of the nitrogen produced in the United States by air separation is
transported by pipeline. Air-separation plants that are built adjacent to the oxy-
gen or nitrogen users’ facilities are usually owned and operated by the user or a
second party. However, plants located at more distant locations are usually
owned and operated by the oxygen producer and the gases transported by pipe-
line to multiple users in the area, eg, the 150-km, 406-mm dia Houston
Ship Channel pipeline system that supplies oxygen and nitrogen from air-
separation plants to chemical plants and oil refineries from Houston to Texas
City, Texas (17). Carbon dioxide and hydrogen are also moved by pipeline in
the Houston area. Other U.S. industrial gas pipeline systems are located near
Gary, Indiana, and along the Mississippi River near the Gulf Coast (18).
Multiple users of oxygen, nitrogen, hydrogen, and carbon monoxide near
Rotterdam, the Netherlands, are supplied by pipeline from a nearby industrial
gas complex. Oxygen has been transported for many years by a French pipeline
connecting Metz with Nancy and extending to Luxembourg and Saarbrucken in
Germany.
The application that has led to increased interest in carbon dioxide pipeline
transport is enhanced oil recovery (see PETROLEUM). Carbon dioxide flooding is
used to liberate oil remaining in nearly depleted petroleum formations and trans-
fer it to the gathering system. An early carbon dioxide pipeline carried by-pro-
duct CO2 96 km from a chemical plant in Louisiana to a field in Arkansas, and
two other pipelines have shipped CO2 from Colorado to western Texas since the
1980s. Feasibility depends on crude oil prices.
Helium is extracted from natural gas in the southwestern United States
and moved by a 685-km, 50-mm dia pipeline to storage in a partially depleted
gas field near Amarillo, Texas, as part of the U.S. government’s helium conserva-
tion program.
1.4. Cryogenic Gases. Some of the most sophisticated pipeline technol-
ogy deals with the transport of liquefied gases with very low boiling points,
ie, cryogenic gases such as oxygen (182:96 C), argon (185:7 C), nitrogen
(195:8 C), hydrogen (252:8 C), and helium (268:9 C). These gases are
856 PIPELINES Vol. 1

liquefied by modern cryogenic methods and shipped in cylinders or special sto-


rage vessels for bulk liquid transport and storage (see CRYOGENICS). However,
development of a system for piping them as liquids has been brought about by
the needs of the space program, superconducting magnets, and other high tech-
nology areas. The use of liquid hydrogen as a rocket fuel and liquid oxygen as the
oxidant has resulted in piping these cryogenic materials to launching and test
sites; the use of liquid helium to cool superconducting magnets led to the devel-
opment of vacuum-jacketed, liquid nitrogen-shielded piping systems (qv). Most of
the lines have been relatively small-diameter systems; however, similar but
unshielded vacuum-jacketed lines up to 356-mm dia (19) and 8184 kPa (1200
psi), which would qualify as pipeline-sized piping, can be manufactured. CVI,
Inc. (Columbus, Ohio) produces pipe for liquid helium consisting of an inner
line of stainless steel that is wrapped with aluminized Mylar and glass fiber
paper and may be shielded by a patented aluminum extrusion cooled with liquid
nitrogen or cold helium gas; the extrusion shield is also wrapped with superinsu-
lating paper. This assembly is completely enclosed, tested, and factory sealed in a
vacuum jacket with a stainless steel outer pipe, as shown in Figure 1 (20). Piping

Shield
spacers-location
staggered from
liquid helium
line spacers
Aluminum–copper
liquid N2 / gaseous He Stainless steel
transfer line jacket pipe

Inner
line
spacers

Aluminum
shielding
extrusion

Liquid helium
transfer line SST

Superinsulation Stainless steel


inner pipe

Figure 1. Vacuum-insulated helium piping.


Vol. 1 PIPELINES 857

systems based on similar technology are also designed for transfer of liquid
oxygen, liquid nitrogen, and liquid hydrogen; liquefied natural gas has also
been transferred in vacuum-jacketed piping. Thus far, transfer lines are short
by comparison with the longer-range pipelines for other applications, but many
thousands of vacuum-jacketed piping installations in the 30–150-m range have
been installed, with one system over 8 km in length.

2. Pipeline Transport of Liquids

The main technical difference between liquid and gas pipeline transport is the
compressibility of the fluid being moved and the use of pumps, rather than com-
pressors, to supply the pressure needed for transport. The primary use for liquids
pipelines is the transport of crude oil and petroleum products.
2.1. Crude Oil and Products Pipelines. The true pioneering in long-
range modern oil pipelines was the construction of two pipelines to move crude
oil from Texas to New Jersey and Pennsylvania during World War II. These were
named the Big Inch (2156-km, 600-mm dia) and the Little Big Inch (2373-km,
500-mm dia) pipelines.
The crude oil delivery system starts with relatively small-diameter
gathering lines from individual producing wells to a main-line pump station,
from where it is pumped through a larger transmission trunkline to a refinery
or other destination. At the refinery, the crude oil is separated into gasoline,
jet fuel, kerosene, distillate fuel oil, etc, and the refined products are
transported by products pipelines to markets, storage, shipping terminals,
etc. In modern lines, all inputs and outputs are metered, monitored, and remo-
tely controlled by supervisory control and data acquisition (SCADA) computer
systems.
Several different refined products are shipped in the same pipeline
(batching) by using control methods to minimize intermingling at different inter-
faces. This is achieved by maintaining high turbulent flow in the pipe. Synthetic
rubber spheres have been used as separators at interfaces but are not used often
in the 1990s. Intermingling occurs at the interfaces, but it can be estimated and
minimized. At the terminal, the comingled fraction is separated from other pro-
ducts and either blended or returned to the refinery. Interfaces between products
are delineated with densitometers. This technique is used to ship different types
of crude oil in the same pipeline.
The liquids pipeline subsidiaries of principal oil companies with both
production and refining facilities generally operate all three types of liquids pipe-
lines: gathering, transmission, and products lines. However, the independent
9000-km, 900-mm dia Colonial Pipeline transports only products. It carries
about 105  106 m3 of products annually. Of the nearly 276,000 km of liquids
pipelines reported in 1991 by FERC, 47% were products lines, 34% were trans-
mission lines, and only 19% gathering lines. This compares with 35, 34, and 31%,
respectively, in the 1970s. The total length of the U.S. operating interstate pipe-
line system has remained relatively constant since 1982 at about 720,000 km,
with roughly 32% represented by liquids pipelines and 62% by natural gas
pipelines (4).
858 PIPELINES Vol. 1

At the beginning of 1992, the largest liquids pipelines in the United States,
based on pipeline length, were Amoco Pipeline Co., 19,096 km; Mobil Pipe
Line Co., 15,026 km; Exxon Pipeline Co., 14,983 km; and Conoco Pipe Line
Co., 12,980 km. Distances do not include 1316 km of the Trans-Alaska Pipeline
with multiple ownership. In both 1991 and 1992, the product pipeline company
with the most product deliveries was Colonial Pipeline with 104,990,000 m3,
more than double the amount delivered by Santa Fe Pacific Pipelines, Inc.
The top pipeline in terms of crude oil deliveries was the Alyeska Pipeline Service
Co., operator of the Trans-Alaska Pipeline System, with movement of
105,735,000 m3 (3).
2.2. Sulfur and Chlorine Pipelines. Underground sulfur is melted by
superheated water and then piped as liquid to the surface with compressed air.
At the surface, molten sulfur is transported by heated pipeline to a storage or
shipping terminal. One such pipeline, located under 15 m of water in the Gulf
of Mexico, is insulated and surrounded by steel casing to which are strapped
two 130-mm dia pipelines that carry return water from the deposit. The super-
heated water is carried from shore to the deposit in a 63.5-mm dia pipe inside the
pipeline that carries the molten sulfur (21).
Chlorine is shipped by pipeline in either gaseous or liquid form; however,
great care must be taken to ensure that liquid pipelines are operated only in
the liquid phase and gaseous pipelines operated only in the gaseous phase.
Liquid chlorine has a high coefficient of thermal expansion and high pressure
develops with increasing temperature; liquid chlorine trapped between two
valves can lead to hydrostatic rupture of the pipeline. Because chlorine is corro-
sive and has a high vapor pressure, it and the pipeline must be dry before the
chlorine enters the pipeline. Operating pressure should not exceed 217 MPa
(31,465 psi) and temperature should be no higher than 1218C (22).
2.3. Trans-Alaska Pipeline. The design and construction of the Trans-
Alaska Pipeline was one of the most difficult and ambitious pipeline projects ever
attempted. Its 1316-km length extends from the producing fields at Prudhoe Bay
across three mountain ranges, where up to 12 m of winter snow and wind chills
of 73 C can be expected, to the Valdez Marine Terminal on the southern coast
of Alaska. Thirteen bridges were built to carry the pipeline across rivers and dif-
ficult terrain. To avoid localized melting of the permafrost by the 32–498C crude
oil and possible instability of the soil, a significant portion of the pipeline was
installed above ground, supported by 78,000 vertical support members embedded
in permafrost to depths ranging from 4.5 to 18 m. The pipeline is insulated and
jacketed to reduce heat losses, and passive refrigeration is used to keep the soil
frozen around the vertical support members which can move up or down, thus
allowing a 0.6-m vertical movement of the pipeline. The pipeline is laid out in
a zigzag fashion to allow a 3.6-m horizontal movement for pipe expansion and
contraction and possible seismic disturbances. The buried portion of the pipeline
is 611-km long, with 6.4 km of it refrigerated. The aboveground portion of the
pipe is at least 1.5 m above the surface to permit caribou migration underneath
and at least 3 m above the surface at the 554 special animal crossings. The pipe-
line is constructed of 1219-mm dia high tensile strength carbon steel, with wall
thickness of either 11.7 or 14.3 mm. Although the Prudhoe Bay oil discovery was
announced in 1968, the initial pipe of the pipeline was not laid until early 1975
Vol. 1 PIPELINES 859

because of many delays. Final weld was made May 31, 1977 (23, 24). Final cost,
including construction interest, was roughly $11 billion.

3. Pipeline Transport of Solids

Pipelines to transport solids are called freight pipelines, of which three different
types exist: pneumatic pipelines, the use of which is known as pneumotransport
or pneumatic conveying; slurry pipelines, which may also be called hydrotran-
sport or hydraulic conveying; and capsule pipelines. When air or inert gas is
used to move the solids in the pipeline, the system is called a pneumatic pipeline
and often involves a wheeled vehicle inside the pipeline, propelled by air moving
through the pipe (25). Slurry pipelines involve the transport of solid particles
suspended in water or another inert liquid. Hydraulic capsule pipelines trans-
port solid material within cylindrical containers, using water flow through the
pipeline for propulsion.
3.1. Pneumatic Pipelines. Pneumatic pipe systems are used to move
blood samples, medicine, and supplies between buildings in hospital complexes;
cash and receipts in drive-up banks; parts and materials in factories; refuse from
apartment complexes; and grain, cement, and many other materials. Most of
these are small diameter and usually short; however, a 17-km, 1220-mm dia
pneumatic pipeline has been used to transport rock in the former Soviet Union
since 1981, and a 3.2-km, 1000-mm dia line has moved limestone from the mine
to a cement plant in Japan since 1983 (22).
3.2. Slurry Pipelines. Finely divided solids can be transported in pipe-
lines as slurries, using water or another stable liquid as the suspending medium.
Flow characteristics of slurries in pipelines depend on the state of subdivision of
the solids and their distribution within the fluid system. Although slurry flow
and conventional liquid flow are both divided into laminar, transitional, and
turbulent flow, the contribution of the solids-to-slurry flow results in a further
characterization as either homogeneous or heterogeneous flow. Homogeneous
flow of slurries occurs when very finely divided solids are distributed uniformly
throughout the suspending medium, eg, bentonite slurries as drilling muds.
These slurries are usually quite concentrated and often have viscosities much
higher than those of the suspending medium itself. Heterogeneous flow occurs
when larger, irregular solid particles are slurried and distribution is not uniform
throughout the suspension; these slurries are usually of lower solids concentra-
tion and viscosities are comparable to that of the suspending liquid.
The difference between homogeneous and heterogeneous flow is due to the
deposition velocity, below which solid particles start to separate from the slurry
and build up in the pipeline, and is related to the degree of turbulence and the
particle fall rate. Homogeneous slurries behave more like single-fluid systems;
deposition velocity is primarily a property of heterogeneous slurries. Deposition
velocity increases with increasing particle density, solids concentration, particle
size, and pipeline diameter. In the transport of heterogeneous slurries, turbulent
flow is an important requirement to prevent solids from building up in the pipe-
line. Slurries may be classified according to particle size, eg, colloidal (<1 mm),
structured (1–50 mm), and finely dispersed (50–150 mm, mostly produced by
860 PIPELINES Vol. 1

grinding). A polydispersed structured category, often encountered with products


of technological processes and produced by dispersion and grinding, is also
defined and described as containing a broader ranger of particles, eg, from
finer particles to coarser particles, and sometimes lumps (26). A maximum
flow velocity to minimize pipe wall erosion must be determined.
Slurry pipeline design is similar to the design of conventional liquid pipe-
lines, except for the slurry preparation stage and, if necessary, an additional
step for separating the suspended solids from the suspending liquid at the
point of use. In some instances, such as the direct firing of a concentrated homo-
geneous coal slurry, this final separation step is unnecessary. In the United
States, the Coal and Slurry Technology Association has focused its attention
on the development of coal–water–fuel (CWF) and its eventual commercializa-
tion, rather than on transportation by slurry pipelines (27).
Coal Slurry Pipelines. The only operating U.S. coal slurry pipeline is the
439-km Black Mesa Pipeline that has provided the 1500-MW Mohave power
plant of Southern California Edison with coal from the Kayenta Mine in north-
ern Arizona since 1970. It is a 457-mm dia system that annually delivers
4:5  106 t of coal, the plant’s only fuel source, as a 48.5–50% slurry. Remote
control of slurry and pipeline operations is achieved with a SCADA computer
system. In 1992 coal delivery cost from mine to power plant was calculated to
be $0:010=tk m ð$0:015=tm
 iÞ (28).
Several coal slurry pipelines were planned for the United States during the
1980s, primarily to deliver low sulfur coal from mines in Montana and Wyoming
to power plants in Texas and other states in the southern United States. None
was built, however, because of vigorous opposition from water conservationists,
who opposed using scarce water resources for the slurrying medium, and the
railroads, who feared competition for the utility coal markets (29,30). In addition,
lower prices for oil and natural gas provided little incentive to develop lower
cost competitive fuels; however, a study for the U.S. Department of Energy
(Pittsburgh Energy Technology Center (PETC)) indicated that coal–water
fuels (CWF) could be produced for less than $1.91/m3 of oil equivalent (31),
and it is expected that there will be renewed plans for coal slurry pipelines (32).
The former Soviet Union constructed a 262-km, 508-mm dia experimental
coal slurry line between the Belovo open-pit coal mine in Siberia’s Kuznets basin
to an electric power plant at Novosibirsk, using technology developed by Snam-
progetti. Testing began in late 1989 and tentative plans call for construction of
two much larger slurry pipelines, each 3000-km long, with capacity to move a
total of 33  106 t=yr to industrialized areas near the Ural Mountains (27,33).
Mineral Slurry Pipelines. Bentonite clay slurries, used as drilling muds in
oilwells, are probably the most universally used mineral slurries. Although they
are not transported by pipelines over great distances, they represent a signifi-
cant contribution to slurry pipeline technology because they led to the develop-
ment of powerful pumps needed to force oilwell drilling muds downhole to the
rotating bit, where the mud cools the bit and also carries rock chips from the
hole back to the surface. The pumps developed for this application contributed
greatly to the development of the huge positive-displacement reciprocating
pumps used for pumping slurries through pipelines (29). Table 1 gives some
examples.
Vol. 1 PIPELINES 861

Table 1. Mineral Slurry Pipelines


Pipeline Location Slurry Length, dia (km,mm) 106 t/yr
M. & Chem. Philip Georgia (U.S.) kaolin 25.6,200
Trinidad Cement Australia clay, limestone 9,200
Calaveras California (U.S.) limestone 27,150 1.36
VALEP Brazil 61% phosphate 120,250 2
SAMARCO Brazil 66% iron ore 400,500 6.35
Da Hong Shan China iron ore
Pena Colorado Mexico iron ore 48,200 1.8
Savage River Tasmania iron ore 85,250 2.3
Waipipi New Zealand iron sands 9.3a

Bougainville copper ore 27,152


Bougainville copper ore 105,102
Hondo Japan ore wastes 71a
South Africa uranium sands 19,150
a
Length.

Fiber Slurry Pipelines. Pipelines to carry suspensions of wood, paper,


sludge, etc, have found commercial acceptance. Most of them are less than
15 km long but have diameters of up to 500 mm. These slurries are often concen-
trated and display viscous plastic properties, although particle sizes may vary;
special pumps are used. One such hydrotransport system carries a cellulose
slurry by pipeline from the plant to a paper plant near Heidenau, Germany.
The 250-mm dia pipeline carries 60 t/d over the 3-km distance to thickeners.
In Sweden, a 3.7-km, 500-mm dia pipeline moves cellulose by pressurized hydro-
transport from a cellulose plant in Wifstaur to a sulfite plant at Fagervik (26).
The former Soviet Union has also been active in pipeline transport of fiber
slurries in the cellulose and paper industries and the movement of municipal
waste from aeration plants. For example, municipal sewage sludge from the
Kozhukhovskaya purification plant has been transported by a 2-km, 300-mm
dia pipeline to settling areas since 1936, and a 70-km, 400-mm dia pipeline
has moved sludge from the Lyublino aeration plant in Moscow since 1954 with
no plugging (26).
3.3. Capsule Pipelines. Capsule pipelines involve the transport of
material inside a closed cylindrical container propelled by water flowing through
the pipeline. This is called hydrotransport and the diameter of the cylindrical
container is approximately 90–95% the diameter of the pipeline. Water is 1000
times more dense than air under standard conditions, so the buoyancy is much
greater than in pneumatic pipelines and permits capsules to be suspended at
relatively low water flow velocities. First proposed during World War II, the con-
cept was resurrected in the 1960s and has been demonstrated amply in several
locations, using various sizes and lengths of pipelines. Considerable research
activities on capsule pipelines have been ongoing in several places, including
Japan, Australia, South Africa, the Netherlands, and the United States (25).
The University of Missouri, Columbia, has the only research center and testing
facilities devoted to capsule pipelines in the United States and has a 131-m,
203-mm dia test capsule pipeline devoted to this activity. The concept has been
expanded to include a coal log pipeline.
862 PIPELINES Vol. 1

The coal log pipeline involves forming logs of coal by compressing and
extruding finely divided coal into cylindrical shapes having a diameter 90–
95% the diameter of the pipeline. The coal logs are then injected continuously
into a water-filled pipeline for transport by the flowing water. The log length
is between 1.5 and 3 times the log diameter and buoyancy is achieved at a
water flow rate of 2.4–3.0 m/s. This reduces the energy required for pumping,
as well as the friction between coal log and pipeline walls. The coal log pipeline
is said to use only one-third as much water for coal transport as coal slurry pipe-
lines, in addition to eliminating the dewatering step. No full-scale coal log pipe-
line has been built, but short test lines and research activities continue, and a
Coal Log Pipeline Consortium has been formed, with both private companies
and governmental agencies participating (34).

4. Pipeline Technology

Pipeline technology involves design, construction, maintenance, and operation.


Although certain aspects of the technology differ under different climatic condi-
tions, whether above or below ground or under water, etc, the basic steps are the
same for liquids pipelines as for gas pipelines.
4.1. Design. Pipeline design begins with a preliminary mapping of the
proposed route, noting areas to be avoided and such obstacles as rivers, railroads,
and highways. For cross-country pipelines, aerial and ground-control surveys are
made to establish the final alignment of the pipeline. Permission to cross all land
parcels must be obtained from landowners, and permits obtained from the proper
authorities to cross rivers, highways, and railroads; where permission is not
obtained, appeal is made to the courts under the right of eminent domain.
Approval for gas pipeline construction must be obtained from FERC and/or
state public utility commissions, and an environmental impact statement must
be filed. Liquid pipelines do not require FERC approval for construction; common
carrier pipelines are generally accorded the right of eminent domain and are
regulated by FERC or state authority.
While detailed routing is being established, mechanical design begins by
establishing pipe size, based on the volume and type of material being trans-
ported, design pressures, pipeline length, and spacing between compressor or
pumping stations. The strongest steel possible is used and pipe wall thickness
is determined by a code-specified design formula that recognizes pipe as an
unfired pressure vessel; the design formula for natural gas pipelines is given
by the industry code ASME B31.8 and also by federal regulation 49 CFR 12.
For steel pipe, the formula is as follows:

P ¼ ð2St=DÞ  F  E  T

where P ¼ design pressure in kPa (psig), and may be limited per federal stan-
dard 192-105 to 75% of the pressure if steel pipe has been subjected to cold
expansion to meet the specified minimum yield strength (SMYS) and then
heated to 4828C or held above 3168C for more than 1 h, other than by welding;
S ¼ yield strength in kPa (psig); t ¼ nominal wall thickness in mm (in.);
Vol. 1 PIPELINES 863

F ¼ design factor, which is 0.72 in rural areas and 0.40 in urban areas, with
intermediate values of 0.60 and 0.50 for gas lines and 0.72 for liquid lines;
E ¼ longitudinal joint factor, which depends on the type of pipe weld seam and
is 1 if seam strength equals pipe strength; T ¼ temperature derating factor,
which is 1 for operating temperatures of 1218C or less but <1 for higher tempera-
tures, such as at compressor discharge; and D ¼ nominal outside diameter in
mm (in.).
This formula determines only the primary pipe (hoop) stress caused by the
internal fluid pressure perpendicular to the pipe wall; however, it also recognizes
secondary hoop stress through the design factor, F, which represents a safety
factor. The maximum design factor of 0.72 recognizes the secondary stresses
caused by the load of the overburden against an imperfect trench bottom,
those produced by thermal expansion or contraction against earth constraint
on buried pipe, and others. The design factor is lowered in populous areas to
provide additional safety and to recognize additional stresses in those areas
compared with rural areas; it is also lowered for installations at river crossings,
highways, etc. The maximum allowable operating pressure (MAOP) of the
pipeline can be affected by the pipe mill test pressure level and/or the hydro-
static strength test after construction is complete; either could underrate the
pipeline. Longitudinal primary tensile stresses are half of the primary hoop
stress. Specifications for valves, fittings, etc, must be written as part of the
design phase. Shut-off valves are required at specific intervals to isolate any
pipeline damage; powered valve-closing devices are commonly installed to
shut off valves automatically in the event of a line break and isolate the
damaged portion of the line. Valves on cross-country pipelines should be of a
full round-opening design to permit passage of devices, called pigs, for line
cleaning, separating water from gas during hydrostatic testing, corrosion
detection, etc. Specifications are made for exterior coatings and cathodic pro-
tection to minimize external corrosion, and for interior coatings or additives
to minimize interior corrosion, as well as decrease flow resistance and improve
transmission of odorants. Interior pipe coatings are usually epoxy paints
applied before pipeline assembly.
Spacing between compressor/pumping stations is usually 80–160 km along
the pipeline to boost pressure lost to internal friction by the gas or liquid being
transported. The compressors/pumps are driven by either reciprocating internal
combustion gas engines of up to 4.5 MW (6000 hp), centrifugal gas turbines
of up to 11.2 MW (15,000 hp), or electric motors of up to 7.5 MW (10,000 hp).
A compressor/pumping station usually consists of several compressors/pumps
operating in parallel.
4.2. Construction. Pipeline construction is a continuous activity
required by supply changes due to decreasing production from older fields, open-
ing up new fields, or bringing in new supplies from Canada; obsolete pipelines
are decommissioned, additions are made to existing pipelines, or new pipelines
are designed and laid to take advantage of new supply sources. Construction
of a cross-country pipeline may be divided into segments (spreads) and proceeds
by the following steps: right-of-way clearing and grading, trenching, pipe string-
ing along the right-of-way, pipe bending to fit the trench bottom, welding, clean-
ing and priming the exterior surface for coating, coating and wrapping for
864 PIPELINES Vol. 1

cathodic protection unless it is mill-coated and only the weld areas need coating,
lowering pipeline into trench, and backfill of earth over the pipeline.
Alternative construction methods are used for crossing obstacles. For river
crossings, floats may be used to carry the pipe across the river with concrete coat-
ings or weights, flooding the floats to sink the pipe to the river bottom where the
weights hold it in place. Land under highways is bored with an earth auger fol-
lowed closely by the pipe or casing, if required, advanced by the same machine.
With advances in horizontal directional drilling, bored river or wetlands cross-
ings of up to 762 m of bore have been made.
When construction is complete, the pipeline must be tested for leaks and
strength before being put into service; industry code specifies the test proce-
dures. Water is the test fluid of choice for natural gas pipelines, and hydrostatic
testing is often carried out beyond the yield strength in order to relieve secondary
stresses added during construction or to ensure that all defects are found. Indus-
try code limits on the hoop stress control the test pressures, which are also lim-
ited by location classification based on population. Hoop stress is calculated from
the formula, S ¼ PD=2t, where S is the hoop stress in kPa (psig); P is the internal
pressure in kPa (psig), and D and T are the outside pipe diameter and nominal
wall thickness, respectively, in mm (in.).
Tests of cross-country pipelines are recommended at 90% of pipe yield
strength in order to avoid having insufficient field-test pressure, which can
limit the MAOP of the pipeline to less than that calculated by the design formula.
According to industry code, this can be achieved only by hydrostatic testing with
nonflammable liquid, such as water. Filtered water is used to displace air left in
the line during construction; a pipeline pig separates the two phases and elimi-
nates air pockets. The water is pressurized to produce the hoop stress desired
and is recommended to be held for a minimum of 8 h to test pipe strength.
When the test is finished, the water is displaced by the pressurized fluid for
which the line was built, ie, natural gas, crude oil, etc. A pig is used to separate
the two phases and remove accumulated water at low points; for natural gas
pipelines, multiple pigs also remove water retained on pipe walls that could
cause operational problems. The pipeline is tied in to distribution and supply
systems, if it passes the tests, and put into service. The final step is a cleanup
and posting of required location markers.
Construction of underwater (submarine) pipelines does not take place
under water. Pipelines are welded onshore and dragged into position by powerful
winches on ships floating on the water surface (for short lines), welded on a
specially constructed lay barge, and lowered to the ocean floor by a stinger
from one end of the barge or welded onshore, floated on pontoons, and towed
to the offshore area where they are lowered into position. For smaller size
pipelines, the lines can be welded onshore and spooled onto large reels, placed
on special ships, and spooled into the offshore trench. In shallow water, or
where endangered by anchors or wave action, submarine pipelines are laid in
trenches in the sea bottom (35). Underwater pipelines are concrete-coated or
weighted to overcome the buoyancy effect; concrete coatings applied over the pri-
mary coating provide additional protection against damage during laying and
against corrosion. Submarine pipelines are being used regularly to transport
oil, natural gas, and other commodities to shore from offshore locations, such
Vol. 1 PIPELINES 865

as the Gulf of Mexico, the North Sea, and the Arabian Gulf. One of the longest
and technically challenging submarine pipeline systems is a 2599-km, 1200-mm
dia pipeline for transporting natural gas across the Mediterranean Sea and the
Strait of Messina from gas fields in Algeria to Italy; pipes are laid in water depths
to 610 m. Offshore and onshore pipelines require different design factors (36).
Trenchless Construction. Increased emphasis on environmental and
ecological factors has presented considerable difficulties for normal pipeline
trenching, even in unpopulated areas, if considered harmful to wildlife in their
native habitats. In densely populated areas, trenching is costly and complex
because of airports, commercial buildings, factories, etc. To circumvent them,
trenchless procedures are being developed for installing fuel lines, water lines,
sewer pipe, etc, or in some cases for steel casings that can serve as conduits
for several smaller diameter pipe systems. Horizontal directional drilling is a
trenchless construction technique that drills under an obstacle, such as a street,
airport runway, or river, and the welded continuous steel pipeline segment is
pulled through the opening after horizontal drilling is complete. Another proce-
dure that was used in Berlin to put a large-diameter casing under a railway
embankment between a train station and the River Spree (37) involves the use
of pneumatic pipe ramming to ram the steel pipe through the soil. Trenchless
construction is used commonly for small gas distribution lines.
Microtunneling involves sinking of two vertical shafts, large enough to
lower a horizontal drilling machine to the level from which to dig a tunnel
between the two vertical shafts to accommodate the pipeline. The technique
has proved useful for installing or rehabilitating sewer pipe systems and often
includes the sinking of additional shafts to provide access to the moling (horizon-
tal drilling) machines. Microtunneling was originally and principally developed
in Europe, but variations are being developed in both the United States and
Japan, and costs are expected to drop rapidly with expanded usage (38).Houston
and Dallas are among the U.S. cities that have applied this technique to revita-
lize underground utilities infrastructures (39–41).
4.3. Maintenance. Pipeline Pigging. After construction, a pipeline
must be tested, inspected, cleaned, mapped for bends, dents, or ovalties, main-
tained during operation, and monitored for leaks or corrosion to ensure safety
of operation. These operations involve pigging, moving devices called pigs
through the pipeline that can carry out these missions. These may be sophisti-
cated electronic devices or something as simple as a rubber ball pumped through
the pipeline to displace fluids (42).
Modern pipeline pigging applications begin with the commissioning of a
pipeline when pigs are first used to remove air prior to hydrostatic testing and
then used to remove any water left after the testing. A gauging (caliper) pig is
used to locate any dents or buckles resulting from the laying or backfilling opera-
tions. Special cleaning pigs are used to remove scale, dirt, etc, from interior
walls. If the pipeline transports multiple products, pigs can be used to separate
them at the liquid interfaces. To prevent loss of pipeline efficiency or increased
corrosion due to deposits of wax, scale, or bacteria, the deposits are removed with
brush pigs or even more sophisticated pin-wheel pigs that remove deposits to a
precise depth (43,44). Smart or intelligent pigs involving electronic sensor
interfaces with computers and miniature video cameras (45), ultrasonics,
866 PIPELINES Vol. 1

magnetic-flux leakage, etc, have been developed to inspect pipeline interiors for
integrity, metal loss, corrosion, and other defects; however, some pipelines have
physical limitations that do not accommodate smart pigs. The potential of smart
pigs to improve pipeline safety is considered to be so great that the Office of
Pipelines Safety published a ruling that would require modification of pipelines
to permit their use where feasible or practical; however, many petitions for
reconsideration were filed and a limited suspension of compliance has been
granted while the situation is under study (46). Several pigging operations can
be put into a single train (47).
Corrosion. Anticorrosion measures have become standard in pipeline
design, construction, and maintenance in the oil and gas industries; the prin-
cipal measures are application of corrosion-preventive coatings and cathodic
protection for exterior protection and chemical additives for interior protection.
Pipe for pipelines may be bought with a variety of coatings, such as tar, fiber
glass, felt and heavy paper, epoxy, polyethylene, etc, either pre-applied or
coated and wrapped on the job with special machines as the pipe is lowered
into the trench. An electric detector is used to determine if a coating gap
(holiday) exists; bare spots are coated before the pipe is laid (see CORROSION
AND CORROSION CONTROL).
Cathodic protection is provided by inducing an electric current on the pipe-
line to ensure that the electric potential of the metallic pipe is less than the earth
surrounding the pipeline after it is laid, by using a sacrificial anode bed with a
more electropositive metal than iron, or by thermoelectric protection, using heat
to generate a direct current to lower the potential of the pipe. Inducing an
electric current involves the use of a rectified alternating current to lower the
potential of the pipe relative to the ground, which also provides a way to monitor
the system by an aerial pipeline survey. Pulsating the direct current produces a
magnetic field with a strength proportional to the current flowing in the pipe;
using field coils to measure the magnetic field strength provides a method of
monitoring the current, even for underground pipe. If the survey indicates
that the magnetic strength is reduced over any segment of the pipeline, it
would indicate change of potential and current loss, such as would occur with
a hole in the pipeline coating, an electric casing short, a separated coupling, or
other event that would permit corrosive action (48).
U.S. Department of Transportation (DOT) statistics on liquids pipelines
operated under the Code of Federal Regulations (49) indicate that corrosion
was the second largest contributor to accidents and failures for the period from
1982 to 1991. These statistics covered an average of 344,575 km of liquids pipe-
lines and were derived from required reports to DOT on all pipeline accidents
involving loss of at least 7.95 m3 of liquid, death or bodily harm to any person,
fire or explosion, loss of at least 0.8 m3 of highly volatile liquid, or property
damage of $5000 or more (50). Similar results were also reported for 1991 in
the 1992 DOT/OPS report on both oil and gas pipeline incidents; 62 out of
210 oil pipeline incidents were due to corrosion, of which 74% were due to
external corrosion (43). For gas pipelines, 16 of all 71 reported incidents were
due to corrosion, of which 63% were reported as due to internal corrosion;
however, internal corrosion of gas pipelines is likely only if CO2 and H2O
and/or H2S are present, as with unprocessed gas in gathering lines.
Vol. 1 PIPELINES 867

4.4. Operation. Operations are controlled from a central computerized


office that maintains communication with compressor stations and flow sta-
tions along the route; reports are also received from weekly or biweekly aircraft
flyovers to inspect for potentially threatening conditions, such as soil erosion,
floods, approaching excavation, or any external evidence of leaks. Flow compu-
ters along the pipeline route monitor and quantify flow from producers entering
the pipeline or flow leaving the line to customers. Pipeline flow, pressure,
and other operating variables can be controlled by commands to compressor
stations. In systems that have remote control of main-line valves, central
control can isolate sections of pipelines with reported or threatened problems
and dispatch personnel to the site. If patrols detect indications of significant
increases in population density in specific areas, the safety factor in the design
formula may be increased in accordance with industry code, and the maximum
allowable operating pressure in those areas is reduced. Operations and
maintenance of pipelines are also covered in ASME B31.8 and federal regula-
tion 49 CFR 192.
Cross-country gas pipelines generally must odorize the normally odorless,
colorless, and tasteless gas in urban and suburban areas, as is required of gas
distribution companies. Organosulfur compounds, such as mercaptans, are
usually used for this purpose, and code requires that the odor must be strong
enough for someone with a normal sense of smell to detect a gas leak into air
at one-fifth the lower explosive limit of gas–air mixtures. The latter is about
5%, so the odorant concentration should be about 1%, but most companies
odorize more heavily than this as a safety precaution.

5. Economic Aspects

Pipelines usually represent the least costly way to move fluid products, including
solid slurries, wherever they compete with truck or rail transportation. Compar-
ing the transport costs for removing municipal sludge in the former Soviet Union
shows a surprising similarity to a similar comparison of transport costs for mov-
ing U.S. crude oil and petroleum (qv) products. The Russian study (26) showed
that pipeline transport of sludge is equal to 25–38% the cost of barge transporta-
tion, 6–13% the cost of rail transport, and 4–5% the cost of truck hauling. By
comparison, pipeline cost/km of moving crude oil and petroleum products in
the United States has been only about 17% of rail costs and 3.5% of truck
transport costs (26,51).
The cost of building a pipeline is usually divided into four categories: mate-
rials (line pipe, fittings, coatings, cathodic protection, etc), right-of-way and
damages, labor, and miscellaneous (surveying, engineering, supervision, admin-
istration and overhead, interest, contingencies, afudc, and FERC filing fees). The
relative contribution of each category depends on the size of pipe, the difficulty of
terrain, and delays and added costs due to environmental factors, but generally
the costs for materials and labor account for about 75% of the entire cost of the
pipeline, with each accounting for about half this amount. The remainder is
engineering, overhead, fees, interest, etc. The relative contribution of each
category can vary widely in difficult terrain or climate, however, as illustrated
868 PIPELINES Vol. 1

1860 3000
1802 2900

682 1100
620 1000
Costs in $1000 / km

Costs in $1000 / mi
558 900

496 800

434 700
372 600

310 500
248 400
186 300

124 200
62 100
102
152
203
254
305
356
401
457
508
559
610
660
711
762
813
864
914
965
1016
1067
Nominal pipe diameter, mm

Figure 2. Gas pipeline construction cost vs pipe size.

by the high labor costs of the Trans-Alaska Pipeline, which were 69.3% of the
total, and ca $1:3  109 interest charges during construction or roughly 15% of
the total pipeline cost (52).
The variation of average pipeline construction costs with increasing size of
line pipe is shown in Figure 2, based on data taken from FERC construction
permit applications from July 1991 to July 1992. The cost of a common carrier
pipeline project must be reported to the FERC no later than six months after
successful hydrostatic testing.
The 10-yr construction costs for land-based gas pipelines show no obvious
trend either in the four individual categories or the high–low range, on a $/km
basis; however, pipeline land acquisition costs rose steadily, with average $/km
cost increasing from $370,881 to $528,355 for proposed U.S. gas pipeline projects
over the 1990–1992 period. The $/km costs for 1991–1992 were material,
$190,379; labor, $200,127; right-of-way (ROW) and damages, $21,938; and mis-
cellaneous, $115,930 (4). The cost of compressor stations for the same period
averaged $1894/kW.
When a natural gas pipeline has been commissioned and is operating,
its transmission costs are the operating expenses plus maintenance expenses.
In 1989, these two cost elements for gas pipelines were $13,573/106 m3 and
$2872/106 m3, respectively; for 1990, the corresponding values were $17,669/
106m3 and $3851/106m3, respectively. The primary operating expense for
natural gas pipelines is the cost of compression and gas transmission by
others, which accounted for ca 40% of the 1989–1992 expenses. The next largest
Vol. 1 PIPELINES 869

operating expenses were compressor station labor and expenses, and gas for com-
pressor fuel, accounting for 30% of the total (4). Pigging and testing costs depend
on pig type, pipeline cleanliness and diameter, extent of corrosion, etc. The
Government Accounting Office survey shows a range of $404–$1492/km for
the cost of smart pig inspections; a similar survey of three foreign smart pig
manufacturers indicated a range of $746–2486/km for a 1609-km, 610-mm
dia pipeline. Hydrostatic test costs are even higher; estimates by the U.S.
DOT are $2890/km for a 508-mm dia hazardous liquids pipeline; this cost
rises to $5523–8794/km if costs of transporting, testing, and disposing of the
test water and the revenue loss due to the pipeline being out of service are
added. The cost of pressure testing and transportation or disposal of test
water is about $6563/km of pipeline, depending on the pipeline diameter, the
terrain crossed, and water scarcity in the area (46).

6. Safety, Environmental, and Ecological Aspects

Data compiled by the U.S. DOT indicate that pipeline transport is the safest
materials transport mode, particularly over long distances. In 1990, fatalities
attributed to oil and gas pipelines were significantly lower (8 out of 4679) for
all materials transporters in the United States, compared with 599 for rail trans-
port and 3281 for motor transport (trucks) (53).
To ensure the safety of gas pipelines, the Natural Gas Safety Act of 1968
was created to mandate federal regulation of gas storage facilities and pipeline
transport of natural gas, with the U.S. Department of Transportation given
exclusive authority to regulate safe operation of natural gas pipeline systems
that fall within the jurisdiction of the Federal Power Commission under the Nat-
ural Gas Act. The Hazardous Liquid Pipeline Safety Act of 1979 and the Hazar-
dous Materials Transportation Act of 1979, and their respective amendments,
gave authority for the U.S. DOT to regulate safety for all pipelines operating
under the Code of Federal Regulations (49). The U.S. DOT continues to regulate
safety on pipelines and has established legal standards for all pipeline phases,
from design to construction, operation, and maintenance. DOT also issues statis-
tics on pipeline accidents from required reports on accidents involving release of
gas or LNG and death or hospitalization or losses in excess of $5000.
Between 1985 and 1991, 1726 natural gas pipeline ruptures and leakages
were reported in the United States. These incidents resulted in 634 injuries
and 131 fatalities. Third-party damage was the most common cause of these inci-
dents, followed by corrosion. The GAO believes that the corrosion-related inci-
dents can be reduced with the use of smart pigs (46). U.S. DOT 1992 accident
statistics showed that 52.5% of U.S. oil spills involving loss of at least 1590 m3
came from pipeline accidents, comparable to the worldwide statistic of 51.5%.
The U.S. DOT regulated 344,575 km of liquids pipelines during the 10-yr
study period and received reports on 1901 accidents during that time; thus the
number of failures per year per 1000 miles was 0.888, of which 27% was due to
corrosion and 31% to outside forces (48).
Most ecological issues involve the pipeline’s construction phase and the
effect that it will have on vegetation, wildlife, topography, population density,
870 PIPELINES Vol. 1

and land use. All pipeline design and construction programs must include an
environmental impact statement which must be submitted and evaluated before
permission to proceed with construction is granted. Construction through ecolo-
gically sensitive obstacles, such as rivers and wildlife areas, is often suspended
during certain times of the year, such as fish spawning and wildlife migration
seasons, unless acceptable alternatives can be found. Archaeological concerns
must also be addressed.
Greater natural gas use for power generation and vehicular fuel use, to help
meet more stringent environmental standards, is expected to expand natural gas
pipeline usage. High technology will influence the role of post-Order 636 gas
pipelines (54). Pipeline transport of coal could become a reality in the United
States, as well as other coal-rich countries, to reduce dependence on imported
oil. Greater use of freight pipelines to reduce pollution from trucks and other
transportation modes may arise (25). Pipeline research activities are expected
to expand through organizations such as Japan’s National Pipeline Research
Society and the Capsule Pipeline Research Center and Gas Research Institute
in the United States.
The concept of transporting heat over great distances with chemical energy
pipelines has been considered (55).

BIBLIOGRAPHY

‘‘Pipelines’’ inECT 3rd ed., Vol. 17, pp. 906–929, by T. P. Whaley and G. M. Long, Institute
of Gas Technology.

CITED PUBLICATIONS

1. P. Crow and A. D. Koen, Oil Gas J., 15–20 (Aug. 9, 1993).


2. L. Parent, ed., The Gas Price Report, private communication, 1994.
3. J. Watts, ed., Pipeline Gas J., 12–36 (Sept. 1992).
4. W. R. True, Oil Gas J., 41–62 (Nov. 23, 1992).
5. F. W. Griffith, Pub Utilities Fortnightly 131(6), 21 (1993).
6. S. J. Davis, Pipeline Gas J., 44–47 (Dec. 1992).
7. ‘‘Transportation of Natural and Other Gas by Pipeline,’’ Code of Federal Regulations,
Hazardous Materials Regulations Board, U.S. Dept. of Transportation, Minimum
Federal Safety Standards, U.S. Printing Office, Washington, D.C., Title 49, Chapt.
1, Part 192.
8. Map of Major Natural Gas Pipelines, U.S. Dept. of Energy, for sale by the Superin-
tendent of Documents, U.S. Printing Office, Washington, D.C. Crude oil and pro-
ducts pipeline maps also available.
9. Pipeline Gas J., 14–16 (Aug. 29, 1992); Gas World Intl., 16–24 (May 1993).
10. J. J. Lee, Oil Gas J. 66, 103 (Nov. 4, 1968).
11. N. Hetland, Pipeline Gas J. 250, 38 (Oct. 1978).
12. D. P. Gregory and J. B. Pangborn, Ann. Rev. Energy 1, 279–310 (1976).
13. C. F. Blazek, R. T. Biederman, S. E. Foh, and W. Jasionowski, Proceedings of the 3rd
Annual Hydrogen Meeting, A Blueprint for Hydrogen’s Future with Enabling Tech-
nologies Workshop, Mar. 18–29, 1992, pp. 4-203–4-221.
Vol. 1 PIPELINES 871

14. J. H. Kelley and R. Hagler, Jr., Int. J. Hydrogen 5, 35–54 (1980).


15. C. Carpetis, in C-J. Winter and J. Nitsch, eds., Hydrogen as an Energy Carrier,
Springer Verlag, New York, 1988, Chapt. 9.
16. J. Pottier, E. Blondin, and A. Garat, in T. N. Veziroglu and A. N. Protsenko, eds.,
Hydrogen Energy Progress VII, Proceedings of the 7th World Hydrogen Energy
Conference, Vol. 2, Moscow, Russia, Sept. 25–29, 1988, 1095–1107.
17. Oil Gas J. 6 (Oct. 10, 1983).
18. Technical data, Air Liquide America, Walnut Creek, Calif., 1994.
19. S. Hensley, Vacuum Insulated Pipe, CVI, Inc., Columbus, Ohio, personal communi-
cation, 1994.
20. Technical data, Cryogenic Standard Products, CVI, Inc., Columbus, Ohio, 1994; U.S.
Pat. 4,233,816 (to CVI, Inc.).
21. J. A. Oates, ed., Pipes and Pipelines Manual and Directory, 3rd. ed., Scientific
Surveys, Ltd., London, 1970, 7–19.
22. Chlorine Pipelines, The Chlorine Institute, Inc., Washington, D.C.
23. Technical data, The Trans Alaska Pipeline, Alyeska Pipeline Service Co., Anchorage,
Alaska, 1992.
24. E. S. Newcomer and P. G. McDevitt, Oil Gas J., 49–57 (Sept. 14, 1992).
25. H. Liu, ‘‘Freight Pipelines for Transporting Solids and Packages,’’ presented at
ASCE 1993, Annual Convention and Exposition, Dallas, Tex., Oct. 26, 1993.
26. A. Y. Smoldyrev and Y. K. Safonov, in W. C. Cooley, ed., Pipeline Transport of Con-
centrated Slurries, Terraspace, Inc., Rockville, Md., trans. by A. L. Peabody, 1979.
27. R. M. Braca, Pipeline Gas J., 32–36 (Jan. 1988).
28. J. K. Anderson, Black Mesa Pipeline, Inc., personal communication, Jan. 5, 1994.
29. E. J. Wasp, Mech. Eng. 101, 38 (Dec. 1979).
30. 25(11), 12–14 (Nov. 1988); 95(1), 16, 5 (Jan. 1990).
31. PETC Review 8, 24–31 (Spring 1993).
32. S. D. Serkin, Coal and Slurry Technology Association, Washington, D.C., personal
communication, Feb. 1994.
33. Oil Gas J., 44 (Sept. 18, 1989).
34. H. Liu, ‘‘Coal Log Pipeline; Economics, Water Use, Right-of-Way, and Environmen-
tal Impact,’’ presented at The 10th Annual Pittsburgh Coal Conference, Sept. 20–24,
1993, Pittsburgh, Pa.
35. R. J. Brown, Oil Gas J., 106–111 (May 2, 1983).
36. M. Hein, Oil Gas J., 146–157 (May 2, 1983).
37. Pipeline Util. Constr., 36 (Oct. 1992).
38. Pipeline Util. Constr., 26–32 (Oct. 1992).
39. Pipeline Util. Constr., 24–26 (Feb. 1993).
40. A. V. Almeida, Civil Eng. 62(9), 71–73; Fluid Abstr. 3(5), 6 (1992).
41. Pipeline Util. Constr., 42–44 (Mar. 1992).
42. P. C. Porter, Pipeline Gas J., 37–47 (Aug. 1993).
43. G. Smith, Pipeline Gas J., 35–40 (Aug. 1992).
44. Pipes Pipelines Int., 5–7 (Nov.–Dec. 1992).
45. Pipeline Gas J., 51 (July 1993).
46. J. Watts, ed., Pipeline Gas J., 55–56 (Dec. 1992).
47. M. S. Keys and R. Evans, Pipeline Gas J., 26–33 (Mar. 1993).
48. R. M. Cameron, Pipeline Gas J., 34–40 (Mar. 1993).
49. Code of Federal Regulations, U.S. Dept. of Transportation, U.S. Printing Office,
Washington, D.C., Title 49D, Part 195.
50. D. J. Hovey and E. J. Farmer, Oil Gas J., 104–107 (July 7, 1993).
51. Capsule Pipeline Res. Center NEWS 2(1) (Spring 1993).
52. Oil Gas J. 76, 63 (Aug. 13, 1978).
872 SULFUR AND HYDROGEN SULFIDE RECOVERY Vol. 1

53. National Transportation Statistics Annual Report, DOT-UNTSC-RSPA, Washington,


D.C., June 1992.
54. L. V. Parent, ed., Pipe Line Prog., 13 (Nov. 1993).
55. N. R. Baker, T. P. Whaley, and co-workers, Transmission of Energy by Open-Loop
Chemical Energy Pipelines, Institute of Gas Technology Final Report, Sandia
Contract No. 87-9181, June 1978, 10–17.

THOMAS P. WHALEY
Consultant
GEORGE M. LONG
Institute of Gas Technology

SULFUR AND HYDROGEN


SULFIDE RECOVERY
1. Introduction

The previous decade has seen a rapid shift in the mode of sulfur production
with native sulfur recovery by the Frasch process being eliminated in the
United States and in steep decline elsewhere (1). The increasing reliance on
sour natural gas (CH4, H2S, and CO2) throughout the World and the pro-
duction of low sulfur fuels in refineries has resulted in nearly all of the World’s
elemental sulfur coming from conversion of H2S. This trend will continue as
very large reserves of oil with associated sour gas and new nonassociated
sour natural gas reservoirs are developed in Kazakhstan and in the Middle
East region (1). In addition, high sulfur-content bitumen recovered from
Canada and Venezuela is expected to become a major source of synthetic
crude oil. The accessible bitumen in Alberta, equivalent in volume to all
crude oil reserves in Saudi Arabia, is upgraded to yield synthetic crude oil
with associated production of large amounts of sulfur. By 2010, it is expected
that production of 3 million bbl/day of synthetic crude oil in northern Alberta
will yield 5 million tons of sulfur per annum (1). With these factors in mind,
the bulk of this article will describe processes for removal of H2S from large
volume gas streams that originate from oil and sour natural gas recovery and
refining.

2. H2S Removal Using Aqueous Ethanolamines and Related Materials

2.1. General Comments. Removal of H2S and CO2 from industrial gas
streams is practiced on a very large scale in the natural gas, refining, and
hydrogen-production industries. Most processes are based on a closed recircu-
lation system that includes absorption and chemical reaction to remove the H2S

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.1921120603011615.a01.pub2
Vol. 1 SULFUR AND HYDROGEN SULFIDE RECOVERY 873

Fig. 1. Overview of amine sweetening process.

and CO2, then regeneration of the treating agent. Chemical solvents, eg, aqu-
eous ethanolamines and a variety of physical and hybrid solvents, are used to
perform this function (2). For chemical solvents, the key aspect of the technol-
ogy (Fig. 1) is that H2S and CO2 are removed from high pressure gas streams in
a contactor at moderate temperatures (15–358C) and the chemistry is reversed
at higher temperature (1208C) in a regenerator yielding low pressure acid gas,
a mixture of H2S and CO2. As may be seen from the general scheme depicted in
Figure 1, heat exchangers are used to heat the rich solvent and to cool the aqu-
eous solvent to ambient temperatures such that residual levels of H2S do not
exceed 4 ppmv in the treated gas. The amount of CO2 removal depends on
the type of solvent used, operating conditions, and physical arrangement of
the contactor. In some processes, selective separation of H2S from CO2 is
achieved (3,4).
2.2. The Chemistry of H2S and CO2 Removal by Alkanolamines. All
amines are weak bases and produce aqueous solutions with pH 10–12, the
actual pH being dependent on the type of amine and the amount in solution
(Fig. 2). Absorption of H2S occurs by solubility and reaction with the free
amine and the protonated amine to yield soluble hydrosulfide species (Fig. 2).
The rates of these reactions are rapid with the result that removal of H2S from
a gas stream depends on mass transfer rates of the H2S from the gas-to-liquid
phases. Thus, the internal structure of an amine contactor is a critical design fea-
ture that along with countercurrent downflow of regenerated amine through the
contactor, engenders highly efficient H2S removal.
874 SULFUR AND HYDROGEN SULFIDE RECOVERY Vol. 1

RNH2 + H2O [RNH3+][OH−] All reactions are


acid–base proton
transfer
H2S H2S
Kinetics: Rapid

[RNH3 +][HS−] +][HS−]


[RNH3 + H2O

Fig. 2. The uptake of H2S by aqueous amines. The efficiency of H2S removal depends on
the concentration of [RNH3þ][OH] (gas–liquid contact requires countercurrent flow of
gas and liquids).

Absorption of CO2 from a gas stream is more complicated as it involves two


pathways, the bicarbonate and the carbamate mechanisms (Fig. 3). The bicarbo-
nate pathway is minor under most circumstances as the first step, formation of
carbonic acid (H2CO3), is equilibrium limited. The formation of the carbamate
(Fig. 3) is rapid. As in the case of H2S removal, the rate-limiting step is the
mass transfer of CO2 from the gas phase. Since at least one H atom must be
bonded to the N atom of the amine for carbamate formation to occur, tertiary
amines, eg, MDEA (Fig. 4) can be used for selective removal of H2S due to the
slow bicarbonate path kinetics, a desirable process for treating off-gas from redu-
cing Claus plant tail gas units.
A large number of ethanolamines (Fig. 4) are now used in the industry with
different blends and formulations for the various industrial gas mixtures under
consideration (2). The choice of solvent depends on a number of factors related to
treating pressure and temperature; energy requirements for regeneration; the
need to leave CO2 in the treated stream; degradation rates of the solvent due
to irreversible reactions of H2S, CO2, and other trace impurities in the feed
gas; and the potential for corrosion in a particular plant.
Key system design parameters are the amine solution strength, the rich
amine loading, which sets circulation rate, the lean amine loading (required to
attain 4 ppm H2S in treated gas), and the regeneration energy required to the
lean loading. Some typical parameters are given in Table 1.

RNH2 + H2O [RNH3+ ][OH–]


CO2 Kinetics: Fast H2CO3 Kinetics: Slow

RNHC–OH [RNH3+ ][HCO3– ] + H2O


=

O ‘bicarbonate’

RNH2
(b)

[RNHCO–][RNH+3]
=

O
‘carbamate’
(a)

Fig. 3. The uptake of CO2 by aqueous amines. The carbamate pathway is favored for
primary and secondary amines only; tertiary amines take up CO2 by bicarbonate route
only.
Vol. 1 SULFUR AND HYDROGEN SULFIDE RECOVERY 875

Primary Amines (R—NH2)


Monoethanolamine (MEA) Diglycolamine (DGA)
H
H
N CH2 CH2 OH N CH 2 CH 2 O CH 2 CH2 OH

H H
‘ethanol’ ‘diglycol’

Secondary Amines (R—N—R)


H
Diethanolamine (DEA) Diisopropanolamine (DIPA)

HO CH2 CH2 N CH2 CH 2 OH HO CH CH2 N CH 2 CH OH

H CH 3 H CH 3
‘isopropanol’

Tertiary Amines (R—N—R)


Triethanolamine (TEA) R Methyldiethanolamine (MDEA)

HO CH2 CH2 N CH2 CH2 OH HO CH2 CH2 N CH2 CH2 OH

CH2 CH2 OH CH3

Fig. 4. Molecular structure of common gas treating amines.

Note that rich loading values are a function of pressure. Values given in Table 1
are good for absorber pressures > 2700 kPa. Rich loading values should be
checked against equilibrium values available in open literature.
2.3. Hybrid and Physical Solvents. As is implied by the name, physi-
cal solvents, eg, glycols and NFM (Fig. 5) remove H2S and CO2 from industrial
gas streams without chemical reaction (5,6). Advantages of these solvents are
they can accept higher amounts of H2S and SO2 per volume of solvent; less
energy is required to regenerate the solvent; and these solvents possess
increased efficiency over alkanolamines for removal of COS and mercaptans,
two minor components sometimes present in industrial gases. A hybrid solvent,
a mixture of an ethanolamine and a physical solvent, is used when a combination
of the properties of two solvents is required (5). Overall, nearly all application
needs can be met by integration of design and solvent choice.
One limitation of both amine and physical solvent processes is that the
loaded solvent dissolves some CH4 or other hydrocarbons. Thus, the regenerated

Table 1. Typical Alkanolamine Parameters


Maximum rich Maximum lean
Concentration, loading, mol loading, mol Regeneration
Amine wt% AG/mol amine AG/mol amine energy, kJ/m3
MEA 15 0.40 0.12 334,440
DEA 25 0.45 0.06 278,700
MDEA 45 0.50 0.01 250,830
DGA 50 0.45 0.05 362,310
876 SULFUR AND HYDROGEN SULFIDE RECOVERY Vol. 1

Hybrid:
Sulfinol - Sulfolane / H2O / amine (DIPA, MDEA )

Hybrisol - MeOH / H2O / amine ( DEA )

Physical:
Selexol - dimethylethers of polyethylene glycols
CH3 O(CH 2 CH2 O)n CH3 (n = 3 – 9)

Morphysorb - N-formyl-morpholine ; N-acetyl-morpholine


(NFM) (NAM)

better removal of other S compounds (e.g., COS and RSH)

higher coabsorption of hydrocarbons (than aqueous amine systems)

Fig. 5. Representative physical sweetening solvent systems.

acid gas contains H2S, CO2, and CH4, and aromatic and larger aliphatic hydro-
carbons if these materials are present in the raw gas. Usually, the hydrocarbon
content of the acid gas does not exceed 1 vol.%, but even this small amount can
have a significant impact on the operation of downstream sulfur recovery
units. Physical solvents typically absorb significantly more hydrocarbons than
chemical solvents.

3. Conversion of H2S to Elemental Sulfur by the Claus Process

3.1. General Comments. The original Claus process envisaged direct


oxidation of H2S (2) using air in a catalytic unit, but the associated temperature
rise could not be controlled when significant amounts of H2S were present.

1 1
H2 S þ O2 ! S8 þ Heat
2 8

This difficulty was overcome by designing a multistage process in which


the heat could be removed sequentially in a number of units (Fig. 6). A modern
Claus sulfur recovery plant (SRU) is a complex system consisting of a high
temperature furnace stage, a series of catalytic converters operating at suc-
cessively lower temperatures, possibly a tail gas unit for improved sulfur
recovery, and an incinerator (7). Overall, the SRU system can be designed to
achieve > 99% conversion of the SRUs inlet H2S to sulfur. When state of the
art tail gas units are applied, > 99.8% sulfur recovery is possible. Part of the
complexity of the Claus process arises from the fact that elemental sulfur exists
in two main forms: S2 at temperatures > 5008C and S8 below that temperature.
The two allotropes readily interconvert as the temperature crosses the
5008C threshold.

4 S2 Ð S8 þ Heat
Vol. 1 SULFUR AND HYDROGEN SULFIDE RECOVERY 877

2H2S: SO2

Fig. 6. Simplified process scheme for the modified-Claus sulfur recovery process.

The bond-breaking process to form S2 accounts for the unusual behavior


for the equilibrium % sulfur recovery as a function of temperature (Fig. 7),
as formation of sulfur at high temperature is endothermic. The overall enth-
alpy changes for both the furnace and catalytic steps of what is usually
referred to as the modified-Claus process are given in Figure 8. Note that the

PRACTICAL 100 98 %
LIMIT 95 %
CONV 3
CATALYTIC
205°C THERMAL STAGE
CONV 2 STAGE
225°C
90 89 %

CONV 1
305°C
% RECOVERY

80 R
R
TGCU c
R
70 c
67 %
98 % c
95 %
c WHB F
99 % + 89 %
60 67 %
900° – 1200°

LIQUID SULFUR

50
200 300 400 600 800 1000 1200

TEMPERATURE (°C)

Fig. 7. Claus sulfur recovery.


878 SULFUR AND HYDROGEN SULFIDE RECOVERY Vol. 1

OXIDATION

original –2 0
CAT.
Claus H 2S + 1/2 O 2 1/8 S8 + H 2O + heat (209 kJ)
process
SINGLE STEP ‘PARTIAL’ OXIDATION

MODIFIED - CLAUS:
SPLIT THE PROCESS INTO TWO STEPS

1. Combustion (Reaction Furnace)


–2 +4
H 2S + 3/2 O 2 high T SO 2 + H 2O + heat (518 kJ)

(2 H 2S + SO 2 high T 3/2 S2 + 2 H 2O)

2. Redox (Catalytic Converter)

–2 +4 CAT. 0
2 H 2S + SO 2 low T
3/8 S 8 + 2 H 2O + heat (108 kJ)

3 H 2S + 3/2 O 2 OVERALL 3/8 S8 + 3 H 2O + heat (626 kJ)

Fig. 8. Claus process overview.

furnace stage, despite producing S2 in an endothermic process, is, overall, exo-


thermic because of the formation of SO2 and H2O. Oxygen input, either via air
or O2 enriched air, can be adjusted to control the product gas H2S/SO2
ratio leaving the furnace. As Figure 8 shows, the furnace produces sulfur as
well as SO2 with the latter used to react with H2S in the downstream catalytic
stages in a 2:1 ratio. This ratio control is usually achieved by analysis of
sulfur species in the tail gas, with feedback of that information to the air/O2
flow control devices.
The catalytic stages are operated at successively lower temperatures
to achieve increased conversion to sulfur (8) (Figs. 6 and 7), but within the
dew point constraint that liquid sulfur must not be formed inside the pore struc-
ture of the catalyst to avoid mass transfer limitations. The simplified equation for
production of sulfur in the catalytic converters is as follows:

3
2 H2 S þ SO2 Ð S8 þ 2 H2 O þ Heat
8

The catalysts used in the converters are based on either alumina or


titania, with many variations and composite materials now applied to achieve
high efficiency and continuous operation over many years. The temperature
regime used in the catalytic converters (< 3508C) is such that S8 is now formed
by an exothermic process. For equilibrium processes, Le Chatelier’s principle
states that a system will move to oppose disturbances to the system. As such,
if sulfur is removed between the catalytic stages by condensation, the system
moves toward increasing conversion. The need for interstage condensation
of liquid sulfur requires gas reheating to avoid subdew point operation in the
Vol. 1 SULFUR AND HYDROGEN SULFIDE RECOVERY 879

Fig. 9. Reaction furnace chemistry: Main reactions.

converters (converters, condensers and re-heaters are designated conv., C and


R in Fig. 7).
3.2. The Principles of the Claus Furnace. The furnace stage of the
Claus process requires an alumina refractory lined reaction chamber and a
waste heat boiler to remove the heat of reaction, generally by production
of steam (Fig. 9). Steam or a heated fluid is produced in the WHB. This high
pressure steam (typically 1050–4200 kPag) can be utilized in other parts of the
plant for heat. This high pressure steam can be utilized in the amine plant to
drive the regeneration units, to drive steam turbines in the STU, or may be
exported. The configuration of the burner and internals of the furnace chamber
are designed to control the shape of the flame, to avoid hot spots developing on
the refractory lining, and to ensure good mixing of air and inlet acid gas. These
factors are particularly important with use of oxygen-enriched air or pure O2
supported combustion as flame temperatures can easily exceed the softening
point of the alumina brick lining (1700–18008C). The simplified chemistry
given in Figure 9 does not adequately describe the very complex combustion sys-
tem of the furnace as H2, CO, CS2, and COS are nearly always found in addition
to the major products of sulfur, H2O, SO2, and unreacted H2S. New work on
Claus furnace chemistry, much of which is summarized in a review article (9),
shows that H2 may arise by a series of partial oxidation processes and dissocia-
tion of H2S in the anoxic zone of the combustion chamber (Figs. 10 and 11). The
formation and retention of H2 in the process gas, a matter related to the design of
the WHB among other factors, is of considerable interest as H2 is a necessary
component for tail gas units that produce H2S for recycle to the front end of
the plant.
The formation of COS and CS2 is related to the presence of CO2 and CH4 or
other hydrocarbons in the acid gas fed to the main burner of the furnace (10,11).
880 SULFUR AND HYDROGEN SULFIDE RECOVERY Vol. 1

Combustion chamber WHB


600oC
O 2 (N2) 2000oC 1300 oC

flame zone
acid gas Anoxic zone
(H2S/CO2/H2O/CH4)

H 2S/O 2 Ratio Products


H 2S + 3/2 O 2 SO 2 + H 2O (1) 0.66

2 H 2S + 5/2 O 2 2 SO 2 + H 2O + H 2 (2) 0.80

2 H 2S + O 2 S 2 + 2 H 2O (3) 2.0

2 H 2S + 1/2 O 2 S2 + H 2O + H 2 (4) 4.0

Fig. 10. Claus furnace: Primary combustion processes (fast).

Carbon dioxide can be a major constituent of acid gas stream, along with CH4
and other hydrocarbons that are present as a result of their solubility in rich
amine systems. The highly reducing conditions of the anoxic region of the com-
bustion chamber lead to the formation of COS and CO from CO2, as follows:

CO2 þ H2 ! CO þ H2 O
CO2 þ H2 S ! COS þ H2 O
2 CO þ S2 ! 2 COS
2 CO2 þ S2 ! CO þ COS þ SO2

Combustion chamber WHB


600°C
O 2 (N2) 2500°C 1300°C
flame

acid gas Anoxic zone


zone

Products
3/2 S2 + 2 H 2O 2 H 2S + SO 2

S 2 + 2 H 2O H 2S + H 2 + SO 2 (favoured)

2 H 2S 2 H 2 + S2

2 H 2 + SO 2 2 H 2O + 1/2 S 2

Fig. 11. Claus furnace: Anoxic, high T processes (sulfur species).


Vol. 1 SULFUR AND HYDROGEN SULFIDE RECOVERY 881

CH 4 C + 2 H2

C + H 2O CO + H 2
Endothermic reactions,
CO + H 2O CO 2 + H 2 a source of hydrogen

Overall : CH 4 + 2 H 2O CO 2 + 4 H 2

Also: CH 4 + 2 S 2 CS 2 + 2 H2S

C + S2 CS 2

Note: Other hydrocarbons also lead to CS2

Fig. 12. Claus furnace: Anoxic, high T processes (hydrocarbons).

Each of these species presents significant challenge to the operator as COS


must be eliminated in the catalytic converters and CO must be destroyed in the
incinerator.
It has been shown that CS2 arises only from hydrocarbons and not from
CO2 as had been assumed for many years (10). The solubility of CH4 and other
hydrocarbons in the loaded amine results in  0.5–1.0 vol.% hydrocarbon leaving
the amine regenerator with the acid gas. Thus, formation of CS2 and even soot
under unfavorable conditions is a perennial difficulty of Claus furnace operation
(Fig. 12). If the acid gas is high in H2S content, the overall temperature of the
anoxic chamber may be high enough for complete destruction of COS and CS2
as it is formed:

CS2 þ 2 H2 O ! 2 H2 S þ CO2
COS þ H2 O ! H2 S þ CO2
CS2 þ SO2 ! CO2 þ 3=2 S2
2 COS þ SO2 ! 2 CO2 þ 3=2 S2

But, the large quantity of CO2 in the acid gas from natural gas operations
results in too low a temperature for complete CS2 and COS destruction. This
situation may have to be rectified in the downstream catalytic stages to avoid
low total sulfur recovery and excessive emissions.
Due to the presence of nitrogen compounds in nearly all crude oils, refinery
Claus plants must be configured to deal with an NH3-containing stream originat-
ing from the refinery hydrotreating units. In major conurbations, transport of
ammonia is highly regulated and banned in some jurisdictions. Therefore, tech-
nology has been developed for conversion of the NH3 to N2 in the partially oxidiz-
ing regime of the Claus furnace typically with temperatures of  12008C
required. In most cases, specialized burners must be employed with injection
of the NH3-containing stream into the furnace chamber so as to avoid conversion
to NOx. The chemistry of this process (Figs. 13 and 14) has come under intense
882 SULFUR AND HYDROGEN SULFIDE RECOVERY Vol. 1

2 NH 3 + 3/2 O 2 N2 + 3 H 2O

2 NH 3 + O2 N2 + H 2 + 2 H 2O Exothermic

2 NH 3 + O2 N2 + 2 H 2 + H 2O

2 NH 3 N2 + 3 H 2 Endothermic

More O2 must be added to the combustion chamber

Fig. 13. Ammonia destruction in the Claus furnace.

scrutiny in the last 10 years (12) with implementation of commercial strategies


that take advantage of the rapid conversion of NH3 to N2 by reaction with SO2
(Fig. 14).
Overall, the furnace stage can be responsible for 65–70% of the sulfur pro-
duced in the entire plant (Fig. 7). As is evident from the preceding discussion, the
high temperature and anoxic conditions of the system lead to a very complex
mixture of gases, in addition to sulfur, SO2, and water. In a ‘‘classic’’ Claus
plant, O2 supply to the furnace is regulated to leave some H2S in the product
gas such that a 2:1 H2S/SO2 ratio is obtained for the catalytic stages. However,
high sulfur recovery and elimination of sulfur-based emissions from the plants
means, that in many cases, O2 supply is regulated to produce an H2S/SO2 ratio
of 4 or greater since many modern tail gas units work most efficiently under this
condition.
A Claus SRU typically contains at least two catalytic reactors (converters)
(Fig. 6) to further promote H2S/SO2 reaction to form sulfur and increase recovery
to 95–97%. These reactors generate low pressure steam that can provide much
of the amine regeneration energy.
3.3. The Principles of Claus Catalysis. The principles of operation
of a Claus converter are well known, but recently it has become clear that a
detailed understanding of the chemistry of this process is helpful in practical
operation of the units (13). A catalyst speeds up a chemical reaction by lowering
the activation energy of the process. In the case of the Claus reaction, it is
thought that adsorption of SO2 onto the surface weakens the bonds in that
molecule and activates it towards attack by H2S (Fig. 15). One ramification of

Oxidation rates: H 2S >> NH 3 ˜ CH 4

Consequences: H 2S + 3/2 O 2 SO 2 + H 2O
(excess)

4 NH 3 + 3 SO 2 2 N2 + 6 H 2O + 3/2 S 2
Rapid
reactions
2 NH 3 + SO 2 N2 + 2 H 2O + 1/2 S 2 + H 2

Fig. 14. Effect of rates of oxidation of ammonia destruction.


Vol. 1 SULFUR AND HYDROGEN SULFIDE RECOVERY 883

Fig. 15. Claus converter chemistry: Main reaction.

this adsorption is that a deficit of SO2 may be observed in the period shortly
after start-up of a new catalyst bed, resulting in the tail gas analyzer ‘‘overad-
justing’’ air or O2 supply to the main burner. It is generally accepted that the
catalyst bed must be preheated to the operating temperature (300–3408C for
the first converter) before introducing process gas to the catalyst. Preheat of
process gas leaving the upstream sulfur condenser is achieved by either O2-
fired combustion systems, heat exchange units or by electric heaters, as pre-
viously described.
The equilibrium limitation of the Claus reaction is illustrated by the Kp
values (Fig. 16) calculated for the 120–5008C range. The actual conversion to

The Claus Reaction


K eq
2 H 2S + SO 2 3/8 S8 + 2 H 2O + HEAT

PS83/8 PH2O2
Kp = K varies with temperature
PH2S2 PSO2

Temperature Kp
120°C 1.26 × 10 8

200°C 4.84 × 10 5 becomes larger at


lower temperatures
300°C 4.41 × 10 3 for exothermic rxns

400°C 175 Le Chatelier’s principle


500°C 16.9

Fig. 16. Equilibrium position of a chemical reaction.


884 SULFUR AND HYDROGEN SULFIDE RECOVERY Vol. 1

CATALYST
Al2O 3
2 H 2S + SO 2 2 H 2O + 3/8 S8 + heat
200 – 315 oC

EXOTHERMIC

EQUILIBRIUM vs KINETICS

%
CONV. RATE

TEMP TEMP

KEEP TEMPERATURE LOW KEEP TEMPERATURE HIGH


vs
TO IMPROVE CONVERSION TO OPTIMIZE RATE

BUT AVOID SULFUR DEW-POINT


( BY 5 – 15°C )

Fig. 17. Claus converter: Principles of operation.

sulfur is dependent on the amount of water already in the process gas from the
upstream amine unit and that produced in the furnace stage. Clearly, it would be
very beneficial to configure a Claus process that includes both water removal and
sulfur condensation after all stages but, as yet, no technology has been developed
that would allow this improvement to be implemented. The fact that a catalyst
has to be used at all illustrates that the gas-phase Claus reaction is too slow
for practical purposes at 120–3508C. Moreover, as depicted in Figure 17,
the practical operation of a Claus converter is actually an interplay between
three factors: equilibrium limitations (Fig. 7); reaction kinetics; and sulfur
dew-point control; with these last two factors being the overriding issues for
commercial operation.
A valuable insight into the catalytic chemistry of the Claus process is
obtained from the observation that sulfate and thiosulfate species can be
found on alumina and titania that have been used in Claus service (14). More-
over, as a general rule of thumb, the amount of sulfate found on the catalyst
increases as the temperature of operation decreases (Fig. 18). In the past, it
was generally accepted that sulfate was an artifact of O2 ingress into the
system either as a result of poor combustion in the furnace stage or as a conse-
quence of direct fired reheaters upstream of the converter. However, sulfate-
contaminated alumina is found in plants in which the furnace operates
efficiently and which use indirect (no O2) reheat of the gas. Of course, unless
the main burner is operating so inefficiently as to allow air/O2 to stream
through the furnace and the WHB, one may well ask the question as to how
O2 could possibly survive a 12008C reaction chamber that always contains
excess fuel (H2S and S2).
Vol. 1 SULFUR AND HYDROGEN SULFIDE RECOVERY 885

2 : 1 H 2 S/SO2 , N2 , H 2 O (trace impurities)

Al2 O3 Al2 O3

2 : 1 H 2 S/SO2 (lower concentration)

1st Converter 2nd Converter 3rd Converter

Sulfate (wt%) 1 2 3 4 >4


liquid S8
on catalyst

T (°C, °F) 320, 608 260, 500 220, 428

Conclusion: sulfate is always found on alumina catalysts

Fig. 18. Sulfate on Claus catalysts: Commercial observations.

A mechanistic scheme for the reaction of H2S and SO2 (Fig. 19) reveals
that sulfate forms as part of the catalytic cycle and should reach a steady-state
concentration on the catalyst surface that is dependent on the temperature and
the amount of H2S in the process gas. Laboratory studies confirmed that sulfate

The Claus Reaction

SO 2 + H2O [ H 2SO3] 2H + + SO 3 2–

SO 32– + 1/8 S 8 S2O 32–

S2 O3 2– + SO 2 S3O 52–

S3O5 2– + H 2 S H S4O4 + OH

H S4 O4 + H 2S [ S3 ] + H 2O + H S 2 O3

H S2 O3 + OH H 2O + S 2 O3 2–

OR H S2 O3 + OH SO 4 2– + H 2 S

SO4 2– + H2S H2 O + S 2 O3 2–

S8 Formation ~ S6
2 [ S3 ] S6 Sx
x = 6, 7, 8, ………26

predominant species

Fig. 19. How is sulfur formed on an Al2O3 Claus catalyst?


886 SULFUR AND HYDROGEN SULFIDE RECOVERY Vol. 1

was always produced during the Claus reaction on alumina and titania under
conditions that rigorously excluded O2.

4. Practical Implications of Sulfate Formation on Claus Catalysts

4.1. O2 Carry-over. The most important implication of natural sulfate


relevant to commercial operation of Claus converters is that the presence of
sulfate is not justification alone for change out of the catalyst during a turn-
around. Key factors in making the decision to replace the catalyst are Claus
reaction performance, hydrolysis reaction performance, catalyst surface area
and available pore volume, and the carbon content (from carbon–sulfur polymer,
carsul) of the catalyst.
Although direct-fired reheaters are becoming less common for a variety of
reasons, a valid question is whether O2 scavenging catalysts, designed to prevent
sulfate formation, should be used in Claus converters. An O2 scavenging catalyst is
usually an iron oxide modified alumina that is designed to remove low levels of O2
from the process gas that, most probably, works via the following chemical cycle:

2 FeII þ 2 Hþ þ 1=2 O2 ! 2 FeIII þ H2 O


2 FeIII þ H2 S ! 2 FeII þ 2 Hþ þ 1=8 S8

In addition to the natural sulfate formation discussed earlier in this article,


it is known that O2 will cause excessive and, perhaps, irreversible sulfate
formation. Since iron oxide promoted alumina is an efficient Claus catalyst in
addition to being an O2 scavenger, many plants employing direct fired reheaters
incorporate an O2 scavenging catalyst as the top layer in each converter down-
stream of such a reheater.
4.2. Ratio Control, CS2 and COS Conversion. The overall chemistry
of the catalytic conversion of H2S/SO2 (Fig. 19) is a controlling factor in determin-
ing how much sulfate resides on a Claus catalyst. The amount of sulfate is impor-
tant because it has been determined that although sulfate is an integral part of
the Claus conversion chemistry because it produces thiosulfate (S2O2 3 ), it also
inhibits conversion of carbon disulfide (CS2) and carbonyl sulfide (COS), species
invariably found in Claus process gas. Since efficient catalytic conversion of CS2
and COS is paramount to H2S/SO2 ratio control due to reformation of H2S in the
hydrolysis reactions below, understanding the factors that control sulfate
amounts, hence catalytic activity for CS2 conversion, are relevant to plant control.

CS2 þ 2 H2 O ! 2 H2 S þ CO2
COS þ H2 O ! H2 S þ CO2

Interestingly, just as predicted by the mechanism illustrated in Figure 19,


the amount of sulfate found on a catalyst is dependent on the H2S/SO2 ratio (and,
indeed, H2S quantity) in the process gas (Fig. 20a). Furthermore, there is a direct
relationship between CS2 conversion and ratio or, of course, sulfate amount
(Fig. 20b). These observations have two very important ramifications to
Vol. 1 SULFUR AND HYDROGEN SULFIDE RECOVERY 887

Fig. 20. (a) Average bed weight % SO42 as a function of H2S:SO2 ratio in the sulfating
environment at 3208C. (b) Steady-state CS2 conversion as a function of H2S:SO2 ratio in
the sulfating environment at 3208C.

operation of Claus plants. The first is that plants operated at deliberately high
H2S/SO2 ratios, as is the case for those with direct oxidation or reducing tail
gas systems, will experience superior CS2 conversions (usually > 85%). Second,
it would seem that the practice of catalyst rejuvenations for sulfate removal at
high ratio for plants that normally operate at a standard 2:1 ratio has no
merit since the sulfate level will simply readjust to its natural sulfate level
almost immediately on reestablishment of the working H2S/SO2 ratio. This
conclusion has been justified by observations in the laboratory (13) that reveal
‘‘normal’’ sulfate levels are reestablished within 2 h of restart of a converter at
the new H2S/SO2 ratio. This comment should not be taken to infer that high
ratio H2S/SO2 soaks are of no value as it has been shown that ‘‘carsul’’ can be
removed by this process.
The level of CS2 conversion is then determined by the H2S/SO2 ratio, the
temperature of operation of the converter, and the type of catalyst. As already
discussed, high ratio operation increases conversion of CS2 and COS, but the con-
verter must also be operated at appropriate temperatures and space velocities in
order to achieve optimum CS2 conversion. In practice, a high quality alumina
will enable 90% CS2 and total COS conversion at 3308C and a 1000 h1 space
velocity. If CS2 is anticipated to be a significant problem, the first converter
888 SULFUR AND HYDROGEN SULFIDE RECOVERY Vol. 1

Fig. 21. The CS2 conversion by direct SO2 reaction. Titania more active than alumina.

could be designed to operate at a lower space velocity. The more usual strategy
has been to utilize titania as all or part of the catalyst in the first converter as it
is considerably more active than alumina for CS2 conversion. Although titania is
very efficient in catalyzing the hydrolytic conversion of CS2 and COS, it has been
established that it also promotes CS2 conversion by reaction with SO2. As may
be seen from the data plotted in Figure 21, titania very effectively promotes
the CS2/SO2 conversion reaction, but alumina is, in comparison, only a weak
catalyst for this process (15). Most often, when titania is used as a composite
catalyst bed with alumina, it is used as the lower layer of catalyst in order to
protect the more expensive catalyst from particulate matter entering with the
process gas.
4.3. Catalyst Degradation by BTX. The formation of sulfate as part of
the Claus reaction catalytic cycle is also an important factor in controlling cata-
lyst degradation by BTX. The abbreviation BTX is the acronym given to a collec-
tion of aromatic hydrocarbons that are sometimes found in Claus process gas
(Fig. 22). These hydrocarbons may be present in the acid gas in refineries and

CH 3 CH 3
CH 3

CH 3
CH 3 CH 3 CH 3

B T X
1st Converter Conditions: 320° – 350°C, Al2 O3 catalyst

“cracking” Soot(Carbon)
BTX +
+ S
Carsul (C-S polymer)

Fig. 22. Deactivation of Claus catalysts by aromatics.


Vol. 1 SULFUR AND HYDROGEN SULFIDE RECOVERY 889

Fig. 23. A mechanism for catalyst degradation by BTX. *Immature carsul; substitution
pattern changed for artistic convenience.

in sour gas processing plants, although their presence tends to have a much
greater consequence in the gas plants where CO2 is present in high concentra-
tions. In these cases, low furnace temperatures or some acid gas by-pass to the
first converter results in some BTX entering the catalytic units. The overall pro-
blem of having BTX in the gas is that chemical interaction with sulfur results in
polymeric carbon–sulfur junk (carsul) accumulating in the pore structure of the
catalyst. Xylenes are, by far, the most troublesome aromatic, but toluene also
results in significant degradation of the catalyst. The options of removing BTX
from an acid gas or operating the reaction furnace at a high enough temperature
to completely destroy these aromatics are technically difficult and expensive (16).
Designers are advised to consider measures in the gas treating unit to reduce
BTX pickup.
Sulfate on the catalyst appears to be the driving force for BTX degradation
because the countercation to sulfate on the surface is a proton (Hþ) (it could also
be an aluminum cation) and this Brønsted acidity (Lewis acidity in the case of
an aluminum cation) initiates reactions with the electron-rich aromatics, BTX
(Fig. 23). Since the positively charged aromatic species are surrounded by a
sea of sulfur, rapid C   S bond formation proceeds with production of carsul
that occludes the catalyst surface. Laboratory experiments confirmed that
sulfate was a key player in this chemistry as it was seen that alumina catalyst
that had not been exposed to Claus conditions (no sulfate) did not promote
reaction with xylene or other aromatics (17). Thus, any oxide that forms a stable
sulfate at Claus reactions conditions will promote both H2S/SO2 conversion and
BTX degradation.
4.4. Tail Gas Processing. A standard Claus sulfur recovery system
consists of a furnace and two or three catalytic converters, which, as is evident
for the theoretical sulfur recovery (Fig. 7), limits total sulfur recovery to
95–98%. Environmental considerations have lead to legislated limits that sti-
pulate >99% recovery in most regions of the World, although the total sulfur
recovery required is often linked to the capacity of the plant. Many regions of
the world now require a minimum 99.8% recovery of all sulfur contained in
the SRU feed gas. The regulations have lead to the development of a variety
of ‘‘tail gas’’ treatment options to achieve recoveries > 99% (2).
890 SULFUR AND HYDROGEN SULFIDE RECOVERY Vol. 1

5. The Claus Reaction at Subdew Point Conditions

It is clear from Figure 7 that operation of a converter bed at 1258C would be


advantageous as equilibrium limitations for conversion to sulfur are mini-
mized at these temperatures. Contrary to regular Claus reactors, the subdew
point reactors are operated at these low temperatures where sulfur will con-
dense. As a result, subdew point systems require two or more catalyst beds
in order that removal of the sulfur can be accomplished off-line for one of
the beds. The basic features of the process are that the inlet H2S/SO2 ratio of
2 is maintained as closely as possible as any deviation from this value will
decrease conversion to sulfur (Fig. 24). As discussed earlier, catalysts work
by adsorption of materials on the surface, reaction of the adsorbed species,
and desorption of the products, sulfur, and water in this case. Clearly, water
will still be desorbed to the gas phase at 1258C, but sulfur will accumulate at
the catalyst surface.
The catalyst bed is regenerated by introducing hot process gas (300–3308C)
into the catalyst bed to evaporate the sulfur from the catalyst pore structure
(Fig. 25). At the end of the uptake cycle, the amount of SO2 and other materials
adsorbed on the surface is a function of that temperature. Most probably, more
SO2 can be held on the catalyst surface at the lower temperatures employed in
the uptake mode (125–1308C). Since the process gas used for regeneration is
rerouted to the other converters and the subdew point beds in uptake mode,
SO2 is removed preferentially from the gas stream as the subdew points beds
are switched from uptake to regeneration and vice versa. Indeed, from a practical
viewpoint, a spike in emissions from the incinerator is observed during the
switch-over period. Obviously, such emission spikes can be minimized with
longer uptake/regeneration cycles.

Fig. 24. The chemistry of subdew point Claus tail gas operation.
Vol. 1 SULFUR AND HYDROGEN SULFIDE RECOVERY 891

Feed tail gas Regeneration gas (>300ºC)


(stop flow) (introduce flow)

Increase bed temperature


Catalyst bed from 130 >300ºC
loaded with
liquid sulfur

Condenser Regeneration gas (~150ºC)

liquid S 8

• Key point - exposes active alumina sites to SO2 coverage


at temperature of regeneration gas

Fig. 25. The principle of regeneration of a catalyst bed.

6. Direct Oxidation of H2S

Another major tail gas process for increasing the total sulfur recovery beyond 99%
are technologies based on direct oxidation of H2S using catalytic methods (Fig. 26)
(18). It is very important to remove as much SO2 as possible from the tail gas before
the direct oxidation step as this species is not treated and, consequently, will
contribute to emissions in the final incinerator step. As in the reducing tail gas
technologies, the plant is operated to achieve a high H2S/SO2 ratio (4–8) before
the tail gas enters the direct oxidation unit. This is accomplished by restricting

Fig. 26. Euro-Claus (Jacobs Nederland B.V.).


892 SULFUR AND HYDROGEN SULFIDE RECOVERY Vol. 1

the air supply to the main burner of the Claus furnace and, in an adaptation
recently introduced to commercial operation, by placing a layer of hydrogenation
catalyst at the bottom of the catalyst bed preceding the direct oxidation reactor.
This new strategy then decreases the quantity of SO2 entering the oxidation reac-
tor, possibly allowing a lower amount of SO2 leaving the converter in comparison to
equilibrium prediction. The chemistry of the hydrogenation on the reducing cata-
lyst in the second converter may be similar to that described for the low tempera-
ture reduction described in the previous section.
The direct oxidation catalyst is based on iron species deposited on silica
although various modifiers are used to limit secondary reactions that result in
SO2 formation. Silica is chosen because it is an oxide that does not form a stable
sulfate and, thus, cannot promote the Claus reaction. This facet of catalyst
behavior is important because use of a Claus active support would result in
reconversion of sulfur to H2S and SO2 by the reverse Claus reaction. The
catalytic cycle of the direct oxidation reaction formally involves an Fe(III)/Fe(II)
couple that is cycled by reaction of Fe(II) species with O2 added in controlled
amounts upstream of the direct oxidation reactor. Apparently, the catalyst can
only be kept in an active state if excess O2 is present in the tail gas. However,
this feature of the process imposes a small limitation on overall recovery as sul-
fur vapor produced in the direct oxidation reaction can be converted to SO2 in
homogeneous or surface-catalyzed chemistry. Overall, direct oxidation tail gas
treatment can result in 98.8–99.5% total recovery dependent on the original
acid gas under treatment.

7. Reduction of Tail Gas Sulfur Species to H2S

Another popular tail gas process that enables >99.8% total sulfur recovery is
reduction of all sulfur species to H2S with recycle of H2S to the front end of
the plant. A key feature of the process is that the plant is controlled to result
in a high H2S/SO2 ratio (4–8) to limit the amount of SO2 entering the tail gas
unit. This technology (Fig. 27) was introduced in the late 1960s at a time when
there was little concern about CO2 emissions. Operation of the reducing gas
generator/reheating unit at the front end of the plant, handling sour water
from the quench unit and recycle of the H2S, utilizing an amine unit, all
have associated energy demands that result in considerable CO2 emission.
Note that the amine unit, not shown in Figure 27, processes the gas from the
top of the quench column. The alumina-CoS/MoS2 catalyst employed in the
hydrogenation unit converts all sulfur species to H2S although in the case of
CS2 and COS, it is thought that conversion occurs via hydrolysis reactions
(Fig. 28). In original versions of the process, the H2 and CO produced in the
Claus furnace was supplemented by operation of a partial combustion system
(labeled reheater in Fig. 27) in which CH4 is burnt with a restricted air supply.
This unit also serves to reheat the gas to  3408C, the temperature required for
operation of the catalytic unit.
In a drive to make the process more efficient, low temperature catalysts
have been introduced that operate at  2408C (19). This innovation has the
very important effect of replacement of the partial oxidation system with an
Vol. 1 SULFUR AND HYDROGEN SULFIDE RECOVERY 893

Indirect re-heater

Hydrogenation Quench
reactor column

> 300 240ºC


steam

Gas cooler

Claus tail gas

condensate
to SWS
∗ Catalyst active at lower temperatures
Fig. 27. Low temperature tail gas reduction.

Fig. 28. Low temperature reducing tail gas chemistry.

indirect reheating unit as supplemental H2 is not required in most plants.


All versions of this tail gas process require the use of selective tertiary amine
systems to reject CO2 from the treated gas and so avoid build-up of CO2 in
the plant.

8. Tail Gas Treatment in an Organic Solvent

The Claus reaction of H2S and SO2 can be carried out in liquid solvent, eg, poly-
ethylene glycols, when a basic catalyst in present is the medium (2). The only
894 SULFUR AND HYDROGEN SULFIDE RECOVERY Vol. 1

process currently in operation was introduced by Institute Francois Petrol


and, although used quite widely in Europe, it has not become popular in North
America or elsewhere. The chemistry of this technology is simply the Claus reac-
tion, so overall efficiency depends on maintaining a 2:1 H2S/SO2 ratio throughout
the system. The liquid sulfur separates at the bottom of the reactor and is drawn
off from the solvent. Periodic cleaning of the solvent is required to remove inor-
ganic salts that build up in the glycol solution. Given the ratio control constraints
and vapor losses from liquid sulfur handling, total sulfur recovery from plants
employing the IFP process is in the range of 99–99.5%.

9. Recovery of Sulfur Values as SO2

Oxidation of Claus tail gas converts all sulfur compounds to SO2. In many
cases, this simply represents the tail gas incineration step before final release
of the tail gas to the atmosphere, but it is possible to capture that SO2 in
various forms (2). Such technologies are not used widely, but may be applied
more extensively in the future. One adaptation involves conversion of the
SO2 to useful inorganic derivative taking the SO2 directly from the tail gas
while another utilizes a diamine solvent to capture the SO2 for recycle to the
front end of the Claus plant. These technologies may be limited by the markets
for the inorganic derivatives in the former case and by issues of economics and
process consequences of returning large volumes of SO2 to the furnace in the
diamine technology.

10. Small Scale Recovery of H2S and Sulfur

A growing number of processes are being introduced to handle gas streams


that produce < 10 ton/day of sulfur in response to increased environmental
awareness (2). The only process applied on a wide scale is based on aqueous
iron systems that hold iron in solution using a sophisticated array of chelating
compounds. The iron species oxidize H2S to elemental sulfur and the iron is
reoxidized by using air in chemistry related to that described in previous sec-
tions for O2 scavenging and the direct tail gas oxidation. Sulfur is usually pro-
duced as a water-wet filter cake and is disposed of in landfill sites. Newer
methods based on gas phase or subdew point catalytic oxidation have been
introduced in Canada. This technology is similar to the direct oxidation tail
gas process, but the catalyst does not require excess air for continuous opera-
tion. Both the aqueous and the catalytic processes can be applied to H2S
removal from natural gas streams as well as to acid gases separated from
the hydrocarbon feed.

11. Acknowledgments

The author would like to thank his colleagues at ASRL, particularly, Dr. Norman
Dowling and Dr. Minming Huang, for support over many years. Also thanks
Vol. 1 SULFUR AND HYDROGEN SULFIDE RECOVERY 895

are due to the reviewers of this article for their improvements to the original
manuscript.

BIBLIOGRAPHY

‘‘Sulfur Removal’’ in ECT 3rd ed., Vol. 22, pp. 267–292, by J. R. West, Texagulf Inc.;
‘‘Sulfur Removal and Recovery’’ in ECT 4th ed., Vol. 23, pp. 432–452, by M. Capone,
Exxon Engineering; ‘‘Sulfur Removal and Recovery’’ in ECT (online), posting date:
December 4, 2000, by M. Capone, Exxon Engineering.

CITED REFERENCES

1. M. Kitto, Sulphur Markets, Proceedings of Sulphur 2004, Barcelona, Spain, October


2004, Published by British Sulphur Publishing, London, U.K.
2. R. N. Maddox and D. J. Morgan, Gas Conditioning and Processing, Vol. 4, Gas Treat-
ing and Sulfur Recovery, Chapt. 3, Campbell Petroleum Series, Norman, Oklahoma,
published January 1998.
3. J. L. Harbison and G. E. Handwerk, Selective Removal of Hydrogen Sulfide Utilizing
Generic MDEA, Laurance Reid Gas Conditioning Conference, University of Okla-
homa, March 2–4, 1987, published by Centre for Continuing Education, University
of Oklahoma.
4. Y. F. Jou, F. D. Otto, and A. E. Mather, Acid Gas Mixture Data for MDEA, Canadian
Gas Processors Association, 4th Quarterly Meeting, Calgary, Alberta, 1986.
5. R. N. Maddox and D. J. Morgan, Gas Conditioning and Processing, Vol. 4, Gas Treat-
ing and Sulfur Recovery, Chapt. 6, Campbell Petroleum Series, Norman, Oklahoma,
published January 1998.
6. A. A. Adeyiga, G. J. Mains, and R. N. Maddox, Vapor–Liquid Equilibrium Coeffi-
cients for Acid Gas Components and Physical Solvents, Proceedings of the Canadian
Chemical Engineering Conference, Calgary, Canada, 1980.
7. P. D. Clark and co-workers, Main Group Chem. News, 7(2), (1998).
8. G. W. Gamson and R. H. Elkins, Chem. Eng. Prog. 49, 203 (1953).
9. I. A. Gargurevich, Ind. Eng. Chem. Res. 44, 7706 (2005) and references cited therein.
10. P. D. Clark, M. Huang, and N. I. Dowling, Chemistry of the Front-End Reaction Fur-
nace: Hydrocarbon Reactions and Formation of Carbon Disulfide, Proceedings of the
Brimstone Engineering Sulfur Recovery Workshop, Vail, Colorado, September 1997.
11. P. D. Clark and co-workers, Ind. Chem. Eng. Res. 40, 497 (2001).
12. P. D. Clark, N. I. Dowling, and M. Huang, Alberta Sulphur Res. Quart. Bull. 34(4)
(1998). (Held in the University of Calgary Library, Calgary, Alberta, Canada.)
13. P. D. Clark, N. I. Dowling, M. Huang, and W. S. Kijlstra, How do Claus Catalysts
Really Work, Proceedings of the Laurance Reid Gas Conditioning Conference,
February, 2002; published by the Centre for Continuing Education, University of
Oklahoma.
14. P. D. Clark, M. Huang, N. I. Dowling, and W. S. Kijlstra, Appl. Catal. A, Gen. 235, 61
(2002).
15. P. D. Clark, M. Huang, and N. I. Dowling, Appl. Catal. B, Environ. 31, 207 (2001).
16. P. Crevier, Options for Removal of BTX From Claus Acid Gas, Proceedings of Sour
Oil and Gas Advanced Technology, Doha, Qatar, November 2004.
17. P. Crevier, P. D. Clark, M. Huang, and N. I. Dowling, Alumina Catalysts: The Good,
The Bad and The Ugly: An Evaluation of Materials in the Marketplace in the
896 CORROSION AND CORROSION CONTROL Vol. 1

Presence of Aromatic Compounds, Proceedings of the Brimstone Engineering Sulfur


Recovery Workshop, Vail, Colorado, September 2002.
18. H. Borsboom, Recent Advances in the SuperClaus and EuroClaus Processes, Pro-
ceedings of the Brimstone Engineering Sulfur Recovery Workshop, Vail, Colorado,
September 2004.
19. M. van Son, G. Bloemendal, and J.-L. Ray, Low Temperature Scot Catalysis and
Operation, Proceedings of the Brimstone Engineering Sulfur recovery Workshop,
Vail, Colorado, September 2005.

P. D. CLARK
University of Calgary

CORROSION AND
CORROSION CONTROL
1. Introduction

Corrosion is the natural degradation of materials in the environment through


electrochemical or chemical reaction. Traditionally, the definition of corrosion
refers to the degradation of metals and has not included the degradation of non-
metals such as wood (rotting) or plastics (swelling or crazing), but increasingly,
natural degradation of any engineering material is being regarded as corrosion.
The vast majority of the technologically significant corrosion involves the
deterioration of metallic materials, and only the corrosion of metallic materials
is discussed here.

2. Electrochemical Nature of Corrosion

Ores are mined and are then refined in an energy intensive process to produce
pure metals, which in turn are combined to make alloys. Corrosion occurs
because of the tendency of these refined materials to return to a more thermody-
namically stable state (1–3). The key reaction in corrosion is the oxidation or ano-
dic dissolution of the metal to produce metal ions and electrons

M k!
1
Mnþ þ n e ð1Þ

The ions, Mnþ, formed by this reaction at a rate, k1, may be carried into a bulk
solution in contact with the metal, or may form insoluble salts or oxides. In order
for this anodic reaction to proceed, a second reaction which uses the electrons
produced, ie, a reduction reaction, must take place. This second reaction, the

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.0315181813151801.a01.pub2
Vol. 1 CORROSION AND CORROSION CONTROL 897

cathodic reaction, occurs at the same rate because the electrons produced by the
anodic reaction must be consumed by the cathodic reaction to maintain electro-
neutrality. Therefore, Ic ¼ Ia, where Ic and Ia are the cathodic and anodic cur-
rents, respectively. The cathodic reaction, in most cases, is hydrogen evolution
or oxygen reduction.
The four elements necessary for corrosion are an aggressive environ-
ment, an anodic and a cathodic reaction, and an electron conducting path
between the anode and the cathode. Other factors such as a mechanical
stress also play a role. The thermodynamic and kinetic aspects of corrosion
determine, respectively, if corrosion can occur, and the rate at which it does
occur.

3. Manifestations of Corrosion

The most common form of corrosion is uniform corrosion, in which the entire
metal surface degrades at a near uniform rate (1 – 4). Often the surface is
covered by the corrosion products. The rusting of iron (qv) in a humid atmo-
sphere or the tarnishing of copper (qv) or silver alloys in sulfur-containing
environments are examples. High temperature, or dry, oxidation, is also
usually uniform in character. Uniform corrosion, the most visible form of cor-
rosion, is the least insidious because the weight lost by metal dissolution can
be monitored and predicted.
An especially insidious type of corrosion is localized corrosion (1–3,5,6),
which occurs at distinct sites on the surface of a metal while the remainder
of the metal is either not attacked or attacked much more slowly. Localized
corrosion is usually seen on metals that are passivated, ie, protected from corro-
sion by oxide films, and occurs as a result of the breakdown of the oxide film.
Generally the oxide film breakdown requires the presence of an aggressive
anion, the most common of which is chloride. Localized corrosion can cause con-
siderable damage to a metal structure without the metal exhibiting any appreci-
able loss in weight. Localized corrosion occurs on a number of technologically
important materials such as stainless steels, nickel-base alloys, aluminum, tita-
nium, and copper.
Two types of localized corrosion are pitting and crevice corrosion. Pitting
corrosion occurs on exposed metal surfaces, whereas crevice corrosion occurs
within occluded areas on the surfaces of metals such as the areas under rivets
or gaskets, or beneath silt or dirt deposits. Crevice corrosion is usually associated
with stagnant conditions within the crevices. A common example of pitting cor-
rosion is evident on household storm window frames made from aluminum
alloys.
Another type of corrosion is dealloying which has also been called parting
or selective leaching. Dealloying (1–4) is the preferential removal of one of
the alloying elements from an alloy resulting in the enrichment of the other
alloying element(s). Common examples are the loss of zinc from brasses (dezin-
cification) and the loss of iron from cast irons (graphitization).
Corrosion may also appear in the form of intergranular attack, ie, preferen-
tial attack of the boundaries between the crystals (grains) in metals and alloys
898 CORROSION AND CORROSION CONTROL Vol. 1

(1–5). Intergranular attack generally occurs because the grain boundary and the
grain have different corrosion tendencies, ie, different potentials. Intergranular
corrosion often leads to a loss in strength or ductility of the metal.
Corrosion also occurs as a result of the conjoint action of physical processes
and chemical or electrochemical reactions (1–5). The specific manifestation of
corrosion is determined by the physical processes involved. Environmentally
induced cracking (EIC) is the failure of a metal in a corrosive environment and
under a mechanical stress. The observed cracking and subsequent failure would
not occur from either the mechanical stress or the corrosive environment alone.
Specific chemical agents cause particular metals to undergo EIC, and mechanical
failure occurs below the normal strength (yield stress) of the metal. Examples are
the failure of brasses in ammonia environments and stainless steels in chloride
or caustic environments.
When a stress is cyclic rather than constant, the failure is termed corrosion
fatigue. Fretting corrosion results from the relative motion of two bodies in con-
tact, one or both being a metal. The motion is small such as a vibration. Erosion
corrosion results from the action of a high velocity fluid impinging on a metal
surface. Metals and alloys can also experience cracking in liquid metal environ-
ments. This form of corrosion is referred to as liquid metal cracking (LMC)
(1,2,7). For example, mercury promotes cracking in highly stressed copper alloys
and high strength aluminum alloys, and liquid copper promotes cracking of
steels and stainless steels. Titanium and nickel alloys are also susceptible to
LMC in specific environments.
Galvanic corrosion (1–4) occurs as a result of the electrical contact of differ-
ent metals in an aggressive environment. The driving force is the electrode
potential difference between the two metals. One metal acts principally as a
cathode and the other metal as the anode. Galvanic corrosion can result from
the presence of a second phase in a metal. An example is manganese sulfide
[18820-29-6], MnS, inclusions in steels. Galvanic corrosion is also an important
consideration for the environmental stability of metal matrix composites (qv)
such as graphite reinforced aluminum. Galvanic corrosion can accelerate many
of the other types of corrosion.
Microbiologically influenced corrosion results from the interaction of micro-
organisms and a metal (3,8–11). The action of microorganisms is at least one of
the reasons why natural seawater is more corrosive than either artificial sea-
water or sodium chloride solutions. Microorganisms attach to the surfaces of
metals and can, for example, act as diffusion barriers; produce metabolites
that enhance or initiate corrosion; act as sinks or sources for species involved
in cathodic reactions, such as oxygen and hydrogen; increase the pH at the sur-
face as a result of photosynthesis; or decrease the pH by production of acid meta-
bolites. A more detailed discussion of the various forms of corrosion may be found
in the literature (1–4).

4. Origin of Corrosion

In the presence of oxygen and water the oxides of most metals are more thermo-
dynamically stable than the elemental form of the metal. Therefore, with the
Vol. 1 CORROSION AND CORROSION CONTROL 899

Table 1. Standard Electrode (Reduction)


Potentials, 258Ca
Electrode E0 , Vb
Liþ þ e ! Li 3:045
Zn2þ þ 2e ! Zn 0:763
Fe2þ þ 2e ! Fe 0:440
Cd2þ þ 2e ! Cd 0:403
Co2þ þ 2e ! Co 0:277
Ni2þ þ 2e ! Ni 0:250
Sn2þ þ 2e ! Sn 0:140
Pb2þ þ 2e ! Pb 0:126
Sn4þ þ 2e ! Sn2þ 0:15
Cu2þ þ e ! Cuþ 0:158
Cu2þ þ 2e ! Cu 0:337
Fe3þ þ e ! Fe2þ 0:771
Agþ þ e ! Ag 0:799
Br2 ðlÞ þ 2e ! 2Br 1:065
Cl2 ðgÞ þ 2e ! 2Cl 1:358
a
Ref. 3.
b
Against standard hydrogen electrode (SHE) set at
E0 ¼ 0:000V

exception of gold, the only metal that is thermodynamically stable in the


presence of oxygen, there is always a thermodynamic driving force for corro-
sion of metals. Most metals, however, exhibit some tendency to passivate, ie,
to form a protective oxide film on the surface that retards further corrosion.
Were it not for the passivation by oxide films, most metals and alloys would
corrode rapidly.
The thermodynamics of electrochemical reactions can be understood by con-
sidering the standard electrode potential, E0, the potential of a reaction under
standard conditions of temperature and pressure where all reactants and pro-
ducts are at unit activity. Table 1 lists a variety of standard electrode potentials.
The standard potential is expressed relative to the standard hydrogen reference
electrode potential in units of volts. A given reaction tends to proceed in the ano-
dic direction, ie, toward the oxidation reaction, if the potential of the reaction is
positive with respect to the standard potential. Conversely, a movement of the
potential in the negative direction away from the standard potential encourages
a cathodic or reduction reaction.
A piece of zinc placed in an aqueous 1 M Zn2þ solution attains a potential of
approximately  0.763 V vs standard hydrogen. If Cu2þ ions are then introduced
into the solution, the zinc begins to corrode because the zinc surface exists at a
potential which encourages the cathodic reaction for copper, ie, the standard
potential of the zinc reaction is negative to that of the copper and the Cu ions
electroplate onto the zinc via the reaction Cu2þ þ 2 e ! Cu. This coupling causes
the zinc potential to shift, ie, to be polarized, in the positive direction and the
reaction Zn ! Zn 2þ þ 2 e then proceeds. The potentials of the two reactions
are polarized toward each other, away from the respective standard potentials.
900 CORROSION AND CORROSION CONTROL Vol. 1

The rates of these two reactions must be the same to preserve electroneutrality,
and additionally, some of the zinc surface must be exposed to the solution. That
is, if a continuous coating of copper were to be electroplated on the zinc surface
such that none of the zinc were exposed, further oxidation would be prevented.
Corrosion occurs even if the two reactants involved are not at standard con-
ditions. In this case, the nonstandard equilibrium potential for each reaction,
often referred to as the reversible potential, can be calculated from the Nernst
equation. Additional information on thermodynamic aspects of corrosion can be
found in the literature (1–4,12,13).
The larger the potential difference between the equilibrium potentials of
the anodic and cathodic reactions involved in corrosion, the greater the driving
force which is more traditionally represented as the change in Gibbs free energy,
DG. The free-energy change for a reaction is related to the difference in the equi-
librium potentials of the two reactions, cathodic minus anodic by

G ¼ nFE ð2Þ

where n is the number of electrons involved in the overall reaction, F is the Fara-
day constant, and E, the cell voltage or the difference in the equilibrium potential
values of the cathode and the anode, respectively, given in volts.
The standard reduction potential of oxygen (taken at pH = 0)

O2 þ 4 Hþ þ 4 e ! 2 H2 O ð3Þ

is 1.229 V, a much more positive number than the metal reactions given in Table 1.
Therefore, the free-energy change for a coupling of any of the metal reactions,
considered as the anode, and the oxygen reaction, considered as the cathode,
would be a large negative number; ie, the reaction is thermodynamically highly
favored to proceed. Reduction of oxygen is the most common cathodic reaction
in the corrosion of metals in neutral and alkaline media. In acidic media the
common cathodic reaction is hydrogen evolution

2 Hþ þ 2 e ! H2 ð4Þ

The standard potential for this reaction is 0.000 V. Most of the metal reac-
tions have standard potentials below this value.

5. Electrochemical Equilibrium Diagrams

The thermodynamic data pertinent to the corrosion of metals in aqueous media


have been systematically assembled in a form that has become known as
Pourbaix diagrams (13). The data include the potential and pH dependence
of metal, metal oxide, and metal hydroxide reactions and, in some cases, complex
ions. The potential and pH dependence of the hydrogen and oxygen reactions
are also supplied because these are the common corrosion cathodic reactions
as discussed above. The Pourbaix diagram for the iron–water system is given
as Figure 1.
Vol. 1 CORROSION AND CORROSION CONTROL 901

1.6 0
–2
Fe3+ –4 2 –?
(B) –6 FeO4
FeOH2+

0.8

Fe (OH)2+
E, V

(A) Fe (OH)3
0 Fe2+

Fe (OH)2

– 0.8

Fe HFeO2–

–1.6

0 2 4 6 8 10 12 14 16
pH

Fig. 1. Pourbaix diagram for the iron–water system at 258C, considering as solid
substances only Fe, Fe(OH)2, and Fe(OH)3, where (), line A, represents the hydrogen
reaction line; (  ), line B, the oxygen line(13). The values 0,2, 4, 6 are molar
concentrations of the ions involved in each reaction.

If the potential of a metal surface is moved below line A, the hydrogen reac-
tion line, cathodic hydrogen evolution is favored on the surface. Similarly a
potential below line B, the oxygen reaction line, favors the cathodic oxygen
reduction reaction. A potential above the oxygen reaction line favors oxygen evo-
lution by the anodic oxidation of water. In between these two lines is the region
where water is thermodynamically stable.
The Pourbaix diagram is constructed from data for a variety of reactions.
The expressions for these reactions can be (1) potential dependent and pH inde-
pendent, (2) pH dependent and potential independent, (3) both pH and potential
dependent, or (4) both pH and potential independent. The dependence is deter-
mined by whether the reaction of interest is electrochemical or chemical and
whether or not it involves the components of water. From this diagram it is pos-
sible to determine if there is a driving force for corrosion at any given potential–
pH combination and what the corresponding cathodic reaction is likely to be.
Regions where insoluble corrosion products form are indicated. These are
regions where passivity is possible. Passivity generally results from the forma-
tion of an oxide film on a metal surface. The film can dramatically reduce
further corrosion of the metal. Regions of immunity, where there is no thermo-
dynamic driving force for the metal to corrode, can also be determined. Figure 2
presents a simplified picture of the Pourbaix diagram of Figure 1 showing the
902 CORROSION AND CORROSION CONTROL Vol. 1

1.6

0.8

E,V
Passivation
0
Corrosion

– 0.8

Corrosion
Immunity

–1.6
0 4 8 12 16
pH

Fig. 2. The thermodynamic regions of corrosion, immunity, and passivation of iron in an


iron–water system assuming passivation by a film of Fe2O3(13).

regions of immunity, possible corrosion, and possible passivation. A complete


collection of electrochemical equilibrium diagrams can be ordered from NACE
International (13).
Pourbaix diagrams are only thermodynamic predictions and yield no infor-
mation about the kinetics of the reactions involved nor are the influences of other
ionic species which may be present in the solution included. Complexing ions,
particularly halides, can interfere with passivation and can influence the posi-
tion of the lines in a Pourbaix diagram. Further, in real corrosion situations
the local chemistry in pits, crevices, and cracks differs from that of the bulk con-
centration and influences the actual electrochemical reactions at those sites.
More comprehensive information on Pourbaix diagrams and the use of them in
corrosion is available (1–3,12,13).

6. Kinetics of Electrochemical Reactions

Even in uniform corrosion, a corroding metal surface has numerous local anodes
and cathodes. The sites of these local reactions may be fixed by microstructural
features or may change as corrosion proceeds. The oxidation reaction at anodic
sites on the metal surface can be represented as in equation 1. A corresponding
reduction reaction must be occurring at cathodic sites. In acidic solutions this
would likely be hydrogen evolution as shown in equation 4. The potentials of
these two reactions would be moved toward each other, away from the respective
equilibrium potentials, and the metal surface would assume an overall uniform
potential. The dotted lines in Figure 3 illustrate the relationships of the indivi-
dual electrochemical reaction kinetics as influenced by electrochemical potential.
Vol. 1 CORROSION AND CORROSION CONTROL 903

e)

A
2
+ +
H
2
(H 2

Anodic oxidation curve


2+
Electrode potential, V (M M + 2 e– )
C

iapplied = ired – iox Cathodic reduction curve


+
(2 H + 2 e– H2)
ioxidation ireduction
(M
Immunity 2+
+
region 2
e

M
)

10 100 1000
Log(current density)

Fig. 3. Hypothetical Evans diagram and polarization curve for a metal corroding in
an acidic solution, where point A represents the exchange current density, i0, for the
hydrogen electrode at equilibrium; point B, the exchange current density at the reversible

or equilibrium potential, Erev, for M !  M þ 2e ; and point C, the corrosion current
density, icorr, at the open-circuit corrosion potential, Ecorr. See also discussion in text.

These lines also illustrate the interaction of the two reactions on the metal sur-
face. Diagrams of this sort are referred to as Evans or mixed potential diagrams.
The terms are used interchangeable in the corrosion community.
The two dashed lines in the upper left hand corner of the Evans diagram
represent the electrochemical potential vs electrochemical reaction rate
(expressed as current density, ie, current divided by surface area) for the oxida-
tion and the reduction forms of the hydrogen reaction. At point A the two are
equal, ie, at equilibrium, and the potential is therefore the equilibrium potential,
E0, for the hydrogen reaction for the specific conditions involved. Note that the
reaction kinetics are linear on these axes. The change in potential for each dec-
ade of log current density is referred to as the Tafel slope (14). The Tafel slope is
related to the activation energy for the reaction. A more detailed treatment of
Tafel slopes can be found elsewhere (1–4,14,15).
The value of the current density where the oxidation and reduction forms of
a reaction cross, ie, at point A or point B in Figure 3, is referred to as the
exchange current density, i0. For any equilibrium point, the exchange current
density is the rate of the two forms of the reaction, ie, cathodic and anodic, occur-
ring at equilibrium. The greater the exchange current density the more kineti-
cally favored is the reaction. The exchange current density for the reduction of
hydrogen on platinized platinum is 103A/cm2, whereas on mercury it is
1013A/cm2, indicating that platinum is a good catalyst for hydrogen evolution.
The exchange current densities for most electrochemical reactions fall in
between these two values. Exchange current density and Tafel slope values for
904 CORROSION AND CORROSION CONTROL Vol. 1

a variety of other electrochemical reactions of interest to corrosion can be found


in the literature (2,3,16). The example in Figure 3 involves hydrogen evolution on
a corroding surface and the exchange current density value for the hydrogen
reaction on that surface must then be applied.
In the lower left part of the Evans diagram, the oxidation and reduction
forms of metal dissolution reactions are given. Assuming the data of this diagram
are accurate, a prediction of corrosion rate and the potential of the corroding
metal surface can be made. The potentials of the hydrogen and the metal reac-
tions are pulled toward one another following the dashed lines, or Tafel slope,
appropriate for the particular polarization; ie, the anodic line is followed for
the metal dissolution reaction, the cathodic line followed for the hydrogen
reaction. Point C where the two cross is the only potential where the rates of
the anodic and the cathodic reactions are equal. This defines the corrosion
potential, Ecorr, and the corrosion current density, icorr, from which the corro-
sion rate can be calculated by Faraday’s law. The other dotted lines were
not considered because the rates at the corrosion potential are so low as to
be insignificant.
The solid line in Figure 3 represents the potential vs the measured (or the
applied) current density. Measured or applied current is the current actually
measured in an external circuit; ie, the amount of external current that must
be applied to the electrode in order to move the potential to each desired point.
The corrosion potential and corrosion current density can also be determined
from the potential versus measured current behavior, which is referred to as
polarization curve rather than an Evans’ diagram, by extrapolation of either or
both the anodic or cathodic portion of the curve. This latter procedure does not
require specific knowledge of the equilibrium potentials, exchange current den-
sities, and Tafel slope values of the specific reactions involved. Thus Evans
diagrams, constructed from information contained in the literature, and polari-
zation curves, generated by experimentation, can be used to predict and analyze
uniform and other forms of corrosion. Further treatment of these subjects can be
found elsewhere (1–4,17).

7. Galvanic Corrosion and Cathodic Protection

Galvanic corrosion results when two or more dissimilar metals or alloys


immersed in the same electrolyte are in electrical contact (1–4). This form of cor-
rosion is one of the most common and most preventable. The metal or alloy in a
galvanic couple having the lower corrosion potential has its potential pulled in
the positive direction by the coupling with the metal or alloy that has the higher
corrosion potential, generally causing the one with the lower potential to experi-
ence accelerated corrosion. Conversely, the metal with the higher corrosion
potential experiences a negative shift in potential as a result of the coupling
causing it to corrode less and to support additional cathodic reaction. Figure 4
is a galvanic series for a variety of metals and alloys listing the corrosion poten-
tials of these materials in seawater. Free corrosion potentials, not coupled or
equilibrium potentials, are given and these allow a determination of which
metal has the lower potential in a galvanic couple. Generally, the larger the
Vol. 1 CORROSION AND CORROSION CONTROL 905

E vs saturated calomel half-cell reference electrode, V


+0.2 0 –0.2 – 0.4 – 0.6 –0.8 –1.0 –1.2 –1.4 –1.6

Magnesium
Zinc
Beryllium
Aluminum alloys
Cadmium
Mild steel cast iron
Low alloy steel
Austenitic nickel cast iron
Aluminum bronze
Naval brass, yellow brass, red brass
Tin
Copper
Pb–Sn solder (50/50)
Admiralty brass, aluminum brass
Manganese bronze
Silicon bronze
Tin bronzes (G & M)
Stainless steel-types 410, 416
Nickel silver
90 –10 Copper–nickel
80–20 Copper–nickel
Stainless steel-types 430
Lead
70 –30 Copper–nickel
Nickel–aluminum bronze
Nickel–chromium alloy 600
Silver bronze alloys
Nickel 200
Silver
Stainless steel-types 302, 304, 321, 347
Nickel–copper alloys 400,K-500
Stainless steel-types 316, 317
Alloy 2C stainless steels cast and wrought
Nickel–iron–chromium alloy 825
Ni–Cr–Mo–Cu–Si alloy C
Titanium
Ni–Cr–Mo alloy C
Platinum
Graphite

Fig. 4. Galvanic series in flowing seawater. Certain alloys may become more active in
low velocity or poorly aerated seawater and the potentials exhibited under these condi-
tions are indicated as darkened potential ranges (1).

potential difference between components in a galvanic couple the faster will


be the corrosion of the one having the lower potential. However, the kinetics
of the particular reactions involved must be evaluated for an accurate deter-
mination of corrosion rate.
Galvanic corrosion can be used to a corrosion advantage. If, eg, a metal such
as zinc or magnesium, having a low corrosion potential in most environments, is
coupled with a steel component, the zinc or magnesium pulls the potential of the
steel down causing the steel to corrode less. When a sacrificial metal or alloy,
called a sacrificial anode, is attached to a structure having a higher corrosion
906 CORROSION AND CORROSION CONTROL Vol. 1

potential to intentionally pull the potential of the higher potential metal down
and thus decrease the corrosion rate, it is called cathodic protection. This method
of corrosion mitigation is common for underground pipelines (qv), residential hot
water heaters, and hulls and tanks of ships. The lowering of potential can also
be achieved by the application of external electrical current, ie, impressed cur-
rent cathodic protection. Electrons are pumped into the structure to lower the
potential and thus discourage anodic (corrosion) reactions. A second electrode
(anode) is needed for this procedure and this electrode is often a highly inert
material such as platinum to prevent it from being corroded away. However,
scrap iron or steel is sometimes used as the impressed current anode. This
approach remains functional until the scrap metal anode corrodes away.
The lower potential metal in a galvanic couple does not always have its cor-
rosion rate accelerated. For metals that form a passive film, coupling with
another metal of higher potential can cause the film-forming metal’s potential
to shift from a value at which it corrodes to one at which it passivates and there-
fore corrodes less. When this is done intentionally the procedure is referred to as
anodic protection, ie, achieving protection by intentionally shifting the potential
in the positive direction. Anodic protection is generally achieved by adding oxidi-
zers to the electrolyte or by an external electrical circuit.
The higher potential metal in a galvanic couple is not always rendered less
corrodible. A passive metal can sometimes be pulled out of its passive region and
into a more corrosive region. A metal can also have its potential pulled down to
the point where hydrogen reaction is possible and this can lead to one of the
forms of EIC called hydrogen induced cracking (HIC). Another example involves
organic coatings on metal structures. These coatings can be damaged by having
the potential of the structure shifted too far in the negative direction. This is one
of the consequences of excessive cathodic protection.

8. Environmental Effects

The environment plays several roles in corrosion. It acts to complete the electri-
cal circuit, ie, supplies the ionic conduction path; provides reactants for the
cathodic process; removes soluble reaction products from the metal surface;
and/or destabilizes or breaks down protective reaction products such as oxide
films that are formed on the metal. Some important environmental factors
include: the oxygen concentration; the pH of the electrolyte; the temperature;
and the concentration of anions.
Reduction of oxygen is one of the predominant cathodic reactions contribut-
ing to corrosion. Awareness of the importance of the role of oxygen was developed
in the 1920s (18). In classical liquid drop experiments, the corrosion of iron or
steel by drops of electrolytes was shown to depend on electrochemical action
between the central relatively unaerated area, which becomes anodic and suffers
attack, and the peripheral aerated portion, which becomes cathodic and remains
unattacked. In 1945, the linear relationship between rate of iron corrosion and
oxygen pressure from 0 to 2.5 MPa (0–25 atm) was shown (19).
The concentration dependence of iron corrosion in potassium chloride
[7447-40-7], sodium chloride [7647-14-5], andlithium chloride [7447-44-8]
Vol. 1 CORROSION AND CORROSION CONTROL 907

16.66

13.33
Loss in weight, mg

10.00

6.66

3.33
0 0.5 1.0 1.5 2.0
Normality of solution

Fig. 5. Corrosion–concentration curves for alkali chlorides where  denotes LiCl; &,
NaCl; and ~, KCl (20).

solutions is shown in Figure 5 (20). In all three cases there is a maximum in cor-
rosion rate. For NaCl this maximum is at 0.5 N (3 wt%). Oxygen solubility
decreases with increasing salt concentration, thus the lower corrosion rate at
higher salt concentrations. The initial increase in the iron corrosion rate is
related to the action of the chloride ion in concert with oxygen. The corrosion
rate of iron reaches a maximum at 708C. As for salt concentration, the
increased rate of chemical reaction achieved with increased temperature is
balanced by a decrease in oxygen solubility.
The corrosion rate of iron in aerated water is also a function of pH and gen-
erally follows the pattern in Figure 6 (21). At pH 4–10 , the rate is controlled by
the availability of oxygen. In more acidic solutions (lower pH) the corrosion rate
is accelerated, and the reduction of hydrogen ion replaces the reduction of oxygen

1.00

0.75
Corrosion rate, mm/yr

0.50

A +
+
+
0.25 + + + +
+++
+
+ +
+
14 13 12 11 10 9 8 7 6 5 4 3 2
pH

Fig. 6. Effect of pH on corrosion of iron in aerated water at room temperature (21). Point A
is where H2 evolution begins. To convert mm/yr to mils per year (mpy), multiply by 39.37.
908 CORROSION AND CORROSION CONTROL Vol. 1

Table 2. Composition of Green Patina on Copper from Different Locationsa


Composition of green patina, %
Location of Age of
structure structure, year CuCO3 CuCl2 Cu(OH)2 CuSO4
urban 30 14.6 9.6 49.8
rural 300 1.4 58.5 25.6
marine 13 12.8 26.7 52.5 2.5
urban–marine 38 4.6 61.5 29.7
a
Refs. 26 and 27.

as the rate controlling cathodic reaction. Many metals follow approximately the
same behavior with the exception of those metals that dissolve to form ampho-
teric ions. Zinc forms the zincate ion, ZnO2
2 , which causes zinc to corrode exces-
sively above a pH 12; whereas Al forms the aluminate ion [11098-82-1], AlO2,
which increases the dissolution rate above a pH of 8.
Chlorides, which are ubiquitous in nature, play an important role in the
corrosion of metals. Chlorides and other anions also play an important role in
localized corrosion, ie, the breakdown of the insoluble protective reaction product
films, eg, passive films, that prevent corrosion of the underlying metal. A variety
of mechanisms attempting to explain the role of chloride in general and in loca-
lized corrosion have been proposed (6,22–25).
Very often the environment is reflected in the composition of corrosion
products, eg, the composition of the green patina formed on copper roofs over a
period of years. The determination of the chemical composition of this green
patina was one of the first systematic corrosion studies ever made (see COPPER).
The composition varied considerably depending on the location of the structure
as shown in Table 2 (26,27).

9. Alloy Composition and Metallurgical Factors

A primary factor in determining corrosion behavior of metals and alloys is their


chemical composition. Alloys having varying degrees of corrosion resistance have
been developed in response to environmental needs. A good example of how cor-
rosion resistance can be successfully changed by altering the composition can be
seen in the alloying of steels. At the lower end of the alloying scale are the less
costly and less corrosion resistant low alloy steels. These are iron-base alloys con-
taining from 0.5–3.0 wt% Ni, Cr, Mo, or Cu and generally small amounts of P, N,
and S. At the higher end of the alloying scale are the more costly and signifi-
cantly more corrosion resistant stainless steels. These latter alloys contain a
minimum of 10.5 weight percent Cr and, depending on the grade, other alloying
elements such as Ni, Mo, and N. The corrosion resistance of these alloys is based
on the protective nature of the surface film, which in turn is based on the physi-
cal and chemical properties of the oxide film.
In addition to alloying there are other metallurgical factors, such as crystal-
lography, grain size and shape, grain heterogeneity, second phases, impurity
inclusions, and residual stress, that influence corrosion. The technologically
Vol. 1 CORROSION AND CORROSION CONTROL 909

important structural materials are polycrystalline aggregates. Each individual


crystal is referred to as a grain. Grain orientation can affect corrosion resistance
as evidenced by metallographic etching rates and pitting behavior (22). Grain
shape may likewise vary greatly depending on the alloy and processing history.
Alloys, particularly in the as-cast condition, generally exhibit chemical inhomo-
geneity such that there is segregation of alloying elements and impurities to the
grain boundary regions. These heterogeneities, which can also develop during
subsequent processing such as welding or heat treatment, can produce different
electrochemical characteristics at the grain boundary relative to the grain inter-
ior and can lead to intergranular corrosion. This problem can be of great practical
importance, especially to wrought stainless steels and nickel alloys. Second
phases, such as ferrite grains in an otherwise austenitic stainless steel and
beta grains in an otherwise alpha brass, can be of considerable importance in
some alloy systems and some forms of corrosion. Residual stresses from cold-
working or other sources can lead to increased corrosion rates (22,24,28), and
are also important in stress–corrosion cracking.
The following sections briefly discusses some of the alloying and metallur-
gical factors in a three technologically important alloy systems. A more complete
discussion of the corrosion behavior of these and other alloy systems and the
influence of metallurgical factors on each is available (4).
9.1. Stainless Steels. The stainless steels, by virtue of the alloying
additions and processing, can be categorized as ferritic (body-centered cubic
structure), austenitic (face-centered cubic structure), duplex (a combination of
ferritic and austenitic), martensitic (body-centered tetragonal or cubic struc-
ture), and precipitation hardened. There are some 180 different alloys that can
be recognized as stainless steels. These alloys contain a minimum of 10.5 wt% Cr
and, depending on the grade, other alloying elements such as Ni, Mo, and N
(5,29). The concentration of particular alloying elements are based on the desired
properties of the stainless steel. Figure 7 shows the compositional and property
linkages for the stainless steels (29).
Figure 7 Compositional and property linkages for the stainless steels (29).
The corrosion resistance of stainless steels results from the formation of a
passive film and, for this reason these materials are susceptible to pitting corro-
sion and to crevice corrosion. The resistance to pitting or crevice corrosion is
improved by additional alloying with Cr, Mo, and N for the austenitic and duplex
stainless steels and alloying the ferritic stainless steels with additional Cr and
Mo. The manufacturers of stainless steels have developed indexes for resistance
to pitting and crevice corrosion based on the composition of the alloy. The index is
known as the pitting resistance equivalent number (PREN). PREN is given by
PREN ¼ %Cr þ 3.3 (%Mo) þ 16 (%N) þ 1.65 (%W). In general, the higher the
value of PREN, the greater the resistance to crevice and pitting corrosion. The
major problem with using a PREN is that it is based solely on composition and
ignores the detrimental effects of microstructure, such as Cr depleted zones and
alloy segregation (29).
The most common and serious metallurgical factor affecting the corro-
sion resistance of stainless steels is termed sensitization. This condition is
caused by the precipitation of chromium-rich carbides at the grain boundaries,
giving rise to chromium depleted grain boundary areas. For example, heating
910 CORROSION AND CORROSION CONTROL Vol. 1

Fig. 7. Compositional and property linkages for the stainless steels (29).

a type 304 stainless steel containing 0.039% carbon for 10 h at 7008C will reduce
the chromium level from 19% to <13% in the region next to the carbide pre-
cipitate (29). The depleted areas are anodic to the grain interior and tend to dis-
solve thereby causing intergranular corrosion. Sensitization can also make
stainless steel more prone to stress–corrosion cracking (SCC). There are several
measures available to mitigate sensitization. Low carbon grades such as AISI
304L and 316L are available that have much less tendency toward sensitization.
Also, alloying additions of titanium or niobium and tantalum can be used to tie
up carbon. Another area of concern with stainless steel alloys is EIC and this
topic is treated in more detail later in this article for stainless steels as well as
for other selected alloys. A more detailed discussion of stainless steels may be
found in the literature (5,29).
9.2. Copper Alloys. Copper and its alloys have good corrosion resis-
tance in many nonoxidizing aqueous environments and atmospheric conditions.
Several copper alloys are exceptionally resistant to certain atmospheres. The
copper–nickel alloys are the most corrosion resistant of the commercial copper
alloys (3). The two most common Cu Ni alloys are 90% Cu 10% Ni and 70%
Cu 30% Ni. They have good resistance to corrosion in both fresh and salt
water (4).
The most known of the copper alloys is brass, which is produced by the addi-
tion of zinc. The zinc improves the mechanical properties by solid solution
strengthening. However, the zinc reduces the corrosion resistance. A variety of
names are associated with brasses with a particular zinc content: 40% Zn Cu
Vol. 1 CORROSION AND CORROSION CONTROL 911

alloys are referred to as Muntz metal; 30% Zn Cu alloys are referred to as yel-
low brass; and 15% Zn Cu alloys are termed red brass (2). Brasses are suscep-
tible to dealloying in the form of dezincification, ie, the preferential loss of zinc
from the alloy. Brasses having zinc concentrations of 15% or greater are prone to
dezincification and dezincification is generally more severe in brasses that have
two metallurgical phases. For the two-phase alloys, greater than 38 wt% Zn,
dezincification usually starts in the high zinc content b-phase and is followed
by dezincification of the lower zinc a-phase (4). Naval brass and admiralty
brass are similar in composition to Muntz metal and yellow brass, respectively,
but have 1 wt% tin added for improved resistance to dezincification. Red brass is
relatively resistant to dezincification but is more susceptible to impingement
attack than, for example, yellow brass (2).
Conditions that favor dezincification include stagnant solutions, espe-
cially acidic ones, high temperatures, and porous scale formation (2). Tin
tends to inhibit dealloying especially in cast alloys (4). Additions of small
amounts of arsenic, antimony, or phosphorus can increase the resistance to
dezincification. These elements are, however, not entirely effective in prevent-
ing the dezincification of the two-phase (a-b) brasses because dezincification of
the b-phase is not prevented (4). Another area of corrosion concern involves
applied or residual stresses from fabrication that can lead to EIC of brasses
in the form of SCC.
9.3. Aluminum Alloys. The corrosion resistance of aluminum, like
stainless steels, is provided by a passive film that protects the surface from degra-
dation by the environment. However, in environments that contain aggressive
anions such as chloride, the passive film can be degraded locally causing film break-
down and, depending on the geometry of the sample, pitting or crevice corrosion.
Copper, silicon, magnesium , zinc, and manganese are some common alloy-
ing additions to aluminum. Most of the alloying elements are added to aluminum
to produce alloys having improved mechanical properties. However, the
strengthening phases that result from the alloying can disrupt the passive
oxide layer on aluminum and lead to localized corrosion. Also, the second
phase constituents can produce local galvanic cells as discussed below. Common
alloying elements that are added to steels and many other metals to improve cor-
rosion resistance, such as Cr and Mo, have a very low solubility in aluminum and
result in second phase precipitates. The presence of the second phase constitu-
ents, in turn, lowers the corrosion resistance. A number of surface modification
techniques that produce single phase, metastable aluminum alloys have been
shown to increase the localized corrosion resistance as compared to pure alumi-
num (30–34). These techniques include ion implantation, laser alloying, and
magnatron sputtering. However, due to capital costs and the sizes of the sample
that can be treated, these techniques have not gained widespread use.
Aluminum alloys are susceptible to intergranular corrosion when precipi-
tates form. Intergranular corrosion in aluminum alloys is caused by a potential
difference, ie, galvanic couple. The formation of the anodic and cathodic areas in
the couple varies depending on the alloying additions (3). Reference (35) is a com-
pilation of potentials reported for intermetalic phases in aluminum alloys.
Stress–corrosion cracking is characteristically intergranular for aluminum
alloys. Wrought high strength aluminum alloys, whether the products are rolled,
912 CORROSION AND CORROSION CONTROL Vol. 1

forged, or extruded, tend to be highly textured because of the manner in which


the secondary phases or inclusions are strung out. Exfoliation corrosion, a type
of intergranular corrosion, proceeds along subsurface paths. The corrosion
products have a greater volume than the alloy and produce a layer type attack.
Selection of alloys with proper tempers can provide resistance to exfoliation
corrosion (4). The texture is also is important to the SCC of high strength alumi-
num alloys.

10. Environmentally Induced Cracking

Environmentally induced cracking is a brittle fracture process caused by the con-


joint action of a mechanical stress and a corrosive environment. There are
several different types of EIC: stress–corrosion cracking (SCC), corrosion
fatigue cracking (CFC), and hydrogen induced cracking (HIC). In SCC and
HIC the stress may be normal stress experienced in operation or residual stress
from welding, heat treatment, and/or cold work. CFC results from cyclic stresses.
In any of these, the corrosion rates are generally very low and the stress level
required for the cracking is generally well below the normal yield stress of the
particular alloy involved. The concentration of the chemical species responsible
for the cracking need not be high. EIC failures are often catastrophic such as
the Pt. Pleasant, West Virginia, bridge collapse that killed 46 people in 1967 (3).
Cracking in steam pipes, underground pipelines, and aircraft have also been
attributed to EIC.
The chemical species that can lead to EIC in each alloy system are fairly
well known although the exact mechanisms of crack initiation and propagation
are not thoroughly understood. The species that promote EIC in one alloy system
do not necessarily promote EIC in others. A discussion of EIC and the chemical
species that promote it can be found in the literature (3).
The cracking in EIC can proceed either intergranularly, ie, between the
grains, or transgranularly, ie, across the grains. Small changes in the environ-
ment can cause a shift from one type to the other as can relatively small changes
in the composition of the alloy. Likewise the crack may be single and unbranched
or have multiple origins and also be branched. Thus neither intergranular or
transgranular cracking is indicative of or excludes EIC as the cause of a service
crack. More detailed information regarding EIC is available (1,3,4,36).
10.1. Copper Alloys. Copper alloys under an applied or residual stress
are susceptible to SCC in environments containing ammonia (qv) or ammonium
compounds (qv). In addition to ammonia, water and oxygen, or another cathodic
reactant, are required to cause SCC. Ammonia and ammonium compounds are
sometimes found in the atmosphere, in cleaning compounds, in fertilizers (qv),
or in chemicals used for water treatment. Brasses containing <15% zinc are
highly resistant to SCC, whereas brasses containing 20–40% zinc are highly sus-
ceptible and this susceptibility increases only slightly as zinc concentration
increases from 20 to 40% (4). SCC is usually intergranular but can occur trans-
granular under certain conditions (2,4).
Trace quantities of nitrogen oxides may also cause SCC. These failures are
probably the result of the nitrogen oxides being converted to ammonium salts on
Vol. 1 CORROSION AND CORROSION CONTROL 913

the brass surface. The premature failure of yellow brass brackets in the humidi-
fier chamber of an air-conditioning system was traced to this latter cause (2). In
another case, SCC of 12% Ni 23% Zn Cu alloy, known as nickel brass, parts of
the Central Office Telephone equipment in Los Angeles occurred within 2 years
of installation for similar reasons. The source of nitrogen was the air in the Los
Angeles area which has high concentrations of nitrogen oxides and suspended
nitrates (see AIR POLLUTION). The nitrates settle as dust on the brass parts (2).
The susceptibility to SCC can be minimized by (1) proper alloy selection; (2) ther-
mal stress relief; (3) avoiding contact with ammonia and ammonia compounds;
and (4) using an inhibitor (2,4).
10.2. Aluminum Alloys. Both the 2000 and the 7000 series aluminum
alloys have experienced significant SCC in service. These high strength alloys in
the wrought form are highly textured, the grains are flattened and elongated into
specific crystallographic directions due to the processing. SCC behavior differs
greatly according to the direction of tensile stress with respect to the texture direc-
tions. The alloys are most vulnerable to SCC if stressed parallel to the short trans-
verse grain direction, ie, parallel to the thinnest dimension of the grain and most
resistant if stressed only parallel to the longest grain dimension. In practice, it is
the short transverse-direction stresses that cause SCC problems. Hence prudent
practice is to avoid designs in which high sustained stresses are imposed across
the short transverse direction. For example, one avoids an interference-fit
fastener, or a taper pin fastener oriented to stress a part across the vulnerable
texture unless the alloy is inherently of low susceptibility to SCC. Table 3 sum-
marizes alloys and tempers in various categories of susceptibility as judged from
specimens having short transverse, ie, most vulnerable, orientations.
In addition to the possibility of selecting alloys having minimum SCC sus-
ceptibility while retaining other properties as needed, there are other steps pos-
sible to reduce the SCC probability: (1) avoid designs that permit water to
accumulate; (2) avoid conditions in which salts, especially chlorides, can concen-
trate; and (3) where available and otherwise acceptable, use a clad alloy (2) (see
METAL SURFACE TREATMENTS).
10.3. High Strength Steels. Steels that owe their strength to heat treat-
ment, whether martensitic, precipitation hardened, or maraging, and whether
stainless or not, are susceptible to SCC in aqueous environments, including
water vapor. The primary factor in determining the degree of SCC susceptibility
of a given steel is its strength. There is no sharply defined threshold strength that
defines a threshold susceptibility, but above 1200 MPa (174,000 psi) yield strength
the problem becomes of increasing concern, until at 1400 MPa (200,000 psi)
susceptibility is generally very high. This does not mean that steels cannot be
used at such strength levels or even higher, but for steels to be used at these
strengths, great care must be exercised in design such that any tensile or bending
stresses are small, or that moisture is excluded from the surface of the steel.
High strength steels are also susceptible to HIC. Cadmium electroplating
is a useful protection measure for steels, but hydrogen either must not be
codeposited with cadmium, or if it is codeposited, as in the cyanide plating
bath, hydrogen must be safely redistributed by thermal treatment before the
high strength steel component is stressed. Otherwise the steel may experience
HIC.
914 CORROSION AND CORROSION CONTROL Vol. 1

Table 3. Susceptibility Classification of Commercial Wrought


Aluminum Alloys in Plate Form—Short Transverse Orientationa
Aqueous
susceptibility Aluminum alloy Temperb
very low 1100 all
3003, 3004, 3005 all
5000, 5050, 5052, 5154 all
5454, 6063
5086 O, H32, H34
6061, 6262 O, T6
Alclad: 2014, 2219, 6061, 7075 all
low 2219 T6, T8
5086 H36
5083, 5456 controlled
6061 T4
6161, 5351 all
6066, 6070, 6071 T6
2021 T8
7049, 7050, 7075 T73
moderate 2024, 2124 T8
7050, 7175 T736
7049, 7075, 7178 T6
appreciable 2024, 2219 T3, T4
2014, 7075, 7079, 7178 T6
5083, 5086, 5456 sensitized
7005, 7039 T5, T6
a
Ref. 37.
b
Standard mill designation.

10.4. Stainless Steels. Austenitic stainless steels undergo SCC when


stressed in hot aqueous environments containing chloride ion. The oxygen level
of the environment can be important, probably through its effect in establishing
the electrode potential of the steel. The total matrix of combinations of stress
level, chloride ion and oxygen concentrations, and temperature that cause
SCC has not been worked out. At about the yield strength stress and ca
2908C, 1 ppm chloride and 1 ppm dissolved oxygen are about the minimum
levels that initiate SCC. At room temperature, SCC by chloride seldom occurs
in austenitic stainless steels except when the steels are heavily sensitized.
When these steels are not heavily sensitized, there is not usually SCC except
at elevated temperatures. Whereas there is no well-defined threshold tempera-
ture for vulnerability to SCC in stainless steels, above 608C, it becomes of
increasing concern.
The relative susceptibilities of the common grades of austenitic stainless
steel to SCC in chloride do not differ greatly, although the steels that can and
do become sensitized are decidedly inferior in this condition. The high purity fer-
ritic grades of stainless steel offer appreciable improvement in SCC resistance as
compared to the austenitic grades, but they are not immune to cracking. Care
must also be exercised to avoid the ductile-to-brittle transition that most steels
undergo as the temperature is decreased and potential embrittlement by s-phase
Vol. 1 CORROSION AND CORROSION CONTROL 915

formation in elevated temperature service. The standard methods for avoiding


chloride SCC in austenitic stainless steels include: avoidance of fabrication
stresses; minimizing chloride ion level; and minimizing oxygen concentration
in the environment. Additionally, where feasible, it is possible to mitigate
against SCC in these alloys by cathodic protection, but excessive cathodic protec-
tion can lead to HIC.

11. Inhibitors

Corrosion inhibitors are substances that slow or prevent corrosion when added to
an environment in which a metal usually corrodes. Corrosion inhibitors are
usually added to a system in small amounts either continuously or intermittently.
The effectiveness of corrosion inhibitors is partly dependent on the metals or alloys
to be protected as well as the severity of the environment. The main factors, which
must be considered before application of a corrosion inhibitor to an aqueous sys-
tem, are the compatibility of the inhibitor and the metal(s), the salt concentration,
the pH, the dissolved oxygen concentration, and the concentration of interfering
species such as chlorides or metal cations. In addition, many inhibitors, most nota-
bly chromates, are toxic and environmental regulations limit use. (38).
Inhibitors act and are classified in a variety of ways (1,3,38,39). The classi-
fications used herein closely follow the discussion in (38). Types of inhibitors
discussed below include (1) anodic, (2) cathodic, (3) organic, (4) precipitation,
and (5) vapor-phase inhibitors.
11.1. Anodic Inhibitors. Passivating or anodic inhibitors produce a
large positive shift in the corrosion potential of a metal. There are two classes
of anodic inhibitors which are used for metals and alloys where the anodic
shift in potential promotes passivation, ie, anodic protection. The first class
includes oxidizing anions that can passivate a metal in the absence of oxygen.
Chromate is a classical example of an oxidizing anodic inhibitor for the passiva-
tion of steels. The second class of anodic inhibitors contains ions which need oxy-
gen to passivate a metal. Tungstate and molybdate, eg, require the presence of
oxygen to passivate a steel. The concentration of the anodic inhibitor is critical
for corrosion protection. Insufficient concentrations can lead to pitting corrosion
or an increase in the corrosion rate. The use of anodic inhibitors is more difficult
at higher salt concentrations, higher temperatures, lower pH values, and in some
cases, at lower oxygen concentrations (38).
11.2. Cathodic Inhibitors. Cathodic inhibitors act to retard or poison
the cathodic reaction or selectively precipitate onto cathodic areas producing dif-
fusion barriers to cathodic reactants, thereby reducing the rate of the cathodic
reaction. There are three types of cathodic inhibitors: (1) hydrogen poisons,
(2) oxygen scavengers, and (3) cathodic precipitates. Hydrogen poisons are chemical
species such as arsenic or antimony that retard the hydrogen reduction reaction.
Because the hydrogen poison slows the cathodic reaction, and because the catho-
dic and anodic reactions must proceed at the same rate, the whole of the corro-
sion process is slowed. A potentially serious drawback upon use of hydrogen
poisons is that the hydrogen on the surface can be more easily absorbed
into the metal or alloy and can lead to HIC in susceptible materials. Oxygen
916 CORROSION AND CORROSION CONTROL Vol. 1

scavengers prevent corrosion by tying up the oxygen in solution, thereby making


it unavailable for the cathodic reaction. The most common oxygen scavengers
used in water at ambient temperatures are sulfur dioxide [7446-09-5] and
sodium sulfate [7757-82-6]. Cathodic precipitate inhibitors such as calcium
carbonate [471-34-1] or magnesium carbonate [546-93-0] precipitate onto the
cathodic areas producing a film that reduces the cathodic activity (38).
11.3. Organic Inhibitors. Generally, organic inhibitors adsorb on the
entire metal surface and impede corrosion reactions (38). Organic inhibitors con-
sist of broad classes of organic compounds. For example, aliphatic organic amines
(qv) adsorb by the surface-active NH2 group that forms a chemisorptive bond with
the metal surface. The hydrocarbon tails orient away from the interface toward
the solution, so that further protection is provided by the formation of a hydropho-
bic network that excludes water and aggressive ions from the metal surface (39).
Organic inhibitors influence both anodic and cathodic reactions to varying
degrees depending on the potential, the chemical structure of the inhibitor, and
the size of the inhibitor molecule. Soluble organic inhibitors produce a protective
layer that is only a few molecules thick, whereas insoluble inhibitors added as dis-
persions can build a film to a thickness of several hundredths of a centimeter (38).
11.4. Precipitation and Vapor-Phase Inhibitors. Precipitation inhi-
bitors are film-forming compounds that produce barrier films over the entire sur-
face. Phosphates and silicates, which are the most common, do not provide the
degree of protection afforded by chromate inhibitors, but are useful in situations
where nontoxic additives are required. Two main drawbacks to the use of phos-
phates and silicates are the dependence on the water composition and the control
required to achieve maximum inhibition (38,39).
Vapor-phase inhibitors are volatile compounds that adsorb onto metal sur-
faces, and retard or prevent corrosion by a variety of mechanisms (38). Inhibitors
such as dicyclohexamine nitrate [3882-06-02] can protect a variety of metals in-
cluding steel, aluminum, and tinplate. A number of vapor-phase inhibitors are
commercially available as powders or tablets. However, vapor-phase inhibitors
attack nonferrous metals to varying degrees, thus the manufacturers’ recom-
mendations should be checked before application. The system to be protected
must be closed to maintain the volatile compound, but objects as large as the
interior of an ocean-going tanker have been treated by this technique.
11.5. Environmentally Compatible Inhibitors. The recent and ex-
pected environmental regulations concerning the release of hazardous materials
and the growing list of materials considered as hazardous are impacting the use
of many effective inhibitors such as chromate. Thus, new research and develop-
ment on environmentally acceptable inhibitors is proceeding. This is a great
challenge as inhibitor and coating users and manufacturers have to replace com-
mercial products that have decades of acceptable field performance in regard to
corrosion (40,41).

12. Coatings for Corrosion Prevention

Coatings are applied to metal substrates to prevent corrosion. Generally,


the coating protects the metal by imposing a physical barrier between the
Vol. 1 CORROSION AND CORROSION CONTROL 917

metal substrate and the environment. However, the coating can also act to pro-
vide cathodic protection or by serving as holding reservoirs for inhibitors. Coat-
ings may be divided into organic, inorganic, and metallic coatings (1–3,42). For a
more in-depth coverage of this topic see (1–3) (see also COATINGS, MARINE; METAL
SURFACE TREATMENTS).
12.1. Coating Types. Organic coatings afford protection by providing a
physical barrier and are often used as holding reservoirs for corrosion inhibitors.
Organic coatings include paints, resins, lacquers, and varnishes (see also BARRIER
POLYMERS; PAINT). These coatings usually comprise four basic constituents: binder,
pigments and fillers, additives and solvents. The properties of the coating
strongly depend on the formulation of the coating (43). The effective application
of an organic coating requires (1) proper surface preparation of the metal
substrate, (2) selection of the proper primer or priming coat, and (3) selection
of the appropriate top coat(s). Poor performance of organic coatings most often
results from improper surface preparation or poor application (1). Maintenance
programs involving periodic inspection and repair are necessary for reliable cor-
rosion protection by organic coatings. Organic coatings probably protect more
metal on a tonnage basis than any other means of corrosion protection. As a
rule, however, organic coatings should not be used in environments that would
rapidly corrode the metal if the coating were compromised. For example, paint
would not be used to protect the inside of a tank car used for shipping hydrochlo-
ric acid because one small coating defect would result in the rapid perforation of
the tank car wall.
Inorganic coatings are also used to provide a barrier between the environ-
ment and the metal (1,2,42). Inorganic coatings include chemical conversion
coatings, glass (qv) linings, enamels (see ENAMELS, PORCELAIN OR VITREOUS),
and cement (qv). Chemical conversion coatings are produced by intentionally
corroding the metal surface in a controlled manner. This is done so as to pro-
duce an adherent corrosion product that protects the metal from further corro-
sion. Anodization of aluminum, one of the more commonly used conversion
coatings techniques, produces a protective aluminum oxide film on the alumi-
num metal. Another example of a chemical conversion coating is phosphatizing
for the protection of automobile bodies. Porcelain enamel coatings, that
are inert in water and resistant to most weather are routinely applied to
steel, cast iron, and aluminum (4). They are commonly seen on appliances
and plumbing fixtures. Glass-lined metals are used in process industries
where there is concern over corrosion or contamination of the product.
For example, glass-lined metals are used in the pharmaceutical industry (4).
Portland cement coatings have been used to protect steel and cast-iron water
pipes (2).
Often metallic coatings, in addition to providing a barrier between the
metal substrate and the environment, provide cathodic protection when the
coating is compromised (2,3). Metallic coatings and other inorganic coatings
are produced using a variety of techniques including hot dipping, electroplat-
ing, cladding, thermal spray techniques, chemical vapor deposition, or by sur-
face modification using directed energy (laser or ion) beams. Two classic
metallic coating techniques and a relatively recent technique are described
below.
918 CORROSION AND CORROSION CONTROL Vol. 1

13. Coating Methods

Hot dipping is one of the oldest methods used for coating metals. In hot dipping,
coatings are applied by dipping the metal in a molten bath commonly of zinc, tin,
lead, or aluminum (2,4). The best known hot dipping procedure is that of coating
steel with zinc to produce galvanized steel. In addition to providing a barrier
between the steel substrate and the environment, the zinc acts as a sacrificial
anode and provides cathodic protection to the underlying steel when the coating
is breached.
Electroplating consists of immersing a metal in a plating bath and
electrochemically plating the solution species onto the metal substrate (1,4).
Electroplating is used to provide coatings for a variety of reasons including
corrosion protection, wear resistance, decoration, and to build up the dimen-
sions of a substrate. Additives to the plating bath can improve coating
properties such as grain size, strength, uniformity, and brightness. The elec-
troplate can be a single metal, an alloy, or a sequence of layers of different
composition. Coating thicknesses can range from on the order of thousandths
of a mil to 20 mils (1). Zinc, nickel, tin, and cadmium are the most commonly
plated materials on a tonnage basis. Electroplated tin is used as a protective
coating for food cans. The electroplated tin provides a physical barrier and
galvanic protection (in most food products) if the coating is compromised.
However, the tin coating cannot provide galvanic protection in the presence
of dissolved oxygen so that food should not remain in tin plated cans after
opening (3).
The ability to modify metal surfaces using directed energy beams is a
new approach being used to improve the corrosion resistance of metals
(31,44). Techniques such as ion implantation (qv), ion beam mixing, and ion
beam assisted deposition (IBAD), as well as laser-surface alloying and proces-
sing have been shown to improve the corrosion behavior of metals (see LASERS).
These techniques are versatile and have been used to modify the cathodic
or anodic reactions, improve the nature of passive films, and produce barrier
coatings (31). Ion implantation has been used to improve the corrosion
resistance of titanium in hot acids and aluminum and steels in aqueous
chloride environments (44). However, due to capital costs and the sizes of
the sample that can be treated, these techniques have not yet gained wide
spread use.

BIBLIOGRAPHY

‘‘Corrosion Inhibitors’’, in ECT 2nd ed., Vol. 6, pp. 317–346, by C. C. Nathan; ‘‘Corrosion
and Corrosion Inhibitors’’, in ECT 3rd ed., Vol. 7, pp. 113–142, by R. T. Foley and B. F.
Brown, The American University; in ECT 4th ed., Vol. 7, pp. 548–572, by Patrick &
Moran, United States Naval Academy, and Faul M. Natishan, Naval Research Labora-
tory; ‘‘Corrosion and Corrosion Control’’ in ECT (online), posting date: December 4,
2000, by Patrick J. Moran, United States Naval Academy and Paul M. Natishan, Naval
Research Laboratory.
Vol. 1 CORROSION AND CORROSION CONTROL 919

CITED PUBLICATIONS

1. M. G. Fontana, Corrosion Engineering, McGraw-Hill Book Co., Inc., New York,


1986.
2. H. H. Uhlig, and R. W. Revie, Corrosion and Corrosion Control, John Wiley & Sons,
Inc., New York, 1985.
3. D. A. Jones, Principles and Prevention of Corrosion, Macmillan Publishing, New
York, 1992.
4. ASM Handbook, ASM International, Materials Park, 1987.
5. A. J. Sedriks, Corrosion of Stainless Steels, John Wiley & Sons, Inc., New York, 1979.
6. Z. Szklarska-Smialowska, Pitting Corrosion of Metals, National Association of Corro-
sion Engineers, Houston, Tex., 1986.
7. H. L. Logan, Corrosion Basics: An Introduction, L. S. V. Delinder, ed., National Asso-
ciation of Corrosion Engineers, Houston, Tex., 1984.
8. S. C. Dexter, Bull. Electrochem. 12, 1 (1996).
9. S. C. Dexter, and P. Chandrasekaran, Biofouling 15, 313 (2000).
10. B. Little, P. Wagner, and F. Mansfeld, Int. Mater. Rev. 36, 253 (1991).
11. R. Pope, B. Little, and R. Ray, Biofouling 16, 83 (2000).
12. M. Pourbaix, Lectures on Electrochemical Corrosion, Plenum Press, New York-
London, 1973.
13. M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous Solutions, NACE,
Houston, Tex., 1974.
14. J. Tafel, Z. Physik. Chem. 50, 641 (1905).
15. J. O. M. Bockris and A. K. N. Reddy, Modern Electrochemistry, Plenum Press, New
York, 1970.
16. H. Kita, J. Electrochem. Soc. 113, 1095 (1966).
17. R. Baboian, Electrochemical Techniques in Corrosion Engineering, NACE, Houston,
Tex., 1986.
18. U. R. Evans, J. Soc. Chem. Ind. 43, 315 (1924).
19. W. H. J. Vernon, J. Sci. Inst. 22, 226 (1945).
20. C. W. Borgmann, Ind. Eng. Chem. 29, 814 (1937).
21. W. Whitman, R. Russell, and V. Atteri, Ind. Eng. Chem. 16, 665 (1924).
22. J. Kruger, Int. Mater. Rev. 33, 113 (1988).
23. H. H. Strehblow, Werkstoffe Korros. 35, 437 (1984).
24. Z. Szklarska-Smialowska, Corrosion 27, 223 (1971).
25. G. S. Frankel, J. Electrochem. Soc. 145, 2970 (1998).
26. W. H. J. Vernon, and L. Whitby, J. Inst. Met. 44, 389 (1930).
27. W. H. J. Vernon, J. Inst. Met. 49, 153 (1932).
28. Corrosion in Action, The International Nickel Co, New York, 1977.
29. A. J. Sedriks, ASM Handbook, ASM International, Materials Park, 2002.
30. V. Ashworth, W. A. Grant, and R. O. M. Procter, Corr. Sci. 16, 661 (1976).
31. E. McCafferty, G. K. Hubler, P. M. Natishan, P. G. Moore, R. A. Kant, and B. D.
Sartwell, J. Mater. Sci. Eng. 86, 1 (1987).
32. W. C. Moshier, and G. D. Davis, J. Electrochem. Soc. C145 (1988).
33. P. M. Natishan, E. McCafferty, and G. K. Hubler, J. Electrochem. Soc. 135,
321 (1988).
34. A. R. Srivatsa, C. R. Clayton, and J. K. Hirvonen, Advances in Coating Technology for
Corrosion and Wear Resistant Coatings, TMS, Warrendale, 1995.
35. R. G. Buchheit, J. Electrochem. Soc. 142, 3994 (1995).
36. L. S. V. Delinder, Corrosion Basics: An Introduction, National Association of
Corrosion Engineers, Houston, Tex., 1984.
920 BIOREMEDIATION Vol. 1

37. E. H. Spuhler, and C. L. Burton,in ALCOA Green Letter, Aluminum Company of


America, Pittsburgh, 1970.
38. N. Hackerman, and E. S. Snavelyin, Corrosion Basics: An Introduction, L. S. V.
Delinder, ed., NACE, Houston, Tex., 1984.
39. McCaffertyE., in J. H. Delinder, ed., Corrosion Control by Coatings, Science Press,
Princeton, N. J. 1979, p. 279.
40. S. R. Taylor, H. S. Isaacs, and E. W. Brooman, Environmentally Acceptable Inhibitors
and Coatings, The Electrochemical Society, Inc., Pennington, 1997.
41. D. C. Hansen, and E. McCafferty, J. Electrochem. Soc. 143, 114 (1996).
42. N. E. Hamner, Corrosion Basics: An Introduction, L. S. V. Delinder, ed., NACE,
Houston, Tex., 1984.
43. J. H. W. D. Wit, in P. Marcus and J. Oudar, eds., Corrosion Mechanisms in Theory
and Practice, Marcel Dekker, Inc., New York, 1995, p. 581.
44. McCaffertyE., P. M. Natishan, and G. K. Hubler, Nucl. Instr. Meth. B56, 639 (1991).

PAUL NATISHAN
Naval Research Laboratory
PATRICK MORAN
U.S. Naval Academy

BIOREMEDIATION
Bioremediation is the process of judiciously exploiting biological processes
to minimize an unwanted environmental impact; usually it is the removal of a
contaminant from the biosphere. Like most definitions in biology, that of biore-
mediation is the subject of some debate. A narrow definition might focus on the
conversion of contaminating organic molecules to carbon dioxide, water, and
inorganic ions, and the oxidation or reduction of contaminating inorganic ions.
A broader definition would include biological processes for ameliorating extremes
of pH, concentrating contaminants so that they can be more easily removed by
physical techniques, converting toxic species to less toxic or less bioavailable
forms that pose less of a threat to the environment, and restoring functional eco-
systems to contaminated or disturbed sites when the contaminants or distur-
bance cannot be removed.
The concept of ‘‘judiciously exploiting biological approaches’’ is also a sub-
ject of debate:

Some would restrict it to providing a nutrient that is otherwise limiting the


most effective growth of organisms catalyzing the desired reaction, whether
it is the degradation of an organic compound, the reduction or oxidation of
an inorganic ion, or the accumulation of a contaminant. This simple
approach has been successful with a range of contaminants. The nutrient
might be a fertilizer providing nitrogen, phosphorus and other essential
minerals, or an electron acceptor such as oxygen.

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.0209151816180914.a01
Vol. 1 BIOREMEDIATION 921

Others would extend the fertilizer concept to the simultaneous addition of


readily biodegradable substrates along with the fertilizer nutrients to
stimulate the growth of contaminant-degrading organisms most rapidly,
and to aid in the rapid utilization of the fertilizer nutrients before they
might be leached from the contaminated area. The specific requirements
for the most efficacious substrates is an area of current research.
An alternative use of added readily degradable substrates is to drive the local
environment toward anaerobiosis so that reactions such as reductive de-
chlorinations or reductive removal of nitro-groups are promoted.
A broader view would include the addition of a substrate to stimulate the
growth of organisms known to degrade the contaminant of interest only
as an incidental part of their metabolism, one might almost say serendipi-
tously. This process is sometimes called co-metabolism, and it too has had
success.
Others would include the addition of materials aimed at increasing the bioa-
vailability of the contaminant to the degrading organisms. The most stu-
died compounds are surfactants, but cations have been reported to
increase the bioavailability of some organic compounds, and sorbents and
clays are also considered. The dispersion of spilled oil on water by the
application of dispersants is perhaps the major commercial use of this idea.
Another important option is the addition of remediating organisms. While the
addition of contaminant-degrading bacteria has not yet had much docu-
mented success with natural products such as hydrocarbons, there is rea-
son to expect that it will be efficacious with pollutants that are more recent
additions to the environment. The planting of specific plants, in the process
known as phytoremediation, is also a promising approach. Not only do the
plants themselves have remediating activities, such as the accumulation of
certain metal ions, but they also have extensive bacterial and fungal popu-
lations associated with their root systems, and inoculation of this rhizo-
sphere is widely practiced. It is, thus, possible that planting seeds with
microbial inoculants will become an option for bioremediation. There is
much talk of genetically modifying organisms so that their remediative
potential is increased, and in the future this may well become an important
option.
The broadest view of ‘‘aiding and abetting’’ includes doing nothing, but
merely watching natural processes occur without further intervention.
This has been termed ‘‘Intrinsic Bioremediation’’, and it too has met
with success. From an environmental point of view, although unfortu-
nately not always from a regulatory viewpoint, it is important that any re-
mediation intervention yield a clear net environmental benefit.
Sometimes very mild stimulation of intrinsic processes may be the most
environmentally responsible option.

Bioremediation overlaps some older biotechnologies. Municipal and indus-


trial wastewater treatment is a well-established industry, and although it can be
distinguished from bioremediation in that the pollutants are under physical
control during treatment, the fundamental biological processes have much in
922 BIOREMEDIATION Vol. 1

common. Similarly composting is a well-established phenomenon, currently


gaining popularity in the municipal solid waste treatment industry, and the
biofiltration of waste gases is becoming a useful technology. Developments of
these technologies, where the contaminant is already under physical control,
will undoubtedly aid the development of bioremediation as an accepted tool for
dealing with similar wastes when they have escaped control. This article focuses
on biological treatments for contaminants when they have escaped into the
environment.
Bioremediation is already a commercially viable technology, with estimates
of aggregate bioremediation revenues of $2–3 billion for the period 1994–2000
(1). There are significant opportunities to enlarge upon this success. Bioremedia-
tion has applications in the gas phase, in water, and in soils and sediments. For
water and soils, the process can be carried out in situ, or after the contaminated
medium has been moved to some sort of contained reactor (ex situ). The former is
generally rather cheaper, but the latter may result in such a significant increase
in rate that the additional cost of manipulating the contaminated material is
overshadowed by the time saved. Bioremediation may explicitly exploit bacteria,
fungi, algae, or higher plants. Each, in turn, may be part of a complex food-web,
and optimizing the local ecosystem may be as important as focusing solely on the
primary degraders or accumulators.
Bioremediation usually competes with alternative approaches to achieving
an environmental goal. Bioremediation is typically among the least expensive
options, but an additional important consideration is that in many cases biore-
mediation is a permanent solution to the contamination problem, since the con-
taminant is completely destroyed or collected. Some of the alternatives
technologies, such as thermal desorption and destruction of organics, are also
permanent solutions, but the simplest, removing the contaminant to a dump
site, merely moves the problem, and may well not eliminate the potential
liability. Furthermore, by its very nature bioremediation addresses the bioavail-
able part of any contamination, and when biodegradation or bioaccumulation
ceases this probably means that the bioavailable part of the contamination has
been addressed. Residual concentrations of contaminants, although perhaps
detectable by today’s sensitive analytical techniques, may in fact have no resi-
dual environmental impact. The same cannot necessarily be said for nonbiologi-
cal technologies, which may leave bioavailable contaminants at low levels.
Bioremediation also has the advantage that is can be relatively nonintru-
sive, and can sometimes be used in situations where other approaches would
be severely disruptive. For example, bioremediation has been used to clean up
hydrocarbon spills under buildings, roads, and airport runways without interfer-
ing with the continued use of these facilities.
On the other hand, bioremediation is usually slower than most physical
techniques, and may not always be able to meet some very strict clean-up stan-
dards. Nevertheless, it is becoming a widely used technology. This article
addresses bioremediation in its broadest sense, focusing on the contaminants
that can be treated, the underlying biological processes that can mitigate the
contamination, and the technologies that have been used, or are being developed,
to treat them.
Vol. 1 BIOREMEDIATION 923

1. General Biological Aspects

The biosphere plays an important role in the great elemental cycles of


the earth (2), and bioremediation must be placed in this context if it is to be
appreciated in its broadest ramifications. One of the underlying fundamental
truths of biological diversity is that if there is free energy available in the
metabolism of a substrate, there is probably a guild of organisms that has
evolved to make use of it. This is particularly germane to the biodegradation
of organic molecules. For example, crude oil seeps to both land and water
have occurred for millennia, and as a consequence, aerobic oil-degrading micro-
organisms are ubiquitous. If biology does not yet take advantage of a source of
free energy, then it can be expected that there will be a strong selection
pressure in favor of any organism that develops an ability to exploit it. This
has been seen with by-products of nylon manufacture, where a Pseudomonas
aeruginosa has gained the ability to degrade the novel compound 6-aminohex-
anoate linear dimer, a by-product of nylon-6 manufacture, as the sole source
of carbon and nitrogen (3). The successful bioremediation of xenobiotic
compounds, such as pesticides and herbicides, may well represent a similar
acquisition of traits.
Not all organic molecules provide a source of free energy, however. Some,
such as small halogenated solvents, provide no significant source of nutrients
or energy, and their aerobic destruction can only occur co-metabolically with
the degradation of a more nutritious substrate (4). The white-rot fungi provide
another variation on this theme. These organisms seem unique in their ability
to degrade lignin, the structural polymer of higher plants. They may not gain
any direct energetic benefit from lignin degradation, but it clearly allows access
to cellulose which is a substrate for growth. Lignin degradation is catalyzed by a
group of extracellular peroxidases that generate nonspecific oxidants, and there
have been several proposals to use these systems for destroying contaminating
organic compounds (5).
With successful bioremediation, organic compounds can eventually be con-
verted to carbon dioxide, water, and biomass. Similarly, nitrogenous molecules,
such as excess ammonia or nitrate in groundwater, can be mineralized to gas-
eous nitrogen. Alternatively they can stimulate the growth of plants, either ter-
restrial or marine, and the plant biomass can eventually be harvested so that the
nitrogen is effectively removed from the local environment. Other nonorganic
contaminants provide a different challenge for bioremediation. A few, such as
mercury and selenium, are volatilized by some biological processes, but it is
not clear that this is always beneficial. In some cases, such as chromium and
arsenic, there is a dramatic difference in environmental toxicity depending on
the redox state of the contaminant. Bioremediation has sometimes focused on
this detoxification, usually by bacterial processes. A more satisfying approach
would be to use a biological process to accumulate and concentrate the contami-
nant so that it can be removed for safe disposal. Fungi, algae, and higher plants
have all been used in these efforts.
Table 1 explains a few of the biological terms that are widely used in dis-
cussing bioremediation, and which are used in the following text.
924 BIOREMEDIATION Vol. 1

Table 1. Some Biological Definitions Relevant to Bioremediation


Term Explanation
aerobic conditions with free oxygen
anaerobic conditions scrupulously free of oxygen
anoxic conditions with very low levels of oxygen
autotrophic growth using atmospheric CO2 as sole source of carbon
co-metabolic biodegradation of a contaminant only fortuitously with
degradation degradation of a true substrate
denitrification the reduction of nitrate to gaseous nitrogen
Eukaryotes organisms with a membrane-bound nucleus; the protozoa,
fungi, plants, and animals
eutrophic very rich nutrient conditions, especially of nitrogen compounds
heterotrophic growth at the expense of complex organic substrates
lignolytic growth of white-rot fungi under conditions where they
synthesize lignin-degrading peroxidases
methanogenic very anaerobic conditions, where carbon dioxide is reduced
to methane
methanotrophic aerobic growth with methane as sole source of carbon
and energy
mineralization conversion of a contaminant to its simplest forms, eg, CO2,
H2O, CH4, N2, Cl
nitrate-reducing, anoxic conditions, where nitrate is reduced to nitrogen gas
denitrifying
nitrification the biological oxidation of ammonia to nitrite and nitrate
oligotrophic very low nutrient conditions
Prokaryotes organisms lacking a membrane-bound nucleus; the bacteria
and archaea
recalcitrant very resistant to biodegradation
reductive the sequential loss of halogen substituents under anaerobic,
dehalogenation, usually methanogenic, conditions
reductive
dechlorination
rhizosphere the soil around plant roots; this zone has different microbial
populations from the bulk soil
sulfate-reducing very anaerobic conditions, where sulfate is reduced,
by sulfate-reducing bacteria, to sulfide
vadose zone the part of the soil above the water table

2. General Technological Aspects

Successful bioremediation hinges upon the effective application of the biology


discussed above. Sometimes the contaminant is on the surface, so access to
it is reasonably simple. Indeed the required technology may be as simple as
broadcast spreaders or sprayers to apply fertilizers, or tilling the soil to
allow good aeration. Of course this is not necessarily as simple as it sounds,
since contaminated sites are often very different from agricultural fields, and
the technology has to be significantly stronger to ‘‘plow’’ the soil. Frequently
the contaminant is below the surface, and applying even simple bioremediation
strategies can be very involved. Table 2 lists some of the technologies in use
today.
Vol. 1 BIOREMEDIATION 925

Table 2. Some Technological Definitions Relevant to Bioremediation


Technology Description
air sparging; aquifer injection of air to stimulate aerobic degradation; may also
sparging; biosparging stimulate volatilization
air stripping injection of air to stimulate volatilization
aquifer bioremediation in situ bioremediation in an aquifer, usually by adding
nutrients or co-substrates
aquifer sparging injection of air into a contaminated aquifer to stimulate
aerobic degradation, may also stimulate volatilization
batch reactor a bioreactor loaded with contaminated material, and run
until the contaminant has been consumed, then emptied,
and the process is repeated
bioactive barrier; a zone, usually subsurface, where biodegradation of a
bioactive zone; biowall contaminant occurs so that no contaminant passes
the barrier
bioaugmentation addition of exogenous bacteria with defined degradation
potential (or rarely indigenous bacteria cultivated in a
reactor and reapplied)
biofilm reactor a reactor where bacterial communities are encouraged on a
high surface area support, biofilms often have a redox
gradient so that the deepest layer is anaerobic while the
outside is aerobic
biofiltration usually an air filter with degrading organisms supported on
a high surface area support such as granulated activated
carbon
biofluffing augering soil to increase porosity
bioleaching extracting metallic contaminants at acid pH
biological fluidized bed; bioreactor where the fluid phase is moving fast enough to
fluidized-bed bioreactor suspend the solid phase as a fluid-like phase
biopile; soil heaping an engineered pile of excavated contaminated soil, with
engineering to optimize air, water, and nutrient control
bioslurping vacuum extraction of the floating contaminant, water, and
vapor from the vadose zone; the air flow stimulates
biodegradation
biostimulation optimizing conditions for the indigenous biota to degrade the
contaminant
biotransformation the biological conversion of a contaminant to some other
form, but not to carbon dioxide and water
biotrickling filter a reactor where a contaminated gas stream passes up a
reactor with immobilized micro-organisms on a solid
support, while nutrient liquor trickles down the reactor
bioventing vacuum extraction of contaminant vapors from the vadose
zone, thereby drawing in air that stimulates the
biodegradation of the remainder
borehole bioreactor; the addition of nutrients and electron acceptor to stimulate
in-well bioreactor biodegradation in situ in a contaminated aquifer
closed-loop bioremediation groundwater recovery, a bioreactor, and low-pressure
reinjection to maximize nutrient use, and maintain
temperature in cold climates
composting addition of biodegradable bulking agent to stimulate
microbial activity; optimal composting generally
involves self-heating to 50–608C
constructed wetland artificial marsh for bioremediation of contaminated water
continuous stirred a completely mixed bioreactor
tank reactor (CSTR)
926 BIOREMEDIATION Vol. 1

Table 2. (Continued)
Technology Description

digester usually an anaerobic bioreactor for digestion of solids and


sludges that generates methane
ex-situ bioremediation usually the bioremediation of excavated contaminated soil
in a biopile, compost system or bioreactor
fixed-bed bioreactor bioreactor with immobilized cells on a packed column matrix
land-farming; application of a biodegradable sludge as a thin layer to a soil
land treatment to encourage biodegradation; the soil is typically tilled
regularly
natural attenuation; unassisted biodegradation of a contaminant
intrinsic bioremediation
phytoextraction the use of plants to remove and accumulate contaminants
from soil or water to harvestable biomass
phytofiltration the use of completely immersed plant seedlings, to remove
contaminants from water
phytoremediation the use of plants to effect bioremediation
phytostabilization the use of plants to stabilize soil against wind and water
erosion
pump and treat pumping groundwater to the surface, treating, and
reinjection or disposal
rhizofiltration the use of roots to immobilize contaminants from a
water stream
rotating biological bioreactor with rotating device that moves a biofilm through
contactor the bulk water phase and the air phase to stimulate
aerobic degradation
sequencing batch periodically aerated solid phase or slurry bioreactor
reactor operated in batch mode
soil-vapor extraction vacuum-assisted vapor extraction

3. Organic Contaminants

3.1. Hydrocarbons. Constituents. Hydrocarbons get into the environ-


ment from biogenic and fossil sources. Methane is produced by anaerobic bac-
teria in enormous quantities in soils, sediments, ruminants and termites, and
it is consumed by methanotrophic bacteria on a similar scale. Submarine
methane seeps support substantial oases of marine life, with a variety of inver-
tebrates possessing symbiotic methanotrophic bacteria (6). Thus, methano-
trophic bacteria are ubiquitous in aerobic environments. Plants generate large
amounts of volatile hydrocarbons, including isoprene and a range of terpenes
(7). These compounds provide an abundant substrate for hydrocarbon-degrading
organisms.
Crude oil has been part of the biosphere for millennia, leaking from oil
seeps on land and in the sea. Crude oils are very complex mixtures, primarily
of hydrocarbons although some components do have heteroatoms such as nitro-
gen (eg, carbazole) or sulfur (eg, dibenzothiophene). Chemically, the principal
components of crude oils and refined products can be classified as aliphatics, aro-
matics, naphthenics, and asphaltic molecules. Representative examples are
shown in Figure 1. The ratios of these different classes varies in different oils,
but a typical crude oil might contain the four classes in a ratio of approximately
Vol. 1 BIOREMEDIATION 927

S
Octane
Dibenzothiophene

Isooctane N
2,2,4-trimethylpentane H

Carbazole

HO

Toluene

A naphthenic acid

Phenanthrene

Tetralin
1,2,3,4-tetrahydronaphthalene

A putative asphaltene

Fig. 1. Some representative hydrocarbons found in crude oil.

30:30:30:10. Most crude oils contain hydrocarbons ranging in size from methane
to molecules with hundreds of carbons, although the lightest molecules are
usually absent in oils that have been partially biodegraded in their reservoir.
When crude oils reach the surface environment the lighter molecules evaporate,
and are either destroyed by atmospheric photooxidation or are washed out of the
atmosphere in rain, and are biodegraded. Some molecules, such as the smaller
aromatics (benzene, toluene, etc) have significant solubilities, and can be washed
out of floating slicks, whether these are at sea, or on terrestrial water tables. For-
tunately the majority of molecules in crude oils, and refined products made from
them, are biodegradable, at least under aerobic conditions.
Biodegradation. Methane and the volatile plant terpenes are fully biode-
gradable by aerobic organisms, and most refined petroleum products are essen-
tially completely biodegradable under aerobic conditions. Estimates for crude oil
biodegradability range up to 90% (8), and the least biodegradable material, prin-
cipally polar molecules and asphaltenes, lacks the ‘‘oily’’ feel and properties that
are associated with oil. These are essentially impossible to distinguish from more
928 BIOREMEDIATION Vol. 1

HO O –O
O

R R R R

Linear alkanes
O–
O
O O

O
–O
Cycloalkanes

Fig. 2. Initial steps in the biodegradation of linear and cyclic alkanes.

recent organic material in soils and sediments, such as the humic and fulvic
acids, and appear to be biologically inert.
Numerous bacterial and fungal genera have species able to degrade hydro-
carbons aerobically and the pathways of degradation of representative aliphatic,
naphthenic and aromatic molecules have been well characterized in at least some
species (8). Other organisms, such as algae and plants, do not seem to play a very
important role in the biodegradation of hydrocarbons. It is a truism that the hall-
mark of an oil-degrading organism is its ability to insert oxygen atoms into the
hydrocarbon, and there are many ways in which this is achieved. Figures 2 and 3
show the most well-studied. Once a hydrocarbon possesses a carboxylate or alco-
hol functionality it is almost invariably a readily degradable compound. A simple
example at the human level is the difference between oleic acid, a high calorie
food, and octadecane, present in mineral oil, which is so inert that it serves as
an intestinal lubricant!
For many years it was assumed that oil biodegradation was an exclusively
aerobic process, since any degradation must involve oxidation. Indeed the very
existence of oil reservoirs indicates that anaerobic degradative processes in
such environments must be very slow. Nevertheless, in recent years it has
become clear that at least some hydrocarbons are oxidized by bacteria under
completely anaerobic conditions, where the oxygen is probably coming from
water. Limited hydrocarbon biodegradation has now been shown under
sulfate-, nitrate-, carbon dioxide- and ferric iron-reducing conditions (Table 3).
The phenomenon is still poorly understood, however, and at present the largest
molecules demonstrated to undergo biodegradation under these conditions are
hexadecane, heptadecene, and phenanthrene. The pathways of degradation are
only beginning to be addressed. Figure 4 shows the intermediates identified in
anaerobic toluene degradation in different organisms. It is noteworthy that
Vol. 1 BIOREMEDIATION 929

OH OH
OH OH OH
O

O⫺
O
Naphthalene
HO OH

OH

OH

HO

HO
Toluene

Fig. 3. Initial steps in the aerobic degradation of naphthalene, as a representative


multiringed aromatic, and toluene. The different initial steps of toluene degradation
are examples of the diversity found in different organisms.

while organisms capable of aerobic oil biodegradation seem to be ubiquitous,


organisms capable of the anaerobic degradation of hydrocarbon have to date
only been found in a few places.
Although the majority of molecules in crude oils and refined products are
hydrocarbons, the U.S. Clean Air Act amendment of 1990 mandated the addition
of oxygenated compounds to gasoline in many parts of the United States. The

Table 3. Hydrocarbons that have been Shown to be


biodegraded Under Anaerobic Conditions
Electron acceptor Substrate
nitrate (to nitrogen) heptadecene
toluene, ethylbenzene, xylene
naphthalene
terpenes
iron(III) (to iron(II)) toluene
manganese(IV) (to Mn(II)) toluene
sulfate (to sulfide) hexadecane, alkylbenzenes
benzene
naphthalene, phenanthrene
CO2 (to methane) toluene, xylene
930 BIOREMEDIATION Vol. 1

HO

Benzyl alcohol

OH

para-Cresol

CoA S

β-PhenylpropionylCoA

Fig. 4. Proposed initial steps in the anaerobic biodegradation of toluene in different


organisms.

requirement is usually that 2% (w/w) of the fuel be oxygen, which requires that
5–15% (v/v) of the gasoline be an oxygenated additive (eg, methanol, ethanol,
methyl tert-butyl ether (MTBE), etc). Although methanol and ethanol are readily
degraded under aerobic conditions, the degradability of MTBE remains some-
thing of an open question. The compound was previously very rare in the envir-
onment, but now it is one of the major chemicals in commerce. At first it seemed
that the compound was completely resistant to biodegradation, but complete
mineralization has now been reported (9). Whether biodegradation can be opti-
mized for effective bioremediation remains to be seen.
Bioremediation. Crude oil and refined products are readily biodegradable
under aerobic conditions, but they are only incomplete foods since they lack any
significant nitrogen, phosphorus, and essential trace elements. Bioremediation
strategies for removing large quantities of hydrocarbon must therefore include
the addition of fertilizers to provide these elements in a bioavailable form.
Air. Hydrocarbon vapors in air are readily treated with biofilters. These
are typically rather large devices with a very large surface area provided by
bulky material such as a bark or straw compost. The contaminated air, perhaps
from a soil vapor-extraction treatment, or from a factory using hydrocarbon sol-
vents, is blown through the filter, and organisms, usually indigenous to the filter
material or provided by a soil or commercial inoculum, grow and consume the
hydrocarbons. Adequate moisture must be maintained for effective operation.
Alternatively, trickling biofilters with recycled water are also in use. Both bac-
teria and fungi readily colonize such filters, and they can be very effective.
Nevertheless, biofilters are usually equipped with a small granulated activated
carbon ‘‘backup’’ filter to handle any sudden pulse loads that might overwhelm
the biological capacity of the filter. Biofilters compete with granulated activated
carbon filters, and are often cheaper because they minimize the cost of the granu-
lated activated carbon, and the energy required to destroy the contaminant and
the granulated activated carbon when the latter is saturated (10). Potential
Vol. 1 BIOREMEDIATION 931

problems include plugging and uneven air or water flow, but successful designs
work for many years with minimal maintenance except the occasional addition of
nutrients and stirring of the bed.
Sea. Crude oil spills at sea are perhaps the most widely covered environ-
mental incidents in the national and international media. Despite their notori-
ety, catastrophic tanker spills and well blow-outs are fortunately rather rare,
and their total input into the world’s oceans is approximately equivalent to
that from natural seeps; significantly more oil reaches the world’s oceans
from municipal sewers (11). Physical collection of the spilled oil is the preferred
remediation option, but if skimming is unable to collect the oil, biodegradation
and perhaps combustion or photooxidation are the only routes for elimination of
the spill. One approach to stimulating biodegradation is to disperse the oil with
chemical dispersants. Early dispersants had undesirable toxicity, but modern
dispersants and application protocols can stimulate biodegradation by increas-
ing the surface area of the oil available for microbial attachment, and perhaps
providing nutrients to stimulate microbial growth (12). Patents have been
issued for dispersant formulations that specifically include nitrogen and
phosphorus nutrients (13), but the products are not currently commercially
available.
Bioremediation by the addition of oil-degrading microbes is often promoted
as a treatment option for floating spills, but this approach has not yet met with
any documented success (13).
Shorelines. The successful bioremediation of shorelines affected by the
spill from the Exxon Valdez in Prince William Sound, Alaska, was perhaps the
largest bioremediation project to date (14,15). More than 73 miles of shoreline
were treated in 1989 and similar amounts of fertilizer were used in 1990. Oil
had typically penetrated into the surface gravel on these shorelines, occasionally
getting as deep as 30 cm into the sediment. Since the gravel was typically very
permeable, oxygen availability was unlikely to be the limiting factor for biodegra-
dation, and indeed this was subsequently shown to be correct. Bioremediation
thus focused on the addition of nitrogen and phosphorus fertilizers to partially
remove the nutrient-limitation on oil degradation. Of course the addition of fer-
tilizers was complicated by the fact that oiled shorelines were washed by tides
twice a day. These tides would have rapidly removed any soluble fertilizer, so
a strategy was sought that would provide nutrients for a significant length of
time. Various approaches to applying fertilizers were tried, including both stan-
dard and slow release nutrients, oleophilic nutrients and solutions of liquid fer-
tilizers. Two fertilizers were used in the full-scale applications; one, an oleophilic
product known as Inipol EAP22 (trademark of CECA, Paris, France), was a
microemulsion of a concentrated solution of urea in an oil phase of oleic acid
and trilaurethphosphate, with butoxyethanol as a cosolvent. This product was
designed to adhere to oil, and to release its nutrients to bacteria growing at
the oil-water interface. The other fertilizer was a slow-release formulation of
inorganic nutrients, primarily ammonium nitrate and ammonium phosphate,
in a polymerized vegetable oil skin. This product, known as Customblen (trade-
mark of Grace-Sierra, Milpitas, California), released nutrients with every tide,
and these were distributed throughout the oiled zone as the tide fell. Fertilizer
application rates were carefully monitored so that the nutrients would cause no
932 BIOREMEDIATION Vol. 1

harm, and the rate of oil biodegradation was stimulated between two- and five-
fold (14,15).
A wide range of fertilizers, including agricultural and horticultural fertili-
zers, and bone and fish meals have been tried at the pilot scale, usually with
at least modest success (13). Some current work is aimed at addressing whether
providing a readily degradable substrate with the fertilizer nutrients helps
immobilize the nutrients in biomass at the oiled site. Of course nutrient-
supplementation is only likely to markedly stimulate the rate of biodegradation
where nutrient levels are naturally low. It is unlikely that fertilizers will have a
dramatic effect in situations where agricultural or municipal run-off maintains
elevated levels of nutrients, such as happens in some estuaries and bays. Here
aeration is likely to be most effective.
Bioremediation by the addition of oil-degrading microbes has been pro-
moted as a treatment option for oiled shorelines, but this approach has not yet
met with any documented success (13).
Areas where there are currently few remediation options include oiled
marshes, mangroves, and coral reefs. These environments are generally easily
damaged by human intrusion and physical cleaning options may not provide
any net environmental benefit. Bioremediation may provide some attractive
options, and some success has been claimed, on a small scale, with fertilizer
applications (13). Marshes and mangroves offer the additional complication
that they are typically anoxic. Perhaps the anaerobic degradation of oil could
be stimulated by inoculation with anaerobic hydrocarbon degrading microbes,
or perhaps gentle aeration or the addition of slow release oxygen compounds,
such as some inorganic peroxides, might stimulate aerobic degradation without
significantly changing the redox balance of these environments. This is an area
where research is very much in its infancy and there are no well-documented
success stories to date.
Bioremediation also offers options for dealing with oiled material, such as
seaweed, that gets stranded on shorelines; composting has been shown to be
effective.
Groundwater. Spills of refined petroleum product on land, and leaking
underground storage tanks, sometimes contaminate groundwater. Bioremedia-
tion is becoming an increasingly popular treatment for such situations.
Hydrocarbons typically have a specific gravity of less than 1, and refined
products usually float on the water table if they penetrate soil that deeply. In
the parlance of the remediation industry, such floating spills are often called
NAPLs (nonaqueous phase liquids). Indeed they are sometimes known as
LNAPLs for light nonaqueous phase liquids, to distinguish them from more
dense materials, such as halogenated compounds, which are more likely to
sink in groundwater. Stand-alone bioremediation is an option for these situa-
tions, but ‘‘pump and treat’’ is the more usual treatment. Contaminated water
is brought to the surface, free product is removed by flotation, and the cleaned
water re-injected into the aquifer or discarded. Adding a bioremediation compo-
nent to the treatment, typically by adding oxygen and low levels of nutrients, is
an appealing and cost-effective way of stimulating the degradation of the resi-
dual hydrocarbon not extracted by the pumping. This approach is becoming
widely used.
Vol. 1 BIOREMEDIATION 933

Benzene Toluene Ethylbenzene

o-Xylene m-Xylene p-Xylene

Fig. 5. The components of BTEX.

Hydrocarbons are not very soluble in water, but the most soluble compo-
nents leach out of a spill if there is continual flushing. Typically only small aro-
matic molecules, the infamous BTEX (benzene, toluene, ethylbenzene, and
xylenes, Fig. 5), are soluble enough to contaminate groundwater. Although
with the advent of oxygenated gasolines, it is expected that these oxygenates
(ethanol, methanol, MTBE (methyl-tert-butyl ether) etc) will also be found in
groundwater. In the past, remediation of such situations has usually used
pump-and-treat methodologies. These methods are slow and may leave reser-
voirs of contaminants in pockets that are poorly connected to the main water
body. Of course the contaminant is biodegradable, and some biodegradation is
probably already occurring when the contamination is discovered. The cheapest
approach to remediation is, thus, to allow this intrinsic process to continue.
Evidence that it is indeed occurring can be found in the selective disappearance
of the most biodegradable compounds in the contaminant mixture, and the con-
comitant disappearance of electron acceptors from the groundwater. Thus oxy-
gen is depleted as the preferential terminal electron acceptor for metabolism,
followed by nitrate, ferric iron, sulfate, and finally CO2 for methanogenesis (16).
Intrinsic bioremediation is becoming an acceptable option in locations
where the contaminated groundwater poses little threat to environmental
health. Nevertheless, although intrinsic bioremediation is appealingly simple,
it may not be the lowest cost option if there are extensive monitoring and docu-
mentation costs involved for several years. In such cases it may well be more cost
effective to optimize conditions for biodegradation.
One approach is to optimize the levels of electron acceptors. Oxygen can be
pumped in as the pure gas, or as air, although this is relatively energy intensive
since oxygen is so poorly soluble. Hydrogen peroxide has been used in some situa-
tions, but there have been problems with biomass plugging near the injection
wells. Slow release formulations of inorganic peroxides, such as magnesium per-
oxide, have recently been used with success (17). Nitrate may be added, although
there are sometimes regulatory limitations on the amount of this material that
may be added to groundwater (18). Ferric iron availability may be manipulated
by adding ligands (19).
934 BIOREMEDIATION Vol. 1

If there are significant amounts of both volatile and nonvolatile contami-


nants, remediation may be achieved by a combination of liquid and vapor extrac-
tion of the former, and bioremediation of the latter. This combination has been
termed ‘‘bioslurping’’, where the act of pumping out the liquid contaminant
phase draws in air at other wells to stimulate aerobic degradation (20). Such
bioremediation requires that there be enough nutrients to allow microbial
growth, and fertilizer nutrients are frequently added at the air injection wells.
Bioslurping has had a number of well-documented successes.
The majority of remediation operations include stopping the source of the
contamination, but in some cases this is impossible, either because of the location
of the spill, or because it is over a large area, and not a point source. In these
situations it may be possible to intercept the flow of contaminated groundwater
off-site, and ensure that no contamination passes. The simplest intervention is a
line of wells for pump and treat, but including a biological component may be
more cost effective. This can range from the installation of a sparge line for aer-
ating the contaminated plume, to installing some form of semicontained bioreac-
tor where nutrients can be applied with some modicum of control. Often these
designs are combined with barriers to ensure that all the contaminated plume
passes through the reactive zone. These designs have a variety of names, includ-
ing biowall, trench biosparge, funnel and gate, bubble curtain, sparge curtain
and engineered trenches and gates. Both aerobic and anaerobic designs have
been successfully installed.
Where there are large volumes of contaminated water under a small site,
it is sometimes most convenient to treat the contaminant in a biological reactor
at the surface. Considerable research has gone into reactor optimization for dif-
ferent situations and a variety of stirred reactors, fluidized-bed reactors, and
trickling filters have been developed. Such reactors are usually much more effi-
cient than in situ treatments, although correspondingly more expensive.
Of course the presence of a liquid phase of hydrocarbon in a soil gives rise to
vapor contamination in the vadose zone above the water table. This can be trea-
ted by vacuum extraction, and the passage of the exhaust gases through a biofil-
ter (see above) can be a cheap and effective way of destroying the contaminant
permanently.
Soil. Hydrocarbon contamination of soils runs the gamut from crude oils
at production well and pipeline spills, to the full slate of refined products at refi-
neries, distribution centers, service stations and accident sites. Significant
hydrocarbon contamination is also often found at manufactured gas plants,
now mainly abandoned, wood treatment facilities, railroad rights of way and
terminals, and various military bases. Sometimes the contamination is the result
of leaking underground storage tanks and pipelines, leading to subsurface con-
tamination, but surface spills also occur. Physical removal of gross contamina-
tion is an obvious first step at all locations, and bioremediation is an appealing
option for remediating residual contamination in many of these sites.
Spills from production facilities and pipelines often involve both oil and
brine, since most oil reservoirs float on top of concentrated brines, and both
are produced in later stages of production. The brine is typically separated
from the oil and re-injected into the reservoir, but some is retained in many
production pipelines. The environmental impact of spilled brine can be quite
Vol. 1 BIOREMEDIATION 935

deleterious. Not only is salt toxic to most plants, and can inhibit many soil
bacteria, but it also can have a major effect on the soil structure by altering
the physical properties of clays. Successful bioremediation strategies must there-
fore include remediating the brine. In wet regions the salt is eventually diluted
by rainfall, but in arid regions, and to speed the process in wetter regions,
gypsum is often added to restore soil porosity.
Many hydrocarbons bind quite tightly to soil components, and are thereby
less available to microbial degradation. The kinetics of binding seem to be com-
plex, and the process of ‘‘aging’’ is only poorly understood. Nevertheless, it seems
clear that hydrocarbons that have been in contact with soil for a long time are not
as available for biodegradation as fresh spills. Several groups of researchers have
suggested the addition of surfactants to overcome this limitation, but this
approach is not yet widely used. A significant potential concern is that the sur-
factant will be degraded in preference to the contaminant of concern.
Intrinsic biodegradation occurs, but it usually only removes the lightest
refined products, such as gasoline, diesel and jet fuel. Active intervention is
typically required. Usually the least expensive approach is in situ remediation,
typically with the addition of nutrients, and the attempted optimization of
moisture and oxygen by tilling. Various approaches to applying fertilizers have
been tried, including both standard and slow-release nutrients, oleophilic nutri-
ents and solutions of liquid fertilizers. Oxygen is a likely limiting nutrient in
many cases, and soil tilling is widely practiced. This in situ bioremediation of
hydrocarbon-contaminated soils is akin to the old practice of ‘‘land-farming’’,
wherein sludges and other refinery wastes were deliberately spread onto soil
and tilled and fertilized to stimulate biodegradation. Although this practice is
now discontinued in the United States, it was quite widely used.
Deeper contamination may be remedied with bioventing, where air is
injected through some wells, and extracted through others to both strip volatiles
and provide oxygen to indigenous organisms. Fertilizer nutrients may also be
added. This is usually only a viable option with lighter refined products.
A recent suggestion has been to use plants to stimulate the microbial degra-
dation of the hydrocarbon (hydrocarbon phytoremediation). This has yet to
receive clear experimental verification, but the plants are proposed to help deli-
ver air to the soil microbes, and to stimulate microbial growth in the rhizosphere
by the release of nutrients from the roots. The esthetic appeal of an active phy-
toremediation project can be very great.
When soil contamination extends to some depth it may be preferable to
excavate the contaminated soil and put it into ‘‘biopiles’’ where oxygen, nutrient
and moisture levels are more easily controlled. Biopiles can also be kept warm
during winter months, increasing the amount of time available for biodegrada-
tion in colder climates. Since the soil is well mixed during the construction of
the pile, there is an opportunity to add selected microbial and fungal strains in
an additional attempt to maximize biodegradation.
Composting by the addition of readily degradable bulking agents is also a
useful option for relatively small volumes of excavated contaminated soil.
Since efficient composting invariably involves self-heating as biodegradation pro-
ceeds, this also offers an option for extending the bioremediation season into the
winter months in cold climates. A potential drawback of composting is that it
936 BIOREMEDIATION Vol. 1

usually increases the volume of contaminated material, but if fully successful the
finished compost can be returned to the site as a positive contribution to soil
quality.
Slurry bioreactors offer the most aggressive approach to maximizing con-
tact between the contaminated soil and the degrading organisms. Both lagoons
and reactor vessels have been used, but the former are often not optimally
designed for all the soil to be partially suspended by the mixing impellers. Con-
tained reactor designs include mixing tank, airlift, and fluidized-bed aeration. A
major advantage of contained slurry bioreactors is the potential ability to opti-
mize nutrients, aeration and degradative inocula as fresh soil is added, and
the control of waste materials, including gases continually. Slurry bioreactors
are usually the most expensive bioremediation option because of the large
power requirements, but under some conditions this cost is offset by the rapid
biodegradation that can occur.
In all these cases it is important to bear in mind that although the majority
of hydrocarbons are readily biodegraded, some, such as the steranes and
hopanes, are very resistant to microbial attack. Estimates of oil biodegradation
range from 60–95% for different crude oils, so fresh spills of crude oils are readily
treated by bioremediation (21). Refined products, such as gasoline, diesel, jet
fuels, and heating oils are usually more biodegradable than typical whole oils,
but the various heavy fractions of crude oils, such as the asphalts, are far less
biodegradable, and are not such attractive targets for bioremediation. Some
crude oils have already been extensively biodegraded in their reservoirs, and
these are also poor targets for bioremediation. An example is Orimulsion, a
heavy oil in water emulsion (70% bitumen) stabilized by low levels of surfactants,
used as a fuel for electricity generation. Similarly, old spills may have already
undergone significant biodegradation and the residue may be relatively biologi-
cally inert. It is thus important to run laboratory studies to ensure that the con-
taminant is sufficiently biodegradable that clean-up targets can be met.
3.2. Halogenated Organic Solvents. Constituents. Halogenated
organic solvents are widely used in metal processing, electronics, dry cleaning
and paint, paper and textile manufacturing, and some representative examples
are shown in Figure 6. These solvents have been used for more than fifty years,
and unfortunately they are fairly widespread contaminants. Unlike the hydro-
carbons, which usually float on water, the halogenated solvents typically have
specific gravities greater than 1, and they generally sink to the bottom of any
groundwater, and float on the bedrock. For this reason they are sometimes
known as DNAPLs for dense nonaqueous phase liquids.
Biodegradation. Halogenated solvents are degraded under aerobic and
anaerobic conditions. The anaerobic process is typically a reductive dechlorina-
tion that progressively removes one halide at a time (Fig. 7). For example, under
methanogenic conditions, carbon tetrachloride is sequentially dechlorinated to
chloroform, dichloromethane, methyl chloride and methane, while trichloroethy-
lene is sequentially reduced to ethylene. Some of these compounds are also deha-
logenated under sulfate-reducing conditions, and under denitrifying conditions
there are reports that the final product can be CO2 (22). Chloromethane and
dichloromethane have been shown to be the sole carbon source for several anae-
robic organisms (23), and it seems there is much to be learned about the
Vol. 1 BIOREMEDIATION 937

Cl
Cl Cl
Cl Cl
Cl Cl Cl

Carbon tetrachloride Tetrachloroethylene


(perchloroethylene)

Cl
Cl Cl
Cl
Cl Cl

1,1,1-Trichloroethane Trichloroethylene

Cl

Vinyl chloride

Fig. 6. Some representative halogenated solvents.

microbial diversity of anaerobic microorganisms capable of dechlorinating


solvents.
The simplest chlorinated alkanes, alkenes, and alcohols (eg, chloromethane,
dichloromethane, chloroethane, 1,2-dichloroethane, vinyl chloride, and 2-chloro-
ethanol) serve as substrates for aerobic growth for some bacteria, but the major-
ity of halogenated solvents cannot support growth (24). Nevertheless these com-
pounds are mineralized under aerobic conditions, albeit with no apparent benefit
to the degrading organism. Indeed, the oxidation appears to be fortuitous and it
occurs during the metabolism of a growth substrate. The phenomenon is there-
fore known as co-metabolism or co-oxidation. Numerous bacteria are able to cat-
alyze the oxidation of trichloroethylene; some use monooxygenases (for example
methane and ammonia oxidizing species); others contain dioxygenases (eg, some
toluene oxidizing species). The difference between these two classes of enzymes is
the fate of the two atoms of molecular oxygen. Monooxygenases insert one oxygen

Cl Cl Cl Cl Cl Cl Cl
CH4
Cl Cl Cl

H2 HCl H2 HCl H2 HCl H2 HCl

Carbon tetrachloride

Cl Cl Cl Cl Cl Cl Cl

Cl Cl Cl

H2 HCl H2 HCl H2 HCl H2 HCl

Tetrachloroethylene

Fig. 7. Reductive dechlorination of carbon tetrachloride and tetrachloroethylene.


938 BIOREMEDIATION Vol. 1

Cl Cl Cl O Cl

Cl Cl
Monooxygenase

Cl Cl OH OH
Cl Cl
Cl Cl Cl
Dioxygenase

Fig. 8. Aerobic activation of tetrachloroethylene.

atom into their substrate, and reduce the other to water. Dioxygenases insert
both atoms into their substrate. The effect of these two types of enzyme is
illustrated in Figure 8. In either case, biodegradation proceeds to complete
mineralization (25).
The biodegradation of trichloroethylene is the most studied since this
is probably the most widespread halogenated solvent contaminant. Several
substrates drive trichlorethylene co-oxidation, including methane, propane, pro-
pylene, toluene, isopropylbenzene, and ammonia (25). The enzymes that metabo-
lize these substrates have subtly different selectivities with regard to the
halogenated solvents, and to date none are capable of co-oxidizing carbon tetra-
chloride or tetrachloroethylene. Complete mineralization of these compounds
can, however, be achieved by sequential anaerobic and aerobic process.
Bioremediation. Air. Biofilters are an effective way of dealing with air
from industrial processes that use halogenated solvents such chloromethane,
dichloromethane, chloroethane, 1,2-dichloroethane and vinyl chloride, that sup-
port aerobic growth (26). Both compost-based dry systems and trickling filter wet
systems are in use. Similar filters could be incorporated into pump-and-treat
operations.
Groundwater and Soil. Halogenated solvents have contaminated soils
and groundwater throughout the industrialized world, and remediation has a
high priority. Since the solvents are so dense, they are typically found on the bed-
rock underlying aquifers. Pumping out the liquid phase is an obvious first step if
the contaminant is likely to be mobile, but in situ bioremediation is a promising
option. Thus, the U.S. Department of Energy is investigating the use of anaero-
bic in situ degradation of carbon tetrachloride in an aquifer some 76 m below
the Hanford, Washington site, with nitrate as electron acceptor, and acetate as
electron donor (22).
Trichloroethylene is the most frequent target of remediation, and as dis-
cussed above, this is only metabolized cometabolically. Remediation operations
thus incorporate the addition of the cometabolized substrate. Methane was
used successfully at the U.S. Department of Energy site at Savannah River,
near Aiken, South Carolina (27) which had both an air-stripping and a biological
component. Horizontal wells were used to pump methane and air below the
contaminant, while an upper horizontal well in the vadose zone was used to
withdraw these gases through the contaminated zone. Optimum biodegradation
performance seemed to come from alternate injection of air and methane in air,
Vol. 1 BIOREMEDIATION 939

and the inclusion of nitrous oxide and triethylphosphate, both gases, to give a
C:N:P ratio of 100:10:1.
Plants may have a role to play in enhancing microbial biodegradation
of halogenated solvents, for it has recently been shown that mineralization of
radiolabelled trichlorothylene is substantially greater in vegetated rather than
unvegetated soils (28), indicating that the rhizosphere provides a favorable
environment for microbial degradation of organic compounds.
Methane has also been used in aerobic bioreactors that are part of a pump-
and-treat operation, and toluene and phenol have also been used as co-substrates
at the pilot scale (29). Anaerobic reactors have also been developed for treating
trichloroethylene. For example, Wu and co-workers (30) have developed a suc-
cessful upflow anaerobic methanogenic bioreactor that converts trichloroethy-
lene and several other halogenated compounds to ethylene.
Groundwater contaminated with other halogenated solvents can also be
treated in aboveground reactors. Aerobic reactors are useful for those compounds
that can support growth. For example, a membrane reactor has been designed
for treating 1,2-dichloroethane (31), and bubble columns and packed bed reactors
have been developed for the aerobic degradation of 2-chloroethanol (32). As men-
tioned above, sequential anaerobic and aerobic reactors are capable of mineraliz-
ing tetrachloroethylene (33).
3.3. Halogenated Organic Compounds. Constituents. Complex
halogenated organic compounds have been widely used in commerce in the last
fifty years. A few representative examples are shown in Figure 9; pentachlorophe-
nol has been widely used as a wood preservative, and also for termite control.

OH
Cl Cl Cln Cln

Cl Cl
Cl
Polychlorinated biphenyl
Pentachlorophenol

O OH OH
O Cl Cl
O– Cl

Cl Cl
Cl Cl
Cl
Hexachlorophene
(2,4-Dichlorophenoxy)acetate

Cl
Cl Cl
Cl Cl

DDT

Fig. 9. Representative halogenated organic contaminants.


940 BIOREMEDIATION Vol. 1

(2,4-Dichlorophenoxy)acetate (2,4-D) is widely used as a broad-leaf herbicide,


DDT was widely used as an insecticide, and hexachlorophene has been widely
used as a germicide. Polychlorinated biphenyls (PCBs) were sold with varying
levels of chlorination for a range of purposes. They ranged from light oily fluids
(with two, three, or four chlorines) to viscous oils (five chlorines) to greases and
waxes (six or more chlorines), and their names indicated the level of chlorination.
Thus Aroclor 1242 (trademark of Monsanto, U.S.), Clophen A30 (trademark
of Farbenfabriken Bayer AG, Germany) and Kanechlor 300 (trademark of
Kane-gafuchi Chemical Industries, Japan) all contained 42% chlorine by weight,
and an average of three chlorines per biphenyl. An important property shared by
all these compounds is their relative resistance to biodegradation, so at first
glance they may not seem a good target for bioremediation. Indeed complex
halogenated organic compounds were widely thought to be almost exclusively
anthropogenic in origin, so that there would have been little time for biodegrada-
tion pathways to evolve. This view is being corrected, for in fact a variety of organ-
isms, particularly marine algae and some fungi, produce significant quantities of
these compounds (34). There is, thus, good reason to expect that halogenated-
organic degrading organisms will be found in the biosphere, and this has been
borne out in practice.
Biodegradation. An important characteristic of degradation is the clea-
vage of carbon–chlorine bonds, and the enzymes that catalyze these reactions,
the dehalogenases, are being characterized (35). The reductive dechlorination
seen with carbon tetrachloride and tetrachloroethylene (see Fig. 7) seems
to be a general phenomenon, and even compounds as persistent as DDT
and the polychlorinated biphenyls are reductively dechlorinated under some
conditions, particularly under methanogenic conditions. Some compounds,
such as pentachlorophenol, can be completely mineralized under anaerobic
conditions, but the more recalcitrant ones require aerobic degradation after
reductive dehalogenation.
Pentachlorophenol can be mineralized aerobically and anaerobically, and
both processes have been exploited for bioremediation. Under methanogenic con-
ditions a reductive dehalogenation, analogous to that seen with halogenated sol-
vents, eventually generates phenol, although some of the intermediate congeners
are quite recalcitrant (36). The phenol is further mineralized to methane and car-
bon dioxide, and sulfate-reducing bacteria may be involved (37).
Several bacterial isolates are able to grow aerobically using pentachloro-
phenol as sole source of carbon and energy, but many grow rather better when
supplemented with a more nutritions substrate. Unfortunately, in many cases it
seems that the more vigorous growth with these substrates does not enhance the
biodegradation of pentachlorophenol (38). The initial step in pentachlorophenol
biodegradation in one Flavobacterium species is an NADPH-dependent oxygeno-
lytic dechlorination, where the para-chloro group is replaced by a hydroxyl to
generate tetrachlorohydroquinone. Other species seem to produce the same
metabolite by a hydrolytic process (38). Fungal degradation of pentachlorophe-
nol, apparently using the lignolytic apparatus, has also been reported (39).
(2,4-Dichlorophenoxy)acetate (2,4-D) has been one of the world’s most pop-
ular herbicides. Although it is somewhat resistant to biodegradation, it is biode-
graded by several bacterial isolates. It is a general truism that the more halogens
Vol. 1 BIOREMEDIATION 941

on a molecule, the slower its biodegradation, and this is borne out with the
related herbicide 2,4,5-T ((2,4,5-trichlorophenoxy)acetate). Nevertheless, bacter-
ial degradation has been seen under both aerobic and anaerobic conditions, the
latter involving reductive dechlorination via 2,4-D. Aerobic degradation removes
acetate from (2,4-dichlorophenoxy)acetate to yield 2,4-dichlorophenol, which is
subsequently hydroxylated to 3,5-dichlorocatechol, followed by ring cleavage
and complete mineralization (40). Genetic engineering has been used to con-
struct strains that are particularly adept at consuming these compounds, but
whether these will overcome the regulatory hurdles to allow their use outside
the laboratory remains to be seen.
DDT (1,1,1-trichloro-2,2-bis(p-chlorophenyl)ethane) is a remarkably resis-
tant molecule, which explains both its efficacy as an insecticide, and its accumu-
lation at the top of the food-chain. Nevertheless, there are indications that it can
be biodegraded, both anaerobically with initial dechlorination and aerobically
with initial ring hydroxylation (40). DDT and its partially degraded congeners
are very hydrophobic, and biodegradation seems to be stimulated by adding sur-
factants. White-rot fungi also degrade DDT under lignolytic conditions, although
there is little mineralization to CO2 (5).
The polychlorinated biphenyls are quite recalcitrant. Some lightly chlori-
nated biphenyls are readily mineralized under aerobic conditions (41), and
indeed the structure of an enzyme that catalyzes the key ring-cleavage oxygena-
tion has recently been determined by x-ray crystallography (42). More chlori-
nated congeners are resistant to aerobic degradation, but they are reductively
dechlorinated under anaerobic conditions. Complete degradation of the commer-
cial mixtures, thus, generally requires an anaerobic process followed by an aero-
bic one. A major issue for the oxidative process is that biphenyl seems to be
required for significant expression of the biodegradative system (41); the chlori-
nated compounds do not induce the enzymes that would degrade them. The bio-
chemistry of the biodegradative process is only beginning to be unraveled, but
already there are suggestions that there is considerable diversity in the enzyme
systems able to degrade these compounds. There is a lot of effort aimed at engi-
neering organisms to degrade polychlorinated biphenyls, and in finding ways to
make these very hydrophobic compounds more bioavailable by the use of chemi-
cal oxidants, such as Fenton’s reagent (43), or surfactants.
Bioremediation. Soil. Pentachlorophenol has been the target of biore-
mediation at a number of wood-treatment facilities, and good success has
been achieved in several applications. It is rarely the sole contaminant, and
is often present with polynuclear hydrocarbons from coal creosote. In situ
degradation has been stimulated by bioventing, where air is injected through
some wells, and extracted through others to both strip volatiles and provide
oxygen to indigenous organisms (44). Just as with the halogenated solvents,
it seems that plants stimulate microbial degradation of pentachlorophenol in
the rhizosphere (45).
The kinetics of such in situ degradation are rather slow, however, and more
active bioremediation is usually attempted. For example, contaminated soil at
the Champion Superfund site in Libby, Montana, was placed into 1-acre land
treatment units in 6-in. layers, and irrigated, tilled, and fertilized. Under
these conditions, the half-lives of pentachlorophenol, pyrene, and several other
942 BIOREMEDIATION Vol. 1

polynuclear aromatic hydrocarbons, initially present at around 100–200 ppm,


were on the order of 40 days (46). This success relied on the indigenous microbial
populations in the soil, but many groups are focusing on the addition of organ-
isms (eg, 47), perhaps immobilized on some sort of carrier. Composting, and bior-
emediation focusing on the use of white-rot fungi, has also met with success at
the pilot scale. Others have used fed-batch or fluidized-bed bioreactors to stimu-
late the biodegration of pentachlorophenol. This allows significant optimization
of the process and increases in rates of degradation by tenfold (48).
A major concern when remediating wood-treatment sites is that pentachlor-
ophenol was often used in combination with metal salts, and these compounds,
such as chromated copper–arsenate, are potent inhibitors of at least some pen-
tachlorophenol degrading organisms (49). Sites with significant levels of such
inorganics may not be suitable candidates for bioremediation.
The phenoxy-herbicide, 2,4-D, has been successfully bioremediated in a soil
contaminated with such a high level of the compound (710 ppm) that it was toxic
to microorganisms (50). There were essentially no indigenous bacteria in the soil.
Success relied on washing a significant fraction of the contaminant off the soil
and adding bacteria enriched from a less contaminated site. Success was
achieved in remediating both soil washwater and soil in a bioslurry reactor
(50). 2,4-D is also effectively degraded in composting, with about half being com-
pletely mineralized, and the other half becoming incorporated in a nonextract-
able form in the residual soil organic matter (51).
The bioremediation of polychlorinated biphenyls in soils in receiving signif-
icant attention because these compounds are quite widely distributed in the
environment, either from leaking electrical transformers or sometimes because
they were applied as part of road maintenance. In the latter case, the contami-
nation usually includes petroleum hydrocarbons, and unfortunately it seems
that the two contaminants inhibit the degradation of each other. Nevertheless,
cultures are being found that can degrade both polychlorinated biphenyls and
petroleum hydrocarbons. There is also interest in the role of rhizosphere organ-
isms in polychlorinated biphenyl degradation, particularly since some plants
exude phenolic compounds into the rhizosphere that can stimulate the aerobic
degradation of the less chlorinated biphenyls.
Groundwater. A successful groundwater bioremediation of pentachloro-
phenol is being carried out at the Libby Superfund site described above. A shallow
aquifer is present at 5.5 to 21 m below the surface, and a contaminant plume
is nearly 1.6 km in length. Nutrients and hydrogen peroxide were added at the
source area and approximately half way along the plume, and pentachlorophenol
concentrations decreased from 420 ppm to 3 ppm where oxygen concentrations
were successfully raised. A membrane oxygen dissolution system has now been
installed to replace the hydrogen peroxide additions, and costs have been substan-
tially lowered without an apparent decrease in remediation performance (52).
Pentachlorophenol is readily degraded in biofilm reactors (53), so bioreme-
diation is a promising option for the treatment of contaminated groundwater
brought to the surface as part of a pump-and-treat operation.
River and Pond Sediments. Much of the work on polychlorinated biphe-
nyls has focused on the remediation of aquatic sediments, particularly from
rivers, estuaries, and ponds. As noted above, a few of the most lightly chlorinated
Vol. 1 BIOREMEDIATION 943

compounds are mineralized under aerobic conditions, but the more chlorinated
species seem completely resistant to aerobic degradation, even by white rot
fungi (54). On the other hand, there is extensive dechlorination of highly
chlorinated forms under anaerobic conditions, particularly methanogenic
conditions. Bioremediation thus requires anaerobic and aerobic regimes.
Intrinsic biodegradation of polychlorinated biphenyls can be recognized by
the changing ‘‘fingerprint’’ of the individual isomers as biodegradation pro-
ceeds (41, 55). The anaerobic dechlorination of the most recalcitrant congeners
can apparently be primed by adding a readily dehalogenated congener, such as
2,5,30 ,40 -tetrachlorobiphenyl (56), but whether this is a realistic approach for in
situ bioremediation remains to be seen. Harkness and co-workers (57) have
successfully stimulated aerobic biodegradation in large caissons in the Hudson
River by adding inorganic nutrients, biphenyl, and hydrogen peroxide, but
found that repeated addition of a polychlorinated-biphenyl degrading bacter-
ium (Alcaligenes eutrophus H850) had no beneficial effect. Essentially no bio-
degradation occurred in the stirred control caissons, but losses on the order of
40% were seen in the caissons that received nutrients and peroxide, regardless
of whether the stirring was aggressive or rather gentle. Whether this approach
can be scaled-up for large-scale use, with a net environmental benefit, remains
to be seen.
3.4. Nonchlorinated Pesticides and Herbicides. Constituents.
Pesticides A vast number of compounds are used as herbicides, pesticides, fungi-
cides, etc, and a few are shown in Figure 10. In order to be effective they must
have their effect before they are degraded in the environment, but on the other
hand they must not be so resistant to degradation that they accumulate where
they are used, or accumulate in food chains. It is unusual for these compounds
to become contaminants where they are applied correctly, but manufacturing
facilities, storage depots and rural airfields where crop-dusters are based have
had spills that can lead to long lasting contamination. Bioremediation is a
promising technology to remediate such sites. There are also some locations
were groundwater has become contaminated by these chemicals, and again
bioremediation may be a cost-effective remediation strategy.
Biodegradation. The vast majority of pesticides, herbicides, fungicides,
and insecticides in use today are biodegradable, although the intrinsic biode-
gradability of individual compounds is one of the variables used in deciding
which compound to use for which task. Some herbicides are acutely toxic to
plants that absorb them, but are so readily degraded in soil that seeds can be
planted at the same time as the herbicide application. Other herbicides are
known to be effective at preventing plant growth for many months, and are
used in situations where this is the desired goal. Very few degradation pathways
of these compounds have been worked out in detail, but some generalizations can
be made.
Compounds with organophosphate moieties, such as Diazinon, Methyl
Parathion, Coumaphos and Glyphosate are usually hydrolyzed at the phos-
phorus atom (40,58). Indeed several Flavobacterium isolates are able to grow
using parathion and diazinon as sole sources of carbon.
Triazines pose rather more of a problem, probably because the carbons are
in an effectively oxidized state so that no metabolic energy is obtained by
944 BIOREMEDIATION Vol. 1

N
NH N Cl
N H
N
N N N
O
O O N
NH P NH2

S
Atrazine Diazinon Amitrole

NO2 NO2 O O2

NH

O2N HO P OH
O
O O OH O
P
S
Methyl Parathion Dinoseb Glyphosate

O
S N O N
H
Aldicarb

Fig. 10. Examples of herbicides and pesticides amenable to bioremediation.

their metabolism. Very few pure cultures of microorganisms are able to degrade
triazines such as Atrazine, although some Pseudomonads are able to use the
compound as sole source of nitrogen in the presence of citrate or other simple
carbon substrates. The initial reactions seem to be the removal of the ethyl or
isopropyl substituents on the ring (41), followed by complete mineralization of
the triazine ring.
Nitroaromatic compounds, such as Dinoseb, are degraded under aerobic
and anaerobic conditions (59). The nitro group may be cleaved from the molecule
as nitrite, or reduced to an amino group under either aerobic or anaerobic condi-
tions. Alternatively, the ring may be the subject of reductive attack. Thus, while
these molecules are sometimes quite long-lived in the environment, they can be
completely mineralized under appropriate conditions (59). Recent work has iso-
lated a Clostridium bifermentans able to anaerobically degrade dinoseb cometa-
bolically in the presence of a fermentable substrate. The dinoseb was degraded to
below detectable levels, although only a small fraction was actually mineralized
to CO2 (60).
Carbamates such as Aldicarb undergo degradation under both aerobic and
anaerobic conditions. Indeed the oxidation of the sulfur moiety to the sulfoxide
and sulfone is part of the activation of the compound to its most potent form. Sub-
sequent aerobic metabolism can completely mineralize the compound, although
this process is usually relatively slow so that it is an effective insecticide,
Vol. 1 BIOREMEDIATION 945

acaricide and nematocide. Anaerobically these compounds are hydrolyzed, and


then mineralized by methanogens (61).
Bioremediation. Groundwater. Atrazine dominated the world herbi-
cide market in the 1980s, and contamination of groundwater has been reported
in several locations in the U.S., Europe, and South Africa. There are several
reports that once in groundwater it is very recalcitrant, suggesting that
atrazine-degrading organisms are not widespread. Nevertheless, successful
biodegradation has been achieved with indigenous organisms in laboratory
mesocosms after a lag phase, and once activity was found, it remained (62). Inter-
estingly the degradation was somewhat slowed by the addition of low concentra-
tions of readily assimilate carbon, such as lactate, and it is not clear how
biodegradation might be stimulated in the field. Nevertheless, it is clear that
intrinsic remediation is likely to lead to the disappearance of atrazine from
groundwaters once atrazine-utilizers have become abundant, and perhaps inocu-
lation with atrazine-metabolizing species will be effective.
If more active treatment is required, such as pump-and-treat, it is possible
that biological reactors will be a cost-effective replacement for activated carbon
filters (63).
Marsh and Pond Sediments. Herbicides and pesticides are detectable
in marsh and pond sediments, but intrinsic biodegradation is usually found to be
occurring. Little work has yet been presented where the biodegradation of these
compounds has been successfully stimulated by a bioremediation approach.
Soil. Herbicides and pesticides are of course metabolized in the soil to
which they are applied, and there are many reports of isolating degrading organ-
isms from such sites. Degradation activities are typically much higher at sites
that have seen product application, indicating that natural enrichment processes
occur. Much current effort is aimed at assessing the diversity of degradative
pathways, and in many cases it seems that several different natural metabolic
pathways can degrade individual pollutants. Little work has yet been presented
where the biodegradation of these compounds has been successfully stimulated
by a bioremediation approach, but inoculation with active organisms may be a
promising approach (64, 65).
It is a general observation that herbicide degradation occurs more readily in
cultivated than fallow soil, suggesting that rhizosphere organisms are effective
herbicide degraders. Whether this can be effectively exploited in a phytoremedia-
tion strategy remains to be seen.
3.5. Military Chemicals. Constituents. The military use a range of
chemicals as explosives and propellants, which are sometimes termed ‘‘energetic
molecules’’. Generally speaking, modern explosives are cyclic, often heterocyclic,
composed of carbon, nitrogen and oxygen. Perhaps the most well known is 2,4,6-
trinitrotoluene (Fig. 11), but RDX (Royal Demolition eXplosive; hexahydro-1,3,5-
trinitro-1,2,3-triazine) and HMX (High Melting eXplosive; octahydro-1,3,5,7-
tetranitro-1,3,5,7-tetrazocine), are even more powerful. N,N-Dimethylhydrazine
is used as a solid rocket fuel. These compounds are sometimes present at quite
high levels in soils and groundwater on military bases and production sites. One
quite infamous problem at the latter is ‘‘pink water’’, a relatively undefined mix-
ture of photodegradation products of TNT. Bioremediation is a promising new
technology for treating sites contaminated with such compounds.
946 BIOREMEDIATION Vol. 1

NO2
O2N NO2
N

N N
O2N NO2
NO2

TNT RDX

O2N
N
NO2
N
N NH2
N
O2N
N
NO2

HMX N,N-Dimethylhydrazine

Fig. 11. Military explosives and a rocket propellant.

Bioremediation may also be an appropriate tool for dealing with chemical


agents such as the mustards and organophosphate neurotoxins (66), but little
work on actual bioremediation has been published.
Biodegradation. Natural nitrosubstituted organic compounds are quite
unusual, and it was once thought that their degradation was principally by abio-
tic processes. As shown above in the case of Dinoseb, however, nitrosubstituted
compounds are subject to a variety of degradative processes. The biodegradation
of TNT is well established (67). Under anaerobic conditions it is readily reduced
to the corresponding aromatic amines and subsequently deaminated to toluene.
As shown in the section on hydrocarbons, the latter can be mineralized under
anaerobic conditions, leading to the potentially complete mineralization of TNT
in the absence of oxygen.
Under aerobic conditions TNT can be mineralized by a range of bacteria and
fungi, often co-metabolically with the degradation of a more degradable sub-
strate. There is even evidence that some plants are able to deaminate TNT reduc-
tively. However, there is also ample evidence that under some conditions, the
TNT is converted to insoluble large molecular weight compounds. This probably
occurs by addition reactions to soil components, such as humic and fulvic acids,
or cellular material in the case of plants. Lignolytic fungi are also yielding pro-
mising results for the degradation of both TNT and ‘‘pink water’’, particularly if
the latter is pretreated with uv irradiation.
RDX and HMX are rather more recalcitrant, especially under aerobic con-
ditions, but there are promising indications that biodegradation can occur under
some conditions, especially composting (67). Several strains of bacteria able to
use RDX (and Triazine) as a sole source of nitrogen for growth have recently
been isolated, and this is an area where rapid progress is being made.
Little work has been reported on the biodegradation of dimethylhydrazine,
but it may become an important target for remediation at some sites (68).
Bioremediation. Groundwater. Nitrotoluenes have been detected in
groundwater in some areas, and intrinsic remediation may be occurring at
Vol. 1 BIOREMEDIATION 947

some sites by anaerobic degradation. Research into whether this can be stimu-
lated with a net environmental benefit is in its very earliest stages, and no
clear evidence for success has been presented.
A commercial technology (69), the SABRE process, treats contaminated
water and soil in a two-stage process by adding a readily degradable carbon
and an inoculum of anaerobic bacteria able to degrade the contaminant. An
initial aerobic fermentation removes oxygen so that the subsequent reduction
of the contaminant is not accompanied by oxidative polymerization.
Soil. Composting of soils contaminated by high explosives is being carried
out at the Umatilla Army Depot near Hermiston, Oregon (70). Soil from muni-
tions washout lagoons is being treated indoors in compost rows of 2,000m3,
and the estimated cost is less than one-third the estimated cost of incineration.
If this is successful, there are 30 similar sites on the National Priority List that
could be treated in a similar way.
3.6. Other Organic Compounds. The majority of organic compounds
in commerce are biodegradable, so bioremediation is a potential option for
cleaning up after industrial and transportation accidents. For example, Ref.
71 reports the successful bioremediation of 23,000 liters of vinyl acetate spilled
from a railroad tank car in Albany, New York, at a cost approximately half that
of excavation and disposal of the 1,100 m3 of contaminated soil. They also
report that in situ biological treatment has been used to remediate spills of
other organics, including acrylonitrile, styrene, 2-butoxyethanol, and ethacry-
late, at other sites where there had been railroad accidents. Bioremediation
is thus already an important tool in remediating accidental spills of organic
compounds.
Bioremediation is also an option when spills of such compounds contami-
nate groundwater. For example, bioremediation seems a feasible treatment for
aquifers contaminated with alkylpyridines (72) and phenol (73).

4. Inorganic Contaminants

4.1. Nitrogen Compounds. Constituents. Nitrogen-containing com-


pounds are of concern for several reasons. Nitrate levels are regulated in ground-
water because of concerns for human and animal health. Ammonia is regulated
in streams and effluents as a potential fish toxicant, and any nitrogenous con-
taminant is a potential problem in water because of its stimulatory effect on
the growth of algae. Other nitrogenous contaminants include cyanides in mine
waters. Fortunately, all are amenable to biological treatment.
4.2. Biodegradation. The biological mineralization of fixed nitrogen is
well studied; ammonia is oxidized to nitrite, and nitrite to nitrate, by autotrophic
bacteria, and nitrate is reduced to nitrogen by anaerobic bacteria. Urea in sew-
age and industrial wastes is readily hydrolyzed to ammonia and CO2 by many
bacteria, and cyanides and cyanates are used as sole sources of carbon and nitro-
gen by some organisms. Wastewater treatment facilities utilize these organ-
isms in assuring that municipal and industrial effluents meet strict water
quality standards, but this biological process is outside the scope of this article.
On the other hand, ammonia and nitrate are essential nutrients for plant
948 BIOREMEDIATION Vol. 1

and bacterial growth, so one option is to use these organisms to take up and use
the contaminants.
Bioremediation. Surface Water. One example of exploiting biology to
handle excess nitrogen in a surface water is the case of the Venice Lagoon in
Italy. About a million tons of sea lettuce (Ulva) grows in the lagoon annually
because of the high levels of nitrogenous nutrients in this relatively landlocked
bay. This material is harvested, composted and sold as a low-cost remediation of
this problem (74). In areas where water quality is sufficiently high, growing even
more valuable crops, such as nori (Porphyra) may be an attractive bioremedia-
tion option.
A more constrained opportunity for nitrate bioremediation arose at the US-
DoE Weldon Spring Site near St. Louis, Missouri. This site had been a uranium
and thorium processing facility, and treatment of the metal had involved nitric
acid. The wastestream, known as raffinate, was discharged to surface inpound-
ments and neutralized with lime to precipitate the metals. Two pits had nitrate
levels that required treatment before discharge, but heavy rains in 1993 threa-
tened to cause the pits to overflow. Bioremediation by the addition of calcium
acetate as a carbon source successfully treated more than 19 million liters of
water at a reasonable cost (75).
Groundwater. One approach to minimizing the environmental impact of
excess nitrogen in groundwater migrating into rivers and aquifers is to intercept
the water with rapidly growing trees, such as poplars, that will use the contami-
nant as a fertilizer.
An alternative approach is to add a readily degradable substrate to the con-
taminated aquifer, in the absence of oxygen, to stimulate bacterial denitrifica-
tion. Soluble substrates such as ethanol have been suggested, but a recent
alternative suggestion is to provide vegetable oil, either down the well itself, or
on a gravel trickling filter. Since the oil is insoluble in water, it becomes immo-
bilized on the aquifer material, allowing denitrification as the water flows
through (76).
4.3. Metals and Metalloids. A wide range of metallic and nonmetallic
elements are present as contaminants at industrial and agricultural sites
throughout the world, both in ground- and surface water, and in soils. They
pose a quite different problem from that of organic contaminants, since they can-
not be degraded so that they disappear. Some metal and metalloid elements have
radically different bioavailabilities and toxicities depending on their redox state,
so one option is stabilization by converting them to their least toxic form. This
can be a very effective way of minimizing the environmental impact of a contami-
nant, but if the contaminant is not removed from the environment, there is
always the possibility that natural processes, biological or abiological, may
reverse the process. Removing the contaminants from water phases is relatively
straightforward, and the wastewater treatment industry practices this on an
enormous scale. Pump-and-treat systems that mimic waste water treatments
are already being used for several contaminants, and less complex systems invol-
ving biological mats are a promising solution for less demanding situations.
Microbial leaching of metals from ores is a promising adjunct to more
aggressive metal recovery technologies (77), but is generally achieved by oxida-
tive processes that generate very acidic waters. It seems unlikely that similar
Vol. 1 BIOREMEDIATION 949

approaches will be of much value in removing contaminant metals and metal-


loids from soils.
In the past, removing metal and metalloid contaminants from soil has been
impossible, and site clean-up has meant excavation and disposal in a secure land-
fill. An exciting new approach to this problem is phytoextraction, where plants
are used to extract contaminants from the soil and harvested.
Immobilization and Toxicity-Minimization. Adsorption. Biomass, often
agricultural by-products, has been widely used as an adsorbent for metals and
other contaminants in water (78), but this is outside the scope of this article.
Microbial Processes. Most elements display a range of solubilities and
biological effects depending on their chemical form. For example, chromium,
although it may be an essential micronutrient for many organisms, is known
to be toxic at higher levels. Indeed chromium has a range of available redox
states that differ significantly in their environmental impact. Cr(VI) is generally
more soluble and more toxic than Cr(III), but a wide range of organisms, both
bacteria and fungi, are able to reduce the former to the latter under both aerobic
and anaerobic conditions. Thereby the environmental impact of the contaminant
is reduced. Since Cr(III) precipitates as insoluble hydroxides under neutral and
alkaline conditions, microbial reduction is a potential treatment for soils and
waters (79).
Similarly selenium is a micronutrient that is toxic at higher levels. It also
has quite different bioavailability in its different redox states, with the elemental
form being the least biologically available. Many microorganisms, both bacteria
and fungi, are able to reduce more oxidized species, especially selenite (Se(IV)) to
the red elemental form, providing an appealing remediation option for this ele-
ment. An alternative approach to remediating selenium contamination is to
encourage methylation to volatile dimethylselenide (80). Although the microbiol-
ogy of this process is not yet very well understood, it seems to be the result of
degradation of the selenium-containing amino acid, selenomethionine.
Arsenic is another element with different bioavailability in its different
redox states. Arsenic is not known to be an essential nutrient for eukaryotes,
but arsenate (As(V)) and arsenite (As(III)) are toxic, with the latter being rather
more so, at least to mammals. Nevertheless, some microorganisms grow at the
expense of reducing arsenate to arsenite (81), while others are able to reduce
these species to more reduced forms. In this case it is known that the element
can be immobilized as an insoluble polymetallic sulfide by sulfate reducing bac-
teria, presumably adventitiously due to the production of hydrogen sulfide (82).
Indeed many contaminant metal and metalloid ions can be immobilized as metal
sulfides by sulfate reducing bacteria.
A rather more specific mechanism of microbial immobilization of metal ions
is represented by the accumulation of uranium as an extracellular precipitate of
hydrogen uranyl phosphate by a Citrobacter species (83). Staggering amounts of
uranium can be precipitated; more than 900% of the bacterial dry weight! Recent
work has shown that even elements that do not readily form insoluble phos-
phates, such as nickel and neptunium, may be incorporated into the uranyl phos-
phate crystallites (84). The precipitation is driven by the production of phosphate
ions at the cell surface by an external phosphatase. Although the process
requires the addition of a phosphate donor, such as glycerol-2-phosphate, it
950 BIOREMEDIATION Vol. 1

may be a valuable tool for cleaning water contaminated with radionuclides. An


alternative mode of uranium precipitation is driven by sulfate-reducing bacteria
such as Desulfovibrio desulfuricans, which reduce U(VI) to insoluble U(IV).
When combined with bicarbonate extraction of contaminated soil, this may pro-
vide an effective treatment for removing uranium from contaminated soil (85).
Microbial processes can also detoxify mercury ions and organic compounds
by reducing the mercury to the elemental form, which is volatile (86). This cer-
tainly reduces the environmental impact of compounds such as methylmercury,
however, such a bioprocess would have to include a mercury capture system
before it could be exploited on a large scale with public support.
Rhizofiltration. Rhizofiltration is the use of plant root systems to remove
contaminating metal ions from water (87). The plants must have substantial root
systems, and success has been reported with both hydroponically grown terres-
trial plants, such as sunflowers (Helianthus annuus), and floating aquatic plants
such as water hyacinth (Eichhornia crassipes), and water milfoil (Myriophyllum
spicatum). A filamentous cyanobacterium, Phormidium sp., has also proven
quite effective at removing metals from water. Understanding rhizofiltration at
the molecular level is at an early stage, but it is already clear that several pro-
cesses are involved. Adsorption of the contaminant to the root surface must occur
first, and presumably happens in all plants, probably to polygalacturonic acids
for cations, and positively charged polypeptides for anions. Some plants actively
translocate the metal ions into their cells and some transport the metal to the
leafy shoots. This can be exploited in phytoextraction of metal ions from soils
(see below). This phenomenon may actually be undesirable in rhizofiltration,
because if the metals are retained in the roots there will be a smaller volume
of contaminated material to treat at the end of the water treatment. Perhaps
of more importance for rhizofiltration is the process of root assisted precipitation,
where metal ions are precipitated on the roots as insoluble inorganic complexes.
For example, lead is precipitated as lead phosphates on roots of Indian Mustard
(Brassica juncea) to a remarkable level; up to 45% of dry weight (87). This phe-
nomenon seems to require active exudation of precipitants from the roots, for it is
dramatically reduced when dead roots are used.
Phytoextraction. The fact that certain plant species are able to survive
on soils with such high levels of metal ions that the growth of most plants is
inhibited has been known for centuries. However, it was not until this century
that it became clear that some of these plants actually accumulated the poten-
tially toxic metals and somehow resisted their toxic effects. These plants,
known as hyperaccumulators, are exemplified by Alpine Pennycress, Thlaspi
caerulescens, which is able to accumulate a wide range of metals, including
cadmium, chromium, copper, lead, nickel, and zinc, more than several thou-
sand fold in its roots, and somewhat less in its leaves (88). Unfortunately for
a use in bioremediation, most hyperaccumulators are small, slow growing
plants. Researchers have, therefore, turned to more rapidly growing plants
which, while they might not be quite as effective as the hyperaccumulators
on a weight basis, might be able to extract more metal in a growing season
(88). Indian mustard, Brassica juncea, was found to be almost as effective as
T. caerulescens and to grow much faster. This plant is now being used in
field trials of phytoextraction.
Vol. 1 BIOREMEDIATION 951

Phytoextraction of metals from soils requires plants that have substantial


root systems to maximize contact with the contaminated soil, and effective trans-
port mechanisms to get the metal from the root to harvestable biomass. Merely
accumulating the metal in or on the roots would be less desirable, because har-
vesting would likely increase the environmental impact of the contaminant by
creating dust. It has been known for some time that metals such as cadmium
are often stored in plant tissues as phytochelatin-complexes. Phytochelatins
are small peptides rich in cysteines that chelate metals via the cysteinyl sulfur
(89). Other elements are stored as metalphytate precipitates (88), and in both
cases seem to be concentrated in the vacuolar sap. But how the metals are trans-
ported in the xylem sap from the root to the shoot is only beginning to be
explored. Cadmium seems to be transported as a soluble salt of organic acids
(90), at least in B. juncea, while nickel is thought to be transported as a soluble
histidine complex in Alyssum species (91). Optimizing this process will be an
important consideration for the commercial success of phytoremediation.
It will also be important to understand the rhizosphere ecology around the
roots of metal accumulating plants fully. Maximizing the bioavailability of the
contaminant metals in this zone may require the optimization of the microbial
communities, or perhaps the addition of soil amendments. There are early indi-
cations that such intervention may be beneficial (88), but research in this area is
at a very early stage.
Phytoremediation is also being developed for dealing with soils contami-
nated with high levels of selenium in California; again B. juncea seems to be par-
ticularly effective in accumulating the contaminant from soil, and all plants
tested were more effective at removing selenate than selenite (92). This is an
interesting contrast to bacterial systems, where selenite reduction is more com-
monly found than selenate reduction.
There have also been reports that some plants, including B. juncea, stimu-
late volatilization of dimethylselenide, although it seems that this is an indirect
effect where the plant roots stimulate bacterial evolution of the gas (93). Presum-
ably the selenium is eventually oxidized in the atmosphere and returned to the
soil as rain. Since much of the world is marginally selenium deficient, such a pro-
cess might have no deleterious environmental effects. A similar volatilization
approach may eventually be used for mercury contamination. Ref. 94 describes
the effective transfer of bacterial mercury reduction genes into a plant. In the
future, this approach might offer an option for cleaning mercury-contaminated
soils, albeit with some form of mercury capture technology.
Phytostabilization. In some cases it may be important to stabilize a con-
taminated site to minimize harmful environmental impacts. Barren soils are
more prone to erosion, leaching, and dust formation than vegetated soils, so
establishing plants as a ground cover may be a very beneficial treatment, even
if the contaminant remains in place. In some cases, planting metal tolerant spe-
cies is enough to establish a plant community, but in other situations, such as
mine waste piles, it may be essential to restore fertility to the soil by adjusting
the pH and adding organic composts. Where phytoextraction requires plants that
actively transport metals to the above-ground biomass, this would be an undesir-
able phenomenon in phytostabilization programs because it would promote the
movement of the contaminating metals into the biosphere.
952 BIOREMEDIATION Vol. 1

It may also be possible to stimulate bioreductions in soils, such as Cr(VI) to


Cr(III), more effectively by growing plants than by adding bacteria or nutrients
to stimulate the microbial process.
Bioremediation. Water. Groundwater can be treated in anaerobic
bioreactors that encourage the growth of sulfate reducing bacteria, where the
metals are reduced to insoluble sulfides, and concentrated in the sludge. For
example, such a system is in use to decontaminate a zinc smelter site in the
Netherlands (95).
Bacterial remediation of selenium oxyanions in San Joaquin, California,
drainage water is under active investigation (96,97), but has not yet been com-
mercialized. Agricultural drainage rich in selenium is also typically rich in
nitrates, so bioremediation must also include conditions that stimulate denitrifi-
cation (98). Phytoextraction of selenium is also being tested, but is not yet being
used on a large scale.
Phytoremediation is also not yet being used commercially, but results at
several field trial suggest that commercialization is not far away. Perhaps the
biggest success to date is the successful rhizofiltration of radionuclides from a
Department of Energy site at Ashtabula, Ohio, where uranium concentrations
of 350 ppb were reduced to less than 5 ppb, well below groundwater standards,
by Sunflower roots (99). Small field tests also suggest that rhizofiltration will be a
remediation option for removing radionuclides, such as 137Cs and 90Sr, from con-
taminated lakes and ponds near the Chernobyl nuclear reactor in Ukraine.
Mine Drainage. Natural drainage water that come into contact with
active and abandoned metal and coal mines can become seriously contaminated
with a range of heavy-metal ions, and/or often become quite acidic, with a
pH near 2. While underground, the water is typically anoxic, and any iron is pre-
sent as soluble ferrous species. When this mixes with aerobic surface water, the
iron precipitates as bright orange ferric hydroxide, and this can have a serious
environmental impact. In recent years it has become clear that the environmen-
tal impact of acid mine drainage can be minimized by the construction of
artificial wetlands that combine geochemistry and biological treatments. These
systems are being designed for a range of wastewaters, most of which fall outside
the scope of this article.
The precipitation of ferric hydroxide is typically biologically mediated by
iron-oxidizing bacteria at acid pH, but is usually rather slow. Abiological oxida-
tion becomes more important at pH values above 5, and this is usually much fas-
ter than the biological process. Most constructed wetlands for treating acid mine
waters thus start with a zone designed to raise the pH. A bed of crushed lime-
stone often suffices to raise the pH significantly, and it is important that this
be kept anoxic to prevent rust precipitation on the limestone, which would pre-
vent further production of alkalinity. Once the pH is near neutral, the water is
discharged into an aerobic wetland to encourage the precipitation of iron and alu-
minum oxides, and the co-precipitation of arsenic, if this is present (100,101).
If heavy metals are present in the mine water, the iron-free water can be
made to flow into an anaerobic part of the constructed wetland, where organic
material, such as compost, manure or sawdust, provides reductants to sulfate-
reducing bacteria that become established. These bacteria reduce sulfate to sul-
fide, which precipitates the heavy metal ions as insoluble sulfides. It is important
Vol. 1 BIOREMEDIATION 953

that this part of the constructed wetland be kept anaerobic, to prevent oxidation
and remobilization of the precipitated metals. For this reason, this part of the
wetland is typically kept flooded and free of aquatic plants that might introduce
oxygen through their roots (100). Finally, an aerobic algal mat can act as a pol-
ishing step to complete the removal of contaminants, particularly manganese
(101,102). Mine drainage that is not acidic may not need such complicated sys-
tems and individual parts of the treatment train described above may suffice in
some situations.
Soil. The first reported field trial of the use of hyperaccumulating plants
to remove metals from a soil contaminated by sludge applications has been
reported (103). The results were positive, but the rates of metal uptake suggest
a time scale of decades for complete cleanup. Trials with higher biomass plants,
such as B. juncea, are underway at several chromium and lead contaminated
sites (88), but data are not yet available.
The bacterial reduction of Cr(VI) to Cr(III) discussed above is also being
used to reduce the hazards of chromium in soils and water (104).

5. Conclusions

Bioremediation has been successfully used to treat a wide range of contaminants,


including crude oils and refined petroleum products, halogenated solvents, pes-
ticides, herbicides, military chemicals, and mine waters. Much of this success has
come by small adjustments of the local environment to encourage the growth of
remediating organisms. Fertilizer addition has been a successful treatment for
terrestrial and marine oil spills, and the addition of co-substrates, particularly
methane, has been a successful treatment for remediating halogenated solvents,
such as trichloroethylene. Composting is proving to be a successful treatment for
a range of contaminants, and constructed wetlands are successfully treating a
range of wastewaters, including those emanating from mines.
One area where there is considerable disagreement between academic
scientists and engineering practitioners of bioremediation is the area of bioaug-
mentation, the addition of selected microorganisms to a site to encourage biode-
gradation. Dozens of companies sell bacteria for this purpose and claim success
in the field, but efforts to demonstrate effectiveness rigorously have met with lit-
tle success (105,57). Perhaps the most startling test was performed by the U.S.
EPA when testing potential inoculants for stimulating oil biodegradation in
Alaska following the Exxon Valdez oil spill (106). Eight products were tested in
small laboratory reactors that allowed substantial degradation of a test oil by the
indigenous organisms of Prince William Sound. All eight microbial inocula had a
greater stimulatory effect on alkane degradation, at least for the first 11 days,
when they were sterilized by autoclaving prior to addition! This suggests that
the indigenous organisms readily out-competed the added products, but that
autoclaving the products released some trace nutrient that was able to stimulate
the growth of the endogenous organisms.
Of course, bioaugmentation may prove more effective with contaminants
that have only recently entered the biosphere, such as methyl-tert-butyl ether,
or with organisms that have a dramatic selective advantage over indigenous
954 BIOREMEDIATION Vol. 1

organisms. For example, modern molecular biology may offer opportunities for
moving effective degradation pathways into organisms native to a contaminated
site, improving biodegradation pathways by broadening the substrate range of
degradative enzymes, or removing regulatory constraints to maximize degrada-
tive activities. There are, however, many technical and regulatory hurdles to be
surmounted before this potential can be tested.
Another area where there is controversy is in the role of surfactants. Many
bacteria, particularly those that degrade hydrocarbons, produce surfactants as
they grow. Some release them into the medium, others incorporate them into
their cell exterior, and there have been elegant experiments to show that inhibit-
ing the production of these compounds inhibits the ability of the bacteria to
degrade oil. There have thus been many suggestions to add surfactants, either
bacterial or synthetic, to stimulate biodegradation. This seems to be beneficial
in the case of some oil dispersants at sea (12), but there have not yet been any
clear demonstrations of efficacy on a large scale in soils. The role of surfactants in
bioremediation is an area that requires further study.
Bioremediation has many advantages over other technologies, both in cost
and in effectively destroying or extracting the pollutant. An important issue is
thus when to consider it, and a series of questions may lead to the appropriate
answer (see Table 4).

Table 4. Will Bioremediation be a Suitable Treatment for a Site Contaminated with


Organic, Nitrogenous, or Organic Contaminants?
Organic
Is the contaminant biodegradable? If the contaminant is a complex mixture of components,
are the individual chemical species biodegradable? If the contaminant has been at the site
for some time, biodegradation of the most readily degradable components may have
already occurred. Is the residual contamination biodegradable?
Are degrading organisms present at the site?
What is limiting their growth and activity? Can this be added effectively?
Are the levels of contaminant amenable to bioremediation? Are they toxic to
microorganisms? Are they so abundant that even substantial microbial activity will
take too long to clean the site?
Are the clean-up standards reasonable? Are biological processes known to degrade sub-
strates down to the levels required?
Nitrogenous
Are appropriate microorganisms present at the site?
What is limiting their growth and activity? Can this be added effectively?
Are the levels of contaminant amenable to bioremediation? Are they toxic to
microorganisms? Are they so abundant that even substantial microbial activity
will take too long to clean the site?
Can the nitrogenous compound be used by plants?
Are the clean-up standards reasonable? Are biological processes known to degrade
substrates down to the levels required?
Inorganic
Can the contaminant be made less hazardous by changing its redox state?
Can the contaminant be brought to a reactor or constructed wetland where biological
systems, microbial or plant, can extract and immobilize the contaminant?
Can plants extract the contaminant from the soil matrix?
Are the clean-up standards reasonable? Are biological processes known to accumulate
contaminants down to the levels required?
Vol. 1 BIOREMEDIATION 955

If the answers to the questions in Table 4 lead to the selection of bioreme-


diation, it then becomes important to assess the success of the bioremediation
strategy in achieving the clean-up criteria. A major disadvantage of bioremedia-
tion is that it is typically rather slower than competing technologies such as ther-
mal treatments. How can regulators and responsible parties gain confidence
during this time that success will indeed be achieved? The National Research
Council (107) has recently addressed this issue, and suggested a three-fold strat-
egy for ‘‘proving’’ bioremediation: (1) a documented loss of contaminants from the
site; (2) laboratory tests showing the potential of endogenous microbes to cata-
lyze the reactions of interest; and (3) some evidence that this potential is
achieved in the field.
Although laboratory tests to demonstrate the potential of endogenous
organisms are relatively straightforward, the other two are often not as simple
as they seem, especially in soils and sediments. Documenting the loss of contami-
nant is often difficult because of the heterogeneous distribution of the contami-
nant in the field; large numbers of samples may be needed in order to be able to
detect statistically significant decreases in absolute concentration. Providing evi-
dence that biodegradation has indeed been stimulated is also a challenge, but if
the contaminant is a complex mixture, such as a crude oil, refined petroleum, or a
mixture of polychlorinated biphenyls, the least degradable compounds in the
mixture can be used as conserved internal markers for quantifying the degrada-
tion of the more degradable components.
For example, hopane(Fig. 12) is a conserved marker in crude oils in at
least the early stages of biodegradation (up to 80% degradation) (108); its con-
centration increases in the residual oil as biodegradation proceeds. Basing esti-
mates of biodegradation on hopane allowed us to quantify the effect of
the successful bioremediation strategy following the Exxon Valdez oil-spill
(14,15). Even in refined products, the least degradable detectable analyte can
serve this role. Thus, the trimethyl-phenanthrenes can be used to estimate
qualitatively the degradation of diesel oils in the environment, even though
these molecules are themselves biodegradable (109). Of course, since these
molecules are themselves biodegradable, estimates of the rates of biodegrada-
tion of more readily biodegradable compounds are systematically underesti-
mated using this approach, but it can still provide very valuable information.
The more halogenated polychlorinated biphenyls can serve a similar role in
assessing the environmental fate of these products (55), and benzene is typi-
cally the last compound in BTEX plumes to be degraded, particularly under
anaerobic conditions.

H
H

Fig. 12. 17 a(H), 21 b (H)-hopane.


956 BIOREMEDIATION Vol. 1

In situations where conserved internal markers cannot be used, such as in


spills of essentially pure compounds, the evidence for enhanced biodegradation
may have to be more indirect. Oxygen consumption, increases in microbial activ-
ity or population, and carbon dioxide evolution have all been used with success.
Finally, a caveat. Despite its documented success in many situations, bior-
emediation may not always be able to meet current clean-up criteria for a parti-
cular site. Some standards are so tight that they are essentially ‘‘detection limit’’
standards, and it is not clear that biological processes will be able to remove con-
taminants to such low levels. For example, the level of contaminant may be so
low that it does not induce the microorganisms to produce the enzymes necessary
for biodegradation. Or perhaps the contaminant is bound to soil or sediment par-
ticles in such a way that it is not available for biodegradation, although it is still
extractable with aggressive solvents in analytical procedures. These are areas
that require further research, but bioremediation will be more likely to fulfill
its promise as an important tool in contaminated site remediation if there is pro-
gress towards standards based on bioavailability and net environmental benefit
from the clean up, rather than on arbitrary absolute standards.

BIBLIOGRAPHY

CITED PUBLICATIONS

1. K. Devine, in R. E. Hinchee, J. A. Kittel, and H. J. Reisinger, eds., Applied Bioreme-


diation of Hydrocarbons, Battelle Press, Columbus, Ohio, 1995, 53–59.
2. S. S. Butcher, R. J. Charlson, G. H. Orians, and G. V. Wolfe, Global Biogeochemical
Cycles, Academic Press, New York, 1992.
3. I. D. Prijambada, S. Negoro, T. Yomo, and I. Urabe, Appl. Environ. Microbiol. 61,
2020–2022 (1995).
4. P. Adriaens and T. M. Vogel, in L. Y. Young and C. E. Cerniglia, eds., Microbial Trans-
formation and Degradation of Toxic Organics, Wiley-Liss, New York, 1995, 435–486.
5. A. Paszczynski and R. L. Crawford, Biotechnol. Progr. 11, 368–379 (1995).
6. J. J. Childress, Amer. Zool. 35, 83–90 (1995).
7. T. D. Sharkey and E. L. Singsaas, Nature 374, 769 (1995).
8. R. C. Prince, Bioremediation of Crude Oil, in R. A. Meyers, ed., Encyclopedia of
Environmental Analysis and Remediation, John Wiley & Sons, Inc., (in press)
(1998).
9. J. P. Salanitro, Curr. Opini. Biotechnol. 6, 337–340 (1995).
10. J. M. Yudelson and P. D. Tinari, in R. E. Hinchee, G. D. Sayles and R. S. Skeen, eds.,
Biological Unit Processes for Hazardous Waste Treatment, Battelle Press, Columbus,
Ohio, 1995, 205–209.
11. National Research Council Oil in the Sea: Inputs, Fates and Effects, National Acad-
emy Press, Washington, D.C., 1985.
12. R. Varadaraj, M. L. Robbins, J. Bock, S. Pace, and D. MacDonald, in Proceedings of
the 1995 International Oil Spill Conference, American Petroleum Institute,
Washington, D.C., 1995, 101–106.
13. R. C. Prince, Critical Reviews Microbiology 19, 217–242 (1993).
14. R. C. Prince, J. R. Clark, J. E. Lindstrom, E. L. Butler, E. J. Brown, G. Winter, M. J.
Grossman, R. R. Parrish, R. E. Bare, J. F. Braddock, W. G. Steinhauer, G. S.
Douglas, J. M. Kennedy, P. Barter, J. R. Bragg, E. J. Harner, and R. M. Atlas, in
Vol. 1 BIOREMEDIATION 957

R. E. Hinchee, B. C. Alleman, R. E. Hoeppel and R. N. Miller, eds., Hydrocarbon


Remediation, Lewis Publishers, Boca Raton, Fla., 1994, 107–124.
15. J. R. Bragg, R. C. Prince, E. J. Harner, and R. M. Atlas, Nature 368, 413–418 (1994).
16. R. C. Borden, C. A. Gomez, and M. T. Becker, Ground Water 33, 180–189 (1995).
17. R. D. Norris, in Ref. 1, 483–487 (1995).
18. G. Battermann and M. Meier-Löhr, in Ref. 1, 155–164.
19. D. R. Lovley, J. C. Woodward, and F. H. Chapelle, Nature 370, 128–131 (1994).
20. B. A. Keet, in Ref. 1, 329–334.
21. S. J. McMillen, N. R. Gray, J. M. Kerr, A. G. Requejo, T. J. McDonald, and G. S. Dou-
glas, in R. E. Hinchee, G. S. Douglas, and S. K. Ong, eds., Monitoring and Verifica-
tion of Bioremediation, Battelle Press, Columbus, Ohio, 1995, 1–9.
22. B. M. Peyton, M. J. Truex, R. S. Skeen, and B. S. Hooker, in R. E. Hinchee, A. Lee-
son, and L. Semprini, eds., Bioremediation of Chlorinated Solvents, Battelle Press,
Columbus, Ohio, 1995, 111–116.
23. D. Kohler-Staub, S. Frank, and T. Leisinger, Biodegradation 6, 229–236 (1995).
24. D. B. Janssen, F. Pries, and J. R. van der Ploeg, Ann. Rev. Microbiol. 48, 163–191 (1994).
25. D. J. Arp, Curr. Opin. Biotechnol. 6, 352–358 (1995).
26. S. J. Ergas, K. Kinney, M. E. Fuller, and K. M. Scow, Biotechnol. Bioengineer. 44,
1048–1054 (1994).
27. K. H. Lombard, J. W. Borten, and T. C. Hazen, in R. E. Hinchee, ed., Air Sparging
for Site Remediation, Lewis Publishers, Boca Raton, Fla., 1994, 81–96.
28. T. A. Anderson and B. T. Walton, Environ. Toxicol. Chem. 14, 2041–2047 (1995).
29. J. A. Oleskiewicz and M. Elektorowicz, J. Soil Contam. 2, 205–208 (1993).
30. W. M. Wu, J. Nye, R. F. Hickey, M. K. Jain, and J. G. Zeikus, in R. E. Hinchee, A.
Leeson, and L. Semprini, eds., Bioremediation of Chlorinated Solvents, Battelle
Press, Columbus, Ohio, 1995, 45–52.
31. L. M. F. Dossantos and A. G. Livingston, Appl. Microbiol. Biotechnol. 42, 421–431
(1994).
32. M. Knippschild and H. J. Rehm, Appl. Microbiol. Biotechnol. 44, 253–258 (1995).
33. J. Gerritse, V. Renard, J. Visser, and J. C. Gottschal, Appl. Microbiol. Biotechnol. 43,
920–928 (1995).
34. G. W. Gribble, J. Chem. Ed. 71, 907–911 (1994).
35. S. Fetzner and F. Lingens, Microbiol. Rev. 58, 641–673 (1994).
36. P. Juteau, R. Beaudet, G. Mcsween, F. Lepine, and J. G. Bisaillon, Can. J. Microbiol.
41, 862–868 (1995).
37. C. Kennes, W. M. Wu, L. Bhatnagar, and J. G. Zeikus, Appl. Microbiol. Biotechnol.
44, 801–806 (1996).
38. K. A. McAllister, H. Lee, and J. T. Trevors, Biodegradation 7, 1–40 (1996).
39. B. C. Okeke, A. Paterson, J. E. Smith, and I. A. Watson-Craik, Letts. Appl. Microbiol.
19, 284–287 (1994).
40. J. Aislabie and G. Lloyd-Jones, Aust. J. Soil. Res. 33, 925–942 (1995).
41. D. A. Abramowicz, Crit. Rev. Biotechnol. 10, 241–251 (1990).
42. S. Han, L. D. Eltis, K. N. Timmis, S. W. Muchmore, and J. T. Bolin, Science 270,
976–980 (1995).
43. B. N. Aronstein and L. E. Rice, J. Chem. Technol. Biotechnol. 63, 321–328 (1995).
44. J. L. Gentry and T. J. Simpkin, in R. E. Hinchee, R. M. Miller, and P. C. Johnson,
eds., In situ Aeration: Air Sparging, Bioventing and Related Remediation Processes,
Battelle Press, Columbus, Ohio, 283–289 (1995).
45. A. M. Ferro, R. C. Sims, and B. Bugbee, J. Environ. Qual. 23, 272–279 (1994).
46. S. G. Huling, D. F. Pope, J. E. Matthews, J. L. Sims, R. C. Sims, and D. L. Sorenson,
in R. E. Hinchee, R. E. Hoeppel and D. B. Anderson, eds., Bioremediation of Recal-
citrant Organics, Battelle Press, Columbus, Ohio, 1995, 101–109.
958 BIOREMEDIATION Vol. 1

47. G. M. Colores, P. M. Radehaus, and S. K. Schmidt, Appl. Biochem. Biotechnol. 54,


271–275 (1995).
48. M. P. Otte, J. Gagnon, Y. Comeau, N. Maatte, C. W. Greer, and R. Samson, Appl.
Microbiol. Biotechnol. 40, 926–932 (1994).
49. A. J. Wall and G. W. Stratton, Water Air Soil Poll. 82, 723–737 (1995).
50. F. Baud-Grasset and T. M. Vogel, in R. E. Hinchee, J. E. Fredrickson and B. C. Alle-
man, eds., Bioaugmentation for Site Remediation, Battelle Press, Columbus, Ohio,
1995, 39–48.
51. F. C. Michel, C. A. Reddy, and L. J. Forney, Appl. Environ. Microbiol. 61, 2566–2571
(1995).
52. C. J. Gantzer and D. Cosgriff, in Ref. 44, 543–549.
53. R. U. Edgehill, Water Res. 30, 357–363 (1996).
54. D. Dietrich, W. J. Hickey, and R. Lamar, Appl. Environ. Microbiol. 61, 3904–3909
(1995).
55. D. L. Bedard and R. J. May, Environ. Sci. Technol. 30, 237–245 (1996).
56. D. L. Bedard, S. C. Bunnell and L. A. Smullen, Environ. Sci. Technol. 30, 687–694
(1996).
57. M. R. Harkness, J. B. McDermott, D. A. Abramowicz, J. J. Salvo, W. P. Flanagan, M.
L. Stephens, F. J. Mondello, R. J. May, H. J. Lobos, K. M. Carroll, M. J. Brennan, A.
A. Bracco, K. M. Fish, G. L. Warner, P. R. Wilson, D. K. Dietrich, D. T. Lin, C. B.
Morgan, and W. L. Gately, Science 259, 503–507 (1993).
58. R. E. Dick and J. P. Quinn, Appl. Microbiol. Biotechnol. 43, 545–550 (1995).
59. R. H. Kaake, D. L. Crawford, and R. L. Crawford, Biodegradation 6, 329–337 (1995).
60. T. B. Hammill and R. L. Crawford, Appl. Environ. Microbiol. 62, 1842–1846 (1996).
61. J. Kazumi and D. G. Capone, Appl. Environ. Microbiol. 62, 2820–2829 (1995).
62. I. Mirgain, G. Green, and H. Monteil, Environ. Technol. 16, 967–976 (1995).
63. C. J. Hapeman, J. S. Karns, and D. R. Shelton, J. Agric. Food Chem. 43, 1383–1391
(1995).
64. D. R. Shelton, S. Khader, J. S. Karns, and B. M. Pogell, Biodegradation 7, 129–136
(1996).
65. J. A. Entry, P. K. Donnelly, and W. H. Emmingham, Appl. Soil Ecol. 3, 85–90 (1996).
66. D. L. Kaplan, Curr. Opinion Biotechnol. 3, 253–260 (1992).
67. T. Gorontzy, O. Dryzga, M. W. Kahl, D. Bruns-Nagel, J. Breitung, E. Von Loew, and
K.-H. Blotevogel, Crit. Rev. Microbiol. 20, 265–284 (1994).
68. N. S. Kasimov, V. B. Grebenyuk, T. V. Koroleva, and Y. V. Proskuryakov, Eurasian
Soil Science 28, 79–95 (1996).
69. U.S. Pat. 5,387,271 (1995),
70. S. Eversmeyer, P. Faessler, and C. Bird, Soil Groundwater Cleanup (Dec. 27–29,
1995).
71. E. Flathman, B. J. Krupp, P. Zottola, J. R. Trausch, J. H. Carson, R. Yao, G. J. Laird,
P. M. Woodhull, D. E. Jerger, and P. R. Lear, Remediation 6, 57–79 (1996).
72. Z. Ronen, J. M. Bollag, C. H. Hsu, and J. C. Young, Ground Water 34, 194–199
(1996).
73. M. H. Essa, S. Farooq, and G. F. Nakhla, Water Air Soil Poll. 87, 267–281 (1996).
74. V. Cuomo, A. Perretti, I. Palomba, A. Verde, and A. Cuomo, J. Appl. Phycol. 7, 479–
485 (1995).
75. G. C. Schmidt and M. B. Ballew, in R. E. Hinchee, J. L. Means, and D. R. Burris,
eds., Bioremediation of Inorganics, Battelle Press, Columbus, Ohio, 1995, 109–116.
76. W. J. Hunter and R. F. Folett, in Clean Water—Clean Environment—21st Century,
Vol. II: Nutrients. American Society of Agricultural Engineers, St Joseph, Mich.,
1995, 79–82.
77. D. K. Ewart and M. N. Hughes, Adv. Inorg. Chem. 36, 103–135 (1991).
Vol. 1 BIOREMEDIATION 959

78. B. Volesky and Z. R. Holan, Biotechnol. Prog. 11, 235–250 (1995).


79. C. E. Turick, W. A. Apel, and N. S. Carmiol, Appl. Microbiol. Biotechnol. 44, 683–688
(1996).
80. W. T. Frankenberger and U. Karlson, Geomicrobiol. J. 12, 265–278 (1994).
81. D. Ahmann, A. L. Roberts, L. R. Krumholz, and F. M. M. Morel, Nature 371, 750
(1994).
82. K. A. Rittle, J. I. Drever, and P. J. S. Colberg, Geomicrobiol. J. 13, 1–11 (1995).
83. L. E. Macaskie, R. M. Empson, F. Lin, and M. R. Tolley, J. Chem. Technol. Biotech-
nol. 63, 1–16 (1995).
84. K. M. Bonthrone, G. Basnakova, F. Lin, and L. E. Macaskie, Nature Biotechnology
14, 635–638 (1996).
85. E. J. P. Phillips, E. R. Landa, and D. R. Lovley, J. Ind. Microbiol. 14, 203–207
(1995).
86. E. Saouter, M. Gillman, and T. Barkay, J. Ind. Microbiol. 14, 343–348 (1995).
87. V. Dushenkov, P. B. A. N. Kumar, H. Motto, and I. Raskin, Environ. Sci. Technol.
29, 1239–1245 (1995).
88. D. E. Salt, M. Blaylock, N. P. B. A. Kumar, V. Dushenkov, B. D. Ensley, I. Chet, and
I. Raskin, Bio-Technology 13, 468–474 (1995).
89. W. E. Rauser, Ann. Rev. Biochem. 59, 61–86 (1990).
90. D. E. Salt, R. C. Prince, I. J. Pickering, and I. Raskin, Plant Physiol. 109, 1427–1433
(1995).
91. U. Kramer, J. D. Cotterhowells, J. M. Charnock, A. J. M. Baker, and J. A. C. Smith,
Nature 379, 635–638 (1996).
92. G. S. Banuelos and D. W. Meek, J. Environ. Qual. 19, 772–777 (1990).
93. A. M. Zayed and N. Terry, J. Plant. Physiol. 143, 8–14 (1994).
94. C. L. Rugh, H. D. Wilde, N. M. Stack, D. M. Thomson, A. O. Summers, and R. B.
Meagher, Proc. Natl. Acad. Sci. USA 93, 3182–3187 (1996).
95. L. J. Barnes, P. J. M. Scheeren, and C. J. N. Buisman, in J. L. Means and R. E.
Hinchee, eds., Emerging Technology for Bioremediation of Metals, Lewis Publishers,
Boca Raton, Fla., 1994, 38–49.
96. J. M. Macy, S. Lawson, and H. DeMott-Decker, Appl. Microbiol. Biotechnol. 40, 588–
594 (1993).
97. L. P. Owens, K. C. Kovac, J. A. L. Kipps, and D. W. J. Hayes, in Ref. 75, 89–94.
98. J. L. Kipps, in J. L. Means and R. E. Hinchee, eds., Emerging Technology for Bior-
emediation of Metals, Lewis Publishers, Boca Raton, Fla., 1994, 105–109.
99. C. M. Cooney, Environ. Sci. Technol. 30, A194 (1996).
100. G. Robb and J. Robinson, Mining Environ. Manage., 19–21 (Sept. 1995).
101. M. E. Dodds-Smith, C. A. Payne, and J. J. Gusek, Mining Environ. Manage., 22–24
(Sept. 1995).
102. J. Bender and P. Phillips, Mining Environ. Manage., 25–27 (Sept. 1995).
103. A. J. M. Baker, S. P. McGrath, C. M. D. Sidoli, and R. D. Reeves, Resource Conserv.
Recycling 11, 41–49 (1994).
104. L. J. DeFilippi, Environ. Sci. Pollut. Control Ser. 8, 437–457 (1994).
105. P. H. Pritchard, Curr. Opinion Biotechnol. 3, 232–243 (1992).
106. A. D. Venosa, J. R. Haines, W. Nisamaneepong, R. Govind, S. Pradhan, and B. Sid-
dique, J. Ind. Microbiol. 10, 13–23 (1992).
107. National Research Council, In situ Bioremediation. When Does it Work? National
Academy Press, Washington, D.C., 1993.
108. R. C. Prince, D. L. Elmendorf, J. R. Lute, C. S. Hsu, C. E. Haith, J. D. Senius, G. J.
Dechert, G. S. Douglas, and E. L. Butler, Environ. Sci. Technol. 28, 142–145 (1994).
109. G. S. Douglas, K. J. McCarthy, D. T. Dahlen, J. A. Seavey, W. G. Steinhauer, R. C.
Prince, and D. L. Elmendorf, J. Soil Contam. 1, 197–216 (1992).
960 BIOREMEDIATION Vol. 1

GENERAL REFERENCES

M. Alexander, Biodegradation and Bioremediation, Academic Press, San Diego, Calif.,


1994.
R. E. Hinchee, J. T. Wilson, and D. C. Downey, eds., Intrinsic Bioremediation, Battelle
Press, Columbus, Ohio, 1995.
R. E. Hinchee, R. M. Miller, and P. C. Johnson, eds., In situ Aeration: Air Sparging,
Bioventing and Related Remediation Processes, Battelle Press, Columbus,
Ohio, 1995.
R. E. Hinchee, J. E. Fredrickson, and B. C. Alleman, eds., Bioaugmentation for Site Reme-
diation, Battelle Press, Columbus, Ohio, 1995.
R. E. Hinchee, A. Leeson, and L. Semprini, eds., Bioremediation of Chlorinated Solvents,
Battelle Press, Columbus, Ohio, 1995.
R. E. Hinchee, G. S. Douglas, and S. K. Ong, eds., Monitoring and Verification of
Bioremediation, Battelle Press, Columbus, Ohio, 1995.
R. E. Hinchee, J. A. Kittel, and H. J. Reisinger, eds., Applied Bioremediation of Hydrocar-
bons, Battelle Press, Columbus, Ohio, 1995.
R. E. Hinchee, R. E. Hoeppel, and D. B. Anderson, eds., Bioremediation of Recalcitrant
Organics, Battelle Press, Columbus, Ohio, 1995.
R. E. Hinchee, C. M. Vogel, and F. J. Brockman, eds., Microbial Processes for Bioremedia-
tion, Battelle Press, Columbus, Ohio, 1995.
R. E. Hinchee, G. D. Sayles, and R. S. Keen, eds., Biological Unit Processes for Hazardous
Waste Treatment, Battelle Press, Columbus, Ohio, 1995.
R. E. Hinchee, J. L. Means, and D. R. Burris, eds., Bioremediation of Inorganics, Battelle
Press, Columbus, Ohio, 1995.
National Research Council, In situ Bioremediation. When Does it Work? National Acad-
emy Press, Washington, D.C., 1993.
Organization for Economic Cooperation and Development, Bioremediation: the Tokyo ’94
Workshop, OECD, Paris, 1995.
L. Y. Young and C. E. Cerniglia, eds., Microbial Transformation and Degradation of Toxic
Organic Chemicals, Wiley-Liss, New York, 1995.
R. E. Hinchee, A. Leeson, L. Semprini, and S. K. Ong, eds., Bioremediation of
Chlorinated and Polycyclic Aromatic Hydrocarbons, Lewis Publishers, Ann Arbor,
Mich., 1994.
R. E. Hinchee, B. C. Alleman, R. E. Hoeppel, and R. N. Miller, eds., Hydrocarbon Biore-
mediation, Lewis Publishers, Ann Arbor, Mich., 1994.
R. E. Hinchee, D. B. Anderson, F. B. Metting, Jr., and G. D. Sayles, eds., Applied Biotech-
nology for Site Remediation, Lewis Publishers, Ann Arbor, Mich., 1994.
J. L. Means and R. E. Hinchee, eds., Emerging Technology for Bioremediation of Metals,
Lewis Publishers, Ann Arbor, Mich., 1994.
R. E. Hinchee, ed., Air Sparging for Site Remediation, Lewis Publishers, Ann Arbor,
Mich., 1994.
R. F. Hickey and G. Smith, eds., Biotechnology in Industrial Waste Treatment and
Bioremediation, Lewis Publishers, Boca Raton, Fla., 1996.
R. H. Kadlec and R. L. Knight, Treatment Wetlands, Lewis Publishers, Boca Raton,
Fla., 1996.

ROGER C.PRINCE
Exxon Research and Engineering Company
PART II: REFINED
PRODUCTS AND FUELS

LIQUEFIED PETROLEUM
GAS
Liquefied petroleum gas (LPG) is a subcategory of a versatile class of petroleum
products known as natural gas liquids (NGLs) that are produced along with and
extracted from natural gas (see NATURAL GAS). LPG is also produced from the refin-
ing of crude oil (see PETROLEUM). Although LPG is commercially defined as propane
[74-98-6], butane [106-97-8], and butane–propane mixtures, commercial availabil-
ity is primarily limited to propane (see HYDROCARBONS). There are two grades of spe-
cification propane, propane HD-5 and special-duty propane. The primary difference
in the two grades is that the Propylene [115-07-1] content of propane HD-5 is
restricted to a maximum of 5 vol%. Propylene (qv) is found only in refinery-
produced propane. The principal uses of LPGs are as fuels and feedstocks (qv)
for the production of motor gasoline and a wide variety of chemicals (see GASOLINE
AND OTHER MOTOR FUELS).
Other natural gas liquids include natural gasoline [8006-61-9], which is com-
posed of the pentanes and heavier components of the natural gas stream, and ethane
[74-84-0]. Most recently ethane has become the principal product of natural gas proces-
sing plants.

1. Properties

In general, LPG specifications involve limits for physical properties. Conse-


quently, the composition of the commercial-grade products varies between

1
Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.1209172120012512.a01
2 LIQUEFIED PETROLEUM GAS Vol. 2

Table 1. Physical Properties of LPG Componentsa


CAS Boiling Vapor Liquid
Registry Molecular point, pressure, density,
Component Number formula 101.3 kPa,b 8C 37.88C, kPab g/Lc
ethane [74-84-0] C2H6 88:6 354.9
propane [74-98-6] C3H8 42:1 1310 506.0
isobutane [75-28-5] C4H10 11:8 498 561.5
n-butane [106-97-8] C4H10 0:5 356 583.0
1-butene [106-98-9] C4H8 6:3 435 599.6
cis-2-butene [590-18-1] C4H8 3.7 314 625.4
trans-2-butene [624-64-1] C4H8 0.9 343 608.2
n-pentane [109-66-0] C5H12 36.0 107 629.2
a
Ref. 1.
b
To convert kPa to psi, multiply by 0.145.
c
At saturation pressure.

wide limits. Physical properties of the principal components of LPG are sum-
marized in Table 1 (1).

2. Manufacture and Processing

LPG recovered from natural gas is essentially free of unsaturated hydro-


carbons, such as propylene and butylenes. Varying quantities of these
olefinsmay be found in refinery production, and the concentrations are
a function of the refinery’s process design and operation. Much of the pro-
pylene and butylene are removed in the refinery to provide raw materials
for plastic and rubber production and to produce high octane gasoline
components.
LPG is recovered from natural gas principally by one of four extrac-
tion methods: turboexpander, absorption, compression, and adsorption.
Selection of the process is dependent on the gas composition and the degree
of recovery of ethane and LPG, particularly from large volumes of lean natu-
ral gas.
2.1. Turboexpander Process. Ethane has become increasingly desir-
able as a petrochemical feedstock resulting in the construction of many plants
that recover the ethane from natural gas at 73 to 93 C. Combinations of
external refrigeration and liquid flash-expansion refrigeration with gas turboex-
pansion cycles are employed to attain the low temperatures desired for high
ethane recovery. Figure 1 is a flow diagram of a one-expander cycle having exter-
nal refrigeration.
Dry inlet gas that has been dehydrated by Molecular sieves or alumina beds
to less than 0.1 ppm water is split into two streams by a three-way control valve.
Approximately 60% of the inlet gas is cooled by heat exchange with the low pres-
sure residue gas from the demethanizer and by external refrigeration. The
remainder of the inlet gas is cooled by heat exchange with the demethanized
Vol. 2 LIQUEFIED PETROLEUM GAS 3

Recompressor
Residue gas
to pipeline

Turbo
Compressor expander

Hot-gas Cold-gas
exchanger

Condensate
exchanger

separator
Dehydrated
gas inlet
Control Propane
valve refrigerant Demethanizer

Chiller

Side heater
Product
exchanger Reboiler

LPG and Product pump


natural-gasoline
liquids product

Fig. 1. One-expand cycle with external refrigeration for high ethane recovery in the
hydrocarbon liquid product.

bottoms product, the reboiler, and the side heater. A significant amount of low
level refrigeration from the demethanizer liquids and the cold residue gas stream
is recovered in the inlet gas stream.
The two portions of the feed stream recombine and flow into the high pres-
sure separator where the liquid is separated from the vapor and is fed into an
intermediate section of the demethanizer with liquid level control. The decrease
in pressure across the level-control valve causes some of the liquid to flash which
results in a decrease in the stream temperature. The pressure of the vapor
stream is decreased by the way of a turboexpander to recover power, thus to
achieve more cooling than would be possible by Joule-Thompson expansion.
The outlet of the turboexpander then is fed into the top of the demethanizer
where the separation of liquid and vapor occurs. The vapor is passed as cold resi-
due to the heat exchanger and the liquid is distributed to the demethanizer top
tray as reflux.
Essentially all of the methane [74-82-8] is removed in the demethanizer
overhead gas product. High recovery of ethane and heavier components as
demethanizer bottoms products is commonplace. The work that is generated
by expanding the gas in the turboexpander is utilized to compress the residue
gas from the demethanizer after it is warmed by heat exchange with the inlet
gas. Recompression and delivery to a natural gas pipeline is performed down-
stream of the plant. A propane recovery of 99% can be expected when ethane
recoveries are in excess of 65%.
4 LIQUEFIED PETROLEUM GAS Vol. 2

Recoveries of 90–95% ethane have been achieved using the expander pro-
cesses. The liquid product from the demethanizer may contain 50 liquid vol %
ethane and usually is delivered by a pipeline to a central fractionation facility
for separation into LPG products, chemical feedstocks, and gasoline-blending
stocks.
2.2. Absorption. Oil absorption is another process used for recovery of
LPG and natural gas liquids from natural gas. Recovery is enhanced by lowering
the absorption temperature to –458C and by keeping the molecular weight of the
absorption oil down to 100. Heat used to separate the product from the absorp-
tion oil contributes to the cost of recovery. Therefore, this process has become less
competitive as the cost of energy has increased. A simplified flow diagram of a
typical oil-absorption process is shown in Figure 2.
The natural gas feed exchanges heat with the residue gas from the absorber
overhead. Ethylene glycol [107-21-1] is injected as an antifreeze and the stream
is cooled further by refrigeration to –378C. The gas from the cooler enters a glycol
separator where the glycol that contains water is separated from the natural gas

Compressor

Cooler
Refrigerating
Oil absorber

Exchanger cooler
separator
Glycol

inlet gas

deethanizer
Rich-oil
Rich-oil
flash

Cooler

Hydraulic
Glycol turbine
injection Exchanger Exchanger
Exchanger
Glycol
regeneration
Butane splitter

Drier Sulfur Sulfur Drier


removal removal
Depropanizer

deethanizer
Lean-oil

Isobutane
Debutanizer

n-Butane

Booster
Residue
natural gas Propane

Natural gasoline
liquids

Fig. 2. Refrigerated absorption process for the production of LPG and natural gasoline
liquids.
Vol. 2 LIQUEFIED PETROLEUM GAS 5

as a liquid phase. A dry glycol is recovered for recycling to the injection point by
distilling the water. The gas and any higher boiling hydrocarbons that are pre-
sent pass to the base of the absorber where the gas comes into contact with
absorption oil which enters at the top of the absorber at 37 C. Approximately
85% of the propane and essentially all of the higher boiling hydrocarbons are
absorbed in the oil. The overhead residue gas from the absorber (34 C) is heat
exchanged with the inlet gas; at 48C and 4.24 MPa (600 psig), the gas flows to
the booster where the pressure is increased to that of the natural gas line.
The rich oil from the absorber is expanded through a hydraulic turbine for
power recovery. The fluid from the turbine is flashed in the rich-oil flash tank to
2.1 MPa (300 psi) and –328C. The flash vapor is compressed until it equals the
inlet pressure before it is recycled to the inlet. The oil phase from the flash passes
through another heat exchanger and to the rich-oil deethanizer. The ethane-rich
overhead gas produced from the deethanizer is compressed and used for produ-
cing petrochemicals or is added to the residue-gas stream.
The bottoms, consisting of absorption oil, absorbed propane, and higher
boiling hydrocarbons, are fed to the lean-oil fractionator. The LPG and the nat-
ural gas liquids are removed as the overhead product from the absorption oil
which is removed as a kettle-bottom product.
The lean oil from the lean-oil fractionator passes through several heat
exchangers and then through a refrigerator where the temperature is lowered
to 37 C. Part of the lean oil is used as a reflux to the lower section of the
rich-oil deethanizer. Most of the lean oil is presaturated in the top section of
the deethanizer, is cooled again to 37 C, and is returned to the top of the absor-
ber, thus completing the oil cycle.
The overhead product from the lean-oil fractionator, consisting of propane
and heavier hydrocarbons, enters the depropanizer. The depropanizer overhead
product is treated to remove sulfur and water to provide specification propane.
The depropanizer bottoms, containing butane and higher boiling hydrocarbons,
enters the debutanizer. Natural gasoline is produced as a bottom product from
the debutanizer. The debutanizer overhead product is mixed butanes, which
are treated for removal of sulfur and water, then fed into the butane splitter.
Isobutane is produced as an overhead product from the splitter and n-butane
is produced as a bottoms product.
2.3. Compression. Compression is the simplest and the least effective
of the four recovery methods. It was the first process used for the recovery of
hydrocarbon liquids from natural gas but is used only in isolated cases. The
most significant application of the compression process is for gas-cycling plants
where the natural gas liquids are removed and the remaining gas is returned to
the production formation. Figure 3 is a schematic of a typical gas-cycle plant.
The pressure used in producing gas wells often ranges from 690– 10,300 kPa
(100–1500 psi). The temperature of the inlet gas is reduced by heat-exchange
cooling with the gas after the expansion. As a result of the cooling, a liquid
phase of natural gas liquids that contains some of the LPG components is formed.
The liquid is passed to a set of simple distillation columns in which the most vola-
tile components are removed overhead and the residue is natural gasoline. The
gas phase from the condensate flash tank is compressed and recycled to the gas
producing formation.
6 LIQUEFIED PETROLEUM GAS Vol. 2

Compressor

Gas returned Exchanger


to formation
Water
cooler

Gas
Expander
well Compressor

Butane Propane

Condensate
separator
Debutanizer

Debutanizer

Condensate
flash tank

Natural gasoline

Fig. 3. High pressure gas-cycling plant with compression.

Condensable liquids also are recovered from high pressure gas reservoirs by
retrograde condensation. In this process, the high pressure fluid from the reser-
voir produces a liquid phase on isothermal expansion. As the pressure decreases
isothermally the quantity of the liquid phase increases to a maximum and then
decreases to disappearance. In the production of natural gas liquids from these
high pressure wells, the well fluids are expanded to produce the optimum
amount of liquid. The liquid phase then is separated from the gas for further pro-
cessing. The gas phase is used as a raw material for one of the other recovery
processes, as fuel, or is recompressed and returned to the formation.
2.4. Adsorption. Adsorption processes have been used to recover hydro-
carbons that are heavier than ethane from natural gas. Although the adsorption
process has applications for the recovery of pentane and heavier hydrocarbons
from lean gas, the percentage recovery of LPG components in these plants
usually is low compared to the normal recovery of LPG in modern turboexpander
or oil-absorption plants.
A simplified flow diagram for the fast-cycle adsorption plant is shown in
Figure 4. Activated carbon, alumina gel, and silica gel are used as adsorbents.
Use of internal insulation in the adsorption towers affords less cycle time. The
complete process cycle consists of three phases: regeneration, cooling, and
adsorption. In Figure 4, the inlet gas is divided. The larger portion flows directly
to the bed in the adsorption cycle; the smaller portion flows first through the bed
in the cooling phase of the cycle, then through the regeneration cycle. The efflu-
ent gas from the cooling cycle is heated further to 260–3158C in a separate fired
Vol. 2 LIQUEFIED PETROLEUM GAS 7

Booster Compressor

Inlet Adsorption
gas towers

Low pressure
High pressure
Regeneration
Adsorption

separator
Cooling

flash
cycle
cycle

cycle

Exchanger

Regeneration
furnace Hydrocarbon liquids:
Cooler LPG and natural
gasoline

Dry gas

Fig. 4. A fast-cycle absorption unit for recovery of hydrocarbon liquids from


natural gas.

furnace before it is injected into the bed in the regeneration cycle. The effluent
from the regeneration cycle contains the condensable hydrocarbons that have
been stripped from the adsorbent and that are removed in the high pressure
separator after the gas that is leaving the regeneration cycle is cooled. The liquid
from the high pressure separator is flashed through to remove light hydrocar-
bons. The flash vapor is compressed and is mixed with the vapor from the high
pressure separator. The pressure on the compressed gas stream is boosted to the
inlet gas pressure and the gas is recycled to the feed to the adsorption cycle. The
liquid from the low pressure flash, which contains LPG and natural gas liquid
components, is processed further in a series of distillation towers to produce pro-
pane, butanes, and natural gasoline.
Less propane and butanes are produced compared to natural gas liquids by
the adsorption process than are obtained normally for the same gas by the oil-
absorption process. Because adsorption efficiency increases with a decrease in
temperature, the adsorption cycle should operate at the lowest temperature
that is economically feasible.
2.5. Purification. The LPG generally requires treatment for removal of
hydrogen sulfide [7783-06-4], H2S, organic sulfur compounds, and water in order
to meet specifications. Several methods are used.
Amine Treatment. The LPG is brought into contact with a 15– 20 wt%
solution of ethanolamine [141-43-5] in water, which removes H2S to specifica-
tion levels. The foul amine solution that contains the H2S is regenerated in
a stripper at low pressure using indirect stream stripping, and the stripped
amine solution is returned to the LPG contactor. The amine solution gene-
rally has little affect on organic sulfur compounds, although diethanolamine
8 LIQUEFIED PETROLEUM GAS Vol. 2

[111-42-2] (DEA) and diglycolamine [929-06-6] (DGA) have been used to obtain
acceptable carbonyl sulfide [463-58-1], COS, levels in propane.
Caustic Treatment. Amine treatment may be followed by a caustic treat-
ment step in which the LPG is brought into contact with 10–20 wt % caustic
solution to remove any residual H2S and to remove mercaptans. The mercaptans
may be stream-stripped from the caustic solution, after which the stripped caus-
tic is recycled to the LPG caustic contactor. Caustics that contain H2S must be
discarded. Various promoters may be added to the caustic to improve the effi-
ciency of the mercaptan removal.
Coalescing. Sand towers or cartridge-type coalescers may be used to
separate any undissolved water from the LPG. Removal of the undissolved
water meets the specification moisture limit for butanes. However, this step
does not produce specification propane.
Solid-Bed Dehydration. Silica gel, bauxite, activated alumina, or molecu-
lar sieves can be used for removing dissolved water to meet propane specifica-
tions. The solid-bed dehydrators are used in a cyclic adsorption process. After
an adsorption cycle has completed, the bed is heated with a purge gas or a vapor-
ized liquid-product stream for regeneration. If the latter is used, the liquid pro-
duct is condensed, separated from the free water, and returned to the process.
After the beds are regenerated, they are cooled and returned to the adsorption
cycle.
Molecular Sieve Treatment. Molecular sieve treaters can be designed to
remove H2S, organic sulfur compounds (including carbonyl sulfide), and water in
one step. Solid-bed units are utilized and regeneration occurs in the same man-
ner as simple, solid-bed dehydrators.
Solid-Bed Caustic Treatment. Solid-bed caustic units utilizing methanol
[67-56-1] injection into the LPG feed stream can be used for carbonyl sulfide
removal. The methanol–caustic solution must be drained periodically from the
beds and discarded. When the solid bed is exhausted, the spent caustic must
be discarded and replaced. The LPG from the treater has a low enough water
content to meet the propane specification.
Fractionation. Direct fractionation also can be used to remove dissolved
water from LPG. The water-rich overhead vapor from the dryer fractionator is
returned to the fractionator as reflux and the water phase is discarded. A dry
LPG product that meets either propane or butane water specifications is pro-
duced as a kettle product from the fractionator.

3. Production and Shipment

Historically, about two-thirds of the LPG produced in the United States came
from natural gas processing and one-third was produced from refinery opera-
tions (2). In 1991, this ratio was 61% from natural gas processing and 39%
from refinery operations. Total production of LPG in 1991 was 76:85  106 m3
ð294:19  106 bblÞ from natural gas processing and 30:08  106 m3 ð189:23
106 bblÞ produced from refinery operations.
Ethane production for 1991 was 30:74  106 m3 ð193:32  106 bblÞ from natu-
ral gas operations and 1:49  106 m3 ð9:34  106 bblÞ from refinery operations for
Vol. 2 LIQUEFIED PETROLEUM GAS 9

Table 2. 1991 U.S. NGL Supply and Demand, 106 m3a


Places Ethane Propane n-Butane i-Butane Pentanes b Total
Supply
gas plants 30.74 28.28 8.72 9.78 18.79 96.30
refineries 1.49 25.27 4.56 0.25 31.57
imports 0.48 5.29 2.04 0.73 1.31 9.85
from (to) stocksc 0.35 0:27 0.06 0.28 0:18 0.23
Total 33.05 58.57 15.37 11.04 19.92 137.95
Demand
chemicals 32.09 24.36 5.23 4.81 66.49
RES and COMd 20.16 0.00 20.16
engine fuel 2.05 0.00 2.06
industrial 0.05 4.61 1.70 6.36
utility 0.61 0.26 0.02 0.89
gasoline 5.53 11.04 12.93 29.50
farm 4.29 0.00 4.29
export 0.01 1.61 0.80 0.03 2.45
other 0.30 1.23 2.08 2.15 5.76
Total 33.06 58.57 15.37 11.04 19.91 137.95
a 3
To convert m to bbl, divide by 0.159.
b
Includes Cx compounds where x  5.
c
Negative sign indicates supply to stocks (inventory) in order to balance supply/demand.
d
RES and COM ¼ research and commercial.

a total production of 32:21  106 m3 ð202:66  106 bblÞ. A summary of total nat-
ural gas liquids (NGL) including LPG supply and demand in the United States is
shown in Table 2.
The progress of LPG utilization has been closely related to progress in
transportation and storage of this fuel. Large volumes of LPG must be trans-
ported from the producing plants to centers of consumption, and transportation
costs are a principal factor in the cost of LPG to the consumer. Large volumes of
LPG usually are transported by high pressure pipelines (qv). As of this writing,
this use of pipelines is increasing rapidly. Large quantities of LPG are trans-
ported in railroad tank cars which have an average capacity of 113:5 m3 =car
(3  104 gal=car), although this use is decreasing. Tank-truck transports having
capacities of about 17.8 m3 (104 gal) also move large quantities of product from
producing plants to distribution centers and from the pipeline terminals to
points of distribution or consumption. Delivery trucks having capacities from
3.8–11.4 m3 (1000–3000 gal) generally are used for the final delivery from the
distribution bulk plant to the storage plants of the larger consumers. Smaller
quantities are shipped in metal cylinders having capacities from 9–45 kg LPG.
These cylinders are filled at the distribution bulk plant and are delivered by
truck to the consumer or to the cylinder dealer or cylinders may be filled for
the customer at small distribution stations.
Tankers and barges are also used for transporting LPG. Tankers and
barges are designed for both high pressure ambient temperature and for low
pressure refrigerated transportation. For larger volumes and long distances,
low pressure refrigerated tankers almost always are used (2).
10 LIQUEFIED PETROLEUM GAS Vol. 2

Ethane usually is transported in high pressure pipelines from the point of


production to the point of consumption. However, for small quantities of ethane,
multitube trunk trailers are used.

4. Economic Aspects

The production and consumption of LPG in the United States increased drama-
tically from its early beginnings in the 1930s until the international energy crises
of the 1970s when rising prices and regulatory restraints resulted in reduced
domestic production. However, total consumption, including imports, resumed
a modest growth characteristic after that time. In 1984, total LPG consumption
in the United States was 76:58  106 m ð481:71  106 bblÞ; by 1991, total con-
sumption was 84:98  106 m3 ð534:50  106 bblÞ. The principal growth segment
has been the increasing use of LPG for petrochemical feedstocks as can be
seen from Table 3.

5. Specifications and Standards

Specifications for the principal LPG products are summarized in Table 4.


Detailed specifications and test methods for LPG are published by the Gas Pro-
cessor’s Association (GPA) (3) and ASTM (4). The ASTM specification for special-
duty propane and GPA specification for propane HD-5 apply to propane that is
intended primarily for engine fuel. Because most domestic U.S. LPG is handled
through copper tubing, which could fail if corroded, all products must pass the
copper strip corrosion test. A test value of No. 1 represents a LPG noncorrosive
to the copper.

6. Storage

Large volumes of LPG are stored to meet peak demand during cold seasons.
LPGs are both volatile and flammable and must be stored and handled in special
equipment. Standards for storing and handling LPG are published by the
National Fire Protection Association (5) and API (6).
Four main types of storage are used: high pressure storage above ground,
low pressure refrigerated storage above ground, frozen earth storage, and under-
ground cavern storage. The capacities of the storage unit vary from 500 mL pres-
sure cylinders to 1:9  106 m3 ð500  106 galÞ underground storage caverns.
Economic factors determine the proper storage for any given requirement.
Above ground pressure-storage tanks usually are designed for a 1720 kPa
(250 psi) working pressure for propane and 860 kPa (125 psi) for butane. Refri-
gerated, aboveground storage tanks usually are designed for a few kilopascals of
pressure. These tanks must be coupled with refrigeration systems to cool the pro-
duct that is to be stored to a temperature equal to the product’s boiling point at
the operating pressure of the tanks. Vapors generally are recondensed by refrig-
eration and returned to the tanks.
Table 3. Sales of LPG 1984–1991, 106 m3a,b
Residential and Motor fuel
Year Chemical commercial heating Engine fuel Industrial Utility gas blending Farm Export Other Total
1984 47.49 15.05 2.80 10.06 1.45 35.71 3.77 2.93 11.82 2,115.08
1985 51.57 16.97 2.78 7.66 0.91 37.09 4.38 3.72 8.90 2,118.98
1986 48.72 16.62 2.48 8.23 2.73 36.36 4.29 2.59 2.90 2,110.92
1987 54.45 18.39 2.39 8.35 0.54 37.69 4.08 2.33 2.96 2,118.17
1988 60.24 18.25 2.21 8.19 1.23 35.15 4.03 2.96 3.97 2,124.23

11
1989 55.52 20.41 2.20 9.90 1.14 33.26 4.44 2.37 6.74 2,123.99
1990 57.57 18.84 2.01 5.80 1.06 32.71 4.30 2.16 3.78 2,118.23
1991 66.49 20.16 2.06 6.36 0.89 29.50 4.29 2.45 5.76 2,128.95
a
Ref. 2.
b
To convert m3 to bbl, divide by 0.159.
12 LIQUEFIED PETROLEUM GAS Vol. 2

Table 4. Commercial Liquefied Petroleum Gas Specificationsa,b


Butane–propane
Parameter Propane Butane mixture Propane HD-5
composition predominantly predominantly butanes and/or 90 liquid
propane and/or butane and/or butylenes plus vol % propane,
propylene butylenes propane and/or 5 liquid vol %
propylene propylene
vapor pressure 1434 482.6 1434 1434
at 37.788C, kPac
temperature at 95% 38:3 2.2 2.2 38:3
evaporation,d8C
total sulfur, ppmw 185 140 140 123
moisture pass pass
free water none none
a
Refs. 3 and 4.
b
Corrosion copper strip test, maximum ¼ No:1.
c
To convert kPa to psi, multiply by 0.145.
d
Values are maximum.

In frozen earth storage of propane, the walls and bottom of a pit in the ground
are frozen and a dome is constructed over the pit. The pressure in the storage
cavern is maintained at nearly atmospheric pressure by refrigeration systems
that cool the product to its boiling point at storage pressure. Heat leaks into the
cavity and vaporizes some of the propane. The vapor that is formed is compressed,
cooled, and returned to the pit as a liquid by the refrigeration system. Because this
storage must operate at temperatures considerably below the freezing point of wet
earth and at atmospheric pressure, it cannot be used for butane storage.
Underground storage caverns, which operate at approximately formation
temperatures and at the corresponding LPG vapor pressure, may be either
mined underground storage caverns or cavities that have been produced in a
salt formation by solution mining. The underground caverns must be of sufficient
depth to develop an overburden pressure greater than the vapor pressure of the
stored liquid. Mined storage caverns are 60–152 m deep, whereas salt formation
caverns may be from 106–1524 m deep. Underground as compared to above-
ground storage is much more economical for storage of large volumes, ie, more
than 2785 m3 (107 gal) of LPG. A washed-out salt cavern costs only 10–50% as
much to develop as typical mined cavern storage. In 1987, underground storage
capacity for LPG in the United States was 79  106 m3 ð493  106 bblÞ (2) and
90% of this was in salt formations.

7. Uses

About 35% of total U.S. LPG consumption is as chemical feedstock for petrochem-
icals and polymer intermediates. The manufacture of polyethylene, Polyethylene
and poly(vinyl chloride) requires huge volumes of ethylene (qv) and propylene
which, in the United States, are produced by thermal cracking/dehydrogenation
of propane, butane, and ethane.
Vol. 2 LIQUEFIED PETROLEUM GAS 13

Residential and commercial fuel demands represent about 24% of total U.S.
LPG consumption. Although this market demand is weather dependent, it has
assumed the characteristics of a mature market. Growth is related to the general
economic trends.
Nearly two-thirds of total butane supply, about 20% of total LPG, is con-
sumed in the manufacture of motor gasoline. However, the environmental man-
dates of the early 1990s have had a negative impact on this market segment.
These mandates have reduced gasoline volatility requirements, effectively redu-
cing the value of the butanes as blending stocks. However, normal butane can be
used as feedstock for production of isobutylene, a key ingredient of ether blend-
stocks, such as methyl tert-butyl ether [1634-04-4] (MTBE) for motor gasoline.
Shifts in U.S. use patterns can be seen in Table 3.

BIBLIOGRAPHY

‘‘Liquefied Petroleum Gas’’ in ECT 2nd ed., Vol. 12, pp. 470–480, by L. Pollack, Phillips
Petroleum Co.; in ECT 3rd ed., Vol. 14, pp. 383–394, by F. E. Selim, Phillips Petroleum
Co.

CITED PUBLICATIONS

1. Engineering Data Book, 10th ed., Gas Processors Supplier’s Association, Tulsa,
Okla., 1987.
2. LP-Gas Market Facts, National LP-Gas Association, Oak Brook, Ill., 1977.
3. Liquefied Petroleum Gas Specifications and Test Methods, Gas Processors Associa-
tion, GPA Publication 2140-92, Tulsa, Okla.
4. ASTM Standard D1835-91, American Society for Testing and Materials,
Philadelphia, Pa., 1992.
5. Storage and Handling of Liquefied Petroleum Gases, National Fire Protection
Association, NFPA 58, Boston, Mass., 1989.
6. Design and Construction of LP-Gas Installations at Marine Terminals, Natural Gas
Processing Plants, Refineries, Petrochemical Plants, and Tank Farms, API Standard
2510, 4th ed., American Petroleum Institute, Washington, D.C., Dec. 1978.

GENERAL REFERENCES

Petroleum Products Handbook, McGraw-Hill Book Co., Inc., New York, 1960.
C. C. McKee, ‘‘The Supply/Demand Outlook for LP-Gas,’’ Proceedings of the 58th Annual
Convention 1979, Gas Processors Association, Tulsa, Okla.
Magic Formula, LP-Gas, Duluth, Minn., Jan. 1980.
Stayin’ Alive, LP-Gas, Duluth, Minn., Jan. 1980.

R. RAY TAYLOR
Phillips Petroleum Company
14 LUBRICATION AND LUBRICANTS Vol. 2

LUBRICATION
AND LUBRICANTS
1. Introduction

Lubrication is a process in which a film of lubricant is inserted between rubbing


surfaces for the purpose of controlling friction and/or to reduce wear of the sur-
faces. These films are designed to minimize contact between the rubbing surfaces
and to shear easily so that the frictional force opposing the rubbing motion is low.
Lubricants may be liquids, solids, gases, or greases. Lubricating oils and greases
contain refined or synthesized base oils from animal, vegetable or mineral
(petroleum) origin, and a variety of additives to improve their lubricating and
other characteristics.
Lubrication is a major component of tribology, defined as the science and
technology concerned with interacting surfaces in relative motion, including fric-
tion, lubrication, wear and erosion (1).
Tribology and lubrication are ancient arts. In his splendid History of Tribo-
logy (2), Professor Dowson traces the development of these arts and sciences, and
describes the outstanding artists and scientists responsible, from the paleolithic
age to the end of the twentieth century. He reports archeological evidence that
bitumen was used to lubricate potters wheels 5000 years ago. Water-lubricated slid-
ing bearings were used in Egypt 2400 BC to transport large objects. A chariot wheel
from 1400 bc was found with traces of tallow as lubricant, and the Chinese had
lubricated metal wheel bearings with leather seals to hold the lubricant in place
in the fourth century bc (2).
The word ‘‘tribology’’ first appeared in Lubrication (Tribology) Education
and Research—A Report on the Present Position and Industries Needs, Depart-
ment of Education and Science (UK), 1966. This is often called The Jost Report,
after H. Peter Jost, the chairman of the British Lubrication Engineering Work-
ing Group, which prepared the report. The word is derived from the Greek
tribein, meaning ‘‘to rub’’, and logos, meaning ‘‘reading’’ or ‘‘study’’. Tribology is
literally the study of rubbing. The Working Group defined it more precisely as
‘‘the science and technology of interacting surfaces in relative motion and the prac-
tices related thereto’’ (3).
The Jost Report was part of an effort to focus attention on the ‘‘scientific,
technological, economic and environmental issues’’ (2) involved in the study
and practice of tribology. Another purpose was to bring together the many,
and often splintered engineering and scientific disciplines that deal with this
technology. Dowson speculates that the dramatic progress in this field in the
final third of the twentieth century may have been significantly influenced by
these efforts in the 1960s (2).
The 1966 report by the British Lubrication Engineering Working Group
demonstrated to industry and government the impact of friction, wear, and lubri-
cation on the nation’s economy, and the value of further research in tribology.
That report showed, eg, that the most significant value of better lubrication
(91%) comes from increased productivity, lower maintenance and replacement
costs, and lower investment cost. Direct energy savings (5%), and savings

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.1221021802151519.a01.pub2
Vol. 2 LUBRICATION AND LUBRICANTS 15

in the cost of lubrication, in manpower and material (4%), account for the
remainder.
Tribology is a multidisciplinary science that embraces lubrication, friction,
wear, properties of lubricants, surface characterization, bearing materials, and
the selection and design of lubricating systems. The lubrication engineer
would add to this list lubricant and coolant selection, plant lubrication and
maintenance programs, and machine condition monitoring.

2. Fundamentals of Lubrication

Tribology, by definition, is concerned with interacting surfaces in relative


motion. It is appropriate, therefore, to begin the discussion of lubrication funda-
mentals by describing the characteristics of tribological surfaces.
2.1. The Nature of Interacting Surfaces in Relative Motion. Tribo-
logical surfaces are the load-bearing surfaces on the moving parts of
machines. They include surfaces on crankshaft rod and main bearings, radial
and thrust bearings on steam and gas turbines, cams and valve lifters, pistons
and cylinders, natural and artificial hip joints, ball and roller bearings,
machine tool slideways, cutting tools, magnetic information storage devices,
and microelectromechanical systems (MEMS). Despite their appearance
and finishing efforts, these surfaces are not perfectly smooth. There are micro-
scopic irregularities; gently sloping hills and valleys called asperities on
them.
If an imaginary surface is drawn through a real surface, such that the
volume of all of the material above the imaginary surface is equal to the
volume of voids below that surface, the roughness of the real surface, Ra,
can be defined as

Ra ¼ ðj y1 j þ j y2 j þ þ j yn jÞ=n ð1Þ

Where Ra is the center line or arithmetic average of the absolute distances, yi,
from the imaginary surface (mean line) for a given sampling length (usually
0.80 mm).
Roughness (Ra) values of machined surfaces range from 0.025 mm for ball
bearing surfaces to 25 mm clearance surfaces on rough machine parts (4–6).
The roughness of computer hard disk surfaces is measured in angstroms (Å) or
nanometers (nm) (7).
The total profile of a surface consists of a ‘‘waviness’’ and a roughness com-
ponent. The parameter Ra, although it is the most common measure of surface
roughness, is insensitive to the shape or waviness of the profile. A more useful
parameter is the root-mean-square (rms) roughness, Rq.

Rq ¼ ½ð y21 þ y22 þ þ y2n Þ=n1=2 ð2Þ

Where Rq is the rms deviation of yi from the mean line for a given sampling
length (4). The rms roughness of computer hard disk surfaces is <2 nm (7).
16 LUBRICATION AND LUBRICANTS Vol. 2

The most repeatable of the roughness parameters is the 10 points


height, Rz.

Rz ¼ ½ðP1 þ P2 þ þ P5 Þ  ðV1 þ V2 þ þ V5 Þ=5 ð3Þ

Where Rz is the average distance between the five highest peaks (Pi) and
the five deepest valleys (Vi) within the sampling length. It is also linked
with the machining parameter S2/8r, where S is the feed rate and r is the
tool radius (4).
Another index of roughness is Rt, the maximum peak-to-valley height.

Rt ¼ Rp þ Rv ð4Þ

Where Rp is the maximum peak height and Rv is the maximum valley depth
within the sampling length (4).
The description of the test disk used in the ASTM D 6425-02 test method for
measuring friction and wear illustrates the impact of surface texture on these
quantities (8) and is shown below and used with ASTM’s permission.

Test Disk, AISI 52100 Steel, 62  1 HRC Hardness. The surfaces of the disk
are lapped and free of lapping raw materials. The topography of the disk will be
determined by four values: 0.005 mm hRzi 0.65 mm; 0.035 mm hRai 0.050 mm;
0.020 mm hRpi 0.035 mm; 0.050 mm hRvi 0.075 mm

Four different measures of surface topography are specified for the disk in
order to get acceptable reproducibility of the test method (9).
Approximate roughness indexes obtained with various metal-working
processes are

Production process Rt, mm Ra, mm


turning 4.00–25.0 0.500–3.0
grinding 2.00–6.0 0.400–0.8
milling 1.50–20.0 0.200–2.0
boring 0.50–20.0 0.050–1.6
honing 0.03–1.0 0.015–0.2
lapping 0.03–0.6 0.015–0.1

Engineering surfaces also differ in composition from the underlying bulk


material. A metal bearing, eg, will have a work hardened layer at the surface,
over which an oxide layer forms, on top of which is an adsorbed layer of moisture
and gases. When two such surfaces come in moving contact, their surface struc-
tures, compositions, and the interaction between opposing asperities accounts
for a major portion of the friction between them and much of the wear that inevi-
tably occurs (10–12).
For example, without the oxide and adsorbed layers, coefficients of friction > 4
have been measured in a vacuum of 0.133 mPa on surfaces cleaned by abrasive
cloth and heated. The coefficient decreased considerably when oxygen was
Vol. 2 LUBRICATION AND LUBRICANTS 17

admitted to a pressure of 0.133 Pa (13). The oxide and adsorbed layers on metal
surfaces can, therefore, be considered as lubricating films.
2.2. Friction. When two of these surfaces are brought together, they
initially touch at the highest asperities. The load, N, normal to the surfaces at
the contact points causes the asperities to deform until the pressure in the result-
ing contact areas just equals the yield pressure, p, of the asperities. The sum of
these contact areas is the real contact area, Ar. The yield pressure is equivalent
to the Brinell Hardness Number (BHN), in consistent units, measured at the
surface of the material (4).

Ar ¼ N=p ¼ N=BHN ð5Þ

The real area of contact is a minute fraction of the total surface area. For
example, with a typical bearing contact stress of 3 MPa and a bronze bearing
asperity yield pressure of 500 MPa, <1.0% of the nominal area would involve
asperity contact.
As the load on the surfaces increases, the asperities continue to deform,
the softer surface more than the other. More of the asperities come in contact
and the real area of contact grows. The opposing surfaces also tend to adhere
or bond to each other in the contact area. The shear strength of these bonds
depends on the time of contact and the difference in composition of the two
surfaces.
Sliding of one of these nonlubricated surfaces across the other requires a
friction force, F, to displace the contacting asperities. This force includes several
components, among them: a shear or adhesion component arising from bonding
of the contacting asperities; a plowing or deformation component, arising from
the interlocking of asperities; a lifting component to raise asperities over the
roughness of the mating surfaces.
The shearing component, Fs, may account for 90% or more of the total fric-
tion force. This component is proportional to the shear strength, s, of the asperity
junctions:

Fs ¼ Ar s ð6Þ

More detailed descriptions of surface texture, surface structure, and compo-


sition and the real area of contact will be found in Refs. (4–6,12–15).
2.3. Coefficient of Friction. The coefficient of friction, f, of a pair of con-
tacting surfaces is defined as the ratio of the total frictional force to the normal
force or load. It can also be expressed as the ratio of shear strength, s, to the yield
pressure, p, at the asperity junctions.

f ¼ F=N ¼ s=p ð7Þ

If there is a lubricating film on the surfaces, the coefficient of friction is the


ratio of the shear strength of the surface film, sf, to the yield pressure, pm of the
substrate or backing material.

f ¼ sf =pm ð8Þ
18 LUBRICATION AND LUBRICANTS Vol. 2

If a shear force is gradually applied to one of two dry, unlubricated surfaces


in contact, the surface will not move until the force is great enough to overcome
the shear strength of the asperity contacts. The ratio of the shear force required
to start motion to the normal force on the surfaces is the static coefficient of
friction. Once motion starts, less force is needed to keep the surface moving
at a constant velocity. The coefficient of friction during sliding is the kinetic or
dynamic coefficient.
The static coefficient measured for a hard steel surface on another hard
steel surface is 0.78. The dynamic coefficient measured for hard steel on hard
steel is 0.42. When a thin film of light mineral oil is applied to these surfaces,
the static coefficient drops to 0.23. The dynamic coefficient with a light oil film
drops to 0.1. Adding a friction modifier to the oil can reduce or reverse the dif-
ference between the two coefficients. Adding stearic acid to the lubricant, eg, for
hard steel on hard steel, reduces the static coefficient to 0.0052, which is lower
than the dynamic coefficient, 0.029 (16).
An extensive Friction and Wear Databank is found in Refs. 17,18. Tables of
the coefficient of friction values for a wide variety of material combinations are
also available in Refs. 19 and 20 and many other sources. These data, however,
should be used with caution. The coefficient of friction varies with changes
in humidity, gas pressure, time, temperature, sliding speed, surface quality,
the shape of the contact region, the method of testing, and other variables.
Where high reliability is needed, the friction should be measured using a proto-
type device under design conditions (20).
2.4. Wear. The principal types of wear in sliding contacts are adhesive,
abrasive, and corrosive wear. Fatigue wear occurs in concentrated contacts (ball
and roller bearings, gears, cams, and automotive valve lifters) under the combi-
nation of sliding and rolling (21).
The Archard equation reported by Rabinowicz (21) gives a simple, quanti-
tative relationship for predicting the adhesive wear rate:

V ¼ kNx=p ð9Þ

where V ¼ wear volume, k ¼ wear coefficient, N ¼ normal load, x ¼ sliding


distance, p ¼ yield stress or indentation hardness.
Values of k for several unlubricated material combinations are shown in
Table 1 (14). Others will be found in Ref. 18 and in Refs. (21–24).

Table 1. Wear Coefficients for Various Sliding Combinationsa


Material combination Wear coefficient, k
zinc on zinc 53  103
low carbon steel on low carbon steel 7  103
copper on copper 11  103
stainless steel on stainless steel 7  103
copper on mild steel 5  104
mild steel on copper 17  105
phenolic resin on phenolic resin 7  106
a
For equation 9, from Ref. 14.
Vol. 2 LUBRICATION AND LUBRICANTS 19

Adhesive wear is material separated or transferred during the shearing


of asperity contacts. These wear particles, and other particulate surface
contaminants that are hard enough to damage the surface, cause abrasive
wear. Abrasive wear is the removal of material by ploughing, cutting, or scratch-
ing. Its rate generally obeys equation 9 and the wear coefficients tend to be
higher than the adhesive coefficients (21). Corrosive wear is the wearing away
of the products of galvanic or chemical corrosion of the surface. There is no sim-
ple equation that characterizes this type of wear.
Current broader discussions of the friction and wear phenomena are found
in Refs. 13, 25, and 26.
2.5. Viscosity. Viscosity, or resistance to flow is the most important
property of a lubricating oil. It was defined by Newton (27) as the ratio of the
shear stress, T, divided by the shear rate, dU/dh, in a fluid during flow.
Newton’s law of viscosity is given in equation 10 (28).

T ¼ Fs =As ¼ m dU=dh ð10Þ

The fluid in contact with the surface of the moving plate has the same velo-
city, U, as the plate. The fluid in contact with the stationary plate has zero velo-
city. There is, therefore, a shear stress, T, on the fluid equal to the force, Fs,
required to keep the plate moving divided by the area, As, of the moving plate,
and is proportional to the velocity gradient, dU/dh, of the fluid. The parameter
Fs is a frictional force and the coefficient, m, is the dynamic viscosity of the fluid
(29). The unit of dynamic viscosity in the System International (SI) is pascal-
second (Pa s). The customary unit is centipoise (cP), which is Pa s  103.
Schoff and Kamarchik (30) describe methods of measuring the viscosity of a
wide variety of materials. Standard methods for measuring the dynamic viscos-
ity of lubricating oil can be found in Refs. 31–34.
Generally, however, it is easier to measure the kinematic viscosity, v, of a
lubricating oil using a capillary viscometer. Kinematic viscosity is the dynamic
viscosity divided by the density, d, of the fluid at the same temperature.

v ¼ m=d ð11Þ

The customary unit for kinematic viscosity is centistoke (cSt), which is


equivalent to millimeters squared per second (mm2/s). The most common method
for accurate measurement of kinematic viscosity over a wide temperature
range is ASTM D 445 (35). Older literature may report kinematic viscosities of
lubricating oils in Saybolt Universal Seconds (SUS). ASTM D2161-04, has been
retained for the purpose of calculating kinematic viscosity in centistokes from
SUS or SFS data (36).
Viscosity–Temperature Relationship. The MacCoull equation, also called the
Walther equation, relates the kinematic viscosity, v, of a liquid to its temperature.

log log Z ¼ A  Bðlog TÞ ð12Þ

Where Z ¼ (v þ 0.7) for 2.00 cSt  v  2  107 cSt; A and B ¼ constants, and
T ¼ absolute temperature, 8K ¼ temperature,8C þ 273.2.
20 LUBRICATION AND LUBRICANTS Vol. 2

It is applicable to homogeneous liquid lubricants with conventionally


refined hydrocarbon base oils and is valid between the cloud point at low tem-
perature and the initial boiling point ( 3408C) at higher temperatures. (The
cloud point is the temperature at which a cloud of wax crystals first appears
when it is cooled under conditions prescribed in ASTM D 2500.) This equation
is the basis of the viscosity–temperature charts described in ASTM D 341 (37).
Oils with ester, phosphate, silicone, and synthesized hydrocarbon base oils
follow the MacCoull (Walther) relationship over the range of 18 to 1758C to
within 5%. Many esters, synthesized hydrocarbons, and low pour point mineral
oils exhibit low temperature (40 to 548C) viscosities substantially below
MacCoull equation predictions (38).
Viscosity Index. Another widely used and accepted measure of the varia-
tion of viscosity with temperature is the viscosity index (VI). The higher the VI,
the lower is the variation of viscosity with temperature. The VIs for oils having
values <100 are calculated by

VI ¼ 100½ðL  UÞ=ðL  HÞ ð13Þ

where U ¼ kinematic viscosity at 408C of the oil whose viscosity index is to be


calculated; L ¼ kinematic viscosity at 408C of a ‘‘0’’ VI oil with the same viscosity
at 1008C as the unknown; H ¼ kinematic viscosity at 408C of a ‘‘100’’ VI oil with
the same viscosity at 1008C as the unknown.
Values of H and L for viscosities of 2 cSt and above are given in ASTM D
2270 (39).
Equation 13 gives confusing results for VI values >100. Viscosity indexes of
100 or greater are calculated by the empirical formula (39):

VI ¼ ½ððantilog NÞ  1Þ=0:00715 þ 100 ð14Þ

where YN ¼ H/U and Y ¼ kinematic viscosity at 1008C of the oil whose viscosity
index is to be calculated.
Viscosity index is sometimes used as a measure of the quality of lubricating
oil, especially for selecting base stocks for automotive engine oils and automatic
transmission fluids. This is not always applicable, however. The role of viscosity
index in base stock selection is described later.
Several aromatic type fluids, polyphenyl ethers, aryl phosphate esters, and
halogenated aromatic hydrocarbons have negative VIs.
Effect of Pressure on Viscosity. The lubricant film pressure in the con-
centrated contact areas of rolling element bearings, gears, cams, etc, can be as
high as 2000–3000 MPa. The viscosity of the film at these pressures could be
a million times higher than that of the lubricating oil at atmospheric pressure,
as illustrated in Fig. 1, or the lubricant may have solidified (38,40). The rate of
viscosity increase with pressure of a liquid lubricant varies with its composition
and chemical structure and with temperature and pressure.
Traction drives, found in some industrial machinery and the toroidal drives
in some continuously variable transmissions (CVTs) depend on this property of
traction fluids to transfer power.
Vol. 2 LUBRICATION AND LUBRICANTS 21

106

0°C
105

°C
33
Absolute viscosity, mPa·s(=cP)

104
°C
99

103

°C
218
102

10

1
0 200 400 600 800 1000
Pressure, MPa

Fig. 1. Viscosity pressure curve for typical petroleum oils, where —— ¼ Paraffinic, - - - -
¼ Alicyclic, " ¼ Solid.

Generalized pressure–temperature–viscosity relationships from extensive


data on petroleum and synthetic oils are described in Refs. (14,41,42).
Newtonian Versus Non-Newtonian Behavior. If the viscosity of a fluid
subjected to shear is independent of the rate of shear or magnitude of the
shear stress, it is a Newtonian fluid. Most industrial lubricating and hydraulic
oils are Newtonian fluids.
If the viscosity changes with shear stress or shear rate, the fluid is non-
Newtonian. This behavior is typical of multigrade engine oils and other oils con-
taining polymeric viscosity improvers.
The rate of decline in viscosity in a non-Newtonian lubricating oil is initially
slow, then reaches a maximum, and finally slows again. At very high shear rates,
the viscosity tends to level out and approach that of the base oil.
Grease also behaves in a non-Newtonian manner. At low shear rate, it acts
like a high viscosity semisolid and ‘‘stays in place’’. In a bearing, under
high shear rate, it acts more like its base oil and supports full fluid lubricating
films.
2.6. Lubrication Regimes. In the full fluid film regimes, the moving,
load-bearing surfaces are completely separated. There is no contact between
them. Resistance to motion arises solely from the internal friction of the fluid,
a function of its viscosity. Adhesive wear is absent. Wear may occur from surface
fatigue or from contamination of the fluid with corrosive or abrasive substances.
The coefficient of friction in a liquid lubricated system is dependent on,
among other things, the lubricant viscosity, the relative speed of the surfaces,
22 LUBRICATION AND LUBRICANTS Vol. 2

Fig. 2. Variation of coefficient of friction with ZN/P.

and the load on the surfaces. In a journal bearing, eg, the lubricant film thick-
ness and the coefficient of friction are functions of the dimensionless Sommerfeld
bearing characteristic number S.

S ¼ ðR2 =C2 ÞZN=P ð15Þ

where P is the average pressure on the bearing surface, W/2RL; Z is the dynamic
viscosity, m, of the fluid; and N ¼ is the rotational speed of the journal, U/2pR.
The relationship between the kinetic coefficient of friction and the dimen-
sionless quantity, ZN/P, is illustrated in Fig. 2 (10). This is known as a
Stribeck curve.
At values of ZN/P greater than 30, lubrication is in the full fluid film
regime. Friction increases with increasing ZN/P because of increasing resis-
tance to flow. As ZN/P is reduced, by reductions in the speed or viscosity, or
by increases in load, the coefficient will reach a minimum value. Further reduc-
tion in ZN/P leads to partial breakdown in the fluid film and lubrication is in the
mixed-film regime. Friction increases as ZN/P decreases in this regime, as more
and more of the load is carried by asperity contact and less by fluid-film pressure.
Finally, a point is reached, as ZN/P gets smaller, where there is no fluid pressure,
the rate of increase in the coefficient of friction starts to level out, and lubrication
is in the thin-film boundary regime (43). Ideally, a bearing is designed to operate
where the coefficient of friction is at its minimum.
Hydrodynamic Lubrication Regime. Rohde (44) and Dowson (45) remind
us that the basic mechanism of fluid-film lubrication was explained by Reynolds
(46) in 1886, based on the earlier work of Petrov (47), and Tower (48).
The fundamental requirements of lubrication in the hydrodynamic regime
are the formation of a wedge-shaped film; and generation of pressure in the film,
sufficient to keep the surfaces apart, by the motion of the surfaces themselves.
Vol. 2 LUBRICATION AND LUBRICANTS 23

Fig. 3. Pressure distribution in a hydrodynamic film. Reprinted with permission from


Ref. 29.

The pressure distribution in the lubricating film of the sliding bearing illu-
strated in Fig. 3 (29), moving with velocity U relative to a slanted stationary pad
in a fluid with viscosity m, and assuming no flow out of the sides of the bearing,
is (49)

p ¼ 6 m Uxðh  h2 Þ=h2 ðh1 þ h2 Þ ð16Þ

The total force P that the bearing will support per unit width is

p ¼ ½6 m UB2 =ðh1  h2 Þ2 ½lnððh1 =h2 Þ  2½ðh1  h2 Þ=ðh1 þ h2 Þ ð17Þ

and the frictional force F required to move the slider at speed U is

F ¼ ½2 m UB=ðh1  h2 Þ½2 lnðh1 =h2 Þ  3ðh1  h2 Þ=ðh1 þ h2 Þ ð18Þ

The analysis of hydrodynamic fluid films assumes laminar flow in the film, ie, a
Reynolds number, Re, <1000.

Re ¼ Uph=m ¼ Uh= ð19Þ

where U ¼ velocity, r ¼ density, h ¼ average film thickness, m ¼ dynamic visco-


sity, n ¼ kinematic viscosity.
Transition from laminar flow to turbulent flow starts when the Re is 1000
and the flow is completely turbulent at a Re of 1600 (50).
In a journal bearing, the wedge is formed because the diameter of the jour-
nal is smaller than that of the bearing. As the journal starts to rotate, its center-
line moves away from that of the bearing. The rotating journal drags, or pumps
the lubricant through the wedge, against the resistance to flow, and increases the
pressure in the fluid until the journal is lifted off the bearing surface. The load
24 LUBRICATION AND LUBRICANTS Vol. 2

carrying capacity W of a journal bearing is (51)

W ¼ ðm UR2 L=C2 Þ12 p e = ð2 þ e2 Þð1  e2 Þ1=2 ð20Þ

where m ¼ lubricant viscosity, U ¼ journal peripheral speed, R ¼ journal radius,


C ¼ clearance ¼ bearing radius  journal radius, L ¼ axial length of the bearing,
e ¼ eccentricity ¼ center offset distance e/clearance C.
Bearings designed for hydrodynamic lubrication include journal and thrust
bearings in steam and gas turbines, and main and rod bearings for automotive
engine crankshafts.
Elastohydrodynamic (EHD) Lubrication Regime. The shapes of sliding
surface bearings designed for hydrodynamic lubrication have a high degree of
conformity, a relatively large contact area, and low unit loading, such that the
effect of pressure on viscosity can be neglected. In contrast, the surfaces in a roll-
ing element bearing and on gear teeth are nonconforming. Surface contact is con-
centrated at a point in ball bearings or along a line in roller bearings and gears.
Contact pressure, therefore, is high enough to cause elastic deformation of the
contacting surfaces, forming a small area of contact. The viscosity of liquid lubri-
cants entering the contact area increases exponentially and may solidify. Since
the viscosity of the lubricant is affected by the elastic deformation of the surfaces,
as well as other fluid properties, this lubrication regime is called elastohydrody-
namic or EHD.
The hydrodynamic pressure developed in the lubricant is sufficient to sepa-
rate the surfaces at the leading edge of the contact area. As the lubricant is
drawn into the contact, its pressure and viscosity increase further, keeping the
surfaces apart. During contact, the fluid acts like an elastic solid, so that it
cannot escape the contact except in the direction of rolling (52).
As the lubricant comes out of the contact area, there is a sharp pressure spike,
followed by a sudden pressure drop so extreme that it causes a bulge in the rolling
surfaces. The minimum film thickness is at the location of this bulge. The pressure
spike and sudden film restriction directly affect the rolling element fatigue life (53).
The Dowson - Higginson equation for minimum film thickness (54,55) is

hmin =R ¼ 1:6ða E0 Þ0:6 ðm U=E0 RÞ0:7 ðP=E0 Þ0:13 ð21Þ

where R is the effective radius of the contacting surfaces, a is the viscosity–


pressure coefficient, E0 is the effective Young’s modulus of the contacting
surfaces, m is the dynamic viscosity at the inlet temperature and atmospheric pres-
sure, U is the effective velocity, and P is the Hertz pressure at the line contact.
Similar equations for film thickness in point contact are given by Dowson
(2) and by Khonsari and Hua (53).
The life of rolling element bearings is related to a film thickness
parameter, l.

l ¼ h=s ð22Þ

where h is the calculated EHD film thickness, and s is the composite surface
roughness.
Vol. 2 LUBRICATION AND LUBRICANTS 25

In the absence of chemically active additives in the lubricating oil, damage


to the bearing surface occurs, and reduces the life of the bearing, if the film thick-
ness becomes less than the height of the surface asperities (l  1). The life of the
bearing increases significantly at values of l > 1.5 (56).
Hydrostatic Lubrication. In hydrostatic lubrication, fluid is pumped
under pressure to the load-carrying bearing. Almost any fluid may be used,
including gases (nitrogen, helium, air), water, and liquid metals.
The principle applications of hydrostatic lubrication are gas bearings, mov-
ing large masses on relatively small bearing areas and for startup of heavily
loaded hydrodynamic bearings.
Squeeze Films. Viscous lubricant films do not immediately collapse
when sliding stops. During the time it takes for these films to be squeezed out
of the contact area, they can support peak loads higher than those supported
in steady-state operation. This time delay also provides damping for shock
loads and shaft vibration. These squeeze films are important in rod bearings of
reciprocating automotive engines, damping in turbomachinery and in the lubri-
cation of skeletal joints (hips etc). Ludema (55) estimates the time delay for a vis-
cous fluid squeezed out of an elliptical contact area as

1=h2 ¼ 1=h20 þ f½2Wða2 þ b2 Þt=3pa3 b3 mg ð23Þ

where a and b are dimensions of the ellipse, h0 is the original film thickness
(for small values of h0 relative to a or b), m is the dynamic viscosity of the fluid,
and W is the load that produces a film of thickness h after time t.
Boundary Film Lubrication Regimes. At lower values of the Sommerfeld
number (eq. 15), ie, when unit loads, P, are very high or when sliding speed, N, or
viscosity, Z, are low, a full-fluid film cannot be generated. The film thickness is
smaller than the height of asperities on the surfaces and the surfaces come in
contact with each other. When there is no fluid pressure, and only a thin film
of lubricant on the surfaces, lubrication is said to be in the boundary regime.
Except in the case of starved lubrication, ie, inadequate or blocked lubricant sup-
ply, lubricating oils and greases usually provide lubrication in the mixed or par-
tial film regime, where the load is supported by a combination of surface asperity
contact and fluid film pressure, at lower Summerfeld numbers.
Lubricants designed to operate in the mixed- or partial-film regime may
contain additives that form lubricating films on the bearing surface by adsorp-
tion, chemisorption, or tribochemical reaction. The additives may be friction
modifiers, antiwear additives, EP additives, or multifunctional additives. These
films are discussed later under lubricant films.
2.7. Micro- or Nanotribology. Micro- or nanotribology began with the
ability to study tribological phenomena at the molecular or atomic level. A key
tool is the atomic force microscope. With this tool a probe, the tip of which has
a radius <10 nm (<109 m), can be moved over a surface to measure frictional
resistance and other phenomena. At this level, the apparent surface interactions
do not obey the same ‘‘laws’’ as those observed on a macroscopic scale (57,58).
Topics in micro- or nanotribology are discussed in Refs. 59 and 60.
A principal application for micro- or nanotribology is computer hard disk
drives. The rms roughness, Rq, of the disk surface is <2 nm. The surface is
26 LUBRICATION AND LUBRICANTS Vol. 2

covered with a 1–2-nm film of a perfluoropolyether-based lubricant, and, as a


desired goal, a hydrodynamic air film thickness (flying height) of 1–3 nm.
Flying height is the distance between the disk and the read-write head–slider
assembly (7).
2.8. Lubricant Films. Layers of lubricating material, deposited on the
rubbing surfaces, control friction, and wear in the boundary portion of the
mixed-film regime. This material is initially dissolved or dispersed in the oil,
and is deposited on the surface by adsorption, chemisorption, or tribochemical
reaction. As portions of these films are rubbed off, they are replaced by additional
material in the oil.
A film of paraffin oil adsorbed on rubbing steel surfaces can reduce the coef-
ficient of friction from 0.78 to 0.23 (17). Such a film will not, however, support
much load or high temperature.
A better strategy for friction modification is to add a surface-active polar
molecule, with a long, nonpolar tail to the paraffin oil. These molecules, known
as ‘‘friction modifiers’’ or ‘‘boundary lubricity additives’’ can be naturally occur-
ring substances derived from animal fats, like lard oil, neatsfoot oil, tallow
and, in time past, sperm whale oil; vegetable oils, like olive oil, palm oil, rapeseed
oil and soybean oil, or they may be synthesized molecules designed to have a
particular structure and chain length.
Typical structures are fatty alcohols, esters, amines, and acids with chain
lengths of 16–22 carbon atoms. These additives are adsorbed or chemisorbed
out of the nonpolar base oil and condense on the surface. The polar end sticks
to the surface, and the nonpolar tails pack in as closely as possible and, because
of the lateral cohesive forces (London forces) between them, form a strong solid
film. This film resists the penetration of asperities and inhibits metal-to-metal
contact (61).
Fatty acids are chemisorbed on a metal surface, forming a metal soap with-
out removing the metal atoms from the lattice structure of the surface. With
stearic acid, eg, the length of each soap molecule is 19 Å, and there can be little
over four such molecules on each 100 Å2 of surface (61).
Adsorbed films and soaps are temperature sensitive. As surface tempera-
ture increases, it reaches the point where the film desorbs, gets disoriented, or
melts. For this reason, some lubricants contain EP or antiwear additives,
which are activated by the heat of friction at points of metal-to-metal contact
during sliding. They react with the surface to form protective layers of solid com-
pounds that prevent welding and catastrophic destruction of the metal surface.
Both antiwear and EP additives usually contain one or more of the elements
chlorine, phosphorus, or sulfur.
These tribochemical reactions involve the controlled corrosion of the metal
surface to prevent more serious damage under extreme loads and temperatures.
This regime is often called the Extreme Pressure or EP lubrication regime.
Examples of the type of chemicals used include (1) Zinc dialkyl or diaryl thiopho-
sphate (ZDDP), an effective antiwear additive used in engine oils and hydraulic
fluids. It has a relatively high activation temperature ( 1508C) and forms a
tough, wear-resistant surface layer. It is not an effective friction modifier.
(2) Mild EP additives with intermediate activation temperatures, such as the
sulfur–phosphorous compounds used in automotive hypoid and industrial gear
Vol. 2 LUBRICATION AND LUBRICANTS 27

systems. (3) Highly sulfurized or halogenated compounds used in some metal


cutting fluids that are activated at relatively low temperature, in some cases
at room temperature.

3. Lubricating Oil Base Stocks

The most important component of a lubricating oil or grease is its base oil.
Although often supplemented by additives, the base oil determines the flow char-
acteristics of the lubricant, its oxidation stability (sludge and deposit-forming
tendency), its volatility, and its corrosion potential.
Most lubricant base oils are mixtures of paraffins (straight- or branched-
chain hydrocarbons), naphthenes (cycloparaffins), and aromatics (alkyl benzenes
and multiring aromatics), typically containing 20–40 carbon atoms per molecule.
If paraffins predominate, the base stock is paraffinic. If naphthenes predominate,
it is a naphthenic base stock. Paraffins and naphthenes are saturates, meaning
that all of the carbon atoms in the hydrocarbon molecule are singly bonded to
another carbon atom or to a hydrogen atom. These lubricant base stocks are
manufactured by the distillation of selected crude oils, followed by further refin-
ing of the lube oil distillates by conventional separation or modern conversion
processes. When refined by conventional separation processes, the type of base
stock is crude specific, meaning that paraffinic base stocks come from paraffinic
crude oils, and naphthenic oils from naphthenic crudes. When refined by modern
conversion processes, base stocks are less crude specific, since these processes
are capable of converting naphthenic and aromatic compounds to paraffins
(see also HYDROCARBONS).
3.1. Base Stock Classification. The American Petroleum Institute
(API) defines a base stock as ‘‘. . . a lubricant component that is produced by a sin-
gle manufacturer to the same specifications (independent of feed source or man-
ufacturer location); that meets the same manufacturer’s specification; and that is
identified by a unique formula, product identification number, or both. Base
Stocks may be manufactured using a variety of different processes including
but not limited to distillation, solvent refining, hydrogen processing, oligomeriza-
tion, esterification, and rerefining. Rerefined stock shall be substantially free
from materials introduced through manufacturing, contamination, or previous
use’’ (62).
‘‘A base stock slate is a product line of base stocks that have different
viscosities but are in the same base stock grouping and from the same manufac-
turer’’ (62).
‘‘A base oil is the base stock or blend of base stocks used in an API-licensed
oil’’ (62).
API has also established five base stock categories, classified according
to saturates content, sulfur content, and VI. The classification is shown in
Table 2 (62).
Base oils with higher saturates content are generally more resistant to,
and easier to protect against, oxidation. They also have higher VIs. Paraffinic
oils, at the same saturates level, have higher VIs than naphthenic oils. Sulfur
compounds produce corrosive material when oxidized. Base stocks with a wide
28 LUBRICATION AND LUBRICANTS Vol. 2

Table 2. API Base Stock Categories


Saturates content, mass % Sulfur content, VI by ASTM
API group by ASTM D 2007 mass %a D 2270
I < 90 and/or > 0.03 80  VI < 120
II  90 and  0.03 80  VI < 120
III  90 and  0.03 VI 120
IV polyalphaolefins (PAO)
V All other base stocks not included in group I, II, III, or IVb
a
Measure sulfur content by the most recent version of one of the following test methods:
ASTM D 1522, ASTM D 2622, ASTM D 3120, ASTM D 4294, ASTM D 4927.
b
Group V includes naphthenic hydrocarbon and re-refined base stocks.

molecular weight range tend to be more volatile than those with a narrow
range.
Base Stocks in API groups I–IV are paraffinic hydrocarbons and are manu-
factured in several viscosity grades. Naphthenic base stocks are in group V
(Refs. 63–70). Typically, the saturates content of group II and III base stocks
is >99%, and the sulfur content is <15 ppm (<0.0015%). Group IV stocks are
100% saturates and contain no sulfur. The viscosity indexes of group II stocks
are typically 100–115, and those of groups III and IV are 120–140. Group IV
base stocks have the best low temperature flow characteristics because they
contain no wax. Naphthenic stocks, with minimal wax contents, also have good
low temperature flow properties. They also have the lowest viscosity indexes.
Oxidation stability of the paraffinic stocks improves with each group number,
as do volatility and deposit, sludge, and soot control.
3.2. Synthetic Base Stocks. The word ‘‘synthetic’’ is not part of the
API classification. It is a marketing term, not a technical term. The application
of the word synthetic is the subject of controversy because of the difficulty of
defining it. It was originally used to differentiate between base stocks made by
conventional crude oil refining processes and those synthesized from other che-
micals. Polyalphaolefins in group IV are still called synthetic, as are the diesters,
polyol esters, polyglycols, etc, in group V. This definition has been blurred, how-
ever, by the availability of base stocks derived from petroleum feed stocks by
modern conversion processes (71). ‘‘Synthetic’’ is an accepted description, eg,
for group III oils.
3.3. Base Stock Manufacturing Processes. Figure 4 lists some of the
refining and conversion processes used to manufacture groups I, II, and III base
stocks (72,73). Crude oil is first fractionated in an atmospheric distillation tower
to produce light gases and fuel products. The residue, or bottoms, from the
atmospheric tower are then fractionated in a vacuum distillation tower to pro-
duce gas oil and lube oil fractions. The vacuum residue is separated with propane
to produce asphalt and deasphalted cylinder oil.
Solvent Extraction. In the solvent extraction process, lube oil feedstock
from the vacuum tower flows upward through a treating tower, countercurrent
to a stream of solvent. The solvent preferentially removes undesirable tars,
resins, asphaltic compounds, and polycyclic naphthenes and aromatics. Solvents
used include phenol, furfural,and N-methyl-pyrrolidone (NMP). Solvent is then
Vol. 2 LUBRICATION AND LUBRICANTS 29

Atmos-
pheric
Crude oil Fuel products Vacuum gas oil
pipe-
still
Vacume
Atmospheric Lube oil fractions
pipe-
residuum
still

Vacuum residume

Lube oil Solvent Hydro- Dewax Group I base stocks


fractions extraction finishing

Raffinate
Lube oil Solvent
hydro- Dewax Group II base stocks
fractions extraction
conversion

Raffinate Wax
Lube oil Solvent hydro- Groups II and III
hydro-
fractions extraction isomerization base stocks
conversion

Lube Wax
Lube oil hydro- hydro- Groups II and III
fractions cracking isomerization base stocks

Fig. 4. Base stock refining options. Used with permission.

stripped from both the aromatic extract and raffinate streams. Solvent extraction
increases the VI and the stability of the raffinate (71,73).
Hydrofinishing. Hydrofinishing is catalytic hydrogenation process that
converts unstable compounds remaining after solvent extraction to stable ones.
It converts some aromatics to naphthenes and removes some sulfur compounds
and other trace materials. It is a relatively mild process compared to the modern
hydrocracking process that causes major molecular changes (74). It is severe
enough, however, to produce naphthenic base oils from unextracted distillates
that do not require labeling as carcinogenic under OSHA’s Hazardous Sub-
stances Communication Standard. This makes them suitable for use as base
oils for metalworking fluids, greases, and other industrial lubricating products.
Dewaxing. Paraffinic distillates contain some high molecular weight,
high melting point paraffin waxes. At temperatures below their melting point,
these waxes crystallize and cause the oil to gel. Several methods are used to
reduce the wax content of lubricating oil base stocks. In one method, the hydro-
fined raffinate is mixed at low temperature with a chilled solvent, in which the
heaviest waxes are insoluble. The wax crystallizes and is then separated from
the mixture by cold filtration. Solvents used include propane, methyl ethyl
ketone (MEK), methyl isobutyl ketone, and a mixture of MEK and toluene.
Another method for removing wax from the raffinate is catalytic dewaxing.
This method uses a shape-selective dewaxing catalyst to crack straight-chain or
slightly branched waxes into naphtha and gas (73).
Isomerization technology is used in the manufacture of group II and III
base stocks. In this process, high melting point straight-chain paraffins are
30 LUBRICATION AND LUBRICANTS Vol. 2

isomerized in a shape-selective catalyst to branched-chain, lower pour-point lube


molecules. The process also produces gasoline and diesel fuels (73,75).
Hydrogen Processing. Hydrogen reforming, hydrocracking, and wax iso-
merization are modern refining processes that convert undesirable components of
lube oil fractions into desirable components, rather than remove them. They are
used instead of solvent extraction on vacuum tower distillates, or on raffinates
from the extraction tower, to produce group II and III lubricating oil base stocks
with higher saturates content, lower sulfur content and higher viscosity index.
The principal reactions taking place in these processes are (71) saturation, ring
opening, reforming (isomerization), cracking, desulfurization, denitrogenation.
Naphthenic and group I paraffinic base stocks are refined by conventional
separation processes and are crude specific. Groups II and III paraffinic stocks
produced by modern conversion processes, hydrogen reforming, hydrocracking,
catalytic dewaxing, and wax isomerization, are less crude specific (See also
PETROLEUM REFINING PROCESSES).
3.4. Polyalphaolefins. Group IV base stocks are poly(a-olefins) (PAO).
They are produced by steam cracking hydrocarbons to produce ethylene; ethy-
lene oligomerization to produce linear a-olefins (1-decene, 1-dodecene, or 1-tetra-
decene); oligomerization of linear a-olefin to form a mixture of dimers, trimers,
tetramers, and higher oligomers; hydrogenation of the unsaturated oligomer.
The characteristics of the finished PAO are affected by the chain length
of the olefin raw material, temperature, time and pressure, catalyst types,
and concentration and distillation of the final product to remove the dimers
(76,77).
3.5. Polyinternal Olefins and Gas–Liquid Base Stocks. The Tech-
nical Association of the European Lubricants Industry, ATIEL, has established
a group VI category for polyinternalolefins (PIO). These are mixtures of linear
and cyclic olefin isomers (78,79).
The conversion of natural gas to liquids (GTL) by the Fischer Tropsch pro-
cess is a promising method for producing high quality lubricating oil base stocks.
Table 3 shows the typical properties of a GTL base stock compared to those of
current base stocks (80).

Table 3. Typical Properties of GTL Base Stocka,b


GTL-5, typical Industry range,
Base stock properties ASTM properties min-max Value
viscosity at 1008C, cSt D445 4.5 4.0–5.0 —
viscosity index D2270 144 120–141 high
pour point, 8C D97 21 24 to 12 low
cold-cranking simulator D5293 816 729–2239 low
at 258C, cP
NOACK wt% D5800 7.8 10.4–14.8 low
composition, mass%
iso-paraffins 100 47.3–80.9 high
monocycloparaffins 0 18.7–28.8 low
polycycloparaffins 0 5.3–22.2 low
aromatics 0 00–1.2 low
a
See Ref. 79.
b
Reprinted with permission.
Vol. 2 LUBRICATION AND LUBRICANTS 31

3.6. Organic Esters. Organic esters synthesized by reacting dibasic


acids with monoalcohols (diesters), or by reacting monoacids with polyhydric
alcohols (polyol esters), have been used as lubricating oil base stocks
for > 50 years. They are branched hydrocarbon molecules with multiple ester
linkages, which give them polarity. Their unique structure gives them excel-
lent thermal and oxidation stability, good low temperature flow char-
acteristics, low volatility, lubricity, detergency and dispersancy, and
biodegradability. Esters have been used exclusively in aircraft turbine engine
oils (jet engine oils) for >40 years. They are also the preferred base stock in
refrigerator compressor lube oils used with R-134 refrigerants. Other applica-
tions for ester base stocks include rotary screw air and process gas compres-
sors, oven chain lubricants and gas engines. Typical structures of ester base
stocks can be found in Refs. 79,81,82.
3.7. Polyglycols. Polyalkylene glycols (PAG) and polyethers are usually
copolymers of ethylene oxide and propylene oxide. The oxide monomer sequence
can be random or blocked and their solubility can be varied from oil soluble
to completely water soluble. Their applications include gear and compressor
lubricants, metalworking fluids, fiber lubricants, and fire-resistant hydraulic
fluids (79,83).
3.8. Vegetable Oil Esters. Triglyceride esters are obtained from
renewable sources: olive, soybean, rapeseed, canola, safflower, sunflower, mea-
dowfoam, castor, and other vegetable oils. They are biodegradable and offer spe-
cific environmental benefits over hydrocarbon-based lubricants. Lubricating oils
made with these base stocks and their derivatives are recommended in
applications where lubricant leaks and spent lubricant can contaminate the
environment. Examples include chain bar lubricants, two-cycle oils for outboard
marine engines, hydraulic fluids for farm machinery, and rail curve greases.
Most vegetable oil esters have a combination of saturated and unsatu-
rated fatty acids attached to the three alcohol groups in glycerine. Highly satu-
rated oils have good oxidation stability and poor low temperature flow
properties. As the amount of saturation decreases, oxidation stability
decreases and the low temperature flow properties improve. Advances in
breeding technology can change fatty acid profiles and alter the physical prop-
erties of vegetable oils (84).
3.9. Biodegradable Base Stocks. Lubricant or base stock biodegrad-
ability is the extent to which the material can be broken down by living things
(microorganisms) into innocuous products such as water, carbon dioxide, in-
organic compounds and methane. The least biodegradable of lubricant base
stocks are silicone oils, polyphenyl ethers, perfluoro alkyl ethers and alkylated
aromatic oils. Naphthenic stocks and base stocks in API groups I–IV also have
relatively poor biodegradability. The most biodegradable base stocks are vegeta-
ble oil esters, followed by polyalkylene glycols, organic esters, and phosphate
esters (85).
3.10. Other Base Stocks. Other chemical compounds used as lubricant
base stocks include polybutenes (84), hydrocarbons obtained by the alkylation of
naphthalene with a-olefins (70,87), alkylated aromatic hydrocarbons (88), sili-
cones (88,89), phosphate esters (90), chlorotrifluoroethylene (91,92), polypheny-
lethers (92), perfluoroalkyl polyether (92,93).
32 LUBRICATION AND LUBRICANTS Vol. 2

3.11. Rerefined Base Stock. In its definition of a base stock, the API
states ‘‘Rerefined stock shall be substantially free from materials introduced
through manufacturing, contamination, or previous use’’ (62).
Products made from rerefined stock are subject to the same stringent refin-
ing, compounding and performance standards applied to virgin oil products.
Rerefined oil may, in fact, have superior oxidation stability than virgin stocks
because the easily oxidized compounds will have been reacted during previous
use and then removed during reprocessing (71,94).
In the rerefining process, used oil is preferably segregated by type, col-
lected, and delivered to the reprocessing facility. The oils are then screened
and inappropriate feed stocks rejected. Solid and other gross contaminants are
then separated by, eg, propane precipitation, alcohol–ketone precipitation,
acid–clay filtration, etc. The filtrate is dehydrated and then refractionated by
vacuum distillation. The distillates are then finished by hydrotreating or clay
filtration or both, and then vacuum distilled to obtain the desired viscosity
grades (71,94).

4. Additives for Lubricating Oil and Grease

Advanced refining and manufacturing technologies have significantly


improved the quality of base oils used in the manufacturing of lubricating
oils and greases. They are more stable, have better low temperature flow prop-
erties, are less volatile, and less corrosive. Alone, however, they do not meet
the requirements of modern lubricating oils and greases. Additives are chemi-
cal substances added to the base oil to impart or improve certain properties.
They include oxidation inhibitors, rust and corrosion inhibitors, pour point
depressants, viscosity (VI) improvers, detergents, dispersants, friction modi-
fiers, EP agents, thickeners, emulsifiers, demulsifiers, bactericides, fungicides,
and tackiness additives. Most oils and greases contain several different addi-
tives, many of them surface active. They can assist each other, resulting in a
synergistic effect, or they can have antagonistic effects. Many additives have
several functions (multipurpose additives). These additives and base stocks
are the elements used by the lubricant designer to meet the increasingly criti-
cal requirements of equipment manufacturers and of the users of lubricating
oils and greases (95,96).
4.1. Oxidation Inhibitors (Antioxidants). Hydrocarbons and many
other components of lubricating oils and greases are subject to oxidation under
the operating conditions in which they are used. The products of oxidation may
be corrosive, they may be varnish, soot or sludge that cause wear, interfere with
proper operation of equipment or plug oil passages, preventing the lubricant
from getting to its intended application.
Preventing or delaying the oxidation of lubricating oils and greases extends
their operating lives. Oil change intervals for automotive engines have increased
10-fold over the past 50 years, even though engine operating temperatures have
increased significantly over the same period. Many machine elements are ‘‘lubri-
cated for life’’. Steam turbine oils remain in use for decades without any oxida-
tion effects. These are major economic benefits for machine operators.
Vol. 2 LUBRICATION AND LUBRICANTS 33

The oxidation process is a chain reaction with several different chemical


actions. It is generally initiated by heat or light leading to the formation of
free radicals, followed by, in the presence of air, the formation of peroxides,
and then a variety of reactions including polymerization. It is catalyzed, or
speeded up, by the activity of metal surfaces. Metals employed in soap thickeners
have been found to catalyze the oxidation of greases (97). Detailed mechanisms
are described in Refs. 98 and 99.
Natural inhibitors (sulfur or nitrogen containing) present in some base
stocks, or their oxygen-containing intermediates can impair the formation of
free radicals. Molecules with carbon–carbon double bonds and aromatic residues
can promote formation of radicals. Refining processes remove easily oxidized
components from the base stock and make it more responsive to oxidation inhi-
bitors. They also remove some natural inhibitors. Synthesized base stocks with
few impurities (and no natural inhibitors) are less prone to oxidation and more
responsive to antioxidants.
A variety of additives are available that terminate the chain reaction in the
early stages of oxidation. There are (97) (1) free-radical scavengers, primary anti-
oxidants that function by breaking the propagation step; (2) hydroperoxide
decomposers, secondary antioxidants that function by inhibiting the branching
step; (3) metal deactivators and corrosion inhibitors that prevent catalysis of
the initiation reaction.
High quality lubricants may contain more than one of each type of these
oxidation inhibitors. The combination of different antioxidants is synergistic.
Each functions via a different mechanism, in a different temperature range,
and at different rates. Each stabilizes different species, and one may act to regen-
erate another (98–100).
There are two types of primary antioxidants or free radical scavengers: Hin-
dered phenols and aromatic amines. Both function by donating a hydrogen atom
to the peroxide radical, ROO., to break the propagation step in the oxidation
mechanism and form relatively inert products (97,98).
In the presence of active antioxidants, oxidation proceeds very slowly until
the inhibitors have no more available hydrogen atoms to contribute. At this
point, the rate of oxidation increases dramatically and other measures are neces-
sary to deal with the oxidation products. These include dispersants and deter-
gent additives (discussed later), filtration, reconstitution of the additive
package and replacing the oil. Oil quality monitoring is often used to follow
the progress of deterioration and determine when an oil change is needed
(100,101).
Figure 5 shows the way hindered phenols react with peroxide radicals to
terminate the oxidation chain reaction (98,99). Chemicals used in this process
include (97,98,102,103) 2,6-di-tert-butyl phenol (DTBP), 2,6-di-tert-butyl-p-cresol
(DBPC or BHT), 3,5-di-tert-butyl-4-hydroxyanisole (BHA), 4,40 -methylenebis
(2,6-di-tert-butyl phenol), 3,5-di-tert-butyl-4-hydroxy-hydrocinnamic acid, C7–
C9 branched alkyl ester, 1,6-hexamethylene bis(3,5-di-tert-butyl-4-hydroxy-
hydrocinnamate).
Examples of aromatic amine antioxidants include alkylated diphenyla-
mines (ADPA), phenyl alpha diphenyl amine (PANA), alkyl substituted PANA,
and trimethyl quinoline derivatives (94,98,99).
34 LUBRICATION AND LUBRICANTS Vol. 2

OH O

ROO ROOH

R R

O O

ROO

R ROO R

Fig. 5. Radical scavenging with hindered phenols. Reprinted with permission.

A variety of mechanisms is reported for the action of amine-type antioxi-


dants. They act at both the initiation and propagation steps and are effective
at lower concentrations and over a wider temperature range than are the hin-
dered phenols (97–99).
There is also evidence that some of the DBPC and PANA in the oil react
directly with oxygen and those portions are not contributing to the termination
of the chain reaction (100,104).
Secondary antioxidants include sulfur containing and organophosphorus
compounds. Typical examples include dialkyl thioesters, metal dithiocarba-
mates, metal dithio phosphates, and aryl phosphites. They inhibit the branching
step by decomposing hydroperoxides. Reyes-Gavilan and Odorisio (97), referring
to several earlier works, report that, in general: (1) Thioethers react with hydro-
peroxides to form sulfoxides, sulfones, and alcohols; (2) zinc dialkyl dithiocarba-
mates react with hydroperoxides to form sulfur acids that ionically catalyze
hydroperoxide decomposition; (3) the reaction of zinc dialkyldithiophosphate
(ZDDP) with hydroperoxides occurs through an O,O0 -dialkyl hydrogen phosphoro
dithioate catalyzed ionic mechanism; (4) hydroperoxide decomposition by phos-
phites is accomplished through a displacement reaction mechanism. The pro-
ducts of this reaction consist of an organic phosphate and an alcohol.
Secondary antioxidants are best used in combination with primary antiox-
idants. They are not as effective as the latter when used alone. There are some
multifunctional additives, in which sulfur is incorporated in the phenolic mole-
cule, that are both primary and secondary antioxidants (97,98).
4.2. Metal Deactivators. Metal deactivators passivate the surface, of
copper, brass, and bronze alloys, or they chelate ions of copper, iron, manganese,
cobalt, and other metals. Passivators, which lay a protective layer over the metal
surface, include benzotriazole, 2-mercaptobenzothiazole, and tolutriazole deriva-
tives. Chelators, which trap metal ions in the bulk of the substrate, include
N,N0 -disalicylidene-1,2-diaminopropane and ethylenediaminetetraacetic acid
(EDTA) (95,98). Both types prevent metal catalysis at the oxidation initiation step.
4.3. Corrosion Inhibitors. Corrosion inhibitors in lubricating oils and
greases form a protective layer on the surface of metals to prevent moisture
and oxygen from easily reaching the surface. Natishan and Moran discuss the
mechanism of corrosion of metals, mainly caused by electrolytic reactions (see
Vol. 2 LUBRICATION AND LUBRICANTS 35

CORROSION AND CORROSION CONTROL). Polyisobutenylsuccinic acid derivatives, amine


phosphates, fatty acid amides of sarcosine, imidazoline derivatives, and sulfo-
nates are a few of the chemicals used in lubricating oils as corrosion inhibitors
(102,103,105).
4.4. Detergents and Dispersants. Detergent and dispersant additives
are used principally in gasoline and heavy duty diesel engine oils, automatic
transmission fluids, and tractor hydraulic fluids. Their primary function is to
keep the engine surfaces clean: they prevent deposits on hot surfaces such as pis-
tons and rings; they neutralize corrosive oxidation products and other contami-
nants; they suspend insoluble oxidation products and debris in the oil, thus
preventing them from clogging vital oil passages. They do not, however, remove
deposits already on the surfaces.
Detergents are made by reacting metal oxides or hydroxides with long-
chain organic acids. The metals include calcium, magnesium, barium, and
sodium. Acids include alkylbenzene sulfonic acid, alkylnaphthalene sulfonic
acid, alkyl phenol, sulfur, or methylene bridged alkyl phenol, alkyl salicylic
acid, and polyisobutenyl phosphonic acid (106).
Detergents may be neutral, basic, or overbased. Basic and overbased deter-
gents have reserve alkalinity to neutralize acidic combustion products (usually
formed from sulfur in the fuel) that leak or ‘‘blow by’’ the rings in an internal
combustion engine. These are made by using excess metal base in the reaction
and then blowing the product with CO2 to form metal carbonate. The carbonate
is held in suspension in a micelle structure (106).
Some detergents, particularly the phenates and salicylates, are detergent
inhibitors, meaning that parts of the molecule are able to inhibit oil oxidation
and bearing corrosion (107).
Dispersants are metal free and therefore ashless. They are higher in mole-
cular weight than detergents and more effective in suspending oxidation
products and debris. Some examples of dispersant chemistry are polyisobutenyl-
succinimides, polyisobutenyl succinate esters, polyaminomethylpolyisobutylphe-
nols, and bis(hydroxypropyl polyisobutenylthiophosphonate) (106).
The polar ends of the dispersant molecules surround and attach themselves
to the insoluble contaminant particles. Their oil-soluble, nonpolar tails keep the
particles in suspension (106,107).
Dispersancy is also built into multifunctional viscosity improvers based on
olefin copolymers, polyacrylates, and styrene–maleic anhydride polymers (106).
4.5. ZDDP: Antiwear Additive, Antioxidant, and Corrosion Inhibitor.
The ZDDP (Zinc Dialkyl DithioPhosphate and Zinc Diaryl DithioPhosphate)
compounds have, for > 50 years, been the most effective wear reducing additives
in engine lubricating oils, hydraulic oils and other lubricants.
When two steel surfaces rub together, and the metal on one surface is in
contact with the metal on the other surface, the frictional heat caused by the
rubbing causes the ZDDP to react chemically with the surfaces to form very
tough wear-resisting boundary lubricating films. These films do not shear easily,
and have little effect on the coefficient of friction. The ZDDP films do rub off,
however, and they are constantly being rebuilt by the chemical reaction of the
metal with ZDDP. These films are particularly effective in concentrated sliding
contacts such as those found in engine valve trains and in vane type hydraulic
36 LUBRICATION AND LUBRICANTS Vol. 2

pumps. Concentrated contacts have a very small contact area which results in
extremely high pressure on the contact area. Some friction modifying additives
work with ZDDP to reduce both friction and wear of the rubbing surfaces.
The ZDDP compounds are multifunctional. They are secondary oxidation
inhibitors, decomposing hydroperoxides in the branching step of hydrocarbon
oxidation. They are also effective bearing corrosion inhibitors.
This compound is manufactured by first reacting an alkyl or aryl alcohol
with phosphorous pentasulfide (P2S5), and then neutralizing the resultant acid
with zinc oxide. The alkyl alcohol may be primary alkyl, branched chain primary
alkyl, secondary alkyl or, rarely, tertiary alkyl. As a rule, the secondary and ter-
tiary alkyl derivatives are more effective as antiwear additives, but are the least
stable. The aryl derivatives generally have the best thermal stability and the
least potent antiwear activity (108,109).
The ZDDP compounds, particularly those with lower thermal stability, can
decompose at temperatures below those of rubbing steel surfaces and become cor-
rosive to copper and copper alloys. This is generally not a problem in internal
combustion engines, but it can be a problem in machinery using, eg, bronze bush-
ings, pistons, or clutch plates. However, by the judicious selection of alcohols, and
careful control of the manufacturing process, ZDDP compounds are made that
perform well in machines such as automatic transmissions and piston type
hydraulic pumps that contain parts made with copper alloys.
Despite its effectiveness as a wear, oxidation and corrosion inhibitor, and
its relatively low cost, there are serious efforts to reduce or eliminate the use
of ZDDP additives in engine oils. When the oil gets in the combustion chamber,
the additive decomposes at high temperature and, because a metal is present,
leaves ash, a solid residue, and other deposits in the chamber. These deposits
interfere with engine efficiency and emissions control. There is also evidence
that ZDDP decomposition products poison the catalysts used to convert CO,
NOx, unburned hydrocarbons and volatile organic compounds to less harmful
emissions. Zinc oxide, phosphorus pentasulfide, and a zinc pyrophosphate
glaze have been found on catalyst surfaces that prevent passage of exhaust
gases into the porous structure of the catalyst (110). (see also EMISSION CONTROL,
AUTOMOTIVE). Lacking an acceptable emission systems protection test, the
International Lubricant Standardization and Approval Committee (ILSAC)
places maximum phosphorus and sulfur limits on the GF-4 engine oil specifica-
tion. These limits are expected to be reduced further in the next generation
specification.
4.6. EP and Antiwear Additives. There is very little distinction
between so-called EP and antiwear additives. Both react with rubbing metal
surfaces to form lubricating films, and they both inhibit wear. ZDDP additives
have relatively high activation energy and are not effective until the surface
temperature gets >1508C. The EP additives are used in relatively high load,
slow speed applications to prevent catastrophic failure of the surface. Mild EP
additives used in automotive and industrial gear oils have lower activation
temperatures than ZDDP. ‘‘Active’’ EP additives used in metal cutting oils, like
elemental sulfur, chlorinated hydrocarbons and some sulfurized esters, can react
at room temperature. Table 4 lists some of the many compounds used as EP
and antiwear additives in lubricating oils (103,111,112).
Vol. 2 LUBRICATION AND LUBRICANTS 37

Table 4. EP and Antiwear Additives


zinc dialkyl and diaryl dithiophosphates antiwear, antioxidant, corrosion inhibitor
zinc dialkyldithiocarbamate antiwear, antioxidant, corrosion inhibitor,
metal deactivator, color stabilizer
molybdenum dialkyldithiophosphate antiwear, antioxidant, friction modifier, EP
molybdenum di-n-butyldithiocarbamate antiwear, antioxidant, friction modifier, EP
antimony dialkyldithiophosphate antiwear, antioxidant, friction modifier, EP
antimony tris(dialkyldithiocarbamate) EP, antiscuff, antioxidant
ethyl acrylate dialkyldithiophosphate ashless antiwear, EP
methylene bis(dibutyldithiocarbamate) ashless EP, antioxidant
2,5-dimercapto-1,3,4-thiadiazol, ashless antiwear, antioxidant
alkyl polycarboxylate
amine phosphate and thiophosphates ashless antiwear, rust inhibition
tricresyl phosphate ashless antiwear, EP
trialkyl phosphates ashless antiwear, EP
alkyl monoacid and diacid phosphates ashless antiwear, EP
tertiarybutylphenyl phosphate ashless antiwear, EP
dialkyl, trialkyl and triaryl phosphites ashless antiwear, EP
sulfurized olefins EP
sulfurized oils and esters EP
esters of thio and dithio phosphoric acid EP
chlorinated paraffins EP

4.7. Friction Modifiers (Boundary Lubricity Additives). Friction is a


force that resists sliding, or slipping, of one surface across another. Static friction
prevents sliding or causes the surfaces to move together, with no relative motion
between the surfaces. When a shearing force high enough to overcome the static
force is applied, the surfaces move relative to each other and the friction is
kinetic. On unlubricated surfaces, and on surfaces lubricated with fluids that
do not contain friction modifying additives, static friction is higher than kinetic
friction. Friction modifiers added to the oil tend to reduce both the static and
kinetic friction forces and to make the static friction lower than the dynamic
friction.
Friction modifiers have two functions; to reduce energy consumption by
machinery by lowering the kinetic friction; or to eliminate shudder in automatic
transmission torque converter clutches, chatter in limited slip automotive differ-
ential gear systems, and stick-slip in machine tool slideways. Shudder, chatter,
and stick-slip are all the same phenomena, frictional vibration caused by oscilla-
tion between static and kinetic friction forces. The oscillation occurs at low rela-
tive speeds between surfaces in contact and when the static friction is higher
than the kinetic friction. Inherent resilience in the applied shearing force also
contributes to stick-slip. With a friction modified lubricant, kinetic friction is
higher than static friction and the frictional vibration is eliminated (113–115).
Like many lubricating oil additive molecules, organic friction modifiers
have a polar end that attaches to the rubbing surface and a long-chain hydrocar-
bon tail. Sulfurized fats, phosphoric and thiophosphoric acid derivatives form
boundary lubricating film layers by chemical reaction with the surface material.
The latter three also function as EP additives. Other friction modifiers, eg, fatty
acids, carboxylic acid derivatives, esters, ethers, amines, amides, and imides
38 LUBRICATION AND LUBRICANTS Vol. 2

form layers of boundary lubricating films by chemisorption. The polar heads on


the surface are attracted to each other by hydrogen bonding and dipole–dipole
interactions and the hydrocarbon tails are lined up perpendicular to the surface
by van der Waals (London) forces.
Another class is friction polymers formed at the asperities under the influ-
ence of contact temperature and load. These include partial complex esters,
methacrylates, unsaturated fatty acids, and sulfurized olefins.
Finally, there are metalloorganic types like molybdenum dialkyldithiopho-
sphate and molybdenum dialkyldithiocarbamate and solid lubricants like gra-
phite, molybdenum disulfide, and PTFE (116).
4.8. Pour Depressants. The pour point is the lowest temperature at
which an oil is observed to flow, when cooled as prescribed in ASTM D97. At
this point, wax crystals that have previously precipitated as solid needles and
platelets agglomerate and cause the oil to gel, preventing its flow. Dewaxing
and hydroisomerization processes remove most, but not all of the waxes (high
molecular weight n-paraffins) from groups I–III base stocks. The resultant
pour points of the base stocks vary from about 15 to about 258C. Adding
a pour depressant can further reduce the pour point another 30–408C. Pour
point depressants are branched-chain polymeric additives that coprecipitate
with the wax crystals and prevent them from agglomerating. The side chains
on these polymers, often called wax crystal modifiers, are about the same length
as the wax molecules (117).
The first pour depressants were alkylated naphthalenes. Phthalic acid dia-
lkyaryl esters followed, and then long-chain alkyl acrylates and methacrylates.
Polymethacrylates are probably the most widely used pour depressants today.
Many other polymeric additives can be tailored to provide pour depressant prop-
erties in addition to their other functions (117–119).
4.9. Viscosity Improvers. Viscosity improvers are polymeric additives
that are more soluble in oil at high temperature than at low temperature. The
dissolved molecules expand or stretch out and considerably increase the viscosity
of the oil at high temperatures. These molecules shrivel up and shrink, contribut-
ing very little to the viscosity of the oil at low temperature. Viscosity improvers
are added to low viscosity base oils, which have the necessary low temperature
flow characteristics, to increase their viscosity and resultant lubricating ability
at operating temperatures.
Multigrade oils, oils containing viscosity improvers, are non-Newtonian
oils. The high temperature viscosity of the oil undergoing high shear rates is
lower than its viscosity at very low shear rates. This viscosity loss may be tem-
porary and the oil will recover its original viscosity when the shear stress is
relieved. Permanent viscosity loss occurs when the polymer itself is sheared.
Automotive engine oil specifications control both temporary and permanent visc-
osity loss caused by shear stress.
Viscosity improving polymers include olefin copolymers from ethylene, pro-
pylene, and butylene monomers; polymethacrylates; styrene–butadiene copoly-
mers; and hydrogenated polyisoprene.
Many of these polymers can be modified to make them multifunctional, giv-
ing them pour depressing and/or dispersant properties in addition to viscosity
improvement (117,120).
Vol. 2 LUBRICATION AND LUBRICANTS 39

4.10. Foam Inhibitors. Foam in lubricating oils consists of bubbles of


air on the liquid surface whose walls are thin liquid films. Strong foaming affects
the lubricating properties of oils and enhances their oxidation. Foam is generated
on mixing air with the oil during agitation and then separation of the air at
the oil surface. Foam stability is a function of oil viscosity and surface tension
at the oil–air interface. It is strongly affected by surface active substances
such as detergents, friction modifiers, EP additives, and corrosion inhibitors.
Finely dispersed polydimethylsiloxane, which is insoluble or has borderline
solubility in hydrocarbon oils, forms a monolayer on the liquid surface, lowering
the surface tension and breaking the foam. Antifoam additives based on this sili-
cone oil are widely used in automotive engine, transmission, and gear oils.
Silicone oils, however, tend to stabilize air emulsions, where the air bubbles
do not separate from the oil or separate very slowly. Entrained air in an oil cir-
culating system is compressible and leads to spongy response in hydraulic
systems and severe cavitation damage in bearings. Alkylacrylate and alkyl-
methacrylate copolymers, while not as effective as silicones for foam suppression,
are preferred where air entrainment is a problem (121–123).
4.11. Demulsifiers. Steam turbine oil, paper machine oil, industrial
gear oils, hydraulic oils, and oils for many other applications are subject to con-
tamination by water. It is important that this water separates easily from the oil
and drops to the bottom of the oil reservoir, where it can be drained off. If it does
not, it can mix with and circulate with the oil, seriously degrading the oils lubri-
cating capability. Demulsifiers are sometimes added to the oil formulation to
facilitate water separation. They concentrate at the oil–water interface and pro-
mote coalescence of the water droplets. Chemicals used include anion-active com-
pounds such as dinonylnaphthalene sulfonates and polyalkoxylated phenols,
polyols, and polyamines (124,125).
4.12. Emulsifiers. Water-in-oil emulsions are used as coolants–lubricants
in metalworking applications and as hydraulic fluids. Medium molecular
weight sodium sulfonates (derived from natural and synthetic raw materials) are
the most widely used emulsifiers in these fluids. Nonionic emulsifiers based on poly-
ethylene oxide are also used, as are long-chain alkylammonium salts (126).

5. Lubricating Oils

Lubricating oils are specifically formulated for virtually every type of machine
and manufacturing process. The base stocks and the type and concentration of
additives used for these oils are selected based on the requirements of the
machinery or process being lubricated, by the quality required by the builders
and the users of the machinery, and by government regulation. Table 5 is a
partial list of the principal applications for which lubricating oils are specifically
formulated:
Each of these oils has a unique set of performance requirements. In addition
to proper lubrication of the machine or process, these requirements include
maintenance of the quality of the lubricant itself, as well as the effect of the lubri-
cant’s use and disposal on energy use, the quality of the environment and on the
health of the user. The performance requirements of a few of these lubricating
oils are discussed below.
40 LUBRICATION AND LUBRICANTS Vol. 2

Table 5. Principal Applications for Lubrication Oils


Automotive lubricating oils Industrial lubricating oils
gasoline engine oils industrial gear oils
for passenger cars and light trucks pneumatic tool lubricating oil
for heavy duty automotive and industrial service high temperature oils
for piston engines in general aviation service air and gas compressor oils
small 2-stroke and 4-stroke gasoline engines for reciprocating compressors
for outboard motors for rotary vane compressors
for scooters, mopeds, and motorcycles for rotary screw compressors
for lawnmowers and small tractors for refrigeration compressors
for chain saws and similar portable equipment machine tool way oils
diesel engine oils textile oils
for heavy duty trucks, agricultural and steam turbine oils
construction vehicles
for industrial cross-head and trunk piston hydraulic fluids
diesel engines
for railroad diesel engines paper machine oils
for marine cross-head and trunk piston food machinery oils
diesel engines
gas engine oils steam cylinder oils
gas turbine oils Metalworking fluids
for aircraft jet engines in commercial aviation for metal cutting
service
for industrial gas turbine engines for metal rolling
automatic transmission fluids for metal drawing, forging,
stamping, etc
gear oils
for automotive manual transmissions
for automotive differentials

5.1. Gasoline Engine Oils for Passenger Cars and Light Trucks.
Internal combustion engines provide the best example of performance require-
ments dictating the formulation and quality of the engine oil. These engines
have lubrication requirements in nearly every regime: hydrodynamic, elastohy-
drodynamic, squeeze film, EP, boundary, and mixed film. They operate at high
oil and surface temperatures, must start up at very low temperatures, are sub-
ject to contamination (by fuels, combustion products, oxidation products, and
dirt), and they must meet stringent fuel economy and exhaust emission stan-
dards. Their lubricating oils contain a wide variety of additives and their base
oil requirements determine the type and quality of base stocks produced in
lube refineries and chemical manufacturing plants. The performance require-
ments of gasoline engine oils for use in automobiles are a good example of the
relationship between the performance requirements and the formulation of
lubricating oils for a specific application.
Gasoline engine oils for use in automobiles in the United States are licensed
and certified to assure that they meet the minimum performance standards
established by industry technical specifications. The Engine Oil Licensing and
Certification System (EOLCS), administered by the API, is a voluntary program
that authorizes oil marketers to use the API Engine Oil Quality Marks–the
API Service Symbol ‘‘Donut’’ and Certification Mark ‘‘Starburst’’ (127). This
Vol. 2 LUBRICATION AND LUBRICANTS 41

program is a cooperative effort between the oil industry and vehicle and engine
manufacturers–Ford, General Motors and Daimler Chrysler; the Japan Auto-
mobile Manufacturers Association (JAMA); and the Engine Manufacturers
Association (EMA).
Performance requirements, test methods, and limits are cooperatively estab-
lished by vehicle and engine manufacturers, technical societies, [eg, the Society of
Automotive Engineers (SAE) and the American Society of Testing and Materials
(ASTM)], and industry associations (such as the American Chemistry Council and
API). An ongoing monitoring and enforcement program to ensure licensees adhere
to industry technical specifications backs APIs Engine Oil Program (128).
The API Service Symbol ‘‘Donut’’ is divided into three parts: (1) The top of
the Donut shows the oil’s performance level for gasoline and/or diesel engines.
The letter ‘‘S’’ followed by another letter (eg, ‘‘SM’’) refers to oil suitable for gaso-
line engines. The letter ‘‘C’’ followed by another letter and/or number (eg, CI-4)
refers to oil suitable for diesel engines. (2) The center identifies the oil’s viscosity
characteristics. The numbers and the letter ‘‘W’’ indicate the SAE viscosity
grade. These grades are defined in SAE J300, May 2004 (129). The low tempera-
ture ‘‘W’’ grades indicate how quickly an engine will crank and how well the oil
will flow to lubricate critical engine parts at low temperatures. The grades with-
out the ‘‘W’’ indicate the oil’s operating temperature flow properties. A multi-
grade oil (eg, 5W-30) provides good flow capability for low temperatures and
adequate viscosity characteristics for high temperature lubrication. It meets
the viscosity requirements of both the ‘‘W’’ grade and the other grade indicated
(129). (3) The bottom of the Donut tells whether the oil has demonstrated energy
conserving properties in a standard test in comparison with reference oil. It may
also include a supplemental category description (eg, CI-4 PLUS).
The API Certification Mark ‘‘Starburst’’ (Fig. 6) is designed to identify
engine oils recommended for a specific application (such as gasoline engine
service). An oil may be licensed to display the Starburst only if the oil satisfies
the most current requirements of the International Lubricant Standardization

Fig. 6. API certification mark.


42 LUBRICATION AND LUBRICANTS Vol. 2

and Approval Committee (ILSAC) minimum performance standard for this appli-
cation. Members of ILSAC include representatives of the Japanese Automobile
Manufacturers Association, DaimlerChrysler Corporation, Ford Motor Company
and General Motors Corporation.
An engine oil meeting the latest API service category may or may not meet
all of the requirements of the latest ILSAC category. When it does meet both
requirements, its container may display both symbols.
The current ILSAC minimum performance standard, suitable for use in
the current and all earlier model year automobiles, is ILSAC GF-4 Standard
for Passenger Car Motor Oils (130). The corresponding API Service Category is
API Service SM. Both were issued in 2004 and superceded the 2001 standards,
ILSAC GF-3 (131) and API Service SL. Those, in turn, superceded ILSAC GF-2
and API Service SJ, issued in 1996 and API Service SH, issued in 1992 and
now obsolete. After April 2005, the Starburst can no longer be used for GF-3 oils.
The performance required by ILSAC for GF-4 Engine Lubricating Oils
is measured in a battery of full scale engine tests and laboratory bench tests
conducted by qualified independent laboratories. Tables 6–11 describe these
engine and bench test requirements for ILSAC GF- 4, GF-3, and GF-2 and for
API Service SM, SL, SJ, and SH (130,131).
Engine test requirements in the ILSAC GF-4 Standard include (1) The
ASTM Sequence IIIG test measures viscosity increase, piston deposits at high

Table 6. Engine Test Requirements, ASTM Sequence IIIG, IIIF, IIIE for Wear, Piston
Deposits, and High Temperature Oxidation
ILSAC ILSAC GF-4 ILSAC GF-3 ILSAC GF-2
CLASSIFICATION:
API SERVICE Proposed API SM API SL API SJ API SH
CATEGORY:
ISSUE DATE: 2004 2001 1996 1992
ASTM Engine SEQUENCE IIIG SEQUENCE SEQUENCE SEQUENCE
Test Sequence IIIF IIIE IIIE
ASTM Test Procedure: ASTM D 5533 ASTM D 5533
kinematic viscosity 150 max 275 max 375 max 375 max
increase at 408C, %
average weighted piston 3.5 min 4
deposits, merits
hot stuck rings none none none none
cam plus lifter wear,
average, mm 60 max 20 max 30 max 30 max
maximum per 64 64
position, mm
piston skirt varnish 9.0 min 8.9 min 8.9 min
oil consumption, L 5.2 max 5.1 max 5.1 max
avg. oil ring land deposits 3.5 min 3.5 min
avg. engine sludge rating 9.2 min 9.2 min
lifter sticking none none
cam þ lifter scuffing none none
low temperature stay in grade or rate and
pumping viscosity at next higher report
end of test by ASTM D grade
4684 (MRV TP-1)
Vol. 2 LUBRICATION AND LUBRICANTS 43

Table 7. Engine Test Requirements, ASTM Sequence VG and VE for Wear, Sludge
and Varnish
ILSAC CLASSIFICATION: ILSAC GF-4 ILSAC GF-3 ILSAC GF-2
API SERVICE CATEGORY: Proposed API SM API SL API SJ API SH
ISSUE DATE: 2004 2001 1996 1992
ASTM ENGINE TEST SEQUENCE VG SEQUENCE SEQUENCE SEQUENCE
SEQUENCE VG VE VE
ASTM Test Procedure: ASTM D 6593 ASTM D 6593 ASTM D 5302 ASTM D 5302
average engine sludge, 7.8 min 7.8 min 9.0 min 9.0 min
merits
average rocker cover 8.0 min 8.0 min 7.0 min 7.0 min
sludge, merits
average engine varnish, 8.9 min 8.9 min 5.0 min 5.0 min
merits
average piston skirt 7.5 min 7.5 min 6.5 min 6.5 min
varnish, merits
oil screen sludge 20 max 20 max 20 max 20 max
(clogging), % area
oil screen debris, rate and rate and
% area report report
hot stuck compression none none none none
rings
cold stuck rings rate and rate and
report report
oil ring clogging, rate and rate and rate and 15
% area report report report
cam lobe wear,
average, mm 127 max 130 max
maximum per 380 max 380 max
position, mm
cam follower pin rate and rate and
wear, cyl #8, report report
avg, mm
ring gap increase, rate and
cyl #1 & #8, avg, mm report
cylinder bore wear rate and
report
ring wear rate and
report

operating temperature and valve train wear. At the end of the test, the low tem-
perature pumping viscosity of the oil is measured. It is run in a 1996/1997 3800 cc
series II General Motors V-6 fuel-injected gasoline engine. After a 25-min
break-in, it operates at 125 bhp, 3600 rpm, 1508C oil temperature for 100
hours (130,132). (2) The ASTM Sequence VG test (ASTM D 6593-04) evaluates
the oil’s ability to prevent sludge and varnish formation. It is run in a 1994
Ford V-8, 4.6 liter fuel injected gasoline engine. The procedure involves 54 cycles
in 216 h, each cycle consisting of three different sets of operating conditions
(130,132,133). (3) The ASTM Sequence IVA test (ASTM D 6891-03) measures
the average camshaft lobe wear in a 100 hour test in a KA24E Nissan 2.4
liter, 4-cylinder engine (130,132,134). (4) The ASTM Sequence VIII test (ASTM
44 LUBRICATION AND LUBRICANTS Vol. 2
Table 8. Engine Test Requirements, ASTM Sequence IVA, VIII, IID and CRC-L-38
For Valve Train Wear, Bearing Corrosion, Shear Stability and Engine Rusting
ILSAC CLASSIFICATION: ILSAC GF-4 ILSAC GF-3 ILSAC GF-2
API SERVICE Proposed API API SL
CATEGORY: SM
ISSUE DATE: 2004 2001
ASTM ENGINE TEST SEQUENCE SEQUENCE API SJ API SH
SEQUENCE IVA IVA
ASTM Test Procedure: ASTM D 6891 ASTM D 6891 1996 1992
average cam wear 90 max 120 max
(7 position average) mm
ENGINE TEST SEQUENCE SEQUENCE CRC L-38 CRC L-38
SEQUENCE VIII VIII
ASTM Test Procedure: ASTM D 6709 ASTM D 6709 ASTM D 5119 ASTM D 5119
bearing weight loss, mg 26 max 26.4 max 40 max 40 max
shear stability (10 hour stay in grade stay in grade stay in stay in grade
stripped viscosity) grade
piston skirt varnish 9 min
ASTM ENGINE TEST (See Ball rust SEQUENCE SEQUENCE
SEQUENCE: bench test) IID IID
ASTM Procedure: ASTM D ASTM D
5844 5844
avg. engine rust rating 8.5 min 8.5 min
lifter sticking none none

Table 9. Engine Test Requirements, ASTM Sequence VIB and VIA for Fuel Economy
ILSAC CLASSIFICATION: ILSAC GF-4 ILSAC GF-3 ILSAC GF-2
API SERVICE CATEGORY: Proposed API SM API SL API SJ
ISSUE DATE: 2004 2001 1996
ASTM ENGINE TEST SEQUENCE VIB SEQUENCE VIB SEQUENCE VIA
SEQUENCE
ASTM Test Procedure: ASTM D 6837a ASTM D 6837a ASTM D 6202a
SAE 0W-20 and 5W-20
viscosity grades
% fuel economy improvement, 2.3 min 2.0 min
Phase 1
% fuel economy 2.0 min 1.7 min 1.4 min
improvement, Phase 2
SAE 0W-30 and 5W-30
viscosity grades
% fuel economy 1.8 min 1.6 min
improvement, Phase 1
% fuel economy 1.5 min 1.3 min 1.1 min
improvement, Phase 2
% total fuel economy improvement 3.0 min
SAE 10W-30 and all other
grades not listed above
% fuel economy improvement, 1.1 min 0.9 min
Phase 1
% fuel economy improvement, 0.8 min 0.6 min 0.5 min
Phase 2
% total fuel economy improvement 1.6 min
a
All fuel economy values, Phases 1 and 2, are determined relative to ASTM Reference Oil BC.
Vol. 2 LUBRICATION AND LUBRICANTS 45

Table 10. Bench Test Requirements for Catalyst Compatibility, Wear and Volatility
ILSAC CLASSIFICATION: ILSAC GF-4 ILSAC GF-3 ILSAC GF-2
API SERVICE CATEGORY: Proposed API SM API SL API SJ API SH
ISSUE DATE: 2004 2001 1996 1992
Catalyst compatibility
phosphorus content, mass %, 0.08 max 0.1 max
ASTM D 4951
SAE 0W-20, 5W-20, 5W-30, 0.1 max
and 10W-30 only
SAE 5W-30 and 10W-30 only 0.12 max
sulfur content, mass %, ASTM
D 4951 or D 2622
SAE 0W and 5W multigrades 0.5 max
SAE 10W multigrades 0.7 max
Wear
phosphorus content, mass %, 0.06 min
ASTM D 4951
NOACK Volatility, ASTM D 5800
evaporation Loss after 1 h 15 max 15 max
at 2508C, %
SAE 0W-XX 22 max NR
SAE 5W-XX 22 max 25 max
SAE 10W-30 22 max 20 max
SAE 15W-40 20 max 18 max
Volatility, Simulated
Distillation, ASTM D 6417
% volatilized to 3718C 10 max 10 max
SAE 0W-XX 17 max NR
SAE 5W-XX 17 max 20 max
SAE 10W-30 17 max 17 max
SAE 15W-40 15 max 15 max

D 6709-03a) evaluates a lubricant’s performance in combating copper, lead and


tin bearing corrosion and measures viscous shear stability at high operating tem-
perature. A 42 C.I.D. single cylinder engine runs for 40 hours at 3150 rpm. The
oil temperature is raised to 1438C using an external oil heater (130,132,135). (5)
The ASTM Sequence VIB test (ASTM D 6837-04) measures the effects of multi-
grade engine oils on the fuel economy of passenger cars and light duty trucks. It
is run in a 1993 Ford 4.6 liter V-8 engine on a dynamometer test stand
(130,132,136).
Maximum phosphorus and sulfur contents are specified in GF-4 for catalyst
compatibility (130,137,138). Minimum phosphorus is specified for wear protec-
tion (132,137). Bench tests are specified for minimum volatility, maximum
high temperature deposits, filterability, engine rusting and foaming characteris-
tics (132,139–145).
Many of the engine test results are visual merit ratings. These are made by
experts trained in the rating process using standards published in Coordinating
Research Council (CRC) Manuals (146,147).
5.2. Heavy Duty Diesel Engine Oils for Trucks, Agricultural, and
Construction Vehicles. The driving forces for diesel engine oil development
46 LUBRICATION AND LUBRICANTS Vol. 2

Table 11. Bench Test Requirements for High Temperature Deposits, Filterability, Foam
Tendency and Stability, Engine Rusting and Flash Point
ILSAC CLASSIFICATION: ILSAC GF-4 ILSAC GF-3 ILSAC GF-2
API SERVICE CATEGORY: Proposed API SM API SL API SJ API SH
ISSUE DATE: 2004 2001 1996 1992
High temperature deposits,
TEOST, ASTM D 6335
Total Deposits, mg 35 max 45 max
Filterability
Engine Oil Water Tolerance
Test, ASTM D 6794
% flow reduction with 50 max 50 max
0.6% water
% flow reduction with 50 max 50 max
1.0% water
% flow reduction with 50 max 50 max
2.0% water
% flow reduction with 50 max 50 max
3.0% water
Engine Oil Filterability Test,
ASTM D 6795
% flow reduction 50 max 50 max 50 max 50 max
Foam tendency/stability,
ASTM D 892 (Option A)
sequence I Oct-00 Oct-00 Oct-00 Oct-00
sequence II 50/0 50/0 50/0 50/0
sequence III Oct-00 Oct-00 Oct-00 Oct-00
high temperature 100/0 100/0 200/50 NR
Engine rusting, Ball Rust Test, (See Engine Test
ASTM D 6557 Sequence IID)
Avg Gray Value 100 min 100 min
Flash point, ASTM D 92, 8C
0W-XX 200 min NR
5W-XX 200 min 200 min
10W-30 200 min 200 min
15W-40 215 min NR

are engine durability, reduction of NOx, and soot emissions and compatibility
with exhaust aftertreatment systems. As with gasoline engine oils, the perfor-
mance of diesel engine oils is continually being upgraded to meet engine manu-
facturers’ needs, consumer requirements and government regulations. Diesel
engine oils are licensed and certified by the API in its Engine Oil Licensing
and Certification System (EOLCS) (130). Performance Requirements, test meth-
ods, and limits are cooperatively established by the ASTM Heavy Duty Engine
Oil Classification Panel (HDEOCP) and the Diesel Engine Oil Advisory Panel
(DEOAP). Both of these groups are made up of members of the Engine Manufac-
turers Association (EMA) and oil and additive company members of API (148).
Table 12 is an example of the performance criteria and the engine and
bench tests used to measure them, for the API ‘‘C’’ category diesel engine oils.
This table illustrates the increasingly rapid growth of new performance require-
ments (categories), the number of tests and the number of new engines required
Vol. 2 LUBRICATION AND LUBRICANTS 47

Table 12. Performance Requirements of 4-Stroke Diesel Engines


API service category CG-4 CH-4 CI-4 PC-10
foaming characteristics (ASTM D892)
foaming/settling, mL, max,
sequence I 10/0 10/0 10/0 10/0
sequence II 20/0 20/0 20/0 20/0
sequence III 10/0 10/0 10/0 10/0
ASTM corrosion test D5968 D5968 D6594
copper increase, ppm, max 20 20 20
lead increase, ppm, max 60 120 120
tin increase, ppm, report 50 50
copper corrosion by ASTM D130, 3 3 3
max rating
NOACK volatility(ASTM D5800) or X
distillation (ASTM D2887)
evaporative loss at 2508C, % max 15
SAE 10W-30, % volatility loss at 20
2508C, max
SAE 15W-40, % volatility loss at 18
2508C, max
or or
volatility by capillary gas
chromotography(ASTM D6417)
SAE 10W-30, % volatility loss at 9.3
3718C, max
SAE 15W-40, % volatility loss at 12.5
3718C, max
shear stability(ASTM D6278)
kinematic viscosity after shearing
SAE XW-30, cSt min 9.3 9.3 X
SAE XW-40, cSt min 12.5 12.5 X
shear stability -90 cycles(ASTM X
ASTM D3945)
low temperature pumpability with
sooted oils
ASTM D4684 (MRV TP-1)Viscosity of
75 h used
oil sample from T-10 test at 208C, 25,000 X
mPa s, max
high temperature/high shear viscosity, 3.5 X
mPa s, min
elastomer compatibility limits X
nitrile
volume change þ5/3 X
hardness þ7/5 X
tensile strength þ10/TMC 1006 X
elongation þ10/TMC 1006 X
silicone
volume change þTMC 1006/3 X
hardness þ5/TMC 1006 X
tensile strength þ10/45 X
elongation þ20/30
polyacrylate
volume change þ5/3 X
hardness þ8/5 X
tensile strength þ18/15 X
elongation þ20/30 X
48 LUBRICATION AND LUBRICANTS Vol. 2

Table 12. ðContinuedÞ


API service category CG-4 CH-4 CI-4 PC-10
FKM
volume change þ5/2 X
hardness þ7/5 X
tensile strength þ10/TMC 1006 X
elongation þ10/TMC 1006 X

for certification of heavy-duty diesel engine oils (149–153). Detailed performance


requirements and engine test details for European diesel oil categories are in
reference (150).
5.3. Gas Turbine Oils. The performance requirements for aircraft jet
engines are described in the military specifications MIL-PRF-7808L and MIL-
PRF-23699F. These specifications require bench tests for physical, chemical,
and performance characteristics, an accelerated endurance test in a turboshaft
engine and, if these requirements are met, a 150-h test on two different models
of aviation gas turbines conducted in a test cell and a 500-h flight evaluation
test (154,155).
Critical bench test performance requirements in MIL-PRF-23699F include
oxidation and corrosion stability tests (OCS) for 72 h each at temperatures of
175, 204, and, 2188C, a thermal stability and corrosivity test at 2748C, a test
for gear load carrying ability, and tests for bearing deposits and bearing corro-
sion. The maximum pour point specified is 548C (155).
The wide operating temperature range needed in this application is the rea-
son that neopentyl polyol ester base stocks are used for aircraft jet engine oils.
MIL-PRF-23699F fluids have a nominal viscosity of 5 cSt at 1008C. The nominal
viscosity of MIL-PRF-7808L fluids is 3 cSt at 1008C.
5.4. Automatic Transmission Fluids. Automatic Transmission Fluids
(ATFs) are highly complex fluids that perform many essential functions in an
automatic transmission (156). They transfer power in the torque converter; pro-
vide hydraulic pressure to operate clutches and shift gears; lubricate bearings,
gears and bushings; remove heat from the transmission; and provide the right
friction profile for proper operation of plate, band, and torque converter clutches.
Automatic transmissions normally operate at temperatures of 75–958C and
start up at temperatures as low as 408C. The ATFs must flow easily at low tem-
peratures, be highly resistant to thermal and oxidative degradation at high tem-
peratures, be noncorrosive towards all transmission components, be resistant to
foaming and have specialized friction and wear properties. They may contain as many
as 15 different additives and a mixture of base stocks in API groups I–IV (114).
The major automobile and truck manufacturers each develop their own
specifications for automatic transmission fluids, often using proprietary test
methods and test materials consistent with their own transmission designs. Cur-
rent examples include General Motors Dexron III fluid, version H; Ford’s Mercon
V and Mercon SP; Daimler Chrysler’s MS-9602 (ATFþ4); Toyota’s Fuel
Saving WS ATF, and Detroit Diesel Allison’s TES-295 (157). These fluids
are not necessarily interchangeable. Performance requirements specified by all
Vol. 2 LUBRICATION AND LUBRICANTS 49

manufacturers include friction characteristics, oxidation stability, low tempe-


rature flow, shear stability, EP and wear protection, antishudder, and friction
durability.
Fluids designed for use in continuously variable transmissions (CVTs)
require special frictional characteristics. There are two types of CVT used in
modern production vehicles: (1) Belt-drive CVTs using a Van Doorne Type steel
push-belt or a link-plate chain running in hydraulically adjustable, variable width
sheaves or pulleys. Torque is transferred between the sheaves and the belt by the fric-
tional force provided by metal–metal contact. (2) Traction-drive CVTs using a mova-
ble roller between two disks with toroid-shaped cavities. Torque is transferred
between the roller and disks by a fluid film in the elastic-plastic (EHD) flow regime.
Fluids with high traction coefficients, two to three times higher than those
of conventional ATFs, are required for efficient operation of the toroidal traction
drive. Fluids that have been evaluated as base stocks for traction fluids include
derivatives of the a-methyl styrene dimer, of isomerized tricyclopentadiene and
other multiring naphthenic structures (158,159).
Transmission fluids for belt-drive CVTs require high metal – metal fric-
tion coefficients to provide adequate torque transfer between the belt and
sheaves. They must also provide good antiwear and antipitting protection.
Since CVTs use controlled slip torque converters and wet clutches for starting
and reverse, these properties must be provided without sacrificing the anti-
shudder and wet clutch performance of modern ATFs. They must also have
the proper viscosity characteristics over the entire operating temperature
range, excellent thermal and oxidation stability and compatibility with other
transmission components.
5.5. Automotive Gear Oils. Automotive and industrial gear systems
are the principal applications for mild EP lubricants. These lubricants are formu-
lated to prevent wear, fatigue pitting, and catastrophic failure (spalling, scoring,
scuffing) of gear tooth surfaces. They are also formulated to protect against ther-
mal and oxidative degradation, rust, cuprous alloy corrosion, foaming, and oil
seal deterioration (160).
Automotive gear lubricants in API category GL-5 are designed for use in
final drive axles (differential gears) in moderate to severe service. Gear oils in
API category MT-1 are intended for use in nonsynchronized manual transmis-
sions in heavy-duty trucks and buses (161). Test requirements for GL-5 gear
lubricants include

CRC L-37 24-h dynamometer test, operated at high torque, low speed condi-
tions, to measure gear scoring and wear (162).
CRC L-42 2-h dynamometer test, operated at high speed with occasional shock
loading, to measure gear scoring (162).
CRC L-60-1 48-hour oxidation test with motored gears to measure viscosity in-
crease and insolubles (162).
CRC L-33 7 days in humidity cabinet after initial motoring phase to rate rust-
ing (162).
ASTM D 130 to measure copper strip corrosion (163).
ASTM D 892 to measure foam tendency (164).
50 LUBRICATION AND LUBRICANTS Vol. 2

Gear Oil viscosity grades are defined by the SAE J306 Automotive Gear Oil
Viscosity Classification (160,161,165).
The MT-1 requirements include additional CRC L-60-1 ratings, ASTM D 5662-
99 seal compatibility tests (166), ASTM D 5579-04 high temperature cyclic dur-
ability test (167), a more severe copper corrosion requirement, and ASTM D
4998-95(2003)e FZG tractor hydraulic fluid wear test (161,162,168).
SAE J2360 and MIL-PRF-2105E, which have identical performance require-
ments, combine the requirements of GL-5 and MT-1 gear oils (160).

Limited slip differentials contain friction clutches that operate to apply


equal torque to each driving wheel under slippery conditions. Lubricants for
these units require a supplemental friction modifier added to the GL-5 lubricant
to prevent chatter and noise when the clutches engage (161).
5.6. Industrial Oil Viscosity Classification and Guide Recommen-
dations. The ISO viscosity classification system for industrial oils, based on
kinematic viscosity at 408C, is shown in Table 13 (169,170). Other viscosity clas-
sification systems include ASTM D 6080 (171), and ANSI/AGMA 9005 (172).
Recommended ISO viscosity grades for hydrodynamic bearings operating
between 15 and 608C are given in Table 14 (173). Table 15 shows viscosity

Table 13. Viscosity System for Industrial Fluid Lubricantsa


Kinematic
viscosity limits,
cSt at 408C
Midpoint viscosity,
Viscosity grade cSt at 408C min max
ISO VG 2 2.2 1.98 2.4
ISO VG 3 3.2 2.88 3.52
ISO VG 5 4.6 4.14 5.06
ISO VG 7 6.8 6.12 7.48

ISO VG 10 10 9.00 11.0


ISO VG 15 15 13.5 16.5
ISO VG 22 22 19.8 24.2
ISO VG 32 32 28.8 35.2
ISO VG 46 46 41.4 50.6
ISO VG 68 68 61.2 74.8

ISO VG 100 100 90.0 110


ISO VG 150 150 135 165
ISO VG 220 220 198 242
ISO VG 320 320 288 352
ISO VG 460 460 414 506
ISO VG 680 680 612 748

ISO VG 1000 1000 900 1100


ISO VG 1500 1500 1350 1650
ISO VG 2200 2200 1980 2420
ISO VG 3200 3200 2880 3520
a
Reprinted with permission.
Vol. 2 LUBRICATION AND LUBRICANTS 51

Table 14. Suggested Viscosity Grade of Oil for Plain Bearings Operating between
15 and 608C
Loading on projected bearing area

Speed range, Light, Up to Medium, Heavy,


rpm 700 kPa 700–1700 kPa >1700 kPa
5,000–10,000 10
2,000–5,000 15
1,000–2,000 22 32, 46
500–1000 32, 46 68, 100
300–500 68, 100 100,150
100–300 100, 150 220, 320 320, 460
50–100 150, 220 220, 320 460, 680, 1000
< 50 220, 320 320, 460 460, 680, 1000

grade recommendations for industrial enclosed gears (174). The recommended


viscosity grade for rolling bearing lubricants as a function of temperature and
dN number are given in Ref. 175. dN number is the product of the bore diameter
and rotational speed.
The lubrication of gears, hydrodynamic thrust and radial bearings, rolling-
contact bearings and other machine elements, such as sintered metal sliding
bearings, screws, wire ropes, seals, valves, springs, and pneumatic components,
is discussed at length in Ref. 176.

Table 15. ISO Viscosity Grade Guide Recommendations for Enclosed Gears
Ambient temperature
Type of gear 10 to 108C 10 to 508C
parallel shaft units: input speeds to 3600 rpm
single reduction: shaft center distancea
up to 20 cm 68, 100 100, 150
20–50 cm 68, 100 150, 220
> 50 cm 100, 150 150, 220
double reduction: shaft center distance
up to 20 cm 68, 100 100, 150
> 20 cm 100, 150 150, 220
triple reduction: shaft center distance
up to 20 cm 68, 100 100, 150
20–50 cm 100, 150 150, 220
>50 cm 150, 220 220, 320
input speeds > 3600 rpm: all types and sizes 46 68
planetary units: diameter of housing
up to 40 cm 68, 100 100, 150
> 40 cm 100, 150 150, 220
bevel gears, spiral or straight: cone distanceb
up to 30 cm 68, 100 150, 220
> 30 cm 100, 150 220, 320
gear motors and shaft mounted units 68, 100 150, 220
a
Shaft center distance ¼ distance between centers of largest gear and its pinion.
b
Cone distance ¼ one-half the outside diameter of the largest gear.
52 LUBRICATION AND LUBRICANTS Vol. 2

5.7. Turbine Oils for Industrial Steam and Gas Turbines. Steam
and gas turbines are prime movers for electric power generation, for large
ocean-going vessels, for pumps and compressors, for processing sugar cane, for
rolling steel, and in many other industrial processes. Lubricating oils for the
thrust and radial shaft bearings on these turbines may also be used for the gen-
erators, reduction gears, compressors, and other machines that the turbines are
driving.
Compared to internal combustion engine oils, turbine oils have a relatively
small number and concentration of additives. Their lives, however, are measured
not in miles or hours, but in decades. A popular advertisement several years ago
showed a young engineer next to a large turbine with the caption The oil in this
turbine was put there before this engineer was born. Steam and gas turbine
bearings are lubricated by an oil circulation system, which includes the oil, a
pump, driven by the turbine shaft, an oil reservoir, a filter, an oil cooler, the bear-
ings, and auxiliary equipment.
Turbine oils must have excellent thermal and oxidation stability, protect
bearing surfaces from corrosion, separate easily from water, and they may not
entrain air or foam excessively. Some may also require load carrying capacity
beyond that afforded by a hydrodynamic lubricating film.
Typical specifications for an ISO 32 grade steam turbine oil and an ISO 32
grade gas and steam turbine oil are shown in Table 16. Specifications for oils
requiring additional load carrying capacity have a minimum ASTM D 5182
FZG failure load stage of 8 or 9 (177–181).

Table 16. Typical Turbine Oil Specifications


Test Steam Steam and gas
Property method turbine oil turbine oil
ISO viscosity grade D 2422 32
flash point, 8C, D 92 215 min
pour point, 8C D 97 12 max
total acid number,
mg KOH/g D 974 2 max
air release, separation
time, min. at 508C D 3427 4 max
foam tendency/ stability D 892
sequence I, mL 10/0
sequence II, mL 20/0
sequence III, mL 10/0
demulsibility D 1401
oil/water/emulsion, mL 40/37/3
separation time, min 20
copper strip corrosion, D 130
3 h at 1008C 1 max
distilled water rust test D 665 Pass
oxidation stability, TOST D 943
time to 2.0 TAN, h 3000 min 7000 min
oxidation stability, RPVOT D 2272
time to 25 psi pressure 500 min 1000 min
drop, min
Vol. 2 LUBRICATION AND LUBRICANTS 53

Premium turbine oils can be formulated for >10,000 h D 943 oxidation lives,
RPVOT lives of >1000 min, D 1401 demulsibility of 41/39/0 in 3 min and an
FZG failure load stage of 12 (178).

6. Metalworking Fluids

Metalworking fluids are lubricants and coolants for metalworking processes.


There are metal cutting or machining processes, which include drilling, turning,
grinding, honing, lapping, milling, broaching, and other metal removal opera-
tions; and deformation processes such as rolling, drawing, extrusion, forging,
stamping, and spinning (see METAL TREATMENTS).
6.1. Types of Metalworking Fluids. Types of metalworking fluids are
found in the following sections (175).
Straight Oils. Blends of lubricity and EP additives in naphthenic base
oils from API group V stocks and, increasingly, paraffinic oils from Groups
I –IV stocks. Additives include sulfur, chlorine, fatty oils, sulfurized, chlori-
nated or sulfo-chlorinated fatty oils, phosphorus, metal salts, and solid lubri-
cants (182).
Soluble Oils. A misnomer; They are milky macroemulsions of naphthenic
or paraffinic hydrocarbon oil and additives in water. Emulsifiers include petro-
leum sulfonates and naphthenates, alkanolamides, amine salts of fatty acids,
and other surfactants. Particle sizes range from 2 to 5 mm and lubricating addi-
tives are the same as those used in straight oils (183–185).
Synthetics. The word ‘‘synthetic’’ has a somewhat different meaning for
metalworking fluids. Here it means an aqueous metalworking fluid that contains
no mineral oil. Synthetic metalworking fluids are solutions of totally water-
soluble chemicals or they are emulsions of synthesized base oils from group V
stocks. Base oils include polyglycols, amides, and organic esters. Particle sizes
are <103 mm. Synthetics may contain borates, molybdates, phosphates, amines
and amides for corrosion protection, surface active agents to promote wetting,
and water soluble lubricity and EP additives (183–185).
Semisynthetics. Microemulsions of oil and additives in water. Emulsi-
fiers include petroleum sulfonates, ethoxylated alcohols, fatty amines and
amides, sulfonamide compounds, and other surfactants. Particle sizes are
between 103 and 1 mm. Lubricity and EP additives include polyol esters, phos-
phate esters, dithiophosphates and carbamates, and others (183–185).
All of the water-based fluids will contain biocides, or the biocide may be
added to the cutting fluid reservoir, or both. Biocides are used to prevent the
growth of bacteria and fungi in the reservoir (186). A list of commonly used che-
micals registered by the U. S. Environmental Protection Agency (EPA) for use as
metalworking fluid preservatives is included in reference (187).
Polymers may be added to all types of fluid to reduce the amount of mist
generated during use.
In general, for high speed cutting operations where cooling is critical, water-
based fluids are preferred. For heavy duty, slow speed cutting, where lubrication
is more important, straight oils are used. However, there are a multitude of
factors involved in cutting fluid selection. They include machinability of the
54 LUBRICATION AND LUBRICANTS Vol. 2

Rake face

Cutting
tool

Clearance face
Direction
of motion
Built
up
edge

Fig. 7. Chip formation in metal cutting.

metal, the mix of cutting operations, recovery and disposal of the spent fluid,
plant size, water quality, type of filtration, and cost (183).
6.2. Metal Cutting and Grinding Applications. Lubrication in metal
cutting and grinding operations differs from lubrication in bearings and gears in
several important ways: (1) The relative motion of the chip and workpiece sur-
faces on the rake face and clearance face of the cutting tool (Fig. 7) tends to
move fluid away from the contact zone rather than into it. Lubrication, therefore,
is always in the boundary regime. Cutting fluid enters the contact zone through a
labyrinth of capillaries formed by asperities on the surfaces. It is drawn in by the
pressure difference between the contact zone and the atmosphere. (2) As the chip
shears off the workpiece, nascent metal surfaces are formed, which rub on the
rake and clearance faces of the cutting tool. Without an oxide coating, these
surfaces are very reactive. Pieces of this active metal weld together to form a
‘‘built-up edge’’, which causes poor quality of the machined surface and wear of
the clearance face. Wear of the rake face occurs behind the built-up edge and
shows up as a crater some distance from the tool tip. Lubricating additives in
the cutting fluid must react with the nascent metal surfaces, at the temperature
on the surfaces, to minimize formation of the built-up edge and prevent wear of
the cutting tool. (3) The additives must also reduce friction force on the chip. The
higher the friction force on the chip, the smaller the shear angle, the longer the
shear plane and the more force it takes to shear the metal and form the chip.
About one-quarter of the work in metal cutting is done to overcome friction
and about three-quarters is done to form the chip. (4) Finally, cooling is extre-
mely important to prevent distortion of the machined surface and to extend
the life of the tool. Tool life decreases exponentially as the temperature at the
tool–chip interface increases (188):

Tyn ¼ K ð24Þ

where T is the tool life (min); y is temperature at the tool–chip Interface (8C); and
n is an exponent dependent on type of tool, with value usually between 20 and 30;
and K is a constant dependent on tool and workpiece material.
The other function of the cutting fluid is to flush chips and metal fines,
called swarf, from the cutting area. This prevents the swarf from damaging
the finished surface.
Vol. 2 LUBRICATION AND LUBRICANTS 55

6.3. Metal-Forming Operations. Metal forming is the process of chan-


ging the shape or size of a piece of metal without removing any material. Opera-
tions include blanking, drawing, extrusion, forming, forging, rolling, spinning,
and stamping.
For example, soluble oil emulsions are sprayed on the backup rolls in alu-
minum hot rolling mills to cool the rolls and lubricate the process. Its principle
purpose is to prevent pickup of aluminum by the work roll surfaces. Friction
control is critical to provide traction to bring the work into the rolls, and to pre-
vent slippage of the rolls on the surface being rolled.
In cold rolling aluminum sheet and foil, narrow boiling, paraffinic distillate
oils fortified with fatty acids or alcohols or fatty esters such as lanolin and butylstea-
rate. Normal paraffins, made by passing an atmospheric distillate like kerosene
through a molecular sieve, and then hydrotreating and fractionating the n-paraffins,
are widely used base stocks for aluminum cold rolling oils (184,189–192).

7. Lubricating Grease

Lubricating grease is a mixture of a fluid lubricant and a thickener that is dis-


persed in oil. It is a non-Newtonian fluid that acts like a sticky solid when there
is no shear stress on it and that flows when a shear force is applied. It is widely
used in ball and roller bearings and other machine elements where liquid lubri-
cants cannot be retained. Automotive applications include wheel bearings, alter-
nator bearings, constant velocity driveshaft joints (CVJ), mechanical clutch
mechanisms, and gear-box bearings. Steel mills have central lubricating systems
filled with multipurpose EP greases that are pumped long distances and applied
through complex dispensing systems. Electric motor bearings, couplings, and roll
neck bearings are some of the applications for grease in this industry (193,194).
Greases are classified by their consistency and by the type of thickener. The
consistency of a grease is measured by cone penetration, ie, the distance, in mm/
10 (penetration number), that a standard cone, acting under the influence of
gravity, will penetrate a grease sample at 258C under test conditions described
in ASTM D 217-02 (195). The National Lubricating Grease Institute (NLGI)
classification system for greases, based on worked penetration, is: (193)
Worked penetration is the penetration of a sample immediately after it has
been subjected to 60 double strokes in a standard grease worker.

NLGI Consistency Worked penetration,


number mm/10 at 258C
000 445–475
00 400–430
0 355–385
1 310–340
2 265–295
3 220–250
4 175–205
5 130–160
6 85–115
56 LUBRICATION AND LUBRICANTS Vol. 2

The thickener in a grease may play as important a role as the oil in


lubrication. The first greases were lime soap greases, made by saponifying
a fatty oil with an slurry of hydrated lime, Ca(OH)2, and then dispersing
the soap in oil and heating to drive off excess water. These greases are highly
resistant to water, but unstable at high temperatures. Soda soap greases,
made with NaOH, are stable at high temperatures, but wash out in moist
conditions. Lithium soap greases resist both heat and moisture. A mixed-
base soap is a combination of soaps, offering some of the advantages of each
type (196).
A complex soap is formed by the simultaneous reaction of an alkali with a
high molecular weight fat or fatty acid and with a low molecular weight organic
or inorganic acid. Non-soap thickeners include clays, silica gels, carbon black,
and synthesized organic materials such as substituted ureas and polyureas
(193,194,196).
Lithium soap and lithium complex greases are the most widely used
multipurpose greases. The fatty acid portion is most often 12-hydroxystearic
acid, made from hydrogenated castor oil. Dimethyl azelate and sebacate have
been used as complexing agents. The dropping point of lithium soap
greases is in the range 180–1908C; of lithium complex greases, 260–3008C.
Dropping point is the lowest temperature at which a grease is sufficiently
fluid to drip, as determined by test method ASTM D 566-02 or D 2265-00. It is
an indication of whether a grease will stay in a bearing at operating
temperatures (197,198).
Polyurea greases are made by reacting amines with isocyanates or diisocya-
nates in base oil. Because they have superior high temperature durability, water
resistance, structural stability and low torque performance, polyurea greases are
used in sealed-for-life bearings in electric motors and alternators and water
pumps, and in constant velocity joints (CVJ).
Bentonite clay (sodium montmorillonite and hectorite) is used as a grease
thickener after being coated with quaternary alkyl ammonium cations
(see CLAYS, USES). Clay thickened greases have good oxidation, excellent thermal,
and mechanical stability. They have been successfully used in applications where
ambient temperatures >2608C. An example is roller bearings on conveyor belts
in glass factories.
The effect of shear rate on the apparent viscosity of three greases is shown
in Fig. 8 (199). Grease has a yield point or yield stress, and does not flow until
that stress is exceeded. Given the high shear rates in most rolling bearings, the
apparent viscosity of grease is nearly the same as the viscosity of its base oil
(193,199).
Table 17 is a grease application guide, comparing the properties of the sev-
eral grease thickener systems (193).
ASTM D4950-01 (2004)e (200) covers lubricating greases suitable for
the periodic relubrication of chassis systems and wheel bearings of passenger
cars, trucks, and other vehicles. It includes specifications for two chassis
lubricants, LA and LB, and three wheel bearing lubricants, GA, GB, and
GC. To facilitate easy and accurate identification of greases in these cate-
gories, NLGI has made available the ‘‘NLGI Certification Mark’’ to be dis-
Vol. 2 LUBRICATION AND LUBRICANTS 57

100
80
60

40
Soda
base
20

Aluminum
10 base
Apparent viscosity, Pa·s

8
6 Calcium
base
4

1
0.8
Viscosity of mineral oil used
0.6

0.4

0.2

0.1
10 20 40 60 100 200 400 600 1000 2000 4000 10,000
Rate of shear, s−1

Fig. 8. Effect of shear rate on the apparent viscosity of three NLGI grade 1 Greases at
258C.

played on grease packaging. These specifications and the certification marks


are also described in the January 2000 revision of SAE recommended practice
J310 (194).

8. Solid-Film Lubrication

There are several applications where liquids do not provide adequate lubrication.
In aerospace applications, eg, wide temperature ranges (240 to 9008C) are
encountered accompanied by high vacuum. The manufacture of glass products
requires lubricants that work at the temperature of molten glass. Lubricants
are also needed that can withstand chemical attack in a corrosive atmosphere
and are resistant to acids, aggressive gases, liquid oxygen, fuels, and solvents.
Electrical contacts and precision machinery that require low start-up friction
Table 17. Grease Application Guidea,b
Calcium, Calcium, Aluminum Calcium Lithium
Properties Aluminum Sodium conventional anhydrous Lithium complex complex complex Polyurea Organo clay
dropping 110 163–177 96–104 135–143 177–204 260 þ 260 þ 260 þ 243 260 þ
point, 8c

usable tempe- 79 max 121 max 93 max 110 max 135 max 177 max 177 max 177 max 177 max 177 max
rature, 8c
water good– poor–fair good– excellent good good– fair– good– good– fair–
resistance excellent excellent excellent excellent excellent excellent excellent
work stability poor fair fair–good good– good– good– fair–good good– poor– fair–good
excellent excellent excellent excellent good
oxidation excellent poor– poor– fair– fair– fair– poor–good fair– good– good
stability good excellent excellent excellent excellent excellent excellent
rust good– good– poor– poor– poor– good– fair– fair– fair– poor–
protection excellent excellent excellent excellent excellent excellent excellent excellent excellent excellent
pumpability poor poor–fair good– fair– fair– fair– poor– good– good– good
(in centra- excellent excellent excellent good fair excellent excellent

58
lized
systems)
oil separation good to fair–good poor– good good– good– good– good– good– good–
good excellent excellent excellent excellent excellent excellent
appearance smooth smooth– smooth smooth smooth smooth smooth smooth smooth and smooth
and clear fibrous and and and and and and buttery and
buttery buttery buttery buttery buttery buttery buttery
other adhesive EP grades EP grades EP grades EP grades inherent EP grades EP grades
properties and available available available available EP and available available
cohesive reversible reversible antiwear
principal thread rolling general military multiservice multiservice multiservice multiservice multiservice high
uses lubricants contact uses for multi- automotive industrial automotive automotive automotive temperature
bearings economy service and and industrial and and (frequent
industrial industrial industrial relube)
a
Multiservice includes rolling contact bearings, plain bearings and others. Reversibility is the ability of a grease to return to its normal grease-like
consistency after temporary exposure to temperatures near or above its dropping point.
b
Reprinted with permission.
Vol. 2 LUBRICATION AND LUBRICANTS 59

and cannot tolerate contamination by lubricating oils and greases are further
examples (201). For these applications, solid lubricating films are applied to
the rubbing surfaces.
The wide range of solid lubricants can generally be classified as either
inorganic compounds or organic polymers (both commonly used in a bonded
coating on a matching substrate), chemical conversion coatings, and metal
films. Since solid-film lubricants often suffer from poor wear resistance and
inability to self-heal any breaks in the film, the search continues for improved
compositions.
8.1. Inorganic Compounds. The most important inorganic mater-
ials are layer – lattice solids in which the bonding between atoms in an indi-
vidual layer is by strong covalent or ionic forces and those between layers are
relatively weak van der Waal’s forces. Because of their high melting points,
high thermal stabilities, low evaporation rates, good radiation resistance,
and effective friction lowering ability, molybdenum disulfide (MoS2)
[1317-33-5], and graphite [7782-42-5] are the preferred choices in this
group. Hexagonal boron nitride and boric acid also provide excellent lubri-
cation (202). Among other layer – lattice solids that find occasional use are
tungsten disulfide (WS2) [12138-09-9], tungsten diselenide (WeS2) [12067-46-8],
niobium diselenide (NbSe2) [12034-77-4], calcium chloride [10108-64-2];
cadmium iodide (CdI) [7790-80-9], and graphite fluoride [11113-63-6]
(Table 18) (203).
Graphite is widely used as a dry powder or as a colloidal dispersion in
water, petroleum oil, castor oil, mineral spirits, or other solvents. The water dis-
persions are used for lubricating dies, tools, metalworking molds, oxygen equip-
ment, and wire drawing. Graphite dispersed in solvents is used for drawing,
extruding, and forming aluminum and magnesium, as a high temperature lubri-
cant for conveyors, and for a variety of industrial applications. Graphite alone is
ineffective in vacuum since adsorbed water normally plays a decisive role in its
lubricating ability. Its film-forming ability can be restored, however, by mixing
with cadmium oxide or MoS2 and most organic materials, so that graphite may
offer effective lubricating action when bonded to the surface with organics. Air
oxidation commonly sets a use limit of 5508C, and high friction may occur in
air with water desorption >1008C (205).
Molybdenum disulfide has increasingly supplanted graphite for three rea-
sons: consistent properties in rigid specifications, independence from need for
adsorbed vapors in providing lubrication, and superior load capacity (205).
Like graphite, MoS2 has a layer–lattice structure in which weak sulfur–sulfur
bonds allow easy sliding between each sulfur–molybdenum–sulfur layer.
Molybdenum disulfide covered by MIL-M-7866 is the most common lubricant
grade: It is purified from molybdenite ore and is essentially free of abrasive
constituents (203).
Petroleum oil and grease dispersions of MoS2 are used extensively in
automotive and truck chassis lubrication and in general industrial use. Dis-
persions are also made with 2-propanol, polyalkylene glycols, other syn-
thetic oils, and water for airframe lubrication, in wire drawing, and for
splines, fastenings, gears, and fittings. Above 4008C, the MoS2 is oxidized to
Table 18. Common Solid Lubricantsa
Acceptable usage temperature, 8C
Minimum Maximum Avg friction coefficient, f
In N2 or In N2 or In N2 or
Material In air vacuum In air vacuum In air vacuum Remarks
molybdenum 240 240 370 820 0.10–0.25 0.05–0.10 low f, carries high load, good overall
disulfide (MoS2) lubricant, Can promote metal
corrosion
polytetrafluoro- 70 –70 290 290 0.02–0.15 0.02–0.15 lowest f of solid lubricants, load
ethylene (PTFE) capacity moderate and decreases
at elevated temperature
fluoroethylene- 70 70 200 200 0.02–0.15 0.02 low f, lower load capacity than PTFE
propylene
copolymer (FEP)
graphite 240 540 unstable in 0.10–0.30 0.02–0.45 low f and high load capacity in air,
vacuum high f and wear in vacuum,

60
conducts electricity
niobium diselenide 370 1,320 0.12–0.40 0.07 low f, high load capacity, conducts
(NbSe2) electricity (in air or vaccum)
tungsten disulfide 240 240 430 820 0.10–0.20 f not as low as MoS2, temp capability
(WS2) in air a little higher
tungsten diselenide 370 1320 same as for Ws2
(WSe2)
lead sulfide (PbS) 480 0.10–0.30 very high load capacity, used primarily
as additive with other solid
lunricants
lead oxide (PbO) 650 0.10–0.30 same as for PbS
calcium fluoride– 430 430 820 820 0.10–0.25 can be used at higher temp than other
barium fluoride above 5408C solid lubricants, high f < 5408C
eutectic 0.25–0.40
(CaF2 –BaF2) above 5408C
antimony trioxide high load capacity, used as corrosion
(Sb2O2) inhibitor in MoS2 lubricants
a
See Ref. 203.
Vol. 2 LUBRICATION AND LUBRICANTS 61

molybdenum trioxide, which has a significantly higher coefficient of friction.


There are indications that, as rubbed films, both MoS2 and graphite may
accelerate corrosion: MoS2 by hydrolysis to form corrosive acids and graphite
by galvanic action.
Various other soft materials without the layer–lattice structure are used
as solid lubricants (209), eg, basic white lead or lead carbonate [598-63-0] used
in thread compounds, lime [1305-78-8] as a carrier in wire drawing, talc
[14807-96-6] and bentonite [1302-78-9] as fillers for grease for cable pulling,
and zinc oxide [1314-13-2] in high load capacity greases. Graphite fluoride is
effective as a thin-film lubricant up to 4008C and is especially useful with a sui-
table binder such as polyimide varnish (205). Boric acid has been shown to have
promise as a self-replenishing solid composite (202,206).
8.2. Organic Polymers. These self-lubricating polymers are used pri-
marily in three ways: as thin films, as self-lubricating materials (see BEARING
MATERIALS), or as binders for lamellar solids (202,204,207). Coatings are typically
applied in powder or dispersion form with thickness ranging upward from 25 mm.
The polymer is then fused to the surface as a coating that provides lubricity,
abrasion and chemical resistance, or release properties.
PTFE is outstanding in this group. In thin films it provides the lowest coef-
ficient of friction (0.03–0.1) of any polymer, is effective from 200 to 2508C, and
is generally unreactive chemically. The low friction is attributed to the smooth
molecular profile of PTFE chains that allows easy sliding (202,204). Typical
applications include chemical and food processing equipment, electrical compo-
nents, and as a component to provide improved friction and wear in other
resin systems.
Other polymers finding self-lubricating use are fluorinated ethylene –
propylene copolymer (FEP), perfluoroalkoxy resin (PFA), ethylene – chlorotri-
fluoroethylene alternating copolymer (ECTFE), and poly(vinylidene fluoride)
(PVDF) (207). With a useful temperature range up to 2008C, outstanding
weatherability, and low friction, FEP finds use in chemical process equip-
ment, roll covers, wire and cable, and as powder in resin bonded products.
Perfluoroalkoxy resin provides somewhat better mechanical properties than
PTFE and FEP at temperatures up to 2508C. The ECTFE provides superior
strength, wear resistance, and creep resistance from cryogenic temperatures
to 1658C. Although fairly expensive, it is effective in its common use as a
corrosion-resistant coating. Also having superior mechanical properties,
PVDF is more commonly used for lining chemical piping and reactor vessels
than as a lubricant.
8.3. Bonded Solid-Film Lubricants. Although a thin film of solid
lubricant that is burnished onto a wearing surface often is useful for break-in
operations, >95% are resin bonded for improved life and performance (210).
Use of adhesive binders permits applications of coatings 5–20 mm thick by
spraying, dipping, or brushing as dispersions in a volatile solvent. The perfor-
mance properties of some commonly used bonded lubricant films are listed in
Table 19 (207,210).
For many moderate-duty films for operating temperatures <80–1208C,
MoS2 is used in combination with acrylics, alkyds, vinyls, and acetate room
temperature curing resins. For improved wear life and temperatures up to
Table 19. Performance Properties of Typical Solid-Film Lubricantsa
Organicb Inorganicb
Thermo set Air dry
Specification: MIL-L-8937 MIL-L-46010 MIL-L-23398 MIL-L-46009 MIL-L-81329 AMS2525A AMS2526A
composition:
lubricant MoS2 MoS2/metallic MoS2/graphite PTFE MoS2 MoS2/graphite MoS2/graphite graphite MoS2
binder phenolic epoxy silicone phenolic silicate
application: spray spray spray spray spray aerosol spray impingement
cure: 1498C 2048C 2608C 2048C ambient ambient 2048C 1498C 1498F
compatibility:
LOX N/A N/A N/A N/A N/A N/A X X X
OZ N/A N/A N/A N/A N/A N/A X X X
rocket fuel X N/A L N/A N/A N/A L X X
jet fuel X X X X X L X X X
hydrocarbons X X X X X L X X X
solvents X X X X X L X X X

62
radiation: Fair N/A N/A N/A N/A V.G. V.G. V.G.
operating temp.
air (high) 2608C 2608C 3718C 2608C 1768C 2048C 6498C þ10938C 4008C
air (low) 2208C 2208C 1578F 2208C 2208C 1858C 2408C 2408C 2208C
vacuum 104Pa 107Pa 107Pa N/A N/A N/A 103Pa 107Pa 107Pa
load capacityc
force test 2500 2500 2500 150 2500 2500
lb gage lb gage lb gage lb lb gage lb gage
Falex Falex Falex LFW-1 Falex Falex
wear lifec
load 1000 1000 1000 150 1000 1000 50 50
lb gage lb gage lb gage lb lb gage lb gage
test Falex Falex Falex LFW-1 Falex Falex Falex Falex
time 60 min 450 min 60 min 120,000 120 min 70 min 2 min 5 min
coefficient of friction <0.1 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1 <0.1
corrosion resistance G. V.G. F E G  F  
a
See Ref. 210.
b
X ¼ Compatible; L ¼ Low; N/A ¼ Not Applicable.
c
Falex tests are all designed for English units and are, therefore, reported as such.
Vol. 2 LUBRICATION AND LUBRICANTS 63

150–3008C, baked coatings are commonly used with thermosetting resins, eg,
phenolics, epoxies, alkyds, silicones, polyimides, and urethanes. Of these, the
MIL-L-8937 phenolic type is widely used (204).
Inorganic binders are used, usually with graphite or MoS2, for extreme
conditions such as high vacuum, liquid oxygen, radiation resistance, and
high temperatures (208). The most common binder systems are silicates, phos-
phates, and aluminates. Some silicon and titanate metalloorganics used for
high temperature binders become inorganic on curing. An emerging class of
ceramic bonded materials for aerospace applications use either graphite, a
CaF2BaF2 eutectic, or proprietary systems, often with a glass frit binder
that is fused into a continuous film (203,208). Plasma spray coating avoids
overheating damage to the substrate metal while achieving the melting
point of at least one component in high temperature film compositions
(203,205,210).
The solid lubricant/binder ratio is a principal performance factor. High
lubricant content usually gives minimum friction, while high binder content
tends to give better corrosion resistance, hardness, durability, and a glossy finish
(210). With commonly used MoS2 –graphite and organic resin binders, the
optimum lubricant/binder ratio usually is from 1:1–4:1. With inorganic bind-
ing agents, the ratio is from 4:1 to as high as 20:1 and increases with high
temperatures.
Substrate Properties. It is clear from equation 8 that higher hardness of
the substrate lowers friction. Wear rate of the film also is generally lower. Phos-
phate undercoats on steel considerably improve wear life of bonded coatings by
providing a porous surface that holds reserve lubricant. The same is true for sur-
faces that are vapor- or sandblasted prior to application of the solid-film lubri-
cant. A number of typical surface pretreatments are given in Table 20 to
prepare a surface for solid-film bonding (208).
Optimum surface roughness usually is 0.05–0.5 mm; a very smooth surface
contains very little lubricant within its depressions, whereas rough peaks pene-
trate the lubricant to promote wear. Improved corrosion resistance may be

Table 20. Typical Pretreatments for Various Substratesa


Substrate Pretreatment
aluminum vapor degrease plus anodize
light abrasive blast plus chromate conversion
copper and its alloys vapor degrease
light abrasive blast plus chromate conversion
iron and steel vapor degrease
abrasive blast plus phosphate
stainless steel vapor degrease
sandblast
passivation
titanium alkaline cleaning
abrasive blast
fluoride phosphate or alkaline anodize
a
See Ref. 208.
64 LUBRICATION AND LUBRICANTS Vol. 2

obtained with a suitable subcoating surface conversion treatment or by inclusion


of inhibitors in the coating.
Plasma-Applied Coatings. Composite coatings consisting of, eg, NiCr as
a binder, Cr2O3 as hardener, BaF2/CaF2 as high temperature lubricant, and
silver as low temperature lubricant have been developed by NASA (211,212).
Applied as a plasma spray, these coatings provide lubrication at temperatures
from 85 up to 10008C (203,212).
Plasma enhanced chemical vapor deposition (PECVD) is used to produce
diamond-like carbon and near-frictionless carbon coatings on steel substrates.
Thin-film carbon coatings applied by this technique are described as ‘‘nanos-
mooth’’, and have excellent friction modifying, wear reduction and antiscuffing
characteristics (213–215).
Vacuum deposition techniques, such as sputtering, ion plating, physical
vapor deposition, pulse laser deposition, ion beam assisted deposition, and
plasma enhanced chemical vapor deposition, produce solid lubricant films that
strongly adhere to the substrate. When these advanced methods are combined
with surface texturing and micropatterning, the films achieve much improved
tribological properties (216).
8.4. Chemical Conversion Coatings. These involve inorganic surface
compounds developed by chemical or electrochemical action (see METAL SURFACE
CONVERSION TREATMENTS). One of the best known treatments for steel is phosphat-
ing to coat the surface with a layer of mixed zinc, iron, and manganese phos-
phates. Other films are anodized oxide coatings on aluminum, oxalate on
copper alloys, and various sulfides, chlorides, and fluorides. Although many
of these films are not strictly solid lubricants, they are often effective for
short-term wear resistance. For long-term effectiveness, they often provide a
porous reservoir for liquid lubricants and increased life of organically bonded
coatings.
Diffusion provides an alternative procedure for generating a chemically
modified surface, eg, sulfide surface films can be formed by immersing steel in
molten mixtures of sulfur-containing salts such as sodium thiosulfate or
sodium sulfide. Similar processes are employed for carburizing, nitriding, bor-
iding, or siliconizing. Metalliding can introduce a new element into many
metal surfaces from a molten fluoride bath. A number of hardening treat-
ments, as well as flame sprayed tungsten and titanium carbides, provide
excellent wear resistance. Some of these also provide good bases for low
shear strength films.
8.5. Metal Films. In many respects, soft metals such as gallium, indium,
thallium, lead, tin, gold, and silver are ideal solid lubricants (203,209). They have
low shear strength, can be bonded strongly to substrate metal as continuous
films, have good lubricity, and have high thermal conductivity. Metal films can
be applied by electroplating or by vacuum processes, eg, evaporation, sputtering,
and ion plating.
Melting points of gallium, indium, and tin are too low, and those of thallium
and lead are borderline when high surface temperatures are generated by high
speeds and loads. Gallium is a special case, ie, it is above its melting point under
most conditions and is too reactive with many metals. It is effective, however,
Vol. 2 LUBRICATION AND LUBRICANTS 65

when applied in a vacuum with AISI 440C stainless steel and with ceramics
(qv) such as boron carbide or aluminum oxide, which can be applied as
undercoats (217).
A number of metal films are used industrially. Copper and silver are elec-
troplated on the threads for bolt lubrication. Slurries of powders of nickel, cop-
per, lead, and silver are also used in commercial bolt lubricants. Tin, zinc,
copper, and silver coatings are used as lubricants in metalworking, where toxi-
city has virtually eliminated lead as a lubricating coating (218). Gold and silver
find limited use on more expensive workpiece materials such as titanium. Silver
films are useful in a variety of other sliding and rolling contacts, in vacuum and
at high temperatures, since silver forms no alloys with steels and is soft at high
temperatures.
Under severe conditions and at high temperatures, noble metal films may
fail by oxidation of the substrate base metal through pores in the film. Improved
life may be achieved by first imposing a harder noble metal film, eg, rhodium or
platinum–iridium, on the substrate metal. For maximum adhesion, the metal
of the intermediate film should alloy both with the substrate metal and the
soft noble-metal lubricating film. This sometimes requires more than one
intermediate layer. For example, silver does not alloy to steel and tends to lack
adhesion. A flash of hard nickel bonds well to the steel, but the nickel tends to
oxidize and should be coated with rhodium before applying silver of 1–5-mm
thickness. This triplex film then provides better adhesion and greatly increased
corrosion protection.

9. Extreme Ambient Conditions

9.1. Gas Lubrication. Despite severe limitations, gas lubrication of


bearings has received intensive consideration for its resistance to radia-
tion, for high speeds, temperature extremes, and use of the working fluid
(gas) in a machine as its lubricant. A primary limitation is, however, the
very low viscosity of gases (219) that leads to a limiting load of only 15 –
30 kPa (2.2 – 4.4 psi) for most self-acting (hydrodynamic) gas bearings and
up to 70 kPa (10 psi) for operation with external gas-lifting pressure in hydro-
static operation.
Gases that have been used for bearing lubrication include air, hydrogen,
helium, nitrogen, oxygen, uranium hexafluoride [7783-81-5], carbon dioxide,
and argon [7440-37-1]. A useful property of gases is that their viscosity, and
hence their capacity to generate hydrodynamic pressure P, increases with tem-
perature whereas the opposite is true for liquids. Gas viscosity is usually inde-
pendent of pressure up to 1 MPa (10 atm).
Hydrodynamic principles for gas bearings are similar to those involved with
liquid lubricants except that gas compressibility usually is a significant factor
(220,221). With gas employed as a lubricant at high speeds, start–stop wear is
minimized by selection of wear-resistant materials for the journal and bearing.
This may involve hard coatings, such as tungsten carbide or chromium oxide
flame plate, or solid lubricants, eg, PTFE and MoS2.
66 LUBRICATION AND LUBRICANTS Vol. 2

Because of the very small bearing clearances in gas bearings, dust


particles, moisture, and wear debris (from starting and stopping) should
be kept to a minimum. Gas bearings have been used in precision spindles,
gyroscopes, motor and turbine-driven circulators, compressors, fans, Brayton
cycle turbomachinery, environmental simulation tables, and memory
drums.
9.2. Liquid Metals. If operating temperatures rise >250–3008C, where
many organic fluids decompose and water exerts high vapor pressure, liquid
metals have found some use, eg, mercury for limited application in turbines;
sodium, especially its low melting eutectic with 23 wt % potassium, as a hydrau-
lic fluid and coolant in nuclear reactors; and potassium, rubidium, cesium, and
gallium in some special uses.
Liquid metal selection is usually limited to the lower melting point metals
such as those mentioned above. Liquid metal viscosity generally is similar to
water at room temperature and approaches the viscosities of gases at high
temperature (Fig. 9). Hydrodynamic load capacity with both liquid metals
and water in a bearing is about one-tenth of that with oil, as indicated in
Table 21.
The sodium–potassium eutectic is commercially available for use as a liquid
over a wide temperature range. Because of its excessive oxidizing tendency in
air, however, its handling and disposal is hazardous; it can be used only in closed
vacuum or in an inert gas atmosphere of helium, argon, or nitrogen. In addition
to the oxidation problem, bearing material selection is critical for liquid metal
bearings. Tungsten carbide cermet with 10–20 wt% cobalt binder gave superior
performance when running against molybdenum under heavy loads at low
speeds at temperatures up to 8158C (222).
A low melting (58C) gallium–indium–tin alloy has been the choice for small
spiral-groove bearings in vacuum for X-ray tubes at speeds up to 7000 rpm (223).
Surface tension 30 times that of oil avoids leakage of the gallium alloy from the
ends of the bearings.
9.3. Cryogenic Bearing Lubrication. Cryogenic fluids, such as liquid
oxygen, hydrogen, or nitrogen, are used as lubricants in liquid rocket propulsion
systems, turbine expanders in liquefaction and refrigeration, and pumps to
transfer large quantities of liquefied gases. Bearings operating in cryogenic
fluids are amply cooled from the standpoint of dissipating the heat generated
from friction. Unfortunately, the low viscosity of the fluids leads to marginal
lubrication.
For wear resistance and low friction, coatings of PTFE or MoS2 generally
have been satisfactory. Use of low thermal expansion filler in PTFE helps mini-
mize cracking and loss of adhesion from metal substrates with their lower coeffi-
cients of expansion.
Because of the low viscosities of cryogenic liquids, rolling element bearings
seem better suited than hydrodynamic bearings for turbo pumps. The AISI 440C
stainless balls and rings generally are preferred for their corrosion resistance
over the more commonly used AISI 52100 steel.
9.4. Nuclear Radiation Effects. Components of a nuclear reactor sys-
tem that are exposed to radiation and that require lubrication include control-
rod drives, coolant circulating pumps or compressors, motor-operated valves,
Vol. 2 LUBRICATION AND LUBRICANTS 67

17.0
Turbine oil

5.0

1.8

1.6

1.4
Viscosity, mPa·s(=cP)

Mercury
1.2

1.0

0.8

0.6

Rubidium
0.4

0.2 Water NaK 77


Potassium

Air
0
0 100 200 300 400 500 600 700
Temperature, °C

Fig. 9. Viscosity versus temperature.

and fuel handling devices, and, of course, are exposed to varying amounts of
ionizing (224).
Degree of damage suffered by a lubricant depends primarily on the total
radioactive energy absorbed, whether it is from neutron bombardment or from
gamma radiation. The common energy unit for absorbed dosage, the gray (Gy),
is equal to 105 J (100 ergs) absorbed per gram of material, or 0.01 Gy ¼ 1 rad.
The first changes observed with petroleum oils (at 104 Gy dosage) is evolution
of hydrogen and light hydrocarbon gas as fragments from the original molecule.
68 LUBRICATION AND LUBRICANTS Vol. 2

Table 21. Typical Design Limits for Fluid-Film Hydrodynamic Bearings


Item Load on projected area, MPaa
oil lubrication
steady load
electric motors 1.4
steam turbines 2.1
railroad car axles 2.4
dynamic load
automobile engine main bearings 24
automobile connecting-rod bearings 34
steel-mill roll necks 34
water lubrication 0.2
gas bearings 0.02
a
To convert MPa to psi, multiply by 145.

Unsaturation results in decreased oxidation stability, cross-linking, polymeriza-


tion, or scission (225).
The trend is for increasing viscosity with increased dose for several petro-
leum oils (225). For many lubricant applications, a dose that gives a 25%
increase in 408C viscosity can be taken as a tolerance limit. Lower radiation
absorption seldom changes the lubricant sufficiently to interfere with its perfor-
mance. Greater dosage results in more rapid thickening, sludging, and operat-
ing trouble (226).
The general range of tolerance limit of 1–4  105 Gy (1–4  108 rad) for pet-
roleum oils in Table 22 tends to be somewhat higher than for synthetic oils (227).
This is surprising in view of the excellent thermal and oxidative stability of
methyl silicones, diesters, silicates, and some other synthetics. An exception
is the high order of stability with synthetic oils consisting of aromatic hydrocar-
bons in which much of the absorbed energy appears to be transferred into
harmless resonance in the aromatic ring structure. This reduces the degree of

Table 22. Radiation Tolerance Limits of Several Oil Types


Tolerance limit, 106Gya for
Oil 25% increase in 408C viscosity
petroleum 1–4
synthetic
diester MIL-L-6085 1.1
synthetic 2.5–4.5
hydrocarbon
phosphate ester 0.4–0.6
poly(propylene oxide) 1.0
alkylbenzene 5
dimethyl silicone <1
methyl phenyl 1
silicone
tetraaryl silicate 0.6
a
To convert Gy to rad, multiply by 100.
Vol. 2 LUBRICATION AND LUBRICANTS 69

damaging ionization and free-radical formation that occurs on a more general


basis with the chain-like structures in paraffinic oils or in the saturated ring
structure of alicyclic oils. Oil life is reduced by both the radiation dose and by
oxidation if oxygen (air) is present at high temperature.
Conventional greases consisting of petroleum oils thickened with
lithium, sodium, calcium, or other soaps suffer significant breakdown of
the soap gel structure at doses above 105 – 106 Gy (107 – 108 rad). Initial
breakdown commonly involves increased softening of the grease to the
point, where it may become fluid. At even higher doses, polymerization of
the oil phase eventually leads to overall grease hardening. Some greases
with radiation-resistant components, eg, polyphenyl ether oil and nonsoap
thickeners, maintain satisfactory consistency for lubrication purposes up to
107 Gy (109 rad).
9.5. Lubrication with Glass. Softening glass is used as a lubricant for
extrusion, forming, and other hot working processes with steel- and nickel-
base alloys up to 10008C; for extrusion and forming titanium and zirconium
alloys; and less frequently for extruding copper alloys (218). Principal types of
glasses used are pure fused silica [7631-86-9], 96% silica– soda – lime, boro-
silicates, and aluminosilicates [1327-36-2]. The glass composition is selected
for proper viscosity, typically 10– 100 Pa s (1000 – 10,000 P) at the mean tem-
perature of the die and workpiece, to serve as a true hydrodynamic lubricant.
Glass may be applied as fibers or powder to the die or hot workpiece, or as a
slurry with a polymeric bonding agent to the workpiece before heating.
Another method involves rolling heated steel billets across glass sheets,
where the glass then wraps itself around the billet before passing to a die
extrusion chamber.
The U.S. Bureau of Mines has employed glass for forming ceramic materials
at high temperatures (228).
9.6. Ionic Liquids. Ionic liquids are organic salts that remain liquid at
room temperature. They have virtually no vapor pressure and, in applications
like aluminum cold rolling, can significantly reduce air emissions if used in
place of hydrocarbon-based lubricants. Forces holding the molecules together
are Coulombic or electrostatic, as opposed to the weaker van der Waals forces
holding most liquids together. A variety of anions such as Cl, BF4, and
PF6, to name a few, may be used. The properties of two of these liquids, 1-
ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide and 1-butyl-3-
methylimidazolium hexafluorophosphate are described in Ref. 229 (see also
MICROWAVE TECHNOLOGY).
9.7. Solid and Liquid Lubricants for Extreme Environments. A
compilation of papers presented at a Symposium on Lubricants for Extreme
Environments is given in Ref. 230.

10. The Lubricants Market

The National Petrochemical and Refiners Association (NPRA) reports lubri-


cating oil sales of 2.4 billion gallons in the United States in 2002. Grease
sales in 2002, according to NLGI International, were 500 million lb. Worldwide
70 LUBRICATION AND LUBRICANTS Vol. 2

Lubricant demand in 2003, without marine oils, was 36 million metric tons.
Lubricating oil base stock production in North America was 71 million bar-
rels in 2003. These numbers are broken down by region and product type in
Table 23 (231,232).
The sales of automotive oils and greases have been fairly level in the period
1998–2002. Sales of industrial oils have declined in this period. The number of
base stock refineries in the United States and Canada has fallen from 40 in
1989–23 in 2004. United State and Canadian base stock capacity has dropped
from 250,000 barrels/day in 2001 to 217,000 barrels/day in 2004. Production of
groups II and III base stocks has grown over the past 10 years, however, from
4 to 42% (231).

11. Environmental and Health Factors (Toxicology)

Conservation, health, safety, and environmental pollution concerns have led to


the creation of wide-reaching legislation such as the U.S. Congress Energy Policy
and Conservation Act, Toxic Substances Control Act of 1976 (233), Resources
Conservation and Recovery Act of 1976, the Oil Recycling Act of 1980, and
subsequent implementation of many rules and regulations such as the OSHA
Hazard Communication Standard. Continuing publications of new and proposed
rules and regulations are available from the EPA (Washington, D.C.) and from
the National Technical Information Service of the U.S. Department of Commerce
(Springfield, Virginia) (234).
Regulations generally prohibit disposal of lubricants in streams,
chemical dumps, or other environmental channels. Over one-half of disposed
lubricants are burned as fuel, usually mixed with virgin residual, and distillate
fuels (235).
Waste aqueous metalworking fluids may be successfully treated by conven-
tional means for removal of tramp oil, surfactants, and other chemical agents to
provide suitable effluent water quality (236).
11.1. Lubricant Recycling. Considerable effort is underway to improve
and expand recycling of lubricating oils. Although typical processes result in
80–90% yield, questions remain regarding initial collection and the separation
of used oil from water and other contaminants. Recycling treatment varies
from simple cleaning to essentially the complete refining process used with vir-
gin oil. Typical steps involved in purifying used petroleum lubricating oil are
indicated schematically in Fig. 10 (237).
11.2. Reclamation. Reclamation involves simple separation of contami-
nants by gravity settling of water and dirt, centrifuging, filtering, and membrane
techniques. With water-soluble cutting oils containing only a few percent oil,
chemical emulsion breakers are first added that consist of sulfuric acid and
then aluminum sulfate as a coagulant. Polymers are sometimes added to speed
the process. The separated oil then is decanted, skimmed, or centrifuged and
commonly is burned. Generally, 1–5% reprocessed waste oil may be added to
fuel and still meet EPA industrial furnace limits, ie, <5 ppm arsenic by mass,
<2 ppm cadmium, <10 ppm chromium, <100 ppm lead, and <4000 ppm halogens
and a flash point >188C (234).
Vol. 2 LUBRICATION AND LUBRICANTS 71

Table 23. The Lubricants Marketa


Worldwide lubricant demand(without marine oils) 2003 %
Asia–Pacific 11.2 31.2
North America 8.4 23.2
Western Europe 4.8 13.4
Central and Eastern Europe 4.8 13.4
Latin America 3.1 8.7
Near and Middle East 1.8 4.9
Africa 1.9 5.2
total worldwide lubricant demand, millions of metric tons 36.0 100.0

U. S. lubricating oil sales, 2002


multigrade crankcase oils 1,008 42
monograde crankcase oils 96 4
automotive transmission and hydraulic oils 264 11
other automotive oils 48 2
total automotive oils, millions of gallons 1,416 59
general industrial oils 312 13
industrial engine oils 144 6
total industrial lubricating oils, millions of gallons 456 19
process oils 480 20
metalworking oils 48 2
total U. S. lubricating oil sales, millions of gallons 2,400 100

North America lubricating grease production, 2002


conventional lithium grease 180 36
lithium complex grease 160 32
polyurea grease 35 7
aluminum soap grease 45 9
calcium soap grease 30 6
sodium and other metallic soap grease 5 1
organo-clay grease 35 7
other non-soap grease 10 2
total North American grease production, millions of pounds 500 100

U. S. refinery base stock production, 2003


paraffinic base stocks 51.0 84
naphthenic base stocks 9.6 16
total U. S. refinery base stock production, million of barrels 60.6 100

Estimated North American base stock production, 2003


paraffinic group I 30.8 43.4
paraffinic group II 28.0 39.4
paraffinic group III 1.4 2.0
polyalphaolefin group IV 1.0 1.4
naphthenic and other group V 9.8 13.8
total estimated North American base stock production, 71.0 100.0
millions of barrels
a
See Refs. 231,232.
72 LUBRICATION AND LUBRICANTS Vol. 2

Used oil
(automotive or industrial)

Reclaim

Heat: 40–90°C
Gravity settle Solids
Surfactants Water
Screening Sludge
Centrifuge

Reprocess

Reclaim methods plus chemicals


Heat: 65–200°C Solids
Filtration Water
Centrifuge Sludge
Other contaminants

Rerefine

Dehydration
Chemical test Water
Distillation/clay 340°C Sludge
Filtration Bottoms
Spent clay

Fig. 10. Oil recycling flow diagram.

11.3. Reprocessing. The simplest operation involves flash distillation


in an evaporator at 100–2008C in partial vacuum to remove water and low boil-
ing contaminants, eg, gasoline and solvents. This is followed by treatment with
fuller’s earth or other activated clay for removing oxidation products and most
additives to produce a purified, light-colored oil which, with suitable additives,
is satisfactory for use as fuel, metalworking base stocks, noncritical lubricants,
and concrete form oil. Some used oils, eg, hydraulic and transformer oils, can
be reprocessed with a portable unit of capacity to their original oil quality
directly at the equipment in which they are being used.
11.4. Rerefining. The technology currently attracting most attention for
producing original quality lubricating oil depends on distillation in thin-film eva-
porators (TFE) (231). Thin-film evaporator processes usually involve a scheme
similar to that shown at the bottom of Fig. 10. The preliminary removal of
water, solvents, and fuel is done in the same fashion as in most other recycling.
Coking and fouling during distillation is avoided because the maximum tempera-
ture is maintained for only 2–5 s as the oil flows down the TFE wall under the
influence of moving wiper blades; this is a small fraction of residence time in the
packing or plates of a more traditional distillation tower. These variations
involve batch operation with a single unit, or sequential distillation in multiple
units to produce several lube fractions.
Vol. 2 LUBRICATION AND LUBRICANTS 73

Older rerefining units used 2–5 kg/L of activated clay at 40–708C and
higher temperatures in place of TFE to clean the oil (238). More elaborate che-
mical and hydrotreating of used engine oils without a distillation step has been
developed by Phillips Petroleum for processing 40,000 m3/year (10  106 gal/
year). Establishment of a reliable feedstock supply is a critical consideration
for larger rerefining plants.
11.5. Toxic and Hazardous Constituents. Questionable constituents
of lubricating oils are polycyclic aromatics in the base oil plus various additives
(239). Of refining steps used in preparing lubricating oil base stocks from toxic
distillates, only effective solvent extraction, severe hydrogenation, or exhaus-
tive fuming sulfuric acid treatment appear adequate to eliminate carcinogeni-
city. Mild hydrofinishing or light solvent extraction reduces carcinogenic
potential, but does not necessarily eliminate it. Increasing the severity of
hydrofinishing can eliminate carcinogenic compounds. Group IV base stocks,
polyalphaolefins, are not expected to be carcinogenic since no polycyclic aro-
matics are present.
Most additives for lubricants present little risk, but the following involve
significant hazards: lead compounds, phenyl 2-naphthylamine, sodium nitrite
plus amines, tricresylphosphate high in the o-cresol isomer, and chlorinated
naphthalenes. A number of sulfur compounds used as additives cause skin irri-
tation; however, properly refined base oils containing usual concentrations of
these additives have a low degree of toxicity. Used motor oil has displayed
increased carcinogenic activity over its new counterpart (234,239). Users should
also avoid contact with lubricants, metalworking oils, and quench oils that are
highly degraded, were in service at extremely high temperatures, or are contami-
nated with toxic metals or bacteria.
The latest government regulations set forth under the Toxic Substan-
ces Control Act and in Public Health Service publications should be
checked before formulating new lubricants. Users of lubricants should request
Material Safety Data Sheets for each substance involved plus certification of
compliance from vendors. Lubricant compounders should insist on similar
information from their suppliers for any additive packages. Manufacturers of
both additives and lubricants commonly make toxicity checks on commercial
products.
11.6. Food Processing. Prior to September, 1998, the U.S. Depart-
ment of Agriculture (USDA), Food Safety Inspection Service (FSIS),
working with the Food and Drug Administration (FDA) maintained a system
to assure minimal risk to the consumer from incidental or unintended
contact with lubricants used in the food and beverage industries (see FOOD
PROCESSING) (240).
Upon satisfactory determination of nontoxicity, the USDA issued one of
three ratings (241): H1 Lubricants: Food-grade lubricants used in the food pro-
cessing environment, where there is a possibility of incidental food contact, eg, as
by splashing or dripping from machinery above an edible product. H2 Lubri-
cants: Nonfood-grade lubricants used on equipment and machine parts in loca-
tions where there is no possibility of food contact, eg, as in sealed gear boxes
or machinery below a product line. H3 Lubricants: Food-grade lubricants,
74 LUBRICATION AND LUBRICANTS Vol. 2

typically edible oils, used to prevent rust on hooks, trolleys and similar equip-
ment. These classes include a number of petroleum and nonpetroleum oils and
greases.
In September, 1998, the USDA issued the Hazard Analysis and Critical
Control Point (HACCP) protocol and eliminated the approval program. This
action shifted the burden of assessing the risk and approval of lubricants to
the food processor or manufacturer. Under the HACCP system, food processing
facilities must show the FSIS inspectors ‘‘documentation substantiating the
safety of a chemical’s (lubricant’s) use in a food processing environment’’
(240,241).
In order to minimize disclosure of proprietary formulation details, lubricant
manufacturers, with the approval of food processors, equipment builders and the
USDA, have developed a system whereby third party certifiers assume the role
previously handled by USDA. This includes preauthorization of nonfood com-
pounds, screening of product formulations, product registration, and the issu-
ance of approval documents (242,243).
The rating system used by USDA for lubricants will continue to be used.
Food-grade lubricants rated H1 or H3 for incidental contact must be generally
regarded as safe (GRAS); used in accordance with the provisions of a prior sanc-
tion or approval; or listed under 21 CFR 178.3570. Lubricants rated H2 for no
allowable contact must not contain poisons, toxins, or other materials that can
cause a health risk.

BIBLIOGRAPHY

‘‘Lubrication and Lubricants’’ in ECT 1st ed., Vol. 8, pp. 495–540, by R. G. Larsen and A.
Bondi, Shell Development center; in ECT 2nd ed., Vol. 12, pp. 557–616, by R. E. Lee, Jr.,
and E. R. Booser, General Electric Co.; in ECT 3rd ed., Vol. 14, pp. 477–526, by E. R.
Booser, General Electric Co.; in ECT 4th ed., Vol. 15, pp. 463–517, by E. R. Booser,
Consultant; ‘‘Lubrication and Lubricants’’ in ECT (online), posting date: December 4,
2000, by E. R. Booser, Consultant.

CITED PUBLICATIONS

1. ASTM G 40-02 Standard Terminology Relating to Wear and Erosion, in Annual


Book of ASTM Standards, Vol. 3.02, ASTM International, Issued Annually.
2. D. Dowson, History of Tribology, 2nd ed., Professional Engineering Publishing,
London and Bury St. Edmunds, 1998.
3. British Lubrication Engineering Working Group, H. P. Jost, Chairman, Lubrication
(Tribology) Education and Research—A Report on the Present Position and Indus-
tries Needs, Department of Education and Science, U.K., 1966.
4. V. V. Dunaevsky, Y. R. Jeng, and J. A. Rudzitis, in E. R. Booser, ed., Tribology Data
Handbook, CRC Press, Boca Raton, Fla., 1997, 415–434.
5. J. A. Broadston, in E. A. Avallone, and T. B. Baumeister III, eds., Marks’ Standard
Handbook for Mechanical Engineers, 9th ed., Section 13.5, 13.75–13.81, McGraw
Hill, New York, 1987.
Vol. 2 LUBRICATION AND LUBRICANTS 75

6. E. Oberg, F. D. Jones, H. L. Horton, and H. H. Riffel, R. E. Green, ed., Machinery’s


Handbook, 24th ed., Industrial Press Inc., New York, 1992, pp. 667–682
7. K. Carnes, Tribol. Lub. Technol. 60(11), 31 (2004).
8. ASTM D 6425-02 Standard Test Method for Measuring Friction and Wear Pro-
perties of Extreme Pressure (EP) Lubricating Oils Using SRV Test Machine, in
Annual Book of ASTM Standards, Section Five, ASTM International, Issued
Annually.
9. M. Woydt, in G. E. Totten, L. D. Wedeven, J. R. Dickey, and M. Anderson, eds.,
Bench Testing of Industrial Fluid Lubrication and Wear Properties Used in
Machinery Applications, STP 1404, ASTM, Conshohocken, Pa., 2001, pp. 199–209.
10. Exxon Booklet DG-5A Principles of Lubrication, Exxon Company, Houston, Tex.,
1992.
11. D. H. Buckley, in E. R. Booser, ed., Handbook of Lubrication: Theory and Practice of
Tribolgy, Vol. II, CRC Press, Boca Raton, Fla., 1984, pp. 17–30.
12. R. S. Fein, in E. R. Booser, ed., Handbook of Lubrication: Theory and Practice of
Tribology, Vol. II, CRC Press, Boca Raton, Fla., 1984, pp. 49–68
13. K. C. Ludema, Friction, Wear and Lubrication: A Textbook in Tribology, CRC Press,
Boca Raton, Fla., 1997, p. 74.
14. P. B. Abel and J. Ferrante, in B. Bhushan, ed., Modern Tribology Handbook, Vol. 1,
CRC Press, LLC, Boca Raton, Fla., 2001, pp. 5–47.
15. B. Bhushan, Principles and Applications of Tribology, CRC Press, LLC, Boca Raton,
Fla., 1999.
16. D. D. Fuller, in E. A. Avallone, and T. B. Baumeister III, eds., Marks’ Standard Hand-
book for Mechanical Engineers, 9th ed., McGraw-Hill, New York, 1987, pp. 3-24–3-33.
17. A. W. Ruff, in B. Bhushan, ed., Modern Tribology Handbook, Vol. 1, CRC Press, LLC,
Boca Raton, Fla., 2001, pp. 523–561.
18. E. R. Booser, ed., Tribology Data Handbook, CRC Press, Boca Raton, Fla., 1997,
pp. 435–444.
19. ASM International and ASTM, Friction and Wear Testing: Sourcebook of Selected
References from ASTM Standards and ASM Handbooks, 1997, pp. 163–167.
20. K. C. Ludema, in E. R. Booser, ed., Handbook of Lubrication: Theory and Practice of
Tribology, Vol. II, CRC Press, Boca Raton, Fla., 1984, pp. 31–48.
21. E. Rabinowicz, in E. R. Booser, ed., Handbook of Lubrication: Theory and Practice of
Tribology, Vol. II, CRC Press, Boca Raton, Fla., 1984, pp. 201–208.
22. V. V. Dunaevsky, in E. R. Booser, ed., Tribology Data Handbook, CRC Press, Boca
Raton, Fla., 1997, pp. 455–461.
23. E. Rabinowicz, Friction and Wear of Materials, John Wiley & Sons, Inc., New York,
1965.
24. J. F. Archard and W. Hirst, Proc. R. Soc. London, Ser. A 236, 397 (1956).
25. K. C. Ludema, in B. Bhushan, ed., Modern Tribology Handbook, Vol. 1, CRC Press,
LLC, Boca Raton, Fla., pp. 205–233.
26. K. Kato and K. Adachi, in B. Bhushan, ed., Modern Tribology Handbook, Vol. 1, CRC
Press, LLC, Boca Raton, Fla., pp. 273–300.
27. I. Newton, Philos. Nat. Princ. Math. 4, 268 (1687).
28. J. W. Murdock, in E. A. Avallone, and T. B. Baumeister III, eds., Marks’ Standard
Handbook for Mechanical Engineers, 9th ed., McGraw-Hill, New York, 1987,
pp. 3-34–3-71.
29. V. L. Streeter, Fluid Mechanics, McGraw-Hill, New York, 1958, pp. 3–15, 141.
30. C. K. Schoff and P. Kamarchik, Jr., ‘‘Rheologival Measurements’’, Kirk-Othmer
Encyclopedia of Chemical Technology, 4th ed., John Wiley & Sons, Inc.,
New York, 1997.
76 LUBRICATION AND LUBRICANTS Vol. 2

31. ASTM D 2983 Standard Test Method for Low-Temperature Viscosity of Automotive
Fluid Lubricants Measured by Brookfield Viscometer, in Annual Book of ASTM
Standards, Section Five, ASTM International, Issued Annually.
32. ASTM D 5293 Standard Test Method for Apparent Viscosity of Engine Oils between
5 and 308C using the Cold-Cranking Simulator, in Annual Book of ASTM Stan-
dards, Section Five, ASTM International, Issued Annually.
33. ASTM D 4684 Standard Test Method for Determination of Yield Stress and Appar-
ent Viscosity of Engine Oils at Low Temperature, in Annual Book of ASTM Stan-
dards, Section Five, ASTM International, Issued Annually.
34. ASTM D 4683 Standard Test Method for Measuring Viscosity at High Shear Rate
and High Temperature by Tapered Bearing Simulator, in Annual Book of ASTM
Standards, Section Five, ASTM International, Issued Annually.
35. ASTM D 445 Standard Test Method for Kinematic Viscosity of Transparent and
Opaque Liquids (the Calculation of Dynamic Viscosity), in Annual Book of ASTM
Standards, Section Five, ASTM International, Issued Annually.
36. ASTM D 2161-04, Standard Practice for Conversion of Kinematic Viscosity to
Saybolt Universal Viscosity or Saybolt Furol Viscosity, in Annual Book of ASTM
Standards, Section Five, ASTM International, Issued Annually.
37. ASTM D 341 Standard Viscosity-Temperature Charts for Liquid Petroleum
Products, in Annual Book of ASTM Standards, Section Five, ASTM International,
Issued Annually.
38. E. E. Klaus and E. J. Tewksbury, in E. R. Booser, ed., Handbook of Lubrication:
Theory and Practice of Tribology, Vol. II, CRC Press, Boca Raton, Fla., 1984,
pp. 229–254.
39. ASTM D 2270, in E. R. Booser, ed., Handbook of Lubrication: Theory and Practice of
Tribology, Vol. II, CRC Press, Boca Raton, Fla., 1984.
40. ASME Pressure-Viscosity Report, American Society of Mechanical Engineers,
New York, 1953.
41. D. Klamann, Lubricants and Related Products, Verlag Chemie, Weinheim,
Germany, 1984, p. 19.
42. R. S. Fein, ASM Handbook, Vol. 18, ASM International, Metals Park, Ohio, 1992,
pp. 81–88.
43. D. D. Fuller, in E. A. Avallone, and T. B. Baumeister III, eds., Marks’ Standard
Handbook for Mechanical Engineers, 9th ed., McGraw-Hill, New York, 1987,
pp. 8-114–8-131.
44. M. Rhode, Fluid Film Lubrication: A Century of Progress, The American Society of
Mechanical Engineers, 1983 (Preface).
45. D. Dowson, in M. Rhode, ed., Fluid Film Lubrication: A Century of Progress, The
American Society of Mechanical Engineers, 1983, pp. 1–51.
46. O. Reynolds, in M. Rhode, Fluid Film Lubrication: A Century of Progress, The
American Society of Mechanical Engineers, 1983, pp. 135–217.
47. N. Petrov, in M. Rhode, Fluid Film Lubrication: A Century of Progress, The
American Society of Mechanical Engineers, 1983, pp. 107–133.
48. B. Tower, in M. Rhode, Fluid Film Lubrication: A Century of Progress, The American
Society of Mechanical Engineers, 1983, pp. 63–93.
49. V. L. Streeter, Fluid Mechanics, McGraw-Hill, New York, 1958, pp. 194–197.
50. A. Harnoy, Bearing Design in Machinery: Engineering Tribology and Lubrication,
Marcel Dekker, New York, 2003.
51. J. Boyd and A. A. Raimondi, in J. J. O’Connor, J. Boyd, and E. A. Avallone, eds.,
Standard Handbook of Lubrication Engineering, McGraw-Hill, New York, 1968,
Chapt. 3, pp. 3-1–3-10.
Vol. 2 LUBRICATION AND LUBRICANTS 77

52. U. J. Möller and U. Boor, Lubricants in Operation, (Translated by R. F. Ferguson)


A. R. Landsdown, ed., VDI Verlag and Mechanical Engineering Publications Ltd.,
London, 1986.
53. M. M. Khonsari and D. Y. Hua, in E. R. Booser, ed., Tribology Data Handbook, CRC
Press, Boca Raton, Fla., 1997, pp. 611–637.
54. D. Klamann, Lubricants and Related Products, Verlag Chemie, Weinheim,
Germany, 1984, Deerfield Beach, Fla., Basel, 1984, 35.
55. K. C. Ludema, Friction, Wear, Lubrication: A Textbook in Tribology, CRC Press,
Boca Raton, Fla., 1996, Chapt. 7, pp. 111–128.
56. T. Jendzurski and C. A. Moyer, in E. R. Booser, ed., Tribology Data Handbook, CRC
Press, Boca Raton, Fla., 1997, pp. 645–668.
57. N. Cantor, Tribol. Lub. Technol. 60(6), 42 (2004).
58. J. Krim, Lub. Eng. 53(1), 8 (1997).
59. B. Bhushan, Principles and Applications of Tribology, John Wiley & Sons, Inc.,
New York, 1999.
60. A. D. Berman and N. Iraelachvili, in B. Bhushan, ed., Modern Tribology Handbook,
Vol. 1, CRC Press, LLC, Boca Raton, Fla., Section II (2001).
61. D. Godfrey, in J. J. O’Connor, J. Boyd, and E. A. Avallone, eds., Standard Handbook
of Lubrication Engineering, McGraw-Hill, New York, 1968, pp. 2-1–2-24.
62. Appendix E, API Base Oil Interchangeability Guidelines for Passenger Car Motor
Oils and Diesel Engine Oils, November 2003 version, in API 1509, API Engine Oil
Licensing and Certification System, 15th ed., April 2002.
63. Cross Oil Refining & Marketing, Base Oils, www.crossoil.com/baseoils 2004.
64. Calumet Product Data Sheets, Calumet Lubricants Co., Indianapolis, 2004.
65. American Refining Group, Inc., Kendex Industrial Products, www.amref.com
Bradford Pa., 2004.
66. Valero Marketing and Supply Co., Valero Lubricant Base Oils, www.valero.com
Paulsboro, N.J., 2004.
67. Motiva STAR and TEXHVI Base Oils, www.motivalubes.com 2004.
68. Chevron Texaco Corp., Chevron Texaco Base Oils, www.chevron.com 2003.
69. CP Chem, Synfluid PAOs, www.chevronphillipschemicalcompany.net 2004.
70. ExxonMobil Chemical, Synthetic Lubricant Base Stocks, www.exxonmobilsynthe-
tics.com 2004.
71. R. F. Haycock and J. E. Hillier, Automotive Lubricants Reference Book, 2nd ed., SAE
International, Warrendale, Pa., 2004, pp. 45–63.
72. T. Hilbert, S. Tabak, I. Cody, and S. Hantzer, Hydrocarbon Asia 50 (May/June
2004).
73. ExxonMobil Corp., Lube Technologies, www.exxonmobil.com/refiningtechnologies/
lubes 2004.
74. G. E. Kellis, in J. J. O’Connor, J. Boyd, and E. A. Avallone, eds., Standard
Handbook of Lubrication Engineering, McGraw-Hill, New York, 1968, pp. 13-
2 –13-19.
75. Chevron Lummus Global, Isodewaxing www.chevron.com 2003.
76. R. L. Shubkin, in E. R. Booser, ed., Tribology Data Handbook, CRC Press, Boca
Raton, Fla., 1997, pp. 37–41.
77. Chevron Phillips Chemical Company, Polyalphaolefins Overview, www.chevronphil-
lipschemicalcompany.net, April 20, 2004.
78. T. Sullivan, Europeans Adopt 6th Base Oil Group, from Lube Report, www.lubere-
port.com Sep. 23, 2004.
79. L. R. Rudnick and R. L. Shubkin, Synthetic Lubricants and High Performance
Functional Fluids, 2nd ed., Marcel Dekker, New York, 1999.
78 LUBRICATION AND LUBRICANTS Vol. 2

80. C. Mangone, Independent Lubricant Manufacturer’s Association, Gas to Liquids—


Conversions Produce Extremely Pure Base Oils, Machinery Lubrication Magazine,
November 2002.
81. Hatco Corporation, Esters in Synthetic Lubricants, www.hatcocorporation.com 2004.
82. J. M. Perez and E. E. Klaus, in E. R. Booser, ed., Handbook of Lubrication: Theory
and Practice of Tribology, Vol. III, CRC Press, Boca Raton, Fla., 1994, pp. 237–252.
83. W. L. Brown, in E. R. Booser, ed., Handbook of Lubrication: Theory and Practice of
Tribology, Vol. III, CRC Press, Boca Raton, Fla., 1994, pp. 253–267.
84. S. S. Lawate, K. Lal, and C. Huang, in E. R. Booser, ed., Tribology Data Handbook,
CRC Press, Boca Raton, Fla., 1997, pp. 103–116.
85. W. J. Bartz, in E. R. Booser, ed., Tribology Data Handbook, CRC Press, Boca Raton,
Fla., 1997, p. 36.
86. J. D. Fotheringham, in E. R. Booser, ed., Tribology Data Handbook, CRC Press, Boca
Raton, Fla., 1997, pp. 94–102.
87. G. H. Dudley, P. A. Brown, and S. Ho, Lubricant Performance with Alkylated
Naphthalene Blendstocks, presented at the STLE Annual Meeting, Toronto, May,
2004.
88. L. R. Rudnick and R. L. Shubkin, Synthetic Lubricants and High Performance Func-
tional Fluids, 2nd ed., Marcel Dekker, New York, 1999.
89. E. D. Brown, in E. R. Booser, ed., Tribology Data Handbook, CRC Press, Boca Raton,
Fla., 1997, pp. 75–79.
90. D. G. Placek and M. P. Marino, in E. R. Booser, ed., Tribology Data Handbook, CRC
Press, Boca Raton, Fla., 1997, pp. 60–74.
91. L. L. Ferstandig, in E. R. Booser, ed., Tribology Data Handbook, CRC Press, Boca
Raton, Fla., 1997, pp. 88–93.
92. L. J. Gschwender, D. C. Kramer, B. K. Lok, S. K. Sharma, C. E. Snyder, and M. L.
Sztenderowicz, ‘‘Liquid Lubricants and Lubrication’’ in B. Bhushan, ed., Modern
Tribology Handbook, Vol. 1, CRC Press LLC, Boca Raton, Fla., 2001, pp. 361–382.
93. G. A. Bell, ‘‘Perfluoroalkylpolyethers,’’ in E. R. Booser, ed., Tribology Data Hand-
book, CRC Press, Boca Raton, Fla., 1997, pp. 80–87.
94. California Integrated Waste Management Board, The Facts About Rerefined Oil,
www.ciwmb.ca.gov, Dec. 3, 2003.
95. D. Klamann, Lubricants and Related Products, Verlag Chemie, Weinheim,
Germany, 1984, pp. 177–184.
96. Exxon Corporation, Encyclopedia For The User of Petroleum Products, Exxon
Lubetext DG-400, 1993.
97. J. L. Reyes-Gavilan and P. Odorisio, NLGI Spokesman 64(11), 22 (2001).
98. C. A. Migdal, in L. R. Rudnick, Lubricant Additives: Chemistry and Applications,
Marcel Dekker, New York, 2003, 1–27.
99. M. Rasberger, in R. M. Mortier, and S. T. Orzulik, eds., Chemistry and Technology of
Lubricants, 2nd ed., Blackie Academic and Professional, London, New York, 1997.
100. W. van Leeuwen, Lubrication Eng. 56(10), 30 (2000).
101. E. R. Booser, ed., Tribology Data Handbook, CRC Press, Boca Raton, Fla., 1997,
Section X., pp. 875–1028.
102. King Industries, NA-LUBE AO Antioxidants, www.kingindustries.com 2004.
103. R. T. Vanderbilt Company, Inc. Vanderbilt Lubricant Additives, Norwalk, CT,
2004.
104. M. Hunter, E. E. Klaus, and J. L. Duda, Lubrication Eng. 49(6), 492 (1993).
105. D. Klamann, Lubricants and Related Products, Verlag Chemie, Weinheim,
Germany, 1984, pp. 212–215.
106. S. Q. A. Rizvi, in L. R. Rudnick, ed., Lubricant Additives: Chemistry and Applica-
tions, Marcel Dekker, New York, 2003, pp. 113–170.
Vol. 2 LUBRICATION AND LUBRICANTS 79

107. R. F. Haycock and J. E. Hillier, Automotive Lubricants Reference Book, 2nd ed., SAE
International, Warrendale, Pa., 2004, pp. 69–77.
108. R. F. Haycock and J. E. Hillier, Automotive Lubricants Reference Book, 2nd ed., SAE
International, Warrendale, Pa., 2004, pp. 77–80.
109. R. A. McDonald, in L. R. Rudnick, Lubricant Additives: Chemistry and Applications,
Marcel Dekker, New York, 2003, pp. 29–43.
110. K. Nice, How Catalytic Converters Work, http://auto.howstuffworks.com.
111. W. D. Phillips, ‘‘Ashless Phosphorus-Containing Lubricating Oil Additives’’ in L. R.
Rudnick, ed., Lubricant Additives: Chemistry and Applications, Marcel Dekker,
New York, 2003, pp. 45–111.
112. L. O. Farng, ‘‘Ashless Antiwear and Extreme-Pressure Additives’’ in L. R. Rudnick,
ed., Lubricant Additives: Chemistry and Applications, Marcel Dekker, New York,
2003, pp. 223–257.
113. H. P. Bloch, Practical Lubrication for Industrial Facilities, The Fairmont Press,
Lilburn, Ga. and Marcel Dekker, New York, London, 2000, p. 20.
114. H. E. Dean, Handbook of Lubrication: Theory and Practice of Tribology, Vol. I, CRC
Press, Boca Raton, Fla., 1983, pp. 45–53.
115. R. F. Watts, 10th International Colloquium on Tribology, Technical Academy Esslin-
gen, Ostfilern, Germany, 1996.
116. D. Kenbeek and T. F. Buenemann, in L. R. Rudnick, Lubricant Additives: Chemistry
and Applications, Marcel Dekker, New York, 2003, pp. 203–221.
117. B. G. Kinker, in L. R. Rudnick, Lubricant Additives: Chemistry and Applications,
Marcel Dekker, New York, 2003, pp. 329–353.
118. R. F. Haycock and J. E. Hillier, Automotive Lubricants Reference Book, 2nd ed., SAE
International, Warrendale, Pa., 2004, pp. 63–64.
119. D. Klamann, Lubricants and Related Products, Verlag Chemie, Weinheim,
Germany, 1984, pp. 193–195.
120. R. F. Haycock and J. E. Hillier, Automotive Lubricants Reference Book, 2nd ed., SAE
International, Warrendale, Pa., 2004, pp. 65–68.
121. L. I. Osipow, Surface Chemistry, Rheinhold Publishing Corporation, New York,
1962, pp. 344–376.
122. D. Klamann, Lubricants and Related Products, Verlag Chemie, Weinheim,
Germany, 1984, pp. 210–211.
123. E. A. Bardasz and G. D. Lamb, in L. R. Rudnick, Lubricant Additives: Chemistry and
Applications, Marcel Dekker, New York, 2003, pp. 387–428.
124. F. L. Lee, and J. J. W. Harris, L. R. Rudnick, Lubricant Additives: Chemistry and
Applications, Marcel Dekker, New York, 2003, pp. 557–586.
125. D. Klamann, Lubricants and Related Products, Verlag Chemie, Weinheim,
Germany, 1984, p. 212.
126. D. Klamann, Lubricants and Related Products, Verlag Chemie, Weinheim,
Germany, 1984, pp. 215–216.
127. API 1509, API Engine Oil Licensing and Certification System, 15th ed., April 2002.
128. American Petroleum Institute, Engine Oil Licensing and Certification System,
http://api-ep.api.org.
129. SAE J300 Engine Oil Viscosity Classification, May 2004, in SAE Handbook, Society
of Automotive Engineers, Warrendale, Pa., Issued Annually.
130. International Lubricant Standardization and Approval Committee, ILSAC GF-4
Standard for Passenger Car Engine Oils, January 14, 2004.
131. International Lubricant Standardization and Approval Committee, ILSAC GF-3
Standard for Passenger Car Engine Oils, July, 2001.
132. Southwest Research Institute, Abstracts of Test Procedures Performed By Fuels and
Lubricants Research Division, www.swri.org.
80 LUBRICATION AND LUBRICANTS Vol. 2

133. ASTM D6593-04 Standard Test Method for Evaluation of Automotive Engine Oils
for Inhibition of Deposit Formation in a Spark-Ignition Internal Combustion Engine
Fueled with Gasoline and Operated Under Low-Temperature, Light-Duty Condi-
tions, in Annual Book of ASTM Standards, Section 5, ASTM International, West
Conshohocken, Pa., Issued Annually.
134. ASTM D6891-03 Standard Test Method for Evaluation of Automotive Engine Oils in
the Sequence IVA Spark-Ignition Engine, in Annual Book of ASTM Standards,
Section 5, ASTM International, West Conshohocken, Pa., Issued Annually.
135. ASTM D6709-03a Standard Test Method for Evaluation of Automotive Engine Oils
in the Sequence VIII Spark-Ignition Engine (CLR Oil Test Engine), in Annual Book
of ASTM Standards, Section 5, ASTM International, West Conshohocken, Pa.,
Issued Annually.
136. ASTM D6837-04 Standard Test Method for Measurement of the Effects of Automotive
Engine Oils on the Fuel Economy of Passenger Cars and Light-Duty Trucks in the
Sequence VIB Spark Ignition Engine, in Annual Book of ASTM Standards, Section 5,
ASTM International, West Conshohocken, Pa., Issued Annually.
137. ASTM D4951-02 Standard Test Method for Determination of Additive Elements in
Lubricating Oils by Inductively Coupled Plasma Atomic Emission Spectrometry, in
Annual Book of ASTM Standards, Section 5, ASTM International, West Consho-
hocken, Pa., Issued Annually.
138. ASTM D2622-03 Standard Test Method for Sulfur in Petroleum Products by
Wavelength Dispersive X-ray Fluorescence Spectrometry, in Annual Book of
ASTM Standards, Section 5, ASTM International, West Conshohocken, Pa., Issued
Annually.
139. ASTM D5800-04 Standard Test Method for Evaporation Loss of Lubricating Oils by
the Noack Method, in Annual Book of ASTM Standards, Section 5, ASTM Interna-
tional, West Conshohocken, Pa., Issued Annually.
140. ASTM D6417-03 Standard Test Method for Estimation of Engine Oil Volatility by
Capillary Gas Chromatography, in Annual Book of ASTM Standards, Section 5,
ASTM International, West Conshohocken, Pa., Issued Annually.
141. ASTM D6335-03b Standard Test Method for Determination of High Temperature
Deposits by Thermo-Oxidation Engine Oil Simulation Test, in Annual Book of
ASTM Standards, Section 5, ASTM International, West Conshohocken Pa., Issued
Annually.
142. ASTM D6794-02 Standard Test Method for Measuring the Effect on Filterability of
Engine Oils After Treatment with Various Amounts of Water and a Long (6-h) Heat-
ing Time, in Annual Book of ASTM Standards, Section 5, ASTM International, West
Conshohocken, Pa., Issued Annually.
143. ASTM D892-03 Standard Test Method for Foaming Characteristics of Lubricating
Oils, in Annual Book of ASTM Standards, Section 5, ASTM International, West Con-
shohocken, Pa., Issued Annually.
144. ASTM D6557-04 Standard Test Method for Evaluation of Rust Preventive Charac-
teristics of Automotive Engine Oils, in Annual Book of ASTM Standards, Section 5,
ASTM International, West Conshohocken, Pa., Issued Annually.
145. ASTM D92-02b Standard Test Method for Flash and Fire Points by Cleveland Open
Cup Tester, in Annual Book of ASTM Standards, Section 5, ASTM International,
West Conshohocken, Pa., Issued Annually.
146. CRC Deposit Rating Manual, Manual 20, Cordinating Research Council, Alpharetta,
Ga., 2002.
147. CRC Distress Rating Manual, Manual 21, Cordinating Research Council, Alpharetta,
Ga., 2002.
148. D. McFall, Lubes-n-Greases 10, 2 (2004).
Vol. 2 LUBRICATION AND LUBRICANTS 81

149. J. A. McGeehan, Diesel Engines Have a Future and That Future is Clean, SAE
Paper 2004-01-1956, Engine Lubricants, Effects of Fuels & Lubricants on Automo-
tive Devices, and Lubricant Applications & New Test Methods, SAE SP-1885,
2004.
150. R. F. Haycock and J. E. Hillier, Automotive Lubricants Reference Book, 2nd ed., SAE
International, Warrendale, Pa., 2004, Chapt. 6, 219–262.
151. ASTM D4485-04, in Annual Book of ASTM Standards, Section 5, ASTM Interna-
tional, West Conshohocken, Pa., Issued Annually.
152. D. McFall, Lubes-n-Greases 10(7), 6 (2004).
153. D. McFall, Lubes-n-Greases, 10(8), 31 (2004).
154. MIL-PRF-7808L Lubricating Oil, Aircraft Turbine Engine, Synthetic Base, NATO
Code Number O-148.
155. MIL-PRF-23699F Lubricating Oil, Aircraft Turbine Engine, Synthetic Base, NATO
Code Number O-156, 1997.
156. A. G. Papay, Lub. Eng. 46(8), 511 (1989).
157. Trends 2003, Infineum USA, L.P.
158. P. A. Willermet, Lub. Eng. 55(2), 39 (1999).
159. T. Matsuoko, et al., Development of Toroidal Traction Drive CVTF for Automobile,
SAE Paper 2002-01-1696, International Spring Fuels & Lubricants Meeting & Exhi-
bition, Reno, Nev., May 6–9, 2002.
160. Lubrizol Ready Reference for Lubricant and Fuel Performance Driveline Lubricants
– Automotive Gear Lubricants, www.lubrizol.com 2005.
161. R. F. Haycock and J. E. Hillier, Automotive Lubricants Reference Book, 2nd ed., SAE
International, Warrendale, Pa., 2004, Chapt. 7, pp. 263–277.
162. Southwest Research Institute Gear oil Testing, www.swri.org May 2004.
163. ASTM D130-04 Standard Test Method for Corrosiveness to Copper from Petroleum
Products by Copper Strip Test, in Annual Book of ASTM Standards, Section 5,
ASTM International, West Conshohocken, Pa., Issued Annually.
164. ASTM D892-03 Standard Test Method for Foaming Characteristics of Lubricating
Oils, in Annual Book of ASTM Standards, Section 5, ASTM International, West
Conshohocken, Pa., Issued Annually.
165. SAE J306 Automotive Gear Lubricant Viscosity Classification Jul 1998 in SAE
Handbook: Society of Automotive Engineers, Warrendale, Pa., Issued Annually.
166. ASTM D 5662-99 Standard Test Method for Determining Automotive Gear Oil Com-
patibility with Typical Oil Seal Elastomers, in Annual Book of ASTM Standards,
Section 5, ASTM International, West Conshohocken, Pa., Issued Annually.
167. ASTM D5579-04 Standard Test Method for Evaluating the Thermal Stability of
Manual Transmission Lubricants in a Cyclic Durability test, in Annual Book of
ASTM Standards, Section 5, ASTM International, West Conshohocken, Pa., Issued
Annually.
168. ASTM D 4498-95(2003)e1 Standard Test Method for Evaluating Wear Characteris-
tics of Tractor Hydraulic Fluids, in Annual Book of ASTM Standards, Section 5,
ASTM International, West Conshohocken, Pa., Issued Annually.
169. ASTM D 2422-97 (2002) Standard Classification of Industrial Fluid Lubricants by
Viscosity System, in Annual Book of ASTM Standards, Section 5, ASTM Interna-
tional, West Conshohocken, Pa., Issued Annually.
170. ISO Standard 3448 Industrial Liquid Lubricants - ISO Viscosity Classification (1992).
171. ASTM D 6080 Standard Practice for Defining the Viscosity Characteristics of
Hydraulic Fluids, in Annual Book of ASTM Standards, Section 5, ASTM Interna-
tional, West Conshohocken, Pa., Issued Annually.
172. ANSI/AGMA 9005–E02, Industrial Gear Lubrication, American Gear Manufac-
turers Association, Alexandria, Va.
82 LUBRICATION AND LUBRICANTS Vol. 2

173. Exxon Application Data Sheet DG-5D, Lubrication of Plain Bearings, Exxon
Company, Houston, Tex., 1997.
174. Exxon Application Data Sheet DG-5G, Lubrication of Enclosed Gears, Exxon
Company, Houston, Tex., 1990.
175. Exxon Application Data Sheet DG-5D, Lubrication of Rolling-Contact Bearings,
Exxon Company, Houston, Tex., 1999.
176. H. P. Bloch, Practical Lubrication for Industrial Facilities, The Fairmont Press,
Lilburn, Ga. and Marcel Dekker, New York, London, 2000, pp. 281–394.
177. I. Macpherson and D. McCoy, Hydraulic and Turbine Oil Specifications, in E. R. Booser,
ed., Tribology Data Handbook, CRC Press, Boca Raton, Fla., 1997, pp. 274–281.
178. V. Bajpai, M. Fletschinger, N. Hoeck, and P. Rohrbach, Additive Technology for
Formulating Modern Turbine Oils, Ciba Specialty Chemicals, Tarrytown, N.Y.,
2004.
179. ASTM D 943-04 Standard Test Method for Oxidation Characteristics of Inhibited
Mineral Oils, in Annual Book of ASTM Standards, Section 5, ASTM International,
West Conshohocken, Pa., Issued Annually.
180. ASTM D 2272-02 Standard Test Method for Oxidation Stability of Steam Turbine
Oils by Rotating Pressure Vessel, in Annual Book of ASTM Standards, Section 5,
ASTM International, West Conshohocken, Pa., Issued Annually.
181. ASTM D 5182-97(2002)e1 Standard Test Method for Evaluating the Scuffing Load
Capacity of Oils (FZG Visual Method), in Annual Book of ASTM Standards, Section
5, ASTM International, West Conshohocken, Pa., Issued Annually.
182. G. J. Folz and H. J. Noble, Metal Removal: Fluid Selection and Application Guide,
in E. R. Booser, ed., Tribology Data Handbook, CRC Press, Boca Raton, Fla., 1997,
pp. 831–839.
183. J. D. Silliman and R. Perich, eds., Cutting and Grinding Fluids: Selection and
Application, 2nd ed., Society of Manufacturing Engineers, Dearborn Mich., 1992.
184. T. G. McClure, in E. R. Booser, ed., Tribology Data Handbook, CRC Press, Boca
Raton, Fla., 1997, pp. 840–853.
185. N. M. Canter, in E. R. Booser, ed., Tribology Data Handbook, CRC Press, Boca
Raton, Fla., 1997, pp. 862–871.
186. A. C. Eachus, in E. R. Booser, ed., Tribology Data Handbook, CRC Press, Boca
Raton, Fla., 1997, pp. 854–861.
187. F. J. Passman, Controlling Microbial Contamination in Metalworking Fluids, SME
Paper MF92-127, Society of Manufacturing Engineers, Dearborn, Mich., www.biode-
terioration-control.com 1992.
188. T. J. Drozda and C. Wick, eds., Tool and Manufacturing Engineers Handbook,
Vol. 1 Machining, Society of Manufacturing Engineers, Dearborn, Mich., 1983,
pp. 4-1–4-60.
189. L. B. Sargent Jr., in Handbook of Lubrication: Theory and Practice of Tribology,
Vol. 1, CRC Press, Boca Raton, Fla., 1983, pp. 321–339.
190. J. O. McLean, M. C. Hawkins, and R. H. Hendrick, Lubrication in the Non-
ferrous Industries; in Standard Handbook of Lubrication Engineering, J. J.
O’Connor, J. Boyd, and E. A. Avallone, eds., McGraw-Hill, New York, 1968,
pp. 42-1 – 42-14.
191. C. Wick, J. T. Benedict, and R. F. Veilleux, Tool and Manufacturing Engineers
Handbook, Vol. 2 Forming, Society of Manufacturing Engineers, Dearborn, Mich.,
1984, pp. 3-1–3-22.
192. W. J. Landry, Use of Normal Paraffins in Cold Rolling of Aluminum, Technigram
D239-J2, Exxon Company U.S.A., 1995.
193. Lubricating Grease Guide, 3rd ed., National Lubricating Grease Institute (NLGI),
Kansas City, Mo., 1994.
Vol. 2 LUBRICATION AND LUBRICANTS 83

194. R. F. Haycock and J. E. Hillier, Automotive Lubricants Reference Book, 2nd ed., SAE
International, Warrendale, Pa., 2004, pp. 295–313.
195. ASTM D 217-02 Standard Test Methods for Cone Penetration of Lubricating Grease,
in Annual Book of ASTM Standards, Section 5, ASTM International, West Consho-
hocken, Pa., Issued Annually.
196. Exxon Booklet DG-400, Encyclopedia for the User of Petroleum Products, Exxon
Company, Houston, Tex., 1993.
197. ASTM D566-02, Standard Test Method for Dropping Point of Grease, in Annual
Book of ASTM Standards, Section 5, ASTM International, West Conshohocken,
Pa., Issued Annually.
198. ASTM D2265-00, Standard Test Method for Dropping Point of Lubricating Grease
Over Wide Temperature Range, in Annual Book of ASTM Standards, Section 5,
ASTM International, West Conshohocken, Pa., Issued Annually.
199. C. J. Boner, Manufacture and Application of Lubricating Greases, Rheinhold Pub-
lishing Corporation, New York, 1954, p. 397.
200. ASTM D4950-01(2004)e1 Standard Classification and Specification of Automotive
Service Greases, in Annual Book of ASTM Standards, Section 5, ASTM Interna-
tional, West Conshohocken, Pa., Issued Annually.
201. D. Klamann, Lubricants and Related Products, Verlag Chemie, Weinheim,
Germany, Deerfield Beach, GL, Basel, 1984, pp. 154–169.
202. A. Erdemir, Solid Lubricants and Self-Lubricating Films, in B. Bhushan, ed.,
Modern Tribology Handbook, Vol. 2, pp. 787–825 (2001).
203. L. C. Lipp, Lub. Eng. 32(11), 574 (1976).
204. J. K. Lancaster, in Handbook of Lubrication: Theory and Practice of Tribology,
Vol. II, CRC Press, Boca Raton, Fla., 1983, 269–290.
205. H. F. Sliney, ASM Handbook, Vol. 18, ASM International, Metals Park, Ohio, 1992,
pp. 113–122.
206. A. Erdemir and co-workers, Lub. Eng. 47(3), 179 (1991).
207. R. M. Gresham, in E. R. Booser, ed., Tribology Data Handbook, CRC Press, Boca
Raton, Fla., 1997, pp. 600–607.
208. E. R. Booser, ed., Handbook of Lubrication: Theory and Practice of Tribology,
Vol. III, CRC Press, Boca Raton, Fla., 1983, pp. 167–181.
209. W. E. Campbell, in F. F. Ling, E. E. Klaus, and R. S. Fein, eds., Boundary Lubrica-
tion: An Appraisal of World Literature, American Society of Mechanical Engineers,
New York, 1969.
210. R. M. Gresham, Lub. Eng. 44(2), 143 (1988).
211. C. DellaCorte, Tribol. Trans. 46(3), 361 (2003).
212. K. Carnes, Tribol. Lub. Technol. 59(11), 14 (2004).
213. A. Erdemir and C. Donnet, ‘‘Tribology of Diamond, Diamond-Like Carbon, and
Related Films’’ in B. Bhushan, ed., Modern Tribology Handbook, Vol. 2, CRC Press
LLC, Boca Raton, Fla., 2001, pp. 871–908.
214. M. F. Alzoubi, O. O. Ajayi, J. B. Woodford, A. Erdemir, and G. R. Fenske, Tribol.
Trans. 44(4), 591 (2001).
215. Near-Frictionless Coating Nears Commercial Applications, www.spacedaily.com,
Argonne, Sept. 3, 2002.
216. C. Donnet and A. Erdemir, Tribol. Lett. 17(3), 389 (2004).
217. D. H. Buckley and R. L. Johnson, ASLE Trans. 6, 1 (1963).
218. J. A. Schey, Tribology in Metalworking: Friction, Lubrication and Wear, ASM Inter-
national, Metals Park, Ohio, 1983.
219. D. F. Wilcock, in E. R. Booser, ed., Handbook of Lubrication: Theory and Practice of
Tribology, Vol. II, CRC Press, Boca Raton, Fla., 1983, pp. 291–300.
220. W. A. Gross, ed., Fluid Film Lubrication, John Wiley & Sons, Inc., New York, 1980.
84 LUBRICATION AND LUBRICANTS Vol. 2

221. M. M. Khonsari, L. A. Matsch, and W. Shapiro, Gas Bearings in GR 14, Vol. III,
pp. 553–575.
222. A. J. Baumgartner, in R. A. Burton, ed., Bearing and Seal Design in Nuclear Power
Machinery, American Society of Mechanical Engineers, New York, 1967.
223. J. Gerkema, ASLE Trans. 28(1), 47 (1985).
224. D. M. Pirro and A. A. Wessol, Lubrication Fundamentals, 2nd ed., Marcel Dekker,
New York, 2001.
225. J. G. Carroll and S. R. Calish, Lub. Eng. 13(7), 388 (1957).
226. E. R. Booser, in J. J. O’Conner, J. Boyd, and E. A. Avallone, eds., Standard Hand-
book of Lubrication Engineering, McGraw-Hill, New York, 1968, pp. 44-1–44-13.
227. R. O. Bolt, in E. R. Booser, ed., Handbook of Lubrication: Theory and Practice of
Tribology, Vol. I, CRC Press, Boca Raton, Fla., 1983, pp. 209–223.
228. J. E. Kelley, T. D. Roberts, and H. M. Harris, A Penetrometer for Measuring the
Absolute Viscosity of Glass, report 6358, U. S. Bureau of Mines, Washington, D.
C., 1964.
229. R. A. Reich, P. A. Stewart, J. Bohaychick, and J. A. Urbanski, Lub. Eng. 59(7),
16 (2003).
230. Solid and Liquid Lubricants for Extreme Environments ASLE Special Publication
SP-15, 1984.
231. 2004–2005 Lubricants Industry Sourcebook, LNG Publishing Company, Inc., Falls
Church, Va., 2004.
232. M. Khonsari and E. R. Booser, Mach. Lub., May 2004.
233. Toxic Substances Control Act, Public Law No. 469, 94th U. S. Congress Chemical
Substance Inventory, U. S. Environmental Protection Agency, Office of Toxic Sub-
stances, Washington, D. C., 1975.
234. J. M. Perez and D. J. Hoke, in E. R. Booser, ed., Tribology Data Handbook, CRC
Press, Boca Raton, Fla., 1997, pp. 1031–1045.
235. J. W. Swain Jr., in E. R. Booser, ed., Handbook of Lubrication: Theory and Practice
of Tribology, Vol. I, CRC Press, Boca Raton, Fla., 1983, pp. 533–549.
236. S. Napier and K. Rich, ‘‘Waste Treatability of Aqueous-Based Synthetic Metalwork-
ing Fluids’’, Lub. Eng. 41(6), 361 (1985).
237. J. W. Swain Jr., Lub. Eng. 39(9), 551 (1983).
238. D. W. Brinkman, Lub. Eng. 43(5), 324 (1987).
239. T. M. Warne and C. A. Halder, Lub. Eng. 42(2), 97 (1986).
240. M. J. Raab, Lub. Eng. 58(2), 16 (2002).
241. Draft Standard NSF 116-2001 Nonfood Compounds Used in Food Processing
Facilities - Food-Grade Lubricants, NSF International, Ann Arbor, Mich., 2001.
242. M. J. Raab, Food-Grade Lubricants: Current Issues presented to New York Section,
STLE, December, 2004.
243. K. Yano, Non-Food Compounds Registration: Past, Present and Future, NLGI Tech-
nical Paper 0308, Presented at 70th NLGI Annual Meeting, NLGI, Kansas City,
Mo., 2003.

JAMES R. DICKEY
Lubricants Consultant
Vol. 2 WAXES 85

WAXES
1. Introduction

Waxes are among the oldest worked materials used by humans. Their value as
versatile construction materials (‘‘man’s first plastic’’) was discovered very early.
Today, waxes are used mostly as additives and active substances. The use of
waxes is expected to increase in the future because of their generally favorable
toxicological and ecological properties.
The historic prototype of all waxes is beeswax. Since it could be obtained
with relatively little effort, it was popular in antiquity, and even now the
term wax is occasionally used in everyday speech as a synonym for beeswax.
However, the scientific and commercial definition of wax covers a much wider
area.
1.1. History of Waxes and Their Applications. Utilization of waxes
was probably begun in prehistoric times, but because of their transitory nature,
no definite archaeological evidence exists. Thus, the utilization of wax and
related substances in mummification and as protective coverings in ancient
Egypt (from ca. 3000 B.C.) represents the earliest scientific proof of the use of
waxes. Ancient written documents contain many indications that waxes found
many different applications. The most well known is the story of Daedalus and
Icarus, who used wax as an adhesive to make wings by attaching feathers to each
other. In antiquity, wax was used as a raw material for modeling, in the produc-
tion of casting molds, as a pigment carrier, and for surface protection.
In colonial times, hitherto unknown waxes, such as carnaúba, candelilla,
and Chinese insect wax, were introduced in Europe. From the occurrence of
wax, COLUMBUS inferred the riches of the Caribbean islands: ‘‘Where there is
wax, there are also thousands of other things’’ (1).
For a long time, not much was known about the chemical nature of wax.
Only in the 18th century did the discovery occur that beeswax is an animal
secretion and not a plant product collected by bees. Wax research was estab-
lished as a scientific discipline in 1823 (2). It became part of the new research
area of soaps, oils, fats, and waxes. The real breakthrough of wax as an impor-
tant raw material, in terms of quantity as well, occurred at the beginning of the
Industrial Revolution. Ozocerite (fossil wax) was mined and refined to give cer-
esin (1875), montan wax was obtained from Eocene lignite (1897), and paraffin
waxes were obtained from crude petroleum. In 1935 the first fully synthetic
waxes were produced by the Fischer–Tropsch process. Polyethylene wax
has been synthesized by the high pressure process since 1939, and became
available by the low-pressure Ziegler process after 1953. On a laboratory
scale polyolefin waxes can also be synthesized by using modern metallocene
catalysts.
Degradation processes for the production of waxes, which start from high
molar mass plastics (mainly polypropylene), have also achieved a certain degree
of importance. Finally, low molar mass substances, which would otherwise have
to be disposed of in plastics production, can be processed or refined to give indust-
rially utilizable waxes.

ß2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim


10.1002/14356007.a28_103
86 WAXES Vol. 2

A large and still increasing number of natural and synthetic waxes and
related substances exists, as well as applications for these materials. No decrease
in overall demand is apparent, so waxes are likely to remain important
during the coming decades. The history of waxes has been described in great
detail in (3).
1.2. Definition. No generally accepted definition exists for the term wax.
All attempts to formulate a precise, comprehensive, and scientifically verifiable
definition of wax must take the large number of waxlike products and the chemi-
cal complexity of individual types into account. Nevertheless, use of the term wax
for different chemical species with common properties is still reasonable.
Typically waxes do not consist of a single chemical compound, but are often
very complex mixtures. Being oligomers or polymers in many cases, the compo-
nents differ in their molar mass, molar mass distribution, or in the degree of side-
chain branching. Functional groups (eg, carboxyl, alcohol, ester, keto, and amide
groups) can be detected in waxes, sometimes several different groups.
The academic definition still quoted in chemistry text books — that waxes
are esters of long-chain carboxylic acids with long-chain alcohols — is no longer
useful. It applies fairly well only to some classical waxes, such as beeswax; others
(eg, petroleum waxes) do not fall in this category. Today, physical and technical
definitions are preferred. Several attempts were made to differentiate between
waxes and other classes of substance, particularly fats, resins, and high molar
mass polymers, by using several criteria. These primarily physical definitions
are to some extent arbitrary and are not generalled accepted. Waxes can also
be classified according to their applications.
Probably the most conclusive definition has been drawn up in Europe by
the Deutsche Gesellschaft für Fettwissenschaft (DGF, German Association for
Fat Science). It was used in modified form in the customs tariff of the European
Union (4,5). According to this definition, waxes must have:

1. A drop point (mp) >40 8C


2. Their melt viscosity must not exceed 10 000 mPa  s at 10 8C above the drop
point
3. They should be polishable under slight pressure and have a strongly
temperature-dependent consistency and solubility
4. At 20  C they must be kneadable or hard to brittle, coarse to finely crystal-
line, transparent to opaque, but not glassy, or highly viscous or liquid
5. Above 40  C they should melt without decomposition
6. Above the mp the viscosity should exhibit a strongly negative temperature
dependence and the liquid should not tend to stringiness
7. Waxes should normally melt between ca. 50 and 90  C (in exceptional cases
up to 200  C)
8. Waxes generally burn with a sooting flame after ignition
9. Waxes can form pastes or gels and are poor conductors of heat and electri-
city (ie, they are thermal and electrical insulators).

In the case of higher molar mass polyolefin waxes, differentiating between


a wax and a thermoplastic polymer is sometimes difficult. This differentiation is
Vol. 2 WAXES 87

Fig. 1. Classification of waxes.

particularly important for applications where a food approval is essential (FDA,


EC, MITI in Japan, etc.). In case of doubt, relevant national regulations must be
consulted.
1.3. Classification. Waxes can be classified according to various criteria
such as origin; chemical, physical, and engineering properties; or applications
(see Section 1.2). The primary differentiation is usually made according to origin
or occurrence and synthesis. A proposal for a classification is shown in Figure 1.
Here, waxes are divided into two main groups: natural and synthetic. Natural
waxes exhibit their wax character without chemical treatment, whereas syn-
thetic waxes generally acquire their waxy nature in the course of synthesis.
There are no clear borderlines.
Natural waxes are formed through biochemical processes and are pro-
ducts of animal or plant metabolism. Natural waxes formed in earlier geological
periods are known as fossil waxes. All petroleum, lignite (montan wax), and peat
waxes belong to this largest category. Fossil waxes occur predominantly as minor
components of oil, coal, and peat. Some deposits with a high wax content were
formed by sedimentation. Ozocerite separated from oil, and pyropissite, one of
the few organic minerals, is a mixture of montan wax and resins. Whereas pet-
roleum waxes are predominantly nonpolar hydrocarbons, lignite and peat waxes
consist of oxygen-containing compounds.
Biological wax synthesis is still taking place in nature. Many plants, parti-
cularly carnaúba palm, and animals – most importantly insects, such as bees –
produce waxes. These are known as nonfossil or recent natural waxes. Natural
88 WAXES Vol. 2

waxes are seldom used industrially in their original form. They are generally
converted into modified natural waxes by refining, eg, by distillation, or extrac-
tion. Chemical processes, such as hydrogenation, bleaching, and oxidation, can
also be used. Physical and chemical treatment can be combined; the objective
of all these processes is to obtain a wax that is as pure as possible.
Partially Synthetic Waxes. If natural waxes or waxlike materials are
modified by chemical reactions such as esterification, amidation, or neutraliza-
tion of acidic waxes (with aqueous alkali or alkaline-earth metal hydroxides),
partially synthetic waxes are obtained. These can be tailored to particular appli-
cations. In esterified waxes the relationship to natural waxes is still easily recog-
nizable. For example, natural montan waxes, which consist mainly of esters of
long-chain acids with long-chain alcohols, can be converted into acid waxes by
cleavage of the esters and oxidation of the alcohols. Subsequent reaction with
dihydric alcohols gives esters that have chemical structures comparable to
those of the starting materials. Amide waxes do not occur naturally to any signif-
icant extent. This group consists of reaction products of fatty acids with ammo-
nia, amines, and diamines. An industrially important amide wax is distearoyl
ethylenediamine.
Alcohol waxes (lanette waxes) also belong to the partially synthetic waxes.
They are mixtures of long-chain alcohols that are optimized for their main area of
use, in ointments and creams for pharmaceuticals and cosmetics, by the addition
of fatty acid esters and emulsifiers. Alcohol waxes can also be used for industrial
emulsions with good long-term stability.
Wool wax (lanolin) is obtained in crude form from sheep’s wool. It is refined
by chemical and adsorptive bleaching, and sometimes by subsequent hydro-
genation or separation of the acidic fractions after hydrolysis. Wool wax is a
mixture of esters, some being derived from branched or cyclic waxy or fatty
acids and alcohols.
Fully synthetic waxes were developed only in the 20th century. Low
molar mass compounds are used as starting materials. The products can
be waxes in the narrower sense of the definition (see Section 1.2), or sub-
stances with only partial wax character. The two main groups of fully syn-
thetic waxes are the Fischer – Tropsch and the polyolefin waxes. These
waxes can be classified according to the starting material used for production
(C1 or C2 chemistry).
C1 Building Blocks. The Fischer– Tropsch (FT) synthesis is described
elsewhere in detail. The oldest FT variant is the ARGE process (Arbeitsge-
meinschaft Ruhrchemie– Lurgi). Synthesis gas is obtained by reaction of
steam with carbon and is then converted catalytically into a broad spectrum
of saturated and unsaturated hydrocarbons. Among these products are the
Fischer– Tropsch waxes. They are characterized by low branching of the hydro-
carbon chain and a narrow molar mass distribution. They are relatively hard
products. The ARGE process has been used commercially by SASOL in the
Republic of South Africa.
Besides carbon, natural gas can also be used as a raw material for C1 syn-
theses. The Shell Middle Distillate Synthesis (SMDS) process utilizes syn-
thesis gas produced by steam reforming of natural gas (6). Fischer–Tropsch
wax fractions are produced and marketed at the new SMDS plant in Bintulu
Vol. 2 WAXES 89

(Sarawak/Malaysia). This process is a variant of the classical FT process and


employs specially developed catalysts.
Polyolefin waxes are, in most cases, products of C2 or C3 chemistry. Low
molar mass a-olefins, usually ethylene, are polymerized. The other route to poly-
ethylene waxes – thermomechanical degradation of polyethylene plastic – is now
of minor importance. The two most important production processes are the high-
pressure and the Ziegler processes. More recently, polyolefin waxes have also
been synthesized by using modern metallocene catalysts.
Polyolefin waxes can be homopolymers or copolymers of C2, C3, C4 (ethy-
lene, propene, and butene), or higher a-olefins. Polypropylene (PP) waxes are
produced by the Ziegler process or by thermomechanical degradation of polypro-
pylene plastic. Polypropylene waxes are generally partially crystalline. In addi-
tion, predominantly amorphous and atactic low molar mass polypropylene (APO
or APAO) with a waxlike character is now produced and marketed by several
companies. Besides this deliberate production, atactic PP (APP) is still produced
as a byproduct of PP plastics.
All FT and polyolefin waxes are nonpolar. Polar waxes are obtained by sub-
sequent reactions, such as oxidation by air or grafting (eg, with maleic anhy-
dride), optionally followed by neutralization to metallic soaps, esterification, or
amide formation. Polar polyolefin waxes are also accessible via direct copolymer-
ization or addition reactions (eg, of olefins with unsaturated carboxylic acids or
esters).
A considerable proportion of industrially processed waxes is used not in
pure form but as wax compounds. For example, the viscosity and mp of paraffin
waxes can be increased to the desired degree by mixing with polyolefin waxes.
1.4. Properties and Uses. The number of possible and actual uses
of waxes is extraordinarily high. The physical and chemical properties of various
types of wax can be combined fairly freely through mixtures and prepa-
rations. Therefore, waxes are represented in almost all possible areas of use,
although to differing extents. Any selection of applications thus has to be some-
what arbitrary.
For the user, the application-oriented function of the wax is important, not
so much its chemical structure. The latter is defined primarily by the character-
istic hydrocarbon chain, the basic building block of almost all waxes. It imparts
hydrophobing properties to waxes. The carbon chain length of fatlike waxes is
n ¼ 16–18 (eg, amide wax); for esterified and paraffin waxes, n ¼ 20–60; and
for polyethylene waxes, less than hundred to several thousand. Besides the
chain length itself (and thus the molar mass), molar mass distribution and
degree of branching also affect the properties of waxes. The melting and soften-
ing points, melt viscosity, and degree of hardness increase with increasing chain
length. High degrees of branching lower the mp and the hardness. Polar groups,
such as carboxyl, ester, and amide groups, strengthen intermolecular forces and
generally raise the mp and hardness.
Particular features of many waxes are their good absorption capacities and
their ability to bind solvents. Hot wax solutions form stable, homogeneous pastes
on cooling. These can be applied to surfaces and polished after evaporation of the
solvent. The resulting wax films are shiny or glossy, hard, and resistant to
mechanical stress.
90 WAXES Vol. 2

When a suitable emulsifier is added, polar waxes can form fine stable dis-
persions in water. Mixed aqueous–organic emulsions can also be produced.
These mostly stable dispersions and emulsions are used for surface protection.
After application, the wax-containing phase dries and a protective, sometimes
even glossy, coating is formed.
In addition to surface protection, hydrophobing, and gloss-giving, other
functions of waxes include surface protection, hydrophobing, gloss-giving,
matting, mold releasing, lubricating, binding, imparting compatibility or flexibil-
ity, regulating viscosity, adjusting consistency, emulsifying, dispersing, and
adjusting drop points. Waxes are also used as combustible and illuminating
materials.
For special applications only a few – frequently only one – of these functions
are important. Some occasionally opposite effects are noteworthy: wax gives pol-
ish and gloss to various surfaces but has a matting effect as an additive in paints
and coatings. Waxes are used as binders in master batches but as release agents
in molding plastics. Table 1 gives some idea of the versatility of waxes but does
not claim to be complete. For the specific suitability of waxes to individual appli-
cations, monographs, company brochures, and prospectuses should be consulted.

Table 1. Branches of Industry in which Waxes are Used


Branch Examples of applications
Adhesives, hot melts viscosity regulation, lubricants, surface hardening
Building modification of bitumen, antigraffiti treatment
Candles fuel, drop point regulation
Ceramics and metal binders for sintering
Cosmetics binders and consistency regulators for ointments,
pastes, creams, lipsticks
Electrical and electronics release agents, insulating materials, etching bases
industries
Explosives stabilization
Foods citrus fruit and cheese coating, chewing gum base,
confectionery
Matches, pyrotechnics impregnation, fuel
Medicine and pharmaceuticals molding and release agents in dental laboratories,
retardants, surface hardening of pills
Office equipment dispersing agents and binders for carbon paper
and selfduplicating paper; antioffset for toners
for photocopiers
Paints and coatings matting, surface protection
Paper and cardboard surface hardening
Plastics lubricants (PVC *), release agents (PA **),
pigment carriers (masterbatch)
Polishes surface protection of leather, floors, cars
Printing inks improvement of rub resistance, slip
Recycling compatibilizing
Rubber industry release agents enhancing rigidity, surface
hardening

*PVC ¼ poly(vinyl choride).


**PA ¼ polyamide.
Vol. 2 WAXES 91

1.5. World Market Volume for Waxes. No reliable global capacity or


production data are available for waxes. The main reasons for this are the
many types of wax and the difficulties in differentiating among them; the com-
paratively low transparency of the market; and the dependence of wax produc-
tion on external factors, such as crude petroleum consumption and annual
harvest of natural waxes. Some statistical data exist for certain markets (eg,
the United States (7)). Thus, the following figures can be regarded only as
rough indications, with a correspondingly large degree of fluctuation, of the mag-
nitude of annual worldwide wax consumption data:

Paraffin waxes, including microcrystalline waxes 3 000 000 t/a


Polyolefin waxes including oxidates and copolymers 200 000 t/a
Fischer–Tropsch waxes 100 000 t/a
Amide wax 50 000 t/a
Montan wax including crude montan wax 25 000 t/a
Carnaúba wax 15 000 t/a
Beeswax 10 000 t/a
Candelilla and other plant waxes 3 000 t/a

Forecasts for future market development anticipate positive growth factors.


Specialties, such as special waxes or compounds for critical applications (eg, in
the electrical, electronics, or ceramics industries), will probably profit more
from this development than waxes, that are tend to be continuously replaced
in their applications. The decreasing demand for waxes in the packaging sector,
for example, is becoming noticeable. The proportion of waxes used in some other
classical applications (eg, carbon paper) is continually declining. However, waxes
generally have favorable ecological and toxicological properties, in addition to the
variety of applications described, so their market chances may outweigh the
risks in the long term.
1.6. International Wax Organizations. In Europe the interests of all
wax producers have been represented by the European Wax Federation since
1979 (EWF, Avenue van Nieuwenhuyse 4, Brussels, Belgium). The EWF is a
member of the Conseil Européen des Federations de l’Industrie Chimique
(CEFIC) and is open to all European wax producers and some outside Europe.
It represents the interests of its members, with regard to both technical ques-
tions such as specifications and legal matters referring in particular to the Eur-
opean Union. In other regions, no organizations exist that cover the entire wax
spectrum.

2. Recent Natural Waxes

2.1. Introduction. Many plant and animal organisms produce waxes


with extraordinarily complex compositions. The main components are usually
those that correspond to the classical definition of waxes – ie, esters of long-
chain aliphatic alcohols and acids, which belong to homologous series in the
ca. C16 –C36 (predominantly C22 –C36) range, those with an even number of car-
bon atoms predominating. Recent natural waxes also contain smaller quantities
92 WAXES Vol. 2

of esters derived from acids and alcohols with an uneven number of carbon atoms
as well as bifunctional components such as diols, hydroxycarboxylic, and dicar-
boxylic acids. Natural waxes contain very small proportions of branched car-
boxylic acids, mainly those with o-methyl branching. Aromatic acids, such as
cinnamic, 4-hydroxycinnamic, and ferulic (4-hydroxy-3-methoxycinnamic)
acids, have also been found in some waxes. Because of the presence of bifunc-
tional components, natural waxes can contain not only the simple, classical
wax esters, but also oligoesters with two or more ester functions and other func-
tional groups in the molecule.
Free acids and alcohols are present in natural waxes in widely varying
quantities. They are generally identical to those present as ester components.
Homologous n-alkanes in the ca. C15 –C37 range, mainly those with an
uneven number of carbon atoms, are also components of natural waxes in propor-
tions of 1–50 %. Branched-chain and olefinic hydrocarbons are present in insig-
nificant quantities. Glycerides (fats), phytosterols, terpenes, resins, long-chain
carbonyl compounds, and flower pigments are also present in very small propor-
tions. Because of the complex mixtures involved and the relatively simple analy-
sis techniques of earlier years, the inconsistency of literature data on the
composition of natural waxes is understandable.
Modern analytical instruments are required for complete analysis of a nat-
ural wax. High-temperature capillary GC, possibly coupled to a mass spectro-
meter, is a suitable the best analytical method.
Beeswax and carnaúba and candelilla waxes have a total annual market
volume of ca. 28 000 t worldwide. These are economically the most important nat-
ural waxes (8).
Among the large number of other known waxes, ouricury, esparto, bamboo,
and Japan wax (actually a special vegetable tallow) are of regional or local impor-
tance. Their availability is estimated to be 3000–4000 t/a (8,9).
In all areas of application for waxes, recent natural waxes are now mostly
used together with modified fossil and synthetic waxes. Applications include pre-
parations for cleaning, polishing, and preserving floors, furniture, and car body-
work surfaces; the candle industry; craft work; production of pharmaceuticals,
cosmetics, and confectionery (observing appropriate legal approval); metal and
ceramic sintering technology; coating techniques; the paint and coating indus-
tries; and the production of carbon paper compounds. The latter, which was
important in terms of quantity, has been replaced almost completely by new
duplicating processes. Despite the market recession this has caused (10), an
annual growth of ca. 4 % is expected in the U.S. market for natural and synthetic
waxes (11).
Special, ion-free or low-ion natural wax raffinates have also been used in
electronics.
In his standard book The Chemistry and Technology of Waxes, A. J.
WARTH included a comprehensive chapter on all known recent natural waxes.
The second edition of the book was published in 1956 and is currently out of
print (12).
2.2. Vegetable Waxes. In contact with the environment, plants form a
compressed epidermis on their surfaces by depositing secondary layers of cellu-
lose, except in the case of inflorescence (13). This mechanical reinforcement is
Vol. 2 WAXES 93

supplemented by deposition of cutin in the outer epidermal layers and an outer


coating of cutin (cuticula) to regulate water content. As a result of climatic
conditions, many plants in tropical regions also store waxes in cutinized mem-
brane layers in the cuticula as an additional protection against evaporation of
water. In many cases, waxes exude outward and form wax coatings, which can
be up to several millimeters thick. Xerophytes, plants found in exceptionally arid
regions, are however not the greatest wax producers (14). Specific soil types, tem-
perature, dry periods, and heavy rainfall, as well as temporal changes in these,
are responsible for enhanced wax formation. Vegetable waxes can be classified
according to their origin as deciduous and coniferous tree waxes, shrub and
herb waxes, and grass and flower waxes, and within these groups into leaf, nee-
dle, stem, and root waxes; bark and skin waxes; and seed and fruit waxes.
Carnaúba Wax. The leaf wax of the carnaúba palm is by far the most
important recent vegetable wax both economically and with regard to applica-
tions (12,13,15,16). The carnaúba palm is a fan palm found mainly in Brazil
and given the botanical name Copernicia cerifera by V. MARTIUS in 1819 in
honor of the astronomer COPERNICUS. It was apostrophized as the life tree of Bra-
zil by V. HUMBOLDT.
History. The existence of carnaúba wax was first reported in 1648 (17),
and some data concerning its composition were published in 1811 in Transac-
tions of the Royal Society, London.
In 1836 the Brazilian MACEDO arranged for chemical investigations to be car-
ried out on carnaúba wax at the Sorbonne in Paris. He began small-scale extrac-
tion and purification of carnaúba wax in 1856. In 1862, 1280 t of carnaúba wax
was exported from Brazil.
Occurrence and Isolation. The carnaúba palm occurs mainly in north
eastern Brazil in the provinces of Bahia, Rio Grande do Norte, Paraiba, Pernam-
buco, and particularly Ceara and Piaui, where there is a total stand of ca. 90106
palms in a surface area of 106 km2 (18,19). Smaller areas exist in southern Brazil,
northern Argentina, Paraguay, and Bolivia. Different climatic conditions in
other tropical and subtropical countries, in which carnaúba palms were planted
as an experiment, allow satisfactory growth but markedly reduced formation of
wax deposits on the leaves.
The palms grow to a height of 6–12 m, occasionally to 20 m, and are said to
achieve a lifetime of up to 200 years. When the palm has reached an age of ca.
eight years, one leaf gives a wax yield of 5–7 g, which is worth harvesting. The
annual yield of wax per palm is ca. 150 g.
To isolate the wax, ca. 20–30 leaves at different stages of development are
removed from each palm twice a year (September and December) (12).
The unopened heart leaves, of which only a few are removed per palm, carry
particularly light-colored wax (prime yellow). Wax on the outer palm leaves is
yellow, gray-green, or gray-brown, depending on climatic conditions during the
vegetation period and on leaf age.
After drying the leaves in the sun or on steam-heated racks, the wax is
chipped off manually or nowadays mostly by machine. The primary purification
process involves melting over water and gives a crude wax that can be purified
further by centrifugation, filtration with additives, or solvent extraction. Addi-
tional bleaching with hydrogen peroxide is also carried out.
94 WAXES Vol. 2

The central ribs are often removed from the leaves before drying. The dried
leaves are then chopped and the chips separated by air classification to give a
mixture of ca. 60 % wax and 40 % chopped plant, which is stirred with water
and oxalic acid to form a paste. The latter is heated to its boiling point and
pressed through a filter while hot. Centrifugation of the filtrate gives virtually
anhydrous wax. The remaining wax is extracted from the dried filter cake
with aliphatic hydrocarbons (eg, heptane).
The crude product is normally supplied in lumps that are obtained by
breaking up the wax blocks into which the melt was cast. Carnaúba wax is
also sold now in other forms.
Types. Internationally recognized classifications and specifications for
genuine, unfalsified types of carnaúba wax were worked out by the American
Wax Importers and Refiners Association according to established methods of
determination (20).
The classification used in Brazil, shown in Table 2, has been valid world-
wide since the beginning of the 1980s.
The highest-value type, T 1, is a prime yellow wax purified by filtration. To
obtain T 3, suitable crude waxes (colored) are filtered and then bleached with
hydrogen peroxide. The T 4 wax is exported predominantly in the form of centri-
fuged or filtered variants.
Chemically refined or derivatized carnaúba waxes are no longer available
commercially because of lack of demand (8). Chemical refining processes, such
as heating with alkali in the presence of paraffins, gave carnaúba wax residues
in addition to the corresponding raffinates (21–27). Where these processes are
used, they account for <1 % of the total volume of carnaúba wax.
Properties. Carnaúba wax is one of the hardest and highest-melting nat-
ural waxes. Because of its fine crystalline structure, its fracture edges are smooth

Table 2. Carnaúba Wax Types and Specifications


T1 T 3 (fatty T 4 (fatty T 4 (fatty T 4 (fatty
(prime gray/ gray/ gray/ gray/
Characteristic yellow) light) crude) centrifuged) filtered)
Minimum mp, 8C 83.0 82.5 82.5 82.5 82.5
Minimum flash point, 8C 310 299 299 299 299
Moisture content, % 0 0 2.0 (max.) 1.0 (max.) 0.1 (max.)
Minimum acid number, 2 4 4 4 4
mg KOH/g
Maximum acid number, 6 10 10 10 10
mg KOH/g
Minimum saponification 78 78 78 78 78
number, mg KOH/g
Maximum saponification 88 88 88 88 88
number, mg KOH/g
Ester number, mg KOH/g 75–85 75–85 75–85 75–85 75–85
Maximum ash content, % 0.04 0.15 0.20 0.20 0.10
Solidification point, 8C 80 79 79 79 79
Drop point, 8C 84 83 83 83 83
Minimum particle size, 95 95 95 95 95
mesh
Vol. 2 WAXES 95

and have a macroscopically homogeneous structure. At room temperature the


wax has a weakly aromatic odor and a characteristic haylike scent (similar to
coumarin) in the molten state (28). Carnaúba wax is compatible with almost
all natural and synthetic waxes and a number of natural and synthetic resins.
It is readily soluble in most nonpolar solvents on warming (12) and is miscible
with them in all proportions above its mp. On cooling, the wax precipitates
from solution to form a solid paste. Carnaúba wax is only partially soluble in
polar solvents, even on warming, and is generally sparingly soluble at room tem-
perature (0.15–0.6 %) (12).
Melting, solidification, and drop points as well as hardness of other waxes
increase on addition of small quantities of carnaúba wax. With paraffins, the
addition of carnaúba wax suppresses their tendency toward coarse crystallinity
and reduces possible tack in the case of soft waxes.
Composition. Many analyses of carnaúba wax has been carried out, with
conflicting results (12,14,15). According to a thorough investigation, carnaúba
wax prime yellow (T 1) has the composition given below (29):

Aliphatic esters 40.0 wt %


Diesters of 4-hydroxycinnamic acid 21.0 wt %
Esters of o-hydroxycarboxylic acids 13.0 wt %
Free alcohols 12.0 wt %
Diesters of 4-methoxycinnamic acid 7.0 wt %
Free aliphatic acids 4.0 wt %
Free aromatic acids 1.0 wt %
Hydrocarbons (paraffins) 1.0 wt %
Free o-hydroxycarboxylic acids 0.5 wt %
Triterpene diols 0.5 wt %
Unsaponifiable components 56.4 wt %
Saponifiable components 39.2 wt %
Aromatics and/or resins 4.4 wt %

The aliphatic esters contain monocarboxylic acids of average chain length


C26 and monohydric alcohols of average chain length C32. The o-hydroxyesters
contained are mixtures of ca. 90 % esters of o-hydroxyacids (C26) and monohydric
alcohols (C32) and 10 % esters of monocarboxylic acids (C28) and a,o-diols
(C30).
Esters containing 4-hydroxy- and 4-methoxycinnamic acids are present
mainly as oligomers and polymers. The monomer units of these are diesters
of the above-named cinnamic acids with mono- and polyhydric alcohols and
o-hydroxycarboxylic acids.
The free alcohols are similar in composition to those in aliphatic esters. The
wax also contains a small proportion of secondary alcohols.
The even-numbered n-monocarboxylic acids, n-monohydric alcohols, and
n-alkanes present in all carnaúba wax types have typical, distinctive chain-
length distributions (see Table 3).
The important difference between Type 1 carnaúba wax and the other types
is that the latter contain no triterpenediols and the esters of 4-hydroxy- and
4-methoxycinnamic acids have higher degrees of polymerization compared
with Type 1. This may be attributed partly to differences in climate but also to
96 WAXES Vol. 2

Table 3. Chain-length Distributions for Straight-chain


Mmonocarboxylic Acids and Monohydric Alcohols, and Paraffins
in Carnaúba Wax
Chain length Acids *, % Alcohols *, % Paraffins *, %
(C-number) (approx.) ** (approx.) ** (approx.) **
16 2.0
18 3.0
20 8.5 0.4
21 0.5
22 7.5 0.5 0.7
23 1.3
24 23.2 0.7 1.6
25 2.6
26 10.5 0.5 3.5
27 8.7
28 16.5 2.0 5.4
29 13.1
30 5.5 10.6 13.9
31 29.3
32 2.0 60.8 9.6
33 7.3
34 0.5 16.4 1.5
35 0.3
36 0.6 0.4

*Values based on 100 % total acid, total alcohols, and total paraffins in each
case (deviations from 100 % ¼ sum of all other acids and alcohols).
**Approximate.

the fact that Type 1 is harvested from genetically younger leaves, whereas the
others come from older leaves that have been exposed to more sunshine.
Uses. T 4 Variants. The wide use of carnaúba wax in the production of
preserving and cleansing agents for floors (floor polishes, ionogenic and noniono-
genic day-bright emulsions, wiping waxes and cleaners, sealing emulsions, spray
cleaners, etc.); and polishes for furniture, cars, and shoes (solvent–water or
straight solvent products) is based on its ability to form pastes and glossy and
repolishable films and its ability to be dispersed. In this classical area a large pro-
portion of T 4 variants of carnaúba wax are used.
Type 1 carnaúba wax and some of Type 3 (after absorptive purification to
remove peroxide components arising from hydrogen peroxide bleaching) are
approved for use in the pharmaceutical industry (eg, as polishes for pills) and
the cosmetics industry (as lipsticks and lip salves).
In the food sector, T 1 and T 3 waxes are used as release agents for bakery
and confectionery products and as additives in chewing gum production.
In polymer processing, carnaúba wax is used in release agent preparations
and, to a small extent, as a lubricant. The varnish industry uses carnaúba wax as
an additive in the interior coating of food containers (8).
In the leather industry, carnaúba wax is used for cutting waxes, cleaning
waxes, and edging inks. It is also approved for coating citrus fruit.
Vol. 2 WAXES 97

Economic Aspects. About 10 000–16 000 t/a carnaúba wax is available


worldwide, of which ca. one-fifth is Type 1. Brazil, the sole exporter, supplies
mainly Europe (ca. 2000 t/a), the United States (3000–4000 t/a), and Japan (8).
About 3000–3500 t/a is for indigenous supply.
Carnaúba wax consumption in the United States decreased markedly from
1955 to 1976 and then remained constant (9). Brazil is increasingly exporting
purified or bleached products in flaked form to achieve higher prices. Harvests,
exports, and price structuring are subject to government control.
Waxes with Similar Compositions. Copernicia australis, a palm that
produces large quantities of leaf wax, is found in Paraguay and around its border
with Argentina (Gran Chaco). It is similar to Copernicia cerifera in many
respects. A few small factories for isolating this palm wax are said to have
been erected since 1945 (30).
The Brazilian cauacu bush, found mainly in Pará province and by the
upper Amazon, produces a wax that is said to be similar to carnaúba wax in che-
mical composition. However, the wax is isolated only sporadically and to a lim-
ited extent (31).
Candelilla Wax. Occurrence and Isolation. The climatic conditions
of the semideserts in southern California, Arizona, south western Texas,
northern Mexico, and parts of Central and South America favor enhanced
wax formation in Euphorbia (E. cerifera, E. antisyphilitica) and Pedilanthus
(P. pavonis, P. aphyllus) species. The wax deposited on the stalks and leaf
stems of these plants, which grow as bushes or shrubs, is known as candelilla
wax (12,14,32,33).
Numerous and dense stands of these plants are found mainly in northern
Mexico in the states of Chihuahua, Coahuila, Durango, Nuevo León, San Luis
Potosi, Tamaulipas, and Zacatecas over a total area of more than 100 000 km2.
These are the only areas in which candelilla wax is obtained (34).
In manual harvesting the plants are pulled out, leaving most of the root
system in the soil. Together with scattered seeds, roots remaining partially in
the soil guarantee the growth of new stands, which takes several years.
In collecting and processing locations, plant material is boiled with ca. 0.2 %
sulfuric acid in open vessels, which are heated using already dewaxed and dried
stalks. Metal grids keep the plants under the surface of the liquid, and addition
of sulfuric acid inhibits emulsification of the wax. The latter collects in molten
form on the surface of the liquid. After the wax is skimmed off into open vats
or barrels, it solidifies to form the crude wax known as cerote. Yields based on
plant material are 3–4 %.
Wax from outer areas of the solidified wax blocks is removed with impuri-
ties that have concentrated there and is subjected to the boiling process again.
The remainder of the wax is purified in a central refinery located in Saltillo by
melting with sulfuric acid and filtering in filter presses. The crude wax broken
down into lumps is the Mexican commercial product (Mexican Standard Grade
Candelilla Wax).
Candelilla wax raffinates are obtained from crude wax by adsorptive puri-
fication or hydrogen peroxide bleaching. They are produced and sold in the Uni-
ted States and Europe. The properties of these products, apart from color, are
identical to those of crude candelilla wax.
98 WAXES Vol. 2

Following the transfer of Mexican candelilla wax production from state


monopoly to private enterprise, candelilla wax raffinates produced in Mexico
will also be put on the market (8).
By deresinification and subsequent oxidative chromic acid bleaching, pro-
ducts that are almost white can be obtained. Because of the high price of genuine
wax, these processes have no practical importance.
Properties and Composition. Candelilla wax is a hard, brittle wax that
is very similar to carnaúba wax with regard to solubility in polar and nonpolar
organic solvents. A large proportion of the resin present in the wax can be
extracted at room temperature with hot 70 % ethanol, acetone, or low-boiling,
chlorinated organic solvents.
The physical and chemical properties, the color (brown–brownish yellow–
pale yellow), and the degree of purity vary depending on climate, time of harvest,
region, and age of harvested plants. The recognized specification for genuine
crude candelilla wax is given in the following (34).

mp 68.5–72.5 8C
Refractive index 1.4550–1.4611
d15
4 0.950–0.990
Acid number 12–22 mg KOH/g
Saponification number 43–65 mg KOH/g
Hydrocarbon content 30.6–45.6 %
Total acid content 20.6–29.0 %
Color amber
Flash point 235.4–248.4 8C
Components insoluble in xylene/toluene (1 : 1) 0.0–0.1 %

The data agree with those previously laid down by the American Wax
Importers and Refiners Association for prime crude candelilla wax (20).
The average composition of candelilla wax gives below is derived from the
large number of values found in the literature.

Hydrocarbons (ca. 98 % paraffins þ 2 % alkenes) 42.0 wt %


Wax þ resin þ sitosteroyl esters 39.0 wt %
Lactones 6.0 wt %
Free wax and resin acids 8.0 wt %
Free wax and resin alcohols (terpene alcohols) 5.0 wt %
Saponifiable components 23.0–29.0 wt %
Unsaponifiable components 71.0–77.0 wt %

Candelilla wax differs fundamentally from carnaúba wax in its high hydro-
carbon content of 45% (max.) and resin content of 20%. Whereas carnaúba wax
contains >80 % wax according to the classical chemical definition, candelilla wax
consists of <35 % of these components.
As in all natural waxes, in candelilla wax the paraffins and wax acids or
alcohols with even-numbered carbon chains have typical and distinctive chain-
length distributions (Table 4).
Vol. 2 WAXES 99

Table 4. Chain-length Distribution of Even-numbered Carboxylic


Acids and Alcohols, and Paraffins in Candelilla Wax
Chain length Acids *, % Alcohols *, % Paraffins *, %
(C-number) (approx.) ** (approx.) ** (approx.) **
22 0.3
23
24 0.1
25
26 0.3 2.4 0.1
27 0.2
28 3.2 17.4 0.5
29 5.6
30 36.4 58.0 1.3
31 80.0
32 48.6 19.0 1.8
33 9.0
34 8.0 0.2
*Values based on resin-free total acid, total alcohol, and total paraffin (de-
viations from 100 % ¼ sum of all other acids, alcohols, and hydrocarbons).
**Approximate.

Uses. Candelilla wax is used in classical cleaning preparations and


polishes (eg, as a component of shoe and other leather polishes) and in furniture,
car, and floor polishes. Because of its particular composition, this wax can be
used in the applications only in combination with other suitable natural and syn-
thetic waxes. Its use in floor polish is very limited because the high resin content
leads to increased dirt pickup.
Candelilla wax is also used in the production of candles, coatings for paper
and cardboard, hotmelt adhesives, and in polymer processing.
It is also approved for use in the cosmetics, pharmaceutical, and food indus-
tries. The main use of candelilla wax, principally in the United States and Eur-
ope, is by the cosmetics industry in the production of lipstick. Here, properties
such as outstanding framework formation with castor oil, gloss, and uniform
shrinkage on cooling (very good demolding after casting lipsticks) are very
important (8). Candelilla wax is also used as a polish for pills and in the produc-
tion of chewing gum and confectionery.
Economic Aspects. For nomadic peasants in the areas of Mexico men-
tioned, proceeds from the candelilla harvest were an important part of their
income for a long time. Because of continuing improvements in infrastructure,
the uneconomical isolation of candelilla wax has decreased since 1950 and
been replaced increasingly by agricultural production. Exports to the main con-
sumer, the United States, fell from 2500 t in 1950 (14) to 1800 t in 1974 (33),
1100 t in 1980, and ca. 310 t in 1992 (35). About 800–1000 t of candelilla wax
is available on the world market, of which ca. 200 t is consumed in Europe (8).
In the United States, demand for candelilla wax is increasing because
of the shift of use from classical areas to those of foods, cosmetics, and
pharmaceuticals (35). Privatization of candelilla wax isolation in Mexico and
100 WAXES Vol. 2

efforts to produce raffinates there will probably lead to greater availability of


these waxes.
Ouricury Wax. Occurrence and Isolation. Ouricury wax is deposited
on the leaf stems of feather palms (Syagros coronata), which grow mainly in the
Brazilian state of Bahia. There it is also known as urucury, licuri, aricuri, nicuri,
and coqueiro.
The isolation process for ouricury wax is very similar to that used for car-
naúba wax, but it is made more difficult by strong adhesion of the wax to the
plant, and mechanical scratching or scraping off is necessary. The annual yield
per palm is ca. 1 kg of crude wax.
Preliminary purification involves melting crude wax in hot water and filter-
ing the melt through filter cloths or suitable sieves (12,14,36). Additional puri-
fication in filter presses is used to obtain USA Pure Refined wax. The crude
wax is sold in lumps (Brazilian Crude).
Properties and Composition. Ouricury wax resembles carnaúba wax
very closely with regard to its hardness, gloss, solubility in polar and nonpolar
solvents, and hardening capacity when used in combination with soft waxes.
Its color is somewhat darker than that of dark carnaúba wax types. The mp is
82.5 8C. The most important properties for genuine commercial products have
been established (20). Thorough investigation of ouricury wax gave the following
composition (37):

Simple aliphatic esters 23.5 wt %


Monoesters of hydroxycarboxylic acids 22.4 wt %
Diesters of hydroxycarboxylic acids 17.2 wt %
Acidic hydroxypolyesters 5.4 wt %
Free acids 8.7 wt %
Free alcohols 3.0 wt %
Resins 14.8 wt %
Water 1.4 wt %
Ash 0.4 wt %

Alkaline saponification of ouricury wax yields varying quantities of total


acids (69–74 %) and unsaponifiable components (26–31 %), as usual in the case
of natural waxes. Despite these facts, ouricury wax, like other vegetable waxes,
can be characterized definitely by the chainlength distributions of its n-paraffin
and acids and alcohol components.
Uses and Importance. Ouricury wax was previously used in floor and
shoe polishes and in carbon paper coloring agents, instead of carnaúba wax,
mainly because of its low price. As a result of increasing price, the importance
of ouricury wax has decreased sharply. According to Brazilian statistics, 39 t
was exported in 1980. In 1991 the consumption in the United States was ca.
4.5 t/a (38).
Sugarcane Wax. Occurrence and Isolation. Sugarcane wax is formed
as a powdery, pale to dark yellow deposit on sugarcane stalks (particularly in
the thickened nodes) in quantities of 0.1–0.25 % based on cane weight, depend-
ing on country of origin. The main sugarcane-growing countries are India (ca.
96106 t/a), Brazil (ca. 80106 t/a), and Cuba (ca. 40106 t/a).
Vol. 2 WAXES 101

Sugarcane wax is generally extracted from the dried filter cake (ca. 1 % based
on sugarcane) (12) formed during sugarcane processing by using alcohols, carbon
disulfide, carbon tetrachloride, or aromatic and aliphatic hydrocarbons (39).
The crude wax content of the filter cake (in weight percent) varies
between the following limits, depending on sugarcane origin and method of
processing:

India 8.0–18.0 wt %
Louisiana 4.4–17.9 wt %
Puerto Rico 12.0–14.0 wt %
Hawaii 9.6–11.0 wt %
Philippines 10.5–11.0 wt %
Cuba 12.4–22.0 wt %
Java 5.0–15.0 wt %
Republic of South Africa 6.9–14.6 wt %
Argentina 11.9–15.5 wt %
Brazil 8.9–17.8 wt %

Crude wax can be freed from most glycerides (fats and oils) by treatment
with acetone, methyl ethyl ketone, methanol, ethanol, 1-propanol and isopropa-
nol, 1-butanol, or ethyl acetate. In a subsequent recrystallization step, preferably
in isopropanol, resinous and colored components can be removed. Further treat-
ment with bleaching earths or oxidative bleaching of the wax melt with air gives
waxes ready for use in a variety of applications. Bleaching processes for prepur-
ified wax using potassium chlorate–sulfuric acid or chromic acid–sulfuric acid
would be effective but are not used for cost reasons.
Properties and Composition. Crude sugarcane wax is a black-brown to
green-brown semisolid with an unpleasant, rancid odor.
Semi- and completely purified products are brown- to yellow-colored hard
waxes, whose characteristics vary considerably. Melting points of 68–81 8C,
acid numbers of 7–22 mg KOH/g saponification numbers of 32–65 mg KOH/g
and proportions of unsaponifiable material of 52–62 % have been published in
the literature. The following values are typical for a semiraffinate (12):

Wax esters (aliphatic and sterol esters) 78–82 wt %


Free fatty and wax acids ca. 14 wt %
Free alcohols (wax alcohols and sterols) 6–7 wt %
Hydrocarbons 3–5 wt %

On comparing crude wax with a raffinate sample (40) in terms of chain-


length distributions of the even-numbered acids and wax alcohols and of the par-
affins, characteristic differences can be seen, particularly with the acids.
Importance and Uses. From an estimated annual worldwide harvest of
ca. 510106 t of sugarcane, ca. 5.1106 t/a of dry filter cake material is theoreti-
cally obtained. From this, ca. 510 000 t of crude sugarcane wax could be
extracted, given an average wax content of ca. 10 % in the filter cake. If refining
losses are assumed to be 50 %, ca. 255 000 t of sugarcane wax raffinate per year
remains theoretically available worldwide.
102 WAXES Vol. 2

In the past, many attempts have been made to produce crude sugar-
cane wax on an industrial scale. As early as 1841, isolation of the wax from
filter cakes was begun, and in 1909 a corresponding extraction process was
patented.
Around 1916, crude sugarcane wax was extracted as a commercial product
in South Africa and Java. In 1958 the wax was produced in two Cuban and one
Australian plant. However, in 1960, production in Australia was halted and only
one plant was still operating in Cuba (39). In India, an extraction plant is
thought to have been operating continuously since 1950 (39). This plant is
attached to a sugar refinery with a daily throughput of 1200 t of sugarcane.
Difficulties resulting from the necessary centralization of production loca-
tions, the necessarily rapid and complete drying of large quantities of filter
cake material (which has a tendency to ferment), and last, but not least, the
increasing costs of extraction have made sugarcane wax raffinate incapable of
competing with carnaúba wax on the world market in terms of price. Conse-
quently, sugarcane wax raffinates have only regional importance.
In producer countries, crude sugarcane wax can be used in floor and shoe
polishes, carbon paper production, impregnating agents, and production of wax
compounds.
Retamo Wax. Occurrence and Isolation. In arid areas of Argentina,
Bulnesia retama grows wild as trees and shrubs. Branches and twigs are coated
with a wax deposit that can be collected mechanically after drying the plant
material harvested in summer. The crude wax is purified by melting with dilute
sulfuric acid and subsequent filtration through cloths. It is sold exclusively on
the home market as retamo wax (12).
Properties and Composition. Retamo wax is a light to medium brown,
odorless, hard wax. It is very similar to other vegetable waxes, such as carnaúba
wax, with regard to its solubility in polar and nonpolar organic solvents.
Data found in the literature, such as mp (76–78 8C), acid number (ca.
49 mg KOH/g), saponification number (ca. 87 mg KOH/g), proportion of hydro-
carbons (15–25 %), and content of unsaponifiable components (38–42 %), char-
acterize the wax less precisely than the typical chain-length distributions of
even-numbered wax acids and alcohols and of paraffins (40).
Uses and Importance. With a decreasing annual production of several
hundred tonnes (1975, ca. 500 t/a), retamo wax is of limited importance on the
Argentinian market. It is used for floor, car, and shoe polishes and in the produc-
tion of colored polishing inks.
Jojoba Oil. According to the DGF definition, jojoba oil is not a wax. How-
ever, its chemical composition allows it to be classified as a ‘‘liquid wax’’ in the
sense of the classical definition of waxes.
Occurrence and Isolation. The jojoba bush (Simmondsia chinensis ¼
Simmondsia californica = Buxus chinensis) grows wild over an area of ca.
250 000 km2, most of which lies in the Sonora desert. The population density
varies from a few to 500 plants per hectare. In certain areas, millions of jojoba
bushes can be found.
The bushes carry fruit capsules that can contain one to three nuts of differ-
ing sizes. The yield from the harvest of wild jojoba bushes can range from a few
nuts to 14 kg of nuts per plant. The average yield is ca. 1.8 kg per bush.
Vol. 2 WAXES 103

The nuts contain 50–60 wt % oil, which can be extracted almost completely
from the deshelled, granulated fruit. Around 35–40 % of a particularly high-
value oil is obtained by cold pressing and a further 15–20 % of a lower-value
oil by extraction of the pressed cake. The oil is purified by simple filtration (41).
Properties and Composition. Jojoba oil is a virtually colorless to
golden yellow, odorless, unsaturated oil, for which the most important character-
istics are as follows:
mp 6.8–7.0 8C
Acid number 2 mg KOH/g
Saponification number 92 mg KOH/g
Iodine number 82 g I2/100 g

A highly crystalline, relatively hard wax is obtained from jojoba oil by


hydrogenation. Saponification of the oil gives 51 % unsaponifiable material and
52 % total acid.
Jojoba oil has the following composition with regard to substance groups:
Wax esters ca. 97.0 wt %
Free alcohols ca. 1.1 wt %
Free acids ca. 1.0 wt %
Sterols ca. 0.9 wt %

Unlike other vegetable waxes, the acids and alcohols in jojoba oil are mono-
unsaturated, straight-chain representatives of the chain-length range C16 –C24.
Among the acids the C18 (ca. 12.5 % of the total acids), C20 (ca. 67 % of the total),
and C22 components (ca. 14.5 % of the total) predominate, and among the alco-
hols, the C20 (ca. 48 % of the unsaponifiable components) and C22 components
(ca. 41 % of the unsaponifiable components). With this unusual composition,
jojoba oil is very similar to sperm oil contained in the skull a bone of the
sperm whale.
Uses and Importance. An extraordinarily high market demand exists
for a sperm oil substitute, which can be satisfied to a great extent by jojoba oil.
This demand is a result of the worldwide prohibition of whaling, even though the
latter is not obeyed by all countries.
The first experimental jojoba plantations were begun in the United States
(Arizona, Texas, and California) to effect infrastructural improvements. Many
other jojoba oil plantations were started in Mexico, Argentina, Australia, Brazil,
Africa, Egypt, Israel, and countries whose climatic conditions allowed desert
farming.
Because management of these plantations is difficult and the time required
to achieve economical harvests is relatively long (eight to ten years), jojoba oil
is available in insufficient quantities at a very high price. Its use is therefore
limited to cosmetics and pharmaceuticals. As soon as reliable quantities can be
supplied at acceptable prices, jojoba oil may attract extraordinarily great interest
as a raw material, particularly in the production of high-performance lubricants
and in modified form in plasticizers and stabilizers. Hydrogenated jojoba oil,
which corresponds to fatty esters such as stearyl stearate in its properties, will
remain noncompetitive with these esters in terms of price and will therefore not
achieve market importance.
104 WAXES Vol. 2

Other Vegetable Waxes. Many other vegetable waxes are known, and
the composition of some of them has been investigated (12). However, they are
either of very limited local importance or of purely academic interest. These
waxes include:

Caranday wax
Raffia wax
Columbia wax
Esparto wax
Alfalfa wax
Bamboo wax
Hemp wax
Douglas fir wax
Cork wax
Sisal wax
Flax wax
Cotton wax
Dammar wax
Cereal wax
Tea wax
Coffee wax
Rice wax
Ocatilla wax
Oleander wax
* Palm waxes

Japan tallow wax, which is available commercially, fulfills some applica-


tion-oriented criteria as a wax but, because of its chemical nature, is a vegetable
tallow and thus belongs to the group of other tallows such as myrica, bayberry,
ucuhuba, Borneo, Malabar, and illipe.
2.3. Animal Waxes. Animal waxes are obtained mainly from bees (Apoi-
dae), particularly the honeybee (Apidae), stingless bee (Meliponinae), and bum-
blebee (Bombinae). Some scale insects (Coccidae) also produce waxes.
Waxes obtained from land mammals (sheep, goats, llamas, and dromed-
aries) and marine animals, as well as those of microbiological origin, are also
known.
Beeswax. Beeswax is the oldest natural wax used by humans and still
the most important animal wax.
Occurrence. Beeswax is an end product of the metabolism of a honeybee
class (Apis mellifica, A. carnica), which belongs to the Apis genus. This species of
honeybee is found worldwide and domesticated to a high degree. Wax is secreted
from the wax glands in the lowest abdominal segments of the worker bee as
white, slightly transparent flakes (virgin wax). From there it is taken to jaw,
plasticized by chewing, and then used in this form as a building material for
incubating combs and honeycombs.
Vol. 2 WAXES 105

If plenty of food is available and honey production is increasing, the wax


glands are stimulated to enhance production for building new honeycombs. If
the beekeeper makes preformed honeycombs available to the bees, honey produc-
tion is increased (13,36,42,43).
A bee population of ca. 30 000 worker bees produces ca. 270 g of wax from
3 kg of sugar or 3.6 kg of honey in one-fifth of a generation – ca. 50 d. A bee popu-
lation consumes ca. 50 kg of pollen and 62 kg of honey per year.
Beeswax is produced in almost all countries of the world. In Europe, wax
produced in a particular country is consumed there and supplemented by
imports. Suppliers are African countries, such as Tunisia, Morocco, Kenya, Tan-
zania, Zambia, and the Central African Republic. North America, Australia, New
Zealand, Russia, and the People’s Republic of China also export beeswax.
Isolation. To obtain wax, the combs are freed from honey as far as possi-
ble by centrifuging. Then the wax is isolated either thermally by using sunlight
or with boiling water. Particularly heavy impurities become enriched in the
aqueous phase. Insoluble impurities remaining in the wax melt can be removed
by filtration through a fine mesh sieve of filter cloths, or by pressure filtration or
centrifuging.
Colored, wax-soluble impurities can often be removed by treating the wax
melt with activated carbon, aluminum or magnesium silicates, or other bleaching
earths. As in pressure filtration and centrifugation, wax-containing residues are
formed from which lower-quality waxes can be isolated by solvent extraction.
The oldest method of lightening the color, which is said to have been practiced
in 1000 B.C., involves bleaching beeswax, that has been processed to thin strips
or shavings (large surface), in air and sunshine.
In the aforementioned purification processes, beeswax remains chemically
unchanged and its characteristic, honeylike odor is retained. The success of the
process depends mainly on the source of the beeswax.
Posttreatment of beeswax is now limited to washing with water or adsorp-
tive purification with bleaching or diatomaceous earths. Bleaching with hydro-
gen peroxide is used out because peroxy compounds are formed (8).
If the process is carried out skillfully, chemical bleaching processes, such as
refining with chromic acid–sulfuric acid, give very light, almost chemically
unchanged products, which have lost the aromatic odor of the starting material.
However, these processes are not employed. The same holds true for those invol-
ving mild bleaching with permanganate and sulfuric acid or the use of chlorine-
containing bleaching agents such as chlorine bleach, sodium chlorite, and chlor-
amine. The latter give waxes that do not have stable colors and often contain
chlorine.
Beeswax is sometimes derivatized by ethoxylation or esterification. The
products obtained do not have any market importance.
Properties. Crude, mechanically purified beeswax contains, besides wax,
flower and pollen pigments and propolis resin, which acts as a cement in honey-
comb construction. Therefore, depending on the source, the wax can have a yel-
low, orange, or dark brown color.
Beeswax is moderately hard. It becomes plastic and kneadable on warming
in the hands, without sticking. It has a noncrystalline, finely grained fracture pat-
tern. On cutting, matt surfaces are produced. Beeswax has a honeylike odor that
106 WAXES Vol. 2

intensifies on melting. It is moderately soluble in polar and nonpolar organic sol-


vents in the cold, and completely soluble when heated to its boiling point.
Chemically bleached beeswaxes are cream to ivory in color and practically
odorless and tasteless. They have a somewhat harder and more brittle consis-
tency than chemically untreated products.
The following limiting values have been established for the properties of
mechanically purified and bleached beeswax (20):

mp 62–65 8C
Acid number 17–24 mg KOH/g
Ester number 72–79 mg KOH/g
Ester number/acid number 3.3–4.2
Saponification turbidity point 65 8C

Extensive investigations show that properties vary even among waxes from the
same source (44).
Composition. The following values are given in the literature for the
approximate composition of beeswax with regard to wax classes:

Wax esters 70–80 %


Free acids 10–15 %
Paraffins 10–20 %

Alkenes, isoparaffins, cholesteryl esters, and pollen pigments are present in


minor amounts (12,45).
Since the typical data for beeswax vary between relatively wide limits, mis-
representation of beeswax is difficult to detect by using these data (46). A large
number of publications exist on the analysis of beeswax (47). Beeswax can be
separated into the various substance classes by chromatographic methods
(48–53). Beeswax of different origin has been shown to have similar composition.
Comparative GC studies, in which ca. 65 % of the waxes investigated are
analyzed (residues mostly involatile di- and triesters), also show great similarity
among beeswax of widely varying origin (54) (see Table 5).

Table 5. Composition of Beeswax of Different Origins* (54)


Eastern Czech
Components Switzerland Brazil China Canada Republic
Hydrocarbons, % 14 15 16 14 13
Fatty (wax) acids, % 12 14 9 (8) – 12 13
Fatty alcohols, % 1 1 1 1
Palmitates, % 22 20 24
35 ** 31 **
Oleates, % 4 3 5
Hydroxymonoesters, % 12 12 11 (4) – 8 13
Total, % 65 65 66 (62) – 70 70
*Values in parantheses were found only in a few cases.
**Palmitates and oleates.
Vol. 2 WAXES 107

Table 6. Percentages of Individual Substance Classed in a Hydrolyzed


Wax Sample from the Honeybee Apis Mellifica (Switzerland) (54)
Chain length Fatty Hydroxy Fatty
(C-number) Hydrocarbons acids fatty acids alcohols
16 17.98 11.60
17
18 2.44 0.71
19
20 0.15
21
22 0.33 1.66 0.12
23 0.25
24 0.10 4.94 0.59 5.59
25 0.58
26 0.11 1.75 1.38 3.37
27 4.42
28 0.08 1.78 0.10 4.37
29 2.64
30 0.07 1.58 9.58
31 3.27
32 1.05 7.98
33 2.24
34 0.94 0.88
35 0.12
36 0.08
Total 14.23 33.02 16.04 31.89

Parallel GC investigations on a genuine beeswax sample and the same


sample after hydrolysis and silylation show that C22 – C36 wax acids are
present predominantly as the free acid, while all C12 – C20 fatty acids are ester-
ified (54).
Gas chromatography of hydrolyzed and silylated beeswax allows a total
analysis in which 95 % of its components can be identified (54) (Table 6).
Studies using GC show that a spectrum of wax samples from Apis mellifica
has the same characteristic composition independent of the breed of bee. Small
differences exist only in the percentage of individual components and substance
classes, so that in general the following ranges of composition are correct for
various classes of substance (54):

Hydrocarbons 13–16 %
Fatty acids 30–35 %
Palmitic acid 15–20 %
Hydroxy fatty acids 13–17 %
Fatty alcohols 30–35 %
Other substances 4–8 %

The remaining substances are present mainly in concentrations of <0.05 %.


By using these facts, beeswax misrepresentation can be detected by a cor-
responding total analysis of the wax even in the presence of small quantities of
blending materials (54).
108 WAXES Vol. 2

Uses. Beeswax, mainly the yellow variety, is one of the raw materials
used in the production of household candles, expensive decorative candles, and
candles for religious purposes. For the latter, the Roman Catholic Church pre-
scribes a minimum beeswax content of 10 %. Beeswax is also used for artistic
and handicraft purposes (wax flowers, wax sculptures, painting, and batik).
Most bleached and purified beeswax is used in cosmetics and pharmaceuticals
(to regulate the consistency of lipsticks, creams, ointments, and suppositories)
and in the food sector (as a release agent and a polish in confectionery produc-
tion). Its use in these areas is favored by corresponding licensing regulations.
A significant proportion of beeswax, ca. 40 % of the total market volume, is
sold back to beekeepers for the production of synthetic honeycombs, in which
blends with paraffins and ceresin, which are not allowed for pharmaceuticals,
are often used.
Economic Aspects. From the worldwide honey production figures for
1976, production of 11 000–19 000 t of beeswax was estimated. Whether the
quantities used in the candle and beekeeping industries were included in the
worldwide consumption figure of ca. 10 000 t for 1981, is not known. At that
time, total imports for the Federal Republic of Germany amounted to 1000 t.
At present (1994), ca. 15 000 t of beeswax is available on the world market.
Ca. 6000 t is used in industry, and ca. 4000 t is sold back to beekeepers (8).
The annual availability of beeswax depends strongly on climatic vari-
ations in exporter countries and bee diseases that break out occasionally. In
1993, for example, a marked drop in exports from the United States occurred
because of catastrophic weather conditions in the Mississippi area (8).
Other Insect Waxes. The following waxes are not market products and
only in some cases have limited regional importance in the countries of origin.
Ghedda Wax. Ghedda wax is produced by some types of wild bee bound
in Asia (Apis indica, A. florea, A. dorsata). It was formerly exported as such but
more frequently now is blended with beeswax from the East Indies.
Ghedda wax has a yellow-brown color and varying characteristic data (eg,
acid number 4–11 mg KOH/g, saponification number 86–130 mg KOH/g, iodine
number 5–11 g I2/100 g, mp 60–66 8C). It has a consistency similar to beeswax
with a fatty feel, which can be attributed to its glyceride content. Ghedda wax
consists of ca. 78–80 % wax esters, 4 % glycerides (fats), 5–6 % free acids,
8–9 % hydrocarbons, and 1–2 % partly resinous impurities (12).
Among the acids are 7- and 16-hydroxypalmitic acids and 7,16-dihydroxy-
margaric acid, which can be used for identification of ghedda wax or other bees-
waxes blended with ghedda wax in modern umpire assays.
Shellac Wax. The resinous exudate of the scale insect Laccifer lacca
(formerly Tachardia lacca) of the Coccidae family is an important commercial
product, known as shellac. Shellac contains 4–5 % wax, which can be obtained
by treatment with dilute soda solution or 90–95 % ethanol (spirit extraction)
as an insoluble component.
Shellac wax is a hard, yellow to brown product. Very light, bleached types
are also known. The widely varying data found in the literature (eg, acid number
12–24 mg KOH/g, saponification number 63–126 mg KOH/g, iodine number
6–9 g I2/100 g, solidifying point 58–80 8C) may be attributed to the different
sources of shellac and different methods of extraction and posttreatment of the
Vol. 2 WAXES 109

wax component. Thus, the composition with regard to substance classes also var-
ies as follows (12):

Wax esters 70–82 %


Free acids 10–24 %
Free alcohols <1 %
Hydrocarbons 1–6 %
Shellac 1–4 %

Investigations (55) of bleached shellac waxes show that the total acid con-
tent of the wax, similar to that of beeswax, consists essentially of a mixture of
even-numbered fatty acids (C12 –C18, 21–26 % of the total acid) and waxy acids
(C28 –C34, mainly C32 and C34). A significant component is trihydroxypalmitic
acid (aleuritic acid), even though it occurs only in quantities of 0.5–1.0 %.
Among the alcohols (C28 –C32) the C28 component predominates (62–65 % of
the total alcohol). The main hydrocarbon components are paraffins with 27, 29,
and 31 carbon atoms.
Chinese Insect Wax. The East Asian wax scale insect (Ericerus pela,
subgenus Lecaniinae, family Coccidae) secretes a wax on the twigs and leaves
of trees and shrubs on which it is a parasite. The wax can be extracted by melting
with hot water. It is hard, brittle, crystalline, and odorless, with a light color, and
consists of 95–97 % wax esters, 0.5–1.0 % free acids, and 2–3 % hydrocarbons
(12). The high proportion of wax esters is significant. The East Indian wax
scale insect (Ceroplastes ceriferus) is said to produce a wax similar to beeswax
in composition (12).
Wool Wax. Sheep secrete a substance from their sebaceous glands to pro-
tect the epidermis and the wool from adverse weather conditions and from their
own acidic perspiration. This substance is incorrectly referred to in the literature
as wool fat. Because of its chemical composition it should definitely be classified
as a wax.
Wool wax is obtained in emulsified form from raw wool by machine washing
with soap liquor. Extraction of wool with suitable organic solvents is not
employed because it would lead to complete degreasing of the wool and thus
make it nonelastic.
Many processes exist for obtaining crude wool wax from the emulsion, some
of them patented (36). Wax is first precipitated along with the acids of the soap
solution by addition of sulfuric acid. The fatty acids remain in the crude product
regardless of whether the precipitates are pressed out, centrifuged, or extracted
with solvents. They can be removed almost completely by dissolving the crude
product (eg, in hexane) and subsequent extraction with a mixture of alcohol
and dilute aqueous sodium hydroxide. Further purification steps (eg, bleaching
with hydrogen peroxide in the presence of phosphoric acid (56)) give neutral wool
wax, which is usually treated subsequently with adsorbents such as bleaching
earth or activated carbon and sold in a highly pure form as Adeps lanae.
Crude wool wax is a greasy, glutinuous, brown-yellow to brown-black sub-
stance with a penetrating goatlike odor (mp 34–38 8C). Neutral wool wax is yel-
low to light brown in color (mp 38–42 8C), with a milder odor, whereas Adeps
lanae is a pale yellow, almost odorless substance (mp 40–42 8C).
110 WAXES Vol. 2

The composition of wool wax varies depending on the origin of the wool,
breed of sheep, climatic conditions, farming method, and type of pasture.
Wool wax usually consists of a mixture of ca. 48 % wax esters, 33 % sterol
esters, 6 % free sterols, 3.5 % free acids, 6 % lactones, and 1–2 % hydrocarbons
(12). The low mp and appearance of wool wax may be attributed to the high pro-
portion of branched acids.
Uses. Wool wax is used mainly as the crude product, as neutral wool wax,
or as hydrogenated neutral wool wax and in the form of wool wax alcohols
(obtainable by hydrolysis) in leather polishes, cosmetics (eg, creams, baby-care
products, and toilet soap), pharmaceuticals (eg, plasters, ointments, and suppo-
sitories), and lubricants. The purified, commercial product is known as wool wax
or lanolin.

3. Montan Wax

3.1. Formation and Occurrence. Montan wax is a vegetable fossil


wax. It forms part of the extractable, bituminous components of lignite and
peat. The composition of bitumen depends on the carbonized plant material
and on the geology and conditions of the deposit. The main components of bitu-
men are waxes, resins, asphaltenes, and dark residues. If the proportion of wax
is >60 %, the bitumen is generally known as crude montan wax. Prerequisites for
formation of this wax were the same or similar to those now necessary for forma-
tion of recent natural waxes (eg, carnaúba wax (57–60)). Lignite, which origi-
nates from plant material of the Eocene period, has a high and economically
utilizable proportion of wax components. Deposits of lignite, which are used for
wax extraction, are found mainly in eastern Germany (Röblingen), Ukraine
(Alexandrija), Russia (Baschkiren), the United States (California), and China.
Other deposits are of only local importance or are not even exploited. The largest
crude montan wax producers are the Montanwerke Romonta in Röblingen (Ger-
many) and the American Lignite Co. in the United States.
3.2. Extraction. The most economically important bitumen-rich lignite
deposits are found in the area around the city of Halle in Germany. An industrial
process for extracting montan wax from this type of coal was based on investiga-
tions bei RIEBECK. The original patent (61,62) for this process, published in 1880,
describes how montan wax can be obtained by solvent extraction. In 1900 the
production of montan wax was begun in Völpke, and later in other locations.
Romonta in Amsdorf (62) is currently the most important production site.
There, lignite with a bitumen content between 10 and 20 % is used for industrial
extraction of montan wax. The process used involves the following steps: size
reduction, drying, extraction, evaporation of the extract, and formulation
(57,60). The yield and composition of the extract are determined by coal quality
(clay and mineral content); physical parameters such as water content, particle
size, and particle-size distribution; and properties of the solvent.
Granulating and Drying. Run-of-mine, pit-wet lignite contains ca. 50 %
water. Before extraction it is ground in roll mills and dried to a residual moisture
level of ca. 15 % in steam heated tubular dryers. The particle-size distribution is
ca. 7 % >8 mm, 75 % 1–8 mm, and 18 % fines.
Vol. 2 WAXES 111

Lignite Extraction and Wax Production. Various machines such as


band, chamber, carousel, or bucket chain conveyor extractors are used for extrac-
tion. Many solvents and solvent mixtures have been tested and used to extract
wax from lignite. They include aliphatic, aromatic, and unsaturated hydrocar-
bons and hydrocarbon mixtures, mono- and polyhydric alcohols, ketones, esters,
ethers, and mixtures of hydrocarbons and alcohols. The yield of extract is highest
when polar solvents (eg, ethanol) or solvent mixtures (eg, ethanol–toluene) are
used. However, the extract then contains a high proportion of polar substances,
such as resin acids, sterols, triterpenes, or dark residues, and only a small pro-
portion of wax. With hydrocarbons the overall yield is lower, but the proportion
of wax in the extract is significantly higher. In industry, only hydrocarbons or
hydrocarbon mixtures are therefore used (eg, toluene at Romonta in Röblingen).
Mixtures of aromatic and aliphatic hydrocarbons and naphthenic benzine are
used by other producers.
If toluene is used, extraction is carried out at 85 8C. After extraction, the
solvent contains ca. 8–10 % dissolved crude montan wax. The mixture is worked
up in a three-stage distillation. The bitumen content is increased to 50 % in a
continuous evaporator and then to 90 % in a circulation evaporator. The remain-
ing solvent is removed by blowing in steam directly or in a falling-film vaporizer.
The wax melt is pelletized or poured into molds. The extracted coal granulates
are freed from solvent by steam stripping and used for electricity production in
a power station. The toluene–water mixture is worked up, and toluene is
recycled to the extraction process.
3.3. Properties and Composition. Crude montan wax is a black-brown,
hard, brittle product with a conchoidal fracture pattern. It cannot be scratched
by fingernails at room temperature. Crude montan wax has a penetrating
terpene-like odor, which becomes very intense on melting. The wax is soluble
in many organic solvents, particularly aromatic or chlorinated hydrocarbons,
even on moderate heating. Crude montan wax consists of a mixture of wax
acids, wax esters, resins, asphaltenes, and dark residues. The ester and acid com-
ponents can be saponified and emulsified by treatment with aqueous alkali
(57,59,60). The waxes form the crystalline component of montan wax. They effect
the relatively sharp transition from solid to liquid state and vice versa. The
resins and asphaltenes have a more amorphous nature and exhibit glassy beha-
vior. The qualitative and quantitative composition of crude montan wax is deter-
mined by the carbonized plants, extent of carbonization, the solvent, and
extraction conditions.
A typical crude montan wax contains ca.:

Wax acids 35 %
Wax alcohols 20 %
Hydroxycarboxylic acids 10 %
Dicarboxylic acids 3%
Wax ketones 1%
Hydrocarbons 2%
Resin acids 15 %
Sterols 10 %
Dark residues 4%
112 WAXES Vol. 2

Characteristic properties are :

Acid number 20–40 mg KOH/g


Saponification number 70–120 mg KOH/g
Iodine number 30–45 g I2/100 g
Drop point ca. 85 8C
Density (100 8C) ca. 0.86 g/cm3
Thermal expansion (20–100 8C) 18 %

Wax Components. Crude montan wax contains 70–75 % waxlike com-


ponents, of which the following have been isolated:

Straight-chain carboxylic acids C16 –C34


Straight-chain alcohols C18 –C34
Straight-chain o-hydroxycarboxylic acids C16 –C32
Straight-chain 2-hydroxycarboxylic acids C16 –C32
Straight-chain o-dicarboxylic acids C22 –C32
Esters of C16 –C34 carboxylic acids and C18 –C34
alcohols
Straight-chain aliphatic hydrocarbons C25 –C33

These compounds form homologous series, in which chain lengths in the


range of C28 –C30 predominate. The chain-length distribution for straight-chain
carboxylic acids and alcohols found by GC analysis is shown in Figure 2.
Resins and Dark Residues. Besides the wax components, crude
montan wax also contains ca. 25 –30 % resins and dark residues. The quantity
and composition of these components are determined mainly by the properties
of the lignite, the extraction solvent, and the extraction temperature. The resin
can be removed from crude montan wax by exhaustive extraction with metha-
nol. In industry, this method is not used to determine resin content because it
is expensive and time-consuming. Instead, the proportion of soluble com-
pounds (ie, resins) is determined by a standardized procedure such as solu-
bility in acetone or in a toluene– ethanol or benzene– ethanol mixture. The
resin is characterized by its chemical and physical properties, of which the

Fig. 2. Chain-length distribution of acids and alcohols in montan wax.


Vol. 2 WAXES 113

following are typical:

Drop point ca. 75 8C


Acid number ca. 33 mg KOH/g
Saponification number ca. 75 mg KOH/g
OH number ca. 70 mg KOH/g
Iodine number ca. 90 g I2/100 g
Sulfur content ca. 2–2.5 %

Determination of all individual components in resin extracts has not yet


proved possible. However, substance classes and typical compounds in these
classes have been identified: eg, triterpenes such as betulinol, sterols, diterpe-
noid acids and their esters, and diterpenoid alcohols and their esters.
3.4. Refining and Derivatization. Because of its dark color and its
resin content, which sometimes lead to problems, direct use of crude montan
wax is limited. For most applications the wax must be refined. Refining involves
three steps: extractive deresinification, oxidative bleaching, and subsequent deri-
vatization (57–60,63).
Deresinification. The resin components of the crude wax are soluble in
many solvents even at low temperature (10–20 8C). Treatment of pulverized
crude wax with methylene chloride, ethyl acetate, or methanol reduces the
resin content to 5–10 %. Other solvents, such as ketones, ethers, hydrocarbons,
mixtures of alcohols and aromatics, liquid SO2 or supercritical CO2, can be used
but have not achieved industrial importance. After removal of the solvent, the
extracted resin contains ca. 10–20 % wax. Deresinified wax can be used directly
or as a mixture with other wax components for various applications. Such pro-
ducts are sold by Romonta, Amsdorf, and the American Lignite Co. Most deresi-
nified crude wax is subjected to oxidative bleaching.
Bleaching. A purification process for the crude wax was sought very early
on. Of the many processes investigated, only three have actually been used in
production (57,60,64–67).
Distillation was the first method for obtaining light, hard waxes from
crude montan wax. The crude wax was heated to 420 8C in stills and thus par-
tially cracked. The distillate therefore consisted of 50–80 % hydrocarbons and
20–50 % wax acids. The hydrocarbons were removed in a further workup step,
leaving the montan wax acids.
Oxidation with Nitric and Sulfuric Acids. In this process, paraffin
had to be added as a diluting agent. The resin was precipitated as the nitroresin
and removed. The dark residues were coked and the wax acids extracted from
this coke. However, the wax still had a certain amount of dark residues and
nitrate-containing products.
Oxidation with Chromic Acid or Chromates in Sulfuric Acid. In
processes described thus far, the wax is damaged considerably and the chainlength
distributions of its constituents are altered. Oxidative bleaching with aqueous chro-
mic–sulfuric acid mixtures, developed by I.G. Farben, is much more suitable. The
starting material is deresinified crude montan wax. The resin components and
dark residues are oxidized to carbon dioxide or low molar mass, water-soluble
compounds in a multistage oxidation process. The wax esters are simultaneously
114 WAXES Vol. 2

hydrolyzed; the wax alcohols formed are oxidized to wax acids; and the hydroxycar-
boxylic acids are oxidized to dicarboxylic acids. Some esters remain unaltered.

where n is an even number between 24 and 32.


A mixture of light wax acids, which are very similar in chain-length distri-
bution to the crude wax, is obtained in a yield of 80–85 % based on crude wax.
Minor components of the mixture are unchanged esters and long-chain dicar-
boxylic acids. The spent oxidizing agent is removed and worked up, and the
bleached wax is freed from residual chromium salts in a subsequent step.
Depending on the quantity of chromic acid, the acid number of the bleached
wax is between 100 and 135 mg KOH/g. Oxidation can be performed batchwise or
continuously. In the Gersthofen process used by Hoechst, the oxidizing agent is
regenerated electrochemically. BASF, Völpke, and other producers use the spent
oxidizing agent in an associated plant for the production of chromium-containing
tanning agents.
Development is still being carried out on refining processes (68) such as alka-
line dehydrogenating oxidation with anhydrous alkali hydroxide. Apparently,
these processes have not been used in industry up to now. An overview of the cur-
rent processes used for extraction and refining of montan wax is given in Figure 3.

Fig. 3. Flow sheet for extraction and refining of montan waxes.


Vol. 2 WAXES 115

Derivatization. Reconstruction of the original wax structure is attempted


by treating the wax acids obtained with dihydric alcohols or Ca(OH)2. By varying
the proportions of free acid and soap and the degree of esterification, the reaction
can be carried out in such a way that the products formed are optimized for a
particular application (57,64,65,69).

CH3
ðCH2 Þn
COOH þ HOCH2
CH2 OH ! CH3
ðCH2 Þn
COO
CH2
CH2
OR

R
H or CH3
ðCH2 Þn
CO


CH3
ðCH2 Þn
COOH þ CaðOHÞ2 ! ðCH3 COOÞ2 Ca2þ
ðCH2 Þn

The most important products in this series are the esters of ethylene glycol
(eg, Hoechst Wax E or Hoechst Wax KSL) and mixtures of esters of 1,3-butane-
diol with calcium montanate (eg, Hoechst Wax OP). Broad variation in properties
can be achieved by using tri- and polyhydric alcohols, aromatic alcohols, aminoal-
cohols, and ethoxylated or fluorinated alcohols. Reactive waxes can be produced
from partial esters of polyols by treatment with dicarboxylic acids, isocyanates,
epoxides, or unsaturated carboxylic acids.

These waxes exhibit interesting properties and have been produced and
used for special applications. If the reaction is not stopped at the reactive wax
stage and a suitable stoichiometry is chosen, oligomeric waxes are produced.
3.5. Uses and Economic Aspects. Montan wax acid and its deriva-
tives are versatile materials (64,65,69–71). They are used in the form of flakes,
powders, pastes with solvents, or aqueous emulsions. The main areas of use are
in polishes for floors, cars, and leather; in plastics processing as lubricants,
release agents, and nucleating agents; in the paper and building industries;
and in wood and metal processing. Montan wax derivatives are also used in
small quantities in pharmaceutical formulations; cosmetics; and the production
of adhesives, resins, and office equipment. In addition to their technical proper-
ties, the low toxicity of montan wax derivatives and the fact that BGA or FDA
approval has already been granted for some derivatives are important criteria
favoring their use.
Production figure of montan wax for 1994 are given below, together with a
breakdown of montan wax consumption in different application areas:

Crude montan wax


Romonta ca. 19 000 t/a
Alpco ca. 2 500 t/a
Ukraine ca. 1 500 t/a
China ca. 1 000 t/a
116 WAXES Vol. 2
Bleached montan waxes
Hoechst Gersthofen ca. 11 000 t/a
Völpker Montanwachs ca. 3 500 t/a
China ca. 500 t/a
Poland <500 t/a
Czechia <500 t/a
Application areas
Plastics, lubricant ca. 7 000 t/a
Polishes, coatings ca. 5 500 t/a
Technical applications ca. 2 500 t/a

4. Petroleum Waxes

4.1. Introduction. The quantity of waxes obtained from crude petroleum


has increased continuously for two reasons: (1) the demand for lubricating oils
with low pour points and (2) the large proportion of paraffinic crudes in total
crude oil production that have to be dewaxed for the production of lubricating
oils. This development has increased the need to find more industrial applica-
tions for petroleum waxes. When the processing of lubricating oil began in the
second half of the 19th century, waxes were still inconvenient side products.
Since then, worldwide consumption of petroleum waxes has increased to
3106 t/a. This consumption corresponds to more than 95 % of the waxes of all
types produced worldwide. Depending on their natural occurrence and their
crystallinity, petroleum waxes are divided into:

1) Macrocrystalline waxes (paraffin waxes)


2) Microcrystalline waxes (microwaxes)

Table 7 gives a detailed classification of paraffin and microcrystalline waxes


based on origin and method of refining (80).
Paraffin waxes are obtained from light and middle lubricating oil cuts of
vacuum distillation. Paraffin waxes also include waxes from heavy lubricating
oil distillates, which are intermediates between macrocrystalline and microcrys-
talline waxes with regard to structure and composition (intermediate waxes).
Microcrystalline waxes originate from vacuum residues and from the
sediments of paraffinic crude oil (settling waxes). Waxes that are liquid at room
temperature are mostly contained in diesel oil or gas oil fractions (! Heating Oils,
! Aviation Turbine Fuels) and can be isolated from them. These are not dealt
with here.
4.2. Macrocrystalline Waxes (Paraffin Waxes). Chemical Composition
and General Properties. Chemical Composition. Paraffin waxes consist
predominantly of mixtures of straight-chain alkanes in a typical distribution of
the homologous series whose molar masses depend on the boiling range of the
lubricating oil distillate from which they are obtained. Long-chain, weakly
branched isoalkanes are present in a much lower proportion, along with a very
small fraction of monocyclic alkanes.
The intermediate waxes have a similar composition, but the molar masses
of the n- and isoalkanes are higher. Intermediate waxes contain a higher
Vol. 2 WAXES 117

Table 7. Classification of Waxes from Crude Petroleum


Origin
light, medium, heavy residues from crude oil
lubricating oil vacuum distillation
distillates
Group (macrocrystalline) microcrystalline settling
paraffin waxes waxes (microwaxes) waxes
Subgroup paraffin intermediate residue waxes pipe waxes
waxes waxes
tank bottom
waxes
Crude products crude waxes (slack waxes) petrolatum raw waxes
Deoiled and scale waxes deoiled slack plastic microwaxes
refined products waxes filtered hard microwaxes
(decolorized)
waxes fully
refined waxes
Side products soft waxes soft petrolatum
from deoiling (microwax slacks)

proportion of cycloalkanes and isoalkanes; the latter more strongly branched


than those in paraffin waxes.
According to the European Wax Federation (EWF), paraffin waxes have a
C-number distribution of n-alkanes from 18 to 45 and a total content of iso-
and cycloalkanes of 0 – 40 %. Typical data for the intermediate waxes are an
n-alkane C-number of 22 to 60 and a total content of iso- and cycloparaffins
of 30 – 60 %.
General Physical Properties. Paraffin waxes are insoluble in water
and sparingly soluble in low molar mass aliphatic alcohols and ethers. They are
more soluble in ketones, chlorohydrocarbons, petroleum spirit, solvent naphtha,
benzene, toluene, xylene, and higher aromatics, especially at elevated tempera-
ture. The solubility decreases markedly with increasing molar mass (higher
melting point) of the waxes.
Chemical Properties. Paraffin waxes are extremely unreactive under
normal conditions. Oxidation reactions occur only at elevated temperatures
(eg, on storage and processing above 100 8C), particularly in the presence of
oxygen and catalytically active metals. These reactions can be recognized
from the burnt odor produced and the yellow to brown coloration of the
waxes. Nevertheless, under certain thermally and catalytically controlled
conditions, these waxes can undergo chemical reactions such as chlorination,
oxidation, dehydrogenation, and cracking, of which chlorination and cracking
are important in industry.
Division into Product Classes. Depending on the degree of refining, par-
affin waxes are divided into the following product classes:

1) Crude waxes, also known as slack waxes


2) Slack wax raffinates (scale waxes)
118 WAXES Vol. 2

Table 8. Variation in Physical Data with Degree of Refining of Paraffin Wax (Starting
Material : Slack Wax from a Medium Machine Oil)
Slack Deoiled Filtered Fully
Characteristics wax wax (decolorized) wax refined wax
Congealing point, 8C 59 62 62 62
Needle penetration, 0.1 mm
at 25 8C 57 18 18 17
at 30 8C 80 26 26 25
at 35 8C 105 36 36 36
Oil content, % 9.9 0.5 0.5 0.4
Viscosity (at 100 8C), mm2/s 6.8 6.1 6.0 5.8
Color brown brown whitish white
Fluorescence very strong weak none

3) Deoiled slack waxes


4) Soft waxes
5) Semirefined waxes
6) Filtered (decolorized) waxes
7) Fully refined waxes

Table 8 gives an overview of the change in physical characteristics with


degree of refining.
Crude waxes (slack waxes) [64742-61-6] consist of a mixture of alkanes,
which can be solid, semisolid, or liquid at room temperature; alkyl-substituted
cyclopentanes and cyclohexanes (naphthenes); and alkyl-substituted aromatics.
They also contain, as impurities, the typical contents of the lubricating oil cuts
from which they originate, such as asphaltenes, resins, olefins, and sulfur and
nitrogen compounds. The oil content of slack waxes is usually between 5 and
12 %, but can be as high as 25 %.
n-Paraffins and isoparaffins predominate in slack waxes in terms of quan-
tity; both are present in continuously distributed homologous series. The chain-
length spectrum (C-number distribution) is determined from the width of the
boiling range, the distribution maximum (C-number maximum) from the boiling
level of the lubricating oil fraction. In isoparaffins, compounds with terminal
methyl branches predominate, followed by other methyl-substituted alkanes.
The concentration of these compounds decreases as the branching point moves
toward the middle of the chain. Other components are compounds with terminal
ethyl branches and multiply branched structures, which can be detected only in
slack waxes from heavy vacuum distillates (81,82).
Table 9 shows the composition of some slack waxes as a function of origin.
Physical Properties. Depending on the origin and production process,
slack waxes are light- to dark-brown, soft, unctuous to semisolid materials with-
out a clear crystal structure because of their high oil and soft wax content. The
congealing points lie between 35 and 65 8C; the drop points, between 35 and
68 8C; the needle penetration at 25 8C (a reciprocal measure of the hardness) is
between 40 and 80 (at greater oil content, even higher); the viscosities at 100 8C
are between 3 and 10 mm2/s; the densities at 70 8C, between 775 and 815 kg/m3;
and the flash points between 190 and 250 8C.
Vol. 2 WAXES 119

Table 9. Composition of Slack Waxes from Various Lubricating Oil Distillates of the
Same origin (80 % Arabian, 10 % German, and 10 % Norwegian Crude Oil)
Slack waxes from
Heavy Light Medium Heavy
Characteristic spindle oil machine oil machine oil machine oil
Congealing point, 8C 47 50 58 62
Oil content, % 9.5 9.0 9.8 8.0
n-Paraffin content, % 73 58 52 39
C-number range 16–36 19–43 21–48 22–58
C-number max. 24 29 34 41
Cmax content, % 17 14 10 8

Scale Waxes. Normally being physically decolorized slack waxes, scale


waxes [90669-78-6] essentially have the same composition as the raw materials
from which they are obtained. Only the dark material contents of the slack waxes
(asphaltenes and resins) are removed by the decolorization process, and the con-
tent of alkylaromatics, and of sulfur and nitrogen compounds, is reduced. In phy-
sical properties, scale waxes are similar to slack waxes. Decolorization renders
the products white to pale yellow.
Deoiled Slack Waxes (Crude Hard Waxes, Raw Waxes). In deoiled
slack waxes [8002-74-2] the predominant proportion of the hydrocarbons that
are liquid (oils) under normal conditions, and a certain proportion of the semiso-
lid ones (soft waxes), have been removed. Liquid and semisolid waxes consist
mainly of low molar mass n- and isoparaffins and of isoparaffins with centrally
located and strongly branched side chains, as well as naphthenes and alkylaro-
matics. Therefore, in deoiled slack waxes the n- and weakly branched isopar-
affins are enriched. By removal of the low molar mass portion a narrower
C-number range results along with a more pronounced C-number maximum.
Table 10 shows these relationships with two waxes from heavy spindle oil
and medium machine oil used as examples.
Physical Properties. In structure and consistency, deoiled slack waxes are
coarse to medium crystalline, brittle to weakly plastic and depending on the
degree of deoiling, hard to very hard. These waxes are light to dark brown and

Table 10. Composition of Deoiled Slack Waxes from Two Lubricating Oil Distillates
(Origin – See Table)
Characteristic Slack wax * Deoiled Slack wax ** Deoiled
Congealing point, 8C 47 50 58 67
Oil content, % 9.5 0.3 9.8 0.4
n-Paraffin content, % 73 84 52 64
C-number range 16–36 19–36 21–48 23–48
C-number max. 24 24 34 34
Cmax content, % 17 19 10 11
* From heavy spindle oil.
** From medium machine oil.
120 WAXES Vol. 2

Fig. 4. Variation of needle penetration with temperature for deoiled slack waxes of dif-
ferent origins. a) From heavy spindle oil (North Sea); b) From light machine oil (North
Sea); c) From heavy machine oil (North Sea); d) Wax from Indonesia.

darken on heating. The congealing points are between 48 and 72 8C; the drop
points between 48 and 75 8C; and the needle penetrations at 25 8C between 10
and 30. At the same degree of deoiling, the course of the penetration – temperature
curve (increase in penetration in the penetrogram (83) essentially depends on the
origin of the waxes (Fig. 4).
Crude soft waxes [64742-67-2] are sometimes known as foot oils (see
Section 4.2.3.2) and are formed in the deoiling of slack waxes. They consist pre-
dominantly of low molar mass n-paraffins and strongly branched isoparaffins;
higher molar mass, weakly branched isoparaffins; naphthenes; and alkylaro-
matics. The n-paraffin content of soft waxes is generally <25 wt %.
Soft waxes are very soft, unctuous to inhomogeneous materials at room
temperature because of their high oil content, which can be 30 wt % or greater.
The oil is included in voids of the crystal lattice of the solid hydrocarbons. Soft
waxes have a light- to dark-brown color and congealing points <40 8C.
Their viscosities at 100 8C are between 3 and 12 mm2/s; their flash points,
between 180 and 220 8C.
Semi-refined, filtered (decolorized), and fully refined waxes are
similar to deoiled slack waxes in the composition of their main components.
Through refining they are freed from more highly condensed alkylaromatics
and naphthenes, olefins, and sulfur and nitrogen compounds. Depending on
the origin, their content of n-paraffins varies between 92 wt % (East-Asian
waxes) and 45 wt % (waxes from heavy machine oil distillates of German origin).
The composition of the refined waxes is determined not only by the origin of the
crude (Table 11), but also by the boiling range of the starting distillate. As the
boiling ranges of the lubricating oils rise, the molar masses of the hydrocarbons
Vol. 2 WAXES 121

Table 11. Composition of Fully Refined Waxes (Congealing Point


62 8C in Each Case) from Different Origins
Wax from machine oil distillate
Characteristic Saudi Arabia Russia East Asia
Congealing point, 8C 62 62 62
Oil content, % 0.3 0.6 0.2
n-Paraffin content, % 52 78 92
C-number range 23–48 23–41 25–38
C-number max. 34 32 30
Cmax content, % 10 14 21

contained in waxes obtained from them increase. The concentration ratio of


iso- and cycloparaffins to n-paraffins also increases. Thus, for example, a refined
wax from Tuismasy (Ural) crudes contains 81.6 % n-paraffins with an average
molar mass of 360 g/mol (average C-number 25.7) and 9.3 % isoparaffins (average
molar mass 384 g/mol; average C-number 27.4). The naphthenes (9.1 %)
consist of 81.5 % monocyclic and 15.5 % bicyclic alkanes (average C-number
28.9) (84).
Physical Properties. Waxes in this group differ from deoiled slack waxes and
from each other in their residual oil content and, thus, in hardness, color, and
color stability.
Semirefined waxes have an oil content of 1.5–3 %, and the needle penetra-
tion at 25 8C can be 20–60. They are somewhat plastic and kneadable, colorless
to white, have good color stability under light, and are virtually odorless and tas-
teless.
Filtered (decolorized) waxes have an oil content of 1.5 % (max.) and a needle
penetration at 25 8C of 10–26. They have a whitish color and are relatively color
stable and generally odorless.
Fully refined waxes (oil content and penetration similar to decolorized
waxes) are crystalline, no longer kneadable, pure white, transparent to slightly
opaque, color stable, light-resistant, odorless, and tasteless. They fulfill purity
criteria for the production of packaging materials for foods and for the formula-
tion of cosmetics and pharmaceuticals. Detailed information on the chemical
composition of fully refined waxes and methods for their analytical determina-
tion can be found in the CONCAWE report (85).
According to a more recent definition, fully refined wax products are those
that fulfill various national purity requirements (see Section 4.2.5). The oil con-
tent, which was once limited to 0.5 % (max.), is no longer a criterion since the oils
contained in the wax also consist of n- and isoparaffins of the required purity.
Depending on the literature, the oil content of fully refined waxes can be 1.5 %
(max.) (86) or 2.0 % (max.) (87).
Occurrence of Raw Materials and Processing. Paraffin waxes are con-
tained in crude petroleum and are obtained during oil refining. Depending on the
source of the petroleum, the content of solid waxes can vary between 2 % (Ruma-
nian origin) and 30 % (Indonesian origin); generally, however, it is between 3 and
15 % in crude petroleum from the main production areas.
122 WAXES Vol. 2

Because of their higher molar masses and thus higher boiling ranges,
waxes become enriched in the residues from atmospheric distillation (long resi-
dues). In processing the residues to lubricating oils, waxes are obtained in all
fractions of the vacuum distillation. Because of their poor solubility, they give
rise to the high pour points of the basic oils and must therefore be removed. The
yields of wax vary depending on the origin and quality requirements of the oils.
For a wax content of crude petroleum of ca. 5 %, 8–18 % solid waxes in the indi-
vidual lubricating oil fractions can be expected.
Slack waxes are produced in the dewaxing of lubricating oil distillates. The
type of process used and process parameters used are according to the desired
quality of the lubricating oils.
The process steps necessary for the production of refined waxes are shown
in Figure 5.
Dewaxing Lubricating Oil Distillates. The principle of dewaxing lubri-
cating oil distillates is based on the different crystallization temperatures of
straight-chain and weakly branched paraffins and the oil phase. The wax-
containing solvent-neutral oils (ie, the raffinates from solvent refining) are
mixed with suitable solvents such as propane, naphtha, chlorohydrocarbons,
ketones, or (most frequently) a toluene–methyl ethyl ketone mixture. The oil–
solvent mixture is subsequently warmed to obtain a homogeneous solution and
then cooled continuously in scraping chillers to obtain the wax crystals in a loose

Fig. 5. Process schematic for the production of paraffin waxes.


Vol. 2 WAXES 123

Fig. 6. Dewaxing of lubricating oil distillates. a) Scraping chiller; b) Cooler; c) Vacuum


rotating filter; d) Solvent recovery.

suspension. In the subsequent filtration, crystallized wax is separated in vacuum


rotary filters, and washed with fresh solvent; the low-wax solvent-neutral oil is
then freed from the solvent by distillation. The solvent-containing wax is either
separated from the solvent by distillation if it is to be obtained as such or fed
directly into the deoiling process for production of hard waxes.
Process conditions are controlled in such a wax that high throughput rates
(ie, short filtration times) and as low an oil content of the slack waxes as possible
are achieved (88).
The selectivity of the dewaxing process is controlled by the cooling tempera-
ture in the scraping chiller and by the oil–solvent ratio. For common lubricating
oil fractions the cooling temperature is between 50 and 20 8C at oil yields of
65–85 %. The most important process parameter is the targeted pour point of the
dewaxed oil. Figure 6 shows a typical dewaxing process.
In dewaxing, all hydrocarbons with crystallization temperatures above the
chosen cooling temperature are removed. n-Paraffins are thus almost completely
separated, along with a suitable fraction of the long-chain, weakly branched
isoparaffins. Even under optimum filtration and washing conditions, impurities
consisting of highly branched isoparaffins and cycloparaffins are present in slack
waxes, thus giving rise to their oil content (89).
A second process for producing lubricating oils with the required low-
temperature properties is catalytic dewaxing. n-Paraffins are preferentially
cracked on zeolite catalysts to give low molar mass hydrocarbons. Wax thus
cannot be obtained from this process. Catalytic dewaxing is said to be very selec-
tive in the case of heavy distillates, whereas in light distillates the branched par-
affins also react (loss of yield). Since no solvents are used in this process it does
not require solvent recovery and is therefore environmentally friendly (90).
Deoiling Slack Waxes. Refining slack waxes begins with deoiling to
obtain harder, higher-value products (hard waxes). The two main processes
are solvent and sweat deoiling. The oils produced as byproducts, which contain
124 WAXES Vol. 2

Table 12. Processes for Deoiling Slack Waxes (95)


Company Process Solvent Starting material
Exxon crystallization methyl ethyl ketone – slack wax – solvent
process (Dilchill methyl isobutyl ketone mixture *
deoiling) (or – toluene)
Edeleanu pulping process 1,2-dichloroethane – slack wax – solvent
(Di-Me deoiling) methylene chloride mixture *
Edeleanu spray deoiling 1,2-dichloroethane slack wax
Texaco pulp process benzene (or benzene – slack wax – solvent
(one- and toluene) mixture *
two-step)
Texaco crystallization methyl ethyl ketone slack wax and slack
process (wax wax – solvent mixture *
fractionation)
Union oil crystallization water-saturated slack wax
process methyl isobutyl
ketone
sweat deoiling slack wax
* From the dewaxing step.

a higher or lower proportion of soft waxes, are known as foot oils or soft waxes,
depending on the consistency and the process used.
Solvent deoiling is the most commonly used process, involving one of
three possible techniques: pulping, crystallization, or spray deoiling. Starting
materials for all deoiling processes are either molten slack waxes (if deoiling is
carried out in a location other than that of slack wax production) or slack wax–
solvent mixtures, formed in dewaxing lubricating oil distillates (see Table 12).
Slack waxes from spindle oils to heavy machine oils, and to some extent pet-
roleum as well (see Section 4.3), can be deoiled by the three solvent deoiling
methods. The yields of deoiled slack waxes depend on their origin, the process
used, and the degree of deoiling. Yields are between 80 % for spindle oil distil-
lates and ca. 60 % for heavy machine oil distillates, based on the slack wax used.
Pulping Process. The inhomogeneous mixture of crystallized wax, oil,
and solvent is diluted or repulped by addition of the same solvent or a solvent
mixture (normally by using part of the wash filtrates produced later). The oils,
soft waxes, and part of the low molar mass hard waxes thus dissolve. The pulp
is cooled with stirring in scraping chillers, and the temperature is adjusted to
that necessary for the desired degree of deoiling. Hard waxes, which are partly
undissolved and have partly recrystallized, are removed on vacuum rotary filters
and washed with solvent; the solvent is subsequently distilled off.
The quality of the hard waxes produced is determined by the deoiling tem-
perature and the quantity of solvent used. With a 1,2-dichloroethane–methylene
chloride mixture as solvent (Edeleanu Process), for example, typical deoiling
temperatures are 10 8C for slack waxes from heavy spindle oil, 5 8C for
those from a light machine oil, and þ15 8C for those from medium to heavy
machine oil.
Crystallization Process. The molten slack wax or solvent-containing
wax crystallizate from the dewaxing process is dissolved completely, or almost
Vol. 2 WAXES 125

completely, in solvents by warming, and the mixture is cooled to a given tem-


perature in one or several steps, depending on the desired degree of deoiling.
The temperature is chosen such that only the hard wax crystallizes. The crystal-
lizate is filtered and washed, and the solvent is removed by distillation.
In both the pulping and the crystallization processes, technical problems
arise in attempting to produce readily filterable hard wax crystals that allow
high filtration capacity with good washability. These problems can be solved in
various ways: eg, through the nature and quantity of the solvent; the construc-
tion of the crystallization apparatus; successive solvent dosing during cooling;
addition of cold solvent to the warm, intensively stirred wax–solvent mixture
(91); and use of filter aids (88). Other improvements have been brought about
by the development of economical cooling and filtration sysems (92).
Spray Deoiling. Molten slack wax is sprayed into a countercurrent of
cold air, and precipitated wax particles are washed with solvents in mixers
whereby the oil is largely diluted. After settling the washed wax particles are
centrifuged and washed, and the adhering solvent is distilled off. The washing
temperature is between 5 and 15 8C (93).
Sweat Deoiling. Molten slack wax is charged to chambers equipped with
sieve bottoms and heating coils. The wax is solidified by cooling and then warmed
very slowly. The oil – and, at higher temperatures, the low-melting soft waxes, as
well – sweat from the wax block and run through the perforated bottom (run-off
slack wax). At the end of the process the remaining hard wax is melted to remove
it from the sweating chamber.
No new plants are being built for this classical deoiling process because of
the low selectivity (poor yields of hard wax), time-consuming warmup, need to
use a batch process, and inapplicability of the method to strongly oil-binding
slack waxes from medium and heavy machine oil distillates. In existing plants
an attempt is being made to improve yields of hard waxes by partially recycling
the run-off slack (94).
Overview of Processes in Current Use. The deoiling processes most
widely used in industry are listed in Table 12.
Of the deoiling plants operating in Germany, half employ solvent deoiling
(using mainly the Edeleanu dichloroethane–methylene chloride process) and the
others use spray deoiling (Edeleanu) and sweat deoiling.
In the United States, more than 90 % of the existing plants use solvent
deoiling. The most common solvents are methyl ethyl ketone and methyl
ethyl ketone –aromatics mixtures. Propane and methyl isobutyl ketone are
also used.
Refining Deoiled Slack Waxes. Deoiled slack waxes whose maximum
oil content is 0.5 %, 0.5–1.5 %, or 2–3 %, depending on the use anticipated, still
contain impurities and are mostly dark in color. To improve their quality they
are purified further by adsorbents (decolorizing) or chemically. The choice of
refining process depends on economic factors and the quality of raffinates
required. Process combinations have also been introduced into industry. The pre-
ferred process in new plants is hydrotreating.
Refining with adsorbents (decolorizing) can be performed batchwise,
semicontinuously, or continuously. The decolorizing temperature can be between
70 and 120 8C, depending on starting material.
126 WAXES Vol. 2

In batch decolorizing, molten wax is stirred in heated vessels with decolor-


izing clays (chemically activated clays, bentonite, bauxite) in one or more steps
until the desired lightening of color is achieved. The used clay is removed in
heated filter presses and thermally regenerated.
Percolation Process. In the percolation process, molten wax flows down-
ward through a tower containing 10–50 t of decolorizing clay, depending on
plant size. After the clay has been exhausted, the wax flow is stopped; wax adher-
ing to the clay is washed off with naphtha; the remaining solvent is driven off with
steam; and the clay is burnt off in a separate tower, activated, and reused (96).
The semicontinuous decolorizing process operates with two or three per-
colator towers, which are in different phases of the adsorption cycle. While the
adsorption process occurs in one tower, in the others the used decolorizing clay
is washed, stripped, removed for regeneration, and replaced by new activated clay.
In the continuous process the regenerated clay flows downward through the
adsorption tower in countercurrent to the molten wax. The loaded clay passes in
the form of a sludge to a washing tower where adsorbed wax is extracted with
naphtha and then to a second tower where naphtha is driven off wich steam.
The clay is then burnt off in a rotary kiln, activated, and fed back to the top of
the adsorption tower after cooling (97).
The decolorized waxes are then stripped with steam in packed columns at
110–170 8C, either under vacuum (0.6–1.6 kPa), under reduced pressure (up to
80 kPa), or under pressure (up to 0.8 MPa) to remove odorous substances. The
weight ratio of steam to wax varies between 0.5 : 1 and 1 : 1 at detention times
up to 15 min. To improve the aging resistance of the waxes, apparatus with alu-
minum or stainless steel lining is recommended for stripping (98).
Depending on the starting material, the wax qualities listed in Table 13 are
obtained by the decolorizing process.
The quantity of decolorizing clay required depends on the origin and desired
quality of the waxes and is normally between 2 and 4 %. Refining with adsor-
bents removes dark substances, condensed alkyl aromatics, compounds contain-
ing hetero atoms, and metallic impurities.
Nevertheless, producing fully refined waxes fulfilling all of the purity
requirements (see Section 4.2.5) by adsorption is possible only using extremely
complex processes (multistage decolorizing) and large quantities of decolorizing
clay, particularly in the case of the high molar mass wax fractions. To produce
these types of high-purity waxes, chemical refining is the process of choice.
Chemical Refining. Concentrated sulfuric acid, fuming sulfuric acid
(oleum), or hydrogen is used as chemical refining agent.

Table 13. Wax Qualities Obtained by Decolorization


Starting material Oil content, % Decolorized product
Slack wax >3 scale wax
Deoiled slack wax 1.5 – 3 semirefined wax
Deoiled slack wax 1.5 (max.) filtered wax
[64742-43-4]
Vol. 2 WAXES 127

Sulfuric acid refining can only be carried out batchwise because of the slow
separation of the wax and acid sludge phases. Molten wax is mixed with sulfuric
acid at 80–140 8C for some time (between 0.5 and 2 h, depending on the origin of
the wax). After separation of the heavier acid sludge, the wax is washed with
alkali, treated with decolorizing clay, and stripped with steam to remove the
last residues of impurities and odorous substances.
All compounds that react chemically with the aggressive acid, such as unsa-
turated aliphatic and aromatic hydrocarbons, metal compounds, and those con-
taining hetero atoms, are removed by sulfuric acid refining. Labile quaternary
carbon atoms are also attacked, with bond cleavage and formation of aliphatic
sulfonic acids.
The process can be optimized by variation of the number of reaction steps,
degree of acidity, reaction temperature, ratio of wax to acid used, reaction time,
and intensity of mixing. The higher the average molar mass of the wax fraction,
the slower is the separation between the wax and the acid sludge phase.
The process has several disadvantages such as the necessity for batch
operations, poor yields of raffinates, formation of polluting byproducts, and
waste gas and corrosion problems. For these reasons, almost all new plants
are based on hydrotreating.
Hydrotreating (Fig. 7). Deoiled slack wax is heated to the required tem-
perature together with fresh and recycyled hydrogen in the preheater (b) and
passed over a sulfur-resistant fixed-bed catalyst (c). After cooling, the reaction
mixture and hydrogen are separated in the high-pressure gas separator (d),
and hydrogen is recycled into the process. After depressurizing in the low-
pressure gas separator (e), the wax passes into the stripper (f), where all the
light crack products, odorous substances, and reaction gases are stripped off com-
pletely with compressed steam under vacuum. The wax is subsequently dried
with nitrogen (g). A combination of metals from groups 6 and 10 on an inert

Fig. 7. Two-step hydrotreating of waxes. a) Deaerator; b) Preheater; c) Reactors;


d) High-pressure gas separator; e) Low-pressure gas separator; f) Stripper; g) Vacuum
dryer; h) Filter.
128 WAXES Vol. 2

carrier is generally used as the catalyst (eg, nickel–tungsten or nickel–molybde-


num on neutral aluminum oxide).
Reaction conditions can vary within wide limits, depending on starting
material, degree of refining required, and composition of the catalyst – eg, between
200 and 350 8C, 20 and 200 bar, and throughputs of 0.2–0.81 L of wax per liter of
catalyst and hour. Yields are almost 100 %, based on starting wax (99,100).
Under the severe reaction conditions of high-pressure hydrogenation, all
aromatic compounds are hydrogenated to naphthenes, all dark substances are
decomposed, and all sulfur and nitrogen atoms are removed as hydrogen sulfide
and ammonia. Metallic impurities become bonded to the catalyst.
The process can be used to refine all classes of wax (paraffin waxes, inter-
mediate waxes, and microwaxes).
Storage, Transportation, Commercial Forms, and Producers. Paraffin
waxes are transported in liquid form (molten) in road tank trucks, rail tankers,
or liquid containers. Solid waxes are marketed in slabs in small and large cartons
and sacks, and as pastilles, flakes, and powder in sacks or big bags. They are also
supplied in fiberboard and steel drums and in molded cartons.
Even refined waxes can become yellow or decompose to form odorous sub-
stances on prolonged heating, particularly in the presence of air or catalytically
active metals. On storage, transportation, and processing, the waxes are there-
fore preferably heated in vessels, fitted with warm water equipment, and the
containers and piping are well insulated. Storage tanks are often blanketed
with inert gas. In some cases, antioxidants are added to fully refined waxes in
quantities up to 0.01 %. Deoiled and refined waxes are differentiated according
to melting range (gradation). Some examples of commercial waxes are given
below.

Europe United States


(Melting gradation in 8C) (AMP* gradation in 8F)
48/50 120–122
50/52 122–124
52/54 126–130
54/56 130–132
56/58 132–134
58/60 134–136
60/62 136–140
62/64 143–145

* AMP = American Paraffin Wax.

Important producers of paraffin waxes include: Shell (United Kingdom,


United States, Germany), Texaco (United States), BP (France, Germany),
Mobil (United States, United Kingdom), Lützendorf (Germany), DEA (Germany),
Wintershall (Germany), CFP (France), Total (France), Empetrol (Spain), Petro-
gal (Portugal). In Germany, for example, the companies either process the
waxes to refined waxes (DEA, Wintershall) themselves or sell them to wax refi-
neries (H. O. Schümann). Waxes are sold mostly by DEA, H. O. Schümann, and
Wintershall.
Vol. 2 WAXES 129

Quality Specifications and Analysis. The quality specifications of paraf-


fin waxes are determined mainly by their uses. They are differentiated in terms
of physical characteristics, chemical composition, purity requirements, and
application properties. The following standard specifications have been pub-
lished for waxes that come in contact with foods:

1. Food and Drug Administration: Code of Federal Regulations (United


States) (101)
2. FAO Food and Nutrition Paper (EC) (102)
3. Specifications of the European Wax Federation (EC) (86)
4. Kunststoffe im Lebensmittelverkehr (Plastics which come into contact with
foods, Recommendations of the German BGA) (103)
5. Hydrocarbons in Food Regulations (United Kingdom) (104)

For waxes in cosmetic and pharmaceutical preparations, standards are found


in the following publications: DAB (Germany), Codes Français (CF) (France), B.P.
(United Kingdom), ISCI Monograph (Japan), U.S. Formulary (United States), and
CTFA International Cosmetics Ingredient Dictionary (United States).
The most common test standards for the determination of physical data are
summarized in Table 14. The most important test standards for determining che-
mical composition can be found in Table 15. Some examples of important test
methods for determination of the purity of waxes are listed below:

Fluorescence 49th communication


of the BGA
Alkali- or acid-reacting DAB, B.P., U.S.P.
substances
Behavior toward sulfuric acid DAB, BGA, CF, U.S.P.
Presence of polycyclic DAB, BGA, FDA
aromatics

Table 14. Standardized Test Methods for Determination


of Physical Data
Determination Method
mp (cooling curve), 8C DIN ISO 3841 ASTM D 87
Congealing point, 8C DIN ISO 2207 ASTM D 938
Needle penetration, 0.1 mm DIN 51 579 ASTM D 1321
Oil content, % DIN ISO 2908 ASTM D 721
Viscosity, mm2/s DIN 51 562/T1 ASTM D 445
Density, kg/m3 DIN 51 757 ASTM D 1298
Refractive index DIN 51 423/T2 ASTM D 1747
Color
ASTM color index DIN ISO 2049 ASTM D 1500
Lovibond color IP * 17
Saybolt color DIN 51 411 ASTM D 156
Odor ASTM D 1833
* IP = Institute of Petroleum, United Kingdom.
130 WAXES Vol. 2

Table 15. Standardized Test Methods for Determining Chemical Composition


Determination Method
Ash content, % DIN EN 7 ASTM D 482
Sulfur content, ppm DIN 51 400/T7 ASTM D 2622
Heavy metal content (As, Cr, Cd, Pb), ppm XRF a analysis
C-number distribution DGF b M-V9 EWF c method
C-number max. DGF b M-V9 IP d draft
n-Paraffin content, % DGF b M-V9 ASTM draft
a
XRF = X-ray fluorescence spectromety.
b
DGF Einheitsmethoden der Deutschen Gesellschaft für Fettwissenschaften [Standardized methods
of the Deutsche Gesellschaft für Fettwissenschaften (German Fat association)].
c
EWF = European Wax Federation.
d
IP = Institute of Petroleum, United Kingdom.

A looseleaf collection of national and international purity regulations for


waxes has been issued by the European Wax Federation and is supplemented
continuously (105).
Some of the methods cited are basic test procedures for gaining accredi-
tation according to the series of standards DIN EN 45 000 and the regulations
of the Deutsche Akkreditierungsstelle Mineralöl GmbH (DASMIN). With
accreditation, DASMIN recognizes the competence of testing laboratories to
test the quality of waxes.
The determination of application properties is described mostly in in-house
methods and includes, for example, oil binding properties, contraction, polish-
ability, abrasion resistance, solvent retention, blocking point, steam permeabil-
ity, flexibility, extendability, and color stability. Melting and congealing
characteristics and latent heats are determined by using differential thermoana-
lysis [standardized DGF method M-III 14 (1989); ASTM-D 4419–84].
Special analytical methods have been developed to determine the chemical
composition of paraffin waxes.
The content of n- and isoparaffins is determined most accurately by using
computer-supported high-temperature gas chromatography (106,107). Here, the
DGF method M-V9 (1988) is the universal standard. This method allows not only
sufficient separation of the n- and isoparaffin signals (depending on the column
length and injector temperature, waxes up to a C-number of ca. 80 can be deter-
mined quantitatively), but also determination of the C-number distribution and
position of the C-number maximum.
The determination of n- and isoparaffins using urea adduct formation
and molecular sieve adsorption has been shown to be a nonquantitative separa-
tion method. Only enrichment of the n-paraffins results. This, however, clearly
diminishes with increasing molar mass and complexity of the waxes investigated
(108).
For individual analysis of the different, paraffinic substance groups,
coupled GC–MS and GC–NMR methods have been described (85,109).
The determination of individual polycyclic aromatic hydrocarbons in waxes
is currently possible into the mg/kg (ppb) region (85,110,111).
Vol. 2 WAXES 131

Table 16. Uses of Paraffin Waxes Worldwide, 1992 (in 1000 t)


United Western of which
Use States Asia Europe Germany
Candles and decorative modeling 100 450 230 55
Paper, cardboard, wood 450 220 120 25
Hot melt adhesives 60 20 20 10
Rubber and cables 40 20 30 10
Industrial products 50 30 50
Pharmaceuticals and cosmetics 20 30 10
Other uses 130 130 40 50 *
* Including industrial products, pharmaceuticals and cosmetics.

Uses. Paraffin waxes are used in very different fields of industry. They
not only are used as such, but are also important components of wax compounds
(ie, mixtures of paraffin waxes with other waxes, plastics, resins, etc.). The con-
sumption figures listed in Table 16 cover the most important areas of use.
Slack waxes that are not processed further to refined waxes or sub-
jected to dehydrogenation or cracking processes – for example, to give a-olefins
(see Hydrocarbons) – are used mostly in the wood and mineral oil processing
industries.
Wood processing requires slack waxes, either molten or in the form of emul-
sions, mainly for the production of particle board and fiberboard. In the United
States the production of artificial logs for barbecues and fires is also important.
In the mineral oil processing industry, slack waxes are used to produce
petroleum jellies, wire-drawing agents, lubricants, spray emulsions, and
anticorrosives.
In addition to wax industry products with a wide area of applications,
slack waxes are also used to produce chlorinated paraffins, papers and card-
boards for technical uses, cable coverings, solid anticorrosives, fire lighters,
and wax torches.
Soft waxes have only a limited market in the production of industrial
petroleum jellies and emulsions (wood and building industry), torches, and fire
lighters and for conditioning fertilizers. The deoiling of soft waxes gives highly
plastic and extendable products (isowaxes).
Refined Waxes [Scale Waxes, Filtered (Decolorized), Semi- and
Fully Refined Waxes]. In Germany the producers of candles and wax goods
(pictures, sculptures, ornaments) are the main users of refined waxes, which
can be processed by all common production methods for candles (molding, draw-
ing, extrusion molding, and powder compressing). In terms of quantity, paraffin
wax is the most important raw material for the candle industry.
The second most important consumer of refined waxes is the wax industry,
which requires paraffin waxes as an important raw material for compounding
wax products for the most widely varying applications.
In the paper, cardboard, and packaging industries, refined waxes are used
in the form of emulsions for paper sizing as antifoams in the production of paper,
cardboard, and corrugated cardboard; and as impregnating and coating waxes
for paper and packaging materials (112).
132 WAXES Vol. 2

The viscosity of heat-seal waxes and hot-melt adhesives is adjusted with


paraffin waxes. In rubber articles, tires, and solid cable coverings, waxes act
as external lubricants/plasticizers, mold-release agents, and rapidly migrating
protection agents of rubber against sunlight. They are therefore used predomi-
nantly for statically stressed rubber articles (profiles, gaskets, linings, and
cable jackets).
Other important applications are in cleansers and polishes; cosmetics
(creams, ointments, lipsticks); matches; plastics (external lubricants); office
papers; and waterproof textiles (dry impregnation).
In the United States, by far the greatest proportion of refined waxes is used
for paper and cardboard finishing, mainly for deep-freeze packaging and wrap-
ping for meat, candy, and flowers; paper cups; yogurt and soft cheese pots; and
corrugated cardboard (113).
In the former Soviet Union, refined waxes still play an important role in the
production of synthetic fatty acids via catalytic oxidation processes. Fatty acids
are mainly used in the soap industry.
4.3. Microcrystalline Waxes (Microwaxes). Chemical Composition
and General Properties. Like paraffin waxes, microcrystalline waxes con-
sist of a mixture of saturated hydrocarbons that are predominantly solid at
room temperature, such as n- and isoalkanes, naphthenes, and alkyl- and
naphthene-substituted aromatics. Unlike paraffin waxes, isoparaffins and
naphthenic compounds predominate here. The microcrystalline structure can
be explained by the presence of strongly branched isoparaffins and naphthenes,
which inhibit crystallization.
General Physical Properties. Microcrystalline waxes are insoluble in
water and most organic solvents at room temperature. They are moderately to
readily soluble in solvents such as chlorohydrocarbons, benzene, toluene,
xylene, solvent naphtha, and turpentine oil, especially at elevated tempe-
rature. Solubility decreases markedly as molar mass increases. Solvents and
oils are retained very strongly by microcrystalline waxes and therefore evapo-
rate very slowly. The quality of some consumer products, such as petroleum
jellies or floor and shoe polish, is determined by this retention capacity of
microwaxes.
Chemical Properties. Microwaxes are more reactive than paraffin
waxes because of the higher concentration of complex branched hydrocarbons
with tertiary and quaternary carbon atoms. These CC bonds are not very
thermally stable (ie the waxes darken and resinify) on prolonged heating. In
addition, they form black tarlike substances on contact with aggressive chemi-
cals such as concentrated sulfuric acid or antimony pentachloride.
The reaction of microwaxes with oxygen at elevated temperature and in the
presence of catalytically acting heavy-metal soaps is used for the production of
oxidized microwaxes.
Division into Product Classes. Depending on the degree of refining,
microwaxes are divided into the following classes:
1. Bright stock slack waxes (petrolatum)
2. Plastic microwaxes
3. Hard microwaxes
Vol. 2 WAXES 133

Table 17. Typical Physical Characteristics of Microwaxes


Bright stock Plastic Hard
Characteristic slack wax microwax microwax
Congealing point, 8C 67 72 75 87
Drop point, 8C 76 78 80 91
Needle penetration, 0.1 mm
at 25 8C 65 30 28 8
at 40 8C 140 85 66 18
Oil content, % 3.8 1.8 1.6 0.4
Viscosity (at 100 8C), mm2/s 18.7 15.7 14.9 16.2
Refractive index (at 100 8C) 1.4505 1.4458 1.4412 1.4410
Color dark brown yellow white white

Table 17 and Figure 8 show some typical characteristics of microwaxes.


Bright stock slack waxes (petrolatum) constitute the most complex
mixture in the group of microcrystalline waxes. The molar masses of the hydro-
carbons can be between 320 and 1200 g/mol. The n-alkane content is seldom
>10 %. Asphaltenes, hydrocarbon resins, and sulfur, nitrogen, and oxygen com-
pounds are present as impurities. The oil content is normally between 3 and
15 %, but it can be as high as 30 %. Bright stock slack waxes are white or yellow
to dark brown, greasy to softly plastic, unctuous, smooth products with a typical
odor, depending on the refining process. They have a very good oil-staining capa-
city, and some grades have a pronounced stickiness. The congealing points are
between 60 and 75 8C; the drop points between 62 and 80 8C; and the hardness
(needle penetration at 25 8C), between 45 and 160, depending on the oil content.
The viscosities (10–30 mm2/s at 100 8C), densities (810–860 kg/m3 at 80 8C), and
flash points (290–320 8C) are clearly higher than those of paraffin waxes.
Plastic Microwaxes. In plastic microwaxes the higher molar mass
n- and isoparaffins and the weakly branched isoparaffins and naphthenes are

Fig. 8. Variation of needle penetration with temperature for various microwaxes


compared with a paraffin wax. a) Bright stock slack wax; b) Plastic microwax; c) Hard
microwax; d) Paraffin wax 60/62.
134 WAXES Vol. 2

enriched compared with the petrolatum from which they are obtained by deoil-
ing. Molar masses are between 450 and 1200 g/mol. The n-paraffin content can be
as high as 40 %. Depending on the degree of refining the impurities present in
petrolatum are removed completely or to a great extent. The oil content is
usually between 1.5 and 5 %.
The plastic microwaxes, although relatively hard, are definitely plastic,
readily moldable, and kneadable. The plasticity is due to the presence of
very long carbon chains in the n-paraffins and terminally branched isopar-
affins. The waxes feel dry to slightly adhesive, and are sold as white to
brown products. Congealing points are 65 –80 8C, and the hardness (needle
penetration at 25 8C) is 10 to 60. Highly refined microwaxes are odorless and
tasteless.
Hard microwaxes also essentially consist of high molar mass alkanes
(predominantly n-alkanes and isoalkanes with short side chains statistically dis-
tributed over the main carbon chain) and naphthenes. Average molar masses are
between ca. 500 and 1200 g/mol. Thus, for example, a refined, hard microwax of
North American origin contains 89 % nonpolar hydrocarbons (n-, iso-, and cyclo-
paraffins) and has an average C-number of 48 (85). The n-paraffin content can be
as high as 70 %.
The extremely microcrystalline to almost amorphous structure of hard
microwaxes is associated with the formation of supercooled melts during the set-
ting process.
The waxes are tough and hard, barely ductile and weakly adhesive.
Depending on the degree of refining, hard white, yellow and brown microwaxes
are on the market, the white to pale yellow products being of highest purity.
They solidify between 80 and 94 8C. The needle penetration at 25 8C is <10.
The hardness of hard waxes decreases only slightly with temperature until
close to their softening point (Fig. 8).
Occurrence of Raw Materials and Processing. Like paraffin waxes,
microwaxes are components of crude petroleum. Because of their high molar
masses, and thus high boiling range and poor solubility, they either are enriched
in the vacuum residues (short residues) from lubricating oil distillation (residual
waxes) or separate during the transportation and storage of crude oils (settling
waxes). Settling waxes form pasty to solid deposits on the walls of pump tubes
and pipelines and on the bottom of storage tanks (pipe and tank bottom
waxes). Vacuum residues and pipe and tank bottom waxes are the raw materials
for recovery of microcrystalline waxes.
Figure 9 shows schematically the process steps involved in the production
of microcrystalline waxes.
In the United States, the most important producer country for microwaxes,
the vacuum residues are subjected to deasphalting, refining, and dewaxing to
produce high-value bright stock oils. Bright stock slack waxes (petrolatum) are
produced in the process as economically important byproducts. However, only in
the rarest cases they are processed further to microwaxes in the lube oil refi-
neries themselves. Petrolatum is either marketed directly or used, for example,
for the production of petroleum jellies. For further processing it is sold to compa-
nies specializing in the production of microwaxes. The same companies also pro-
cess settling waxes purchased from oil companies.
Vol. 2 WAXES 135

Fig. 9. Process schematic for the production of microwaxes.

Deasphalting is the first step in the processing of tank bottom waxes,


pipe waxes, and residues, which have been freed from water, mechanical impu-
rities, and volatile components. When propane is used as the deasphalting agent,
considerable proportions of resins, polycyclic aromatics, and nitrogen and sulfur
compounds are removed. To improve the selectivity with as low a propane
throughput as possible, rotating disk contactors are now used instead of packed
or sieve bottom columns. In addition, the extraction process is carried out in two
steps (114).
Deasphalting of residues is followed by solvent extraction and dewaxing.
The products formed here are the bright stock oils and bright stock slack
waxes (petrolatum).
Deoiling. The crude and bright stock slack waxes thus obtained are
deoiled by fractional crystallization if they are to be processed further to
microwaxes. Preferred solvents are methyl ethyl ketone and methyl iso-
butyl ketone, often as a mixture with benzene or toluene. Propane and naphtha,
frequently used earlier, have become less important because of their low
selectivity.
136 WAXES Vol. 2

The crude waxes (deasphalted settling waxes) and bright stock slack waxes
are mixed either molten or as a pulp (if the deoiling is to follow immediately after
dewaxing the brightstock oils), without isolation of the slack wax with fresh,
warm solvents and warmed until only the less soluble (high molar mass)
waxes remain undissolved. These waxes are then filtered, washed, and freed
from solvent (hard microwax fraction). To obtain plastic microwaxes the filtrate
is cooled further. The properties of reprecipitated and refiltered waxes are deter-
mined by the filter temperature. The low molar mass, very plastic, and soft
microwax foot oils [64742-67-2] remains in the filtrate. If the hard microwaxes
are not extracted, a one-step crystallization process is sufficient for obtaining
plastic microwaxes.
Deoiled microwaxes are still dark yellow to dark brown and not completely
odorless, so they must subsequently be refined to satisfy a broad application range.
Refining. Refining is carried out either with adsorbents (decolorizing) or
with hydrogen (hydrotreating). Refining with concentrated sulfuric acid, which is
possible in principle, has not become important economically because of the high
loss of material.
For refining microwaxes with adsorbents, the same process principles apply
as for decolorizing paraffin wax (see page 33). Since microwaxes are more diffi-
cult to decolorize than paraffin waxes, countercurrent percolation is used predo-
minantly. Here, the wax melt continually comes in contact with fresh
decolorizing clay (115). Bauxite is also frequently used as a decolorizing agent.
Hydrotreating is carried out as described for paraffin waxes, in one or two
steps in reactors packed with sulfur-resistant, heavy-metal-doped mixed catalysts
(see page 34). Reaction conditions are ca. 350 8C and 150 bar, depending on the
properties of the catalyst, at a space velocity of 0.3 L of wax per liter of catalyst
and hour. Hydrogenated products are treated further by adsorptive decolorizing
or stripping to completely remove odorous materials and volatile components.
In the United States, a whole range of microwaxes is produced according to
the processed described. Products differ in their congealing point, drop point,
hardness, viscosity, and color. They are sold in the liquid (molten) form and as
solids in the form of slabs, pellets, and powder.

Table 18. Producers of Microwaxes in the United States and Commercial Products
Producer Production site Commercial products
Bareco Div. of Kilgore, Tex. Petrolite Crown, Be Square,
Petrolite Corp. Barnsdale, Okla. Mekon, Stawax, Cardis,
Victory, Ultraflex
Mobil Oil Co. Beaumont, Tex. Wax Rex, Mobilwax
Paulsboro, N.J.
Quaker State Oil Emlenton, Pa. Microwax, Superflex
Ref. Corp. Farmers Valley,
Pa. Newell, W.V.
Shell Oil Co. Deer Park, Tex. Shellwax
Wood River, Ill.
Sun Oil Co. Tulsa, Okla. Sunoco Wax, Sunolite
Witco Chem. Corp. Bradford, Pa. Multiwax
Vol. 2 WAXES 137

Commercial Products and Producers. Table 18 shows the most impor-


tant producers of microwaxes in the United States and the corresponding
commercial products.
In Europe, H. O. Schümann (Germany), Mobil (United Kingdom), CFR/
Total, Mobil, and BP (France), Shell and Witco (Holland), Repsol (Spain), and
Petrogal (Portugal), for example, produce microwaxes by deoiling bright stock
slack waxes.
Quality Specifications and Analysis. The quality required depends on the
industrial use. In principle, the specifications given for paraffin waxes apply here.
Basic product specifications can be found in the following sources:

1. Food and Drug Administration; Code of Federal Regulations (United


States) (101)
2. European Wax Federation (EC) (86)
3. Kunststoffe im Lebensmittelverkehr; Empfehlungen des Bundesgesund-
heitsamtes Deutschland (Plastics for Contact with Foods; Recommenda-
tions of the Federal Health Authority, Germany) (103)

The EWF specifies microwaxes for use in packaging materials for foods, as
shown in Table 19.
In the use of microwaxes to produce food packaging materials, and cheese
and chewing gum waxes, purity and physiological nontoxicity are particularly
important. For microwaxes the absence of specific polycyclic aromatics is required.

Table 19. Microwaxes for Food Packaging


Determination Method Specification
Congealing point, 8C DIN ISO 2207 ASTM D 938 65 (min.)
Needle penetration, 0.1 mm DIN 51 579 ASTM D 1321 <50
Extractables, % ASTM D 721 2.5 (max.)
Viscosity at 100 8C, mm2/s DIN 51 562/T1 ASTM D 445 >13
Density at 100 8C, kg/m3 DIN 51 757 ASTM D 1298 790 – 840
Refractive index at 100 8C DIN 51 423/T2 ASTM D 1747 1.430 – 1.450
Color DIN ISO 2049 ASTM D 1500 2.0 (max.)
Odor ASTM D 1833 weak
Ash content, % DIN EN 7 ASTM D 482 <0.05
Sulfur content, ppm DIN 51 400/T7 ASTM D 2622 <300 a
<4000 b
Heavy-metal content (As, Cd, XRF analysis c <1
Cr, Pb), ppm
C-number distribution of DGF M-V9 (88) IP 372–85 23 – 85
n-paraffins
Non-n-paraffin content, % DGF M-V9 (88) 25 – 75
UV-Absorption $ 172.886 FDA
at 280 – 289 mm 0.15 (max.)
at 290 – 299 mm 0.12 (max.)
at 300 – 359 mm 0.08 (max.)
at 360 – 400 mm 0.02 (max.)
a
Hydrotreated.
b
Refined with sulfuric acid/decolorizing clay.
c
XRF ¼ X-ray fluorescence spectrometry.
138 WAXES Vol. 2

These substances can occur as impurities in insufficiently refined waxes, and some
of them have been found to be carcinogenic. Polycyclic aromatics can be detected
by UV extinction after a special enrichment process (detection limit in micro-
waxes 107 g per gram of wax substance). Microwaxes have been shown to be
toxicologically and physiologically harmless in extensive animal experiments,
when UV adsorption values specified according to test regulations of the FDA
(101) and the BGA (103) are not exceeded.
The physical characteristics of microwaxes are determined according to the
same standardized methods used for paraffin waxes (see Table 14). In addition,
some special analytical methods have been introduced, for example:

Determination of gloss and gloss stability ASTM D 2895


Determination of blocking point ASTM D 3234
Determination of sealing stability TAPPI [Tech. Assoc.
Pulp & Paper Ind.
(United States)] T 642
Determination of water vapor ASTM D 988
permeability
DIN 53 413
Determination of specific penetration DIN 53 482
resistance
Determination of dielectric loss factor DIN 53 483

Oil-binding capacity, adhesiveness, and light stability are additional qual-


ity factors.
Uses. Bright Stock Slack Waxes. The greatest proportion of bright
stock slack waxes is processed further to microwaxes. Important consumers
are the wax-producing industry, which requires crude and refined bright stock
slack waxes for compounding adhesive waxes, cheese waxes, chewing gum
base, cosmetic preparations, sealing and cable compounds, impregnating insulat-
ing materials and pesticide sprays; and the mineral oil processing industry,
which produces anticorrosives, cavity and underfloor protection agents for
motor vehicles, and petroleum jellies. Other uses include the production of arti-
ficial fire logs and oxidized petrolatum, which is suitable for the formulation of
rust preventives.
Oxidized petrolatum and microwaxes. are mainly produced industrially in
the United States. Air at normal or at slight overpressure is blown through the
wax, which is heated to 150–180 8C until the desired degree of oxidation is
reached. Darkening of the wax is avoided by adding fresh wax to the reaction
zone. Oxidized microwaxes are specified and traded according to their melting
range and degree of oxidation (expressed in terms of acid and saponification
numbers). The acid and ester groups, which are distributed statistically over
the hydrocarbon chains, can be reacted with inorganic and organic bases to
form soaps that can be dispersed in water (emulsified waxes). These compounds
also impart to the oxidized waxes an excellent adhesion to metals.
Plastic Microwaxes. The main consumer of plastic microwaxes is the
wax industry. Plastic microwaxes are added to wax compounds that require
either pronounced plastic or oil- and solvent-staining properties for their indus-
trial applications.
Vol. 2 WAXES 139

Typical applications include the following:

1. Paper and cardboard coating for packaging meat and sausage products,
cereals, bread, candy, ice cream, cookies, yogurt, and frozen food
2. External lubricant (sliding) for extrusion of plastics
3. Formulation of cheese waxes and chewing gum bases
4. Decorative for candles, production of wax pictures and sculptures
5. Polishes and grinding agents
6. Lost wax casting and dental compounds
7. Foam regulators for detergents
8. Protectives against crack formation in tires and rubber articles (sun-proof
waxes) (116)
9. Additives for explosives and propellant materials.

Hard microwaxes have numerous applications based on their high melt-


ing points and great hardness; favorable hardness–temperature behavior; and
good solvent retention. They are used as release agents for pressing plastics
and fiberboard, in the production of ceramic articles and in molding polyur-
ethane foam, in solvent-containing cleansing agents, in grinding and polishing
pastes, and flat varnishes and paints.
The Committee on Toxicity (COT) in the United Kingdom has recommended
that only hard microwaxes be approved for use in the production of cheese waxes,
poultry defeathering compounds, and chewing gum (117).
On an industrial scale, hard microwaxes are oxidized to an acid number
of 30 and are used in this form to produce emulsions (polishes and release
agents) and as additives for paints and coatings, printing inks, and office
transfer papers.
4.4. Legal Aspects. Apart from environmental effects, petroleum waxes
can be absorbed by the human organism via foods (paraffins either are contained
in them naturally, have been added to them, or have migrated there from the
packaging materials), from chewing gum (only on swallowing or the simulta-
neous consumption of fat-solubilizing chocolates), and through the use of lipstick.
To protect human health, national and international regulations and recommen-
dations have been enacted to limit the oral consumption of waxes.
The use of waxes in packaging materials is subject of EEC legislation (118).
In the appendix lists (lists of additives), waxes are permitted, provided that spe-
cific migration limits of 0.3 mg per kilogram food for waxes treated with sulfuric
acid or decolorizing clay and 3 mg per kilogram for high-pressure hydrotreated
waxes are not exceeded.
Extensive literature on the migration of paraffins into foods has been pub-
lished (119–122).
The permitted paraffins must correspond to published specifications (see
Section 4.2.5).
In Germany the addition of petroleum waxes to food and animal feed is not
allowed. The use of waxes as cheese coatings and in chewing gum is controlled by
supplementary regulations.
In the United Kingdom the Ministry of Agriculture, Fisheries, and Food
(MAFF) proposes a prohibition of petroleum waxes as food additives in England,
140 WAXES Vol. 2

Scotland, and Northern Ireland (117). Only specified microwaxes may still be
added to foods for a limited period.
In the United States, not indication has been given of any intention to ban
paraffins in food.
4.5. Ecology. Paraffin and microcrystalline waxes are insoluble in
water and can be separated as solids, together with sludge, in oil separators
and sewage treatment plants. In the German catalogue of substances that are
hazardous to water, they are therefore listed in water hazard class 0 (Wasserge-
fährdungsklasse, WGK 0). Since paraffins have a water solubility of <1 mg/L,
they have no toxic effects on aquatic organisms. Used paraffin-containing packa-
ging materials can be combusted without any problem in waste incineration
plants. No polluting substances are formed, and the high calorific value of the
paraffins improves the energy balance of the plants.
The biodegradability of petroleum waxes satisfies OECD guideline 301 B
(1981) (123). With the exception of microwaxes, biodegradabilities of more than
60 % after 28 d were measured. Bacterial degradation mechanisms are described
in (124). According to this, paraffin-containing packaging materials can also be
composted (125–127). No substances that are toxic or hazardous to groundwater
are formed.
The biodegradation of waxes under anaerobic conditions (covered landfills)
proceeds much more slowly.
A large number of investigations have been carried out on the recycling of
used wax-containing packaging materials. The water-insoluble, hydrophobic pet-
roleum wax tends to form particles in the paper mills and give rise to blockages
in the sizing equipment (128).
In Europe, ca. 20106 t of waste paper was recycled as a raw material for
paper production in 1991. This figure includes 0.35106 t of wax-containing
paper and cardboard with average wax content of 12 %. Assuming an even dis-
tribution, this gives a wax content of 0.21 % wax in wastepaper (129). The prac-
tice has shown that in modern wastepaper processing plants, wax-containing
packaging materials do not interfere with the dilution process,but nearly
increase the quantity of waste products. These waste products can be used to pro-
duce water-resistant cardbord.
4.6. Economic Aspects. In 1992, 9.5105 t of paraffin and microcrys-
talline waxes were produced in the United States, 8105 t in Asia, 6.5105 t in
Western Europe, 2.6105 t in Germany alone, 2.0105 t in Eastern Europe. and
4.0105 t in the rest of the world.
In industrialized countries, a stagnation (probably a decrease) in petroleum
wax production is anticipated for the coming years, independent of the growth in
demand. This is because of the further development of car engine and tribology
technology (reduced specific lubricant consumption and longer periods between
oil changes), an increase in the proportion of partially or fully synthetic and
biodegradable lubricants, the recycling of used oils, and the closing down of
excess capacity in distillation plants despite an increase in the number of
motor vehicles and the distances traveled (130).
However, in Asia, particularly China and Malaysia, an increase in wax pro-
duction to 1.5106 t in 2000 is predicted, because these waxes can be inexpen-
sively produced from indigenous wax-rich crude petroleum by direct dewaxing.
Vol. 2 WAXES 141

According to careful estimates, worldwide petroleum wax production will be


ca. 3.8106 t in 2000 (40 % in Asia, 25 % in the United States, 17 % in Western
Europe).
The worldwide consumption of petroleum waxes was ca. 2.8106 t in 1992,
of which (in million tonnes) 0.5 was in Western Europe, 0.2 in Eastern Europe,
0.85 in the United States, and 0.9 in Asia.
4.7. Candles. According to German commercial quality class
RAL 040 A2 (Issue 02/1993), candles are sources of heat and light and consist of
a wick and a solid combustible material surrounding the wick. Therefore tea
lights, lights for graveyards, and oil lamps are also included in this category.
The first candles have been traced to shortly after the birth of Christ. With
Christianity, and later with the rise of principalities and the upper middle class,
the use of candles developed rapidly. Whereas initially, tow wicks, beeswax,
tallow, and spermaceti were used for candle production, from the 19th century
onward the use of cotton wicks, petroleum wax, and stearin permitted the
problem-free handling of candles.
Today, refined (petroleum) wax is the main raw material used in candle pro-
duction. Candles are also made from stearin, beeswax, composites (mixtures of
various waxes), and ceresin (mixtures of paraffin and hard waxes). Candles
are dyed with wax-soluble organic aniline dyes (full dyeing), organic pigments,
and (less often) natural pigments (dip dyeing). Fashionable candles with a
colored dipping-paint coating are also available. Wicks consist of twisted cotton
threads and are sold as round or flat wicks of different strengths. The wicks are
twisted and prepared so as to guarantee uniform burning, good wick curling posi-
tion, and good suction capacity for the molten combustible material and to pre-
vent afterglow of the wick once the candle has been extinguished. Candles are
sometimes perfumed.
A range of mechanical and hand-worked processes exist for industrial pro-
duction of candles. In the drawing process a candle strand is produced by passing
the wick several times through the hot wax bath. The strand then passes onto
the cutting machine, where it is cut to the desired length of the candle. With con-
tinuously operating candle-drawing machines, production of more than 10 000
candles per hour is possible. In the molding process, molten wax is poured into
molds, in the middle of which the wicks are held taut. Molding machines are
available in different sizes. Modern rotating molding machines have an hourly
capacity up to 10 000 candles. The powder pressing process uses powdered wax
as the starting material, which is either molded together with the wick in an
extruder to form a candle strand or pressed into molds in a stamp press. At
the same time, the wick is introduced with a tubular needle. Another process
exists, known as the dunking process. Here, the wick is dipped repeatedly into
molten wax.
Blank candles are made pointed by using milling cutters and are coated
with dyed dip-coating material if necessary.
The quality control of candles involves, for example, determining their
weight and dimensions, burning period, and burning properties (no drips or
soot). The production of candles in Germany (East and West) was 80 000 t in
1988, and 17 000 t were imported. Thus, with an export of 22 000 t, a total of
75 000 t was consumed within Germany.
142 WAXES Vol. 2

5. Fischer–Tropsch Paraffins

Production. The sole producer of Fischer–Tropsch paraffins is South


African Coal, Oil and Gas (SASOL) in the Republic of South Africa (! Coal Lique-
faction, Chap. 2.2.). The formerly used ARGE (Ruhrchemie/Lurgi consortium)
process, which employed iron catalysts arranged as a fixed bed in multitubular
flow reactors, was changed to the SASOL slurry bed proces (SSBP), which
afforded considerable cost savings. Production capacity has been doubled (131).
In the slurry or bubble column reactor, finely divided iron catalyst (average
particle size 50 mm) is suspended in a Fischer–Tropsch wax with low viscosity
at the reaction temperature, and synthesis gas (a mixture of carbon monoxide
and hydrogen) is bubbled through. At reaction temperatures of 220–240 8C
and pressures around 2 MPa, n-alkanes and n-alkenesare formed preferentially,
with up to 40 % crude wax in the synthesis product.
Crude wax is separated from the other products of synthesis by fractional
condensation. The low-boiling constituents (gasoline and diesel fuel) are removed
by distillation under atmospheric pressure, followed by vacuum distillation to
separate the soft waxes. Besides n-alkanes, distillation bottoms also contain
alkenes, hydrocarbons with hydroxyl and carbonyl groups, and colored compo-
nents. For purification and stabilization, the bottom products are subjected to
a hydrofining step employing a nickel catalyst to yield a white hard wax that
is practically free of alkenes, aromatics, functional hydrocarbons, and sulfur
compounds.
Soft waxes are also subjected to hydrogenation.
Properties. Fischer–Tropsch waxes consist essentially of n-paraffins
with chain lengths between 20 and 50 carbon atoms. Products with an average
molar mass of 400 g/mol are marketed as soft, and those with an average molar
mass of 700 g/mol as hard waxes. The paraffins have a fine crystalline structure
and, because of the narrow molar mass distribution, a small melting range and
very low melt viscosities. Congealing point, density, and hardness increase with
increasing mean molar mass. The low molar mass compared with polyolefin
waxes is the reason for a certain displaceability of the crystal layers relative to
each other and, associated with this, the polishability.
Synthetic paraffins are fully compatible with refined waxes, polyolefin
waxes, and most vegetable waxes. They are soluble at elevated temperature in
the usual wax solvents (eg, naphtha, turpentine, and toluene) to give clear solu-
tions. Addition of 10–20 % synthetic paraffin to other waxes increases their con-
gealing point and hardness without significantly influencing melt viscosity. In
wax pastes, the tendency of Fischer–Tropsch waxes to form microcrystals
increases solvent retention. Some typical data for Fischer–Tropsch waxes are
listed in Table 20.
Uses. Fischer–Tropsch waxes are used in plastics processing as lubri-
cants for poly(vinyl chloride) and polystyrene, as well as mold-release agents;
as melting point improvers, hardeners, and viscosity reducers in hot melts and
candles; and, because of their good polishability, for the production of cleaning
agents and polishes.
Micronized waxes improve the abrasion resistance of paints and printing
inks.
Vol. 2 WAXES 143

Table 20. Typical Properties of Fischer – Tropsch Soft and


Hard Paraffins
Sasol Sasol Wax H2
Characteristic Wax M Vestowax SP 1002
Drop point, 8C 106–112
Needle penetration, 0.1 mm 1–3
Color white white
Molar mass, g/mol 400 700
Density (at 23 8C), g/cm3 0.94 0.94
Viscosity (120 8C), mPa  s <20 <20

Oxidized waxes containing fatty acids and fatty acid esters are produced
from synthetic paraffin by oxidation and partial saponification. Major areas of
application are as mold-release agents in plastics processing, in polishes and
cleaning agents, and as auxiliaries in the textile and paper industries.
Trade Names. Fischer–Tropsch waxes are marketed, eg, as Sasol Wax
(Sasol Marketing Co., Johannesburg, South Africa) and Vestowax SH/SP (Hüls
AG, Marl, Germany).

6. Polyolefin Waxes

6.1. Production and Properties. Polyethylene Waxes by High-


Pressure Polymerization. High-pressure polyethylene (PE) waxes are pro-
duced, like high-pressure polyethylene plastic, at high pressure and elevated
temperature in the presence of radical formers. As waxes, their molar masses
are considerably lower than those of plastics. The molar mass range is adjusted
during polymerization by the addition of regulators.
High-pressure polyethylene waxes are partially crystalline and therefore
consist mainly of branched molecular chains in which shorter side chains, such
as ethyl and butyl, predominate. They generally have low densities (low-density
polyethylene waxes, LDPE waxes). Less branched polyethylene waxes with
higher crystallinity and density (high-density polyethylene waxes, HDPE
waxes) can be produced by increasing the pressure.
High-pressure polyethylene waxes with molar masses between 3000 and
20 000 g/mol [weight-average molar mass M  w, determined by gel permeation
chromatography (GPC)] dominate the market.
The development of high-pressure PE waxes proceeded in parallel to
that of LDPE plastics. In 1939, waxlike ethylene polymers were formed by
using a variant of the ICI high-pressure polymerization process (133). Indus-
trial production of high-pressure ethylene waxes was started in the 1940s
(134,135). Parallel to this, the technology of thermal depolymerization – the
thermal degradation of polyethylene plastics to polyethylene waxes – was
developed.
With the rapid growth of petrochemistry and the expansion of polyethylene
plant capacities, the economic importance of PE waxes also increased rapidly.
Waxes with very versatile property profiles could be synthesized by varying
144 WAXES Vol. 2

the density and molar mass of the homopolymers and functionalizing polyethy-
lene by copolymerization with various monomers (eg, vinyl acetate or acrylic
acid) or by melt oxidation.
Because PE waxes are rather inexpensive, have improved applicability and
consistency in quality, and are in constant supply they have displaced expensive
natural waxes (eg, carnaúba and montan waxes) in many areas.
The high-pressure process is not applicable for production of polypropy-
lene waxes, because only soft or oily products can be obtained. These waxes
are therefore produced by the Ziegler process or by depolymerization of high-
density polypropylene.
Production. The technology involved in the production of high-pressure
PE waxes (see Fig. 10) is analogous to that used for high-pressure polyethylene.
The only differences are in the product finishing process. Unlike viscous plastic
melts, which must be granulated underwater in granulators, for example, the
mobile wax melt can be converted into powder by spraying or into granules by
using dicers.
The reaction vessels are stirred autoclaves or tubular reactors, the design
of which can vary considerably.
Reaction Mechanism. The high-pressure polymerization of ethylene
is a highly exothermic radical chain reaction, which is initiated by reaction
of ethylene with radicals formed by decomposition of an initiator (136).
Chain propagation occurs through addition of further ethylene units. Chain
growth is terminated, for example, by reaction with a regulator molecule

Fig. 10. Schematic of production of polyethylene waxes by the high-pressure process in


stirred autoclaves. a) Precompressor; b) Postcompressor; c) Autoclave reactor; d) High-
pressure separator; e) Low-pressure separator.
Vol. 2 WAXES 145

such as propene:

CH2 ĊH2 þ CH2 


R CH3 ! R
CH CH
CH2 þ CH3
ĊH
CH3

Polymerization is terminated by recombination or disproportionation of two


macroradicals:

Long-chain branching occurs by intermolecular chain-transfer reactions:

A particular feature of polyethylene wax synthesis by the high-pressure


process is the increased formation of short ethyl and butyl side chains. These
are formed by intramolecular radical transfer:

At the high pressures necessary for polymerization, ethylene is in a super-


critical state. Polymerization therefore takes place in a one-phase system. After
leaving the reactor, the reaction mixture is decompressed in high- or low-
pressure separators, and unreacted ethylene evaporates. The wax remains as a
melt and ethylene is recycled.
Reaction Conditions. The structure, molar mass, and thus properties of
polyethylene wax are determined mainly by reaction pressure, reaction tempera-
ture, type and quantity of initiator and molar mass regulator, and reactor type
and geometry (137–141).
In the homo- and copolymerization of ethylene the reaction pressure is
usually 150–320 MPa. Somewhat lower pressures are used in autoclaves than
in tubular reactors. Other process variants involve much lower pressures of
70 MPa (max.) with isopropanol as the molar mass regulator (139,140).
Higher pressure favors chain propagation and thus leads to very short
residence times in the reactor. It also inhibits chain-transfer reactions so that
with increasing pressure the degree of branching decreases and the density,
146 WAXES Vol. 2

crystallinity, hardness, and mp increase. In tubular reactors, rising pressure


increases the polymer yield, whereas in autoclaves this effect is smaller.
For homopolymerizaion the reaction temperature is 200–350 8C and for copo-
lymerization 200–300 8C. High reaction temperature favors chain-transfer reac-
tions. For this reason the polyethylene waxes formed have many short-chain
and only a few long-chain branches (142) and thus low densities. Higher-density
waxes with lower degrees of branching are formed at lower reaction temperature.
Organic peroxides and molecular oxygen (the latter exclusively in tubular
reactors) are used as initiators. They decompose into radicals under polymeriza-
tion conditions. To ensure that the peroxides are metered reproducibly, they are
dissolved in organic solvents.
Hydrogen and almost all organic compounds can act as molar mass regula-
tors (chain terminators). Hydrogen, lower alkanes (eg, propane), lower alkenes
(eg, propene or butene), alkyl aromatics, lower aldehydes (eg, propionaldehyde),
and lower alcohols (eg, isopropanol) or mixtures of these substances are mainly
employed. The concentration and reactivity of these regulators determine the
average molar mass, the degree of branching, and thus the density of polyethy-
lene wax. The activity of the regulator increases with increasing temperature
and decreases with increasing pressure (136).
Reactor type and geometry also significantly influence polymer structure
and properties. Because of the backmixing occurring in autoclaves, spherical
molecules with many long-chain branches are predominantly formed. In tubular
reactors with plug flow, long, straight molecules with little long-chain branching
predominate (143). High-density, highly crystalline polyethylene waxes can
therefore be produced only in tubular reactors.
Just as the geometries of the autoclave and the tubular reactor differ, so do
operating procedures. In the case of autoclaves the initiator solution and the
molar mass regulator are charged directly to the preheated pressurized vessel
by using high-pressure pumps. In tubular reactors the initiator (eg, air) and
the molar mass regulator are added to the reaction mixture before the compres-
sion stage or at the entrance to the reactor. With autoclaves, comonomers can be
charged to different parts of the reaction vessel (see Fig. 10).
Product Finishing. Shaping of the molten polyethylene can be performed
directly after removal of gaseous products in high- and low-pressure separators
and subsequent fine degassing and filtration. Since no metal-containing catalysts
are used in high-pressure polymerization of ethylene, no catalyst removal is
necessary. Low molar mass fragments and residues of the initiator and molar
mass regulator evaporate mainly in the separators. Only an extremely small pro-
portion is incorporated into the polymer.
Granules can be produced by using strip dicers. The wax melt flows in thin
strips on a continuous, water-cooled metal band and, after solidification, is
chopped with a rotating knife. Pastilles can be obtained by metered dripping of
the melt from a distributor head onto this type of cooling band.
A particular characteristic of wax melts is their low melt viscosity. Thus,
spraying the melt to form powders of spherical particles is also possible. By
using special nozzle arrangements (binary nozzles) with nitrogen as the atomiz-
ing gas, micronized waxes (very fine powders) with particle sizes in the micro-
meter range can be produced.
Vol. 2 WAXES 147

Powders and micronized waxes can also be obtained by grinding granules in


jet mills. With soft or very viscoelastic waxes, cooling with dry ice or liquid nitro-
gen is necessary. Coarse particles must be removed by subsequent classification.
Like all organic dusts (eg, coal or flour), wax powders, and particularly
micronized waxes, are highly susceptible to dust explosions. Appropriate safety
precautions must be taken during processing, in particular grounding all instal-
lations to avoid spark formation through electrostatic charge buildup.
Properties. The transition from polyethylene plastic to polyethylene wax
is flexible. If the molar mass is lowered, the thermoplastic materials is gradually
changed to a wax. The DGF definition of wax (see Section 1.2) gives an approx-
imate boundary: For waxes, an upper limit to the melt viscosity of ca.
20 000 mm2/s at 120 8C is defined, which corresponds to an average molar mass
(weight-average molar mass M  w) of ca. 37 000 g/mol.
The properties of polyethylene waxes are determined strongly by M  w as a
measure of the average chain length and by the degree of branching as a
measure of the shape of the molecules. Melt viscosity increases with increas-
ing molar mass. Crystallinity, hardness, mp, and solidification point increase
as the degree of branching decreases. These data are important for application-
oriented properties.
The molar mass can be determined by GPC. This method gives the weight-
average (M  w) and the number-average (M  n) molar mass. The ratio M  w/M n is a
measure of molar mass distribution and is known as the polydispersity index
(! Polymerization Processes, Chap. 2.1.1.). For high-pressure polyethylene
waxes currently produced, this value lies between 2.0 and 2.5. The waxes have
a much narrower molar mass distribution than high-pressure polyethylene, for
which the values are 3–8 (tubular reactor) and 12–16 (autoclave) (136).
Table 21 gives a series of structure and property data for three high-pres-
sure polyethylene waxes with increasing density from two different producers
(some of this represents unpublished data supplied by the authors).

Table 21. Structure and Properties of High-Pressure Polyethylene Waxes A – C and D –


F from Two Different Producers
Characteristic A B C D E F
3
Density (23 8C), g/cm (DGF-M-III 2a) 0.92 0.92 0.945 0.92 0.92 0.94
CH3 per 1000 C atoms 26 22 17 32 23 20
Branches per molecule 3.4 1.8 1.0 3.7 1.9 1.8
Crystallinity (from IR data), % 54 59 67 52 62 64
Central double bonds per 1000 C atoms 0.1 0.4 0.4 0 0 0
Terminal double bonds per 1000 C atoms 0.1 0.1 0 1.3 1.2 1.3
Side-chain double bonds per 1000 C atoms 0.6 0.4 0.2 0.9 0.5 0.4
Ball indentation hardness (23 8C), bar 200 370 550 314 413 482
(DGF-M-III 9a)
Solidification point, 8C (DGF-M-III 4a) 96 99 106 93 101 104
Drop point, 8C (DGF-M-III 3) 106 112 116 106 111 113
Melt viscosity (120 8C), mm2/s (DGF-M-III 8) 1310 1160 1200 780 925 890
Weight-average molar mass M  w, g/mol 6140 5930 6000 5200 5500 5500
Number average molar mass M  n, g/mol 2920 2390 2440 2490 2380 2660
 
Mw/Mn (polydispersity index) 2.1 2.5 2.5 2.1 2.3 2.1
148 WAXES Vol. 2

The following conclusions can be drawn from these data:


1. The degree of branching is generally low with ca. ten branches (max.) per
molecule. The side chains are mostly ethyl and butyl groups with a very
small proportion of long-chain branches (144–146).
2. The density, crystallinity, hardness, solidification point, and drop point
increase as degree of branching decreases (147). All density-dependent
properties vary correspondingly.
3. High-pressure PE waxes contain a small proportion of double bonds, whose
distribution in the macromolecule depends on the production process, as
the large differences between the levels of central double bonds show.
The formation of double bonds is attributed to depolymerization of chain
radicals at high reaction temperature (136,148), eg:

High-density polyethylene waxes are colorless, white to transparent and


form clear melts. Like other waxes, they dissolve in nonpolar solvents (eg,
aliphatic, aromatic, and chlorinated hydrocarbons) on heating and generally
crystallize as very fine particles on cooling. Depending on the type and concen-
tration of the wax, they then form mobile dispersions or paste-like gels, which
frequently exhibit thixotropic properties.
Copolymeric Polyethylene Waxes by High-Pressure Polymerization. In
the high-pressure process, many other monomers can copolymerize with ethy-
lene giving rise to considerable changes in product properties (139,141). For
copolymerization, autoclaves with their intensive mixing, constant temperature,
and consequently stable reaction process are particularly suitable. In industry,
vinyl acetate and acrylic acid are used mainly as comonomers, giving waxes
with higher polarity and lower crystallinity. Here also, the degree of branching
decreases with increasing pressure and decreasing temperature. Density, mp,
and hardness increase accordingly.
Table 22 gives characteristic data for two important copolymeric high-
pressure polyethylene waxes with vinyl acetate and acrylic acid as comonomers.
Ethylene–vinyl acetate copolymer is a relatively polar but nevertheless
hydrophobic wax that can be dispersed in organic solvents particularly well. It
is used as an additive in metallic automotive paints, as a dispersing agent in pig-
ment concentrates, and as a component of hot melts.
Ethylene–acrylic acid copolymer is suitable for aqueous systems because it
can be emulsified readily and is chemically and thermally intensive. Its emul-
sions are used mainly in modern floor polishes and mold-release agents.
Polyolefin Waxes by Ziegler–Natta Polymerization. Like high molar
mass polyethylene plastics, PE waxes can be produced by the Ziegler low-pres-
sure process using organometallic catalysts, in addition to the radical process
at high pressure and temperature. Like high-density polyethylene, Ziegler
waxes have a mainly linear molecular structure. They can contain short side
chains but not long ones like high-pressure polyethylene waxes.
Vol. 2 WAXES 149

Table 22. Physical and Chemical Data for Copolymeric High-pressure Polyethylene
Waxes
Ethylene – vinyl Ethylene – acrylic
acetate copolymer acid copolymer
Characteristic wax wax
Weight-average molar mass M  w, g/mol ca. 6800 ca. 6100
Number-average molar mass M  n, g/mol 3000 3000
Melt viscosity (120 8C), mm2/s (DGF-M-III 8) 2000 1300
Solidification point, 8C (DGF-M-III 4a) 85 93
Drop point, 8C (DGF-M-III 3) 96 102
Ball indentation hardness (120 8C), 120 410
bar (DGF-M-III 9a)
Acid number, mg KOH/g 45
Vinyl acetate content, % 10

For direct synthesis of polymer waxes from propylene or higher a-olefins,


only the Ziegler–Natta process is suitable.
Ziegler PE waxes currently on the market have a molar mass between 800
and 8000 g/mol (M  n). They thus bridge the gap between Fischer–Tropsch and
paraffin waxes and HDPE thermoplastics. Besides controlling chain length,
the polymerization process allows adjustment of the degree of crystallinity so
that both hard–brittle, high-melting products and softer, flexible products with
low melting points can be obtained.
The soluble, soft wax fractions inevitably produced in the suspension pro-
cess for HDPE are more similar to paraffin waxes than to PE waxes because of
their low molar masses. For normal hard wax applications, they can be used only
after processing by distillation or extraction (149).
Production. For the production of Ziegler–Natta polyolefin waxes, Zieg-
ler catalysts that give polymers with low average chain lengths and narrow
chain-length distributions, and are sufficiently active under the special condi-
tions of wax synthesis, are particularly suitable. The low degree of polymeriza-
tion typical of waxes is achieved by carrying out polymerization in the presence
of hydrogen as a molar mass regulator at comparatively high temperature
(usually between 100 and 200 8C, sometimes even higher) (150–154). Aliphatic
hydrocarbons are used as reaction medium. Because of the high reaction tem-
perature the waxes are formed in solution (solution polymerization). If the pro-
duct viscosity is sufficiently low the molten polymer formed can itself function as
the solvent (bulk polymerization) (155,156). Unlike modern HDPE processes in
which the catalyst is not removed, in the case of wax synthesis the catalyst must
usually be decomposed and filtered off (157,158). This applies at least to the het-
erogeneous titanium catalysts still used today (see below) and is necessary
because waxes have higher purity requirements than PE plastics as a conse-
quence of the applications for which they are used. After the solvent has been
distilled off, the wax melt is shaped (eg, by spraying or pastille formation).
Catalyst Systems. Classical Ziegler catalysts, consisting of titanium tet-
rachloride and alkylaluminum compounds, can be used for wax synthesis, but
their activity is low (155,156,159–161). The degree of branching and thus also
150 WAXES Vol. 2

the density, drop point, and hardness can be adjusted by copolymerization, by


the special type of catalyst preparation (160), and by varying the polymerization
temperature (155,161).
The state of the art for production of PE homo- and copolymer waxes
involves, as in plastics production, the use of supported catalysts, which con-
tain titanium atoms as the active species and magnesium compounds as the
carrier material. Catalysts derived from titanium tetrachloride and magne-
sium chloride, oxide, hydroxide (151– 154), or alkoxide (150), for example,
are suitable.
Catalysts based on Ti–Mg compounds can also be used in the production of
polypropylene waxes. Through an appropriate choice of catalyst, the degree of
crystallinity (isotacticity) can be varied within wide limits and thus adapted to
a particular application. Flexible products with average crystallinity can be
obtained (162). Highly crystalline polypropylene waxes can now be produced eco-
nomically by using stereoregulating silanes as additional catalyst components
(163,164). Formerly, these waxes could be produced only by thermal degradation
of high molar mass polypropylene.
Indications are that the homogeneous metallocene–alumoxane catalyst sys-
tems (165,166), which became known at the end of the 1970s and nowadays are
used for plastics synthesis in industry, will also become important in wax synth-
esis. Since the 1980s, many patent applications have appeared in this field, relat-
ing to both PE (167–172) and PP (173–175) waxes. Metallocene catalysts have a
number of advantages compared with titanium catalysts. They are more active
and can thus be employed in lower quantities, so catalyst removal is not neces-
sary. Polymerization can occur at lower temperature, and the process is therefore
more economical. Above all, the specific polymerization properties and great
structural variability of metallocene systems are likely to give polyolefin waxes
with optimized or completely new property profiles.
Properties. The most important molecular structure parameters, which
determine the macroscopic property profile of PE waxes, are the average molar
mass (average chain length) and the degree of branching. Melt viscosity rises
exponentially with increasing molar mass (Fig. 11). Unbranched, homopolymeric
PE waxes have a high degree of crystallinity as a result of their undistorted

 w and melt viscosity for PE waxes.


Fig. 11. Relationship between average molar mass M
Vol. 2 WAXES 151

Table 23. Structure and Properties of Typical Commercial Ziegler Polyolefin Waxes
A B C D E
Ethylene Ethylene Ethylene Ethylene Propylene
homopolymer copolymer homopolymer homopolymer homopolymer
wax wax wax wax wax
Number-average 1600 2200 850 5000 3000
molar mass M  n,
g/mol (GPC)
Weight-average 4800 6300 920 17 000 18 000
molar mass
M w, g/mol (GPC)
Molar mass 3.0 2.9 1.1 3.4 6.0
distribution
(M w/M n)
Density (23 8C), 0.97 0.93 0.96 0.97 0.88
g/cm3 (DGF-M-
III 2a)
Heat of fusion, 240 120 230 200 60
J/g (DSC)
Viscosity (140 8C), 400 600 10 25 000 1700 *
mPa  s
Drop point, 8C 125 118 107 135 ** 158 **
(DGF-M-III 3)
Needle penetration, <1 3 2 <1 <1
0.1 mm
(DGF-M-III 9b)
* Temperature of measurement 170 8C.
** Softening point ring/ball (DGF-M-III 13).

linear molecular structure. They are brittle and hard (needle penetration num-
bers <1) and have high densities and melting points (up to ca. 127 8C). These
products can be obtained only by Ziegler polymerization. Radical polymerization
always gives branched waxes.
By using 1-alkenes (generally propene) as copolymers, defects can be intro-
duced into the polymer chain in the form of short-chain branches, which lead to a
decrease in crystallinity. The density, hardness, drop point, and heat of fusion
are also decreased, while flexibility and plasticity increase.
In Table 23, some important physical properties of typical commercially
available Ziegler polyolefin waxes are compared.
Ziegler waxes are colorless and odorless and form clear, transparent melts
with high thermostability. Like most other waxes, they are soluble on warming
in nonpolar aliphatic and aromatic solvents without leaving a residue, and on
cooling often crystallize out as pastes. Commercial products have a melt viscosity
range from ca. 10 to 30 000 mPa  s (at 140 8C). Their densities are between 0.92
and 0.97 g/cm3.
Degradation Polyolefin Waxes. By heating in the absence of air, the
molecular chain of high molar mass polyolefins can be cleaved into smaller mole-
cules with waxlike character. The starting materials are usually low- and high-
density polyethylene, isotactic polypropylene, and polybutene. The production
capacities of degradation polyolefin waxes have diminished since ca. 1980 in
152 WAXES Vol. 2

favor of the built products. While the proportion of degradation PE waxes has
diminished, that of PP waxes has increased. Despite the relatively unfavorable
energy balance of the degradation process, it is still used because of its technolo-
gical simplicity and, for PE waxes, occupies third place with regard to the pro-
duction quantity after the high-pressure and low-pressure PE waxes. The
degradation process was developed approximately in parallel to high-pressure
polymerization (176–178).
Production. Polyethylene is a thermally stable polymeric hydrocarbon.
In the absence of oxyen it can withstand temperatures up to ca. 290 8C (179).
At higher temperature, thermal degradation begins with a decrease in the
molar mass; from 400 8C, degradation occurs with high rate and yield (179).
The cleavage products consist almost exclusively of a wax fraction with a
molar mass distribution that is barely wider than that of high-pressure PE
waxes. Low-boiling and gaseous crack products are formed only in small quanti-
ties. Individual process steps are shown in Figure 12.
The starting materials for thermal degradation to polyolefin waxes are
LDPE and HDPE (180) and HD polypropylene (181–184). Degradation of copo-
lymeric polyolefins (185) and polybutenes (186) has also been described but has
not achieved great economic importance.
In the industrial process (180), polyolefin pellets are melted under nitrogen
in a single-screw extruder and extruded into a steel tube heated to ca. 400 8C.
The residence time is 15–30 min. The product is then cooled by evaporative
cooling in ca. 5 min to <250 8C. It is freed from low molar mass components by
degassing in a separator. The product can then be finished as described for
high-pressure PE wax.
The degree of branching of degradation waxes formed, and thus all
structure-dependent properties such as crystallinity, density, hardness, and

Fig. 12. Process schematic for the production of degradation polyolefin waxes.
Vol. 2 WAXES 153

mp, are controlled by the choice of starting material (LDPE or HDPE, stabilizer
free). The average molar mass depends on residence time in the hot-tube reactor
and the average molar mass of the starting material. The degradation tempera-
ture affects mainly the reaction time and thus the throughput rate.
Batchwise, thermal degradation in autoclaves has also been described
(182,185). It is performed in certain cases in the presence of hydrogen and hydro-
genation catalysts to saturate double bonds formed on chain cleavage (176).
Like other polyethylene waxes, degradation waxes can be modified by air
oxidation and thus rendered emulsifiable. Preliminary hydrogenation of double
bonds can be necessary to avoid cross-linking reactions during oxidation.
Because of their double bonds, degradation waxes are particularly
suited to grafting with unsaturated carboxylic acids (eg, with maleic anhy-
dride (187,188)).
Little information is available in the literature on the mechanism of ther-
mal degradation of polyolefins to form waxes (179). Presumably, radicals form
from the few oxygen-containing groups in the polyolefin and initiate a radical
chain reaction, in which the chain is cleaved through radical transfer and a dou-
ble bond is formed:

Since tertiary hydrogen atoms are particularly susceptible to radical trans-


fer, the degradation reaction begins at the carbon atoms of the branches, to form
fragments containing side-chain double bonds. Only later are linear chains
cleaved, to form central and terminal double bonds.
Properties. In Table 24, properties (in some cases unpublished results
supplied by the authors) of three degradation and three high-pressure polyethy-
lene waxes, with about the same average molar mass to decreasing branching
and increasing density, are compared. The two types of wax can be seen not to
differ significantly. For both, the same relationships are valid between molar
mass and melt viscosity and between degree of branching and density, crystalli-
nity, hardness, and solidification and drop points. The width of the average
molar mass distribution (polydispersity) is also comparable. A significant differ-
ence is that degradation polyethylene waxes contain considerably more double
bonds than comparable polyethylene waxes formed by high-pressure polymeriza-
tion. Determination of the type and level of double bonds from the IR spectrum
154 WAXES Vol. 2

Table 24. Characteristics of Three Degradation and Three High-pressure Polyethylene


Waxes
Degradation PE wax High-pressure PE wax
Characteristic G H I A B C
3
Density (23 8C), g/cm 0.92 0.93 0.95 0.92 0.93 0.945
(DGF-M-III 2a)
CH3 per 1000 C atoms 32 16 8 26 22 17
Branches per molecule 3.5 0.5 0 3.4 1.8 1.0
Crystallinity (from 43 63 74 54 59 67
IR data), %
Central double bonds per 1.2 1.5 2.0 0.1 0.4 0.4
1000 C atoms
Terminal double bonds per 3.3 2.6 2.6 0.1 0.1 0
1000 C atoms
Side-chain double bonds per 1.7 1.1 0.5 0.6 0.4 0.2
1000 C atoms
Ball indentation hardness 450 500 600 200 370 550
(23 8C), bar (DGF-M-III 9a)
Solidification point, 8C 96 100 110 96 99 106
(DGF-M-III 4a)
Drop point, 8C (DGF-M-III 3) 105 112 121 106 112 116
Melt viscosity (120 8C), mm2/s 930 1230 1300 1310 1160 1200
(DGF-M-III 8)
Weight-average molar 5400 6040 6100 6140 5930 6000
mass M  w, g/mol
Number average molar 2400 2160 2440 2920 2390 2440
mass M  n, g/mol
 
Mw/Mn (polydispersity index) 2.3 2.8 2.5 2.1 2.5 2.5

can therefore be used for analytical identification of degradation polyethylene


waxes. The technical properties are affected only slightly by double bonds.
Compared with degradation polyethylene waxes, degradation polypropy-
lene waxes usually have extremely high melting points and hardness because
they are produced from highly crystalline, isotactic high-density polypropylene.
Some characteristics of a relatively low molar mass degradation polypropylene
wax are listed below:

mp (DSC) 130–160 8C
Ball indentation hardness (23 8C) 500–1900 bar
(DGF-M-III 9a)
Melt viscosity (200 8C) (DGF-M-III 8) 800–1900 m2/s
Weight-average molar mass (GPC, 17 000–37 000 g/mol
standard: PP)

Degradation polyolefin waxes are colorless, white to transparent, and form


clear melts. On warming, they dissolve in nonpolar solvents such as aliphatic,
aromatic, and chlorinated hydrocarbons, and on cooling, they crystallize as dis-
persions or gels (pastes).
Polar Polyolefin Waxes. Polar polyolefin waxes generally contain
carboxyl and ester groups, which give them special technical properties, in
Vol. 2 WAXES 155

particular emulsifiability in aqueous media. They can be produced by a variety of


methods:

1. Oxidation of nonpolar PE waxes (melt oxidation) or PE plastics (oxidative


degradation) with air
2. Radical high-pressure polymerization of ethylene with oxygen-containing
comonomers (acrylic acid, acrylates, vinyl acetate, etc.)
3. Radical grafting of polar unsaturated monomers onto nonpolar PE and PP
waxes

To a great extent, PE waxes are converted into wax oxidates by treatment


with atmospheric oxygen in the melt at ca. 150–160 8C (189–191). Oxygen func-
tions (carboxyl, ester, carbonyl, hydroxyl groups, etc.) are introduced into the
wax molecule by complex mechanisms involving peroxy intermediates and
chain-shortening reactions (192). The usual commercial products with acid num-
bers between 25 and 30 mg of KOH per gram contain ca. 3–4 wt % oxygen,
mainly in the form of carboxyl and ester groups.
Oxidation of the molten wax is limited to waxes whose molar mass is not
too high. In the case of longer-chain starting materials, intensive dispersion of
air in the melt, which is necessary for the reaction to run smoothly, is inhibited
by the melt’s high viscosity. In this case, waxes can be oxidized in inert dis-
persion agents. Since a (statistical) chain cleavage is always associated with
oxidation (ie, a lowering of molar mass), PE plastics can also be converted
into a polar wax by this route (oxidative degradation). For example, molten
polyethylene in an aqueous dispersion can be converted into emulsifiable
waxes by passing in air at 130– 160 8C with intensive stirring. These waxes
are harder and have higher drop points than the usual melt oxidates with
the same acid number because of their relatively high molar masses (193).
The reaction can be accelerated by addition of peroxides. High-density poly-
ethylene gives hard brittle products. Tough, flexible waxes, which can be
used as antislip components of floor polishes, can be obtained from LDPE con-
taining vinyl acetate (194).
Oxidative degradation of polyethylene can also be carried out as a gas–solid
reaction. In this case, no dispersing agent is required. Air, oxygen-enriched air,
or pure oxygen in the presence of radical initiators is reacted with the substrate
in powder form a few degrees below the softening point (195–198). Because a low
temperature (<130 8C) must be maintained, the oxidation period is compara-
tively long. For example, it requires at least 20 h to achieve acid numbers of
25–30 mg of KOH per gram. Use of air–ozone mixtures as the oxidizing agent
has also been described (199,200). Another process variant involves oxidation
of polyethylene powder with air in aqueous suspension under pressure (201).
The properties of some representative commercial wax and plastic oxidates are
given in Table 25.
Oxidized PE waxes can be modified further by esterification (202), amida-
tion (203–205), saponification, and other derivatization reactions (206) and
thus adapted to special technical requirements.
Polyolefin waxes can also be converted into emulsifiable products by
grafting unsaturated polar compounds (maleic anhydride, etc.) (161,207–210).
156 WAXES Vol. 2

Table 25. Properties of Typical Commercial Oxidized Polyolefin Waxes


Characteristic A B C D E
Raw material high-pressure Ziegler HDPE HDPE ethylene –
PE wax PE wax vinylacetate
copolymer
Acid number, mg KOH/g 22 18 16 60 21
(DGF-M-IV 2)
Saponification number, 50 38 103 80
mg KOH/g (DGF-M-IV 2)
Viscosity (120 8C), mPa  s 400 200 8500 (150 8C) 300 3000
Drop point, 8C (DGF-M-III 3) 100 116 140 111 100
Needle penetration, 0.1 mm 5 5 <0.5 1 2
(DGF-M-III 9b)
Density, g/cm3 0.96 0.98 0.98 1.02 0.96
(DGF-M-III 2a)

For polypropylene wax (211–213), this is the only method of introducing oxygen
functions that is used in practice, since air oxidation would lead to low molar
mass, colored, soft degradation products.
6.2. Uses. Nonpolar Polyolefin Waxes. Polyolefin waxes are used in
widely differing applications, mostly as auxiliaries in products and produc-
tion processes. The areas of use of high-pressure, Ziegler, and degradation
waxes overlap extensively. More details can be found in leaflets and bro-
chures of various production companies. Important areas of use include the
following:

1. In Additive and Pigment Master batches. PE and PP waxes are carrier


materials and binders in concentrates used as plastics additive and for col-
oration of plastics.
2. As Additives to Improve Rub Resistance and as Slip Agents (214). PE waxes
are incorporated into printing inks after micronizing by milling or spray-
ing, or in the form of dispersions.
3. In Paints and Coatings, PE waxes are used for matting effects and to
increase resistance to scratching (215,216).
4. In Plastics Processing, PE waxes are used as lubricants and release agents
in molding.
5. In Polishes. PE waxes are components of heat-resistant pastes for floor,
shoe, and car polishes.
6. In Hotmelt Coatings. PE and PP waxes are heat-sealable components of
coatings and adhesives for paper and metal sheeting because they improve
insulation, gloss properties) and heat resistance.
7. For Corrosion Protection. PE waxes are used as hydrophobing components
for temporary corrosion protection of motor vehicles, machines, and instru-
ments.
8. Photocopying. PE and highly crystalline PP waxes are used in toner pre-
parations for photocopying machines.
9. In the Rubber Industry, PE waxes are used as release agents in processing.
Vol. 2 WAXES 157

Also, PE and PP waxes are used as components of insulating compounds, as


carriers and binders for mechanically and chemically reacting self-duplicating
paper and wax crayons, to coat granular fertilizer, and to increase the heat resis-
tance of candles.
Polar Polyolefin Waxes. Polar polyolefin waxes are used mainly in the
form of aqueous emulsions (dispersions). Depending on the nature of the emulsi-
fier system used, nonionic, nonionic–ionic, anionic, and cationic formulations can
be produced. Emulsions are produced by mixing molten waxes and emulsifiers
with hot water. On rapid cooling with intensive stirring, wax droplets solidify
and form stable, finely divided dispersions. High-melting waxes must be emulsi-
fied under pressure. After the emulsion is applied, water evaporates and leaves a
homogeneous wax film, often with a high gloss, on the treated surface. This effect
is used in floor polishes where polar waxes are applied in combination with acry-
late- and styrene-based polymer dispersions. Waxes obtained by oxidative degra-
dation of plastics are primarily used in this area because they form harder-
wearing films due to their greater hardness than lower molar mass melt oxi-
dates. Polyethylene wax emulsions are also used as leather and shoe polishes
and in many other areas of industry. In textiles, they are used for wash and
wear finishes and to improve sewing properties. They are used as hydrophobing
agents in paint dispersions, as release agents in the building industry, and in
surface finishing of paper products. Citrus fruit is provided with a glossy surface
and protected from drying out through the application of wax films (217). In the
plastics processing industry, polar PE waxes are used as lubricants and release
agents.
6.3. Economic Importance. As indispensable additives with a secure
raw material supply and broad application potential, polyolefin waxes play an
important role, mainly in the industrial, but also in the consumer market sectors.
The quantity of polyolefin waxes annually consumed worldwide is ca.
200 000 t (1995), including oxidized and grafted products. High-pressure waxes
(PE and polar copolymers) account for ca. 70 %. The remainder consists essen-
tially of Ziegler products (PE, PP). Degradation waxes (mainly PP) are less
important in terms of quantity.
Ca. 12 000 t of polar waxes are produced worldwide by oxidative degrada-
tion of PE plastics.
The main consumers of nonpolar polyolefin waxes are the producers of pig-
ment concentrates and of plastic and rubber articles, as well as the paint and
printing ink industry. Polar waxes are used mainly in the form of emulsions in
the textile, polish, and paper industries.
Producers and Trade Names. High-Pressure Waxes. Allied-Signal,
United States (A-C Polyethylene). Eastman Kodak, United States (Epolene
Wax). BASF, Germany (Polyethylenwachs BASF, Luwax). Leuna, Germany
(LE-Wachs).
Ziegler Waxes. Hoechst, Germany (Hoechst-Wachs). Mitsui Petrochemical
Ind., Japan (Hi-wax). Hüls, Germany (Vestowax). Petrolite, United States
(Polywax).
Degradation Waxes. Sanyo Chemical Industries, Japan (Viscol, Youmex).
Mitsui Petrochemical Ind., Japan (NP-Types). Lion Chemical, Korea (L-C Wax).
Eastman Kodak, United States (Epolene). Ceralit, Brazil (Cerit Poly).
158 WAXES Vol. 2

7. Toxicology

On the basis of experience accumulated over thousands of years, waxes are gene-
rally considered harmless, nontoxic, environmentally friendly products. The
results of scientific toxicological and ecotoxicological tests on natural waxes, mon-
tan waxes, polyethylene waxes, and other wax products generally support this.
Natural Waxes. From the group of animal and vegetable waxes, carnaúba
wax was subjected to toxicological tests. In a two-generation feeding experiment,
1 % wax (max.) based on the feed, was administered first to pregnant rats and then
to their first-generation offspring over a period of 90 d. The results gave no indica-
tion of dose-dependent effects in the rats treated compared to the untreated con-
trol group (218).
Montan Waxes and Their Derivatives. The toxicological properties of
acid, ester, and partially saponified ester waxes (partially neutralized with
calcium hydroxide) derived from montan wax are determined mainly by the indi-
gestibility of the products, as in the case of natural waxes. Besides determining
acute oral toxicity, extensive feeding experiments on rats and dogs over periods
of 90 d, 140 d, and 2 a have been carried out with montan wax derivatives (219).
The LD50 values are >15 000 and 20 000 mg per kilogram of body weight, respec-
tively. According to the usual classification, these products are considered prac-
tically nontoxic and harmless (220).
Long-term feeding experiments led to no macroscopically or microscopically
recognizable toxicological damage to the animals, even at very high dosages of
5 % in the feed (Table 26).
Dermatological examination of montan wax derivatives in the percutaneous
test and the mucous membrane test on rabbits (Hoechst waxes S, E, and OP) did
not lead to any irritation of the skin or mucous membranes.
In the Ames mutagenicity test with and without external metabolic activa-
tion, a montanic acid–ethylene glycol ester wax (Hoechst wax E) did not show
any mutagenicity (tested dosages 4–5000 mg per plate) (221).
The ecotoxicological behavior of montan waxes is attributed to their water
insolubility. Mixtures with water are attacked very slowly by microorganisms
and cause no perceptible biological oxygen demand (BOD) in water bodies and
sewage treatment plants. The waxes are separated from wastewater as ballast
materials together with sludge. In the determination of biodegradability accord-
ing to OECD test method 302 B (static test/Zahn–Wellens test), the finely dis-
persed montan wax derivatives in wastewater show good elimination of
chemical oxygen demand (! Ecology and Exotoxicology). They are separated
with sludge and are nontoxic to fish and bacteria.
In the case of wax emulsions, when the usual test methods for determina-
tion of COD and BOD are used, a very slow oxygen consumption can be detected,
which is higher than the oxygen demand of the emulsifiers (219). In wastewater,
these emulsions break up as a result of great dilution, altered pH, and the pre-
sence of calcium and magnesium ions. The findings described for pure wax–
water mixtures also apply to the agglomerates formed.
Polyolefin Waxes. Polyethylene waxes, oxidized polyethylene waxes, and
polypropylene waxes are similar to those based on montan wax and natural waxes
with regard to toxicology. The LD50 values for rats are >5000 and 15 000 mg per
Vol. 2 WAXES 159

Table 26. Acute Oral, Subchronic, and Chronic Toxicity of Montan Wax Derivatives
Subchronic
toxicity, Chronic
Acute oral Subchronic 140-d toxicity, 2-a
toxicity, toxicity, 90-d no-effect no-effect
Chemical LD50, no-effect level *, level *, level *,
Wax composition mg/kg mg/kg mg/kg mg/kg
Acid wax montanic acid >15.000
(Hoechst C16 – C36 (rat)
wax S)
Ester wax esters of >20 000 50 000 50 000
(Hoechst montanic (mouse) (dog) (rat)
wax E) acid with
ethanediol
Ester wax esters of >15 000 50 000 (dog) 50 000
(Hoechst montanic (rat) (rat)
wax KPS) acid with
ethanediol
and 1,3-
butanediol
Ester wax esters of 50 000 (rat)
(Hostalub montanic
WE 4) acid with
glycerol
Partially esters of >20 000 50 000 50 000
saponified montanic (mouse) (dog) (rat)
ester wax acid with 1,
(Hoechst 3-butanediol
wax OP) and calcium
salts of
montanic
acid
* Proportion by weight in the feed.

kilogram of body weight, respectively. In a long-term toxicity test (a 9-month feed-


ing experiment on rats), polyethylene waxes caused no toxic damage to the animals
even at the highest dosage of 8.5 % in the feed. In a 90-d feeding experiment on
rats, oxidized polyethylene wax up to acid numbers 70 in dosages of 0.2, 1, 1.6,
and 5 % in the feed produced no adverse effects in the animals. Histopathological
tests showed no morphologically recognizable damage or changes caused by the
waxes. No accumulation was observed in tissues, liver, or lymph glands. Fat and
iron metabolism and spermatogenesis and oogenesis were unaffected (Table 27)
(219). Practically no negative effects were found in dermatological tests on rabbits.
Polyethylene waxes and oxidized polyethylene wax show good COD elimination
under sewage treatment plant conditions and no toxicity to fish or bacteria.
Paraffins and Microwaxes. Petroleum waxes have been used for
decades as food additives, as wax coatings for cheese, in chewing gum, for
upgrading food packaging, and in cosmetics and pharmaceutical products, with-
out damaging human health. The toxicological properties of paraffins and micro-
waxes extracted from crude petroleum depend essentially on their composition
and purity, particularly their residual aromatics content.
160 WAXES Vol. 2

Table 27. Acute oral, subchronic, and chronic toxicity of polyolefin waxes
Subchronic Subchronic
Acute oral toxicity, 90-d toxicity, 9-month
toxicity, LD50, (rat) no-effect (rat) no-effect
Wax mg/kg level *, mg/kg level *, mg/kg
Polyethylene wax (Hoechst >15 000 85 000
wax PE 520)
Polyethylene wax oxidates
(Hoechst wax PED 121) 16 000
(Hoechst wax PED 522/ >15 000 50 000
PED 136)
Polypropylene wax (Hoechst >5000
wax PP 230)
* Proportion by weight in the feed.

In a study of the toxicity of petroleum waxes, rats were fed with several
paraffin products for 2 years. At a dosage of 10 % wax in the feed, no toxicolo-
gical or carcinogenic effects were observed. Even on repeated application of
petroleum waxes to the skin of mice and rabbits, no carcinogenic effects
were detected (222).
Toxicological tests have been carried out with paraffins and microwaxes are
mainly 90-d feeding experiments on rats. In these, varying toxicological effects
were observed. For low molar mass paraffins, accumulation in the tissues and
increased organ weights of the liver, kidney, spleen and lymph glands were
observed. The most important histopathological findings were granulomata in
the liver and histiocytosis in the lymph glands. These effects were not found
with high molar mass waxes.
No toxic effects were found at a dose of 0.6 mg per kilogram of body weight
and day for low molar mass paraffins (C-number maximum 25) and 1.8 g per
kilogram of body weight per day for high molar mass paraffin (microwaxes,
C-number maximum 42).
The latest results are published in BIBRA reports (223,224), CONCAWE
reports (225), and reports of the Toxicology Forum (226,227).
Amide Waxes. The most well known amide wax, bis(stearoyl)- bis-
(palmitoyl)ethylenediamide, has an LD50 of >20 000 mg/kg (mouse). In determi-
nation of the chronic toxicity in a 2-a feeding experiment on rats, a non-effect
level of 5 % in the feed was established (218).
Approval for Food Contact. On the basis of their low toxicity, a range
of natural waxes, montan wax derivatives, amide waxes, polyethylene waxes,
paraffins, and microcrystalline waxes have been approved for food contact by
the health authorities of many countries [eg, the German BgVV (formerly
BGA), the U.S. FDA, and in Australia and Japan]. The products are also con-
tained in the draft version of the EC additives list (Synoptic Document no. 7 of
May, 15, 1994). Some of the waxes are approved in the European Directives
95/2/EC and 95/3/EC for food contact.
Registration According to EINECS/TSCA/AICS/DSL. Waxes are
registered under the CAS numbers given in Table 28 in the European Inventory
Vol. 2 WAXES 161

Table 28. Registration According to EINECS, TSCA, AICS, DSL


Type of wax CAS registry no.
Natural waxes
Beeswax [8012-89-3]
Carnaúba wax [8015-86-9]
Candelilla wax [8006-44-8]
Waxes based on montan wax
Montan wax fatty acids [68476-03-9]
Ethylene esters of montan wax fatty acids [73138-45-1]
1-Methyl-1,3-propanediyl esters of [73138-44-0]
montan wax fatty acids
Calcium salts of montan wax fatty acids [68308-22-5]
Montan wax glycerides [68476-38-0]
Montan wax (crude montan wax) [8002-53-7]
Polyolefin waxes
Polyethylene wax [9002-88-4] *
Ethylene [74-85-1]
Polypropylene wax [9003-07-0] *
Propylene [115-07-1]
Polyethylene wax oxidates [68441-17-8] *
Petroleum waxes
Paraffin and hydrocarbon waxes [8002-74-2]
Microcrystalline waxes [63231-60-7]
Amide wax
N,N0 -Ethylenedistearamide [110-30-5]
* In EINECS, polymers are registered only under the CAS registry no. of
the monomer.

of Existing Chemical Substances (EINECS), the TSCA inventory in the United


States, the Australian AICS inventory, and the Canadian DSL inventory.
Waxes are not in the MAK value list for 1995.
Labeling According to GefStoffV, ChemG, and EC Guidelines. If
they are pure and are not in the form of preparations with components for
which labeling is compulsory (eg, emulsifiers), waxes do not have to be
labeled according to GefStoffV (October 26, 1993), the ChemG, Technische
Regeln für Gefahrstoffe (technical regulations governing hazardous sub-
stances, TRGS) 200, and the EC Guideline 67/548/EC and adaptations and
amendments.
Status According to OSHA. Paraffins are regulated by OSHA. Labeling
is prescribed on packaging of paraffins (right-to-know label), in which warning of
paraffin vapors is given. Paraffin vapors are classified as eye and respiratory irri-
tants; they have the following exposure limits:

OSHA-PEL 2 mg/m3
ACGIH-TLV 2 mg/m3

Other waxes are not regulated by OSHA.


162 WAXES Vol. 2

REFERENCES

1. Chr. Columbus, Bordbuch, November, 29, 1492. Reprint: edition Erdman, Thienemann,
Stuttgart 1983.
2. M. E. Chevreul: ‘‘Recherches chimiques sur les corps gras d’origine animale,’’ Frankreich
1823.
3. Hoechst AG: ‘‘Vom Wachs,’’ Hoechster Beiträge zur Kenntnis der Wachse, Frankfurt
1959 ff.
4. DGF-Einheitsmethoden, Abteilung M: Wachse und Wachsprodukte, Stuttgart
1975.
5. Erläuterungen des Rates für die Zusammenarbeit auf dem Gebiet des Zollwesens
3404/1 (1992) und 3404/2 (1988): International Customs Tariff, Harmonized System
HS 3404: ‘‘Artificial Waxes (Including Water-Soluble Waxes); Prepared Waxes, not
Emulsified or Containing Solvents’’.
6. J. Haggin: ‘‘Asian Natural Gas Pipeline Proposed for Easing Energy Feedstocks
Strains,’’ Chem. Eng. News (1994) no. 6, 17–18.
7. ‘‘Wax Data,’’ Newsletter for Marketers and Consumers of Petroleum, Natural and
Synthetic Waxes, Rauch Associates Inc., Bridgewater, New Jersey (published 13
times annually).
8. Kahl & Co., Otto-Hahn-Str. 2, 22946 Trittau, personal communication, 1994.
9. ‘‘Animal and Vegetable Waxes–A Review,’’ Wax Data 1993, Rauch Associates Inc.,
P.O. Box 6802, Bridgewater, N.J. 08807.
10. Chem. Mark. Rep. 244, (1993) no. 12, 7, 11.
11. ‘‘Study Forecasts Growth for Natural and Synthetic Waxes,’’ Wax Data 1994, Rauch
Associates Inc., P.O. Box 6802, Bridgewater, N.J. 08807.
12. A. H. Warth: The Chemistry and Technology of Waxes, Reinhold Publ. Corp.,
New York 1956.
13. R. Büll et al.: ‘‘Vom Wachs,’’ vol. 1; Hoechster Beiträge zur Kenntnis der Wachse, vol. 1
(contributions 1–8); vol. 2 (contributions 1–10), Hoechst AG, Werk Gersthofen,
Augsburg 1958–1970.
14. E. B. Kurtz, Jr., Fette Seifen Anstrichmittel 61 (1959) no. 6, 505.
15. E. J. Fischer, W. Presting: Laboratoriumshandbuch für die Untersuchung tech-
nischer Wachs-, Harz- und Ölgemenge, VEB Wilhelm Knapp Verlag, Halle/Saale
1958.
16. L. Ivanovszki: Wachs-Enzyklopädie, Verlag für Chem. Industrie H. Ziolkowsky,
vol. 1 Augsburg, 1954; vol. 2, 1960.
17. W. Piso, Historia Naturalis Brasiliae (1648).
18. J. B. De Moraes: Ensaios sobre à Carnanbeira, Nat. Ölinstitut, Rio de Janeiro 1942.
19. J. B. Nogueira, R. D. Machado: Glossario da Palmeiras, Nat. Ölinstitut, Rio de
Janeiro 1950.
20. Am. Wax Importers and Refiners Assoc.: Blue Book, New York 1973.
21. Chem. Fabrik Dessau, DE 581 130, 1931.
22. Dtsch. Hydrierwerke, DE 658 749, 1932.
23. Degussa, DE 938 501, 1942.
24. E. Eisenring, CH 271 936, 1950.
25. W. A. Sprague, US 1 658 062, 1926.
26. I.G. Farbenind., US 1 989 626, 1932.
27. S. C. Johnson & Son, US 2 527 481, US 2 531 785, 1948.
28. G. Kronbach, Seifen Öle Fette Wachse 99 (1973) 97.
29. L. E. Vandenburg, E. A. Wilder, J. Am. Oil Chem. Soc. 44 (1967) 659.
30. S. Markley, Econ. Bot. 9 (1955) 39.
Vol. 2 WAXES 163

31. Seifen Öle Fette Wachse 86 (1960) 422.


32. E. A. Wilder: ‘‘Chemical Specialties Manufacturers Assn.,’’ 61st. Mid-Year Meeting,
Washington 1975.
33. G. Illmann, Fette Seifen Anstrichmittel 81 (1979) no. 8, 322.
34. The Candelilla Wax Industry FONCAM; Fideicomisos Relacionados con la Cera de
Candelilla; Banco Nacional de Crédito Rural, S.A. Mexico, company brochure.
35. Chem. Mark. Rep. 243 (1993) no. 2, 21.
36. C. Lüdecke, L. Ivanovszky: Taschenbuch für die Wachsindustrie, 4th ed.,
Wissenschaftl. Verlagsgesellschaft, Stuttgart 1958, pp. 33–36.
37. L. J. N. Cole, J. B. Brown, J. Am. Oil Chem. Soc. 37 (1960) 359.
38. Chem. Mark. Rep. 241 (1992) no. 6, 10, 11.
39. P. J. Manohar Rao, Coop. Sugar/India 13 (1981) no. 2, 1.
40. H. Schmidt: ‘‘Analytik von Naturwachsen am Beispiel des Zuckerrohrwachses,’’
DGF-Vortragstagung, Mainz 1967.
41. A. Schwab: Jojoba, ein hochwertiges Pflanzenöl aus der Wüste, Ledermann-Verlag,
Bad Wörishofen 1981.
42. E. Zander: Handbuch der Bienenkunde in Einzeldarstellung, Verlag E. Ulmer,
Stuttgart 1946.
43. K. V. Frisch: Aus dem Leben der Bienen, Springer Verlag, Heidelberg 1956.
44. G. Titschak, Fette Seifen Anstrichmittel 71 (1969) no. 5, 369–379.
45. R. J. Holloway, J. Am. Oil Chem. Soc. 46 (1969) 189–190.
46. A. P. Tulloch, J. Am. Oil Chem. Soc. 50 (1973) 269.
47. A. P. Tulloch, Bee World 61 (1980) 47.
48. A. P. Tulloch, Lipids 5 (1970) 247.
49. A. P. Tulloch, Chem. Phys. Lipids 6 (1971) 235.
50. A. P. Tulloch, L. L. Hoffmann, J. Am. Oil Chem. Soc. 49 (1971) 696.
51. K. Stransky, M. Streibl, Collect. Czech. Chem. Commun. 36 (1971) no. 6, 2267.
52. K. Stransky, M. Streibl, V. Kubelka, Collect. Czech. Chem. Commun. 36 (1971) no. 5,
2281.
53. K. Stransky, K. Ubik, M. Streibl, Collect. Czech. Chem. Commun. 37 (1972) no. 12,
4099.
54. H. Brüschweiler, H. Felber, F. Schwager, Fett Wiss. Technol. 91 (1989) no. 2, 73–79.
55. H. Schmidt: ‘‘Analytische Untersuchungen an Schellackwachs,’’ DGF-Vortragsta-
gung, Freiburg 1981.
56. C. A. Anderson, J. Am. Oil Chem. Soc. 40 (1963) 333–336.
57. V. Vcelák: Chemie und Technologie des Montanwachses, Verlag der tschechoslowa-
kischen Akademie der Wissenschaften, Prague 1959.
58. A. H. Warth: The Chemistry and Technology of Waxes, Reinhold Publ. Corp.,
New York 1956.
59. E. J. Fischer, W. Presting: Laboratoriumsbuch für die Untersuchung techni-
scher Wachs-, Harz- und Ölgemenge, VEB Wilhelm Knapp Verlag, Halle (Saale)
1958.
60. A. Lissner, A. Thau: Die Chemie der Braunkohle, vols. 1 and 2, VEB Wilhelm Knapp
Verlag, Halle (Saale) 1952.
61. E. Riebeck, Dissertation, Freiberg 1880.
62. Von Boyen, DE 101 373, 1897.
63. VEB Braunkohlekombinat Röblingen: Festschrift, ‘‘75 Jahre Herstellung von
Montanwachs,’’ Röblingen 1972.
64. R. Büll: Informationsschrift ‘‘Vom Wachs,’’ 2 vols, Hoechst AG, Frankfurt-Höchst
1959.
65. BIOS Final Report 1759 and 1831, Technical Information and Document Unit,
London 1947.FIAT Final Report 737, Field Information Agency, London 1946.
164 WAXES Vol. 2

66. VEB Braunkohlekombinat Gustav Sobotka, DD 135 089, 1979 (H. J. Grauen et al.).
67. Hoechst AG, DE 954 903, 1953 (R. Schirmer); DE 957 508, 1955 (R. Schirmer);
DE 1 007 918, 1955 (R. Schirmer); DE 2 855 263, 1979 (A. Riedel); DE 2 915 764,
1979 (H. Stetter); DE 2 915 802, 1979 (H. Stetter); DE 2 915 755, 1979 (H. Stetter).
68. BASF, DE 3 726 514, 1987 (L. Schuster).
69. O. Malitschek: Hoechst Wachse, Ein Leitfaden für Wachsanwender, Hoechst AG,
SBU Wachse und Additive, Werk Gersthofen 1990.
70. R. Gächter, H. Müller: Kunststoffadditive, Hanser Verlag, München 1993.
71. P. Piesold: ‘‘Montanwachs als Kunststoffadditiv,’’ in J. Edenbaum (ed.): Plastics,
Additives and Modifiers Handbook, Van Nostrand Reinhold, New York 1992,
pp. 757–772.

GENERAL REFERENCES

72. DGF-Einheitsmethoden: Untersuchung von Fetten, Fettprodukten und verwandten


Stoffen, Abteilung M: ‘‘Wachse und Wachsprodukte,’’ Wissenschaftl. Verlagsges.,
Stuttgart 1975.
73. H. Bennet: Industrial Waxes, vol. 2, Compounded Waxes, Chem. Publ. Comp.,
New York 1963.
74. A. H. Warth: The Chemistry and Technology of Waxes, Reinhold Publ. Co., New York
1956.
75. J. Teubel, W. Schneider, R. Schmiedel: Erdölparaffine, VEB Dtsch. Verlag f. Grund-
stoffind., Leipzig 1965.
76. F. Asinger: Die petrochemische Industrie, Akademie-Verlag, Berlin 1971, part I,
pp. 33–70.
77. C. Zerbe: Mineralöle und verwandte Produkte, 2nd ed., Springer Verlag, Berlin 1969,
part 1, pp. 469 –488; part 2, pp. 581–593.
78. R. Sayers: Wax–An Introduction, Gentry Books Ltd., London 1983.
79. M. Freund et al.: Paraffin Products, Elsevier Sc. Publ. Co., Amsterdam–Oxford–
New York 1982.

SPECIFIC REFERENCES

80. European Wax Federation (EWF): Glossary of Wax Terminology, Brussels 1981.
81. C. Kajdas, Seifen Öle Fette Wachse 96 (1970) 251–260.
82. C. Kajdas, Seifen Öle Fette Wachse 112 (1986) 471–474.
83. G. Matscholl, Coating 14 (1981) 117–118.
84. I. A. Michailov, Chem. Tech. (Leipzig) 20 (1968) 351–353.
85. ‘‘Assessment and Comparison of the Composition of Food-Grade White Oils and
Waxes Manufactured from Petroleum by Catalytic Hydrogenation Versus Conven-
tional Treatment,’’ CONCAWE Rep. 84/60 (1984) 18–26, 33–46.
86. European Wax Federation (EWF): Specifications of Petroleum Waxes, Brussels 1991.
87. F. V. Synder: Presentation at National Petroleum Refiners Assoc., Houston, Texas
1979.
88. Röhm GmbH, Spektrum 39, company brochure, Darmstadt 1992.
89. CONCAWE 1991: Report no. 91/ 33, 58–62 CONCAWE, Brussels.
90. R. J. Taylor, Oil Gas J. 91 (1993) no. 4, 48–50.
91. D. A. Gudelis et al., Hydrocarbon Process. 52 (1973) no. 9, 141.
92. M. Soudek, Hydrocarbon Process. 53 (1974) no. 12, 64, 66.
Vol. 2 WAXES 165

93. Hydrocarbon Process. 59 (1980) no. 9, 201.


94. L. M. L. Tobing, Hydrocarbon Process. 56 (1977) no. 9, 143–145.
95. Hydrocarbon Process. 59 (1980) no. 9, 180, 181, 192, 200, 201, 206.
96. Hydrocarbon Process. 59 (1980) no. 9, 218.
97. Shell Oil, US 4 701 254, 1987 (L. W. Maas, R. A. Geimann, C. T. Adams).
98. F. Wolf et al., Erdöl Kohle Erdgas Petrochem. 24 (1971) 79–82, 153–155.
99. W. Himmel et al., Erdöl Kohle Erdgas Petrochem. 39 (1986) 408–414.
100. P. Polanek et al., Erdöl Erdgas Kohle 109 (1993) 324–326.
101. U.S. Depart. of Health, Education and Welfare; Food and Drug Admin. (FDA),
Part 21, Paragraph 172.886.
102. FAO Food and Nutrition Paper 52, Rome 1992.
103. BGA-Regulation XXV.
104. Mineral Hydrocarbons in Food Regulations, SI 1966, No. 1073 (1966).
105. European Wax Federation: Summaries of National Regulations Affecting Wax
Usage, Brussels 1991.
106. A. Case, Adhes. Age 33 (1990) 28–31.
107. A. Aduan et al., J. Pet. Res. 6 (1987) 63–76.
108. G. Michalczyk, Fette Seifen Anstrichmittel 87 (1985) 481–486.
109. M. P. Goutam et al., Seifen Öle Fette Wachse 118 (1992) 1277–1281.
110. D. T. Kaschani, DGMK-Projekt 242, Hamburg 1992.
111. G. Grimmer et al., Chromatographia 9 (1976) 30–40.
112. Chem. Mark. Rep. 26 (1980) no. 5, 7.
113. Chem. Purch. 5 (1980) 74–85.
114. R. Hournac, Hydrocarbon Process. 60 (1981) no. 1, 207–211.
115. A. E. Dustan, Soap Cosmet. Chem. Spec. 56 (1980) no. 12, 42B–42E.
116. K. Marche, Cleveland Rubber Group Meeting, Los Angeles 1984.
117. Ministery of Agriculture, Fisheries and Food News Release 278/93, London 1993.
118. EEC-Directive 90/128 (EC 1990) and appendix lists.
119. D. Monzie, Document of Chemistry and Technology of Petroleum Waxes (CTP)
No. 776 (1976), Special report of the Institute of Petroleum, London.
120. R. Franz et al., Z. Lebensm. Technol. Verfahrenstech. 43 (1992) 291–296.
121. D. van Battum, TNO-Rep. B 77/2694 A (1979) 26–30.
122. J. Rustige, Presentation at Toxicology Forum;Special Meeting on Mineral Hydrocar-
bons, Green College, Oxford 1992.
123. A. O. Hanstveit, TNO-Rep. R 90/1986 (1990) 18–40.
124. DGMK Forschungsbericht 461–01, Hamburg 1991.
125. J. Davie: ‘‘Polymers, Laminations and Coatings Conference,’’ TAPPI Proc. 1992,
179–183.
126. A. O. Hanstveit, TNO-Rep. R 90/243 a (1991) 18–46.
127. R. Wenzel-Hartung, Erdöl Erdgas Kohle 108 (1992) 424–426.
128. J. G. E. McEwen, Progr. Pap. Recycl. 1992, 11–20
129. J. T. E. Lemmen, TNO-Rep. PK 92.1051 (1992) 33, 38–46.
130. A. Slaughter, Pet. Rev. 98 (1993) 258–260.
131. Hydrocarbon Process. 72 (1993) no. 8, 36.

GENERAL REFERENCES

132. A. Davidsohn, B. M. Milwidsky: Polishes, Leonard Hill, London 1968. A. Renfrew, P.


Morgan; Polyethylene, 2nd ed., Iliffe & Sons, London; Interscience Publ., New York
1960.
166 WAXES Vol. 2

SPECIFIC REFERENCES

133. I.G. Farbenind., DE 745 425, 1939 (H. Hopff, S. Goebel).


134. I.G. Farbenind., DE 864 150, 1942 (R. Kern).
135. Allied Chemical, US 2 504 400, 1947 (M. Erchak Jr.).
136. G. Luft, Chem. Ing. Tech. 51 (1979) 960–969.
137. BASF, DE 1 720 273, 1967 (O. Büchner et al.); DE 1 927 527, 1969 (H. Pfannmüller,
F. Urban, W. Hensel, O. Büchner); DE 1 951 879, 1969 (K. Stark, H. Gropper, F.
Urban, H. Pfannmüller); DE 2 034 534, 1970 (F. Urban, O. Büchner, W. Hensel, K.
Stark).
138. Eastman Kodak, US 2 772 259, 1953 (H. J. Hagemeyer Jr.); GB 861 277, 1959 (D. C.
Hull, R. J. Schrader).
139. Allied Chemical, US 2 683 141, 1952 (M. Erchak Jr.).
140. Allied Chemical, BE 780 023, 1971 (J. E. Dench, H. Knutson, M. K. Seven).
141. Allied Chemical, DE 2 233 360, 1971 (J. E. Dench, H. Knutson, M. K. Seven).
142. M. J. Roedel, J. Am. Chem. Soc. 75 (1953) 6110–6112. W. M. D. Bryant, R. C. Voter,
J. Am. Chem. Soc. 75 (1953) 6113–6118. F. W. Billmeyer, J. Am. Chem. Soc. 75
(1953) 6118–6122.
143. R. Kuhn, H. Krömer G. Rossmanith, Angew. Makromol. Chem. 40/41 (1974)
361–389.
144. G. Baum, J. Boelter, Fette Seifen Anstrichmittel 70 (1968) 15–19.
145. N. Nakajima, C. F. Stark, R. D. Hoffmann, Trans. Soc. Rheol. 15 (1971) 647–662.
146. E. Brauer, H. Wiegleb, M. Helmstedt, W. Eifert, Plaste Kautsch. 24 (1977) 463–468.
147. H. Hendus, G. Schnell, Kunststoffe 51 (1961) 69–74.
148. G. Ehrlich, G. A. Mortimer, Adv. Polym. Sci. 7 (1970) 386–448.
149. Chusei Oil Co., US 5 223 122, 1992 (Y. Katayama).
150. Hoechst, DE-OS 1 929 863, 1969 (B. Diedrich, K. Rust, R. Klein).
151. Hoechst, DE-OS 1 942 734, 1969 (B. Diedrich, R. Klein).
152. Mitsui Petrochemical, DE 2 241 057, 1972 (T. Tomoshige).
153. Mitsui Petrochemical, EP-A 0 021 700, 1980 (T. Ueda).
154. Leuna-Werke, DE-OS 4 235 404, 1992 (P. Adler, J. Krüger, M. Gebauer, R.-P.
Pfitzer).
155. Scholven-Chemie, DE-AS 1 301 524, 1965 (K. Schmitt et al.).
156. VEBA-Chemie, DE-OS 1 520 927, 1963 (K. Schmitt, F. Gude).
157. VEBA-Chemie, DE-OS 1 645 409, 1965 (K. Schmitt, G. Baum, F. Gude,
W. Nagengast).
158. Hoechst, DE-OS 2 257 917, 1972 (M. Engelmann, R. Höltermann).
159. VEBA-Chemie, DE-OS 1 645 423, 1965 (K. Schmitt, F. Gude).
160. VEBA-Chemie, DE-OS 1 520 925, 1963 (K. Schmitt, F. Gude).
161. Scholven-Chemie, DE-OS 1 520 914, 1962 (K. Schmitt, G. Keller).
162. Hoechst, DE-OS 2 329 641, 1973 (M. Engelmann).
163. BASF, EP-A 480 190, 1991 (R. Schlund et al.).
164. Hoechst, EP-A 584 586, 1993 (G. Hohner).
165. F. Kübler, New Sci. (1993) Aug., 28.
166. R. Mühlhaupt, Nachr. Chem. Tech. Lab. 41 (1993) 1341.
167. Mitsui Petrochemical, JA 61–236 804, 1985 (T. Tsutsui, M. Kioka, N. Kashiwa).
168. Mitsui Petrochemical, JA 62–129 303, 1985 (T. Tsutsui, A. Toyota, N. Kashiwa).
169. Hoechst, DE-OS 3 743 322, 1987 (H. Lüker).
170. Exxon Chemical Patents, US 4 914 253, 1988 (M. Chang).
171. Hoechst, DE-OS 4 217 378, 1992 (H.-F. Herrmann et al.).
Vol. 2 WAXES 167

172. Hoechst, EP 602 509, 1993 (R. Hess, H. Voigt, L. Böhm, H.-F. Herrmann).
173. Hoechst, DE-OS 3 743 320, 1987 (A. Winter, M. Antberg, J. Rohrmann).
174. Hoechst, DE-OS 3 904468, 1989 (A. Winter et al.).
175. Mitsui Toatsu Chemicals, EP-A 563 834, 1993 (Y. Sonobe, M. Mizutani, T. Shiomura,
N. Hirayama).
176. BASF, DE 922 618, 1952 (H. Hopff, H. Eilbracht).
177. Eastman Kodak, FR 1 289 766, 1960 (H. W. Coover, Jr., M. A. McCall, J. E. Guillet).
178. W. G. Oakes, R. B. Richards, J. Chem. Soc. 1949, 2929–2935.
179. A. Schwarz, G. Cramer in Polyethylene, 2nd ed., Iliffe & Sons, London; Interscience
Publ., New York 1960, pp. 232–239.
180. BASF, DE-AS 1 940 686, 1969 (H.-G. Trieschmann et al.).
181. Allied Chemical, US 3 345 352, 1964 (J. J. Baron, J. P. Rahus).
182. Eastman Kodak, US 2 835 659, 1957 (J. E. Guillet).
183. Eastman Kodak, US 3 519 609, 1967 (R. L. McConnell, D. A. Weemes).
184. J. E. Guillet, H. W. Coover, SPE J. 16 (1960) 311–314.
185. Eastman Kodak, US 3 562 788, 1967 (D. A. Weemes, R. L. McConnell).
186. H. W. Coover, R. L. Combs, J. E. Guillet, D. A. Weemes, J. Appl. Polym. Sci. 11
(1967) 1271–1281.
187. BASF, DE 1 364 747, 1962.
188. Eastman Kodak, US T 869 011, 1966 (M. O. Bronson).
189. A. Thalhofer in R. Vieweg, A. Schley, A. Schwarz (eds.): Kunststoff-Handbuch,
vol. VI, Polyolefine, Hanser Verlag, München 1969, p. 161.
190. VEB Leuna-Werke, DD 283 730, 1988 (I. Klimek et al.).
191. Mitsui Petrochemical, DE-OS 2 126 725, 1971 (A. Mori et al.).
192. N. M. Emanuel, E. T. Denisov, Z. K. Maizus: Liquid-Phase Oxidation of Hydrocar-
bons, Plenum Press, New York 1967.
193. Hoechst, DE-OS 2 035 706, 1970 (H. Korbanka, K.-H. Stetter).
194. Hoechst, EP 28 384, 1980 (H. Korbanka et al.).
195. Allied Chemical Corp., DE-OS 1 495 938, 1963 (R. W. Bush, C. L. Kehr).
196. Allied Chemical Corp., DE-OS 1 520 008, 1964 (C. L. Kehr).
197. Allied Chemical Corp., DE-OS 1 570 652, 1965 (G. L. Braude, S. T. Phoa).
198. BASF, DE-OS 3 112 163, 1981 (A. Hettche et al.).
199. BASF, DE-OS 3 720 952, 1987 (L. Schuster et al.).
200. BASF, DE-OS 3 720 953, 1987 (L. Schuster et al.).
201. BASF, DE-OS 3 238 652, 1982 (R. W. Reffert et al.).
202. Hoechst, DE-OS 2 201 862, 1972 (H. Korbanka, K.-H. Stetter).
203. Th. Goldschmidt, EP 0 257 506, 1987 (E. Ruf).
204. Th. Goldschmidt, EP 0 258 724, 1987 (E. Ruf).
205. Th. Goldschmidt, EP 0 259 652, 1987 (E. Ruf).
206. Hüls, EP-A 0 451 352, 1990 (A. Kühnle, V. Schrenk, M. Vey, E. Wolf).
207. Mitsui Petrochemical, DE-OS 285 3862, 1978 (A. Tachi, T. Tomoshige, H. Furuta, N.
Matsuzawa).
208. Petrolite Corp., US 3 590 076, 1967 (W. J. Heintzelmann, M. I. Naiman).
209. Eastman Kodak Co., US 4 028 436, 1975 (R. T. Bogan, C. M. Shelton).
210. Mitsui Petrochemical, EP 0 024 034, 1980 (T. Tomoshige, H. Furuta, A. Tachi, N.
Kawamoto).
211. Eastman Kodak Co., US 3 480 580, 1965 (F. B. Joyner, R. L. McConnell).
212. Eastman Kodak Co., US 3 481 910, 1965 (M. O. Brunson).
213. Eastman Kodak Co., US 5 290 954, 1992 (T. D. Roberts, T. W. Smith, C. A. Galvin).
214. K. Rieger, Fette Seifen Anstrichmittel 73 (1971) 231.
215. K. Rieger, G. Grimm, Adhäsion 11 (1977) 311.
168 WAXES Vol. 2

216. R. Köhler, Fette Seifen Anstrichmittel 87 (1985) 214.


217. J. Wildgruber, Fette Seifen Anstrichmittel 84 (1982) 75.
218. Results of investigation of Food and Drug Research Ltd., private communication of
J. Maehlmann & Cie., Fortaleza, Brazil 1982.
219. J. Lange, J. Wildgruber, Fette Seifen Anstrichmittel 78 (1976) 62–69.
220. W. S. Spector: Handbook of Toxicology, W. B. Saunders, Philadelphia–London
1956.
221. Hoechst, Pharma Research Toxicology, unpublished results, Frankfurt 1986.
222. P. Shubik et al.: Toxicology and Applied Pharmacology, vol. 4, Supplement,
Academic Press, New York–London 1962.
223. BIBRA 1992: ‘‘A 90 Day Feeding Study in the Rat, With Six Different Mineral Oils
and Three Different Mineral Waxes,’’ BIBRA Project no. 3, 1010, BIBRA Toxicology,
Carshalton, Surrey, U.K.
224. BIBRA 1993: ‘‘A 90 Day Feeding Study in the Rat, With Two Mineral Waxes,’’ (an
Intermediate and a Micro-paraffin Blend), BIBRA Project no. 3, 1205, BIBRA
Toxicology, Carshalton, Surrey, U.K.
225. CONCAWE 1993: ‘‘Report no. 93/56, White Oil and Waxes, Summary of 90 Day
Studies,’’ CONCAWE, Brussels.
226. Toxicology Forum 1992: Special Meeting on Mineral Hydrocarbons Sept. 21–23,
1992, Green College, Oxford, U.K. The Toxicology Forum, Washington, D.C.
227. Toxicology Forum 1993: Mineral Hydrocarbons, 1993 Winter Meeting, Feb. 15–17,
1993, Toxicology Forum, Washington, D.C.

UWE WOLFMEIER
HANS SCHMIDT
FRANZ-LEO HEINRICHS
JOSEF WILDGRUBER
GERD HOHNER
Hoechst Aktiengesellschaft, Werk Gersthofen,
Augsburg, Federal Republic of Germany
GEORG MICHALCZYK
DEA Mineralöl AG,
Hamburg, Federal Republic of Germany
WOLFGANG PAYER
KLAUS BOEHLKE
Hoechst Aktiengesellschaft, Werk Ruhrchemie,
Oberhausen Federal Republic of Germany
WOLFRAM DIETSCHE
BASF Aktiengesellschaft,
Ludwigshafen, Federal Republic of Germany
Vol. 2 ASPHALT 169

ASPHALT
Asphalt [8052-42-4] is defined by the American Society for Testing and Materials
(ASTM) (1) as a dark brown to black cementitious material in which the predo-
minating constituents are bitumens that occur in nature or are obtained in pet-
roleum processing. Bitumen is a generic term defined by ASTM as a class of black
or dark-colored (solid, semisolid, or viscous) cementitious substances, natural or
manufactured, composed principally of high molecular weight hydrocarbons, of
which asphalts, tars, pitches, and asphaltites are typical.
Canadian and European practice (2–4) and geologists and archaeologists in
the United States use bitumen or asphaltic bitumen as a synonym for asphalt,
and apply asphalt to the mixture of bitumen and inorganic matter that is used
for paving purposes. On the other hand, pitches and tars are derived from the
destructive distillation of coal, crude oils, and other organic materials.
ASTM (1) further classifies asphalts or bituminous materials as solids, semi-
solids, or liquids. Semisolid and liquid asphalts predominate in commercial prac-
tice today. Prior to 1907, the principal portion of asphalt used occurred naturally;
since that time most asphalts have been produced from the refining of petroleum.
Air-blown petroleum asphalts (5) of diverse hardness became available in the early
1900s, and as their use increased, the demand for native products diminished.
More recently, asphalts have been defined as the manufactured materials
that are produced during petroleum processing whereas bitumens have been
defined as the naturally occurring materials (5,6). That is, petroleum and related
materials can be divided into various class subgroups:

Natural Materials Manufactured Materials Derived Materials


petroleum wax oils
heavy oil residuum resins
mineral wax asphalt asphaltenes
bitumen (native asphalt) tar carbenes
asphaltite pitch carboids
asphaltoid coke
migrabitumen synthetic crude oil
bituminous rock
bituminous sand
kerogen
natural gas

The term tar sands is a misnomer; tar is a product of coal processing. Oil
sands is also a misnomer but equivalent to usage of ‘‘oil shale.’’ Bituminous
sands is more correct; bitumen is a naturally occurring asphalt. Asphalt is a
product of a refinery operation, usually made from a residuum. Residuum is
the nonvolatile portion of petroleum and often further defined as atmospheric
(bp > 350 C) or vacuum (bp > 565 C). For convenience, the terms ‘‘asphalt’’
and ‘‘bitumen’’ will be used interchangeably in this article.
Asphalts characteristically contain very high molecular weight molecular
polar species, called asphaltenes, which are soluble in carbon disulfide, pyridine,
aromatic hydrocarbons, chlorinated hydrocarbons, and tetrahydrofuran.

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.0119160819160509.a01
170 ASPHALT Vol. 2

1. Naturally Occurring Materials

1.1. Native Asphalt (Bitumen). Construction activities in ancient


times depended on native asphalts (2,3,7–13). The product found on the sur-
face of the Dead Sea may have been used by the Egyptians for mummification.
In fact, the word mummy is thought to have derived from mumiia or Bitumen
of Judea. Standard grades of bitumens apparently were known to the Sumer-
ians; olive oil was used as the flux and fibers were added, if necessary, to pre-
vent flow of the mixture. Clay-stabilized mastics were used as flooring and for
waterproofing aprons. The presence of asphalt seepages signals a probable
crude-oil reservoir. Several centuries later, mention is made of Trinidad
asphalt by Raleigh during his visit to the island in 1595.
1.2. Lake Asphalt. Trinidad Lake asphalt was first reported used for
paving in the United States about 1874 (14), and the use of the asphalt from Ber-
mudez Lake in Venezuela soon followed. Trinidad asphalt (3,7) exists in a num-
ber of deposits on the island of Trinidad, near the northeast coast of Venezuela.
The largest deposit, and the one of chief commercial importance, is an asphalt
lake occupying about 0.4 km2 and of uniform composition to a depth of 87 m.
As the bitumen is mined, the openings fill in, the level appears constant and
an overturning movement of the lake from center to edge keeps the composition
constant. This asphalt has been used for pavement constructions and other appli-
cations throughout the world.
Trinidad asphalt has a relatively uniform composition of 29% water and
gas, 39% bitumen soluble in carbon disulfide, 27% mineral matter on ignition,
and 5% bitumen that remains adsorbed on the mineral matter. Refining
is essentially a process of dehydration by heating the crude asphalt to ca
1658C. The refined product averages 36% mineral ash with a penetration at
258C of about 2 (0.2 mm), a softening point (ring and ball method) of 998C,
a flash point (Cleveland open cup) of 2548C, a sulfur content of 3.3%, and a
saponification value of 45 mg KOH/g. The mineral matter typically contains
particle sizes from 40% finer than 10 mm to 10% from the 30 –75 mm range.
Because of the influence of the fine mineral particles, many of them colloidal
in size, the viscosity of Trinidad Lake asphalt is higher than that of normal
bitumens.
Asphalt from Bermudez Lake (7) previously was exported in bulk in cargo
vessels to the United States. In the early 1900s, it was used extensively for
paving and waterproofing, but has not been available since the 1940s.
1.3. Rock Asphalt. Rock asphalt (15) deposits contain from 5–25%
asphalt (Table 1). When extracted the asphalt varies from a soft-flux material
to hard penetration. The mineral content, usually of a sandstone or limestone
nature, has many gradations. Rock asphalt mixtures, ground and sized,
are placed and rolled to form pavement surfaces. Some have superior antiskid
properties in wearing surfaces, particularly the sandstone type. Rock
asphalts are found in Texas, Alabama, Oklahoma, Colorado, California, and
Kentucky. The composition of Kentucky asphalt is suitable for paving in its
natural condition if selected at bitumen contents of 7–9% (16). Some of
the well-known and long-used asphalt deposits of Europe (3,7) are located at
Vol. 2 ASPHALT 171

Table 1. Typical Composition and Properties of Rock Asphalts


Property Kentuckya Oklahomab Alabamac Texasd Sicilye Germanyf
bitumen 7–9 12.5 4.5 10–20 8–9 4.7–9.5
content, wt%
softening point 28 38 30–36
(ring and
ball), 8C
penetration too soft over 360 15–25 4 too soft soft
of 100 g
at 258C for
5 s, mm/10
type of sand to sand oolitic limestone marly limestone
mineral sandstone limestone limestone
a
Crushed to fine size, less than 6.4 mm. Spread and rolled at temperatures above 15.68C for
pavements.
b
Product blended from various deposits or sand added.
c
Quarried, crushed, sized from 0.95 cm to dust and fluxed with softer asphalt for use as cold-applied
pavement surfacing.
d
Mixed with trap rock and asphalt flux to obtain mix for cold rolling as pavement surfacing.
e
Selected grades ground and used for paving purposes.
f
Enriched by adding asphalt and marketed for use as pavements or mastic.

Seyssel, France; Ragusa in Sicily; Val-de-Travers, Switzerland; and Vorwohle,


Germany.
Asphalt (bitumen) also occurs in various oil sand (also called tar sand)
deposits which occur widely scattered through the world (17) and the bitumen
is available by means of various extraction technologies. A review of the proper-
ties and character of the bitumen (18) suggests that, when used as an asphaltic
binder, the bitumen compares favorably with specification-grade petroleum
asphalts and may have superior aging characteristics and produce more water-
resistant paving mixtures than the typical petroleum asphalts.

2. Manufactured Materials

Since the early 1900s most of the asphalts produced from the refining of petro-
leum have been used primarily in paving and roofing applications. The advent of
motorized transportation led to increased asphalt manufacture from petroleum
in order to provide binders for hard-surfaced pavements.
Petroleum asphalts, compared to native asphalts, are organic with only
trace amounts of inorganic materials. They derive their characteristics from
the nature of their crude origins with some variation possible by choice of
manufacturing process. Although there are a number of refineries or refinery
units whose prime function is to produce asphalt, petroleum asphalt is primar-
ily a product of integrated refineries (Fig. 1). Crudes may be selected for
these refineries for a variety of other product requirements and the asphalt
(or residuum) produced may vary somewhat in characteristics from one
172 ASPHALT Vol. 2

refinery-crude system to another and even by cut-point (Table 2) and asphalt


content (Fig. 2) (5,6). The approximate asphalt yields (%) from various crude
oils are as follows:

Nigerian light 7 L. A. Basin 44


Arabian light 14 California Valley 55
North Slope 21 California Coastal 65
Arabian heavy 31 Boscan 79

Louisiana Bachaquero
Gravity, API 13.1 2.8
Sulfur, wt % 0.9 3.7
Nitrogen, wt % 0.4 0.6
Carbon, wt % 15.8 27.5
Nickel, ppm 20 100
Vanadium, ppm 8 900
Pour point, 8C 55

Fig. 1. Residua are obtained by removal of the volatile constituents of the feedstock at
atmospheric pressure or at reduced pressure and can be converted to asphalt by various
methods.
Table 2. Properties of Various Residua Produced from Tia Juana, Venezuela Light Crude Oil
Property Values
boiling range, 8C whole crude >220 >295 >345 >400 >455 >505
yield on crude, vol % 100.0 70.2 57.4 48.9 39.7 31.2 17.9
gravity, API 31.6 22.5 19.4 17.3 15.1 12.6 7.1
specific gravity 0.8676 0.9188 0.9377 0.9509 0.9652 0.9820 1.0209
sulfur, wt % 1.08 1.42 1.64 1.78 1.93 2.12 2.59
carbon residue 6.8 8.1 9.3 11.2 13.8 21.6
(Conradson), wt %
nitrogen, wt% 0.33 0.39 0.45 0.60
pour point, 8C 20 9 1 7 16 24 49
viscosity

173
kinematic,
mm2 =sð ¼ cStÞ
388C 10.2 83.0 315 890 3100
998C 9.6 19.6 35.0 77.0 220 7959
Furol (SFS), s
508C 70.6 172 528
998C 37.6 106 3760
Universal (SUS), 57.8 96.8 165 359 1025
s at 998C
metals
vanadium, ppm 185 450
nickel, ppm 25 64
iron, ppm 28 48
174 ASPHALT Vol. 2

Fig. 2. Different crude oils contain different amounts of residua and the properties of
specific ‘‘cut point’’ residua may be different, eg, for the 5658C residua:

2.1. Straight Run Asphalt. In crude-oil refining, the crude oil at 340–
4008C is injected into a fractionating column (5,6,19,20). The lighter fractions are
separated as overhead products and the residuum is straight-reduced asphalt.
Crude oil containing about 30% asphalt can be refined completely in an
atmospheric unit to an asphalt product. However, most crude oil cannot be
distilled satisfactorily to an asphalt product at atmospheric pressure because
of the presence of substantial proportions of high boiling gas oil fractions.
Thus, as a supplement to the atmospheric process, a second fractionating
tower (a vacuum tower) is added (Fig. 1).
This two-stage process is particularly applicable to crude oils containing
15–30% asphalt. Vacuum distillation units are continuous flow-through units
and include a heating unit, the vapor–liquid flash separation zone, the fractio-
nating zone, and the vacuum producing equipment. The heated feedstock is
pumped into a flash zone where the more volatile constituents vaporize.
The asphalt is the nonvolatile residual fraction which is removed continuously.
Manufacturing asphalt by straight reduction does not change the chemical
nature of asphalt, other than its viscosity.
2.2. Propane Asphalt. As noted above, crude oils contain different
quantities of residuum (Fig. 2) and, hence, asphalt. Asphalt is also a product of
the propane deasphalting and fractionation process (5,6,21,22) which involves
the precipitation of asphalt from a residuum stock by treatment with propane
under controlled conditions. The petroleum charge stock is usually atmo-
spheric-reduced residue from a primary distillation tower.
Propane is usually used in this process although propane–butane mixtures
and pentane have been used with some variation in process conditions and
Vol. 2 ASPHALT 175

Fig. 3. Schematic representation of propane deasphalting and fractionation process.

hardness of the product. Propane deasphalting is used primarily for crude oils of
relatively low asphalt content, generally <15%. Asphalt produced from this pro-
cess is normally blended with other asphaltic residua for making paving asphalt.
The process (Fig. 3) is a countercurrent liquid–liquid extraction. The feed-
stock is introduced near the top of an extraction tower and the liquid propane
near the bottom, using solvent-to-oil ratios from 4:1 to 10:1. The deasphalted
oil–propane solution is withdrawn overhead and the asphalt from the bottom,
and each is subsequently stripped of propane.
Temperature, solvent ratio, and pressure each have an effect upon the split
point or yield of the oil and asphalt components (Table 3). Contrary to straight
reduction which is a high temperature and low pressure process, propane deas-
phalting is a low temperature and high pressure process.
There are small differences in the properties of asphalts prepared by
propane deasphalting and those prepared by vacuum distillation from the
same stock (Fig. 4). Propane deasphalting also has the ability to reduce a resi-
duum even further and produce an asphalt product having a lower viscosity,
higher ductility, and higher temperature susceptibility than other asphalts
(23). However, such properties might be anticipated to be very much crude
oil dependent. Propane deasphalting is conventionally applied to low asphalt-
content crudes, which are generally different in type and source from those
processed by distillation of higher yield crudes.
2.3. Air-Blown Asphalt. Air-blowing is an exothermic process that is
dependent upon several variable process parameters (24). In the process, an
asphalt (flux) is converted to a harder product by air contact at 200–2758C.
Dehydrogenation and polymerization are involved, and oxygen is not retained
in the asphalt product except in a very minor amount (25). In theory, the oxygen
in the air combines with the hydrogen in the asphalt to evolve water vapor. This
leaves unsaturation for cross-linking. However, the chemistry may be much
176 ASPHALT Vol. 2

Table 3. Typical Yields and Characteristics of Products Obtained from the Propane
Deasphalting Process
Crude source
Process Pa. Okla. Ill. Le Duc, Canada Iraq
reduced crude
gravity, 8APIa 25.1 19.3 13.7 12.1 7.1
kinematic viscosity at 98.98C, 47.2 82.8 378 441 1050
mm2 =sð¼ cStÞ
Saybolt viscosity, s 220 385 1760 2050 4880
crude, vol% 24.0 26.1 11.3 15.2 16.4
deasphalted oil
yield, vol % 83.7 75.6 50.7 47.1 34.1
gravity, 8APIa 26.6 23.6 23.4 22.7 20.2
kinematic viscosity at 98.98C, 30.4 29.8 40.7 42.6 38.2
(mm2 =s ð¼ cStÞ)
Saybolt viscosity, s 144 141 190 199 179
precipitated asphalt
yield, vol% 16.3 24.4 49.3 52.9 65.9
specific gravity at 15.68C 0.953 1.023 1.030 1.054 1.064
softening point 62.2 62.8 80 62.8
(ring and ball), 8C
penetration of 100 g 23 21 0 16
at 258C for 5 s, mm/10
operating conditions
throughput, kL/db 800
charge stock reduced crude
gravity charge, 8APIa 20.0
temperature of deasphalting 65
tower, 8C
solvent-to-oil ratio, by volume 6:1
pressure, kPac 2170 (300)
API gravity at 15:6 C ¼ sp gr at141:5
a
15:6=15:6 C  131:5.
b
To convert kL/d to barrels/d, multiply by 6.25.
c
To convert kPa to psi, multiply by 0.145.

more complex than the simple theory predicts (24). For example, asphaltenes are
produced during the air-blowing process (Table 4) and the evidence suggests
that those constituent species of the asphalt which have higher proportions of
heteroatoms (ie, nitrogen, oxygen, and sulfur) react preferentially with oxygen
leaving less polar substances unreacted (26,27). This also raises the specter of
incompatibility of the resulting products. For example, if the more polar species,
ie, the resins, are preferentially oxidized to produce more asphaltenes, then the
means by which the asphaltenes are dispersed in the oily medium is removed.
Incompatibility should be anticipated.
A variety of other substances can provide the same reaction: sulfur yields
hydrogen sulfide; chlorine yields hydrogen chloride. In some cases, some of the
bonds created are quite weak resulting in, after an induction period, a phenom-
enon termed ‘‘fallback.’’ When fallback occurs, usually at a time when the har-
dened or oxidized asphalt is stored at or near the original processing or
reaction temperature, softening of the asphalt is the result (28–31).
Vol. 2 ASPHALT 177

Fig. 4. Comparison of propane deasphalting with vacuum distillation of asphalt from


Lagunillas, Venezuela crude.

Table 4. Fractional Changes in Hardness of Asphalta upon Air-Blowing


Initial Air-blown Air-blown Air-blown
Properties
softening point 54.4 85 96.1 173.3
(ring and ball), 8C
penetration of 100 g at 36 13 9 1
258C for 5 s, mm/10
Fractions, wt %
asphaltenes 14.8 26.9 31.4 51.3
(hexane insolubles)
hard resins 45.5 36.6 36.1 19.6
soft resins 25.0 22.3 20.9 16.9
oils 12.3 11.9 10.0 11.1
waxes 2.5 2.0 1.8 1.6
Total 100.1 99.7 100.2 100.5
a
Straight-reduced Arkansas asphalt.
178 ASPHALT Vol. 2

Fig. 5. Schematic representation of the batch process for air-blowing.

Air-blown asphalts, more resistant to weather and changes in temperature


than the types mentioned previously are produced by batch and continuous
methods. Air-blown asphalts, of diverse viscosities and flow properties with
added fillers, polymers, solvents, and in water emulsions, provide products for
many applications in the roofing industry.
Batch Process. In the batch process (Fig. 5), the feedstock is preheated in
a tube furnace or heater placed between the feedstock storage and the blowing
vessel. The air supply is provided by a variety of blowers or compressors and a
vertical-tower vessel is preferable for air-blowing. Knockout drums, water scrub-
bers, incinerators, furnaces, and catalytic burning units have been used for fume
disposal (32). Steam is used for safety and to ensure positive fume flow to the
incinerator.
Temperature is the most important variable and preheating is generally
necessary to 200–2308C. After air has been introduced, there is a gradual tem-
perature rise because of the exothermic reaction, until some means is applied to
hold the temperature such as a water or steam spray on the asphalt surface to
maintain a temperature of approximately 2608C. The end point can be predicted
by periodic testing of the softening point.
Continuous Process. Continuous air-blowing (Fig. 6) offers the following
potential advantages: lower equipment and maintenance costs for the same
production capacity; shorter blowing times with air-efficient equipment; less pre-
heat capacity requirements; better control and operation; lower blowing losses;
and lower operating labor requirement. It is particularly preferable when several
principal products, such as coatings, saturants, and built-up roofing asphalts,
can be processed from a single feedstock. When a comparison is made with a
batch operation on the basis of the same daily capacity, the saving in blowing
time is most apparent. Practical use of continuous air-blowing has been made
in the manufacture of paving binders from soft vacuum residua and in the
Vol. 2 ASPHALT 179

Fig. 6. Schematic representation of a continuous air-blowing process.

manufacture of roofing asphalts. Blends of intermediate grades from a harder


air-blown asphalt product such as unfilled coating asphalt (100–1158C softening
point) and a softer air-blown product such as roll saturant (40–458C softening
point) are often used.
Continuous oxidizers are usually operated at a constant temperature
(2608C) and a constant liquid level with the production rate and product charac-
teristics controlled by air rate and charging rate.
Catalytic Asphalt. The term catalytic asphalt has been accepted although
this is not truly a catalytic process. The agents used are consumed in the
reaction. The general effect is the reduction of blowing time as well as a change
of the softening point–penetration relationship (33), which would normally be
obtained from the same feedstock. Reduction of blowing time is an economic
incentive, whereas the change in product properties permits manufacture to
specifications not possible otherwise and the use of a wider variety of crude
sources for flux.
Many agents have been proposed and patented including copper sulfate
(34), zinc chloride (35), ferric chloride (36), aluminum chloride (36), and phos-
phorus pentoxide (37); ferric chloride, zinc chloride, and phosphorus pentoxide
have been most widely used. The addition of these agents may vary from
0.1 to 3%, depending upon the feedstock and the desired characteristics of the
product (Table 5) and all asphalt feedstocks do not respond to catalysts in the
same way. Differences in feedstock composition are important qualifiers in deter-
mining the properties of the asphalt product. The important softening point-
penetration relationship, which describes the temperature susceptibility of
an asphalt, also varies with the source of the feedstock. Straight-reduced,
air-blown, and air-blown catalytic asphalts from the same crude feedstock also
vary considerably.
2.4. Thermal Asphalt. Thermal or cracked asphalts differ from other
asphalts in that they are products of a cracking process. They have relatively
180 ASPHALT Vol. 2

Table 5. Air-Blowing of Fluxes With and Without Ferric Chloride and Phosphorus
Pentoxide Catalysts
California stock Arkansas stock
No 0.8% 0.5% No 2.2%
Property catalyst FeCl3 6 H2 O P2O5 catalyst FeCl3 6 H2 O 1.0% P2O5
blowing 221.1 221.1 221.1 254.4 237.8 237.8
temperature, 8C
blowing time, min 135 44 103 105 46 128
softening pointa , 8C 87.8 86.1 86.1 86.7 83.3 87.8
penetration of 100 g 27 51 53 27 62 59
at 258C for 5 s,
mm/10
penetration of 200 g 13 30 28 15 38 38
at 08C for 60 s,
mm/10
penetration of 50 g at 74 48 109 110
468C for 5 s, mm/10
ductility at 258C, cm 3.5 3.4 4.8 3.3 3.9 5.3
a
Ring and ball method.

high specific gravity, low viscosity, and high temperature susceptibility, and they
contain cokelike bodies (carbenes) as indicated by the Oliensis test (38). Thermal
asphalts have significant application as saturants for cellulosic building products
such as insulation boards, brick-finish siding, and fiber soil pipes.
The thermal cracking process (Fig. 7) involves preheating of the charge
to 480–6008C and then discharging into a reaction vessel or chamber under

Fig. 7. Schematic representation of thermal cracking process.


Vol. 2 ASPHALT 181

Table 6. Three Types of Liquid Asphalt Made by Cutting Back with Petroleum Diluents
Type asphalt Diluent type Diluent, % Viscosity rangea , mm2 =s ð¼ cStÞ
slow curing (SC) gas oil 0–50 70–6000
medium curing (MC) kerosene 15–45 30–6000
rapid curing (RC) naphtha 15–45 70–6000
a
The liquid cutback asphalts are prepared in a number of viscosity grades, ranging generally from 70
to 6000. The grade number indicates the viscosity at 608C.

pressures up to 1480 kPa (200 psig). Cracking decomposition takes place to form
both lighter and heavier products, including a certain amount of fixed carbon.
Distillation is used to separate the overall products into gas, gasoline, middle
distillate, and an asphaltic residuum.
2.5. Blended Asphalt. Any particular, refinery may stock two grades of
asphalt, one at each end of the viscosity spectrum of the entire product grade
requirements. Intermediate grades are prepared by blending (proportioning)
the extremes (39). The preparation of asphalts in liquid form by blending (cutting
back) an asphalt with a petroleum distillate fraction is customary. There are
three general types of cutback asphalt which differ mainly in the diluent used
(Table 6). The slow-curing type is often called road oil, Road oil which is made
by direct reduction, but most are commonly blended. The medium curing and
rapid curing products usually have four grades within a given type and differ
mainly in the amount of diluent used and in the kinematic viscosity of the
blended product.
Preparation is accomplished by simple blending of the diluent into the hot
base asphalt. This is generally accomplished in tanks equipped with coils for air
agitation or with a mechanical stirrer or a vortex mixer. Line blending in a batch
circulation system or in a continuous fashion (40) is used where the volume
produced justifies the extra facilities. A continuous, line-blending system is
applicable to the manufacture of cutback asphalts and asphalt cements (Fig. 8).
2.6. Asphalt Emulsions. An emulsion is one immiscible liquid dis-
persed in the other in the form of very fine droplets from about 1–25 mm and
an average of 5 mm diameter. In the most common asphalt emulsion, the oil-in-
water type, the asphalt is the dispersed (internal) phase, and water is the contin-
uous (external) phase. Most of the important properties of an emulsion are
dependent upon the amount and type of emulsifying agent used. The more pop-
ular cationic emulsions (41,42) are made with cation-active agents which provide
emulsion droplets with positive charges. The anionic agent is usually the sodium
or potassium salt of a fatty acid (43). Nonionic cellulose derivatives are also used
to increase the viscosity of the emulsion if needed. The acid number of asphalt is
an indicator of its emulsifiability and reflects the presence of high molecular
weight asphaltogenic or naphthenic acids. Diamines, frequently used as cationic
agents, are made from the reaction of tallow acid amines with acrylonitrile,
followed by hydrogenation.
Emulsion processes (Fig. 9) vary and the emulsions are made in rapid-,
medium-, and slow-setting types for diverse application techniques in the road-
building industry. Emulsions are most commonly formulated with rapid-setting
properties and are designated as a spraying type, ie, for road seal-coating
182 ASPHALT Vol. 2

Fig. 8. Schematic representation of a continuous asphalt blender.

operations. Slow-setting emulsions are usually used for soil stabilization or in


mulching applications.
Industrial emulsions have applications outside the road-building industry.
They are made with harder grades of asphalt and contain clays, casein, gelatin,
or blood albumin as peptizing agents. Certain clays, such as bentonite, are good
emulsion dispersants and impart a buttery consistency to the emulsion. These
emulsions have found a wide variety of applications, such as in surface coating
of asphalt pavements, for built-up roofs, and for other weatherproof coverings.

Fig. 9. Flow diagram of an asphalt emulsion plant.


Vol. 2 ASPHALT 183

3. Composition and Properties

Determination of the components of asphalts has always presented a challenge


because of the complexity and high molecular weights of the molecular constitu-
ents (6,44,45). The principle behind composition studies is to evaluate asphalts in
terms of composition and performance. Such studies are always suspect when
this is not the overriding goal of the work.
The influence of the composition of asphalt has been recognized, for many
years, as being an important factor in controlling the performance of such mate-
rials. Furthermore, rheological properties have always been associated with com-
position but, in order to utilize compositional data effectively, more definitive
correlations between composition and properties are needed (46–48).
The methods employed can be conveniently arranged into a number of
categories: (1) fractionation by precipitation; (2) fractionation by distillation;
(3) separation by chromatographic techniques; (4) chemical analysis using spec-
trophotometric techniques (infrared, ultraviolet, nuclear magnetic resonance,
x-ray fluorescence, emission, neutron activation), titrimetric and gravimetric
techniques, elemental analysis; (5) molecular weight analysis by mass spectro-
metry, vapor pressure osmometry, and size-exclusion chromatography.
However, for the past 30 years fractional separation has been the basis for
most asphalt composition analysis (Fig. 10). The separation methods that have
been used divide asphalt into operationally defined fractions. Four types of
asphalt separation procedures are now in use: (1) chemical precipitation in

Fig. 10. The relationship between a crude oil, two residua from the crude oil, and the
propane asphalt.
184 ASPHALT Vol. 2

which n-pentane separation of asphaltenes is followed by chemical precipitation


of other fractions with sulfuric acid of increasing concentration (ASTM D2006);
(2) solvent fractionation: separation of an ‘‘asphaltene’’ fraction by the use of
1-butanol followed by dissolution of the 1-butanol solubles in acetone. The
acetone solution is chilled, forcing the precipitation of paraffinics from the cold
acetone-soluble cyclics; (3) adsorption chromatography using a clay–gel proce-
dure where, after removal of the asphaltenes, the remaining constituents are
separated by selective adsorption–desorption on an adsorbent (ASTM D2007
and D4124); (4) size-exclusion chromatography in which gel-permeation chroma-
tographic (gpc) separation of asphalt constituents occurs based on their asso-
ciated sizes in dilute solutions (ASTM D3593).
The fractions obtained in these schemes are defined operationally or proce-
durally. The amount and type of asphaltenes in an asphalt are, for instance,
defined by the solvent used for precipitating them. Fractional separation of
asphalt does not provide well-defined chemical components. The materials sepa-
rated should only be defined in terms of the particular test procedure.
The component of highest carbon content is the fraction termed carboids-
Carboids and consists of species that are insoluble in carbon disulfide or in pyr-
idine (5,6,49). The fraction that has been called carbenes Carbenes contains
molecular species that are soluble in carbon disulfide and soluble in pyridine
but which are insoluble in carbon tetrachloride and in benzene (Fig. 11) (5,6,49).
Asphaltenes seem to be relatively constant in composition in residual
asphalts, despite the source, as determined by elemental analysis (6). Determi-
nation of asphaltenes is relatively standard, and the fractions are termed
n-pentane, n-hexane, n-heptane, or naphtha-insoluble, depending upon the
precipitant used (5,6,49). After the asphaltenes are removed, resinous fractions
are removed from the maltenes-petrolenes usually by adsorption on activated
gels or clays. Recovery of the resin fraction by desorbtion is usually nearly
quantitative.

Fig. 11. Simplified representation of the separation of petroleum into six principal
fractions.
Vol. 2 ASPHALT 185

The forerunner of the modern methods of asphalt fractionation was first


described in 1916 (50) and the procedure was later modified by use of fuller’s
earth (attapulgite (1337-76-4]) to remove the resinous components (51). Further
modifications and preferences led to the development of a variety of fractionation
methods (52–58). Thus, because of the nature and varieties of fractions possible
and the large number of precipitants or adsorbents, a great number of methods
can be devised to determine the composition of asphalts (5,6,44,45). Fractions
have also been separated by thermal diffusion (59), by dialysis (60), by electroly-
tic methods (61), and by repeated solvent fractionations (62,63).
The composition data should always be applied to in-service performance
(64,65) in order to properly evaluate the behavior of the asphaltic binder under
true working conditions.
3.1. Colloidal State. The principal outcome of many of the composition
studies has been the delineation of the asphalt system as a colloidal system at
ambient or normal service conditions. This particular concept was proposed in
1924 and described the system as an oil medium in which the asphaltene fraction
was dispersed. The transition from a colloid to a Newtonian liquid is dependent
on temperature, hardness, shear rate, chemical nature, etc. At normal service
temperatures asphalt is viscoelastic, and viscous at higher temperatures. The
disperse phase is a micelle composed of the molecular species that make up
the asphaltenes and the higher molecular weight aromatic components of the
petrolenes or the maltenes (ie, the nonasphaltene components). Complete pepti-
zation of the micelle seems probable if the system contains sufficient aromatic
constituents, in relation to the concentration of asphaltenes, to allow the asphal-
tenes to remain in the dispersed phase.
Many attempts have been made to characterize the stability of the colloidal
state of asphalt at ordinary temperature on the basis of chemical analysis in
generic groups. For example, a colloidal instability index has been defined as
the ratio of the sum of the amounts in asphaltenes and flocculants (saturated
oils) to the sum of the amounts in peptizers (resins) and solvents (aromatic
oils) (66):

ðasphaltenes þ saturated oilsÞ


Ic ¼
ðresins þ aromatic oilsÞ

As this ratio increases so does the gel character of the asphalt cement, and
its colloidal stability increases. The properties of the asphalt (softening point,
ductility, embrittlement temperature, thermal susceptibility, elastic recovery,
shearing susceptibility, etc) vary significantly with the colloidal instability
index and hence with composition.
Many investigations of relationships between composition and properties
take into account only the concentration of the asphaltenes, independently of
any quality criterion. However, a distinction should be made between the asphal-
tenes which occur in straight run asphalts and those which occur in blown
asphalts. Remembering that asphaltenes are a solubility class rather than a
distinct chemical class means that vast differences occur in the make-up of
this fraction when it is produced by different processes.
186 ASPHALT Vol. 2

3.2. Elemental Analysis. Asphalt is not composed of a single chemical


species, but is rather a complex mixture of organic molecules that vary widely in
composition from nonpolar, saturated hydrocarbons to highly polar, highly
condensed aromatic ring systems. Although asphalt molecules are composed
predominantly of carbon and hydrogen, most molecules contain one or more of
the heteroatoms nitrogen, sulfur, and oxygen, together with trace amounts of
metals, principally vanadium and nickel. The heteroatoms, although a minor
component compared to the hydrocarbon moiety, can vary in concentration
over a wide range depending on the source of the asphalt. Because the heteroa-
toms often impart functionality and polarity to the molecules, their presence may
make a disproportionately large contribution to the differences in physical prop-
erties among asphalts from different sources.
Generally, most asphalts are 79–88 wt % C, 7–13 wt% H, trace-8 wt% S,
2–8 wt% O, and trace-3 wt% N (Table 7). Trace metals such as iron, nickel,
vanadium, calcium, titanium, magnesium, sodium, cobalt, copper, tin, and
zinc, occur in crude oils. Vanadium and nickel are bound in organic complexes
and, by virtue of the concentration (distillation) process by which asphalt is
manufactured, are also found in asphalt.
Asphalt typically contains oxygen (<2%), nitrogen (<2%), and sulfur
(<7%), as can be measured by many methods of elemental analysis. The het-
eroatom content tends to increase with increasing boiling point within an
asphalt residue. The heteroatom content of a typical asphalt is significant,
especially when considered on the molecular level. For example, if the mean
molecular mass of the asphaltene fraction is 1000, then one heteroatom per
molecule in the asphaltene would require the following proportions of heteroa-
toms: sulfur 3.2%, nitrogen 1.4%, and oxygen 1.6%. Such concentrations of het-
eroatoms are commonly found in whole asphalts and are typically exceeded in
the asphaltene fractions. The heteroatom content seems to vary more than
the hydrogen:carbon (H:C) ratio of the asphaltene fraction from different
crudes (5,6).
Techniques are available for quantitative identification of the six principal
types of heteroatom compounds in asphalt: carboxylic acids, 2-quinolones,
phenols, pyrroles, amides, and pyridines (67).
Many investigators have also measured the trace metal content of asphalts
(68). The catalytic behavior of vanadium has prompted studies of the relation
between vanadium content and an asphalt’s sensitivity to oxidation (viscosity

Table 7. Elemental Analyses of Representative Petroleum Asphalts


B-2959 B-3036 B-3051 B-3602
Code source Mexican Blend Arkansas–Louisiana Boscan California
carbon, % 83.77 85.78 82.90 86.77
hydrogen, % 9.91 10.19 10.45 10.94
nitrogen, % 0.28 0.26 0.78 1.10
sulfur, % 5.25 3.41 5.43 0.99
oxygen, % 0.77 0.36 0.29 0.20
vandium, ppm 180 7 1380 4
nickel,ppm 22 0.4 109 6
Vol. 2 ASPHALT 187

ratio). The significance of metals in the behavior of asphalts is not yet well under-
stood or defined.
3.3. Molecular Weight. The molecular weights of the individual frac-
tions of asphalt have received more attention, and have been considered to be
of greater importance, than the molecular weight of the asphalt itself. Asphal-
tenes display a wide range of molecular weights, from 500 to at least 2500,
depending upon the method of measurement. Asphaltenes associate in dilute
solution in nonpolar solvents (69) giving higher molecular weights than is actu-
ally the case on an individual molecule basis. The molecular weights of the resins
are somewhat lower than those of the asphaltenes and usually fall within the
range 500 to 1000 (70). This is not only because of the absence of association
but also because of a lower absolute molecular size. The molecular weights of
the oil fractions (ie, the asphalt minus the asphaltenes and minus the resins)
is usually less than 500, often 300 to 400.
3.4. Acid Number. Asphalt contains a small amount of organic acids
and saponifiable material and the acid values of asphalt, 0.1–2.8 mg potassium
hydroxide per gram of asphalt, are related to emulsification conditions and ease
of dispersion. The acid content is largely determined by the percentage of
naphthenic (cycloparaffinic) acids of higher molecular weight that are originally
present in the crude oil. With increased hardness, asphalt from a particular
crude oil normally decreases in acid number as more of the naphthenic acids
are removed during the distillation process.
3.5. Rheology. Asphalt is a viscoelastic material whose rheological
properties reflect crude type and, to a lesser extent, processing. The ability of
the asphalt to perform under many conditions depends on flow behavior
(71,72). Asphalt films or coatings showing no appreciable change from original
conditions are usually desired, ie, they should allow some structural movement
without permanent deformation. Some procedures for measuring viscoelastic
properties of asphalt have been described (2,73–75). Liquid asphalts, in effect,
substitute diluent (or water emulsification) for heat to allow pumping, spraying,
or application of the material at lower temperature.
The viscosity of hydrocarbons and temperature are related by the Walther
equation (76):

log logð100Z0 Þ ¼ a  m log T

where Z0 is the limiting viscosity at low shear rates, T is the absolute tempera-
ture, a and m are constants reflecting the intercept level and slope (or measure of
susceptibility of viscosity to temperature, respectively). On the other, hand, the
general relationship for viscosity, temperature, and shear rate is


log log 100Z0 1  CDN ¼ a  m log T

where D is shear rate; C is a function of limiting viscosity, limiting slope, and


other constants; and N is the limiting slope.
Typical profiles for the different general families of asphalts (cutbacks or
liquid materials, paving asphalt cements and the harder roofing and industrial
188 ASPHALT Vol. 2

Fig. 12. A viscosity-temperature chart, mm2 =s ¼ cSt; – – –, industrial asphalts; — —


cutback asphalts;  paving asphalts.

materials which are usually graded by softening point) can be generated by the
use of the equation (Fig. 12). At lower temperatures (608C and lower) and/or
higher shear rates, which are typical of asphalt service conditions after incor-
poration in a roof or pavement, semisolid and solid asphalts display an increas-
ing elastic component which relates viscosity with shear rate (73,77,78). The
constant high viscosity at lower shear rates is the limiting viscosity. Viscosities
in the area where viscosity changes with shear rate are generally termed appar-
ent viscosities.
Curves for the viscosity data, when displayed as a function of shear rate
with temperature, show the same general shape with limiting viscosities at
low shear rates and limiting slopes at high shear rates. These curves can be com-
bined in a single master curve (for each asphalt) employing vertical and horizon-
tal shift factors (77–79). Such data relate reduced viscosity (from the vertical
shift) and reduced shear rate (from the horizontal shift).
A number of viscometers have been developed for securing viscosity data at
temperatures as low as 08C (58,59). The most popular instruments in current use
are the cone plate (ASTM D3205), parallel plate, and capillary instruments
(ASTM D2171 and ASTM D2170). The cone plate can be used for the determina-
tion of viscosities in the range of 10 to over 109 Pa  s (1010 P) at temperatures
of 0–708C and at shear rates from 103 to 102 s1 . Capillary viscometers are com-
monly used for the determination of viscosities at 60–1358C.
Tests developed for measurement of viscoelastic properties are directly
usable in engineering relations. Properties can be related to the inherent struc-
ture of bituminous materials. The fraction of highest molecular weight, the
asphaltenes, is dispersed within the asphalt and is dependent upon the content
and nature of the resin and oil fractions. Higher aromaticity of the oil fractions
or higher temperatures leads to viscous (sol) conditions. A more elastic (gel)
Vol. 2 ASPHALT 189

condition results from a more paraffinic nature and is indicated by large elastic
moduli or, empirically, by a relatively high penetration at a given softening
point. Empirically, the penetration index PI(2) and penetration temperature sus-
ceptibility PTS have been used to measure the degree of dispersion.

30
PI ¼  10
1 þ 90  PTS

log 800  log penetration at 25 C; 100 g; 5 s


PTS ¼
ðsoftening point; ring and ball; CÞ  25 C

This equation is based on the approximation that the penetration is 800 at the
softening point, but the approximation fails appreciably when a complex flow is
present (80,81). However, the penetration index has been, and continues to be,
used for the general characteristics of asphalt; for example asphalts with a PI
less than 2 are considered to be the pitch type, from 2 to þ2, the sol type,
and above þ2, the gel or blown type (2). Other empirical relations that have
been used to express the rheological-temperature relation are fluidity factor; a
Furol viscosity V, at 1358C and penetration P, at 258C, relation of (V–P)P/100;
and penetration viscosity number PVN again relating the penetration at 258C
and kinematic viscosity at 1358C (82,83).
Asphalts develop an internal structure with age, steric hardening (3), in
which viscosity can increase upon aging without any loss of volatile material
(73, 83). Those with a particularly high degree of gel structure exhibit thixotropy.
3.6. Durability. The term ‘‘durable’’ has several meanings, but in the
present context it is used to describe an asphalt that possesses the necessary che-
mical and physical properties required for the specified pavement performance,
being resistant to change during the in-service conditions that are prevalent dur-
ing the life of the pavement.
Asphalts are used as protective films, adhesives, and binders because of
their waterproof and weather-resistant properties. One particularly valuable
property, the ability of the asphalt to undergo movement without fracture, can
occur because of the viscous (sol) nature. In addition, asphalts have long and con-
tinuous satisfactory service because of their slow rate of hardening from heat,
oxidation, fatigue, and weathering (84–87). Exposed asphalt films harden par-
tially from a loss of volatile oils and to a greater extent from the formation of
additional asphaltene material, with the concurrent loss of petrolene-maltene
constituents through oxidation. Such chemical change undoubtedly is catalyzed
by uv irradiation (86–90). The oxidation may form water-soluble degradation
products which are removed from the asphalt film (89,90), but the conversion
of the petrolene–maltene constituents to lower molecular weight asphaltene
material (27) cannot be ignored.
Failure of an asphalt film is reflected in the appearance of typical crack
patterns because of a decrease in the plasticity of the asphalt through insuffi-
cient lower molecular weight constituents remaining in the continuous phase.
Other manifestations of failure are loosening of the bond, such as loosening of
190 ASPHALT Vol. 2

granules in roofing shingles to actual loss of the film, or the binder in pavement
mixtures might become friable and crack. A stiffer asphalt, under uniform
loading conditions, could reduce pavement deflection, extend fatigue life, and
allow less flow deformation. A softer material would normally allow a longer
weathering life before the maltene–asphaltene composition becomes critical
in service. Usually the softest material allowed by initial service needs is
selected.
The water resistance of asphalt films is also a manifestation of durability.
Asphalts that have a low content of soluble salts show a low water absorption.
The pickup of water is primarily a surface phenomenon that causes the film to
soften leading to blistering. Even with a high rate of absorption, asphalt films
show little loss of bond to surfaces on continued immersion in water, and con-
tinue to protect metals from corrosion for long periods of time. Bacteria and
fungi can attack the very low molecular weight portion of bituminous materials.
However, deterioration of the harder bituminous films as a result of such agents
seems insignificant (91,92).
Mineral fillers are often added to asphalts to influence their flow properties,
reduce costs, and are commonly used as stabilizers in roofing coatings at concen-
trations up to 60 wt% (93). Mineral-filled films show improved resistance to
flow at elevated temperatures, improved impact resistance, and better flame-
spread resistance. Fillers may increase the water absorption of asphalts. Fillers
commonly used are ground limestone, slate flours, finely divided silicas, trap
rocks, and mica; they are often produced as by-products in rock-crushing opera-
tions. Opaque fillers offer protection from weathering. Asbestos (qv) filler has
special properties because of its fibrous structure, high resistance to flow, and
toughness. It has been used in asphalt paving mixes to increase the resistance
to movement under traffic (94) and in roofing materials for fire-retardant
purposes.

4. Specifications and Tests

In 1903 an American Society for Testing and Materials (ASTM) Committee on


Road and Paving Materials was formed to develop test methods and specifica-
tions for highway materials. Test methods for volatilization, penetration, and
bitumen were developed by the Office of Public Roads and were adopted by
ASTM in 1911.
In the volatilization test, a 20-g sample was heated for 5 h at 1638C in a tin
box 6 cm in diameter and 2 cm deep. The loss in weight was determined and the
consistency of the residue was an optional requirement of that time. To provide a
greater depth of sample the method was revised to use a 50-g sample and a tin
box 5.5 cm in diameter and 3.5 cm deep.
For the asphalt cements produced at that time the adoption of the volati-
lization and penetration tests provided some degree of control of excessive
changes during plant mixing that might be reflected in more durable asphalts.
The adoption of the method for bitumen was intended to provide a means for
identifying Trinidad asphalt by observing the amount and color of the insoluble
ash.
Vol. 2 ASPHALT 191

Fig. 13. Relation of physical properties to the yield of residuum from Tia Juana,
Venezuela light crude oil.

During the 1920s the following three national specifications for asphalt
cements were published: (1) Federal specifications, adopted in 1925; (2) the
American Association of State Highway Transportation Officials (AASHTO)
specifications, adopted in 1924, revised in 1926; and (3) ASTM specifications,
adopted in 1922 to 1926, withdrawn in 1939, and re-issued in 1947.
The Federal specifications stipulated that only those asphalts that had been
demonstrated by service tests as satisfactory for the intended use would be
accepted. The specifications also indicated the type and location of construction
and the relative amount of traffic for each of the grades. The AASHTO specifica-
tions indicated that the use of each grade depended on the type of road, climate,
and traffic. The ASTM suggested the type of construction for which each grade
would be used.
With minor exceptions the requirements for the physical and chemical
properties of asphalt were essentially the same for the three national specifica-
tions and included: penetration and ductility at 258C; flash point; % loss at
1638C; penetration of residue as a % of original; solubility in carbon disulfide;
solubility in carbon tetrachloride; specific gravity at 258C; and softening point.
Remembering that the properties of residua vary with cut-point (Table 2),
ie, the vol% of the crude oil (Fig. 13) helps the refiner produce asphalt of a
192 ASPHALT Vol. 2

specific type or property. There are several properties that are usually controlled
in asphalt specifications:

Property Test
safety flash point
purity solubility, ash, water content
composition naphtha insolubles, various separation techniques, distillation
tests, homogeneity, wax content
rheology penetration, viscosity, softening point, float, ductility, flow,
impact or shock, break point, temperature susceptibility ratios
durability thin-film oven, aging index, weathering
density or specific specific gravity
gravity
special properties bond or adhesion, compatibility, stain, storage stability,
chemical resistance, etc

Asphalts are usually specified in several grades for the same industry, dif-
fering in hardness or viscosity (95). However, with the changing nature and com-
position of crude oil feedstocks over the past two decades, performance and
supply factors are also an important consideration (95).
Specifications for paving asphalt cements usually include five grades differ-
ing in either viscosity or penetration level at 608C (Table 8) (ASTM D496).
Susceptibility of viscosity to temperature is usually controlled in asphalt cement

Table 8. ASTM Requirements for Asphalt Cement Viscosity


Viscosity gradeb
Tests on residue from rolling
thin-film oven testa AR-1000 AR-2000 AR-4000 AR-8000 AR-16000
c
viscosity, 608C, Pas 100 25 200 50 400 100 800 200 1600 400
viscosity, 1358C, min, 140 200 275 400 550
mm2 =s ð¼ cStÞ
penetration of 100 g 65 40 25 20 20
at 258C for 5 s, min,
mm/10
original penetration 40 45 50 52
at 258C, min %
ductility, 258C, 5 cm/min, 100d 100 75 75 75
min, cm
tests on original asphalt: 205 219 227 232 238
flash point,
Pensky-Martens closed
tester, min, 8C
solubility in trichloroethylene, 99.0 99.0 99.0 99.0 99.0
min %
a
Thin-film oven test may be used but the rolling thin-film oven test shall be the referee method.
b
Grading based on residue from rolling thin-film oven test.
c
To convert Pas to P, multiply by 10.
d
If ductility is less than 100, material will be accepted if ductility at 15.58C is 100 minimum at a rate
of 5 cm/min.
Vol. 2 ASPHALT 193

by viscosity limits at a higher temperature such as 1358C and a penetration or


viscosity limit at a lower temperature such as 258C.
Paving cutbacks are also graded at 608C by viscosity, with usually four to
five grades of each type. For asphalt cements, the newer viscosity grade designa-
tion is the midpoint of the viscosity range. The cutback’s grade designation is the
minimum kinematic viscosity for the grade, with a maximum grade viscosity of
twice the minimum.
Roofing and industrial asphalts are also generally specified in various hard-
nesses usually with a combination of softening point and penetration to distin-
guish grades. Asphalts for roof construction are specified in ASTM D312
grades and asphalts for damp-proofing and waterproofing in ASTM D449.
Temperature susceptibility is usually controlled in these requirements by
specifying penetration limits or ranges at 258C and other temperatures as well
as softening point ranges at higher temperatures. The ASTM D312 asphalts for
built-up roof constructions are differentiated according to application depending
primarily on pitch of the roof and to some extent on whether or not mineral
surfacing aggregates are specified. The damp-proofing grades reflect above or
below grade construction, primarily, and whether or not a self-healing property
is incorporated.
4.1. Test Procedures. Most tests applied to petroleum asphalts are
empirical in nature (45). The ASTM tests are not the only ones applied to
asphalt testing. Private tests of a propriety nature have been used within com-
panies. Some of these tests are now being made public and occur from time
to time in literature reports. Such tests may become a part of the ASTM
standards.
The significance of a particular test is not always apparent by reading
the procedure, and sometimes can only be gained through working familiarity
with the test (96). The following tests are commonly used to characterize
asphalts.
Bitumen Insoluble in Paraffin Naphtha (AASHTO T46). This test desig-
nated by the American Association of State Highway and Transportation
Officials (AASHTO) is used to indicate the content of naphtha-insoluble asphal-
tenes in an asphalt. Other solvents such as n-heptane (ASTM D3279), n-hexane,
and n-pentane have been substituted for the naphtha solvent.
Bitumen Soluble in Carbon Disulfide (ASTM D4). Asphalt is defined as a
mixture of hydrocarbons that are completely soluble in carbon disulfide. Trichlor-
oethylene or 1,1,1-trichloroethane have been used in recent years as safer sol-
vents for this purpose. The procedure for these and other solvents for asphalt
with little or no mineral matter are described in ASTM D2042.
Breaking Point (FRAAS, IP 80/53). This test of the Institute of Petro-
leum is an approximate indication of the temperature at which an asphalt
possesses no ductility and would reflect brittle fracture conditions.
Bond and Adhesion (ASTM D1191). This test, designed for use on crack
and joint sealers, is used primarily to determine whether a jointing material
possesses an arbitrary amount of bonding strength at low temperatures where
portland cement concrete is being used.
Compatibility (ASTM D1370). When a coating asphalt and a saturating-
grade asphalt are used together, as in prepared roofing, this test indicates
194 ASPHALT Vol. 2

whether they are likely to bleed, show strike-through, or disbond under stress at
the coating felt interface.
Distillation (ASTM D402). Approximate amounts of volatile constituents
are determined by this test which is particularly applicable to cutback asphalt
and road oils.
Ductility (ASTM D113). The ductility of an asphalt is expressed as the
distance in cm which a standard briquet can be elongated before breaking.
Ductility is a combination of flow properties and reflects both cohesion and shear
susceptibility.
Emulsified Asphalts (ASTM D244). This standard covers a variety of
tests for the composition, handling, nature and classification, storage, use, and
specifying of asphalt emulsions used primarily for paving purposes.
Flash Point (ASTM D92). The Cleveland open cup method is most com-
monly used although the Tag open cup (ASTM D3143) is applicable to cutbacks.
Flash point is an indication of fire hazard and the test is frequently used to
indicate whether a given product has been contaminated with materials of
lower flash point.
Float Test (ASTM D139). The consistency properties of an asphalt at a
very low applied force are indicated by this test. The test is normally used for
those asphalts that are too soft for the penetration test.
Homogeneity. Visual tests for homogeneity are generally used to judge
uniformity.
Penetration (ASTM D5). This is a commonly used consistency test. It
involves the determination of the extent to which a standard needle penetrates
a properly prepared sample of asphalt under definitely specified conditions of
temperature, load, and time. The distance that the needle penetrates in units
of mm/10 measured from 0 to 300, is the penetration value. Soft asphalts have
high penetration values.
Sampling (ASTM D140). This standard provides guidance for the sam-
pling of asphalts, liquid and semisolid, at point of manufacture, storage, or delivery.
Softening Point (Ring and Ball Method, ASTM D36). The softening
point of an asphalt may be defined as that temperature at which an asphalt
attains a particular degree of softness under specified conditions of test. It is com-
monly used to classify industrial and roofing asphalt grades. A method (ASTM
D2398) has been developed to provide a single bath medium for the full range
covered by the two media in ASTM D36.
Specific Gravity (ASTM D70). For solid and semisolid asphalts a
pycnometer is generally used and a hydrometer is applicable to liquid asphalts
(ASTM D3142).
Spot Test (AASHTO T102). The test distinguishes asphalts that contain
bodies poorly tolerated by the asphalt system.
Stain (ASTM D1328). This test measures the amount of stain on paper
or other cellulosic materials by asphalt. Variations of the cigarette paper stain
procedure include the Barber stain, usually conducted at 54.48C and 3.9 MN
(400 g-force). Talc stain tests are also used.
Temperature Susceptibility. This is the term used to designate the
change in consistency of an asphalt with changes in temperature. The Walther
equation slope constant m can be used as a fundamental measure.
Vol. 2 ASPHALT 195

Temperature–Volume Correction (ASTM D1250). Tables are provided


to allow the conversion of volumes of asphaltic materials from one temperature
to another or, as generally used, to adjust volumes to a temperature of 15.68C,
the standard basis of measurement in the United States. The value
commonly taken for mean coefficient of expansion is 0.00036 in the range
15.6–121.18C.
Viscosity. This is a measure of resistance to flow. A number of instruments
are in common use with asphalt for this purpose. The vacuum capillary (ASTM
D2171) is usually used to classify paving asphalt cements at 608C, although
it is applicable to materials in the range 4:2  20; 000 Pas (42–200,000 P).
Kinematic capillary instruments (ASTM D2170) are used in the 60–1358C
temperature range for both liquid and semisolid asphalts in the range of
30  100; 000 mm2 =s ð¼ cStÞ. Saybolt tests (D88) are also used in this tempera-
ture range and at higher temperature (ASTM E102). At lower temperatures the
cone and plate instrument (ASTM D3205) has been used extensively in the
viscosity range 102  109 Pas (103 –109 P). Other techniques include use of
the sliding plate microviscometer and the rheogoniometer (see RHEOLOGY).
Thin-Film Oven Test (ASTM D1754). This test has the purpose of deter-
mining the hardening effect of heat and air on a static film of asphalt when
exposed in a thin film. An analogous procedure is the Rolling Thin-Film Test
(ASTM D2872) which has the same purpose but utilizes a moving film exposed
for 75 min at 1638C.
Water Content (ASTM D95). This test covers the water content of
asphalt by distillation using a Dean-Stark receiver.
Wax Content. The Deutsche Industrie Normen (DIN) method utilizes
destructive distillation of the asphalt, followed by freezing out of the wax in
the distillate fractions.
Weathering (ASTM D529). This test evaluates the relative weather
resistance of asphalts used for protective-coating applications, especially for roof-
ing. No direct measure of outdoor life or service can be obtained from this test.
Methods for preparing test panels (ASTM D1669) and failure end-point testing
(D1670) are available.
Finally, these are not the only tests used for determining the property and
behavior of an asphaltic binder. As in the petroleum industry (5,6), a variety of
tests are employed that evolved through local, or company, usage (97–99) with-
out having passed through the ASTM system of approval. Tests are also neces-
sary to determine the most practical design of asphalt-aggregate mixtures (100).

5. Uses

Properties, and therefore uses, depend upon the method of manufacture


(Table 9). The Asphalt Institute lists a multitude of uses for asphalt (101), includ-
ing hydraulics (dam facings, canal linings, pond linings) (102), recreation (sub-
strate for artificial surfaces, tennis courts, running tracks) (103), agriculture
(mulches, underground water barriers, stockyard paving) (104–107), transporta-
tion (railroad ballast treatment and roadbeds) (108), and metals (ore leaching
pads, ore and coal briquetting, etc).
196 ASPHALT Vol. 2

Table 9. Properties of Asphalts


Straight-run, Air-blown
Property residual Thermal residual
softening point (ring and ball), 8C 46 113 93
penetration of 100 g at 258C 90 0 20
for 5 s, mm/10
ductility at 258C, 5 cm/min, cma 150þ too hard 3.2
specific gravity, 15.6/15.68C 1.03 1.12 1.05
mean coefficient of cubical
expansion/8C
15.6–65.68C 0.00063 0.00058 0.00063
15.6–2328C 0.00068 0.00063 0.00068
specific heat, J=ðkgKÞb
4.48C 1675 1549 1633
93.38C 1968 1842 1926
204.48C 2345 2177 2303
thermal conductivity 0.16 0.16 0.16
at 26.78C, W=ðmKÞ
permeability constant
at 258C, kgm=ðm2 sN=m2 Þ
water vapor 0:62  1:93  1015 1:1  1015 1:25  2:4  1015
oxygen 0:08  1015
water absorption of 10-milc films
on aluminum panels, wt%
50 wks 1.5–10
100 wks 2.5–16.5
surface tension, mN=m ð ¼ dyn=cmÞ
258C 34 32
1008C 27 28
dielectric strength, spherical 11  45  106 36  106 30  35  106
electrodes, V/m
dielectric constant, 50 Hz at 208C 2.7 3.0 2.7
a
A þ sign after a number indicates a minimum value.
b
To convert J to cal, divide by 4.184.
c
10 mil ¼ 6:254 mm.

5.1. Straight-Run Asphalt. The largest use of straight-run asphalts


is in the paving industry (101) where they serve primarily as binders in paving
mixes and as bases in liquid asphalts used as seal coatings, surface treatments,
road mixes, and soil stabilizers. The most important technical innovation in
asphalt paving has been to utilize asphalt throughout the entire pavement struc-
ture (termed total asphalt) to provide more efficient distribution of traffic stres-
ses to the subgrade and provide better protection of the base from intrusion of
outside materials such as water and soil. It has been established that 2.5 cm of
asphaltic concrete can replace at least 5 cm of aggregate base for this purpose. In
these mixes the binder contents usually vary from 4.0–10% depending on the
aggregate gradation and nature and the construction.
The hot mixes are designed by using a standard laboratory compaction
procedure to develop a composition reflecting established criteria for volume
percent air voids, total volume percent voids between aggregate particles, flow
and stability, or compressive strength. Tests such as the Marshall, Unconfined
Vol. 2 ASPHALT 197

Compression, Hubbard-Field, Triaxial Procedure, or the Hveem stabilometer


method are used (109).
If straight-run asphalts are reduced to a hardness below 300 penetration
(30 mm), they are termed asphalt cements. For hot-mix paving, either the
AC-10 viscosity grade or the next harder AC-20 grade is commonly specified.
Propane-precipitated asphalts have properties similar to the straight-run
asphalts, assuming the same crude source, but are also obtainable in harder
grades with higher softening points. The possible degree of reduction is much
greater than with the distillation process. A practical limitation is the ability
to remove and handle the precipitated asphalt fraction. These materials are
often employed in fesco board manufacture and as a component in self-sealing
roofing shingle adhesive.
5.2. Air-Blown Asphalts. Air-blown asphalts can be used in prepared
roofings (110, 111) as well as for saturating the felts used in asphalt shingles
and mineral-surfaced roll roofing. The shingle saturant is commonly a flux air-
blown to the 50–608C softening point level with a 50–90 penetration (5–9 mm)
at 258C, and the coating is the same flux air-blown to the 100–1158C softening
point level with a penetration of 16–25 (1.6–2.5 mm). The coating asphalt is
commonly filled with 50–60% mineral stabilizer, most of which passes a 74 mm
(200 mesh) screen, and then is applied at approximately 35% of the finished
weight of roofing shingles. This coating is also lightly applied to the back of
the felt for protection. In addition to providing the waterproof surfacing, the coat-
ing also bonds the colored granules to the saturated felt. The opacity of the gran-
ule protects the coating from the hardening influence of light catalyzed
reactions. The combination provides economical but highly durable mineral-sur-
face shingles or roll roofing.
Built-up roofing constitutes several plies of a saturated roofing felt (low
melt, flexible asphalt saturant) with each ply mopped in place and the structure
covered by air-blown asphalts of from 608 to 1058C softening point, with the
hardness selected depending primarily on roof slope. These roofs are usually
surfaced with mineral aggregates.
A wide variety of industrial products are made from air-blown asphalts,
such as laminants, pipeline coatings, waterproofings, potting compounds, and
joint fillers. Additionally, the oxidized asphalts can be cut back with various sol-
vents to produce a variety of protective coatings and paints. These coatings are
often reinforced with mineral fillers to produce cold-applied roof coatings,
adhesives, sound deadeners, railroad protective coatings, and undercoatings
for automobiles.
5.3. Thermal Asphalt. Thermal asphalt products are in low supply
because the thermal process has been virtually replaced by catalytic cracking
processes. Thermal pitches, because of their high viscosity temperature suscept-
ibility, are very hard at ordinary temperatures (Table 9), but become quite fluid
when heated. They are especially useful as binders and as saturants for fiber-
boards, and as sizings for fiberboard when ground or dispersed and introduced
directly into the wet furnish. Their weatherability on exposure in thin films is
such that they are usually not satisfactory for use as protective coatings.
5.4. Liquid Asphalt. Liquid asphalt products comprise cutback asphalts
and emulsions. A number of grades of different viscosities are available, which
198 ASPHALT Vol. 2

permit application from ambient temperatures to 1508C. The lower viscosity


products are used for dust-laying purposes and as tack coats, prior to
laying asphalt surface courses. The heavier grades are used for mix-in-place
road mixes.
The properties of asphalt emulsions (ASTM D977 and D2397) allow a
variety of uses. The rapid-setting grades are used for surface treatment and
seal coat or surface treatments requiring cover aggregate. The medium-setting
grade has higher stability to break on contact with surfaces and is used for
mix-in-place macadam aggregates. It may also be used in plant mixes to pro-
duce a cold-laid macadam aggregate. Both of these types are less stable than
the slow-setting type which has sufficient stability so that a finely divided
material such as cement may be mixed without breaking the emulsion. The
slow-setting grade is suitable for mixing with very finely divided dense-graded
aggregates, or for the so-called slurry seal. The slow-setting grade is also used
as a tack coat, and as a prime for dust laying and for soil treatment mixtures
(Table 10).
In addition to the paving grades of emulsions, a variety of products are
available that are suitable for industrial uses, for paper sizings, and for use in
the roofing application field. Asphalts emulsified with colloidal clays are espe-
cially suitable for outdoor applications because of their excellent weathering
properties. Some of these products have been adapted for use with roofing sys-
tems, either with brush or spray application.

6. Modified Asphalts

The deterioration of asphalt pavements under heavy traffic loads and/or under
hostile weather conditions, coupled with the inability of many modern crude
oil feedstocks to meet the challenge of producing specification-grade asphalts,
has resulted in the search for asphalt modifiers. These are materials that can
be added to the asphaltic binder to improve service life through enhancement
of a particular property or providing a stronger bond between the binder and
the aggregate thereby reducing the susceptibility of the mix to, for example,
water damage (112,113).
A variety of materials has been proposed to modify the properties of
asphaltic binders to enhance the properties of the mix (112), including fillers
and fibers to reinforce the asphalt–aggregate mixture (114), sulfur to strengthen
or harden the binder (115,116), polymers (98,117–121), rubber (122), epoxy–
resin composites (123), antistripping agents (124), metal complexes (125,126),
and lime (127,128). All of these additives serve to improve the properties
of the binder and, ultimately, the properties of the asphalt–aggregate mix.
The addition of fossil fuel derived materials has also been investigated. Of
particular interest are the materials which contain functional groups that are
capable of enhancing the asphalt–aggregate bond. Modifiers such as those iso-
lated from the products of coal liquefaction (129) and the components of shale
oil and even shale oil residua (130) have been used. The latter is especially inter-
esting since the concept focuses on the use of basic nitrogen functions in the shale
oil which enhance the adhesion of the asphalt to the aggregate. Test strips
Vol. 2 ASPHALT 199

Table 10. General Uses of Emulsified Asphalta


D977b
RS-1, RS-2, MS-1, MS-2h, SS-1, SS-1h,
Type of construction CRS-1c CRS-2c HFMS-1 HFMS-2h CSS-1c CSS-1hc
Bituminous-aggregate mixtures
for pavement bases and surfaces
plant mix (hot) D2629 Xd
plant mix (cold)
open-graded aggregate X
dense-graded aggregate X X
travel plant (mixed-in-place)
open-graded aggregate X X
dense-graded aggregate X X
sand X X
sandy soil X X
slurry seal X X
Bituminous-aggregate applications
treatments and seals
single-surface treatment X X
(chip seal)
multiple-surface treatment X X
sand seal X X X
penetration macadam
large voids X
small voids X
Bituminous applications
fog seal Xe Xf Xf
prime coat-penetrable surface Xf Xf
tack coat Xe Xf Xf
dust binder Xf Xf
mulch treatment Xf Xf
crack filler X X X
Maintenance mix
immediate use X X
a
Only those grades of emulsified asphalt in general use have been indicated herein. It is possible that
under certain variation of aggregates and/or climatic conditions, additional selections might be appro-
priate. Where the use of emulsified asphalt for applications other than those listed in the table are
contemplated, the emulsion supplier should be consulted.
b
Unless otherwise noted.
c
Cationic D2397.
d
ASTM D2629 permits use of other emulsion grades by note; grades of emulsion other than MS-2h
may be used where experience has shown that they give satisfactory performance.
e
Diluted with water by manufacturer.
f
Diluted with water.

of shale oil modified asphalt have been laid on highways in various parts of the
United States.

7. Aggregates

Although much of the focus tends to be on the composition and behavioral char-
acteristics of the asphalt (ie, the binder or cement), the properties of the aggre-
gate also play a significant role in determining the ultimate properties of the
asphalt–aggregate mix. There are many factors to consider and they include:
200 ASPHALT Vol. 2

the functionalities in the asphalt (131), absorption of the asphalt into the
pores of the aggregate (132), the aging properties of the asphalt when it is
on the aggregate (133), as well as methods for coating the asphalt on to the
aggregate (134).
Attempts have also been made to carry out surface modifications of the
aggregate to enhance interactions with the asphalt (135) and other workers
have made attempts to measure or predict the strength and type of asphalt–
aggregate bonds (136,137). However, it must also be remembered that mix
design parameters play an important role in determining the performance of
asphalt–aggregate mixes (138–142).

8. Health and Safety

The word asphalt has been carelessly used in that it is not adequately differen-
tiated from thermally degraded materials, especially coal tar and derivatives. It
is essential to differentiate asphalt from these materials which contain known
carcinogens and health hazards. For this reason, the use of cracked asphalts
must be treated with caution.
There are a variety of tests for asphalt emissions (Table 11). No signifi-
cant air pollution problems are associated with emissions from hot paving
operations using several asphalt cements (143). In fact, the concentrations of
gaseous substances and emissions from paving asphalt cement have been
found to be in very low concentration and within existing EPA and OSHA
standards, even when the ambient air sampling was done under confined
conditions (144–146). In addition, the high molecular weight polynuclear aro-
matic constituents that occur in asphalt have been studied as health hazards
(147–151). The general conclusion from these studies is that surveillance
should be continued although asphalt has not been shown to be a material of
significant hazard.
Steps to minimize potential safety hazards in the handling of asphalt are set
forth by the American Petroleum Institute (152). These include: (1) sudden pressure
increases from hot asphalt in contact with moisture in enclosed tanks or transports;
(2) exposure to air at 1508C or above; (3) local overheating above heating coils;
flashing of asphalt volatiles in the presence of an ignition source or possible auto-
ignition; and (4) hydrogen sulfide from high temperature operations.

Table 11. Analysis Techniques for Asphalt Emissions


Sample Analytical technique
air mass spectrometry
volatile hydrocarbons (C1–C6) gas chromatography
carbon monoxide infrared spectroscopy
volatile sulfur compounds gas chromatography
nitrogen dioxide uv-vis spectroscopy at 550 nm
aldehydes uv spectroscopy
phenols uv spectroscopy
particulates optical microscopy
polynuclear aromatics (PNA) gas chromatography
Vol. 2 ASPHALT 201

BIBLIOGRAPHY

‘‘Asphalt’’ in ECT 1st ed., Vol. 2, pp. 164–199, by A. P. Anderson, Shell Oil Company, and
V. A. Endersby, Shell Development Company; in ECT 2nd ed., Vol. 2, pp. 762–806, by A. J.
Hoiberg, The Flintkote Company, L. W. Corbett, Esso Research & Engineering Company,
and R. B. Lewis, Shell Oil Company; in ECT 3rd ed., Vol. 3, pp. 284–327, by J. V. Evans,
Amoco Oil Company.

CITED PUBLICATIONS

1. Terms Relating to Materials for Roads and Pavements, ASTM Designation D8,
American Society for Testing and Materials, Philadelphia, Pa., 1976.
2. P. H. Pfeiffer, The Properties of Asphaltic Bitumen, Elsevier Publishing Company,
Amsterdam, The Netherlands, 1950.
3. H. Abraham, Asphalts and Allied Substances, 6th ed., Vol. 1, Van Nostrand Co. Inc.,
Princeton, N.J., 1960, Chapt. 2.
4. A Glossary of Petroleum Terms, The Institute of Petroleum, London, UK, 1961.
5. J. G. Speight, Fuel Science and Technology Handbook, Marcel Dekker, Inc.,
New York, 1990.
6. J. G. Speight, The Chemistry and Technology of Petroleum, 2nd ed., Marcel Dekker,
Inc., New York, 1991.
7. J. S. Miller, Jr., in E. A. Dunstan, ed., The Science of Petroleum, Vol. 4, Oxford
University Press, New York, 1938.
8. B. Redwood, Treatise on Petroleum, 5th ed., Vol. 1, Griffin, London, 1976, plate 7,
pp. 53, 208–212.
9. Herodotus, Persian Wars, Vol. 6, Modern Library, New York, p. 119.
10. A. Sequin, Proc. 2nd World Pet. Congr. 4, 56 (1937).
11. R. J. Forbes, Bitumen 5, 9, 41, 63 (1935).
12. R. J. Forbes, Bitumen and Petroleum in Antiquity, E. J. Brill, Leiden. The
Netherlands, 1936.
13. W. A. Attwooll and D. C. Broome, Trinidad Lake Asphalt, 2nd ed., Baynard Press,
London, UK, 1954.
14. Proc. Assoc. Asphalt Paving Technol. 43A, (1974).
15. Statistical Abstracts of the United States, U.S. Department of Commerce, Washington,
D.C., 1961, p. 716.
16. W. R. Jillson, Pan Am. Geol. 41, 251 (1924).
17. N. Berkowitz and J. G. Speight, Fuel 54, 138 (1975).
18. J. C. Petersen, Fuel Sci. Technol. Int. 6, 255 (1988).
19. W. L. Nelson, Petroleum Refining Engineering, 4th ed., McGraw-Hill Book Co., Inc.,
New York, 1958.
20. C. H. Nielson, Distillation in Practice, Chapman and Hill Ltd., London, UK, 1956.
21. H. M. Chelton and R. N. Traxler, Proc. 5th World Pet. Congr., New York, 1959, p. 247.
22. A. E. Leybourne, III and H. E. Schweyer, Ind. Eng. Chem. Process Des. Dev. 1, 127
(1962).
23. I. Ishai and Y. A. Tuffour, Proc. Assoc. Paving Technol. 56, 599 (1987).
24. M. A. Quddus and F. Khan, Fuel 69, 509 (1990).
25. L. W. Corbett and R. E. Swarbrick, Proc. Assoc. Asphalt Paving Technol. 29, 104 (1960).
26. A. J. Hoiberg and W. E. Garris, Jr., Ind. Eng. Chem. Anal. Ed. 16, 294 (1944).
27. S. E. Moschopedis and J. G. Speight, Fuel 57, 235 (1978).
28. W. L. Nelson, Oil Gas J. 53(16), 158 (Aug. 23, 1954).
202 ASPHALT Vol. 2

29. D. B. Pruess, Pet. Eng. 30, C-6 (Oct. 1958).


30. R. L. Purvin, Pet. Eng. 22, C-28 (Apr. 1950).
31. W. H. Shearon and A. J. Hoiberg, Ind. Eng. Chem. 41, 2672 (1949).
32. W. L. Nelson, Oil Gas J. 46(25), 99 (Oct. 25, 1947).
33. A. J. Hoiberg, Proc. Assoc. Asphalt Paving Technol. 19, 225 (1950).
34. Aust. Pat. 153,506 (June 10, 1938), R. Lichtenstein.
35. U.S. Pat. 1,782,186 (Nov. 18, 1930), G. Abson (to Chicago Paving Laboratory Inc.).
36. U.S. Pat. 2,200,914 (May 14, 1940), R. E. Burk and C. H. Whitacre (to Standard Oil
Company of Ohio).
37. U.S. Pat. 2,450,756 (Oct. 5, 1948), A. J. Hoiberg (to Lion Oil Company).
38. G. L. Oliensis, Proc. Assoc. Asphalt Paving Technol. 6, 88 (1935).
39. D. W. Gagle, Refining Eng. 30, C-15 (Mar. 1958).
40. L. Lowry, Pet. Eng. 29, C-50 (1957).
41. E. W. Mertens, L. D. Coyne, and E. D. Rogers, Am. Soc. Test. Mater. Spec. Tech.
Publ. 294, 68 (1960).
42. H. Al-Abdulwahhab, F. Bayomy, and A. Al-Halhouli, Trans. Res. Rec. 1106(1), 71
(1987).
43. W. A. McIntosh, Ind. Eng. Chem. 44, 1656 (1952).
44. K. H. Altgelt, Die Makromolekulare Chemie 88, 75 (1965).
45. J. B. Green, J. W. Reynolds, and S. K-T Yu, Fuel Sci. Technol. Int. 7, 1327 (1989).
46. D. A. Anderson, J. C. Petersen, and D. W. Christensen, Proc. Assoc. Asphalt Paving
Technol. 55, 250 (1986).
47. D. S. Decker and J. L. Goodrich, Proc. Assoc. Asphalt Paving Technol. 58, 503 (1989).
48. U.S. Pat. 4,904,305 (1990), L. Zanzotto and co-workers (to Nova Husky Research
Corp., Canada).
49. W. A. Gruse and D. R. Stevens, Chemical Technology of Petroleum, 3rd ed.,
McGraw-Hill Book Co., Inc., New York, 1960, p. 585.
50. J. Marcusson, Z. Agnew Chem. 29, 1, 21 (1916).
51. L. W. Corbett, Anal. Chem. 36, 1967 (1964).
52. L. W. Corbett, Anal. Chem. 41, 576 (1969).
53. W. R. Middleton, Anal. Chem. 39, 1839 (1967).
54. F. S. Rostler and H. W. Sternberg, Ind. Eng. Chem. 41, 598 (1949).
55. F. S. Rostler and R. M. White, Proc. Assoc. Asphalt Paving Technol. 31, 35 (1962).
56. R. N. Traxler and H. E. Schweyer, Oil Gas J. 52, 158 (1953).
57. A. C. Pitchford and W. M. Axe, Pet. Rep. 6(3), B-43 (1961).
58. E. C. Knowles and H. Levin, Ind. Eng. Chem. Anal. Ed. 13, 314 (1941).
59. R. A. Gardner, J. Chem. Eng. Data 4, 155 (1959).
60. R. N. Traxler, Chem. Rev. 19, 119 (1936).
61. L. H. Csanyi and B. S. Bassi, Proc. Assoc. Asphalt Paving Technol. 27, 52 (1958).
62. E. S. Hillman and B. Barnett, Proc. Am. Soc. Test. Mater. 37, 559 (1937).
63. Grader, Oe. u Kohle 38, 867 (1942).
64. J. V. Evans and L. E. Ott, in Ref. 46, p. 353.
65. P. W. Jennings and J. A. S. Pribanic, in Ref. 45, p. 1269.
66. C. Gaestel, R. Smadja, and K. Lamminan, Rev. Gen. des Routes et Aerodromes 466,
85 (1971).
67. J. C. Petersen and H. Plancher, Anal. Chem. 47, 112 (1975).
68. R. Fish, Anal. Chem. 56, 2452 (1984) and references cited therein.
69. S. E. Moschopedis, J. F. Fryer, and J. G. Speight, Fuel 55, 227 (1976).
70. J. A. Koots and J. G. Speight, Fuel 54, 179 (1975).
71. W. O. Yandell, in O. E. Briscoe, ed., Asphalt Rheology: Relationship to Mixture,
ASTM Special Publication No. 941, ASTM, Philadelphia, Pa., 1985.
72. M. Tia and B. E. Ruth, in Ref. 71.
Vol. 2 ASPHALT 203

73. A. W. Sisko, Highw. Res. Rec. 67, 27 (1965).


74. K. A. Lammiman and J. E. Roberts, Lab. Pract. 10, 816 (1961).
75. R. Roque, M. Tia, and B. E. Rutto. in O. E. Briscoe, ed., Asphalt Rheology: Relation-
ship to Mixture, ASTM Special Publication No. 941, ASTM, Philadelphia, Pa., 1985.
76. C. Walther, Oil Kohle 1, 71 (1933).
77. J. F. Brodnyan, Highw. Res. Board Bull. 192, 1 (1958).
78. F. H. Gaskins, Trans. Soc. Rheol. 4, 265 (1960).
79. J. L. Williams, R. F. Landel, and J. D. Ferry, J. Am. Chem. Soc. 77, 3701 (1955).
80. R. N. Traxler, Ind. Eng. Chem. 30, 322 (1938).
81. R. H. Lewis and J. Y. Welborn, Public Roads 21, 1 (1940).
82. R. N. Traxler, H. E. Schweyer, and J. W. Romberg, Ind. Eng. Chem. 36, 823 (1944).
83. C. Van der Poel, in M. Reiner, ed., Building Materials, Their Elasticity and Inelas-
ticity, Interscience Publishers, New York, 1954, p. 373.
84. R. N. Traxler, Asphalt, Its Composition, Properties and Uses, Reinhold Publishing
Co., New York, 1961, Chapt. 6.
85. A. J. Hoiberg, Bituminous Materials: Asphalts, Tars, and Pitches, Vols. 1, 2, and 3,
Wiley-Interscience, New York, 1964–1966.
86. K. G. Martin, Am. Chem. Soc., Div. Pet. Chem. Prepr. 16, D111 (1971).
87. J. C. Petersen, in Ref. 86, p. D27.
88. L. R. Kleinschmidt and H. R. Snoke, J. Res. Nat. Bur. Stand. 63C, 31 (1959).
89. S. H. Greenfield, J. Res. Nat. Bur. Stand. 63C, 287 (1960).
90. J. W. H. Oliver and H. Gibson, in Ref. 86, p. D60.
91. J. O. Harris, Microbiological Studies Reveal Significant Factors in Oil and Gas
Pipeline Back-Filled Ditches, Technical Bulletin 102, Kansas State University,
Manhattan, Kans., 1959.
92. K. G. Martin, Deterioration of Bituminous Roofing Fabrics, Commonwealth Scienti-
fic and Industrial Research Organization, Technical Paper 11, Division of Building
Research, Melbourne, Aust., 1961.
93. S. H. Greenfeld, The Effects of Mineral Additives on the Durability of Coating Grade
Roofing Asphalts, Building Materials and Structures Report 147, National Bureau of
Standards, Washington, D.C., 1956.
94. D. A. Tamburro, A. T. Blekicki, and J. H. Kietzman, in Ref. 55, p. 151.
95. J. L. Goodrich and H. L. Dimpfl, Proc. Assoc. Asphalt Paving Technol. 55, 57 (1986).
96. R. L. Dunning, in Ref. 46, p. 33.
97. A. S. Noureldin and L. E. Wood, in Ref. 47, p. 256.
98. J. Zielinski and A. Bukowski, J. Thermal Anal. 32, 1797 (1987).
99. K. W. Lee and M. I. Al-Jarallah, Trans. Res. Rec. 1096, 81 (1986).
100. J. Bonnot, Trans. Res. Rec. 1096, 42 (1986).
101. A Brief Introduction to Asphalt and Some of Its Uses, Manual Series No. 5, The
Asphalt Institute, College Park, Md., 1975.
102. Asphalt in Hydraulics, Manual Series No. 12, The Asphalt Institute, College Park,
Md., 1976.
103. Athletics and Recreation on Asphalt, Information Series No. 147, The Asphalt
Insitute, College Park, Md.
104. C. M. Hansen and A. E. Erickson, Am. Chem. Soc. Div. Pet. Chem. Prepr. 12,
C164–C169 (1968).
105. L. C. Brunstrum, L. E. Ott, and T. L. Speer, Proc. 7th World Pet. Congr. 7 (1968).
106. L. C. Brunstrum, L. E. Ott, and T. L. Speer, Proc. Am. Pet. Inst. Sect. 3, (1968).
107. Asphalt Mulch Treatment, Information Series IS-161, The Asphalt Institute, College
Park, Md., 1960.
108. Railway Roadbeds for Tomorrow, Information Series IS-137, The Asphalt Institute,
College Park, Md.
204 ASPHALT Vol. 2

109. Mix Design Methods for Asphaltic Concrete, Manual Series No. 2, The Asphalt
Institute, College Park, Md., 1974.
110. R. N. Traxler, Asphalt, Reinhold Publishing Co., New York, 1961.
111. Roofing and Shingles; Asphalt-Prepared, Mineral Surfaced, Federal Specification
SS-R-521, General Services Administration, Washington, D.C.
112. W. A. Higgins, Trans. Res. Rec. 1115, 89 (1987).
113. J. A. Scherocman, K. A. Mesch, and J. J. Proctor, in Ref. 46, p. 213.
114. U.S. Pat. 4,634,622 (1987), K. L. Jenkins and co-workers (to Manville Corp.).
115. U.S. Pat. 4,769,288 (1988), D. Saylak (to Texas A & M University).
116. W. Akili and G. J. Courval, in Ref. 71.
117. U.S. Pat. 4,639,273 (1987), D. W. Gilmore and T. G. Kugele (to Morton Thiokol Inc.).
118. L. D. Coyne, Proc. Assoc. Asphalt Paving Technol. 57, 545 (1988).
119. C. Lenoble, Am. Chem. Soc. Div. Pet. Prepr. 35(3), 541 (1990).
120. T. S. Shuler, J. H. Collins, and J. P. Kirkpatrick, in Ref. 71.
121. H. W. Murray, G. N. King, and J. B. Prudhomme, in Ref. 71.
122. H. B. Takallon, R. G. Hicks, and D. C. Esch, in Ref. 99, p. 68.
123. C. H. Lhymn, J. Mater. Sci. 8, 1019 (1989).
124. H. W. Busching and co-workers, in Ref. 46, p. 120.
125. U.S. Pat. 4,721,529 (1988), T. E. Mullins (to Unichem International Inc.).
126. J. Epps and co-workers, in Ref. 99, p. 106.
127. T. W. Kennedy, Bulletin No. 325, National Lime Association, Arlington, Va.
128. M. Stroup-Gardiner and J. Epps, in Ref. 112, p. 12.
129. H. H. Yoon and A. R. Tarrer, Am. Chem. Soc. Div. Fuel Chem. Prepr. 33(4), 975
(1988).
130. U.S. Pat. 4,325,738 (1982), H. Plancher and J. C. Petersen (to U.S. Department of
Energy).
131. C. W. Curtis, Y. W. Jeon, and D. J. Clapp, in Ref. 45, p. 1225.
132. J. A. Guin, H. S. Chang, and S. C. Yen, Am. Chem. Soc. Div. Fuel Chem. Prepr. 35(3),
370 (1990).
133. J. F. McKay, in Ref. 132, p. 496.
134. U.S. Pat. 4,692,350 (1987), A. R. Clarke, C. L. Martin, and R. H. Bowering (to Mobil
Oil Corp.).
135. R-F. Ju, Y-L. Lee, S. H. Lin, and W. J. Burke, Colloids and Surfaces 49, 395
(1990).
136. R. T. Podoll and K. C. Irwin, in Ref. 132, p. 346.
137. A. R. Tarrer and V. P. Wagh, in Ref. 132, p. 361.
138. F. M. Bayomy and J. A. Naji, in Ref. 47, p. 362.
139. R. L. Davis, in Ref. 71.
140. R. Roque, M. Tia, and B. E. Ruth, in Ref. 71.
141. N. W. McLeod, in Ref. 71.
142. P. S. Kandhal and W. C. Koehler, in Ref. 71.
143. Asphalt Hot-Mix Emission Study, Research Report RR-75-1, The Asphalt Institute,
College Park, Md., 1975.
144. R. P. Hangebranck, D. J. von Lehmden, and J. E. Meaker, Sources of Polynuclear
Hydrocarbons in the Atmosphere, Publication No. 999-AP-33, U.S. Public Health
Service, Washington, D.C., 1967.
145. Particulate Polycyclic Organic Matter, National Academy of Sciences, Washington,
D.C., 1972.
146. J. M. Colucci and C. R. Begeman, Environ. Sci. Technol. 5, 145 (Feb. 1971).
147. Petroleum Asphalt and Health, API Medical Research Rep. EA 7103, American
Petroleum Institute, Washington, D.C., Dec. 1, 1971.
148. Health Lab. Sci. 11, 218 (July 1974).
Vol. 2 PETROLEUM COKE 205

149. C. H. Baylor and N. K. Weaver, Arch. Environ. Health 17, (Aug. 1968).
150. W. Luinsky, Anal. Chem. 32, 684 (1960).
151. W. C. Hueper, Occupational Tumors and Allied Diseases, Charles C Thomas,
Springfield, Ill., 1942, p. 69.
152. Guide for the Safe Storage and Handling of Heated Petroleum-Derived Asphalt
Products and Crude Oil Residue, Publication 2023, American Petroleum Institute,
Washington, D.C., Mar. 1977.

JAMES G. SPEIGHT
Western Research Institute

PETROLEUM COKE
1. Introduction

Petroleum coke is a grayish-black solid residue with a high carbon content. It is


obtained during the thermal conversion process in crude oil processing. Although
petroleum coke has been known for more than 100 years, both production and its
specific application and improvement have seen an increasing boom only during
the last 50 years. It is anticipated that up to the year 2000 40  106 t/a of petro-
leum coke will be produced in oil refineries. The majority of this quantity is
obtained in delayed coking plants. In other facilities for processing crude oil resi-
due components, investments are made at present in flexicoking units, whereas
fluid cokers are still run but hardly set up in new production.

2. Physical and Chemical Properties

Physical Properties. Petroleum coke is a grayish-black solid matter of


very varied composition. Generally, a distinction is made between petroleum
coke from fluid and flexicoking plants and that derived from delayed coking pro-
cesses. The petroleum coke from delayed cokers is also called Petroleum coke-
green–>green coke. The fluid and flexicokers supply a fine-grained petroleum
coke that has a high level of isotropy and in general an ash content of between
1 and 4 wt %. The grain consistency of these cokes is high, the content of volatile
combustible matter (VCM) is low and hardly exceeds 5 wt %. In contrast petro-
leum coke from delayed cokers is normally characterized by both high porosity
and anisotropy. The VCM content is between 7 and 14 wt %; it is only set to
< 5 wt % for needle coke production. The ash content is low, generally
< 0.1 wt %. The bulk density of normal-quality petroleum coke is around 0.8 –
0.9 g/cm3. If however, the feed stream to the delayed coker is rich in high mole-
cular asphalt, a non-porous isotropic petroleum coke is the result which has a
bulk density of 1.2 – 1.4 g/cm3. The formation of this shot coke quality can be
counteracted by machine adjustment.

ß2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim


10.1002/14356007.a19_235
206 PETROLEUM COKE Vol. 2

Another basic form of petroleum coke is calcined coke. Calcined coke is a


product derived from green coke. Here, the hydrocarbons have been removed
by heating green coke under reducing conditions in kilns or hearths to
> 1300 8C. Classification of petroleum coke types derived from delayed cokers
into:

– Shot coke
– Sponge coke (fuel coke)
– Sponge/honeycomb coke (regular or anode-grade coke)
– Needle coke (graphite or premium coke)

is not defined precisely (1). A differentiation with regard to isotropy and crystal
structure is logical. There is a graded transition from:

Shot coke (mainly isotropic; amorphous, no crystal structures formed;


hardly any pores)

to

needle coke (high anisotropy; regular crystal structures, containing a great


number of fine pores, crystallite size of the order of 4 – 7 nm).

Petroleum coke results from a thermal cracking and polymerization reaction.


After going through the mesophase a solid crystalline form is formed, which,
contrary to indications (2) does not plastify again. Coking only occurs in the
VCM particles contained. If their fraction exceeds 15 %, the release of this
petroleum coke at high temperatures gives the impression of a plastic state
of aggregation. By treating petroleum coke with steam and hot water, those
components that are slightly water soluble can be washed and steamed out
of the petroleum coke.
Chemical Properties and Composition. Under normal conditions petro-
leum coke has a low reactivity. At higher temperature (up to 1400 8C, calcina-
tion conditions) an expansion of the crystallite size up to 2.5 – 3.5 nm takes
place because of the cracking of side chains. At the same time the original
hydrogen content of ca. 3.5 wt % is reduced to below 0.1 wt %. Under graphitiza-
tion conditions (ca. 2900 8C) good crystallized forms of petroleum coke (needle
coke character) are transformed into graphites. These graded changes in
quality of petroleum coke with the temperature demonstrate best the structure,
character, and classification of the product.
The dominant element in petroleum coke is carbon. The hydrogen content
is around 3.5 wt %. An overproportional amount of sulfur and nitrogen reaches
the petroleum coke from the feed components (S content of petroleum coke
0.1 – 5 wt %; N content 0.2 – 1.5 wt %). The V and Ni contents of the feed
streams remain entirely in the petroleum coke. The Na, Ca, and Mg content
in the petroleum coke is the sum of the content of these cations in the crude
oil and of their concentration in the cutting water (Section 9). According to
the 21st ATP (Adaptation of Technical Progress of the European Substance
Vol. 2 PETROLEUM COKE 207

Directive) and CONCAWE, product dossier no. 93/105 (3), petroleum coke is
considered a nondangerous product. The level of polycyclic aromatics in petro-
leum coke is about 250 mg/kg. Benzo[a]pyrene makes up about 25 mg/kg. The
amount of heavy metals apart from V and Ni is insignificant (< 1 mg/kg). In
general the quality of the feed components and the production conditions
2
determine the quality of the product.1

3. Production

In 1996 about 35  106 t of petroleum coke were produced worldwide. This is


about 10 % of the quantity of blast-furnace coke (hard coke) produced from hard
coal (4). More than 95 % of the petroleum coke is produced in delayed cokers,
about a third of the feed streams is incurred as petroleum coke. Residues and
crack components (see Section 2) are used in delayed cokers in more than 55 refi-
neries. Due to the reaction conditions only about 5 – 10 wt % of the feed material is
converted into petroleum coke in the case of fluid and flexicokers. The number of
these production units is much lower. In 1996 13 fluid cokers and 5 flexicokers
were in production worldwide. Most petroleum coke is produced in North America,
followed by Western Europe. Refineries producing anode grade coke about two-
thirds of the delayed coker complexes in production are listed in Table 1. Anode
grade is the most important product with regard to margin and value.
3.1. Production Processes. Delayed Coking. Coking is an effective
way for an integrated refinery to process the residues of crude oil. Coking pro-
cesses are normally carried out to minimize coke and maximize liquid products.
The first atmospheric coking processes were started in 1928. Coking, in general,
has become more important over the years because tradtional heavy fuel oil
demand has declined and the demand for lighter products has increased. This
has made it more attractive to convert the heavy fuel oil components into better
marketable light, clean components. The present major source of production of
marketable coke is the delayed coking process developed by Standard Oil. Con-
ventional delayed coking is shown in the flow diagram in Figure 1. A delayed
coke complex is depicted in Figure 2.
The refinery residues are heated in a furnace (b) up to about 485 – 500 8C.
After passing a fractionator tower where middle distillates are topped off, then
they are fed into one of the coking chambers. The energy gained in the furnace
passages is sufficient to release the cracking reaction in the filling time of the
coking chamber (delayed coking). The reaction zone is also called mesophase,
because the transition from liquid residue molecules to solid crystalline coke
structure occurs here. The pressure in the coking chamber can be varied
between 2 and 6 bar. In the furnace, steam is dosed to the residue to inhibe
the formation of coke. The vapors in the coking chamber — gases, naphtha,
middle distillates, and a heavy coker gas oil — are led to the fractionator for
separation, fractionated and forwarded for stabilizing post-treatment. The
heavy coker gas oil is recycled as coker feed or is used in other refinery pro-
cesses: eg, after go-fining (a hydrogenation process) as feed for fluid catalytic
cracking (6).
208 PETROLEUM COKE Vol. 2

Table 1. Refineries Producing Anode Grade Petroleum Coke (5)


Apparent capacity
Owner Location State green coke (t/d)
North America
Arco Los Angeles California 2500
Arco Cherry Point Washington 2500
Alb Petro-Canada Edmonton Canada 430
BP Oil Alliance Louisiana 850
BP Oil Lima Ohio 700
BP Oil Toledo Ohio 400
Chevron El Segundo California 3000
Chevron Porth Arthur Texas 1800
Chevron Salt Lake City Utah 350
Conoco Ponca City Oklahoma 800
Crown Houston Texas 450
Derby Wichita Kansas 200
El Paso Refining El Paso Texas 130
Exxon Baton Rouge Louisiana 4800
Farmland Coffeyville Kansas 600
Koch Carbon Corpus Christi Texas 375
La Gloria Tyler Texas 230
Landmark Fruita Colorado 180
Marathon Robinson Illinois 800
Mobil Beaumont Texas 1500
Powerine Santa Fe California 550
Sas Consumers Coop Regina Canada 360
Shell Norca Louisiana 1000
Sun Tulsa Oklahoma 300
Texaco Bakersfield California 550
Texaco Anacortes Washington 100
Unocal Rodeo California 1250
Oceania/South America
Pertamina Dumai Indonesia 850
Yacimientos Petroliferos La Plata Argentina 1000
Yacimientos Petroliferos Lujan De Cujo Argentina 1500
Western Europe
DMP Mineroel Burghausen Germany 680
Esso AG Karlsruhe Germany 1200
Ruhr Oel Gelsenkirchen Germany 1200
Wintershall Lingen Germany 420
Raffinor AS Mongstad Norway 480
Repsol La Coruna Spain 550
Conoco Immingham United Kingdom 2150

While one coking chamber is being filled in the cycle time, the other cham-
ber is emptied. A typical volume of a modern coke drum is about 1000 m3 with a
range in diameter of 5 to 8 m and in height of 25 to 35 m. After steaming and cool-
ing of the coke chambers, the coke is removed by drilling and cutting with high
pressure water (up to 300 bar pressure). The green coke obtained is conveyed to
its application after the draining step: either as fuel of feed for calcining. Table 2
shows the different green coke qualities in relation to use.
Vol. 2 PETROLEUM COKE 209

Fig. 1. Flow sheet of delayed coking. a) Fractionator; b) Furnace; c) Coke drum;


d) Crusher; e) Coke dewatering bin; f) Water tank; g) Coke – water pit.

Fig. 2. Photograph of a refining delayed coker complex: furnace, coke drums, fractiona-
tor, and dewatering bins.
210 PETROLEUM COKE Vol. 2

Table 2. Green Coke Qualities in Relation to Use


Heating medium Regular calcinate feed Needle coke feed
Sulfur content, wt % >2 1.0 – 3.0 0.1 – 0.8
V/Ni content, mg/kg 200 – 500 50 – 200 < 50
Ca/Na content, mg/kg 50 – 300 20 – 100 < 50
VCM, wt % 9 – 14 7–9 5–7

The following components can be used as feed for delayed cokers (7):
1. Crude residues (atmospheric or vacuum),
2. Crack components (visbroken tar, cycle oil, decant oil, or thermal crack tar),
3. Deasphalting residues,
4. Coal oils, and
5. Used plastic materials (recycling)
The products of delayed coking processes are (basis: vacuum residue or crack
components):

1. Gas/LPG (ca. 13 wt %)
2. Naphtha (ca. 11 wt %)
3. Middle distillate (ca. 45 wt %)
4. ‘‘Green coke‘‘ petroleum coke (ca. 31 wt %)
Delayed coking is carried out in order to free refinery from the product
‘‘heavy fuel oil‘‘ which is not easily marketable. In terms of investment delayed
coking is a much cheaper technology than flexicoking or hydrocracking (factor 3
to 5). A further argument in favor of delayed coking can be the revalorization
potential for the carbon products ‘‘regular calcinate‘‘ or ‘‘needle coke‘‘ produced.
Moreover, delayed coking shows good product results, which, however, like all
thermal conversions, have ‘‘middle distillates‘‘ as main product.
Figure 3 shows the product yields for the North Sea crude ‘‘Brent‘‘. All ther-
mal conversion processes reduce the crude residues higher than middle distil-
lates, especially the delayed coker unit.Goal of nearly all delayed coker units is
to minimize coke yield and obtain as many distillate fractions as possible. How-
ever, when the objective is to produce green coke that shall serve as raw material
for higher quality coke products (regular calcinate and needle coke) a specific
feed optimum must be found that by no means results in a minimum coke yield.
The distillate yields and thus the capacities of delayed coking can be
increased by (7,8):
1. Preconversion of feed components (6)
2. Addition of light middle distillate to the feed stream
3. High coke drum outlet temperature/high coker furnace outlet temperature
(COT)
4. Low operating pressure
5. Low fresh feed: recycle ratio
6. Addition of coil steam
Vol. 2 PETROLEUM COKE 211

Fig. 3. Product yields with thermal conversion processes.

Some quantitative effects are shown in Figure 4.


The following technical improvements have been achieved for the produc-
tion of petroleum coke in modern delayed coking plants. They result in increased
plant reliability, improved work safety and better coke quality as well as
increased production capacity (10–13):

1. Onstream decoking (allows decoking of one of the four furnace coils during
current production)

Fig. 4. Relationship between plant operating conditions and plant yield; parameters:
coke drum pressure = 2.5 bar; feed/recycle ratio = 1 : 1.3; coke drum outlet tempera-
ture = 440 8C. (9).
212 PETROLEUM COKE Vol. 2

Table 3. Time Allotted for Different Steps in Typical 12-h and 24-h Coking cycles
Time, h
Operation step 24-h cycle 12-h cycle
a
Switch drum 0.5
b
Steam out to fractionator 0.5
Steam out to blowdown 1.0
b
Slow water cooling 1.0
b
Full water rate 5.0
Drain water and remove top head 3.5 1.5
Drop bottom head and feed line 0.5 0.5
Hydraulic decoking 4.0 2.0
Replace heads and feed line 1.0 0.5
Steam purge and pressure test 1.0 0.75
Drum heat up 4.0 2.25
Slippage allowance 2.0 1.0
Total 24 h 12 h
a
Total time for both operation steps: 0.5.
b
Total time for both operation steps: 3.0.

2. Homogeneous petroleum coke quality (requires temperature and recycle


ratio ramps over the cycle time to adjust continuous coking conditions)
3. Optimum coke chamber filling (requires mass balance of feed and products
and measurement of the coke chamber level)
4. Batch computerization (allows automatic checking and starting of opera-
tion steps without losing time)
5. Automatic coke cutting with vibration alarms at coke drum wall
6. Automatic hydraulic coke drum bottom head
7. Combination tool for drilling and cutting
8. Hydraulic feed line
9. Hydraulic coke drum bottom head closing

With these technical measures the cycle time for delayed coking can be reduced
from about 24 to 12 h (14–17). The time required for different steps in typical 12-
h and 24-h coking cycles is listed in Table 3.
Fluid Coking. The continuous fluid coking process utilizes fluidized solids
technique developed for fluid catalytic cracking (! Oil Refining, Chap. 3.2),
except that no catalyst is used. A flow scheme is shown in Figure 5.Residue is
feed to the reactor (a). The cracking reactions occur at 500 – 550 8C in a fluidized
bed of coke particles. The coke formed in the reactor flows continuously to the
heater, where it is heated up to 600 – 650 8C by partial combustion in a second
fluidized bed. Part of the heated coke particles are returned to the reactor to sup-
ply the energy to the endothermic crack reactions and to maintain the reactor
temperature. The remainder of the coke after cooling is removed as a stream
of fine particle ‘‘petroleum coke’’. Fluid coke is only occasionally used for
production of anodes (18).
Flexicoking. The flexicoking process combines fluid coking with gasifica-
tion. A flow scheme of flexicoking is shown in Figure 6. Flexicoking is a residue
Vol. 2 PETROLEUM COKE 213

Fig. 5. Flow sheet of fluid coking. a) Reactor; b) Scrubber; c) Burner; d) Air blower.

Fig. 6. Flow sheet of flexicoking. a) Reactor; b) Scrubber; c) Heater; d) Gasifier; e) Cyclones;


f) Airblower.
214 PETROLEUM COKE Vol. 2

conversion process that combines coke gasification with conventional fluid


coking. Three fluidized beds are in operation at the reactor, heater, and gasifier.
Shrinking markets for the coke obtained in substantial amounts as byproduct of
delayed coking and fluid coking due to restrictions of sulfur emission encouraged
the development of the flexicoking process. In flexicoking most of the coke pro-
duced is converted to low-Btu fuel gas in the gasification reactor. This fuel gas
can be desulfurized to approximately the equivalent of a 0.25 wt % sulfur fuel
oil. Thus, particularly for feed stocks with a high metal and sulfur content, flex-
icoking can be an advantage, because it produces marketable liquid products
plus fuel gas.
Thermal conversion of the feed takes place in the reactor (a) where the feed
is atomized through a series of nozzles into a bed of fluidized coke particles. Coke
occurs in a thin film on the fluidized particles within the bed. The coke bed is
fluidized by the product vapors and by stripping steam injected at the bottom
of the reactor. Typical bed temperatures are 510 – 540 8C. Vapor products from
the conversion reactions in the bed go overhead through cyclone separators,
which remove most of the entrained coke and return it to the reactor bed. The
cyclone outlets discharge the vapor product directly into a scrubber (b) where
heavy liquid is used to scrub out the remaining coke dust and condense uncon-
verted high boiling fractions. The dust laden liquid is recycled as ‘‘slurry cycle‘‘ to
the reactor with the feed. The scrubbed vapor is sent to the coker fractionator,
where the stream is split into gas, naphtha, distillate and heavy gas oil streams.
Coke produced in the reactor is gasified by steam and air in the gasifier (d),
which operates at a bed temperature of approximately 955 8C. The heater (c)
serves as a surge vessel and transfers heat from the gasifier overhead gas to
coke which in turn supplies the heat of reaction in the reactor. The heater bed
temperature is approximately 610 8C. Coke is continuously circulated between
the three vessels to transfer heat and maintain vessel inventories. Heat for the
endothermic cracking reactions in the reactor is supplied by coke combustion in
the gasifier. In addition to the coke fines recovered from the heater overhead gas,
product coke can be withdrawn from the heater bed. All liquid or gas streams
require further treating (19).
3.2. Calcination. About 40 % of the petroleum coke derived from delayed
coking processes is stabilized by calcination. Calcination includes the elimina-
tion, burning, and coking of water and higher boiling hydrocarbons contained
in petroleum coke. Finally a carbon product remains that contains sulfur and
nitrogen plus small amounts of vanadium, nickel, sodium, calcium, iron, and
silicon.
At present calcining is carried out at about 40 calciner sites (see Table 4).
While the calciners in North America are usually run by independent companies
or attributed to aluminum works, the calciners in other regions are production
units of the refineries.
Calcining is carried out in rotary kiln and rotary hearth plants. While the
rotary kiln largely predominates in America, equal amounts of kiln and hearth
plants are used in Europe.
In both types of calcining complexes the water and VCM contents of green
coke are eliminated by heating up to 1400 8C. The heat required for calcining
the green coke is primarily provided by burning the VCM released from the
Vol. 2 PETROLEUM COKE 215
Table 4. Petroleum Coke Calciners (19)
Production
Owner Location State Type capacity, 103 t/a
North America
Alcan Arvida Quebec kiln 215
Alcan Kitimat British Columbia kiln 80
Alcan/Intalco Edmonton Alberta kiln 165
Arco Cherry Point Washington hearth 675
Arco Los Angeles California kiln 315
Calciner Industries Chalmette Louisiana kiln 200
Calciner Industries Gramercy Louisiana kiln 230
Calciner Industries Norco Louisiana kiln 170
Calciner Industries Purvis Mississippi kiln 80
Conoco Lake Charles Louisiana kiln 260
Great Lakes Carbon Enid Oklahoma kiln 470
Great Lakes Carbon Port Arthur Texas kiln 635
Kaiser Aluminium Mead Washington kiln 80
Landmark Petroleum Fruita Colorado kiln 240
Reynolds Metals Baton Rouge Louisiana kiln 620
Reynolds Metals Lake Charles Louisiana kiln 200
Union Chemicals Rodeo California kiln 295
Venco Lake Charles Louisiana kiln 400
Venco Moundsville West Virginia kiln 400
South America
Copetro La Plata Argentina kiln 200
Petrocoque Cubatao Brazil kiln 400
Africa
Egyptalum Nag Hamadi Egypt hearth 135
Asia/Oceania
Government unknown China kiln 455
KOA Oil Marifu Japan kiln 115
Petrocokes Mizushima Japan kiln 80
Various various India kiln 240
Petramina Dumai Indonesia kiln 270
Europe
Conoco Immingham United Kingdom kiln 550
DMP Mineraloel Burghausen Germany hearth 200
Esso AG Karlsruhe Germany hearth 130
Government Sisak Yugoslavia kiln 50
Government various CIS hearth/kiln 2015
Rafinor Mongstad Norway hearth 170
Repsol La Coruna Spain hearth 180
Veba Oel Gelsenkirchen Germany kiln 200
Wintershall Lingen Germany kiln 130

coke calcination process. Combustion air is admitted to the calciner. In normal


operation only the quantity of air sufficient for complete combustion of the
VCM release from the coke is added, no free oxygen occurs in the off-gas.
Rotary Kiln Calciner. The rotary kiln calciner (Fig. 7) consists of a rotat-
ing heated steel tube (e) with a refractory lining. Because of the gradient and the
rotation of the tube, the green coke descends in the calciner and heat flow rises in
countercurrent. The calcined coke leaving the kiln is discharged into a rotary
216 PETROLEUM COKE Vol. 2

Fig. 7. Schematic of the rotary kiln calciner. a) Screen; b) Crusher; c) Weigh feeder;
d) Green coke sampling; e) Rotary kiln; f) Cooler; g) Sampling calcined coke; h) Incinera-
tor; k) Stack.

cooler (f). The flue gases are cooled in a waste-heat boiler (i) and cleaned in an
electrostatic precipitator (j).
Rotary Hearth Calciner. The rotary hearth calciner (Fig. 8) was originally
developed by the Wise Coal and Coke company. Green coke charged at the

Fig. 8. Schematic of the rotary hearth calciner. a) Feed bin; b) Rotary hearth;
c) Calcinate cooler; d) Waste-heat boiler; e) Electrostatic precipitator; f) Stack.
Vol. 2 PETROLEUM COKE 217

periphery of the hearth travels around the calciner (b) in a series of concentric
paths. At the completion of each revolution the coke is moved by a rabble arm
set in the roof into the next concentric path. This process is repeated, until
at the final rabble, the coke is discharged into the soaking pit. From the
soaking pit the coke can be discharged into a rotary cooler (c). During
normal operation the retention time for the coke in the hearth and soaking pit
is approximately 2 h (21).

4. Uses

Worldwide annual consumption of industrial carbon is about 20106 t/a, exclud-


ing metallurgical coke (see Table 5). Two-thirds of this margin is calcinate from
petroleum coke (22,28).
4.1. Green Petroleum Coke. More than half the petroleum coke pro-
duced is marketed as green coke. This green coke is a cheap fuel used primarily
in the cement industry or for electricity generation. Qualities with a sulfur con-
tent below 1.5 % are used as a supplementary substance for hard coke produc-
tion, if the coal has too high a VCM content. In addition, this green coke is
used in the production of carbides, too.
Green coke is sold worldwide. The biggest trans-shipment center with some
million tons of storage capacity is Rotterdam.
4.2. Calcined Petroleum Coke. Anode-Grade Coke (Regular Calci-
nate). Most calcined petroleum coke is used for production of carbon and
Söderberg anodes (Table 5). Moreover, as aluminum production with Söderberg
anodes is declining due to environmental aspects, the requirement will shift
more to carbon anodes. The coke quality required is called anode grade coke or
regular calcinate (12).
Worldwide aluminum production was 24.5  106 t in 1995; 60 % of the
aluminum production capacities are located in North America and Europe.

Table 5. Production of industrial carbon (World 1994)


Production, Average product Market value,
Carbon Products t/a value, $/t 109 $/a
From petroleum coke
Carbon anodes 8  106 500 4.0
Söderberg anodes 4  106 300 1.2
Graphit electrodes 1  106 3500 3.5
Partly from petroleum coke
Carbon cathodes 5  05 1000 0.5
From other products
Carbon black 6  106 1000 6
Activated carbons 4  105 2500 1
Carbon construction 1  105 500 0.5
materials, etc.
Electrographite powder 8  104 1000 0.08
Carbon fibers 1  104 5  104 0.5
Synthetic diamonds 50 10  106 0.5
Total & 20  106 & 18.0
218 PETROLEUM COKE Vol. 2

Aluminum production is expected to increase about 9 % per year. With a


consumption of about 400 kg regular calcinate per tonne of aluminum the world-
wide regular calcinate demand in the aluminum industry is more than
10  106 t/a.
Regular calcinate in the fraction 0.2 – 1.0 mm is used for titanium dioxide
production.
Needle Coke. Petroleum coke is also used for the production of graphite
electrodes (Table 5). The quantity employed for this purpose has been stagnating
for years. This high-grade quality is called needle coke. Although the proportion
of electrosteel production in overall steel production is steadily increasing, a ser-
ies of optimizing steps have continuously reduced the electrode consumption per
tonne of steel. Whereas in 1985, 7 kg electrode material was needed per tonne of
steel, modern furnaces can manage with less than 2 kg/t. For this reason the
consumption of electrodes worldwide is still around 1  106 t/a.
Needle coke qualities are used as part components for cathode blocks, side-
wall blocks, and ramming pastes for the production of aluminum. A further
application of this quality is its use in steel carburization.
A series of special industrial uses of carbon material, which are negligible in
terms of quantity, make use the thermal and chemical resistance, the thermal
and electrical conductivity, and graphitization of petroleum coke. The amount
required for this purpose is normally covered by carbon dust separated during
the production of carbon anodes and electrodes.

5. Quality Aspects

The quality of coke is chiefly influenced by the feed material used in the petro-
leum refining. Of second and third importance are the coking and the calcining
conditions. Structure, density and VCM contents can vary significantly as can
the sulfur, metals, and ash contents.
Since the mid 1980s an international standardization of testing methods for
carbon materials has been developed. ISO, DIN or ASTM specifications or drafts
exist for nearly all parameters. Standardization has almost been achieved even
in the case of coefficient of thermal expansion measurement (CTE), for a long
time tested by proprietary methods of electrode manufacturers. The situation
that national standards (ASTM, DIN) are incorporated in the international
ISO specifications makes itself positively noticeable.
5.1. Green Coke. The quality of normal green coke mainly used as fuel
is defined by the sulfur content. Additional parameters are the water content (for
the quantity calculation), the ash, and VCM content. In the case of green coke
used for higher grade products the level of metals and the grain size distribution
are further quality criteria.
5.2. Regular Calcinate. The quality of calcined petroleum coke pro-
duced in Europe is superior to that produced in other continents. Due to author-
ity requirements and improvements in efficiency, regular calcinate in Europe
has undergone a quality improvement towards needle coke: the sulfur contents,
reactivity results, and CTE figure are lower, and the vibrated bulk density
is higher. Some non-European manufacturers are already producing similar
Vol. 2 PETROLEUM COKE 219

qualities. Typical European quality characteristics are listed below. The figures
in brackets refer to quality requirements outside Europe (23–26).
Sulfur content, wt % max. 1.7 (3)
Ni/ V content, ppm max. 120 – 150 (200)
Na/ Ca/ Si content, ppm max. 100
Vibrated bulk density (1 – 2mm), g/cm3 0.80 – 0.85 (0.75 – 0.80)
Air reactivity (1 – 1.4 mm), wt %/min max. 0.15
CO2 reactivity (1 – 1.4 mm), wt % max. 12
CTE factor (carbon anodes), 10-6 /8C 3.5 – 4.5 (5.5)
Hydrogen content, wt % 0.05 – 0.09
Real density, g/cm3 2.06 – 2.10
Ash content, wt % 0.05 – 2.20
Crystallite size Lc, nm 2.5 – 3.2
Specific electric resistivity, mO m 460 – 520
Grain stability, wt % min. 70
Dust forming factor, wt % max. 0.02
Water/oil content, wt % max. 0.3
(from antidust additive)

Except for the SO2 emission the sulfur content only has slight side effects on
aluminum electrolysis.
The goal in aluminum electrolysis is to have a small consumption of anodes
with a good electricity yield. Good production results are 95 % electricity yield
with an anode consumption of < 410 kg per tonne aluminum. These figures can
be reached if the anodes meet the specifications for:

1. High carboxy (CO2) reactivity residue


2. High air reactivity residue
3. Low specific electric resistivity
4. Low thermal conductivity

With the capacity increase of aluminum smelters, larger anodes have come
into use. Cracking of the anodes due to thermal shock has been observed occasion-
ally in these cases. Good thermal shock resistance of the anodes is therefore desir-
able. Improvements in thermal shock behavior will occur if CTE, the Young’s
modulus, and the flexural strength are decreased, and the thermal conductivity
is increased.
5.3. Needle Coke. In the production of steel by the arc furnace process
about 1  106 t/a graphite electrodes are needed. Charging stock for the electro-
des are needle coke (about 85 %), and pitch binder for backing and impregnation.
Needle coke is the highest grade product of calcinated petroleum coke. Because of
the massive thermal and mechanical stresses in the arc furnaces, the highest
demands are placed on the electrode properties. Moreover, subsequent modified
graphitization due to the baking of the electrodes result in additional stricter
quality requirements.
Modern lengthwise graphitization is considerably more economic and
only needs a third of the time required by conventional Acheson graphi-
tization, but the more drastic reaction conditions demand superior quali-
ties from the needle coke. For this reason needle coke specifications have
220 PETROLEUM COKE Vol. 2

greatly improved over the last 15 years. Room for further improvement seems
only marginal.
For needle coke the chemical properties (low contents of S, Ni, V, Na, Ca, Si)
and the following physical characteristics must be optimum:

1. High real density and high vibrated bulk density


2. High grain stability and high proportion of coarse grain
3. Low coefficient of thermal expansion

In addition the microcrystalline structure is important, because all physical


parameters depend on it. A high level of anisotropy, regular micropores between
the graphite lattice plane, a high crystallite size, especially after graphitization,
and crystal areas, which contrary to graphite have to be offset against each
other, are indispensable. In order to obtain these needle coke qualities, three
manufacturing areas have to be optimized:

1. Feed selection for delayed coking (low sulfur and low metal residues with
high contents of 3, 4, and 5-ring aromatics and low asphaltene content)
2. Delayed coking conditions (longer cycle, higher recycle ratio, right furnace
coil out temperature ramp)
3. Calcination conditions (heating rate and final calcination temperature)

The quality analysis of needle coke is special because two parameters are specified
that can only be analyzed after graphitization: The CTE and the puffing factor; ie,
the irreversible lengthwise expansion during graphitization (requirement: < 2 %).
The quality comparison has been more complicated up to now by the fact that
the preparation, graphitization, and measurement of CTE were proprietary com-
pany methods. Standardization however is nearly concluded (last step: proposal of
CTE–DIN method, an English translation exists as ISO draft) (27).
The quality requirements of electrodes depend on the electrode diameter
and also on the type of arc furnace (d.c. or a.c. operation). That is why different
types of needle coke are available. Some properties of ‘‘super premium‘‘ and
‘‘intermediate premium‘‘ qualities are listed in Table 6. The CTE values refer
to the measuring range 20 – 200 8C after preparation and graphitization accord-
ing to the DIN method.
There are ongoing tests to produce ‘‘ultra premium‘‘ needle coke with CTE
levels of 0.25  106/ 8C. Production constancy, possible production quantity and
processing characteristics of this quality still need to be evaluated.

Table 6. Typical Range of Needle Coke Types


Super Premium Intermediate
Property needle coke Premium needle coke
CTE (20 – 200 8C), 10-6/8C 0.35 0.6
Sulfur content, wt % 0.2 – 0.4 0.4 – 0.6
Vibrated bulk density, g/cm3 0.83 – 0.90 0.83 – 0.90
V/ Ni content, mg/kg < 10 < 10
Real density, g/cm3 2.13 – 2.15 2.13 – 2.14
H content, wt % 0.03 – 0.05 0.03 – 0.05
Vol. 2 PETROLEUM COKE 221

An additional challenge in the use of electrodes in the arc furnace process is


the design and production of the nipples (screw connections between the indivi-
dual electrodes). Thorough pilot tests, test runs, and a guaranteed production
constancy for needle coke are the basis for delivery agreements.

6. Environmental and Safety Aspects

6.1. Green Coke. Although green coke contains polycyclic aromatics


their total concentration of about 250 mg/kg is below the critical level of 1000
mg/kg. Also the content of the key component benzo[a]pyrene (ca. 25 mg/kg) is
not critical. Green coke is therefore classified as a nontoxic substance (3).
No classification and labeling as regards toxicity, carcinogenicity, flamm-
ability, and environmental effects is required. Contents of Ni, V, Na, Ca, and
Fe originating from the crude oil residues used as feed are accumulated in
green coke. The levels amount to ca. 10 – 500 mg/kg. Small amounts of metals
and anions are introduced with the cutting water, which makes up 5 – 15 % of
green coke (cutting water content: Na, Ca, Cl, SO42). Other heavy metals
(Hg, Cd, Pb, Cr, Cu, Co, As) are not traceable (level < 1 mg/kg). Hydrocarbons
and metals in green coke cannot be elutriated with water (S4 eluate method).
The following classification is employed for the combustion and explosion
behavior [measurement of Deutsche Montan-Technologie (DTM), mining test
circuit, Dortmund]:
Combustion behavior BZ 3 (local combustion or glowing
without spreading)
Glow temperature 300 8C
Ignition temperature 700 8C
Dust explosiveness contents > 60 g/m3 and with ignition
source of > 10 000 J

Green coke is normally stored in open piles with side walls as protection
against the wind. In order to reduce the dust discharge, green coke piles are
sprinkled with water. The potential dust emission is visibly improved by
means of dosing antidust additives (polyacrylate types) to the sprinkling water.
The general explosive boundary value at work of 6 mg/m3 for fine dust con-
centrations is to be complied with.
6.2. Calcined Petroleum Coke. All hydrocarbons in green coke are
eliminated by calcination. The remaining hydrogen content is < 0.1 %. Therefore,
no polycyclic aromatics are traceable. The content of S, N, and metals is virtually
unchanged. With water, only part of the salt content of the cutting water can be
elutriated. No classification and labeling is required.
Due to the dryness of the calcinate, irritation of eyes and throat by the dust
is possible. The general boundary value at work (max. 6 mg/m3 for fine dust con-
centration) is to be complied with.
Calcined coke is usually stored in closed silos or bins to keep the material
dry. To reduce dust formation, the calcinate is sucked out during transport and
loading and the dust held back by filter stations. Sprinkling the calcinate
with residue parts of 0.2 – 0.4 wt % (old process) or an aqueous emulsion of
222 PETROLEUM COKE Vol. 2

dust binder prevents environmental problems from dust. The dust forming fac-
tor (‘‘R+D-Carbon‘‘-method) must be < 0.02 wt %. The combustion and explosive
behavior of calcined coke presents no problem:

Combustion behavior BZ1 (no combustion)


Glow temperature no glowing up to 450 8C
Ignition temperature no ignition up to 850 8C
Dust explosive no

Since no combustible dust – air mixture is formed, calcined dust is a material


that has no security zone classification (zone 11) and does not come under the
risk management regulation.

REFERENCES

1. R. De Biase, J. D. Elliot, ‘‘Coking Process Reflects Trends, Innovations,’’ Oil Gas J.


(1982) April 19, 81.
2. Ch. L. Montell, Carbon Graphite Handbook, Interscience Publishers, New York
1968.
3. CONCAWE, reports nos. 93/105 and 95/95,’’ The Classification and Labelling of
Petroleum Substances according to the EU dangerous Substances directive,’’
Brussels, July 1995.
4. Statistik der Kohlewirtschaft, Essen 1995.
5. PACE, Petroleum Coke Quarterly(1992 and 1996).
6. H. Predel, ‘‘Delayed Coking,’’ Erdöl Kohle Erdgas Petrochem. 43 (1990) 145–146.
7. Conoco Inc., US 4 455 219,1984.
8. Conoco Inc., US 4 518 487,1985.
9. H. Predel, ‘‘Theoretical and Practical Methods for Coke Yield Calculations and
Optimisation,’’ Light Metals 1992, TMS, San Diego, p. 601 – 609.
10. N. Liebermann,Troubleshooting in Process Operations, Pen Well Publishing Comp.,
1985, 16–63.
11. H. Predel, ‘‘New Developments in Delayed Coker Operations,’’ Carbon Conference,
Essen, 22–26.6.1992.
12. A. Stefani, ‘‘Debottleneck Delayed Cokers for greater Profitability,’’ Hydrocarbon
Process. (1996) June, 99–103.
13. Mobil Oil Corporation,US 4 118 281, 1978.
14. J. D. Elliot, R. Klick, ‘‘Trends in Delayed Coker Design,’’ Hydrocarbon Technology
International, Summer 1994, p. 17 – 23.
15. J. D. Elliot, ‘‘Latest Coker Designs Increase Liquid Yields, Reduce Emissions,’’ Oil
Gas J. (1993) Nov. 8.
16. J. D. Elliot, ‘‘Maximize Distillate Liquid Products,’’ Hydrocarbon Process. (1992) Jan.
17. NPRA (National Petroleum Refiners Association), ‘‘Refiners exchange experiences
on FCC problems, coking operations,’’ Oil Gas J. (1994) May, 110–115.
18. M. R. Parrish, D. G. Hammond,V. A. Citarella, ‘‘Fluid Coking: A continuous, flexible
and reliable conversion process,’’ Hydrocarbon Technology International, Spring
1996, p. 25 – 31.
19. Refining 1996, Hydrocarbon Process. 75 (1996) Special Edition, Nov., 100.
20. S. M. Hume, Influence of raw material properties on the reactivity of carbon anodes,
Aluminum-Verlag, Düsseldorf 1993, 180–185, 197–205.
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 223

21. H. Predel, ‘‘Steps for Green Coke Calcination-Mathematic Model and Practical Tests
and Experiences,’’ Light Metals 1996, TMS, p. 499–507.
22. W. K. Fischer, G. Collin, J. W. Stadelhofer (1990), ‘‘Technischer Kohlenstoff—
Herstellung und Anwendung,’’ DGMK-Haupttagung, Sept. 1921, p. 423.
23. H. Blumberg, ‘‘Fibers for Composites Status Quo and Trends,’’ Chemical Fibers
Internat. 47 (1997) Feb., 36–41.
24. C. Jonville, J. C. Thomas,C. Dreyer, ‘‘The Influence of Coke Source on Anode
Performance,’’ JOM 47 (1995) Aug. 23–24.
25. S. M. Hume, W. K. Fischer, R. C. Perruchoud, B. Welch, ‘‘A Model for Petroleum
Coke Reactivity,’’ Light Metals 1993, TMS, p. 525–534.
26. F. Keller, U. Mannweiler, ‘‘Anode Performance: The Link between Coke Properties
and Aluminum Production Cost,’’ PACE, Petroleum Coke Quarterly 1995.
27. M. W. Meier: Cracking Behavior of Anodes, R+D Carbon Ltd., Sierre 1996.
28. J. Mochida, K. Fujimoto, T. Oyama in P. A. Thore (ed.): ‘‘Chemistry in the Production
and Utilization of Needle Coke,’’ Chemistry and Physics of Carbon, vol. 25, Marcel
Dekker, New York – Basel – Hongkong 1997, p. 111–212.

HEINRICH PREDEL
MiRO GmbH & Co. KG, Karlsruhe
Federal Republic of Germany

GASOLINE AND OTHER


MOTOR FUELS
1. Introduction

Gasoline and other motor fuels comprise the largest single use of energy in the Uni-
ted States, and in 2002 accounted for 22% of all energy usage (1). The cost of this
energy has been and is expected to continue to be a primary factor in the national
economy. Moreover, the fraction of resources from which these fuels come that is
provided by foreign sources is a matter of political concern. The fraction of total
crude oil produced domestically shrunk from 75% in 1970 to 39% in 2002 (2) and
is predicted to continue to drop (see NOMENCLATURE IN THE PETROLEUM INDUSTRY).
Imported products also continue to grow. In 2002, the United States imported
130 106 L/day of gasoline and gasoline components. In the 1970s, two Organization
of Petroleum Exporting Countries (OPEC) embargoes resulted in rapid increases
in the price of crude oil and therefore motor fuels. These increases triggered pro-
grams designed to develop alternative sources of fuels such as coal (qv), oil shale
(qv), and natural gas (see NATURAL GAS). In 2001, as a result of lower price volatility
and improved energy efficiencies, the inflation adjusted cost of driving an average
passenger car is about 40% lower than it was in 1960 (3). Today, alternative trans-
portation fuels are being considered as replacements for traditional petroleum
based fuels based on their perceived potential to improve the security of energy
supply and to reduce emissions of greenhouse gases. Among alternative fuels

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.0701191508150308.a01.pub2
224 GASOLINE AND OTHER MOTOR FUELS Vol. 2

under consideration are hydrogen, natural gas and biofuels such as ethanol and
esterified edible oils. (see also AIR POLLUTION; EXHAUST CONTROL, AUTOMOTIVE).
1.1. General Aspects of Manufacture of Motor Fuels. All motor fuel
in the United States is manufactured by private companies. Many of these are
vertically integrated. That is, the same company finds the crude oil or buys it
from a producing government, refines it into finished products, and then sells
these products to independent retailers who specialize in that company’s blended
products or sells them at company operated service stations. There are also a
significant number of companies that participate in only some aspects of the
business cycle such as refining or marketing.
Four groups are involved in the production or use of motor fuels in the
United States: (1) manufacturers of the vehicles; (2) manufacturers and/or mar-
keters of the fuels; (3) purchasers and users of fuels and vehicles; and (4) federal,
state, and local regulatory agencies. Each has a different role.
Vehicle manufacturers must build cars and trucks that operate well on
available fuels. They also specify the fuel requirements of their vehicles. Fuel
marketers must produce fuels that operate in both new and old vehicles. The con-
sumer, the purchaser of the fuel and the vehicle, wants the fuel to be affordable,
readily available, and able to provide a high level of performance. Regulators
check that fuels are labeled properly and meaningfully and that no unwarranted
claims are made. They also regulate emissions from vehicles and set fuel speci-
fications based on environmental considerations.
Appropriate specifications are set when the vehicle and the fuel are viewed
as a system, and when all four groups, or stakeholders, work together to define
cost effective and efficient solutions that satisfy the needs of all stakeholders.
The American Society of Testing and Materials (ASTM), now known
as ASTM International was founded in 1902 to promote just such a process.
ASTM Committee D-2 provides a forum for regulators, vehicle manufacturers,
fuel producers, and consumers to develop and recommend nonbinding standards
for petroleum products. Among the types of ASTM standards are Specifications
and Test methods. Specifications define a precise set of requirements to be satis-
fied by a material. Test methods define procedures for measuring qualities
and characteristics of a material. Specifications for gasoline are contained in
ASTM D 4814 (4) and for automotive diesel fuel in ASTM D 975 (5). These stan-
dards are updated periodically.
ASTM committees must be balanced in that the number of voting produ-
cers must not be greater than the number of voting nonproducers. For petroleum
products, nonproducers are regulators, consumers, and equipment manufac-
turers. Committee chairs must be nonproducers. Although standards must be
approved by a majority, all negative votes must be carefully considered and a
response made. In practice, because all issues are fully discussed and the discus-
sions are based on hard data, very few negative votes are cast when standards
are submitted for final approval. Although ASTM specifications do not have the
force of law, many states have adopted them as state regulation, thus forcing a
minimum quality level in the field.
ASTM specifications cover those aspects of fuel properties that ensure good
performance in gasoline and diesel powered vehicles. Other fuel properties are con-
trolled for environmental reasons. These environmental related specifications are
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 225

set by government agencies on the national (Environmental Protection Agency),


state (eg, California Air Resources Board) and even local levels.
The legal authority to set specifications based on environmental needs
derives from the Clean Air Act, which was last amended in 1990. Fuel composi-
tion regulations may be set to provide direct benefits to the environment (eg,
lower vapor pressure) or to allow emission control systems to operate properly
(eg, lead free gasoline). Environmental regulations are generally developed in
an open public process where all stakeholders have an opportunity to comment
and provide input. Government agencies, which make the final decision, must
show that their decisions protect the public health and welfare, are cost-effective
and fair to all parties.
In June, 1998, a coalition of automotive manufacturers attempted to cir-
cumvent the traditional standard setting mechanism by publishing the World
Wide Fuel Charter (WWFC). Four sets of specifications were defined for gasoline
and diesel, based on four levels of vehicle technology and environmental regula-
tions. Predictably, the specifications reflected the view of the automotive manu-
facturers and no members of the petroleum industry have adopted the WWFC.
Regulatory agencies have not adopted the WWFC either, preferring to work
within existing groups and procedures.

2. Gasoline

Gasoline, the preferred fuel for the Otto engine, evolved from an unwanted by-
product to an indispensable mainstay of modern life in only a few short years. In
1900, gasoline was little more than a disposal problem for the infant petroleum
industry in the production of stove and lamp oils, lubricating oils, and greases.
The first prototype four-stroke, internal combustion, spark-ignited engine was
built by Nicolaus Otto in 1876 in Germany. This engine, which ran on gasoline, pro-
vided enormous advantages over earlier internal combustion designs in terms of
weight, power, and efficiency. By 1890, almost 50,000 of these engines had been
sold. As the popularity of these engines grew at an exponential rate, the demand
for gasoline grew accordingly. Gasoline demand grew by a factor of five between
1907 and 1915.
Figure 1 shows the growth in gasoline demand in the United States since
1950 (6). Demand is largely determined by the growth in the number of cars,
the kilometers of paved roads available for driving, population, and economic
growth. Figure 1 also shows the number of passenger vehicles and the average
yearly distance driven per vehicle (7). Most Americans drive to work, to shop,
and for recreation. In 2002, American consumers burned 503  109L of gasoline.
The average passenger vehicle is driven about 19,000 km/yr, up from about
15,000 km/yr in 1950. Despite stated goals of reducing the number of kilometers
that people drive to work through increased use of mass transit and car-pooling,
the public’s preference for personal driving has remained strong.
A primary factor in moderating the demand for gasoline has been the dra-
matic improvement in automotive fuel economy from the mid-1970s through the
mid-1980s. Figure 2 shows the trend in fuel economy for new cars since 1955
and for light duty trucks from 1980 (8). Improvement is the result of governmental
226 GASOLINE AND OTHER MOTOR FUELS Vol. 2

500 25

Gasoline demand, passenger


400 20

Travel per vehicle


vehicles 300 15

200 10

100 5

0 0
1950 1960 1970 1980 1990 2000
Year

Fig. 1. Growth in annual gasoline demand, in GL/yr; millions of passenger cars; and
travel per vehicle, in km/yr. Passenger vehicles,  106, gasoline demand, GL/yr,
and travel per vehicle, 1000 km/yr.

regulations that were a direct outcome of the oil shortages of the 1970s.
Average fuel economy has remained virtually constant since the mid-1980s, as
no new regulations have been imposed. This clearly shows that the current cost
structure of gasoline is not a significant impediment to people’s desire to drive,
and that, in general, the consuming pubic would rather have larger, more powerful
family vehicles than smaller more efficient ones.
Energy outlooks are developed annually by the Department of Energy. In
their most recent forecast for the years 2002–2025, energy demand for light
duty vehicles is projected to grow about 1.9% per year between 2002 and 2025,

14.0

12.0
Fuel economy, km/L

10.0

8.0

6.0

4.0

2.0

0.0
1955 1965 1975 1985 1995 2005
Year

Fig. 2. Fleet average fuel economy for new cars and light duty trucks. Passenger cars
light trucks.
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 227

1.9% per year for light commercial trucks and 2.2% per year for freight trucks (9).
These projections include growth in travel and modest improvements in fuel
economy. The recent trend of using technology to improve performance rather
than efficiency will continue; average horsepower is projected to grow by 24%
while fuel economy will improve by only 6%. It is expected that by 2025, 2.1%
of light duty vehicles will be fueled by alternative fuels. It should be recognized
that these projections are sensitive to many factors, including overall economic
growth and fuel costs.
2.1. Requirements of Good Gasoline. To satisfy high performance
automotive engines, gasoline must meet exacting specifications, some of which
are varied according to location and based on temperatures or altitudes. The fuel
must evaporate easily and burn completely when the spark plug fires in each
cylinder. Early detonation of the fuel in the cylinder can cause destructive engine
knock. The fuel must be chemically stable. It should not form gums or other poly-
meric deposit precursors. There should be no particulate contaminants or entra-
ined water. Contaminants must be prevented from the point of manufacture in
the refinery all the way through the distribution system until the fuel is metered
from the vehicle tank into the engine.
Octane. Octane is probably the single most recognized measure of gaso-
line quality. The octane value of gasolines are posted on service station dispen-
sers, and most drivers recognize a fuel in which the octane is too low. Broadly
speaking, octane is a measure of the combustion characteristics of gasoline. In
an Otto cycle engine, gasoline vapor in the combustion chamber must start to
burn only when the spark plug ignites after the charge has been compressed.
Then it must burn with a well-defined flame front that travels all the way across
the combustion chamber. Low octane gasoline has a tendency to preignite. As the
flame front sweeps across the combustion chamber, the unburned portion of the
fuel (end gas) heats up under rapidly rising temperature and pressure condi-
tions. The fuel–air mixture undergoes chemical reactions that may cause it to
autoignite and detonate the entire remaining mixture. Instead of being pushed
down smoothly on the power stroke, the piston is given a hard instantaneous rap
to which it cannot respond because of the large mechanical inertia present in the
crankshaft and other pistons. This rapid energy release causes pressure fluctua-
tions in the cylinder which result in a loud metallic noise commonly called knock.
In addition to producing an objectionable sound, knock can reduce the amount of
useful work that is extracted from the engine. Power is dissipated in the pressure
waves, and increased heat is radiated to the cylinder walls and into the cooling
water. Under extreme conditions of prolonged knock, overheating and even
engine damage can occur. The damage is typically caused by catastrophic melt-
ing of piston crowns or head gaskets.
Knock is caused by unwanted chemical reactions in the combustion cham-
ber. These reactions are a function of the specific chemical species which make
up the fuel and the environmental conditions to which the fuel is subjected
during the compression and power stroke in the engine. Therefore, both the
chemical makeup of the fuel and the engine design parameters must be consi-
dered when trying to understand knock.
Chemical Factors. Because knock is caused by chemical reactions in the
engine, it is reasonable to assume that chemical structure plays an important
228 GASOLINE AND OTHER MOTOR FUELS Vol. 2

Table 1. Chemical Factors Affecting Knock


Change in knocking
tendency

Change in chemical structure Increase Decrease


longer paraffin chains x
isomerizing normal paraffins x
aromatizing normal paraffins x
alkylating aromatics x
saturating aromatic rings x

role in determining the resistance of a particular compound to knock. Reactions


that produce knock are generally free-radical chain-type reactions which are
different from those that occur in the body of the flame; the former occur at
lower temperatures and are called cool flame reactions. Peroxy and hydroperoxy
radicals play important roles in the knock process. A number of good reviews
have discussed the details of the chemical mechanisms (10). Ignition delay
(tau) has also been used for description of the chemical tendency to knock (11).
The chemical factors affecting knock are shown in Table 1.
A lengthy and systematic investigation into the cause and prevention of
knock, culminated in the use of lead alkyl compounds such as tetraethyllead
[78-00-2] (TEL) to decrease the knocking tendency of fuels dramatically. Lead
alkyls function as antiknocks by decomposing in the combustion process to
form lead oxides which terminate the free-radical chain mechanisms leading to
knock. The problem with lead additives is the presence of lead in the exhaust and
the consequent potential health effects on the population, especially children,
living near roadways.
Lead also poisons the noble metal catalysts used on all gasoline powered
motor vehicles sold in the United States today. In 1974, gasoline marketers
started to sell unleaded gasoline and lead was totally banned by EPA in 1995
except for use in aviation gasoline and racing fuels. Today, lead is used widely
only in certain countries in Africa and plans are in place to stop even that
small use.
In 1957, Ethyl Corp. announced a new antiknock compound, methylcyclo-
pentadienylmanganese tricarbonyl [12108-13-3] (MMT, trademark of Ethyl
Corp.). MMT is almost as effective as lead on a per gram of metal basis, but
because manganese was more expensive than lead, MMT was not widely used
until limits were placed on the lead content of gasoline. The Clean Air Amend-
ments of 1990 banned the use of MMT in reformulated gasoline (gasoline blended
to reduce emissions). Automotive manufacturers generally oppose the use of
MMT in new vehicles because they claim that the additive increases emissions
by affecting catalysts, control systems and spark plugs (12). Although EPA
allows MMT use in conventional gasoline, little or none is used in the United
States today. It has been widely used in Canada although its use there is also
being challenged by automotive manufacturers.
Vehicle Factors. Because knock is a chemical reaction, it is sensitive to
temperature and reaction time. Temperature can, in turn, be affected either by
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 229

external factors such as the wall temperature or by the amount of heat released
in the combustion process itself, which is directly related to the density of the
fuel–air mixture. A vehicle factor which increases charge density, combustion
chamber temperatures, or available reaction time promotes the tendency to
knock. Engine operating and design factors which affect the tendency to produce
knocking are

Increased Knock Decreased Knock


Higher compression ratio Increased turbulence
Advanced spark schedule Exhaust gas recycle
Higher coolant temperature Cooled air charge
Turbocharging or supercharging High altitudes
Combustion chamber deposits High humidity
Engine loading

Generally, anything which increases the efficiency or power output of the


engine increases the tendency to knock. Higher compression ratio increases
the initial temperature of the fuel–air charge and therefore the peak combus-
tion temperatures. Combustion chamber deposits are highly insulating and
reduce the amount of heat rejected to the coolant, thereby increasing combustion
chamber temperatures. Advancing the time in the cycle at which the spark
plug fires provides more time for the end gas to undergo preknock reactions.
One of the easiest fixes for a knocking engine is to retard the spark timing. The
engine loses a little power but the knocking can almost always be eliminated.
In a similar manner, designing the combustion chamber to promote shorter burn
times can effectively reduce knocking tendencies. Putting the spark plug near
the center of the combustion chamber is one way. Another is to promote turbu-
lence during burning thus increasing flame speed.
Ambient conditions can also affect knocking tendency. At high altitudes, air
density is lower and the combustion of the fuel air charge generates less heat.
High humidity dilutes the oxygen with water vapor and also interferes with
some of the knock reactions leading to lower knocking tendencies.
Measuring Octane. Two different values need to be considered when dis-
cussing octane measurements. One is the knocking tendency of the fuel, called
the fuel octane number. The other is the knocking tendency of the vehicle, called
octane number requirement.
It is impractical to measure a fuel’s octane number in a full-sized vehicle.
Thus laboratory techniques have been developed through ASTM to measure
these numbers in well-defined engines. The standard knock engine is a single
cylinder cooperative fuels research (CFR) engine. This engine has a compression
ratio which can be varied between 3 and 30. It also has four carburetor bowls
so that reference and test fuels can be switched easily. It is equipped with a
magneto-restrictive type detonation sensor that measures the rapid change in
pressure characteristic of knock. Two primary reference fuels (PRFs) have
been defined: 2,2,4-trimethylpentane [540-84-1], commonly referred to as iso-
octane, has an octane number of 100 and n-heptane [142-82-5] has an octane
230 GASOLINE AND OTHER MOTOR FUELS Vol. 2

Table 2. Octane Test Operating Conditions


Research octane no. Motor octane no.
Condition (ASTM D 2699) (ASTM D 2700)
engine speed, RPM 600 900
inlet air temperature, 8C 51.7 38
mixture temperature, 8C 149
spark advance, 8BTDCa 13 14–26
a
Spark advance for Motor method is a function of compression ratio.
BTDC ¼ before top dead center.

number of zero. Octane numbers of primary reference fuels between zero and 100
are formed by volumetric mixtures of isooctane and n-heptane. For example, a
50:50 mixture of the two has an octane number of 50. Primary reference fuels
for measuring octane numbers above 100 contain specified amounts of tetra-
ethyllead. For instance, isooctane containing 0.238 g TEL/L has a defined octane
number of 108.
The octane value of an unknown fuel sample is determined by comparing its
knocking tendency to various primary reference fuels. Its measured octane is
equal to the octane of the PRF which has the same knocking intensity. Knock
intensity is controlled to an average value by varying the compression ratio of
the CFR engine. In practice, the exact value of a fuel’s octane number is deter-
mined to the nearest 0.1 octane number by interpolation from two PRFs that are
no more than two octane numbers apart.
The CFR engine is operated at two conditions to simulate typical on-road
driving conditions. The less severe condition measures research octane number
(RON); the more severe one measures motor octane number (MON). Table 2
summarizes the operating conditions for each test.
The difference between RON and MON for a particular fuel is called the
sensitivity. By definition, the RON and MON of the primary reference fuels are
the same and the sensitivity is zero. For all other fuels, the sensitivity is almost
always greater than zero. Generally, paraffins have low sensitivities whereas
olefins and aromatics have sensitivities ranging up to 10 and higher.
The octane numbers of many pure compounds have been measured and
reported in the literature. Probably the most comprehensive project was carried
out under the auspices of the American Petroleum Institute (13). Table 3 lists
RON and MON values for a number of representative compounds. Some aromatic
compounds cannot be tested neat in the knock engine, so these are evaluated at
levels of 20%, and the equivalent octane number is calculated. The values for
oxygenates in Table 2 have been reported elsewhere (14).
These two test methods and the octane numbers of the fuels measured
are ultimately used to evaluate the performance of vehicles. The two methods
provide useful information about how fuels perform in cars. The RON corre-
sponds to light load, low speed conditions, whereas the MON corresponds to
heavier loads, and high speed severe driving conditions.
The octane number requirement (ONR) of a car is the octane number which
causes barely audible, ie, trace knock when driven by a trained rater. The Coor-
dinating Research Council (CRC), a research organization funded jointly by the
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 231

Table 3. Octane Numbers of Pure Compounds


Research Motor
Compound O.N. O.N.
paraffins
n-pentane 61.8 63.2
n-hexane 24.8 26.0
n-heptane 0 0
2-methylbutane 92.3 90.3
2-methylpentane 73.4 73.5
2-methylhexane 42.4 46.4
olefins
2-methyl2-butene 97.3 85.5
1-pentene 90.9 77.1
1-hexene 76.4 63.4
1-heptene 54.5 50.7
aromaticsa
toluene 124 112
o-xylene 120 102
ethylbenzene 124 107
propylbenzene 127 129
oxygenates
ethanol 130 96
methyl tert-butyl ether (MTBE) 118 100
ethyl tert-butyl ether (ETBE) 118 102
methyl tert-amyl ether (TAME) 111 98
a
Aromatics were measured in a 20% blend with a 60/40
mixture of 2,2,4-trimethylpentane and n-heptane.

American Petroleum Institute (API) and six members of the automotive industry
(General Motors, Ford, Daimler-Chrysler, Toyota, Volkswagon and Honda), has
defined test procedures for measuring ONR. Each car is driven under a set of
light and heavy accelerations until the most sensitive driving mode is deter-
mined. Then a series of fuels is run in the car until trace knock is determined.
Auto manufacturers have developed knock detecting and limiting devices
called knock sensors which are installed in most engines today. The sensor is a
piezoelectric crystal which is tuned to the natural frequency of engine knock
sound (see SENSORS). When knock is detected, a signal is sent to the on-board
computer which retards the spark timing. In this way, knocking noise is elimi-
nated until the maximum spark retard is reached. The tradeoff, however, is
reduced power. Full throttle acceleration times can be increased by up to 10%
when these cars use low octane fuel. Measuring octane number requirements
in cars with knock sensors involves either determining when octane is sensed
and spark retard is initiated, or sensing when the maximum level of retard
has been reached. CRC has defined a test procedure for measuring the octane
response of cars equipped with knock sensors (15).
It has long been known that the design of a car affects the way in which it
responds to RON and MON. The response of a car to RON and MON can be char-
acterized by a parameter called severity, which is a measure of its relative res-
ponse to RON and MON. In the 1930s, most cars responded only to RON. That is,
it did not really matter what the MON of a fuel was. As long as the RON was
232 GASOLINE AND OTHER MOTOR FUELS Vol. 2

above a minimum value, the car would not knock. Cars of the 1990s generally
have a severity of about 1, meaning that a 1 number change in RON can be offset
by a corresponding and opposite change in MON to provide equivalent knocking
protection. Octane numbers which are posted on gasoline dispensers in all states
show the antiknock index which is the average of the RON and MON values of
the gasoline being sold. Recent data suggest that modern cars respond mostly to
RON and that consideration should be given to changing the way that octane is
posted on gasoline dispensers (16,17).
In designing engines, automotive manufacturers must decide what kind
of fuel that engine should use, from low to high octane. Compression ratio is
the primary determining factor in a vehicle’s octane requirements. High perfor-
mance, high compression ratio engines need high octane or premium fuel. It is
also known that the octane requirement of a vehicle increases as a vehicle ages.
This phenomena, known as octane requirement increase (ORI), is caused by the
buildup of fuel and lubricant based deposits in the combustion chamber. These
deposits are highly insulating and increase the combustion wall temperature,
thereby increasing the tendency to knock. Generally, ORI reaches an equili-
brium value of 6–9 octane numbers over the first 24–32,000 kilometers of opera-
tion, although higher increases are not uncommon. In some cases, as cars age
and start to burn more oil, octane requirements can increase again.
Over the years, engine compression ratios have undergone a number of
changes. These increased rapidly through the 1950s and 1960s. The advent of
unleaded gasoline and the demise of lead as a cheap source of octane, brought
about a drop in fuel octane numbers and compression ratios. In the early
1980s, both compression ratios and average octane numbers requirements
increased as a result of customer demand for more powerful cars.
The particular fuel octane number which a gasoline marketer chooses to
produce and sell is determined by a number of factors including the cost of
producing the various levels of octane, the number of customers requiring the
various octane levels, and competitive offerings. Determining customer require-
ments is the most difficult. The surveys conducted by CRC measure vehicle ONR
using trained raters. Trained raters are much more sensitive to knocking noise
than the average customer. Consequently, octane requirements percieved by
customers are generally lower than trained raters. This customer/rater delta
has been the subject of a number of studies (18,19) and the average difference
has been estimated to be between 3–5 octane numbers. Typically, a marketer de-
cides that the premium grade should satisfy a certain fraction of customers,
usually well over 90%. The octane values for the regular and/or intermediate
grades are determined in a similar manner.
The optimum level of gasoline octane is a complex function of a number of
factors involving the customer as well as the automotive and petroleum indus-
tries. Increasing octane values may require significant investment and higher
operating costs on the part of the oil industry that could raise the price of gaso-
line to the consumer. On the other hand, vehicle manufacturers could take
advantage of the availability of higher octane gasolines by raising the compres-
sion ratios of their engines which could in turn increase fuel economy due to
higher thermal efficiency. Higher compression ratio engines may cost more to
build, but the consumer would save in fuel costs over the life of the car.
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 233

Volatility. The properties of a gasoline which control its ability to evapo-


rate are critical to good operation of a vehicle. In an Otto cycle engine, the fuel
must be in the vapor state for combustion to take place. The volatility or vapor-
ization characteristics of a gasoline are defined by three ASTM tests: Reid vapor
pressure (RVP), the distillation curve, and the vapor/liquid ratio (V/L) at a given
temperature, ASTM D 4953, D 86, and D 2533, respectively. These three tests
are sufficient to describe the volatility related characteristics of gasoline and
predict its driveability performance. Moreover, these are the specifications that
industry and government use to define volatility characteristics. Proper choice of
these parameters ensures that cars operate well under all ambient conditions.
RVP is a vapor pressure measurement at a fixed air/liquid ratio of 4 and a
temperature of 388C. ASTM has defined a number of tests which can be used to
measure RVP: D 4953, D 5190, D 5191 and D 5482.
The distillation (qv) test, D 86, is a batch distillation using a specified
heating rate. A 100-mL sample is distilled and the vapor temperatures at which
various percentages of the sample have condensed are recorded. The data are
reported as the temperature at which 10, 20, 30%, etc, of the sample has been
evaporated. The amount of heavy material which does not evaporate is measured
at the end of the test and the amount of light material which was not condensed
is calculated by difference from the total condensate and the residue. The D 86
test approximates a one-plate distillation with limited recycle. It has been com-
pared to the kind of distillation the gasoline undergoes in a carburetor and intake
manifold. For the more detailed description of the volatility characteristics need-
ed to design and run a refinery there are other tools which can be used to generate
true boiling point (TBP) curves. TBP are essentially distillation curves run at
high efficiency and reflux. One method uses a glass distillation column with
15 theoretical plates and a reflux ratio of 5. Another uses a gas chromatography
technique to identify most or all of the individual compounds in the sample and
then uses a computer-based chemical library to mathematically construct a
distillation curve using known or estimated vapor pressures, fugacities, etc.
Distillation data may be expressed in two ways: the percent evaporated at a
given temperature (Exxx); or the temperature for a given percent evaporation
(Tyy). Because Exxx values blend linearly, these are generally preferred by
refiners and blenders. Gasoline performance specifications have been reported
in both ways. ASTM specifications generally prefer the Tyy format.
The vapor/liquid ratio tests measure the amount of vapor formed from
a given volume of liquid at a given temperature at atmospheric pressure. A
common measure used in specifying gasoline is the temperature at which the
vapor/liquid ratio is 20 (TV/L ¼ 20). Although V/L can be measured experimen-
tally, it is a difficult and time consuming test to carry out, and techniques
have been developed to calculate it from RVP and D 86 values.
When designing fuel volatility targets, gasoline blenders must strike a
balance between various driveability performance characteristics. Driveability
refers to the ability of a car to start easily, accelerate and idle smoothly, and res-
pond to changes in throttle position as expected. Too much volatility can cause
as many problems as too little volatility. Targets must be matched to local ambi-
ent temperature conditions. As for other performance features, a good match
between vehicle and fuel design is important to proper performance. Volatility
234 GASOLINE AND OTHER MOTOR FUELS Vol. 2

requirements, like octane, are a strong function of vehicle design. A great deal of
data on the volatility requirements of new cars is collected in cooperative pro-
grams run by CRC. Trained raters drive cars under strictly controlled conditions
and evaluate the driveability performance on a range of fuels. Customers are
generally less sensitive to problems than the raters. Each company decides which
level of protection to provide. ASTM also uses these data to set target levels
for the various fuel parameters.
Startability. In order to achieve combustion in an Otto cycle engine, the
air/fuel ratio in the combustion chamber must be near the stoichiometric ratio.
Unfortunately, when the engine is first started, the walls of the combustion
chamber and the intake manifold are not hot enough to vaporize much fuel.
Therefore, the vehicle is designed to meter extra fuel and less air to the engine
upon start up so that there is adequate vapor in the engine to support combus-
tion. From the fuel’s perspective, a fuel which is easily vaporized and which con-
tains a large percentage of light compounds such as butane is desirable to achieve
good starting. The ability of a fuel to achieve good starting can be correlated with
RVP and a measure of the front end of the distillation curve, either E70 or T10.
Usually, minimum levels of RVP such as 60 kPa (0.6 bar) and E70 minimum of
10% (or a T10 maximum of 608C) are satisfactory for good startability in most
winter locations (20). Obviously, fuel specifications change with ambient tem-
peratures. At higher temperatures, lower RVP and front-end volatilities are
adequate to provide good starting characteristics.
Vapor Lock. At the other end of the spectrum from starting is vapor lock,
a problem of too much volatility. Vapor lock occurs when too much of the fuel
evaporates and either starves the engine for fuel or provides too much fuel to
the engine. It occurs on days that are warmer than usual and when the car
has reached full operating temperatures. A typical situation would be when a
car is stuck in heavy traffic; the engine reaches high temperatures and the
fuel reaches its boiling point. Most cars today are fuel injected, have fuel
pumps located in the tank and fuel lines which are under positive pressure.
They are therefore less prone to vapor lock than older carbureted vehicles.
Vehicle manufacturers minimize vapor lock by keeping the fuel system
cool and under positive pressure. Fuel manufacturers minimize the problems
by seasonal volatility blending. As might be expected, the front end of the gaso-
line controls vapor lock. The temperature to obtain a V/L ratio of 20 is referred
to in ASTM D 4814 as a way to control the vapor locking tendencies of gasoline.
Warm-Up. Warm-up refers to that period of operation beginning imme-
diately after the car has started and continuing until the engine has reached
normal operating temperatures, usually after 10 minutes or so of operation.
During this period, the vehicle designer wants to get the vehicle equivalence
ratio to stoichiometric as soon as possible to minimize emissions. On the other
hand, if the mixture is leaned out too soon, the car experiences poor driveability
during the warm-up period. Fuel system design is critical during this period.
With older, single point fuel metering systems such as carburetors and throttle
body injectors, there is generally liquid fuel on the walls of the intake manifold
during warm-up. If this liquid only reaches the cylinders in bursts the vehicle
may experience unwanted surges. Multipoint fuel injectors minimize many of
these problems but makes the scheduling of the fuel pulses critical.
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 235

Manufacturers use powerful on-board computers to implement sophisticated and


complex algorithms to control operation of fuel injectors during this critical period.
From the fuel’s perspective, the middle of the distillation curve plays the
largest role in achieving good warm-up performance. Under the typical vehicle
operating regime, the front end of the fuel totally evaporates. The back end of
the fuel, containing the heaviest molecules, has trouble evaporating at this point.
The molecules boiling between about 100 and 1508C are the most important.
Here again, the actual levels required are a strong function of ambient tempe-
rature. The critical periods are usually the Spring and Fall, when temperatures
are between 0 and 158C. At very cold temperatures, the RVP and front-end
volatility are so high that there is adequate fuel vaporization. Also, because
cold weather emission control has not been a principal concern, vehicles can be
calibrated to maintain rich air/fuel ratios longer than in warmer weather. Simi-
larly, in hot weather, the engine heats up faster and maldistribution is not as
much of a problem.
The most common expression for controlling driveability is known as the
driveability index (DI), which has the form:

DI ¼ 1:5 T10 þ 3 T50 þ T90

ASTM has adopted DI as one of the standard requirements for good quality
gasoline. DI should be less than 5978C (1250 8F) during warm weather and less
than 5698C (12008F) in cold weather. It has been shown that gasolines blended
with ethanol do not follow this relationship. For fuels having high values of DI,
the presence of ethanol degrades performance relative to non-oxygenated fuels,
even at equal levels of T10, T50 and T90 and an offset is required (21). The exact
magnitude of the offset and the form of the equation have not been resolved, and
ASTM still recommends using the equation shown above.
Icing. At temperatures within 58C of freezing and under conditions of
high humidity, ice can form in the intake system of vehicles with carburetors
or throttle body fuel injectors. When the gasoline in these systems evaporates,
it cools the incoming air because of the latent heat of vaporization. If the air is
near its saturation point, ice crystals can form on the throttle plate, on the choke
plate, or in the venturi throat. The ice can cause rough idle if it accumulates
around the edges of the throttle plate when it is nearly closed. In the extreme,
ice can clog the carburetor jets and stall the car completely. After the car is fully
warmed-up there is generally enough heat in the intake system to prevent any
ice buildup. The universal adoption of port fuel injection has eliminated this
concern from today’s cars.
Back-End Volatility. The portion of the gasoline that boils above 1508C
is referred to as the back end. Molecules in this region have high energy density
and provide a significant contribution to fuel economy. Too much material
in this boiling range, however, can cause problems. This material is hard to
volatilize and when the engine is cold, tends to accumulate on the walls of
the cylinder. From there it can be washed into the oil sump and dilute the
oil. Generally, as the engine heats up this material evaporates. However, if
there are too many back ends in the gasoline, then not all may boil off and
236 GASOLINE AND OTHER MOTOR FUELS Vol. 2

Table 4. ASTM Vapor Pressure and Distillation Requirements


Vapor pressure, T10, T90, max, End point, Driveability
Class max, kPa max, 8C T50, 8C 8C max, 8C index, max, 8C
AA 54 70 77–121 190 225 597
A 62 70 77–121 190 225 597
B 69 65 77–118 190 225 591
C 79 60 77–116 185 225 586
D 93 55 66–113 185 225 580
E 103 50 66–110 185 225 569

the performance of the lubricant may be degraded. Very heavy molecules, such
as those having more than 12 carbon atoms, may contribute to combustion
chamber deposits. Condensed ring aromatics are particularly strong contribu-
tors to these deposits (22).
Vehicle volatility requirements are a strong function of ambient tempe-
ratures. As described above, at any given ambient temperature, volatility
should be within a certain range. Too high and vehicles may experience
vapor lock. Too low and vehicles may experience hard starting and poor drive-
ability. ASTM has defined six volatility classes for good driveability and six
classes for vapor lock protection. These specifications are shown in Tables 4
and 5 respectively.
Each month, each state is assigned a class from Table 4 depending on the
low temperatures experienced, and a class from Table 5 depending on historical
high temperatures. Some states, because of widely varying ambient conditions
may contain two or more areas. Together, these two ratings define the volatility
range for good gasolines over the course of a year.
Cleanliness. Good gasoline must be both chemically and physically clean.
Chemical cleanliness means that it does not contain nor react under conditions
of storage and use to form unwanted by-products such as gums, sludge, and
deposits. Chemical cleanliness is assured by controlling the hydrocarbon compo-
sition and by appropriate additives. Physical cleanliness means that there are
no undissolved solids or large amounts of free water in the gasoline.
After production in the refinery, gasoline may undergo two types of oxi-
dative degradation. The first occurs under ambient temperature conditions and

Table 5. ASTM Vapor Lock Protection Requirements


Vapor liquid ratio

Vapor lock Test temperature, V/L,


protection class 8C max
1 60 20
2 56 20
3 51 20
4 47 20
5 41 20
6 35 20
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 237

long periods of time, as long as six months, and is reflected in storage stability.
The second has an impact on high temperature stability and is a phenomenon
of the high temperatures and short residence times encountered as the fuel
makes its way through the vehicle intake system. Both mechanisms share some
similar chemical features in that they involve oxidation of fuel compounds and
free-radical polymerization to form high molecular weight gums. The gums and
gum precursors formed under storage conditions can lead to damaging deposits
in fuel injectors or intake valves.
Compounds associated with poor stability are olefins in general and conju-
gated diolefins in particular. A compound such as 1,3-cyclopentadiene is known
to be particularly detrimental. Compounds having poor stability can be produced
in cracking processes such as thermal cracking, including visbreaking and
steam cracking, and catalytic cracking. The presence of compounds such
as sulfur and nitrogen can also degrade stability and promote gum formation.
Vehicle manufacturers know that copper promotes oxidation of the fuel and are
careful to minimize the use of copper in the fuel system materials.
A number of laboratory tests are used to predict chemical stability. The
amount of existent gum in a gasoline is determined by ASTM D 381. This method
involves evaporating a sample by a jet of heated air. The residue is weighed,
solubles are extracted with n-heptane, and the sample is reweighed. The total
is called unwashed gum and the insoluble portion is called solvent-washed
gum. ASTM D 4814 specifies that gasoline should contain less than 5 mg/100 mL
of solvent-washed gum. Most gasolines contain less than 3 mg/100 mL.
Although ASTM D 381 measures the gum existing in gasoline at a parti-
cular point in time, it does not indicate how much more might be formed during
storage at the refinery, during various modes of transportation, in service sta-
tion tanks, or in vehicle tanks. The oxidative stability test (ASTM D 525) was
developed to provide a rapid means of predicting future gum formation. This
method consists of placing a sample of the gasoline in a bomb, pressurizing it
to 690 kPa (100 psi) with oxygen, and maintaining it at a temperature of
1008C. The pressure inside the bomb is monitored continuously. The oxygen
reacts with the gasoline slowly at first, but eventually the pressure drops sharply,
14 kPa (2 psi) or more within a 15 minute interval, indicating a breakpoint. The
time from the start of the test to the breakpoint is the induction period and is a
measure of the oxidative stability of the gasoline. ASTM D 4814 specifies that
gasoline induction period must exceed 240 minutes, although most good gaso-
lines have induction periods in excess of 960 minutes, the duration of the test.
Another ASTM test method, Potential Gum (D 873), combines the existent
gum and the oxidation stability tests to measure potential gum. A sample of
gasoline is subjected to the oxidation stability test for 960 min, filtered to remove
particulates, and then subjected to an existent gum test. The potential gum is
expressed as the total (unwashed) gum in this test.
Other tests to predict stability of gasoline have been developed and reported
in the literature. One, developed by the U.S. military, stores gasoline at elevated
(438C) temperatures for up to 12 weeks and measures existent gum at the end of
that period (23). Another measures existent gum in the presence of copper. The
copper catalyzes oxidation and may be a better estimator of the stability of
gasoline at high temperature/low residence time conditions.
238 GASOLINE AND OTHER MOTOR FUELS Vol. 2

Laboratory simulation tests have been developed which attempt to provide


a better simulation of engine conditions than the analytical tests for predicting
high temperature stability and intake system deposits, especially for gasolines
containing additives. A test rig has been described (24,25) which could be used
in a laboratory to predict the deposits formed in a carburetor and on intake
valves. Such a test could be used for certification of fuels or for comparing
the potency of competing additives. Nevertheless, these tests have not proven to
be accurate measures of the deposit forming tendencies of fuels, and vehicle
tests must always be carried out.
In summary, a number of tests are used to predict the chemical stability of
gasolines. Although storage stability may be reasonably well predicted by labora-
tory tests, no test has been developed to predict deposit forming tendencies. This
may be a reflection of the various regimes that the fuel undergoes between pro-
duction and combustion. Even in the intake system itself, the time/temperature
history of a fuel in a fuel injector is much different than it is on an intake valve.
There is no a priori reason why one test should be expected to correlate with all
of these cases. For this reason the best tests are not short laboratory tests but
long expensive ones such as long-term storage, and engine/vehicle-based deposit
tests for fuel injector and intake valve deposit formation. Much more funda-
mental work on the mechanisms of deposit formation must be carried out before
short tests can be expected to be successful.
Other Requirements. There are a number of other specific features that
must be present (or absent) in gasoline. Some of these are achieved by blending
or manufacturing; others are achieved through the use of additives. Sulfur
should not be present as foul smelling mercaptans. The gasoline should not pro-
mote rust in pipeline, service station tanks, or vehicle metals. It should be clear
and bright in appearance and should not pick up water either as a haze or as
an emulsion. Free water in gasoline contributes to rust and corrosion. Gasoline
should not contain more than trace amounts of carbonyls which can dissolve
elastomeric seals and diaphragms, nor should it contain phosphorus bearing
additives which can damage sensitive catalytic converters.
2.2. Manufacture. Crude oil can be easily separated into its principal
products, ie, gasoline, distillate fuels, and residual fuels, by simple distillation.
However, neither the amounts nor quality of these natural products matches
demand. The refining industry has devoted considerable research and engineer-
ing effort as well as financial resources to convert naturally occurring molecules
into acceptable fuels. Industry’s main challenge has been to devise new ways to
meet the tremendous demand for gasoline without, at the same time, overprodu-
cing other petroleum products.
Distillation. Petroleum refining begins with the distillation of crude oil
into a number of different fractions. In many cases, two distillations are carried
out in units called pipestills. The first is at atmospheric pressure and tempera-
tures up to 4008C. The second distillation is carried out at reduced pressure and
fractionates the heaviest material from the atmospheric pipestill. Table 6 shows
typical boiling ranges for the various crude oil fractions and typical yields from
Arab Light, a common crude oil. The yield and properties of these straight-run
or virgin fractions are a function of the type of crude oil being distilled. Virgin
naphtha can be used as gasoline, except that its octane value is very low
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 239

Table 6. Typical Properties of Crude Oil Fractions


Boiling Yield from crude, %
Fraction range, 8C
gas <0 <1
light virgin naphtha 0–100
heavy virgin naphtha 100–200 18 (L þ H)
gas oil/kerosene 200–400 33
residue >400 48

(78 RON/75 MON). Whereas the potential yield of gasoline directly from crude oil
is less than 20%, the demand is about 50%. The heavy material must be convert-
ed to lighter material and the octane of many of the existing streams improved by
changing chemical composition.
Catalytic Cracking. Although it has long been known that heating crude
oil fractions could break or crack the compounds into smaller molecules, the
development of suitable catalysts and processing designs has made catalytic
cracking the premier refinery process for changing the molecular structure of
the crude (see CATALYSIS). Catalytic cracking generates higher yields than ther-
mal cracking as well as superior quality products. Over 50% of the gasoline in
the United States is obtained by catalytic cracking which uses a fluidized bed
of powdered or small diameter catalysts that are continuously regenerated
in an adjacent vessel called a regenerator (see FLUID CATALYTIC CRACKING (FCC)
UNITS, REGENERATION). The fluidized-bed catalytic cracking (FCC) process was
first commercialized in 1942 by Standard Oil Co. (New Jersey) at its Baton
Rouge refinery (26) using powdered silica/alumina as a catalyst. Fluidized beds
offered the ability to use powdered, high surface area catalysts, to operate con-
tinuously, to regenerate easily, and to use short contact times which increased
yield and selectivity. Since 1942, many improvements have been made in process
design, catalyst formulation, and the ability to handle heavier feeds (27–29).
Today’s catalysts are zeolites having highly controlled pore size and surface
area.
The principal class of reactions in the FCC process converts high boiling,
low octane normal paraffins to lower boiling, higher octane olefins, naphthenes
(cycloparaffins), and aromatics. FCC naphtha is almost always fractionated into
two or three streams. Typical properties are shown in Table 7. Properties of

Table 7. Typical Properties of FCC Naphtha


Light FCC Intermediate Heavy FCC
Property naphtha FCC naphtha naphtha
boiling range, 8C <105 105–160 160–220
RON 91 88 91
MON 79 77 81
aromatics, vol% 10 35 65
olefins, vol% 60 20 15
240 GASOLINE AND OTHER MOTOR FUELS Vol. 2

specific streams depend on the catalyst, design and operating conditions of the
unit, and the crude properties.
Two undesirable aspects of FCC naphtha quality are that it may contain
unacceptably high amounts of foul smelling mercaptans, and that its thermal
stability may be too low. Mercaptans are usually found in the light FCC naphtha
and may be removed or converted to sulfides and disulfides by a sweetening pro-
cess such as Merox, developed by UOP. Thermal stability is improved in sweet-
ening processes through removal of cresylic and naphthenic acids. It may be
further improved by clay treating and by addition of oxidation inhibitors such
as phenylene diamine.
Thermal Cracking. Certain cracking conversion processes are carried out
without catalysts. Heavy residuum streams in the refinery can be cracked ther-
mally to produce coke and a mixture of lighter products. If naphtha is produced,
it may require extensive treating before it can be used directly in gasoline or used
as feed to other processes. Depending on the specific process conditions
and design these processes are known as delayed coking, visbreaking , Fluid
Coking, and Flexicoking (the last two developed by Exxon). Heavy distillate
may also be thermally cracked using high (8008C) temperature steam. This pro-
cess, steam cracking, is used to generate olefins for use in chemicals plants, but
also generates material in the naphtha range which, if the quality is appropriate,
may be used in gasoline.
Reforming. Catalytic reforming is a process to increase the octane of
gasoline components. The feed to a reforming process is naphtha (usually virgin
naphtha) boiling in the 80– 2108C range. The catalysts are platinum on alu-
mina, normally with small amounts of other metals such as rhenium. Because
the catalysts contain noble metals, the feed must have very low levels of sulfur,
nitrogen, and heavy metals such as lead and arsenic. Reactors may operate
in continuous or semicontinuous mode, the difference being the method in
which the catalyst is regenerated. Various designs have been developed by
companies such as UOP (Platforming and CCR), Exxon (Powerforming), and
Amoco (Ultraforming).
Depending on the catalysts and operating conditions, the following types
of reactions occur to a greater or lesser extent: (1) heavy paraffins lose hydrogen
and form aromatic rings; (2) cycloparaffins lose hydrogen to form corresponding
aromatics; (3) straight-chain paraffins rearrange to form isomers; and (4) heavy
paraffins are hydrocracked to form lighter paraffins.
Reformers generate highly aromatic, high octane product streams, and a
great deal of hydrogen (qv). Reformate can have RON values over 100 and
MON values of 90. The hydrogen, which can be used to improve the quality of
many other refinery streams, is an extremely valuable product of the reformer.
Some of the negative aspects of reformate are production of benzene, poly-
nuclear or multiring aromatics (PNAs), and light gas (C1 –C4) (see also BTX
PROCESSING). Benzene is a recognized carcinogen and its concentration in gaso-
line is regulated, whereas PNAs can contribute to combustion chamber deposits.
PNAs may be removed by distilling the whole reformate and discarding the
heaviest fractions.
Alkylation. Alkylation (qv) is the chemical combination of two light
hydrocarbon molecules to form a heavier one and involves the reaction of butenes
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 241

and butanes in the presence of a strong acid catalyst such as sulfuric or hydro-
fluoric acid. The product is a heavier multibranched isoparaffin. Propene and
the various pentenes may also be used, to produce C7 or C9 isoparaffins, respec-
tively. Process designers need to be careful not to let three molecules combine to
form C12 molecules because these have too high a molecular weight and probably
would contribute to poor combustion and higher emissions. Octane numbers for
alkylate are usually above 95 for both RON and MON, and the reaction has the
effect of lowering the RVP of the available pool, a plus when refiners are trying to
reduce it. The principal competitors to alkylation are the various etherification
processes which also use C4 and C5 olefins as feeds. As a result of environmental
concerns associated with use of ethers in gasoline, alkylation is becoming a more
desirable process.
There are environmental concerns over the use of HF catalyst. Although the
refining industry has taken steps to reduce the likelihood of an accidental release
and to minimize the environmental impact in the event of a release, most new
units use sulfuric acid catalysts.
Isomerization. Isomerization is a catalytic process which converts normal
paraffins to isoparaffins. The feed is usually light virgin naphtha and the cata-
lyst platinum on an alumina or zeolite base. Octanes may be increased by over
30 numbers when normal pentane and normal hexane are isomerized. Another
beneficial reaction that occurs is that any benzene in the feed is converted to
cyclohexane. Although isomerization produces high quality blendstocks, it is
also used to produce feeds for alkylation and etherification processes. Normal
butane, which is generally in excess in the refinery slate because of RVP con-
cerns, can be isomerized and then converted to alkylate or to methyl tert-butyl
ether (MTBE) with a small increase in octane and a large decrease in RVP.
Hydrogen Processing. Hydrogen is probably the most valuable refinery
chemical in terms of its ability to improve the quality of refinery streams. It can
be used to remove unwanted species such as sulfur and nitrogen from gasoline
and diesel. Hydrogen treating improves the ability to resist oxidative degrada-
tion and reduces the carbon-forming potential of diesel fuels. Under more severe
conditions and with the proper catalysts, hydrogen can saturate olefinic and
aromatic bonds. Modern desulfurization process which rely on hydrogen are
designed to remove sulfur and avoid unwanted reactions which increase hydro-
gen consumption and can lower octane by saturating olefins and aromatics. The
largest producer of hydrogen in a typical refinery is the naphtha reformer, but
hydrogen can also be produced by reforming methane into CO2, CO, and H2.
Blending Agents. Blending agents are components of gasoline that are
used at levels up to 20% and which are not natural components of crude oil.
As of this writing, all blending agents are oxygenated compounds such as ethers
(qv) and alcohols and are used for one or more of a variety of reasons, including to
increase the octane of the fuel, to reduce vehicle emissions, to reduce dependency
on imported crude oil, and/or to reduce emissions of greenhouse gases by using
renewable resources.
Ethers, such as MTBE and methyl tert-amyl ether (TAME), are made by a
catalytic process from methanol (qv) and the corresponding isomeric olefin.
These ethers have excellent octane values and compete on an economic basis
with alkylation for inclusion in gasoline. Another ether, ethyl tert-butyl ether
242 GASOLINE AND OTHER MOTOR FUELS Vol. 2

(ETBE) is made from ethanol (qv) and isobutylene (see BUTYLENES). The cost and
economic driving forces to use ETBE vs MTBE or TAME are a function of the raw
material costs and any tax incentives that may be provided because of the etha-
nol that is used to produce it. In the United States, MTBE has been found in
groundwater and drinking water supplies, where it imparts a noticeable taste
and smell at levels as low as 10 ppb. For this and other reasons, MTBE has
been banned in California, New York and Connecticut. Consequently, the use
of MTBE in gasoline is expected to decline.
Alcohols such as methanol, ethanol, and tert-butyl alcohol [75-64-9] (TBA)
C4H10O, have also been used as gasoline blending agents. The only alcohol used
in large volumes is biomass-derived ethanol. When ethanol is used as a gasoline
blending agent at a level of 10 vol%, the mixture is known as gasohol. Gasohol
actually contains only 9.5 vol% pure ethanol because the ethanol must be dena-
tured using 5% unleaded gasoline before it can legally be shipped anywhere. As of
this writing, gasohol receives a 5.2 ¢/gal (1.4 ¢/L) federal tax subsidy in the
United States. Some states, mostly farm states, also provide additional tax
rebates for blenders of ethanol. These incentives significantly improve the eco-
nomics of ethanol use.
Ethanol gasoline mixtures have a tendency to separate into two phases in
the presence of water and ethanol is corrosive in its own right. Therefore, pipe-
line companies do not allow ethanol blends to be transported through the normal
pipeline distribution system (see PIPELINES). Denatured ethanol is transported
in trucks, barges or tank cars and blended at terminals. Ethanol also has signi-
ficant nonlinearity when blended with hydrocarbons, especially for RVP and
volatility. Adding only a few percent ethanol can raise RVP of a blend by approxi-
mately 1 psi. In order to achieve a final blend with the desired properties,
refiners must prepare a special mixture called an RBOB (Reformulated gasoline
Blendstock for Oxygenated Blending). Due to ethanols nonlinear blending char-
acteristics, RBOBs must be prepared for specific ethanol contents (ie, 5.7%, 7.7%
or 10%). If used for Federal RFG, refiners must make a hand blend of RBOB and
ethanol and measure its properties before shipping the RBOB. For California
RFG, refiners may use a mathematical model to predict the properties of fully
formulated gasoline containing ethanol.
Despite these limitations, gasohol has achieved a significant market share.
In 2002, ethanol production accounted for 1.6% of the total U.S. gasoline supply
(30). In large farm states, such as Iowa and Nebraska, gasohol sales were 60%
and 38%, respectively, of the total gasoline market (31,32).
According to Section 211 of the Clean Air Act, EPA must approve use of any
new blending agent in gasoline. ‘‘New’’ means that the agent is not ‘‘substantially
similar’’ to unleaded gasoline used to certify 1974 vehicles. In 1981, EPA ruled
that gasolines containing up to 2 wt% oxygen were substantially similar (33).
All aliphatic ethers and alcohols could be used, except that methanol had to
be blended with equal parts of higher alcohols. In 1991, EPA increased the
approved level of oxygenates to 2.7 wt% oxygen for aliphatic ethers and alcohols,
except methanol (34). A complete list of allowable oxygenates may be found in
the literature (35).
Another factor which must be considered when blending oxygenates is the
effect of a large amount of a single component on the volatility and therefore
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 243

190

Temperature, ˚C

140

90

40
0 20 40 60 80 100
Percent evaporated

Fig. 3. Distillation curves for ( ) nonoxygenated, and oxygenated, ie, ( ) 15%


MTBE (2.7% oxygen), and ( ) 10% ethanol (3.7% oxygen), fuels.

the driveability characteristics of the fuel blend. In addition to desired cut-


points, the volatility curve of a good fuel should be smooth and not have any
bumps or flat spots. Figure 3 shows the distillation curves for three fuels, one
made from hydrocarbons, one containing 15% MTBE, and one containing 10%
ethanol. The oxygenated fuels have distinct flat spots which may cause drive-
ability problems in sensitive vehicles.
Additives. Gasoline additives are used to improve the performance of
the fuel either because the hydrocarbon components themselves contain some
deficiency or because it is more effective to add a small amount of additive than
to change the composition of the gasoline. Lead antiknock additives were a good
case in point. Octane can be increased either by processing steps such as reform-
ing, which requires significant investment and increased operating costs, or by
using additives which is considerably less expensive. Additives are added
in parts per million levels to distinguish them from blending agents which are
added in the percents. Most additives are added at terminals as gasoline is
being pumped from storage tanks. Adjustable pumps are used to inject the appro-
priate amount of additive.
Dyes. Dyes may be added to gasoline to impart color for a number of rea-
sons. Originally, these compounds were used to identify leaded gasoline so that
it would not be used for other inappropriate purposes, such as solvents. Dyes can
also be used to identify different gasoline grades so that pipeline companies can
244 GASOLINE AND OTHER MOTOR FUELS Vol. 2

separate tenders, and so that service stations can easily check that the correct
grade was placed in the underground tanks (aq). Dyes are usually based on
azo chemistry and are added in concentrations below 10 ppm. Dyes are rarely
used in gasoline today and pipelines rely on other measures such as density to
separate tenders.
Antioxidants. Antioxidants (qv) are used in gasoline to improve storage
stability. Poor storage stability may sometimes be traced to high concentrations
of olefins, especially conjugated diolefins. In many cases, an antioxidant is added
to the FCC naphtha in the sweetening process. The antioxidant may be added in
large enough amounts so that the blended gasoline contains the proper concen-
tration. Antioxidants are almost always one of two chemical types: phenylenedia-
mines or hindered phenols (qv). These were first used in gasoline around 1930
(36) by Universal Oil Products.
Metal Deactivators. Small amounts of metals in gasoline can contribute
to enhanced rates of oxidation. Copper is especially potent in this regard. In order
to prevent metal catalyzed oxidation, chelating agents (qv) are often added to
gasoline. The most commonly used additive is N,N0 -disalicylidene-1,2-propane-
diamine [94-91-7], C17H18N2O2 (37) at concentrations up to 10 ppm. Special
care must be taken if any of the refinery streams are treated with a copper-
based process. Automobile manufacturers sometimes use higher levels of metal
deactivators in factory fill gasoline, because this gasoline may sit in vehicle tanks
for a long period of time until the vehicle is sold.
Corrosion Inhibitors. Corrosion-causing free water is almost always
present in at least small amounts in all parts of the gasoline distribution system.
It can come about when hot, water saturated gasoline cools in storage tanks or
pipelines. Salt water is found in ship holds, and rainwater is found in under-
ground tanks. Acidic or caustic water may be carried over inadvertently from
various processes in the refinery. Careful monitoring of service station storage
tanks is an important part of good operating practice to prevent unwanted
water from contaminating customers’ vehicle tanks (see CORROSION AND CORROSION
CONTROL).
All corrosion inhibitors in use as of this writing are oil-soluble surfactants
(qv) which consist of a hydrophobic hydrocarbon backbone and a hydrophilic
functional group. Oil-soluble surfactant-type additives were first used in 1946
by the Sinclair Oil Co. (38). Most corrosion inhibitors are carboxylic acids (qv),
amines, or amine salts (39), depending on the types of water bottoms encoun-
tered in the whole distribution system. The wrong choice of inhibitors can lead
to unwanted reactions. For instance, use of an acidic corrosion inhibitor when the
water bottoms are caustic can result in the formation of insoluble salts that can
plug filters in the distribution system or in customers’ vehicles. Because these
additives form a strongly adsorbed impervious film at the metal liquid interface,
low liquid concentrations are usually adequate. Concentrations typically range
up to 5 ppm. In many situations, pipeline companies add their own corrosion
inhibitors on top of that added by refiners.
Rust protection is measured using any one of a number of tests: National
Association of Corrosion Engineers (NACE) rust test (40), ASTM D 665, and
MIL-I-25017 B/C (41). These all involve immersing a steel spindle in a mixture
of gasoline and water for a specified time and evaluating the resulting corrosion.
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 245

ASTM D 4914 lists the Copper Strip Corrosion Test which protects against the
corrosive effects of fuel sulfur on copper.
Antiicing Additives. Additives to control icing are rarely used in the
United States because vehicle design changes have made icing problems almost
nonexistent.
Detergent Additives. Deposits may build up in those parts of an engine
wetted by fuel: fuel injector, intake valves, and the combustion chamber itself.
Both fuel and vehicle design contribute to deposit formation. Proper choice of
additive or additives can prevent deposit buildup and in some cases can clean
up existing deposits. The exact mechanisms of deposit formation are poorly
understood. Moreover, the fuel undergoes a different time/temperature history
in each location leading to different types of deposits. An additive that is effective
in controlling one type of deposit may not work at all, or may even contribute to
increased deposits, in another engine location. For instance, solvent oils can be
particularly effective in preventing the buildup of intake valve deposits, but too
much solvent oil causes excessive combustion chamber deposits and leads to high
ORI. Detergents were first developed for carburetors, and the first was intro-
duced by Standard Oil Co. Of California in 1954 (42). Few if any carbureted
vehicles have been made since the mid-1980s, and today’s vehicles all use fuel
injectors to meter the fuel.
Fuel Injector Detergents. Port fuel injectors generally experience tempe-
ratures around 1008C. Deposits can accumulate in the small annulus in the tip
of the injector through which fuel must flow. As the annulus become plugged,
fuel flow becomes erratic, atomization degrades, and control of stoichiometry
becomes more difficult. Although well-designed injectors can tolerate plugging
levels of up to 30%, variability among injectors can cause problems even at
much lower levels. This is because the vehicle computer controls stoichiometry
for all cylinders on average. If one cylinder is running rich, then the other cylin-
ders run lean so that average stoichiometry is correct. Even partially plugged
injectors lead to poor driveability and higher emissions as well as loss of fuel
economy and engine power.
Although the exact chemical mechanism which leads to injector deposits
has not been identified, it is generally believed that heavy hydrocarbons oxidize
and polymerize during vehicle hot soak. The resulting gums and resins trap
other particles present in the crankcase blowby and in the EGR. Olefins and
diolefins are known to be significant contributors, whereas the role of sulfur is
still unresolved (43–45). Polar compounds have been shown to contribute to
injector deposits. In one experiment, a dirty deposit forming fuel was clay filtered
and the resulting filtrate had very low deposit-forming tendencies (46).
A number of additives have been developed which not only prevent injec-
tor deposits from forming, but which also remove deposits already in place. The
best additives can unclog clogged injectors in less than one tankful of gasoline.
Polybutene succinimides, polyisobutylene amines, polyether amines, and certain
lower molecular weight amines have been shown to be effective injector deter-
gents (43). Phenylenediamine, which is known to improve the oxidative storage
stability of gasoline, does not prevent the reactions which lead to fuel injector
deposits. Performance of additives in controlling fuel injector deposits is mea-
sured using ASTM D 5598. This is a vehicle based test and involves driving a
246 GASOLINE AND OTHER MOTOR FUELS Vol. 2

vehicle with a 2.2 L Chrysler turbocharged engine on a dynamometer for


16,000 km and measuring fuel injector flow compared to baseline values.
Intake Valve Detergents. In the mid-1980s some vehicle manufacturers
started to experience levels of intake valve deposits which caused driveability
degradation, especially before the engine was fully warmed up. These problems
were associated with deposit levels lower than those which had caused problems
in the past. It has been hypothesized that the deposits act as hydrocarbon
sponges, especially during acceleration when the mixture is usually enriched
somewhat in order to provide adequate power. If the deposits, because of physical
and chemical makeup, absorb some of the extra fuel that is passing by, the vehi-
cle may stumble and hesitate because the mixture is too lean. Cars built to low
emission standards generally have the stoichiometry set as lean as possible dur-
ing startup and warm-up, and any upset which leans out the mixture even
further is not well tolerated.
Intake Valve Deposits (IVD). IVD have been studied extensively and a
number of vehicle design and fuel factors have been determined which are prin-
cipal contributors. As for other deposit-forming problems, the mechanism of
deposit formation is not known in much detail, but the broad outlines are fairly
well understood. The primary difference between injector deposits and intake
valve deposits is the temperature regime. Intake valves generally operate at a
temperature as high as 3008C (47), much higher than injectors. The hot soak por-
tion of the driving cycle is not as important for development of IVD as it is for
injector deposits. Also, there is a small but constant flow of lubricating oil
down the stem of the intake valve which may influence deposit formation.
Vehicle factors that contribute to IVD are typically related to whether or
not the valve is kept wet by liquid gasoline during operation. Design factors
which encourage wetting of the valve tend to minimize deposits. Valve rotation,
which rotates the valves slightly on each cycle, also tends to reduce deposits.
Valves that do not rotate develop deposits on the back side which sees little fuel
from the injectors. Oil flow down the valves can either help or hurt. If the oil flow
is too low, the valve may stay dry, whereas oil flow that is too high encourages
buildup of oil related deposits. Injector spray pattern is also important. An even,
continuous spray is the best for minimizing deposits, whereas intermittent injec-
tors tend to promote higher deposit levels.
Fuel factors which contribute to IVD have been shown to include olefins
(48,49), alcohols, and cracked stocks (50).
Additive packages have been developed that do an excellent job of prevent-
ing IVD. The key to effective operation is to keep the valve wet so that the addi-
tive can prevent deposit buildup. Most packages include a combination of
detergent/dispersant and a carrier oil. If no carrier oil is present, then the fuel
may evaporate off the valve too rapidly for the package to be effective. When
the valves do not rotate, the portion of the valve which has the highest deposit
level is the back side which is not constantly wet.
Additive effectiveness is measured a number of ways, but the most mean-
ingful measures use expensive vehicle testing. ASTM D 5500 involves running a
1985 BMW 318i on the road for 16,000 km, after which the engine head is re-
moved and each valve weighed. No laboratory simulation test has been shown
to be a good predictor of IVD and additive effectiveness.
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 247

In 1997, EPA implemented detergency requirements for all gasoline sold in


the United States. Detergent packages must be certified by gasoline blenders and
distributors for nationwide use, for limited geographic use, or for only certain
specific blends. Fuels with severe properties (eg, sulfur, aromatics, olefins, T90)
must be tested and must demonstrate good detergency performance for fuel
injector plugging (ASTM D 5598) and intake valve deposit control (ASTM D
5500).
Combustion Chamber Deposits. As engines age, carbonaceous deposits
build up in the combustion chamber-piston crown and top land, head, and valve
faces. These deposits, which reach equilibrium after about 20,000 km, are typi-
cally 60% carbon, 20% oxygen, 2% nitrogen, and the balance made up of elements
from lube oil (51). The heaviest portion of the fuel, especially multi-ring aro-
matics may be major contributors to the combustion chamber deposits (CCD).
Combustion chamber deposits have extremely good insulating properties
and therefore have a number of effects on vehicle operation. Since less heat
is rejected to the walls, peak flame temperatures are higher, fuel economy is bet-
ter, NOx emissions are higher, and octane number requirements are higher.
CCDs are probably the biggest factor contributing to ORI.
Additives used as fuel injector or IVD detergents may also contribute to
CCD. Generally, all detergent additives contribute to higher levels of CCD, and
their contribution depends on dosage and specific chemistry.
A large study published by the Coordinating Research Council in 2000
examained the effects of CCD on vehicle operation (52). Seven vehicles of each
of four models were run for 24,000 km. Emissions were measured at the start
and end, and CCD measurements were made at the end of testing. Comparing
clean and equilibrated conditions, NOx and CO2 emissions, and fuel economy
were higher for the equilibrated condition. Two different additives produced
higher levels of CCD than nonadditized fuel. NOx and CO2 emissions and fuel
economy were also higher for the additized fuels. The study could not determine
the nature of the relationship between CCD level and emissions or fuel economy.
Development of additives to reduce CCD or to limit CCD growth has been
attempted (53), but testing by EPA has failed to substantiate a significant benefit
for one additive versus another.
Demulsifiers. Because free water is present in many parts of the distri-
bution system, special care must be taken to prevent emulsions (qv) from form-
ing. Emulsions generally refer to thick oil in water emulsions that are very stable
and form a separate phase at the fuel-water interface. Emulsions do not burn
very well, and they always carry large quantities of dirt and rust particles with
them. These particles plug vehicle fuel filters and cause a car to stop running.
Haze is a dilute water-in-oil emulsion in which the water droplets are dispersed
throughout the gasoline but which does not form a separate phase. Haze and
emulsions may form during high shear conditions such as pumping, or during
low shear conditions such as sloshing in the hold of a ship or barge. Gasoline
without additives does not generally form emulsions, but the use of surfactants
to control deposit formation and limit corrosivity can lead to gasoline having very
severe emulsion forming behavior. Demulsifiers are highly surface active chemi-
cals or mixtures having limited solubility in both fuel and water (54). The hydro-
philic portion may be polyethylene or polypropylene and the hydrophobic portion
248 GASOLINE AND OTHER MOTOR FUELS Vol. 2

may be a long-chain alkylphenol or alcohol. The specific demulsifiers chosen


depend on the base gasoline, the rest of the additive package, and the type of
water encountered in the distribution system. Demulsifiers may have to work
for water bottoms that range from pure rainwater, to salty seawater, to caustic
water bottoms.
In addition to ASTM D 1094, there are a number of widely used industry
tests to measure the water handling properties of gasoline. One uses a Waring
Blender to simulate high shear (54), whereas another uses a wrist action shaker
to simulate mild shear conditions (55).
2.3. Blending and Distribution. When blending gasoline from its com-
ponents, refinery operators must balance a number of factors in the most
economical way. First, the components must be used at the same rate at which
these are produced or else the refinery either runs out of material or drowns in
excess components. Secondly, each gasoline fuel grade must be produced to the
specifications set by marketers and regulators. The specifications should not be
exceeded in a way which increases manufacturing cost. Third, blend targets
must take into account the fact that the gasoline might not be sold immediately
and may travel in pipelines, barges, or tankers and then sit in a distribution
terminal. The time between production in the refinery and sale to the customer
can be as long as one month. Finally, since all testing has a certain amount of
variability, the blender must decide how much margin or ‘‘giveaway’’ to build
in to his targets. A large margin is costly but ensures that if a customer tests
quality at the time of transfer, or if a regulator checks compliance with regula-
tions, the fuel will likely meet standards. Conversely, with a small margin, the
blender runs the risk that a retest of the fuel will show it to be out of specifica-
tions. Procedures for adjudicating these kinds of product quality disputes are
defined in ASTM D 3244.
Most refineries accomplish the difficult task of planning and executing
blending through the use of sophisticated linear programming algorithms. The
linear programs are run in a few different time scales ranging from weekly to
yearly to help refineries plan their seasonal operations. In the winter, gasoline
demand is down and heating oil demand is up; in the summer, gasoline demand
is at its peak.
Many gasoline properties, especially octane and RVP, do not blend linearly.
Whereas the volatility properties of blends can be well approximated using non-
ideal properties, it has not been possible to predict the octane values of blends
from first principles. It is known that when various refinery streams are blended
together, the octane of the blend is usually different from the volume weighted
average octane of the individual components. This deviation can be either posi-
tive or negative. For instance, blending alkylate and reformate almost always
results in an octane number that is lower than the weighted average of the indi-
vidual components. This is known as an octane blending debit. Likewise, mixing
FCC naphtha and virgin naphtha almost always results in a blended octane
higher than expected, or an octane blending bonus (56). Experiments using
blending agents show that the addition of most agents is nonlinear and depends
strongly on the composition of the base gasoline.
Proper prediction of the octane of refinery blends is important because
octane has traditionally been one of the most expensive gasoline properties and
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 249

94 LBV1 94

92 92
Octane number

90 90

A ON1
88 88

LBV2

B 86
86

ON2

84 84
0 0.2 0.4 0.6 0.8 1
Fraction of component 1

Fig. 4. Octane blending behavior where (—) represents the measured octane response
curve, ( ) the tangent to the curve, and (– – –) the linear connection between the
octane number of components 1 and 2. The line AB corresponds to the octane blending
bonus of this composition.

raising pool octane often entails significant investment and increased operating
costs. Also, it is possible to meet targets for the different grades by properly choos-
ing blend stocks to take advantage of octane bonuses available from the non-
linear blending characteristics.
Blending behavior of a binary mixture may be characterized by a linear
blending value (LBV). Figure 4 shows the response curve of a hypothetical
two-component mixture. The LBV for each of the components at any composition
fc is defined by the tangent at that point according to the formula.

LBV 1 ¼ ON c þ ð1  f1 Þ ðdON=df1 Þc
LBV 2 ¼ ON c  f1 ðdON=df1 Þc

where f1 is the weight fraction of component 1, (dON/df)c is the slope of the


blending curve at composition c, and ONc is the octane number of the blend at
composition c.
As can be seen from Figure 4, LBVs for these components are not constant
across the ranges of composition. An interaction model has been proposed (57)
250 GASOLINE AND OTHER MOTOR FUELS Vol. 2

which assumes that the lack of linearity results from the interaction of pairs of
components. An approach which focuses on the difference between the weighted
linear average of the components and the actual octane number of the blend
(bonus or debit) has also been developed (58). The independent variables in
this type of model are statistical functions (averages, variances, etc) of blend
properties such as octane, olefins, aromatics, and sulfur. The general statistical
problem has been analyzed (59) and the two approaches have been shown to be
theoretically similar though computationally different.
Most of the octane blending values reported in the literature use a slight
variation on this theoretically sound approach. The composition and octane of
the base fuel are assumed to be fixed and the second component is assumed to
be added. Using the same nomenclature, the blending octane number (BON) of
component 2 is defined as

BON 2 ¼ ðON c  ON 1 f1 Þ=f2

The advantage of this definition is that it does not depend on measuring the
tangent of the response curve, although the variation in the value of the blending
octane number is greater. Typically, BONs are measured at an 80/20 mixture.
This technique is also useful when trying to measure the octane of a compound
such as butane or methanol that is difficult or impossible to measure in its pure
state.
Most refineries develop individual octane blending equations which do a
good job of predicting that refinery’s blending behavior. In order to use these
equations in refinery planning and operations, these may be linearized in a
piecewise fashion.
Until the 1960s, most gasoline blending was done in batches. All the com-
ponents were stored in large tanks, a recipe was defined by a master blender, and
an empty tank was filled according to the recipe. After the blend was finished, a
sample was sent to the laboratory for testing. If all properties were within spe-
cifications, the blend was released. If not, certain components would be added
to fix the blend properties. In the extreme, a blend could be downgraded from
premium to regular, it could be totally reblended, or it could be sold at a discount
to a wholesale customer willing to use off-spec material.
Today, many, if not all of the steps in blending and certifying gasoline have
been automated. Computers define preliminary recipes. Although ASTM certi-
fication tests are typically laboratory based, some have been adapted and auto-
mated for use in the refinery. As a blend is being made in the refinery, samples
may be taken automatically as often as every 20 minutes. These samples are fed
to special instruments which measure critical properties such as octane and vola-
tility. If the integrated values of the blend are not within specifications, then the
automated blenders adjust the recipe.
In a large refinery, an average blend might be 15  106 L , having an aver-
age flow rate though the blender of 106 L/h. When in-line testing is used, a large
number of repeat tests conducted on the blend as it is made, and the properties of
the blends may be known to a high degree of statistical confidence. The final
blends may be pumped directly into a pipeline or a barge.
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 251

Sometimes, regulations require that blend properties be measured on a


sample of the final blend. In these cases, a sample is taken from the blend
tank to the refinery laboratory where it is analyzed again. The single laboratory
measurement is then used as the official property of the blend. The outcome of
this process is a value that is much less precise than if all the in-line analyzer
results would be used.
Gasoline blends are shipped from the refinery to a storage terminal. At the
terminal, additives are metered into the gasoline as it is loaded onto tank trucks
for shipping to individual service stations. The trucks, which have 40,000 L
capacity, have 4–5 compartments so that they can deliver different grades at
the same time. Service station tanks are buried underground and have a capacity
of 12,000–15,000 L. High volume service stations may have underground tanks
as large as 45,000 L. Submerged turbine pumps transfer the gasoline from the
tanks to dispensers at the dispensing islands. Tanks have been made of a
number of materials, although the most popular today is reinforced fiberglass.
As a result of environmental concerns about gasoline leakage from underground
tanks, many new installations have double-wall construction with leak detectors
between the two walls.
2.4. Fuel Economy. Fuel economy, typically expressed as distance dri-
ven per volume of fuel consumed, ie, in km/L (mi/gal), is measured over two driv-
ing cycles specified by the Federal Test Procedure (60). One cycle simulates city
driving and consists of relatively low speed (32 km/h) driving, and includes a
portion where the car starts after having equilibrated at ambient conditions
for 16 hours. The second simulates highway driving conditions and includes
higher speeds and fewer starts, stops, and accelerations. The statutory fuel econ-
omy standards are based on a harmonic average of the city and highway tests
assuming that 45% of distance is accumulated under highway conditions and
55% is accumulated under city driving conditions.
Fuel economy is measured using a carbon balance method calculation. The
carbon content of the exhaust is calculated by adding up the carbon monoxide,
carbon dioxide, and unburned hydrocarbons (qv) concentrations. Then using
the percent carbon in the fuel, a volumetric fuel economy is calculated. If the
heating value of the fuel is known, an energy specific fuel economy in units
such as km/MJ can be calculated as well.
The most important fuel property in terms of volumetric fuel economy is the
volumetric heat content of the fuel. Fuel economy has been shown to respond
linearly and with a slope of almost unity to changes in energy content of the
fuel. In turn, fuel energy content is most influenced by density and oxygen con-
tent. Density is positively correlated with heat content whereas oxygen is nega-
tively correlated. Figure 5 shows the relationship between energy content and
fuel economy for a group of twenty 1989 vehicles tested on a wide range of gaso-
lines with and without oxygen (61).
2.5. Environmental Issues. Gasoline properties discussed up to this
point have generally dealt with those properties that contribute to good vehicle
operation from a mechanical perspective. Other properties are controlled for
environmental reasons and those are discussed in this section. Regulatory autho-
rity to control fuel properties for environmental reasons derives from the Clean
Air Act and its various amendments.
252 GASOLINE AND OTHER MOTOR FUELS Vol. 2

9.3

9.2

9.1
Fuel economy, km/L

8.9

8.8

8.7

8.6

8.5
30 30.5 31 31.5 32 32.5 33
Net heating value, MJ/L

Fig. 5. Fuel economy vs energy content for various fuels. Each data point represents the
average of twenty 1989 vehicles. To convert MJ/L to Btu/gal, multiply by 2:78  108 .

The Clean Air Act directs EPA to set National Ambient Air Quality Stan-
dards (NAAQS) for a number of pollutnats such as ozone, CO and NOx, and
sets or directs EPA to set vehicle emissions standards and fuel composition
standards to help meet these NAAQSs. Because of its especially serious air
pollution problems, the state of California is allowed to impose its own strict
set of standards under the Clean Air Act.
The first of these standards set by EPA went into effect in 1968 and man-
dated that the vapors from the vehicle crankcase be routed back through the
engine and burned. Since then, the standards have continued to grow stricter.
Table 8 shows the federal exhaust emission standards and Table 9 shows the
more stringent California standards. A novel feature of the latest federal and
California standards is that they must be met by each manufacturer on their
sales weighted average. This means that certain vehicles may emit more than
others, as long as the average meets the standard. For the federal program, man-
ufacturers must certify vehicles in various bins, which define standards for non-
methane hydrocarbons (NMHC), CO, NOx and PM, and the sales weighted fleet
must meet the NOx average shown in Table 8. In a similar way, the California
program defines various emission classes, and sales weighted averages must
meet a standard defined in terms of nonmethane organic gases (NMOG).
NMOG is very similar to the EPA’s NMHC.
The emissions standards have forced changes in vehicle hardware as shown
in Table 7. Many of these changes in hardware have resulted in changes in fuel
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 253

Table 8. Federal Light Duty Exhaust Emission Standards (80,000 km)


Emissions, g/kma
Year HC CO NOx Major hardware changes
pre-control 9.4 56 3.9
1970 2.5 21 lean combustion
1972 1.8 17
1973 1.9 exhaust gas recycle
1975 0.94 9.4 oxidation catalysts
1977 1.2
1980 0.25 4.4
1981 2.1 0.6 three-way catalysts, oxygen sensors
1994–6 0.16b 2.1 0.25
2001 0.047c 2.1 0.12
2004–07d 0.047 2.1 0.031
a
To convert to g/mi, multiply by 1.609.
b
Nonmethane hydrocarbons.
c
Nonmethane organic gases, the total mass of exhaust hydrocarbon and oxygenated
compounds, excluding methane.
d
Fleet average standards must be met for NOx, based on emissions in ‘bins’. Values shown
for NMOG and CO correspond to the Bin 5 standards.

Table 9. California Light Duty Vehicle Exhaust Emission Standardsa


Vehicle typeb NMOG, g/km CO, g/km NOx,g/km

LEV 0.047 2.1 0.031


ULEV 0.025 1.1 0.031
SULEVc 0.006 0.62 0.012
ZEV 0.0 0.0 0.0
Fleet average
Year NMOG, g/km
1999 0.070
2000 0.045
2001 0.044
2002 0.042
2003 0.039
2004 0.033
2005 0.030
2006 0.029
2007 0.027
2008 0.025
2009 0.024
2010 0.022
a
Standards at 80,000 km unless shown.
b
LEV ¼ low emissions vehicle.
ULEV ¼ ultra low emissions vehicle.
SULEV ¼ super low emissions vehicle.
ZEV ¼ zero emissions vehicle.
c
192,000 km.
254 GASOLINE AND OTHER MOTOR FUELS Vol. 2

as well. For example, starting in 1975, vehicle manufacturers installed noble


metal catalysts in order to meet federal standards and because the catalysts
could not tolerate lead, all fuel manufacturers were mandated to sell at least
one grade of unleaded gasoline. As demand for unleaded gasoline grew, new pro-
cessing capacity was required and different blends developed.
In addition to setting standards for exhaust emissions, the government set
standards for evaporative emissions. These refer to hydrocarbons that escape
from the vehicle when fuel evaporates; while the car is operating (running
losses); while it is sitting and not being operated (diurnal emissions), immedi-
ately after operation (hot soak); or while it being refueled. In order to control
evaporative emissions, auto manufacturers have installed canisters of activated
charcoal in their vehicles since 1972. The charcoal traps any hydrocarbon vapors
from the fuel tank or from the intake system. During engine operation, the
adsorbed hydrocarbons are stripped off the charcoal by a countercurrent air
stream and fed into the engine and burned.
Although the charcoal canisters are about 95% effective, fuel volatility still
impacts the mass of vapors that break through the canister. Therefore, EPA
mandated that starting in the summer of 1992, RVP levels be reduced below
the levels specified in ASTM D 4814. Class C regions, generally the northern part
of the country, are limited to a maximum RVP of 62 kPa (9.0 psi) vs an ASTM
limit of 79 kPa (11 psi), and the southern Class B regions are limited to a maxi-
mum RVP of 54 kPa (7.8 psi) vs 69 kPa (9.0 psi) for ASTM.
The Clean Air Act Amendments of 1990 (67) introduced a new concept in
emissions reduction: reducing exhaust emissions by controlling the composition
of the fuel. Whereas previous regulations had lowered exhaust emissions by
setting standards for new vehicles, this law mandated gasoline marketers to
change gasoline composition so that emissions from existing vehicles would be re-
duced. Reduction targets of 15% and at least 20% were set for 1995 and 2000,
respectively. Reductions are to be measured against 1990 vehicles and industry
average gasoline. The reductions are for hydrocarbons (summer ozone season
only), and for air toxics (year round). Air toxics are defined as the sum of the
emissions of benzene [71-43-2], 1,3-butadiene [106-99-0], formaldehyde [50-00-0],
acetaldehyde [75-07-0], and polycyclic organic material. Reformulated gasoline
(RFG) may not result in any increase in NOx emissions. Additionally, RFG must
contain no more than 1% benzene and at least 2% oxygen.
Each refiner has the flexibility to choose the specific formulation to produce
based on the economics of the individual refineries. RFG meeting the statutory
requirements must be sold in the nine areas of the country which have the worst
ozone (qv) problem. In addition, all other areas of the country which exceed the
ozone NAAQS may adopt the RFG requirements. In 2003, approximately 30% of
the gasoline sold in the United States was subject to RFG rules.
In the winter, the Clean Air Act mandates that gasoline in all areas which
exceed the NAAQS for CO must contain at least 2.7% oxygen. This is based on
the assumption that adding oxygen to the fuel reduces CO emissions.
Compliance with the RFG regulations is measured by using a formula
to calculate the emissions from a given fuel. This formula, the Complex Model,
predicts average exhaust and evaporative emissions as a function of gasoline
chemical and physical parameters. In anticipation of the need to develop this
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 255

equation, and because very little data were available, the three domestic auto
companies and fourteen petroleum companies formed a joint research program
in late 1989 called the Auto/Oil Air Quality Improvement Research Program
(AQIRP). This four-year, $40 million program had the objective of ‘‘conducting
a research and testing program to develop data on potential improvements in
vehicle emissions and air quality, primarily ozone, from reformulated gasoline,
various other alternative fuels, and developments in vehicle technology’’ (63).
The program developed a great deal of information on the effects of fuel pro-
perties on exhaust and evaporative emissions (64–66). The data collected by
AQIRP represent the largest single body of information collected on this subject.
Over two thousand individual emission tests were conducted between 1990 and
1992. EPA used this and other data to generate an equation to predict emissions
as a function of fuel properties.
Final regulations were published after much consultation with the oil and
automotive industries, and with consumer and environmental groups. Specific
reduction targets were set for RFG, and gasoline manufacturers were given the
option of complying with the regulations on every gallon or on average. In gene-
ral, if averaging is used, the standard is more severe (1.5% for VOC, NOx, toxics)
and the performance of any one blend may not be too far below the regulatory
level (within 2.5% for VOC, NOx).
Federal RFG and wintertime oxygenated fuels must contain compounds
that add oxygen to the fuel (2% and 2.7% by weight respectively). These require-
ments have been met for the most part by using MTBE and to a lesser extent
ethanol. MTBE use has caused concerns because of underground tank leakage
and groundwater contamination. Some states have banned MTBE, and where
oxygenates are required by federal mandate, ethanol has been used exclusively.
There is considerable political pressure to repeal the portion of the CAA and asso-
ciated regulations that require oxygenate use. Technically, oxygenates have little
or no impact on emissions from modern vehicles, and do not reduce either ozone
or CO levels in the atmosphere (67).
The state of California has taken a different conceptual approach to redu-
cing emissions through control of gasoline composition. Instead of defining a
performance target, ie, 25% reduction, California has defined composition targets
which are aimed at achieving emissions reductions. The third round of regu-
lations, known as Phase 3, are shown in Table 10 and took effect on January 1,

Table 10. California Phase 3 Gasoline Compostion


Values for averagers

Property Limit per liter Average Cap


a
RVP, kPa 48 NA 50
sulfur, ppm 20 15 40
aromatics, vol% 25 22 35
olefins, vol% 6 4 10
T90, 8C 152 146 166
T50, 8C 101 95 104
benzene, vol% 0.8 0.7 1.1
a
To convert kPa to psi, multiply by 0.145.
256 GASOLINE AND OTHER MOTOR FUELS Vol. 2

2004. Gasoline producers must meet these targets on every liter of gasoline. If
desired, a company may choose to meet yearly average targets, but these are
slightly more severe than the per liter specifications, and the ranges which
may be used for averagers are restricted.
In order to provide refiners some flexibility in meeting requirements,
California developed the Predictive Model, which defines trade-offs among the
specifications, while maintaining equivalent emissions performance. Refiners
use the Predictive Model to define alternative specifications for their operation.
These alternative specifications, which can be changed as often as desired, are
used in place of the values in Table 9. For instance a refiner may choose a lower
sulfur level, say 10 ppm, to offset higher olefins, say 9%.
California regulations have been deemed equivalent to the Federal RFG
rules and supersede them except for the use of oxygenates which must be used
in areas that would otherwise require Federal RFG (Los Angeles, San Diego, and
the San Joaquin Valley). Since California has banned use of MTBE and other
ethers, ethanol is the only oxygenate used in the state.
In recent years, other government agencies have regulated gasoline speci-
fications and composition. Some examples are:

Area Controls
Detroit, Mich. Low RVP
Pittsburgh, Pa. Low RVP
Atlanta, Ga. Low RVP, low Sulfur
Birmingham, Ala. Low RVP, low Sulfur
Phoenix, Ariz. CA or Federal RFG, oxygen
not required
New York State MTBE banned
Connecticut MTBE banned

By one count, 20 different sets of formulations or regulations have been


defined across the United States. Many of these so-called boutique fuels are
sold in only a small geographic area. Distribution of so many different grades
through the fungible pipeline and terminal system has the potential to overload
the system and to make temporary supply disruptions or shortages more prob-
able in the event of a refinery upset.
Gasoline composition is also regulated in Europe as shown in Table 11.
These values were defined by the European parliament after analysis of air
quality and fuel effects on automotive emissions.
Gasoline composition regulations are being considered in many other parts
of the world as well. The options range from the basic, such as eliminating lead,

Table 11. European Gasoline Compostion Regulations


2000 2005 2009
sulfur, max, ppm 150 50 10
aromatics, max, wt% 42 35 35
benzene, max, wt% 1.0 1.0 1.0
olefins, max, wt% 18 18 18
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 257

to full reformulations such as those adopted in the United States and Europe.
The choice of appropriate regulations should be based on many factors, including
vehicle technology, the status of air quality, and cost effectiveness relative to
other emission control steps.

3. Diesel Fuel

As a fuel for internal combustion engines, diesel fuel ranks second only to gaso-
line. In 2001, total U.S. demand for diesel fuel for transportation was about 150
billion L, about 29% of the demand for gasoline (68). Use of diesel fuel for high-
way vehicles accounted for 81% of the total, off-highway vehicles for 6%, ships for
5%, locomotives for 7%, and military use for 1% (69). Diesel cars, thought at one
time to be very promising, have encountered significant customer resistance in
the United States. There are significant obstacles to overcome. Consumers have
perceived diesels as noisy, smoky, and hard to start in winter. The fuel economy
benefit that diesel engines can offer is not as attractive when gasoline prices are
relatively low. Unites States automotive manufacturers are considering diesels
as a way to meet corporate average fuel economy standards and still be able to
sell large, powerful SUVs. Increased penetration of diesel engines in the light
duty market means that they will have to meet exceedingly strict emission stan-
dards and will have to compete with gasoline-battery hybrids for those interested
in higher fuel economy.
In many countries, diesels engines do not have the negative image that
they do in the United States. Furthermore, tax differentials on fuels and vehi-
cles, as well as overall high fuel prices can make diesel economically attractive.
In France, for example, diesel engines constituted 56% of new car passenger and
light duty truck sales in 2002, vs about 0.1% in the United States.
3.1. Combustion in Diesel Engines. Unlike the spark-ignited gaso-
line engine, the diesel engine, first used by Rudolf Diesel in the 1890s to burn
finely powdered coal dust, employs compression ignition. Liquid fuel was employ-
ed soon after. Through the action of the pistons, air alone is drawn into the cylin-
ders and compressed. Near the end of the compression stroke, fuel is injected into
the cylinder and after a short delay, is ignited by the high temperature gene-
rated during compression. The fuel must be finely atomized and well mixed
with the air for complete combustion to take place. The ignition delay is useful
because it allows the fuel to be injected and mixed before combustion starts. Its
length depends largely on the composition of the fuel, and is described by cetane
number.
Diesel engines offer a number of advantages over gasoline engines. Diesel
engines are able to operate at high (16–22) compression ratios which improves
energy efficiency. Because combustion starts at many different sites and there is
no flame progressing across the combustion chamber, there are no knock con-
cerns. The engine operates with unthrottled air flow, which also improves effi-
ciency. Because the air flow is unthrottled, changing fuel flow results in a wide
range of stoichiometry, always in the lean region. Lean operation also means
that HC and CO emissions are generally low for diesels. If too much fuel is fed
to the engine to boost power, visible black smoke and soot are emitted.
258 GASOLINE AND OTHER MOTOR FUELS Vol. 2

Diesels also suffer from a number of disadvantages relative to gasoline


engines. Diesels are generally hard to start in cold weather. Many engines use
glow plugs to heat up the combustion chamber so that the fuel can ignite. It
used to be common for heavy-duty diesels to be left running in cold weather
because they are so hard to restart. Although diesel engine-out emissions of
NOx are not much different from those of equivalent gasoline engines, diesel
engines do not have exhaust catalysts so the tailpipe emissions are much higher.
Additionally, diesel engines tend to be heavier and more expensive than gasoline
engines of the same power due to the need for more structural strength and
more complex fuel system.
In assessing the possible health effects of diesel exhaust, EPA concluded
that diesel exhaust is likely to pose a lung cancer hazard to humans (70). Com-
ponents of diesel exhaust that contribute to this assessment include aldehydes,
benzene, 1,3 butadiene polyaromatic hydrocarbons (PAH), nitro-PAH, and parti-
culates. EPA’s assessment was not able to quantify the risk, nor was it able to
identify the hazards posed by specific components. EPA also pointed out that
the data was based on engines built before the mid-1990s. The next generation
of diesel engines will be much cleaner and the health impacts are likely to be
much lower.
There are two categories of diesel engines: direct injection (DI) and indirect
injection (IDI). In DI engines, the fuel is injected directly into the combustion
chamber. In IDI engines, there is a small prechamber into which the fuel is
injected. The fuel starts to ignite in the prechamber and the hot burning gases
are forced out into the main combustion chamber through a small passage. IDI
engines may operate at higher speeds and use lower pressure injector systems
which tend to be less expensive. Most new engines, both light and heavy duty,
use DI designs to take advantage of higher power and better fuel economy.
A new approach to diesel engine design is called Homogeneous Charge
Compression Ignition (HCCI). In an HCCI engine, the fuel air mixture is homo-
geneous as in a spark ignition engine and in contrast to a traditional diesel
engine, where there is significant nonhomogeneity. However, the combustion is
auto-ignited and takes place very rapidly without a flame front. The stoichio-
metry of the mixture is lean to keep NOx emissions low. This type of engine
has the potential to have very low emissions as well as high efficiency, and com-
bines the best features of gasoline and diesel engines. A key technical hurdle is to
make sure that combustion is smooth and complete even under high speed and
high load conditions. The fuel requirements are being developed, and it is possi-
ble that the fuel will also fall somewhere between traditional gasoline and diesel
properties.
3.2. Requirements for Good Diesel Fuel. Diesel fuel is used in a
wide variety of vehicular engines ranging from small passenger cars to large
trucks and construction equipment. There are actually three grades of diesel fuel
defined in ASTM D 975, the specification for diesel fuels. The first is Grade
1-D, suitable for high speed engines which operate under widely varying con-
ditions of speed and load. Grade 1-D also has excellent low temperature pro-
perties. Grade 2-D is a general-purpose diesel suitable for use either in
automotive or nonautomotive applications. It can be used in high speed engines
involving relatively high loads and uniform speeds. Grade 4-D is much more
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 259

viscous and is used in low and medium speed engines having sustained loads at
substantially constant speed. Most cars and trucks use 2-D, a general-purpose
grade. 1-D, a more volatile, lower density, lower aromatic fuel, used in cold
weather and in municipal buses. Starting in 1993, EPA has lowered the regulat-
ed sulfur level of diesel fuel twice and ASTM has defined two new sets of grades-
low sulfur (LSD) and ultra low sulfur (ULSD). These are described in more detail
below.
Ignition Quality. The ability of diesel fuel to burn with the proper charac-
teristics is described by its cetane number, a measure of ignition delay. Exces-
sively long ignition delays (low cetane number) cause rough engine operation,
misfiring, incomplete combustion, and poor startability. Because the fuel starts
to burn later in the cycle, pressure rise is more rapid without increasing the net
work from the cycle. Power may be reduced and combustion may be incomplete
leading to high emissions if the fuel does not have time to burn in the expansion
stroke.
The procedure for measuring the cetane number of diesel fuel (ASTM D 613)
is similar to that used for measuring gasoline octane number. Cetane [544-76-3]
(n-hexadecane), C16H34, is defined as having a cetane number of 100; a-methyl-
naphthalene [90-12-0], C11H10, is defined as having a cetane number of 0. 2,2,4,-
4,6,8,8-Heptamethylnonane [4390-04-9] (HMN), C16H34, which can be produced
in high purity, is used as the low reference fuel and has a cetane number of 15.
Blends of cetane and HMN represent intermediate ignition qualities according to
the formula:

cetane number ¼ % cetane þ 0:15 ð% HMNÞ

The cetane engine is a variable compression single cylinder engine very


much like the octane engine. The engine is run at 900 rpm and injection is
timed to start at 138 before top dead center (BTDC). The compression ratio is
adjusted so that the test fuel starts to ignite at exactly top dead center (TDC),
for an ignition delay of 138 or 2.4 ms. Reference fuels are chosen which bracket
the sample and the cetane number of the sample is estimated by interpolation
between the two reference fuels.
The cetane number of a fuel depends on its hydrocarbon composition. In
general, normal paraffins have high cetane numbers, isoparaffins and aromatics
have low cetane numbers, and olefins and cycloparaffins fall somewhere in
between. Most diesel fuels marketed in the United States have cetane numbers
ranging between 40 and 50. Most manufacturers specify a minimum cetane num-
ber of 40–45.
Cetane number is difficult to measure experimentally. Therefore, various
correlation equations have been developed to predict cetane number from fuel
properties. One such equation may be found in ASTM D 4737 to calculate a
cetane index (CI). ASTM D 975 allows use of CI as an approximation if cetane
numbers are not available.
In an effort to improve the precision and cost of measuring cetane number,
ASTM has adopted a laboratory device called IQT (Ignition Quality Tester) as
method D 6890. This device measures ignition delay in a constant volume bomb,
under carefully controlled conditions. Results are converted to cetane numbers
260 GASOLINE AND OTHER MOTOR FUELS Vol. 2

by means of a correlation equation. Compared to the cetane engine, the IQT


method is less expensive, more precise and faster. D 6890 has not been approved
as an alternative to the cetane engine measurement (ASTM D 613) because the
bias between the two tests is not fully understood and characterized.
Cold Temperature Properties. Diesel fuel must be able to be pumped and
to flow through all filters and injectors at the lowest temperature that may be
encountered in use. When the temperature is lowered, wax molecules in the fuel
start to crystallize. This temperature is known as the wax appearance point or
cloud point. These temperatures, which are generally the same, are measured by
ASTM D 2500 and D 3117, respectively. If the temperature is lowered still
further, the fuel gels and does not flow. This is the pour point and is measured
by ASTM D 97. These tests measure the ability of a fuel to operate in a diesel
engine. Generally, the cloud point of a fuel is 4–68C above the pour point,
although fuels having differences of 118C are not uncommon. The true operability
temperature is somewhere in between the two; cloud point is too high and pour point is
too low. Many engine manufacturers recommend fuels having pour points of 68C
below the lowest temperature at which the engine is expected to operate.
Some additives have the ability to lower the pour point without lowering
the cloud point. A number of laboratory scale flow tests have been developed to
provide a better prediction of cold temperature operability. They include the
cold filter plugging point (CFPP), used primarily in Europe, and the low tempera-
ture flow test (LTFT), used primarily in the United States. Both tests measure
flow through filter materials under controlled conditions of temperature, pres-
sure, etc, and are better predictors of cold temperature performance than either
cloud or pour point for additized fuels.
If the fuel temperature is below its pour point, the fuel has difficulty flowing
out of the storage tank on the vehicle. Diesel powered vehicles generally do not
have in-tank pumps. If, however, the fuel temperature is above its pour point
but below its cloud point, the following situation may occur. The engine starts
but as operation continues, wax crystals begin to collect on the fuel filter, plug-
ging it after a few minutes and stopping the vehicle totally. Vehicle manu-
facturers can take a number of steps to minimize these problems. Some are
relatively expensive, such as heated filters, tanks, and engine blocks. Others
are relatively simple although not as effective. These include making sure that
the fuel lines have no kinks or sharp bends in them and routing lines through
sheltered locations.
Vehicle testing is the best way to determine low temperature requirements.
These tests can be carried out in environmentally controlled facilities under
strictly defined conditions or in actual use. Many companies and industry groups
carry out customer tests in cold climates such as northern Canada or Finland.
ASTM D 975 recognizes that ‘‘it is unrealistic to specify low temperature pro-
perties that will ensure satisfactory operation at all ambient conditions’’. There-
fore, it specifies that the purchaser and supplier of fuel should agree on low
temperature property requirements at the time of purchase.
Volatility. Volatile light fractions in diesel fuel help to provide easy engine
starting but are generally low in cetane number and energy content. Heavy frac-
tions, which have good cetane and energy content, can contribute to deposit
formation and hard starting if present in too high concentrations. Desirable
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 261

quality characteristics are obtained by careful blending of refinery streams. The


temperatures at which 10, 50, and 90% of the fuel evaporate in the ASTM D 86
distillation test are used as controls to provide good volatility. In general, most
diesel fuels have an initial distillation temperature above 1608C and a 90% point
of 290–3608C, depending on fuel grade.
Viscosity. For optimum performance of diesel engine injector pumps, the
fuel should have the proper viscosity. Too low viscosity results in excessive in-
jector wear and leakage. Viscosity that is too high may cause poor atomization
of the fuel upon injection into the cylinders.
Diesel fuel kinematic viscosity is measured by ASTM D 445, and is reported
in units of mm2/s at 408C. Desired viscosity is a function of fuel grade and ranges
from a minimum of 1.3 mm2/s for 1-D to a maximum of 24 mm2/s for 4-D.
Density. The greater the density of diesel fuel, the greater its heat content
per unit volume and therefore the greater its power or fuel economy. Because
diesel fuel is purchased on a volume basis, density is often stipulated in purchase
specifications and measured on delivery. A common measurement of density is
API gravity (ASTM D 287), which is measured easily and accurately using cali-
brated hydrometers. The relationship between specific gravity (SG) and API
gravity is

API at 15:6 C ¼ ð141:5=SGÞ  131:5

Flash Point. The flash point of a fuel indicates the temperature below
which the fuel can be handled without danger of fire. This is the temperature to
which the fuel must be heated to create sufficient vapors above the surface of the
liquid that they can be ignited in the presence of an ignition source. The flash
point of diesel fuel is measured by ASTM D 93, using a closed-cup Penske-
Martens tester. The fuel is heated at a rate of 5.68C/min and a test flame is intro-
duced into the test chamber at 30 s intervals. The fuel temperature reached
when the flame ignites the vapors is called the flash point. Specifications for
flash point vary with grade; the lowest value is 388C for grade 1-D. Controlling
flash point is important in order to prevent the vapor space in storage and vehicle
tanks from being in the explosive range. Setting the flash point at 388C protects
most storage vessels from exploding.
Carbon Residue. The tendency of a diesel fuel to form carbon deposits in
an engine can be roughly predicted by the Ramsbottom Coking Method (ASTM D
524), which determines the amount of carbon residue left after evaporation and
chemical decomposition of the fuel at elevated temperatures for a specified length
of time. The Ramsbottom method involves heating the last 10% residue of the
ASTM distillation at 5498C for 20 minutes. The result is reported as percent
carbon on 10% residuum. For use in high speed diesel engines operating over
a range of loads and speeds, ASTM specifications call for no more than 0.15%
Ramsbottom carbon residue. Because ignition control additives may interfere
with the tests without actually leading to deposit formation in the engine, these
laboratory tests should be made on fuel before any addition of ignition control
agent.
Sulfur. Sulfur in diesel fuel should be kept below set limits for both en-
vironmental and operational reasons. Operationally, high levels of sulfur can
262 GASOLINE AND OTHER MOTOR FUELS Vol. 2

lead to high levels of corrosion and engine wear owing to emissions of SO3 that
can react with condensed water during start-up to form sulfuric acid. From an
environmental perspective, sulfur burns to SO2 and SO3, the exact split being
a function of temperature and time in the combustion chamber. SO3 can react
with water vapor in the exhaust which can further react to form sulfates, leading
to high levels of particulates. Although the SO3 fraction may be small, diesel fuel
has traditionally contained sulfur concentrations as high as 5000 ppm. As parti-
culate emission standards for diesels became more stringent, engine manufac-
turers made the case that they could not meet the standards unless sulfur
levels were reduced. Through negotiations between the Engine Manufacturers
Association and API, it was agreed that sulfur content of highway diesel could
be reduced, and EPA specified a maximum level of 500 ppm starting in October,
1993. In December 2000, EPA promulgated new emissions standards for diesel
engines and concluded that these standards could only be met if diesel fuel for
on-road engines were also reduced. Starting in 2006 and phasing in fully by
2010, diesel fuel for on-road engines must have a sulfur level below 15 ppm.
Starting in 2010, diesel fuel for nonroad engines, such as farm and construction
equipment, must also be below 15 ppm. Manufacturers are expected to
target refinery release limits of approximately 8 ppm to take into account the
sulfur picked up in pipelines and other parts of the distribution system. This margin
may change over time as experience is gained in shipping low sulfur products.
Ash Content. The fuel injectors of diesel engines are designed to very
close tolerances and are sensitive to any abrasive material in the fuel. Therefore,
the maximum permissible ash content of the fuel is specified. This is measured
by ASTM D 482, which consists of burning a small sample of the fuel in a weighed
container until all the carbonaceous matter has been consumed. The permissible
amount of ash is between 0.01 and 0.1 wt% depending on the grade of diesel.
Low speed engines operating at constant speed and load can tolerate higher
levels of ash.
Aromatics Content. Aromatic compounds have very poor ignition quality
and, although they are not specifically limited in ASTM D 975, there are practi-
cal limitations to using high aromatic levels in highway diesel fuel. In the United
States, where gasoline demand represents about one-half of the crude barrel, and
where heating oil demand is relatively low, there is great pressure to be able to
blend aromatic FCC streams into diesel fuel. Average aromatic levels in the
United States are about 33%. In addition to having poor fuel quality, aromatics
also contribute to exhaust emissions. The federal government began effectively
limiting aromatic content to below 40% starting in October, 1993 by specifying
a minimum cetane index of 40. California limits aromatic levels below 10%
beginning in the same time period, also because of emissions concerns. It is
expected that average aromatic levels in ULSD will be lower than current levels
because the refinery processes used to reduce sulfur will also saturate some
aromatics. The magnitude of this change is difficult to predict.
Stability. Diesel fuel can undergo unwanted oxidation reactions leading to
insoluble gums and also to highly colored by-products. Discoloration is believed
to be caused by oxidation of pyrroles, phenols, and thiophenols to form quinoid
structures (71). Eventually, these colored bodies may increase in molecular
weight to form insoluble sludge.
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 263

Gums can lead to deposits in critical injector orifices that can degrade atomi-
zation and combustion performance of the engine. The chemistry involved in the
formation of these gums appears to be similar to that which occurs in gasolines.
Stability is measured using ASTM D 2274. In this test, a sample is heated at
958C for 16 hours while oxygen is bubbled through the liquid. After cooling,
the insoluble material is filtered and washed. The amount remaining is reported
as gums.
Lubricity. Diesel fuel has traditionally had adequate lubricity to protect
fuel pumps and injectors from excessive wear. Changes in equipment and fuel
processing have caused concern that lubricity may no longer be adequate. Pres-
sures in modern unit fuel injectors may be as high as 170 MPa (25 kpsi) and
tolerances for direct injection equipment have become much tighter. Lower visco-
sity fuels have poorer lubricity, and processing steps that lower sulfur may also
remove trace levels of surface active compounds that provide natural lubricity.
Lubricity is measured in two different tests: ASTM D 6078, the scuffing
load ball-on-cylinder evaluator (SLBOCLE); and ASTM D 6079, the high fre-
quency reciprocating rig (HFRR). In response to concerns voiced by engine and
equipment manufacturers, ASTM will likely adopt a maximum specification of
520 mm average wear scar in the HFRR test. Fuels which do not meet this
specification will have to be additized, either at the refinery or at the terminal.
ASTM specifications for different grades of diesel fuel are summarized
in Table 12.
3.3. Diesel Fuel Manufacture. The biggest factors in determining how
diesel fuel is blended in a given refinery are the availability of high cetane stocks.
In order of decreasing ignition quality, the hydrocarbon types rank in the follow-
ing order: normal paraffins, olefins, cycloparaffins, branched paraffins, and aro-
matics. Because straight-run distillates contain the greatest amount of normal
paraffins and cycloparaffins and the least amounts of branched paraffins and
aromatics, these are the preferred stocks for diesel blending. Cracked stocks,
which are relatively rich in aromatics, are less desirable from the standpoint
of ignition quality. However, these have high energy density and good cold
temperature properties. The amount of cracked material allowable in diesel

Table 12. Specifications for Diesel Fuel (ASTM D 975)


Grade

Fuel Property 1-D 2-D 4-D


flash point, 8C (min.) 38 52 55
T90, 8C 288, max 282-338
kinematic viscosity, mm2/s 1.3-2.4 1.9-4.1 5.5-24.0
carbon residue, % (max) 0.15 0.35
ash, % (max) 0.01 0.01 0.10
sulfur, % (max)a 0.50 0.50 2.00
cetane number (min) 40 40 30
a
Sulfur values are for regular grade diesel (RSD). Low sulfur diesel (LSD)
has a specification of 0.05%, and ultra low sulfur diesel (ULSD) has a sulfur
specification of 0.0015%.
264 GASOLINE AND OTHER MOTOR FUELS Vol. 2

fuel depends largely on the cetane number specification. In the United States,
where a high level of cracking is necessary to meet gasoline demand, the large
supply of cracked fractions and the relatively small supply of straight-run distil-
lates make substantial use of cracked stocks economically necessary. This has
been made possible through the use of cetane improvers to improve cetane and
through the use of hydrogenation to improve stability.
Ignition Improvers. In order to meet the increasing demand for diesel fuel
and to allow use of blendstocks having good low temperature properties but
low cetane number, diesel manufacturers frequently use cetane improvers in
the fuel. Cetane improvers work in just the opposite way that antiknock addi-
tives do. During the preignition period, the improvers generate free-radical
species that promote faster onset of combustion and reduce the ignition delay
properties of the base fuel. Various types of chemicals have been shown to be
effective, including nitrates and nitrites, nitro and nitroso compounds, and per-
oxides (71). Alkyl nitrates are by far the most common and are used in concen-
trations up to 0.3%. Although the effectiveness varies with base fuel composition,
0.1% in a typical fuel would give an increase of five numbers (72).
Stability Improvers. Diesel fuels that contain high amounts of cracked
stocks generally have poorer stability than virgin diesel. If the fuel is hydro-
treated to remove sulfur, the stability is vastly improved. Traditional antioxi-
dants such as hindered phenols are not particularly effective in preserving color
although these still prevent other oxidation reactions. Stabilizers are amines
or other nitrogen-containing compounds that prevent sediment formation by
interfering with the oxidation reactions that occur between heteroatoms
and available oxygen. These act as radical traps and/or peroxide decomposers.
Corrosion Inhibitors. The corrosion inhibitors used in diesel fuel are gene-
rally similar to those used in gasoline and, like the latter, produce an effect pri-
marily by surface action. If amine additives are used for detergency, these may
provide some corrosion protection as well.
Detergent Additives. Diesel engine deposits are most troublesome in the
fuel delivery system, ie, the fuel pump and both fuel side and combustion side of
the injectors. Small clearances and high pressures mean that even small
amounts of deposits have the potential to cause maldistribution and poor atomi-
zation in the combustion chamber. The same types of additives used in gasoline
are used in diesel fuel. Low molecular weight amines can also provide some cor-
rosion inhibition as well as some color stabilization. Whereas detergents have
been shown to be effective in certain tests, the benefit in widespread use is not
fully agreed upon (73).
Cold Flow Improvers. The cold flow properties of diesel fuel may be
improved by the use of additives or blending agents. As is the case for gasoline,
additives are materials present in low concentrations (ppm), whereas blending
agents are present in levels of a few percent or more. Blending agents work
by diluting the concentration of wax crystal forming paraffins to a point
below which these paraffins are no longer a problem. Kerosene and grade 1-D
diesel fuel may be used as effective diluents. Gasoline, which also improves
cold flow properties, is unsafe and dangerous to use as a blending agent
because it can drastically lower the flash point of the mixture and lead to poor
lubricity.
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 265

Additives are often the most economical way to improve the cold flow pro-
perties of middle distillates. These additives have been called by a number of
different names including wax crystal modifiers, mid-distillate flow improvers,
and wax antisettling additives. All additives work by modifying the crystalliza-
tion process in some way. Some change the shape of the wax crystals so that the
crystals are less likely to plug fuel system filters. Others work by promoting
nucleation of many small crystals instead of the growth of relatively few large
crystals (74). Various types of chemistry have been shown to be effective in im-
proving low temperature flow properties of diesel fuels. Examples are ethylene–
vinyl ester copolymers, chlorinated hydrocarbons, and polyolefins. The exact
nature of most commercially available additives is proprietary. Treatment rates
and the specific additive formulation used are a strong function of the fuel
composition and depend in large part on the distribution of molecular weights
of the normal paraffins in the fuel and the concentration of high molecular
weight paraffins (75). Fuels having a wide distribution of molecular weights
are the most responsive to flow improver additives.
Three types of cold flow additives have been described (76). The first are
pour point depressants, which are primarily low molecular weight ethylene–
vinyl ester copolymers. These work by cocrystallizing with the paraffin wax and
preventing further paraffin addition to existing crystals. The second type of addi-
tive is the cloud point depressant (CPD). CPDs are low molecular weight poly-
mers where the paraffin segments have been designed to interact with the
paraffins in the fuel to delay the onset of crystallization. The third type of addi-
tive is called an operability additive (OA). OAs are multicomponent additives
optimized for given fuel properties and show an improvement in the cold filter
plugging point and low temperature flow tests as well as lowering the pour point.
Lubricity Improvers. Lubricity performance of modern low sulfur, low aro-
matic diesel fuel in modern diesel engines is a concern. Some of these concerns
surfqaced in Sweden with the introduction of Swedish Class 1 and Class 2 diesel
in the early 1990s. A number of additives are available which can improve
lubricity as measured in ASTM D 7078 mand D 6079 at concentrations of
100–200 ppm. Use of fatty acid methyl esters has also been shown to improve
lubricity.
Dyes. Azo type dyes are fairly common in diesel fuel, and red dye is used
to identify fuel for nonroad applications, which is not subject to highway taxes.
This practice makes it easy to identify cheating and to enforce tax regulations.
Any red diesel found in nonroad equipment is evidence of cheating. Clear diesel
found in nonroad equipment usually means that on-road fuel was downgraded at
some point in the distribution system.
Antifoamants. When diesel fuel is splashed into tanks, it can foam up to
such an extent that the foam can splash onto the customer or onto the ground.
Antifoamants are silicon based additives that reduce foaming tendnecies so that
a high fueling rate can be maintained. Treat rates are usually below 30 ppm.
Oxygenates. Oxygenated materials have been considered for addition to
diesel fuels for the same reasons these compounds are added to gasoline. Put-
ting oxygen in the fuel should lead to better combustion and reduced emissions
of CO and, more importantly, particulates. However, use of oxygenates has
also been shown to increase emissions of nitrogen oxides. A number of different
266 GASOLINE AND OTHER MOTOR FUELS Vol. 2

oxygenates have been tested, including esters; ethers (qv); and alcohols. The
most common oxygenates used or advocated for use in diesel fuel are fatty
acid methyl esters (FAME). FAME can be made from vegetable oils such as
soy (SOME), or rapeseed oil (RSME), or from animal fat, such as cooking
grease. The term biodiesel generally represents FAME made from a renewable
oil. If pure biodiesel is used, it is called B100. Other concentrations, such as B5
or B20, can be used too.
FAME has a number of advantages. It has a high cetane number and good
lubricity. Emissions of hydrocarbons, and particulates are significantly lower
than standard diesel fuel. Finally, use of biodiesel can reduce emissions of green-
house gases.
FAME also has a number of disadvantages. Emissions of NOx are higher
than commercial diesel. Cold flow and stability may also be worse.
Generally, engine manufacturers have supported FAME use at levels below
5%, if the material meets the requirements of ASTM D 975 and if the ester meets
the requirements of ASTM D 6751. Use of biodiesel has been limited primarily
because of its high cost.
3.4. Diesel Environmental Regulations. Emission standards have
been set for heavy-duty vehicles in much the same manner as they have been
set for gasoline engines. Because heavy-duty vehicles are primarily diesels, the
focus is on diesel engine emissions. Standards have been written in units of
grams per brake-horsepower-hour (g/bhph ¼ g/kW h  1.34), which normalize
the emissions according to the total energy output of an engine over the specified
driving cycle. In contrast to light-duty vehicle testing where testing is carried
out on the total vehicle, heavy-duty engines are certified in tests on an engine
dynamometer. A series of accelerations is carried out and the emissions are
measured. Table 13 shows U.S. emissions standards. For heavy-duty engines,
the most difficult emissions standards to meet are total particulates and NOx.
When standards were relatively high, they were met by engine redesign without
fuel changes. Since 1994, fuel changes have also been specified. Many studies
have been published on the interactions between engine design, fuel composition,

Table 13. Federal Heavy-Duty Truck Exhaust Emission


Standards, (g/kW h)a
Year HC CO NOx HC þ NOx Particulates
1970 4.89 47.4
1974 30 11.9
1979 1.1 18.7 7.5
1984 0.97 11.6 8.0 3.7
1985 0.97 11.6 8.0
1988 0.45
1990 4.5
1991 3.7 0.19
1994 0.07
1998 3.0
2007b 0.10 0.15 0.007
a
To convert g/kW h to g/bhp-hr, multiply by 1.340.
b
2007 standards for NOx phase in over three years.
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 267

and emissions, and the conclusions are not always in agreement. It is generally
agreed that sulfur contributes to sulfate particulate emissions and that lowering
sulfur helps to meet the particulate standards. Higher aromatics, especially
PAHs (polycyclic aromatic hydrocarbons) are known to contribute to organic
carbon based particulates. Particulate emissions have also been shown to be
affected by fuel density (77). The impact of cetane has been shown to be positive
and negative. In older heavy duty engines, higher cetane was generally cor-
related with lower levels of NOx and particulates. In modern heavy duty engines,
this effect seems to be negligible. In modern European light duty diesels, higher
cetane can actually increase particulate emissions.
It is also generally agreed that in terms of engine design parameters, there
is an inverse relationship between NOx and particulates. Rapid, complete combu-
stion reduces particulate emissions but also promotes the formation of NOx.
Exhaust gas recycle (EGR), which lowers NOx emissions, generally causes
particulate emissions to increase.
It is expected that diesel engine manufacturers will use exhaust aftertreat-
ment devices to meet the NOx and particulate emissions standards that take
effect in 2007–2010. Particulates will likely be controlled by catalyzed traps; NOx
will be controlled either by NOx reduction catalysts that operate under lean con-
ditions or by catalysts that use a reductant such as urea. In anticipation of these
devices, and in anticipation that the devices would only work with extremely low
levels of sulfur, EPA enacted the ULSD regulations described above.
California has taken a slightly different approach to diesel fuel composition
standards from that of the federal government. In October, 1993, California limi-
ted diesel fuel to no more than 10% aromatics and 500 ppm sulfur. Alternative
formulations are possible if these are shown to have equivalent NOx emissions to
a base reference fuel. In addition to the specifications that apply to the commer-
cial fuel, other aspects of the reference fuel composition are tightly controlled.
The fuel must also have a minimum cetane number of 48 without cetane impro-
vers. Fuel manufacturers test the emissions from proposed alternatives and com-
pare them to the emissions from the reference fuel. For instance, manufacturers
may be able to produce higher aromatic fuel if the fuel also has higher cetane
number. In 2003, most diesel fuel in California used an alternative formulation.
Based on a small sample, average aromatics were 20%, average cetane was 51
and average sulfur was 40 ppm in the Los Angeles area in the Summer of 2003
(78). Starting in 2006, California is adopting the federal sulfur limit of 15 ppm
in addition to its other requirements.
In Europe, the European Union has defined diesel composition standards as
shown in Table 14.

Table 14. European Diesel Specifications


2000 2005 2009
sulfur, max, ppm 350 50 10
T95, max, 8C 360 360 360
cetane number 51 51 51
PNA, max, wt% 11 11 11
density, kg/m3 845 845 845
268 GASOLINE AND OTHER MOTOR FUELS Vol. 2

4. Alternative Fuels

Alternative fuels fall into two general categories. The first class consists of fuels
that are made from sources other than crude oil but that have properties the
same as or similar to conventional motor fuels. In this category are fuels made
from coal, shale and natural gas through Fischer-Tropsch synthesis. (see FUELS,
SYNTHETIC, GASEOUS FUELS). In the second category are fuels that are different from
gasoline and diesel fuel and which use redesigned or modified engines. These
include compressed natural gas (CNG), liquefied petroleum gas (LPG) and
hydrogen. The most probable use of hydrogen would be in fuel cells vehicles,
both of which will be discussed elsewhere.
During World War II, Germany developed large-scale production of fuel
from coal using Fischer-Tropsch synthesis. In the United States, in the 1970s,
interest in producing alternative fuels was generated by upheavals in the price
and availability of crude oil. In addition to coal, a great deal of research was
carried out to develop processes to develop fuels from oil shale and natural gas.
Production of synthetic fuels from coal, oil shale, and methane involves
changing the chemical structure of the raw material, especially the hydrogen–
carbon ratio. In coal, the hydrogen–carbon ratio is too low and the molecular
weight of the raw material is too high to burn cleanly. Coal can be converted
to syngas (CO/H2) and then converted to liquids via Fischer-Tropsch synthesis.
Alternatively, it may be liquefied directly with the addition of hydrogen. Oil
shale is retorted in large reactors or in situ to convert the solid organic material
(kerogen) into a synthetic crude oil. The oil is processed to resemble crude frac-
tions. Methane may be converted to hydrocarbon liquids or to methanol via gasi-
fication to syngas (CO/H2) followed by Fischer-Tropsch reactions. A principal
factor in all of these processes is the amount of energy needed to run the process
and the disposition of the waste products. In the case of shale oil, the spent rock
takes up more volume than the original shale and has to be disposed of in an
environmentally safe manner.
Decisions on commercialization depend on a number of factors, including
the price of the new energy source relative to the price of crude oil, need for
energy self-sufficiency, and environmental considerations. The factors are differ-
ent for each of the potential sources. In the case of natural gas, conversion to
Fischer-Tropsch liquids must compete with the production of liquefied natural
gas (LNG). Environmental considerations include pollution from burning the
fuel itself, pollution from the whole production cycle, and production of gases
that are involved in the enhanced greenhouse effect (see AIR POLLUTION). With
the dramatic improvements in the low emissions capability of gasoline vehi-
cles, the incremental improvement offered by some alternative fuels can be very
expensive and has tended to limit their introduction. Today, most alternative
fuels are being considered for their ability to reduce emissions of greenhouse
gases and the desire to reduce dependence on petroleum based fuels.
Fischer-Tropsch liquids made from natural gas (GTL) seem to be the most
likely fuel in this category to reach the market. GTL plants can produce a very
high quality fuel in the diesel boiling range. Sulfur and aromatics are essentially
zero, and cetane numbers can be as high as 75. The most efficient use of this
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 269

material will probably be as a blendstock in conventional diesel, and not as a fuel


in its own right.
4.1. Compressed Natural Gas. Compressed natural gas (CNG), which
contains mostly methane, is used in almost 1.5 million passenger cars and light
duty trucks worldwide and offers a number of clear environmental advantages. It
has a high octane number, has the potential to be clean burning, and is currently
shipped through pipelines to many parts of the country. Optimized engines
would have high compression ratios to take advantage of methane’s high octane
number. These vehicles store the CNG in high pressure cylinders made of light-
weight spun carbon fiber (see CARBON AND GRAPHITE FIBERS) and use specially
designed fuel injectors. Dedicated CNG vehicles are projected to have improved
energy efficiency and lower emissions of hydrocarbons, carbon monoxide and
carbon dioxide.
Unfortunately, CNG also has a number of disadvantages. The energy
density of methane is very low and it cannot be compressed enough to provide
equal volumetric energy density to gasoline. Producers of automobiles are
faced with either eliminating trunk space to provide equal range or compromis-
ing on both range and storage space. New infrastructure would have to be put
into place to refuel CNG vehicles. A service station refueling island would
could cost $500,000 or more, and refueling a car would take about 10 minutes.
A home compressor capable of overnight refueling would cost anywhere from
$3,000–8,000.
CNG has some potential environmental debits as well. To achieve maxi-
mum energy efficiency, CNG vehicles would probably operate on the lean side
of stoichiometry, and this makes catalytic control of NOx difficult. Methane is
also a potent contributor to enhanced global warming, being anywhere from
20–50 times worse than CO2. Exhaust emissions of methane and leaks in the dis-
tribution system must be tightly controlled. The exact leakage rate is not well
known but estimates range from less than 1% to almost 7%.
In the United States, CNG’s initial use is in captive, centrally refueled fleets
which require limited range, such as delivery vans, taxis, or school buses. Such
vehicle fleets can afford the capital cost of refueling equipment and can tolerate
the slightly longer refueling times. Fleet experience with CNG has been some-
what disappointing. CNG trucks and buses cost more than their diesel counter-
parts and have had poor reliability and high maintenance costs. In order to
reduce exhaust emissions, it may make more sense to use a limited budget
to purchase new, low emitting diesels, or to retrofit existing diesels with parti-
culate taps.
4.2. Liquefied Petroleum Gas. Liquefied petroleum gas (LPG) is a gen-
erally available fuel that is a mixture of mostly propane and butane. The exact
proportion depends on marketplace conditions and alternative outlets for the two
main components. LPG has a number of advantages over both gasoline and
methane. It has a higher energy density than methane, because LPG is a liquid.
It is easier starting than gasoline because of its higher vapor pressure. Gene-
rally, LPG vehicles vaporize the fuel before it is fed into the engine. In cold
weather, if the butane content is too high, the fuel may have trouble vaporizing.
The largest problem with LPG is that the supply is not large enough to make a
significant dent in the total fuel demand picture.
270 GASOLINE AND OTHER MOTOR FUELS Vol. 2

BIBLIOGRAPHY

‘‘Gasoline and Other Motor Fuels’’ in ECT 3rd ed., Vol. 11, pp. 652–695, by J. C. Lane,
Ethyl Corp.; ‘‘Refinery Processes, Survey’’ under ‘‘Petroleum’’ in ECT 3rd ed., Vol. 17,
pp. 183–256, by C. E. Jahnig, Consultant; in ECT 4th ed., Vol. 12, pp. 34–388, by Albert
M. Hochhauser, Exxon Research and Engineering Co.; ‘‘Gasoline and Other Motor Fuels’’
in ECT (online), posting date: December 4, 2000, by Albert M. Hochhauser, Exxon
Research and Engineering Co.

CITED PUBLICATIONS

1. Annual Energy Review 2002, Report No DOE/EIA-0384(2002), Energy Information


Administration, US Dept. Of Energy, Washington, D.C., Oct. 2003.
2. ibid., pp. 127, 131.
3. ibid., pp. 61, 173.
4. 2004 Annual Book of ASTM Standards, Vol. 5.02, ASTM International, West
Conshohocken, Pa., 2004.
5. 2004 Annual Book of ASTM Standards, Vol. 5.01, ASTM International, West
Conshohocken, Pa., 2004.
6. Annual Energy Review 2002, Report No DOE/EIA-0384(2002), Energy Information
Administration, U.S. Dept. Of Energy, Washington, D.C., Oct. 2003, p. 152.
7. Highway Statistics Summary to 1995, U.S. Dept. Of Transportation, Federal
Highway Administration, Office of Highway Policy Administration, Washington,
D.C., April 1997, Table VM-201a; Highway Statistics 1997, U.S. Dept. of Transporta-
tion, Federal Highway Administration, Office of Highway Policy Administration,
Washington, D.C., Oct. 1998, Table VM-1; Highway Statistics 1999, U.S. Dept. Of
Transportation, Federal Highway Administration, Office of Highway Policy Adminis-
tration, Washington, D.C., Oct. 2000, Table VM-1; Highway Statistics 2001, U.S.
Dept. of Transportation, Federal Highway Administration, Office of Highway Policy
Administration, Washington, D.C., Oct. 2002, Table VM-1 Highway Statistics 2002,
U.S. Dept. Of Transportation, Federal Highway Administration, Office of Highway
Policy Administration, Washington, D.C., Nov., 2003, Table VM-1.
8. S. K. Skinner, National Transportation Statistics, Annual Report, 1990, U.S. Dept. of
Transportation, Washington, D.C., 1990, p. 124. U.S. Dept. Of Transportation, Bureau
of Transportation Statistics, National Transportation Statistics 2002, BTS02-08,
Washington, D.C., U.S. Government Printing Office, Dec. 2002, Table 4–23.
9. Annual Energy Outlook 2004 Report # DOE/EIA-0383(2004), Energy Information
Administration, U.S. Dept. of Energy, Washington, D.C., Jan., 2004.
10. J. B. Heywood, Internal Combustion Engine Fundamentals, McGraw-Hill Book Co.,
Inc., New York, 1988.
11. A. M. Douaud and P. Ezyat, SAE Tans. 87, 294–308 (1978).
12. J. D. Benson and G. Dana, SAE Paper No. 2002-01-2894, SAE, Warrendale, Pa., 2002.
13. Knocking Characteristics of Pure Hydrocarbons, ASTM Special Publication No. 225,
American Society of Testing Materials, Philadelphia, Pa., 1958.
14. G. H. Unzelman, Oil Gas J., 33–37 (Apr. 10, 1989).
15. CRC Octane Acceleration Workshop/Interlaboratory Study, CRC Report No. 634,
Coordinating Research Council, Alpharetta, Ga., 2003.
16. G. T. Kalghatgi, SAE Paper No. 2001-01-3584, SAE, Warrendale, Pa., 2001.
17. G. T. Kalghatgi, SAE Paper No. 2001-01-3585, SAE, Warrendale, Pa., 2001.
18. E. S. Corner, A. M. Hochhauser, and H. F. Shannon, SAE Trans. 87, 1151–1527 (1978).
Vol. 2 GASOLINE AND OTHER MOTOR FUELS 271

19. W. E. Bettoney and co-workers, SAE Trans. 89(4), 4198–4212, (1980).


20. Y. Nakajima and T. Sato, SAE Paper No. 700612, 1970.
21. 2003 CRC Intermediate Temperatue Volatility Program, CRC Report No. 638, Coordi-
nating Research Council, Alpharetta, Ga., 2004.
22. K. Owen and T. Coley, Automotive Fuels Handbook, Society of Automotive Engineers,
Warrendale, Pa., p. 116.
23. J. N. Bowden and D. W. Brinkman, Stability Survey of Hydrocarbon Fuels, Report
BETC/17784, U.S. Dept. of Energy, Washington, D.C., 1979.
24. A. A. Johnston and E. Dimitroff, SAE Paper No. 660783, SAE, Warrendale, Pa., 1966.
25. H. W. Marbach and co-workers, SAE Paper No. 790204, 1979.
26. A. D. Reichle, Chem. Eng. Progress, 70–74, Sept. 1990.
27. J. R. Murphy, Oil Gas J., 49–58, (May 18, 1992).
28. A. A. Avidan, Oil Gas J., 59–66, (May 18, 1992).
29. A. A. Murcia, Oil Gas J., 68–71, (May 18, 1992).
30. Annual Energy Review 2002, Report No DOE/EIA-0384(2002), Energy Information
Administration, U.S. Dept. of Energy, Washington, D.C., Oct. 2003, p. 42.
31. Iowa Department of Natural Resources website, www.iowadnr.com/news/03jan/
e10.html
32. Monthly Motor Fuel Consumption, Nebraska Dept. of Revenue, Motor Fuels Division,
Lincoln, Neb., Nebraska Energy Office, Lincoln, Neb.
33. U.S. Environmental Protection Agency, Fed. Reg. 46(144) (July 28, 1981).
34. U.S. Environmental Protection Agency, Fed. Reg. 56(28), 5352 (Feb. 11, 1991).
35. Alcohols and Ethers, A Technical Assessment of Their Application as Fuels and Fuel
Components, API Publication 4261, 3rd ed., American Petroleum
Institute, Washington, D.C., 2001.
36. C. J. Pedersen, Ind. Eng. Chem. 48(10), 1881, Oct. 1956.
37. P. Polss, Hydrocarbon Process. 52(2), 61 (Feb. 1963).
38. E. W. Unruh and F. M. Watkins, Oil Gas J. 47 63 (June 17, 1948).
39. L. M. Gibbs, SAE Trans., J. Fuels Lubricants 99(4), 618–638, (1990).
40. National Association of Corrosion Engineers, NACE Standard TM0172, Houston,
Tex., 1972.
41. U.S. Dept. of the Air Force, Specification MIL-F-25017d for Fuel Soluble Corrosion
Inhibitor, Wright Patterson AFB, Dayton, Ohio, May 1981
42. H. W. Sigworth and J. Q. Payne, SAE Paper No. 405, SAE, Warrendale, Pa., 1954.
43. J. D. Benson and P. A. Yaccarino, SAE Trans 95(6), 562–579 (1986).
44. R. C. Tupa, SAE Paper No. 872113, SAE, Warrendale, Pa., 1987.
45. D. L. Hilden, SAE Trans. 97(3), 3.847–3.862 (1988).
46. C. Kim, S. I. Tseregounis, and B. E. Scruggs, SAE Trans. 96, 7.617–7.629 (1987).
47. G. T. Kalghatgi, SAE Trans. J. Fuels Lubricants 99(4), 639–667 (1990).
48. T. Nishizaki and co-workers, SAE Paper No. 790203, SAE, Warrendale, Pa., 1979.
49. T. J. Bond, F. S. Gerry, and W. Wagner, SAE Paper No. 892115, SAE, Warrendale,
Pa., 1989.
50. B. Bitting and co-workers, SAE Trans. 96(7), 639–655 (1987).
51. G. T. Kalghatgi, SAE Paper No. 952443, SAE Warrendale, Pa., 1995.
52. Effects of Combustion Chamber Deposits on Vehicle Emissions and Fuel Economy,
CRC Report No. 624, Coordinating Research Council, Alpharetta, Ga., 2000.
53. N. R. Balysky and co-workers, SAE Paper No. 2001-01-1997, SAE Warrendale,
Pa., 2001.
54. R. C. Tupa and C. J. Dorer, SAE Trans 95(6), 340–374, 1986.
55. T. J. Sheahan, C. J. Dorer, and C. O. Miller, SAE Paper No. 690516, SAE,
Warrendale, Pa., 1969
56. W. E. Morris, Oil Gas J. 84(3), 63 (Jan. 20, 1986).
272 GASOLINE AND OTHER MOTOR FUELS Vol. 2

57. W. E. Morris, The Interaction Approach to Gasoline Blending, paper presented at


NPRA Annual Meeting, San Antonio, Tex., Mar. 23–25, 1975.
58. W. C. Healy, Jr., C. W. Maassen, and R. T. Peterson, A New Approach to Blending
Octanes, paper presented at 24th Mid-year Meeting of the American Petroleum Insti-
tute’s Division of Refining, New York, May 27, 1959.
59. R. D. Snee, Technometrics 23(2), 119–130 (May 1981).
60. Fed. Reg. FR40CFR part 600 51(206), 37851 (Oct. 24, 1986).
61. A. M. Hochhauser and co-workers, SAE Paper No. 930138, SAE, Warrendale,
Pa., 1993
62. Clean Air Act Amendments of 1990, PL 1639, U.S. Environmental Protection Agency,
Washington, D.C., Oct. 26, 1990.
63. V. R. Burns and co-workers, SAE Trans. J. Fuels Lubricants 100(4), 687–714 (1991).
64. A. M. Hochhauser and co-workers, SAE Trans. J. Fuels Lubricants 100(4), 748–788
(1991).
65. W. J. Koehl and co-workers, SAE Trans. J. Fuels Lubricants 100(4), pp. 789–802 (1991).
66. R. M. Reuter and co-workers, SAE Trans. J. Fuels Lubricants 101(4), 463–484 (1992).
67. Ozone Forming Potential of Reformulated Gasoline, National Research Council,
National Academy Press, Washington, D.C., 1999.
68. Annual Energy Review 2002, Report No DOE/EIA-0384(2002), Energy Information
Administration, U.S. Dept. of Energy, Washington, D.C., Oct. 2003, p. 152.
69. Ibid. p. 155.
70. Health Assessment Document for Diesel Engine Exhaust, U.S. Environmental Protec-
tion Agency, Office of Research and Development, National Center for Environmental
Assessment, Washington, D.C., EPA/600/8-90/057F, 2002.
71. R. C. Tupa and C. J. Dorer, SAE Trans 95(6), 340–374, 1986.
72. T. R. Coley and co-workers, SAE Paper No. 861524, SAE, Warrendale, Pa., 1986.
73. J. E. Benethum and R. E. Windsor, SAE Trans. J. Fuels Lubricants 100(4), 803–809
(1991).
74. N. Feldman, SAE Paper No. 730677, SAE, Warrendale, Pa., 1973.
75. S. R. Reddy and M. L. McMillan, SAE Trans. 90, 3598 (1981).
76. J. E. Chandler, F. G. Horneck, and G. I. Brown, SAE Paper No. 922186, SAE,
Warrendale, Pa., 1992.
77. W. W. Lange, SAE Trans., J. Fuels Lubricants 101(4), 1233–1256, (1991).
78. North American Fuel Survey, Alliance of Automotive Manufacturers, Southfield,
Mich., 2003.

GENERAL REFERENCES

J. B. Heywood, Internal Combustion Engine Fundamentals, McGraw-Hill Book Co., Inc.,


New York, 1988.
K. Owen and T. Coley, Automotive Fuels Handbook Second Edition, Society of Automotive
Engineers, Warrendale, Pa., 1995.
Automotive Handbook, 5th Edition, Robert Bosch GmbH, Stuttgart, Germany, 2000.
L. M. Gibbs, SAE Trans. J. Fuels Lubricants 99(4), 618–638 (1990).
G. T. Kalghatgi, SAE Trans. J. Fuels Lubricants 99(4), 639–667 (1990).
R. Tupa and C. J. Dorer, SAE Trans. 95(6), 340–374 (1986).

ALBERT M. HOCHHAUSER
Exxon Mobil Research and Engineering Co.
Vol. 2 AVIATION TURBINE FUELS 273

AVIATION TURBINE FUELS


1. History

Turbine-powered aircraft were used by both British and German air forces for
military operations during the latter part of World War II. The British engines
were operated on illuminating kerosine containing 1 % light lubricating oil (jet
engine oil) (1). The German engines burned a wider boiling mixture of gasoline
and heavier distillate fractions. Fuel development in Germany stopped at war’s
end. Continuing British research has tightened some requirements over the
years, yet today’s primary British specifications still describe kerosine-type
fuels for both commercial and military aircraft, the main difference being in
the additive content.
The initial United States military fuel was a low freezing point kerosine
called JP-1, but its limited availability forced the development of a wider boiling
fuel that included gasoline as well as kerosine fractions (2). This ‘‘wide-cut‘‘ fuel
called JP-4 was, until recently, the major military grade. However, safety concerns
associated with its flammability led to its phase-out in favor of the kerosine based
grade JP-8. The switch from JP-4 to JP-8 was largely completed by the end of 1996.
Commercial aircraft have been burning mostly kerosine-type fuels since
1953 when the early turboprop aircraft came into service, and this has remained
the major civil fuel in the turbojets and later the turbofans which power most of
the world’s civil aircraft fleet. The principal grades are Jet A in the United
States, TS-1 in Russia and Jet A-1 in the rest of the world. The wide-cut fuel
Jet B, analogous to the military grade JP-4, is now rarely used in civil aviation.

2. Fuel Types and Specifications

The most widely used civil aviation turbine fuel grades are defined by a small
number of key specifications as listed in Table 1. These specifications are similar
in most respects although a small number of significant differences remain.
The US Jet A fuel, defined by specification ASTM D1655 (3), differs from Jet
A-1 only in freezing point. The British DEF STAN 91-91 specification (formerly
well known as DERD 2494 whilst under the jurisdiction of the UK Ministry of
Defence) (4) is recognized as a civil jet fuel specification and is used as such in
many countries around the world. The Russian GOST 10227 specification also
defines a grade designated RT which is very similar to western Jet A-1, as is
the Chinese no.3 Jet Fuel.
In addition to these prime specifications, the International Air Transport
Association (IATA) maintains fuel ‘‘Guidance Material’’ (5) for Jet A-1 and
Jet B grades for the use of its member airlines. Because of the worldwide nature
of commercial aviation, the close correspondence between specifications is a
necessity to discourage the proliferation of many different national or local
specifications. The major jet fuel suppliers subscribe to a document entitled
‘‘Aviation Fuel Requirements for Jointly Operated Systems’’ (AFQRJOS) which
defines a ‘‘Check List’’ which comprises the most restrictive requirements of the

ß2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim


10.1002/14356007.a03_201
274 AVIATION TURBINE FUELS Vol. 2

Table 1. Major civil jet fuel grades and their specifications


Grade Area of usage Major specification
Jet A-1 world-wide ASTM D1655, DEF STAN
91-91 (DERD 2494)
Jet A USA only ASTM D 1655
TS-1 Russia and former GOST 10227
Eastern Bloc countries
Jet Fuel no.3 China GB 6537
Jet A-1 Canada CAN/CGSB-3.23

ASTM D 1655, DEF STAN 91-91 fuel specifications and IATA Guidance Mate-
rial. Thus fuel meeting the Check List automatically meets all the requirements
of the parent specifications. This document was developed to cater for the situa-
tion where many different fuel suppliers deliver fuel into the common storage
facilities which predominate at most international airports.
Further efforts are underway to minimize the differences between the
major fuel specifications with the eventual aim of developing a single, interna-
tional specification to define aviation turbine fuel. An International Civil Jet
Fuel Specification Liaison Group was formed in 1989 to promote this activity
and a growing number of national specification bodies are contributing to its
progress.
2.1. Specification Requirements. Details of the major fuel specifi-
cations for Jet A-1 type fuel are shown in Table 2. Although test methods in
most western countries are either identical or very similar, no international
standardization exists and direct comparisons between limits may be misleading
(refer to the notes applying to Table 2). Test method designations published by
the principal standards organizations are given in Table 3.
2.2. Fuel Properties. Unlike most chemicals but like other petroleum
fuels, aviation turbine fuels are not defined by composition, molecular structure,
and purity. Instead their characteristics are limited by boundary conditions
imposed by inspection tests. Some of these tests are fundamental, others empiri-
cal; many reflect a solution to past engine or aircraft problems. However, any fuel
meeting all requirements is acceptable for aircraft operation. The specifications
are dynamic, but changes are conservative because of the potential risk to avia-
tion operations. Long equipment life ( 20 years) and certification restrictions
(see Chap. 5) also mitigate against drastic specification changes.
The individual specification properties or requirements can be divided
rather arbitrarily into bulk and trace properties. Bulk properties are those
requiring significant composition changes, involving more than 5 vol% of the
total fuel, to effect a change in the property. Trace properties tend to respond
to changes < 1 vol% or even 1 ppm or less. Changes in bulk properties affect pro-
duct yield, whereas changes in trace properties do not. The grouping has no
direct bearing on fuel performance. Instead it helps assess effects of crude com-
position and manufacturing steps on product yield (8).
Bulk Properties. Density. The mass – volume relationship is important
because the quantity of fuel loaded onboard an aircraft is metered volumetri-
cally, but fuel consumption is measured by mass. This relationship is also used
Table 2. Major Fuel Specifications for Civil Jet A-1 Type Fuel
United States United Kingdom International Russiaa Chinab
Specification ASTM D1655 DEF STAN 91-91 Checklist GOST 10227-86 GB 6537-94
Issue 96b 2 16 Amendment 4 effective June 95
Date Jul. 96 May 96 Jun. 96 Apr. 96 Aug. 94
Grade Jet A-1 Jet A-1 Jet A-1 TS-1 No.3 Jet Fuel
Composition
Acidity, total, mg KOH/g (max) 0.10 0.015 0.015 0.009c 0.015
Aromatics, vol % (max) 25 22.0 b 22.0 b 22 mass % 20.0
Olefins, vol % (max) NRe NRe NRe iodine no.2.5 max 5.0
Sulfur, total, mass % (max) 0.30 0.30 0.30 0.25 0.20
Sulfur, mercaptan (thiol), 0.003 0.003 0.003 0.005 0.002
mass % (max)
or
Doctor test negative negative negative NRe negative

Volatility

275
Distillation:
Initial boiling point (IBP), 8C (max) NRe report report 150 report
10 % vol recovered at, 8C (max) 205 205 205 165 205
20 % vol recovered at, 8C (max) NRe NR report NRe report
50 % vol recovered at, 8C (max) report report report 195 232
90 % vol recovered at, 8C (max) report report report 230 report
Final boiling point (FBP), 8C (max) 300 300 300 98 % (250 8C) 300
Residue, vol % (max) 1.5 1.5 1.5 NRe 1.5
Loss, vol % (max) 1.5 1.5 1.5 NRe 1.5
Flash pointf , 8C (min) 38 40 (38) 40 (38) 33 (28) 38
Density at 15 8C , kg/m3 775–840 775–840 775–840 775 (min) (20 8C) 775–830 (20 8C)

Fluidity
Freezing point, 8C (max)  47  47  47  50 g  47
Viscosity at 20 8C , mm2/s (min) NR NR NR 1.25 1.25
Viscosity at  20 8C, mm2/s (max) 8.0 8.0 8.0 8 ( 40 8C ) 8.0
Table 2. (Continued)
United States United Kingdom International Russiaa Chinab

Combustion
Specific energy (net), MJ/kg (min) 42.8 42.8 42.8 42.9 42.8
Smoke point, mm (min) 25 25 25 25 25
or
Smoke point, mm (min) 18 19 19 NRe 20
and naphthalenes, vol % (max) 3.0 3.0 3.0 NRe 3.0
or
Luminometer no. (min) 45 45 45 NRe 45

Corrosion
Copper corrosion (max) 1 1 1 passes 1
Silver corrosion (max) NRe 2h 2 NRe 1

Stability
Thermal stability: not comparable

276
JFTOT test temp., 8C 260i 260 260 260
Filter pressure drop, mm Hg (max) 25 25 25 3.3 kPa
Tube deposit (visual) rating LT 3 LT 3 LT 3 LT 3
Peacock/abnormal deposits fail fail fail fail

Contaminants
Copper content ** (max) NRe NRe NRe NRe 150 mg/kg
Existent gum, mg/100mL (max) 7 7 7 5 7
Water reaction
Interface rating (max) 1b 1b 1b 1 1b
Separation rating NRe NRe NRe NRe report
Microseparometer (MSEP) (min) NRe 85 (þ SDA ¼ 70) 85 (þ SDA ¼ 70) NRe NRe
Particulates, mg/L (max) NRe NRe NRe none report
Ash content, mass % (max) NRe NRe NRe 0.003 NRe
Electrical conductivityk *, pS/m 50–450 50–450 50–450 50–600 50–450 (20 8C)
Additives
Antioxidant (hydroprocessed fuels) optional mandatory mandatory mandatory optional
Antioxidant (other fuels) optional optional optional optional optional
Metal deactivator optional optional optional NRe NR e
Static dissipator optional mandatory mandatory optional optional
Fuel system icing inhibitor permitted permitted permitted permitted permitted
Lubricity improver permitted permitted permitted mandatory in HP permitted
a
Specification limits are in terms of Russian GOST test methods and may not be directly comparable with limits obtained using ASTM test methods.
b
Specification limits are in terms of Chinese GB/T or SH/T test methods and may not be directly comparable with limits obtained using ASTM test methods.
c
Value in terms of mg KOH/g calculated from GOST limit which is in terms of mg KOH/100 mL.
d
25.0 % max aromatics are permitted provided that hydrogen content is reported also.
e
NR = no requirement under this heading.
f
Flash point limits are in terms of ASTM D56 (TAG) test method; limits in parantheses are in terms of other, local, test methods.
g
The GOST test method records wax appearence point on cooling which is 3 8C below the wax disappearance point on warming, as recorded in the ASTM
method. Thus the specification limits for TS-1 and Jet A-1 are equivalent.
h
It is planned to delete the silver corrosion clasification requirement by June 1997.
i
If the test at 260 8C produces a failure, a pass at a test temperature of 245 8C is acceptable.
j
Applies only if fuel produced by copper-sweetening process.
k

277
Conductivity limits apply only if fuel contains static dissipator additive.
278 AVIATION TURBINE FUELS Vol. 2

Table 3. National Standard Aviation Fuel Test Methods


Test name ASTM (3) IP (6) ISO (7) GOSTa *
Flash point (TAG closed cup) D56 6356
Distillation D86 123 3405 : 1988 2177
Corrosion (copper strip) D130 154 2160 : 1990 6321
Color (Saybolt) D156
Vapor pressure (Reid) D323 69 3007 : 1986 1756
Existent gum D381 131 6246 : 1981 1567
Kinematic viscosity D445 71 3104 : 1994 33
Water reaction D1094 289 6250 : 1982 27154
Sulfur (lamp method) D1266 107 19121
Density (hyrometer) D1298 160 3675 : 1993 3900
Aromatics D1319 156 3837 : 1993 6994
Smoke point D1322 57 3014 : 1993 4338
Sulfur (high temperature method) D1552
Luminometer no. D1740 17750
Naphthalene hydrocarbons D1840 17749
Particulate contamination (field method) D2276 216
Freezing point D2386 16 3013 : 1974 5066
Sulfur (X-ray method) D2622
Electrical conductivity D2624 274 6297:1984 25950
Sulfur, mercaptan D3227 342 3012 : 1991 17323
Undissolved water D3240
Thermal oxidation stability (JFTOT) D3241 323 6249 : 1984 17751
Total acidity D3242 354 5985
Estimated heat of combustion D3338
Estimated hydrogen content D3343
Hydrogen content by NMR D3701 338
Peroxide no. D3703
Flash point (SetaFlash closed tester) D3828 303 3679/3680 : 1983
Water separation (MSEP) D3948
Density (Digital meter) D4052 365
Sulfur (ND-XRF method) D4294
Filter flow at low temperature D4305
Estimated heat of combustion D4529 381 3648 : 1994 11065
Heat of combustion (bomb) D4809 21261
Doctor test D4952 30 5275 : 1979
Vapor pressure (automatic) D5190
Vapor pressure (mini method) D5191
Particulate contamination (lab method) D5452
Sulfur (XRF method) D5453 336 8754:1992
Freezing point (auto/optical) D5901
Freezing point (auto/phase transition) D5972
Corrosion (silver strip) 227
Copper content 225
Sulfur (Wickbold combustion) 243 4260 : 1987
a
Although ASTM, IP and ISO methods are essentially identical, GOST methods may be quite different.

extensively in commercial sales of fuel. Although in the past there have been sev-
eral different ways of defining this relationship, such as specific gravity, relative
density, or API gravity, all the major jet fuel specifications now define density at
15 8C in units of kg/m3. Changes of density with temperature are given in (9).
Density is commonly measured using hydrometers or digital density meters.
Vol. 2 AVIATION TURBINE FUELS 279

In-line density measurement is also possible and offers the opportunity for fuel
delivery by mass instead of volume.
Specific Energy. Specific energy or heat of combustion is directly related
to density. The fuel’s specific energy, its density, the aircraft’s performance char-
acteristics, and the predicted flight conditions are employed to calculate the
amount of fuel required for a given flight or conversely to estimate the aircraft’s
range with maximum fuel aboard. Specific energy is determined by measuring
the total heat released by a unit mass of fuel and subtracting from this value
the heat of condensation of the water vapor formed during the combustion. How-
ever, this measurement is time consuming, so several empirical relationships
have been developed to calculate specific energy from combinations of other
inspection tests. The most common is based on the product of aniline point and
density, ASTM D 4529 (3); others use the density, boiling range, and aromatic
content, ASTM D 3338 (3), or the hydrogen content and density, IP 355 (6). All
three methods include corrections for fuel sulfur content. Although these calcu-
lation methods are used routinely, a method of measurement, ASTM D 4809 (3),
is still specified as the referee method for use in cases of dispute.
Volatility. A fuel’s volatility or its tendency to vaporize affects evapora-
tion rate from tanks, including aircraft fuel tanks. Volatility can cause vapor lock
in fuel systems when the combined fuel and vapor volume exceeds pump capa-
city; cavitation can result when the fuel vapor pressure exceeds the liquid system
pressure between gear teeth or in plain bearings. Volatility also directly affects
flammability and causes safety problems both inside and outside of tankage (10).
Because of the complex composition of turbine fuel, its volatility cannot be
described by a single fundamental property. Therefore, volatility is normally
expressed using a combination of boiling range, flash point, or vapor pressure.
The volatility of kerosine type fuels is controlled by flash point and distilla-
tion range (Table 2). Wide-cut fuel is regulated by a combination of distillation
range and vapor pressure.
Combustion. Combustion in a modern jet engine is a highly turbulent,
high-pressure,and high-intensity mixing process. Engine designers carefully
control air flow, fuel flow, and fuel atomization to maximize combustion efficiency
and to minimize smoke, carbon deposits, and the formation of partially oxidized
gases, such as carbon monoxide. Fuel affects all these as well as the intensity of
radiation emitted by the flame [11–13]. No single fuel property requires a greater
degree of empiricism than the estimation of fuel combustion characteristics.
Smoke point, and to a lesser extent the luminometer, have been favored because
as actual combustion tests, they respond to a combination of other fuel properties
including volatility, composition, and viscosity. However, a direct comparison of
these laminar, wick-fed flames with the high pressure, turbulent combustor
flame is not possible and thus other alternative controls also appear in the spe-
cifications. Naphthalenes content was added because multi-ring aromatics have
shown adverse combustion effects, in terms of flame radiation and smoke forma-
tion, out of proportion to their concentration (14). During the 1980s, the use of
hydrogen content, which can be measured very accurately by NMR (ASTM
D 3701), gained favor as a possible replacement for the less precise measure-
ments of aromatics and naphthalenes contents. However, it was found that
the total hydrogen content of the fuel could not predict reliably the combustion
280 AVIATION TURBINE FUELS Vol. 2

behavior of all fuels(15) and it is now unlikely that it will be adopted for
civil fuels, although it continues to find limited use in certain military fuel
specifications.
Total aromatic content (ASTM D1319, (3) ), covered under ‘‘Elastomer Com-
patibility’’, does not correlate particularly well with engine combustion, possibly
because of the particular separation and classification of hydrocarbons by the
method (16).
In general, correlations between engine combustion performance and fuel
inspection tests will always be difficult to establish because of the differing
performance of the same fuel in different engines, the extreme expense of
controlled engine tests, and the long time required to obtain engine experience
in the commercial world, where only a limited quantity of minimum quality
fuel is burned (16).
Elastomer Compatibility. A major requirement of jet fuels is their com-
patibility with materials of construction, particularly with the elastomers used in
diaphragms, fuel tanks, fuel tank coatings and sealants, linings, and a wide vari-
ety of static and dynamic seals in all parts of the fuel system. The fuel property
most indicative of elastomer compatibility has been total aromatic content.
Low-Temperature Operability. The extremely low temperatures at
high altitudes can cause several different fuel problems. During fuel cooling,
hydrocarbon molecules precipitate as wax crystals. n-Hydrocarbons are the
most insoluble, and in fact measurement of their type and concentration can
indicate the insoluble fraction expected at low temperature (17).
The freezing point of the fuel is defined as the temperature at which the
last wax crystal melts on warming the fuel following a specified cooling process.
Traditionally, this has been determined visually but now several automated
methods exist which employ variously, light scattering, phase transition mea-
surement or flow through a fine filter (ASTM D 4305) (3).
With the increasing use of twin-engined aircraft on long over-water routes,
the so-called extended operations (ETOPS) flights, and with the introduction of
very long range aircraft such as the Boeing 747–400, there has been increasing
focus on fuel low-temperature properties. Studies underway may question the
usefulness of ‘‘freezing point’’ as currently measured and could lead to its repla-
cement by a more meaningful, flow-based test.
Low fuel temperature can also affect engine starting, which depends
heavily on fuel atomization. Adequate atomization is determined largely by
fuel viscosity. Typical viscosities versus temperature are plotted in Figure 1.
Because fuels within the same grade have the same viscosity-temperature
slope, a single maximum viscosity at a standard temperature adequately controls
this property.
Trace Properties. Instead of describing trace properties individually,
a better overview is obtained by grouping these properties into the following
performance categories.
Oxidative Stability. Jet fuel’s exposure to oxidative conditions over a
wide temperature and time range can result in a wide variety of problems (19).
Ambient temperature oxidation can form peroxides that degrade certain
elastomers, eg Buna N, which are used in aircraft fuel systems. Long-term
storage can cause soluble and insoluble oxidation products labeled ‘‘gums’’.
Vol. 2 AVIATION TURBINE FUELS 281

Fig. 1. Typical viscosities vs. temperature (18).

Peroxide formation is usually seen in fuels which have been produced in the
refinery by catalytic hydrogen processes (see Chap. 3) which remove certain
trace compounds containing sulfur or nitrogen which are natural anti-oxidants.
Anti-oxidancy is restored by the addition of oxidation inhibitor additives
(Section 3.2). Gum formation is rare but is monitored and controlled by the
Existent Gum test.
High-temperature oxidation is of major concern. Oxidation products–gums,
lacquers and insoluble deposits – can form in the hot areas of an aircraft’s fuel
system causing a number of problems (20). The heat exchanger of the fuel-cooled
oil-cooler can lose efficiency; flow control valves in the fuel control unit can stick
or jam, coke deposits can form in the burner feed arms which can, in the worst
case, cause plugging of the fuel injector nozzles. These deposits represent such a
small fraction of the fuel that the oxygen dissolved in the fuel is sufficient to
cause their formation. Although deoxygenation is effective on a laboratory
scale, in large scale practice it is impossible to reduce the oxygen content of
the fuel to a level low enough to prevent oxidation. The problem is attacked by
designing aircraft fuel systems to minimize hot spots; indeed, thermal manage-
ment is one of the most important criteria in modern engine and aircraft design.
In addition, the thermal stability of fuel supplies is controlled by a test in the fuel
specification (Table 2) using an instrument called the Jet Fuel Thermal Oxida-
tion Tester (JFTOT), which is a small-scale simulation of a heat exchanger in
which formation of lacquer and deposit are measured (ASTM D 3241) (3). Cer-
tain metals, notably copper, are oxidation catalysts and when dissolved in the
fuel at extremely low levels, eg, 20 mg/kg (ppb), can promote thermal instability
and deposit formation. Traces of copper can be picked up by the fuel from copper-
containing alloys sometimes encountered in the refinery or during transporta-
tion. The deleterious effects of this dissolved copper can be alleviated by dosing
the fuel with a metal deactivator additive (Section 3.2).
Corrosivity. Corrosion of fuel system metals can be a problem. Copper or
bronze in early fuel pump bearings and other components needed protection (1).
282 AVIATION TURBINE FUELS Vol. 2

Silver surfaces in highly loaded portions of variable displacement pumps were


attacked and ultimately removed by shear forces (21). Dihydrogen sulfide and
elemental sulfur were identified as culprits. As little as 0.5 mg/kg (ppm) of either
material caused the difficulties. As a result, all the major fuel specifications con-
tain a test for copper corrosion. The DEF STAN 91-91 (DERD 2494) specification
has, in addition, a test for silver corrosion
Solutions to other corrosion difficulties have addressed the offending fuel
constituent rather than the metal involved. Thiols cause cadmium corrosion
(1) and are limited in fuel specifications. Sulfur, which includes all organosulfur
compounds, causes high-temperature turbine blade corrosion, particularly in the
presence of sodium (22). However, such low sulfur concentrations cause corrosion
that virtually complete sulfur removal from the fuel would be needed to solve the
problem, and blade metallurgy was changed instead. Therefore the current sul-
fur specification maxima are not based on any corrosion criteria.
Organic acidity, primarily as naphthenic acids, has been blamed for some
diesel engine corrosion, but the low-acidity level permitted by the jet fuel speci-
fications appears to have prevented similar aviation difficulties. The primary role
of organic acids is their involvement in the formation of surface-active agents as
discussed in the next paragraph.
Water Separation. Free or immiscible water is a potential major hazard
in aircraft. When cooled below 0 8C the resulting ice can block orifices, screens, or
filters. In large quantities water will ‘‘put out the fire’’ by interrupting fuel flow.
Water can lead to microbial growth in stagnant parts of the fuel system. There-
fore, every effort is made to remove all immiscible water by special filter-separa-
tors before the aircraft is fueled. Satisfactory operation of these filters depends
on the maintenance of internal hydrophobic surfaces. Trace levels of surfactants
render these surfaces ineffective (23). Although the organic acids normally in jet
fuel are not strong surfactants, they react with sodium hydroxide or other basic
materials. The resultant salts or soaps can have detrimental effects (1), (24), (25).
The lower organic acidity maximum in the British specification offers greater
protection in this regard than does the higher level in ASTM D 1655, but surfac-
tants are not limited to those formed by organic acids, and additional tests are
used for surfactant detection and control.
The Water Reaction test gives a semiquantitative indication of insoluble
interface material formed when fuel is shaken with buffered water. In a much
more sensitive test a fuel–water emulsion is pumped through two layers of
coated fiberglass from a filter-separator. The ability of the filter element to
strip all water from the emulsion is measured.
The earliest test, the Separometer (ASTM D 2550 (3)), was developed to
detect low levels of sodium sulfonate (33); but this test as well as the later ver-
sions, the Miniseparometer (ASTM D 3602) and the Microseparometer (ASTM
D 3948), are widely used in the United States and elsewhere to test for the pre-
sence of various surfactants. Because of the difficulties in correlating results
from these tests with field performance, there is only limited industry agreement
on the definition of acceptable fuel by these tests. ASTM D 2550 and D 3602 have
now been discontinued.
Static Electricity. Whenever a hydrocarbon liquid flows with respect to
another surface, an electrical charge is generated in the liquid and an equal but
Vol. 2 AVIATION TURBINE FUELS 283

opposite charge is imposed on that surface. One set of charges is swept along
with the fuel while the opposite charges, which accumulate along the wall sur-
face, usually leak to ground (earth). This charge separation results in a rise in
voltage in the moving fuel (26). High speed pumping of aviation fuel through
pipes and filters can generate very high charges and when the charged fuel
enters a tank, a substantial voltage difference may be produced between the sur-
face of the liquid and the tank walls and this may result in a static discharge (27).
If the vapor space above the liquid is in the flammable range, such a spark dis-
charge can cause a fire or explosion.
Charge build-up can be reduced by reducing the flow-rates of the fuel or
designing and constructing facilities — pipework, filters, valves, etc. – to mini-
mize charge generation and/or provide time for the charge to relax to ground
before entering a tank (28). Another approach is to add a static dissipater addi-
tive to the fuel (Section 3.2) to increase the electrical conductivity of the fuel to
facilitate charge relaxation.
2.3. Nonspecification Properties. Some aviation fuel properties are
not limited by specifications but are useful in equipment design or performance.
Absolute Vapor Pressure. Reid vapor pressure, specified for Jet B fuel,
is not an absolute vapor pressure because it is measured with a vapor–liquid
ratio of 10 : 1; it is determined only at 38.8 8C, and also it includes the partial
pressures of water vapor and air. With low-vapor-pressure products, such as
Jet A or A-1, absolute or true vapor pressure is used instead. Figure 2 shows typi-
cal true vapor pressure-temperature curves for different jet fuel grades (18).
Water Miscibility. Even the very low concentration of dissolved water
in aviation turbine fuels is hazardous if this water separates and freezes in a

Fig. 2. True vapor pressures (18).


284 AVIATION TURBINE FUELS Vol. 2

Table 4. Typical Nonspecification Properties (18)


20 8C 40 8C 60 8C
Water solubility, vol %
Jet A, A-1 0.0049 0.0088 0.0163
Jet B 0.0076 0.0138 0.0255
Specific heat, kJ kg1 K1
Jet A, A-1 1.95 2.04 2.13
Jet B 2.06 2.14 2.23
Thermal conductivity, Wm1 K1
Jet A, A-1 0.1155 0.1124 0.1085
Jet B 0.1155 0.1124 0.1085
Dielectric constant at 400 Hz
Jet A, A-1 2.13 2.10 2.07
Jet B 2.06 2.03 2.00
Surface tension, mN/m
Jet A, A-1 23.3 21.7 20.2
Jet B 21.8 20.5 18.4

critical aircraft location. Because dissolved water cannot be avoided in vented sys-
tems and is not removed by airport fuel handling equipment, aircraft fuel systems
either heat the fuel above 0 8C or depend on anti-icing additives (see Section 3.2).
Typical water saturation levels for commercial jet fuels are listed in Table 4 (18).
Lubricity. Various aircraft fuel system components – fuel pumps, servo
valves, fuel control units – rely on the fuel to lubricate their moving parts. The
effectiveness of a jet fuel as a lubricant in such equipment is referred to as its
‘‘lubricity’’. Differences in component design and materials result in varying
degrees of equipment sensitivity to fuel lubricity. Similarly, jet fuels vary in
their level of lubricity. The chemical and physical properties of jet fuel cause
it to be a relatively poor lubricating material under high temperature and load
conditions where lubrication may depend on adsorbed films of trace polar com-
pounds present in the fuel, so-called ‘‘boundary lubrication’’. Severe hydroproces-
sing (see Chap. 3) removes trace components, resulting in fuels which tend to
have lower lubricity than straight-run or wet-treated fuels. This lack of lubricity
can be remediated by the addition of corrosion inhibitor/lubricity improver addi-
tives (Section 3.2). Such additives are mandatory in most military jet fuel speci-
fications but they have not found widespread use in civil aviation. It is generally
acknowledged that the ultimate solution to the fuel lubricity ‘‘problem’’ is to
design fuel system equipment such that it is insensitive to fuel lubricity. Civil
fuel specifications do not contain a requirement for a minimum level of lubricity,
not least because there is no method of test for this property which is universally
accepted. One method, the Ball-on-Cylinder Lubricity Evaluator (BOCLE) ASTM
D 5001 (3), can be used for identifying possible problem fuels and is used by
equipment manufacturers to monitor special low-lubricity fluids used in develop-
ment testing of fuel system equipment, and by the USAF to monitor the lubricity
additive concentration in their fuel supplies.
Other Properties. Detailed data on other properties, including gas
solubility, flammability, surface tension, heat of vaporization, enthalpy, specific
heat, thermal conductivity, and dielectric constant, are contained in the CRC
Vol. 2 AVIATION TURBINE FUELS 285

Handbook of Aviation Fuel Properties (18). Some typical values for several prop-
erties are given in Table 4.

3. Production

3.1. Fuel. Since its earliest days virtually all aviation turbine fuel has
been refined from selected petroleum crudes by atmospheric distillation followed
by limited processing to reduce thiol or total sulfur. Crude selection is vital for
such simple processing because the only other variable available for controlling
bulk properties is adjusting the product’s boiling range. Desirable crudes are
low in sulfur (sweet crudes), low in n-hydrocarbons (nonwaxy crudes), low in aro-
matic content, and high in product boiling between 160 and 290 8C.
Adjusting either end of the boiling range has predictable results. Lowering
the initial boiling point improves combustion and low-temperature properties
but decreases handling safety (see Table 5). Raising the final boiling point
degrades combustion and low-temperature performance and has no effect on
handling safety. Because the crude also yields both lighter and heavier fractions,
the demand for these products affects distillation cut points and therefore the
volume of jet fuel available (29).
Total sulfur can be reduced by relatively low-pressure hydrotreating. Thiol
sulfur is reduced (sweetened) by low-pressure hydrotreating or wet chemical
treating, such as Merox, Linde, Doctor, or Bender treating. Flash point may be

Table 5. Alteration of Boiling Range


Lowering initial temperature Raising final temperature
Increases
smoke point density
hydrogen content aromatic content
J/mass naphthalenes
vapor pressure viscosity
Decreases
density smoke point
flash point hydrogen content
aromatic content J/mass
naphthalenes
freezing point
J/vol
viscosity
Leads to
better combustion poorer combustion
quality quality
better low-temperature poorer low-temperature
properties properties
poorer handling safety unchanged handling
safety
higher gravimetric lower gravimetric
specific energy specific energy
lower volumetric higher volumetric
specific energy specific energy
286 AVIATION TURBINE FUELS Vol. 2

adjusted so that the same product can be sold as either aviation turbine fuel or
commercial illuminating or heating kerosine.
Some heavier fractions are hydrocracked to make jet fuel, but the 160 –
290 8C fraction from catalytic cracking is not suitable for jet fuel unless cracking
is followed by high-pressure hydrotreating. A limited amount of kerosine is
hydrotreated to reduce aromatic content and increase smoke point by aromatic
saturation. In the United States the finished jet fuel is often filtered through
towers of activated Attapulgus clay to adsorb surfactants and to remove particu-
late fines.
3.2. Additives. All aviation fuel additives must be tested and approved.
The extent of testing is at the discretion of the aircraft and engine manufacturers
who issue the approval. Additive composition plays a major part in approval pro-
cedures, with ash-forming materials receiving the greatest scrutiny. Recom-
mended Practice D 4054 (ASTM) (3) covers the evaluation of compatibility
between additives and aircraft fuel system materials. Approved additives and
their permitted levels are normally listed in the fuel specifications. Unapproved
additives cannot legally be used in aircraft operation.
Oxidation Inhibitors. Oxidation inhibitors prevent peroxide formation
and improve storage stability in certain fuels (! Antioxidants). They are gener-
ally secondary and tertiary hindered phenols and are listed in the fuel specifica-
tion by chemical composition. They are effective primarily during storage, not
during high-temperature use. Antioxidants are mandatory for hydroprocessed
fuels in the major military fuel specifications and in civil Jet A-1 covered by
the DEF STAN 91-91 (DERD 2494) specification and IATA and Joint System
Checklist documents, but only optional in the ASTM D 1655 specification. Dosage
rate is 17 – 24 mg/L. Annual consumption of oxidation inhibitors in aviation tur-
bine fuel is difficult to estimate because the same additives are widely used in
other petroleum products.
Metal Deactivators. Metal deactivators are intended to passivate trace
metals by chelation (! Antioxidants). The only approved deactivator is N,N0 -dis-
alicylidene-1,2-diaminopropane [94-91-7]. It is permitted up to 5.7 mg/L, but is
rarely used.
Corrosion Inhibitors and Lubricity Improvers. Unlike the preceding
additives, corrosion inhibitor formulations are proprietary. They usually contain
dibasic acids that function as liquid phase rust inhibitors. The polar ends of the
molecule adsorb onto metal surfaces and in ground handling and distribution
systems, e.g., steel pipelines, they act to protect surfaces from corrosion. In air-
craft fuel systems, the tendency to adsorb on to metal surfaces is utilized for
lubricity improvement rather than corrosion inhibition. Because these additives
are surface-active (commonly referred to as ‘‘surfactant’’), they can display the
adverse side-effect of impairing the water separation properties of the fuel.
Furthermore, their surfactancy means that they can be lost from the fuel during
distribution (as, of course, they should be if they are to protect metal surfaces
from corrosion); consequently, if added for lubricity enhancement, these addi-
tives are often injected at the airport, just prior to aircraft fueling. Lists of
approved additives are published in the Qualified Products Lists (QPLs) asso-
ciated with the U.S. MIL-I-25017 and UK DEF STAN 68-251 specifications for
this type of additive.
Vol. 2 AVIATION TURBINE FUELS 287

Static Dissipater Additives (SDAs), also referred to as Conductivity


Improving Additives, are added to jet fuel to increase electrical conductivity,
and thereby reduce the tendency of a fuel to accumulate dangerous electrostatic
charge during handling (28). There are only two additives approved for use in
aviation fuel, Shell ASA-3 and Octel (formerly Du Pont) Stadis 450. Production
of the Shell additive ceased in 1991 leaving Stadis 450 as the only additive cur-
rently available. The concentration is adjusted to obtain an electrical conductiv-
ity within the range 50–450 pS/m but with a maximum permitted dosage of 3 mg/
L. A further addition up to 2 mg/L is permitted if conductivity has been lost dur-
ing fuel distribution. The minimum conductivity limit is set to ensure charge
relaxation sufficient to alleviate static discharges, while a maximum limit is
applied to avoid causing anomalies in aircraft fuel tank contents gauges which
operate on an electrical capacitance principle. The use of SDA is mandatory in
military JP-4 and JP-8 fuels and in civil Jet A-1 outside the United States; its
use is optional in Jet A fuel within the United States.
Anti-icing additive, also referred to as Fuel System Icing Inhibitor
(FSII), prevents the freezing of trace levels of free water in fuel at the low tem-
peratures encountered in aircraft fuel tanks at high altitudes and avoids block-
age of fuel lines and filters by ice crystals.). Traditionally, ethylene glycol
monomethyl ether (2-methoxyethanol) [109-86-4], has been used for this purpose
but, owing to its toxicity, this has largely been superseded by diethylene glycol
monomethyl ether (2-(2-methoxy)ethanol) [111-77-3]. Normal dosage rate is
between 0.1–0.15 vol %. Such use is mandated in military fuels but in civil avia-
tion its use is restricted to small aircraft which do not have heaters in their fuel
systems. The additive, which is only sparingly soluble in the fuel, preferentially
dissolves in any free water present (partition coefficient about 200 : 1) where it
acts as an antifreeze. This affinity for water can be problematic in fuel distribu-
tion systems and additive loss must be considered and possibly corrected. FSII
has the additional property of acting as a biostat, preventing the growth of bac-
teria and fungi in fuel tanks (see below).
Biocides. Microorganisms that metabolize hydrocarbons can be present
in both ground and aircraft fuel systems (1). Certain anaerobic bacteria that
reduce sulfate to hydrogen sulfide have caused copper and silver corrosion.
The formation of slimes, fungal mats, and corrosive byproducts has caused
wing tank corrosion and capacitance gage failures. The only generally approved
fuel-soluble biocide is Biobor JF (Hammonds), a blend of 95 % dioxaborinanes and
5 % naphtha. Recommended dosage to kill microorganisms is 226 mg/L; mainte-
nance level is 108 mg/L. An alternative material, which is approved for most
applications, is Kathon FP1.5 (Rohm & Haas) which is based on a mixture
of a methyl-isothiazolinone [2682-20-4] and a chloromethyl isothiazolinone
[26172-55-4]. Treatment levels are between 100 and 300 mg/L. Although treat-
ment with biocides kills the microbial growths in fuel tanks, subsequent tank
cleaning is still essential to remove all the dead material which might otherwise
lead to filter blockage.
Thermal Stability Additives. In the early 1990s, the USAF perceived
that future aircraft designs would demand a fuel with enhanced thermal oxida-
tive stability. They decided that the most economic way of achieving this would
be to use fuel additives to increase the thermal stability of the existing fuel type,
288 AVIATION TURBINE FUELS Vol. 2

JP-8, rather than developing a special fuel. It was calculated that an increase of
100 8F (56 8C) in the thermal stability level of the fuel would be required and this
led to the new fuel being referred to as ‘‘JP-8 þ 100’’. The extensive research pro-
gram to develop and test such additives was referred to as ‘‘HiTTS’’, standing for
High Temperature Thermal Stability (30). The main problems at high tempera-
ture are fouling of the injector feed arm and coking in the fuel supply rings to the
augmenter (afterburner or reheat) section. An additive package comprising a
detergent/dispersant, an antioxidant, and metal deactivator proved to be very
effective in preventing deposit formation in the hot areas of existing aircraft
fuel systems. Already the benefits in terms of reduced maintenance costs are
being appreciated. However, because the dispersant/detergent element of the
additive package is very surface active, it cannot be added to the fuel in storage
because it would encourage water and dirt pick-up and also disarm the filter/
water separators in the airport fueling system. Instead, the additive package
is injected into the fuel as the aircraft is being fueled.

4. Handling, Storage, and Transportation

System Descriptions. It is essential to prevent aviation fuel from being


contaminated with other petroleum products between the time it is manufac-
tured in the refinery until the moment it is put on-board an aircraft. Conse-
quently fixed storage facilities are designed to keep aviation fuel physically
isolated from other products. However, large pipelines, marine transports and
sometimes even road tankers are multiproduct systems. Each transportation
mode therefore has an elaborate protocol to protect the quality of all the products
it carries. Pipelines carry compatible products both preceding and following jet
fuel; the interface between products is cut into the non-aviation grade. Marine
and road transports depend on cleaning procedures, with the amount and type
of cleaning dependent on the preceding product. When aviation fuel has been
transported through a non-dedicated (i.e., a multiproduct) system, samples are
taken for recertification testing in which those properties most sensitive to con-
tamination, eg, flash point, freezing point, gum, etc., are retested and the results
compared with the original values reported on the refinery Certificate of Quality
to confirm that no inter-product contamination has occurred. At the airport, the
fuel is delivered from storage to the aircraft either via underground pressurized
piping (hydrant system) through mobile hydrant dispensers or by specialized
tank trucks called fuelers. Both types of fueling vehicle are equipped with fil-
ter/water separators or filter monitors to ensure only clean, dry fuel is pumped
into the aircraft.
Contamination Removal Equipment. Multiproduct transportation
systems and long distances between refinery and airport furnish opportunities
for water, rust, dust, surfactants, and other contaminants to enter the fuel.
Specialized equipment and procedures remove these unwanted materials (31).
Water and solids settle in storage, assisted by suction lines that float just
beneath the liquid level. Filter/water separators (32) and filter monitors (33)
remove free water and fine solids, while activated Attapulgus clay filters can
be used to adsorb surfactants.
Vol. 2 AVIATION TURBINE FUELS 289

The operation of this equipment is checked regularly with a variety of


test equipment. Solids are detected and measured per ASTM D 2276 and
ASTM D 3830 (3). Surfactant removal by clay filters can be observed with the
Microseparometer (ASTM D 3948 (3)). A variety of tests exist for the detection
of low levels of undissolved or free water, with most of the tests containing a
water-sensitive dye; ASTM D 3240 (3) measures water levels 1 ppm.

5. Legal Aspects

All aircraft, including all flight components, are certified for flight by a govern-
ment agency which promulgates and enforces the certification rules. These rules
have the force of law. Aircraft and engines use fuel meeting certain specifications
during certification. These specifications become part of the aircraft’s operating
certificate. The fuel must therefore completely meet the certification specification
even though a deviation may not affect operation adversely. Normally aircraft
and engine manufacturers designate industry fuel specifications (e.g., ASTM
D 1655 (3) ), but manufacturers can certify to their own proprietary specification.
In such a case any points of difference from the industry norms must be specified.
Once certified, aircraft must be operated by the rules of the country in which
they are registered.

6. Environmental Aspects

Aircraft exhaust emissions are controlled by regulations issued by the Interna-


tional Civil Aviation Organisation (ICAO). These regulations define the per-
mitted levels of emissions of nitrogen oxides (NOx), carbon monoxide,
unburned hydrocarbons, and smoke for each engine type. At present, emissions
are controlled only for the landing/take-off (LTO) cycle, which is defined as
engine idle, taxi out, take-off, and climb to 900 m, descent from 900 m, taxi in,
and idle. Exhaust emissions at altitude are not restricted although there are
numerous studies underway to determine the effects of high altitude emissions
on global warming and ozone depletion (34–36). Future regulation is expected
but this is likely to be met by further advances in combustion chamber design
rather than from significant changes in the composition or properties of aviation
fuel.

7. Economic Aspects

Compared with gasoline and diesel fuel production, aviation turbine fuel repre-
sents a small proportion of the crude barrel. In 1995, worldwide production of
kerosine was 285  106 t of which ca.190  106 t was for jet fuel with the remain-
der for non-aviation uses (heating, lighting, cooking, etc.). This total kerosine
volume represents ca. 9 % of world oil consumption (37). World air travel is pre-
dicted to grow, on average, by 5.5 % per year over the next decade and 4.9 % over
the coming twenty years (38–40). Highest growth rates are in the ‘‘Pacific Rim’’
290 AVIATION TURBINE FUELS Vol. 2

countries. By 2006, air travel will rise by nearly 75 %; by 2016, it will reach two-
and-a-half times current levels. Fuel volumes will also grow significantly but at a
lower rate due to improvements in aircraft fuel efficiency. The United States is
the largest single consumer of jet fuel with about 50 % of the total world volume;
this share is expected to decrease in the future as the Far Eastern economies con-
tinue to grow.
A forecast of world aviation fuel demand in 106 t is given below (37):

1990 1993 1995 2000 2005 2010 2014


161 175 189 255 330 420 500

8. Future Trends

8.1. Petroleum-Derived Fuels. Despite the predicted increase in


demand for aviation fuel, there should be no overall supply problem. Both refin-
ing capacity and processing technology are available to supply these volumes.
Whether the quality of jet fuel will change significantly in the future is open to
debate. In the early 1980s, predictions tended to be gloomy with the expectation
that fuel quality would inevitably deteriorate due to decreasing availability of
sweet crudes and the increasing use of heavy and high sulfur crudes. This led
to design studies aimed at developing engines which could cope with the resul-
tant poorer combustion quality (41). However, these predictions were not borne
out during the 1980s and 1990s because new oilfields continued to be discovered.
Advances in process technology will be able to maintain, or even improve on, fuel
quality for the foreseeable future.
Conventional petroleum crude will continue to form the major source for jet
fuel production well into the next century. Experimental production of jet fuel
from oil shale in the United States during the 1980s, although successful, was
terminated because it was economically non-viable in a world of continuing
low crude oil prices. However, this technology remains on the shelf and could
be resurrected should the economic climate change significantly. Production of
jet fuel from tar sand derived crudes is practiced in Canada, but gives only a
very small proportion of the total fuel production.
Synthesis of liquid fuels from natural gas can be achieved by the Shell Mid-
dle Distillate Synthesis process (SMDS) (42,43). Whilst producing kerosine of a
very high quality in terms of combustion performance, the SMDS product itself is
unlikely to be suitable as a jet fuel because its complete lack of aromatics causes
concern in the areas of compatibility with elastomers and lubricity. However, the
product remains an excellent blending component for upgrading lower quality
conventional kerosines which would otherwise be unsuitable as jet fuel.
8.2. Alternative Fuels. Kerosines derived from coal using the
Fischer – Tropsch process have been synthesized in South Africa by SASOL
for a number of years. The process produces kerosine which could be con-
sidered suitable as jet fuel although it has never been supplied as such.
Although energy intensive, increasing jet fuel demand in southern Africa has
prompted the producers of this synthetic product to seek approval from the
Vol. 2 AVIATION TURBINE FUELS 291

aircraft and engine manufacturers for its use as jet fuel. However, like SMDS
kerosine, many of the coal-derived fractions contain no aromatics and will need
to be blended with other streams before being acceptable. Although this produc-
tion route has been confined to South Africa, at some time in the future it may be
applied in other countries with large coal reserves, e.g., China. Oxygenated
liquids like alcohols or esters have been examined as potential aviation fuels or
fuel constituents. The pure materials penalize aircraft operation by their much
lower specific energy content, whereas blends with petroleum fuels may suffer
from water extractability and system material compatibility difficulties (29).
Thus the widespread adoption of oxygenated compounds, particularly methanol,
as commercial aviation fuel appears unlikely.
It is often said that the long-term future may lie with cryogenic fuels such as
liquid methane or liquid hydrogen. Successful flights have been made using
methane in Russia, while the ‘‘Cryoplane’’ project (44) is aimed at proving the
feasibility of a liquid-hydrogen powered aircraft. However, a major barrier to
these cryogenic fuels is that they cannot be used in conventional aircraft, and
a new generation of aircraft specifically designed for them would need to be
developed. Furthermore, completely new production, storage, distribution, and
handling systems would have to be constructed all around the world to ensure
supply at all major airports. Proponents of cryogenic fuels tend to underestimate
the capital investment involved in establishing a supply infrastructure and it is
questionable whether these fuels will ever become commercially viable (45).
Methane may be more advantageously used as the starting material in the man-
ufacture of conventional liquid hydrocarbon fuels rather than as a liquid fuel
itself (46). Liquid hydrogen has excellent combustion qualities and produces no
CO2 and, at first sight, appears to be more environmentally friendly than conven-
tional kerosine-based jet fuel. However, unless it is produced via nuclear or
hydroelectric power, its production from petroleum resources would continue
to consume energy and produce CO2. Moreover, combustion of hydrogen in air-
craft engines would still not avoid the production of NOx and, furthermore, pro-
duces 2.5 times as much water vapor as kerosine, which will also contribute to
the greenhouse effect.
The massive investments required for new fuels together with the long lead
time in design and certification will encourage the use of fuels similar to those
currently used for the longest possible time.

REFERENCES

1. M. Smith: Aviation Fuels, G. T. Foulis Ltd., Henley-on-the-Thames 1970, 495 pp.


2. V. B. Guthrie (ed.): Petroleum Products Handbook, McGraw-Hill, New York 1960,
pp. 5–1 to 5–28.
3. 1997 Annual Book of ASTM Standards, vols 05.01–05.03, ASTM, West Consho-
hocken, PA 1997.
4. Defence Standard 91-91(DERD 2494), Issue 2, Ministry of Defence, London, 1996.
5. IATA Guidance Material for Aviation Turbine Fuels., 2n ed., IATA Montreal 1994.
6. Standard Methods for Analysis and Testing of Petroleum and Related Products, In-
stitute of Petroleum, London, 1997.
292 AVIATION TURBINE FUELS Vol. 2

7. International Organisation for Standardisation, Case Postale 56, CH-1211 Geneva


20, Switzerland.
8. Manual on the Significance of Tests for Petroleum Products MNL 1, 6th ed., ASTM,
1993, pp. 34–53.
9. Petroleum Measurement Tables, vol. – XII ASTM, West Conshohocken, PA.
10. ‘‘Aviation Fuel Safety 1975,’’ CRC Report 482, Co-ordinating Research Council,
Atlanta 1975.
11. C. A. Moses, D. W. Naegeli: American Society for Mechanical Engineers Preprint 79-
GT-178, March 1979.
12. J. M. Casper, E. E. Ekstedt, W. J. Dodd, M. W. Shayeson, NASA Report CR-159855,
Cleveland Sept. 1980.
13. R. E. Vogel, D. L. Troth, A. J. Verdouw, U.S. Air Force Report AFAPL-TR-79-2072,
Wright-Patterson, Ohio, April 1980.
14. E. A. Droegemueller, R. K. Nelson: SAE Preprint 47A, SAE, Warrendale, Pa.,
April 1958.
15. T. T. Bowden, J. H. Pearson, The Effect of Fuel Composition Upon Combustion Perfor-
mance in a Rolls-Royce Tyne Combustor, ASME Paper 85-GT-39, ASME, March 1985.
16. National Research Council, Aviation Turbine Fuels-An Assessment of Alternatives,
National Academy Press, Washington, D.C., 1982.
17. W. A. Affens, R. N. Hazlett, D. E. Smith, NRL Report 8692, Naval Research Lab.,
Washington, D.C., July 1983.
18. ‘‘Aviation Fuel Properties,’’ CRC Report 530, Coordinating Research Council,
Atlanta 1983, 113 pp.
19. ‘‘CRC Literature Survey on the Thermal Oxidation Stability of Jet Fuel,’’ CRC Re-
port 509, Coordinating Research Council, Atlanta, April 1979.
20. ‘‘Thermal Oxidation Stability of Aviation Turbine Fuels,’’ ASTM Monograph 1,
ASTM, West Conshohocken PA 1991.
21. S. A. W. Thompson, S. R. Hills, N. P. H. Brown, A. Lewis, J. Inst. Petr. (London) 50
(1964) 72.
22. V. Patarini, N. S. Bornstein, M. A. DeCrescente: ASME Preprint 78-GT-82, Ameri-
can Society of Mechanical Engineers, New York April 1978.
23. J. A. Bert, H. R. Porter: Paper 11a–63, American Petroleum Institute, Washington,
D.C., May 13 1963.
24. R. N. Hazlett, H. W. Carhart, Filtr. Sep. 9 (1972) 456.
25. ‘‘Water/Fuel Separation Characteristics: A Bibliography,’’ CRC Report 460, Coordi-
nating Research Council, Atlanta July 1973.
26. W. M. Bustin, W. G. Dukek: Electrostatic Hazards in the Petroleum Industry,
Research Studies Press, London 1983.
27. ‘‘Generation and Dissipation of Electrostatic Charge During Aircraft Fueling-A
Selected Literature Survey,’’ CRC Report 466, Coordinating Research Council,
Atlanta 1964.
28. ASTM D 4865 -96: 1996.
29. K. H. Strauss: AIAA Preprint AIAA-81-0770, May, 1981.
30. S. P. Heneghan, S. Zabarnick, D. R . Ballal, W. E. Harrison, ‘‘JP-8 þ 100: The Devel-
opment of High-Thermal-Stability Jet Fuel,’’ ASME Journal of Energy Resources
Technology 118 (1996) September, 170–179.
31. Manual of Aviation Fuel Quality Control Procedures, MNL 5, 2nd ed., ASTM, West
Conshohocken, PA, 1995.
32. Specifications and Qualification Procedures - Aviation Jet Fuel Filter/Separators, API
Bulletin 1581, 3rd ed., publisherName>American Petroleum Institute, May 1989.
33. Specifications and Qualification Procedures-Aviation Fuel Filter Monitors with
Absorbent Type Elements, Institute of Petroleum, January 1995.
Vol. 2 HEATING OIL 293

34. L. J. Archer, Aircraft Emissions and the Environment: COX, SOX, HOX and NOX,
Oxford Institute for Energy Studies, 1993.
35. Air Transport and the Environment, IATA Seminar, Washington, March 1993.
36. Air Transport and Environmental Protection, Airbus Industrie, March 1993.
37. Aviation Fuels Business Worldwide 1996, Petroleum Economist, London April 1996.
38. ‘‘1997 Current Market Outlook,’’ Boeing Commercial Airplane Group, March 1997.
39. Outlook for Commercial Aircraft 1995–2014, McDonnell-Douglas, August 1996.
40. Global Market Forecast 1995–2014, Airbus Industrie, March 1995.
41. J. P. Longwell, J. Grobman, J. Eng. Power 101 (1979) 161.
42. The Shell Middle Distillate Synthesis Process, Shell Selected Paper, November 1989.
43. P. J. A. Tijm, J. M. Marriott, The Markets for Shell Middle Distillate Synthesis
Products, Alternative Energy ’95, Vancouver, May 1995.
44. ‘‘The Light Stuff,’’ Chemistry in Britain 1997 29–31.
45. A. M. Momenthy, Aviation Fuel in a Changing World, Boeing, January 1994.
46. O. Hadaller, A. M. Momenthy, Fuel Reformulation and Jet Fuel, Boeing, November
1994.

GEOFF J. BISHOP
Shell Aviation Ltd., London, United Kingdom

HEATING OIL
1. Domestic Heating Oil

Domestic heating oil (domestic fuel oil, industrial gas oil, Heizöl EL) is an
extra-light liquid distillate fuel produced from crude oil in refineries (see ! Oil
Refining). In some countries the product can contain a certain percentage of fatty
acid methyl ester (FAME). This fuel is used for commercial and industrial pro-
cesses, and mainly for heating purposes in working areas and domestic accom-
modations. The fuel is burned in atomizing or vaporizing burners.
The first industrial applications of heating oil were reported at the begin-
ning of the 20th century. In the United States, industrial gas oil was being
used to produce hot water and steam for residential heating as early as the
1930s. The major expansion of domestic heating oil and oil firing in Europe
has occurred since the end of the 1950s. Today domestic heating oil and natural
gas (in some countries electricity) are the most common energy carriers in the
domestic heating market of many European countries, in Northern America,
and in Japan. Due to emission legislation and technical requirements different
products are available in different countries. Basically heating oils can be divided
into three classes:

1. Kerosene
2. Standard heating oil
3. Low-sulfur heating oil

ß2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim


10.1002/14356007.a12_617.pub2
294 HEATING OIL Vol. 2

Table 1. Elemental Composition of a Typical Domestic


Heating Oil
Element Concentration
Carbon 86.5 mass %
Hydrogen 13.3 mass %
Sulfur < 0.2 mass %
Nitrogen 50 – 400 ppm
Oxygen traces

The large volume and share of industrial gas oil in the total refinery stream
(20 to 30 %) makes this product economically significant.
1.1. Composition. Domestic heating oil is an extra-light liquid fuel
composed almost entirely of paraffinic, naphthenic, and aromatic hydrocarbons.
In addition to these hydrocarbons, it contains small quantities of sulfur com-
pounds and still smaller (ppm) amounts of nitrogen and oxygen compounds.
Since the beginning of the 21st century special heating oil qualities have been
developed, which are almost free of sulfur and nitrogen. The chemical character-
ization of the fuel is limited to group analysis by classes of compound. The pro-
duct structure depends on the origin of the crude oil and the processing
employed. The elemental composition, however, does not vary significantly.
The boiling range is between 160 and 390 8C (kerosene: 300 8C). A typical domes-
tic heating oil has the elemental composition given in Table 1.
1.2. Quality Aspects. Quality aspects are mostly described by physical
parameters or special test methods for mineral oil products. Basically for all
three different qualities (kerosene, standard heating oil, low-sulfur heating oil)
the requirements can be classified as follows.

1. Requirements due to national legislation (tax, etc.)


2. Requirements because of the climatic situation
3. Performance parameters with similar test methods and limits in different
countries
4. Performance parameters, where national limits vary in a wide range
5. Performance parameters, where test methods and limits have to be devel-
oped and/or harmonized

Density. The density depends on temperature and for heating oil is spe-
cified at 15 8C. Density alone is of little significance as an indication of the burn-
ing characteristics of heating oil in modern atomizing type burners. However, in
vaporizing-type burners soot and CO emissions frequently increase when fuels
are used whose density and viscosity are close to the specified limits. That is
the reason why in some countries (UK and Japan) kerosene with its lower den-
sity and boiling range is used as heating fuel for vaporizing-type burners. Den-
sity limits vary in a wide range from country to country. When used in
conjunction with other properties the density is of value in mass–volume rela-
tionships and in calculating the specific energy (heating value, net calorific
value) of an oil.
Vol. 2 HEATING OIL 295

Fig. 1. Effect of temperature on viscosity.

Net Calorific Value. The net or inferior calorific value Hi (MJ/kg; synon-
ymous with lower heating value, see ! Gas Production, Chap. 8.2.1.1) does not
vary significantly between different heating oils. Hi can be calculated from the
density at 15 8C (d15, kg/m3) and the sulfur content w(S) (mass %):

d15
Hi ¼ 52:92  11:93   0:29  wðSÞ
1000
Viscosity. Viscosity limits are specified to help maintain uniform fuel
flow in appliances with gravity flow, and to provide satisfactory atomization
and constant flow rate through the small nozzles of domestic burners. A viscosity
of 2.5 to 3 mm2/s is needed for good spray formation in conventional atomizing
burners. This can be achieved by preheating of the oil to a temperature of
60 8C maximum. At the same time, the oil throughput in a conventional nozzle
is reduced by preheating because of thinner film formation in the nozzle itself.
The effect of temperature on viscosity is shown in Figure 1.
Distillation. The distillation test shows the volatility of a fuel and the
ease with which it can be vaporized. It is of greater significance for oils that
are to be burnt in vaporizing type burners than for those burnt in the atomizing
type. For example, a low 10 % value assures easy starting of a vaporizing type
burner. Nevertheless, a homogeneous distillation curve is necessary to reduce
emissions of CO and soot. A typical distillation curve of a domestic heating oil
is depicted in Figure 2. Kerosene type fuels however vaporize at significantly
lower temperatures.
Flash Point. The flash point of a fuel is an indication of the maximum
temperature at which it can be stored and handled without serious fire hazard.
The minimum permissible flash point is usually regulated by law and can there-
fore differ from country to country.
Sulfur Content. The sulfur content is usually limited by national legis-
lations. The sulfur content in domestic heating oil has been reduced significantly
over the years. In Europe an EC directive allows sulfur content to be as high as
0.2 %. This value is reduced to 0.1 % starting from 2008. The amount of low-
temperature corrosion products formed in modern energy efficient boilers can
296 HEATING OIL Vol. 2

Fig. 2. Typical distillation curve of domestic heating oil.

be influenced by the sulfur content of the oil. In condensing boilers which is the
most efficient burner technology but not very common for domestic heating oil in
contrast to natural gas equipment — the pH value of the condensate depends on
the sulfur content. To promote these energy saving appliances in many countries
a second quality with significantly lower sulfur content has been introduced to
the market and standardized.
Low-Temperature Behavior. In heating oil crystallization of high
molecular mass n-paraffins (waxes) can occur at low ambient temperature. The
wax crystals can block fuel pipes and especially the fuel filter. Three test meth-
ods are used to assess the low-temperature behavior.
Pour Point. The pour point gives the lowest temperature at which a fuel
oil can be stored and still is capable to flow under very weak forces. The determi-
nation of the pour point alone is not sufficient to assess the cold flow behavior in
practical application.
Cloud Point. The cloud point is the temperature at which heating oil
begins to become turbid at a defined cooling rate because of the formation of
small wax crystals.
Cold Filter Plugging Point (CFPP). The CFPP is the temperature at
which, under defined conditions, a heating oil sample can no longer pass a
defined filter within a specified time. The CFPP can be improved by additives
(see Section 1.3.2). Neither cloud point nor cold filter plugging point alone can
sufficiently describe the cold flow behavior in real installations but the combina-
tion of both does.
Carbon Residue. The carbon residue of a fuel is a measure of the carbo-
naceous material that is left after all volatile fuel components are vaporized in
the absence of air. It is a rough approximation of the tendency of a fuel to
form deposits in vaporizing burners where the fuel is vaporized in an air-defi-
cient atmosphere. To obtain measurable values of carbon residue in the lighter
distillate heating oils, first 90 % of the oil must be removed by distillation, then
the carbon residue concentrated in the remaining 10 % bottoms is determined.
Ash. The amount of ash is the quantity of noncombustible material in an
oil. The amount of ash in domestic heating oil is almost zero if heating oil directly
from the refinery is investigated. However, traces of metals can be found in heat-
ing oil, as steel equipment is mainly used for transportation and storage (1).
Vol. 2 HEATING OIL 297

Water and Sediment. If a domestic heating oil contains appreciable


amounts of water and sediment this will lead to fouling of facilities for handling
the oil, and difficulties in burner mechanisms will arise. Sediment may accumu-
late in storage tanks and on filter screens or burner parts, resulting in obstruc-
tion to flow of oil from the tank to the burner. Water in distillate fuels can cause
corrosion of tanks and other metal equipment.
Storage Stability. The aging of domestic heating oil is caused by oxida-
tion and radical polymerization reactions, which cause darkening and then sedi-
ment formation. The phenomenon of natural aging of middle distillates has been
treated in many publications. However, a comprehensive, detailed and consistent
theory of the aging mechanism was not found so far. A detailed summary of more
than 330 publications dealing with stability of middle distillates is given in (2).
The following items can deteriorate the stability of domestic heating oil:

Heat
Light
Oxygen
Catalytic effects of metals (copper)

Thermal Stability. If the thermal stability of a heating oil is not suffi-


cient, deposits may be formed in hot spots in the burner — usually between
the oil preheater and the nozzle. In modern burners with reduced NOx emissions
and low throughput, temperatures significantly higher than 100 8C have been
measured after shut-down of the burner. This is due to back radiation from
the heating chamber. It is known from many laboratory tests that the aging pro-
cess (sediment formation) of the oil increases significantly between 100 and
120 8C. To circumvent this, burner manufacturers tend to develop burners
with maximum temperature of 100 8C especially after shut-down of the burner.
If the oil contains copper ions, the thermal stability of the fuel is reduced by
approximately 20 K. A new laboratory test to assess the thermal stability of a
fuel has recently been developed in Germany (3).
Lubricity. As sulfur components contribute to the lubricity of the fuel,
problems may occur with desulfurized heating oil qualities. Therefore lubricity
requirements have been implemented in some standards for low-sulfur heating
oil.
1.3. Production. Product and Processing. Since the early 1970s, pro-
duction of petroleum-based products has been influenced by an increasing range
of crude oils used and by a restructuring of refinery production facilities.
There is a persistent trend towards minimizing the amount of residual pro-
ducts and maximizing the quantity of middle distillates and gasolines. This
change has become possible as the following equipment has been incorporated
in the refinery process:

1. Vacuum distillation equipment,


2. Thermal gas-oil units,
3. Catalytic crackers, and alternatively
4. Hydrocrackers.
298 HEATING OIL Vol. 2

Fig. 3. Production of domestic heating oil (simplified flow diagram).

To reduce the sulfur content of domestic heating oil, hydrodesulfurization equip-


ment has been installed in many refineries. A simplified flow diagram of the pro-
duction of domestic heating oil is given in Figure 3. The following distillation cuts
are the main components for the blending of heating oil:
1. Kerosene
2. Light gas oil (LGO)
3. Heavy gas oil (HGO)
4. Hydrodesulfurized gas oil (HDS-GO)

The necessity and extent of blending domestic heating oil from different
components depends on the crude oil or the mixture of crude oils employed, on
the refinery facilities and the product mix that has to be achieved, and on the
specifications which have to be met.
Upgrading with Additives. Product quality and application requirements
can quite often only be met by employing special purpose additives, which are
added to the final product in the required concentrations. Some types of additive
are usually introduced into the product in the refinery. The use of additives for
quality improvement of heating oil is approved according to many heating oil
standards. These additives may be applied in suitable concentrations, provided
they are not known to exhibit any harmful side effects. Recommendations, which
should help to select suited additives, are published in (4).
Flow Improvers. Flow improver additives improve the cold flow proper-
ties of domestic heating oil. These additives are added to the product stream in
many refineries. Ash-free polymers are used as flow improvers. At a temperature
slightly higher than the cloud point, the additive forms crystallization nuclei,
to which the wax crystals become absorbed. Further growth of the wax crystals
is reduced by adsorption of the additive on the crystal surface. Thus, both the
crystal size and the tendency to agglomerate are lowered. The small crystals
Vol. 2 HEATING OIL 299

formed pass through the filters of the heating system. Therefore, the CFPP can
be lowered significantly by adding flow improvers. The dosage depends on the
base fuel and on the CFPP value which is to be achieved. Typical concentrations
are 50–500 ppm.
Flow improvers have to be added to the fuel at temperatures above the
cloud point; addition after crystal formation is not effective.
Metal Deactivators. In special cases, metal deactivators are also added
to industrial gas oil to attain long-term storage stability of more than three
years. The negative influence of copper (5,6) on storage stability can be reduced
by these additives.
Other types of additives are sometimes introduced into the customer’s sto-
rage tank but not into the fuel in the refinery.
Combustion Improvers. The purpose of these additives is to suppress
carbon formation during combustion. Combustion improvers are predominantly
organometallic compounds, ferrocenes have proved particularly effective in
domestic heating oil. Although burner technology has advanced considerably
with respect to near-stoichiometric combustion with low carbon formation, com-
bustion improvers can still effect and stabilize optimal combustion over long per-
iods. The application of combustion improver additives is not necessary in
modern blue flame burners with reduced NOx emissions.
Additives for Increased Storage Stability. The application of metal
deactivators and antioxidants can improve the storage stability of the fuel. The
importance of these types of additives is increasing because the average storage
time of heating oil in domestic storage tanks increases due to reduced oil con-
sumption (higher efficiency equipment, better insulation of houses) while the
size of the storage tanks has been kept constant.
Reodorants. Some additive packages contain reodorants to improve the
odor of heating oil. The unpleasant smell is caused particularly by certain sulfur-
containing and unsaturated components. If there is no leakage in the oil heating
system, the application of reodorants is not necessary.
Biocides. In some cases water can be found in the bottom of storage
tanks. Conditions at the interface between water and oil allow the growth of
microorganisms. The result is very often sediment formation. The addition of sui-
table biocides stops the growth of microorganisms but the best way is to remove
water regularly from the bottom of the tank.
1.4. Uses and Heating Efficiency. Domestic heating oil is mainly
burnt for heating purposes and for the heating of drinking water in burner/boiler
systems. The substantial rise in energy prices, the growing awareness of air pol-
lution and the international discussion on greenhouse gases have changed the
assessment of heating systems. It is possible to save a lot of energy in the domes-
tic heating sector if modern heating systems replace the old-fashioned systems.
A modern heating system has to fulfill the following demands:

1. Minimal pollutant load on the environment


2. Optimal matching of capacity to demand
3. High efficiency on an annual basis
4. Reliable operation
300 HEATING OIL Vol. 2

5. Comparatively attractive product prices


6. Compliance with all standards
7. Long-term availability of the energy source employed

The pollution of the environment could be reduced considerably by develop-


ment and installation of more efficient heating systems (reduced oil consump-
tion) and of low-emission burners. Generally, the fuel is evaporated to a spray
before combustion. In modern low-emission burners, up to 30 % of the flue gas
is recycled to the fuel spray. The partial pressure of oxygen and the maximum
temperature in the flame is reduced by this technology which allows to meet
strict NOx emission limits. As the oil spray is completely vaporized before it
reaches the combustion zone, the formation of soot and particulate matter
emission (PM10) can be avoided. Thus, similar advantages as in the combus-
tion of natural gas can be achieved. A good summary of basic knowledge for
air-fuel mixing, evaporation and combustion technology with special regard
to liquid fuels is given in (7).
Since investments like improved thermal insulation in existing buildings
and especially the better insulation standards of new buildings gave the expected
energy saving, oil burners designed for lower throughput have been developed. In
new buildings the heating system is often dimensioned according to the demand
of hot drinking water rather than by the demand of space heating. The minimum
power for conventional pressure jet burners is 15 kW (ca. 1.5 L/h) because of
the dimension of the nozzle. Other spray formation systems have been developed
to allow the construction of burners with reduced throughput. However, not
many of these systems have been installed yet.
The boiler efficiency on an annual basis has been improved significantly
during the last 25 years by better insulation of the boiler walls, reduced internal
cooling loss and by reduction of the flue gas loss. The annual efficiency of a mod-
ern low-temperature burner–boiler unit is more than 93 % (basis: net calorific
value). The radiation loss and the internal cooling loss account for less than
1 % each, while the flue gas loss is influenced by the flue gas temperature and
the amount of excess air.
To maximize the efficiency condensing boilers have been developed. The
annual efficiency of condensing boilers is between 100 and 103 % (basis: net
calorific value). The market share of condensing boilers differs from country to
country but it is steadily increasing. The construction and materials used in con-
densing boilers are very much dependent on the fuel specification, especially the
sulfur content. If the sulfur content is below 50 ppm, it is possible to use similar
heat-exchanger materials as in gas condensing boilers. If the sulfur content is
higher, the materials used in the condensing areas of the boiler and the flue
gas pipe have to be specially resistant to corrosion by acidic condensate. The
pH value of typical condensate varies from 2 to 3 depending on the construction
of the burner and boiler, the sulfur content of the fuel and the amount of conden-
sate (6). For low sulfur heating oil the pH value is between 2.8 and 4, which is
similar to gas condensing boilers (9).
In some countries the basis for the calculation of the flue gas loss is the
gross calorific value of the heating oil, in other countries the calculation is
based on the net calorific value. In the latter case, the flue gas loss can be
Vol. 2 HEATING OIL 301

Fig. 4. Flue gas loss (basis: net calorific value).

calculated from the difference in temperature between flue gas and ambient air
(DT) and the concentration of CO2 or O2 in the flue gas:
 
0:5
flue gas loss ¼ DT þ 0:007
½CO2 

or
 
0:68
flue gas loss ¼ DT þ 0:007
ð21  ½O2 Þ

These formulae cannot be used for condensing boiler technologies because the
latent heat of condensation of the water (produced in combustion) is utilized in
addition to the sensible heat. Therefore, the flue gas loss and the efficiency of a
boiler should be calculated with the gross or superior calorific value, Hs
(synonymous with higher heating value, see ! Gas Production, Chap. 8.2.1.1).
The flue gas loss can be taken from Figures 4 and 5.
Domestic heating oil is used in small power stations (with Diesel engines)
for simultaneous production of electricity and hot water or steam for heating

Fig. 5. Flue gas loss (basis: gross calorific value).


302 HEATING OIL Vol. 2

purposes (10). In research projects and pilot applications domestic heating oil has
been used to produce hydrogen and carbon monoxide as feedstock for high tem-
perature fuel cells (11).
1.5. Product Specifications. There is no international or European
standard for domestic heating oil. The minimum requirements of domestic heat-
ing oil in some national standards are given in Table 2 and 3. Some countries
have introduced standards which differ mainly in sulfur content. Other countries
have set different limits for cold flow properties in winter and summer. Generally
the lighter qualities (eg, kerosene type fuels, Grade 1 in USA) are specially suited
for vaporizing burners.
In 2003 a new European Standard (EN 14213) has been published. This
standard specifies requirements and test methods for marketed and delivered
fatty acid methyl esters (FAME) to be used either as a heating fuel at 100 % con-
centration, or as a blending component for the production of heating fuel. How-
ever, mixtures have not been standardized yet. An increasing demand for
bio-fuels as components of automotive fuels can be observed especially in Europe.
As in some countries the logistics of diesel and heating oil are not completely
separated, it has to be specified to what extent FAME is allowed to be a compo-
nent of heating oil.
1.6. Storage and Transportation. A particular characteristic of
domestic heating oil is the flash point, established at >55 8C. This value leads
to classification of the fuel in Hazard Class III A (GefStoffV). There are no special
safety restrictions concerning storage or transportation of the product.
During transportation and storage domestic heating oil has to be kept at
such temperatures that the sedimentation of wax crystals is avoided. Therefore
the cold flow behavior of the fuel has to be considered for storage and transporta-
tion. This can be a problem especially in underground storage tanks, when the
pipes to the burner are not sufficiently insulated. In Germany the storage tem-
perature must be above 0 8C. Lower temperatures are allowed if a special fuel
with satisfactory low temperature-properties is used.
1.7. Environmental Protection. The importance of the sulfur content
of domestic heating oil for the consumer lies in the emission of sulfur dioxide
and thus the problem of environmental pollution. In the EU, the sulfur content
of domestic heating oil is now limited to 0.2 mass % (Austria 0.1 mass %). The EU-
sulfur directives (1999/32/EC) requires to reduce that limit to 0.1 mass % by
2008. In some countries or regions there are special fuels available with sulfur
content <0.05 mass % or even <50 mg/kg (0.005 mass %, low-sulfur heating
oil). In general the average sulfur content has been decreasing continuously.
Further reduction of the sulfur content of the fuel and thus reduced
SO2 emissions when burning the fuel can be accomplished with desulfurization
processes, if these facilities are available in the refinery. However, removal of
sulfur requires hydrogen and is performed at pressures of 2.5 to 7 MPa with
the result of increased energy consumption and CO2 emissions. On the other
hand this is more than compensated by higher annual efficiencies of low tem-
perature or condensing boilers, because fouling of the heat exchanger surfaces
is reduced.
The nitrogen content of domestic heating oil is not specified in national
standards. It depends on the crude oil and the processing employed. Some
Table 2. Product Specifications of Domestic Heating Oil in Various EU Countries (Minimum Requirements)
Austria Belgium France Germany Great Britain Italy Sweden
ÖNORM C1109 NBN T NBN EN 590 M 15-008 DIN 51603-1 BS 2869-2 UNI 6579 SS 15 54 10
52-716
Issue 2003 2000 2005 1997 2003 1998 1998 1998
Remarks Low sulfur Standard Low sulfur Low sulfur Standard Class C1 Class D Summer/Winter
Remarks Diesel EN 590 Kerosene
Property Requirements Requirements Requirements Requirements Requirements Requirements Requirements Requirements Requirements

Density at max. 0.860 0.820– 0.820– 0.830 – max. 0.860 — 0.815– stated
158C. g/cm3 0.880 0.845 0.880 0.875
Gross calorific — min. 44.4 — — — — —
value, MJ/kg
Net calorific min. 42.0 — — min. 42.6 — — —
value, MJ/kg
Flash point > 55 min. 55 >55 min.55 > 55 min. 43.0 min. min. 55 min. 56
Pensky- 56
Martens, 8C
Viscosity 2.8 – 6.0 max .6.0 max 9.5 max. 6.00 — — —
kinematic at
20 8C, mm2/s
Viscosity — — 2.0 – 4.5 — — 1.50 – 5.50 2.0–7.4 2.0–4.0/1.5–3.0
kinematic at

303
40 8C, mm2/s
Viscosity — — — — — — —
kinematic at
38 8C, mm2/s
Distillation — — — — — max. 2 —
recovery at
1508C, vol%
Distillation — — — — min. 15, max. 60 — — —
recovery at
2008C, vol%
Distillation — — — — — — min. 10
recovery at
2308C, vol%
Distillation — max. 65 max. 65 < 65 < 65 max. 65 < 65 —
recovery at
2508C, vol%
Distillation — — — — — — —
recovery at
3008C, vol%
Distillation min. 90 min. 85 min. 85 min. 85 min. 85 min. 85 min. 85 —
recovery at
3508C, vol%
Table 2. (Continued)
Austria Belgium France Germany Great Britain Italy Sweden
ÖNORM C1109 NBN T 52-716 NBN EN 590 M 15-008 DIN 51603-1 BS 2869-2 UNI 6579 SS 15 54 10
Issue 2003 2000 2005 1997 2003 1998 1998 1998
Remarks Low sulfur Standard Low sulfur Low sulfur Standard Class C1 Class D Summer/Winter
Remarks Diesel EN 590 Kerosene
Property Requirements Requirements Requirements Requirements Requirements Requirements Requirements Requirements Requirements

Distillation — — min. 95 — — — — —
recovery at
3608C, vol%
Distillation — — — — — — min. 95
recovery at
3708C, vol%
Distillation
temperature, 10 %
recovered, 8C
Distillation
temperature,
90 %
recovered, 8C
Distillation
temperature,

304
95 %
recovered, 8C
Pour point, 8C — — max. 8 — — max. 6 —
Cloud point, 8C — — max. 2 max. 3 — — max. 0/22
Cold filter max. 8 max. 10 max. 0/15 — max. 12 max. 4 / 12 max. 0 max.10/32
plugging point (CP 3)/11 (CP 2)/
(CFPP), 8C 10 (CP 1)
Carbon residue max. 0.15 max. 0.10 max. 0.30 max. 0.3 max. 0.30 max. 0.2
(on 10%
residue),
%(m/m)
Sulfur content, — max. 0.2 max. 0.20 0.0050 – 0.20 max. 0.04 max. 0.20 max. 0.20 < 0.10
%(m/m)
Sulfur content, max. 50 50–1000 max. 50 max. 50
mg/kg
Water content, max. 200 max. 200 max. 200 max. 200 max. 200 max. 200 — max. 200
mg/kg
Particulate — — — — — — —
contamination,
mg/L
Total max. 24 max. 24 max. 24 — max. 24 max. 24 — max. 24
contamination,
mg/kg
Ash, %(m/m) max. 0.01 — max. 0.01 — max. 0.01 max. 0.01 max.0.01 max. 0.01
Neutralisation — — — — — —
number, mg
KOH/100 mL
Copper — max. 1 class 1 — — class 1 class 1 — max. 1
corrosion
Sediment, %(m/m) — — — — max. 0.01 — max. 0.05
Mechanical — — — — — —
contamination,
% (m/m)
Water and — — max. 0.10 — — max. 0.05 —
sediment, vol%
Color red — red — — — —
Color ASTM — — — — — —
D 1500
Oxidation — — max. 2.5 — — — — —
stability,
mg/100 mL

305
Cetane index — — — — — — —
Organic — — — — — — max. 6
chlorine, mg/kg
PCB, mg/kg — — — — — — max. 2
Nitrogen, mg/kg — — — — — — stated

* See
Table 3. Product Specifications of Domestic Heating Oil in Some Non-Eu Countries
Japan Norway Switzerland USA
Issue JIS Gasoil SN ASTM
No.3 181160-2 D396
Remarks 2002 1999
Remarks Kerosene Production CH- Euro- Grade 1 Grade 2
Standard Quality Quality
Property Requirements Requirements Requirements Requirements Requirements Requirements
Density at 158C. g/cm3 0.820–0.860 0.820–0.855 0.820–0.860 max. 0.850 max. 0.876
Gross calorific value, MJ/kg — min. 45 —
Net calorific value, MJ/kg — min. 42.6 min. 42.6
Flash point Pensky-Martens, 8C min. 40 min. 60 min. 55 min. 55 min. 38 min. 38
Viscosity kinematic at 20 8C, mm2/s — — —
Viscosity kinematic at 40 8C, mm2/s 2.0–4.0 2–4 max. 4 1.3–2.1 1.9–3.4
Viscosity kinematic at 38 8C, mm2/s — — —
Distillation recovery at 1508C, vol% — — —

306
Distillation recovery at 2008C, vol% — — —
Distillation recovery at 2308C, vol% — — —
Distillation recovery at 2508C, vol% max.65 — max. 65
Distillation recovery at 3008C, vol% — — —
Distillation recovery at 3508C, vol% min. 85 — min. 85
Distillation recovery at 3608C, vol% — min. 90 —
Distillation recovery at 3708C, vol% min. 95 — —
Distillation temperature, 10 % recovered, 8C max. 215
Distillation temperature, 90 % recovered, 8C max. 288 282–338
Distillation temperature, 95 % recovered, 8C max. 270
Pour point, 8C — — — max. 18 max. 6
Cloud point, 8C max. 0 max. 2 max. 3
Cold filter plugging point (CFPP), 8C max. 11 max. 10 12/11/10
Carbon residue (on 10% residue), %(m/m) max. 0.15 max. 0.05 max. 0.05
Sulfur content, %(m/m) max. max. 0.05 max. 0.2
0.0499
Sulfur content, mg/kg max. 80
Water content, mg/kg max. max. 150 max. 200
100
Particulate contamination, mg/L max.8.0 — —
Total contamination, mg/kg — max. 12 max. 24
Ash, %(m/m) max.0.01 max. 0.01 max. 0.01
Neutralisation number, mg KOH/100 mL max.0.5 — —
Copper corrosion 1 — —
Sediment, %(m/m) — — —
Mechanical contamination, % (m/m) — — —
Water and sediment, vol% — — —
Color — — —
Color ASTM D 1500 <=1.5 — —
Oxidation stability, mg/100 mL max.2.5 — —
Cetane index min. 45 — —
Organic chlorine, mg/kg — — —
PCB, mg/kg — — —

307
Nitrogen, mg/kg — ÖKO-oil max. 100
308 HEATING OIL Vol. 2

Table 4. Emissions of Oil Heating Systems (Atomizing Burners)


Average Modern Best available
Emission data technology technology
Carbon monoxide, mg/MJa 7 – 76 (13) 7 1.4
Nitrogen oxidesb**, mg/MJa 29 – 50 (13) 30 17
Unburnt hydrocarbons, 1.4 <1 <1
mg/MJa
PM10, mg/MJa 0.05 – 1.2 <0.3 <0.1
Sulfur oxidesc*, mg/MJa 70 70 70
Carbon dioxide, kg/MJa 0.074 0.074 0.074
a
Net calorific value of fuel.
b
Nitrogen content of reference fuel¼140 mg/kg.
c
Depending on sulfur content of fuel (here 0.15 mass %).

desulfurization processes may also reduce the nitrogen content of the fuel. Typi-
cally, the nitrogen content is between 50 and 200 mg/kg, occasionally it can be as
high as 400 mg/kg. For low-sulfur heating oil or kerosene the nitrogen content
often is below 50 mg/kg. The fuel nitrogen contributes to the emissions of nitro-
gen oxides (NOx) during combustion. 50 to 100 % of the fuel nitrogen are con-
verted to NOx depending on the type of burner (12).
Typical emission levels for oil heating (Table 4) are dependent on the tech-
nology employed (13). As could be expected, modern boiler and burner technology
results in lower emission values of CO, NOx, and particulate matter (PM10) than
older technologies.
The emissions of components that are traditionally not determined have
been investigated (14). The study encompasses emissions of N2O, methane, ben-
zene, toluene, xylene, and PCDD/PCDF (polychlorinated dibenzo-p-dioxin/diben-
zofuran). The emission levels of these components are very low. Oil-fired systems
do not contribute significantly to total emissions of these components.
The emissions of NOx and SO2 could be reduced by secondary treatment of
the flue gas. The technology for that flue gas treatment is available on the mar-
ket but is usually not used for domestic installations, because the emission limits
can be met by modern combustion systems alone.
Recently particulate matter emissions have become more important. It is
well accepted that with modern combustion technology in which the fuel is eva-
porated before being combusted the particulate matter emissions are similar to
those of gas appliances.
1.8. Economic Aspects. The consumption of heating oil in selected
countries is given in Table 5 (17). The sizes of the US market and the Japanese
market are similar to the market in Germany. Market data are available on
www.euro-fuel.net. Data for the US market can be taken from (15).
The consumption of heating oil has been decreasing in many countries. This
is caused by better insulation of houses and more efficient boiler systems and by
the fact that in some countries oil heating systems have been switched to other
systems. In many countries natural gas is the main competitor, in other coun-
tries electrical heat-pumps are gaining market share. In contrast, in some coun-
tries the number of oil installations has increased by replacing direct electrical
heating.
Vol. 2 HEATING OIL 309

Table 5. Consumption of Domestic Heating Oil in Selected Countries (103 t)


Country 1996 1997 1998 1999 2000 2001 2002 2003 2004
Germany 38 420 36 292 34 631 29 633 27 875 31 888 28 544 28 064 25 424
Austria 1956 1907 1897 1852 1592 1961 1761 1979 1690
Belgium 6479 6490 6295 5825 5418 5736 4999 5538 5497
Denmark 2021 1880 1806 1778 1669 1727 1453 1462 1397
Spain 8130 7587 8310 9468 9207 9628 9534 10 235 10 843
Finland 2446 2361 2485 2435 2150 2367 2285 2372 2180
France 17 262 16 725 17 303 19 960 15 583 17 257 15 597 16 022 16 042
Greece 3283 2784 3581 3638 3961 4379 4528 5057 4788
Ireland 1324 1383 1556 1710 1965 1850 1565 1400 1368
Italy 7027 6820 6452 6627 6289 5797 5223 5559 5333
Luxemburg 366 368 376 355 301 335 311 303 319
Holland 1675 1660 1602 1604 1246 1136 1153 1209 1191
Portugal 783 1074 1299 1048 1266 1223 1070 984 1177
UK 7631 7326 7245 6682 6583 6598 6129 6142 4927
Sweden 3105 2214 2175 2313 2969 1950 3129 631 1588
EU (15) 101 928 96 871 96 993 91 928 88 074 94 032 87 271 86 957 83 764
Iceland 331 317 345 363 408 324 331 326 404
Norway 2073 2003 2026 2114 1791 1810 1812 2062 1870
Switzerland 4939 5388 5315 4734 4299 5274 4856 4562 4454
Turkey 2912 3172 3054 3475 3723 3258 3258 3293 3436
Total 112 183 107 751 107 735 102 614 98 295 104 698 97 528 97 200 93 928

2. Residual Fuel Oil

Residual fuel oil (RFO) is a refinery product derived from petroleum crudes.
Since the introduction of oil conversion technologies in the mid-1960s most
RFO has been produced by blending vacuum distillation residues with residues
from thermal crackers (e.g., visbreakers). Material with a marketable viscosity
range results from further addition of catcracked components as well as gas oil
fractions of varying density derived from thermal processing. Heavy oils derived
from lignite and coal will not be discussed here because these are only of local
interest.
2.1. Composition and Properties. Residual fuel oil is an extremely
complex material as a consequence of its origin in the cracking process. In addi-
tion to paraffinic, naphthenic, and aromatic hydrocarbons (some with molecular
masses approaching 20 000), RFO contains nitrogen and oxygen compounds as
well as organometallic substances that contribute to its ash content. A typical
elemental analysis shows a distribution (in wt %) of 87.5 % C, 10.6 % H, 1.8 % S,
and 0.1 % ash (C : H ratio 8.25); these values may be assumed valid for perform-
ing combustion calculations.
RFOs are colloidal systems of highly dispersed macromolecular components
(asphaltenes) in solution equilibrium with a continuous oil phase (the maltenes).
Just like the maltenes, asphaltenes may be regarded as natural fuel components,
although with a much higher C : H ratio. Most of the metallic ash components are
present in oil-soluble form, either as porphyrins or as part of the asphaltene
phase. Metal atoms are in most cases bonded to the heteroelements sulfur,
oxygen, and nitrogen.
310 HEATING OIL Vol. 2

Table 6. Metal Content of Residual Fuel Oil (in mg Per Kilogram Atmo-
spheric Residue) as a Function of Source
Crude oil source Vanadium Sodium Nickel
Nigeria/Forcados 1 8 10
North Sea/Brent 6 1 3
Libya/Brega 6 2 10
Middle East/Oman 30 5 10
Middle East/Arab Heavy 75 15 27
Venezuela/Tia Juana Pesado 426 22 56

Crude oil is separated into its various fractions by fractional distillation, so


RFO tends to be enriched in metals. The most important ash components are
vanadium, sodium, and nickel, as well as calcium and magnesium (all in the
ppm range). The metal content depends on the geographic origin of the crude
oil (see Table 6).
Numerous other metals are present in trace amounts (mg/kg), but may be
ignored for commercial applications. Most metal analyses today result from
X-ray fluorescence determinations performed directly on the RFO.
The principal distinction between RFO and domestic heating oil is that the
former must be preheated prior to use. Compared to coal, RFO offers the advan-
tages of relatively straightforward handling, favorable burning characteristics,
and a high specific heat content. Furthermore, the relatively low ash content
of RFO largely eliminates problems related to the disposal of ash and filter dust.
2.2. Production. Depending on the proposed application, RFO compo-
nents as described in Section 2.1 are blended in the refinery to meet appropriate
product specifications. The characteristic that most clearly distinguishes
one grade of RFO from another is nominal viscosity at 100 8C (or 80 8C in English-
speaking countries). Rheologic factors preclude the use of 50 8C viscosity data.
Standard grades of RFO for steam boilers and industrial furnaces have visc-
osities of 20–50 mm2/s, whereas the marine market prefers material in the
10–55 mm2/s range (all at 100 8C). Viscosities between 100 and 3000 mm2/s are
typical for oil gasification units and for internal use within a refinery.
Economic considerations are responsible for the fact that suppliers seldom
employ additives in RFO. Additive addition remains the responsibility of the end
user, and is limited to metalcontaining, oil-soluble substances that improve com-
bustion characteristics (e.g., particulate control by suppressing soot formation)
and minimize low-temperature corrosion at cold junctions (through dew-point
reduction). Compounds of magnesium and other alkaline earth metals are occa-
sionally added to raise the melting point of the ash.
2.3. Uses. The most important uses of RFO have already been noted in
Section 2.2. On a world basis, most of the product is used in steam generators
and industrial furnaces (ca. 400  106 t/a) or marine applications (85106 t/a).
Currently, RFO represents ca. 20–22 % of the ‘‘cut of the barrel’’ (based on
data from the Western world), though demand in the industrialized countries
is diminishing. Interest in RFO as a tuyere-injection ‘‘reducing agent’’ for
blast furnaces varies according to the coal–residual oil cost ratio, because from
a technological standpoint the two materials can be regarded as equivalent.
Vol. 2 HEATING OIL 311

Increasing application of oil conversion technologies coupled with environmen-


tal factors suggests that consumption of RFO in the industrialized nations
will continue to decline relative to so-called ‘‘white products’’ (middle distillates
and gasoline). The following market share figures (RFO as a fraction of total
oil consumption) for 1988 provide additional insight: United States, 9 %;
Federal Republic of Germany, 10 %; France, 13 %; United Kingdom, 29 %;
Japan 31 %.
Usually RFO is combusted in furnaces with any of several types of burners
that are distinguished by the method employed for atomization. The most impor-
tant types are:

1. Jet atomizers of the nozzle or lance type (operating pressures up to


4.0 MPa, atomizing viscosities of ca. 10 mm2/s, preheating to 140 8C)
2. Rotary cup burners (0.5 MPa, ca. 15–25 mm2/s, 110 8C)
3. Air or steam atomizers (operating parameters analogous to rotary cup bur-
ners, air or steam flows of 5–7 wt % relative to fuel throughput)

The choice of burner type depends on the design of the combustion facility.
Key factors include adequate residence time for combustion of the asphaltenes
and low specific heat load, together with near-stoichiometric combustion to mini-
mize NOx emission.
Some marine users of RFO continue to employ steam boiler–turbine com-
binations, although diesel engines now account for 96 % of the market world-
wide, and their use continues to increase. Diesel engines are clearly more
economical to operate than turbines, especially under partial load conditions.
With the aid of turbocharging and waste heat recovery, a maximum thermal effi-
ciency of ca. 50 % can now be achieved at a specific fuel consumption as low as
150 g/kWh. Today’s RFO can be utilized without difficulty in both 4-stroke trunk
piston engines (minimum output: 160 kW per cylinder at 750 min1) and
2-stroke cross head engines (maximum output: 3500 kW/cylinder at 80 min1).
Pretreatment of RFO for use as a motor fuel entails centrifugal separation (puri-
fier/clarifier) and fine filtering (10–20 mm). In addition, effective preheating is
necessary (to 160 8C with an injection viscosity of 9–12 mm2/s and an injection
pressure ahead of the jet of as much as 140 MPa). Taken together, these
measures assure optimal combustion as well as adequate protection against
abrasive wear.
2.4. Quality Specifications, Testing, and Analysis. For technologi-
cal and economic reasons reliable standards must be available for RFO classifi-
cation, and product specifications must be clear and acceptable to all parties
involved. To date, standards have been developed for residual fuels for industrial
uses only on a national basis. The most important distinctions between various
product grades relate to viscosity, density, flash point, and sulfur and ash con-
tent. Standards applicable within the United States are covered by ASTM-D
396–84; for the United Kingdom, BS 2869; for the Federal Republic of Germany,
DIN 51063 (part 3 and, in the future, part 5, which specifies a 1 % sulfur content);
for Switzerland, SN 181160/2–86; for Austria, öNORM C 1108–84; and for
Japan, JIS K K 2205.
Standards for RFO intended for marine use are given in ! Marine Fuels.
312 HEATING OIL Vol. 2

Analytical values for RFO vary widely depending on applicable comprehen-


sive standards and local requirements. They are a function not only of refinery
processing but also of the origin of the crude oil. Considerable progress toward
uniformity has been achieved in recent years, largely because of increasing
awareness and knowledge with respect to environmental protection.
Density. Density depends on temperature, thus RFO density data are
uniformly reported at 15 8C to facilitate conversion of volume into the commer-
cial measure mass (in tonnes). Industrial residual fuels are not normally sub-
ject to a density limit (typical values: ca. 995 –1020 kg/m3). However, density is
important for marine applications (maximum 925–1010 kg/m3) because it is a
limiting factor in the effective use of centrifugal separators. Density is also
related to aromatic content, which in turn affects ignition characteristics in a
combustion or diesel engine. Further refinement of ISO 8217 is expected to
result in incorporation of a density limit, particularly for material with a
viscosity <80 mm2/s.
Kinematic Viscosity. This parameter also depends on temperature. The
reported ‘‘nominal viscosity’’ should be determined at a temperature of at least
80 8C, preferably 100 8C. Kinematic viscosity is important in both the industrial
and marine sectors because it largely determines the level of preheating required
for RFO transfer and for its input into a burner or diesel engine. Reliable and
economic operation requires the availability of appropriate viscosity measure-
ment and control systems.
Flash Point (Closed Cup). Flash point information is important in
ensuring safe transport and storage. In the case of residual fuels, flash point
determinations above 65 8C are normally carried out according to ASTM-D 93
(Pensky –Martens). Flash point specifications in the various norms range from
a minimum of 60 8C (marine fuels) to 70–100 8C (inland applications).
Sulfur Content. Sulfur content is largely a function of the origin of the
corresponding crude oil. Residual fuel oil desulfurization at the refinery presents
serious technological problems, and is currently not economical. The sulfur
restrictions imposed by various countries must usually be met by proper choice
of crude oil. Low-sulfur crude is available only from a few sources (North Sea,
North Africa), and for this reason has a higher market value than ordinary
crude. The most important commercial RFO gradations with respect to sulfur
content are 1.0, 1.5, 2.0, and 2.5 % for the industrial market, and up to 5.0 %
for marine applications.
The principal problems posed by sulfur arise from its combustion products,
sulfur dioxide (an environmental hazard) and sulfur trioxide (conversion rate to
sulfur trioxide is 1–3 % because of combustion with surplus oxygen). Sulfur tri-
oxide increases the sulfuric acid dew point in the off-gases. The resulting conden-
sate, together with water, causes low-temperature corrosion of metallic surfaces.
Conradson Carbon Residue (CCR): Asphaltene Content. The CCR
test provides insight into the tendency of a fuel to coke. In the case of RFO,
this test is expected to be replaced in the near future by the Micro Carbon Resi-
due Test (ASTM D 4530) because of the relatively wide variation in CCR test
results (14–20 %). However, even with the expected improvement in precision
and a considerable reduction in effort, the results are likely to remain of
marginal value. This is because the test methods themselves are based on
Vol. 2 HEATING OIL 313

low-temperature carbonization, whereas the process that occurs in a motor or a


burner is an overstoichiometric combustion, an entirely different phenomenon.
Asphaltene content is more significant, although it is not the subject of any
standards. By definition, asphaltenes are substances that are precipitated from an
oil sample by addition of a 30-fold excess of n-heptane, and that are soluble in pure
toluene. Asphaltenes have a molecular mass in the range of 5000–20 000, with
particle sizes of 1–10 nm. Asphaltenes often agglomerate because of polar forces.
Ash Content: Catalyst Fines. The ash from a residual fuel is defined as
the oxide residue remaining after complete conversion of the fuel (ashing tem-
perature 775  25 8C). The ash content of commercial RFO is typically between
0.02 and 0.1 wt %. The most important ash components from a technical stand-
point are those containing vanadium, sodium, nickel, aluminum, and silicon.
Sodium reacts with sulfur to produce sodium sulfate, Na2SO4 (mp 880 8C).
Vanadium oxidizes to vanadium pentoxide, V2O5 (mp 670 8C). A eutectic mix-
ture of these two compounds melts at 565 8C. In addition, provided the condi-
tions are appropriate (e.g., a critical Na : V molar ratio in the fuel of 1 : 3),
Na2O and V2O5 can combine to give vanadyl vanadates. The most commonly
observed forms are vanadate I (Na2O V2O4 5 V2O5, mp 559 8C) and vanadate
II (5 Na2O V2O4 11 V2O5, mp 577 8C). The combination of the aforementioned
materials leads to deposits on the heat transfer sections of boilers and in the
combustion chambers of diesel engines or turbo-chargers. Vanadate melts
cause high-temperature corrosion of such thermally stressed components as
superheater surfaces and exhaust valves. Countermeasures include careful
choice of construction materials and appropriate component design (eg, maxi-
mum steam temperatures in RFO-fired boilers of 540 8C, exhaust valve cooling).
Small amounts of aluminum and silicon are introduced into RFO in the
form of fine zeolite residues (‘‘cat fines’’) during catalytic cracking. With a particle-
size distribution up to ca. 20 mm coupled with distinct oxidizing characteristics
these materials are responsible for significant amounts of abrasive wear, espe-
cially in diesel motors. Nickel residues are converted by combustion into both
water-soluble (nickel sulfite, nickel sulfate) and water-insoluble substances
(nickel oxide, nickel carbonyl).
Limits with respect to the metals vanadium and aluminum have now been
added to the norms for RFO. ISO 8217 specifies a maximum vanadium content of
150–600 mg/kg depending on the product classification, and a maximum alumi-
num content (regarded as an indicator of the level of cat fines) of 30 mg/kg.
Net and Gross Calorific Value Net calorific values of most residual
fuels are in the range of 39.82–41.53 MJ/kg (11.06–11.54 kWh/kg). Systematic
investigations by MARDER have shown a direct relationship between the net
calorific value of a petroleum fuel and its density and sulfur content. Net calorific
values can be calculated with sufficient accuracy by the use of the following equa-
tions (modified Marder equations, neglecting the contribution of ash and based
on anhydrous material):

LHV ¼ 52:91911:930  103  d15 0:293  SðMJ=kgÞ


¼ 14:7003:314  103  d15 0:081  SðkWh=kgÞ

where d 15 is the density at 15 8C (kg/m3) and S is sulfur content (wt%)


314 HEATING OIL Vol. 2

Table 7. Environmental Consequences of Fuel Oil Combustion


Reaction Product concentration in flue
Starting materials products gas (based on 3 % O2)
Metals, solid inorganic ash particulates, 50– 200 mg/m3
foreign matter (0.1 %)
Carbon (ca. 87.5 %) soot particulates, 50– 200 mg/m3
Carbon – combustion air CO 50–100 mg/m3
Carbon – combustion air CO2 inert, though contributing to the
‘‘greenhouse effect’’
Hydrogen – combustion H2O inert, though contributing to the
air (ca. 10.6 %) ‘‘greenhouse effect’’
Nitrogen – combustion 95 % NO
air (0.2–0.5 %)
5 % NO2 NOx; S 700–1100 mg/m3
Sulfur – combustion air 97 % SO2
3 % SO3 SOx; S 1700–8500 mg/m3

Sediment Content. Sediments consist of inorganic or organic foreign


matter remaining after extraction of RFO with boiling toluene (ISO 3735), Sedi-
ment by Extraction). Apart from inorganic materials, sediment arises through
aging or from polymerization or condensation of various oil components. Sedi-
ment content is a subject of numerous standards; maximum observed values
for RFO are ca. 0.25 wt %. Just like the asphaltenes, sediments can be the
cause of filter clogging.
Water Content. RFO is essentially anhydrous when it leaves the refin-
ery. Nevertheless, multiple transfers at sea and the formation of condensate dur-
ing storage may result in product with a water content of 0.1–0.5 %. This can
cause foaming and spattering when the material is heated above 100 8C and
can also lead to burner problems.
2.5. Storage, Transportation, and Handling. From a legal standpoint
and with respect to the risk of fire or water contamination, RFO is not a danger-
ous substance (flash point 70 –100 8C, ignition temperature > 220 8C, explosive
limits in air 0.5–6.5 vol%). Thus, RFO can be stored in single-walled tanks
with a capacity of 50–50 000 m3 without special precautions (e.g., fire extinguish-
ing equipment, protective dikes). For safety reasons the storage temperature
should be kept 10–15 8C below the flash point. Hydrogen sulfide may accumulate
in the vapor space of an RFO storage tank, so vapor inhalation and skin contact
should be avoided. Cleanup should be effected with ordinary cleansing agents;
solvents should not be employed.
Water transport (ocean, inland waterways) may utilize bulk carrier or
barge (500–1000 t); land transport is normally by railroad tank car (30–50 t)
or tank truck (24–28 t). Transport temperature must be maintained high enough
so that the maximum pumping viscosity of 600 –800 mm2/s is not exceeded (eg, at
least 45 8C with a nominal viscosity at 100 8C of 50 mm2/s). Unloading normally
entails a centrifugal or screw-type pump, some RFO tank trucks are equipped to
employ pressure supplied from exhaust gas either with or without an auxiliary
Vol. 2 HEATING OIL 315

compressor. On-site transfer generally involves either permanently installed


pipelines or oil-tight hoses with appropriate couplings.
2.6. Environmental Protection. Residual fuel oil itself is not a source
of environmental problems, but its combustion products are. The relationship
between RFO components and air contamination contribution may be deduced
from Table 7, which presents a schematic overview of the combustion process
in the absence of flue gas treatment.
Environmental considerations have led to worldwide efforts to enact legisla-
tion for drastically reducing the emission of harmful materials. The resulting emis-
sion limits can be met by consumers of large amounts of RFO only with the aid of
flue gas treatment facilities, installation of which is proceeding rapidly. The most
common secondary treatment measures at steam generating plants include:

1. Electrotatic precipitators (particulate removal)


2. Desulfurization plants (DESOX: wet gypsum method of SOx removal)
3. Denitrification plants (DENOX: selected catalytic design for NOx removal)

These secondary treatment methods may be supplemented by primary


treatment directly at the burner. This is intended to minimize the formation of
hazardous materials — especially CO and NOx — directly at their source. Staged
air/fuel combustion plays an important role in primary treatment, as does flue
gas recycling (‘‘low NOx burners’’).
The EEC recommendation of November 24, 1988 sets the following limits
with respect to liquid RFOs as a function of facility size:

SOx ¼ 500–1700 mg/m3


Particulates ¼ 50 mg/m3
NOx ¼ 450 mg/m3

Desulfurization is mandatory above 400 MW thermal output (new con-


struction), EEC member states are expected to adopt these guidelines by
July 1, 1990. The guidelines are intended to apply to existing as well as new
facilities, and should assist the states in meeting their obligation to reduce
total emissions. The United States, Japan, and the Scandinavian countries
have adopted pollution control legislation in which targets are based upon
local emission data

REFERENCES

GENERAL REFERENCES

1. H. M. Spiers, J. W. Rose, J. R. Cooper: Technical Data on Fuel, 7th ed. British


National Committee, Scottish Academic Press, Edinburgh 1977.
2. W. Hansen: Ölfeuerungen-Brennstoff, Technische Einrichtungen, Anwendungen,
2nd ed., Springer Verlag, Berlin, Heidelberg, New York 1970.
3. Deutsche Babcock AG, H. P. Niepenberg (ed.): Industrie-Ölfeuerungen, Oberhausen,
Federal Republic of Germany, 1968.
316 HEATING OIL Vol. 2

4. G. H. Clark: Industrial and Marine Fuels, Butterworths, London 1988.


5. S. H. Kägler (ed.): Neue Mineralölanalyse, vol. 1, Spektroskopie, 2nd ed., Hüthig
Verlag, Heidelberg 1987.
6. C. Zerbe: Mineralöle und verwandte Produkte, 2nd ed., Springer Verlag, Berlin,
Heidelberg 1969.
7. Shell International Trading Co., Shell Marine Service, Know Your Fuels, Shell
Marine Fuel Specifications, London 1987
8. Brennstoff, Wärme, Kraft, VDI Verlag, Düsseldorf.
9. Die Industriefeuerung, H. P. Niepenberg (ed.), Vulkan Verlag, Essen.
10. Reports of the VGB in: VGB Kraftwerkstechnik Technische Vereinigung der
Großkraftwerksbetreiber, Essen.
11. Institut für Wirtschaftliche Oelheizung E. V. (IWO): ‘‘Heizöl EL Produkt und
Anwendung,’’Hamburg 2003. www.iwo.de/sites/iwo/iwo_fachbroschueren/pdf/HEL_
Broschuere_06_11_2003.pdf
12. P. Derst: ‘‘Alterungsmechanismen bei Mitteldestillaten. Eine Literaturrecherche,
‘‘DGMK Forschungsbericht 484, Hamburg 1995.
13. DIN 51371 (in preparation), Bestimmung der thermischen Stabilität von Heizöl EL.
14. DGMK-Forschungsbericht 646, Kriterienkatalog für Heizöl EL Additive, DGMK
2005.
15. J. Lohmar, S. Laurs, O. Akin, P. Beiss, Sanitär und Heizungstechnik 60 (1995)
no. 12, 55–63.
16. J. Lohmar, S. Laurs, O. Akin, P. Beiss, Sanitär und Heizungstechnik 61 (1996) no. 1,
54–74.
17. K. Lucka: ‘‘Beschreibung der Reaktionsphänomene in Brennstoff-Luft-Gemischen
unterhalb der Selbstzündung,’’ Habilitationsschrift (in preparation), RWTH Aachen
2006.
18. C. Küchen, Wärmetechnik 39 (1994) 590–601.
19. DGMK-Forschungsbericht 601, Ermittlung eines technisch begründeten
Grenzwertes für Schwefelgehalte im Heizül EL zur Förderung des Einsatzes von
Brennwerttechnologie in Ölheizungsanlagen, DGMK 2002.
20. Handbuch ‘‘Kraft–Wärme-Kopplung,‘‘ VDI–GET.
21. L. Hartmann, K. Lucka, H. Köhne: ‘‘Mixture preparation by cool flames for diesel
reforming technologies,’’ J. Power Sources 118 (2003) 286–297.
22. U. Feisst, Wärmetechnik 35 (1990) 107–122.
23. C. Hübner et al., Erdöl Erdgas Kohle Petrochem. 112 (1996) 170–174.
24. F. Pfeiffer, M. Struschka, G. Baumbach, Erdöl Erdgas Kohle Petrochem. 113 (1997)
137–140.
25. Energy Information Administration, Official Energy Statistics of the U.S. Govern-
ment http://tonto.eia.doe.gov/dnav/pet/pet_cons_821use_dcu_nus_a.htm.

CHRISTIAN KÜCHEN
Institut für Wirtschaftliche Oelheizung
Hamburg, Germany
KNUT SPITZMÜLLER
Hamburg, Germany
Vol. 2 NAPHTHENIC ACIDS 317

NAPHTHENIC ACIDS

The term naphthenic acid, as commonly used in the petroleum industry, refers
collectively to all of the carboxylic acids present in crude oil. Naphthenic acids
[1338-24-5] are classified as monobasic carboxylic acids of the general formula
RCOOH, where R represents the naphthene moiety consisting of cyclopentane
and cyclohexane derivatives. Naphthenic acids are composed predominantly
of alkyl-substituted cycloaliphatic carboxylic acids, with smaller amounts of
acyclic aliphatic (paraffinic or fatty) acids. Aromatic, olefinic, hydroxy, and diba-
sic acids are considered to be minor components. Commercial naphthenic acids
also contain varying amounts of unsaponifiable hydrocarbons, phenolic com-
pounds, sulfur compounds, and water. The complex mixture of acids is derived
from straight-run distillates of petroleum, mostly from kerosene and diesel frac-
tions (see PETROLEUM).
Naphthenic acids have been the topic of numerous studies extending over
many years. Originally recovered from the petroleum distillates to minimize
corrosion of refinery equipment, they have found wide use as articles of com-
merce in metal naphthenates and other derivatives. A comprehensive overview
of the uses of naphthenic acid and its derivatives can be found in References 1
and 2. A review of the extensive research on carboxylic acids in petroleum
conducted up to 1955 is available (3), as is a more recent review of purification,
identification, and uses of naphthenic acid (4).

1. Chemical Structure

The name naphthenic acid is derived from the early discovery of monobasic
carboxylic acids in petroleum, with these acids being based on a saturated single-
ring structure. The low molecular weight naphthenic acids contain alkylated
cyclopentane carboxylic acids, with smaller amounts of cyclohexane derivatives
occurring. The carboxyl group is usually attached to a side chain rather than
directly attached to the cycloalkane. The simplest naphthenic acid is cyclopen-
tane acetic acid [1123-00-8], (1, n ¼ 1).

CH2
H2C CH (CH2)n COOH
H2C CH2

(1)

Naphthenic acids are represented by a general formula CnH2n-z O2, where n


indicates the carbon number and z specifies a homologous series. The z is equal to
0 for saturated, acyclic acids and increases to 2 in monocyclic naphthenic acids,
to 4 in bicyclic naphthenic acids, to 6 in tricyclic acids, and to 8 in tetracyclic
acids. Typical structures for the homologues of naphthenic acids are shown
in Figure 1. Naphthenic acids in the range of C-7 to C-12 consist mainly of
monocyclic acids. The more complex acids contain larger proportions of multicyc-
lic condensed compounds.

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.1401160802180905.a01
318 NAPHTHENIC ACIDS Vol. 2

Monocyclic Bicyclic Tricyclic Tetracyclic


z=2 z=4 z=6 z=8

COOH COOH

R COOH R COOH R R

COOH COOH

R COOH R COOH R R

COOH

R COOH R

Fig. 1. Typical naphthenic acid structures, where R ¼ alkyl. For the acyclic case z ¼ 0
and the structure is simply R COOH.

The fundamental nature of naphthenic acids was determined by the 1960s


(3,5,6). As recently as 1955, only two naphthenic acids with as many as
10 carbon atoms had been positively identified. Later, more extensive laboratory
studies revealed an astonishing variety of organic acids present in crude oil,
including fatty acids as low in molecular weight as acetic acid, as well as
saturated and unsaturated acids based on single and multiple five- and six-
membered rings. One study identified 1500 different organic acids in a single
California crude oil, ranging in molecular weight between 200 and 700. The peak
of the molecular weight distribution is between 300 and 400 (7). Early studies
concentrated on identification of individual low molecular weight acids; more
recent work has focused largely on geochemical correlations and biodegradation
mechanisms of petroleum (8). C-14 to C-20 acyclic isoprenoid acids and C-10 to
C-11 monocyclic naphthenic acids have been isolated from California crude pet-
roleum (9–12). California crudes have also yielded C-22 to C-24 steroid acids,
affirming the biological origin of petroleum (13).

2. Physical and Chemical Properties

Naphthenic acids are viscous liquids, with phenolic and sulfur impurities present
that are largely responsible for their characteristic odor. Their colors range from
pale yellow to dark amber. An odor develops upon storage of the refined
acids. Naphthenic acids have wide boiling point ranges at high temperatures
(250–3508C). They are completely soluble in organic solvents and oils but are
insoluble ð <50 mg=LÞ in water. Commercial naphthenic acids are available
in various grades and are marketed by acid number, impurity level, and color.
Vol. 2 NAPHTHENIC ACIDS 319

Table 1. Properties of Commercial Naphthenic Acid


Grade
Property Crude Refined Highly refined
acid number, mg KOH=g 150–200 220–260 225–310
acid number (oil-free) 170–230 225–270 230–315
unsaponifiables, wt% 10–20 4–10 1–3
phenolic compounds, wt% 2–15 0.1–0.4 0.05–0.4
water, wt% 0.3–1.0 0.01–0.1 0.01–0.08
specific gravity at 208C 0.95–0.98 0.95–0.98 0.95–0.98
viscosity at 408C, mPa  sð ¼cPÞ 40–80 40–100 50–100
color, Gardner black 6–8 5–6
refractive index, n20
D
1.482 1.478 1.475
avg. mol wt (oil-free) 240–330 210–250 180–250

Typical compositions and properties of three grades of naphthenic acids are


given in Table 1. The product of refractive index and specific gravity, approxi-
mately 1.40–1.45, can be used to distinguish naphthenic acids from unsaponifi-
ables (ca 1.2) and from other acidic compounds found in petroleum such as
aliphatic acids (ca 1.3) and phenols (>1:5) (14).
Chemically, naphthenic acids behave like typical carboxylic acids with
similar acid strength as the higher fatty acids. Dissociation constants are
on the order of 105 to 106 (15). They are slightly weaker acids than low
molecular weight carboxylic acids such as acetic, but are stronger acids than
phenol and cresylic acid. Whereas the principal use of naphthenic acids has
been in the production of metal salts, they also can react to form esters,
amine salts, amides, imidazolines, and many other derivatives. Derivatives
of naphthenic acid often have advantageous physical properties over fatty
acid derivatives, particularly their high stability to oxidation and solubility
in hydrocarbons.
Naphthenic acid corrosion has been a problem in petroleum-refining opera-
tions since the early 1900s. Naphthenic acid corrosion data have been reported
for various materials of construction (16), and correlations have been found
relating corrosion rates to temperature and total acid number (17). Refineries
processing highly naphthenic crudes must use steel alloys; 316 stainless
steel [11107-04-3] is the material of choice. Conversely, naphthenic acid
derivatives find use as corrosion inhibitors in oil-well and petroleum refinery
applications.

3. Occurrence

Naphthenic acids are normal constituents of nearly all crude oils, but not all
crudes contain sufficient quantities of usable acids to make recovery an economic
process. Heavy crudes from geologically young formations have the highest acid
content, and paraffinic crudes usually have low acid content. The acid content
320 NAPHTHENIC ACIDS Vol. 2

Table 2. Acid Content of Various Crudesa


Crude oil source Petroleum acids, wt%
Pennsylvania 0.03
West Texas 0.4
Gulf Coast 0.6
California 1.5
Russia, Balakhany light 1.0
Russia, Balakhany heavy 1.6
Romania, waxy 0.2
Romania, asphaltic 1.6
Venezuela, Lagunillas 1.2
a
Ref. 3.

of crude petroleum varies from 0–3%, with crudes from California, Venezuela,
Russia, and Romania having the highest content. Smaller amounts are found
in U.S. Gulf Coast crudes, whereas little or no naphthenic acids are found in
Pennsylvania, Iraq, or Saudi Arabia crudes. Typical concentrations are shown
in Table 2. Minor amounts of naphthenic acids are also found in bituminous
oil sands, but these are not economically recoverable. Identification of naphthe-
nic acids in water from oil-bearing strata is being examined as a potential
method of petroleum exploration (18).
Naphthenic acids occur in a wide boiling range of crude oil fractions, with
acid content increasing with boiling point to a maximum in the gas oil fraction
(ca 3258C). Jet fuel, kerosene, and diesel fractions are the source of most
commercial naphthenic acid. The acid number of the naphthenic acids
decreases as heavier petroleum fractions are isolated, ranging from 255 mg
KOH/g for acids recovered from kerosene and 170 from diesel, to 108 from
heavy fuel oil (19). The amount of unsaturation as indicated by iodine number
also increases in the high molecular weight acids recovered from heavier
distillation cuts.

4. Manufacture

The commercial production of naphthenic acid from petroleum in based on the


formation of sodium naphthenate. Although the separation of naphthenic acids
from hydrocarbon fractions is a relatively simple process, extraction from crude
petroleum is not feasible because of the low percentage of acids in the oil,
the large volumes of fluids to be handled, and the abundance of other caustic-
extractable compounds found in most crudes. The low water solubility of high mole-
cular weight naphthenate soaps and the tendency of sodium naphthenate to emul-
sify also contribute to making direct recovery of acids from crude oil difficult.
Although naphthenic acids can be removed from crude oil and heavy fractions by
reaction with caustic soda prior to and during their distillation, the acids are prob-
ably unusable since they are mixed (as salts) with large quantities of asphalt or
residual pitch. The numerous patents on this subject are basically modifications
to an early patent for distilling oil in the presence of sodium hydroxide (20). The
Vol. 2 NAPHTHENIC ACIDS 321

alkali residues are extracted with alcohol to remove unsaponifiables, followed by


acidification to recover the naphthenic acids (21).
Naphthenic acids are generally obtained by caustic extraction of petroleum
distillates boiling between 200 and 3708C. A continuous process has been
developed for removing naphthenic acids from refinery streams by caustic wash-
ing (22). Caustic extraction also removes other acidic components of the petro-
leum fraction, including phenol and cresols (cresylic acid), mercaptans, and
thiophenols. In addition to reducing corrosion in the refinery, the caustic wash
is necessary to improve the burning qualities, storage stability, and odor of
the finished kerosene and diesel fuels. The petroleum fractions are extracted
with dilute (2–10%) sodium hydroxide since the sodium naphthenate salts
are emulsifying agents. Stronger caustic strengths increase the solubility of
hydrocarbon oils (unsaponifiables) in the sodium naphthenate. The Fiber-
Film contacting process patented by Merichem reduces emulsification and
caustic carryover during removal of naphthenic acid from petroleum fractions
and achieves low acid number specifications in a single stage (23,24). Noncaus-
tic processes for recovery of naphthenic acids from petroleum distillates,
including ammonia (qv) (25), triethylene glycol (26), ion-exchange resins (27),
and aluminosilicate zeolites (28), have also been reported but not commercia-
lized (see ION EXCHANGE).
The aqueous sodium naphthenate phase is decanted from the hydrocarbon
phase and treated with acid to regenerate the crude naphthenic acids. Sulfuric
acid is used almost exclusively, for economic reasons. The wet crude naphthenic
acid phase separates and is decanted from the sodium sulfate brine. The volume
of sodium sulfate brine produced from dilute sodium naphthenate solutions is
significant, on the order of 10 L per L of crude naphthenic acid. The brine
contains some phenolic compounds and must be treated or disposed of in an
environmentally sound manner. Sodium phenolates can be selectively neutra-
lized using carbon dioxide and recovered before the sodium naphthenate is
finally acidified with mineral acid (29). Recovery of naphthenic acid from aqu-
eous sodium naphthenate solutions using ion-exchange resins has also been
reported (30).
The other acidic compounds extracted by caustic remain in the crude
naphthenic acid after acidification of the sodium naphthenate has occurred.
These phenolic and sulfur compounds are objectionable because of their odor
and color-forming properties. Although numerous methods for naphthenic acid
purification have been suggested, only those which combine low cost and rela-
tively simple operation are used commercially. Crude naphthenic acids are
dried and distilled under reduced pressure to produce the refined products.
Distillation removes some of the phenolic and unsaponifiable impurities, but
the refined products still contain significant levels of impurities boiling at the
same temperature. Because it is difficult and costly to remove all of the hydrocar-
bon impurities, commercial refined naphthenic acids typically contain 5–10%
unsaponifiables. Unsaponifiables can be reduced to <5% by distilling the hydro-
carbons from the alkali salts of the naphthenic acids (31). Naphthenic acids can
also be purified by treatment with concentrated sulfuric acid to precipitate
sludge-forming materials, which reduces the content of phenolics (32). The
naphthenic acid is then distilled for final purification. Liquid–liquid extraction
322 NAPHTHENIC ACIDS Vol. 2

processes reduce unsaponifiable content by extracting naphthenic acids into


water–alcohol mixtures (33,34).
Interest in synthetic naphthenic acid has grown as the supply of natural
product has fluctuated. Oxidation of naphthene-based hydrocarbons has been
studied extensively (35–37), but no commercially viable processes are known.
Extensive purification schemes must be employed to maximize naphthene con-
tent in the feedstock and remove hydroxy acids and nonacidic by-products
from the oxidation product. Free-radical addition of carboxylic acids to olefins
(38,39) and addition of unsaturated fatty acids to cycloparaffins (40) have also
been studied but have not been commercialized.

5. Production

Nameplate capacities of naphthenic acid producers in North America are 9000


metric tons of crude and refined acid at Merichem (Tuscaloosa, Ala.), and 3600 t
of crude acid at Hewchem (Gulfport, Miss.) (41). However, actual production
capacity may vary widely as a result of the mix of feedstocks being processed.
Some feedstocks require significantly greater processing time to achieve high
grade finished product. Naphthenic acid products are shipped in tank cars,
tank trucks, and drums under DOT 9137/UN 3082 identification numbers. Prin-
cipal producers in Japan are Sanko Yuka Kogyo, Yamato, Taniguchi Oil, and
Union Oil (a subsidiary of Idimitsu); in Europe, Nord Import, Oleochimie, Imper-
ial Oil, and Corn Van Loocke.

6. Economic Aspects

Roughly 5500–6000 t of naphthenic acid were consumed in North America


in 1992. Less than 500 t of finished product demand was met by imports
in spite of the large quantities of naphthenic acid produced overseas. The
average unit value was $1.56/kg. After much price volatility in the late
1970s and early 1980s, prices have risen moderately in the 1990s. For fob
bulk shipments, current refined acid prices range from $1.94–2.09/kg
for acid numbers 220 to 300. Crude acid prices range from $0.99–1.17/kg for
acid numbers 150 to 180 (41).
Naphthenic acid availability exceeds demand, although some minor mar-
ket disruptions occurred in North America during the early 1990s. In 1990, the
U.S. Environmental Protection Agency (EPA) put into place low sulfur require-
ments for highway diesel fuel to be met by October 1993. To meet the specifica-
tion, several feedstock producers were prompted to hydrotreat their diesel
fractions, since caustic extraction does not reduce sulfur to the required
level. Hydrotreating destroys naphthenic acids, so the result was a decrease
in feedstock availability. Raw materials from previously unused feedstock
sources in Asia, Europe, and South America were then imported into North
America to achieve market balance. Long-term yearly feedstock supply is
expected to meet market growth projections with large discoveries of high
naphthenic-content oil.
Vol. 2 NAPHTHENIC ACIDS 323

7. Analysis

Naphthenic acid concentration in crude oil and distillates is typically measured


by titrating with potassium hydroxide for neutralization number using potentio-
metric (ASTM D664) or colorimetric (ASTM D974) methods. The same procedures
are used to measure the acid number of crude or refined naphthenic acids after
recovery from the petroleum fraction. The neutralization or acid numbers are
expressed numerically as mg KOH (formula wt ¼ 56:1) required to neutralize
the acidity in 1 g of sample. This value for the acid on an impurity-free basis is
readily converted to equivalent weight by dividing 100 the formula weight of
KOH (56,100) by the acid number. Naphthenic acid having a high acid number
indicates a low molecular or equivalent weight product. Crude oil or a petroleum
fraction would have a very low acid number, since the acid content is low.
Naphthenic acids comprise a highly complex mixture containing hundreds
of compounds that are impossible to separate into individual components, even
by high resolution gas chromatography. Progress in obtaining information on the
structure of naphthenic acids has been retarded because of this complexity. Ana-
lysis by gas chromatography of methyl esters is useful in detecting naphthenic
acids adulterated with synthetic or vegetable oil-based fatty acids. Infrared spec-
troscopy is unable to distinguish clearly between naphthenic and fatty acids.
Two mass spectrometric methods have been developed which improve the
characterization of naphthenic acids. Chemical ionization using fluoride ion as
a reactant gas (42) and fast-atom bombardment (43) both use the negative ion
detection mode. The two methods detect the ðM  1Þ ion in carboxylic acids,
including high molecular weight species with no fragmentation. These methods
allow the identification of naphthenic acids based on carbon number and z-series
distributions, providing a fingerprint useful in correlating the sources of crude
oils.

8. Health and Safety Factors

Naphthenic acids are only slightly toxic to mammals but are toxic to fish,
bacteria, and wood-destroying insects. The lethal oral dose for humans is
approximately 1 L (44), and the oral LD50 is 3.0–5.2 g/kg in rats (45). The deaths
appeared to result from gastrointestinal disturbances. Naphthenic acid is non-
mutagenic by the Ames mutagenicity test (46), and is not listed as a carcinogen
by the International Agency for Research on Cancer, the National Toxicology
Program, or the U.S. Occupational Safety and Health Act.

9. Commercial Uses

More than two-thirds of the naphthenic acid produced is used to make metal
salts, with the largest volume being used for copper naphthenate, consumed
in the wood preservative industry (see WOOD). Metal salts used as paint driers
accounted for only 16% of the naphthenic acid market in 1993. This is a
dramatic contrast with 1977 usage, when 75% of the naphthenates went into
324 NAPHTHENIC ACIDS Vol. 2

the paint drier market. An overall view of the 1993 naphthenic acid market in
North America shows the following uses:

Use Percentage
wood 40.1
oil field 27.7
paint 15.8
tires 8.5
miscellaneous 7.9

Oil field uses are primarily imidazolines for surfactant and corrosion inhi-
bition (see PETROLEUM). Besides the lubrication market for metal salts, the mis-
cellaneous market is comprised of free acids used in concrete additives, motor oil
lubricants, and asphalt-paving applications (47) (see ASPHALT; LUBRICATION AND
LUBRICANTS). Naphthenic acid has also been studied in ore flotation for recovery
of rare-earth metals (48).
Naphthenic acid is ideal for synthesizing metal carboxylates that re-
quire a ligand with some oxidative stability, solubility in hydrocarbons
and oils, and insolubility in water. The general chemical structure of the
naphthenate salts is Mxþ(OOCR)x where x ¼ metal charge and molar stoichio-
metry of naphthenic acid to the metal M. Table 3 lists commercially available
metal naphthenates and their uses. Metal carboxylates are used in the oil-
based (alkyd) paint industry as catalysts to accelerate the drying of coatings;
hence the name driers was given to them. The metal naphthenates were origi-
nally used to replace the corresponding linoleates, resinates, and tallates.
These naphthenate salts have the advantages of high metal content, low viscos-
ity, better solubility in hydrocarbons and oils, and greater stability to oxidation.

Table 3. Metal Naphthenate Uses


CAS Registry
Name Number Applications
copper naphthenate [1338-02-9] wood and textile preservative, catalyst
zinc naphthenate [12001-85-3] wood and textile preservative, lubricant,
wetting agent
cobalt naphthenate [61789-51-3] paint drier, tires, ink drier, catalyst
manganese naphthenate [1336-93-2] paint drier, catalyst, fuel additive
lead naphthenate [61790-14-5] paint drier, wetting agent, lubricant
additive
calcium naphthenate [61789-36-4] paint drier, catalyst, lube additive
iron naphthenate [1338-14-3] paint drier, fuel additive, catalyst
zirconium naphthenate [72854-21-8] paint drier, electrophotographic developer
cerium naphthenate [68514-63-6] paint drier, catalyst, fuel additive
vanadium naphthenate [68815-09-8] paint drier, catalyst, corrosion inhibitor
sodium naphthenate [61790-13-4] emulsifier, ore flotation, leather
potassium naphthenate [66072-08-0] emulsifier, plant growth modifier
aluminum naphthenate [61789-64-8] gelling agent, pigment wetting
Vol. 2 NAPHTHENIC ACIDS 325

Each metal naphthenate has a specific function and is typically used in


combination with other driers. This market is declining as a result of the
1990 Clean Air Act that reduces volatile organic compound (VOC) emissions.
The paint industry is responding by replacing their oil-based paints with
waterborne formulations.
An expanding market for naphthenic acid is in wood and textile preserva-
tives (see TEXTILES; WOOD). Copper and zinc naphthenate are used to prevent dry
rot, as well as fungi and insect attack in wood. Zinc naphthenate is typically used
in the log home industry because it provides a colorless coating on the wood. The
naphthenates are environmentally preferred replacements for creosote, penta-
chlorophenol, and chromated copper–arsenic. Copper naphthenate prevents tex-
tile rotting at lower concentrations than copper oleates or tallates, in part
because the naphthene moiety itself has fungicidal properties (49). In North
America, the 1992 production of 8% copper naphthenate solution was estimated
to be 2600–2900 t at a unit value of $2.20–2.45/kg. The estimated market growth
for this application is 12–14%/yr.
Another market application for naphthenic acid is the tire industry, where
cobalt naphthenate is used as an adhesion promoter. Cobalt naphthenate
improves the bonding of brass-plated steel cords to rubber, presumably by sup-
pressing the de-zincification of brass (50). Its first reported use was in 1970 and
the first patent for its use was issued in 1975 (51). About 900 t of cobalt
naphthenate is used worldwide as an adhesion promoter, half of it in North
America. The unit value fluctuates between $8.75–13.25/kg because of the vola-
tility of cobalt prices. Although it is the industry standard, the use of cobalt
naphthenate is declining with the advent of more economical high metal-
containing substitutes.
In the other market areas, lead naphthenates are used on a limited basis in
extreme pressure additives for lubricating oils and greases. Sodium and potas-
sium naphthenates are used in emulsifiable oils, where they have the advantage
over fatty acid soaps of having improved disinfectant properties. Catalyst uses
include cobalt naphthenate as a cross-linking catalyst in adhesives (52) and
manganese naphthenate as an oxidation catalyst (35). Metal naphthenates are
also being used in the hydroconversion of heavy petroleum fractions (53,54)
and bitumens (55).
The surface-active properties of many naphthenic acid derivatives allow
them to function in a variety of applications, including corrosion inhibitors,
emulsifiers, and defoamers (see CORROSION AND CORROSION CONTROL; EMULSIONS;
FOAMS). These properties can be exploited in performance-oriented applications
such as petroleum recovery, asphalt emulsification, and concrete aeration
where chemical composition is not critical (see ASPHALT; PETROLEUM). Surfactants
may be prepared from the acids themselves or from derivatives such as amides or
esters. The general surfactant applications are dominated by the ethoxylated
naphthenic acids and naphthenyl amides and alcohols. Ethoxylated and sulfated
naphthenic acid derivatives are highly surface-active and have substantial deter-
gency power. The petroleum industry utilizes naphthenic acid amine derivatives
both as surfactants for enhanced (tertiary) oil recovery (56) and as corrosion inhi-
bitors (57). Amides and imidazolines of naphthenic acid used as corrosion inhibi-
tors are noted for oil solubility and flow characteristics that are superior to those
326 NAPHTHENIC ACIDS Vol. 2

of similar fatty acid-based derivatives. Naphthenyl amides and imidazolines


based on polyamines such as diethylenetriamine are also reported in such
diverse applications as bitumen emulsifiers, bactericides (58), and lubricant
additives (see LUBRICATION AND LUBRICANTS). The surfactant properties of naphthe-
nyl amides improve adhesion of aggregate rock to the asphalt binder (59) and
improve mechanical strength and plasticity in concrete (60). Alkaline flooding
of oil sands improves bitumen extraction by producing anionic surfactants in
situ from the naturally occurring naphthenic acids (61).
Simple esters are readily formed by the reaction of naphthenic acids with
monohydric alcohols, olefins, or ethylene or propylene oxides. More complex
esters are prepared by partial or complete esterification of polyhydric alcohols.
Naphthenic acid esters have been repeatedly cited as replacements for phtha-
lates as plasticizers for PVC resins. Alkyl and glycol naphthenate esters
improve the flexibility and workability of resins and are valued for their low
volatility, compatibility, stability, and resistance to kerosene extraction (62).
Naphthenic acid esters of multifunctional alcohols such as pentaerythritol
(63) and poly(ethylene glycol)s have been cited as lubricants or additives for
fuels, and for improved oil recovery (64). Glycol esters and other esters of
naphthenic acid are used as hydraulic fluids (65), plasticizers, and surfactants,
and are finding increasing use in the fat-liquoring process in leather tanning
(66). Triethylene glycol naphthenate improves dispersion of carbon black in
water-borne conductive primer coatings (67).
Naphthenyl alcohols are formed by reduction of the acids or their simple
esters. They are valuable as surfactants, solvents, and components of lubricants.
The acid halides are of value mainly as chemical intermediates (1).

BIBLIOGRAPHY

‘‘Naphthenic Acids’’ in ECT 1st ed., Vol. 9, pp. 241–246, by V. L. Shipp, Socony-Vacuum
Oil Co; in ECT 2nd ed., Vol. 13, pp. 727–734, by S. E. Jolly, Sun Oil Co.; in ECT 3rd ed.,
Vol. 15, pp. 749–753, by W. E. Sisco, W. E. Bastian, and E. G. Weierich, CPS
Chemical Co.

CITED PUBLICATIONS

1. E. S. Lower, Specialty Chem. 7, 76 (1987); 7, 282 (1987); 8, 174 (1988); 9, 135 (1989);
9, 267 (1989).
2. E. S. Lower, Pigment Resin Technol. 15(5), 4 (1986); 15(6), 4 (1986); 15(7), 11 (1986);
15(8), 6 (1986); 15(9), 8 (1986); 15(10), 9 (1986); 15(11), 7 (1986); 15(12), 6 (1986).
3. H. L. Lochte and E. R. Littmann, The Petroleum Acids and Bases, Chemical Publish-
ing Co. Inc., New York, 1955.
4. G. Narmetova, B. Khamidov, N. Ryabova, and E. Aripov, Purification, Identification,
and Use of Naphthenic Acid, Fan, Tashkent, former USSR, 1983.
5. N. Zelinsky and A. Chukasanov, Ber. 57, 42 (1924).
6. J. Von Braun, The Science of Petroleum, Vol. II, Oxford University Press, London,
1938, 1007–1015.
Vol. 2 NAPHTHENIC ACIDS 327

7. W. Seifert, in W. Herz, H. Grisebach, and G. Kirby, eds., Progress in the Chemistry of


Organic Natural Products, Vol. 32, Springer-Verlag, New York, 1975, 1–49.
8. A. Mackenzie, G. Wolff, and J. Maxwell, in M. Bjoroy and co-workers, eds.,
Advances in Organic Geochemistry, John Wiley & Sons, Ltd., Chichester, U.K.,
1981, p. 637.
9. J. Cason and A. Khodair, J. Org. Chem. 32, 3430 (1967).
10. J. Cason, Tetrahedron 21, 471 (1965).
11. J. Cason and K. Liauw, J. Org. Chem. 30, 1763 (1965).
12. J. Cason and A. Khodair, J. Org. Chem. 31, 3618 (1966).
13. W. Seifert, A. Gallegos, and R. Teeter, J. Am. Chem. Soc. 94, 5880 (1972).
14. H. Schutze, B. Shive, and H. Lochte, Ind. Eng. Chem. Anal. Ed. 12, 262 (1940).
15. J. Petkovic and M. Ristic, J. Serb. Chem. Soc. 52, 641 (1987).
16. J. Heler, Mater. Protection 2, 90 (1963).
17. R. Piehl, Mater. Performance 27, 37 (1988).
18. V. Shvets,Org. Geokhim. Vod Poisk. Geokhim., Mater. Mezhdunar Kongr. Org.
Geokhim., 8th 1977, 47 (1982).
19. E. Pyhälä, Z. Angew. Chem. 27, 407 (1914).
20. U.S. Pat. 623,006 (April 11, 1899), F. Berg.
21. U.S. Pat. 1,804,451 (May 12, 1931), T. Andrews and C. Lauer (to the Texas Co.).
22. D. Todd and F. Rac, Hydrocarbon Process. 46(8), 115 (1967).
23. B. Norris and J. Files, An. Congr. Lat.-Am. Eng. Equip. Ind. Pet. Petroquim. 1st
1979, II/91-102 (1980).
24. R. Vazquez, Pet. Int. (Tulsa, Okla.) 48(5), 18 (1990).
25. U.S. Pat. 4,634,519 (Jan. 6, 1987), M. Danzik (to Chevron Research Co.).
26. M. Mardanov and co-workers, Azerb. Neft. Khoz., 54 (1980).
27. U.S. Pat. 2,936,424 (Dec. 6, 1960), G. Ayres and M. Krewer (to Pure Oil Co.).
28. R. Kuliev and co-workers, Sb. Tr.-Akad. Nauk. Az. SSR 16, 76 (1987).
29. U.S. Pat. 2,357,252 (Aug. 29, 1944), H. Berger, E. Nygaard, and H. Angel
(to Socony-Vacuum Oil Co.).
30. G. Defer, A. Abadie, and H. Roques, Trib. CEBEDEAU 28, 379 (1975).
31. U.S. Pat. 5,011,579 (Apr. 30, 1991), G. Davis (to Merichem Co.).
32. U.S. Pat. 2.081,475 (May 25, 1937), D. Carr (to Union Oil Co. of California).
33. U.S. Pat. 2,391,729 (Dec. 25, 1945), W. McCorquodale (to Sun Oil Co.).
34. U.S. Pat. 2,610,209 (Sept. 9, 1952), E. Honeycutt (to Sun Oil Co.).
35. U.S. Pat. 4,227,020 (Oct. 7, 1980), B. Zeinalov and co-workers.
36. B. Zeinalov and co-workers, Azerb. Neft. Khoz. 9, 34 (1991).
37. U.S. Pat. 2,519,309 (June 29, 1948), W. Denton (to Socony-Vacuum Oil Co.).
38. USSR Pat. 1,154,266 (May 7, 1985), B. Zeinalov and co-workers (to Institute of
Petrochemical Processes).
39. S. Sadykh-Zade and R. Dzhalilov, Dokl. Akad. Nauk. Azerb. SSR 22, 17 (1966).
40. U.S. Pat. 2,534,074 (May 2, 1947), A. Shmidl (to Standard Oil Development Co.).
41. Chemical Mktg. Rep. 244, 45 (Sept. 27, 1993).
42. I. Dzidic, A. Somerville, J. Raia, and H. Hart, Anal. Chem. 60, 1318 (1988).
43. T. Fan, Energy & Fuels 5, 371 (1991).
44. W. Rockhold, A.M.A. Arch. Ind. Health 12(5), 477 (1955).
45. Dangerous Prop. Ind. Mater. Rep. 7, 105 (1987).
46. R. Crebelli and co-workers, IARC Sci. Publ. 59, 289 (1984).
47. Brit. Pat. 1,086,390 (Oct. 11, 1967), O. Pordes and N. Bangert (to Shell International
Research).
48. Y. Ge and Q. Wang, Kuangye Gongcheng 9, 56 (1989).
49. P. Marsh, G. Greathouse, K. Bollenbacher, and M. Butler, Ind. Eng. Chem. 36, 176
(1944).
328 NAPHTHENIC ACIDS Vol. 2

50. W. van Ooij, W. Weening, and P. Murray, Rubber Chem. Technol. 54, 227 (1981).
51. U.S. Pat. 3,897,583 (July 29, 1975), C. Bellamy (to Uniroyal SA).
52. J. Donalson, S. Clark, and S. Grimes, eds., Cobalt in Catalysts, Cobalt Development
Institute, London, 1990, Chapt. 8.
53. Eur. Pat. 512,778 (Nov. 11, 1992), J. Harrison, A. Bhattacharya, M. Patel, and D.
Rao (to Texaco Development Corp.).
54. P. LePerchec and co-workers, Prepr.-Amer. Chem. Soc., Div. Pet. Chem. 38, 401
(1993).
55. H. Chen and co-workers, AOSTRA J. Res. 5, 33 (1989).
56. U.S. Pat. 3,861,467 (Jan. 21, 1975), B. Harnsberger (to Texaco Inc.).
57. U.S. Pat. 3,997,469 (Dec. 14, 1976), C. Howle (to Nalco Chemical Co.).
58. U.S. Pat. 3,788,989 (Jan. 29, 1974), P. Adams and A. Petrocci (to Millmaster Onyx
Corp.).
59. Ger. Pat. 1,103,223 (Feb. 15, 1955), M. Lhorty (to Bataasfe Petroleum).
60. USSR Pat. 1,054,322 (Nov. 15, 1983), V. Shestakov and co-workers.
61. L. Schramm and co-workers, AOSTRA J. Res. 1, 5 (1984).
62. B. Zeinalov and co-workers, J. Polymer Sci. Part C 33, 353 (1971).
63. A. Ismailov and B. Salimova, Khim. Tekhnol. 9, 79 (1967).
64. Rom. Pat. 51,410 (Nov. 7, 1968), F. Popescu and co-workers (to Romania Institute for
Drilling and Extraction Research).
65. U.S. Pat. 2,512,255 (June 20, 1950), W. Gilbert, R. Kline, and C. Montgomery (to
Gulf Research and Development).
66. B. Kochetygov and P. Levenko, Kozh. Obuv. Prom. 10, 42 (1983).
67. J. Carmine and R. Ryntz, J. Coatings Technol. 66(836), 93–98 (1994).

GENERAL REFERENCES

M. Naphthali, Chemie, Technologie, und Analyse der Naphthensäuren, Wissenschaftliche


Verlagsgesellschaft, Stuttgart, Germany, 1927.
M. Naphthali, Naphthensäuren und Naphthensulfosäuren, Wissenschaftliche Verlagsge-
sellschaft, Stuttgart, Germany, 1934.
C. Ellis, The Chemistry of Petroleum Derivatives, The Chemical Catalog Co., Inc.,
New York, 1934, 1062–1091.
A. N. Sachanen, The Chemical Constituents of Petroleum, Reinhold Publishing Corp.,
New York, 1945, 315–330.
W. A. Gruse and D. R. Stevens, The Chemical Technology of Petroleum, 2nd ed., McGraw-
Hill Book Co., Inc., New York, 1942, 97–105.
E. R. Littman and J. R. M. Klotz, Chem. Rev. 30(1), 97–111 (1942).
W. Maass, E. Buchspiess-Paulentz, and F. Stinsky, Naphthensäuren und Naphthenate,
Verlag für Chemische Industrie H. Ziolkowsky, Augsburg, Germany, 1961.

JAMES A. BRIENT
PETER J. WESSNER
Merichem Company
MARY NN DOYLE
Shepherd Chemical Company
Vol. 2 FUELS, SYNTHETIC, LIQUID 329

FUELS, SYNTHETIC, LIQUID

The creation of liquids to be used as fuels from sources other than natural
crude petroleum (qv) broadly defines synthetic liquid fuels. Hence, fuel
liquids prepared from naturally occurring bitumen .deposits qualify as syn-
thetics, even though these sources are natural liquids. Synthetic liquid
fuels have characteristics approaching those of the liquid fuels in commerce,
specifically gasoline, kerosene, jet fuel, and fuel oil (see AVIATION TURBINE
FUELS; G ASOLINE AND OTHER MOTOR FUELS). For much of the twentieth century,
the synthetic fuels emphasis was on liquid products derived from coal upgrad-
ing or by extraction or hydrogenation of organic matter in coke liquids, coal
tars, tar sands, or bitumen deposits. More recently, however, much of the
direction involving synthetic fuels technology has changed. There are two
reasons.
The potential of natural gas, which typically has 85–95% methane, has
been recognized as a plentiful and clean alternative feedstock to crude oil (see
GAS, NATURAL). Estimates (1–3) place worldwide natural gas reserves at ca 1
1014 m3 ð3:5  1015 ft3 Þ corresponding to the energy equivalent of ca 11011 m3
(637 109bbl) of oil. As of this writing, the rate of discovery of proven natural
gas reserves is increasing faster than the rate of natural gas production. Many
of the large natural gas deposits are located in areas where abundant crude
oil resources lie such as in the Middle East and Russia. However, huge reserves
of natural gas are also found in many other regions of the world, providing oil-
deficient countries access to a plentiful energy source. The gas is frequently
located in remote areas far from centers of consumption, and pipeline costs can
account for as much as one-third of the total natural gas cost (1) (see PIPELINES).
Thus tremendous strategic and economic incentives exist for on-site gas conver-
sion to liquids.
In general, the proven technology to upgrade methane is via steam reform-
ing to produce synthesis gas, CO þ H2. Such a gas mixture is clean and when
converted to liquids produces fuels substantially free of heteroatoms such as
sulfur and nitrogen. Two commercial units utilizing the synthesis gas from
natural gas technology in combination with novel downstream conversion pro-
cesses have been commercialized.
The direct methane conversion technology, which has received the most
research attention, involves the oxidative coupling of methane to produce higher
hydrocarbons (qv) such as ethylene (qv). These olefinic products may be
upgraded to liquid fuels via catalytic oligomerization processes.
A second trend in synthetic fuels is increased attention to oxygenates as
alternative fuels (4) as a result of the growing environmental concern about
burning fossil-based fuels. The environmental impact of the oxygenates,
such as methanol (qv), ethanol (qv), and methyl tert-butyl ether (MTBE) (see
OCTANE IMPROVERS) is still under debate, but these alternative liquid fuels
are gaining new prominence. The U.S. Alternative Fuels Act of 1988, .and
the endorsement of oxygenate fuels that act contains, clearly underscore the
idea that economics is no longer the sole consideration with regard to alterna-
tive fuels production (5).

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.12091721080114.a01
330 FUELS, SYNTHETIC, LIQUID Vol. 2

Despite reduced prominence, coal technology is well positioned to


provide synthetic fuels for the future. World petroleum and natural gas pro-
duction are expected ultimately to level off and then decline. Coal gasifica-
tion to synthesis gas is utilized to synthesize liquid fuels in much the same
manner as natural gas steam reforming technology. Although as of this writ-
ing world activity in coal liquefaction technology is minimal, the extensive
development and detailed demonstration of processes for converting coal to
liquid fuels should serve as solid foundation for the synthetic fuel needs of
the future.
Coal, tar, and heavy oil fuel reserves are widely distributed throughout
the world. In the Western hemisphere, Canada has large tar sand, bitumen
(very heavy crude oil), and coal deposits. The United States has very large
reserves of coal and shale. Coal comprises ca 85% of the U.S. recoverable fossil
energy reserves (6). Venezuela has an enormous bitumen deposit and Brazil
has significant oil shale (qv) reserves. Coal is also found in Brazil, Colombia,
Mexico, and Peru. Worldwide, the total resource base of these reserves is
immense and may constitute >90% of the hydrocarbon resources in place
(see FUEL RESOURCES).
The driving force behind the production of combustible liquids before 1900
was the search for low cost lighting. Gas produced during coal distillation was
used to light homes at the end of the eighteenth century (7). Large-scale use of
coal, which began in England in the nineteenth century, led to significant reduc-
tions in the costs of hydrocarbon liquids. The production of coal tar, and the
separation therefrom of various coal liquids concomitant to the production of illu-
minating gas, probably predates production from the coking operations asso-
ciated with iron production. The coal tars produced in gas works may have
been the first synthetic liquids turned to fuel use in quantity.
Proof of the existence of benzene in the light oil derived from coal tar (8)
first established coal tar and coal as chemical raw materials. Soon thereafter
the separation of coal-tar light oil into substantially pure fractions produced a
number of the aromatic components now known to be present in significant
quantities in petroleum-derived liquid fuels. Indeed, these separation pro-
cedures were for the recovery of benzene –toluene –xylene (BTX) and related
substances, ie, benzol or motor benzol, from coke-oven operations (8) (see
BTX PROCESSING).
By the middle of the nineteenth century it was realized, both in England
and in the United States, that kerosene, or coal oil, distilled from coal, could
produce a luminous combustion flame. Commercialization was rapid. By the
time of the U.S. Civil War, 87 m3 =yr (23,000 gal/yr) of lamp oil was produced
in the United States from the distillation not only of coal, but also of oil shale and
natural bitumen. In 1859, high gravity, low sulfur crude oil was discovered in the
United States. This produced high quality kerosene with minimal processing,
and the world’s first oil boom erupted. Until the end of the nineteenth century,
kerosene was the only substance of value extracted from natural crude oil. It cost
too much for heating purposes, but was used widely for lighting until replaced by
electricity. Refiners slowly learned to use the residues from kerosene production,
and as the market for lamp oil collapsed, gasoline increased in value. The
widespread use of liquids as fuels dates from that time.
Vol. 2 FUELS, SYNTHETIC, LIQUID 331

Liquid fuels possess inherent advantageous characteristics in terms


of being more readily stored, transported, and metered than gases, solids, or
tars. Liquid fuels also are generally easy to process or clean by chemical
and catalytic means. The energy densities of clean hydrocarbon liquids may
be very high relative to gas, solid, or semisolid fuel substances. Moreover,
liquid fuels are the most compatible with the twentieth century world fuel
infrastructure because most fuel-powered conveyances are designed to function
only with relatively clean, low viscosity liquids. In general, liquid hydrocarbon
fuels possess an intermediate hydrogen-to-carbon content. Production of
synthetic fuels from alternative feedstocks to natural petroleum crude oil is
based on adjusting the hydrogen-to-carbon ratio to the desired intermediate
level.
There is an inherent economic penalty associated with producing
liquids from either natural gas or solid coal feedstock. Synthetic liquid
fuels technologies are generally not economically competitive with crude oil
processing in the absence of extraneous influences such as price supports or
regulations.

1. Indirect Liquefaction/Conversion to Liquid Fuels

Indirect liquefaction of coal and conversion of natural gas to synthetic liquid


fuels is defined by technology that involves an intermediate step to generate
synthesis gas, CO þ H2 . The main reactions involved in the generation of
synthesis gas are the coal gasification reactions:

Combustion
C þ O2 Ð CO2 H298k ¼ 394 kJ=molð94:2 kcal=molÞ ð1Þ
Gasification
C þ 1=2O2 Ð CO H298k ¼ 111kJ=molð26:5Þkcal=mol ð2Þ
C þ H2 OðgÞ Ð CO þ H2 H298k ¼ þ131kJ=molð31:3Þkcal=mol ð3Þ
C þ CO2 Ð 2CO H298k ¼ þ172kJ=molð41:1Þkcal=mol ð4Þ

water gas shift


CO þ H2 OðgÞ Ð CO þ H2 H298k ¼ 41kJ=molð9:8kcal=molÞ ð5Þ

the methane steam reforming reactions: Partial oxidation


CH4 þ 1=2O2 Ð CO þ 2H2 H298k ¼ þ36kJ=molð8:6Þkcal=mol ð6Þ
Reforming
CH4 þ H2 OðgÞ Ð CO þ 3H2 H298k ¼ þ206kJ=molð49:2Þkcal=mol ð7Þ

and the water gas shift reaction (eq. 5), used to increase the H2/CO ratio of the
product synthesis gas.
Coal gasification technology dates to the early nineteenth century but has
been largely replaced by natural gas and oil. A more hydrogen-rich synthesis gas
is produced at a lower capital investment. Steam reforming of natural gas is
332 FUELS, SYNTHETIC, LIQUID Vol. 2

F-T wax Distillate +


conversion specialty products
Fischer-
Tropsch

F-T oil Gasoline +


Coal refining distillate
Synthesis
gas
Natural
gas Methanol-to-
gasoline Gasoline

Methanol
synthesis

Methanol-to- Olefin Gasoline +


olefins oligomerization distillate

Fig. 1. Routes to liquid fuels from natural gas and coal via synthesis gas. F-T is the
Fischer-Tropsch process.

applied widely on an industrial scale (9,10) and in particular for the production of
hydrogen.
The conversion of coal and natural gas to liquid fuels via indirect technology
can be achieved by the routes shown in Figure 1. Two pathways from synthesis
gas can be taken. Both have been commercialized. One pathway involves cou-
pling with Fischer-Tropsch technology to produce fuel range hydrocarbons
directly or upon further processing. Using coal feedstock, this route has been
commercialized in South Africa since the 1950s and a process using natural
gas was commercialized in Malaysia by Shell Oil Co. in 1993. An alternative
route relies on the production of methanol from synthesis gas and subsequent
transformation of the methanol to fuels using zeolite catalyst technology intro-
duced by Mobil Oil Corp. This route was commercialized using indigenous natu-
ral gas in New Zealand in 1985.
1.1. Coal Upgrading via Fischer-Tropsch. The synthesis of methane
by the catalytic reduction of carbon monoxide and hydrogen over nickel .and
cobalt .catalysts at atmospheric pressure was reported in 1902 (11).

CO þ 3 H2 ! CH4 þ H2 OðlÞ H298 K ¼ 250 kJ=mol ð59:8 kcal=molÞ ð8Þ


2 CO þ 2 H2 ! CH4 þ CO2 H298 K ¼ 247 kJ=mol ð59:0 kcal=molÞ ð9Þ

In the early 1920s Badische Anilin- und Soda-Fabrik announced the specific cat-
alytic conversion of carbon monoxide and hydrogen at 20–30 MPa (200–300 atm)
and 300–4008C to methanol (12,13), a process subsequently widely industria-
lized. At the same time Fischer and Tropsch announced the Synthine process
(14,15), in which an iron catalyst effects the reaction of carbon monoxide and
hydrogen to produce a mixture of alcohols, aldehydes, ketones, and fatty acids
at atmospheric pressure.
Vol. 2 FUELS, SYNTHETIC, LIQUID 333

In the classical normal pressure synthesis (16), higher hydrocarbons are


produced by net reactions similar to those observed in the early 1900s, but at
temperatures below the level at which methane is formed:

n CO þ 2n H2 ! Cn H2n þ n H2 O þ heat ð10Þ

2n CO þ n H2 ! Cn H2n þ n CO2 þ heat ð11Þ

In the mid-1930s improvements in catalysts and techniques (17–19) culminated


in the licensing of the process to Ruhrchemie to produce liquid hydrocarbons and
paraffin waxes using precipitated cobalt-on-kieselguhr catalysts. Subsequently,
a medium pressure synthesis was developed (20) at 0.5–2 MPa (5–20 atm)
using dispersed cobalt catalysts which improved hydrocarbon yields by
10–15%. The yields of paraffin wax, in particular, could be increased to 45% of
the total liquid product. Hydrotreating of catalyst (required in the normal pres-
sure process) was avoided, and catalyst life was extended (21–23). There is a
marked influence of pressure on product yields. Beyond the optimum pressure
of about 2 MPa (20 atm), paraffin yield decreases. Little change is found in the
gasoline and gas oil yields, however.
Furthermore it was discovered that reasonable yields could be obtained
using precipitated iron catalysts at 1–3 MPa (10–30 atm), and that very high
melting waxes could be synthesized at 10–100 MPa (100–1000 atm) over ruthe-
nium catalysts. At the same time a related process, the oxo synthesis, was
announced (24) (see OXO PROCESS). Early in World War II, the iso-synthesis
process was developed for the production of low molecular weight isoparaffins
at high temperatures and pressures over thoria and mixtures of alumina and
zinc oxide (25–28). In the early 1960s polymethylenes were synthesized using
activated ruthenium catalysts at high pressures.
Industrial operation of the Fischer-Tropsch synthesis involved five steps:
(1) synthesis gas manufacture; (2) gas purification by removal of water and
dust, and hydrogen sulfide and organic sulfur compounds; (3) synthesis of hydro-
carbons; (4) condensation of liquid products and recovery of gasoline from
product gas; and (5) fractionation of synthetic products. Only the synthesis reac-
tor and its method of operation were unique to the process. For low pressure
synthesis the reactor incorporated elaborate bundles of cooling tubes immersed
in the catalyst, whereby circulating water removed the heat of reaction, limiting
the conversion to methane which produced high temperatures. In the pressure
process, bundles of concentric tubes, with catalyst arranged in the annuli,
through and around which cooling water flowed, served as conversion units. In
both systems, the conversion units each contained about 10 m3 (ca 350 ft3) of cat-
alyst, and were rated at a capacity of ca 4.8 m3 (30 bbl) of liquid product per day.
During World War II, nine commercial plants were operated in Germany,
five using the normal pressure synthesis, two the medium pressure process, and
two having converters of both types. The largest plants had capacities of ca
400 m3/d (2500 bbl/d) of liquid products. Cobalt catalysts were used exclusively.
Developments Outside Germany. In the late 1930s experimental work
in England (29–31) led to the erection of large pilot facilities for Fischer-Tropsch
studies (32). In France, a commercial facility near Calais produced ca 150 m3
334 FUELS, SYNTHETIC, LIQUID Vol. 2

(940 bbl) of liquid hydrocarbons per day. In Japan, two full-scale plants were also
operated under Ruhrchemie license. Combined capacity was ca 400 m3 (2500 bbl)
of liquids per day.
In the mid-1930s Universal Oil Products reported (33,34) that gasoline of
improved quality could be produced by cracking the high boiling fractions of
Fischer liquids, and a consortium, the Hydrocarbon Synthesis, Inc., entered into
an agreement with Ruhrchemie to license the Fischer synthesis outside Germany.
In 1955 the South African Coal, Oil, and Gas Corp. (Sasol) commercialized
the production of liquid fuels utilizing Fischer-Tropsch technology (35). This
Sasol One complex has evolved into the streaming of second-generation plants,
known as Sasol Two and Three. The Sasol One process, shown in Figure 2a (36),
combines fixed-bed Ruhrchemie-Lurgi Arge reactor units with fluidized-bed

Coal
Mine Power station Air separation

Steam Oxygen

Phenols Lurgi gasification

NH3

Phenosolvan Tar refining


Gas liquor Tar oils

To effluent treatment Raw gas Tar products

Lurgi gas purification Rectisol


CO2
H2S Pure gas
85% H2 + CO
13% CH4
2% N2 + CO2

H2 + CO

Sasol
Fluid
Fixed Fischer-Tropsch
bed
bed hydrocarbon
Synthol
synthesis processes
Fluid bed
Synthol

Tail gas
Product recovery Methane CH4 Product
to pipeline
reforming recovery
20 MJ/m3 STP
LPG
LPG
Gasoline
Ethylene
Diesel O2 Steam
Product refinery Gasoline
Fuel oils Product
refinery Diesel
Waxes
Fuel oils
Chemicals
Jet fuel
Chemicals
(a) (b)

Fig. 2. Process flow sheet of (a) Sasol One and (b) the Sasol synfuel process for Sasol Two
and Three (36). LPG is liquefied petroleum gas; other terms are defined in the text. To
convert J to cal, divide by 4.184.
Vol. 2 FUELS, SYNTHETIC, LIQUID 335

Table 1. Product Selectivities for Commercial


Fixed-Bed and Synthol Unitsa
Product Fixed-bed Synthol
methane 2.0 10
ethylene 0.1 4
ethane 1.8 4
propylene 2.7 12
propane 1.7 2
butenes 3.1 9
butanes 1.9 2
C5 and higher 83.5 51
soluble chemicals 3.0 5
water-soluble acids 0.2 1
a
Ref. 36.

Synthol process technology (37). For Sasol One, 16,000 t/d of coal is crushed and
gasified with steam and oxygen. After a number of gas purification steps in which
by-products and gas impurities are removed, the pure gas is processed in both
fixed- and fluidized-bed units simultaneously. Table 1 gives product selectivity
comparisons of fixed-bed and Synthol operations. Conversion to hydrocarbons
is higher in the Synthol unit and the H2/CO ratio is also higher. Because
the fixed-bed Arge reactor favors the formation of straight-chain paraffins,
there is greater production of diesel and wax fractions than the Synthol unit.
The Arge reactor products have lower gasoline octane number but higher diesel
cetane number relative to Synthol. The high wax production using the Arge reac-
tor was disadvantageous at the time owing to market limitations of wax fuels.
Sasol One produced a vast array of chemical and fuel products, including
gasoline at 1:5  106 t=yr.
The 1973 oil crisis resulted in the Sasol Two unit which started up in early
1980 followed by the nearly identical Sasol Three unit two years later. Figure 2b
gives the schematic flow diagrams for the Sasol Two and Three processes. Sasol
Two uses 36 Lurgi gasifiers in parallel to process ca 31,000 t/d of sized coal. By-
product effluents and gas impurities are removed in Rectisol (sulfur compounds
and CO2 removal), Phenosolvan .(oxygen compounds and ammonia removal),
and tar separation units. Synthol fluid-bed units were employed because of the
product distribution and ease of design scale-up. Approximately 80,000 t/d of coal
are needed for the two plants. Composition and manufacturing information for
Sasol Fischer-Tropsch catalysts are trade secrets, but the catalyst is widely
accepted as being an alkali-metal promoted iron-based material.
More recently, Sasol commercialized a new type of fluidized-bed reactor and
was also operating a higher pressure commercial fixed-bed reactor (38). In 1989,
a commercial scale fixed fluid-bed reactor was commissioned having a capacity
similar to existing commercial reactors at Sasol One (39). This effort is aimed
at expanded production of higher value chemicals, in particular waxes (qv) and
linear olefins.
Properties. Fischer-Tropsch liquid obtained using cobalt catalysts is
roughly equivalent to a very paraffinic natural petroleum oil but is not so com-
plex a mixture. Straight-chain, saturated aliphatic molecules predominate but
336 FUELS, SYNTHETIC, LIQUID Vol. 2

monoolefins may be present in an appreciable concentration. Alcohols, fatty


acids, and other oxygenated compounds may represent less than 1% of the
total liquid product. The normal pressure synthesis yields ca 60% gasoline,
30% gas oil, and 10% paraffin (mp 20–1008C). The medium pressure synthesis
yields 35% gasoline, 35% gas oil, and 30% paraffin. The octane rating of the gaso-
line is too low for direct use as motor fuel (40).
Most of the German gasoline production was blended into motor fuels using
benzene derived from coking. The gas oil could be used directly as a superior die-
sel fuel; some was also used in soap (qv) manufacture. The paraffin (referred to
as gatsch) was used primarily for the synthesis of fatty acids and hard soaps. The
propane and butane gases were also used as motor fuels. Some propylene and
butylenes were polymerized over phosphoric acid to high octane gasoline, and
some olefins to lubricating oils. Typical values for the composition of the techni-
cal-scale reaction products of the normal and medium pressure synthesis are
available (41).
1.2. Natural Gas Upgrading via Fischer-Tropsch. In the United
States, as in other countries, scarcities from World War II revived interest in
the synthesis of fuel substances. A study of the economics of Fischer synthesis
led to the conclusion that the large-scale production of gasoline from natural
gas offered hope for commercial utility. In the Hydrocol process (Hydrocarbon
Research, Inc.) natural gas was treated with high purity oxygen to produce the
synthesis gas which was converted in fluidized beds of iron catalysts (42).
Shell Middle Distillate Synthesis. The Shell middle distillate synthesis
(SMDS) process developed by Shell Oil Co., uses remote natural gas as the feed-
stock (43–45). A simplified flow scheme is given in Figure 3. This two-step pro-
cess involves Fischer-Tropsch synthesis of paraffinic wax called the heavy
paraffin synthesis (HPS). The wax is subsequently hydrocracked and hydroi-
somerized to yield a middle distillate boiling range product in the heavy paraf-
fin conversion (HPC). In the HPS stage, wax is maximized by using a
proprietary catalyst having high selectivity toward heavier products and by
the use of a tubular, fixed-bed Arge-type reactor. The HPC stage employs

Fuel gas
Including LPG

Syngas
HPS Flash

Tops / naphtha

Distil- Kerosene
H2 HPC
lation
Gas oil

Fig. 3. The Shell middle distillate synthesis (SMDS) process. HPS ¼ heavy paraffin
synthesis. HPC ¼ heavy paraffin conversion.
Vol. 2 FUELS, SYNTHETIC, LIQUID 337

10 10
Product composition, wt %

Product composition, wt %
8 8

6 6

4 4

2 2

0 0
0 10 20 30 40 0 10 20 30 40
Number of carbons Number of carbons

(a) (b)

Fig. 4. Product compositions as a function of carbon number for the Shell middle distil-
late synthesis process: (a) the Fischer-Tropsch product following HPS, and (b) the final
hydrocracking product following HPC. See text (45).

a commercial hydrocracking catalyst in a trickle flow reactor. The effect of


hydrocracking light paraffins is shown in Figure 4. The HPC step allows for
production of narrow range hydrocarbons not possible with conventional
Fischer-Tropsch technology.
After years of bench-scale and pilot-plant studies, construction was begun
on a gca 1600 m3/d (10,000 bbl/d) unit in Sarawak, Malaysia, by Shell in a joint
venture with Mitsubishi and the Malaysian government. Plant commissioning
was in early 1993 at a capital investment of ca $600  700  106 . The plant
uses natural gas from offshore fields and is located adjacent to the existing Malay-
sian liquefied natural gas (LNG) plant. The production of liquid transportation
fuels via SMDS cannot compete economically with fuels derived from crude oil.
However, economics vary greatly with site location, and subsidies from the
Malaysian government, eg, reduced natural gas cost, brought this plant to com-
mercialization. In addition, premium selling prices for the high quality products
made from SMDS are a primary influence on commercialization potential (44).
A similar process to SMDS using an improved catalyst is under development
by Norway’s state oil company, den norske state olijeselskap AS (Statoil) (46). High
synthesis gas conversion per pass and high selectivity to wax are claimed. The pro-
cess has been studied in bubble columns and a demonstration plant is planned.
Properties. Shell’s two-step SMDS technology allows for process flexibil-
ity and varied product slates. The liquid product obtained consists of naphtha,
kerosene, and gas oil in ratios from 15:25:60 to 25:50:25, depending on process
conditions. Of particular note are the high quality gas oil and kerosene.
Table 2 gives SMDS product qualities for these fractions.
The products manufactured are predominantly paraffinic, free from sulfur,
nitrogen, and other impurities, and have excellent combustion properties. The
very high cetane number and smoke point indicate clean-burning hydrocarbon
liquids having reduced harmful exhaust emissions. SMDS has also been
proposed to produce chemical intermediates, paraffinic solvents, and extra
high viscosity index (XHVI) lubeoils (see LUBRICATION AND LUBRICANTS) (44).
338 FUELS, SYNTHETIC, LIQUID Vol. 2

Table 2. SMDS Product Qualitya,b


Parameter SMDS Specification
Gas oil
cetane number 70 40 to 50
cloud point, 8C 10 10 to þ20
Kerosene
smoke point, mm 45 19
freezing point, 8C 47 47
a
Ref. 44.
b
The tops/naphtha fraction is similar to straight-run material.

1.3. Liquid Fuels via Methanol Synthesis and Conversion. Metha-


nol is produced catalytically from synthesis gas. By-products such as ethers, for-
mates, and higher hydrocarbons are formed in side reactions and are found in
the crude methanol product. Whereas for many years methanol was produced
from coal, after World War II low cost natural gas and light petroleum fractions
replaced coal as the feedstock.
Methanol-to-Gasoline. The most significant development in synthetic
fuels technology since the discovery of the Fischer-Tropsch process is the Mobil
methanol-to-gasoline (MTG) process (47–49). Methanol is efficiently transformed
into C2 –C10 hydrocarbons in a reaction catalyzed by the synthetic zeolite ZSM-5
(50–52). The MTG reaction path is presented in Figure 5 (47). The reaction
sequence can be summarized as

n=2 ½2 CH3 OH !
 CH3 OCH3 þ H2 O !Cn H2n !n½CH2 ð12Þ

where [CH2] represents an average paraffin–aromatic mixture.

70

60
Product distribution, wt %

50

40

30

20

10

0
10 –4 10 –3 10 –2 10 –1 1 10
Space time (LHSV)–1, h

Fig. 5. Methanol-to-hydrocarbons reaction path at 3718C, where (4) is methanol; (U),



dimethyl ether; ( ), water; (&), paraffins (and C6 þ olefins); ( ), aromatics; and (),
C2 –C5 olefins. LHSV ¼ Liquid hourly space velocity.
Vol. 2 FUELS, SYNTHETIC, LIQUID 339

Table 3. MTG Product Distributiona,b


Hydrocarbon Distribution, wt%
methane 1.0
ethane 0.6
ethylene 0.5
propane 16.2
propylene 1.0
i-butane 18.7
n-butane 5.6
butenes 1.3
i-pentane 7.8
n-pentane 1.3
pentenes 0.5
C6 þ aliphatics 4.3
benzene 1.7
toluene 10.5
ethylbenzene 0.8
xylenes 17.2
C9 aromatics 7.5
C10 aromatics 3.3
C11 þ 0.2
aromatics
a
Reaction conditions of 3718C and LHSV of 1:0 h1 .
b
100% conversion.

How the initial C  C bond is formed from the C1 progenitor is unknown


and much debated (48,53–55). Light olefins are key intermediates in the reaction
sequence. These undergo further transformation, ultimately forming aromatics
and light paraffins. Table 3 lists a typical MTG product distribution. A unique
characteristic of these products is an abrupt termination in carbon number at
around C10. This is a consequence of molecular shape–selectivity (56–58), a
property of ZSM-5. The composition and properties of the C5 þ fraction are
those of a typical premium aromatic gasoline. Interestingly, C10 also is the end
point of conventional gasoline.
The MTG process was developed for synfuel production in response to the
1973 oil crisis and the steep rise in crude prices that followed. Because methanol
can be made from any gasifiable carbonaceous source, including coal, natural gas,
and biomass, the MTG process provided a new alternative to petroleum for liquid
fuels production. New Zealand, heavily dependent on foreign oil imports, utilizes
the MTG process to convert vast offshore reserves of natural gas to gasoline (59).
Two versions of the MTG process, one using a fixed bed, the other a fluid
bed, have been developed. The fixed-bed process was selected for installation
in the New Zealand gas-to-gasoline (GTG) complex, situated on the North Island
between the villages of Waitara and Motonui on the Tasman seacoast (60). A sim-
plified block flow diagram of the complex is shown in Figure 6 (61). The plant
processes over 3:7  106 m3 =d ð130  106 SCF=dÞ of gas from the offshore Maui
field supplemented by gas from the Kapuni field, first to methanol, and thence
to 2:3  103 m3 =d (14,500 bbl/d) of gasoline. Methanol feed to the MTG section
is synthesized using the ICI low pressure process (62) in two trains, each with
a capacity of 2200 t/d.
Utility and Auxiliary Liquid
Water Water Cooling instrument steam nitrogen
U 1,270 m3/h treatment water
t air boiler plant
i
l
i
t
i
e 34,000 m3/h Plant Steam system MTG reactor
s BFW
treatment

Steam Crude High vapor pressure


550 t/h methanol Fuel gas gasoline
Storage
storage bullets
Fuel gas Additives

P
r
o Methanol Heavy
Natural gas BFW MTG MTG HGT
c synthesis gasoline Blending
e 155,000 m3/h treatment 4,400 t/d reactors distillation units
two trains storage
s

340
s

Light Gasoline
gasoline
Light gasoline storage
E
f
f
l Sanitary Effluent
u sewage treatment
e
n
t

Discharge
144 m3/h

Fig. 6. Simplified block flow diagram for the New Zealand gas-to-gasoline (GTG) plant (61). To convert m3/h to gal/min, multiply by 4.40.
HGT¼heavy gasoline treatment facility; MTG ¼ methanol-to-gasoline; BFW ¼ boiler feed water.
Vol. 2 FUELS, SYNTHETIC, LIQUID 341

A flow diagram of the MTG section is shown in Figure 7. Methanol feed,


vaporized by heat exchange with reactor effluent gases, is converted in a first-
stage reactor containing an alumina catalyst to an equilibrium mixture of metha-
nol, dimethyl ether (DME), and water. This is combined with recycle light gas,
which serves to remove reaction heat from the highly exothermic MTG reaction,
and enters the reactors containing ZSM-5 catalyst. As indicated in Figure 7, five
parallel swing reactors are used. Four reactors are on feed while the fifth is
under regeneration. The multiple-bed configuration is used to minimize pressure
drop as well as to control product selectivity. Reaction conditions are 360–4158C,
2:17  103 kPa (315 psia), and 9/1 recycle/fresh feed ratio. The overall thermal
efficiency of the plant is ca 53%.
A fluid-bed version of the MTG process has been developed (60,63–65) and
demonstrated in semiworks scale of 15.9 m3/d (100 bbl/d), but has not been
commercialized to date (ca 1993). Heat management of the exothermic MTG
reaction is greatly facilitated by use of fluid-bed reactors. The turbulent bed,
with its excellent heat-transfer characteristics, ensures isothermality through
the reaction zone and permits steam generation by direct exchange with steam
coils in the bed. A schematic diagram appears in Figure 8. The reactor system
consists of three principal parts: the reactor, the catalyst regenerator, and an
external catalyst cooler. The reactor is also equipped with internal heat-exchan-
ger tubes. Methanol is converted in a single pass at 380–4308C, 276–414 kPa
(40–60 psia). The methanol feed rate is 500–1050 kg/h. The fluid-bed demonstra-
tion was carried out in 1982–1983 (66).
Properties. Table 4 contains typical gasoline quality data from the
New Zealand plant (67). MTG gasoline typically contains 60 vol% saturates, ie,
paraffins and naphthenes; 10 vol% olefins; and 30 vol% aromatics. Sulfur and
nitrogen levels in the gasoline are virtually nil. The MTG process produces ca
3–7 wt% durene [95-93-2] (1,2,4,5-tetra-methylbenzene) but the level is reduced
to ca 2 wt% in the finished gasoline product by hydrodealkylation of the durene
in a separate catalytic reactor.
Methanol-to-Olefins and Olefins-to-Gasoline-and-Distillate. Because the
MTG process produces primarily gasoline, a variation of that process has been
developed which allows for production of gasoline and distillate fuel (68). This
process integrates two known technologies, methanol-to-olefins (MTO) and
Mobil olefins-to-gasoline-and-distillate (MOGD). The MTO/MOGD process sche-
matic is shown in Figure 9. The combined process produces gasoline and distillate
in various proportions and, if needed, olefinic by-products.
In the MTO process, methanol is converted over ZSM-5 giving high (up to ca
80 wt% hydrocarbons) olefin yields and low ethylene and light saturate yields.
The low ethylene yields are desirable in achieving high distillate yields using
MOGD. Figure 5 shows that the production of olefins rather than gasoline
from methanol is governed by the kinetics of methanol conversion over ZSM-5
catalyst (69). Generally, catalyst and process variables which increase methanol
conversion decrease olefins yield.
The MTO process employs a turbulent fluid-bed reactor system and typical
conversions exceed 99.9%. The coked catalyst is continuously withdrawn from
the reactor and burned in a regenerator. Coke yield and catalyst circulation
are an order of magnitude lower than in fluid catalytic cracking (FCC). The
DME effluent header
Preheater
vaporizer
circuit

Crude
CH3OH

To duplicate
reactor/effluent
exchanger trains
DME
reactor

Recycle
compressor

342
Hot excess ZSM-5 reactors
effluent four on line
from ZSM-5 one in regeneration
Purge gas to reactors
fractionation

Effluent/recycle
exchanger

Cooled effluent
from ZMS-5 reactor
Gasoline to
fractionation
Cooler

Water

Recycle gas header

Fig. 7. A methanol-to-gasoline (MTG) fixed-bed process flow diagram. DME ¼ dimethyl ether.
Vol. 2 FUELS, SYNTHETIC, LIQUID 343

BPR

Vent
xxxxx
Methanol

Condenser
Two-stage
scrubber
Water

Regenerator
Flare

Reactor
preparation
Feed

Air
+ Filter

Cooler
N2

C4 gas
N2 TC
Light gas
compressor
N2

Debutanizer
Vaporizer
Separator
Superheater
N2
Gasoline
Water

Fig. 8. Fluid-bed MTG demonstration plant schematic diagram. BPR ¼ Back pressure
regulator; TC ¼ temperature controller.

MTO process was first scaled up in a 0.64 m3/d (4 bbl/d) pilot plant and a success-
ful 15.9 m3/d (100 bbl/d) demonstration plant was operated in Germany with U.S.
and German government support.
The MOGD process oligomerizes light olefins to gasoline and distillate
products over ZSM-5 zeolite catalyst. Gasoline and distillate selectivity is >95%
of the feed olefins and gasoline/distillate product ratios can vary, depending on

Table 4. MTG Gasoline Qualitya


Parameter Average Range
3
density at 158C, kg/m 730 728–733
Reid vapor pressure, kPab 86.2 83.4–91.0
octane number
research 92.2 92.0–92.5
motor 82.6 82.2–83.0
durene content, wt% 2.0 1.74–2.29
induction period, min 325 260–370
distillation, % evaporation
at 708C 31.5 29.5–34.5
at 1008C 53.2 51.5–55.5
at 1808C 94.9 94.0–96.5
distillation end point, 8C 204.5 196–209
a
Ref. 67.
b
To convert kPa to psia, multiply by 0.145.
344 FUELS, SYNTHETIC, LIQUID Vol. 2

Methanol

MTO

Fractionation Fuel gas


and (C2 olefin recovery
compression optional)
MTO
gasoline

MOGD
Hydrogen

Distillate
hydrotreater Fractionation
Fuel gas

Diesel Gasoline LPG

Fig. 9. Methanol-to-olefins (MTO) and Mobil olefins-to-gasoline (MOGD) and distillate


process schematic.

process conditions, from 0.2 to >100. High octane MTO gasoline is separated
before the MOGD section and blended with MOGD gasoline. Some MOGD gasoline
may be recycled. The distillate product requires hydrofinishing. Generally, the
process scheme uses four fixed-bed reactors, three on-line and one in regeneration.
A large-scale MOGD refinery test run was conducted by Mobil in 1981.
Properties. The gasoline product from the integrated MTO/MOGD pro-
cess is predominately olefinic and aromatic. The gasoline quality (ca 89 octane)
is comparable to FCC gasoline. Typical distillate product properties are given in
Table 5. After hydrofinishing, the distillate product is mostly isoparaffinic and
has high cetane index, low pour point, and negligible sulfur content. MOGD

Table 5. MTO/MOGD Distillate Properties


Parameter Total distillate Jet fuel Diesel fuel
quantity, vol% 100 30 70
density, g/mL 0.792 0.774 0.800
pour point, 8C 50 30
freeze point, 8C 60 60
flash point, 8C 60 50 100
cetane number 50 52
smoke point, mm 25 25
aromatics, vol% 4 5
sulfur, ppm 50
Vol. 2 FUELS, SYNTHETIC, LIQUID 345

diesel fuel has somewhat lower density than typical conventional fuels (0.8 vs
0.86). Low aromatics levels contribute to a stable jet fuel with very little smoke
emission during combustion. MOGD diesel and jet fuels meet or exceed all
conventional specifications.

2. Direct Conversion of Natural Gas to Liquid Fuels

The capital costs associated with indirect natural gas upgrading technology
are high, thus research and development has focused on direct conversion of
natural gas to liquid fuels. Direct conversion is defined as upgrading methane
to the desired liquid fuels products while bypassing the synthesis gas step, ie,
direct transformation to oxygenates or higher hydrocarbons. Direct upgrading
routes which have been extensively studied include direct partial oxidation to
oxygenates, oxidative coupling to higher hydrocarbons, and pyrolysis to higher
hydrocarbons. Owing to the inert nature of methane, the technology is limited
by the yields of desired products which in turn affects the process economics.
Only one direct oxidative methane conversion process has been commer-
cialized. A plant at Copsa Mica (Romania) in the 1940s (70) produced formal-
dehyde directly from methane and air by partial oxidation. This plant is
no longer in operation. Plants to produce acetylene .from methane by high tem-
perature pyrolysis routes have been commercialized.
Generally, the most developed processes involve oxidative coupling of
methane to higher hydrocarbons. Oxidative coupling converts methane to ethane
and ethylene by

2 CH4 þ 1=2 O2 !H3 CCH3 þ H2 O ð13Þ

H3 CCH3 þ 1=2 O2 !H2 C¼


¼CH2 þ H2 O ð14Þ

The process can be operated in two modes: co-fed and redox. The co-fed
mode employs addition of O2 to the methane/natural gas feed and subse-
quent conversion over a metal oxide catalyst. The redox mode requires the
oxidant to be from the lattice oxygen of a reducible metal oxide in the reac-
tor bed. After methane oxidation has consumed nearly all the lattice oxy-
gen, the reduced metal oxide is reoxidized using an air stream. Both
methods have processing advantages and disadvantages. In all cases, how-
ever, the process is run to maximize production of the more desired ethylene
product.
Direct conversion of natural gas to liquids has been actively researched.
Process economics are highly variable and it is unclear whether direct natural
gas conversion technologies are competitive with the established indirect pro-
cesses. Some emerging technologies in this area are presented herein.
2.1. ARCO Gas-to-Gasoline Process. A two-step process using oxida-
tive coupling to upgrade natural gas to liquid fuels has been proposed by
ARCO (Atlantic Richfield Co.) (71,72). A simplified process scheme is given in
Figure 10. Methane is passed through a redox-mode oxidative coupling reactor
which generates C2 þ hydrocarbons such as ethylene. The olefinic products are
346 FUELS, SYNTHETIC, LIQUID Vol. 2

CO2 to atmosphere

Propane
Untreated gas Feed gas Butane
fractionation Condensate

Recycle gas
Air
Redox Water
CO2 to atmosphere

Olefin Product
conversion fractionation
High octane gasoline

Fig. 10. Simplified flow diagram depicting the ARCO gas-to-gasoline process for a con-
ceptual gasoline production plant (72).

then oligomerized over a zeolite catalyst in a second reactor to produce gasoline


and distillate. Unreacted methane is recycled. ARCO claims 25% conversion
of methane with 75% C2 þ selectivity (ethylene to ethane ratio up to 10) in the
oxidative coupling first stage and 95% ethylene conversion with 70% selectivity
to gasoline and distillate in the olefin oligomerization second stage. This technol-
ogy has been developed through bench-scale and pilot-plant stages.
2.2. OXCO Process. The OXCO process for upgrading natural gas has
been proposed by the Commonwealth for Scientific and Industrial Research Orga-
nization (CSIRO) in Australia (73,74). This process involves C2 þ pyrolysis and
oxidative coupling of natural gas in a two-stage reactor; the entire concept is
shown schematically in Figure 11. The methane in natural gas is separated
and oxidatively coupled in a fluidized-bed reactor operating in co-fed mode to
produce ethylene and ethane. The higher alkanes from the natural gas as well
as the product ethane from the first stage are injected into an oxygen-free
pyrolysis stage to make additional ethylene. The heat from the coupling reactor
is used in the pyrolysis reaction. The overall carbon conversion to unsaturates
plus CO2 per pass is 30% with an overall carbon selectivity to unsaturates of
86%. The ethylene may be subsequently upgraded by oligomerization to liquid
fuels. This process, which produces higher yields of ethylene and has a more
favorable heat balance than conventional oxidative coupling technology, has
been demonstrated in 30- and 60-mm fluidized-bed reactors.
Properties. Liquid fuels derived from oxidative coupling/olefin oligomeri-
zation processes would be expected to have properties similar to those derived
from olefin oligomerization pathways such as MTO/MOGD.

3. Oxygenate Fuels

Alcohols and ethers, especially methanol, ethanol, and methyl tert-butyl ether
[1634-04-4] (MTBE), have been widely used separately or in blends with gaso-
lines (reformulated gasoline) and other hydrocarbons to fuel internal combustion
CO2 H2O

Gas
treatment

Heat Compression
recovery

Alkane
C2+ Alkanes C 2H6 Unsaturates
pyrolysis Hydrocarbon Olefin
recovery oligomerization

347
Oxidative
CO2, H2O coupling Diesel
CH4/CO/H2
Sulfur compounds Gasoline
800–850 C

Natural gas Feed gas CH4 Recycle CH4 CO2


Methanation
pretreatment

O2

Air
separation

Fig. 11. The OXCO process for natural gas conversion (74).
348 FUELS, SYNTHETIC, LIQUID Vol. 2

Table 6. Fuel Oxygenates Propertiesa


Blending octane, Heat of Specific Boiling
Oxygenate 1/2(RON þ MON)b combustion, MJ/Lc gravity point, 8C
methanol 101 18.0 0.79 64.6
ethanol 101 21.3 0.79 78.5
2-propanol 106 26.4 0.79 82.4
2-butanol 99 28.3 0.80 99.5
tert-butyl alcohol 100 28.1 0.80 82.6
MTBEd 108 30.2 0.75 55.4
ETBEe 111 32.5 0.74 72.8
TAMEf 102 31.2 0.77 86.3
gasoline 87 34.8 0.74
a
Ref. 5.
b
RON ¼ research octane number; MON ¼ motor octane number.
c
To convert MJ/L to Btu/gal, multiply by 3589.
d
MTBE ¼ methyl tertbutyl ether.
e
ETBE ¼ ethyl tertbutyl ether.
f
TAME ¼ tertamyl methyl ether.

engines. Fuel properties of key oxygenates are presented in Table 6 (5). These
compounds, as a class, may be considered to be partially oxidized, ie, each has
a mole of oxidized hydrogen. They differ from the hydrocarbons that make up
gasoline principally in lower heating values and in higher vaporization heat
requirements. This constitutes a serious disadvantage to the substitution
of oxygenates, especially lower alcohols, for motor gasoline. For example,
the heating value of methanol is about half that of gasoline on an equivalent
volume basis. Other properties which greatly influence the potential of oxyge-
nates as fuels include octane performance, solubility in gasoline, effect on gaso-
line vapor pressure, sensitivity to water, and evaporative/exhaust emissions.
Oxygenate fuels tests are often debated because the tests employed were devel-
oped for conventional gasolines.
The addition of small percentages of oxygenates to gasoline can produce
large gains in octane. Thus, as blending components in gasoline, oxygenates
improve octane quality. As neat fuels for spark-ignition engines, octane values
for oxygenates are not useful in determining knock-limited compression ratios
for vehicles because of the lean carburetor settings relative to gasoline. Neither
do these values represent the octane performance of oxygenates when blended
with gasoline.
In part because neat alcohols are insufficiently volatile to enable a cold
engine to start, even at moderate temperatures, the use of neat alcohols for auto-
motive motor fuel is problematic. Manufacturers have, however, reported that
alcohol-powered cars, after being started and warmed up, can have the same
or better driveability as gasoline cars (75–77). As for gasoline vehicles, port
fuel injector fouling has occurred in some methanol vehicles and has affected dri-
veability and emissions. Other problems related to high alcohol content gasoline
in conventional engines include vapor lock and corrosion. Flexible-fuel vehicles
(FFV), which can operate on either neat methanol or gasoline, or mixtures
thereof, are being evaluated.
Vol. 2 FUELS, SYNTHETIC, LIQUID 349

Gasoline blends containing oxygenates change the emissions characteris-


tics of a motor vehicle designed for gasoline. Oxygenates and oxygenate-blends
approved for use by the U.S. government are expected to have desirable emis-
sions features as automotive fuels, and governmental environmental mandates
and regulations have necessitated increased examination and implementation
of oxygenates as fuels. As of this writing, however, no process can produce alco-
hols or ethers at equivalent or lower cost per volume than gasolines derived from
natural petroleum.
3.1. Methanol. Methanol production in the 1990s is dominated by reac-
tion of synthesis gas produced from natural gas. The economics of producing
methanol as fuel are highly variable and site-specific. The natural gas feedstock
has a broad range of values depending on location. Delivered costs for methanol
are probably double that for gasoline from petroleum. Impurities, including
water, vary according to the synthesis process employed. The term methyl fuel
.describes certain methanol products that may also contain significant quantities
of water and higher alcohols as well as other oxygenated compounds. Water
removal from such mixtures by distillative processes is generally complicated
because of the formation of azeotropes. Methanol is used to produce MTBE,
another oxygenate fuel.
Methanol as a fuel has been proposed in various ratios with gasoline.
In gasoline formulations having relatively low methanol content, eg, M3
(3% methanol) and M15 (15% methanol), solubilizers are used and stringently
dry conditions must be maintained. High methanol content fuels, M85
(85% methanol) and M100 (neat methanol), have special engine requirements.
The use of high methanol content fuels is limited by methanol cost plus poor
compatibility with the existing gasoline infrastructure.
Methanol is more soluble in aromatic than paraffinic hydrocarbons. Thus
varying gasoline compositions can affect fuel blends. At room temperature, the
solubility of methanol in gasoline is very limited in the presence of water.
Generally, cosolvents are added to methanol–gasoline blends to enhance water
tolerance. Methanol is practically insoluble in diesel fuel.
Concerns about using methanol–gasoline blend fuels include problems with
vapor lock, cold start, and warmup. Oxygenate–gasoline blends, and in particu-
lar those containing methanol, have unusual volatility characteristics and
cannot be accurately characterized using test methods developed for gasoline.
Vapor pressure is an important volatility parameter that is adversely affected
by the addition of methanol. Gasoline blends containing oxygenates form noni-
deal solutions with varying characteristics. In general, methanol–gasoline
blend fuels exhibit increases in Reid vapor pressure (RVP) over that of gaso-
line itself. This effect contributes to vapor lock and evaporative emissions.
The data available for assessing methanol’s impact on exhaust emissions, and
consequently on air quality, are limited. Formaldehyde (qv) has been reported
in the exhausts of cars fueled with straight methanol or methanol–gasoline
blends.
The use of methanol as a motor fuel has been discussed since the 1920s.
Straight methanol has long been a preferred fuel for racing engines because of
the much higher compression ratios at which methanol may be combusted rela-
tive to gasolines. This is translatable for racing purposes at equivalent power
350 FUELS, SYNTHETIC, LIQUID Vol. 2

outputs to engines of considerably reduced weight. However, fuel consumptions


are roughly three times that of gasoline on a km/L (mi/gal) basis, and extremely
high emissions of unburned fuel and carbon monoxide can result (78).
In Germany in the early 1950s, a 50:50 mixture of methanol and 2-propanol
was blended with gasoline, first at a level of 7.5% and later at 1.5% (79). Com-
plaints about stalling, power loss, and phase separation caused the ratio to be
changed to 60:40::methanol:2-propanol but this apparently aggravated the pro-
blems. The practice was discontinued in 1970 when a tax was placed on alcohol.
In the United States, the Clean Air Act of 1970 imposed limitations on
composition of new fuels, and as such methanol-containing fuels were required
to obtain Environmental Protection Agency (EPA) waivers. Upon enactment of
the Clean Air Act Amendments of 1977, EPA set for waiver unleaded fuels con-
taining 2 wt% maximum oxygenates excluding methanol (0.3 vol% maximum).
Questions regarding methanol’s influence on emissions, water separation, and
fuel system components were raised (80).
In 1979 Sun Oil Co. was granted a waiver for a gasoline blend containing
2.75 vol% methanol and an equal volume of tert-butyl alcohol (TBA) (2 wt%
total oxygen). Cosolvents such as TBA were shown to reduce adverse effects of
methanol on volatility and water tolerance. ARCO obtained EPA waiver in
1981 for a 3.5 wt% oxygen fuel blending containing Oxinol, also comprising
equal parts of methanol and TBA. In 1985 a waiver was granted to Du Pont,
Inc. for a gasoline blend containing 5 vol% maximum methanol with at least
2.5 vol% higher alcohol cosolvents. The waiver incorporated a water tolerance
or phase separation requirement (81).
The most extensive worldwide program on methanol blend gasoline was in
Italy where from 1982 to 1987 a 1:9  104 m3 =yr ð5  106 gal=yrÞ plant produced
a mixture containing 69% methanol. The balance contained higher alcohols. This
mixture was blended into gasoline at the 4.3% level and marketed successfully as
a premium gasoline known as Super E (82).
Methanol, a clean burning fuel relative to conventional industrial fuels
other than natural gas, can be used advantageously in stationary turbines
and boilers because of its low flame luminosity and combustion temperature.
Low NOx emissions and virtually no sulfur or particulate emissions have
been observed (83). Methanol is also considered for dual fuel (methanol plus
oil or natural gas) combustion power boilers (84) as well as to fuel gas tur-
bines in combined methanol/electric power production plants using coal gasi-
fication (85).
Owing to its properties, methanol is not recommended for aircraft or marine
fuel uses. Methanol cannot be used in conventional diesel-powered vehicles with-
out modifications to the fuel system and engine. Simple methanol–diesel blends
are not possible because of insolubility. Heavy-duty diesel engines have been
adapted to use neat methanol by many U.S. manufacturers, and several are
being used in field demonstrations (82).
3.2. Ethanol. Ethanol is produced both from ethylene (qv) derived
from the cracking of petroleum fractions and by the fermentation of sugars
.derived from grains or other biomass (see SUGAR). Many of its relevant properties
are similar to those of methanol. Although ethanol may be a more desirable fuel
or fuel component than methanol, its significantly higher cost (volume basis)
Vol. 2 FUELS, SYNTHETIC, LIQUID 351

may outweigh these advantages. Broad implementation of ethanol-containing


fuels would require government action, eg, in the form of subsidies to farmers
and fuel waivers.
The term gasohol .has come into wide usage to identify, generally, a blend of
gasoline and ethanol, with the latter derived from grain. The term may also be
applied to blends of methanol or other alcohols in gasolines or other hydrocar-
bons, without regard to sources of components.
Brazil’s Alcohol Program. In Brazil, the enactment of legislation in 1931
made ethanol addition to gasoline compulsory at a level of 5% (86). Excess
molasses and sugar were converted to alcohol in distilleries attached to sugar
mills as a means to stabilize sugar prices. Production of fuel ethanol in the
1990s is mostly from biomass.
Starting in the city of Sao Paulo in 1977, and extending to the entire state of
Sao Paulo in 1978, a gasohol incorporating 20% ethanol was mandated. Brazil’s
National Alcohol Program (Proalcool) set an initial goal of providing the 20% fuel
mixture nationwide by 1980–1981 and a system of special tax, warranty, and
price considerations were enacted to advance the aims of Proalcool.
For a considerable period, >90% of the new cars in Brazil operated on E96
fuel, or a mixture of 96% ethanol and 4% water (82). The engines have high
compression ratios (ca 12:1) to utilize the high knock resistance of ethanol and
deliver optimum fuel economy. In 1989 more than one-third of Brazil’s 10 million
automobiles operated on 96% ethanol/4% water fuel. The remainder ran on
gasoline blends containing up to 20% ethanol (5).
Gasohol in the United States. Over 90% of the fuel ethanol in the United
States is produced from corn. Typically, 0.035 m3 (1 bushel) of corn yields 9.5 L
(2.5 gal) of ethanol. Ethanol is produced by either dry or wet milling (87). Selec-
tion of the process depends on market demand for the by-products of the two
processes. More than two-thirds of the ethanol in the United States is produced
by wet milling. Depending on the process used, the full cost of ethanol after
by-product credits has been estimated to be between $0.25–0.53/L ($1–2/gal)
for new plants (88). Feedstock costs are a significant factor in the production
of fuel ethanol. A change in corn price of $0.29/m3 ($1.00/bushel) affects the
costs of ethanol by $0.08/L ($0.30/gal).
Ethanol can also be produced from cellulose (qv) or biomass such as wood
(qv), corn stover, and municipal solid wastes (see FUELS FROM BIOMASS; FUELS
FROM WASTE). Each of these resources has inherent technical or economic
problems. The Tennessee Valley Authority (TVA) is operating a 2 t/d pilot
plant on converting cellulose to ethanol.
After the oil embargo in 1973, gasohol use was stimulated by tax incentives.
An application for EPA waiver of gasohol fuels (up to 10 vol% ethanol) was
granted in 1979. From 1981 to 1983 the California Energy Commission field
tested alcohol-powered cars equipped with a gasoline-assist starting system, ie,
having an onboard auxiliary supply of volatile fuel for cold start tests. In 1989
about 8% of U.S. gasoline contained 10% ethanol plus a corrosion inhibitor (82).
As of this writing, government waivers of RVP standards for gasohol fuels are
being considered (89).
3.3. Methyl t-Butyl Ether. MTBE is produced by reaction of isobutene
and methanol on acid ion-exchange resins. The supply of isobutene, obtained
352 FUELS, SYNTHETIC, LIQUID Vol. 2

from hydrocarbon cracking units or by dehydration of tert-butyl alcohol, is lim-


ited relative to that of methanol. The cost to produce MTBE from by-product iso-
butene has been estimated to be between $0.13 to $0.16/L ($0.50–0.60/gal) (90).
Direct production of isobutene by dehydrogenation of isobutane or isomerization
of mixed butenes are expensive processes that have seen less commercial use in
the United States.
More than 95% of MTBE produced worldwide is used to blend with gasoline.
In 1987 U.S. production of MTBE exceeded 3:8  106 m3 =yr ð1  109 gal=yrÞ (82).
The worldwide capacity for MTBE is increasing, especially in the United States
and Europe, and has been projected to exceed production for years to come.
MTBE’s gain in prominence as a fuel-blend component is a result of inher-
ent technical advantages over other oxygenates, especially the lower alcohols.
MTBE has a high blending octane number (Table 6) although this number
varies somewhat with gasoline composition. The low vapor pressure relative
to the lower alcohols results in no increase in RVP for MTBE-gasoline blends
and consequently better evaporative emission and vapor lock characteristics.
No phase separation occurs in blends with other fuels. MTBE, in blends
of < 20 vol% with gasoline, does not deleteriously affect other fuel or driving
characteristics such as cold start, fuel consumption, and engine materials
compatibility.
MTBE has been used in motor fuels in Europe since the early 1970s and is
undergoing rapid growth, particularly in the United States. MTBE-blended gaso-
line containing up to 11 vol% MTBE received EPA waiver in 1981. Later legisla-
tion increased the MTBE waiver up to 15 vol%. In 1987–1988 Colorado began
mandating use of winter oxygenate-based fuels in the Denver region. About 90%
of the fuel in this period used a gasoline blend containing 8 vol% MTBE and in
1988–1989 the fuel was required to contain at least 2% oxygen (11 vol% MTBE).
Based on the success of this program and EPA assessments that CO reductions of
10–20% over the next decade were possible with oxygenate-blend fuels, numer-
ous state governments enacted legislation requiring the use of these fuels in win-
ter and in cities having high ozone (smog) concentrations. The Clean Air Act
Amendments of 1990 have mandated the use of reformulated gasolines, espe-
cially in serious ozone problem areas, by 1995.
The effectiveness of MTBE, however, is under discussion (91). Based on
Denver, Colorado vehicle emissions data from 1981 to 1991 and theoretical mod-
els, Colorado scientists have claimed that the use of MTBE-blended fuels had no
statistically significant effect on atmospheric CO levels, but increased pollutants
such as formaldehyde. A drop in CO levels in Denver during this time period was
attributed to fleet turnover of older, more polluting cars being replaced by newer
cars having cleaner burning engines. In addition, health problems associated
with direct exposure to MTBE in Fairbanks, Alaska has resulted in EPA exemp-
tion of the oxygenated fuel requirement in that area (91).

4. Direct Liquefaction of Coal

Direct liquefaction, the production of liquids from feed coal in a single processing
scheme without a synthesis gas intermediate step, includes two routes for the
Vol. 2 FUELS, SYNTHETIC, LIQUID 353

upgrading of coal: hydrogenation and pyrolysis. In hydrogenation, the conversion


of coal to liquids having higher hydrogen-to-carbon ratio involves the addition of
hydrogen. Generally, the additional hydrogen required is added either from
molecular hydrogen or from a hydrogen-donor solvent such as tetralin. Processes
classified under pyrolysis are those which produce liquids by removal of carbon.
This occurs when coal is thermally processed under inert or reducing atmo-
spheric conditions. The use of hydrogen in a pyrolytic process to increase yields
of distillate products is known as hydropyrolysis. Coal carbonization to produce
metallurgical coke involves much the same chemistry as pyrolysis.
4.1. Coal and Coal-Tar Hydrogenation. If paraffinic and olefinic
liquids are extracted from solid fuel substances, the hydrogen content of the resi-
dual material is reduced even further, and the residues become more refractory.
The yields of liquids so derivable are generally low, even when a significant frac-
tion of the hydrogen is extractable. Thus production of fuel liquids from nonli-
quid fuel substances such as coal and coal tars may be enhanced only by the
introduction of additional hydrogen in a synthesis process. The principal differ-
ences in the processes are from the modes in which hydrogen is introduced and
the catalysts used.
Hydrogenation of coal and other carbonaceous matter using high pressure
hydrogen has been patented (92), and subsequently the Nobel Prize in chemistry
was won for this accomplishment. By 1922, a 1 t/d plant was operating and using
hydrogen at 10 MPa (100 atm) and 4008C to treat brown coal tar to give a liquid
that comprised 25 wt% gasoline boiling at 75–2108C and 40 wt% middle oil,
210–3008C. The pitch residue had a specific gravity of 1.04, and a solidification
point of 158C. The degree of liquefaction was shown to increase with decrease in
oil rank (93). Liquid products were of low quality, being high in oxygen, nitrogen,
and sulfur content, owing to low hydrogenation rates and polymerization of pri-
mary products (94).
In 1935 an ICI coal hydrogenation plant at Billingham, U.K., produced ca
136,000 t/yr motor fuel from bituminous coal and coal tar. By 1936, 272,000 t/yr
of motor fuel were produced by improved hydrogenation of brown coal and coal
tar at a facility constructed at Leuna and some 363,000 t/yr was being produced
in three other German plants (95). Two years later the total German output from
these facilities was ca 1:4  106 t=yr (96). The number of coal hydrogenation
plants in Germany increased during World War II to 12, with total capacity of
about 4  106 t=yr (100,000 bbl/d) of aviation and motor gasolines.
Experimental plants for hydrogenating coal or coal tar were operated in
Japan, France, Canada, and in the United States before or during World War
II. Much of that technology has remained proprietary. In general, coal-in-oil
slurries containing iodine .or stannous oxalate catalyst were subjected to
liquid-phase hydrogenation at pressures of 25–70 MPa (250–700 atm). Liquids
produced were fractionated, and the middle oils were then subjected to vapor-
phase hydrogenation over molybdenum-, cobalt-, or tungsten sulfide-on-alumina
catalysts (97). About 1 t of crude motor fuel was recovered from 4.5 t of coal, from
which all necessary hydrogen and power requirements for the production were
also obtained.
Developments in the United States. A large number of proprietary coal
hydrogenation process variants have been proposed. Much of the technology
354 FUELS, SYNTHETIC, LIQUID Vol. 2

originally directed to the catalytic hydrogenation of coals and coal tars in


Germany has been applied to the hydrorefining of petroleum fractions, but
U.S. commercial interest in coal hydrogenation was offset by the relative abun-
dance of domestic petroleum up to World War II.
The huge demand for liquid fuels during World War II prompted the pas-
sage of the Synthetic Liquid Fuels Act of 1944. There were various programs
relating to demonstration plants to produce liquid fuels from coal, oil shale,
and other substances, including agricultural and forestry products. The Bureau
of Mines had begun work on coal liquefaction in 1936, at which time a 45 kg/d
experimental coal hydrogenation unit was constructed (98). The expanded pro-
gram, after 1944, culminated in the construction and operation of a 45 t/d coal
hydrogenation demonstration plant at Louisiana, Missouri, in 1949 (99), where
a variety of problems and processing variations were investigated (100,101). Cost
studies (102) showed that production of gasoline from coal hydrogenation could
not compete with using natural petroleum as a gasoline source. The demonstra-
tion plant operations were terminated in 1953.
Work on coal hydrogenation continued by the Bureau of Mines on a labora-
tory scale (103–105). In one of these variants (106) coal-oil pastes admixed with
catalyst in tubular reactors were hydrogenated at high pressure and low resi-
dence times to give improved yields of liquid products. The original thrust of
the work was to hydrodesulfurize coal economically to produce environmentally
acceptable boiler fuel (107). In the mid-1970s, a process sponsored by the Bureau
of Mines named Synthoil (108) was developed, but the efforts were terminated by
1978 owing to limited catalyst lifetimes.
H-Coal Process. The H-coal process (Hydrocarbon Research, Inc., HRI,
subsidiary of Dynalectron Corp.), for the conversion of coal to liquid products
(109), is an application of HRI’s ebullated-bed technology for the conversion of
heavy oil residues into lighter fractions. Coal is dried, pulverized, and slurried
with coal-derived oil (110). The coal-oil slurry is charged continuously
with hydrogen to a reactor of unique design (111) containing a bed of ebullated
catalyst, where the coal is hydrogenated and converted to liquid and gaseous
products. The liquid product is a synthetic crude oil that can be converted to
gasoline or heating oil by conventional refining processes. Alternatively, under
milder operating conditions, a clean fuel gas and low sulfur fuel oils may be pro-
duced. The relative yields of these products depend on the desired sulfur level in
the heavy fuel oil. In general, reaction products are separated by fractionation
and absorption (qv). Unreacted coal may be fed into a fluid coker that produces
gas, gas oil, and dry char. The coker gas oil, along with gas oils separated from
the main reactor effluent, may be subjected to hydrocracking for conversion to
lighter products.
In 1976, Ashland Oil (Ashland Synthetic Fuels, Inc.) was awarded the
prime contract to construct a 540 t/d H-coal pilot plant adjacent to its refinery
at Catlettsburg, Kentucky, by an industry–government underwriting consor-
tium. Construction was completed in 1980 (112). The pilot-plant operation
ended in early 1983.
Properties. The properties of naphtha, gas oil, and H-oil products from an
H-coal operation are given in Table 7. These analyses are for liquids produced
from the syncrude operating mode. Whereas these liquids are very low in sulfur
Vol. 2 FUELS, SYNTHETIC, LIQUID 355

Table 7. Properties of Syncrude from H-Coal Processa


Boiling range
Property Initial to 1908C 190–3438C 343–5248C Total
b
specific gravity, (8API) 0.767 (53.0) 0.915 (23.2) 1.05 (3.5) 0.863 (32.4)
vol% of total 40.0 54.2 5.8 100.0
analysis, wt%
carbon 84.5 88.8 89.4 87.3
hydrogen 13.6 11.0 10.2 11.9
oxygen 1.7 0.6
nitrogen 0.1 0.1 0.1 0.1
sulfur 0.1 0.1 0.3 0.1
Total 100 100 100 100
a
Ref. 111.
b 141:5
API ¼  131:5 where r is specific gravity.
r

compared with typical petroleum fractions, they are high in oxygen and nitrogen
levels. No residual oil products (bp > 540 C) are formed.
Solvent-Refined Coal Process. In the 1920s the anthracene oil fraction
recovered from pyrolysis, or coking, of coal was utilized to extract 35–40% of
bituminous coals at low pressures for the purpose of manufacturing low cost
newspaper inks (113). Tetralin .was found to have higher solvent power for
coals, and the I. G. Farben Pott-Broche process (114) was developed, wherein a
mixture of cresol and tetralin was used to dissolve ca 75% of brown coals at 13.8
MPa (2000 psi) and 4278C. The extract was filtered, and the filtrate vacuum dis-
tilled. The overhead was distilled a second time at atmospheric pressure to sepa-
rate solvent, which was recycled to extraction, and a heavier liquid, which was
sent to hydrogenation. The bottoms product from vacuum distillation, or solvent-
extracted coal, was carbonized to produce electrode carbon. Filter cake from the
filters was coked in rotary kilns for tar and oil recovery. A variety of liquid pro-
ducts were obtained from the solvent extraction-hydrogenation system (113). A
similar process was employed in Japan during World War II to produce electrode
coke, asphalt (qv), and carbonized fuel briquettes (115).
In the United States there was little interest in solvent processing of coals.
A method to reduce the sulfur content of coal extracts by heating with sodium
hydroxide and zinc oxide was, however, patented in 1940 (116). In the 1960s
the technical feasibility of a coal deashing process was studied (117), and a
pilot plant able to process ca 45 t/d was completed in late 1974 (118).
A flow diagram of the solvent-refined coal or SRC process is shown in
Figure 12. Coal is pulverized and mixed with a solvent to form a slurry contain-
ing 25–35 wt% coal. The slurry is pressurized to ca 7 MPa (1000 psig), mixed
with hydrogen, and heated to ca 4258C. The solution reactions are completed
in ca 20 min and the reaction product flashed to separate gases. The liquid is fil-
tered to remove the mineral residue (ash and undissolved coal) and fractionated
to recover the solvent, which is recycled.
The liquid remaining after the solvent has been recovered is a heavy resi-
dual fuel called solvent-refined coal, containing less than 0.8 wt% sulfur and
0.1 wt% ash. It melts at ca 1778C and has a heating value of ca 37 MJ/kg
Sulfur
Coal Dissolver

Recycle
Slurry hydrogen
preheater
Raw gas
Slurry
mix tank
Coal slurry

Pulverizer

Absorber Sulfur
plant

Product slurry Absorber liquid

356
Hydrogen
Solvent
recovery
unit
Product and solvent
Filter

Shipment

Solvent–refined coal

Fig. 12. Solvent-refined coal process (119).


Vol. 2 FUELS, SYNTHETIC, LIQUID 357

(16,000 Btu/lb), regardless of the quality of the coal feedstock. The activity of the
solvent is apparently more important than the action of gaseous hydrogen in this
type of uncatalyzed hydrogenation. Research has been directed to the use of pet-
roleum-derived aromatic oils as start-up solvents (118).
In the early 1970s production of low sulfur, ashless (solid) boiler fuel was
the preferred commercial application (119). This basic process (SRC-I) yielded
small amounts of liquid oil products with additional processing. Liquid
output was significantly increased by the coal-oil-gas (COG) refinery concept
(120–122) which incorporated high degrees of hydroconversion and hydrotreat-
ing. A SCR-II process has been developed, in which hydrocracking occurs in the
solution (hydrogenation) vessels (123). A low viscosity fuel oil is the primary dis-
tillate product in this case, although naphtha and LPG are also recovered.
Two pilot plants have been built and operated to demonstrate the feasibility
of the SRC process. These included a 6 t/d plant at Wilsonville, Alabama (vide
infra) and a 50 t/d plant at Ft. Lewis, Washington which was operated from
1974 to 1981.
In an effort to obtain higher value products from SRC processes, a hydro-
cracking step was added to convert resid to distillate liquids. The addition
of a hydrocracker to the SRC-I process was called nonintegrated two-stage
liquefaction (NTSL). The NTSL process was essentially two separate processes
in series: coal liquefaction and resid upgrading. NTSL processes were inefficient
owing to the inherent limitations of the SRC-I process and the high hydrocracker
severities required.
Properties. The properties of the liquid fuel oil produced by the SRC-II
process are influenced by the particular processing configuration. However, in
general, it is an oil boiling between 177 and 4878C, having a specific gravity of
0.99–1.00, and a viscosity at 388C of 40 SUs (123). Pipeline gas, propane and
butane (LPG), and naphtha .are also recovered from an SRC-II complex.
Exxon Donor Solvent-Coal Liquefaction Process. The EDS process
.from Exxon is a hydrogenation process using a donor solvent for the direct con-
version of a broad range of coals to liquid hydrocarbons (124). In the process
sequence, shown in Figure 13, the feed coal is crushed, dried, and slurried
with hydrogenated recycle solvent (the donor solvent) and fed to the reactor
with hydrogen. The reactor is an upward plug-flow design operating at
430–4808C and at ca 14 MPa (2000 psi) total pressure.
The reactor effluent is separated by conventional distillation into recycle
solvent, light gases, C4 to 5378C bp distillate, and a heavy vacuum bottoms
stream containing unconverted coal and ash. The recycle solvent is hydrogenated
in a separate reactor and sent back to the liquefaction reactor.
The heavy vacuum bottoms stream is fed to a Flexicoking unit. This is a
commercial (125, 126) petroleum process that employs circulating fluidized
beds at low (0.3 MPa (50 psi)) pressures and intermediate temperatures, ie,
480–6508C in the coker and 815–9808C in the gasifier, to produce high yields
of liquids or gases from organic material present in the feed. Residual carbon
is rejected with the ash from the gasifier fluidized bed. The total liquid product
is a blend of streams from liquefaction and the Flexicoker.
The EDS process was developed starting from 1976 in a 10-year joint under-
taking between DOE and private industry (127). Under the direction of Exxon
358 FUELS, SYNTHETIC, LIQUID Vol. 2

Catalytic

Solvent Gas Steam


Hydrogen
hydrogenation reforming

Coal H2 H2O

Liquid
H2 Liquefaction Distillation
products

Heavy
bottoms
slurry

H2O
Flexicoking
Air Fuel gas

Ash
residue

Fig. 13. Exxon donor solvent process (124).

Co. USA, a 250 t/d pilot plant was operated at Baytown, Texas. Operation of this
unit began in 1980 and was completed by late 1982.
Properties. Pilot-unit data indicate the EDS process may accommodate
a wide variety of coal types. Overall process yields from bituminous, subbitumi-
nous, and lignite coals, which include liquids from both liquefaction and
Flexicoking, are shown in Figure 14. The liquids produced have higher nitrogen
contents than are found in similar petroleum fractions. Sulfur contents reflect
the sulfur levels of the starting coals: ca 4.0 wt% sulfur in the dry bituminous
coal; 0.5 wt% in the subbituminous; and 1.2 wt% sulfur in the dry lignite.
Table 8 shows that the naphthas produced by the EDS process have
higher concentrations of cycloparaffins and phenols than do petroleum-derived
naphthas, whereas the normal paraffins are present in much lower concentra-
tions. The sulfur and nitrogen concentrations in coal naphthas are high com-
pared to those in petroleum naphthas.
Gas oil fractions (204–5658C) from coal liquefaction show even greater
differences in composition compared to petroleum-derived counterparts than
do the naphtha fractions (128). The coal-gas oils consist mostly of aromatics
(60%), polar heteroaromatics (25%), asphaltenes (8–15%), and saturated
compounds (< 10%). Petroleum-gas oils, on the other hand, contain more than
50% saturated compounds, less than 5% polar heteroaromatics, and no asphal-
tenes. Furthermore, the aromatics of petroleum-gas oils have longer side chains.
Coal Liquefaction at Wilsonville. Starting in 1974 the Advanced Coal
Liquefaction R&D Facility at Wilsonville, Alabama operated a 6 t/d pilot plant
and studied various coal liquefaction processing schemes. The facility, cospon-
sored by the DOE, the Electric Power Research Institute (EPRI) and Amoco
Oil Co, was shut down in early 1992.
Vol. 2 FUELS, SYNTHETIC, LIQUID 359

50

40
Liquid yields, wt % dry coal

30

20

A B C D E

10

0
Bituminous Subbituminous Lignite
Coal type

Fig. 14. Preferred liquefaction-coking liquid yields in the EDS process for various coals
where represents Flexicoking liquids and &, liquefaction liquids (124). A, Ireland (West
Virginia); B, Monterey (Illinois); C, Burning Star (Illinois); D, Wyodak (Wyoming); and E,
Big Brown (Texas).

Initial operation at the Wilsonville pilot plant was in SRC-I mode and later
evolved into a two-stage process (129) by operation in NTSL mode. NTSL limita-
tions described previously combined with high hydrogen consumptions resulted
in subsequent focus on a staged integrated approach, which was to be the basis
for all further studies at Wilsonville.
The integrated two-stage process (ITSL) combined short contact time lique-
faction in one reactor with ebullated-bed hydrocracking in a second stage (130).
The short contact time conditions permitted better hydrogen transfer from the
solvent rather than from the gas phase. The hydrocracking step operated at
lower severity resulting in lowered gas make and improved hydrogen efficiency.
Recycle solvent was generated from the hydrocracked distillates and coupled the
two reaction stages. Results of ITSL processing of Illinois No. 6 coal at Wilson-
ville are given in Table 9. Distillate yields and coal throughput for ITSL were
higher than those obtained by NTSL.
Further developments of the ITSL process resulted in incremental gains in
distillate yields (131). Reconfigured integrated two-stage liquefaction (RITSL)
involved placing the solvent deasher after the hydrocracker thus producing
a recycle solvent consisting of deashed resid and distillate. This resulted in
reduction of feed to the deasher and reduced organic rejection. Close coupled
Table 8. Composition of Naphthasa from Various Sourcesb
EDS coal liquefaction Petroleum naphthas
Component, wt% Illinois coal Wyodak coal Cycloparaffinicc Paraffinicd
saturated compounds 69.9 60.8 77.9 81.3
paraffins 13.4 19.4 36.0 58.2
cycloparaffins 56.5 41.4 41.9 23.1
olefinse 0.5 3.7
aromatics 17.0 28.6 22.1 18.7
benzenes 11.7 25.1 20.8 17.2
indans, tetralins 5.1 3.5 1.0 1.2
indenes 0.2 0.0
naphthalenes 0.01 0.0 0.3 0.3
phenols 12.6 6.9 traces traces
Total 100.0 100.0 100.0 100.0
sulfur 0.57 0.10 0.035 0.049
nitrogen 0.15 0.18 0.0001 0.0001
oxygen 1.60 2.49
a
C5 to 2048C bp.
b
Ref. 128.
c
Prudhoe Bay.
d
Arab Light.
e
Values are approximate.

Table 9. Wilsonville Plant Operating Conditions and Yields for ITSL and CTSL Modesa
Mode of operationb
Parameter ITSL CTSL
Operating conditions
run number 7242BC; 243JK/244B 253A
catalyst Shell 32M Shell 317
First stagec
average reactor temperature, 8C 460; 432 432
space velocity 690; 450d 4.8e
f
pressure, MPa 17; 10–17 17.9
Second stage
average reactor temperature, 8C 382 404
space velocity, feed/catalyste 1.0 4.3
catalyst age, resid/catalyst 278–441; 380–850 100–250
Yieldsg
C1 –C3 gas 4; 6 6
C4 þ distillate 54; 59 70
resid 8; 6 1
hydrogen consumption 4.9; 5.1 6.8
Other
hydrogen efficiency, C4 þ distillate=H2 consumed 11; 11.5 10.3
distillate selectivity, C1 –C3/C4 þ distillate 0.07; 0.10 0.08
energy content of feed coal reject 24; 20–23 20
to ash concentrate, %
a
Feed is Illinois No. 6 coal.
b
CTSL ¼ catalytic two-stage liquefaction; ITSL ¼ integrated two-stage liquefaction.
c
First stage is thermal for ITSL.
d
Value given is coal space velocity at temp >371 C in kg/m3.
e
Value given is in h1 .
f
To convert MPa to psia, multiply by 145.
g
Wt % on a moisture- and ash-free (MAF) coal basis.

360
Vol. 2 FUELS, SYNTHETIC, LIQUID 361

integrated two-stage liquefaction (CC-ITSL) linked the two reactors and removed
several operations between the two stages. A deleterious effect of these two pro-
cessing modes was increased hydrogen consumption over ITSL.
From 1985 to 1992, development activity at Wilsonville was on catalytic
two-stage liquefaction (CTSL). CTSL, initiated by HRI (132), consists of catalytic
processing in two ebullated-bed reactors which lower reaction temperatures and
increase distillate yields, up to 78% yield. CTSL results from Wilsonville for
Illinois No. 6 coal are also given in Table 9. Distillate yields were shown to be
significantly higher for CTSL over ITSL; however, hydrogen consumption was
somewhat increased.
Properties. CTSL distillates have qualities comparable to or better than
No. 2 fuel oil and have good hydrogen content and low heteroatom contents. Dis-
tillates having a higher boiling point distribution from Wilsonville CTSL opera-
tion (131) showed 26.88 API gravity with heteroatom levels of 0.11 wt% sulfur,
<1 wt% oxygen, and 0.16 wt% nitrogen.
4.2. Coal Pyrolysis. Pyrolysis is the destructive distillation of coal in
the absence of oxygen typically at temperatures between 400 and 5008C (133).
As the temperature of carbonaceous matter is increased, decomposition ulti-
mately occurs. Melting and dehydration may also occur. Coals exhibit more or
less definite decomposition temperatures, as indicated by melting and rapid evo-
lution of volatile components, including potential fuel liquids, during destructive
distillation (134). Table 10 summarizes an extensive survey of North American
coals subjected to laboratory pyrolysis. The yields of light oils so derived average
no more than ca 8.3 L/t (2 gal/short ton), and tar yields of ca 125 L/t (30 gal/short
ton) are optimum for high volatile bituminous coals (135).
Coal pyrolysis has been studied at both reduced and elevated pressures
(136), and in the presence of a variety of agents and atmospheres (137). Although
important to the study of coal structure and reactions, coal pyrolysis, as a means
to generate liquids, has proved to have limited commercial value.
COED Process. Sponsored by the Office of Coal Research of the U.S.
Department of the Interior, the COED process was developed by FMC Corp. as
Project Char-Oil-Energy Development (COED) through 1975 (138–140). Bench-
scale experiments led the way to construction in 1965 of a process development
unit employing multistage, fluidized-bed pyrolysis to process 45 kg/h (141). Cor-
related studies included hydrotreating of COED oil (142), high temperature
hydrodesulfurization of COED char, and investigations of char-oil and char-
water slurry pipelining economics (143). A pilot plant capable of processing up
to 33 t/d and hydrotreating 4.7 m3/d (30 bbl/d) was started up in 1970 (144),
and was operated successfully for a number of years (145).
The COED concept (139), designed to recover liquid, gaseous, and solid fuel
components, consists of four stages. Heat is generated by the reaction of oxygen
with a portion of the char in the last pyrolysis stage and is also introduced by the
air combustion of gas to dry feed coal. The number of stages in the pyrolysis, and
the operating temperatures in each, may be varied to accommodate high volatile
bituminous and subbituminous feed coals with widely ranging caking or agglom-
erating properties.
Oil condensed from the released volatiles from the second stage is filtered
and catalytically hydrotreated at high pressure to produce a synthetic crude oil.
Table 10. Average Yields and Range of Yields of Fischer Assay of Various Coalsa
Coke, % Tar, L/tc Light oil, L/tc Gas, m3/td Water, %
Rank of coal Average Range Average Range Average Range Average Range Average Range
semianthracite 3.2 0.14
low volatile bituminous 89.7 85.8–93.3 39.6 29.0–58.4 4.69 3.36–7.41 59.8 54.4–66.6 3.2 1.1–6.6
medium volatile bituminous 83.3 77.4–90.4 86.9 44.6–117.8 7.68 4.92–10.58 66.0 47.3–76.2 4.1 2.8–7.0
high volatile A bituminous 75.5 68.8–81.4 142.1 105.3–187.2 10.53 6.81–15.09 67.0 57.5–80.2 6.0 3.0–9.2
high volatile B bituminous 70.4 66.0–73.2 139.4 111.8–198.3 10.03 7.13–15.82 68.3 56.4–82.3 11.1 10.2–13.1
high volatile C bituminous 67.1 65.4–68.6 124.2 85.1–178.5 8.65 5.93–12.47 61.2 53.0–70.4 15.9 12.0–19.1
high volatile C bituminous 59.1 94.3 84.6–112.2 7.59 6.26–8.88 90.4 23.4

362
or subbituminous A
subbituminous A 81.9 81.0–82.8 6.21 6.12–6.26
subbituminous B 57.6 54.8–59.9 70.8 60.7–76.8 6.12 5.24–7.13 90.4 62.2–93.8 27.8 23.3–30.4
lignite 36.5 69.9 30.8–124.2 5.47 2.90–8.69 71.4 44.0
cannel 58.8 44.1–69.0 338.1 247.0–498.2 23.28 16.84–34.13 61.5 51.0–72.1 3.7 2.0–4.8
a
Ref. 135.
b
As-received basis; maximum temperature, 5008C.
c
To convert L/t to gal/short ton, divide by 4.6.
d
To convert m3/t to ft3/short ton, multiply by 29.4.
Vol. 2 FUELS, SYNTHETIC, LIQUID 363

Table 11. Properties of COED Char Producta


Property Utah coal Illinois No. 6
volatile matter, wt% 6.1 2.7
fixed carbon, wt% 80.2 77.0
ash, wt% 13.7 20.3
higher heating value, MJ/kg,b dry 28.6 25.6
elemental analysis, wt%, dry
carbon 81.5 73.4
hydrogen 1.3 0.8
nitrogen 1.5 1.0
sulfur 0.5 3.4
oxygen 1.5 1.0
chlorine 0.006 0.1
ironc 0.28
a
Ref. 139.
b
To convert MJ/kg to Btu/lb, multiply by 430.
c
Included in ash above.

Medium heat-content gas produced after the removal of H2S and CO2 is suitable
as clean fuel. The pyrolysis gas produced, however, is insufficient to provide the
fuel requirement for the total plant. Residual char, 50–60% of the feed coal, has a
heating value and sulfur content about the same as feed coal, and its utilization
may thus largely dictate process utility.
Properties. The properties of char products from two possible coal feeds,
a low sulfur Western coal, and a high sulfur Midwestern coal, are shown in
Table 11. The char derived from the low sulfur Western coal may be directly sui-
table as plant fuel, with only minor addition of clean process gas to stabilize its
combustion. Flue gas desulfurization may not be required. Flue gas from the
combustion of the char derived from the high sulfur Illinois coal, however,
requires desulfurization before it may be discharged into the atmosphere.
Typical COED syncrude properties are shown in Table 12. The properties of
the oil products depend heavily on the severity of hydroprocessing. The degree of
severity also markedly affects costs associated with hydrogen production and
compression. Syncrudes derived from Western coals have much higher paraffin
and lower aromatic content than those produced from Illinois coal. In general,
properties of COED products have been found compatible with expected indus-
trial requirements.
Occidental Petroleum Coal Conversion Process. Garrett R&D Co.
(now the Occidental Research Co.) developed the Oxy Coal Conversion process
based on mathematical simulation for heating coal particles in the pyrolysis
unit. It was estimated that coal particles of 100-mm diameter could be heated
throughout their volumes to decomposition temperature (450–5408C) within
0.1 s. A large pilot facility was constructed at LaVerne, California, in 1971. This
unit was reported to operate successfully at feed rates up to 136 kg/h (3.2 t/d).
Hot product char carries heat into the entrained bed to obtain the high
heat-transfer rates required. Feed coal must be dried and pulverized. A portion
of the char recovered from the reactor product stream is cooled and discharged as
product. The remainder is reheated to 650–8708C in a char heater blown with
air. Gases from the reactor are cooled and scrubbed free of product tar. Hydrogen
364 FUELS, SYNTHETIC, LIQUID Vol. 2

Table 12. Typical COED Syncrude Propertiesa


Property Utah A-seam Illinois No. 6 seam
b
specific gravity, (8 API) 0.934 (20) 0.929 (22)
pour point, 8C 16 18
flash point, closed cup, 8C 24 16
viscosity, at 388C, mm2/s (¼cSt) 8 5
ash, wt% 0.01 0.01
moisture, wt% 0.1 0.1
metals, ppm 10 10
elemental analysis, wt%
C 87.2 87.1
H 11.0 10.9
N 0.2 0.3
O 1.4 1.6
S 0.1 0.1
ASTM distillation initial bp, 8C 138 88
10% 221 134
30% 277 199
50% 349 270
70% 416 316
90% 493 362
end point (95%) 510 397
hydrocarbon type analysis, liquid vol %
paraffins 23.7 10.4
olefins 0 0
naphthenes 42.2 41.4
aromatics 34.1 48.2
a
Ref. 139.
b 141:5
API ¼  131:5 where r is specific gravity.
r

sulfide is removed from the gas, and a portion is recycled to serve as the entrain-
ment medium.
Properties. A high volatile western Kentucky bituminous coal, the tar
yield of which by Fischer assay was ca 16%, gave a tar yield of ca 26% at a pyr-
olysis temperature of 5378C (146–148). Tar yield peaked at ca 35% at 5778C and
dropped off to 22% at 6178C. The char heating value is essentially equal to that of
the starting coal, and the tar has a lower hydrogen content than other pyrolysis
tars. The product char is not suitable for direct combustion because of its 2.6%
sulfur content.
The TOSCOAL Process. The Oil Shale Corp. (TOSCO) piloted the low
temperature carbonization of Wyoming subbituminous coals over a two-year per-
iod in its 23 t/d pilot plant at Rocky Falls, Colorado (149). The principal objective
was the upgrading of the heating value in order to reduce transportation costs on
a heating value basis. Hence, the solid char product from the process represented
50 wt% of the starting coal but had 80% of its heating value.
Furthermore, 60–100 L (14–24 gal) oil, having sulfur content below 0.4 wt%,
could be recovered per metric ton coal from pyrolysis at 427–5178C. The
recovered oil was suitable as low sulfur fuel. Figure 15 is a flow sheet of the
Rocky Flats pilot plant. Coal is fed from hoppers to a dilute-phase, fluid-bed pre-
heater and transported to a pyrolysis drum, where it is contacted by hot ceramic
Vol. 2 FUELS, SYNTHETIC, LIQUID 365

Flue gas to
atmosphere

Balls
Scrubber
Coal
Gas
Separator
Ball
Surge heater Air Naphtha
hopper
Accumulator
Pre Gas oil
hea
ted

Ball elevator
coa
l Hot
ball
Lift pipe

s Resid
Pyrolysis
drum Trommel
as
eg
t flu
Ho

W
ar
m
ba
lls
Hot
char Char
cooler

Char to
sales

Fig. 15. TOSCOAL process (149).

balls. Pyrolysis drum effluent is passed over a trommel screen that permits
char product to fall through. Product char is thereafter cooled and sent to
storage. The ceramic balls are recycled and pyrolysis vapors are condensed
and fractionated.
Properties. Results for the operation using subbituminous coal from
the Wyodad mine near Gillette, Wyoming, are shown in Table 13. Char yields
decreased with increasing temperature, and oil yields increased. The Fischer
assay laboratory method closely approximated the yields and product assays
that were obtained with the TOSCOAL process.
The volatiles contents of product chars decreased from ca 25–16% with tem-
perature. Char (lower) heating values, on the other hand, increased from ca
26.75 MJ/kg (11,500 Btu/lb) to 29.5 MJ/kg (12,700 Btu/lb) with temperature.
Chars in this range of heating values are suitable for boiler fuel application
and the low sulfur content (about equal to that of the starting coal) permits direct
combustion. These char products, however, are pyrophoric and require special
handling in storage and transportation systems.
Properties of the tar oil products are given in Table 14. The oils change only
slightly with change in the retorting temperature; sulfur levels are low. The
fraction boiling up to 2308C contains 65 wt% of phenols, cresols, and cresylic
acids.
366 FUELS, SYNTHETIC, LIQUID Vol. 2

Table 13. TOSCOAL Retorting of Wyodak Coala,b


Retort temperature
Component 4278C 4828C 5218C
char 524.5 505.8 484.4
gas, C3 and lighter 59.5 78.4 63.0
oil, C4 and heavier 57.0 71.5 93.1
waterc 351.0 351.0 351.0
Total 992.0 1006.7 991.5
Recovery, % 99.2 100.7 99.1
a
Ref. 150.
b
Product yields, kg/t, of as-mined coal.
c
Value assumed from Fischer assay and moisture content. The addition of steam
to the process prevented accurate measurement of water produced in retorting.

Table 14. TOSCOAL Oil Propertiesa


Retort temperature
b
Properties 4278C 4828C 5218C
analysis, wt%
carbon 81.4 80.7 80.9
hydrogen 9.3 9.1 8.7
oxygen 8.3 9.4 9.3
nitrogen 0.48 0.7 0.7
sulfur 0.43 0.2 0.2
chlorine 0.0 0.0 0.0
ash 0.0 0.2 0.1
Total 99.91 100.3 99.9
heating values
gross, MJ/kgc 38.59 37.72 37.13
net, MJ/kgc 36.61 35.75 35.26
specific gravity, (8API)d
primary oil 1.015 (7.9) 1.040 (4.5) 1.061 (1.9)
calculated, with C4 and heavier 0.978 (13.2) 0.985 (12.1) 1.027 (6.2)
components of gas added
pour point, 8C 32 38 35
Conradson carbon, wt% 7.6 9.9 11.4
distillation,e 8C
2.5 vol % 212 216 199
20 vol % 302 288 235
50 vol % 407 413 385
viscosity, SUs
at 828C 122 123 128
at 908C 63 66 69
a
Ref. 150.
b
Feed coal was different from that used for 482 and 5218C.
c
To convert MJ/kg to Btu/lb, multiply by 430.
d 141:5
API ¼  131:5 where r is specific gravity.
r
e
Combination of true boiling point and D1160 distillations.
Vol. 2 FUELS, SYNTHETIC, LIQUID 367

Hydrogen Naphthalene
–252.5°C 218°C

Coal gas

Hydrogen 3,4–Dimethylphenol
sulfide 225°C
–61.8°C

Light oil

Benzene Picene
80.09°C 519°C

Coal tar

–300 –200 –100 0 +100 +200 +300 +400 +500


Temperature, °C

Fig. 16. Boiling ranges of carbonization products (8).

TOSCO tar oils have high viscosity and may not be transported by conven-
tional pipelines. Heating values of product gas on a dry, acid gas-free basis are in
the natural gas range if butanes and heavier components are included.
Coal Carbonization. In the by-product recovery of a modern coke oven,
coal tar is removed first by cooling the gases emanating, and light oil is removed
last by scrubbing the gas with solvents. Other products, including ammonia, phe-
nols, pyridine, or naphthalene, may be recovered between these operations. The
constituents of coal tar, light oil, and gas usually overlap considerably, ie, the
fractional condensation does not effectively separate individual components.
Assuming the lowest boiling coal tar constituent to be benzene (bp 80.098C),
and the highest boiling to be naphthalene, the overlapping compositions of gas,
light oil, and tar may be as shown graphically in Figure 16. Many chemical
compounds have been identified (8) in these substances. Included are most of
the significant constituents of petroleum-derived fuel liquids, although only a
few components are present in sufficient quantity to make commercial recovery
feasible.
The precise compositions of the light oil and coal tar recovered from coke-
oven gas is a distinct function of the design of the recovery system, as well as of
the properties of the starting coal. In general, 12.5–16.7 L/t (3–4 gal/m light oil
per short ton) of coal carbonized is recovered from high temperature coke-oven
operations. Light oil may contain 55–70% benzene, 12–20% toluene, and
4–7% xylene. Unrecovered light oil appearing in the effluent coal gas may
comprise ca 1 vol% and contribute ca 5% of the gas’s heating value. Refining of
light oil consists mainly of sulfuric acid washing, followed by fractional
distillation.
Large-scale recovery of light oil was commercialized in England, Germany,
and the United States toward the end of the nineteenth century (151). Industrial
coal-tar production dates from the earliest operation of coal-gas facilities. The
principal bulk commodities derived from coal tar are wood-preserving oils,
368 FUELS, SYNTHETIC, LIQUID Vol. 2

road tars, industrial pitches, and coke. Naphthalene is obtained from tar oils by
crystallization, tar acids are derived by extraction of tar oils with caustic, and tar
bases by extraction with sulfuric acid. Coal tars generally contain less than 1%
benzene and toluene, and may contain up to 1% xylene. The total U.S. production
of BTX from coke-oven operations is insignificant compared to petroleum product
consumptions.

5. Other Processes

5.1. Shale Oil. In the United States, shale oil, or oil derivable from oil
shale, represents the largest potential source of liquid hydrocarbons that can
be readily processed to fuel liquids similar to those derived from natural petro-
leum. Some countries produce liquid fuels from oil shale. There is no such indus-
try in the United States although more than 50 companies were producing oil
from coal and shale in the United States in 1860 (152,153), and after the oil
embargo of 1973 several companies reactivated shale-oil process development
programs (154,155). Petroleum supply and price stability has since severely cur-
tailed shale oil development. In addition, complex environmental issues (156)
further prohibit demonstration of commercial designs.
5.2. Heavy Oil. The definitions used to distinguish among naturally
occurring heavy petroleum oils, bitumens, asphalts, and tars, are subject to
broad variations. More than 10% of the world’s current crude oil production
has an API gravity below 208, or a specific gravity greater than 0:93415 15 . Oils
having sp gr1515 > 0:904  0:934 (20–258API) are considered heavy oils in most
classifications. However, Safaniyah crude oil produced in Saudi Arabia
having sp gr1515 of 0.893 (278API) carries the designation Arabian heavy, and
in petroleum parlance is generally referred to as heavy crude oil. Yet its pro-
duction method does not differ from that of Arabian medium or Arabian light
crude oils.
Energy in the form of injected water or CO2 may be supplied to increase the
rate of production of light crude oils. Application of heat to the reservoirs, eg,
using hot water, steam, heated CO2, fireflood, or in situ combustion, however,
is generally associated with the production of heavier, viscid crudes.
Heavy crude oil is widely distributed, and it is difficult to estimate reserves
separate from normal crude oil reserves or from tar sands deposits. Estimates of
petroleum reserves frequently include a large heavy oil component, which can
only be produced at significantly higher cost than light oil.
Most heavy oil production is concentrated in California, Canada, and Vene-
zuela. There is significant production of heavy oil in California from the Kern
River field near Bakersfield and in Canada from the Cold Lake deposit in
Alberta. Production generally involves steam drives, or the injection of steam
into reservoirs through special wells in prescribed sequences. Oil–water
mixtures are recovered, and often separated water is treated and reinjected.
Heavy oil may be upgraded through two main routes: coking and hydropro-
cessing. Virtually all established upgrading schemes involve some variant of
those two routes. The challenges in upgrading and refining are from the low
Vol. 2 FUELS, SYNTHETIC, LIQUID 369

hydrogen content and specific gravity and high sulfur, nitrogen, and metals
content of the heavy oil.
5.3. Tar Sands. Tar sands are considered to be sedimentary rocks
having natural porosity where the pore volume is occupied by viscous, petro-
leum-like hydrocarbons. The terms oil sands, rock asphalts, asphaltic
sandstones, and malthas or malthites have all been applied to the same
resource. The hydrocarbon component of tar sands is properly termed bitumen.
Distinctions between tar sands’ bitumens and heavy oils are based largely
on differences in viscosities. The bitumen in oil sand has a specific gravity of less
than 0.986 g/mL (128API), and thus oil sands may be regarded as a source of
extremely heavy crude oil. Whereas heavy oils might be produced by the same
techniques used for the lighter crude oils, the bitumens in tar sands are too vis-
cous for these techniques. Consequently these oil-bearing stones have to be
mined and specially processed to recover contained hydrocarbon.
Tar sands have been reported on every continent except Australia and
Antarctica. The best known deposits are the Athabasca of Canada, where
almost 60,000 km2 in northeastern Alberta is underlain with an estimated
138  109 m3 ð870  109 bblÞ of recoverable bitumen (157). The Alberta deposits
may contain up to 215  109 m3 ð1350  109 bblÞ of bitumen reserves. Venezuela
may have the largest accumulations in the world; the Orinoco heavy-oil belt has
been estimated by some (157) to contain as high as 636  109 m3 ð4000 109 bblÞ.
The Olenek reserves in the former USSR may contain ca 95  109 m3 ð600
109 bblÞ. The United States is estimated to have deposits of about 4:5
109 m3 ð28  109 bblÞ.
The Great Canadian Oil Sands, Ltd. (GCO) (Sun Oil Co.) has been operat-
ing a plant at Fort McMurray, Alberta, Canada, since 1967. Initially, some 8050
t/d (55,000 bbl/d) of synthetic crude oil was produced from coking (158) with the
project expanding to 9220 t/d (63,000 bbl/d). Since 1978, Syncrude Canada has
been producing ca 22,000 m3/d (140,000 bbl) synthetic crude oil by fluid coking
from their plant at Cold Lake, Alberta, Canada (159) with expansion planned
for ca 35,000 m3/d (225,000 bbl/d).

6. Economic Aspects of Synthetic Fuels

As of this writing, processes for production of synthetic liquid fuels by upgrading


natural gas, coal, or heavy oil are generally not directly competitive with crude
oil upgrading (160). The key controlling factors in the economics are crude oil
price and availability. Many economic analyses for synthetic liquid fuels give a
crude oil price target whereupon the alternative technology becomes attractive,
but these studies sometimes neglect the fact that the natural gas, coal, and heavy
oil prices often track those of crude oil. In addition, conversion of a refractory gas
(methane) or solid (coal) to liquids is a greater technical challenge than that of
processing crude oil. Thus there are processing cost penalties which inevitably
exist even after considerable technological development. Nevertheless, synthetic
fuels technology is projected to play a primary role in providing liquid fuels once
crude oil depletion is of concern. Economic competitiveness plays a reduced role
370 FUELS, SYNTHETIC, LIQUID Vol. 2

in commercialization only when environmental legislation mandates the use of


certain fuels such as oxygenates.
The commercialization potential of synthetic fuels technology relies on site-
specific economic and political factors. This complex network of factors may
include capital costs, crude oil price, product yields and value, government
subsidies, strategic impact, alternative uses for the feed, and environmental
and geographical constraints. Whereas no direct coal liquefaction process has
gone to commercial stage, technologies involving indirect conversion of natural
gas or coal have been commercialized. In all cases, special conditions allowed
the technology to progress. In the Sasol project at South Africa, coal upgrading
was possible due to factors such as no indigenous petroleum, minimal environ-
mental standards, and cheap labor (160). The New Zealand GTG process became
economically feasible owing to high oil prices, abundance of indigenous natural
gas, and government commitment to energy self-sufficiency (59). Government
support and long-term strategic benefits were also keys to Shell’s SMDS project
in Malaysia.
At 1994 crude oil prices of ca $94–125/m3 (ca $15–20/bbl), conversion of
natural gas to liquid fuels exists only in unique situations. For natural gas
upgrading via New Zealand-type technology, economics by Mobil for a 1987
plant start-up on the U.S. Gulf Coast (161) indicated an investment of
$895  106 was required for a 2:3  103 m3 =d (14,600 bbl/d) gasoline production
unit. Thus this and other natural gas-to-fuels processes are highly capital-inten-
sive and capital recovery remains the dominant factor even with incremental
advances in conventional technology. This is especially the case using indirect
upgrading of natural gas because the cost of synthesis gas manufacture may
account for more than 50% of the total process capital cost (44). An analysis by
Shell of the SMDS process published in 1988 showed capital expense for a 1600
m3/d (10,000 bbl/d) plant to be $300  106 for a developed site in an industrialized
country and $600  106 for a developing site in a developing country (44). Direct
upgrading of natural gas to gasoline and distillate by oxidative coupling plus ole-
fin oligomerization has been evaluated to be ca 10% costlier in capital than
upgrading via indirect technologies (162).
Natural gas upgrading economics may be affected by additional factors.
The increasing use of compressed natural gas (CNG) directly as fuel in vehicles
provides an alternative market which affects both gas price and value (see
GASOLINE AND OTHER MOTOR FUELS; GAS, NATURAL). The hostility of the remote
site environment where the natural gas is located may contribute to additional
costs, eg, offshore sites require platforms and submarine pipelines.
The economic feasibility of coal upgrading, and in particular direct coal
liquefaction, are closely tied to crude oil price and capital expense. H-coal tech-
nology was evaluated as a base case in 1981 and the results showed economic
feasibility was possible only at crude oil price of $630/m3 ($100/bbl) or greater
(160). A more recent analysis by HRI on coal/oil coprocessing technology indi-
cated the required light crude price was $138–182/m3 ($22–29/bbl) for economic
feasibility (163). The cost of capital could add over $60/m3 ($10/bbl) to the cost of
products. A study of EDS, H-coal, ITSL, and two-stage Wilsonville systems
showed capital costs for a 30,000 t/d plant processing Illinois No. 6 coal to
run between $4100$4700  106 (131). Required break-even selling prices for
Vol. 2 FUELS, SYNTHETIC, LIQUID 371

products from these technologies ranged from $226–314/m3 ($36–50/bbl). The


two-stage system was the most economical at $229.94/m3 ($36.56/bbl). An eva-
luation of coprocessing Cold Lake vacuum bottoms using Alberta subbituminous
coal in a 3200 m3/d (20,000 bbl/d) synthetic crude oil production unit indicated a
selling price of $189–220/m3 ($30–35/bbl) was necessary for the process to be
competitive (164). In general, the economics of direct coal liquefaction depend
more on the high cost of liquefaction rather than the cost for upgrading product
coal liquids (165).
Factors which may affect the cost of coal upgrading are environmental con-
siderations such as toxicity, hazardous waste disposal, and carcinogenic proper-
ties (131). These and other environmental problems from process streams,
untreated wastewaters, and raw products would figure significantly into the
cost of commercialization.

BIBLIOGRAPHY

‘‘Fuels, Synthetic Liquid’’ in ECT 1st ed., Vol. 6, pp. 960–983, by J. H. Arnold and H. Pich-
ler, Hydrocarbon Research, Inc.; ‘‘Carbon Monoxide–Hydrogen Reactions’’ in ECT 2nd
ed., Vol. 4, pp. 446–489, by H. Pichler and A. Hector, Carl-Engler und Hans-Bunte-Insti-
tut für Mineralöl und Kohleforschung der Technischen Hochschule Karlsruhe; ‘‘Fuels,
Synthetic (Liquid)’’ inECT 3rd ed., Vol. 11, pp. 447–489, by C. D. Kalfadelis and H.
Shaw, Exxon Research and Engineering Co.

CITED PUBLICATIONS

1. H. Mimoun, New J. Chem. 11, 4 (1987).


2. U. Preuss and M. Baerns, Chem. Eng. Technol. 10, 297 (1987).
3. ‘‘Liquid Fuels From Natural Gas,’’ Petrole Informations, API 34-5250, 96 (May
1987).
4. Chem. Eng. News, 25 (Aug. 14, 1989).
5. E. E. Ecklund and G. A. Mills, Chemtech, 549 (Sept. 1989).
6. L. Haar, in W. P. Earley and J. W. Weatherly, eds., Advances in Coal Utilization
Technology IV, Institute of Gas Technology, Chicago, 1981, 787–952.
7. L. Shnidman, in H. H. Lowry, ed., Chemistry of Coal Utilization, Vol. 2, John Wiley
& Sons, Inc., New York, 1945, 1252–1286.
8. W. L. Glowacki, in Ref. 7, 1136–1231; E. O. Rhodes, in Ref. 7, 1136–1231.
9. J. R. Rostrup-Nielsen, Steam Reforming Catalysts, Teknisk Forlag A/S, Copenha-
gen, 1975.
10. Catalyst Handbook, Wolfe Scientific Texts, London, 1970.
11. P. Sabatier, Catalysis, Then and Now, Part II, Franklin Publishing Co., Englewood,
N.J., 1965.
12. Fr. Pat. 571,356 (May 16, 1924), (to Badische Anilin- und Soda-Fabrik).
13. Fr. Pat. 580,905 (Nov. 19, 1924), (to Badische Anilin- und Soda-Fabrik).
14. F. Fischer and H. Tropsch, Ber. 56, 2428 (1923).
15. F. Fischer, Die Umwandlung der Kohle in Ole, Borntraeger, Berlin, 1923, p. 320.
16. F. Fischer and H. Tropsch, Ber. 59, 830, 832, 923 (1926).
17. F. Fischer, Brennstoff-Chem. 11, 492 (1930).
372 FUELS, SYNTHETIC, LIQUID Vol. 2

18. F. Fischer and H. Koch, Brennstoff-Chem. 13, 61 (1932).


19. F. Fischer and K. Meyer, Brennstoff-Chem. 12, 225 (1931).
20. F. Fischer and H. Pichler, Brennstoff-Chem. 20, 41, 221 (1939).
21. K. Fischer, Comparison of I. G. Work on Fischer Synthesis, Technical Oil Mission
Report, Reel 13, Library of Congress, Washington, D.C., July 1941.
22. H. Pichler, Medium Pressure Synthesis on Iron Catalyst, (Pat. Appl.), Technical
Oil Mission Report, Reel 100, Library of Congress, Washington, D.C., 1937–
1943.
23. H. Pichler, Medium Pressure Synthesis on Iron Catalyst, Technical Oil Mission
Report, Reel 101, Library of Congress, Washington, D.C., June 1940.
24. U.S. Pat. 2,327,066 (Aug. 17, 1943), O. Roelen.
25. F. Fischer, Ole Kohle 39, 517 (1943).
26. H. H. Storch, N. Golumbic, and R. B. Anderson, The Fischer-Tropsch and Related
Synthesis, John Wiley & Sons, Inc., New York, 1951.
27. W. G. Frankenburg, V. I. Komarewsky, and E. D. Rideal, Advances in Catalysis,
Vol I., Academic Press, Inc., New York, 1948, 115–156.
28. H. H. Storch, in Ref. 7, p. 1797.
29. O. C. Elvins and A. W. Nash, Fuel 5, 263 (1926).
30. O. C. Elvins, J. Soc. Chem. Ind. (London) 56, 473T (1927).
31. A. Erdeley and A. W. Nash, J. Soc. Chem. Ind. (London) 47, 219T (1928).
32. W. W. Myddleton, Chim. Ind. 37, 863 (1937); J. Inst. Fuel 11, 477 (1938); Colliery
Guardian 157, 286 (1938).
33. G. Egloff, Brennst.-Chem. 18, 11 (1937).
34. G. Egloff, E. F. Nelson, and J. C. Morrell, Ind. Eng. Chem. 29, 555 (1937).
35. F. Mako and W. A. Samuel, in R. A. Meyers, ed., Handbook of Synfuels Technology,
McGraw-Hill, Inc., New York, 1984, pp. 2-5–2-43.
36. J. C. Hoogendoorn, Phil. Trans. R. Soc. Lond. A 300, 99 (1981).
37. W. B. Johnson, Pet. Ref. 35 (Dec. 1956).
38. M. E. Dry, ‘‘Fischer-Tropsch Synthesis Over Iron Catalysts,’’ paper presented at
1990 Spring AIChE National Meeting, Orlando, Fla., Mar. 18–22, 1990.
39. B. Jager and co-workers, in Ref. 38.
40. A. E. Sands, H. W. Wainwright, and L. D. Schmidt, Ind. Eng. Chem. 40, 607 (1948);
A. E. Sands and L. D. Schmidt, Ind. Eng. Chem. 42, 2277 (1950).
41. H. Pichler, Technical Oil Mission Report, Reel 259, Library of Congress,
Washington, D.C., 1947, frames 467–654.
42. P. C. Keith, Oil Gas J. 45(6), 102 (1946).
43. Oil Gas J., 74 (Feb. 17, 1986).
44. M. J. v. d. Burgt and co-workers, in D. M. Bibby and co-workers, eds., Methane
Conversion, Elsevier Science, Inc., New York, 1988, 473–482.
45. I. E. Maxwell and J. E. Naber, Catal. Lett. 12, 105 (1992).
46. P. T. Roterud and co-workers, in Ref. 38.
47. C. D. Chang and A. J. Silvestri, J. Catal. 47, 249 (1977).
48. C. D. Chang, Catal. Rev. Sci. Eng. 25, 1 (1983).
49. C. D. Chang and A. J. Silvestri, Chemtech 17, 624 (1987).
50. U.S. Pat. 3,702,886 (1972), R. J. Argauer and G. R. Landolt (to Mobil Oil Corp.).
51. G. T. Kokotailo and co-workers, Nature 272, 437 (1978).
52. D. H. Olson, G. T. Kokotailo, and S. L. Lawton, J. Phys. Chem. 85, 2238 (1981).
53. C. D. Chang, in Ref. 44, 127–143.
54. G. J. Hutchings and R. Hunter, Catal. Today 6, 279 (1990).
55. F. Bauer, ZfI-Mitt. 156, 31 (1990).
56. P. B. Weisz and V. J. Frilette, J. Phys. Chem. 64, 382 (1960).
57. S. M. Csicsery, ACS Monograph 171, 680 (1976).
Vol. 2 FUELS, SYNTHETIC, LIQUID 373

58. N. Y. Chen, W. E. Garwood, and F. G. Dwyer, Shape Selective Catalysis in Industrial


Applications, Marcel Dekker, Inc., New York, 1989.
59. C. J. Maiden, in Ref. 44, 1–16.
60. J. E. Penick, W. Lee, and J. Maziuk, ACS Symp. Ser. 226, 19 (1983).
61. J. Z. Bem, in Ref. 44, 663–678.
62. P. L. Rogerson, in Ref. 35, pp. 2-45–2-73.
63. A. Y. Kam, M. Schreiner, and S. Yurchak, in Ref. 35, pp. 2-75–2-111.
64. D. Liederman and co-workers, Ind. Eng. Chem. Proc. Des. Devel. 17, 340 (1978).
65. H. R. Grimmer, N. Thiagarajian, and E. Nitschke, in Ref. 44, 273–291.
66. K. H. Keim and co-workers, Erdol. Erdgas, Kohle 103, 82 (1987).
67. K. G. Allum and A. R. Williams, in Ref. 44, 691–711.
68. A. A. Avidan, in Ref. 44, 307–323.
69. C. D. Chang, Catal. Rev.-Sci. Eng. 26 (3&4), 323 (1984).
70. M. M. Holm and E. H. Reichl, Fiat Report No. 1085, Office of Military Government
for Germany (U.S.), Mar. 31, 1947.
71. J. A. Sofranko, ‘‘Gas to Gasoline: The ARCO GTG Process,’’ paper presented at
Bicentenary Catalysis Meeting, Sydney, Australia, Sept. 1988.
72. J. A. Sofranko and J. C. Jubin, ‘‘Natural Gas to Gasoline: The ARCO GTG Process,’’
paper presented at International Chemical Congress of Pacific Basin Societies,
Honolulu, Hawaii, Dec. 1989.
73. J. H. Edwards, K. T. Do, and R. J. Tyler, in Ref. 72.
74. J. H. Edwards, K. T. Do, and R. J. Tyler, in E. E. Wolf, ed., Methane Conversion by
Oxidative Processes, Van Nostrand Reinhold, New York, 1992, 429–462.
75. R. J. Nichols, ‘‘Applications of Alternative Fuels,’’ SAE Special Publication SP-531,
Society of Automotive Engineers, Warrendale, Pa., Nov. 1982.
76. R. A. Potter, ‘‘Neat Methanol Fuel Injection Fleet Alternative Fuels Study,’’ paper
presented at Fourth Washington Conference on Alcohol, Washington, D.C., Nov.
1984.
77. N. D. Brinkman, Ener. Res. 3, 243 (1979).
78. T. Powell, ‘‘Racing Experiences with Methanol and Ethanol Based Motor-Fuel
Blends,’’ paper 750124, Automotive Engineering Congress and Exposition, Society
of Automotive Engineers, Detroit, Mich., Feb. 1975.
79. American Petroleum Institute, Task Force EF-18 of the Committee on Mobile
Source Emissions, Alcohols—A Technical Assessment of Their Application as Fuels,
Publication No. 4261, API, New York, July 1976.
80. U.S. Environmental Protection Agency, Fed. Reg. 46(144) (July 28, 1981).
81. U.S. Environmental Protection Agency, Fed. Reg. 50(12), 2615 (Jan. 17, 1985).
82. G. A. Mills and E. E. Ecklund, Chemtech, 626 (Oct. 1989).
83. KVB, Inc., KVB Report Number 72-804830-1998, Vol. 2, California Energy Commis-
sion, Irvine, Calif., Mar. 1985, 1–2.
84. A. J. Weir and co-workers, ‘‘Methanol Dual-Fuel Combustion,’’ paper presented at
1987 Joint Symposium on Stationary Combustion NOx Control, New Orleans, La.,
Mar. 23–26, 1987.
85. S. B. Alpert and D. F. Spencer, Methanol and Liquid Fuels From Coal—Recent
Advances, Electric Power Research Institute, Palo Alto, Calif., 1987.
86. V. Yand and S. C. Trindade, Chem. Eng. Prog., 11 (Apr. 1979).
87. Alcohols: Economics and Future U.S. Gasoline Markets, Information Resources, Inc.,
Washington, D.C., 1984.
88. H. L. Muller and S. P. Ho, ‘‘Economics and Energy Balance of Ethanol as Motor
Fuel,’’ paper presented at 1986 Spring AIChE National Meeting, New Orleans,
La., Apr. 1986.
89. Chem. Eng. News, 7 (Nov. 2, 1992).
374 FUELS, SYNTHETIC, LIQUID Vol. 2

90. ARCO Chemical Co., Testimony to the Colorado Air Quality Control Commission on Pro-
posed Regulation No. 13 (Oxygenate Mandate Program), Denver, Colo., June 4, 1987.
91. Chem. Eng. News, 28 (Apr. 12, 1993).
92. U.S. Pat. 1,251,954 (Jan. 1, 1918), F. Bergius and J. Billwiller.
93. F. Fischer and H. Tropsch, Ges. Abhandl. Kenntis Kohle 2, 154 (1918).
94. F. Bergius, Pet. Z. 22, 1275 (1926).
95. Gas World 104, 421 (1936).
96. Pet. Times 42, 641 (1939).
97. H. H. Storch and co-workers, U.S. Bur. Mines. Tech. Pap. 622 (1941).
98. A. C. Fieldner and co-workers, U.S. Bur Mines Tech. Pap. 666 (1944).
99. M. L. Kastens and co-workers, Ind. Eng. Chem. 41, 870 (1949).
100. J. L. Wiley and H. C. Anderson, U.S. Bur. Mines Bull. 485, I J. L. Wiley and H. C.
Anderson, U.S. Bur. Mines Bull. 485, (1950), II J. L. Wiley and H. C. Anderson,
U.S. Bur. Mines Bull. 485, (1951), III J. L. Wiley and H. C. Anderson, U.S. Bur.
Mines Bull. 485, (1952).
101. C. C. Chaffee and L. L. Hirst, Ind. Eng. Chem. 45, 822 (1953).
102. Bituminous Coal Staff, U.S. Bur. Mines Rep. Invest. 5506 (1959).
103. E. L. Clark and co-workers, Ind. Eng. Chem. 42, 861 (1950).
104. L. L. Newman and A. P. Pipilen, Gas Age 119(10), 16 (1957); 119(11), 18 (1957).
105. U.S. Pat. 2,860,101 (Nov. 11, 1958), M. G. Pelipetz (to the United States of America).
106. S. Akhtar, S. Friedman, and P. M. Yavorsky, U.S. Bur. Mines Tech. Prog. Rep. 35 (1971).
107. S. Akhtar and co-workers, ‘‘Process for Hydrodesulfurization of Coal,’’ paper
presented at 71st National AIChE Meeting, Dallas, Tex., Feb. 20, 1972.
108. B. Linville and J. D. Spencer, U.S. Bur. Mines Inf. Cir. 8612 (1973).
109. U.S. Pat. 3,321,393 (May 23, 1967), S. C. Schuman, R. H. Wolk, and M. C. Chervenak
(to Hydrocarbon Research, Inc.).
110. Coal Age, 101 (May 1976).
111. G. A. Johnson and co-workers, ‘‘Present Status of the H-Coal Process,’’ paper 30, IGT
Coal Symposium, Chicago, 1973.
112. J. E. Papso, in Ref. 35, pp. 1-47–1-63.
113. ‘‘High Pressure Hydrogenation at Ludwigshafen-Heidelberg,’’ Vol. IA, FIAT Final
Report No. 1317, ATI No. 92,762, Central Air Documents Office, Dayton, Ohio, 1951.
114. H. H. Lowry and H. J. Rose, U.S. Bur. Mines Inf. Cir. 7420 (1947).
115. A. Baba and co-workers, Rep. Resources Res. Inst. Jpn., (22) (1955).
116. U.S. Pat. 2,221,866 (Nov. 19, 1940), H. Dreyfus.
117. D. L. Kloepper and co-workers, Solvent Processing of Coal to Produce a De-ashed
Product, Contract 14-01-0001-275, OCR Report No. 9, U.S. Government Printing
Office, Washington, D.C., 1965.
118. V. L. Brant and B. K. Schmid, Chem. Eng. Prog. 68(12), 55 (1969).
119. B. K. Schmid and W. C. Bull, ‘‘Production of Ashless, Low-Sulfur Boiler Fuels From
Coal,’’ paper presented at ACS Division of Fuel Chem. Symposium on Pollution
Control, New York, Sept. 12, 1971.
120. Demonstration Plant, Clean Boiler Fuels From Coal, OCR R&D report no. 82,
Interim report no. 1, Vols. 1–3, Ralph M. Parsons Co., Los Angeles, Calif., 1973–1975.
121. M. E. Frank and B. K. Schmid, ‘‘Economic Evaluation and Process Design of a Coal–
Oil–Gas (COG) Refinery,’’ paper presented at Symposium on Conceptual Plants
for the Production of Synthetic Fuels From Coal, AIChE 65th Annual Meeting,
New York, Nov. 26, 1972.
122. U.S. Pat. 3,341,447 (Sept. 12, 1967), W. C. Bull and co-workers (to the United States
of America and Gulf Oil Corp.).
123. B. K. Schmid and D. M. Jackson, ‘‘The SRC-II Process,’’ paper presented at Third
Annual International Conference on Coal Gasification and Liquefaction, University
of Pittsburgh, Aug. 3–5, 1976; D. M. Jackson and B. K. Schmid, ‘‘Production of
Vol. 2 FUELS, SYNTHETIC, LIQUID 375

Distillate Fuels by SRC-II,’’ paper presented at ACS Div. of Ind. and Eng. Chem.
Symposium, Colorado Springs, Col., Feb. 12, 1979.
124. W. R. Epperly and J. W. Taunton, ‘‘Exxon Donor Solvent Coal Liquefaction Process
Development,’’ paper presented at Coal Dilemma II ACS Meeting, Colorado Springs,
Colo., Feb. 12, 1979.
125. D. E. Blaser and A. M. Edelman, ‘‘Flexicoking for Improved Utilization of Hydrocar-
bon Resources,’’ paper presented at API 43rd Mid-Year Meeting, Toronto, Canada,
May 8, 1978.
126. S. F. Massenzio, in Ref. 35, pp. 6-3–6-18.
127. T. A. Cavanaugh, W. R. Epperly, and D. T. Wade, in Ref. 35, pp. 1-3–1-46.
128. L. E. Swabb, Jr., G. K. Vick, and T. Aczel, ‘‘The Liquefaction of Solid Carbonaceous
Materials,’’ paper presented at The World Conference on Future Sources of Organic
Raw Materials, Toronto, Can., July 10, 1978.
129. E. L. Huffman, Proceedings of the Third Annual International Conference on Coal
Gasification and Liquefaction, Pittsburgh, Pa., 1976.
130. H. D. Schindler, J. M. Chen, and J. D. Potts, Final Technical Report on DOE Con-
tract No. DE-AC22-79ET14804, Department of Energy, Washington, D.C., 1983.
131. H. D. Schindler, Final Technical Report on DOE Contract No. D-AC01-87ER30110,
Vol. 2, Department of Energy, Washington, D.C., 1989.
132. A. G. Comolli and co-workers, Proceedings of the DOE Direct Liquefaction Contrac-
tors’ Review Meeting, Pittsburgh, Pa., 1986.
133. M. G. Thomas, in B. R. Cooper and W. A. Ellingson, eds., The Science and Technol-
ogy of Coal and Coal Liquefaction, Plenum Press, New York, 1984, 231–261.
134. M. J. Burges and R. V. Wheeler, Fuel 5, 65 (1926).
135. W. A. Selvig and W. H. Ode, U.S. Bur. Mines. Bull. 571, (1957).
136. H. C. Howard, in Ref. 7, Vol. 1, 761–773.
137. Ibid., Suppl. Vol., 340–394.
138. J. F. Jones and co-workers, Chem. Eng. Prog. 62(2), 73 (1966).
139. J. A. Hamshar, H. D. Terzian, and L. J. Scotti, ‘‘Clean Fuels From Coal by the COED
Process,’’ paper presented at EPA Symposium on Environmental Aspects of Fuel
Conversion Technology, St. Louis, Mo., May 1974.
140. C. D. Kalfadelis and E. M. Magee, Evaluation of Pollution Control in Fossil Fuel
Conversion Processes, Liquefaction, Section 1, COED Process, EPA-650/2-74-009e,
Environmental Protection Agency, Washington, D.C., 1975.
141. R. T. Eddinger and co-workers, Char Oil Energy Development, Office of Coal
Research R&D Report No. 11, Vol. 1 (PB 169,562) and Vol. 2 (PB 169,563), U.S.
Government Printing Office, Washington, D.C., Mar. 1966.
142. J. F. Jones and co-workers, Char Oil Energy Development, Office of Coal Research
R&D Report No. 11, Vol. 1 (PB 173,916) and Vol. 2 (PB 173,917), U.S. Government
Printing Office, Washington, D.C., Feb. 1967.
143. M. E. Sacks and co-workers, Char Oil Energy Development, Office of Coal Research
Report 56, Interim Report No. 2, GPO Cat. No. 163.10:56/Int.2, U.S. Government
Printing Office, Washington, D.C., Jan. 1971.
144. J. F. Jones and co-workers, Char Oil Energy Development, Office of Coal Research
R&D Report No. 56, Final Report, GPO Cat. No. 163.10:56, U.S. Government Print-
ing Office, Washington, D.C., May 1972.
145. J. F. Jones and co-workers, Char Oil Energy Development, Office of Coal Research
R&D Report No. 73, Interim Report No. 1, GPO Cat. No. 163.10:73/Int 1, U.S.
Government Printing Office, Washington, D. C., Dec. 1972.
146. A. Sass, ‘‘The Garrett Research and Development Company Process for the Conver-
sion of Coal into Liquid Fuels,’’ paper presented at 65th Annual AIChE Meeting,
New York, Nov. 29, 1972.
147. Oil Gas J., 78 (Aug. 26, 1974).
376 FUELS, SYNTHETIC, LIQUID Vol. 2

148. A. Sass, Chem. Eng. Prog. 70(1), 72 (1974).


149. F. B. Carlson and co-workers, Chem. Eng. Prog. 69(3), 50 (1973).
150. F. B. Carlson, L. H. Yardumian, and M. T. Atwood, ‘‘The TOSCOAL Process for Low
Temperature Pyrolysis of Coal,’’ paper presented at American Institute of Mining,
Metallugical, and Petroleum Engineers, San Francisco, Calif., Feb. 22, 1972, and
to American Institute of Chemical Engineers, New York, Nov. 29, 1972.
151. W. Tiddy and M. J. Miller, Am. Gas J. 153(3), 7 (1940).
152. M. J. Gavin, U.S. Bur. Mines Bull. 210, (1922).
153. M. J. Gavin and J. S. Desmond, U.S. Bur. Mines Bull. 315 (1930).
154. H. Shaw, C. D. Kalfadelis, and C. E. Jahnig, Evaluation of Methods to Produce Avia-
tion Turbine Fuels From Synthetic Crude Oils-Phase I, Technical Report AFAPL-TR-
75-10, Vol. I, Air Force Aero Propulsion Laboratory, Wright-Patterson Air Force
Base, Dayton, Ohio, Mar. 1975.
155. C. D. Kalfadelis, Evaluation of Methods to Produce Aviation Turbine Fuels From Syn-
thetic Crude Oils-Phase II, Technical Report AFAPL-TR-75-10, Vol. II, Air Force Aero
Propulsion Laboratory, Wright-Patterson Air Force Base, Dayton, Ohio, May 1976.
156. Colony Development Operation, An Environmental Impact Analysis for a Shale Oil
Complex at Parachute Creek, Colorado, Vols. 1–3, Denver, Colo., 1974.
157. H. L. Erskine, in Ref. 35, pp. 5-3–5-32.
158. R. D. Hynphreys, F. K. Spragins, and D. R. Craig, ‘‘Oil Sands—Canada’s First
Answer to the Energy Shortage,’’ Proceedings 9th World Petroleum Congress, Tokyo,
Japan, May 11, 1975, Vol. 5, p. 17.
159. C. W. Bowman, R. S. Phillips, and L. R. Turner, in Ref. 35, pp. 5-33–5-79.
160. M. Crow and co-workers, Synthetic Fuel Technology Development in the United
States—A Retrospective Assessment, Praeger Publishing, New York, 1988.
161. S. Yurchak and S. S. Wong, ‘‘Mobil Methanol Conversion Technology,’’ Proceedings
IGT Asian Natural Gas Seminar, Singapore, 1992, 593–618.
162. J. L. Matherne and G. L. Culp, in Ref. 73, 463–482.
163. J. E. Duddy, S. B. Panvelker, and G. A. Popper, ‘‘Commercial Economics of HRI
Coal/Oil Co-Processing Technology,’’ paper presented at 1990 SummerAIChE
National Meeting, San Diego, Ca., 1990.
164. M. Ikura and J. F. Kelly, ‘‘A Techno-Economic Evaluation of CANMET Coprocessing
Technology,’’ Proceedings Annual International Pittsburgh Coal Conference,
1990, 719–728.
165. J. G. Sikonia, B. R. Shah, and M. A. Ulowetz, ‘‘Technical and Economic Assessment
of Petroleum, Heavy Oil, Shale Oil and Coal Liquid Refining,’’ paper presented at
Synfuels’ 3rd Worldwide Symposium, Washington, D.C., Nov. 1–3, 1983.

GENERAL REFERENCE

American Petroleum Institute, Alcohols and Ethers-A Technical Assessment of Their


Application as Fuels and Fuel Components, API Publication 4261, American Petroleum
Institute, Washington, D.C., July 1988.

SCOTT HAN
CLARENCE D. CHANG
Mobil Research and Development
Corporation
Vol. 2 FUELS, SYNTHETIC, GASEOUS 377

FUELS, SYNTHETIC,
GASEOUS

Substitute or synthetic natural gas (SNG) has been known for several centu-
ries. When SNG was first discovered, natural gas was largely unknown as a
fuel and was more a religious phenomenon (see NATURAL GAS) (1). Coal (qv)
was the first significant source of substitute natural gas and in the early
stages of SNG production the product was more commonly known under varia-
tions of the name coal gas (2,3). Whereas coal continues to be a principal source
of substitute natural gas (4) a more recently recognized source is petroleum
(qv) (5).

1. Gas from Coal

Coal can be converted to gas by several routes (2,6 – 11), but often a parti-
cular process is a combination of options chosen on the basis of the pro-
duct desired, ie, low, medium, or high heat-value gas. In a very general
sense, coal gas is the term applied to the mixture of gaseous constituents
that are produced during the thermal decomposition of coal at tempera-
tures in excess of 5008C ( > 930 F), often in the absence of oxygen (air) (3).
A solid residue (coke, char), tars, and other liquids are also produced in
the process:

Ccoal þ heat!Cchar þ tar=liquid þ CO þ CO2 þ H2

The tars and other liquids (liquor) are removed by condensation leaving
principally hydrogen, carbon monoxide, and carbon dioxide in the gas phase.
This gaseous product also contains low boiling hydrocarbons (qv), sulfur-con-
taining gases, and nitrogen-containing gases including ammonia (qv) and
hydrogen cyanide. The solid residue is then treated under a variety of condi-
tions to produce other fuels which vary from a purified char to different
types of gaseous mixtures. The amounts of gas, coke, tar, and other liquid pro-
ducts vary according to the method used for the carbonization (especially the
retort configuration), and process temperature, as well as the nature (rank) of
the coal (3,11).
The recorded chronology of the coal-to-gas conversion technology beganin
1670 when a clergyman, John Clayton, in Wakefield, Yorkshire, produced in
the laboratory a luminous gas by destructive distillation of coal (12). At the
same time, experiments were also underway elsewhere to carbonize coal to pro-
duce coke, but the process was not practical on any significant scale until 1730
(12). In 1792, coal was distilled in an iron retort by a Scottish engineer, who used
the by-product gas to illuminate his home (13).
The conversion of coal to gas on an industrial scale dates to the early nine-
teenth century (14). The gas, often referred to as manufactured gas, was pro-
duced in coke ovens or similar types of retorts by simply heating coal to

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.0701190519160509.a01
378 FUELS, SYNTHETIC, GASEOUS Vol. 2

vaporize the volatile constituents. Estimates based on modern data indicate that
the gas mixture probably contained hydrogen (qv) (ca 50%), methane
(ca 30%), carbon monoxide (qv) and carbon dioxide (qv) (ca 15%), and some
inert material, such as nitrogen (qv), from which a heating value of approxi-
mately 20.5 MJ/m3 (550 Btu/ft3) can be estimated (6).
Blue gas, or blue-water gas, so-called because of the color of the flame upon
burning (10), was discovered in 1780 when steam was passed over incandescent
carbon (qv), and the blue-water gas process was developed over the period 1859–
1875. Successful commercial application of the process came about in 1875 with
the introduction of the carburetted gas jet. The heating value of the gas was low,
ca 10.2 MJ/m3 (275 Btu/ft3), and on occasion oil was added to the gas to enhance
the heating value. The new product was given the name carburetted water gas
and the technique satisfied part of the original aim by adding luminosity to gas
lights (10).
Coke-oven gas is a by-product fuel gas derived from coking coals by the pro-
cess of carbonization. The first by-product coke ovens were constructed in France
in 1856. Since then they have gradually replaced the old and primitive method of
beehive coking for the production of metallurgical coke. Coke-oven gas is pro-
duced in an analogous manner to retort coal gas, with operating conditions,
mainly temperature, set for maximum carbon yield. The resulting gas is, conse-
quently, poor in illuminants, but excellent as a fuel. Typical analyses and heat
content of common fuel gases vary (Table 1) and depend on the source as well
as the method of production.
In Germany, large-scale production of synthetic fuels from coal began in
1910 and necessitated the conversion of coal to carbon monoxide and hydrogen.

Water gas reaction Ccoal þ H2 O!CO þ H2

The mixture of carbon monoxide and hydrogen is enriched with hydrogen from
the water gas catalytic (Bosch) process, ie, water gas shift reaction, and passed

Table 1. Analyses of Fuel Gases


Gas composition, vol%
Heat value,a
3
Type of fuel gas CO CO2 H2 N2 O2 CH4 Illuminants MJ/m
blast-furnace 27.5 10.0 3.0 58.0 1.0 0.5 3.8
producer (bituminous) 27.0 4.5 14.0 50.9 0.6 3.0 5.6
blue-water 42.8 3.0 49.9 3.3 0.5 0.5 11.5
carburetted water 33.4 3.9 34.6 7.9 0.9 10.4 8.9b 20.0
retort coal 8.6 1.5 52.5 3.5 0.3 31.4 2.2c 21.5
coke-oven 6.3 1.8 53.0 3.4 0.2 1.6 3.7d 21.9
natural
mid-continent 0.8 3.2 96.0 36.1
Pennsylvania 1.1 67.6 31.3e 46.0
a
To convert MJ/m3 to Btu/ft3, multiply by 26.86.
b
6.7 vol% C2H4 plus 2.2 vol% C6H6.
c
1.1 vol% each of C2H4 and C6H6.
d
2.7 vol% C2H4 plus 1.0 vol% C6H6.
e
31.3 vol% C2H6.
Vol. 2 FUELS, SYNTHETIC, GASEOUS 379

over a cobalt–thoria catalyst to form straight-chain, ie, linear, paraffins, olefins,


and alcohols in what is known as the Fisher-Tropsch synthesis.

cobalt
n CO þ ð2n þ 1Þ H2 ! C H
catalyst n 2nþ2
þ n H2 O
iron
2n CO þ ðn þ 1Þ H2 ! Cn H2nþ2 þ n CO2
catalyst

cobalt
n CO þ 2n H2 ! Cn H2n þ n H2 O
catalyst

iron
CO þ n H2  ! Cn H2n þ n CO2
catalyst

cobalt
n CO þ 2n H2 ! Cn H2nþ1 OH þ ðn  1Þ H2 O
catalyst

iron
ð2n  1Þ CO þ ðn þ 1Þ H2  ! Cn H2nþ1 OH þ ðn  1Þ CO2
catalyst

In Sasolburg, South Africa, a commercial plant using the Fischer-Tropsch


process was completed in 1950 and began producing a variety of liquid fuels and
chemicals. The facility has been expanded to produce a considerable portion of
South Africa’s energy requirements (15,16).
In 1948, the first demonstration of suspension gasification was success-
fully completed by Koppers, Inc. The product gas was of 11.2 MJ/m3
(300 Btu/ft3) calorific value and consisted primarily of a mixture of hydrogen
and oxides of carbon. In the United States, so-called second-generation coal
gasification processes came into being as a result of the recognized need to
develop reliable, domestic energy sources to replace the rapidly diminishing
supply of conventional fuels (see FUEL RESOURCES) (3,9). More recently, the
biological conversion of coal and synthesis gas (carbon monoxide–hydrogen
mixtures) into liquid fuels by methanogenic bacteria has received some
attention (17–19).
1.1. Gas Products. The originally designated names of the gaseous
mixtures are used herein, with the understanding that since their introduction
there may be differences in means of production and in the make-up of the gas-
eous products. Properties of fuel gases are available (20). There are standard
tests to determine properties and character of gaseous mixtures. These tests
are accepted by the American Society for Testing and Materials (ASTM), by
the British Standards Institution (BSI), by the Institute of Petroleum (IP), and
by the International Standards Organization (ISO) (1,10,20).
Low Heat-Value Gas. Low heat-value (low Btu) gas (7) consists of a mix-
ture of carbon monoxide and hydrogen and has a heating value of less than 11
MJ/m3 (300 Btu/ft3), but more often in the range 3.3–5.6 MJ/m3 (90–150 Btu/ft3).
The gas is formed by partial combustion of coal with air, usually in the presence
of steam (7).

2 Ccoal þ O2 !2 CO
Ccoal þ H2 O!CO þ H2
CO þ H2 O!CO2 þ H2
380 FUELS, SYNTHETIC, GASEOUS Vol. 2

This gas is of interest to industry as a fuel gas or even, on occasion, as a raw


material from which ammonia and other compounds may be synthesized.
The first gas producer making low heat-value gas was built in 1832. (The
product was a combustible carbon monoxide–hydrogen mixture containing
ca 50 vol% nitrogen). The open-hearth or Siemens-Martin process, built in
1861 for pig iron refining, increased low heat-value gas use (see IRON). The
use of producer gas as a fuel for heating furnaces continued to increase
until the turn of the century when natural gas began to supplant manufac-
tured fuel gas.
The combustible components of the gas are carbon monoxide and hydrogen,
but combustion (heat) value varies because of dilution with carbon dioxide and
with nitrogen. The gas has a low flame temperature unless the combustion air
is strongly preheated. Its use has been limited essentially to steel (qv) mills,
where it is produced as a by-product of blast furnaces. A common choice of equip-
ment for the smaller gas producers is the Wellman-Galusha unit because of its
long history of successful operation (21).
Medium Heat-Value Gas. Medium heat-value (medium Btu) gas (6,7)
has a heating value between 9 and 26 MJ/m3 (250 and 700 Btu/ft3). At the
lower end of this range, the gas is produced like low heat-value gas, with the
notable exception that an air separation plant is added and relatively pure oxy-
gen (qv) is used instead of air to partially oxidize the coal. This eliminates the
potential for nitrogen in the product and increases the heating value of the pro-
duct to 10.6 MJ/m3 (285 Btu/ft3). Medium heat-value gas consists of a mixture of
methane, carbon monoxide, hydrogen, and various other gases and is suitable as
a fuel for industrial consumers.
High Heat-Value Gas. High heat-value (high Btu) gas (7) has a heating
value usually in excess of 33.5 MJ/m3 (900 Btu/ft3). This is the gaseous fuel that
is often referred to as substitute or synthetic natural gas (SNG), or pipeline-
quality gas. It consists predominantly of methane and is compatible with natural
gas insofar as it may be mixed with, or substituted for, natural gas.
Any of the medium heat-value gases that consist of carbon monoxide and
hydrogen (often called synthesis gas) can be converted to high heat-value gas
by methanation (22), a low temperature catalytic process that combines carbon
monoxide and hydrogen to form methane and water.

CO þ 3 H2 !CH4 þ H2 O

Prior to methanation, the gas product from the gasifier must be thoroughly pur-
ified, especially from sulfur compounds the precursors of which are widespread
throughout coal (23) (see SULFUR AND HYDROGEN SULFIDE RECOVERY). Moreover, the
composition of the gas must be adjusted, if required, to contain three parts hydro-
gen to one part carbon monoxide to fit the stoichiometry of methane production.
This is accomplished by application of a catalytic water gas shift reaction.

CO þ H2 O !
 CO2 þ H2
The ratio of hydrogen to carbon monoxide is controlled by shifting only part
of the gas stream. After the shift, the carbon dioxide, which is formed in the gasi-
fier and in the water gas reaction, and the sulfur compounds formed during gasi-
fication, are removed from the gas.
Vol. 2 FUELS, SYNTHETIC, GASEOUS 381

The processes that have been developed for the production of synthetic nat-
ural gas are often configured to produce as much methane in the gasification step
as possible thereby minimizing the need for a methanation step. In addition,
methane formation is highly exothermic which contributes to process efficiency
by the production of heat in the gasifier, where the heat can be used for the
endothermic steam–carbon reaction to produce carbon monoxide and hydrogen.

C þ H2 O!CO þ H2

1.2. Carbonization. Next to combustion, carbonization represents one


of the largest uses of coal (2,24–26). Carbonization is essentially a process for
the production of a carbonaceous residue by thermal decomposition, accompa-
nied by simultaneous removal of distillate, of organic substances.

Corganic !Ccoke=char=carbon þ liquids þ gases

This process may also be referred to as destructive distillation. It has been


applied to a whole range of organic materials, more particularly to natural pro-
ducts such as wood (qv), sugar (qv), and vegetable matter to produce charcoal
(see FUELS FROM BIOMASS). However, in the present context, coal usually yields
coke, which is physically dissimilar from charcoal and appears with the more
familiar honeycomb-type structure (27).
The original process of heating coal in rounded heaps, the hearth process,
remained the principal method of coke production for over a century, although an
improved oven in the form of a beehive was developed in the Durham-Newcastle
area of England in about 1759 (2, 26, 28). These processes lacked the capability to
collect the volatile products, both liquids and gases. It was not until the mid-
nineteenth century, with the introduction of indirectly heated slot ovens, that
it became possible to collect the liquid and gaseous products for further use.
Coal carbonization processes are generally defined according to process
operating temperature. Terms are defined in Table 2.

Table 2. Coal Carbonization Methodsa


Carbonization Final tempe-
process rature, 8C Products Processes
low temperature 500–700 reactive coke and rexco (7008C) made in
high tar yield cylindrical vertical
retorts; coalite (6508C)
made in vertical tubes
medium temperature 700–900 reactive coke with town gas and gas coke
high gas yield, or (obsolete); phurnacite,
domestic briquettes low volatile steam coal,
pitch-bound briquettes
carbonized at 8008C
high temperature 900–1050 hard, unreactive coal foundry coke (9008C);
for metallurgical use blast-furnace coke
(950–10508C)
a
Ref. 29.
382 FUELS, SYNTHETIC, GASEOUS Vol. 2

Low Temperature Carbonization. Low temperature carbonization, when


the process does not exceed 7008C, was mainly developed as a process to supply
town gas for lighting purposes as well as to provide a smokeless (devolatilized)
solid fuel for domestic consumption (30). However, the process by-products
(tars) were also found to be valuable insofar as they served as feedstocks
for an emerging chemical industry and were also converted to gasolines, heat-
ing oils, and lubricants (see GASOLINE AND OTHER MOTOR FUELS; LUBRICATION AND
LUBRICANTS) (31).
Coals preferred for the low temperature carbonization were usually lignites
or subbituminous, as well as high volatile bituminous, coals (see LIGNITE AND
BROWN COAL). These yield porous solid products over the temperature range
600–7008C. Certain of the higher rank (caking) coals were less suitable for
the process, unless steps were taken to destroy the caking properties, because
of the tendency of these higher rank coals to adhere to the walls of the carboniza-
tion chamber.
The options for efficient low temperature carbonization of coal include
vertical and horizontal retorts which have been used for batch and continuous
processes. In addition, stationary and revolving horizontal retorts have
also been operated successfully, and there are also several process options
employing fluidized or gas-entrained coal. Coke production from batch-type
carbonization of coal has been supplanted by a variety of continuous retorting
processes which allow much greater throughput rates than were previously
possible. These processes employ rectangular or cylindrical vessels of sufficient
height to carbonize the coal while it travels from the top of the vessel to the
bottom and usually employ the principle of heating the coal by means of a
countercurrent flow of hot combustion gas. Most notable of these types of
carbonizers are the Lurgi-Spulgas retort and the Koppers continuous steaming
oven (2).
High Temperature Carbonization. When heated at temperatures in
excess of 7008C (12908F), low temperature chars lose their reactivity through
devolatilization and also suffer a decrease in porosity. High temperature carbo-
nization, at temperatures >900 C, is, therefore, employed for the production of
coke (27). As for the low temperature processes, the tars produced in high tem-
perature ovens are also sources of chemicals and chemical intermediates (32).
A newer concept has been developed that is given the name mild gasifica-
tion (33). It is not a gasification process in the true sense of the word. The process
temperature is some several hundred degrees lower, hence the term mild, than
the usual gasification process temperature and the objective is not to produce a
gaseous fuel but to produce a high value char (carbon) and liquid products. Gas is
produced, but to a lesser extent.
Documented efforts at cokemaking date from 1584 (34), and have seen
various adaptations of conventional wood-charring methods to the production
of coke including the eventual evolution of the beehive oven, which by the
mid-nineteenth century had become the most common vessel for the coking of
coal (2). The heat for the process was supplied by burning the volatile matter
released from the coal and, consequently, the carbonization would progress
from the top of the bed to the base of the bed and the coke was retrieved from
the side of the oven at process completion.
Vol. 2 FUELS, SYNTHETIC, GASEOUS 383

Some beehive ovens, having various improvements and additions of waste


heat boilers, thereby allowing heat recovery from the combustion products, may
still be in operation. Generally, however, the beehive oven has been replaced by
wall-heated, horizontal chamber, ie, slot, ovens in which higher temperatures
can be achieved as well as a better control over the quality of the coke. Modern
slot-type coke ovens are approximately 15 m long, approximately 6 m high, and
the width is chosen to suit the carbonization behavior of the coal to be processed.
For example, the most common widths are ca 0.5 m, but some ovens may be as
narrow as 0.3 m, or as wide as 0.6 m.
Several (usually 20 or more, alternating with similar cells that contain
heating ducts) of these chambers are constructed in the form of a battery over
a common firing system through which the hot combustion gas is conveyed to
the ducts. The flat roof of the battery acts as the surface for a mobile
charging car from which the coal (25–40 t) enters each oven through three open-
ings along the top. The coke product is pushed from the rear of the oven through
the opened front section onto a quenching platform or into rail cars that then
move the coke through water sprays. The gas and tar by-products of the process
are collected for further processing or for on-site use as fuel.
Most modern coke ovens operate on a regenerative heating cycle in order to
obtain as much surplus gas as possible for use on the works, or for sale. If coke-
oven gas is used for heating the ovens, the majority of the gas is surplus to
requirements. If producer gas is used for heating, much of the coke-oven gas is
surplus.
The main difference between gas works and coke oven practice is that, in a gas
works, maximum gas yield is a primary consideration whereas in the coke works the
quality of the coke is the first consideration. These effects are obtained by choice of a
coal feedstock that is suitable to the task. For example, use of lower volatile coals
in coke ovens, compared to coals used in gas works, produces lower yields of
gas when operating at the same temperatures. In addition, the choice of heating
(carbonizing) conditions and the type of retort also play a principal role (10,35).
1.3. Gasification. The gasification of coal is essentially the conversion
of coal by any one of a variety of processes to produce combustible gases
(7,8,11,22,36–38). Primary gasification is the thermal decomposition of coal to
produce mixtures containing various proportions of carbon monoxide, carbon
dioxide, hydrogen, water, methane, hydrogen sulfide, and nitrogen, as well as
products such as tar, oils, and phenols. A solid char product may also be pro-
duced, and often represents the bulk of the weight of the original coal.
Secondary gasification involves gasification of the char from the primary
gasifier, usually by reaction of the hot char and water vapor to produce carbon
monoxide and hydrogen.

Cchar þ H2 O ! CO þ H2

The gaseous product from a gasifier generally contains large amounts of carbon
monoxide and hydrogen, plus lesser amounts of other gases and may be of low,
medium, or high heat value depending on the defined use (Table 3) (39,40).
The importance of coal gasification as a means of producing fuel gas(es) for
industrial use cannot be underplayed. But coal gasification systems also have
384 FUELS, SYNTHETIC, GASEOUS Vol. 2

Table 3. Gas Composition Requirements for Substitute Natural Gas and for Power
Generation
Product requirement
Characteristic Synthetic natural gas Power generation
methane content high, less synthesis required low, probably means no tars
H2/CO ratio high, less shifting required low, CO more efficient fuel
moisture content high, steam required for shift low, lower condensate treatment
costs
outlet low, maximizes methane high, precludes tar formation
temperature minimizes sensible heat loss provides for steam generation
leads to high cold gas efficiency reduces cold gas efficiency
gasifier oxidant O2 only, cost of N2 removal air or O2, low heating gas value
excessive acceptable fuel

undesirable features. A range of undesirable products are also produced which


must be removed before the products are used to provide fuel and/or to generate
electric power (22,41).
Chemistry. Coal gasification involves the thermal decomposition of coal
and the reaction of the carbon in the coal, and other pyrolysis products with oxy-
gen, water, and hydrogen to produce fuel gases such as methane by internal
hydrogen shifts
Ccoal þ Hcoal !CH4

or through the agency of added (external) hydrogen

Ccoal þ 2 H2 !CH4

although the reactions are more numerous and more complex as can be seen in
Table 4.
If air is used as a combustant, the product gas contains nitrogen and,
depending on design characteristics, has a heating value of approximately
5:6  11:2  103 MJ=m3 (150–300 Btu/ft3). The use of pure oxygen, although
expensive, results in a product gas having a heating value of 11–15 MJ/m3
(300–400 Btu/ft3) and carbon dioxide and hydrogen sulfide as by-products.
If a high heat-value gas (33.5–37.3 MJ/m3 (900–1000 Btu/ft3)) ie, SNG, is
the desired product, efforts must be made to increase the methane content.
Shift conversion reaction

CO þ H2 O!CO2 þ H2
CO þ 3 H2 !CH4 þ H2 O
2 CO þ 2 H2 !CH4 þ CO2
CO þ 4 H2 !CH4 þ 2 H2 O

The gasification is performed using oxygen and steam (qv), usually at elevated
pressures. The steam–oxygen ratio along with reaction temperature and
Vol. 2 FUELS, SYNTHETIC, GASEOUS 385

Table 4. Gasification Reactions


Reaction DH, kJa
Gasification zone, 595–12058C
C þ CO2 !2 CO 9.65b
CO þ H2 O!CO2 þ H2 1:93c
C þ H2 O!CO þ H2 7.76d
C þ 2H2 !CH4 5:21
C þ 2 H2 O!2 H2 þ CO2 5.95
Combustion zone, >1205 C
2 C þ O2 !2 CO 22:6
C þ O2 !CO2 6:74
Methanation reactions
CO þ 3 H2 !CH4 þ H2 O 12:9
CO2 þ 4 H2 !CH4 þ 2 H2 O 11:0
2 C þ 2 H2 O!CH4 þ CO2 0.62
Others reactions
2 H2 þ O2 !2 H2 O
CO þ 2 H2 !CH3 OH
2 C þ H2 !C2 H2
CH4 þ 2 H2 O!CO2 þ 4 H2
a
To convert kJ to kcal, divide by 4.184.
b
Boudouard reaction.
c
Water gas shift reaction.
d
Water gas reaction.

pressure determine the equilibrium gas composition. The reaction rates for these
reactions are relatively slow and heats of formation are negative. Catalysts may
be necessary for complete reaction (2,3,24,42,43).
Process Parameters. The most notable effects in gasifiers are those of
pressure (Fig. 1) and coal character. Some initial processing of the coal feedstock

60

50 H2
Composition, vol %

40 20
Heating value, MJ/m3

30

CO
20
CH4 15
10

0
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Gasification pressure, MPa

Fig. 1. Variation of (—) gas composition and () heating value with gasifier pressure.
To convert MJ/m3 to Btu/ft3, multiply by 26.86. To convert MPa to psi, multiply by 145.
386 FUELS, SYNTHETIC, GASEOUS Vol. 2

may be required. The type and degree of pretreatment is a function of the


process and/or the type of coal.
Depending on the type of coal being processed and the analysis of the gas
product desired, some or all of the following processing steps may be required:
(1) pretreatment of the coal (if caking is a problem); (2) primary gasification of
the coal; (3) secondary gasification of the carbonaceous residue from the primary
gasifier; (4) removal of carbon dioxide, hydrogen sulfide, and other acid gases;
(5) shift conversion for adjustment of the carbon monoxide–hydrogen mole
ratio to the desired ratio; and (6) catalytic methanation of the carbon monox-
ide–hydrogen mixture to form methane. If high heat-value gas is desired, all
of these processing steps are required because coal gasifiers do not yield methane
in the concentrations required.
An example of application of a pretreatment option occurs when the coal
displays caking or agglomerating characteristics. Such coals are usually not
amenable to gasification processes employing fluidized-bed or moving-bed reac-
tors. The pretreatment involves a mild oxidation treatment, usually consisting
of low temperature heating of the coal in the presence of air or oxygen. This
destroys the caking characteristics of coals.
Gasification technologies for the production of high heat-value gas do not all
depend entirely on catalytic methanation, that is, the direct addition of hydrogen
to coal under pressure to form methane.
The hydrogen-rich gas for hydrogasification can be manufactured from
steam by using the char that leaves the hydrogasifier. Appreciable quantities
of methane are formed directly in the primary gasifier and the heat released
by methane formation is at a sufficiently high temperature to be used in the
steam–carbon reaction to produce hydrogen so that less oxygen is used to
produce heat for the steam–carbon reaction. Hence, less heat is lost in the
low temperature methanation step, thereby leading to a higher overall process
efficiency.
There are three fundamental reactor types for gasification processes: (1) a
gasifier reactor, (2) a devolatilizer, and (3) a hydrogasifier (Fig. 2). The choice of a
particular design is available for each type, eg, whether or not two stages should
be involved depending on the ultimate product gas desired. Reactors may also be
designed to operate over a range of pressure from atmospheric to high pressure.
Gasification processes have been classified on the basis of the heat-value of
the produced gas. It is also possible to classify gasification processes according to
the type of reactor and whether or not the system reacts under pressure. Addi-
tionally, gasification processes can be segregated according to the bed types,
which differ in the ability to accept and use caking coals. Thus gasification pro-
cesses can be divided into four categories based on reactor (bed) configuration: (1)
fixed bed, (2) moving bed, (3) fluidized bed, and (4) entrained bed.
In a fixed-bed process the coal is supported by a grate and combustion
gases, ie, steam, air, oxygen, etc, pass through the supported coal whereupon
the hot produced gases exit from the top of the reactor. Heat is supplied intern-
ally or from an outside source, but caking coals cannot be used in an unmodified
fixed-bed reactor. In the moving-bed system (Fig. 3), coal is fed to the top of the
bed and ash leaves the bottom with the product gases being produced in the hot
zone just prior to being released from the bed.
Vol. 2 FUELS, SYNTHETIC, GASEOUS 387

C + H2O CO + H2
Coal Gas
CO + H2O CO2 + H2 + Heat

(a)

Coal
C + 2 H2 CH4 Gas
Steam and
H2-rich gas
(b)

Hydrogen
Coal + H2 CH4+ C Gas
Coal

(c)

Fig. 2. Chemistry of (a) gasifier; (b) hydrogasifier; and (c) devolatization processes. The
gaseous product of (a) is of low heating value; that of (b) and (c) is of intermediate heating
value.

Steam,
Coal oxygen,
Gas Coal or air

Coal

Gas

Steam, Steam,
oxygen, Ash oxygen, Ash Slag
or air or air

(a) (b) (c)

Fig. 3. Gasifier systems: (a), moving bed (dry ash); (b), fluidized bed; and (c), entrained
flow.
388 FUELS, SYNTHETIC, GASEOUS Vol. 2

The fluidized-bed system (Fig. 3) uses finely sized coal particles and the bed
exhibits liquid-like characteristics when a gas flows upward through the bed.
Gas flowing through the coal produces turbulent lifting and separation of
particles and the result is an expanded bed having greater coal surface area to
promote the chemical reaction. These systems, however, have only a limited abil-
ity to handle caking coals (see FLUIDIZATION).
An entrainment system (Fig. 3) uses finely sized coal particles blown into
the gas steam prior to entry into the reactor and combustion occurs with the
coal particles suspended in the gas phase; the entrained system is suitable for
both caking and noncaking coals.
Gasifier Options. The standard Wellman-Galusha unit, used for noncak-
ing coals, is an atmospheric pressure, air-blown gasifier (Fig. 4) fed by a two-com-
partment lockhopper system. The upper coal-storage compartment feeds coal
intermittently into the feeding compartment, which then feeds coal into the gasi-
fier section almost continuously except for brief periods when the feeding com-
partment is being loaded from storage. In small units up to 1.5 m internal
diameter ash is removed by shaker grates. In larger units, ash is removed con-
tinuously from the bottom of the gasifier into an ash-hopper section by a revol-
ving grate.
The grate is constructed of flat, circular steel plates set one above the other
with edges overlapping. The grate, revolving eccentrically within the gasifier,
causes ash to fall from the coal bed as the space between the grate and the
shell increases, and then pushes the ash down into the ash-hopper as the
space decreases. The smaller size units are brick-lined, although the larger
sizes are unlined and water-jacketed. Combustion air, provided by a fan, passes
over the warm (828C) jacket water causing the water to vaporize to provide the
necessary steam for gasification. Gas leaves from above the fuel bed at 428–
5388C for bituminous coal.
On the other hand, the agitator type of Wellman-Galusha unit, used for
gasification of any type of coal, uses a slowly revolving horizontal arm which
also spirals vertically below the surface of the fuel bed to retard channeling
and maintain a uniform bed. Use of the agitator not only allows operation with
caking coals but also can increase the capacity of the gasifier by about 25% for
use with other coals.
The Winkler gasifier (Fig. 5) is an example of a medium heat-value gas pro-
ducer which, when oxygen is employed, yields a gas product composed mainly of
carbon monoxide and hydrogen (43).
In the process, finely crushed coal is gasified at atmospheric pressure in a
fluidized state; oxygen and steam are introduced at the base of the gasifier. The
coal is fed by lockhoppers and a screw feeder into the bottom of the fuel bed. Sin-
tered ash particles settle on a grate, where they are cooled by the incoming oxy-
gen and steam; a rotating, cooled rabble moves the ash toward a discharge port.
The ash is then conveyed pneumatically to a disposal hopper.
The gas, along with entrained ash and char particles, which are subjec-
ted to further gasification in the large space above the fluid bed, exit the
gasifier at 954– 10108C. The hot gas is passed through a waste-heat boiler to
recover the sensible heat, and then through a dry cyclone. Solid particles are
removed in both units. The gas is further cooled and cleaned by wet scrubbing,
Vol. 2 FUELS, SYNTHETIC, GASEOUS 389

Fig. 4. A Wellman-Galusha agitator-type gas producer.

and if required, an electrostatic precipitator is included in the gas-treatment


stream.
The Koppers-Totzek process is a second example of a process for the pro-
duction of a medium heat-value gas (44,45). Whereas the Winkler process
390 FUELS, SYNTHETIC, GASEOUS Vol. 2

Condensate Product gas

Green coal
Process steam

Saturated steam

Feed water

Process steam
O2 or air

Process steam
O2 or air

O2 or air
Process steam

Ash

Fig. 5. The Winkler gas producer.

employs a fluidized bed, the Koppers-Totzek process uses an entrained flow


system. In the Koppers-Totzek process, dried and pulverized coal is conveyed
continuously by a screw into a mixing nozzle. From there a high velocity stream
of steam and oxygen entrains the coal into a gasifier. The gasifier (Fig. 6) is a
cylindrical vessel with a refractory-lining that is designed to conduct a selected
amount of heat to a surrounding water jacket in which low pressure process
steam is generated. The lining is thin (about 5 cm) and made of a high alumina
cast material. In a two-headed gasifier two burner heads are placed 1808 apart
at either end of the vessel. Four burner heads, 908 apart, are used in a four-
headed gasifier. The largest gasifiers are 3–4 m diameter at the middle, tapering
to 2–3 m at the burner ends and are about 19 m long. The reactor volume is
about 30 m3 for the two-headed design, and 64 m3 for the larger, four-headed
models.
Vol. 2 FUELS, SYNTHETIC, GASEOUS 391

Gas
flow
Boiler
Low
feed
pressure
water
steam

Four-headed
gasifier

Coal
Steam
Oxygen

Boiler feed
water

Ash discharge

Fig. 6. The Koppers-Totzek gas producer.

The process is carried at moderate (slightly above atmospheric) pressures,


but at very high temperatures that reach a maximum of 19008C. Even though
the reaction time is short (0.6–0.8 s) the high temperature prevents the occur-
rence of any condensable hydrocarbons, phenols, and/or tar in the product gas.
The absence of liquid simplifies the subsequent gas clean-up steps.
Normally ca 50% of the coal ash is removed from the bottom of the gasifier
as a quenched slag. The balance is carried overhead in the gas as droplets which
are solidified when the gas is cooled with a water spray. A fluxing agent is added,
if required, to the coal to lower the ash fusion temperature and increase the mol-
ten slag viscosity.
Conversion of carbon in the coal to gas is very high. With low rank coal,
such as lignite and subbituminous coal, conversion may border on 100%, and
for highly volatile A coals, it is on the order of 90–95%. Unconverted carbon
appears mainly in the overhead material. Sulfur removal is facilitated in the pro-
cess because typically 90% of it appears in the gas as hydrogen sulfide, H2S, and
10% as carbonyl sulfide, COS; carbon disulfide, CS2, and/or methyl thiol, CH3SH,
are not usually formed.
The production of synthetic natural gas can be achieved by use of the Lurgi
gasifier (Fig. 7), which is similar in principle to the Wellman-Galusha unit and is
designed for operation at pressures up to 3.1 MPa (450 psi) (46). Three distinct
reaction zones are identifiable in a pressurized (1.9–2.9 MPa (280–425 psi))
Lurgi reactor: (1) the drying/devolatilization and pyrolysis zone, 370–5958C,
nearest the coal feed end, commences the process by converting the coal to a reac-
tive char; (2) the gasification zone (595–12058C); and (3) the combustion zone
(>1205 C), nearest the discharge end, which provides the heat requirements
for the endothermic steam–carbon reaction. Equations and reaction enthalpies
for the last two zones are given in Table 4.
392 FUELS, SYNTHETIC, GASEOUS Vol. 2

Feed coal

Recycle tar
Coal lock

Drive
Steam

Scrubbing
cooler

Distributor Gas

Grate

Grate Water jacket


drive

Steam +
oxygen

Ash lock

Fig. 7. The pressurized Lurgi gas producer.

The operating conditions in the gasifier (temperature and pressure) and the
reaction kinetics (residence time, concentration of the constituents, and rate con-
stants) determine the extent of conversion or approach to equilibrium.
The coal is fed through a lockhopper mounted on the top of the gasifier
where a rotating distributor provides uniform coal feed across the bed. When
processing caking coals, blades attached to the distributor rotate within
the bed to break up agglomerates. A revolving grate supports the bed at the
bottom and serves as a distributor for steam and oxygen. Solid residue is
removed at the bottom of the gasifier through an ash lockhopper. The entire
gasifier vessel is surrounded by a water jacket in which high pressure steam
is generated.
Vol. 2 FUELS, SYNTHETIC, GASEOUS 393

Crude gas leaves from the top of the gasifier at 288–5938C depending on the
type of coal used. The composition of gas also depends on the type of coal and
is notable for the relatively high methane content when contrasted to gases
produced at lower pressures or higher temperatures. These gas products can
be used as produced for electric power production or can be treated to remove
carbon dioxide and hydrocarbons to provide synthesis gas for ammonia, metha-
nol, and synthetic oil production. The gas is made suitable for methanation, to
produce synthetic natural gas, by a partial shift and carbon dioxide and sulfur
removal.
As in most of the high heat-value processes, the raw gas is in the medium
heat-value gas range and can be employed directly in that form. Removing the
carbon dioxide raises the heating value, but not enough to render the product
worthwhile as a high heat-value gas without methanation.
Methanation of the clean desulfurized main gas (less than 1 ppm total sul-
fur) is accomplished in the presence of a nickel catalyst at temperatures from
260–4008C and pressure range of 2–2.8 MPa (300–400 psi). Equations and reac-
tion enthalpies are given in Table 4.
Hydrogenation of the oxides of carbon to methane according to the above
reactions is sometimes referred to as the Sabatier reactions. Because of the
high exothermicity of the methanization reactions, adequate and precise cooling
is necessary in order to avoid catalyst deactivation, sintering, and carbon deposi-
tion by thermal cracking.
Catalytic methanation processes include (1) fixed or fluidized catalyst-bed
reactors where temperature rise is controlled by heat exchange or by direct cool-
ing using product gas recycle; (2) through wall-cooled reactor where temperature
is controlled by heat removal through the walls of catalyst-filled tubes; (3) tube-
wall reactors where a nickel–aluminum alloy is flame-sprayed and treated to
form a Raney-nickel catalyst bonded to the reactor tube heat-exchange surface;
and (4) slurry or liquid-phase (oil) methanation.
To enable interchangeability of the SNG with natural gas, on a calorific,
flame, and toxicity basis, the synthetically produced gas consists of a minimum
of 89 vol% methane, a maximum of 0.1% carbon monoxide, and up to 10% hydro-
gen. The specified minimum acceptable gross heating value is approximately
34.6 MJ/m3 (930 Btu/ft3).
In a combined power cycle operation, clean (sulfur- and particulate-free) gas
is burned with air in the combustor at elevated pressure. The gas is either low or
medium heat-value, depending on the method of gasification.
The hot gases from the combustor, temperature controlled to 9808C by excess
air, are expanded through the gas turbine, driving the air compressor and gener-
ating electricity. Sensible heat in the gas turbine exhaust is recovered in a waste
heat boiler by generating steam for additional electrical power production.
The use of hot gas clean-up methods to remove the sulfur and particulates
from the gasified fuel increases turbine performance by a few percentage
points over the cold clean-up systems. Hot gas clean-up permits use of the sen-
sible heat and enables retention of the carbon dioxide and water vapor in the
gasified fuel, thus enhancing turbine performance. Further, additional power
may be generated, prior to combustion with air, in an expansion turbine as the
394 FUELS, SYNTHETIC, GASEOUS Vol. 2

Gasifier
Oxygen
Steam Raw gas
Coal

Mill Water
Waste
Gasification heat
zone system

Purification
Mixer

Wastewater
Quench
zone

Coal Settler
slurry

Water
Ash lock

Slag

Fig. 8. The Texaco gasification process.

hot product fuel gas is expanded to optimum pressure level for the combined
cycle. Future advances in gas turbine technology (turbine inlet temperature
above 16508C) are expected to boost the overall combined cycle efficiency
substantially.
More recently, advanced generation gasifiers have been under develop-
ment, and commercialization of some of the systems has become a reality (36,
41). In these newer developments, the emphasis has shifted to a greater through-
put, relevant to the older gasifiers, and also to high carbon conversion levels and,
thus, higher efficiency units.
For example, the Texaco entrained system features coal–water slurry
feeding a pressurized oxygen-blown gasifier with a quench zone for slag cooling
(Fig. 8). In fact, the coal is partially oxidized to provide the heat for the gasifica-
tion reactions. The Dow gasifier also utilizes a coal–water slurry fed system
whereas the Shell gasifier features a dry-feed entrained gasification system
which operates at elevated temperature and pressure. The Kellogg Rust Wes-
tinghouse system and the Institute of Gas Technology U-Gas system are repre-
sentative of ash agglomerating fluidized-bed systems.
In response to the disadvantage that the dry ash Lurgi gasifier requires
that temperatures have to be below the ash melting point to prevent clinkering,
improvements have been sought in the unit; as a result the British Gas-Lurgi
GmbH gasifier came into being. This unit is basically similar to the dry ash
Lurgi unit insofar as the top of the unit is identical but the bottom has been mod-
ified to include a slag quench vessel (Fig. 9). Thus the ash melts at the high tem-
peratures in the combustion zone (up to 20008C) and forms a slag that runs into
Vol. 2 FUELS, SYNTHETIC, GASEOUS 395

Lock hopper

Distributor
Gas outlet

Steam/
Gasifier oxygen
(refractory inlet
lined, water
cooled)

Quench
Slag water
quench inlet
vessel

Slag
Quench
lock
water
hopper
outlet

Fig. 9. The British Gas-Lurgi slagging gasifier.

the quench chamber, which is in reality a water bath where the slag forms gran-
ules of solid ash. Temperatures and reaction rates are high in the gasification
zone so that coal residence time is markedly reduced over that of the dry
ash unit.
In summary, these second-generation gasifiers offer promise for the future
in terms of increased efficiency as well as for use of other feedstocks, such as
biomass. The older, first-generation gasifiers, however, continue to be used.
396 FUELS, SYNTHETIC, GASEOUS Vol. 2

1.4. Combustion. Coal combustion, not being in the strictest sense a


process for the generation of gaseous synfuels, is nevertheless an important
use of coal as a source of gaseous fuels. Coal combustion, an old art and probably
the oldest known use of this fossil fuel, is an accumulation of complex chemical
and physical phenomena. The complexity of coal itself and the variable process
parameters all contribute to the overall process (8,10,47–50).
There are two principal methods of coal combustion: fixed-bed combustion
and combustion in suspension. The first fixed beds, eg, open fires, fireplaces, and
domestic stoves, were simple in principle. Suspension burning of coal began in
the early 1900s with the development of pulverized coal-fired systems, and by
the 1920s these systems were in widespread use. Spreader stokers, which were
developed in the 1930s, combined both principles by providing for the smaller
particles of coal to be burned in suspension and larger particles to be burned
on a grate (10).
A significant issue in combustors in the mid-1990s is the performance of the
process in an environmentally acceptable manner through the use of either low
sulfur coal or post-combustion clean-up of the flue gases. Thus there is a marked
trend to more efficient methods of coal combustion and, in fact, a combustion
system that is able to accept coal without the necessity of a post-combustion
treatment or without emitting objectionable amounts of sulfur oxides, nitrogen
oxides, and particulates is very desirable (51,52).
The parameters of rank and moisture content are regarded as determining
factors in combustibility as it relates to both heating value and ease of reaction as
well as to the generation of pollutants (48). Thus, whereas the lower rank coals
may appear to be more reactive than higher rank coals, though exhibiting a
lower heat-value and thereby implying that rank does not affect combustibility,
environmental constraints arise through the occurrence of heteroatoms, ie, non-
carbon atoms such as nitrogen and sulfur, in the coal. At the same time, anthra-
cites, which have a low volatile matter content, are generally more difficult to
burn than bituminous coals.
Chemistry. In direct combustion coal is burned to convert the chemical
energy of the coal into thermal energy, ie, the carbon and hydrogen in the coal
are oxidized into carbon dioxide and water.

2 Hcoal þ O2 !H2 O
After burning, the sensible heat in the products of combustion can then be con-
verted into steam that can be used for external work or can be converted directly
into energy to drive a shaft, eg, in a gas turbine. In fact, the combustion process
actually represents a means of achieving the complete oxidation of coal.
The combustion of coal may be simply represented as the staged oxidation
of coal carbon to carbon dioxide

2 Ccoal þ O2 !2 CO
2 CO þ O2 !2 CO2

with any reactions of the hydrogen in the coal being considered to be of secondary
importance. Other types of combustion systems may be rate-controlled as a
result of the onset of the Boudouard reaction (see Table 4).
Vol. 2 FUELS, SYNTHETIC, GASEOUS 397

The complex nature of coal as a molecular entity (2,3,24,25,35,37,53) has


resulted in the chemical explanations of coal combustion being confined to the
carbon in the system. The hydrogen and other elements have received much
less attention but the system is extremely complex and the heteroatoms, eg,
nitrogen, oxygen, and sulfur, exert an influence on the combustion. It is this lat-
ter that influences environmental aspects.
For example, the conversion of nitrogen and sulfur, during coal combustion,
to the respective oxides during combustion cannot be ignored:

Scoal þ O2 !SO2
2 SO2 þ O2 !2 SO3
2 Ncoal þ O2 !2 NO
2 NO þ O2 !2 NO2
Ncoal þ O2 !NO2

The sulfur and nitrogen oxides that escape into the atmosphere can be converted
to acids by reaction with moisture in the atmosphere (see also AIR POLLUTION;
ATMOSPHERIC MODELS)

Sulfurous acid SO2 þ H2 O!H2 SO3

Sulfuric acid 2 SO2 þ O2 !2 SO3


SO3 þ H2 O!H2 SO4

Nitrous acid NO þ H2 O!H2 NO2

Nitric acid 2 NO þ O2 !2 NO2


NO2 þ H2 O!HNO3

Combustion Systems. Combustion systems vary in nature depending on


the nature of the feedstock and the air needed for the combustion process (54).
However, the two principal types of coal-burning systems are usually referred to
as layer and chambered. The former refers to fixed beds; the latter is more spe-
cifically for pulverized fuel.
Fixed or Slowly Moving Beds. For fuel-bed burning on a grate, a distil-
lation effect occurs. The result is that liquid components which are formed vola-
tilize before combustion temperatures are reached; cracking may also occur. The
ignition of coal in a bed is almost entirely by radiation from hot refractory arches
and from the flame burning of volatiles. In fixed beds, the radiant heat above the
bed can only penetrate a short distance into the bed.
Consequently, convective heat transfer determines the intensity of warm-
ing up and ignition. In addition, convective heat transfer also plays an important
part in the overall flame-to-surface transmission. The reaction of gases is greatly
398 FUELS, SYNTHETIC, GASEOUS Vol. 2

accelerated by contact with hot surfaces and, whereas the reaction away
from the walls may proceed slowly, reaction at the surface proceeds much
more rapidly.
Fluidized Beds. Fluidized-bed combustion occurs in expanded beds
(Fig. 3) and at relatively lower (9258C) temperatures; high convective transfer
rates exist resulting from the bed motion. Fluidized systems can operate under
substantial pressures thereby allowing more efficient gas clean-up. Fluidized-bed
combustion is a means for providing high heat-transfer rates, controlling sulfur,
and reducing nitrogen oxide emissions from the low temperatures in the combus-
tion zone.
There are, however, problems associated with pollution control. Whereas
the sulfur may be removed downstream using suitable ancillary controls, the sul-
fur may also be captured in the bed, thereby adding to the separations and
recycle problems. Capture during combustion, however, is recognized as the
ideal and is a source of optimism for fluidized combustion.
A fluidized bed is an excellent medium for contacting gases with solids, and
this can be exploited in a combustor because sulfur dioxide emissions can be
reduced by adding limestone, CaCO3, or dolomite, CaCO3  MgCO3, to the bed.

2 SO2 þ O2 !2 SO3


SO3 þ CaCO3 !CaSO4 þ CO2

or
2 SO2 þ O2 þ 2 CaCO3 !2 CaSO4 þ 2 CO2

The spent sorbent from fluidized-bed combustion may be taken directly to dispo-
sal and is much easier than the disposal of salts produced by wet limestone
scrubbing. Alternatively, the spent sorbent may be regenerated using synthesis
gas, CO/H2.
CaSO4 þ H2 !CaO þ H2 O þ SO2
CaSO4 þ CO!CaO þ CO2 þ SO2

The calcium oxide product is supplemented with fresh limestone and returned to
the fluidized bed. Two undesirable side reactions can occur in the regeneration of
spent lime leading to the production of calcium sulfide:

CaSO4 þ 4 H2 !CaS þ 4 H2 O
CaSO4 þ 4 CO!CaS þ 4 CO2

which results in the recirculation of sulfur to the bed.


Entrained Systems. In entrained systems, fine grinding and increased
retention times intensify combustion but the temperature of the carrier and
degree of dispersion are also important. In practice, the coal is introduced at
high velocities which may be greater than 30 m/s and involve expansion from
a jet to the combustion chamber.
Types of entrained systems include cyclone furnaces, which have been
used for various coals. Other systems have been developed and utilized for
Vol. 2 FUELS, SYNTHETIC, GASEOUS 399

the injection of coal–oil slurries into blast furnaces or for the burning of
coal–water slurries. The cyclone furnace, developed in the 1940s to burn coal
having low ash-fusion temperatures, is a horizontally inclined, water-cooled, tub-
ular furnace in which crushed coal is burned with air entering the furnace tan-
gentially. Temperatures may be on the order of 17008C and the ash in the coal is
converted to a molten slag that is removed from the base of the unit. Coal fines
burn in suspension; the larger pieces are captured by the molten slag and burn
rapidly.

2. Gas from Other Fossil Fuels

As of this writing natural gas is a plentiful resource, and there has been a
marked tendency not to use other fossil fuels as SNG sources. However, pet-
roleum and oil shale (qv) have been the subject of extensive research efforts.
These represent other sources of gaseous fuels and are worthy of mention
here.
2.1. Petroleum. Thermal cracking (pyrolysis) of petroleum or fractions
thereof was an important method for producing gas in the years following its
use for increasing the heat content of water gas. Many water gas sets operations
were converted into oil-gasification units (55). Some of these have been used for
base-load city gas supply, but most find use for peak-load situations in the
winter.
In the 1940s, the hydrogasification of oil was investigated as a follow-up to
the work on the hydrogasification of coal (56,57). In the ensuing years, further
work was carried out as a supplement to the work on thermal cracking (58,59),
and during the early 1960s it became evident that light distillates having end
boiling points < 182 C and containing no sulfur could be catalytically reformed
by an autothermic process to pure methane (60). This method was extensively
used in England until natural (North Sea) gas came into use.
Prior to the discovery of plentiful supplies of natural gas, and depending on
the definition of the resources (1), there were plans to accommodate any short-
falls in gas supply from solid fossil fuels and from gaseous resources by the con-
version of hydrocarbon (petroleum) liquids to lower molecular weight gaseous
products.

CHpetroleum !CH4

Thermal Cracking. In addition to the gases obtained by distillation of


crude petroleum, further highly volatile products result from the subsequent pro-
cessing of naphtha and middle distillate to produce gasoline, as well as from
hydrodesulfurization processes involving treatment of naphthas, distillates,
and residual fuels (5,61), and from the coking or similar thermal treatment of
vacuum gas oils and residual fuel oils (5).
The chemistry of the oil-to-gas conversion has been established for several
decades and can be described in general terms although the primary and second-
ary reactions can be truly complex (5). The composition of the gases produced
from a wide variety of feedstocks depends not only on the severity of cracking
400 FUELS, SYNTHETIC, GASEOUS Vol. 2

but often to an equal or lesser extent on the feedstock type (5,62,63). In general
terms, gas heating values are on the order of 30–50 MJ/m3 (950–1350 Btu/ft3).
Catalytic Processes. A second group of refining operations which contri-
bute to gas production are the catalytic cracking processes, such as fluid-bed cat-
alytic cracking, and other variants, in which heavy gas oils are converted into
gas, naphthas, fuel oil, and coke (5).
The catalysts promote steam reforming reactions that lead to a product gas
containing more hydrogen and carbon monoxide and fewer unsaturated hydro-
carbon products than the gas product from a noncatalytic process (5). Cracking
severities are higher than those from thermal cracking, and the resulting gas is
more suitable for use as a medium heat-value gas than the rich gas produced by
straight thermal cracking. The catalyst also influences the reactions rates in the
thermal cracking reactions, which can lead to higher gas yields and lower tar and
carbon yields (5).
The basic chemical premise involved in making synthetic natural gas from
heavier feedstocks is the addition of hydrogen to the oil:

CH3 ðCH2 Þn CH3 þ ðn þ 1ÞH2 !ðn þ 2ÞCH4

In general terms, as the molecular weight of the feedstock is increased, similar


operating conditions of hydrogasification lead to decreasing hydrocarbon gas
yields, increasing yields of aromatic liquids, with carbon also appearing as a
product.
The principal secondary variable that influences yields of gaseous products
from petroleum feedstocks of various types is the aromatic content of the feed-
stock. For example, a feedstock of a given H/C (C/H) ratio that contains a large
proportion of aromatic species is more likely to produce a larger proportion of
liquid products and elemental carbon than a feedstock that is predominantly par-
affinic (5).
Another option for processing crude oils which are too heavy to be hydroga-
sified directly involves first hydrocracking the crude oil to yield a low boiling pro-
duct suitable for gas production and a high boiling product suitable for hydrogen
production by partial oxidation (57). Alternatively, it may be acceptable for car-
bon deposition to occur during hydrogasification which can then be used for heat
or for hydrogen production (64,65).
Partial Oxidation. It is often desirable to augment the supply of naturally
occurring or by-product gaseous fuels or to produce gaseous fuels of well-defined
composition and combustion characteristics (5). This is particularly true in areas
where the refinery fuel (natural gas) is in poor supply and/or where the manu-
facture of fuel gases, originally from coal and more recently from petroleum, has
become well established.
Almost all petroleum fractions can be converted into gaseous fuels, although
conversion processes for the heavier fractions require more elaborate technology
to achieve the necessary purity and uniformity of the manufactured gas stream
(5). In addition, the thermal yield from the gasification of heavier feedstocks is
invariably lower than that of gasifying light naphtha or liquefied petroleum
gas(es) because, in addition to the production of hydrogen, carbon monoxide,
and gaseous hydrocarbons, heavy feedstocks also yield some tar and coke.
Vol. 2 FUELS, SYNTHETIC, GASEOUS 401

As in the case of coal, synthetic natural gas can be produced from heavy oil
by partially oxidizing the oil to a mixture of carbon monoxide and hydrogen

2 CHpetroleum þ O2 !2 CO þ H2

which is methanated catalytically to produce methane of any required purity.


The initial partial oxidation step consists of the reaction of the feedstock with
a quantity of oxygen insufficient to burn it completely, making a mixture consist-
ing of carbon monoxide, carbon dioxide, hydrogen, and steam. Success in par-
tially oxidizing heavy feedstocks depends mainly on details of the burner
design (66). The ratio of hydrogen to carbon monoxide in the product gas is a
function of reaction temperature and stoichiometry and can be adjusted, if
desired, by varying the ratio of carrier steam to oil fed to the unit.
To make synthetic natural gas by partial oxidation, virtually all of the
methane in the product gas must be produced by catalytic methanation of carbon
monoxide and hydrogen. The feed is mixed with recycled carbon and fed,
together with steam and oxygen, to a reactor in which partial combustion
takes place. Heat from the reaction gasifies the rest of the feed, and by-product
coke is formed. The heavier feedstocks tend to produce more carbon than can be
consumed through recycling; thus some must be withdrawn. After carbon
separation by water scrubbing, the synthesis gas that is available can be con-
verted into hydrogen or into synthetic natural gas by methanation.
Steam Reforming. When relatively light feedstocks, eg, naphthas having
ca 1808C end boiling point and limited aromatic content, are available, high
nickel content catalysts can be used to simultaneously conduct a variety of
near-autothermic reactions. This results in the essentially complete conversions
of the feedstocks to methane:

CH3 ðCH2 Þ3 CH3 þ 2 H2 O!4 CH4 þ CO2

Because of limitations on the activity of practical catalysts, this reaction


must be carried out in stages, the first of which is carried out at 425–4808C and
1.5–2.9 MPa (200–400 psi) and amounts approximately to the following reaction:

C5 H12 þ 3 H2 O!3 CH4 þ CO2 þ 3 H2 þ CO


In ensuing catalytic stages, usually termed hydrogasification and methanation
(not to be confused with the noncatalytic, direct hydrogasification processes
described above), the remaining carbon monoxide and hydrogen are reacted to
produce additional methane.
2.2. Oil Shale. Oil shale is a sedimentary rock that contains organic
matter, referred to as kerogen, and another natural resource of some conse-
quence that could be exploited as a source of synthetic natural gas (67–69). How-
ever, as of this writing, oil shale has found little use as a source of substitute
natural gas.
2.3. Biomass. Biomass is simply defined for these purposes as any
organic waste material, such as agricultural residues, animal manure, forestry
residues, municipal waste, and sewage, which originated from a living organism
(70–74).
402 FUELS, SYNTHETIC, GASEOUS Vol. 2

Biomass is another material that can produce a mixture of carbonaceous


solid and liquid products as well as gas:

Corganic !Ccoke=char=carbon þ liquids þ gases

Whereas biomass has not received the same attention as coal as a source of gas-
eous fuels, questions about the security of fossil energy supplies related to the
availability of natural and substitute gas have led to a search for more reliable
and less expensive energy sources (75). Biomass resources are variable, but it has
been estimated that substantial amounts (up to 20  106 mJð20  1015 BtuÞ) of
energy, representing ca 19% of the annual energy consumption in the United
States. In addition, environmental issues associated with the use of coal have
led some energy producers to question the use of large central energy generating
plants. However, biomass may be a gaseous fuel source whose time is approach-
ing (see FUELS FROM BIOMASS).
The means by which synthetic gaseous fuels could be produced from a vari-
ety of biomass sources are variable and many of the known gasification technol-
ogies can be applied to the problem (70,71,76–82). For example, the Lurgi
circulatory fluidized-bed gasifier is available for the production of gaseous pro-
ducts from biomass feedstocks as well as from coal (83,84).

3. Gas Treating

The reducing conditions in gasification reactors effect the conversion of the sul-
fur and nitrogen in the feed coal to hydrogen sulfide, H2S, and ammonia, NH3.
Some carbonyl sulfide, COS, carbon disulfide, CS2, mercaptans, RSH, and hydro-
gen cyanide, HCN, are also formed in the gasifier. These compounds, along with
carbon dioxide, are removed simultaneously, either selectively or nonselectively,
from the gas stream in the clean-up stages of the process using commercially
available physical or chemical solvents and scrubbing agents (1,5,85–88).
Solvents used for hydrogen sulfide absorption include aqueous solutions of
ethanolamine (monoethanolamine, MEA), diethanolamine (DEA), and diisopro-
panolamine (DIPA) among others:

2 RNH2 þ H2 S!ðRNH3 Þ2 S
ðRNH3 Þ2 S þ H2 S!2 RNH3 HS
2 RNH2 þ CO2 þ H2 O!ðRNH3 Þ2 CO3
ðRNH3 Þ2 CO3 þ CO2 þ H2 O!2 RNH3 HCO3
2 RNH2 þ CO2 !RNHCOONH3 R

These solvents differ in volatility and selectivity for the removal of H2S,
mercaptans, and CO2 from gases of different composition. Other alkaline sol-
vents used for the absorption of acidic components in gases include potassium
carbonate, K2CO3, solutions combined with a variety of activators and solubili-
zers to improve gas–liquid contacting.
Vol. 2 FUELS, SYNTHETIC, GASEOUS 403

Whereas most alkaline solvent absorption (qv) processes result in gases of


acceptable purity for most purposes, it is often essential to remove the last traces
of residual sulfur compounds from gas streams. This is in addition to ensuring pro-
duct purity such as the removal of water, higher hydrocarbons, and dissolved ele-
mental sulfur from liquefied petroleum gas. Removal can be accomplished by
passing the gas over a bed of molecular sieves (qv), synthetic zeolites commercially
available in several proprietary forms. Impurities are retained by the packed bed,
and when the latter is saturated it can be regenerated by passing hot clean gas or hot
nitrogen, generally in a reverse direction (see also CATALYSTS, REGENERATION).
By-product water formed in the methanation reactions is condensed by
either refrigeration or compression and cooling. The remaining product gas, prin-
cipally methane, is compressed to desired pipeline pressures of 3.4–6.9 MPa
(500–1000 psi). Final traces of water are absorbed on silica gel or molecular
sieves, or removed by a drying agent such as sulfuric acid, H2SO4. Other desic-
cants may be used, such as activated alumina, diethylene glycol, or concentrated
solutions of calcium chloride (see DESICCANTS).

BIBLIOGRAPHY

‘‘Manufactured Gas’’ in ECT 1st ed., Vol. 8, pp. 765–800, by W. H. Fulweiler, Consultant;
‘‘Gas, Manufactured’’ in ECT 2nd ed., Vol. 10, pp. 353–442, by M. A. Elliot, Illinois Insti-
tute of Technology, and H. R. Linden, Institute of Gas Technology; ‘‘Carbon Monoxide–
Hydrogen Reactions’’ in ECT 2nd ed., Vol. 4, pp. 446–489, by H. Pichler and A. Hector,
Carl-Engler-und Hans-Bunte-Institut für Mineralöl und Kohleforschung der Technischen
Hochschule Karlsruhe; ‘‘Gaseous’’ under ‘‘Fuels, Synthetic’’ in ECT 3rd ed., Vol. 11,
pp. 410–446, by J. Huebler and J. C. Janka, Institute of Gas Technology.

CITED PUBLICATIONS

1. J. G. Speight, ed., Fuel Science and Technology Handbook, Marcel Dekker, Inc.,
New York, 1990.
2. J. G. Speight, The Chemistry and Technology of Coal, Marcel Dekker, Inc.,
New York, 1983.
3. R. A. Hessley, in Ref. 1, 645–734.
4. E. J. Parente and A. Thumann, eds., The Emerging Synthetic Fuel Industry,
Fairmont Press, Atlanta, Ga., 1981.
5. J. G. Speight, The Chemistry and Technology of Petroleum, 2nd ed., Marcel Dekker,
Inc., New York, 1991.
6. A. Kasem, Three Clean Fuels from Coal: Technology and Economics, Marcel Dekker,
Inc., New York, 1979.
7. L. L. Anderson and D. A. Tillman, Synthetic Fuels from Coal, John Wiley & Sons,
Inc., New York, 1979.
8. A. D. Dainton, in G. J. Pitt and G. R. Millward, eds., Coal and Modern Coal Proces-
sing: An Introduction, Academic Press, Inc., New York, 1979.
9. D. M. Considine, Energy Technology Handbook, McGraw-Hill Book Co., Inc.,
New York, 1977.
10. A. Francis and M. C. Peters, Fuels and Fuel Technology, Pergamon Press, Inc.,
New York, 1980.
404 FUELS, SYNTHETIC, GASEOUS Vol. 2

11. J. L. Johnson, in M. A. Elliott, ed., Chemistry of Coal Utilization, 2nd suppl. vol.
John Wiley & Sons, Inc., New York, 1981, Chapt. 23.
12. A. Elton, in C. Singer and co-workers, eds., A History of Technology, Vol. 4, Oxford
University Press, Oxford, U.K., 1958, Chapt. 9.
13. L. Shnidman, in H. H. Lowry, ed., Chemistry of Coal Utilization, John Wiley & Sons,
Inc., New York, 1945, Chapt. 30.
14. C. M. Jarvis, in Ref. 12, Vol. 5, Chapt. 10.
15. P. F. Mako and W. A. Samuel, in R. A. Meyers, ed., Handbook of Synfuels Technol-
ogy, McGraw-Hill Book Co., Inc., New York, 1984, Chapt. "2-1".
16. Oil Gas J. 90(3), 53 (1992).
17. K. T. Klasson and co-workers, C. Akin and J. Smith, eds., Gas, Oil, Coal, and
Environmental Biotechnology II, Institute of Gas Technology, Chicago, 1990, p. 408.
18. B. D. Faison, Crit. Revs. Biotechnol. 11, 347 (1991).
19. S. R. Bull, Energy Sources 13, 433 (1991).
20. R. C. Reid, J. M. Prausnitz, and T. K. Sherwood, The Properties of Gases and
Liquids, McGraw-Hill Book Co., Inc., New York, 1977.
21. Wellman Galusha Gas Producers, research bulletin no. 576A, McDowell-Wellman
Engineering Co., Cleveland, Ohio, May 1976.
22. R. F. Probstein and R. E. Hicks, Synthetic Fuels, pH Press, Cambridge, Mass.,
1990.
23. J. S. Sinninghe Damste and J. W. de Leeuw, Fuel Processing Technol. 30, 109 (1992).
24. N. Berkowitz, Introduction to Coal Technology, Academic Press, Inc., New York,
1979.
25. R. K. Hessley, J. W. Reasoner, and J. T. Riley, Coal Science, John Wiley & Sons, Inc.,
New York, 1986.
26. M. O. Holowaty and co-workers, in R. A. Meyers, ed., Coal Handbook, Marcel
Dekker, Inc., New York, 1981, Chapt. 9.
27. W. Eisenhut, in Ref. 11, Chapt. 14.
28. F. W. Gibbs, in Ref. 12, Vol. 3, Chapt. 25.
29. G. J. Pitt and G. R. Millward, eds., in Ref. 8, p. 52.
30. L. Seglin and S. A. Bresler, in Ref. 11, Chapt. 13.
31. E. Aristoff, R. W. Rieve, and H. Shalit, in Ref. 11, Chapt. 16.
32. D. McNeil, in Ref. 11, Chapt. 17.
33. C. Y. Cha and co-workers, Report No. DOE/MC/24268-2700 (DE89000967), United
States Department of Energy, Washington, D.C., 1988.
34. F. M. Fess, History of Coke Oven Technology, Gluckauf Verlag, Essen, Germany,
1957.
35. M. A. Elliott, ed., in Ref. 11.
36. D. Hebden and H. J. F. Stroud, in Ref. 11, Chapt. 24.
37. R. A. Meyers, ed., in Ref. 15.
38. D. R. Simbeck, R. L. Dickenson, and A. J. Moll, Energy Prog. 2, 42 (1982).
39. W. W. Bodle and J. Huebler, in Ref. 26, Chapt. 10.
40. L. K. Rath and J. R. Longanbach, Energy Sources 13, 443 (1991).
41. S. Alpert and M. J. Gluckman, Ann. Rev. Energy 11, 315 (1986).
42. R. A. Meyers, ed., in Ref. 26.
43. J. H. Martin, I. N. Banchik, and T. K. Suhramaniam, Report of the Committee on
Production of Manufactured Gases, report no. 1GU/B-76, London, 1976.
44. J. F. Farnsworth, Ind. Heat. 41(11), 38 (1974).
45. J. F. Farnsworth, Proceedings Coal Gas Fundamentals Symposium, Institute of Gas
Technology, Chicago, 1979.
46. J. C. Hoogendoorn, Proceedings of the Ninth Pipeline Gas Symposium, Chicago,
Oct. 31–Nov. 2, 1977.
Vol. 2 FUELS, SYNTHETIC, GASEOUS 405

47. R. Essenhigh, in Ref. 11, Chapt. 19.


48. M. A. Field and co-workers, Combustion of Pulverized Coal, British Coal Utilization
Research Association, Leatherhead, Surrey, U.K., 1967.
49. A. Levy and co-workers, in Ref. 26, Chapt. 8.
50. N. Chigier, Combustion Measurements, Hemisphere Publishing Corp., New York,
1991.
51. United States Congress, Public Law 101-549, An Act to Amend the Clean Air Act to
Provide for Attainment and Maintenance of Health Protective National Ambient Air
Quality Standards, and for Other Purposes, Nov. 15, 1990.
52. United States Department of Energy, Clean Coal Technology Demonstration Pro-
gram, DOE/FE-0219P, U.S. Dept. of Energy, Washington, D.C., Feb. 1991.
53. J. E. Funk, in J. A. Kent, ed., Riegel’s Handbook of Industrial Chemistry, Van Nos-
trand Reinhold Co., New York, 1983, Chapt. 3.
54. F. J. Ceely and E. L. Daman, in Ref. 26, Chapt. 20.
55. J. M. Reid, Proceedings SNG Symposium 1, Institute of Gas Technology, Chicago,
May 12–16, 1973.
56. F. J. Dent, Gas J. 288(12), 600, 606, 610 (1956).
57. F. J. Dent, Gas World, 144(11), 1078, 1080 (1956).
58. H. R. Linden and E. S. Pettyjohn, Research Bulletin No. 12, Institute of Gas Tech-
nology, Chicago, 1952.
59. G. B. Schultz and H. R. Linden, Research Bulletin No. 29, Institute of Gas Technol-
ogy, Chicago, 1960.
60. F. J. Dent, Proceedings of the Ninth International Gas Conference, The Hague
(Scheveningen), the Netherlands, Sept. 1–4, 1964.
61. J. G. Speight, The Desulfurization of Heavy Oils and Residua, Marcel Dekker, Inc.,
New York, 1981.
62. B. B. Bennett, J. Inst. Fuel 35(8), 338 (1962).
63. H. R. Linden and M. A. Elliot, Am. Gas J. 186(2), 22 (1959).
64. Oil Gas J. 71(7), 36, 37 (1973).
65. Oil Gas J. 71(15), 32, 33 (1973).
66. C. J. Kuhre and C. J. Shearer, Oil Gas J. 71(36), 85 (1971).
67. P. Nowacki, Oil Shale Technical Data Handbook, Noyes Data Corp., Park Ridge,
N.J., 1981.
68. V. D. Allred, ed., Oil Shale Processing Technology, Center for Professional Advance-
ment, East Brunswick, N.J., 1982.
69. C. S. Scouten, in Ref. 1.
70. J. S. Robinson, Fuels from Biomass: Technology and Feasibility, Noyes Data Corp.,
Park Ridge, N.J., 1980.
71. J. L. Jones and S. B. Radding, eds., Thermal Conversion of Solid Wastes and
Biomass, Symposium Series No. 130, American Chemical Society, Washington,
D.C., 1980.
72. M. P. Kannan and G. N. Richard, Fuel 69, 747 (1990).
73. L. Jimenez, J. L. Bonilla, and J. L. Ferrer, Fuel 70, 223 (1991).
74. L. Jimenez and F. Gonzalez, Fuel 70, 947 (1991).
75. Energy World 145(3), 11 (1987).
76. M. P. Sharma and B. Prasad, Energy Management (New Delhi) 10(4), 297 (1986).
77. A. A. C. M. Beenackers and W. P. M. van Swaaij, Thermochemical Processing of
Biomass, Butterworths, London, 1984, p. 91.
78. G. J. Esplin, D. P. C. Fung, and C. C. Hsu, Can. J. Chem. Eng. 64, 651 (1986).
79. S. Gaur and co-workers, Fuel Sci. Technol. Int. 10, 1461 (1992).
80. M. A. McMahon and co-workers, Preprints, Div. Fuel Chem. 36(4), 1670 (1991).
81. K. Dura-Swamy and co-workers, Preprints, Div. Fuel Chem. 36(4), 1677 (1991).
406 OCTANE ENHANCERS Vol. 2

82. J. T. Hamrick, Preprints, Div. Fuel Chem. 36(4), 1986 (1991).


83. R. Reimert and co-workers, Bioenergy 84, Vol. 3, Elsevier Applied Science Publish-
ers, London, p. 102.
84. P. K. Herbert and J. C. Loeffler, Proceedings, Opportunities in the Synfuels Industry:
SynOps ’88, Energy and Environmental Research Center, University of North
Dakota, Grand Forks, 1988, p. 141.
85. H. A. Grosick and J. E. Kovacic, in Ref. 11, Chapt. 18.
86. A. V. Slack, in Ref. 11, Chapt. 22.
87. H. E. Benson, in Ref. 11, Chapt. 25.
88. C. H. Taylor in Ref. 15, Chapt. 3–13.

JAMES G. SPEIGHT
Consultant

OCTANE ENHANCERS
1. Introduction

Historically, the quality of motor gasoline and the technical development of the
spark ignition Otto engine were closely connected. Since its invention in 1876,
the Otto engine technology has been continuously improved, a process which is
still going on.
Of particular importance for this development was the parallel improve-
ment in gasoline knock resistance. Knock resistance is one of the most important
gasoline properties and its improvement allowed the engine manufacturers to
increase the compression ratio. This in turn resulted in higher engine efficiency
and improved exhaust emission performance. The knock resistance is measured
in terms of gasoline octane numbers in accordance with worldwide standardized
test procedures.
Figure 1 (see next page) shows, for Germany as an example, the close rela-
tionship between octane number and engine compression ratio from the turn of
the century until now. Over this period a variety of measures were applied to
improve refinery technologies to meet the increasing engine octane demand.
These measures incorporated the integration of complex processing plants,
from which a range of gasoline component streams with different octane levels
became available. In 1929, alkyllead compounds were introduced in the United
States as efficient octane enhancers. Thereafter these were used generally to
improve the gasoline octane quality until environmental concerns led to their
use being stepwise restricted worldwide. The resulting shortfall of octane was
compensated by improvements in refinery processes, as well as by the use of
alternative octane enhancers such as fuel oxygenates. Starting in the 1970s a
number of fuel oxygenates became available and were increasingly used for
gasoline blending.

ß2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim


10.1002/14356007.a18_037
Vol. 2 OCTANE ENHANCERS 407

Fig. 1. Trend of research octane number and compression ratio in Germany.

More recently, efforts to reduce air pollution led to the enforcement of more
restrictive exhaust emission limits for vehicles. In this context, passenger cars
with catalytic converters were introduced. Since the addition of lead to gasoline
results in the deactivation of the catalytic converters, unleaded gasoline of high
octane quality was consequently made available.
1

2. Technical Information

2.1. Combustion in Otto Engines. During the compression stroke the


air – fuel mixture is compressed to 1.5 – 2.5 MPa and the temperature rises up to
1000 K, depending on the engine compression ratio. The mixture composition
and the engine condition are controlled in such a way that the upper flammabil-
ity limit of the mixture is not exceeded i.e., no self-ignition occurs. Shortly before
the piston reaches the upper dead point, combustion is initiated by an electrical
discharge between the electrodes of the spark plug. The appropriate spark tim-
ing is essential for engine efficiency. Best efficiency is achieved when maximum
chemical energy released from the combusted charge is transformed into
mechanical energy during the expansion stroke. The spark transports the
necessary ignition energy to form a self-sustained flame kernel in the air –
fuel mixture. Subsequently this flame kernel expands and flame fronts propa-
gate spherically into the combustion chamber. The speed of flame propagation
depends on a variety of parameters.
First of all the process is affected by the initial status of the air – fuel mix-
ture with respect to temperature, pressure, and turbulence. Since the combus-
tion process is exothermic, temperature and pressure increase rapidly. This
accelerates the flame propagation, which in turn affects the status and turbu-
lence of the burned gas as well as the unburned charge ahead of the flame.
The entire combustion consists of a complex interaction of fluid dynamics, reac-
tion kinetics, radical diffusion, and radiation processes.
2.2. Knock Phenomena. Average flame speeds in the engine combus-
tion process range from 10 to 30 m/s. In contrast, flame speeds of 300 to 500 m/s
are measured under knocking conditions. The knock phenomena typically occur
at the end of the normal combustion process (Fig. 2). They originate from zones
408 OCTANE ENHANCERS Vol. 2

Fig. 2. Pressure profile versus crank angle during combustion. A) Normal combustion;
B) Knocking combustion.

of unreacted charge where the critical conditions in terms of temperature and


pressure are exceeded. This causes spontaneous self-ignition which in turn results
in a sharp pressure increase.
The detailed processes occur very rapidly and involve the complex coupling
of fluid dynamic and reaction kinetic processes. Despite a large number of experi-
mental and theoretical studies the phenomenon itself is not fully understood (1).
The phenomenological process, however, is readily detectable. It is charac-
terized by a typical high-frequency noise emitted from engine parts. Knocking
combustion results in high mechanical stress and thermal overload of the engine.
Finally, it leads to severe damage to pistons, piston rings, seals, and/or cylinder
heads.
2.3. Octane Number Measurement. The knock resistance of gasolines
is measured by their research octane number (RON) and motor octane number
(MON). Both are determined in single cylinder engines according to standar-
dized test procedures. The design of the engines, their mode of operation, and
the calibration and measuring methods are complex; detailed descriptions are
given in ASTM D 2699 (RON) and ASTM D 2700 (MON) and in equivalent ISO
and EN standards (! Automotive Fuels Chap. 3.1.1). Measurement of MON is
performed under higher engine severity than the RON measurement in terms
of intake air, temperature, and engine speed. The MON describes high speed
and high load knock behavior in productional engines whereas the RON is a
measure for knock performance under acceleration conditions.

3. History

Until the mid-1920s there was only one gasoline grade available in most coun-
tries, predominantly produced from straight-run crude oil products which were
selected according to their composition of aliphatic and aromatic components. In
order to meet the increasing octane demand, eg, indicated by the introduction of
a premium grade in 1924, several refinery processes were developed. Today, a
Vol. 2 OCTANE ENHANCERS 409

typical refinery produces up to twenty individual product streams used for gaso-
line blending (! Automotive Fuels, Chap. 4.1). The octane enhancement of tetra-
ethyllead (TEL) was discovered in 1922 in the United States by T. MIDGLEY, JR.
(2). It was produced by the Ethyl Gasoline Corporation and first introduced onto
the U.S. market in 1929. In Germany it became available shortly before World
War II as an octane enhancer. TEL and TML (tetramethyllead) are so-called
‘‘lead fluids’’ and are still used in some parts of the world. However, leadalkyl
compounds have disappeared from gasoline in most countries due to environ-
mental concerns and because lead deactivates catalytic converters.
Other octane enhancers–except fuel oxygenates–never achieved any signif-
icant degree of application. The current dominance of unleaded gasolines in the
market, however, causes octane shortfalls. This situation has led to renewed
attention to additional octane sources, including both new and known oxyge-
nates and alternative octane enhancers. Therefore, octane enhancers may play
a more pronounced role in the future provided that environmental aspects are
satisfied, such as toxicological properties and full compatibility with modern
exhaust emission control systems.

4. Types of Octane Enhancers

Octane enhancers are substances which result in octane improvements when


added to gasolines. This definition also includes all hydrocarbon-based refinery
streams of sufficient octane quality. These streams, however, must be regarded
as conventional gasoline blendstocks and are not covered in this article, in spite
of the fact that some are traded as octane enhancers eg, aromatic refinery frac-
tions such as toluene.
Alternative ways to improve octane quality involve the use of special chemi-
cal compounds. Since these compounds have to be readily soluble in gasolines
they must be hydrocarbon-based. They may, however, contain functional groups
with heteroatoms such as oxygen, nitrogen, sulfur, phosphorus, halogens, or
metal atoms. Depending on the compound concentration in the fuel, components
and additives can be distinguished. Components are compounds blended in quan-
tities which also affect gasoline properties other than octane number, such as
heating value, density, or vapor pressure. In contrast, additives are applied in
the ppm-concentration range so that other fuel properties are essentially unaf-
fected. This discrimination reflects the fact that additives are much more efficient
than components. Additives consist exclusively of organometallic compounds.
When combusted they form solid oxidation products (ash) in addition to gaseous
combustion products.

5. Ashless Octane Enhancers

Components are always added in higher quantities, and must, therefore,


consist of ‘‘ashless’’ compounds. Only oxygen- or nitrogen-containing functional
groups are suitable for octane enhancement. Sulfur-, phosphorus-, or halogen-
410 OCTANE ENHANCERS Vol. 2

containing compounds either promote knocking or achieve only insignificant


octane improvements.
5.1. Blend Octane Numbers. When a component is blended to a base
gasoline, the resulting octane number of the mixture cannot be calculated from
the octane numbers of the neat gasoline and the component. Furthermore, the
octane number of the mixture is not proportional to the component concentra-
tion. In order to characterize the octane efficiency of a component, therefore,
‘‘blend octane numbers’’ (B–ON) are defined according to the following equation:

ONM  ONBG ð1  xÞ
B  ON ¼
x
where B–ON is the blend octane number of the component; ONM the octane
number of the mixture; ONBG the octane number of the base gasoline; and x is
the volume fraction of the component in the blend. All parameters on the right-
handside of this equation must be determined experimentally. This equation
attributes the entire change in octane number to the component. Graphically
(Fig. 3), it reflects a linear extrapolation from the base gasoline ONBG, via the
mixture ONM, to a fictive B–ON value of the neat component. If the calculated B–
ON is compared with the measured octane number of the neat component (ONC),
two alternatives are possible. If the B–ON is larger than the measured ONC, the
mixture has a ‘‘synergistic’’ property, if the B–ON is smaller than the ONC, the
mixture shows an ‘‘antagonistic’’ behavior.

Fig. 3. Graphic interpretation of blend formula.


Vol. 2 OCTANE ENHANCERS 411

Fig. 4. Example of MON blend behavior of straight-run gasoline – MTBE mixtures.

An example of the nonlinear MON blend behavior and the determination of


the B-MON value is shown in Figure 4 for straight-run gasoline – MTBE mix-
tures. This example shows that the extrapolated B-MON value based on
x ¼ 0.2 is higher than the experimentally determined MON value of neat
MTBE, indicating a synergistic effect of MTBE.
In general, B-ON values of octane enhancers and refinery components
depend on the component concentration, as well as the octane level and the
composition of the base gasoline. These dependencies on the base gasoline are
relatively small for refinery components. For some octane enhancers, however,
in particular oxygenates such as alcohols, the B-ON values vary significantly
with gasoline composition. Therefore, in refinery blending, different B-ON data
are used for the same oxygenate when blended to different gasoline grades. Nor-
mally, B-ON data derived from paraffinic gasolines are higher than those from
gasolines with high olefinic and/or aromatic contents. Most components achieve
higher B-RON than B-MON values. Both depend on the component concentration
as can be seen in Figure 4 for B-MON. In practice, this effect can be neglected
because the concentration range used is normally small.
Some other properties such as density, volatility, and vapor pressure also
exhibit nonlinear blend behavior in mixtures. Blend properties are determined
analogously to B-ON and play an important role in the evaluation of the eco-
nomic potential of a component.
5.2. Nitrogen-Containing Components. The influence of nitrogen-
containing compounds on octane performance has been studied extensively (3).
Depending on the functional group and the molecular structure, they provide
either proknock (ie, they promote knocking) or antiknock properties.
412 OCTANE ENHANCERS Vol. 2

Table 1. Blend-octane Numbers of Nitrogen-containing Components


(Values Determined from 2 wt % Addition to Premium Base Gasoline)
Component B-RON B-MON
3,4-Dimethylaniline 370 320
3,5-Dimethylaniline 340 310
p-Toluidine 340 305
p-Ethylaniline 320 300
Diphenylamine 310 300
Aniline 310 290
p-tert-Butylaniline 300 260
N-Methylaniline 280 250
Indoline 300 150
N, N-Dimethylaniline 95 84

Whereas organic nitrates and aliphatic amines were generally found to pro-
mote knocking, aromatic amines achieve relatively high octane enhancements.
Examples of B-ON values determined for aniline and some aniline derivatives
are shown in Table 1. All measurements were performed in the same base gaso-
line with a component addition of 2 vol%. The octane enhancement efficiency var-
ies with molecular structure. Relative to aniline, differences are small when
hydrogen atoms of the aromatic ring are substituted by methyl or ethyl groups.
Substitution with longer aliphatic chains results in reduced efficiency. Whereas
primary and secondary amines have good octane enhancing properties, tertiary
amines eg, N,N-dimethylaniline are ineffective.
A serious disadvantage of nitrogen-containing components results from the
fact that nitrogen is partially oxidized in the combustion process. This increases
nitrogen oxide emissions in the exhaust gases. There is also experimental evi-
dence that some of the components may cause deposit formations in fuel intake
manifolds and/or combustion chambers, when used at the required concentra-
tions. Of the listed components, only N-methylaniline was used as an octane
enhancer for a short period after World War II in Germany.
5.3. Oxygen-Containing Components. Organic oxygen compounds
such as alcohols, ketones, ethers, esters, and phenols are commonly characterized
as ‘‘oxygenates’’ or ‘‘fuel oxygenates’’. The majority of the oxygenates act as octane
enhancers; their efficiency depends, however, on the functional group and on the
individual structure of the molecule. Some derivatives of esters, ketones, and phe-
nols achieve relatively high B-ON values, as indicated in Table 2. Because of other
inappropriate properties eg, high boiling points, high acidity, low material com-
patibility, and high toxicity they are not used as octane enhancers.
Of the oxygenates, only alcohols and ethers are used to any significant
extent. They are preferred over other octane enhancers because of lower costs,
availability, handling, and their physical and chemical properties, which are
similar to gasoline components. Names, abbreviations, formulae, and molecular
mass of the most important alcohols and ethers suitable for fuel blending are
listed in Table 3.
EC Directive on Fuel Oxygenates. The use of oxygenates in gasolines
was established during the 1970s and has developed rapidly to a large-scale
Vol. 2 OCTANE ENHANCERS 413

Table 2. Blend Octane Numbers of Selected Oxygenates


Component B-RON B-MON
3,4-Dimethylphenol 180 180
Phenol 190 160
m-Cresol 160 150
o-Cresol 160 140
Dimethyl succinate 130 130
Acetylacetone 120 110
Acetone 107 98

business. Prior to any official regulation, the German automobile and oil indus-
try reached an interindustrial agreement to restrict the methanol content of fuel
to 3 vol% with the obligation to use an adequate cosolvent component (eg, TBA) to
ensure appropiate water tolerance of methanol-containing fuels. For the EC the
European Council passed a directive in December 1985, regulating the use of
oxygenates in gasolines after several years of intensive negotiations of the min-
istries of commerce supported by experts (4). Meanwhile, the technical annex of
the directive has been implemented in the gasoline standards of all EC member
states.

Table 3. Important Fuel Oxygenates


Abbre- CAS registry
Oxygenate viation Formula Mr no.
Alcohols
Methanol MeOH CH3OH 32.04 [67-56-1]
Ethanol EtOH C2H5OH 46.07 [64-17-5]
1-Propanol PrOH C3H7OH 60.97 [71-23-8]
Isopropanol (2-propanol) IPA C3H7OH 60.97 [67-63-0]
1-Butanol NBA C4H9OH 74.12 [71-36-3]
sec-Butanol (2-butanol) SBA C2H5CH(OH)CH3 74.12 [78-92-2]
Isobutanol (2-methyl-1-propanol) IBA (CH3)2CHCH2OH 74.12 [78-83-1]
tert-Butanol (2-methyl-2-propanol) TBA (CH3)3COH 74.12 [75-65-0]
Ethers
Dimethyl ether (oxybismethane) DME CH3OCH3 46.07 [115-10-6]
Methyl tert-butyl ether (2-methoxy- MTBE CH3OC(CH3)3 88.15 [1634-04-4]
2-methyl-propane)
Ethyl tert-butyl ether (2-ethoxy- ETBE C2H5OC(CH3)3 102.18 [637-92-3]
2-methylpropane)
Diisopropyl ether (2,20 - DIPE (CH3)2CHOCH 102.18 [108-20-3]
oxybispropane) (CH3)2
tert-Amyl methyl ether TAME CH3OC(C2H5) 102.18 [994-05-8]
(2-methoxy-2-methyl-butane) (CH3)2
Isopropyl tert-butyl ether [2-methyl- PTBE (CH3)2CHOC 116.28 [17348-59-3]
2-(1-methylethoxypropane)] (CH3)3
Di-sec-butyl ether DSBE C2H5CH(CH3) 130.23 [6863-58-7]
(2,20 -oxybisbutane) OCH(CH3)C2H5
sec-Butyl-tert-butyl ether (2- BTBE C2H5CH(CH3) 130.23 [32970-45-9]
(1,1-dimethyl-ethoxybutane) OC(CH3)3
414 OCTANE ENHANCERS Vol. 2

Table 4. EC Directive on Fuel Oxygenates


A, * B, **
Oxygenate vol% vol%
Methanol (stabilizers to be added) 3 3
Ethanol (stabilizers may be added) 5 5
Isopropanol (IPA) 5 10
tert-Butanol (TBA) 7 7
Isobutanol (IBA) 7 10
Ethers, with five or more C atoms 10 15
Other oxygenates, defined 7 10
Total oxygen content, wt % 2.5 3.7

* A = limits which have to be accepted by all EC countries.


** B = limits which may be adopted by individual member states but above which labeling is
mandatory.

The purpose of the directive was to ensure:

1. The performance of the existing car population


2. The use of gasoline extender and octane-enhancing components in addition
to materials derived from crude oil
3. The compensation of octane shortfall due to lowering of alkyllead
concentration
4. A harmonization of gasoline quality in Europe

The annex lays down definitions and blend compositions of oxygenates. The
most important information is summarized in Table 4, providing the maximum
volumetric additions of individual oxygenates and the maximum gravimetric
oxygen content of the blends.
Blend Octane Numbers of Oxygenates. The octane efficiency of oxyge-
nates cannot be characterized in a simple way because it depends on a variety of
factors. Therefore, B-ON ranges are usually defined to account for their possible
variation. This makes it difficult to compare octane efficiencies of different
oxygenates.
To overcome this problem, Veba Oil conducted a series of investigations
with the objective to characterize the octane blend behavior of some oxygenates.
In the progress of this work, the gasoline sensitivity (SEN, ie, the difference
between RON and MON) was found to be a suitable correlation parameter (5).
Empirically it may be considered as an approximate gauge for the gasoline com-
position; a high paraffinic content corresponds to low SEN, high aromatic and/or
olefinic contents result in high SEN.
The measurements were performed in a variety of different unleaded gaso-
lines, the compositions of which were widely varied. Oxygenate concentrations
ranged from 10 to 20 vol%. Examples of the graphic correlations based on linear
regression of B-ON data versus gasoline SEN are shown in Figure 5 for methanol
and in Figure 6 for PTBE. Both B-RON and B-MON decrease with increasing
gasoline SEN. Similar blend behavior was found for the majority of the oxyge-
nates investigated. Typically, lower alcohols such as methanol and ethanol
show a strong dependence of B-ON on SEN. For higher alcohols and ethers
Vol. 2 OCTANE ENHANCERS 415

Fig. 5. Correlation between B-ON and SEN for methanol.

this dependence is less pronounced; in a few cases, eg, for TBA, no significant
dependence was observed.
Other Properties of Oxygenates. Beside their octane efficiency, there are
other properties of oxygenates which also affect the gasoline quality. Again their
blend behavior may not be linear so that blend properties must be determined
analoguously to the B-ON procedure. Selected properties of alcohols and ethers
used as oxygenates are listed in Table 5.
Density. Densities of alcohols are generally higher than the specified
density region for gasolines, whereas those of ethers range within it. Blend den-
sities of alcohols and ethers show only moderate differences when compared to
the densities of the neat components.
Vapor Pressure. Due to their polarity, lower alcohols such as methanol
and ethanol possess blend vapor pressures (B-VP) which exceed the vapor pres-
sure of the neat components, because association of alcohol molecules is hindered
in the mixture. Therefore, their B-VP values strongly depend on the component
concentration. In contrast, B-VP data of ether components hardly show any
dependence on concentration. They are significantly lower than those of lower
alcohols and are similar to the vapor pressure of the neat components.
Volatility. The addition of oxygenates affects the boiling characteristics of
the gasolines. When ethers are blended, only moderate changes in the distillation
416 OCTANE ENHANCERS Vol. 2

Fig. 6. Correlation between B-ON and SEN for PTBE.


.
curves are observed. Blends of lower alcohols, however, cause significant defor-
mations of the distillation curves due to the formation of azeotropes. This rises
the front-end volatility, ie, an increased distillate fraction in the temperature
range of the alcohol boiling points is obtained.

Table 5. Properties of Oxygenates


Heat of Composition, wt %
com-bustion,
Oxygenate mp, 8C bp, 8C d20
4 n20 kJ/mol RVP * C H O
MeOH  97.7 64.70 0.7912 1.3284  725.7 0.81 37.48 12.58 49.93
EtOH 114.1 78.29 0.7894 1.3614 1367.6 0.70 52.13 13.13 34.73
IPA  87.8 82.26 0.7755 1.3772 2006.9 0.72 59.95 13.42 26.63
SBA 114.7 100.00 0.8069 1.3972 2660.6 64.81 13.60 21.59
IBA  88.5 107.66 0.8016 1.3958 2668.5 0.63 84.81 13.60 21.59
TBA 25.6 82.80 0.7866 1.3877 2644.0 0.64 64.81 13.60 21.59
DME 141.5 24.80 0.6689 1.3018 1460.5 52.14 13.13 34.73
MTBE 108.6 55.20 0.7406 1.3690 3362.1 0.65 68.13 13.72 18.15
ETBE  94.0 73.10 0.7399 1.3756 4014.9 0.29 70.53 13.81 15.66
DIPE  85.5 68.30 0.7235 1.3681 4042.6 0.33 70.53 13.81 15.66
TAME 86.30 0.7700 1.3885 4010.8 0.77 70.53 13.81 15.66
PTBE 87.50 0.7400 1.3791 4667.2 0.20 72.35 13.88 13.77
* RVP Reid vapor pressure ASTM D 323.
Vol. 2 OCTANE ENHANCERS 417

Production. The important processing data of alcohols and ethers are


summarized in (6). Further details of production and economics are described
in (7).
Alcohols. Methanol is produced catalytically from synthesis gas. Apart
from its use as a gasoline blending component, large amounts of methanol are
also employed as feedstock for the etherification with isobutene to form MTBE
(see Methyl Tert-Butyl Ether).
Synthetic ethanol is produced by the hydration of ethylene with an acid cat-
alyst such as sulfuric acid and phosphoric acid in reactor columns or in ion-
exchanger fixed-bed reactors. Bioethanol is produced by fermentation of mono-
saccharides such as glucose, fructose, mannose, or xylose derived from molasses,
sugarcane, sugar, corn, or potatoes. Costs of bioethanol are higher than those of
synthetic alcohol because the process economics depend essentially on the feed-
stock price.
Higher alcohols ie, 2-propanol, 1-butanol, 2-butanol, and 2-methyl-2-
propanol can also be produced by hydration with acid ion-exchanger resins
(8). Tert-butanol (TBA) is formed as a coproduct in the production of propylene
oxide by epoxidation of propene with tert-butyl hydroperoxide (TBHP)
(! Propylene Oxide). In the case of high MTBE demand in the market, TBA is
sometimes dehydrated to isobutene which is then fed to the MTBE process (9).
Ethers. The ethers are chiefly manufactured by exothermic reaction of
branched olefins bearing the double bond at the branching point with alcohols
such as methanol, ethanol, 2-propanol, or butanol according to:

[for MTBE m ¼ n ¼ p = 1; for PTBE m ¼ n ¼ 1, p ¼ 3; for tert-amyl methyl ether


(TAME) m ¼ 2, n ¼ p ¼ 1].
MTBE is the most important ether used as an oxygenate. MTBE capacity
(see Fig. 7) and plant construction have increased steadily because MTBE blend-
ing offers simple and economic flexibility to fill any octane gap in the refiner’s

Fig. 7. Worldwide MTBE capacity trends (1989 survey).


418 OCTANE ENHANCERS Vol. 2

gasoline pool (10). More than 40 % of the total onstream capacity of MTBE is used
as merchant MTBE, which is available to supply refiners in need of additional
octane quality (11). For production of MTBE, see ! Methyl Tert-Butyl Ether,
Chap. 4.
Reaction of ethanol with isobutene produces ethyl tert-butyl ether (ETBE),
which has similar properties to MTBE. This process offers an outlet of bioethanol
(12,13). Reaction of 2-propanol with isobutene yields PTBE (14,15), an oxygenate
with excellent blend properties such as low vapor pressure and low oxygen con-
tent in combination with high octane quality.
Based on a similar technology methylbutenes (amylenes) of C4 to C6
streams can be etherified with methanol to TAME (16,17), a blending ether
with a high octane number. The BP Etherol process (18) uses unsaturated C5
compounds of cracker fractions. Branched olefins react with methanol to form
higher ethers. However, most of the branched olefins have terminal double
bonds with low reactivity. To increase the yield of 3-methyl-2-butene (isoamy-
lene) the double bond is shifted in the presence of a palladiumimpregnated
cation-exchanger resin. The isomerization proceeds according to:

Isoamylene can then be reacted with methanol to form TAME.


Environmental Protection. The compulsory use of oxygenates in gaso-
lines is becoming widespread in the United States, because several investigations
have shown that oxygenates reduce carbon monoxide emissions. This effect offers
a means of bringing CO emissions down to required standards. Depending on the
area and season in the United States, the minimum oxygen content in the fuel is
in the range of 2.0 – 3.0 wt% oxygen (Clean Air Act Amendment) (19). Addition of
TBA and methanol may reduce the CO emissions up to 40% and the hydrocarbon
emissions up to 20% in comparison with conventional gasoline (20). Insufficient
supply of oxygen in air – fuel mixtures causes incomplete combustion of the fuel,
accompanied by CO formation. When oxygen is supplied by oxygenates that are
blended to the gasoline, this chemically available oxygen reduces the formation
of carbon monoxide. Governments of several states made efforts to establish the
relationship between fuel oxygen content and CO exhaust concentrations. In
1988, the state of Colorado mandated a minimum of 1.5 wt% oxygen in gasoline
from January to March; this program was continued from November 1988 until
March 1989, this time a minimum of 2.0 wt% oxygen was required. The 1988 pro-
gram was estimated to lower CO air concentrations by 8 – 10% along the Front
range of the Rocky Mountains (10). In the United States emission measurements
with different vehicles show CO reduction up to 32%, during the Federal Test
Procedure (FTP) cycle (a standardized driving cycle for emission tests) (21).
The tests included vehicles with different exhaust and closed-loop systems and
demonstrated a statistically significant CO reduction due to the oxygenated
Vol. 2 OCTANE ENHANCERS 419

Fig. 8. CO emissions FTP 75.

fuels, depending on the system and oxygen content (Fig. 8). The gasolines were
blended with MTBE and ethanol. Emissions of unburned hydrocarbons and
nitrogen oxides were virtually unaffected.
Under European conditions, studies were made with ethers and alcohols on
the influence of oxygen content on the limited pollutants such as carbon monox-
ide, hydrocarbons, and nitrogen oxide (5). Blend components were methanol with
IPA – TBA as cosolvents and the ethers MTBE, PTBE, and BTBE. Another
investigation was conducted by the European Fuel Oxygenates Association
(EFOA) using mixtures of ethers and alcohols to study the dependence on oxyge-
nates systematically (22). Oxygen content was varied up to 3.6 wt %. The results
showed that oxygenates reduce CO emissions to a large degree. The reduction
increases with oxygen content but is not a function of the type of oxygenate.
Toxicology and Occupational Health. MTBE. Similar investigations of
toxicity and other potential health effects of MTBE were performed. Animal test-
ing to date has indicated that after ingestion, inhalation, or even injection of
MTBE, it is mostly excreted unchanged (4).
PTBE. The Ames test does not show any mutagenity. According to an
acute toxicity study in rats, PTBE is considered to be nontoxic up to a dose of
2000 mg per kilogram body weight.
Storage and Transportation. The higher ethers MTBE, PTBE, and
TAME are stable in the presence of air under atmospheric conditions. Formation
of peroxides is not observed during contact with oxygen. Containers, sealing
materials, and also elastomers in automobiles have to meet the same require-
ments as for contact with gasoline.

6. Metal-Containing Additives

A large number of organometallic compounds are known to act as octane enhan-


cers. Generally, compounds containing Pb, Cu, Ni, Th, Fe, and Mn achieve high
octane improvements, whereas those containing Co, Zn, Bi, V, Se, and Ce are
420 OCTANE ENHANCERS Vol. 2

slightly less effective. For practical use, most of them must be excluded because
of high costs, instability, toxicity, impact on catalytic converters, or other unde-
sired properties.
Their use in gasolines also requires an official approval in most European
countries and in the United States. In Germany, the application of metal-
containing additives–other than lead–is restricted by the gasoline lead act
(Benzin – Blei – Gesetz).
The octane enhancement of organometallic additives is explained by a cat-
alytic effect, because during combustion the metal compound is oxidized to form
very fine solid particles of metal oxide. These particles have large surface areas
with active centers which can react with free radicals formed by partial oxidation
of the gasoline. Therefore the spontaneous combustion is delayed and knocking
does not occur. The consumed centers are reactivated by oxygen and can react
again with free radicals. Due to this cycle only small concentrations of additive
are necessary to cause high octane enhancement.
A metal-free component is unable to delay the spontaneous combustion cat-
alytically. The organic compounds are consumed irreversibly by the intermediate
reactants which are responsible for knocking. Therefore the effectivity of anti-
knocking components is limited and high concentrations are needed to get a
higher octane quality.
A disadvantage of metal-containing additives results from the formation of
solid combustion products that lead to deposit accumulation in the combustion
chambers. For the lead fluids this problem was solved by addition of lead scaven-
gers. However, this is unfeasible for most other organometallic additives, thus
their use must be restricted to very low concentrations. Additives that achieve
sufficient octane improvements in the ppm concentration range and that meet
other technical and economical requirements are methylcyclopentadienylmanga-
nese tricarbonyl (MMT) and dicyclopentadienyliron (ferrocene).
6.1. Alkyllead Compounds. The strong octane-boosting effect of tetra-
ethyllead (TEL) was discovered in 1921 in the United States (23). Its wide use
began in the United States in 1929 and in Germany shortly before World
War II. Curiously, investigations at that time regarded lead as relatively envir-
onmentally friendly.
The effect of lead is based on an intervention in the mechanism of the com-
bustion process. In precombustion reactions, organic peroxides are formed,
which decompose to form free radicals. The latter initiate knocking reactions.
These highly reactive radicals can be deactivated by very finely divided lead com-
pounds formed from the alkyllead compounds in the combustion chamber
(24,25).
In addition to tetraethyllead, which was used exclusively at first, tetra-
methyllead (TML) came into use from 1960 onward. Mixed alkyllead compounds,
such as diethyldimethyllead or ethyltrimethyllead, and mixtures of TEL and
TML were also employed, because a broad-spectrum effect can be achieved
through a combination of different characteristics (Table 6). Thus, the compara-
tively low boiling point of TML is good for increasing the front octane number.
(Fig. 9).
The increase in octane number that can be achieved with alkyllead com-
pounds depends strongly on fuel composition. The lead susceptibility, a measure
Vol. 2 OCTANE ENHANCERS 421

Table 6. Comparison of TEL and TML


Tetraethyllead Tetramethyllead
Chemical and physical
properties (C2H5)4Pb (CH3)4Pb
Molecular mass, g/mol 323.5 267.3
Lead content, % 323.5 267.3
Density, g/cm3 1.66 1.995
bp, 8C 200 (decomp.) 110
Vapor pressure 34 31.5  102
(20 8C), Pa

of the response behavior to the addition of lead, decreases in the following order:
paraffins  cycloparaffins > aromatics > olefins > diolefins (& 0). Accordingly,
fuel components rich in paraffins, such as straight-run gasoline, alkylate, and
isomerate, react very well to lead addition, whereas olefin-rich catalytically
cracked gasoline has a poor response (Fig. 10).
In connection with the addition of alkyllead compounds to gasoline, the
term ‘‘lead bonus’’ was sometimes used. It meant that highly leaded fuels, in com-
parison with unleaded or low-lead ones, have a disproportionately high road
octane number, although the RON and MON determined in test engines are
equally high (26–30). Investigations carried out with the introduction of
unleaded fuels showed, however, that low-lead and unleaded fuels behave almost
identically with regard to the road octane number (31) in most cases.

Fig. 9. Effect of tetraethyllead and tetramethyllead on the gasoline front octane number
(FON).
422 OCTANE ENHANCERS Vol. 2

Fig. 10. Effect of tetraethyllead on some base fuels (octane quality of the base fuel types
are widely varying in practice). a) Olefin-rich fuel; b) Aromatic-rich fuel; c) Paraffin-rich
fuel.

Usually, alkyllead compounds have always been used in combination


with scavengers. These are organochlorine and -bromine compounds, namely,
1,2-dichloroethane and 1,2-dibromoethane. In the combustion chamber, scaven-
gers cause the formation of lead chloride and lead bromide whose good volatility
at high temperature ensures that ca. 70–80 % of the lead is not deposited there.
However, investigations in 1989 have shown that scavengers can be removed with-
out adverse affects when the lead concentration in the fuel is low (eg, 0.15 g/L).
On the other hand, treatment of leaded fuels with an antioxidant is neces-
sary. Tetraethyllead, in particular, is prone to oxidation over long storage peri-
ods, leading to the separation of solid residues that can lead to blockages in the
fuel system.
Originally, the addition of lead to gasoline was considered to be safe to the
environment. This opinion has steadily changed, however (32–34), so that a
decrease in the maximum permitted lead content has been legislated virtually
worldwide. In the United States, Japan, Canada, Australia, Europe, and other
areas leaded fuels have practically disappeared from the market. From a toxico-
logical point of view, however, not only the lead but also the scavengers con-
nected with it are suspected to be health risks, because these can form traces
of highly toxic halogenated dibenzodioxins and -furans in the combustion cham-
ber (35). However, not only the poor compatibility of leaded fuels with the envir-
onment but also their incompatibility with catalytic converter technology
(catalyst poisoning) has driven the move to unleaded gasoline in many parts of
the world.
Vol. 2 OCTANE ENHANCERS 423

Alkyllead compounds protected also unhardened valve seats, present in


older engines. Here, lead compounds formed in the combustion chamber inhib-
ited erosion and wear (‘‘recession’’) of the valve seats in two ways: (1) by avoiding
high-temperature corrosion and (2) by their effect as solid lubricants. However,
nontoxic alkali-metal additives are now available for this purpose, e.g., Power-
shield (Lubrizo) Formula Shell (Shell) Valve Master (Du Pont). Practically the
only suppliers of alkyllead compounds are Ethyl and Octel.
6.2. Methylcyclopentadienylmanganese tricarbonyl. Methylcyclo-
pentadienylmanganese tricarbonyl (MMT) [12108-13-3], is an organometallic
additive used as an octane enhancer at low concentration. MMT is a metallo-
cene compound in which the delocalized electrons of the cyclopentadienyl
group are partially donated to the manganese atom and form covalent bonds.
Carbon monoxide has a lone electron pair on the carbon atom which is donated
to the manganese atom to form the carbonyl bond. In Canada it has been
blended to unleaded gasoline for many years at a concentration of
0.0165 g Mn per liter of gasoline. In the United States, MMT is used in combi-
nation with alkylleads in leaded gasoline. Ethyl Corporation (Orangeburg) is
the U.S. producer of MMT.
History. Ethyl Corporation discovered MMT in 1954 and introduced it into
the U.S. market as a supplement to TEL in 1958. Evaluation in unleaded gasoline
was made in 1969. From 1974 to 1979 MMT was used in unleaded gasoline.
The effect of MMT was tested on a series of in-use three-way catalysts. Cat-
alyst activity was measured and compared to the activity of the catalyst without
MMT operation. Steady-state and light-off temperature measurements indicated
a reduction in catalyst efficiency as exposure to MMT was increased (37).
The EPA predicted that over the useful life of vehicles, a greater number of
MMT-fueled vehicles would not meet their designed hydrocarbon emission stan-
dard compared with corresponding vehicles fueled without MMT. Based on these
results and in accordance with Clean Air Act Amendments, the EPA banned
MMT in 1979 for unleaded gasoline (38).
In 1983 production of MMT was ca. 1800 t/a.
Trade names: AK-33 X, Antiknock – 33, Combustion improver – 2.
Physical Properties (39). MMT is sparingly soluble in water (70 ppm at
25 8C) and soluble in hydrocarbon solvents. MMT is a dark orange (straw color)
liquid with a herbaceous odor; mp  2.2 8C, bp 232.8 8C.
Chemical Properties. When heated to decomposition MMT emits toxic
fumes of CO.
Production. The methylcyclopentadiene dimer [26472-00-4] is added to
molten sodium suspended in diethylene glycol dimethyl ether at 185 to 190 8C.
At this temperature, the dimer splits into the monomer and reacts with sodium
to form sodium methylcyclopentadienide:

Anhydrous, flaked manganese chloride is added to the reaction mixture.


The reaction forms bis(methylcyclopentadienyl)manganese with sodium chloride
as byproduct. In the next step carbonylation under a pressure of 4.3 to 4.5 MPa
424 OCTANE ENHANCERS Vol. 2

Table 7. Toxicity Data of MMT* (40)


Species
Test Rat Mouse Dog Rabbit Guinea pig Cat
Oral
LD50, mg/kg 50 230
LDLo, mg/kg 620 95 900
Skin
LD50, mg/kg 140
Inhalation
LD50, mg/m3 (4 h) 76 58.6
LCLo, mg/m3 489 1237 1237 223
(2 h) (1 h) (1 h) (7 h)
Intraperitoneal
LD50, mg/kg 23 152
Intravenous
LDLo, mg/kg 10
LD50, mg/kg 6600

*American Conference of Governmental Industrial Hygienists—Threshold Limit Values:


0.1 mg Mn/m3 (skin) time weighted average.

at 190 8C forms MMT. Pure MMT is subsequently separated by fractional


distillation.

Octane Enhancement. MMT can enhance between 0.5 and 1.0


(RON þ MON)/2 per 0.01 g Mn/L, depending on gasoline composition, octane
level, and leaded or unleaded basestock.
Toxicology and Occupational Health. MMT is a poison by ingestion,
inhalation, skin contact, intravenous, and intraperitoneal routes (Table 7). MMT
is included in the EPA’s Extremely Harzardous Substances List. Manganese and
some compounds are on the Community Right To Know List.
6.3. Ferrocene. In the early 1930s iron pentacarbonyl was used com-
mercially as an octane-enhancing additive in Europe in concentrations similar
to those of lead at that time. During combustion in engines iron pentacarbonyl
formed excessive amounts of iron oxides which caused engine wear and influ-
enced spark plug operation.
Ferrocene is another iron compound with similar octane-enhancing proper-
ties. Ferrocene [102-54-5] or dicyclopentadienyliron, Fe (C5H5)2, belongs to the
metallocenes; the iron atom is centered between two parallel cyclopentadiene
rings. Ferrocene was first synthesized by MILLER, TEBOTH and TREMAINE in
Vol. 2 OCTANE ENHANCERS 425

1952. The sandwich or doublecone structure was first elucidated by G. WILKINSON


and E. O. FISCHER.
Physical Properties. Important physical data of ferrocene are listed
below (41):

Mr 186.04
Density 1.49 g/cm3
mp 173 8C
bp (at 102.3 kPa) 249 8C
Triple point 183 8C
Vapor pressure
40 8C 0.0066 kPa
100 8C 0.34 kPa
Heat of fusion (at 175 8C) 17.8 kJ/mol
Heat of vaporization (at 175 8C) 47.2 kJ/mol
Heat of sublimation (at 25 8C) 70.2 kJ/mol
Start of decomposition 465 8C

Ferrocene is readily soluble in all hydrocarbon and hydrocarbon –


oxygenate fuel blends and does not alter specified gasoline properties except
for octane numbers.
Chemical Properties. Ferrocene is resistant to water, air, light, and
heat.
Production. The modern electrochemical synthesis of ferrocene consists
of two stages:

1. Electrochemical step (NaBr added as conductivity enhancer)

Fe þ 2 C2 H5 OH!FeðOC2 H5 Þ2 þ H2

2 Chemical conversion

FeðOC2 H5 Þ2 þ 2 C5 H6 ! FeðC5 H5 Þ2 þ 2 C2 H5 OH

Total:

Fe þ 2 C5 H6 ! FeðC5 H5 Þ2 þ H2

In the first stage, metallic iron is oxidized anodically in ethanol to


iron(II)ethoxide. At the cathode, the acidic protons of ethanol are reduced and
form gaseous hydrogen which evolves out of the liquid. The iron(II)ethoxide is
sparingly soluble and precipitates on the rotating cathode. A fixed scraper sepa-
rates the iron(II)ethoxide from the cathode surface and suspends it into the
liquid. In the second stage the ethoxide reacts with cyclopentadiene to form fer-
rocene and ethanol, which is recycled to the first stage. Ferrocene is filtered from
the solution and purified by sublimation.
Octane Enhancement. The octane improvements depend–as in the case
of other octane enhancers–on the ferrocene concentration and on the composi-
tion of the base gasoline. Figure 11 represents a compilation of octane
426 OCTANE ENHANCERS Vol. 2

Fig. 11. Ferroncene octane enhancement in leaded gas olines.

measurements performed in about 50 different base gasolines with ferrocene


concentrations up to 30 ppm (42). The octane response of RON is higher than
that of MON.
The composition of the different unleaded fuel grades tested was widely
varied. The fuel blends included all major refinery components as well as avail-
able oxygenates within their maximum volumetric addition presently specified in

Table 8. Toxicity data of ferrocene (40)


Species
Test Rat Mouse
Oral
LD50, mg/kg 1320 832
Intraperitoneal
LD50, mg/kg 500 335
Intravenous
LD50, mg/kg 178
Vol. 2 OCTANE ENHANCERS 427

Europe. Generally it was found that the lower the octane number of the base
gasoline, the higher the octane gain. Several measurements of octane response
versus concentration have shown that response is more pronounced in the low
ferrocene concentration region and that levelling-off effects are obtained at
higher concentrations.
Toxicology. The toxicological data of ferrocene are listed in Table 8.

REFERENCES

1. CEC Symp. Knocking of Combustion Engine, Wolfsburg, Germany Nov. 1981.


2. T. Midgley Jr., Ind. Eng. Chem. 31 (1939) no. 5, 504 – 506.
3. M. H. Graham, L. Hess: Organic Nitrogen Compounds as Antiknock Additives
Report IR 52 – 12 Ethyl-Corporation, June 1952.
4. M. Allion, EFOA 4th Conf., Brussels 1990.
5. K. Gottlieb, H.-K. Schädlich, P. Schug: ‘‘Erzeugung von oktanzahlsteigernden
höheren Ethern und ihre anwendungstechnische Erprobung unter praxisnahen
Bedingungen’’, BMFT-Forschungsbericht 85–034 (1989).
6. L. F. Hatch, S. Matar: From Hydrocarbons to Petrochemicals, Gulf Publ. Co.,
Houston 1981.
7. SRI reports: P. L. Morse, Rep. no. 53, ‘‘Synthetic Ethanol and Isopropanol’’, (1969);
H. W. Scheeline, S. Terada, S. L. Soder, Rep. no. 131, ‘‘Methyl-Tertiary-Butyl-Ether
and Tertiary-Butyl-Alcohol’’, (1974); P. L. Morse, Rep. no. 111, ‘‘Methanol as a Che-
mical Raw Material’’, (1979), Stanford Research Institute, Menlo Park, California.
8. J. Schulze, M. Homann: C4-hydrocarbons and Derivatives, Springer Verlag, Berlin,
1989.
9. H. E. Eilerts de Haan: ‘‘MTBE and Methanol Blends in Europe’’,EFOA 3rd Conf.,
ARCO Chemical Europe, Frankfurt 1988.
10. G. H. Unzelmann, Oil Gas J., 87 (1989) Apr. 17, 44.
11. Oxygenated Fuels Association, Inc., Feb. 10, 1989, Communication to EFOA.
12. H. Bruderreck, BML-Report FE 83 NR 003 (1985).
13. D. Schliephake (ed.): Nachwachsende Rohstoffe, Verlag J. Kordt, Bochum 1986.
14. VEBA OEL, EP 006 315, 1982 (K. Gottlieb, H. Bruderreck, F. H. Wehmeier).
15. T. Kaitale, E. Merikallio, E. Rautianen, EFOA 2nd Conf., Rome 1987.
16. W. J. Piel: National Conference on Octane & Oxygenates, San Francisco (1989).
17. Oil Gas J. 86 (1988) Nov. 21, 41.
18. H. Wittcoff, CHEMTECH 1990, Jan., 48.
19. EFOA, Newsletter no. 4, July 1989.
20. ARCO, Report on Methanol, 1983.
21. J. A. Gething, J. S. Welstand, J. C. Horn, SAE-paper 890 216 (1989).
22. H. Guttmann, K. P. Schug: ‘‘Results of Exhaust Emission Measurements with
Oxygenates Motor Gasolines’’, SAE-paper no. 900 274, 1990.
23. D. Blackmore, A. Thomas (eds.): Fuel Economy of the Gasoline Engine, Macmillan,
New York 1977, pp. 79.
24. R. Sharma, J. Bardwell, Combust. Flame 9 (1965) 106 – 108.
25. K. Salooja, Combust. Flame 9 (1965) 211 – 217.
26. A. Bell, J. Keene, K. Reders, SAE-Paper no. 710 624, 1971.
27. D. Gray, A. Azhari, SAE-Paper no. 720 084, 1972.
28. H. Bigley, J. Benson, SAE-Paper no. 730 013, 1973.
29. N. Mariotti, A. Ramsden, Int. Automob. Tech. Congr., Pap. B 2–1, 1974.
30. D. Hornbeck, R. Labruyere, L. Stinson, G. Unzelmann, SAE-Paper no. 750 936, 1975.
428 METHYL TERT-BUTYL ETHER Vol. 2

31. K. Reders, W. Dabelstein, A. Reglitzky, W. Lange, Mineralöltechnik (1989) no. 13,


38.
32. H. Schellhaus, W. Obländer, K. Neuschwander, Schriften. Ver. Wasser- Boden-
Lufthyg. 42 (1974) 245 – 258
33. Conference Report on International Symposium, Airborne Lead in Perspective,
National Academy of S ciences, Conference Report, Washington, D.C. 1972.
34. EEC Commission: Health Aspects of the Environmental Pollution with Lead,
Amsterdam 1972/1973.
35. E. Bessey, Report on the BMFT (Bundesministerium für Forschung und Technologie)–
FVV (Forschungsvereinigung Verbrennungskraftmaschinen)–DGMK (Deutsche
Gesellschaft für Mineralöl und Kohle) Gemeinschaftsprojekt ‘‘Emission halogenier-
ter KW,’’ DGMK-Conference Report 8901, 1989, pp. 85 – 119.
36. G. Schoonveld, R. Riley, S. Thomas, S. Schiff, Autom. Eng. 95 (1987) no. 1, 72 – 76.
37. R. G. Williams, W. L. H. Watkins, R. G. Griffins, SAE-paper no. 890 582 (1989).
38. Environmental Protection Agency; Regulation of Fuel and Fuel Additives, MMT
Suspension of Enforcement, Ann Arbor 1979.
39. R. K. Hinderer, Am. Ind. Hyg. Assoc. J. 40 (1979) 164.
40. N. I. Sax, R. J. Lewis: Dangerous properties of Industrial Material, 7th ed., van
Nostrand – Reinhold, New York 1988.
41. Chemische Betriebe Pluto, Ferrocene-Product-Information, Herne, Germany 1989.
42. K. P. Schug, H.-J. Guttmann, A.-W. Preuss, K. Schädlich, SAE-paper no. 900 154
(1990).

KUNO SCHÄDLICH
PETER SCHUG
VEBA OEL AG, Gelsenkirchen Federal
Republic of Germany
WERNER DABELSTEIN
ARNO REGLITZKY
Deutsche Shell AG, Hamburg Federal
Republic of Germany

METHYL TERT-BUTYL
ETHER
1. Introduction

Methyl tert-butyl ether, 2-methoxy-2-methylpropane (MTBE) [1634-04-4], Mr


88.15, was first synthesized (by the classical Williamson ether synthesis) and
characterized in 1904 (1).

ß2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim


10.1002/14356007.a16_543
Vol. 2 METHYL TERT-BUTYL ETHER 429

Extensive studies in the United States during World War II demonstrated


the outstanding qualities of MTBE as a high-octane fuel component (2). Even so,
it was not until 1973 that the first commercial plant went on stream in Italy.
Reduction of the lead content in gasoline in the mid-1970s led to a drastic
increase in demand for octane enhancers, with MTBE being used increasingly in
this role. Political decisions about gasoline quality (eg, lower aromatics content,
lower vapor pressure and a defined oxygen content) especially in the United
States had led to a significant increase in MTBE consumption in the 90th. In
1997 the world wide MTBE production reached some 19  106 t (44). After
years with two-digit growth rates (1990–1995) the increase in MTBE consump-
tion is expected to be less than 2 % per year in the near future.

2. Physical and Chemical Properties

Physical Properties. Methyl tert-butyl ether is a colorless, readily


mobile liquid with a characteristic terpene-like odor. Its most important physical
properties are listed in Table 1:
Vapor pressure, density, and solubility in water, as well as the composition
and boiling points of azeotropes with water and methanol, are given in Tables 2

Table 1. Selected Physical Properties of MTBE


mp  108.6 8 C
bp 55.3 8 C
n20
D 1.3692
Dielectric constant (20 8C) 4.5
Viscosity (20 8C) 0.36 mPa
s
Surface tension (20 8C) 20 mN m1
Specific heat (20 8C) 2.18 kJ kg1 K1
Heat of vaporization (at bp) 337 kJ/ kg1
Heat of formation (25 8C)  314 kJ mol1
Heat of combustion  34.88 MJ kg1
Flash point (Abel – Pensky)  28 8C
Ignition temperature (DIN 51 794) 460 8 C
Explosion limits in air 1.65 – 8.4 vol %
Critical data
tcr 224.0 8C
pcr 3.43 MPa

Table 2. Vapor Pressure, Density, and Water Miscibility of MTBE


Miscibility
Temperature, Vapor pressure,* Density,
3
8C kPa kg/m Water in MTBE, wt % MTBE in water, wt %
0 10.89 761.3 1.19 7.3
10 17.59 751.0 1.22 5.0
20 27.21 740.7 1.28 3.3
30 40.88 730.4 1.36 2.2
40 59.69 720.1 1.47 1.5
*Antoine Parameters: A ¼ 6.254909, B ¼ 265.4, C ¼ 242.517 log p= A  B / (C + W); [W] ¼ 8C
430 METHYL TERT-BUTYL ETHER Vol. 2

Table 3. Binary Azeotropes with MTBE


Azeotrope bp, 8C MTBE content, wt %
MTBE – water 52.6 96*
MTBE – methanol 51.6 86
MTBE – methanol (1.0 MPa) 130 68
MTBE – methanol (2.5 MPa) 175 54
* Condensate separates into two phases.

and 3 (3). Methyl tert-butyl ether has unlimited miscibility with all ordinary
organic solvents and all hydrocarbons.
Chemical Properties. Methyl tert-butyl ether is very stable under alka-
line, neutral, and weakly acidic conditions. In the presence of strong acids, it is
cleaved to methanol and isobutene. Depending on reaction conditions the latter
can form isobutene oligomers.

3. Resources and Raw Materials

MTBE is produced by the reaction of isobutene, contained in C4-fractions, and


Methanol. At present, isobutene from the following sources is used as feedstock
for MTBE production (45):

1) isobutene in raffinate 1, which is formed as a coproduct of butadiene


production from steam cracker C4 fractions (see Table 4 for typical
composition).
2) pseudoraffinate 1, which is obtained by selective hydrogenation of buta-
diene in mixed C4 fractions from steam crackers (composition, see Table 4)
3) isobutene contained in the C4 fraction of fluid catalytic crackers (FCC–C4’s;
for typical composition, see Table 4). FCC–C4 is used as feedstock for ca.
29 % of MTBE production
4) isobutene from dehydrogenation of isobutane, which is obtained both
from refineries and from field butanes after isomerization (35 % of
MTBE feedstocks); and

Table 4. Typical Composition of C4 Hydrocarbon Streams from Steam Crackers


(Raffinate 1) and Fluid Catalytic Crackers (FCC – C4)
Compound Raffinate 1, wt % Pseudoraffinate 1, wt % FCC – C4, wt %
Isobutane 4 2 36
n-Butane 12 7.5 13
Isobutene 44 24 15
Butene-1 24 39 12
cis-Butene-2 6 8 9
trans-Butene-2 9 19 14
Butadiene-1,3 0.5 0.0 0.3
Balance 0.5 0.5 0.7
Vol. 2 METHYL TERT-BUTYL ETHER 431

5) isobutene by dehydration of tert-butanol, a coproduct of propylene oxide


synthesis (Halcon/Arco process, see ! Propylene Oxide) (ca. 15 % of
MTBE feedstocks)

Raffinate 1 and pseudoraffinate 1 provide isobutene for approximately 21 % of


world’s total MTBE production (46).
As demand for MTBE increases, the first source to be exploited will prob-
ably be free FCC–C4 reserves. Any further expansion – and this also holds for
the raffinate 1 route – is thus tied to new cracker construction. Field butanes
will grow more than proportionally in importance. The relative share of tert-
butanol as an isobutene source for MTBE production, on the other hand, is
expected to decline because it is formed as a coproduct and thus linked to future
propylene oxide demand. Raffinate 1 and FCC–C4 can be utilized directly in
MTBE synthesis. By contrast, isobutane must be dehydrogenated. The same
holds true for field butanes after isomerization of the n-butane fraction to isobu-
tane. A number of industrial processes have been established during the last dec-
ades (4,47). For the primary butane isomerization, the Butamer process is most
commonly used. For the isobutane dehydrogenation, the commercially available
processes Oleflex (UOP), Catofin (ABB Lummus Crest, Inc.), STAR (Phillips
Petroleum Co.), and FBD–4 (Snamprogetti SpA), are at present of industrial
importance.
To produce MTBE from tert-butanol (14), isobutene must first be obtained
by elimination of water from the alcohol before the olefin can be used for ether
production.
Methanol, the second reactant in MTBE synthesis, is produced at a typical
purity of > 99.9 % and is used directly for ether synthesis without further puri-
fication. Comparison of today’s methanol capacities, which amounts to 32  106 t/
a, and the methanol demand of 26  106 t/a (48) would allow an additional
increase of MTBE production of ca. 16  106 t/a. Some 25 % of the present metha-
nol output is consumed by MTBE.

4. Production

Methyl tert-butyl ether can be obtained by the acid-catalyzed addition of metha-


nol to isobutene (17). Suitable catalysts are solid acids such as bentonites (18),
zeolites (8,49,50) and – commonly used in industrial world scale MTBE-produc-
tion units – macroporous acidic ion-exchange resins (20). The reaction is weakly
exothermic with a heat of reaction of  37.7 kJ/mol.
The kinetics of MTBE formation have been investigated intensively
(9,51,52). It has been shown recently, that the observed reaction rates can be
described sufficiently both by a kinetic model according to a Langmuir–Hinshel-
wood mechanism and by an Eley–Rideal (ER) approach (53). However, the ER-
model seems to be the most probable one. The commonly used kinetic model has
been developed by REHFINGER et al. (54). Because of equilibrium limitations only
92 % conversion can be achieved with equimolar amounts of isobutene and
methanol at 333 K. An excess of methanol not only increases the conversion of
isobutene but also suppresses its dimerization and oligomerization. Dimerization
432 METHYL TERT-BUTYL ETHER Vol. 2

of isobutene is the most important side reaction of MTBE synthesis (55,56). At a


molar methanol excess as low as 10 %, the selectivity for MTBE is practically
100 %.
In recent years, in addition to the Snamprogetti (57) and Hüls (now Oxeno)
processes (10,58,59) processes developed by Arco (42), IFP (60), and CDTECH
(ABB Lummus Crest and Chemical Research Licensing) have been established (31).
Other industrial processes have been developed by DEA (formerly Deutsche
Texaco) (13), Shell (Netherlands) (14), Phillips Petroleum (60), and Sumitomo
(15). More than 140 MTBE plants with a total installed capacity of ca.
20  106 t/a are on stream (44,46).
Process Description. Most commercially available processes are com-
parable and consist of a reaction and a refining section.
Reaction Section. All processes have in common the reaction of isobutene
with a certain molar excess of methanol on a macroporous acidic ion exchanger
at 50–90 8C. In the Snamprogetti, Hüls, and Arco processes, the pressure in the
reaction section is chosen so that the reactants over the catalyst are entirely in
the liquid phase over the catalyst bed, ie, between 1.0 and 1.5 MPa. This
enhances catalyst life time by reducing the polymerization of isobutene at
the catalyst’s surface and promotes the selectivity for the MTBE formation.
For example, only ca. 100–200 ppm by weight of isobutene is converted to
diisobutene (2.2.4-trimethylpentene-1 and -2) in the ‘‘Hüls-MTBE-Process’’.
Design of the reaction section is largely determined by the heat of reaction,
which must be removed. The heat of reaction depends on the isobutene concen-
tration in the feedstock. In the three processes mentioned (Snamprogetti, Hüls,
Arco) adiabatic reactors are always employed with a FCC–C4 feedstock. If raffi-
nate 1 is used as feedstock, Snamprogetti and Oxeno prefer tubular reactors for
reasons of process engineering, whereas Arco employs recycle reactors. In the
adiabatic reactors employed by IFP (Institute Français du Pétrol) or CDTECH,
which also requires a recycle at higher isobutene concentration, an isobutene
conversion of approximately 85 % is achieved. The heat of reaction is removed
by partial evaporation of the C4 hydrocarbons in the reactor as the reaction pro-
ceeds. This process design, however, gives rise to a certain loss in selectivity for
MTBE. A final isobutene conversion of ca. 97 % is achieved by a subsequent cat-
alytic distillation.
In most industrial plants, isobutene conversion of 95–97 % is sufficient.
Residual butenes are mainly used for the manufacture of alkylate gasoline,
recycled to the cracker, or simply burned. If they are to be utilized for other che-
mical purposes such as the production of polymer-grade butene-1, the degree of
isobutene conversion must be significantly increased. To overcome the conver-
sion limit of 95 to 97 % posed by the chemical equilibrium, a higher excess of
methanol in the feed stream can be used. Unfortunately, this excess methanol
remains in the MTBE product. It can, nevertheless, be recovered by distillation,
eg, as a methanol-poor minimum boiling azeotrope with MTBE (23). Not only
does this give MTBE with purities >99.7 % (balance mainly tert-butanol pro-
duced by a reaction of isobutene with water dissolved in the feedstock), it also
results in isobutene conversions of ca. 98 %. In the Oxeno-MTBE process isobu-
tene conversions of >99.9 % could be achieved in a two-stage process (see Fig. 1).
This conversion corresponds to ca. 300 ppm per weight of residual isobutene in
Vol. 2 METHYL TERT-BUTYL ETHER 433

Fig. 1. Two-stage Oxeno-MTBE-Process. First stage: a) Multitubular reactor; b) Adia-


batic reactor; c) First C4 distillation tower (debutanizer). Second stage: d) Secondary adia-
batic reactor; e) Second C4 distillation tower Methanol recovery: f) Methanol extraction;
g) Methanol tower.

the spent C4 fraction. Usually, if the feedstock is a raffinate 1 stream, the spent
C4 fraction is called raffinate 2.
This high conversion can also be obtained by using highly sulfonated acidic
resins in the reaction section followed by an additional catalytic distillation col-
umn in the refining section.
Refining Section. Catalytic distillation (CD) or reactive distillation (RD)
refers to a process, where both catalytic reaction and distillation are carried
out simultaneously in the debutanizer column. From the viewpoint of reaction
engineering, this column acts as a two-phase countercurrent flow, fixed-bed cat-
alytic reactor. Usually CD is defined as a process, in which a heterogeneous cat-
alyst is located in a distillation column. A more common term for this operation is
RD. This term does not distinguish between a homogeneous or heterogeneous
catalyzed reaction in distillation columns.
The most important advantage of using CD for MTBE synthesis lies
in the elimination of equilibrium limitation of isobutene conversion as a
result of continuous removal of the reaction product MTBE from the reaction
mixture.
The partly reacted mixture from the reaction section – which is usually in
chemical equilibrium – enters the CD column below the catalyst packing zone to
ensure the separation of the high-boiling component MTBE from the feed
stream. The catalyst packing is installed in the upper mid portion of the column
with normal distillation sections above and below.
The CD-based MTBE process was first developed by CR & L using a
patented catalyst support system (commercialized as CD MTBE or CD
ETHEROL) (16,61). The catalyst used in the CD columns is essentially the
434 METHYL TERT-BUTYL ETHER Vol. 2

same as that employed in the reaction section. It is an acidic ion-exchange resin,


which is supported by a fiberglass cloth reinforced with stainless steel wire mesh.
These fiberglass cloths are rolled to catalyst bales and stacked on sieve trays in
the column.
A CD-process has been developed also by UOP and Hüls in 1992, utilizing
Koch Engineering’s KataMax-technology (62) (commercialized as ETHERMAX
process). In this kind of process technology the catalytic section of the CD-column
uses a conventionally structured, wire-cloth distillation packing, which contain
pellets of strong acidic ion-exchange resin (63). The benefit in using this type
of packing is a very good distribution of liquid and vapor phase at a low pressure
drop, an efficient contact of reactants with catalyst pellets, and instantaneous
distillative removal of reactants. Therefore, the packing exhibits excellent
separation performance combined with an efficient mass and heat transfer for
chemical reaction. An analogous catalytic packing system have been developed
by Sulzer (commercialized as KataPak) (64) and Montz (commercialized as Mul-
tiPak) (65).
CD-technology has the important advantage over conventional unit design
(reaction section followed by debutanizing in a distillation column) of lower capi-
tal investment costs, if a very high conversion of isobutene has to be achieved.
Because in the CD-process both chemical reaction and distillation are carried
out in the same device, at least one process step is eliminated. Beyond this,
the exothermic heat of reaction can be used to vaporize the reaction mixture.
Therefore, the maximum temperature in the reaction zone is limited to the boil-
ing point of the mixture, so that an efficient and secure temperature control is
possible.
One evident disadvantage of the CD-process technology is the fact, that
since the debutanizer column must be taken out of operation during exchange
of the catalytic packing, also the whole MTBE unit must be shut down. In con-
trast, in the conventional design, where the reaction section usually consists of
two or more fixed bed reactors (eg, in the two stage Hüls-MTBE-Process), a con-
tinued production can be performed at a lower feed rate, while the catalyst is
exchanged. In the CD-technology it must also be ensured that a small excess
of methanol exists along the reaction zone. In the absence of methanol the
exothermic dimerization of butenes take place at high reaction rates. The ensu-
ing sharp temperature rise (‘‘hot spot’’) will cause irreversible catalyst deactiva-
tion and catalyst damage.
Nevertheless, since the mid-1980s most of the MTBE plants, which have
been built, were equipped with a CD-column.
In the removal of nonconverted C4 hydrocarbons from MTBE by distillation,
excess methanol appears in the distillate, because of formation of an methanol-C4
azeotrope. The methanol content in the azeotrope depends on pressure and C4-
composition, eg, at 0.6 MPa the azeotrope contains ca. 2 to 4 wt % methanol. This
methanol can completely be removed from the C4-stream in an extraction column
using a countercurrent water wash. The methanol can be recovered by distilla-
tion and is recycled to the reaction section.
If methanol losses of ca. 200 ppm by weight are acceptable, methanol recov-
ery from the distillate with molecular sieves is also appropriate (17). This
200 ppm residual methanol together with dimethyl ether, also formed in trace
Vol. 2 METHYL TERT-BUTYL ETHER 435

amounts, can be separated from the butene fraction in a subsequent molecular


sieve adsorption unit (18).
Construction Materials. Because corrosive media are not used anywhere
and do not result from catalyst disintegration, the entire plant is fabricated
from standard carbon steel. Moreover under normal operating conditions, resi-
dues, process wastewater, and gaseous emissions are not formed.
Investment costs (inside battery limits, IBL) for a 200 000 t/a MTBE plant
(Oxeno-MTBE-Process) in the Federal Republic of Germany, running on raffina-
te I feedstock, amount to 14  106 DM.

5. Environmental Protection

The small amounts of byproducts in MTBE need not be removed if the product
is to be used for gasoline. The catalyst is regenerable, and no environmental pro-
blems arise from its disposal.
Because MTBE is soluble in water, suitable precautions against
groundwater contamination must be taken in loading and storage areas.
Recently published complains about groundwater contamination and the
reactions taken by public authorities in parts of the United States arose
from careless storage of the product. For the environmental significance of
MTBE, see Section 11. For emission control in the storage and handling of
MTBE.

6. Quality Specifications

The usual purity of commercial MTBE is 98–99 wt %. The byproducts, tert-buta-


nol and diisobutenes, as well as excess, residual methanol, do not have a detri-
mental effect on octane number improvement through the use of MTBE.
Depending on the quality of the C4 feedstock mixture, the MTBE product may
also contain C5 and C6 hydrocarbons. A removal of these components by distilla-
tion is not necessary in most cases.
A typical MTBE composition adopted worldwide for use in the fuel sector is
given below:

MTBE 98–99 wt %
Alcohols (methanol, tert-butanol) 0.5–1.5 wt %
Hydrocarbons (C5, C6-hydrocarbons, 0.1–1.0 wt %
diisobutenes)
Water 50–1500 ppm by weight
Total sulfur max. 10 ppm by weight
Residue on evaporation max. 10 ppm by weight

For special applications (see Section 10.2), a high-purity MTBE (ether con-
tent > 99.98 % wt) is marketed under the trade name Driveron S by Oxeno. High
purity MTBE is also available from Shell and ARCO.
436 METHYL TERT-BUTYL ETHER Vol. 2

7. Chemical Analysis

Pure MTBE is analyzed by gas chromatography, preferably in capillary col-


umns with a highly polar stationary phase, eg, TCEP [1,2,3-tris-(2-cya-
noethoxy)-propane], Carbowax 20 M, or DX-1. For GC analysis of MTBE-
containing fuels, an oxygen-specific detector (O-FID) (24) or a column combina-
tion technique (21,22) can be employed. For other references on analysis of
MTBE, see (23,24,25).

8. Storage and Transportation

Being noncorrosive, MTBE can be handled in the same way as fuel, and the exist-
ing distribution system can also be used for fuels containing MTBE. The ether
has an indefinite storage life even in the presence of air. Because water is mis-
cible with MTBE (though only slightly), dry storage is required. The product can
be stored in nonpressurized containers. Carbon steel can be used as container
material, as well as aluminum, brass, copper, polyethylene, or polypropylene.
Teflon, Buna-N, and other fuel-resistant plastics and rubbers can be used for
seals; the use of vinylidene fluoride-hexafluoropropene copolymer (Viton) is not
recommended.
Methyl tert-butyl ether has a vapor pressure of 61 kPa at 40 8C. Emission
from storage facilities can be controlled or prevented by ordinary measures.
The usual safety precautions for highly inflammable liquids must be employed.
Extinguishing agents for fire fighting are powder, carbon dioxide, and
alcohol-resistant foam such as Tutogen L. With alcohol-resistant foams, a higher
application rate is necessary for MTBE than for pure hydrocarbons.
Transport classifications are as follows:

ADR/RID II class 3, no. 3 b; Name of substance:


33/2398 Methyl-tert-butylether
ADNR II class 3, no. 3 b
GGVSee/IMDG class 3.1
UN no. 2398
ICAO Code/IATA-DGR II class 3
UN no. 2398

9. Legal Aspects

The principal reason for the current importance of MTBE is environmental leg-
islation (especially in the United States, Japan, and Western Europe). Although
laws do not require the use of MTBE, it is often the best way to satisfy their
provisions.
In the European Community, the Council Directive 85/536/EEC, December 5,
1985 (33), does not require special labeling at filling stations for fuels containing
up to 15 vol % MTBE.
Vol. 2 METHYL TERT-BUTYL ETHER 437

In the United States, the use of oxygenates in gasoline is governed by EPA


waivers from the Clear Air Act. Sun Refining received a provisional waiver in
1988, permitting the use of up to 15 vol% MTBE (2.7 wt% oxygen) (34).
Legislation has developed since oxygenates in fuel have become generally
recognized. To reduce carbon monoxide emission and comply with provisions of
the Clean Air Act, some U.S. cities and states mandate the use of ca. 2.7 wt% oxy-
gen in fuel (mainly as MTBE) in winter months.

10. Uses

10.1. As an Octane Enhancer. More than 95% of MTBE produced


are used in the gasoline pool. In the 1980s – 1990s huge production capa-
cities have been erected in order to cover the demand of MTBE. This process
has been driven especially by the CAA regulation in the United States. In
addition to many other favorable characteristics, the current importance
of MTBE is based primarily on its exceptionally good octane-enhancing
properties when used as a gasoline blendstock (28,31). These antiknock
properties are especially important because the use of cheap but toxic
alkyllead compounds has been restricted or banned by law, both on environ-
mental grounds and to permit the use of exhaust catalytic converters.
Depending on the composition of the base gasoline, blend octane numbers
of 115 – 125 (research octane number, RON) and 92– 120 (motor octane
number, MON) can be achieved. The highest blend values are obtained
with saturated, paraffin-rich gasolines; the lowest, with olefin-rich gasolines.
The lead and aromatics content also influences the blend octane numbers of
MTBE. Because of the relatively low boiling point of MTBE, the effect on the
front-end octane number (FON or R-100 8C-ON, ie, the research octane num-
ber of the fraction with a boiling range < 100 8C) is especially pronounced (see
Fig. 2) (29).
Besides increasing the octane number, addition of MTBE to fuel has other
positive effects. It does not necessitate any modifications to existing vehicles. The
fuel vapor pressure (Reid vapor pressure, RVP) is decreased, so that vapor emis-
sions during automobile fueling and operation are reduced. Addition of MTBE
cuts down exhaust emissions, particularly carbon monoxide, unburned hydrocar-
bons, polycyclic aromatics, and particulate carbon. Although MTBE has a some-
what lower heat of combustion than gasoline, addition of up to 20 vol% neither
impairs motor power nor increases fuel consumption. Easier cold starting and
prevention of carburetor icing are other advantages. With regard to the hot-
weather drivability of modern automobiles, no difference exists between fuels
containing MTBE and those without it (30). Because of the miscibility of water
with MTBE, the fuel’s cloud point is significantly lowered. The MTBE-blended
fuels are compatible with all materials used in automobile manufacture (eg,
gaskets; lacquers; metals in the carburetor, injection pump, or elsewhere; and
elastomers).
10.2. Other Uses. Methyl tert-butyl ether is also used in the petrochem-
ical industry. Production of isobutene by splitting MTBE is the only application
that has been used on an industrial scale.
438 METHYL TERT-BUTYL ETHER Vol. 2

Fig. 2. Ranges of octane improvement by addition of MTBE to an unleaded gasoline with


RON ¼ 88, MON ¼ 81, and R-100 8C-ON ¼ 77.

By reversal of its formation reaction, MTBE can be cracked to isobutene and


methanol on acidic catalysts at > 100 8C (32). The methanol obtained as a copro-
duct is recycled to MTBE synthesis. For economic and ecological reasons, the
MTBE-cracking route and that involving the splitting of tert-butanol obtained
by direct hydration of isobutene (33) will continue to displace the conventional
sulfuric acid extraction process for the isolation (recovery) of pure isobutene
from C4 streams (34). At present, isobutene production with MTBE as feedstock
is employed by Exxon Chemical and Sumitomo.
Methyl tert-butyl ether itself can be used in a number of chemical reactions,
for example, the production of methacrolein and methycrylic acid (35) and of iso-
prene (36). For other possible applications, see (37–42).
The lack of acidic hydrogen atoms makes MTBE a suitable solvent for che-
mical reactions such as Grignard reactions. The preferred grade for this purpose
is high-purity MTBE with a residual alcohol content < 500 ppm by weight, such
as the product marketed by Oxeno under the trade name Driveron S. Because of
its negligible tendency to form peroxides, high ignition temperature, and narrow
explosion limits, MTBE is a good solvent for analytical use (42). It is also used as
an extractant, for example, in solvent dewaxing of hydrocarbon oils (42,43).

11. Toxicology and Occupational Health

11.1. Toxicological Profile. As regards its widespread application


and the amounts used, MTBE ranges within the top 2% of all tested chemical
Vol. 2 METHYL TERT-BUTYL ETHER 439

substances. According to all of the toxicity data reviewed, in respect of European


Union regulations MTBE has been classified as a skin irritant.
Acute Toxicity. Tests with different animal species on oral, dermal, and
inhalation application routes indicate a low acute toxicity for MTBE (oral LD50
well above 2000 mg per kilogram bodyweight, dermal LD50 > 10 000 mg per kilo-
gram bodyweight, inhalation LC50 values in the range from 85–142 mg/L for 4 h).
At high concentrations anesthesia is the most immediate effect (CNS depres-
sion). At nonlethal concentrations local irritations, typical for organic solvents,
have been observed besides transient behavioral effects.
MTBE was found to be only slightly irritant to the rabbit eye but causes
reversible moderate to severe skin effects in rabbits. Animal studies showed no
sensitizing potential for neat MTBE.
Chronic Toxicity. The main effects observed in repeated dose toxicity
studies via the oral and inhalation routes were CNS depression, irritation of
the respiratory and gastrointestinal tract and enlargement of the liver. For the
oral route the no observed adverse effect level (NOAEL) in a 90 day test is
300 mg/kg bodyweight per day while it is 800 mL/m3 or 2880 mg/m3 in 90 days
inhalation tests.
Genotoxicity, Mutagenicity, Carcinogenicity. Neither the structure
of MTBE nor the comprehensive number of available in vitro and in vivo muta-
genicity tests with MTBE and its metabolite tert-butanol provide any indica-
tion of possible genotoxic or mutagenic effects. The other metabolite,
formaldehyde, has a proven potential of DNA damage, but will be rapidly
detoxified when formed in the body. Therefore available test data strongly
suggest that with regard to MTBE itself and its metabolites no genotoxic
activity should be assumed by MTBE exposure. The absence of genotoxic poten-
tial is important when discussing possible carcinogenic effects. MTBE produces
tumors in rats and mice at high concentrations. No observed adverse effect
concentrations (NOAEC) of 400 mL/m3 by inhalation and 250 mg/kg body-
weight at oral dose are derived from the tests results. The tumors
found have been formed via nongenotoxic mechanisms, and are most probably
not relevant to humans. Based on the available data several independent
bodies, such as the International Association of Research on Cancer (IARC,
WHO-agency), the European Union Working Group on Classification and
Labelling of Dangerous Substances, and other bodies recently have not come
to the conclusion that MTBE is a human carcinogen. Based on the available
data MTBE is not classifiable with respect to its carcinogenic potential for
humans.
Reproductive Toxicity. Tests on reproductive toxicity and developmen-
tal development shows that MTBE is not significantly toxic to reproduction nor
to fetal development.
11.2. Human Experience. As the odor threshold of MTBE in air
(0.053 mL/m3 = 0.19 mg/m3) as well as odor and taste thresholds in water
(15 mg/L and 40 mg/L, respectively) are very low, at low exposure levels effects
on humans may be transient and subjectively caused by taste and odor. At
180 mg/m3 no objective symptoms on the CNS and no signs of irritation of the
respiratory tract were reported. Although it is not reported, repeated exposure
on skin may result in de-fatting of the skin.
440 METHYL TERT-BUTYL ETHER Vol. 2

As indicated by all reported human experiences there is no evidence of any


adverse effect caused by low exposure levels.
11.3. Health Risk Characterization, Occupational Health. Since the
exposure concentrations for workers involved in production of MTBE, blending,
distribution and handling of gasoline containing MTBE, service-station atten-
dants and garage workers, and also for consumers filling MTBE containing gaso-
line in cars are significantly lower than the no effect levels in toxicity tests,
MTBE production and use indicates no concern for human health.
However, compliance by workers with the short- and long-term occupa-
tional exposure limits and protection against skin contact is necessary to be pro-
tective against the described effects of MTBE. The lowest recommended
occupational exposure limits are 270 mg/m3 (75 mL/m3) as the short-term and
90 mg/m3 (25 mL/m3) as the long-term value.

12. Ecotoxicology

12.1. Ecotoxicity Data. The result of a large amount of acute aquatic


toxicity test data available for organisms at different trophic levels indicates
low toxicity to aquatic organisms. All measured EC50 and LC50 values for micro-
organisms, algae, invertebrates and fish are well above 100 mg/L. Because of the
fact that the measured concentrations of MTBE in the environment are signifi-
cantly lower than the no effect levels in the mentioned tests and the calculated no
effect levels for sediment organisms and terrestrial organisms like earth worms,
it is verified that the risk for environmental organisms is low.
12.2. Behavior in the Environment. A calculation with McKay level 1
partitioning model estimates that in case of a spillage most of the MTBE (95%)
will distribute in the atmosphere. There MTBE is readily degraded by photoche-
mically produced hydroxyl radicals. The half-life for photodegradation in the air
is about 3–7 days.
Aerobic and anaerobic degradation in water and groundwater depends
strongly on the conditions. With inoculum of predominantly domestic sewage
nearly no biodegradation occurs within 28 days in water in a standard test.
However, with adapted bacteria MTBE is readily biodegradable under aerobic
conditions.
As MTBE is relatively easily soluble in water (about 50 g/L) and will not
adsorb on soil, it is easily washed into the groundwater. Biodegradation in con-
taminated soil and groundwater (remediation) is discussed in Chapter 13.
Due to its physical-chemical properties and metabolism in organisms
MTBE is not expected to bioaccumulate and this has been shown to be correct
by a bioaccumulation test.

13. Removal of MTBE from Impacted Groundwater

As a component of gasoline, MTBE can enter subsurface environments through a


number of release scenarios including leaking underground fuel tanks (LUFT),
dispensing equipment, accidental spills during tank refills, leaking fuel
Vol. 2 METHYL TERT-BUTYL ETHER 441

pipelines, minor on-going spill incidents, and larger volume surface spills. In
many cases, gasoline retail stations and production terminals are sources of
MTBE in groundwater. The concentration of MTBE in groundwater and the
extent of groundwater contamination depend largely on the volume and duration
of a gasoline release and whether remediation has been initiated at the site. As
with any remedial effort for most contaminants, source removal is critical for the
success of remediation systems at MTBE-impacted sites. A quick response fol-
lowing an accidental release of an MTBE-blended gasoline is of great importance
regardless of the selected remediation method since it decreases the size of the
area requiring characterization and the volume of groundwater needing treat-
ment thereby minimizing site characterization and remediation costs.
13.1. Overview of MTBE Remediation Technologies. The selection
of a remedial strategy at an MTBE-impacted site depends on the volume of con-
taminated soil and groundwater, contaminant concentrations and mass, site-spe-
cific hydrogeological characteristics, and risk-based or regulatory-driven cleanup
objectives. Of particular interest at MTBE-impacted sites is the behavior of
MTBE relative to other gasoline components in subsurface environments.
The properties of MTBE greatly influence its fate in the environment, gov-
ern transport mechanisms, phase partitioning and biological interactions, and
impact the success of remedial technologies. Key physiochemical properties of
MTBE were presented in Tables 2 and 3. As compared with toluene, a common
gasoline component and groundwater contaminant (! Toluene, Chap. 2.1.),
MTBE has a lower density and boiling point, but a higher vapor pressure and
much higher water solubility. The properties of MTBE suggest the following
regarding its fate in subsurface environments: (1) MTBE does not readily sorb
to soil particles in aquifers; (2) MTBE volatilizes readily from non-aqueous
phase liquid (NAPL) sources; and (3) MTBE strongly partitions into the aqueous
phase and preferentially remains in water. As a consequence, MTBE can be
highly mobile in subsurface environments relative to other gasoline constituents.
Contarary to early reports, most of the technologies used historically at
gasoline-contaminated sites are effective at removing MTBE from groundwater.
Successful technologies include groundwater extraction followed by air
stripping, advanced oxidation or granular activated carbon systems. In-situ tech-
nology options include air sparging, multi-phase extraction, soil vapor extrac-
tion, bioremediation and phytoremediation. In addition to single technologies
traditionally used at gasoline-contaminated sites, emerging technologies, modifi-
cations to existing technologies or the use of a treatment train involving more
than one technology can greatly improve MTBE removal efficiency and/or reduce
life cycle remediation costs.
13.2. Phase Transfer Technologies. Phase transfer technologies
move MTBE from the water phase to either the gas phase (eg, air stripping) or
the solid phase (eg, granular activated carbon) without chemically transforming
it. Consequently, additional treatment steps to destroy MTBE (eg, off-gas treat-
ment from air strippers) or waste disposal (eg, spent granular activated carbon)
may be required.
Ex-Situ Phase Transfer Technologies. Air Stripping. Air stripping
is a physical process for removing volatile organic compounds from water. In
general, the effectiveness of air stripping increases with a compound’s Henry’s
442 METHYL TERT-BUTYL ETHER Vol. 2

constant and the operating air-to-water ratio. MTBE’s Henry’s constant is much
lower than that of common groundwater contaminants such as toluene (0.018–
0.026 vs. 0.27, (66)). Thus, high air-to-water ratios (100 to 250) are needed to
remove MTBE from water. Air stripping has been used to successfully remove
MTBE at concentrations that are typically associated with groundwater contam-
ination from LUFTs (67). Evaluation of MTBE performance data from eight case
studies of low profile and packed tower air strippers showed that air stripping
has the potential to be more widely used in both drinking water and remediation
applications (68). For each of the studies evaluated, MTBE removal efficiencies
exceeded 90%. Because state and/or local air quality regulations sometimes
required stripper off-gas treatment (granular activated carbon, thermal and cat-
alytic oxidation, advanced oxidation), process costs can increase significantly.
Granular Activated Carbon (GAC). Although MTBE has a low affinity
for sorption to solids, several types of activated carbon including coconut shell or
coal-based GAC can be used to remove low concentrations of MTBE in water
(200 mg/L) cost-effectively (67,69). GAC can be used to treat low flows of water
from private wells with low MTBE concentrations. For treatment of higher con-
centrations of MTBE, GAC can be used as a polishing step in a treatment train
following air stripping or chemical oxidation (69). Background water quality and
co-contaminant concentrations influence MTBE removal efficiencies. High con-
centrations of natural organic matter (NOM) and other gasoline constituents
in influent water streams compete with MTBE for GAC sorption sites thereby
increasing GAC usage rates and process costs.
Other Ex-Situ Phase Transfer Technologies. Results from laboratory
and pilot scale studies suggest that synthetic resins have the potential for suc-
cessful remediation of MTBE-impacted sites (67). One of the advantages
of using synthetic resins is the ease of regeneration relative to GAC. One of
the limitations of resins is the experimentally observed reduction in MTBE sorp-
tion due to the presence of other gasoline constituents, such as benzene, toluene,
ethylbenzene, and xylene (BTEX) compounds and tert-butanol (TBA, tert-
butylalcohol), in contaminated water. This is analogous to the effects of NOM
and BTEX on GAC sorption as discussed above. The use of synthetic resins is
costly relative to GAC but resins can be designed to achieve a higher degree of
selectivity and trend to have a longer lifetime than GAC.
In addition to synthetic resins, other sorbents such as high silica zeolites
have been shown to be effective in removing MTBE from contaminated water (70).
Other technologies, such as membranes and solvent extraction, can be used
to remove MTBE from contaminated groundwater (71,72,73) but only a few
bench- and pilot-scale studies have been conducted.
In-Situ Phase Transfer Technologies. Demonstrated in-situ phase
transfer technologies include soil vapor extraction (SVE), multi-phase extraction
(MPE), in-situ air sparging (IAS) and phytroremediation. Emerging phase
transfer technologies include steam injection/vacuum extraction and six-phase
heating.
Soil Vapor Extraction (SVE) and Multi-Phase Extraction (MPE).
Soil vapor extraction (SVE) is commonly used for the remediation of vadose
zones at gasoline-contaminated sites. In order to remove MTBE from both
saturated and unsaturated zones, other remediation technologies can be used
Vol. 2 METHYL TERT-BUTYL ETHER 443

in combination with SVE. The success of SVE has been documented at MTBE-
impacted sites when implemented either as the sole technology or in conjunction
with other groundwater treatment technologies (74–76) The combination of SVE
and groundwater extraction, known as multi-phase extraction, is effective in
removing MTBE from contaminated aquifers (75) and is routinely used at
many sites. Effectiveness with MTBE can be enhanced if the remediation system
is operated under high vacuum.
In-Situ Air Sparging (IAS). In-situ air sparging involves the injection
of air below the groundwater table to transfer contaminants from the aqueous
phase into the gaseous phase. The injection of air has the additional advantage
of introducing oxygen to subsurface environments thereby potentially enhancing
aerobic contaminant biodegradation rates. Laboratory air sparging column stu-
dies in different soils showed that the rate of MTBE volatilization is very rapid in
sandy soils (77). Results from another laboratory study showed that 85% of
MTBE in water equilibrated with gasoline was successfully removed using air
sparging (78). MTBE recovery rates were shown to increase with increased air
injection rates and water saturation levels. Results from a number of field stu-
dies revealed that MTBE groundwater concentrations at gasoline-contaminated
sites were reduced by 88 to 99% after two years of air sparging (79). As a general
rule, if the site hydrogeology is conducive to successful use of IAS, e.g., soil por-
osity is satisfactory, this technology could be very successful at removing MTBE
from groundwater.
Phytoremediation. Phytoremediation involves the use of plants or trees
to either remove or degrade groundwater contaminants. Phytoremediation
occurs by plant uptake of the contaminants followed by translocation and accu-
mulation into plant tissues, or transpiration through leaves. In some cases, the
breakdown of contaminants can take place either within plant tissues or in the
rhizosphere (80). Phytoremediation is most effective for contaminants with log
Kow values in the range of 0.5 to 3.0 (81). Since MTBE has a log Kow of 1.2, phy-
toremediation is expected to be an effective technology at MTBE-impacted sites.
Several plant and tree species have been recently reported to uptake MTBE
including alfalfa plants (82), poplar trees (80,83,84) and Eucalyptus trees
(84,85). In laboratory studies, poplar trees were shown to remove 30% of the
MTBE in test solutions at concentrations of 300 to 1600 mg/L in a week (81).
Results of several laboratory studies suggest that phytoremediation has the
potential to be effective for the removal of MTBE from groundwater, especially
at sites where the contaminated groundwater is shallow.
13.3. Transformation or Destruction Technologies. The underlying
principle of transformation or destruction technologies is the transformation of
MTBE in water to another chemical form either partially (eg, acetone) or comple-
tely (carbon dioxide). Transformation or destruction technologies can either be
implemented ex-situ following groundwater extraction (eg, chemical oxidation,
biological treatment) or in-situ (eg, in-situ chemical oxidation, in-situ
bioremediation).
Ex-Situ Transformation Technologies. Advanced Oxidation Processes
(AOPs). Advanced oxidation processes involve the use of ozone-peroxide,
ozone – UV, peroxide – UV, Fenton’s reaction (hydrogen peroxide in com-
bination with iron(II)salts), high-energy electron beam irradiation (E-beam)
444 METHYL TERT-BUTYL ETHER Vol. 2

or sonication-hydrodynamic cavitation. Most of these processes are characterized


by the production of hydroxyl radicals. AOP technologies have been tested for
MTBE destruction in both laboratory and pilot-scale studies. In addition, some
full-scale applications have been reported (86–91). Common problematic bypro-
ducts include bromate due to the reaction of ozone with bromide, and TBA, tert-
butyl ether (TBE), acetone and formaldehyde from the use of UV–H2O2 (91).
Some of these compounds are either more toxic than MTBE or more resistant to
oxidation (87,88).
AOPs are believed to be most efficient for groundwater with MTBE concen-
trations ranging between 0.1 and 80 mg/L (87). Although the use of AOPs is effec-
tive at treating high concentrations of MTBE, AOP systems can be expensive and
can lead to the formation and accumulation of toxic or undesirable byproducts.
More research to evaluate the cost-effectiveness and performance of AOPs at
field-scale levels is needed.
Biological Treatment. Alkyl ethers, such as MTBE, are relatively diffi-
cult to degrade because of the high energy required by microorganisms to cleave
the ether bond and the resistance of the branched carbon structure to microbial
attack (92). Several early studies reported that MTBE is biologically recalcitrant
under most environmental conditions (93–95). Recent laboratory and field stu-
dies, however, have shown that a number of bacterial and fungal cultures from
various environmental sources are capable of degrading MTBE under aerobic
(96,97) and anaerobic (98–102) conditions. Microorganisms can either utilize
MTBE as a sole source of carbon and energy, or can degrade it cometabolically
following growth on short-chain alkanes such as propane.
The biodegradation of MTBE has been demonstrated using both suspended
growth and fixed film bioreactors (103–109). In a growth bioreactor microorgan-
isms were able to reduce MTBE concentrations from 2400 mg/L to 1.6 mg/L with
an average removal rate of 96.2% over the course of the study.
Several types of ex-situ reactors can be used for MTBE treatment including
fluidized bed, porous pot, membrane and trickling filter bioreactors. Membrane
bioreactors sustain high densities of microorganisms and are therefore suited for
treating high strength waste streams as well as waste streams containing slowly
biodegraded compounds such as MTBE. In one study, the use of a membrane
bioreactor reduced MTBE from high (2000 mg/L) to very low (40 g/L) concentra-
tions during normal operation (108). The potential for success of ex-situ bioreac-
tors has increased over the last several years with the isolation of pure cultures
capable of degrading MTBE and TBA at relatively rapid rates (96). For example,
two pure cultures designates PM1 and ENV735 are being used commercially in
fluidized bed reactors (FBRs). Such systems show great promise for biologically
treating MTBE contaminated water streams.
In-Situ Transformation Technologies. In-Situ Chemical Oxidation.
In-situ chemical oxidation (ISCO) is based on the same principles of advanced
oxidation. Most ISCO technologies involve the injection of oxidants into
groundwater to produce hydroxyl radicals. Common oxidants include peroxide,
permanganate, ozone, and Fenton’s reagent. In a successful field scale demon-
stration, hydrogen peroxide was injected at a site in New Jersey. MTBE
concentrations dropped from 6000 mg/L to below detection limits in less
than a year (111). Injection of Fenton’s reagent into groundwater at a site in
Vol. 2 METHYL TERT-BUTYL ETHER 445

Texas reduced MTBE concentrations by 83.4% (112). One of the advantages


of ISCO is the potential to destroy MTBE in a relatively short time frame.
Technology limitations include effective oxidant delivery and the potential of
byproduct formation.
In addition to the injection of ozone as an oxidant, technologies involving
ozone sparging are currently being tested. Fine bubbles with a high surface-to-
volume ratio are injected into the saturated zone to extract dissolved MTBE from
contaminated groundwater. Ozone contained within the bubble and thin film
around the bubble is expected to react rapidly and destroy MTBE in the gas
phase.
In-Situ Bioremediation ðsee also BioremediationÞ. In-situ bioreme-
diation involves the use of indigenous or exogenous microorganisms to either
destroy or immobilize contaminants. In-situ bioremediation strategies could
involve direct metabolism, cometabolism, bioaugmentation or some combination
thereof. Results from pilot studies showed that bioremediation approaches invol-
ving oxygen injection and/or bioaugmentation have a strong potential for success
at MTBE-impacted sites (110,113). In addition, a recent study has demonstrated
the success of propane injection and bioaugmentation at an operating service
station in New Jersey where a 93% reduction in MTBE concentration was
achieved within two months of system operation (108).
Fortum Oil and Gas Oy and Oxeno Olefinchemie GmbH, two member com-
panies of EFOA (The European Fuel Oxygenates Association — www.efoa.org)
have been contributing to the Chapters 11, 12, and 13.

REFERENCES

1. C. Henry, Chem. Zentralbl. (1904) 1065.


2. H. C. Barnett et al., Natl. Advis. Comm. Aeronaut., Wartime Report ACR No. E 4
H 03 (1944).
3. Ullmann, 4th ed., 17, 58.
4. K. C. Rock, R. O. Dunn: ‘‘Economics of Producing Butene Intermediate Feedstocks
from Butanes’’, Energ. Prog. 8 (1988) no. 4, 191–195.
5. D. J. Miller: ‘‘Arco’s Processes for Oxygenated Fuels–MTBE/TAME and ETBE’’,
Dewitt 1989 Petrochemical Review, Houston, March 28–30, 1989.
6. F. Ancilotti et al., J. Mol. Catal. 4 (1978) 37–48.
7. A. Gicquel, B. Torck, J. Catal. 31 (1983) 27–31.
8. J. M. Adams et al., Clays Clay Miner. 34 (1986) no. 5, 597–609.
9. F. Ancilotti et al., J. Catal. 46 (1977) 49–57.
10. F. Obenaus: ‘‘Wege zur besseren Nutzung der C4-Kohlenwasserstoffe’’, DGMK-
Meeting Karlsruhe, February 8, 1982.
11. Th. A. Beck, D. J. Miller: Arco MTBE–The Maximum in Flexibility, Dewitt 1988,
Petrochemical Review, Houston, TX., March 23–25, 1988.
12. P. H. O. Dixon et al., Catalytic Destillation Technology and MTBE-Production, 1989
NPRA Annual Meeting, San Francisco, March 19–21, 1989.
13. DEA (formerly Deutsche Texaco), DE 27 06 465, 1987 (H. Humberg, H. Wegner).
14. Shell NL, DE-OS 29 11 077, 1979 (H. Groeneveld, A. Kwantes).
15. Sumitomo Chemical Company, EP-A 0 206 594, 1986 (S. Arakawa et al.).
16. CR & L, EP-A 0 008 860, 1979 (L. A. Smith Jr.).
446 METHYL TERT-BUTYL ETHER Vol. 2

17. E. Trubac: ‘‘Oxygenate Removal from MTBE Raffinate’’, lecture presented at the
AIChE Meeting, Minneapolis, Minn., Aug. 16–19, 1987.
18. Union Carbide Corp., EP-A 0 229 994, 1987 (M. Nagji).
19. Hüls AG, DE 2 629 769, 1976 (F. Obenaus et al.).
20. W. Schneider, J.-Ch. Frohne, H. Bruderreck, J. Chromatogr. 245 (1982) 71–83.
21. J. Winskowski, Erdöl Kohle Erdgas Petrochem. 41 (1988) no. 1, 30–36.
22. J. M. Levy, J. A. Yancey, J. High Resolut. Chromatogr. Commun. HRC CC J. High
Resolut. Chromatogr. 9 (1986) no. 7, 383–387.
23. AFNOR (Association Française de Normalisation) M 07–084, Dec. 1984, Norme
experimentale.
24. DIN 51 413, Parts 4 and 5.
25. BSI (British Standards Institution) proposed method JP-STG-6 C (The Institute of
Petroleum).
26. EEC Council Directive 85/536/EEC, Dec. 5, 1985 (Column 3).
27. Chem. Eng. News, Oct. 24, (1988) 12.
28. G. H. Unzelmann, Oil Gas J. (1989, April 10) 33–37.
29. F. Obenaus, W. Droste, Erdöl Kohle Erdgas Petrochem. 33 (1980) no. 6, 271–275.
30. M. Denham, M. Mays, K. Schug, M. Sposini, lecture presented at the 8th Interna-
tional Symposium on Alcohol Fuels, Nov. 13–16, Tokyo 1988.
31. W. J. Piel, Energ. Prog. 8 (1988) no. 4, 201–204.
32. Fattore et al., Hydrocarbon Process. 60 (1981) Aug., 101. August (1981) 101.
33. Chem. Eng. N.Y. 90 (1983) no. 25, 60.
34. H. Kröper, K. Schlömer, H. M. Weitz, Hydrocarbon Process 48 (1969) no. 9, 195–198.
35. Nippon Shokubai Kagaku Kogyo Co., EP-A 0 304 867, 1988 (K. Kinumi et al.).
36. Kuraray Co, JP-Kokai 60 193 932 <85 193 932>, 1985 (Y. Ninagawa et al.). Sumi-
tomo Chem., US 3 574 780, 1971 (Y. Watanabe et al.).
37. Degussa AG, EP 0 050 869, 1981 (B. Lehmann).
38. Anic S.p.A., DE-OS 3 130 428, 1982 (F. Ancillotti et al.). BASF AG, DE-OS
27 40 590, 1979 (F. Merger et al.).
39. BASF AG, DE-OS 29 11 466, 1980 (H. Naarmann et al.).
40. BASF AG, DE-OS 29 28 098, 1981 (W. Hoffmann).
41. Bayer AG, DE-OS 30 02 203, 1981 (G. Bonse et al.).
42. H. Grangette, C. Bernasconi, S. Randriamahefa, R. Gallo, Inf. Chim. 246 (1984)
Jan./Feb., 135–140.
43. Exxon Research and Engineering Co., EP-A 00 88 603, 1983 (T. H. West).
44. Chem. Week (1997) Dec. 3, 60.
45. M. L. Morgan, DGMK Conference ‘‘C4 Chemistry–Manufacture and Use of
C4-Hydrocarbons’’, Oct 6–8 (1997) Aachen.
46. K. D. Miller, 23rd Annual Dewitt Petrochemical Review, March 31–April 2 (1998)
Houston.
47. P. R. Sarathy, G. S. Suffride, Hydrocarbon Process. (1993) Feb, 43–51.
48. Chem. Week (1997) Oct. 29, 44.
49. P. Chu, G. H. Kühl, Ind. Eng. Chem. Res. 26 (1987) 366.
50. A. Kogelbauer et al., J. Catal. 152 (1995) 122–129.
51. K. Sundmacher, U. Hoffmann, Chem. Eng. Sci. 49 (18) (1994) 3077–3089.
52. T. Zhang, R. Datta, Ind. Eng. Chem. Res. 34 (1995) 730–740.
53. D. Parra et al., Chem. Eng. Sci. 49 (24A) (1994) 4563–4578.
54. A. Rehfinger, U. Hoffmann, Chem. Eng. Sci. 45 (1990) no. 6, 1605–1617.
55. F. Obenaus, Oil Gas J. 77 (1979) 76–77.
56. A. Rehfinger, U. Hoffmann, Chem. Eng. Technol. 13 (1990) 150–156.
57. S. Rossini et al., DGMK-Conference ‘‘C4 Chemistry–Manufacture and Use of
C4-Hydrocarbons’’, Oct. 6–8 (1997), Aachen.
Vol. 2 METHYL TERT-BUTYL ETHER 447

58. Hydrocarbon Process. 72 (1996) Nov., 113.


59. W. Droste, Tech. Mitt. 80 (1987) 277–282.
60. Hydrocarbon Process. 75 (1996) Nov., 114.
61. L. A. Smith, M. N. Huddleston, Hydrocarbon Process. (1982) March, 121–123.
62. J. L. DeGarmo et al., Chem. Eng. Prog. (1992), March, 43–50.
63. Koch Engineering Company, EP-A 0 428 265 , 1991 (A. P. Gelbein, M. Buchholz).
64. Gebr. Sulzer AG, EP 0 396 650 , 1992 (R. Shelden, J.-P. Stringaro).
65. DE-OS 19 701 045 , 1998 (A. Górak, L. Kreul).
66. J. Zogorski, A. Morduchowitz, A. Baehr, B. Baumann, D. Conrad, R. Drew, N. Korte,
W. Lapham, J. Pankow, E. Washington: Fuel Oxygenates and Water Quality.
Coordinated by Interagency Oxygenation Fuel Assessment, Office of Science and
Technology Policy, Executive Office of the President, Washington, DC, 1997.
67. California MTBE Research Partnership: Treatment technologies for removal of
methyl tert-butyl ether from drinking water. National Water Research Institute,
Fountain Valley, CA 2000.
68. R. Deeb, S. Sue, D. Spiers, M. Kavanaugh, Proceedings of the Air and Water Man-
agement Association’s 94th Annual Conference and Exhibition, Orlando, FL 2001.
69. California MTBE Research Partnership: Treating MTBE-impacted drinking water
using granular activated carbon. National Water Research Institute, Fountain Val-
ley, CA (2002).
70. M. Anderson, Environ. Sci. Technol. 34 (2000) no. 4, 725–727.
71. L. Vane, F. Alvarez, B. Mullis, Environ. Sci. Technol. 35 (2001) no. 2, 391–397.
72. A. Keller, B. Bierwagen, Environ. Sci. Technol. 35 (2001) no. 9, 1875–1879.
73. J. Choi, I. Song, W. Lee, J. Membr. Sci. 166 (2000) 159–175.
74. D. Creek, J. Davidson, Proceedings of the 1998 Petroleum Hydrocarbons and
Organic Chemicals in Ground Water: Prevention, Detection, and Remediation,
National Ground Water Association, Westerville, OH 1998, pp. 560–568.
75. P. Johnson, R. Ettinger, in C. Ward, J. Cherry, M. Scalf (eds.): Subsurface Restora-
tion, Ann Arbor Press, Inc., Chelsea, MI 1997.
76. United States Environmental Protection Agency: MTBE fact sheet #2: Remediation
of MTBE contaminated soil and groundwater. EPA 510-F-98–002, 1998.
77. A. Mortensen, K. Jensen, T. Sonnenborg, E. Arvin, Groundwater Monit. Rem. (2000)
87–95.
78. C. Bruce, C. Gilbert, R. Johnson, P. Johnson, in G. Wickramanayake, R. Hinchee
(eds.): Physical, Chemical, and Thermal Technologies: Remediation of Chlorinated
and Recalcitrant Compounds, Battelle Press, Columbus, OH 1998, pp. 293–298.
79. California MTBE Research Partnership. Overview of MTBE remediation technolo-
gies. National Water Research Institute, Fountain Valley, CA 2002.
80. E. Rubin, A. Ramaswami, Water Res. 35 (2001) no. 5, 1348–1353.
81. J. Burken, J. Schnoor, Environ. Sci. Technol. 32 (1998) no. 21, 3379–3385.
82. Q. Zhang, L. Davis, L. Erickson, Environ. Sci. Technol. 35 (2001) no. 4, 725–731.
83. M. Hong et al., Environ. Sci. Technol. 35 (2001) no. 6, 1231–1239.
84. L. Newman et al., Prepr. Extend. Abstr. Am. Chem. Soc. 40 (2000) no. 1, 288–289.
85. L. Newman et al., J. Soil. Contam. 7 (1998) no. 4, 531–542.
86. J. Acreo et al., Environ. Sci. Technol. 35 (2001) no. 21, 4252–4259.
87. S. Cater, M. Stefan, J. Bolton, A. Safarzadeh-Amiri, Environ. Sci. Technol. 34 (2000)
no. 4, 659–662.
88. P. Chang, T. Young, Water Res. 34 (2000) no. 8, 2233–2240.
89. J. Kang, H. Hung, A. Lin, M. Hoffman, Environ. Sci. Technol. 33 (1999) no. 18,
3199–3205.
90. A. Safarzadeh-Amiri, Water Res. 35 (2001) no. 15, 3706–3714.
91. M. Stefan, J. Mack, J. Bolton, Environ. Sci. Technol. 34 (2000) no. 4, 650–658.
448 METHYL TERT-BUTYL ETHER Vol. 2

92. G. White, N. Russell, E. Tidswell, Microbiol. Rev. 60 (1996) 216–232.


93. Y. Fujiwara, T. Kinoshita, H. Sato, I. Kojima, Yugagaku 33 (1984) 111–114.
94. H. Jensen, E. Arvin, in F. Arendt, M. Hinsenveld, W. van den Brink (eds.): Contami-
nated Soil ’90, Kluwer Academic Publishers, Dordrecht 1990, 445–448.
95. J. Suflita, M. Mormile, Environ. Sci. Technol. 27 (1993) 976–978.
96. R. Deeb, K. Scow, L. Alvarez-Cohen, Biodegradation 11 (2000) no. 2–3, 171–186.
C. Yeh, J. Novak, Water Environ. Res. 67 (1995) 828–834.
97. A. Stocking et al., Biodegradation 11 (2000) no. 2–3, 187–201.
98. P. Bradley, F. Chapelle, J. Landmeyer, Appl. Environ. Microbiol. 67 (2001) no. 4,
1975–1978.
99. P. Bradley, F. Chapelle, J. Landmeyer, Environ. Sci. Technol. 35 (2001) no. 23,
4643–4647.
100. K. Finneran, D. Lovley, Environ. Sci. Technol. 35 (2001) no. 9, 1785–1790.
101. J. Landmeyer, F. Chapelle, H. Herlong, P. Bradley, Environ. Sci. Technol. 35 (2001)
no. 6, 1118–1126.
102. United States Environmental Protection Agency: Natural attenuation of MTBE in
the subsurface under methanogenic conditions. EPA/600/R-00/006, 2000.
103. K. Acuna-Askar, A. Englande, C. Hu, G. Jin, Water Sci. Technol. 42 (2000) no. 5–6,
153–161.
104. D. Dupasquier, S. Revah, A. Auria, Environ. Sci. Technol. 36 (2002) no. 2, 247–253.
105. N. Fortin, M. Deshusses, Environ. Sci. Technol. 33 (1999) no. 17, 2980–2986.
106. N. Fortin, M. Deshusses, Environ. Sci. Technol. 33 (1999) no. 17, 2987–2991.
107. P. Hatzinger, R. Steffan, S. Drew, Contaminated Soil, Sediment and Water Journal,
AEHS, 2001, 81–84.
108. R. Steffan, P. Hatzinger, Y. Farhan, S. Drew, Proceedings of the Petroleum Hydro-
carbons and Organic Chemicals in Ground Water. November 14–16, Houston, Texas
2001, pp. 252–263.
109. S. Vainberg, A. Togna, P. Sutton, R. Steffan, Submitted to J. Env. Eng. 2001, in press.
110. J. Salanitro et al., Environ. Sci. Technol. 34 (2000) no. 19, 4152–4162.
111. United States Environmental Protection Agency. Field applications of in-situ reme-
diation technologies: Chemical oxidation. EPA 542-R-98–008, 1998.
112. J. Leethem, Contaminated Soil, Sediment and Water Journal, AEHS, 2001, 54–58.
113. R. Wilson, D. Mackay, K. Scow, Environ. Sci. Technol. 36 (2002) no. 2, 190–199.

UDO PETERS
FRANZ NIERLICH
EKKEHARD SCHULTE-KÖRNE
Oxeno Olefinchemie GmbH, Marl,
Federal Republic of Germany
MICHAEL SAKUTH
Hüls Infracor GmbH, Marl,
Federal Republic of Germany
RULA DEEB
MARYLINE LAUGIER
MICHAEL KAVANAUGH
Malcolm Pirnie, Inc., Emeryville, USA
MARTTI SUOMINEN
Fortum Oil and Gas Oy, Finland
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 449

COMBUSTION SCIENCE
AND TECHNOLOGY
1. Introduction

Fuel combustion is a complex process, the understanding of which involves


knowledge of chemistry (structural features of the fuel), thermodynamics (feasi-
bility and energetics of the reactions), mass transfer (diffusion of fuel and oxidant
molecules), reaction kinetics (rate of reaction), and fluid dynamics of the process.
Therefore, the design of combustion systems involves utilizing information and
data generated in a range of disciplines. Often, the design of practical combus-
tion systems is based on experience rather than on fundamental mechanistic
understanding. However, for certain fuels such as methane, the combustion
mechanism is better understood than for other more complex fuels such as
coal. To accommodate the variety of approaches used to solve practical problems,
this article is divided into two subsections: combustion science and combustion
technology.

2. Combustion Science

2.1. Definitions and Terminology. Higher Heating Value. The heat-


ing value of a fuel is the amount of heat released during its combustion and is
expressed in two forms: higher heating value (HHV) or gross calorific value
and lower heating value (LHV) or net calorific value. The higher heating value
of a fuel is the heat of combustion at constant pressure and temperature(usually
ambient) determined by a calorimetric measurement in which the water formed
by combustion is completely condensed (1–3). The lower heating value is the
similarly measured or defined heat of combustion in the absence of water con-
densation. The higher heating value is most often used in combustion
and flame calculations. The difference between HHV and LHV is numerically
equal to the corresponding enthalpy difference, due to latent heat of vaporiza-
tion. For example, in the complete combustion of methane with O2 at 298 K,
CH4 ðgÞ þ 2 O2 ðgÞ ! CO2 ðgÞ þ 2 H2 O ðgÞ, for one mole of fuel the enthalpy
change is H ¼ 802:1 kJ=mol. This value is equal to the lower heating value
of methane, and the higher heating value for the above reaction is obtained by
adding the heat of vaporization of water at 298 K (44.01 kJ/mol). Thus the higher
heating value of methane is 802:1 þ 2ð44:01Þ ¼ 890:1 kJ=mol of CH4 (212.7 kcal/
mol). The evaluation of DH for use in thermochemical calculations is more gen-
erally performed with the use of tables such as the Joint Army Navy Air Force
(JANAF) Tables.
FAR or AFR. The composition of a mixture of fuel and air or oxidant is
often specified according to the Fuel to Air Ratio (FAR), and can be expressed
on a mass, molar, or volume basis. The FAR is normalized to the stoichiometric
composition by defining the equivalence ratio f as in equation 1, where
mf ¼ mass of fuel, kg; and mo ¼ mass of oxidizer, kg.

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.0315130219080118.a01
450 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

Vapor pressure
curve Nonflammable

Φ
Rich

Φ=1 Explosion
Mist combustion Flammable
possible
Lean

Nonflammable
TL TU AIT
Temperature

Fig. 1. Effect of temperature on limits of flammability of a pure liquid fuel in air, where
TL ¼ lean (or lower) flash point; TU ¼ rich (or upper) flash point; and AIT ¼ auto ignition
temperature (4,5).



mf
m
f ¼  o ð1Þ
mf
mo stoich

If f < 1, the mixture is said to be lean (in fuel) and the products of combustion
contain unreacted or excess O2. If f > 1 the mixture is said to be rich (in fuel) and
the products of combustion contain CO and possibly H2 because of incomplete
combustion caused by the oxygen deficiency (1–3).
Flammability Limits. Any given mixture of fuel and oxidant is flam-
mable (explosive) within two limits referred to as the upper (rich) and lower
(lean) limits of flammability (Fig. 1). Most mixtures are flammable when the
fuel to air volume ratio lies between 50 and 300% of the stoichiometric ratio
(2). The stoichiometric ratio is the exact theoretical ratio of fuel to air required
for complete combustion. The flammability region widens with increasing
temperature, and usually with increasing pressure, although the effect of
pressure is less predictable. The region within the flammability limits of a
fuel – air mixture can be divided into two subregions, the slow oxidation region
and the explosion region, separated by the spontaneous ignition tempe-
rature. To determine this temperature for liquid fuels, standard tests are
used in which the liquid fuel is dropped into an open-air container heated to
a known temperature. The lowest temperature at which visible or audible evi-
dence of combustion is observed is defined as the spontaneous ignition
temperature (1).
Flash Point. As fuel oil is heated, vapors are produced which at a certain
temperature ‘‘flash’’ when ignited by an external ignition source. The flash point
is the lowest temperature at which vapor, given off from a liquid, is in sufficient
quantity to enable ignition to take place. The flash point is in effect a measure
of the volatility of the fuel. The measurement of flash point for pure liquids is
relatively straightforward. However, the measured value may depend slightly
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 451

on the method used, especially for liquid mixtures, since the composition of the
vapor evolved can vary with the heating rate. Special problems arise when
the liquid contains a mixture of fuels and/or fuels and inhibitors. In these
cases the vapor composition will be different from the liquid composition. Conse-
quently, successive tests with the same sample can lead to erroneous results
because the composition of the liquid in the sample holder changes with time
as a result of fractional distillation (4–7).
Ignition. To understand the phenomenon of ignition it is necessary to con-
sider the following concepts: ignition source, gas temperature, flame volume, and
presence of quench wall surfaces. In general, there are two main methods of
igniting a flammable mixture. In the self-ignition method, the mixture is heated
slowly so that the vapor released as the temperature is raised ignites sponta-
neously at a particular temperature. In the forced ignition method, a small quan-
tity of combustible mixture is heated by an external source and the heat released
during the combustion of this portion results in propagation of a flame. The
external ignition source can be an electric spark, pilot flame, shock wave, etc.
For ignition to take place the following conditions should be satisfied: (1) the
amount of energy supplied by the ignition source should be large enough to over-
come the activation energy barrier; (2) the energy released in the gas volume
should exceed the minimum critical energy for ignition; and (3) the duration of
the spark or other ignition source should be long enough to initiate flame propa-
gation, but not too long to affect the rate of propagation. Ignition models fall into
two categories: the thermal model explains the ignition as resulting from supply-
ing the mixture with the amount of heat sufficient to initiate reaction. In the che-
mical diffusion model the main role of the ignition source is attributed to the
formation of a large number of free radicals in the preheat zone, where their dif-
fusion to the surrounding region initiates the combustion process. The thermal
model is applied more widely in the literature and shows better agreement with
experimental data.
At ambient temperatures, reaction rates for gaseous mixtures of fuel and
oxidant are extremely slow. As temperature is increased gradually, slow oxida-
tion begins and as a result of the exothermic reactions that occur, the tempera-
ture keeps increasing. With further temperature increase, the reaction rate
suddenly increases causing rapid combustion reactions to occur. This is provid-
ing, of course, that the rate of heat release is greater than the rate of heat loss
through the container walls. The temperature at which the heat released by the
reaction exceeds the heat loss is commonly referred to as the ignition tempera-
ture. The spontaneous ignition temperature (SIT), on the other hand, is the low-
est temperature at which ignition occurs. Increasing the pressure results in a
decrease in the spontaneous ignition temperature (1) (Fig. 2 and Table 1). This
temperature is highly dependent on the material of construction, apparatus con-
figuration, and test procedure, therefore reported test values vary widely (2,6,7).
From the chemical diffusion model standpoint, the usual values of ignition
temperatures of mixtures often do not correlate well with other flame properties.
Nevertheless, the following highly simplified, even speculative, qualitative des-
cription may be useful in thinking about flames. Early in the ignition or induc-
tion period, the attack of fuel by O2 is initially slow, but generates free radicals
452 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

Cool flames

Ignition delay
10 3 10 2 10 1 10 –1

Pressure
T1 T2 T3

Explosion
Slow oxidation
P3
P2
Long delay
explosion
P1
Temperature,°C

Fig. 2. Slow oxidation, spontaneous ignition, and explosion as a function of pressure and
temperature variations in hydrocarbon mixtures (1).

(OH , H, O, HO2, hydrocarbons) and other intermediate species (CO, H2, and
partial oxidation and decomposition products of hydrocarbons, if present). For
some time, there may be little or no temperature rise, the energy essentially
being stored in the free radicals. In this stage the reactions may be similar to
those that occur in very slow and nearly isothermal oxidation and cool flames
observed in very rich hydrocarbon–O2 mixtures, usually at lower temperatures
and pressures. These reactions often stop at the production of stable oxidation
and decomposition products, such as aldehydes, peroxides, lower hydrocarbons,

Table 1. Spontaneous Ignition Temperaturesa


Fuel SIT, K
propane 767
butane 678
pentane 558
hexane 534
heptane 496
octane 491
nonane 479
decane 481
hexadecane 478
isooctane 691
kerosine (JP-8 or Jet A)b 501
JP-3 511
JP-4 515
JP-5 506
a
Ref. 1.
b
See Aviation and other gas turbine fuels.
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 453

etc, but under some conditions may lead to explosions. In a variety of chain
reactions, the fuel and any intermediates are rapidly attacked by radicals,
some of which are also undergoing very fast chain-branching reactions. Evi-
dence for the behavior of such free radicals is provided by studies of the
mechanism of H2/O2 and similar explosions. These free radicals can multiply
rapidly and attain high transient concentrations, initiating chain reactions
that can eventually lead to the explosion of the flammable mixture. They are
eventually consumed to form stable species by three-body recombination reac-
tions in which most of the heat release occurs. An example of such a reaction,
where M is any atom or molecule in the gas, is represented by

H þ OH þ M ! H2 O þ M H ¼ 498 kJ=mol ð119 kcal=molÞ

Numerous other reactions of this type, including a sequence involving HO2,


exist and the relative importance of each reaction depends on the mixture com-
position. The rates of these termolecular processes increase with increasing
pressure but have little or no temperature dependence.
When the partial pressures of the radicals become high, their homogeneous
recombination reactions become fast, the heat evolution exceeds heat losses, and
the temperature rise accelerates the consumption of any remaining fuel to pro-
duce more radicals. Around the maximum temperature, recombination reactions
exhaust the radical supply and the heat evolution rate may not compensate for
radiation losses. Thus the final approach to thermodynamic equilibrium by
recombination of OH, H, and O, at concentrations still many times the equili-
brium value, is often observed to occur over many milliseconds after the maxi-
mum temperature is attained, especially in the products of combustion at
relatively low (<2000 K) temperatures.
The radicals and other reaction components are related by various equili-
bria, and hence their decay by recombination reactions occurs in essence as
one process on which the complete conversion of CO to CO2 depends. Therefore,
the hot products of combustion of any lean hydrocarbon flame typically have a
higher CO content than the equilibrium value, slowly decreasing toward the
equilibrium concentration (CO afterburning) along with the radicals, so that
the oxidation of CO is actually a radical recombination process.
In most hydrocarbon flames there are many radical consumption reac-
tions that interfere with the chain-branching reactions, and the peak radical
concentrations are therefore always lower than in analogous flames of H2, H2/
CO, or moist CO. Consequently, the overall reaction rates for hydrocarbon
flames at a given temperature are lower than the overall reaction rates for
H2, H2/CO, or moist CO flames. In general, variations in the concentrations
of such radicals are believed to interfere with the flame chemistry and can
account for inhibition of flames by various additives, notably halogen-
containing substances. Such additives can narrow flammability limits and
reduce burning velocities, even when they cause little or no reduction in the
ignition temperature. Those additives containing bromine are particularly
effective, and brominated organic compounds are extensively used in extin-
guishing devices to suppress unwanted diffusion flames (fires). Although
most of the chemistry summarized here has evolved from the study of flames
454 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

and explosions in premixtures, it also applies semiquantitatively to many


nonpremixed systems.
The fundamental parameters in the two main methods of achieving ignition
are basically the same. Recent advances in the field of combustion have been in
the development of mathematical definitions for some of these parameters. For
instance, consider the case of ignition achieved by means of an electric spark,
where electrical energy released between electrodes results in the formation of
a plasma in which the ionized gas acts as a conductor of electricity. The electrical
energy liberated by the spark is given by equation 2 (1), where V ¼ the potential,
V; I ¼ the current, A;  ¼ the spark duration, s; and t ¼ time, s.
Z Q
E¼ VI dt ð2Þ
0

The electrical energy is rapidly transformed into thermal energy, and because
the temperature of the ionized gas is generally above 300 K, the ignition
delay time is short compared with the spark duration, . Ignition only takes
place if the electrical energy exceeds the critical value, Ec, and if this energy is
liberated within a critical volume, vc (1,8,10,11).
Ignition can also be produced by a heated surface. During the process of
heat transfer from a hot surface to a flammable mixture, reactions are initiated
as the temperature rises and the combination of additional heat transfer from
the surface and heat release by chemical reactions can lead to ignition of the
mixture.
In experiments in which a heated cylinder was introduced into a propane–
air mixture, it was found that approximately 11 ms were required for ignition.
During this time, heat released by chemical reactions is negligible. This time
interval before ignition is referred to as the ignition delay or induction time
and is a function of the physical properties of the fuel, the rate of heat conduc-
tion, and the pressure of the combustion chamber (1,3,8,10,11). Ignition is char-
acterized by a rapid change in temperature (eq. 3), where tignition ¼ ignition delay
time, s; Ea ¼ activation energy, J/mol; R ¼ universal gas constant, J/(K  mol);
and Ti ¼ ignition temperature, K.
 
Ea
tignition a exp ð3Þ
RTi

For liquid fuels, ignition delay times are of the order 50 ms at 700 K and 10 ms
at 800 K. At low temperatures most of the ignition delay is the result of slow,
free-radical reactions, and a distinction between the initiation and explosion
periods within the ignition delay time can be made. With increasing ignition
temperature for a given mixture, these times become comparable and at
temperatures as high as 1500 K, both times may be of the order of 104 s. Con-
sequently, the reaction zone in the flame of a mixture is observed to be one
continuous event (12–14).
Another important concept is that of the critical ignition volume. During
the propagation of the combustion wave, the flame volume cannot conti-
nually grow beyond a critical value without an additional supply of energy.
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 455

The condition that controls the critical volume for ignition is reached when the
rate of increase of flame volume is less than the rate of increase of volume of the
combustion products. In this condition a positive exchange of heat between the
flame and the fresh mixture is achieved.
For a point spark source, the flame volume is initially spherical and the cri-
tical ignition volume is determined by calculating the rate of change of flame
volume with respect to radius compared to the rate of change of volume of the
combustion products (eq. 4)
   
d 4  d 4
p ðr þ eÞ3 r3  p r3 ð4Þ
dr 3 dr 3

where r ¼ radius of flame, m; e ¼ thickness of flame front, m; and rc ¼ radius of


the critical spherical volume for ignition, m. This gives equations 5 or 6.

2 er þ e2  r2 ð5Þ
pffiffiffi
rc ¼ e þ 2 ð6Þ

For a line spark source, the flame volume is initially cylindrical with the
cylinder length equal to the separation distance between the electrodes. Thus,
for a cylindrical flame, rc ¼ e, and the critical ignition volumes are equation 7
for a spherical flame and equation 8 for a cylindrical flame where vc ¼ critical
ignition volume, m3/kg; e ¼ thickness of flame front, m; and d ¼ flame height, m.
 pffiffiffi3
4pe3 1 þ 2
c ¼ ð7Þ
3

vc ¼ pe2 d ð8Þ

In order to control and monitor the amount of energy required to achieve


ignition, the concept of the minimum ignition energy as the smallest quantity
of energy that will ignite a mixture is defined (15–17). The minimum ignition
energy is calculated using the assumption that sufficient energy must be sup-
plied from the exterior to the critical volume, vc, to raise the mixture from its
initial temperature, Ti, to the flame temperature, Tf (1). The critical energy is
given by equation 9: where E ¼ critical energy, J; vc ¼ critical ignition volume,
m3/kg; r ¼ density, kg/m3; cp ¼ specific heat capacity at constant pressure,
J/(kg  K); Tf ¼ flame temperature, K; and Ti ¼ initial temperature, K.
 
E ¼ vc r cp Tf  Ti ð9Þ

Because of heat being transferred to the exterior of the critical volume, only
a fraction of the total spark energy, E, is utilized as ignition energy.
Minimum ignition energies are determined experimentally by means of a
combustion bomb with two electrodes. Both the ignition energy and the quen-
ching distance can be varied in this device. By using high performance (usually
456 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

air-gap) condensers, the majority of the stored energy will appear in the spark
gap. The stored energy can be calculated using equation 10 (1,2,9), where
E ¼ total spark energy released/stored, J; C ¼ capacity of the condenser; and
V ¼ voltage just before the spark is passed through the gas, V.

CV 2
E¼ ð10Þ
2

The spark must always be produced by a spontaneous breakdown of the gas


because an electronic firing circuit or a trigger electrode would either obviate the
measurement of spark energy or grossly change the geometry of the ignition
source (1,2,9).
Once ignition has been achieved, flame propagation can be controlled. This
is achieved by the use of the concept of the quenching distance, defined as the
minimum orifice diameter, wall separation, or mesh spacing just sufficient to
prevent flame propagation (Fig. 3). Experiments carried out on spark ignition
of mixtures between parallel plates show that, with sufficiently powerful sparks,
a flame develops in the immediate neighborhood of the spark but does not propa-
gate through the mixture unless the plates are separated by more than the
quenching distance (1,2).
Cool Flames. Under particular conditions of pressure and temperature,
incomplete combustion can result in the formation of intermediate products
such as CO. As a result of this incomplete combustion, flames can be less exother-
mic than normal and are referred to as cool flames. An increase in the pressure or

3.5

3
Quenching distance, mm

2.5

1.5

0.5 1 1.5 2
Fuel/air equivalence ratio

Fig. 3. Quenching distance as function of equivalence ratio for hydrocarbon mixtures


with air (1), where x ¼ methane, & ¼ propane, 4 ¼ propylene, and ¼ ethylene.
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 457

temperature of the mixture outside the cool flame can produce normal sponta-
neous ignition (1).
Flame Temperature. The adiabatic flame temperature, or theoretical
flame temperature, is the maximum temperature attained by the products
when the reaction goes to completion and the heat liberated during the reaction
is used to raise the temperature of the products. Flame temperatures, as a func-
tion of the equivalence ratio, are usually calculated from thermodynamic data
when a fuel is burned adiabatically with air. To calculate the adiabatic flame
temperature (AFT) without dissociation, for lean to stoichiometric mixtures,
complete combustion is assumed. This implies that the products of combustion
contain only carbon dioxide , water, nitrogen, oxygen, and sulfur dioxide.
Actual temperatures in practical flames are lower than calculated values as
a result of the heat losses by radiation, thermal conduction, and diffusion. At
high temperatures, dissociation of products of combustion into species such as
OH, O, and H reduces the theoretical flame temperature (7). Increasing the pres-
sure tends to suppress dissociation of the products and thus generally raises the
adiabatic flame temperature (4).
Adiabatic flame temperatures agree with values measured by optical tech-
niques, when the combustion is essentially complete and when losses are known
to be relatively small. Calculated temperatures and gas compositions are thus
extremely useful and essential for assessing the combustion process and predict-
ing the effects of variations in process parameters (4). Advances in computational
techniques have made flame temperature and equilibrium gas composition calcu-
lations, and the prediction of thermodynamic properties, routine for any fuel-
oxidizer system for which the enthalpies and heats of formation are available
or can be estimated.
Flame Types and Their Characteristics. There are two main types of
flames: diffusion and premixed. In diffusion flames, the fuel and oxidant are
separately introduced and the rate of the overall process is determined by the
mixing rate. Examples of diffusion flames include the flames associated with can-
dles, matches, gaseous fuel jets, oil sprays, and large fires, whether accidental or
otherwise. In premixed flames, fuel and oxidant are mixed thoroughly prior to
combustion. A fundamental understanding of both flame types and their struc-
ture involves the determination of the dimensions of the various zones in the
flame and the temperature, velocity, and species concentrations throughout
the system.
The development of combustion theory has led to the appearance of several
specialized asymptotic concepts and mathematical methods. An extremely strong
temperature dependence for the reaction rate is typical of the theory. This makes
direct numerical solution of the equations difficult but at the same time accurate.
The basic concept of combustion theory, the idea of a flame moving at a constant
velocity independent of the ignition conditions and determined solely by the
properties and state of the fuel mixture, is the product of the asymptotic
approach (18,19). Theoretical understanding of turbulent combustion involves
combining the theory of turbulence and the kinetics of chemical reactions (19–23).
Laminar Premixed Flames. The structure of a one-dimensional premixed
flame is well understood (1). By coupling the rate of heat release from chemical
reaction and the rate of heat transfer by conduction with the flow of the
458 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

unburned mixture, an observer moving with the wave would see a steady lami-
nar flow of unburned gas at a uniform velocity, Su, into the stationary wave or
flame. Hence Su is defined as the burning velocity of the mixture based on the
conditions of the unburned gas. The thickness of the preheating zone and the
equivalent reaction zone is found to be inversely proportional to the burning velo-
city. By considering the heat release from the chemical reaction, it is possible to
calculate the thickness of the effective reaction zone. For example, at atmo-
spheric pressure, for hydrocarbon flames, the thicknesses of the preheat and
reaction zones are found to be 0.7 and 0.2 mm, respectively (1–3). A propagating,
premixed flame is in essence a thin wave in which the temperature rise is con-
tinuous and rapid. The thicknesses of the wave depend on f and on pressure; for
a typical hydrocarbon–air mixture at f ¼ 1 at 101.3 kPa (1 atm), the entire rise,
from the initial temperature to the adiabatic temperature, occurs in 0.01 mm,
and the concentration and temperature gradients are accordingly very steep.
Su has been approximated for flames stabilized by a steady uniform flow of
unburned gas from porous metal diaphragms or other flow straighteners. How-
ever, in practice, Su is usually determined less directly from the speed and area
of transient flames in tubes, closed vessels, soap bubbles blown with the mixture,
and, most commonly, from the shape of steady Bunsen burner flames. The
observed speed of a transient flame usually differs markedly from Su. For exam-
ple, it can be calculated that a flame spreads from a central ignition point in an
unconfined explosive mixture such as a soap bubble at a speed of (ro/rb)Su,
in which the density ratio across the flame is typically 5–10. Usually, the
expansion of the burning gas imparts a considerable velocity to the unburned
mixture, and the observed speed will be the sum of this velocity and Su.
By applying the conservation equations of mass and energy and by neglect-
ing the small pressure changes across the flame, the thickness of the preheating
and reaction zones can be calculated for a one-dimensional flame (1).
There are a number of sources of instability in premixed combustion
systems (23,24).

1. System instabilities involve interactions between flows in different parts of


a reacting system. Although these instabilities can cause turbulence, they
cannot be analyzed on their own and the whole system should be consid-
ered as a unit (11,25).
2. Acoustic instabilities involve the interactions of acoustic waves with the
combustion processes. These instabilities are of high frequency and can
be significant in certain situations but can successfully be avoided by sui-
table design (11,24).
3. Taylor instabilities involve effects of buoyancy or acceleration in fluids with
variable density; a light fluid beneath a heavy fluid is unstable by the
Taylor mechanism. The upward propagation of premixed flames in tubes
is subject to Taylor instability (11).
4. Landau instabilities are the hydrodynamic instabilities of flame sheets that
are associated neither with acoustics nor with buoyancy but instead involve
only the density decrease produced by combustion in incompressible flow.
The mechanism of Landau instability is purely hydrodynamic. In principle,
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 459

Landau instabilities should always be present in premixed flames, but in


practice they are seldom observed (26,27).
5. Diffuse-thermal instabilities involve the relative diffusion reactants
and heat within a laminar flame. These are the smallest-scale instabilities
(11).

Laminar flame instabilities are dominated by diffusional effects that can


only be of importance in flows with a low turbulence intensity, where molecular
transport is of the same order of magnitude as turbulent transport (28). Flame
instabilities do not appear to be capable of generating turbulence. They result in
the growth of certain disturbances, leading to orderly three-dimensional struc-
tures which, though complex, are steady (1,2,8,9).
Turbulent Premixed Flames. Combustion processes and flow phenomena
are closely connected and the fluid mechanics of a burning mixture play an
important role in forming the structure of the flame. Laminar combusting
flows can occur only at low Reynolds numbers, defined as

rud
RE ¼ ð11Þ
m

where, r ¼ density, kg/m3; u ¼ velocity, m/s; d ¼ diameter, m; and m ¼ kinematic


viscosity, kg/ms. When Re > Recr the laminar structure of a flow becomes
unstable and when the Reynolds number exceeds the critical value by an order
of magnitude, the structure of the flow changes. Along with this change of struc-
ture from an orderly state to a more chaotic state, the following parameters begin
to fluctuate randomly: velocity, pressure, temperature, density, and species con-
centrations. Overall, with increasing Reynolds number, laminar flow becomes
unstable as a result of these fluctuations and breaks down into turbulent flow.
Laminar and turbulent flames differ greatly in appearance. For example, while
the combustion zone of a laminar Bunsen flame is a smooth, delineated and thin
surface, the analogous turbulent combustion region is blurred and thick (1).
Turbulent flame speed, unlike laminar flame speed, is dependent on the
flow field and on both the mean and turbulence characteristics of the flow,
which can in turn depend on the experimental configuration. Nonstationary
spherical turbulent flames, generated through a grid, have flame speeds of the
order of or less than the laminar flame speed. This turbulent flame speed
tends to increase proportionally to the intensity of the turbulence.
In high speed dusted, premixed flows, where flames are stabilized in the
recirculation zones, the turbulent flame speed grows without apparent limit, in
approximate proportion to the speed of the unburned gas flow. In the recircula-
tion zones the intensity of the turbulence does not affect the turbulent flame
speed (1).
In the reaction zone, an increase in the intensity of the turbulence is related
to the turbulent flame speed. It has been proposed that flame-generated turbu-
lence results from shear forces within the burning gas (1,28). The existence of
flame-generated turbulence is not, however, universally accepted, and in uncon-
fined flames direct measurements of velocity indicate that there is no flame-
generated turbulence (1,2).
460 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

The balanced equation for turbulent kinetic energy in a reacting turbulent


flow contains the terms that represent production as a result of mean flow shear,
which can be influenced by combustion, and the terms that represent mean flow
dilations, which can remove turbulent energy as a result of combustion. Some of
the discrepancies between turbulent flame propagation speeds might be
explained in terms of the balance between these competing effects.
To analyze premixed turbulent flames theoretically, two processes should
be considered: (1) the effects of combustion on the turbulence, and (2) the effects
of turbulence on the average chemical reaction rates. In a turbulent flame, the
peak time-averaged reaction rate can be orders of magnitude smaller than the
corresponding rates in a laminar flame. The reason for this is the existence of
turbulence-induced fluctuations in composition, temperature, density, and heat
release rate within the flame, which are caused by large eddy structures and
wrinkled laminar flame fronts.
A unified statistical model for premixed turbulent combustion and its sub-
sequent application to predict the speed of propagation and the structure of plane
turbulent combustion waves is available (29–32).
Laminar Diffusion Flames. Generally, a diffusion flame is defined as one
in which no mixing of the fuel and oxidant takes place prior to emission from the
burner. However, it can also be defined as a flame in which the mixing rate is
sufficiently slow compared to the reaction rate, that the mixing time controls
the burning rate. Since a continuous spectrum of flames exists between the per-
fectly premixed flame and the diffusion flame, the term diffusion flame is
reserved for those flames in which there is a total separation between fuel and
oxidant (1,5). Some of the practical advantages of diffusion flames include high
flame stability, safety (as there is no need for the storage of the combustible mix-
ture), intense radiation and hence high heat exchange with the surroundings. It
is difficult to give a general treatment of diffusion flames largely because no sim-
ple measurable parameter, analogous to the burning velocity for premixed
flames, can be used to characterize the burning process (3).
Many different configurations of diffusion flames exist in practice (Fig. 4).
Laminar jets of fuel and oxidant are the simplest and most well understood diffu-
sion flames. They have been studied exclusively in the laboratory, although a com-
plete description of both the transport and chemical processes does not yet exist (2).
The discussion of laminar diffusion flame theory addresses both the gaseous
diffusion flames and the single-drop evaporation and combustion, as there are
some similarities between gaseous and liquid diffusion flame theories (2). A fre-
quently used model of diffusion flames has been developed (34), and despite some
of the restrictive assumptions of the model, it gives a good description of diffusion
flame behavior.
The Displacement Distance theory suggests that since the structure of the
flame is only quantitatively correct, the flame height can be obtained through
the use of the displacement length or ‘‘displacement distance’’ (35,36) (eq. 12),
where h ¼ flame height, m; V ¼ volumetric flow rate, m3/s; and D ¼ diffusion
coefficient.

V
h¼ ð12Þ
2pD
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 461

a a

f a f a a f a a f a a f a

(a) (b) (c) (d) (e)


Side views

a a
f a a f a a f a
a
a a

Top views

Fig. 4. Various configurations (a–e) used to obtain gaseous diffusion flames where
a ¼ air and f ¼ fuel (33).

In addition to the Burke and Schumann model (34) and the Displacement
Distance theory, a comprehensive laminar diffusion flame theory can be written
using the equations of conservation of species, energy, and momentum, including
diffusion, heat transfer, and chemical reaction.
Evaporation and burning of liquid droplets are of particular interest in
furnace and propulsion applications and by applying a part of the Burke and
Schumann approach it is possible to obtain a simple model for diffusion flames.
Combustion chemistry in diffusion flames is not as simple as is assumed in
most theoretical models. Evidence obtained by adsorption and emission spectro-
scopy (37) and by sampling (38) shows that hydrocarbon fuels undergo appreci-
able pyrolysis in the fuel jet before oxidation occurs. Further evidence for the
existence of pyrolysis is provided by sampling of diffusion flames (39). In general,
the preflame pyrolysis reactions may not be very important in terms of the gross
features of the flame, particularly flame height, but they may account for the for-
mation of carbon while the presence of OH radicals may provide a path for NOx
formation, particularly on the oxidant side of the flame (39).
Combustion chemistry in diffusion flames also accounts for the smoke
formation process. Characteristic behavior of smoking diffusion flames includes:
(1) the first appearance of smoke is generally at the flame tip, the width of the
smoke trail increasing as the amount of smoke increases; (2) the appearance and
quantity of smoke increases with increasing flame height; (3) the smoking ten-
dency changes with oxidant flow rate and the oxygen content of the oxidant
stream; (4) the smoking tendency is a function of fuel type; and (5) the smoking
tendency generally increases with increasing pressure (39).
Diffusion Flames in the Transition Region. As the velocity of the fuel jet
increases in the laminar to turbulent transition region, an instability develops at
the top of the flame and spreads down to its base. This is caused by the shear
forces at the boundaries of the fuel jet. The flame length in the transition region
462 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

is usually calculated by means of empirical formulas of the form (eq. 13): where
l ¼ length of the flame, m; r ¼ radius of the fuel jet, m; v ¼ fuel flow velocity, m/s;
and C1 and C2 are empirical constants.

r
l¼ ð13Þ
C2
C


Turbulent Diffusion Flames. Laminar diffusion flames become turbulent


with increasing Reynolds number (1,2). Some of the parameters that are affected
by turbulence include flame speed, minimum ignition energy, flame stabilization,
and rates of pollutant formation. Changes in flame structure are believed to be
controlled entirely by fluid mechanics and physical transport processes (1,2,9).
Consider the case of the simple Bunsen burner. As the tube diameter
decreases, at a critical flow velocity and at a Reynolds number of about 2000,
flame height no longer depends on the jet diameter and the relationship between
flame height and volumetric flow ceases to exist (2). Some of the characteristics of
diffusion flames are illustrated in Figure 5.
The preferred models for predicting the behavior of turbulent-free shear
layers involve the solution of the turbulent kinetic energy equation in order to
obtain the local turbulent shear stress distribution (1,2,9). These models are
ranked according to the number of simultaneous differential equations that
need to be solved. The one-equation model considers the turbulent kinetic energy
equation alone, whereas the two-equation model considers the turbulent kinetic
energy equation plus a differential equation for the turbulence length scale,
or equivalently, the dissipation rate for turbulent kinetic energy. These equa-
tions are solved along with the conservation equations (momentum, energy,
and species) to model turbulent flows.
The physics and modeling of turbulent flows are affected by combustion
through the production of density variations, buoyancy effects, dilation due to
heat release, molecular transport, and instability (1–3,5,8). Consequently, the
conservation equations need to be modified to take these effects into account.
This modification is achieved by the use of statistical quantities in the conserva-
tion equations. For example, because of the variations and fluctuations in the
density that occur in turbulent combustion flows, density weighted mean values,
or Favre mean values, are used for velocity components, mass fractions,
enthalpy, and temperature. The turbulent diffusion flame can also be treated
in terms of a probability distribution function (pdf ), the shape of which is
assumed to be known a priori (1).
In general, comprehensive, multidimensional modeling of turbulent com-
bustion is recognized as being difficult because of the problems associated with
solving the differential equations and the complexities involved in describing the
interactions between chemical reactions and turbulence. A number of computa-
tional models are available commercially that can do such work. These include
FLUENT, FLOW-3D, and PCGC-2.
The various studies attempting to increase our understanding of turbulent
flows comprise five classes: moment methods disregarding probability density
functions, approximation of probability density functions using moments,
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 463

Diffusion flames Transition region Fully developed turbulent flames

Envelope of
flame length
Flame height, hF

Envelope of
break-point
0
Increasing nozzle velocity

(a)

60 24

50 20
Total height, on-part flames
Distance from nozzle, cm

Upper part of flame blows off

40 16
Flame separation begins

in.
30 Total height, 12
off-part flames

20 8

Break-point height
10 Bottom 4
of off-part
flames

0 100 200 300 400 500 600 700 800


Nozzle velocity, U, cm/s

(b)

Fig. 5. Effects of nozzle velocity on flame appearance in laminar and turbulent flow:
(a), flame appearance; (b), flame height and break-point height (40).

calculation of evolution of probability density functions, perturbation methods


beginning with known structures, and methods identifying coherent structures.
For a thorough review of turbulent diffusion flames see References (41–48).
2.2. Fundamentals of Heterogeneous Combustion. The discussion
of combustion fundamentals so far has focused on homogeneous systems. Hetero-
geneous combustion is the terminology often used to refer to the combustion of
liquids and solids. From a technological viewpoint, combustion of liquid hydro-
carbons, mainly in sprays, and coal combustion are of greatest interest.
464 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

Most theories of droplet combustion assume a spherical, symmetrical dro-


plet surrounded by a spherical flame, for which the radii of the droplet and the
flame are denoted by rd and rp, respectively. The flame is supported by the fuel
diffusing from the droplet surface and the oxidant from the outside. The heat
produced in the combustion zone ensures evaporation of the droplet and conse-
quently the fuel supply. Other assumptions that further restrict the model
include: (1) the rate of chemical reaction is much higher than the rate of diffusion
and hence the reaction is completed in a flame front of infinitesimal thickness;
(2) the droplet is made up of pure liquid fuel; (3) the composition of the ambient
atmosphere far away from the droplet is constant and does not depend on the
combustion process; (4) combustion occurs under steady-state conditions;
(5) the surface temperature of the droplet is close or equal to the boiling point
of the liquid; and (6) the effects of radiation, thermodiffusion, and radial pressure
changes are negligible.
In order to obtain an expression for the burning rate of the droplet, the fol-
lowing parameters are needed: physical constants such as the specific heat, and
the thermal conductivity of the droplet, the radius of the flame, and the temp-
erature of the flame. To determine these quantities, heat conduction, diffusion,
and the kinetics of the chemical processes associated with droplet combustion
need to be analyzed. This is achieved mathematically by solving the equations
of mass continuity, mass continuity for components, and the energy equation.
The solving of these equations can be facilitated if the following simplifying
assumptions are made: the flame surrounding the droplet is a diffusion flame
and, by definition, is formed where the fuel and oxidant meet in stoichiometric
proportions; the temperature of this flame is very close to the adiabatic flame
temperature; and the heat required for evaporation of the droplet and the heat
loss to the surroundings through the burned gas are small and can therefore be
neglected. These equations are usually solved in spherical coordinates for a one-
dimensional case. However, since the flame is relatively thick, and the droplet is
relatively small, the one-dimensional model of the process may not be a particu-
larly accurate representation. Nevertheless, the values obtained for burning
rates provide useful information (9).
The burning rate of the droplet (kg/s), and its rate of change of radius are
related by:
drd
me ¼ 4pr2d rf ð14Þ
dt
where me ¼ burning rate of the droplet, kg/s; rf ¼ density of fuel, kg/m3; and
rd ¼ radius of droplet, m. This equation can be simplified, assuming rt =rd ’
constant and me ’ rd .
dðd2d Þ
¼ K ¼ constant ð15Þ
dt
Hence, the constant K is termed the ‘‘burning-constant of the droplet.’’ Integra-
tion of the equation 15 produces the droplet burning law:

d2d ¼ d20  Kt ð16Þ

where d0 ¼ the initial diameter of the droplet, m, and t ¼ burnout time, s.


Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 465

The amount of data available on droplet combustion is extensive. However,


the results can be easily summarized, because the burning rate constants for the
majority of fuels of practical interest fall within the narrow range of 7 to
11 103 cm2 =s. An increase in oxygen concentration results in an increase in
the burning rate constant. If the burning takes place in pure oxygen, the values
for burning rate are increased by a factor of about 2.0, compared to when the
burning takes place in air (9).
The convective gas flow around a burning particle affects its burning rate. It
has been postulated that in the absence of convection, the burning rate is inde-
pendent of pressure. Forced convection, on the other hand, is believed to increase
the burning rate.
During the final stages of the combustion of a droplet, coke remains, and
although it represents a relatively small percentage of the mass of the original
oil droplet, the time taken for the heterogeneous reaction between the oxygen-
depleted combustion air and the coke particle is generally the slowest of all
the combustion steps (9).
The reaction between a porous solid, such as a coke sphere, and a gas, such
as oxygen, occurs in the following stages: (1) the main reactant species diffuse
thoroughly through the boundary layer toward the solid surface and the products
of reaction diffuse away from the surface; (2) diffusion and simultaneous chemi-
cal reaction take place within the pores of the solid proceeding from the external
surface toward the interior, and gaseous products diffuse in the opposite direc-
tion; and (3) at the participating surfaces, the reacting gas chemisorbs, some
intermediate species are formed, then the final products of the reaction desorb
from the surface. Thus the observed reaction rate is a function of the individual
resistances–boundary layer diffusion, pore diffusion, and chemical kinetics,
and the rate controlling process is the slowest of them or a combination of
these processes. Even though a number of gas reactions may take place at the
surface of a burning carbonaceous solid, the reaction forming CO is most often
assumed, C þ 1=2 O2 ! CO. Coke combustion is treated mathematically like
char combustion.
In a practical combustion chamber, the droplets tend to burn in the form of
sprays, hence it is important to understand the fundamentals of sprays. In the
most simple case, a fuel spray suspended in air will support a stable propagating
laminar flame in a manner similar to a homogeneous gaseous mixture. In this case,
however, two different flame fronts are observed. If the spray is made up of very
small droplets they vaporize before the flame reaches them and a continuous
flame front is formed. If the droplets are larger, the flame reaches them before
the evaporation is complete, and if the amount of fuel vapor is insufficient for
the formation of a continuous flame front, the droplets burn in the form of iso-
lated spherical regions. Flames of this type are referred to as heterogeneous lami-
nar flames. Experimental determination of burning rates and flammability limits
for heterogeneous laminar flames is difficult because of the motion of droplets
caused by gravity and their evaporation before the arrival of the flame front.
In modern liquid-fuel combustion equipment the fuel is usually injected into
a high velocity turbulent gas flow. Consequently, the complex turbulent flow and
spray structure make the analysis of heterogeneous flows difficult and a detailed
analysis requires the use of numerical methods (9).
466 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

The combustion of a coal particle occurs in two stages: (1) devolatilization


during the initial stages of heating with accompanying physical and chemical
changes, and (2) the subsequent combustion of the residual char (49). The burn-
ing rate or reactivity of the residual char in the second stage is strongly depen-
dent on the process conditions of the first stage. During pulverized coal
combustion the devolatilization step usually takes about 0.1 s and the residual
char combustion takes on the order of 1 s. Since char combustion occurs over
90% of the total burning time, its rate can affect the volume of the combustion
chamber required to attain a given heat release and combustion efficiency. The
rate of devolatilization, and the amount and nature of volatiles can significantly
affect the ignition process and hence the onset of char combustion. During devo-
latilization or thermal decomposition, rupture of various functional groups
bonded to the macromolecular structure leads to the evolution of gases and the
formation and opening of pores. Also, depending on its thermoplastic properties,
a coal particle may undergo softening and swelling resulting in a change in size
and physical characteristics. Diffusion of oxygen to, and within, a char particle
depends on its physical structure and accessible surface area (char morphology).
Attempts have been made to explain the combustion rates of chars in terms of
their morphologies (50–52).
The ignition mechanism is rather complex and is not well understood in
terms of actually defining the ignition temperature and reaction mechanisms.
The ignition temperature is known, however, not to be a unique property of
the coal and depends on a balance between heat generated and heat dissipated
to the surroundings around the coal particle. Measuring the ignition character-
istics is complicated by the fact that they are strongly dependent on the physical
arrangement of the particles, eg, single particle, clouds of coal dust, or coal piles.
Reported ignition temperatures range from 303 to 373 K in the case of sponta-
neous ignition of coal piles at ambient temperature to 1073–1173 K in the
case of single coal particle ignition (53,54). Characteristics such as coal type,
particle size and distribution of mineral matter, and experimental conditions
such as gas temperature, heating rate, oxygen, and coal dust concentration are
some of the important factors that influence values obtained for ignition tem-
peratures. Ignition of coal particles can occur either homogeneously, with igni-
tion of the volatiles released and the subsequent ignition of the char surface,
or heterogeneously, with both volatiles and the char surface igniting simulta-
neously. Heating rate and particle size affect the mode of ignition. The early
theories of ignition were based on an energy balance per unit volume of reactive
mixture (55,56). Later, a generalized theory of flame propagation in laminar coal
dust flames was proposed (57). Good reviews in this area include References (58–60).
A variety of techniques has been used to determine ignition temperatures:
fixed beds, the crossing point method, the critical air blast method (61), photo-
graphic techniques (62,63), entrained flow reactors (64), electric spark ignition
(65), luminous glow observations (66), plug flow reactors (67), shock tubes (68),
and thermogravimetric analysis (69). The techniques mostly used are constant
temperature methods (66,70,71) in which coal particles are introduced into a
preheated furnace maintained at a fixed temperature. If ignition of the coal par-
ticles is not observed by the appearance of a glow, flame, or sharp temperature
rise, the test is then conducted at a higher temperature and the procedure is
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 467

repeated until the ignition of the sample is observed. It has been a matter of con-
troversy as to whether the value of the critical temperature determined using
this method for these low temperature tests represents the ignition tendency
or the combustibility of the coal during the high heating rates encountered in
a pulverized coal flame.
The structure of residual char particles after devolatilization depends
on the nature of the coal and the pyrolysis conditions such as heating rate,
peak temperature, soak time at the peak temperature, gaseous environment,
and the pressure of the system (72). The oxidation rate of the char is primarily
influenced by the physical and chemical nature of the char, the rate of diffusion
and the nature of the reactant and product gases, and the temperature and pres-
sure of the operating system. The physical and chemical characteristics that
influence the rate of oxidation are chemical structural variations, such as the
concentration of oxygen and hydrogen atoms (73–75), the nature and amount
of mineral matter (76–79), and physical characteristics such as porosity, particle
size, and accessible surface area. The rate of diffusion of the reactant gas is
governed by the temperature and pressure of the operating system.
The rate limiting step in the combustion of char is either the chemical
kinetics (adsorption of oxygen, reaction, and desorption of products) or diffusion
of oxygen (bulk and pore diffusion). Variations in the reaction rate with tempera-
ture for gas–carbon reactions have been grouped into three main regions or
zones depending on the rate limiting resistance (80) (Figs. 6 and 7).
The overall reaction rate is determined by equation 17,

1
q¼ pg ð17Þ
1 1
þ
kdiff ks

where q ¼ rate of removal of carbon atoms per unit external surface area, kg/
m2s; pg ¼ partial pressure of oxygen in the free stream, Pa; kdiff ¼ diffusional

Fig. 6. The three ideal zones (I–III) representing the rate of change of reaction for a por-
ous carbon with increasing temperature where a and b are intermediate zones, Ea is acti-
vation energy, and Et is true activation energy. The effectiveness factor, Z, is a ratio of
experimental reaction rate to reaction rate which would be found if the gas concentration
were equal to the atmospheric gas concentration (80).
468 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

Interior of porous solid

Boundary layer
Zone 3

Concentration of fluid reactant


Zone 1

Zone 2

0 R
Distance from the center of
the solid reactant

Fig. 7. The concentration of the reacting fluid as a function of the distance from the
center of the solid reactant in various reaction zones.

reaction rate coefficient, kg/(m2  Pa); and ks ¼ surface reaction rate coefficient,
kg/(m2  Pa).
In Zone 1, the low temperature zone, the reaction rate is slow and the
concentration of the reactant gas is uniform throughout the interior of the solid.
The overall rate is controlled by the chemical reactivity of the solid, the kinetics
are not influenced by mass transfer, and the kinetics are the intrinsic kinetics.
The measured activation energy is equal to the true activation energy. In a
higher temperature zone, Zone 2, the concentration of the gaseous reactant
becomes zero at a point somewhere between the surface and the center of the
particle, and the reaction rate is controlled by both the chemical kinetics and
the diffusion rate of the gaseous reactant. The measured activation energy in
this zone is one-half of the true activation energy. At still higher temperatures,
in Zone 3, the chemical reactivity is so high that oxygen is consumed as soon as it
reaches the surface of the solid and the concentration of the gaseous reactant
approaches zero at the surface, indicating that the reaction rate is controlled
by bulk diffusion of the reacting gas. The measured activation energy approaches
a value of zero in this zone.
Burning times for coal particles are obtained from integrated reaction rates.
For larger particles (>100 mm) and at practical combustion temperatures, there
is a good correlation between theory and experiment for char burnout. Experi-
mental data are found to obey the Nusselt ‘‘square law’’ which states that the
burning time varies with the square of the initial particle diameter (tc  D2:0 o ).
However, for particle sizes smaller than 100 mm, the Nusselt relationship
seems to predict higher burning times than those observed and it has been
noted that burning times are proportional to the initial diameter of the particles
raised to the power of 1.4 (tc  D1:4
o ) (54).
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 469

3. Combustion Technology

Technology addresses the more applied, practical aspects of combustion, with an


emphasis on the combustion of gaseous, liquid, and solid fuels for the purpose of
power production. In an ideal fuel burning system (1) there should be no excess
oxygen or products of incomplete combustion, (2) the combustion reaction should
be initiated by the input of auxiliary ignition energy at a low rate, (3) the reaction
rate between oxygen and fuel should be fast enough to allow rapid rates of heat
release and it should also be compatible with acceptable nitrogen and sulfur
oxide formation rates, (4) the solid impurities introduced with the fuel should
be handled and disposed of effectively, (5) the temperature and the weight of
the products of combustion should be distributed uniformly in relation to the par-
allel circuits of the heat absorbing surfaces, (6) a wide and stable firing range
should be available, (7) fast response to changes in firing rate should be easily
accommodated, and (8) equipment availability should be high and maintenance
costs low (49–51).
3.1. Combustion of Gaseous Fuels. In any gas burner some mechan-
ism or device (flame holder or pilot) must be provided to stabilize the flame
against the flow of the unburned mixture. This device should fix the position of
the flame at the burner port. Although gas burners vary greatly in form and com-
plexity, the distribution mechanisms in most cases are fundamentally the same.
By keeping the linear velocity of a small fraction of the mixture flow equal to or
less than the burning velocity, a steady flame is formed. From this pilot flame,
the main flame spreads to consume the main gas flow at a much higher velocity.
The area of the steady flame is related to the volumetric flow rate of the mixture
by equation 18 (81,82)

V_ mix ¼ Af Su ð18Þ

where V_ mix ¼ volumetric flow rate, m3/s; Af ¼ area of the steady flame, m2; and
Su ¼ burning velocity, m/s.
The volumetric flow rate of the mixture is, in turn, proportional to the rate
of heat input (eq. 19):

V_ mix ðHHV Þ ¼ Q_ ð19Þ

where V_ mix ¼ volumetric flow rate, m3/s; HHV ¼ higher heating value of the
fuel, J/kg; and Q_ ¼ rate of heat input, J/s.
In the simple Bunsen flame on a tube of circular cross-section, the stabiliza-
tion depends on the velocity variation in the flow emerging from the tube. For
laminar flow (parabolic velocity profile) in a tube, the velocity at a radius r is
given by equation 20:
 
v ¼ const R2  r2 ð20Þ

where v ¼ laminar flow velocity, m/s; R ¼ tube radius, m; r ¼ flame radius, m;


and const ¼ experimental constant.
470 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

The maximum velocity at the axis is twice the average, whereas the velocity
at the wall is zero. The effect of the burner wall is to cool the flame locally and
decrease the burning velocity of the mixture. This results in flame stabilization.
However, if the heat-transfer processes (conduction, convection, and radiation)
involved in cooling the flame are somehow impeded, the rate of heat loss is
decreased and the local reduction in burning velocity may no longer take
place. This could result in upstream propagation of the flame.
To make the flame stable against the flow in a thin annular region near the
rim, the flow velocity vr should be made equal to the burning velocity at some
radius r. This annulus serves as a pilot and ignites the main flow of the mixture,
ie, the flame gradually spreads toward the center. In most of the mixture flow,
vr > Su which results in a stable flame. With increasing mixture flow, the height
and area of the flame increase. Measurement of the area of a stable Bunsen flame
is the basis for the method most commonly used to determine Su (81–83).
By feeding the mixture through a converging nozzle, the velocity profile
may be made nearly flat or uniform. A Bunsen flame in such a flow has a smaller
range of stability but the mechanism is essentially the same and the flame very
closely approximates a cone. If the apex angle of the flame is , then Su can be
obtained from equation 21
 

Su ¼ r sin ð21Þ
2

where Su ¼ burning velocity, m/s; vr ¼ mixture velocity at the nozzle exit, m/s;
and  ¼ the apex angle of the flame in degrees.
If the tube diameter is appreciably larger than the quenching distance, Su
will exceed vr in some parts of the flowing mixture due to a lack of quenching,
and the flame will then propagate down the tube as far as there is mixture to
consume. This undesirable condition is referred to as flashback. If, on the
other hand, vr exceeds Su in the mixture flow, the flame lifts from the port and
blows off. This condition is referred to as blowoff and like flashback should be
avoided (Fig. 8). The velocity gradient at the wall, gw, is defined as
 
dv
gw ¼
dr r¼R

For example, in a laminar or Poiseuille flow in a round tube of radius R,


gw ¼ 4gav =R, and for a given initial mixture composition, flashback (or blowoff)
will occur at the same value of gw in tubes of various sizes, whereas the corre-
sponding average velocity at flashback (or blowoff) is proportional to R. Typical
velocity gradient values for stoichiometric methane–air flames are at flashback
about 400 s1 and at blowoff 2000 s1 . Thus, if the mixture is burned on a 1-cm
diameter tube, the average velocity of flashback is 400 0:5=4 ¼ 50 cm=s and at
blowoff the average velocity is 250 cm/s, or the range of stability would be
roughly one and a half to seven times the burning velocity. At flashback, gw is
at a maximum around f ¼ 1. If the burner is operated such that the surrounding
inert atmosphere is inert, gw is at a maximum at blowoff.
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 471

3200
A
3000

Theoretical flame temperature, K


2800 D

B
2600
C
2400

2200

0 500 1000 2000 3000


Preheated air temperature, K

Fig. 8. Flashback and blowoff critical velocity gradients for a natural gas flame (10),
where A is N2 =O2 ¼ 2, p ¼ 1:01 MPa (10 atm); B, N2 =O2 ¼ 2, p ¼ 101:3 kPa (1 atm); C,
air, p ¼ 1:01 MPa; and D, air, p ¼ 101:3 kPa.

The behavior of rich mixtures is complicated by the entrainment of air at


the burner port that sustains combustion of hot combustion products of the
primary flame near the port. The blowoff velocity is found to increase continu-
ously with f, or richer mixtures are more stable with respect to blowoff. They
also have a lesser tendency toward flashback. Hence, a Bunsen flame has more
latitude for stable operation if the primary mixture is rich. For this reason
many appliance burners that involve assemblies of such flames are routinely
adjusted by first making the primary mixture so rich that soot just forms in
the burned gas (yellow-tipping), and then increasing the air until the yellow
luminosity disappears. The primary equivalence ratio is then perhaps 1.5 or
more; the rich products of that primary flame are burned in the secondary dif-
fusion flame in the surrounding air, or the faintly luminous outer mantle of a
Bunsen flame.
Most of the commercial gas–air premixed burners are basically laminar-
flow Bunsen burners and operate at atmospheric pressure. This means that
the primary air is induced from the atmosphere by the fuel flow with which it
mixes in the burner passage leading to the burner ports, where the mixture is
ignited and the flame stabilized. The induced air flow is determined by the fuel
flow through momentum exchange and by the position of a shutter or throttle at
the air inlet. Hence, the air flow is a function of the fuel velocity as it issues from
the orifice or nozzle, or of the fuel supply pressure at the orifice. With a fixed fuel
flow rate, the equivalence ratio is adjusted by the shutter, and the resulting
induced air flow also determines the total mixture flow rate. The desired air–
fuel volume ratio is usually seven or more, depending on the stoichiometry.
Burners of this general type with many multiple ports are common for domestic
furnaces, heaters, stoves, and for industrial use. The flame stabilizing ports in
such burners are often round but may be slots of various shapes to conform to
the heating task.
472 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

Atmospheric pressure industrial burners are made for a heat release


capacity of up to 50 kJ/s (12 kcal/s), and despite the varied designs, their
principle of stabilization is basically the same as that of the Bunsen burner.
In some cases the mixture is fed through a fairly thick-walled pipe or casting
of appropriate shape for the application and the desired distribution of the
flame. The mixture issues from many small and closely spaced drilled holes, typi-
cally 1–2 mm diameter, and burns as rows of small Bunsen flames. It may be
ignited with a small pilot flame, spark, or heated wire, usually located near
the first holes to avoid accumulation of unburned mixture before ignition. The
rated heat release for a given fuel–air mixture can be scaled with the size and
number of holes. For example, for 2-mm diameter holes it would be 10–100 J/
(hole) or, in general, 0.3–3 kJ/cm2s (72–720 cal/cm2s) of port area, depending
on the fuel. The ports may also be narrow slots, sometimes packed with corru-
gated metal strips to improve the flow distribution and reduce the tendency to
flashback.
Gas burners that operate at high pressures are usually designed for high
mixture velocities and heating intensities and therefore stabilization against
blowoff must be enhanced. This can be achieved by a number of methods such
as surrounding the main port with a number of pilot ports or using a porous dia-
phragm screen.
In order to achieve high local heat flux the port velocity of the mixture
should be increased considerably. In burners that achieve stabilization using
pilot ports, most of the mixture can be burned at a port velocity as high as 100
Su to produce a long pencil-like flame, suitable for operations requiring a high
heat flux.
High local heat flux can also be obtained with Bunsen flames using mix-
tures with high burning velocities as in H2/O2 and C2H2/O2 torches. Their stabi-
lization mechanism is essentially the same as it is for slower burning mixtures
but the port or nozzle of the torch is usually much smaller, in part to avoid tur-
bulence in the mixture flow. The consequences of flashback are also much more
severe since most such mixtures are detonatable, and the premixing chamber or
tube must accordingly be more rugged. For this reason, many large hydrogen–
oxygen and hydrocarbon–oxygen flames are not premixed. They actually consist
of assemblies of closely spaced diffusion flames, produced from separately fed but
contiguous fuel and oxidizer flows. In such surface-mixing burners, the surface is
an array of very small and closely packed alternating fuel and oxidizer ports. The
arrangement and number of the ports and the complexity of the required mani-
folding of the reactant passages vary with the application and the desired geome-
try of the burned gas flow. With a very fast burning fuel-oxidizer combination,
the individual diffusion flames may be so short that the assembly approximates
a large, flat, premixed flame, as is the case with some rocket engine injectors
(84,85). The most frequently used modern burners are the circular type, capable
of burning oil and gas.
It is often desired to substitute directly a more readily available fuel for the
gas for which a premixed burner or torch and its associated feed system were
designed. Satisfactory behavior with respect to flashback, blowoff, and heating
capability, or the local enthalpy flux to the work, generally requires reproduction
as nearly as possible of the maximum temperature and velocity of the burned
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 473

gas, and of the shape or height of the flame cone. Often this must be done pre-
cisely, and with no changes in orifices or adjustments in the feed system.
If the substitute fuel is of the same general type, eg, propane for methane,
the problem reduces to control of the primary equivalence ratio. For nonaspiring
burners, ie, those in which the air and fuel supplies are essentially independent,
it is further reduced to control of the fuel flow, since the air flow usually consti-
tutes most of the mass flow and this is fixed. For a given fuel supply pressure and
fixed flow resistance of the feed system, the volume flow rate of the fuel is inver-
pffiffiffiffiffi
sely proportional to rf . The same total heat input rate or enthalpy flow to the
flame simply requires satisfactory reproduction of the product of the lower heat-
pffiffiffiffiffi
ing value of the fuel and its flow rate, so that WI  Qp = rf remains the same. WI
is the Wobbe Index of the fuel gas, and is a commonly used criterion for inter-
changeability in adjusting the composition of a substitute fuel. The units of WI
are variously given, but, if used consistently, are unimportant since only ratios of
the Wobbe Indices are ordinarily of interest. Sometimes rf is taken as the specific
gravity relative to some reference gas, eg, air, or average molecular weights may
be used.
The Wobbe Index criterion also applies to substitution with aspirating or
atmospheric pressure burners in which the volume flow of primary air induced
pffiffiffiffiffi
by momentum exchange with the fuel increases with rf . Because the volu-
metric air requirement for a given f is nearly proportional to the heating
value of the fuel, an adjustment of Qp and ff to the same WI results in about
the same stoichiometry of the primary flame. For example, if propane is an avail-
able substitute for methane or natural gas, it is common practice to prepare a
mixture of approximately 60% propane–40% air (which of course is well above
the upper flammability limit)pffiffiffiffiffiffiffiffiffi
to use
ffi as the fuel supply; though the heating
value of the mixture is 1:53= 2:36 ¼ 1:0 so that its Wobbe Index is the same.
Though there would be slight differences in the stoichiometry of the flame, aris-
ing from the air mixed with the propane, and in Su of the final mixture, substitu-
tion with the same supply conditions would be quite satisfactory. On the other
hand, if a mixture of the same heating value as that of methane were used
(39.2% propane in air) at the same supply pressure, the flame would be much
leaner and generally unsatisfactory (84,85).
There are direct substitutions of possible interest that would not be feasible
without drastic changes in the feed system or pressure. Thus if the available sub-
stitute for natural gas is, eg, a manufactured gas containing much CO, there
would almost always be a mismatch of the WIs unless the fuel could be further
modified by mixing with some other gaseous fuel of high volumetric heating
value (propane, butane, vaporized fuel oil, etc). Moreover, if there are substantial
differences in Su, eg, as a result of the presence of considerable H2 as well as CO
in the substitute gas, the variation in flame height and flashback tendency can
also make the substitution unsatisfactory for some purposes, even if the WI is
reproduced. Refinements and additional criteria are occasionally applied to mea-
sure these and other effects in more complex substitution problems (10,85).
Turbulence in the flow of a premixture flattens the velocity profile and
increases the effective burning velocity of the mixture; eg, at a pipe-Reynolds
number of 40,000 the turbulent burning velocity is several times the laminar
burning velocity and it can be perhaps fifty times larger at very high Reynolds
474 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

numbers. A turbulent flame is always somewhat noisy, the apparent flame sur-
face becomes diffuse owing to the fluctuations in the actual or flame surface
about its average position, and its stability tends to be less predictable. The
instantaneous flame surface may be thought of as wrinkled by velocity variations
in turbulent flow, or by the average distribution over a greater thickness (or
time). Although the resulting enhancement of the mixture consumption rate
may be considerable, turbulence is often considered undesirable in Bunsen-
type flames. For this and other reasons, a large number of burner ports of
small characteristic dimension, rather than a single large port, are frequently
used to assure laminar flow to the individual flames. However, turbulence has
an essential role in facilitating the mixing of fuel, oxidizer, and flame products,
and serves an important function in the various types of flame-stabilizers of prac-
tical importance.
3.2. Combustion of Liquid Fuels. There are several important liquid
fuels, ranging from volatile fuels for internal combustion engines to heavy hydro-
carbon fractions, sold commercially as fuel oils. The technology for the combus-
tion of liquid fuels for spark-ignition and compression-ignition internal
combustion engines is not described here. The emphasis here is primarily on
the combustion of fuel oils for domestic and industrial applications.
In general, the combustion of a liquid fuel takes place in a series of stages:
atomization, vaporization, mixing of the vapor with air, and ignition and main-
tenance of combustion (flame stabilization). Recent advances have shown the
atomization step to be one of the most important stages of liquid fuels combus-
tion. The main purpose of atomization is to increase the surface area to volume
ratio of the liquid. This is achieved by producing a fine spray. The finer the ato-
mization spray the greater the subsequent benefits are in terms of mixing, eva-
poration, and ignition. The function of an atomizer is twofold: atomizing the
liquid and matching the momentum of the issuing jet with the aerodynamic
flows in the furnace (86–88). Some of the more common designs of atomizers
are shown in Figure 9.
Atomizers for large boiler burners are usually of the swirl pressure jet or
internally mixed twin-fluid types, producing hollow conical sprays. Less common
are the externally mixed twin-fluid types (89,90).
The principal considerations in selecting an atomizer for a given application
are turn-down performance and auxiliary costs. An ideal atomizer would possess
all of the following characteristics: ability to provide good atomization over a
wide range of liquid flow rates; rapid response to change in liquid flow rate; free-
dom from flow instabilities; low power requirements; capability for scaling (up or
down) to provide desired flexibility; low cost, light weight, easily maintained, and
easily removed for servicing; and low susceptibility to damage during manufac-
ture and installation (89,90). There are differences in the structures of the sprays
between atomizer types which may affect the rate of mixing of the fuel droplets
with the combustion air and, hence, the initial development of a flame.
For distillate fuels of moderate viscosity 30 mm2/s (¼ cSt), at ordinary
temperatures, simple pressure atomization with some type of spray nozzle is
most commonly used. Operating typically with fuel pressure of 700–1000 kPa
(7–10 atm) such a nozzle produces a distribution of droplet diameters from
10–150 mm. They range in design capacity of 0.5–10 cm3/s or more and the
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 475

Liquid Plain-orifice

Liquid Liquid
Simplex Internal-mixing
air assist
Air

Primary
Duplex
Secondary
Liquid
External-mixing
Air air assist
Secondary Dual-orifice
Primary
Liquid Liquid
Liquid Plain-jet
Liquid Fan spray airblast
Air

Supply Liquid Prefilming


Spill Spill return Air airblast
Disk Cup

(a) (b) (c)

Fig. 9. (a) Pressure atomizers; (b) rotary atomizer; and (c) twin-fluid atomizers (89,90).

pumping power dissipated is generally less than 1% of the corresponding heat


release rate. A typical domestic oil burner nozzle uses about 0.8 cm3/s of No. 2
fuel oil at the design pressure. Although pressure-atomizing nozzles are usually
equipped with filters, the very small internal passages and orifices of the smal-
lest units tend to be easily plugged, even with clean fuels. With decreasing fuel
pressure the atomization becomes progressively less satisfactory. Much higher
pressures are often used, especially in engine applications, to produce a higher
velocity of the liquid relative to the surrounding air and accordingly smaller dro-
plets and evaporation times. Other mechanical atomization techniques for the
production of more nearly monodisperse sprays or smaller average droplet size
(spinning disk, ultrasonic atomizers, etc) are sometimes useful in burners for
special purposes and may eventually have more general application, especially
for small flows.
Conventional spray nozzles are relatively ineffective for atomizing fuels of
high viscosity, such as No. 6 or residual fuel oil (Bunker C) and other viscous,
dirty fuels. In order to transfer and pump No. 6 fuel oil, it is usually heated to
about 373 K, at which temperature its viscosity is about 40 mm2/s. Relatively
large nozzle passages and orifices are necessary to accommodate the possibility
of suspended solids. Atomization of such fuel is often accomplished, or at least
assisted, by the use of atomizing air, pumped at high velocity through adjacent
passages in or around the liquid injection ports. Much of the relative velocity
required to shear the liquid and form droplets is thus provided by the atomizing
air; its mass flow is usually comparable with the fuel flow and thus a small frac-
tion of the stoichiometric combustion air, although it is sometimes called primary
air. In a typical high pressure, air-atomizing nozzle designed for injecting resi-
dual oil in a gas turbine combustor, the atomizing air is supplied by an auxiliary
compressor with a power usage of about 1% of the combustion heat release rate.
476 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

Dry steam, if available, may also be used in a similar way, as is common practice
in the furnaces of power plant boilers using residual oil (89,90).
Air atomization with low pressure and relatively low velocity air is also
used in some burners for low viscosity distillate oils. In most aircraft gas turbines
some, or even a large part, of the atomization is done in this way by a small
fraction of the warm, compressed combustion air supplied in swirling flow
around the fuel nozzles. Imparting swirl to at least some of the air flow around
fuel injectors of all types is a common feature of many burners and combustors;
in some, swirl is introduced on a larger scale in all of the primary combustion air.
The velocity gradients or shear in the resulting vortexlike flow promote mass
transfer or mixing, including the recirculation of hot products of combustion to
the rich mixture or suspension in the low pressure core that contributes to sta-
bilization of the primary combustion zone (25). The angular or swirl velocity
imparted to the air and the strength of such flows are of course limited by the
available pressure drop; eg, in gas turbine combustors the allowable pressure
loss is usually <4% of the absolute pressure.
Combustion of fuel oil takes place through a series of steps, namely vapor-
ization, devolatilization, ignition, and dissociation, which finally lead to attaining
the flame temperature. Vaporization and devolatilization of the fine spray of fuel
droplets take place as physiochemical processes in the combustion chamber. The
vaporization temperature for fuel oil is in the range of 311–533 K, depending on
the grade of the fuel. Devolatilization takes place at about 700 K. The final flame
temperature attained is between 1366–1918 K. Complete combustion of an oil
droplet can occur in 2–20 ms depending on the size of the droplet. A typical char-
acteristic of an oil flame is its bright luminous nature which is the result of incan-
descent carbon particles in the fuel rich zone.
The study of the combustion of sprays of liquid fuels can be divided into two
primary areas for research purposes: single-droplet combustion mechanisms and
the interaction between different droplets in the spray during combustion with
regard to droplet size and distribution in space (91–94). The wide variety of ato-
mization methods used and the interaction of various physical parameters have
made it difficult to give general expressions for the prediction of droplet size and
distribution in sprays. The main fuel parameters affecting the quality of a spray
are surface tension, viscosity, and density, with fuel viscosity being by far the
most influential parameter (95).
The following general expression (eq. 22) is commonly used to describe the
droplet size distribution in a spray:

dn ¼ ar a expðbr b Þ dr ð22Þ

where dn ¼ number of droplets with radii between r and r þ dr. The constants a,
b, a, and b are independent of r and are usually determined empirically. The best-
known special case of this general equation is the Rosin-Rammler distribution
(95–97).
Theoretical modeling of single-droplet combustion has provided expressions
for evaporation and burning times of the droplets and the subsequent coke par-
ticles. A more thorough treatment of this topic is available (88,91–93,98).
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 477

Experimental techniques used for studying the combustion of single


droplets can be divided into three groups: suspended droplets, free droplets,
and porous droplets, with ongoing research in all three areas (98).
3.3. Combustion of Solid Fuels. Solid fuels are burned in a variety of
systems, some of which are similar to those fired by liquid fuels. In this article
the most commonly burned solid fuel, coal, is discussed. The main coal combus-
tion technologies are fixed-bed, eg, stokers, for the largest particles; pulverized-
coal for the smallest particles; and fluidized-bed for medium size particles
(99,100) (see COAL).
Fixed-Bed Technology. Fixed-bed firing of coal by means of stokers con-
sists of a solid bed of large (2–3 cm) coal particles on moving grates with combus-
tion air passing through the grates and ash removal from the end of the grate.
The use of a grate limits the application of this technique to small units, as grates
are restricted to a maximum size of about 100 m2 for structural reasons. Mechan-
ical stokers can be classified into the following groups, based on the method of
introducing fuel to the furnace: spreader stokers, underfeed stokers, water-
cooled vibrating-grate stokers, and chain-grate and traveling-grate stokers.
For a thorough review of stokers see Reference 81.
Pulverized-Coal Firing. This is the most common technology used for coal
combustion in utility applications because of the flexibility to use a range of coal
types in a range of furnace sizes. Nevertheless, the selection of crushing, combus-
tion, and gas-cleanup equipment remains coal dependent (54,100,101).
Prior to being fed to a pulverized fuel burner, coal is ground to a size gen-
erally specified such that at least 70% passes a 200 mesh screen (75 mm) and less
than 2% is retained on a 52 mesh screen (300 mm). The top size is determined by
the classifying component of the crushing mill, oversize material being retained
for further grinding (54,100,101).
Suspensions of pulverized coal or coal dust in air can be explosive, hence, it
is essential to have adequate guidelines and procedures to ensure safe and stable
operation during pulverized-coal (PC) firing. Some of these guidelines include:
(1) coal dust should never be allowed to accumulate except in specified storage
facilities; (2) coal and dust suspensions in air should not exist except in drying,
pulverizing, or burner equipment and the necessary transportation ducts; (3) the
furnace and its setting should be purged before introducing any light or spark;
(4) before introducing the fuel into the furnace, a lighted torch or spark-produ-
cing device should be in operation; (5) a positive flow of secondary air should be
maintained through the burners into the furnace and up the stack; and (6) a posi-
tive flow of primary air-coal to the burner should be maintained (81,82).
As for oil and gas, the burner is the principal device required to successfully
fire pulverized coal. The two primary types of pulverized-coal burners are circu-
lar concentric and vertical jet-nozzle array burners. Circular concentric burners
are the most modern and employ swirl flow to promote mixing and to improve
flame stability. Circular burners can be single or dual register. The latter type
was designed and developed for NOx reduction. Either one of these burner
types can be equipped to fire any combination of the three principal fuels, ie,
coal, oil and gas. However, firing pulverized coal with oil in the same burner
should be restricted to short emergency periods because of possible coke forma-
tion on the pulverized-coal element (71,72).
478 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

The self-igniting characteristics of pulverized coal vary from one coal to


another, but for most coals it is possible to maintain ignition without auxiliary
fuel when firing above the capacity of the boiler. The igniters may have to be acti-
vated in the following cases: (1) when firing pulverized coal with volatile matter
less than about 25%, (2) when firing excessively wet coal, and (3) when feeding
coal sporadically into the pulverizer (81,102–104).
Compared to natural gas and oil, complete combustion of coal requires
higher levels of excess air, about 15% as measured at the furnace outlet at
high loads, and this also serves to avoid slagging and fouling of the heat absorp-
tion equipment.
The process of coal ignition in the flame involves a number of steps. Initi-
ally, the pulverized coal is heated by convection as the flame jet entrains and
mixes with the furnace gases and also by radiation from the hotter furnace
gases. On heating to temperatures above about 773 K, the coal starts to decom-
pose, and evolves a mixture of combustible gases such as CO, H2, and hydrocar-
bons (CnHm) as well as noncombustible gases such as CO2 and H2O. At
temperatures of about 1173 K most of the volatile matter has been evolved
and, given adequate mixing of air in the jet, its combustion will sustain the igni-
tion of the flame. The char residue remaining after the devolatilization is then
burned relatively slowly in the flame and furnace. Char combustion has been
the subject of intensive investigation since the early 1930s and is one of the
least understood areas in coal combustion. Good reviews on char combustion
are available (49,105,106). For more information on the industrial applications
of coal, References (107–110) are recommended and for a thorough review of
coal devolatilization see References (111–114) (see COAL CONVERSION PROCESSES).
As pulverized-coal combustion potentially has a significant impact on the
environment, the 1980s saw the employment of techniques such as coal washing
and beneficiation to reduce the emissions of fly-ash, SOx, and water-soluble
metallic oxides. Fly-ash emissions can be reduced by means of electrostatic pre-
cipitators and fabric fillers, with efficiencies higher than 99.8%. SOx emissions
are reduced considerably by means of gas scrubbing which employs water slur-
ries of lime and limestone. Staged combustion is an effective method of reducing
NOx emissions from coal combustion. In this method, combustion takes place in
two zones: a low temperature fuel-rich zone and a high temperature fuel-lean
zone, where hydrocarbons and CO are afterburned. In the fuel-rich zone the
fuel bound nitrogen is converted to N2 rather than to NOx as in the case at
high values of f.
The environmental impact associated with pulverized-coal firing has given
rise to efforts to develop other combustion technologies such as fluidized beds or
the use of coal-water slurry fuels (CWSF), which can be burned as substitutes for
certain liquid fuels (115–117). CWSFs were developed as alternatives to more
expensive and increasingly scarce conventional hydrocarbon fuels. In their
most common form, they are composed of 70–75% by weight of coal (usually
high volatile A bituminous), 24–29% water, and 1% chemical additives. The prin-
cipal potential market for CWSF is as a replacement for residual fuel oil, ie,
heavy fuel oil and No. 6 fuel oil, in utility and large industrial boilers. CWSFs
offer all the advantages of liquid fuels in addition to the cost advantages asso-
ciated with the use of coal. The main challenge in the utilization of CWSFs is
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 479

obtaining stable mixtures that can be successfully atomized and burned. Much
research has been carried out in the area of CWSF atomization but this phenom-
enon is far from being well understood. On the combustion front, novel techni-
ques such as the coupling of a high intensity acoustic field have been employed
to enhance the convective processes occurring during the combustion of CWSF
(116,117).
Fluidized-Bed Technology. In fluidized-bed combustion of coal, air is fed
into the bed at a sufficiently high velocity to levitate the particles. This velocity is
referred to as the minimum fluidizing velocity, umf. At this velocity, the volume
occupied by the bed increases abruptly and the bed exhibits some of the charac-
teristics of a fluid. The two predominant designs of fluidized beds are bubbling
and recirculating, with most theories of fluidization being based on the simpler
bubbling bed concept.
Fluidized combustion of coal entails the burning of coal particles in a hot
fluidized bed of noncombustible particles, usually a mixture of ash and limestone.
Once the coal is fed into the bed it is rapidly dispersed throughout the bed as it
burns. The bed temperature is controlled by means of heat exchanger tubes.
Elutriation is responsible for the removal of the smallest solid particles and
the larger solid particles are removed through bed drain pipes. To increase com-
bustion efficiency the particles elutriated from the bed are collected in a cyclone
and are either re-injected into the main bed or burned in a separate bed operated
at lower fluidizing velocity and higher temperature.
Fluidized beds are ideal for the combustion of high sulfur coals since the
sulfur dioxide produced by combustion reacts with the introduced calcined lime-
stone to produce calcium sulfate. The chemistry involved can be simplified and
reduced to two steps, calcination and sulfation.

Calcination
CaCO3 ! CO2 þ CaO

Sulfation
SO2 þ CaO þ 12 O2 ! CaSO4

The main steps associated with coal combustion (heating, devolatilization,


volatiles combustion, and char burnout), occur sequentially to some extent; how-
ever, there is always some overlap between the stages. Char burnout is the slow-
est step so there is practical interest in determining the factors that influence its
rate. In order to determine the char combustion rate and time, it is necessary to
understand the interaction between the rate of oxygen diffusion to the reacting
surface and the inherent chemical kinetics of char oxidation. In the case of flui-
dized beds the use of a simplified rate coefficient overestimates the burnout time
substantially. The enhancement of mass transfer through the boundary layer as
the result of an applied velocity must be considered in order to predict char com-
bustion times under conditions relevant to fluidized-bed combustion. Char com-
bustion in fluidized beds is believed to be controlled by both diffusional and
chemical kinetic parameters, ie, mixed control. This indicates that models
attempting to predict char burnout times in fluidized beds must consider both
oxygen diffusion rates and inherent chemical kinetics (117–120).
480 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

The main stages of coal combustion have different characteristic times in


fluidized beds than in pulverized coal combustion. Approximate times are a
few seconds for coal devolatilization, a few minutes for char burnout, several
minutes for the calcination of limestone, and a few hours for the reaction of
the calcined limestone with SO2. Hence, the carbon content of the bed is very
low (up to 1% by weight) and the bed is 90% CaO in various stages of reaction
to CaSO4. About 10% of the bed’s weight is made up of coal ash (91). This distri-
bution of 90/10 limestone/coal ash is not a fixed ratio and is dependent on the ash
content of the coal and its sulfur content.
Devolatilization and combustion occur close to the coal inlet tubes. How-
ever, because of rapid mixing in the bed the composition of the solids in the
bed may be assumed to be uniform.
Atmospheric Pressure Fluidized-Bed Boilers. A typical bubbling flui-
dized-bed is usually 1.2 m deep in its expanded or fluidized condition. Normally,
the heat-transfer surface is placed in the bed in the form of a tube handle to
achieve the desired heat balance and bed operating temperature. For fuels
with low heating values the amount of surface can be minimal or absent. Coal-
fired bubbling-bed boilers normally incorporate a recycle system that separates
the solids leaving the economizer from the gas and recycles them to the bed. This
maximizes combustion efficiency and sulfur capture. Normally, the amount of
solids reacted is limited to about 25% of the combustion gas weight. For highly
reactive fuels this recycle system can be omitted. Bubbling-beds that burn coal
usually operate in the range of 2.4 to 3 m/s superficial flue gas velocity at max-
imum load. The bed material size is 590 mm and coarser, with a mean size of
about 1000–1200 mm.
Circulating fluidized-beds do not contain any in-bed tube bundle heating sur-
face. The furnace enclosure and internal division wall-type surfaces provide the
required heat removal. This is possible because of the large quantity of solids that
are recycled internally and externally around the furnace. The bed temperature
remains uniform, because the mass flow rate of the recycled solids is many times
the mass flow rate of the combustion gas. Operating temperatures for circulating
beds are in the range of 816 to 8718C. Superficial gas velocities in some commercially
available beds are about 6 m/s at full loads. The size of the solids in the bed is usually
smaller than 590 mm, with the mean particle size in the 150–200 mm range (81).
Some of the advantages of fluidized beds include flexibility in fuel use, easy
removal of SO2, reduced NOx production due to relatively low combustion tem-
peratures, simplified operation due to reduced slagging, and finally lower costs
in meeting environmental regulations compared to the conventional coal burning
technologies. Consequently, fluidized-bed combustors are currently under inten-
sive development and industrial size units (up to 150 MW) are commercially
available (Fig. 10).
The modeling of fluidized beds remains a difficult problem since the
usual assumptions made for the heat and mass transfer processes in coal combus-
tion in stagnant air are no longer valid. Furthermore, the prediction of bubble
behavior, generation, growth, coalescence, stability, and interaction with heat
exchange tubes, as well as attrition and elutriation of particles, are not well
understood and much more research needs to be done. Good reviews on various
aspects of fluidized-bed combustion appear in References 121 and 122 (Table 2).
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 481

Oxidant flow
Jets near distributor
Bubble formation u – umf
and coalescence
Exchange
Emulsion phase umf

Solid circulation
Entrainment in bubble wake
Elutriation by bursting of bubbles
SO2
sorbent
Fines
recycle
Spent
Coal
sorbent

Air

Fig. 10. The main processes taking place in a fluidized bed (92). Heat transfer to
immersed tubes is 30% by radiation.

Table 2. Operating Conditions for an Atmospheric Pressure Fluidized-Bed Combustora


Representative
Process value Comments
excess air, % 30 values selected to maintain CO emissions at
acceptable levels
bed height, m 1:5 trade-off between pressure drop and gas
residence time
bed temperatures, K 1100 higher values favor higher combustion
efficiencies, higher rates of NO reduction
by char; high values reduce SO2 capture
calcium/sulfur 2:5 value may be reduced by use of more active
(stoichiometric ratio) stone, lower gas velocities, deeper beds
gas velocity, m/s 2 high values are favored by high energy
release rates per unit plan area; lower
values are favored by higher efficiencies
of combustion and SO2 capture
sorbent particle top 3 low values are favored by shorter reaction
size, mm times; size determined by consideration
of elutriation at operating velocities
coal particle top size, 3 trade-off between elutriation rate and
mm reaction time
sorbent residence 5 104 determined by stone reactivity and
time, s Ca/S ratio
gas residence time in  14 an additional residence time of 1–2 s is
bed, s available in the free board
coal particle burning 400 dependent on temperature, size, and
time, s excess air
solid circulation time, s 2 short relative to solid reaction time
a
Ref. 122.
482 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

3.4. Design Considerations in Fossil Fuel Combustion Systems.


One of the most important considerations in the design of a combustion chamber
for a boiler is the fuel that is to be burned in the chamber. Although all fuels burn
and release heat during combustion, the rate at which a fuel burns and releases
heat, and the impurities associated with the fuel have to be considered.
Furnaces for Oil and Natural Gas Firing. Natural gas furnaces are rela-
tively small in size because of the ease of mixing the fuel and the air, hence the
relatively rapid combustion of gas. Oil also burns rapidly with a luminous flame.
To prevent excessive metal wall temperatures resulting from high radiation
rates, oil-fired furnaces are designed slightly larger in size than gas-fired units
in order to reduce the heat absorption rates.
Furnaces for Pulverized Coal Firing. The main differences between boi-
lers fired with coal and those fired with oil or natural gas result from the presence
of mineral matter in coals. The volume of the coal-fired furnace is higher because of
the longer residence time required for the complete combustion of coal particles, the
requirement of a controlled combustion rate to reduce NOx formation, the provision
for a larger heat-transfer surface area resulting from decreased heat-transfer rates
because of ash deposits on the surfaces, and increased spacing of heat-transfer
tubes to reduce flue gas velocities and thereby erosion of heat-transfer surfaces.
Even when firing coal, depending on the reactivity of the coal (rank), the size of
the combustion chamber required can vary (Fig. 11).

Plan Plan
Plan Plan area 1.26W area 1.29W
Plan
area 1.08W area 1.16W = 1.56 = 1.63
area W
= 1.15 = 1.25
= 1.00

D 1.06D 1.08D 1.24D 1.26D

1.52h 2.1h

1.17h
1.15h
h
1.45H
1.30H
1.07H
1.05H
H

A B C D E

Fig. 11. Effect of coal rank on furnace sizing (constant heat output) (82), where
W ¼ width, D ¼ depth, and h and H are the heights indicated. A represents medium
volatile bituminous; B, high volatile bituminous or subbituminous; C, low sodium lignite;
D, medium sodium lignite; and E, high sodium lignite.
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 483

The combustion chamber of a modern steam generator is a large water-


cooled chamber in which fuel is burned. Firing densities are important to ensure
that the chamber wall metal temperatures do not exceed the limits of failure of
the tubes. Firing densities are expressed in two ways: volumetric combustion
intensities and area firing intensities. The volumetric combustion intensity is
defined by equation 23,

Jf hf
Iv ¼ ð23Þ
Vc P

where Iv ¼ volumetric combustion intensity, kJ/(m3  h); Jh ¼ fuel feed rate, kg/h;
hf ¼ heating value of the fuel, kJ/kg; and Vc ¼ volume of the combustion cham-
ber, m3.
The area firing density is defined by equation 24,

J f hf
Ia ¼ ð24Þ
Ac

where, Ia ¼ area firing density, kJ/(m2  h); Jh ¼ fuel feed rate, kg/h; hf ¼ heating
value of the fuel, kJ/kg; and Ac ¼ crosssection of a fluid-bed distributor plate or
grate area of a mechanical stoker or plan area in pulverized-coal combustors.
Table 3 provides some design parameters for fossil fuel burners.
3.5. Environmental Considerations. Atmospheric pollutants released
by combustion of fossil fuels fall into two main categories: those emitted di-
rectly into the atmosphere as a result of combustion and the secondary pollu-
tants that arise from the chemical and photochemical reactions of the primary
pollutants.

Table 3. Comparison of Design Parameters for Fossil Fuel Boilers


Coal

Parameter Gas Oil Grate Fluid bed Pulverized coal

heat rate, mW (t) 0.03–3000 0.03–3000 0.3–30 up to 30 30–3000


volumetric 250–450 250–450 250–750a up to 2000a 150–250
combustion (based on
intensity, kW/m3 bed volume)
area combustion 280–500 280–500 2000 3000 7500
intensity, kW/m2
fuel firing density, 30–100 250 15–30
kg/m3h
kg/m2h 6–11 6–11 40–250 up to 500 up to 1000
practical 1000–1600 1100–1700 1200–1300 850–950 1600–1700
combustion
temperature, 8C
combustion time, s 10 101 20  25 101 up to 5000 100–500 1  2
particle heating <1 103 –104 104 –105
rate, 8C/s
a
Based on the total combustion volume which includes space between the bed and the
convective tubes.
484 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

The main combustion pollutants are nitrogen oxides, sulfur oxides, carbon
monoxide, unburned hydrocarbons, and soot. Combustion pollutants can be
reduced by three main methods depending on the location of their application:
before, after, or during the combustion. Techniques employed before and after
combustion deal with the fuel or the burned gases. A third alternative is to mod-
ify the combustion process in order to minimize the emissions.
Nitrogen Oxides. From the combustion of fuels containing only C, H, and
O, the usual air pollutants or emissions of interest are carbon monoxide,
unburned hydrocarbons, and oxides of nitrogen (NOx). The interaction of the
last two in the atmosphere produces photochemical smog. NOx, the sum of NO
and NO2, is formed almost entirely as NO in the products of flames; typically 5
or 10% of it is subsequently converted to NO2 at low temperatures. Occasionally,
conditions in a combustion system may lead to a much larger fraction of NO2 and
the undesirable visibility thereof, ie, a very large exhaust plume.
NO is formed to some extent from N2 and O2 in flame products when N
atoms are produced at a significant rate. Above 1700 K, the important step in
the much studied Zeldovitch (thermal) mechanism is the production of N
atoms by:

O þ N2 ! NO þ N

This is followed by a very fast reaction:

N þ O2 ! NO þ O

When ½NO  ½NOeq , as is usually true in practice, its formation is essentially


irreversible, and its rate is proportional to [NO][N2] with a large temperature
dependence, an activation energy of 316 kJ/mol (75.5 kcal/mol). Unfortunately,
the rate becomes appreciable just in the range of typical hydrocarbon–air
flame conditions. If it is also assumed that ½O ¼ ½Oeq , the observed rate in
most lean-flame products in which N2 is roughly 75 mol % of the gas can be
approximated by
   
d½NO 3:3 1018 68700
¼ exp ðxO2 Þ1=2
dt T T

Here, xO2 is the mole fraction of O2 in the products at temperature T, and the rate
is given in ppm/ms. The exponential implies a large effective activation energy of
570 kJ/mol, the sum of that for the O–N2 reaction and half the dissociation
energy of O2. In typical hydrocarbon–air flames, the rate of NO formation by
the thermal mechanism can be shown to be about 8 ppm/ms, or in a 10 ms resi-
dence time the thermal NO would be about 80 ppm. If preheating the mixture
were to raise the gas temperature by 100 K, the rate of NO production would
be nearly tripled, making the NO concentration unacceptable. Conversely, the
rate can be reduced by the same amount by a 100 K reduction in temperature
by precooling or heat abstraction from the flame itself, or by dilution of the mix-
ture with excess air, steam, or other inert gas such as recirculated, relatively cool
exhaust gas. Control of thermal NOx thus involves reduction of the maximum
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 485

attainable temperature, or the residence time at high temperature, or both. Such


measures, however, always entail some compromise in stability and control, and
possibly also in the efficiency of the combustion process. The afterburning of CO
tends to be quenched by rapid temperature reduction, and the resulting increase
in the emission of CO must be balanced against the desired NOx reduction. Heat
abstraction or cooling of the flame always occurs to some extent by radiation from
the highly luminous flames produced by pulverized-coal or oil combustion, as is
typical in boilers and similar furnaces. When heat is rejected, eg, by a boiler fluid,
and not returned or recuperated to the unburned mixture, the maximum tem-
perature and thermal NOx formation will be reduced. An extension of this effect
has been applied to achieve low NOx emissions in some furnaces and boilers in
which combustion occurs in a very rich, relatively low temperature primary
stage, followed by heat abstraction by convection as well as radiation to reduce
the gas enthalpy (two-stage combustion). Secondary products and any excess air
are then introduced to complete the combustion and, owing to the previous heat
transfer, the maximum temperature attainable in that stage will never approach
the adiabatic flame temperature. Much soot, which is responsible for the radia-
tive heat loss, may be present in the rich primary flame products. To avoid smoke
from such two-stage processes, care must be taken to assure its oxidation in the
second stage. In practice [O]/[O]eq is seldom unity as assumed. Though [O] is
decreased in the burned gas, its average value may be several times [O]eq and
NO formation may be correspondingly higher than predicted from the [O]eq.
Similarly, the very high radical concentration, eg, [O], in the reaction zone of a
flame often leads to almost instantaneous NO production, even though the tem-
perature is still relatively low and the residence time is relatively short. Other
fast reactions involving transient flame species producing N atoms, for example,
CH þ N2 ! HCN þ N, can also contribute to the production of some NO. In any
case, the NO inevitably formed by these species is called prompt NO. The total
concentration of prompt NO is usually not large, 10–50 ppm depending on the
composition of the flame, but is significant if very low NOx emission levels are
sought.
A different and often more serious source of NOx is chemically bound nitro-
gen in the fuel, eg, NH3, amines, nitrites, pyridine, a fraction of which is always
converted to NO. Most coals contain at least 1% N, of which 50% or more is
retained in gaseous or liquid fuels derived from coal. The N content of distillate
and gaseous petroleum fuels from most sources is usually very low, but it can
reach 0.5% and if such a fuel were burned with air under stoichiometric or
lean conditions, the conversion of its fuel N to NO in the flame would yield up
to 400 ppm of NO in the burned gas. NOx production levels from fuel N and
NOx control measures are now well established from correlations of data from
flames and combustors. Formation occurs in the flame reaction zone by OH radi-
cal oxidation of intermediate species formed by the decomposition of the fuel N,
eg, NH2, NH, N, which if unoxidized would in a short time simply form stable N2.
Whether NO or N2 formation prevails depends on the flame conditions as well as
on the concentrations of the intermediates. At high levels of fuel N, NO can also
be converted directly to N2. In general, the yield (mol of NO/mol of fuel N) is
much higher in lean and stoichiometric flames than in rich flames, and asymp-
totically approaches unity at low fuel N concentrations. It decreases with
486 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

increasing fuel N, and at some level that depends on f, the NO concentration


becomes constant, albeit rather high, and any additional fuel N is converted
rapidly to N2, from which NO can then be formed only by the relatively slow ther-
mal process with oxygen.
Thus if combustion can be effected in two stages, with or without the inter-
mediate heat rejection for thermal NOx control discussed above, the conversion of
fuel N to NO can be largely circumvented by first, a primary stage at f ¼ 1:5  2
with a modest residence time to allow formation of N2 in the hot primary pro-
ducts, followed by rapid addition of secondary air to complete the combustion
at an effective f  0:8. There will of course be a maximum in the temperature
near f ¼ 1 in the course of the secondary air addition, but if the residence
time at that condition is minimized, the production of thermal NO will also
be minimized.
Combustion system developments for reducing NOx formation include: low
NOx burners, staged burning techniques, and flue gas recirculation (FGR). Some
postcombustion techniques for reducing NOx include: selective noncatalytic
reduction (SNCR) and selective catalytic reduction (SCR). In either technology,
NOx is reduced to nitrogen (N2) and water (H2O) through a series of reactions
with a chemical agent injected into the flue gas. The most common chemical
agents used commercially are ammonia and urea for SNCR and ammonia for
SCR systems. Most ammonia-based systems have used anhydrous ammonia
(NH3) as the reducing agent. However, because of the hazards of storing and
handling NH3, many systems use aqueous ammonia at 25–28% concentration.
Urea can be stored as a solid or mixed with water and stored in solution (81).
These ideas form the basis of most approaches to NOx control with N-
containing fuels. In principal, they are readily applicable to the modification of
certain combustors in which the desired divisions in the combustion process
exist for other reasons. Although such improvements have been demonstrated,
it is difficult in practice to make the required revisions in the air and fuel distri-
bution without adverse effects on other emissions or on performance. It has also been
shown that when steam is used to reduce thermal NOx production, the formation
of NOx from fuel N is enhanced, or the reduction is less than otherwise expected.
Sulfur Oxides. Oxides of sulfur are also pollutants of concern. When pre-
sent in the fuel as inorganic sulfides or organic compounds, sulfur is converted
almost completely to SO2 in the products of complete combustion. There are no
known techniques for the elimination of this conversion process in flames, and
emission control measures necessarily involve either desulfurization of the fuel
or removal of the SO2. Up to 10% of the SO2 is oxidized to SO3 at low tempera-
tures in most combustion processes, and the total sulfur oxides emission is often
given as the sum of the concentrations of SO2 and SO3, or SOx. The presence of
H2O in the combustion products is inevitable and at low temperatures SO3 com-
bines with H2O to form H2SO4, which is both a highly corrosive agent to heat
exchange surfaces and a highly undesirable stack emission.
Soot. Emitted smoke from clean (ash-free) fuels consists of unoxidized
and aggregated particles of soot, sometimes referred to as carbon though it
is actually a hydrocarbon. Typically, the particles are of submicrometer size
and are initially formed by pyrolysis or partial oxidation of hydrocarbons in
very rich but hot regions of hydrocarbon flames; conditions that cause smoke
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 487

will usually also tend to produce unburned hydrocarbons with their potential
contribution to smog formation. Both may be objectionable, though for different
reasons, at concentrations equivalent to only 0.01–0.1% of the initial fuel.
Although their effect on combustion efficiency would be negligible at these levels,
it is nevertheless important to reduce such emissions.
Neither soot nor unburned hydrocarbons are found in the products of a lean
or stoichiometric premixed flame with air or O2, although hydrocarbons may be
formed or survive unburned in lean flames partially quenched, eg, by the cold
wall of a combustion chamber. A moderately rich flame also should yield only
the water-gas equilibrium products (and N2); but with increasing equivalence
ratio, at some f still well below the upper flammability limit, the burned gas
becomes faintly luminous with precipitated soot particles that increase in num-
ber density and luminous intensity with further increase in f. The appearance of
the condensed phase (soot) is connected with appreciable nonequilibrium concen-
trations of fairly stable low molecular-weight hydrocarbons, notably acetylene,
from the decomposition of unoxidized fuel. These polymerize (condense with
elimination of H2) to form high molecular-weight products, mostly with ring
structures. If these intermediates are not consumed, eg, by OH, in simultaneous
and competing oxidation reactions, they grow until nucleation, and eventually
precipitation occurs as visibly radiating soot, typically over a period of several
milliseconds after the main flame reaction zone. The particle growth and compet-
ing oxidation then continue in the burned gas.
The composition, properties, and size of soot particles collected from flame
products vary considerably with flame conditions and growth time. Typically the
C–H atomic ratio ranges from two to five and the particles consist of irregular
chains or clusters of tiny spheres 10–40 nm in diameter with overall dimensions
of perhaps 200 nm, although some may agglomerate further to much larger sizes.
Whether soot particles form at all and grow depends on f, the fuel type, and
other variables, eg, the growth is easier and faster at a given f with fuels of
higher C–H ratio and at elevated pressure. Given sufficient residence time to
attain equilibrium at the burned gas condition, the soot and hydrocarbons
would eventually be consumed. In practice, their rapid oxidation occurs in a sec-
ondary flame in which the hot primary products are burned with the required
excess air, which is added by diffusion or by more intensive mixing. However,
if a large excess of air is added too rapidly, the cooling can, in effect, quench
the oxidation of both unburned hydrocarbons and the accompanying soot,
which would then persist as visible smoke. The blackness of the smoke depends
on the size and number density of the particles when quenched, their further
aggregation, etc. Some may also survive as much smaller invisible particles, or
condensation nuclei. For thorough reviews on the mechanisms of pollutant for-
mation see References (123–125).
Diffusion Flame Chemistry. Since most combustion systems employ mix-
ing-controlled diffusion flames, which are characterized by very high pollutant
emissions, it is imperative to look into the chemistry occurring in diffusion
flames. In a typical diffusion flame the mixture composition in the reaction
zone is close to the stoichiometric proportion and the temperature is at a maxi-
mum resulting from the large volume of this zone, thus NOx production is
favored. If, however, the surrounding gas cools the combustion products rapidly,
488 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

further reactions of CO and NO are eliminated. This fixes the concentrations of


these pollutants at unfavorable levels. Furthermore, the fuel diffuses into the
combustion zone through the burned gases and thus is heated in the absence
of oxygen. This creates ideal conditions for the formation of soot and the reduc-
tion of the CO2 produced in the combustion zone to CO. Additionally, diffusion
flames have low combustion intensity and efficiency and hence release large
amounts of unburned hydrocarbon emissions. In general, despite the fact that
the structure of the diffusion flame is more complex and difficult to analyze,
the same basic description of soot formation and oxidation should apply to diffu-
sion flames as for premixed flames.
Emissions Control. From the combustion chemistry standpoint, lean
mixtures produce the least amount of emissions. Hence, one pollution prevention
alternative would be to use lean premixed flames. However, lean mixtures are
difficult to ignite and form unstable flames. Furthermore, their combustion
rates are very low and can seldom be applied directly without additional mea-
sures being taken. Consequently the use of lean mixtures is not practical.
Another potential solution is the use of catalytic combustors, which produce
extremely low levels of emissions by the use of combustion catalysts such as pla-
tinum. The main disadvantage of catalytic combustors, however, is their high cost.
More advanced techniques for emissions control include electrical or plasma
jet augmentation of flames based on radical production. Since in two-phase, het-
erogeneous combustion the flames are always diffusion flames on the microscale,
ie, the individual droplets or particles burn as diffusion flames, and since at the
characteristic times for evaporation, decomposition and burning of individual
particles can be comparable with the characteristic times for mixing and pollu-
tant formation, prevaporization or gasification of the fuel can reduce pollutant
emissions. For this reason catalytic systems for liquid-fuel decomposition and
coal gasification are being considered seriously as alternatives to conventional
combustion technology (126–128).

BIBLIOGRAPHY

‘‘Fuels (Combustion Calculation),’’ in ECT 1st ed., Vol. 6, pp. 913–935, by H. R. Linden,
Institute of Gas Technology; in ECT 2nd ed., Vol. 10, pp. 191–220, by D. M. Himmelblau,
The University of Texas; ‘‘Burner Technology,’’ in ECT 3rd ed., Vol. 4, pp. 278–312, by
G. E. Moore, Consultant; in ECT 4th ed., Vol. 6, pp. 1049 –1092, by R. Sharifi, S. A.
Pisupati, and A. W. Scaroni, Pennsylvania State University.

CITED PUBLICATIONS

1. N. Chigier, ‘‘Energy,’’ Combustion and Environment, McGraw-Hill, Inc., New York,


1981.
2. W. Bartok and A. F. Sarofim, Fossil Fuel Combustion: A Source Book, John Wiley &
Sons, Inc., New York, 1991.
3. J. A. Barnard and J. N. Bradley, Flame and Combustion, 2nd ed., Chapman and
Hall, London and New York, 1985.
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 489

4. W. E. Baker, P. A. Cox, P. E. Westine, J. Kulesz, and R. A. Strehlow, Explosion


Hazards and Evaluation, Elsevier, Amsterdam, The Netherlands, 1983.
5. R. A. Strehlow, Combustion Fundamentals, McGraw-Hill, Inc., New York, 1984.
6. D. J. McCracken, Hydrocarbon Combustion and Physical Properties, Rep. No. 1496,
Ballistic Research Laboratories, Sept. 1970.
7. M. G. Zabetakis, Flammability Characteristics of Combustible Gases and Vapors,
Technical Bulletin, U.S. Bureau of Mines, Washington, D.C., 1965, p. 627.
8. I. Glassman, Combustion, Academic Press, Inc., New York, 1987.
9. J. Chomiak, Combustion; A Study in Theory, Fact and Application, Gordon and
Breach Science Publishers, Montreux, Switzerland, 1990.
10. B. Lewis and G. von Elbe, Combustion, Flames and Explosion of Gases, 2nd ed.,
Academic Press, Inc., New York, 1951.
11. F. A. Williams, Combustion Theory, The Benjamin/Cummings, Publishing Co. Inc.,
Menlo Park, Calif., 1985.
12. K. K. Kuo, Principles of Combustion, John Wiley & Sons, Inc., New York, 1986.
13. R. C. Flagen and J. H. Seinfeld, Fundamentals of Air Pollution Engineering,
Prentice-Hall, Inc., Englewood Cliffs, N.J., 1988.
14. H. B. Palmer and J. M. Beer, eds., Combustion Technology: Some Modern Develop-
ments, Academic Press, New York and London, 1974.
15. M. V. Blank, P. G. Guest, G. von Elbe, and B. Lewis, Third Symposium, Combustion,
Flame and Explosion Phenomena, Williams & Wilkens, Baltimore, Md., 1949.
16. H. C. Barnett and R. R. Hibbard, Basic Considerations in the Combustion of Hydro-
carbon Fuels with Air, NASA Technical Report, 1959, p. 1300.
17. H. F. Colcote, C. A. Gregory, C. M. Barnett, and R. B. Gilmer, Ind. Eng. Chem. 44,
2656 (1952).
18. Ya. B. Zeldovich, G. I. Brenblatt, V. B. Librovich, and G. M. Mackhviladze, The
Mathematical Theory of Combustion and Explosions, Consultants Bureau, 1985.
19. V. R. Kuznetsov and V. A. Sabel’nikov, Turbulence and Combustion, Hemisphere
Publishing Corp., Washington, D.C., 1990.
20. M. Y. Hussaini, A. Kumar, and R. G. Voigt, Major Research Topics in Combustion,
Springer-Verlag, Inc., New York, 1992.
21. C.-M. Brauner and C. Schmidt-Laine, Mathematical Modeling in Combustion and
Related Topics, Martinus Hijhoff Publishers, Dordrecht, The Netherlands, 1988.
22. R. M. C. So, H. C. Mongia, and J. H. Whitelaw, Turbulent Reactive Flow Calcula-
tions, special issue of Combustion Science and Technology, Gordon and Breach
Science Publishers, Inc., Montreux, Switzerland, 1988.
23. F. A. Williams, in W. Bartok and A. F. Sarofim, eds., Turbulent Reacting Flows In
Fossil Fuel Combustion: A Source Book, John Wiley & Sons, Inc., New York, 1991.
24. F. A. Williams, in J. H. S. Lee and C. M. Cuirao, eds., Laminar Flame Instability and
Turbulent Flame Propagation, In Fuel-Air Explosions, University of Waterloo Press,
Waterloo, Ontario, Canada, 1982.
25. J. M. Beer and N. A. Chigier, Combustion Aerodynamics, Applied Science, London;
John Wiley & Sons, Inc., New York, 1972.
26. L. D. Landau, Acta. Phys. Chim. (USSR) 19, 77 (1944).
27. L. D. Landau and E. M. Lifschitz, Mechanics of Continuous Media, Moscow, Russia,
1953; Eng. trans. Addison-Wesley Publishing Co., Inc., Reading, Mass., 1959.
28. A. C. Scurlock and J. H. Grover, 4th International Symposium on Combustion, The
Combustion Institute, Pittsburgh, Pa., 1953, p. 645.
29. K. N. C. Bray, Equations of Turbulent Combustion, Report No. 330, University of
Southampton, UK, 1973.
30. K. N. C. Bray, 17th International Symposium on Combustion, The Combustion In-
stitute, Pittsburgh, Pa., 1978, 57–59.
490 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

31. K. N. C. Bray and P. A. Libby, Phys. Fluids 19, 1687–1701 (1976).


32. K. N. C. Bray and J. B. Moss, Comb. Flame 30, 125 (1977).
33. H. C. Hottel, in Ref. 28, p. 97–113.
34. S. P. Burke and T. E. W. Schumann, Ind. Eng. Chem 20, 998–1004 (1928).
35. W. Jost, Explosion and Combustion Processes in Gases, McGraw-Hill, Inc., New
York, 1946.
36. W. Jost, Diffusion, Steinkopff, Darmstadt, Germany, 1957.
37. A. G. Gaydon and H. G. Walfhard, Flames, 2nd ed., Macmillan, New York, 1960.
38. T. Takeno and Y. Kotani, in L. A. Kennedy, ed., Turbulent Combustion, Prog. Astro-
naut. Aeronaut., Vol. 58, AIAA, 1978, 19–35.
39. M. Gerstein, Diffusion Flames, in Ref. 2.
40. H. C. Hottel, 3rd International Symposium on Combustion, The Combustion Insti-
tute, Pittsburgh, Pa., 1948, 254–266.
41. P. A. Libby and F. A. Williams, eds., Turbulent Reacting Flows, Springer-Verlag,
Berlin and New York, 1980, 1–43, 219–236.
42. A. M. Mellor, in Ref. 31, 377–387.
43. F. A. Williams, in W. E. Stewart, W. H. Ray, and C. C. Conley, eds., Current
Problems in Combustion Research, Dynamics and Modeling of Reactive Systems,
Academic Press, Inc., New York, 1980.
44. P. A. Libby and F. A. Williams, AIAA J. 19, 261–274 (1981).
45. P. A. Libby and F. A. Williams, Annu. Rev. Fluid Mech. 8, 351–379 (1976).
46. R. W. Bilger, Energy and Combustion Science, Student ed. 1, Pergamon Press,
Oxford, 1979, 109–131.
47. D. B. Spalding, Some Fundamentals of Combustion, Butterworth & Co., Ltd.,
London, 1955.
48. R. Borghi, Turbulent Combustion Modeling, Progress in Energy and Combustion
Science, Vol. 14, No. 4, Pergamon Press, Elmsford, N.Y., 1988, 245–292.
49. I. W. Smith, 19th International Symposium on Combustion, The Combustion Insti-
tute, Pittsburgh, Pa., 1982, p. 1045.
50. R. B. Jones, B. B. McCourt, C. Morley, and K. King, Fuel 64, 1460 (1985).
51. N. Oka, T. Murayama, H. Matsuoka, and S. Yamada, Fuel Process. Tech. 15, 213
(1987).
52. M. Shibaoka, Fuel 48, 285 (1969).
53. J. M. Kuchta, V. R. Rowe, and D. S. Burgess, U.S. Bureau of Mines Report 8474,
Washington, D.C., 1980, p. 1.
54. R. H. Essenhigh, in M. A. Elliot, eds., Fundamentals of Coal Combustion, In Chem-
istry of Coal Utilization, 2nd Suppl. Vol., John Wiley & Sons, Inc., New York, 1981.
55. N. N. Semenov, Chemical Kinetics and Chain Reactions, Oxford University Press,
London, 1935.
56. F. Kamenetski, Diffusion and Heat Exchange in Chemical Kinetics, Princeton
University Press, Princeton, N.J., 1955; trans. from Russian by Thanel.
57. K. Annamalai and P. Durbetaki, Comb. Flame 29, 193 (1977).
58. K. Annamalai, Trans. ASME 101, 576 (1979).
59. L. D. Smoot, M. D. Horton, and G. A. Williams, 16th International Symposium on
Combustion, The Combustion Institute, Pittsburgh, Pa., 1976, p. 375.
60. R. H. Essenhigh, M. Misra, and D. Shaw, Comb. Flame 77, 3 (1989).
61. H. E. Blayden, W. Noble, and H. L. Riley, Gas J. 2, 81 (1934).
62. H. K. Griffin, J. P. Adams, and D. F. Smith, Ind. Eng. Chem. 21, 808 (1929).
63. G. P. Ivanova and V. Babii, Thermal Eng. 13, 70 (1966).
64. A. A. Orning, Proceedings of Conference on Pulverized Fuel, Vol. 1, Institute of Fuel,
London, 1947, p. 45.
65. D. L. Carpenter, Comb. Flame 1, 63 (1957).
Vol. 2 COMBUSTION SCIENCE AND TECHNOLOGY 491

66. H. M. Cassel and I. Liebman, Comb. Flame 3, 467 (1959).


67. J. B. Howard and R. H. Essenhigh, Comb. Flame 9, 337 (1965).
68. M. A. Nettleton and R. Stirling, Comb. Flame 22, 407 (1974).
69. L. Tognotti, A. Malotti, L. Petarca, and S. Zanelli, Comb. Flame 44, 15 (1985).
70. M. R. Chen, L. S. Fan, and R. H. Essenhigh, 20th International Symposium on Com-
bustion, The Combustion Institute, Pittsburgh, Pa., 1984, p. 1513.
71. C. P. Gomez, Ignition of and Combustion of Coal and Char Particles; a Differential Ap-
proach, M.S. dissertation, Pennsylvania State University, University Park, Pa., 1982.
72. A. W. Scaroni, M. R. Khan, S. Eser, and L. R. Radovic, Ullmann’s Encyclopedia of
Industrial Chemistry, Vol. A7, VCH Publishers, New York, 1986, p. 245.
73. J. D. Blackwood and F. K. McTaggart, Aust. J. Chem. 12, 533 (1959).
74. R. G. Jenkins, S. P. Nandi, and P. L. Walker, Jr., Fuel 52, 288 (1973).
75. D. W. Van Krevelen, Coal Typology–Chemistry–Physics–Constitution, Elsevier
Publications, New York, 1961, p. 219.
76. E. J. Badin, Coal Combustion Chemistry—Correlation Aspects, Elsevier, New York,
1984, Chapt. 6, p. 68.
77. D. W. McKee, in P. L. Walker, Jr. and P. A. Thrower, eds., Chemistry and Physics of
Carbon, Vol. 16, Marcel Dekker, Inc., New York, 1981, p. 1.
78. B. A. Morgan and A. W. Scaroni, International Conference on Coal Science, Interna-
tional Energy Agency, Sydney, Australia, 1985, p. 347.
79. P. L. Walker, Jr., M. Shelef, and R. A. Anderson, in Ref. 79, Vol. 4.
80. P. L. Walker, Jr., F. Rusinko, Jr., and L. G. Austin, in D. D. Eley, P. W. Selwood, and
P. B. Weisz, eds., Advances in Catalysis, Vol. 2, Academic Press, Inc., New York,
1959, p. 133.
81. S. C. Stultz and J. B. Kitto, eds., Steam, Its Generation and Use, 40th ed., Babcock
and Wilcox Co., Barberton, Ohio, 1992.
82. J. E. Singer, Combustion-Fossil Power Systems, Combustion Engineering, Windsor,
Conn., 1981.
83. American Gas Association, Gas Engineers Handbook, Industrial Press, New York, 1978.
84. S. S. Penner and B. P. Mullins, Explosions, Detonations, Flammability and Ignition,
AGARD Monograph, Pergamon Press, Inc., New York, 1959.
85. M. W. Thring, The Science of Flames and Furnaces, John Wiley & Sons, Inc., New
York, 1962.
86. A. Williams, Comb. Flame 21, 1 (1973).
87. A. Williams, Combustion of Sprays and Liquid Fuels, Elek Science, London, 1976.
88. A. Williams, Combustion of Liquid Fuels Sprays, Butterworths & Co., Ltd., London,
1990.
89. A. H. Lefebvre, Gas Turbine Combustion, Hemisphere Publishing Corp., Washington,
D.C., 1983.
90. A. H. Lefebvre, Atomization and Sprays, Hemisphere Publishing Corp., Washington,
D.C., 1989.
91. A. C. Fernandez-Pello and C. K. Law, in Ref. 51, p. 1037.
92. A. L. Randolf, A. Markino, and C. K. Law, 21st International Symposium on Com-
bustion, The Combustion Institute, Pittsburgh, Pa., 1988, p. 601.
93. J. C. Lasheras, L. T. Yap, and F. L. Dryer, in Ref. 72, p. 1761.
94. T. Niioka and J. Sato, in Ref. 94, p. 625.
95. A. H. Levebvre, Airblast Atomization, Progress in Energy and Combustion Science,
Vol. 6, Pergamon Press, Oxford, UK, 1980, 233–261.
96. N. K. Rizk and A. H. Levebvre, J. Fluids Eng. 97(3), 316–320 (1975).
97. A. Rizkalla and A. H. Levebvre, AIAA J. 21(8), 1139–1142 (1983).
98. C. J. Lawn and co-workers, The Combustion of Heavy Fuels Oils, in C. J. Lawn, ed., The
Principles of Combustion Engineering for Boilers, Academic Press, Inc., New York, 1987.
492 COMBUSTION SCIENCE AND TECHNOLOGY Vol. 2

99. T. F. Wall, The Combustion of Coal as Pulverized Fuel Through Swirl Burners, in
Ref. 100.
100. M. A. Field, D. W. Gill, B. B. Morgan, and P. J. W. Hawksley, Combustion of Pulver-
ized Coal, BCURA, UK, 1967.
101. H. H. Lowry, ed., Chemistry of Coal Utilization, Suppl. Vol., John Wiley & Sons, Inc.,
New York, 1963.
102. L. D. Smoot and D. T. Pratt, eds., Pulverized Coal Combustion and Gassification,
Plenum Press, New York, 1979.
103. L. D. Smoot, Coal and Char Combustion, in Ref. 2.
104. B. R. Cooper and W. A. Ellingson, The Science and Technology of Coal Utilization,
Plenum Press, New York, 1984.
105. M. F. R. Mulcahy and I. W. Smith, Rev. Pure. Appl. Chem. 19, 81 (1969).
106. N. M. Laurendeau, Prog. Energy Comb. Sci. 4, 221 (1978).
107. Modern Power Station Practice, 2nd ed., Pergamon Press, Oxford, UK, 1971.
108. L. D. Smoot and P. J. Smith, Coal Combustion and Gasification, Plenum Press, New
York, 1985.
109. A. Stambuleanu, Flame Combustion Process in Industry, Abacus Press, Tunbridge
Wells, UK, 1979.
110. Prog. Energy Comb. Sci., Special Issue 10, 81–293 (1984).
111. J. B. Howard, Fundamentals of Coal Pyrolysis and Hydropyrolysis, in Ref. 56.
112. C. Y. Wen and E. Stanley Lee, eds., Coal Conversion Technology, Addison-Wesley
Publishing Co., Reading, Mass. 1979.
113. R. H. Essenhigh, in Ref. 61, p. 372.
114. A. W. Scaroni, P. L. Walker, and R. G. Jenkins, Fuel 60, 70–76 (1981).
115. P. Ramachandran, A. W. Scaroni, and R. G. Jenkins, I. Chem., E. Symp. Ser. 107,
128–219 (1987).
116. P. Ramachandran, A. W. Scaroni, G. Reethof, and S. Yavuzkurt, 13th International Con-
ference on Coal and Slurry Technology, Slurry Technology Association, 1988, 241–247.
117. G. Huang and A. W. Scaroni, Fuel 71, 159–164 (1992).
118. J. R. Howard, ed., Fluidized Beds: Combustion and Applications, Elsevier, New
York, 1983.
119. R. Schweiger, Fluidized Bed Combustion and Application Technology, The 1st Inter-
national Symposium, Hemisphere Publishing Corp., New York, 1987.
120. H. R. Hoy and D. W. Jill, The Combustion of Coal in Fluidized Beds, in Ref. 100.
121. H. B. Palmer and C. F. Cullis, The Formation of Carbon from Gases, in P. L. Walker,
ed., Chemistry and Physics of Carbon, Marcel Dekker, New York, 1976.
122. S. Kaliaguine and A. Mahay, eds., Catalysts on the Energy Scene, Elsevier,
New York, 1984.
123. J. M. Beer, in Ref. 59, 439–460.
124. A. F. Sarofim and J. M. Beer, in Ref. 30, 189–204.
125. K. C. Taylor, Automobile Catalytic Converters, Springer-Verlag, New York, 1984.
126. H. G. Wagner, in Ref. 30, p. 3.
127. K. H. Homann, in Ref. 70, p. 857.
128. P. Kesselring, Catalytic Combustion, in F. Weinberg, ed., Advanced Combustion
Methods, Academic Press, Inc., New York, 1986.

REZA SHARIFI SARMA


V. PISUPATI
ALAN W. SCARONI
Pennsylvania State University
Vol. 2 EMISSION CONTROL, AUTOMOTIVE 493

EMISSION CONTROL,
AUTOMOTIVE
1. Introduction

There are over 210 million U.S. automobiles and trucks on the road, consuming
large amounts of gasoline and producing commensurately large amounts of gas-
eous exhaust. The exhaust consists primarily of carbon dioxide [124-38-9], water,
unburned hydrocarbons (qv), carbon monoxide [630-08-0], oxides of nitrogen
(NOx), remaining oxygen, and nitrogen (see GASOLINE AND OTHER MOTOR FUELS).
The latter three atmospheric pollutants have been regulated since the 1970s
by the U.S. government and more stringently by the State of California. Automo-
bile companies have developed fuel metering and exhaust systems using the cat-
alytic converter to meet emission regulations. Carbon dioxide emissions are
indirectly controlled by corporate average fuel economy (CAFÉ) standards for
passenger cars and small trucks.
Prior to emission control, passenger car and truck emissions together were
the largest contributors to atmospheric pollution in the United States (1). By
2004, the Tier 2 exhaust emission standards mandated by the Clean Air Act
Amendments of 1990 required automobiles and small trucks to reduce emissions
of hydrocarbons by 98%, carbon monoxide by over 90% and oxides of nitrogen by
98% as compared to 1970 pre-control emissions. California, having more severe
pollution events, has established a system of the most stringent emission control
regulations in the world. The California standards can be adopted by other
states.
The key components of the spark-ignited engine emission control systems
are the three-way catalytic converter and closed loop oxygen sensor (qv) (See
SENSORS) controlled fuel metering system. Exhaust gas recirculation is used to
reduce formation of engine NOx and provides antiknock values. Over 600 million
vehicles have been equipped with the catalytic converter which, has been rated
among the top 10 engineering breakthroughs of the twentieth century (2).
Emission control is achieved without negatively affecting fuel economy or
performance.
Diesel engine emission control technology is not discussed in detail as very
few light-duty diesel engines are now in service, and although the number
is expected to rise in this coming decade, the emissions control technology
has not been selected. Light-duty and medium-duty diesel engine powered vehi-
cles have to meet the same Tier 2 standards (Table 1) as gasoline-fueled engines
including a strict particle emission standard. Heavy-duty diesel engine trucks
must meet very stringent standards starting with the 2007 model year.

2. Emission Regulation and Testing

United States Federal regulations require automobile manufacturers obtain


a certificate of conformity indicating compliance with all applicable emission
standards over the vehicle’s useful life period. The Tier 2 useful life for car

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.0524080113151514.a01.pub2
494 EMISSION CONTROL, AUTOMOTIVE Vol. 2

and light-duty vehicles is 160,000 km or 10 years or 11 years and 193,000 km for


heavier light-duty vehicles.
2.1. Clean Air Act Amendments. The Clean Air Act Amendments of
1970 set strict emission standards requiring 90% reduction of emissions of
hydrocarbons, carbon monoxide and oxides of nitrogen. The resulting statutory
emissions standard was 0.41 grams per mile of hydrocarbons, 3.4 grams per mile
of carbon monoxide and 0.4 grams per mile of oxides of nitrogen. A series of
interim standards were sequentially set that finally reached this level. The
Clean Air Act Amendments of 1990 required more stringent standards and
resulted in Tier 1 and Tier 2 standards—the later being initiated in 2004 through
2007 for light-duty vehicles and fully implemented for heavy light-duty vehicles
in 2009. The Tier 2 standards essentially required all passenger cars and light-
duty trucks (LDT) including sports utility vehicles (SUV) up to 3,856 kg (8,500 lb)
gross vehicle weight (GVW) meet the same level standards as passenger cars for
193,000 kilometers or 10 years with a 25, 50, 75 and 100% phase-in schedule over
4 years. Heavier trucks and passenger carrying vehicles up to 4,536 kg (10,000 lb)
GVW have to meet the same standards for 193,000 km or 11 years with a 50,
100% phase-in schedule starting in 2008. Table 1 summarizes the Tier 2 stan-
dards. Tier 2 also set fuel sulfur content in gasoline to an average of 30-ppm
with a cap at 80-ppm. For complete list of Federal emission standards, go to
www.epa.gov/otaq/standards.htm (the Tier 2 Program) and State of California
LEV-II emission standards www.arb.ca.gov. A new concept of ‘‘bins’’ was intro-
duced. The heavier LDT and SUV vehicles can initially meet higher emissions

Table 1. US EPA-TIER 2 Emission Standardsa,b


Bin # NOx NMOG CO HCHO PM Commentsc
10d 0.372 0.097/0.143 2.61/3.98 0.011/0.017 0.056 a, b, c, d
9d 0.186 0.056/0.112 2.61 0.011 0.037 a, b, e
8 0.124 0.078/0.097 2.61 0.011 0.012 b, f
7 0.093 0.056 2.61 0.011 0.012
6 0.062 0.056 2.61 0.011 0.006
5 0.043e 0.056 2.61 0.011 0.006
4 0.024 0.044 1.31 0.007 0.006
3 0.019 0.034 1.31 0.007 0.006
2 0.012 0.006 1.31 0.0025 0.006
1 0.000 0.000 0.00 0.000 0.000
a
Ref. 3.
b
Light-duty and medium-duty vehicles less than 4536 kg gross vehicle weight. Tier 2 light-duty full
useful life exhaust emission standards (grams per kilometer). Full useful life: 10 yr. or 160,000 km up
to 3856 GVW; 11 yr. or 193,000 km up to 4536 GVW for passenger carrying vehicles.
c
Corporate average NOx value.
d
Temporary bins expire in 2006 (for LDVs and LLDTs) and 2008 (for HLDTs)
Notes:
a. Bin deleted at end of 2006 model year (2008 for HLDTs).
b. The higher temporary NMOG, CO and HCHO values apply only to HLDTs and expire after 2008.
c. An additional temporary higher bin restricted to MDPVs is discussed in reference.
d. Optional temporary NMOG standard of 0.143 g/km applies for qualifying LDT4s and MDPVs only.
e. Optional temporary NMOG standard of 0.112 g/km applies for qualifying LDT2s only is discussed
in reference.
f. Higher NMOG standard is deleted at end of 2008 model year.
Vol. 2 EMISSION CONTROL, AUTOMOTIVE 495

of temporary Bins 9 and 10 but eventually have to meet Bin 6, 7, 8 or lower. Any
vehicles sold by a manufacturer in one year in the higher bins must be offset by
an equivalent number of vehicles sold by a manufacturer in that same year that
meet the cleaner emissions of Bins 1, 2, 3, and 4, so as to counterbalance and meet an
overall average 0.07 g/mile NOx emission standard of Bin 5. California LEV2
standards are similar to those of Tier 2 with more emphasis on corporate average
HC emissions. For California automobile manufacturers can choose to comply
with voluntary emissions system durability of 15 years or 240,000 kilometers
at slightly relaxed standards. Extended emission system durability requirements
reflect much the improved and longer engine life now found in actual vehicle use
as a result of clean unleaded gasoline. Highly corrosive leaded gasoline has been
completely banned in the United States since 1996.
The U.S. Tier 2 and California LEV-II program will dominate advances in
emission control in this next decade.
2.2. Test Procedure. Originally vehicle manufacturers had to actually
run test track vehicles for 80,000 km as durability vehicles (3). Periodic exhaust
emissions tests using the U.S. Federal Test Procedure (FTP) determined emis-
sions at several mileages between 6,400 km and 80,000 km and a deterioration
factor (DF) was computed. For example, a calculated DF factor of 1.1 for the dur-
ability vehicle could be applied to the emissions found on any number of 6,400 km
emission data vehicles that represented various vehicle models within that
engine family to predict compliance with the in-use standard at 80,000 km.
Thus not all versions of vehicles had to be run for 80,000 km. Nevertheless,
this original certification procedure was laborious and expensive. A simplified
modern verification has been proposed for new emission durability procedures
using a new standard road cycle (SRC) for whole vehicle durability or an alter-
native method of new standard engine bench cycle (SBC) for accelerated aging of
the catalytic converter and oxygen sensor and other exhaust system components
(4). The objective is to use durability procedures that are at least as severe as
actual vehicle on-road use. The SRC would be used for whole vehicle durability
using 7 laps (each different) in a 3.7 mile test track course or on a mileage accu-
mulation dynamometer for the full useful life mileage. The SBC would be used to
age catalytic converter, oxygen sensor and exhaust system under accelerated test
conditions on a slave engine installed on an engine test bench. The aged compo-
nents are then installed and tested on an actual new vehicle model undergoing
compliance testing. The SBC uses a mathematical approach based on the
Arrehenius equation to determine aging time and exhaust gas temperature to
achieve catalyst and O2 sensor aging effects equivalent to that found with the
SRC whole vehicle durability cycle that are at least as severe as actual in-use
vehicle aging. Auto manufacturers are permitted to develop their own SRC
and SBC and have them approved. Tier 2 emission regulations require field
surveys by automobile manufacturers to assure that vehicles in actual use are
in compliance with the emission standards.
The Federal Test Procedure (FTP) specifies that a test vehicle is fueled with
commercial unleaded gasoline and be stored in a temperature environment of
between 20 and 298C for at least 12 hours immediately prior to the emission
test. The vehicle is placed on a chassis dynamometer which is calibrated for
the vehicle weight and road load. The vehicle is started cold and driven for
496 EMISSION CONTROL, AUTOMOTIVE Vol. 2

41 minutes on a prescribed cycle of accelerations, cruises, decelerations, idle


periods, a 10-minute shutdown (called the hot soak), and a hot start and period
of rerun that matches the driving conditions of the first 505 seconds of the cold
start. Total distance traveled is 17.8 km (11.115 miles). Fig. 1a is the USA FTP-75
Test Cycle for standard emission tests. Fig. 1b is the supplementary Federal Test
Cycle US06 High Speed/High Load Cycle which reaches a maximum speed of
129.2 kmh (80.3 mph) implemented to assure emissions are controlled during
higher rates of acceleration and higher speed. Exhaust emissions are sampled,

Length 1874 seconds - Distance = 11.04 miles - Average Speed = 21.2 mph
70
Cold start phase Transient phase Hot start phase
60 505 seconds 864 seconds 505 seconds
Vehicle speed, mph

50

40

30

20

10

0
0

132
198
264
330
396
462
528
594
660
726
792
858
924
990
66

1056
1122
1188
1254
1320
1386
1452
1518
1584
1650
1716
1782
1848
Test time, secs
(a)

Sample Period = 596 seconds - Distance = 8.01 miles - Average Speed = 48.37
90
80
70
Vehicle speed, mph

60
50
40
30
20
10
0
115
138
161
184
207
230
253
276
299
322
345
368
391
414
437
460
483
506
529
552
575
46
69
92
0
23

Test time, secs

(b)

Fig. 1. (a) EPA federal test procedure. (b) US 06 or supplemental FTP driving schedule.
Vol. 2 EMISSION CONTROL, AUTOMOTIVE 497

analyzed and computed with a constant volume sampling system (CVS). All
engine exhaust gases from test start to test end are fed into the (CVS) system
where it is diluted with large amount of air drawn in by the CVS constant
flow pump. Since CVS flow rate is constant throughout the test the air drawn
in modulates inverse to vehicle exhaust flow. Total CVS system flow is recorded
for each test period. Diluted exhaust samples are continuously collected from the
well-mixed diluted exhaust CVS stream for each of the following portions of
the entire test: The first cold start bag contains the continuous diluted gas
sample from the first 505 seconds of the test. The second transient portion bag
is the diluted sample from 506 to 1372 seconds. The engine ignition is turned
off for a 10-min hot soak. The third hot start bag is the diluted gas sample
after the 10-min hot soak from the point of engine ignition of the hot restart
for the last 505 seconds of the test. Each of the three bags are measured for con-
centration of hydrocarbons (HC), carbon monoxide (CO) and oxides of nitrogen
(NOx) using, respectively, a flame-ionization detector (FID) HC analyzer, an in-
frared CO analyzer, and a chemiluminescent NOx analyzer. A CO2 nondispersive
infrared (ndir) is used as input to calculate dilution. The concentration measure-
ments are converted into mass emissions per unit of distance utilizing the
known inputs of distance traveled and total measured CVS system flow. FTP
weighting factors are assigned for each phase to give a composite mass per
unit of distance traveled values for each pollutant for the total test—for example:
0.25 g/km HC; 2.0 g/km CO and 0.25 g/km NOx.
It is interesting to analyze the accumulation of emissions from a vehicle
under test using the US FTP. Fig. 2 shows the typical cumulative grams (g) of
HC emissions with respect to time from a Tier 2 vehicle undergoing the FTP.
Fig. 2 shows the catalyst bed temperature for a typical vehicle meeting the
U.S. Tier 2 emission standards (5). HC accumulation rate is high during the
first 50 seconds from engine start before the catalyst has started to function.
Very little accumulation of HC occurs after the engine and catalyst is up to
operating temperature as can be seen in Fig. 2. NOx emissions occur during cold
and hot starts and coincident with high speed modes. It can be clearly understood
Tier 2 engines will have to be designed to heat up the catalytic unit in about 10
to 15 seconds. In fact, Tier 2 compliant engine design factors being examined to

Fig. 2. Automotive emissions importance of cold start Tailpipe HC emissions (FTP bag 1)
(19). Courtesy Ford Motor Company.
498 EMISSION CONTROL, AUTOMOTIVE Vol. 2

produce hotter exhaust gas are: (1) reduced exhaust system thermal mass;
(2) lean air/fuel cold start; (3) retarded ignition timing; and (4) in some cases vari-
able valve timing. The catalytic converter is located close to the exhaust manifold
(close coupled) to heat up faster. New catalytic units have been developed with:
(5) increased geometric surface area per unit of volume; (6) reduced thermal
mass; and (7) improved thermal/physical properties of the catalyst layer. Clearer
fuels with very low sulfur content are also needed.
These factors are explained in Section 5.

3. Exhaust Gas Composition

The exhaust composition from gasoline/air combustion is dependent on many


factors. Total combustion in the internal combustion engine is not possible
even when excess oxygen is present (6). The quality and formation of the air/
fuel mixture as well as the design of the combustion chamber influence the com-
bustion process, as do engine power and ignition timing. However in modern
engines, for emission control, the main factor affecting exhaust gas composition
is the air/fuel mixture or ratio as is shown in Fig. 3 (7) for a standard gasoline

0.8 16 3200

0.7 14 CO2 2800

0.6 12 2400

NO
0.5 10 2000
Emissions, mol %

Emissions, ppm
kg/kWh
BSFC

0.4 8 1600

BSFC
0.3 6 1200
CO O2

0.2 4 800

0.1 2 HC 400
H2

0 0 0
10 12 14 16 18 20
A/F mass ratio

1.4 1.3 1.2 1.1 1.0 0.9 0.8


Fuel equivalence ratio

Fig. 3. Effect of mixture strength on exhaust gas composition (dry basis) and brake
specific fuel consumption (BSFC) for an unsupercharged automotive-type engine using
indolene fuel, H/C ¼ 1.86, where the ignition is tuned to achieve maximum best torque
(MBT), the brake mean effective pressure (BNEP) is 386 kPa at 1200 rpm (7).
Vol. 2 EMISSION CONTROL, AUTOMOTIVE 499

20

H 2O
18
4.0

16
3.0
1.0

14
CO2
Combustion product, mol %

2.0
12 2.0

10 3.0

4.0
8 1.0

1.0
6
2.0
CO
4.0 3.0
3.0 4.0 1.0
4
2.0 4.0
O2
1.0
2 H2

0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3


Rich Fuel equivalence ratio Lean

Fig. 4. Theoretical mole percent of the principal combustion products of hydrocarbon


fuels for fuel hydrogen:carbon ratios from 1, eg, C6H6, to 4, eg, CH4, wet basis, where
(—) represents H2O; (- - -) CO2; (–  – ) CO; (–   –  ) O2; and (        ) H2 (8).

fuel where the hydrogen to carbon ratio (H/C) is 1.86. The values for a stoichio-
metric air/fuel mixture are given at l ¼ 1.0. The exhaust gas composition will
change substantially as the H/C ratio changes (8), as shown in Fig. 4 (8–10).
Sources of exhaust unburned hydrocarbons are crevices in the combustion
chamber, such as gaps between the piston and cylinder wall, where the com-
bustion flame cannot burn. Unburned hydrocarbon composition is dictated by
fuel composition (11). Carbon monoxide results from areas of insufficient oxygen.
Oxides of nitrogen are produced in the high temperature flame zones during
combustion by reaction of nitrogen molecules and oxygen atoms thermally pro-
duced from oxygen and oxygen-containing species, according to the Zeldovich
mechanism (6,12).
Hydrocarbons and carbon monoxide emission can be minimized by lean air/
fuel mixtures (Fig. 3), but lean air/fuel mixtures maximize NOx emissions. Very
lean mixtures (>20 air/fuel) result in reduced CO and NOx, but encounter
increased HC due to unstable combustion at the lean limit. Improvements in
500 EMISSION CONTROL, AUTOMOTIVE Vol. 2

lean-burn engines have extended the lean limit. Rich mixtures, containing excess
fuel and insufficient air, produce high HC, CO and H2 emissions in the exhaust.
Very rich mixtures have been used in some engine applications at high engine
speeds and loads where the cooling effect of gasoline as it is vaporized in the
cylinder is used to limit high cylinder temperature. A recent supplementary
US FTP procedure called US FTP 06 Fig. 1b is designed to limit such excessive
emissions during high speed driving modes. Rich engine calibrations result in
toxic CO exhaust concentrations of 4 to 5% or more.
The best engine power is achieved slightly rich and the best fuel economy is
achieved slightly lean whereas the stoichiometric air/fuel is the optimal point for
best power, fuel economy and emissions.
Over 150 hydrogen and carbon species are present in the exhaust mix of a
gasoline fueled engine (13–17) including methane, various paraffins, olefins,
aldehydes (qv), aromatics, and polycyclic hydrocarbons as well as unburned gaso-
line. Competitive catalytic oxidative reactions favor oxidation of the latter four
classes before paraffins and methane. Analyses of older catalysts after long in-
use on vehicles reveal that loss of catalytic performance is higher for the methane
and paraffin hydrocarbons than for more reactive species.
Sulfur dioxide is present from the combustion of sulfur contained in the
gasoline. The average U.S. gasoline sulfur content had been 350-ppm (20-ppm
SO2 in exhaust) ranging upwards to 1000-ppm in some sections of the U.S.
From 2004 to 2006 the gasoline fuel sulfur specification will gradually change
to an average of 30-ppm with 80-ppm maximum cap.
Exhaust gas also contains small amounts of hydrogen cyanide and ammo-
nia depending on the air/fuel ratio.

4. Emission Control System

Fig. 5a is the schematic of the major components of the TWC Closed-loop Fuel
Metering System. The catalytic converter is located at the engine exhaust mani-
fold (in a close coupled position for Tier 2 compliant engines). An O2 sensor is
located ahead of the catalyst unit inlet and can be positioned in the converter
inlet shell or in the exhaust manifold. A second O2 sensor is located at the cata-
lytic converter outlet. The inlet O2 sensor generates the operating control signal
to the electronic control unit (ECU). The outlet O2 sensor provides a means for
trimming the control point to correct for system aging changes over the life of the
engine. Both the inlet and outlet O2 sensors also perform a required emission con-
trol quality on-board diagnostic (OBD) function which is explained in Section 8.
The operating O2 sensor is exposed to raw exhaust emitted from the engine and
generates an electrical signal indicative of the concentration of oxygen in the
exhaust. The signal is sent to the ECU. The ECU also obtains an inlet airflow
measurement signal from an engine inlet combustion air measurement meter.
Inlet air flow varies in response to the throttle position and engine load (manifold
vacuum). In addition, the ECU (1) obtains information from the throttle position
switch and engine temperature sensor, (2) has a stored library of engine speed
and load maps, (3) calculates the stoichiometric amount of fuel to be injected
into the measured amount of inlet air, and (4) activates the fuel injector opening
Vol. 2 EMISSION CONTROL, AUTOMOTIVE 501

Fig. 5. (a) Basic components of the TWC closed-loop fuel metering system. (b) The oxy-
gen sensor controls the air/fuel mixture to an average mixture. (Courtesy of Robert Bosch).

time to deliver the proper amount of fuel to a location just ahead of the inlet
valve. After combustion in the cylinder, the exhaust gas is measured by the O2
sensor (3 millisecond response time) producing a millivolt signal indicating
rich (insufficient oxygen) or lean (excess oxygen) combustion products. The ECU
responds to deviations from the stoichiometric control point by correcting the
quantity of fuel injected in the next fueling sequence. Thus the air/fuel ratio is
constantly being adjusted slightly rich and slightly lean of the stoichiometric
mixture (Fig. 5b). The three-way conversion (TWC) catalyst receives exhaust
gas that reflects this constant change back and forth in intake air/fuel mixture
that is stoichiometric centered, but because of the nature of the control system
502 EMISSION CONTROL, AUTOMOTIVE Vol. 2

deviates widely in periodic fashion. The TWC is designed to operate under these
conditions to convert NOx by reduction and HC and CO by oxidation by over 95%.

5. Three-Way Catalytic Converter: Physical

5.1. Catalytic Converter. The modern catalytic converter consists of


three main components: the catalytic unit, a metal shell, and a resilient mount-
ing media. The converter shell assembly surrounds and encloses the fragile cera-
mic or delicate metallic catalytic unit so that exhaust gas can easily flow into the
inlet through the catalytic channels and out of the outlet. The steel shell serves
to protect and insulate the catalytic unit from the elements and harsh road phy-
sical environments. The converter shell material is usually iron/chrome Series
409 muffler-grade stainless steel that is resistant to oxidation and external
corrosion. For the ceramic catalytic unit version, a compliant resilient layer is
positioned between the outside of the catalytic unit and the inside of the conver-
ter shell. The compliant layer grips the catalytic unit with sufficient force to pre-
vent movement of the catalytic unit within the canister under all conditions of
vibration and road shock forces, and compensates for thermal expansion differ-
ences of the catalytic unit and shell materials. Several types of compliant layers
have been used and all have spring-like properties under compression that provide
the necessary gripping force to prevent movement. The compliant layer mount-
ing system has proven durable for the life of the vehicle. Converter design, phy-
sical properties, and flow and pressure drop characteristics are amply described
in the literature (19–23). Tier 2 close coupled converters are designed to be more
thermal resistant due to higher temperature exposure in the manifold location.
A pelleted catalytic converter was used by General Motors and a few other
companies until the mid-1980s. It achieved a high level of development (18).
However, high thermal mass limited rapid heat up needed to achieve the latest
stringent emission standards and durability requirements. The monolithic cata-
lytic converter has become universally used by automobile manufacturers
because it is a smaller, lighter unit and can be mounted in any orientation
with low pressure drop. Originally it was thought that the catalyst would have
to be replaced sometime during engine life. The replacement of pellets was
considered more easily managed. Experience has shown that the monolithic
honeycomb catalytic converter remains active and fully functional for the life
of the vehicle.
5.2. Catalytic Unit. Automotive catalysts for gasoline engines are
known as three-way conversion (TWC) catalysts because it destroys all three
regulated pollutants: HC, CO and NOx. TWC catalysts were introduced for the
1977 model year Volvo sold in California. By the mid-1980s they predominated
throughout the U.S. and Canada. As clean vehicle technology spread throughout
the world TWC catalysts became the dominant system for gasoline spark-ignited
engine automobiles and trucks. Oxidation catalysts are part of emission control
systems for small engines used for lawnmowers, and hand-held 2-stroke engines
emissions control for chainsaws and lawn and garden equipment. The active
component of a catalytic unit is a thin activated catalyst layer that is applied uni-
formly through all flow channels of the monolithic substrate. The catalytic unit is
Vol. 2 EMISSION CONTROL, AUTOMOTIVE 503

designed to provide sufficient geometric surface so that all exhaust gas pollu-
tants contact hot catalyst surfaces as they pass through its channels (20–23).
In order to rapidly heat to operating temperature upon engine start the catalytic
unit has low thermal mass and good heat-exchange and mass-transfer properties
to extract heat and efficiently exchange pollutants to the catalyst surfaces and
products back to the exhaust gas stream.
Activated Catalyst Layer. All catalytic chemical reactions occur on cata-
lyst sites within the activated catalyst layer. The layer contains all the cataly-
tically active ingredients consisting of precious metals, base metal promoter
additives, stabilized alumina support particles, and oxygen storage compo-
nents. The activated catalyst layer of a TWC is applied uniformly to all channel
surfaces of the high geometric surface area substrate (ceramic or metallic). The
activated catalyst layer consists of finely divided high BET (see explanation
below) surface area particles that are porous and thermally resistant. The com-
monly used process of applying the catalyst layer to the substrate is one of pre-
paring a water slurry of the particles and components and applying the slurry
uniformly in a thin layer to the all surfaces of the catalyst substrate by: (1) dip-
ping the substrate into a slurry bath; (2) pouring the slurry through the chan-
nels of the substrate; and (3) drawing the slurry up through the substrate by
vacuum. The unit is dried and calcined. Small 2 to 5 nanometer precious
metal crystals are uniformly dispersed on and within the porous particles of
the catalyst layer. The thickness of the catalytic layer is usually between 25
and 100 mm. The substrate or monolith honeycomb body typically has a geometric
surface area (the surface of all channels contained in the substrate) of between
35.1 and 43.7 cm2/cm3 so that l.6 L catalytic unit volume would contain 5.6 to
7.0 m2 of geometric surface area. The activated catalyst layer increases the
total surface to 7000 m2/L, ie, about three orders of magnitude. The total surface
area consists of the combined internal and external surface of all the minute
particles contained in the activated catalyst layer. Total surface area is mea-
sured by the Brunauer-Emmet-Teller (BET) technique, which involves N2
adsorption at low temperature (24).
The minute porous particles are typically aluminum oxide [1344-28-1],
Al2O3. These particles can have BET surface area of 100 to 300 m2/g. The ther-
mal and physical properties of alumina crystalline phases vary according to the
starting phase (aluminum hydroxide or hydrate) and thermal treatment (see
ALUMINUM COMPOUNDS, ALUMINUM OXIDE).
Alumina is used because it is relatively inert and provides the high surface
needed to efficiently and uniformly disperse the expensive active catalytic
components in a thin effective catalyst layer throughout the entire catalytic
unit. However, no one alumina phase possess the thermal, physical, and chemi-
cal properties ideal for the perfect activated catalyst layer. A great deal of
research has been carried out in search of modifications that can make one or
more of the alumina crystalline phases more suitable. For instance, components
such as ceria, baria, lanthana, stronthana, or zirconia are added to enhance the
thermal characteristics of the alumina. Most recent thermal resistant improve-
ments, dictated by high temperature exposure of catalysts directly coupled to
the engine (close coupled), include improved thermal properties of ceria/zirconia
combinations (25–27).
504 EMISSION CONTROL, AUTOMOTIVE Vol. 2

A nonstabilized high surface area alumina sinters severely when exposed to


9008C hydrothermal conditions encountered in engine exhaust. Sintering is a
process by which the small internal pores in the particles coalesce and lose
large fractions of the total surface area. This process is to be avoided because
it can occlude some of the precious metal catalyst sites. The network of small
pores and passages for exhaust gas transfer collapses thus restricting free gas
exchange into and out of the activated catalyst layer resulting in thermal deac-
tivation of the catalyst.
The activated catalyst layer must possess two additional properties. It must
adhere tenaciously to the substrate surface under a variety of extreme conditions
including rapid thermal changes, high flow, and moisture condensation, evapora-
tion, or freezing. It must have an open porous structure to permit easy gas pas-
sage into the catalyst layer and back into the main exhaust stream. It must
maintain this porous structure up to exhaust temperature above 9008C. In
fact, recent catalyst layer formulations maintain over 60% of original design
open porous structure and surface area after exposure to hydrothermal condi-
tions of 1050o to 11008C.
Precious Metal Catalyst. Precious metals are deposited (impregnated) on
and within the small particles throughout the TWC-activated layer. Rhodium
plays an important role in the reduction of NOx, and is combined with platinum
and/or palladium for oxidation of HC and CO. Palladium only TWC catalyst have
been developed. Various methods of precious metal impregnation are used
including pre-impregnation of the stabilized alumina powder or post-impregna-
tion after the activated layer has been applied to the substrate. Only a small
amount of these expensive materials is used (28) (see PLATINUM-GROUP METALS).
The metals are dispersed on the high surface area as precious metal salt solu-
tions, and then reduced to small metal crystals (size range 2 to 10 nanometers
in diameter) by various techniques. Catalytic reactions occur on the precious
metal surfaces. A typical Palladium catalyst with 20% dispersion could contain
3.0 1021 catalyst sites within one liter of catalyst (29). The small precious metal
crystals can agglomerate through in-use thermal processes into much larger dia-
meter structure with consequent loss of catalyst site surface area. An important
catalytic site factor is ‘‘strong metal to support interaction’’ (SMSI) plays an
important role in influencing the electronic behavior of the catalyst site (30).
Some exhaust gas reactions, for instance the oxidation of alkanes, require larger
Pt crystals than other reactions, such as the oxidation of CO (31).
The small precious metal crystals can exist as metal crystallites or as metal
oxides, both of which can be catalytic (28). Rhodium oxide has a tendency to react
with alumina to form a solid solution (32). To minimize this reaction, zirconia
is used with the aluminia (33). Publications regarding the TWC function of
precious metal catalysts abound (34–38).
Catalyst Support Unit. Terms used to describe the high geometric area
support material upon which the active catalyst layer is placed are substrate,
monolith, honeycomb, and unitary body. Both ceramic and metallic forms are
used for automobile catalysts and each possesses unique properties. The proper-
ties each has in common are a large amount of geometric surface per unit of
volume, and inert surfaces that do not react with the catalyst layer but allow
secure attachment of the catalyst layer.
Vol. 2 EMISSION CONTROL, AUTOMOTIVE 505

Ceramic Substrate. The ceramic substrate is typically a cordierite


compound [12182-53-5] material made from reacting silicon dioxide, talc, and
kaolin at high temperature (12008C). The cordierite form is a micro-cracked
body with a very low coefficient of thermal expansion that provides excellent
thermal-shock resistance and is thermal-dimensionally stable. The manufac-
turing process is one that mixes, mulls and kneads the three primary materials
with water into a sort of dough. The dough is formed into a round billet of about
24 to 36 mm diameter and 1 to 1.5 meters long and forced by an extruding
machine through a complex die to form a high geometric area structure. The
extruded piece, now in a wet clay-like form, is dried and fired in a kiln at
about 12008C to form the cordierite. The outside or circumferential dimension
is formed by the die which shrinks to final dimension during the drying and cal-
cining operation. Similarly, the length is cut to dimension. Variations of the
process exist.
Some important substrate physical properties are given in Table 2. In the
past the most common cell structure used was 62 cells/cm2, 0.17 mm wall thick-
ness, 27.3 cm2/cm3 geometric surface area, and 0.43 g/cm3 bulk density as opti-
mal to meet Tier 1 emission standards. Cell structure selected for the more
stringent U.S. Tier 2 and California LEV2 standards are of higher geometric sur-
face area and lower bulk density—for instance 93 and 140 cell/cm2 with thinner
walls of 0.11 and 0.06 mm wall thickness to create higher geometric surface area
of 34.4 and 43.7 cm2/cm3 and lower bulk density of 0.34 and 0.24 g/cm3 respec-
tively. The combined factors of higher geometric surface area and lower bulk
density allow for faster heat up of the catalytic unit to operating temperature
during cold start where the important property of thermal mass is related to
bulk density and specific heat. Care must be given to flow pressure drop, physical
strength, and general ruggedness in an appropriate mounting system design
(20–23,39–41).
Metallic Substrate. Metallic substrates are formed from thin metal
sheets of metal alloys containing Fe, Cr, and Al stabilized with Y or Ce that
have excellent oxidation resistance over the complete range of exhaust tempera-
tures (42). Melting temperature is in the same range as ceramic cordierite. A fea-
ture is that when heated in air, an aluminum oxide surface layer is formed. The
small amount of yttria or ceria present stabilizes the aluminum oxide surface to
resist spalling of the alumina that may otherwise occur. The ceria-stabilized
alloy develops a whisker-like surface. The activated catalyst layer has been
found to adhere very well to either surface form.
The metallic substrate is designed to provide a cell density similar to that
of the ceramic counterpart, and can be constructed in versions of 93, 124 and
155 cells/cm2. However, the wall thickness is thinner, ie, 0.05 to 0.15 mm thick
(43). The 0.05 mm thick wall material can be constructed in versions of 186 to
248 cell/cm2. The unique advantage of metal substrate over its ceramic counter-
part is that either the same geometric surface can be made into a smaller volume,
or that the same volume and geometric surface area will provide lower pressure
drop (44). The 186 to 248 cells/cm2 metal substrates have a potential perfor-
mance advantage over ceramic because of large increases in geometric surface
area that are needed for Tier 2 emission standards. Metal substrate based cata-
lytic converters find application on high performance vehicles such as Porsche,
Table 2. Physical Properties of Catalytic Converter Substrates
Cell structure [wall thickness, mm / cell density, cells/cm2  square cell shape]

0.115/62
Items Unit 0.17/62 [hexagonal] 0.11/93 0.09/93 0.064/140 0.05/180
material porosity % 35 35 35 35 27 35
wall thickness [nominal] mm 0.17 0.115 0.11 0.09 0.064 0.05
cell pitch [nominal] mm 1.27 1.36 1.04 1.04 0.85 0.73
open frontal area % 75.0 83.8 80.0 83.4 85.6 86.8

506
hydraulic diameter mm 1.10 1.25 0.93 0.95 0.79 0.68
geometric surface area cm2/cm3 27.3 26.9 34.4 35.1 43.5 50.8
pressure dropa % 100 72 130 119 168 217
bulk density g/cm3 0.43 0.29 0.35 0.29 0.28 0.24
bulk density ratio %b 100 67 81 67 65 55
specific heat cal/g/8K <— 0.2 —>
thermal conductivity cal/g/sec/8K <– 3 103 —>
a
Pressure drop values are calculated with the following conditions: Substrate size ¼ 101.6 D 101.6 L; Exhaust gas temp. ¼ 8508C; Gas flow ¼ 5 Nm3/min.
b
Ratio against 0.17mm wall thickness / 62 cell/cm3 substrate. Courtesy of NGK Ceramics, USA, Inc.
Vol. 2 EMISSION CONTROL, AUTOMOTIVE 507

DaimlerChrysler Viper and others (42). The advantage here is lower exhaust gas
pressure drop can result in increased horse-power (1 or 2 hp at high engine
speed) and performance. Other special niche applications exist.
5.3. Heat and Mass Transfer. The heat and mass exchange processes
of exhaust gas to the catalytic surfaces takes place under hot pulsating exhaust
flow conditions. Turbulent flow conditions (Reynolds numbers above 2000) exist
in the exhaust manifold and inlet cone of the catalytic converter in response to
the exhaust stroke of each cylinder (about 6 to 25 times per second) times the
number of cylinders under fluctuating gas pressure. Upon entering the cell pas-
sages of the catalytic unit laminar flow conditions (Reynolds number below 2000)
are developed in a distance of 0.6 cm (45–48).
Heat and mass transfer are driven by temperature and concentration dif-
ferences, respectively, between catalyst surfaces and the bulk gas. Effective
mass transfer of reactants and products commence at operating temperatures
over 2508C as catalyst reactions take place with the catalytic layer.
The process of catalyst heating and initiation of the catalytic functions
involves three or four distinct regions as shown in Fig. 6. Depending on the loca-
tion of the catalytic unit in the exhaust system, total thermal mass before and
including the catalytic unit, and exhaust flow and temperature, it can take
from 10 to 120 seconds for the catalyst unit to reach ignition temperature of
approximately 250–3008C (Region I) where heat exchange processes dominate.
At temperatures above the ignition point, catalyst function increases directly
with temperature increase as catalyst activity increases sharply (Region II)
involving both heat exchange and reactant/product mass transfer processes.
Exothermic reactions contribute to temperature rise within the catalyst layer.
Some catalysts reach a point where the sharp increase in catalytic activity
with inlet temperature abruptly takes on an undesirable mild positive slope
(Region III). Then a point is reached at which catalytic performance improves
only slightly in response to further increases in temperature (Region IV) where
reactant mass transfer processes dominate.

Fig. 6. Conversion vs temperature.


508 EMISSION CONTROL, AUTOMOTIVE Vol. 2

100 CO

HC
80

Conversion, %
60

40

20

100 200 300 400


Temperature, °C

Fig. 7. Effect of temperature on catalyst performance using synthetic exhaust: 0.45%


CO, 12% CO2, 500 ppm NO,0.15% H2, 3% O2, 200 ppm C3H6, 10% H2O, and remainder
N2. The T50 for CO and HC are 2308C and 2708C, respectively. See text.

In Region I (below the ignition point) the process is one of heat exchange.
Catalytic activity is prevented by lack of sufficient activation energy and because
of chemisorption of other species. Within Region II, the specific activity of the
catalyst is the rate-limiting step; this is the kinetically controlled region. Highly
active catalysts have a lower ignition point and exhibit large increases in
catalytic performance associated with small increases in catalyst temperature.
The behavior of a catalyst in Region II is important to the selection of an
automotive emission catalyst because catalyst light-off is a prime metric in
achieving adequate emission control early in the FTP as shown in Fig. 7.
Region III is not present in most catalysts. It depends on the porous struc-
ture design of the catalyst layer and is often only found in used or spent catalyst.
If present, the rate-limiting step is pore diffusion control. Catalysts clogged with
masking agents such as lube oil ash and/or accumulations of metal oxides from
gasoline organic metallic additives, ie, iron or manganese, would be expected
to show Region III behavior. Region IV is a mass-transfer limited region, ie,
catalytic reactions occur so fast that the rate-limiting factor is the rate of transfer
of reactants and products to and from the surface of the catalyst. Mathematical
models describing the entire process are found in several references (45–48).

6. Three-Way Catalytic Converter: Chemical Reactions


and Surface Chemistry

Automobile exhaust catalysts are perfect examples of materials that accelerate


chemical reactions toward thermodynamic equilibrium but are not consumed
by the reactions. Reactions are completed on the catalyst surface and the pro-
ducts leave. Thus the auto exhaust 3-way catalyst performs its functions over
and over again for the life of the vehicle. The catalyst permits reactions to
occur at considerably lower temperatures. For instance, CO reacts with oxygen
in bulk gas phase above 7008C at a substantial rate. An automobile exhaust
Vol. 2 EMISSION CONTROL, AUTOMOTIVE 509

catalyst enables the CO oxidation reaction to occur at a temperature of about


2508C and at a much faster rate and in a smaller reactor volume. This is also
the case for the oxidation of hydrocarbons and the reduction of NOx.
6.1. Selectivity. The reduction of NOx in an automotive three-way
catalyst involves a catalyst property called selectivity. Selectivity occurs
when several reactions are thermodynamically possible but one reaction pro-
ceeds at a faster rate than another. The catalyst chooses the route with lowest
activation energy. For instance the preferred three-way catalyst reduces NO
to N2 and not to N2O or NH3. In the case of the three-way catalyst, CO, HC
and H2 are potential reductants of NO. On the other hand, O2 is present,
which oxidizes CO, HC and H2. If oxidation reactions are too rapid, no
reductant would be available to reduce NO. Using modern TWC catalyst, how-
ever, NO reduction is substantially completed before reductants are consumed
by O2.
6.2. Catalyst Metals. Two classes of metals have been examined for
potential use as catalytic materials for automobile exhaust control. These consist
of some of the transitional base metal series, for instance cobalt, copper, chro-
mium, nickel, manganese, and vanadium; and the precious metal series consist-
ing of platinum [7440-06-4], Pt; palladium [74409-05-3], Pd; rhodium [7440-16-6]
Rh; iridium, [7439-88-5], Ir and ruthenium [7440-18-8], Ru. Specific catalyst
activities are shown in Table 3.
The precious metals possess much higher specific catalytic activity than do
the base metals. In addition, base metal catalysts sinter upon exposure to high
engine exhaust gas temperatures thereby losing low temperature ignition

Table 3. Specific Reaction Rates


CO þ O2a C2H4 þ O2b
Catalyst Temperature, 8C Rc Temperature, 8C Rc
Pd (wire) 300 500 300 100
Pt (wire) 300 100 300 10
Au (wire) 300 15 300 0.03
Co3O4 200 25 þ 5 300 0.33
CuO 200 11 300 0.6
CuCr2O4 200 5 300 0.8
LaCoO3 200 2.3 400 0.53
BaCoO3 300 5.3 400 0.1
SnO2 300 5.2 400 0.4
MnO2 300 3.4 300 0.04
LaMnO3 300 2 400 0.3
La0.5Sr0.5MnO3 300 1.2 400 0.26
La0.5Pb0.3MnO3 300 0.5 400 <0:05
Fe2O3 300 0.4 400 0.06
FeCr2O4 300 0.33
Cr2O3 300 0.03 300 0.006
NiO 300 1.5
ZrO2 300 0.013 400 0.002
a
1% O2, 1% CO, 0% H2O.
b
1% O2, 0.1% C2H2, 0:1% H2O.
c
Units in terms of CO2 formation mL/(min  m2 ).
510 EMISSION CONTROL, AUTOMOTIVE Vol. 2

and catalyst performance. Base metals deactivate due to reactions with sulfur
compounds at the low temperature end of auto exhaust environments. Base
metal catalysts would need to be considerably larger. Consequently, the larger
volume would take more time to heat up quickly enough to achieve the catalytic
performance demanded by stringent emission standards. Base metal catalytic
components that provide positive synergistic effects on catalyst performance
when combined with precious metal are called promoters. Reducible base
metal oxides are very important to three-way catalyst designs as explained in
a following section.
6.3. Diffusion of Pollutants to Active Catalyst Sites. A representa-
tion of gas transport from bulk gas to a catalyst reaction site is shown in Fig. 8. A
gaseous pollutant such as CO diffuses from the main gas stream into the porous
catalyst matrix. The rate of diffusion depends on molecular weight of the mole-
cule—heavy gaseous hydrocarbons diffuse slower and hydrogen has the fastest
rate of diffusion. There is a net CO flow from the main flow stream to the catalytic
surface because CO is at a higher concentration in the exhaust stream and,
once catalyst reactions take place, of lower concentration within the catalyst
layer. The CO is attracted to a precious metal catalyst site that is temporarily
unoccupied. For the same reason, O2 molecules are attracted to the site. The
O2 molecule is chemisorbed on the precious metal surface where oxygen molecule
bond stretching occurs, leading to dissociation into oxygen atoms and electron
sharing with the clean precious metal catalyst site surface. The reaction of CO
with adsorbed O atoms proceed to form CO2, which in the product state is
released and reenters the exhaust stream as there is less CO2 in the exhaust
stream and more within catalytic layer so there is a net flow of CO2 away from
the catalyst surface.

CO CO2

Macropore

Alumina

Micropores

Fig. 8. Catalyst pore and reaction. The CO diffuses into a precious metal site &; reacts
with O2; and leaves as CO2.
Vol. 2 EMISSION CONTROL, AUTOMOTIVE 511

6.4. Chemical Reactions. The principal chemical reactions of a TWC


catalyst are
Oxidation reactions

H2 þ 1=2 O2 ! H2 O

CO þ 1=2 O2 ! CO2

Cm Hn þ ðm þ n=4Þ O2 ! m CO2 þ n=2 H2 O

TWC reaction at stoichiometric A/F mixtures

CO þ NO ! 1=2 N2 þ CO2

Cm Hn þ 2ðm þ n=4Þ NO ! ðm þ n=4Þ N2 þ n=2 H2 O þ m CO2

H2 þ NO ! 1=2 N2 þ H2 O

CO þ H2 O ! CO2 þ H2 water gas shift


 n
Cm Hn þ m H2 O þ m CO þ m þ H2 steam reforming
2
Other NO-reduction reactions
2 NO þ 5 CO þ 3 H2 O ! 2 NH3 þ 5 CO2

2 NO þ 5 H2 ! 2 NH3 þ 2 H2 O

NO þ hydrocarbons ! N2 þ H2 O þ CO2 þ CO þ NH3


2 NO þ CO ! N2 O þ CO2
at 200 8C ðbelow exhaust gas temperatureÞ
2 NO þ H2 ! N2 O þ H2 O
Fuel sulfur reactions
S þ O2 ! SO2

SO2 þ 1=2 O2 ! Pt SO3

>6008C
 Al2 ðSO4 Þ3
3 SO3 þ Al2 O3 !

Al2 ðSO4 Þ3 þ 12 H2 ! PM catalyst 3 H2 S þ Al2 O3 þ 9 H2 O

SO2 þ 3 H2 ! PM catalyst H2 S þ 2 H2 O

Where PM is precious metal.


In the case of oxidation catalysts used for small engines or in any other
similar applications only the Oxidation reactions and Fuel sulfur reactions apply.
512 EMISSION CONTROL, AUTOMOTIVE Vol. 2

6.5. Surface Chemistry and Reaction Mechanisms. Three-way cat-


alytic oxidation of HC and CO must proceed in balance with the reduction of NOx
by CO, H2 or HC. CO has been determined to be the predominant reductant of
NO in a complex reaction series on a metal catalyst surface site. First two mole-
cules of NO molecules are adsorbed on the catalyst site surface where they dis-
sociate forming N2 and leaving two O atoms remaining on the catalyst site. CO
(nonabsorbed species) is required to remove the O atoms to complete the reaction
cycle otherwise the cycle would be interrupted (49). A TWC catalyst must be able
to partition enough CO present in the exhaust for each of these reactions and
provide a surface that has preference for NO adsorption. For instance, Rhodium
has a slight preference for NO adsorption rather than O2 adsorption; Pt prefers
O2 adsorption. Rh does not catalyze the unwanted NH3 reaction as does Pt, and
Rh is more sinter-resistant than Pt (6). Double layered and multilayered cata-
lysts are an effective means to partition the reactions, ie, an inner layer if Pt,
Pd or Pt/Pd and an outer layer of Rh have been found particularly effective in
this respect. Pd only TWC catalysts have been developed that are very thermally
resistant for close coupled application in the new Tier 2 compliant vehicles. The
catalyst layers can also be sequenced axially along the channel for instance with
a Rh layer deposited on the inlet half followed by the Pt, Pd or Pt/Pd layer on
the outlet half of the channel. Recently single layer structured designs have
been reported effective.
A major performance factor for TWC catalysts is coping with the fluctuating
air/fuel conditions common with O2 sensor controlled fuel metering systems
Fig. 6. In actual practice, the air/fuel mixture will fluctuate rich and lean of
the stoichiometric point by þ  0.3 to 1.0 air/fuel mixture ratio one to three
times per second. Actually very little time is spent at the preferred stoichiometric
air/fuel mixture point Thus there is too much oxygen present for one half of the
time and too little for the other half. Incorporation of cerium oxide or other
reducible oxides components, including Palladium oxide, solves this problem
(53). For example, ceria reactions under these fluctuating air/fuel conditions are:
Under lean conditions with excess O2

Ce2 O3 þ 0:5 O2 ! 2 CeO2

Under rich conditions with insufficient O2

2 CeO2 þ CO ! CO2 þ Ce2 O3

The cerium þ3 species captures O when in excess that would otherwise exit the
tailpipe and cerium þ4 releases O for oxidization reactions when in short supply.
These reactions provide a counterbalance chemical force toward maintaining
stoichiometric conditions within the catalyst layer even though the dynamic of
the fuel metering system provide cycles ‘‘rich – lean – rich – lean’’ in response
to the O2 sensor feedback control signal.
Nevertheless, the fuel metering system dynamic (or lag time) has a pro-
found effect on any three-way catalyst formulation. Fig. 9a shows a single-
point fuel injection system with the injector located in the main tube of the
inlet manifold and Fig. 9b show a multi-point fuel injection system with a fuel
Vol. 2 EMISSION CONTROL, AUTOMOTIVE 513

Fig. 9. (a) Single-point fuel injection (TBI). 1 Fuel, 2 air, 3 throttle valve, 4 intake
manifold, 5 injector, 6 engine. (b) Multipoint fuel injection. 1 Fuel, 2 air, 3 throttle valve,
4 intake manifold, 5 injectors, 6 engine. (Courtesy of Robert Bosch).
514 EMISSION CONTROL, AUTOMOTIVE Vol. 2

Fig. 10. The quality of the closed-loop system is very important to catalyst function.
(a) Closed loop 0.3 A/F at 1.5 Hz (multi-point fuel injection). (b) Closed loop 0.5 A/F
at 1.0 Hz (single-point fuel injection). (c) Closed loop 1.0 A/F at 1.0 Hz (closed-loop
carburetor) (51). For (b) and (c) Simulated Utilizing Frequency Generator.

injector located at each inlet valve. Fig. 10 is the catalytic performance evalua-
tion of the same TWC when evaluated on an engine sweep test equipped with a
variable closed-loop feed back O2 sensor controlled fuel metering system so that
different fuel metering systems can be simulated: (a) Actual multi-point fuel
injection. (b) Simulated single-point fuel injection. (c) Simulated closed loop car-
buretor. It is proven here that advanced fuel metering is needed to extract the
optimal catalytic performance of a three-way catalyst (54).
The CO oxidation reactions steps proceed as factors. An O2 molecule is
chemisorbed on Pt metal surface as well as two CO (absorbed) species, resulting
in CO2 formation that enters the exhaust stream as mentioned above.

7. Factors Related to Long Term Durability

7.1. Catalyst Durability. The actual in-use overall durability of the three-
way catalyst has proven to be very good even under extreme operating condi-
tions. A durability test of an advanced three-way catalytic converter that was
aged on the MVEG-Cycle (European automobile manufacturers aging cycle
that mimics autobahn driving) for 160,000 km. The decline in HC, CO and
NOx emissions control performance is minor over 20,000 to 160,000 km. The
result shows that very robust three-way catalysts have been developed.
Automobile catalysts last for the life of the vehicle and still function well at
the time the vehicle is scrapped. However, there is potential for decline in total
catalytic performance from exposure to very high temperatures, accumulation of
catalyst poisons, or loss of the active layer (52–57).
Vol. 2 EMISSION CONTROL, AUTOMOTIVE 515

7.2. Thermal Processes. Exposure to temperatures above 10508C can


cause the active layer surfaces to sinter, collapse, agglomerate, shrink, and pos-
sibly exfoliate. Collapse of the open porous structure can occlude precious metal
sites and prevent free exhaust gas passage to these sites. Also, the small precious
metal crystals can unite and grow into larger crystals, thereby diminishing the
total precious metal surface and thus the total number of active catalyst sites. If
catalyst temperatures exceed 12008C, the substrate softens and perhaps shrinks.
At about 14658C, the substrate melts and is destroyed.
High catalyst temperatures are caused by two types of factors. The first
is engine operating conditions and the second is due to exothermic chemical
reactions within the catalysts. Ignition timing, air/fuel mixture, and engine
load govern the exhaust gas temperature and thus the catalyst inlet tempe-
rature. High engine speed and load can result in exhaust temperature up to
9008C. When the catalyst consumes exhaust hydrocarbons, carbon monoxide
or hydrogen exhaust components, these exothermic reactions result in
increased catalyst temperatures. For instance the combustion of 1% CO
would result in 878C adiabatic temperature rise of the exhaust gas (58); the
combustion of 1000-ppm C6H6 would yield 1038C temperature rise. If an
engine was calibrated rich and air was added to the exhaust prior to an oxida-
tion catalyst with an air pump an exothermic reaction could result in 300 to
7008C exhaust gas temperature rise. The three-way catalyst closed-loop feed-
back system does not normally have an air pump and avoids such mishaps.
Misfire or partial misfire results in non-combusted fuel/air mixtures which
undergo catalytic combustion in the catalyst and can produce very high
exothermic reactions.
Misfire or partial misfire of the air/fuel charge in the combustion chamber
results in unburned fuel mixed with air passing to the catalyst, where complete
combustion occurs. The temperature rise can be considerable. Two misfiring
cylinders in a four-cylinder engine could cause the temperature in the catalyst
to approach the softening temperature (12008C) of the substrate or melting
temperature (14658C) (59).
There are many reasons for misfire. The principal causes are: (1) any inter-
ruption of ignition energy, for instance, caused by faulty or wet ignition wiring;
and (2) extreme lean or rich noncombustible air/fuel mixtures. Modern engines
have electronic ignition, very durable and moisture proof ignition wiring and
connections, and highly improved fuel metering systems, and fuel shut off during
deceleration, all of which have greatly lessened the chances and occurrence of
misfire. Shorted-out ignition failure due to wet wiring has become rare.
Unleaded fuel does not corrode, clog, or wear spark plugs, as leaded gasoline
quickly did, and extends combustion quality and spark plug life by 4 to 10 times
or more. Thus the ignition, fuel metering systems and combustion quality
are vastly improved over pre-1975, pre-catalyst equipped vehicles. Catalysts
have now been designed to resist thermal degradation by incorporation of base
metal oxide stabilizers (25–27,60,61).
7.3. Poisons and Inhibitors. Catalyst poisons and inhibitors can come
from the fuel, the lubricating oil, from engine wear and corrosion products, and
from air ingestion of various materials. The two types are: poisons of catalyst
active sites; and, masking agents.
516 EMISSION CONTROL, AUTOMOTIVE Vol. 2

Fuels. The most serious catalyst poison resulted from combustion of a tet-
raethyl lead (TEL) [78-00-2] and ethylene dibromide or chloride (EDB,EDC),
cocktail mixture additive to gasoline to improve octane. Octane enhancement
of gasoline is the only positive feature of TEL with many negative consequences.
Exhaust of leaded gasoline produces ultra-fine air borne lead perfect for inhala-
tion deep into the lung directly contributing to mental development impairment
of children (unrecoverable loss of up to 10 IQ points in urban areas) and serious
detriments to classroom learning processes. Serious life-shortening health pro-
blems for adults and children are also documented (62,63). Combustion of the
TEL cocktail produces acid components that saturate lube oil resulting in severe
engine corrosion and wear resulting in shortened engine life. Continuous dete-
rioration of spark plugs resulted in decline of ignition and combustion quality
and shortened spark plug life. The entire exhaust system was corroded from
the inside out resulting in reduced exhaust system life.
Leaded gasoline (about 10% of world gasoline) is still used in many parts of
the world—Africa, former Soviet Union countries and Middle East countries.
Leaded gasoline use in North America is banned. The only remaining U.S. source
of leaded gasoline is aviation grade gasoline and thus mechanics and a sector of
the population near airports are still exposed to this poison. Also, tetraethyl lead
in containers is reported available in auto supply stores for individuals for their
own vehicles. Catalytic converters are permanently poisoned and rendered inef-
fective from small amounts due to contamination of leaded gasoline in unleaded
gasoline—30 mg Pb/gal caused rapid decline in catalyst performance in a short
period. The mechanism of lead poisoning of automobile exhaust catalysts has
been identified by Ford researchers (64).
Other metal-based fuel additives (Fe, Mn) also create engine debris and accu-
mulate within the engine and exhaust system. Catalysts have been found clogged
with inlet face surface debris/deposits. See reference 5 for a comprehensive analy-
sis of the engine, oxygen sensor and catalyst component functions where one set of
engine components was exposed to clear unleaded gasoline and an identical set
exposed to a Mn-additive containing gasoline. The study is complete with reveal-
ing photographs. The components were interchanged between the respective
engines used for clear unleaded and Mn-additive in a sound scientific study to
reveal: (1) engine out emissions effects; (2) negative effects of O2 sensor perfor-
mance; and (3) declines in catalyst performance. The negative effects appear to
be a direct consequence of accumulation of Mn deposits on and within the O2
sensor and catalyst (5). In recognition of the negative effects of fuel-borne metal-
organic additives, the World-Wide Fuel Charter (WWFC), a document produced in
cooperation with all world engine manufacturers and automobile manufacturers,
specify metal content of unleaded gasoline (Fe, Mn, Pb and other) be non-detectible
for all world in-use vehicle markets (65).
7.4. Fuel Sulfur. Fuel sulfur is responsible for inhibition of TWC
catalyst performance. Fuel sulfur combustion produces sulfur oxides–(SO2/
SO3). Sulfur compounds react with catalyst layer materials and catalyst sites
in a complex manner. Fuel sulfur has been found to contribute to a decline
in HC, CO, and NOx catalyst performance directly associated with the level of
fuel sulfur in two separate studies (66). Sulfur is stored on catalyst sites and
within the catalyst layer in various forms and is released under certain
Vol. 2 EMISSION CONTROL, AUTOMOTIVE 517

conditions. One mechanism is the release of H2S yielding a rotten egg odor to the
exhaust. A small amount of nickel oxide incorporated into the TWC captures H2S
and releases it inoffensively later as SO2 under higher temperature operating
conditions (50,67). The U.S. EPA Tier 2 specification for gasoline sulfur content
is 30-ppm average with a maximum cap of 80-ppm to start phasing-in from 2004
to 2006. When the lower sulfur gasoline phase-in is complete it is expected that
all current on- and off-road in-use vehicles will improve emissions control perfor-
mance by up to 30%. The World Wide Fuel Charter (65) recommends similar fuel
sulfur specifications. It is noted that TWC catalysts avoided a major fuel sulfur
problem with respect to the in the U.S. light-duty emission control program. Oxi-
dation catalysts, originally used from 1975 through the early 1980s, oxidized SO2
to SO3 resulting in H2SO4 emissions in the order of 0.2 g/km—100 times higher
than TWC catalysts. Had oxidation catalysts continued to be needed for emis-
sions control fuel sulfur content would have had to be addressed in the early
1980s.
7.5. Lubricating Oil. Masking agents deactivate catalysts by plugging
the open porous surface of the catalytic layer. Oil ash compounds resulting from
lubricating oil combustion accumulate on catalyst layer surfaces, or lubricating oil
emitted from the exhaust valve stems deposit on the surfaces of the catalyst where
the organic portion is oxidized leaving inorganic ash compounds of zinc, phosphor-
ous, calcium and barium (see LUBRICATION AND LUBRICANTS) on the catalyst surface.
Significant quantities can accumulate depending on lube oil consumption. The ash
compounds tend to partially or fully block some of the surface pores of the catalyst
layer called pore mouth plugging. It only takes a small amount of material on the
surface of the catalyst to restrict the free passage of exhaust gases into and out of
the porous catalyst layer. A zinc pyrophosphate glaze had been found to form on
the catalyst surface under certain temperature conditions (68,69) masking large
areas of catalyst surface and preventing passage of gases into and out of the cat-
alyst porous structure. Materials such as calcium inhibit the formation of this
glaze. Because modern engines using clean unleaded fuel have a much longer
life, masking mechanisms are becoming a more important in catalyst performance
decline for high mileage engines. Metal contents of lubricating oils are currently
being studied and it is expected they will be reduced to match engine needs fueled
with Tier 2 low sulfur unleaded gasoline (70).
7.6. Catalyst Layer Loss. Partial loss of the catalyst layer via attrition,
erosion, or loss of adhesion and exfoliation could result in some loss of catalyst
performance. The monolithic catalyst is designed to resist all of these mechan-
isms. Nevertheless, catalyst materials have been found in roadside dust extend-
ing a few meters from the roadway indicating some loss is occurring. Analyses of
the leading inlet face edge of catalytic units have revealed evidence of small loss
via a gas erosion process at the leading edge. If this becomes an issue it can be
addressed by a hardening procedure.

8. Oxygen Sensor and the Closed Loop Fuel Metering System

The first commercial application of the TWC catalyst and closed loop fuel meter-
ing system using an oxygen sensor was on the 1977 Volvo (four models) sold in
518 EMISSION CONTROL, AUTOMOTIVE Vol. 2

the California market (71–74). The catalyst was invented by Engelhard (50).
Volvo, Bosch, and Engelhard cooperated to bring the system to production over
a 2.5 year period. Several other car companies adopted the system the following
year, and by the 1990s almost all cars sold in the United States and Canada
utilized it. The fully developed system is described in the literature (see SENSORS).
It is now used on almost all light-duty, medium-duty and heavy-duty SI engine
vehicles produced in the world.
A brief description of the system follows. The function of the oxygen sensor,
electronic control unit (ECU), and fuel metering system is to deliver a stoichio-
metric air/fuel mixture to the engine cylinder combustion process, ie, there
should be neither excess air nor excess fuel. The main purpose is to provide opti-
mal exhaust conditions for the TWC catalyst to reduce NOx while at the same
time to oxidize HC and CO within the same catalyst bed forming nonpollutants
N2, H2O, and CO2. In actual operation, the O2 sensor continuously senses
whether there is an excess or lack of O2 present in the exhaust products. The
O2 sensor signal is continuously monitored by the ECU. The ECU also receives
a continuous air flow measurement signal from a mass-flow inlet airflow
measurement device (usually a hot wire anemometer) (75,76); calculates the
proper amount of fuel for the measured amount of air, and controls each fuel
injector opening time to deliver the calculated amount of fuel. The ECU uses
the O2 sensor signal as a correction signal to the computed amount of fuel to
be injected into a measured flow of input air. This control approach is known
as a feed-back system and by its nature is always correcting after the cylinder
combustion event has taken place. The result is a correction cycle above and
below the stoichiometric air/fuel mixture centering at stoichiometric. The TWC
is designed to operate effectively in such an environment. It is found that mini-
mizing the deviation from the stoichiometric control point and minimizing the
frequency with which it occurs is very beneficial to the performance of any
TWC catalyst (51).
8.1. Oxygen Sensor. The O2 sensor is also referred to as lambda sonde
or lambda sensor (lambda is the Greek letter used to denote the air/fuel ratio),
and as an exhaust gas oxygen sensor (EGO). The O2 sensor is a concentration
cell that responds to a changes in exhaust oxygen content in 3 milliseconds yield-
ing a millivolt output signal, ie, 50 mV at i ¼ 1.05 (lean) to 900 mV at i ¼ 0.99
(rich) as shown in Fig. 11. The voltage changes sharply in the immediate vicinity
of the stoichiometric ratio (i ¼ 1.0). The sensor consists of (qv) ceramic porous
solid electrolyte that is ionically conductive at operating temperatures >3508C.
The outside of the ceramic is coated with platinum electrodes: one electrode
exposed to the exhaust gas, the second to ambient air. Between these electrodes
is a zirconia solid electrolyte. The voltage generated depends on the oxygen con-
centration difference between each electrode. The exhaust porous platinum elec-
trode is a catalyst at the exhaust gas surface and equilibrates the mixture,
consuming, by catalytic action, unreacted oxygen and CO, H2,and HC, yielding
a net amount of oxygen present. Near the stoichiometric point, ie, i ¼ 1.0, the
difference between atmospheric O2 and exhaust O2, and thus the voltage gener-
ated, approach a maximum. Thus a control point sensor signal of about 450 to
500 mv is used to detect whether the exhaust is above or below the stoichiometric
control point (75–77). In order for the sensor to function properly, it needs to be
Vol. 2 EMISSION CONTROL, AUTOMOTIVE 519

800

600
Voltage, Vm

400

200

0
0.6 0.8 1.0 1.2 1.4 1.6
A/F ratio, λ

Fig. 11. The A/F ratio ad the lambda sensor’s voltage signal. Courtesy of Robert Bosch.

heated to above 3508C. To accommodate the need for quick O2 sensor catalyst
function, a heated version of the O2 sensor is used.
8.2. Computer Controller. The computer controller also known as the
electronic control unit (ECU), is used for various engine management and
other vehicle functions. It is rightly claimed that the TWC closed-loop system
brought the ECU into dominant use. The ECU has a stored library of engine
speed and load maps. The input signals of mass inlet air measurement,
engine temperature, throttle position and O2 sensor in conjunction with stored
data are employed to assist the engine in cold starting and to calculate the
proper amount of fuel to be injected by each fuel injector for its current oper-
ating condition. The ECU control system uses the O2 sensor signal as a switch-
ing device using the logic ‘‘rich! Go lean!’’ or ‘‘lean! Go rich!’’ Advanced closed
loop systems use the input of a second O2 sensor located downstream of the
catalyst for minor correction to the system due to long term system drift. An
advanced Honda system using an observer O2 sensor to anticipate rate of
change (rate adjustment) has had a dramatic effect in maximizing the perfor-
mance of the TWC. Utilizing these improvements was a key to a Honda emis-
sion control system able to meet the California super ultra-low emission
vehicle standard (SULEV) and partial zero emission vehicle standard
(PZEV) with a single bed TWC catalyst.
The oxygen sensor closed loop system automatically compensates for
changes in fuel composition or air density. For instance, the system corrects for
stoichiometry differences in winter and summer grades of gasoline or those
fuels containing various amounts of oxygenated hydrocarbons such as MTBE
or ethyl alcohol. Also the stoichiometric air/fuel mixture is maintained even
when the vehicle climbs from sea level to high altitudes where the air density
is lower.
520 EMISSION CONTROL, AUTOMOTIVE Vol. 2

The oxygen sensor signal indication for the stoichiometric point can be a bit
different for certain gases, for instance H2 has a high rate of diffusion. Natural
gas and methanol have high H2 in the exhaust (see Fig. 4). This has to be taken
into account for flexible fuel vehicles that operate on two different fuels; ie
methanol and gasoline.
8.3. On-Board Diagnostics. An on-board diagnostic system (OBD-II) is
required on all new vehicles since 1994. The object is to assure and monitor con-
tinued functional performance of combustion quality (engine misfire), the perfor-
mance of the catalytic converter, and the evaporative emission control system.
The ECU would record incidents of failure by displaying a notice to the vehicle
driver/owner and for use in annual vehicle inspection purposes.
An OBD-II system, during an assessment event, compares the output signal
of two O2 sensors: (1) the standard operating sensor is located in front of the
TWC catalyst; and (2) a second OBD-II O2 sensor located downstream of the cat-
alytic converter (or behind the first converter when more than one is used). The
assessment event is conducted by the ECU which schedules a series of distinct
air/fuel ratio step changes from rich to lean and back to rich air/fuel at a defined
frequency of 2 Hz. During an assessment event, a fully functional TWC catalyst
will produce a characteristic OBD-II O2 sensor signal indicating proper catalyst
performance. An impaired or failed TWC catalyst would produce a different
characteristic OBD-II O2 sensor signal indicating catalyst performance below a
desired rate. The component of the TWC being measured is the ceria function as
described earlier in Section 6 and in Refs. 49,78. A fully functional TWC catalyst
containing activated ceria would capture O2 when in excess and release O2 when
in short supply. Thus the OBD-II O2 sensor located downstream of the catalyst
would sense a stoichiometric A/F mixture and produce a steady 450 to 500 milli-
volt signal indicating that the catalyst function is normal. However, if the OBD-II
O2sensor produced a millivolt signal of 50 to 700 at 2 Hz that mirrored or
approached that of the operating O2 sensor signal, it would indicate that the
TWC catalyst ceria function was damaged or no longer able to perform the oxy-
gen storage and release reactions and by inference had lost catalyst functional
performance (79–82).

9. Other Emission Control Systems

9.1. Exhaust Gas Recirculation. NOx control within the engine can be
achieved by recycling a fraction of the exhaust gases into the engine air inlet
stream. The resultant mixture upon combustion results in lower peak combus-
tion temperature and lowers the amount of thermal NOx produced within the
combustion flame. The reaction of N2 þ O ! NOx is strongly dependent on the
combustion flame temperature (75,83). The degree of NOx suppression is depen-
dent on the amount of exhaust gas recirculation (EGR). EGR reduces flame speed
of the mixture, and thus also provides a certain antiknock quality to the combus-
tion process. The impact of EGR on engine parameters has been detailed (84).
EGR in modern engines are quite transparent to the driver/operator with
respect to engine power and performance. In the 1970s vehicle engine perfor-
mance had been noticeably degraded due to early EGR systems; especially at
Vol. 2 EMISSION CONTROL, AUTOMOTIVE 521

idle, under load at low speed, and during cold start. This negative public memory
lingers but is only of historic importance.
9.2. Evaporative Emissions. Gasoline fumes from the on-board fuel
tank, from fuel left in the fuel delivery system, and from running losses are regu-
lated by U.S. EPA standards. Gasoline is a wide mixture of hydrocarbon species
from high volatility butane (C-4) to lower volatility C-8 to C-10 hydrocarbon com-
pounds. High volatility fractions are needed for choked engine cold start. Winter
and summer grade fuels vary the volatile fractions accordingly.
Evaporative sources are (1) Diurnal emissions due to daily fluctuations in
ambient temperature where evaporative emissions depend on the combined
vapor pressure of the fuel hydrocarbon components and the fuel tank tempera-
ture; (2) Hot soak emissions due to fuel left in the fuel metering system or the
inlet manifold after the engine is turned off. Engine and exhaust system thermal
energy heats the left over fuel causing evaporation and emission loss; (3) Running
loss emissions which occur during vehicle operating modes such as low speeds on
a hot day when the rate of vapor generation in the entire fuel system exceeds
engine fueling demand; and (4) filling vapor losses during fuel tank filling
where the liquid gasoline displaces the fuel tank vapor. Fuel tank filling
losses are controlled by a fuel filling pump device at the gasoline filling station
(85,86).
The current EPA evaporative emission regulation prohibits more than 0.95
to 1.2 grams per test to escape to the atmosphere depending on the vehicle class
(3). The test is a 3-day diurnal temperature cycle and an immediate hot soak in a
shed enclosure following the dynamometer run.
The vehicle evaporative control system typically consists of a charcoal can-
ister that connects to the intake manifold. At engine shut off a valve permits
hydrocarbon fumes to be absorbed and stored by a carbon canister. When the
engine is operating, the stored hydrocarbons are purged into the inlet manifold
where they enter the combustion process.
9.3. Crankcase Emissions. Exhaust gases enter the crankcase by
what is known as blow-by gases that escape the combustion chamber past the
piston rings or inlet and outlet valve stem seals. The exhaust gases would
enter the atmosphere via the crankcase breather tube. As the engine wears,
blow-by increases. A positive crankcase ventilation (PCV) valve is required by
regulation (with specified test procedure) to assure these blow-by gases are fed
into the inlet manifold so that they pass through the engine and exhaust emis-
sion control system (87,88).

10. Emissions Control for Alternative Fuels

The National Energy Policy Act of 1992 requires that nonpetroleum-based trans-
portation fuels steadily replace petroleum fuels in the United States. Such fuels
include natural gas (see NATURAL GAS), liquid petroleum gas (qv) (LPG), methanol,
ethanol and hydrogen (qv).
10.1. Natural Gas. Natural gas fueled passenger cars and light-duty
trucks are produced by U.S. automobile manufacturers (89). Emission control
systems are developed for natural gas fueled engines. The preferred emission
522 EMISSION CONTROL, AUTOMOTIVE Vol. 2

control TWC catalyst is based on Pd and Rh rather than Pt and Rh. Engine out
emissions are shown in Fig. 4 for 4/1 H/C ratio. Natural gas combustion produces
larger amounts of H2 and H2O, lesser amounts of CO and CO2, and CH4 is the
predominant HC emission species with formaldehyde as minor. Formaldehyde is
carcinogenic and has it own regulated emission standard (Table 1). TWC cata-
lysts are quite effective for formaldehyde emission control. The U.S EPA and
California emission standards exclude CH4 emissions in the definitions of
NMOG (nonmethane organic gas) and NMHC (nonmethane hydrocarbons) emis-
sion standards. CH4 has low reactivity in the atmosphere and low ozone (qv)
formation potential (90). Pd-based TWC catalysts require exhaust gas tempera-
ture over 4508C to oxidize CH4. The symmetrical methane molecule is the most
refractory of the hydrocarbon family and resists catalytic reactions on traditional
TWC catalysts below 4508C. The CH4 oxidation performance of Pd and Pd-Rh
TWC catalysts are quickly impeded by very small amounts of sulfur in natural
gas. Interesting catalyst work in this area can be found in work by R. Hicks (91).
Natural gas fuel metering systems must assure sufficient CO be present to per-
mit efficient reduction of NOx with CO. For this purpose a 2/1 ratio of CO/NOx
appears sufficient.
10.2. Liquid Petroleum Gas (LPG). LPG is a transportation fuel used
for special vehicle and equipment applications. Emission control systems are
developed and typically use Pt-based (Pt-Rh) TWC catalysts which exhibit
lower light off temperature for propane (C3H8), the major HC species in LPG.
10.3. Alcohol Fuels. Emission control systems have been developed for
methanol and ethanol transportation fuels which normally contain 15% gasoline
and are known respectively as M85 and E85 alcohol fuels. Engines have been
designed for these fuels and emission control systems have been modified for effi-
cient emission control (92–94). M85 fuel exhaust formaldehyde emissions are the
order of 0.60 g/m. Pd-Rh TWC catalysts are found very effective in controlling
formaldehyde emission. E85 fuel exhaust emissions contain acetaldehyde and
formaldehyde. Standard TWC catalyst formulations effectively control aldehyde
emissions. Ethanol is added to gasoline as an oxygenated additive to meet fuel
composition requirements for certain non-attainment or non-compliance regions
in the United States. The closed loop TWC catalyst fuel metering system is able
to accommodate the change in air/fuel stoichiometry induced by the addition of
alcohols without loss in its ability to control exhaust emission.

11. Diesel Engine Emission Control

Light-duty, high speed, diesel compression ignition (CI) engine powered vehicles
must meet the same emission standards as gasoline spark ignited (SI) engines
(Table 1). The CI engine has been greatly improved over that last decade with
respect to fuel efficiency and engine performance. Diesel engines are found in
light and medium trucks but in very few passenger cars. Penetration into
the light-duty vehicle market is expected to increase sharply over the next dec-
ade. Light-duty diesel vehicles have reached 50% of sales in France and a sub-
stantial percentage in most European countries because of improvements in
fuel efficiency and dramatic improvements in performance due to excellent
Vol. 2 EMISSION CONTROL, AUTOMOTIVE 523

engine torque at low speeds. Diesel fuel has more energy per liter than gasoline
(35,785 vs 31,845 kJ/L). Fuel efficiency for light-duty diesel vehicles can range
from 13% to 24% better than for similar gasoline vehicles in similar service
(39). Higher gasoline prices and growing concern for global warming could
increase attractiveness of diesel engines for the North American market.
11.1. Emission Standards. Emission standards for light- and medium
engines are shown in Table 1. Diesel engine has to meet all standards without
exception. The most difficult standards for diesel engines are the particulate
matter (PM) and NOx standards whereas HC and CO are easily removed with
diesel oxidation catalysts.
11.2. Nature of Diesel Exhaust. Modern high speed compression igni-
tion engines used for passenger cars and LDV are equipped with turbochargers
and electronically controlled fuel injection utilizing a common rail fuel feed line,
individual electronically actuated injectors and very high pressure fuel injection.
The individual injectors provide full flexibility of fueling timing and fueling rates
within the start of injection and finish of injection event; ie, (1) a small amount of
fuel can be added early in the compression stroke referred to as pilot injection;
(2) the main amount of fuel can be injected in one squirt, in several squirts and
with modulated fuel rates or shapes that are adjustable during the fueling event;
and (3) post ignition injection can be scheduled to activate downstream emis-
sion control regeneration when necessary. Base engine emissions from these
diesel engine have lower NOx and PM mass. However, ultra-fine particles (30
to 100 nanometer diameter in size) have not been reduced with any of these
improvements.
Combustion air is inducted into the engine in response to engine RPM and
fueling rate without a throttle. Excess air is present for every condition of engine
speed and load. Mass emissions of carbon monoxide and hydrocarbons are low
compared to a SI gasoline engine. NOx generation is high even though overall
air/fuel is very lean because most combustion is diffusion flame combustion of
fuel droplets rather than homogeneous combustion.
11.3. Emission Control Technology. The light-duty vehicle emission
control system that will conform to Tier 2 emission light-duty vehicle emission
standards is currently under development. In 2006 the vehicles would have to
meet at lease Bin 8 NOx 0.124 and PM 0.0124 g/km (Table 1). There two major
challenges: 1) control of NOx and 2) control of particle emissions:
NOx Control. Two approaches under consideration are: (1) Use of large
fractions of EGR (see Section 9) to prevent formation of thermal NOx; or (2)
Catalytic approaches such as NOx adsorbers. Catalytic reduction of NOx in
the oxidizing conditions of diesel exhaust is difficult at the 90% effective reduc-
tion rates needed. One effective approach that is under intense development
is one that stores and releases NO2. In this respect the NOx adsorber is much
like a TWC catalyst the latter stores and releases O2 whereas the NOx adsorber
stores and releases NO2. The feat is accomplished by oxidizing NO ! NO2 on a
Pt component of the NOx Adsorber and storing it chemically as Ba(NO3)2 for
instance. Periodically, the engine is calibrated slightly rich of stoichiometric
for a few seconds and NO2 is released and catalytically reduced by the Rh com-
ponent in a TWC catalytic reaction. This technology requires ultra-low sulfur
diesel fuel.
524 EMISSION CONTROL, AUTOMOTIVE Vol. 2

Particle Emission Control. Catalyst-based diesel particulate filter


(CB-DPF) are a fully developed and proven technology for removal of all soluble
organic and elemental carbon particles. They remove over 99% of the dangerous
fine carbon particle fraction that has the most serious health concern. Several
types of CB-DPFs exist. The exhaust gas containing particles are passed through
the filter where the particles are removed and collected. The collected carbonac-
eous particles have to be removed by combustion either in a continuous or peri-
odic fashion. The collected particles can be slowly oxidized by several means. One
method involves the oxidation of NO ! NO2 on a Pt component contained within
the inlet channel surface catalytic layer of the CB-DPF. NO2 react chemically
with carbon pariticle forming NO and CO2. Reducible oxides such as ceria are
incorporated to transfer oxygen atoms to carbon. In another method NO2 is gen-
erated by a separate inlet Pt catalyst and delivered to the non-catalyzed filter
where the carbon particles are oxidized. At a temperature of about 375 to
4008C direct catalytic oxidation of carbon with oxygen takes place. Above
5508C the thermal supported oxidation of collected soot proceeds. This techno-
logy will likely be part of any Tier 2 light-duty diesel engine emission control sys-
tem for diesel engines sold in the 2006 or 2007 model year depending upon
whether ultra-low sulfur diesel fuel becomes generally available. Currently
CB-DPF technology is planned to be incorporated into heavy-duty on-road diesel
engines by the 2007 model year to meet the heavy-duty diesel engine 2007 rule
when low sulfur diesel fuel is required to available.

BIBLIOGRAPHY

‘‘Automotive Exhaust Control’’ in ECT 2nd ed., Vol. 2, pp. 814–839, by W. B. Innes and
K. Tsu, American Cyanamid Co.; ‘‘Exhaust Control, Automotive’’ in ECT 3rd ed., Vol. 9,
pp. 494–510, by C. M. Heinen, Chrysler Corp.; in ECT 4th ed, Vol. 9, pp. 982–1022, by
John J. Mooney, Engelhard Corp.

CITED REFERENCES

1. M. P. Walsh, Plat. Met. Rev. 30(3), 106–115, 1986.


2. C. Csere, Car Driver, 60–64 (Jan. 1988).
3. Code of Federal Regulations, CFR 40, Parts 80, 85 and 86, Control of Air Pollutioln
from New Motor Veicles: Tier 2 Motor Vehicles Emission Standards and Gasoline
Sulfur Requirements, Washington, D.C., FR pp. 6732–6870, February 10, 2000.
4. Code of Federal Regulations, CFR 40 Part 86, 86.1863–07 Appendix V to Part 86—
Standard Road Cycle (SEC) and Appendix VII—Standard Bench Cycle (SBC). FR,
Vol. 67, No 64, April 2, 2004, pp. 17564–17568.
5. R. McCabe, G. Guo and C. P. Hubbard, Effects of MMTR Fuel Additive on Emission
System Components: Comparison of Clear- and MMTR fueled Escort Vehicles
from the Alliance Study, SAWE-2004-01-1084, Society of Automotive Engineers,
Warrendale, Pa., 2004.
6. J. T. Kummer, Prog. Ener. Combus. Sci. 6, 177–199 (1981).
7. B. A. D’Alleva, Procedure and Charts for Estimating Exhaust Gas Quantities and
Composition, General Motors Laboratory Report 372, Warren, Mich., 1960.
Vol. 2 EMISSION CONTROL, AUTOMOTIVE 525

8. N. A. Heinen and D. J. Patterson, Combustion Engine Economy, Emissions, and


Controls, University of Michigan, Ann Arbor, Mich., Chapt. 5, 1992.
9. L. Eltinge, Fuel-Air Ratio and Distribution from Exhaust Gas Composition, SAE
680114, Society of Automotive Engineers, Warrendale, Pa., 1968.
10. J. B. Edwards, Combustion: The Formation and Emissions of Trace Species, Ann
Arbor Science Publishers, Ann Arbor, Mich., 1974, p. 16.
11. Auto/Oil Air Quality Improvement Research Program, SAE SP-920, Society of Auto-
motive Engineers, Warrendale, Pa., 1992, 16 pp.
12. P. Blumberg and J. T. Kummer, Combust. Sci. Technol. 4, 73 (1971).
13. N. Pelz and co-workers, The Composition of Gasoline Engine Hydrocarbon
Emissions—An Evaluation of Catalytic and Fuel Effects, SAE 902074, Society of
Automotive Engineers, Warrendale, Pa., 1990.
14. P. R. Shore, D. T. Humphries, and O. Hadded, Speciated Hydrocarbon Emissions
from Aromatic, Olefinic and Parafinic Model Fuels, SAE 930373, Society of Automo-
tive Engineers, Warrendale, Pa., 1993.
15. G. J. den Otter, R. E. Malpas, and T. D. B. Morgan, Effect of Gasoline Reformulation
on Exhaust Emissions in Current European Vehicles, SAE 930372, Society of Auto-
motive Engineers, Warrendale, Pa., 1993.
16. J. Laurikko and N-O. Nyland, Regulated and Unregulated Emissions from Catalyst
Vehicles at Low Ambient Temperatures, SAE 930946, Society of Automotive Engi-
neers, Warrendale, Pa., 1993. Good reference for low ambient temperature emissions.
17. P. R. Shore and R. S. de Vries, On-Line Hydrocarbon Speciation Using FTIR and
CIMS, SAE 922246, Society of Automotive Engineers, Warrendale, Pa., 1992.
18. D. W. Wendland and W. R. Matthes, Visualization of Automotive Catalytic Converter
Internal Flows, SAE 861554, 1986; D. W. Wendland, P. L. Sorrell, and J. E. Kreucher,
Sources of Monolithic Catalytic Converter Pressure Loss, SAE 912372, Society of
Automotive Engineers, Warrendale, Pa., 1991.
19. M. Shelef and R. McCabe, ‘‘Twenty-five years after Introduction of Automotive Cata-
lysts: What Next?’’ Cat. Today 62, 44 (2000).
20. J. Howitt, Thin Wall Ceramics as Monolithic Catalyst Supports, SAE 910611,
Society of Automotive Engineers, Warrendale, Pa., 1991.
21. L. S. Socha, J. P. Day, and E. M. Barnett, Impact of Catalytic Support Design Para-
meters on FTP Emissions, SAE 892041, Society of Automotive Engineers, Warren-
dale, Pa., 1989.
22. H. Yamamoto and co-workers, Warm-Up Characteristics of Thin Wall Honeycomb
Catalysts, SAE 910611, Society of Automotive Engineers, Warrendale, Pa., 1991.
23. J. P. Day and L. S. Socha, Jr., The Design of Automotive Catalyst Supports for
Improved Pressure Drop and Conversion Efficiency, SAE 910371, Society of Automo-
tive Engineers, Warrendale, Pa., 1991.
24. R. J. Farrauto and M. C. Hodson, ‘‘Catalyst Characterization,’’ Encyclopedia of
Physical Science and Technology, Vol. 2, Academic Press, Inc., New York, 1987.
25. J.-P. Cuif et al, High Temperature Stability of Ceria-Zirconia Supported Pd Model
Catalysts. SAE 980668, Society of Automotive Engineers, Warrendale, Pa., 1998.
26. T. Kanazawa et al, Development of Three-way Catalyst Using Composite Alunina-
Ceria-Zirconia, SAE 2003-01-0811, Society of Automotive Engineers, Warrendale,
Pa., 2003.
27. B. Williamson et al, Palladium and Platinum/Rhodium dual-Catalyst NLEV and
Tier IIa Close Coupled Emission Solutions. SAE-2001-01-0923, Warrendale, Pa.,
2001.
28. J. T. Kummer, ‘‘The Use of Noble Metals in Automobile Exhaust Catalyst,’’ Chicago
Symposium, 1985.
29. Zicheng Hu, calculation, Engelhard Corp.
526 EMISSION CONTROL, AUTOMOTIVE Vol. 2

30. S. Tauster, Strong Metal to Surface Interaction, Accounts of Chemical Research,


Vol. 20, p. 384, 1987.
31. Y. T. Yu Yao, Indust. Eng. Chem. Prod. Res. Dev. 19, 293 (1980).
32. H. C. Yao, H. K. Stephen, and H. S. Gandhi, J. Catal. 61, 547 (1980).
33. H. C. Yao, S. Japar, and M. Shelef, J. Catal. 50, 407 (1977).
34. S. H. Oh, P. J. Mitchel, and R. M. Siewert, J. Catal. 132, 287 (1991).
35. R. F. Hicks, C. Rigano, and B. Pang, Catal. Lett. 6, 271 (1990).
36. T. Yamada, K. Kayano, and M. Funibike, The Effectiveness of Pd for Converting
Hydrocarbons in TWC Catalysts, SAE 930253, Society of Automotive Engineers,
Warrendale, Pa., 1993.
37. J. C. Dettling and W. K. Lui, A Non-Rhodium Three-Way Catalyst for Automotive
Applications, SAE 920094, Society of Automotive Engineers, Warrendale, Pa., 1992.
38. C. N. Montreiul, S. D. Williams, and A. A. Adanczyk, Modeling Current Generation
Catalytic Converters: Laboratory Experiments and Kinetic Parameter Optimization—
Steady State Kinetics, SAE 920096, Society of Automotive Engineers, Warrendale,
Pa., 1992.
39. K. Hughes et al, Impact of UltraThinwall Catalyst Sustrates for Tier2 emission
Standards, SAE 2003-01-0658, Society of Automotive Engineers, Warrendale, Pa.,
2003.
40. K. Hughes and W. Witte, Ultra Thin Wall Stubstrates for Performance in FTP and
US06, SAE 2002-03-0356, Society of Automotive Engineers, Warrendale, Pa., 2002.
41. Y. Aoki et al, Product Design and Development of Ultra Thin Wall Ceramic Catalytic
Substrate, SAE 2002-01-0350, Society of Automotive Engineers, Warrendale, Pa.,
2002.
42. S. Pelters, F. W. Kaiser, and W. Maus, The Development and Application of Metal
Supported Catalyst for Porsche’s 911 Carrera 4, SAE 890488, Society of Automotive
Engineers, Warrendale, Pa., 1989.
43. F. W. Kaiser and S. Pelters, Comparison of Metal Supported Catalysts with Different
Cell Geometries, SAE 910837, Society of Automotive Engineers, Warrendale, Pa.,
1991.
44. P. Oser, Novel AutoCatalyst Concepts and Strategies for the Future with Emphasis
on Metal Supports, SAE 880319, Society of Automotive Engineers, Warrendale, Pa.,
1988.
45. S. H. Oh and D. L. Van Ostrom, A Three Dimensional Model for the Analysis of Tran-
sient Thermal and Conversional Characteristics of Monolithic Catalytic Converters,
SAE 880282, Society of Automotive Engineers, Warrendale, Pa., 1988.
46. J. C. W. Kuo, C. R. Morgan, and H. G. Lasson, Mathematical Modelling of CO and
HC Catalytic Converter Systems, SAE 710289, Society of Automotive Engineers,
Warrendale, Pa., 1971.
47. C. N. Montreuil, S. D. Williams, and A. A. Adanczyk, Modeling Current Generation
Catalytic Converters; Laboratory Experiments and Kinetic Parameter Optimization—
Steady State Kinetics, SAE 920096, Society of Automotive Engineers, Warrendale,
Pa., 1992.
48. R. Becker et al., Prediction of Catalytic Performance During Light-off Phase with
Different Wall Thickness, Cell Densitly, and Cell Shape, SAE 2001-01-0930, Society
of Automotive Engineers, Warrendale, PA, 2001.
49. S. Tauster, Personal analysis and explanation, Engelhard Corp., Iselin, N. J., 1992.
50. U.S. Pat. 4,552,733 (Nov. 12, 1985), C. Thompson, J. Mooney, C. D. Keith, and
W. Mannion (to Engelhard Corp.)
51. C. D. Falk and J. J. Mooney, Three-Way Conversion Catalysts-Effect of Closed Loop
Feed-back Control and Other Parameters on Catalyst Efficiency, SAE 800462,
Society of Automotive Engineers, Warrendale, Pa., 1980.
Vol. 2 EMISSION CONTROL, AUTOMOTIVE 527

52. K. Otto, W. B. Williamson, and H. Gandhi, Ceram. Eng. Sci. Proc. 2(6), (May/June,
1981). Good review of various catalyst deactivation processes.
53. H. S. Gandhi, W. B. Williamson, and J. L. Bomback, Appl. Catal. 3, 79–88 (1982).
54. E. Jobson and co-workers, Deterioration of Three-Way Automotive Catalysts, Part I—
Steady State and Transient Emission of Aged Catalyst, SAE 930937, Society of Auto-
motive Engineers, Warrendale, Pa., 1993.
55. S. Lundgren and E. Jobson, Deterioration of Three-Way Automotive Catalysts,
Part II—Oxygen Storage Capacity at Exhaust Conditions, SAE 930944, Society of
Automotive Engineers, Warrendale, Pa., 1993.
56. L. A. Carol, N. E. Newman, and G. S. Mann, High Temperature Deactivation of
Three-Way Catalyst, SAE 892040, Society of Automotive Engineers, Warrendale,
Pa., 1989.
57. N. Miyoshi and co-workers, Development of Thermal Resistant Three-Way Catalyst,
SAE 891970, Society of Automotive Engineers, Warrendale, Pa., 1989.
58. J. K. Hockmuth and co-workers, Hydrocarbon Traps for Controlling Cold
Start Emissions, SAE 930739, Society of Automotive Engineers, Warrendale,
Pa., 1993.
59. C. D. Tyree, Emission Level in Catalyst Temperature as a Function of Ignition-
Induced Misfire, SAE 920298, Society of Automotive Engineers, Warrendale, Pa.,
1992.
60. J. E. Kubsh, J. S. Rieck, and N. D. Spencer, in A. Crucq, Ed., Catalysis and Automo-
tive Pollution Control II, Elsevier, New York, 1991, p. 125.
61. N. Miyoshi and co-workers, Development of Thermal Resistant Three-Way Catalysts,
SAE 891970, Society of Automotive Engineers, Warrendale, Pa., 1989.
62. W. Rogan and J. Ware, Exposure to Lead in Children—How Low is Low Enough?
New Engl. J. of Med 348:16, 17 Apr 2003.
63. Needleman and co-workers, Deficits in Psychologic and Classroom Performance of
Children with Elevated Dentine Lead Levels, New Engl. J. Med. 300(13), (Mar. 29,
1979).
64. H. S. Gandhi and co-workers, ‘‘Affinity of Lead for Moble Metals on Different
Supports,’’ Surface Interface Analy. 6(4), (1984).
65. World-Wide Fuel Charter, Alliance of Automobile Manufacturers, 1401 H St.,
Washington, DC 20005. www.autoalliance.org, December 2002.
66. AAMA / AIAM Study on te Effects of Fuel Sulfur on Low Emission Vehicle Criteria
Pollutants, American Automobile Manufacturers Association and Association of
International Automobile Manufacturers, Decemberr 1997. CRC Sulfur/LEV
Program, CRC Project No. E-42, Coordinating Research Council Inc., December
27, 1997.
67. U.S. Pat. 5196390 (Mar. 23, 1993), S. Tauster, J. Dettling, and J. Mooney (to
Engelhard Corp).
68. W. B. Williamson and co-workers, Catalyst Deactivation Due to Glaze Formation
from Oil-Derived Phosphorus and Zinc, SAE 841406, Society of Automotive Engi-
neers, Warrendale, Pa., 1984.
69. K. Inoue, T. Kurahashi, and T. Negishi, ‘‘Effect of Phosphorus and Ash Con-
tent of Engine Oil on Deterioration of Monolithic Three-Way Catalysts and
Oxygen Sensors’’ SAE 92054, Society of Automotive Engineers, Warrendale, Pa.,
1992.
70. Tier 2/LEV II Emission Control Technologies For Light-Duty Gasoline Vehicles,
Manufacturers of Emission Controls association, 1660 L St NW, Washington DC
20036. August 2003 www.meca.org
71. G. T. Engh and S. Wallman, Development of the Volvo Lambda-Sond System, SAE
770295, Society of Automotive Engineers, Warrendale, Pa., 1977.
528 EMISSION CONTROL, AUTOMOTIVE Vol. 2

72. J. J. Mooney, C. E. Thompson, and J. C. Dettling, Three-Way Conversion Catalysts—


Part of the New Emission Control System, SAE 770365, Society of Automotive
Engineers, Warrendale, Pa., 1977.
73. K. Oblander, J. Abthoff, and H. D. Schuster. Der Driewegkatalysator-eine Abgasrei-
nigungstechologie fur Kraftfahrzenge mit Ottomotoren (Three-Way Catalyst—An
Exhaust Cleaning Technology for Automobiles with Internal Combustion Engines),
VDI-Berichte 630, VDI-Verlag, Dusseldorf, Germany, 1984.
74. J. Abthoff and co-workers, MTZ Motortechnische Zeitschrift 53, 11 (1992).
75. Automotive Handbook, 5th ed., Robert Bosch GmbH, Stuttgart, Germany, 2000,
pp. 474, 518.
76. C. O. Probst, Bosch Fuel Injection & Engine Management, Robert Bentley,
Cambridge, Mass., 1989.
77. I. Gorille, N. Rittmansberger, and P. Werner, Bosch Electronic Fuel Injection with
Closed Loop Control, SAE 750368, Society of Automotive Engineers, Warrendale,
Pa., 1975.
78. G. B. Fisher and co-workers, The Role of Ceria in Automotive Exhaust Catalysis and
OBD-II Catalyst Monitoring, SAE 931034, Society of Automotive Engineers,
Warrendale, Pa., 1993.
79. Technical Support Document Revisions to the Malfunction and Diagnostic System
Requirements Applicable to 1994 and later New California Passanger Cars, Light-
Duty Trucks, and Medium Duty Vehicles with Feedback Fuel Control Systems
(OBD-II), California Air Resources Board, Sacramento, Sept. 14, 1989.
80. J. W. Koupal, M. A. Sabourin, and W. V. Clemmens, Detection of Catalyst Failure
On-Vehicle Using the Dual Oxygen Sensor Method, SAE 91061, Society of Automo-
tive Engineers, Warrendale, Pa., 1991.
81. W. Clemmens, M. Sabourin, and T. Rao, Detection of Catalyst Failure Using On-Board
Diagnostics, SAE 90062, Society of Automotive Engineers, Warrendale, Pa., 1992.
82. S. H. Oh, Thermal Response of a Monolith Catalyst Converter During Sustained
Misfiring: A Computational Study, SAE 881591, Society of Automotive Engineers,
Warrendale, Pa., 1988.
83. A. A. Quader, Why Intake Charge Dilution Decreases Nitric Oxide Emissions from
Spark Ignited Engines, SAE 710009, Society of Automotive Engineers, Warrendale,
Pa., 1971.
84. J. J. Gumbleton and co-workers, Optimizing Engine Parameters with Exhaust Gas
Recirculation, SAE 740104, Society of Automotive Engineers, Warrendale, Pa., 1974.
85. Ref. 24, Chapt. 25.89, SAE J171.
86. T. Cam, K. Cullen, and S. L. Baldus, Running Loss Temperature Profile, SAE
930078, Society of Automotive Engineers, Warrendale, Pa., 1993; H. M. Haskew,
W. R. Cadman, and T. F. Liberty, Evaporative Emissions Under Real-Time Condi-
tions, SAE 891121, Society of Automotive Engineers, Warrendale, Pa., 1989.
87. C. B. Tracy and W. W. Frank, Fuels, Lubricants, and Positive Crankcase Ventilation
System, SAE PT112, 451, Society of Automotive Engineers, Warrendale, Pa., 1963–
1966.
88. Ref. 4, 1991 SAE Handbook, Vol. 3. Chapt. 25.68, SAE J900, Society of Automotive
Engineers, Warrendale, Pa., 1991.
89. W. M. Burkmyre, W. E. Liss, and M. Church, Natural Gas Converter Performance
and Durability, SAE 930222, Society of Automotive Engineers, Warrendale, Pa.,
1993.
90. Technical staff report on Reactivity Adjustment Factors (RAF), California Air
Resources Board, 1993.
91. R. Hicks and co-workers, Structure Sensitivity of Methane Oxidation over Platinum
and Palladium; J. Catal., 280–306 (1990).
Vol. 2 EMISSION CONTROL, INDUSTRIAL 529

92. J. J. Mooney, J. G. Hansel, and K. R. Burns, Three-Way Conversion Catalysts on


Vehicles Fueled with Ethanol–Gasoline Mexuture, SAE 790428, Society of Automo-
tive Engineers, Warrendale, Pa., 1979.
93. T. Suga and Y. Hamazaki, Development of Honda Flexible Fuel Vehicle, SAE 922276,
Society of Automotive Engineers, Warrendale, Pa., 1992.
94. J. K. Hochmuth and J. J. Mooney, Catalytic Control of Emissions from M-85 Fueled
Vehicles, SAE 930219, Society of Automotive Engineers, Warrendale, Pa., 1993.

GENERAL REFERENCES

R. B. Bird and co-workers, Transport Phenomena, John Wiley & Sons, Inc., New York, 1965.
Code of Federal Regulations, Title 40, Part 86, Washington, D.C., July 1, 1992.
G. C. Bond and G. Webb, eds, ‘‘Catalysis,’’ Royal Soc. Chem., London 6 (1984).
D. D. Eley, H. Pines, and P. B. Wiez, eds., Advances in Catalysis, Vol. 33, Academic Press,
Inc., New York, 1985.
H. Heinman and J. J. Carberry, eds., Catalysis Reviews—Science and Engineering,
Vol. 26, Marcel-Dekker, New York, 1984.
C. N. Satterfield, Heterogeous Catalysis in Practice, 1st ed., McGraw-Hill Book Co., Inc.,
New York, 1980.
C. D. Falk and J. J. Mooney, Three-Way Conversion Catalysts—Effect of Closed Loop Feed-
back Control and Other Parameters on Catalyst Efficiency, SAE 800462, Society of
Automotive Engineers, Warrendale, PA., 1980.
Tier2/LEV Emission Control Technologies For Light-Duty Gasoline Vehicles, Manufac-
turers of Emission Controls Association, 1660 L St., NW, Washington DC 20035
www.meca.org. August 2003.

JOHN J. MOONEY
Environmental and
Energy Technology and
Policy Institute

EMISSION CONTROL,
INDUSTRIAL
1. Introduction

Limits for exhaust emissions from industry, transportation, power gene-


ration, and other sources are increasingly legislated (1,2). One of the prin-
cipal factors driving research and development in the petroleum (qv)
and chemical processing industries in the 1990s is control of industrial
exhaust releases. Much of the growth of environmental control technology
is expected to come from new or improved products that reduce such
air pollutants as carbon monoxide [630-08-0] (qv), CO, volatile organic

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.0524080102051807.a01.pub2
530 EMISSION CONTROL, INDUSTRIAL Vol. 2

compounds (VOCs), nitrogen oxides NOx, or other hazardous air pollutants


(see AIR POLLUTION). The mandates set forth in the 1990 amendments to the
Clean Air Act (CAAA) push pollution control methodology well beyond what,
as of this writing, is in general practice, stimulating research in many areas
associated with exhaust system control. In all, these amendments set specific
limits for 189 air toxics, as well as control limits for VOCs, nitrogen oxides,
and the so-called criteria pollutants. An estimated 40,000 facilities, including
establishments as diverse as bakeries and chemical plants are affected by the
CAAA (3).
There are 10 potential sources of industrial exhaust pollutants which may be
generated in a production facility (4): (1) unreacted raw materials; (2) impurities in
the reactants; (3) undesirable by-products; (4) spent auxiliary materials such as
catalysts, oils, solvents, etc; (5) off-spec product; (6) maintenance, ie, wastes and
materials; (7) exhausts generated during start-up and shutdown; (8) exhausts gen-
erated from process upsets and spills; (9) exhausts generated from product and
waste handling, sampling, storage, and treatment; and (10) fugitive sources.
Exhaust streams generally fall into two general categories, intrinsic and
extrinsic. The intrinsic wastes represent impurities present in the reactants,
by-products, coproducts, and residues as well as spent materials used as part
of the process, ie, sources (1–5). These materials must be removed from the
system if the process is to continue to operate safely. Extrinsic wastes are
generated during operation of the unit, but are more functional in nature.
These are generic to the process industries overall and not necessarily inherent
to a specific process configuration, ie, sources (6–10). Waste generation may
occur as a result of unit upsets, selection of auxiliary equipment, fugitive
leaks, process shutdown, sample collection and handling, solvent selection, or
waste handling practices.

2. Control Strategy Evaluation

There are two broad strategies for reducing volatile organic compound (VOC) emis-
sions from a production facility: (1) altering the design, operation, maintenance, or
manufacturing strategy so as to reduce the quantity or toxicity of air emissions pro-
duced, or (2) installing after-treatment controls to destroy the pollutants in the gen-
erated air emission stream (5). Whether the exhaust stream contains a specific
hazardous air pollutant, a VOC, a nitrogen oxide, or carbon monoxide, the best
way to control the pollutant is to prevent its formation in the first place. Many tech-
nologies are being developed that seek to minimize the generation of undesirable
by-products by modifying specific process materials or operating conditions.
Whereas process economics or product quality may restrict the general applicabil-
ity of these approaches, an increased understanding of the mechanisms and condi-
tions by which a pollutant is created is leading to significant breakthroughs in
burner design and operation (for nitrogen oxide control), equipment design, main-
tenance, and operation for fugitive and vent VOC emission control, and product
and waste storage and handling design and operation (for VOC emission control).
One source category of increasing importance is that associated with fugitive emis-
sions from equipment leaks.
Vol. 2 EMISSION CONTROL, INDUSTRIAL 531

3. Fugitive Emissions from Equipment Leaks

Emissions can occur from process equipment whenever there is contact


between the process chemicals and the atmosphere. Fugitive emissions repre-
sent this category of emissions associated with the unplanned or unmanaged,
continuing or intermittent emissions of chemicals to the atmosphere, typically
due to leakage from general plant equipment such as pumps, valves, pressure
relief valves, flanges, agitators, and compressors (6). A leaking pump might lose
>10 lb of product/day; a leaking compressor nearly 10 times this amount (see
Table 1).
Fugitive emissions from chemical plants in the United States have been
estimated to total >300,000 tons every year; about a one-third of the total organic
emissions from U.S. chemical plants (7). Other studies indicate that as much as
72% of the total VOC emissions from a facility were attributable to equipment
leaks (2). Controlling such emissions can also reduce employee exposure to
hazardous chemicals, and enhance the safety of a unit.
Because these emissions generally occur randomly and may vary in inten-
sity over time, they are difficult to predict, much less manage. Since a chemical
process unit may contain anywhere from 5000 to 100,000 such components, mea-
surements of equipment leak emissions actually represent a ‘‘snapshot’’ of the
leaking process. Predicting whether any one component will leak is difficult,
but the statistical predictability over large numbers of components in any one
facility produces a fairly consistent picture of the distribution of leak rates.
The distribution of loss rates varies predictably, with only a few components
leaking significantly, (often defined as ‘‘leakers’’) many components leaking
notably, and most others leaking a little. At 17 refineries investigated by the
EPA, it was found that significant equipment leaks approached 5% of the equip-
ment inspected (8,9).
Considering that these losses generally represent valuable raw materials
or final products lost, controlling such emissions can represent a substantial
increase in performance to the operating facility.
Rates of product loss from individual components at facilities depend on
factors such as (5,6):

 Equipment design and standard of installation.


 Age of components.
 Access to component by operators.
 Time since last major maintenance turnaround.
 Specific inspection and maintenance procedures.
 External environmental variability (temperature, pressure) and stress.
 Specific process temperatures and pressures.
 Nature of chemicals in process unit.
 Rates and variability of production operation.

The approaches used to reduce equipment leak emissions fall into two
categories: (1) modifying or replacing existing equipment with a more effective
Table 1. Fugitive Emissions—Sources and Control Effectivenessa
LDAR Program control
effectiveness, %

Number Approximate Monthly Quarterly


in an typical Average ‘‘Leaker’’ control monitoring, monitoring,
Equipment petrochemical emission emission Modification efficiency, 10,000 ppmv 10,000 ppmv MACT
type unit rate, lb/day rate, lb/day % leak definition leak definition requirements
pumps 5–20 1.1 12.8 sealless design 100 69 45 75
closed-vent 90
system

532
dual mechanical seal 100
with barrier fluid
valves 150–600 0.32 4.1 sealless design, 100 87 67 92
bellows
compressors 1–3 12 85 closed-vent system 90 93
dual mechanical seal 100
with barrier fluid
open-ended 7–30 0.11 0.63 blind, cap, plug, or 100
lines second valve
flanges and 700–3000 0.10 0.60 weld or replace 100 33
connectors
a
Based upon information provided in Refs. 9 and 10.
Vol. 2 EMISSION CONTROL, INDUSTRIAL 533

design, and (2) implementing a leak detection and repair (LDAR) program. The
focus of such a program will vary depending on component type, the service (gas,
light liquid or heavy liquid) and previous leak history (10,11) (see Table 1).
3.1. Pumps. Chemicals transferred by pumps can leak at the point of
contact between the moving shaft and the stationary casing. Thus, pumps
generally require a seal at the point where the shaft penetrates the housing in
order to isolate the pumped fluid from the environment.
Equipment modifications for pumps include (1) routing leaking vapors to a
closed-vent system, (2) installing a dual mechanical seal containing a barrier
fluid, or (3) replacing the existing pump with a sealless type (canned-motor, mag-
netic drive, and diaphragm pumps). The control efficiency of sealless pumps and
a dual mechanical seal is essentially 100%, assuming both the inner and outer
seal do not fail simultaneously.
3.2. Valves. Most common valve designs contain a valve stem that oper-
ates to restrict or limit fluid flow. Typically, the stem is sealed by a packing gland
or O-ring to prevent leakage of process fluid to the atmosphere. Emissions from
valves occur at the stem or gland area of the valve body.
Emissions from process valves can be eliminated when the valve stem can
be isolated from the process fluid, (ie, using sealless valves). Two types of sealless
valves, diaphragm valves and sealed bellows, have been developed. The control
efficiency of both types is essentially 100%.
3.3. Compressors. Compressors transport gases through a process unit
in much the same way that pumps transport liquids. Compressors are typically
driven with rotating or reciprocating shafts. The sealing mechanisms for com-
pressors are similar to those for pumps.
Emissions from compressors may be reduced by collecting and controlling
the emissions from the seal using a closed-vent system or by improving seal
performance by using a dual mechanical seal system similar to pumps. The
dual mechanical seal system has an emissions control efficiency of 100%, assum-
ing both the inner and outer seal do not fail simultaneously.
3.4. Pressure Relief Devices. Pressure relief devices are safety devices
commonly used in chemical facilities to prevent operating pressures from exceed-
ing the maximum allowable working pressures of the process equipment. Equip-
ment leaks from pressure relief devices occur when material escapes from the
pressure relief device during normal operation, either because the device does
not properly reseat after a release, from general deterioration of the seal, or
when the process pressure is operating too close to the set pressure of the PRV.
There are two primary equipment modifications that can be used for control-
ling equipment leaks from pressure relief devices: (1) a closed-vent system, or
(2) use of a rupture disk in conjunction with the PRV. The control efficiency of
the closed vent system is dependent upon a number of operational factors.
When Rupture disks can be installed upstream of a PRV they can prevent fugitive
emissions through the PRV seat. The control efficiency of a rupture disk/PRV
combination is essentially 100% when operated and maintained properly. Rup-
ture disks cannot be used on processes where solids or polymers might collect
around the disk, reducing its ability to release at the design pressure.
3.5. Flanges and Connectors. Connectors and flanges are the most
common pieces of equipment in a chemical process unit. They are used to join
534 EMISSION CONTROL, INDUSTRIAL Vol. 2

sections of piping and equipment wherever pipes may require isolation or


removal, such as at a valve or pump. Flanges are bolted, gasket-sealed connec-
tors and are normally used for pipes with diameters of 2.0 in. or greater. The pri-
mary causes of flange leakage are poor installation, aging and deterioration
of the sealant, and thermal stress. Flanges can also leak if improper gasket
material is chosen. Threaded fittings (connectors) are normally used to connect
piping and equipment having diameters of 2.0 in. or less. Seals for threaded
fittings are made by coating the male threads with a sealant before joining
it to the female piece. Leakage can also occur as the result of poor assembly or
sealant application, or from thermal stress on the piping and fittings.
In cases where connectors are not required for safety, maintenance, process
modification, or periodic equipment removal, emissions can be eliminated by
removing the connector and welding the pipe or piece together.
3.6. Leak Detection and Repair Program. The LDAR program is a
structured program to monitor the equipment on a regular schedule and repair
the equipment identified as leaking (10). A LDAR program is more effective on
valves and pumps than it is for flanges and connectors, because these sources
have a greater leak potential relative to their fewer numbers.
In a typical LDAR program, a portable screening device is used to identify
(monitor) pieces of equipment that have the potential to emit amounts of
material at levels that warrant reduction of the emissions through simple repair
techniques. Odors, area monitoring or remote sensing has also been used to
identify significant leakers. These programs are best applied to equipment
types that can be repaired on-line, resulting in immediate emissions reduction.
The control efficiency of an LDAR program is dependent on three factors: (1) the
leak definition, (2) the monitoring frequency of the LDAR program, and (3) the
final leak frequency after the LDAR program is implemented. The leak definition
is the screening value measured by a portable screening device above which a
leak is indicated Screening values are measured as concentrations in parts per
million by volume (ppmv).
The monitoring frequency is the schedule at which the piece of equipment
must be monitored. It may vary from as short as a week for components that
have a propensity to leak (pumps) and to the initial period during installation
and testing to several years for low leaking components such as flanges (12).
The leak frequency is the percentage of components found to be leakers. It has
been shown that with implementation of a LDAR program, the leak frequency
will decrease substantially, as defective or worn out components are identified
and replaced.
Once an undesirable material is created, the most widely used approach
to exhaust emission control is the application of add-on control devices (13).
For organic vapors, these devices can be one of two types, combustion or capture.
Applicable combustion devices include thermal incinerators (qv), ie, rotary
kilns, liquid injection combusters, fixed hearths, and fluidized-bed combus-
tors; catalytic oxidization devices; flares; or boilers/process heaters. Primary
applicable capture devices include condensers, adsorbers, and absorbers,
although such techniques as precipitation and membrane filtration are finding
increased application. A comparison of the primary control alternatives is
shown in Table 2.
Table 2. Emission Control Technologies
Reduction
Technology effectiveness Recovery Waste generation Advantages Disadvantages
activated 90–98% chemical recovery spent carbon or good for wide variety carbon replacement,
carbon possible with regenerant of VOCs regeneration costs,
adsorption regeneration potential for bed fires
adsorption 75–90% þ chemical recovery spent solvent or simple operation not efficient at low
in wet possible through regenerant concentration
scrubbers decanting/distillation
vapor 50–80% chemical recovery liquid wastes, needs simple operation, low removals
condensation possible through off-gas treatment effective for high applicability limits to
decanting/treatment VOC concentration some VOCs, high
power costs
thermal 99% heat recovery NOx generation, handles any VOC high operating costs,
oxidation CO2 generation concentration capital costs,

535
temperatures, and
maintenance
catalytic 95–99% heat recovery spent catalyst simple systems, lower fouling of catalysts,
oxidation regeneration T than thermal temperature limits
acids and economical operation
alkalines
biofiltration 65–99% none some contaminated low installation and large area required for
water and inert operating cost. Minimal installation. Low
solids waste and air pollutant removals for heavy
generation. Particularly organics. Some
suited for odor reduction. pretreatment required.
potential for upsets with
variable exhaust stream.
536 EMISSION CONTROL, INDUSTRIAL Vol. 2

The most desirable of the control alternatives is capture of the emitted


materials followed by recycle back into the process. However, the removal effi-
ciencies of the capture techniques generally depend strongly on the physical
and chemical characteristics of the exhaust gas and the pollutants considered.
Combustion devices are the more commonly applied control devices, because
these are capable of a high level of removal efficiencies, ie, destruction for a
variety of chemical compounds under a range of conditions. Although installation
of emission control devices requires capital expenditures, these may generate
useful materials and be net consumers or producers of energy. The selection
of an emission control technology is affected by nine interrelated parameters:
(1) temperature, T, of the inlet stream to be treated; (2) residence time; (3) pro-
cess exhaust flow rate; (4) auxiliary fuel needs; (5) optimum energy use; (6) pri-
mary chemical composition of exhaust stream; (7) regulations governing
destruction requirements; (8) the gas stream’s explosive properties or heat of
combustion; and (9) impurities in the gas stream. A process flow diagram for
the consideration of these parameters is shown in Figure 1. Given the many fac-
tors involved, an economic analysis is often needed to select the best control
option for a given application.
Capture devices are discussed extensively elsewhere (see AIR POLLUTION
CONTROL METHOD). Oxidation devices are either thermal units that use
heat alone or catalytic units in which the exhaust gas is passed over a catalyst
usually at an elevated temperature. These latter speed oxidation and are able
to operate at temperatures well below those of thermal systems. One developing
technology that is a combination of the capture and oxidaton approaches is
known as biofiltration.

4. Biofiltration Systems

Because control efficiencies of 90–98% or above are often required in emission


control applications, the use of well-known add-on control technologies such as
thermal and catalytic oxidizers are often favored over new technologies (14).
However, biofiltration is a new technology that is being increasingly consi-
dered for use in certain chemical process industries. A biofiltration system uses
microbes to remove and consume the pollutants from a contaminated exhaust air
stream. The term Biofiltration can apply to a variety of such systems, which
include bioreactors, biotrickling filters, bioscrubbers, biomembranes and bioad-
sorption systems (15,16). These systems vary in the design of the beds, and the
amount and type of auxiliary equipment.
Almost any substance, with the help of microbes, will decompose given the
proper environment. This is especially true for organic compounds, but certain
microbes also can consume inorganic compounds such as hydrogen sulfide and
nitrogen oxides. The most common usage is in addressing odor nuisance where
it is not necessary to meet the high control efficiencies. Although destruction effi-
ciencies as high as 99% (17) have been reported for hydrogen sulfide, In some
cases it may be enough to maintain destruction efficiencies as low as 65%. The
biofiltration systems, have been widely used in the early 1990s in Europe, which
required controlling nuisance from odorous compounds at concentrations
Vol. 2 EMISSION CONTROL, INDUSTRIAL 537

Characterize Physical properties Chemical properties


exhaust
stream Flow Composition
Temp Impurities
Variability Variability
Peaks Explosivity
Heating Moisture
value

Consider alternative Pollution prevention


treatment / disposal Pollution control
strategies
Source reduction Carbon adsorption
Process changes Solvent absorption
Raw material change Vapor condensation
Reuse / recycle Thermal oxidation
Consider normal and Administrative charges
peak VOC concentration

Pretreatment

VOC > UEL yes Safety concerns: Particulate removal


consider source reduction Impurity removal
(upper explosion limit) or product recovery Moisture adjustment
Concentrate

VOC > 25% LEL yes Potential safety concerns


(lower explosive limit) depending on O 2 conc.
and organics

no

Dilution and / or
additional
monitoring needed

no
Auxiliary heat needed

yes Auxiliary burner


Heat exchanger
Recirculated air

Calculate bed volume Depends on


needed to achieve
designed destruction of Catalyst selected
most difficult compound Normal and peak flows
Compounds
Removal efficiency
needed

Calculate temperature
rise through bed

Are catalyst poisons Poisons


present Consider poisons in
Masking agents
Fouling agents Selection of catalyst
Consider regeneration Inhibitors Pretreatment scheme
alternatives Sintering agents Regeneration approach

Develop
system

Fig. 1. Process flow diagram for the selection of an exhaust control system.
538 EMISSION CONTROL, INDUSTRIAL Vol. 2

<50 ppmv. Unfortunately, not all organics are degraded at the same rate. For
that reason, the destruction efficiency of a bioreactor depends heavily on the
composition of the organic feed stream. Large quantities of heavier organics
can depress the average VOC destruction efficiency even while the oxidation
efficiency of simpler compounds remains very high (18).
The EPA has indicated that where applicable, both the capital cost of a bio-
filter installation and its operating costs can be just a fraction of the cost of a
traditional emission control devices.
An efficient Biofiltration system should only use small amounts of electrical
power and some micronutrients in order to operate, and will generate water and
CO2 as by-products. Material that is indigestible becomes residue. However,
because microbes are sensitive to environmental conditions (temperature,
humidity, pH) as well as to the pollutant concentration and type, a biofiltration
system is only suitable for selected applications (15).
Biofilters, usually use beds (media on which microbes live) made from natu-
rally occurring organic materials (yard cuttings, peat, bark, wood chips or
compost) that are slowly consumed by the microbes. Sometimes oyster shells
(for neutralizing acid build-up) and fertilizer (for macronutrients) are mixed
with bed media (see Fig. 2). These organic beds usually can supply most of the
macronutrients needed to sustain the microbial activity. Under normal operating
conditions, the beds must be replaced every 2–5 years (16), depending on the
choice of bed material. However, in handling excessively moist exhaust streams,
sloughing of the beds may clog the discharge system and require replacement of
the beds more often (19).
A fan is used to collect contaminated air from a building or process, but if
the air is too hot, too cold, too dry, or too dirty, it may be necessary to pretreat the
contaminated air stream to obtain optimum conditions before introducing it into
a biofiltration system.
Variables that effect the operation and efficiency of a biofiltration sys-
tem include: temperature, pH, moisture, pollutant mix, pollutant concen-
tration, macronutrient feeding, residence time, compacted-bed media, and
gas channeling. These are crucial variables for which optimum conditions
must be determined, controlled and maintained. Below are variables and lim-
itations that affect the performance of all biofiltration systems, regardless of
process type.

Fig. 2. Basic biofilter.


Vol. 2 EMISSION CONTROL, INDUSTRIAL 539

4.1. Temperature. Probably the most important variable effecting bio-


filtration operations is temperature. Most microbes can survive and flourish in
a temperature range of 60–1058F (18). However, a short flow of hot air can
totally destroy the system, while cold air can slow down and reduce the efficiency
of the system. Where hot gases may be treated, the emissions would need to be
cooled, often by passing them through a humidifier that can cool the gases and
add moisture to the vapor stream.
4.2. Moisture. Microbes need moisture to survive and moisture creates
the film that absorbs pollutants from an air stream where they can be assimi-
lated by microbes. Low moisture problems can be corrected by passing emissions
through a humidifier. Low moisture, for short periods, while not killing the
microbes, will reduce efficiency (15). As water is removed or produced as a by-
product, water may accumulate in the biofiltration system. Excess water needs
to be removed to reduce pressure loss and minimize load on the bed. Sloughing of
the organic filter media has also been observed (19).
4.3. Acidity. Most biofiltration systems perform best when the bed pH is
near 7, or neutral. Some pollutants form acids when they decompose. Examples
of these compounds are: hydrogen sulfide, organic sulfur compounds, and halo-
gens (chlorine, fluoride, bromine, and iodine). The Biofiltration systems that
treat emissions that contain sulfur or sulfur compounds perform best when the
pH is in the range 1–2 pH (20). But, in most cases, if the operation will lower the
pH, it will eventually destroy the microbes. There are several techniques avail-
able to neutralize beds. One of the simplest techniques is to mix oyster shells
with bed media. The shells will eventually dissolve and have to be replaced
(16). Other solutions may be to introduce a dilute solution of soda ash into the
bed either through the top of the bed or through hoses inside the bed.
4.4. Microbe Population. It takes some time for the microbial popula-
tion to acclimate to the contaminated air stream and achieve optimum removal
efficiency. Similarly, if the system is left idle, the population of microbes will
change and efficiency will be reduced (21). A microbial population treating a
variable exhaust stream will also have reduced efficiency since fewer microbes
favoring any particular organic compound will be developed.
Biofilter Advantages:

 Installation and operation costs are usually low.


 Secondary waste generation (spent solvents, contaminated water, or
carbon), pollutants (NOx, CO etc) and other greenhouse gases are minimal.
 Natural-bed media used in biofilters usually need only to be replaced every
2–5 years.
 High destruction efficiencies for problematic odorous compounds such alde-
hydes, organic acids, nitrous oxide, sulfur dioxide, and hydrogen sulfide.

Biofilter Disadvantages:

 Large land requirement for traditional design.


 No continuous internal liquid flow in which to adjust bed pH or to add
nutrients.
540 EMISSION CONTROL, INDUSTRIAL Vol. 2

 Excessive conditions of the exhaust stream may require pretreatment to


remove water, solids or alter temperature.
 Bed replacement can take 2–6 weeks, depending on bed size.
 Information on recovery of the system after an upset remains limited.
 Low destruction efficiencies for some organic compounds.
 Exhaust organic concentrations in excess of 500 ppmv may overtax the
system (19).

5. Oxidization Devices

5.1. Thermal Oxidation. Thermal oxidation is one of the best known


methods for industrial waste gas disposal. Unlike capture methods, eg, carbon
adsorption, thermal oxidation is an ultimate disposal method destroying the
objectionable combustible compounds in the waste gas rather than collecting
them. There is no solvent or adsorbent of which to dispose or regenerate. On
the other hand, there is no product to recover. A primary advantage of thermal
oxidation is that virtually any gaseous organic stream can be safely and cleanly
incinerated, provided proper engineering design is used (22).
A thermal oxidizer is a chemical reactor in which the reaction is activated
by heat and is characterized by a specific rate of reactant consumption. There are
at least two chemical reactants, an oxidizing agent and a reducing agent. The
rate of reaction is related both to the nature and to the concentration of reac-
tants, and to the conditions of activation, ie, the temperature (activation), turbu-
lence (mixing of reactants), and time of interaction.
Thermal oxidation relies on a homogenous gas-phase reaction condition.
Exhaust emissions from industrial sources usually contain organic compounds
(the reducing agents) well mixed with oxygen (the oxidizing agent). Imparting
the necessary, uniform temperature for reaction within this mixture is of pri-
mary importance in the design of the oxidizer and related equipment. Proper
activation requires establishing the minimum required temperature
(650–8008C) for an adequate time (0.1–0.3 s). General design consideration is
given to minimizing heat input and reactor size under the constraints of time,
turbulence, and temperature.
Thermal oxidization devices are widely used, and generally provide a
high degree of assurance that the process oxidizes the material in the exhaust
gas. The high temperature operation causes other problems, however, espe-
cially compared to alternatives such as catalytic oxidation. The thermal
oxidation devices often incur higher fuel costs because of the higher tem-
peratures necessary, and require exotic high temperature materials, because
the high temperatures entailed can bring about serious mechanical design
problems as a result of operating in the temperature range in which metal
creep takes place (1,23). In addition, equipment durability is reduced by
the extent of thermal cycling. Thermal oxidation systems are susceptible to
thermal stress effects which result in distortion of the ductwork and heat-
exchange surfaces, creating the potential for cracks and leaks. A further
consequence of these high temperatures is that a thermal oxidizer may
Vol. 2 EMISSION CONTROL, INDUSTRIAL 541

produce nitrogen oxides (NOx), and sometimes yield undesirable by-products


such as dioxins from chlorinated materials (5).
Some of the problems associated with thermal oxidizers have been attri-
buted to the necessary coupling of the mixing, the reaction chemistry, and the
heat release in the burning zone of the system. These limitations can reportedly
be avoided by using a packed-bed flameless thermal oxidizer that is under deve-
lopment. This system relies on radiant heat from a large heat sink to raise
the temperature of the exhaust gas to its ignition temperature. The heat sink,
a ceramic matrix, is preheated radiatively using an electric preheating element,
or a natural gas preheater, prior to introducing the exhaust gas. Because the
system temperature is reasonably constant, NOx generation within the flame
is minimized. High (99.99%) VOC reductions at low contact times were reported
(24). However, as with all thermal oxidation systems, this system is most effec-
tive for higher concentration exhaust streams, and requires significant auxiliary
heat to treat low concentration streams.
5.2. Catalytic Oxidization. A principal technology for control of
exhaust gas pollutants is the catalyzed conversion of these substances into
innocuous chemical species, such as water and carbon dioxide. This is typically
a thermally activated process commonly called catalytic oxidation, and is a pro-
ven method for reducing VOC concentrations to the levels mandated by the
CAAA (see CATALYSIS). Catalytic oxidation is also used for treatment of industrial
exhausts containing halogenated compounds.
As an exhaust control technology, catalytic oxidation enjoys some signi-
ficant advantages over thermal oxidation. The former often occurs at tempe-
ratures that are less than half those required for the latter, consequently
saving fuel and maintenance costs. Lower temperatures allow use of exhaust
stream heat exchangers of a low grade stainless steel rather than the expensive
high temperature alloy steels. Furthermore, these lower temperatures tend to
avoid the emissions problems arising from the thermal oxidation processes
(25,26).
Critical factors that need to be considered when selecting an oxidation sys-
tem include (27) (1) waste stream heating value and explosive properties. Low
heating values resulting from low VOC concentration make catalytic systems
more attractive, because low concentrations increase fuel usage in thermal sys-
tems; (2) waste gas components that might affect catalyst performance. Catalyst
formulations have overcome many problems owing to contaminants, and a guard
bed can be used in catalytic systems to protect the catalyst; (3) the type of fuel
available and optimum energy use. Natural gas and No. 2 fuel oil can work well
in catalytic systems, although sulfur in the fuel oil may be a problem in some
applications (28). Other fuels should be evaluated on a case-by-case basis; and
(4) space and weight limitations on the control technology. Catalysts are favored
for small, light systems.
There are situations where thermal oxidation may be preferred over cata-
lytic oxidation: for exhaust streams that contain significant amounts of catalyst
poisons and/or fouling agents, thermal oxidation may be the only technically
feasible control; where extremely high VOC destruction efficiencies of difficult
to control VOC species are required, thermal oxidation may attain higher perfor-
mance; and for relatively rich VOC waste gas streams, ie, having 20–25% lower
542 EMISSION CONTROL, INDUSTRIAL Vol. 2

explosive limit (LEL), the gas stream’s explosive properties and the potential for
catalyst overheating may require the addition of dilution air to the waste gas
stream (27).
Whereas the catalytic converter has been used in automobiles to control air
pollutants only since 1975 (5), catalytic oxidation of industrial exhaust emissions
began in the late 1940s, and is a reasonably mature technology (29). Initially it
was used only in circumstances where an extremely serious odor problem was
associated with an industrial system, or where the concentration of organic sol-
vents in the gases to be discharged to the air was high enough that these could
be burned and the heat utilized in the process (23). By the mid-1950s there
were several dozen catalytic incinerators in California, primarily in Los Angeles
county, the first sizable area within the United States to experience a serious air
pollution problem. Early applications of this technology involved some serious
odor, eye irritation, or visible organic emission problems resulting from halogen
poisoning and catalyst fouling (30).
The chemical industry was the first to utilize catalytic oxidation exten-
sively for emission control, building units capable of treating up to 50 m3/s
(100,000 scfm) of exhaust gas containing VOCs. Catalytic systems accounted
for roughly one-fourth of the $200 million market for VOC control systems in
1992, and over one thousand catalytic oxidization devices were in place by the
end of that year (5).
Catalysts. A catalyst has been defined as a substance that increases
the rate at which a chemical reaction approaches equilibrium without becoming
permanently involved in the reaction (31). Thus a catalyst accelerates the
kinetics of the reaction by lowering the reaction’s activation energy (5), ie, by
introducing a less difficult path for the reactants to follow. For VOC oxidation,
a catalyst decreases the temperature, or time required for oxidation, and hence
also decreases the capital, maintenance, and operating costs of the system (see
CATALYSIS).
A key feature of a catalyst is that the catalytic material is not consumed by
the chemical oxidation reactions, rather it remains unaltered by the reactions
that occur generally on its surface and thus remains available for an infinite
number of successive oxidation reactions.
Many chemical elements exhibit catalytic activity (5) which, within limits,
is inversely related to the strength of chemisorption of the VOCs and oxygen,
provided that adsorption is sufficiently strong to achieve a high surface coverage
(32). If the chemisorption is too strong, the catalyst is quickly deactivated as
the active sites become irreversibly covered. If the chemisorption is too weak,
only a small fraction of the surface is covered and the activity is very low (32)
(Fig. 3).
Catalysts vary both in terms of compositional material and physical struc-
ture (33). The catalyst basically consists of the catalyst itself, which is a finely
divided metal (29,32,34); a high surface area carrier; and a support structure.
Three types of conventional metal catalysts are used for oxidation reactions:
single- or mixed-metal oxides, noble (precious) metals, or a combination of the
two (34).
The precious metal or metal oxide imparts high intrinsic activity, the
carrier provides a stable, high surface area for catalyst dispersion, and the
Vol. 2 EMISSION CONTROL, INDUSTRIAL 543

Adsorption Adsorption
too weak too strong

Catalytic activity

Strength of adsorption

Fig. 3. Catalytic activity as a function of adsorption strength (32). (Courtesy of Oxford


University Press.)

mechanical support gives a high geometric surface area for physical support and
engineering design features (35). Only the correct combination of these compo-
nents provides suitable performance and long catalyst life of a properly designed
catalytic system (36).
Metal Oxides. The metal oxides are defined as oxides of the metals occur-
ring in Groups 3–12 (IIIB–IIB) of the periodic table. These oxides, characterized
by high electron mobility and the positive oxidation state of the metal, are gen-
erally less active as catalysts than are the supported nobel metals, but the oxides
are somewhat more resistant to poisoning. The most active single-metal oxide
catalysts for complete oxidation of a variety of oxidation reactions are usually
found to be the oxides of the first-row transition metals, V, Cr, Mn, Fe, Co, Ni,
and Cu.
Noble Metals. Noble or precious metals, ie, Pt, Pd, Ag, and Au, are fre-
quently alloyed with the closely related metals, Ru, Rh, Os, and Ir. These are
usually supported on a metal oxide such as a-alumina, a-Al2O3, or silica, SiO2.
The most frequently used precious metal components are platinum [7440-06-
4], Pt, palladium [7440-05-3], Pd, and rhodium [7440-16-6], Rh. The precious
metals are more commonly used because of the ability to operate at lower tem-
peratures. As a general rule, platinum is more active for the oxidation of paraf-
finic hydrocarbons; palladium is more active for the oxidation of unsaturated
hydrocarbons and CO (34).
Each precious metal or base metal oxide has unique characteristics, and
the correct metal or combination of metals must be selected for each exhaust
control application. The metal loading of the supported metal oxide catalysts
is typically much greater than for nobel metals, because of the lower inher-
ent activity per exposed atom of catalyst. This higher overall metal loading,
however, can make the system more tolerant of catalyst poisons. Some com-
pounds can quickly poison the limited sites available on the noble metal
catalysts (34).
544 EMISSION CONTROL, INDUSTRIAL Vol. 2

Carrier. The metal catalyst is generally dispersed on a high surface


area carrier, ie, the carrier is given a washcoat of catalyst, such that very
small (2–3-nm diameter) precious metal crystallites are widely dispersed over
the surface area, serving two basic functions. It maximizes the use of the costly
precious metal, and provides a large surface area thereby increasing gas contact
and associated catalytic reactions (33).
Proper selection of the carrier is critical (35). For example, in most cases,
the carrier of a precious metal catalyst is a high surface area alumina having
an effective surface area in the order of 120 m2/g of material. Alumina is often
used because of its unique phase transformation properties. Various phases of
the aluminum hydroxides exist as a function of temperature and starting
phase. For a catalyst to be stable, the correct starting phase of alumina must
be selected for the projected commercial operating temperature. Otherwise the
alumina may undergo a transition during operation resulting in a carrier of
less surface area, and hence less catalytic activity (1).
Efforts to redesign catalyst formulations involve both catalyst and wash-
coat. One thrust of this research is to manipulate the catalytic surface so that
it can handle larger quantities of catalyst poisons and to incorporate more
catalytic sites, redistributed within the washcoat to make them more accessible
to exhaust molecules. Altering the composition of the alumina washcoat by
including various nonprecious metal oxides, such as oxides of barium, cerium,
and lanthanum as stabilizers, is being looked at to promote catalyst activity
before sintering and stabilize precious metal dispersion (33). Reformulation
efforts are aided by use of computer controls and cleaner reactants and continu-
ous monitoring, all of which help make exhaust composition more predictable (1).
The Support Structure. After the catalytic element is placed on the high
surface area carrier, it is deposited on a mechanical support structure which
determines the form of the catalyst. The support structure may have many
forms, such as spheres, pellets, woven mesh, screen, honeycomb, or other cera-
mic matrix structures designed to maximize catalyst surface area (13). Some of
the advantages of these different supports are listed in Table 3.
The pelleted and honeycomb support structures are most widely used; hon-
eycombs are the most commonly employed. Pelleted structures are generally
spherical beads or cylinders having diameters ranging from 0.16–0.64 cm. The

Table 3. Conventional Catalyst Bed Geometriesa


Geometry Advantages Disadvantages
metal ribbons low pressure drop; high less active than ceramic-
surface-to-volume supported catalysts
spherical pellets can be used in both fixed and high pressure drop; attrition
fluidized beds problem
ceramic rods low pressure drop low surface-to-volume ratio
ceramic honeycomb low pressure drop; high may have nonuniform
surface-to-volume ratio catalyst coating
metal honeycomb low pressure drop; high less active than ceramic-
surface-to-volume ratio; supported catalysts
high mechanical strength
a
Ref. 29.
Vol. 2 EMISSION CONTROL, INDUSTRIAL 545

pellets are assembled into a packed bed containing large numbers of these pellets
through which the exhaust passes. The honeycomb supports are monolithic
structures having numerous parallel channels through which the exhaust
passes, the channel sizes ranging from 8–50 cells/cm2 of catalyst frontal
area. Each cell has a width opening ranging from 0.29 to 0.13 cm, respectively.
Some commercial honeycombs are available from 1.6 to 100 cells/cm2 (35). The
shape in the individual honeycomb channel is unlimited, eg, circle, square, and
triangle.
Although more expensive to fabricate than the pelleted catalyst, and us-
ually more difficult to replace or regenerate, the honeycomb catalyst is more
widely used because it affords lower pressure losses from gas flow; it is less likely
to collect particulates (fixed bed) or has no losses of catalyst through attrition,
compared to fluidized bed; and it allows a more versatile catalyst bed design
(33), having a well-defined flow pattern (no channeling) and a reactor that can
be oriented in any direction.
The honeycomb structure is either fabricated of ceramic or stainless steel.
The high surface area carrier and catalytic precious metal crystallites are coated
onto the walls of the channels in the honeycomb. The honeycomb catalyst blocks
generally range form 15 to 30-cm square at depths from 5 to 10 cm. These blocks
are packed into larger modules containing many catalyst blocks. Flow through a
honeycomb catalyst structure is shown in Figure 4 (33,37). A typical 30 cells/cm2
honeycomb structure has about 4600 m2 of geometric wall area per cubic meter of
catalyst volume (33). The actual shape of the individual honeycomb channel is
unlimited, eg, it may be circular, square, triangular, etc. In addition, the channel
density can be varied. Commercial honeycombs are available that range from 1.5
to 100 cells/cm2 (10 to 600 cells(in.2)) (35).
Only by using the carrier can the catalyst be sufficiently active because
the majority of applications require 10–100 m2/g of surface area (35). Surface
areas for a typical monolith support structure and a carrier are given (35).

Active precious metal site


Individual
catalyst cell Exhaust
gas

High surface
area

Wall of honeycomb

Exhaust
flow
(a) (b)

Fig. 4. Schematic of the flow through (a) a honeycomb catalyst structure and (b) a cross-
section of a honeycomb channel.
546 EMISSION CONTROL, INDUSTRIAL Vol. 2

Mechanistic Models. A general theory of the mechanism for the complete


heterogeneous catalytic oxidation of low molecular weight vapors at trace
concentrations in air does not exist. As with many catalytic reactions, however,
certain observations have led to a general hypothesis (32).
The overall process of any catalytic reaction is a combination of mass-
transfer, describing transport of reactants and products to and from the interior
of a solid catalyst, and chemical reaction kinetics, describing chemical reaction
sequences on the catalyst surface. The most cost-effective catalytic oxidation
systems require use of a solid catalyst material having a high specific surface
area, ie, high surface area per net weight of catalyst. The presence of many
small pores necessarily introduces pore transport diffusion resistance as a factor
in the overall, or global, kinetics. The overall process consists of (32,38): (1) trans-
port of reactants from the bulk fluid through the gas film boundary layer to
the surface of the catalytic particle; (2) transport of reactants into the catalyst
particle by diffusion through the catalyst pores; (3) chemisorption of at least
one reactant on the catalyst surface; (4) chemical reaction between chemisorbed
species or between a chemisorbed species and a physisorbed or fluid-phase reac-
tant; (5) desorption of reaction products from the catalyst surface; (6) diffusive
transport of products through the catalyst pores to the surface of the catalyst
particle; and (7) mass transfer of products through the exterior gas film to the
bulk fluid.
In principle, any of these steps or some combination can be rate controlling.
In practice, temperature plays a primary role in determining the rate-controlling
stage. Any comprehensive analysis of actual catalytic oxidation systems of prac-
tical interest must include a quantitative understanding of the relative effects of
mass transfer (steps 1,2,6,7) and surface reaction (steps 3–5). The temperature
relationship of these two mechanisms is shown in Figure 5 (32,33,35). As a
catalyst is heated, conversion of the pollutant is negligible until a critical

Catalytically supported
homogenous reaction
Reaction rate

Mass diffusion
controlled

Surface kinetics
controlled

Temperature

Fig. 5. Reaction rate profile as a function of temperature (20).


Vol. 2 EMISSION CONTROL, INDUSTRIAL 547

temperature is reached, then the rate of conversion increases rapidly with rising
temperature. This is referred to as the kinetically limited region. Conversion
increases in this region because catalytic reaction rates increase with tempera-
ture, until the catalyst’s normal operating temperature is achieved. Then the
conversion rate increases only slightly with further temperature rise in the
mass-transfer limited region. At some advanced temperature, the conditions
reach a point where thermal oxidation begins to play a role, and the rate of con-
version again increases rapidly.
In the mass-transfer limited region, conversion is most commonly increased
by using more catalyst volume or by increasing cell density, which increases
the catalytic wall area per volume of catalyst. When the temperature reaches
a point where thermal oxidation begins to play a role, catalyst deactivation
may become a concern.
Reaction Rate. The kinetics for a single catalytic reaction can be
modeled as

rm ¼ kðTÞf ðCÞn ð1Þ

where rm is the rate of the main reaction; k(T) is the rate constant, a function of
temperature, T; f(C) is the function of reactant and product concentration, C;
and n is the effectiveness factor, which accounts for pore-diffusional resistance
(39). The form of the terms k(T) and f(C) depends on the kinetic model for the
system. Kinetic models for the catalytic oxidation can either be empirical or
mechanistic.
Empirical Models. In the case of an empirical equation, the model is a
power law rate equation that expresses the rate as a product of a rate constant
and the reactant concentrations raised to a power (32), such as

rm ¼ kCa1 Cb2 ð2Þ

where rm is the rate constant; C1 is the concentration of reactant 1; C2 is the


concentration of reactant 2; and a and b are empirically determined reaction
orders.
For combustion of simple hydrocarbons, the oxidation reactions appear to
follow classical first-order reaction kinetics sufficiently closely that practical
designs can be established by application of the empirical theory (23). For exam-
ple, the general reaction for a hydrocarbon:

Cx Hy þ ðx þ y=2Þ O2 ! x CO2 þ ðy=2Þ H2 O ð3Þ

can be represented by the rate equation

rm ¼ ðdC=dtÞ ¼ kC ð4Þ

where C ¼ hydrocarbon concentration, rm ¼ rate of change of contaminant con-


centration, t ¼ time, and k ¼ reaction rate constant, which must be determined
experimentally from the burning of various organic materials. The pattern of
548 EMISSION CONTROL, INDUSTRIAL Vol. 2

variation with t is predictable from kinetic theory and follows the Arrhenius
equation,

k ¼ AexpðE=RTÞ ð5Þ

where A is the Arrhenius collision constant, DE the activation energy, and R is


the universal gas constant. A catalyst increases the rate of reaction by adsorb-
ing gas molecules on catalytically active sites. The catalyst may function
simply to bring about a higher concentration of reactive materials at the
surface than is present in the bulk gas phase, which has the effect of increasing
the collision constant, A, or the catalyst may modify a molecule of adsorbed
gas by adding or removing an electron or by physically opening a bond. This
has the effect of decreasing the activation energy, DE, in the Arrhenius equa-
tion. In either circumstance, it is necessary for the reactive materials to reach
the active catalyst surface by diffusion through the gas phase, and for the reac-
tion products to leave the surface. For the conditions encountered in most
hydrocarbon emission control applications, the oxygen partial pressure is
much larger than the organic reactant partial pressure, and can be treated
as a constant.
Mechanistic kinetic expressions are often used to represent the rate data
obtained in laboratory studies, and to explain quantitatively the effects
observed in the field. Several types of mechanisms have been proposed. These
differ primarily in complexity, and on whether the mechanism assumes that
one compound that is adsorbed on the catalyst surface reacts with the other
compound in the gas phase, eg, the Eley-Rideal mechanism (38); or that both
compounds are adsorbed on the catalyst surface before they react, eg, the
Langmuir-Hinshelwood mechanism (40).
The volatile organic compounds on the list of hazardous air pollu-
tants under the CAAA have been classified into four main categories: (1) pure
hydrocarbons (qv), (2) halogenated hydrocarbons, (3) nitrogenated hydrocarbons
(see CYANIDES), and (4) oxygenated hydrocarbons. The compounds in these groups
are characterized by the following oxidation reactions (41):
Hydrocarbons
C6 H6 þ 9 O2 ! 6 CO2 þ 6 H2 O ð6Þ

Halogenated hydrocarbons

CCl4 þ O2 þ 2 H2 O ! CO2 þ 4 HCl þ O2 ð7Þ

Nitrogenated hydrocarbons

2 HCN þ 3 O2 ! 2 CO2 þ 2 H2 O þ 2 N2 ð8Þ

Oxygenated hydrocarbons

1
C4 H8 O þ 5 O2 ! 4 CO2 þ 4 H2 O ð9Þ
2
Vol. 2 EMISSION CONTROL, INDUSTRIAL 549

Table 4. Ignition Temperatures for 90% Conversion


Component Temp, 8C
hydrogen 93
acetylene 177
carbon monoxide 218
cyclohexanone 218
propylene 232
toluene 232
2-propanol 260
ethylene 260
benzene 260
xylene 260
ethanol 260
methyl ethyl ketone 274
ethyl acetate 288
cyclohexane 288
n-hexane 316
methyl isobutyl ketone 316
propane 399
methane 427a
a
Ref. 22.

Temperature reaction rate profiles for representatives compounds are available


(36,41). Particularly important are the operating temperatures required before
destruction is initiated. Chemical reactivity by compound class from high to
low is (42) alcohols > cellsolves/dioxane > aldehydes > aromatics > ketones >
acetates > alkanes > chlorinated hydrocarbons. In general, within a class the
higher the molecular weight, the higher the relative destructibility. All of
these compound classes, except chlorinated hydrocarbons, can be destroyed
with 98–99% efficiency at sufficiently low space velocities and/or high enough
inlet temperatures (43). Table 4 (37) presents oxidation temperatures for a
number of hydrocarbons.
Historically, the destruction efficiency for chlorinated hydrocarbons is quite
low. In addition, tests conducted after the chlorinated hydrocarbon is treated
show that the catalyst is partially deactivated. More recent advancements in cat-
alyst technology have resulted in the development of a number of catalysts and
catalytic systems capable of handling most chlorinated hydrocarbons under a
variety of conditions (34).
Mixture Effects. Care must be taken in determining the oxidation
kinetics for a mixture of chemicals (44). In principle, given one set of conditions
and a two-component mixture, the overall conversion of one component A may be
controlled by mass transfer to the catalyst surface and the conversion of another
component B by surface-reaction kinetics. Of course, the controlling regime
(mass transfer or reaction) can change with temperature. Thus for two indepen-
dent parallel reactions the combined effect of diffusional and reaction rate resis-
tances can have a considerable influence on the relative rate of the two reactions.
Additionally, a third, fourth, or nth component can conceivably affect the other
components by, eg, competing more successfully for active surface sites than B
while simultaneously influencing the mass transfer of A. Thus even for a simple
550 EMISSION CONTROL, INDUSTRIAL Vol. 2

two- or three-component mixture, interpretation of observed results can be


difficult. Extrapolation of mixture behavior from single-component data is
ill-advised.
In a mixture of n-hexane and benzene (44), the deep catalytic oxidation
rates of benzene and n-hexane in the binary mixture are lower than when
these compounds are singly present. The kinetics of the individual compounds
can be adequately represented by the Mars-VanKrevelen mechanism. This
model needs refinements to predict the kinetics for the mixture.
One important consideration in any catalyst oxidation process for a complex
mixture in the exhaust stream is the possible formation of hazardous incomplete
oxidation products. Whereas the concentration in the effluent may be reduced to
acceptable levels by mild basic aqueous scrubbing or additional vent gas treat-
ment, studying the kinetics of the mixture and optimizing the destruction
cycle can drastically reduce the potential for such emissions.
Design and Operation. The destruction efficiency of a catalytic oxidation
system is determined by the system design. It is impossible to predict a priori the
temperature and residence time needed to obtain a given level of conversion of a
mixture in a catalytic oxidation system. Control efficiency is determined by pro-
cess characteristics such as concentration of VOCs emitted, flow rate, process
fluctuations that may occur in flow rate, temperature, concentrations of other
materials in the process stream, and the governing permit regulation, such as
the mass-emission limit. Design and operational characteristics that can affect
the destruction efficiency include inlet temperature to the catalyst bed, volume
of catalyst, and quantity and type of noble metal or metal oxide used.
Catalytic oxidation systems are normally designed for destruction efficien-
cies that range from 90 to 98% (42). In the early 1980s, typical design require-
ments were for 90% or higher VOC conversions. More recently, however, an
increasing number of applications require 95 to 98% conversions to meet the
more stringent emission standards (35).
Operational Considerations. The performance of catalytic incinerators
(43) is affected by catalyst inlet temperature, space velocity, superficial gas velo-
city (at the catalyst inlet), bed geometry, species present and concentration, mix-
ture composition, and waste contaminants. Catalyst inlet temperatures strongly
affect destruction efficiency. Mixture compositions, air/gas (fuel) ratio, space velocity,
and inlet concentration all show marginal or statistically insignificant effects (45).
Operating Temperature. The operating temperature needed to achieve
a particular VOC destruction efficiency depends primarily on the species of pol-
lutants contained in the waste stream, the concentration of the pollutants, and
the catalyst type (29). One of the most important factors is the hydrocarbon
species. Each has a catalytic initiation temperature that is also dependent on
the type of catalyst used (29).
For a given inlet temperature, the quantity of supplied heat may be pro-
vided by (13) the heat supplied from the combustion of supplemental fuel, the
sensible heat contained in the emission stream as it enters the catalytic system,
and the sensible heat gained by the emission stream through heat exchange with
hot flue gas (13). Three types of systems for catalytic oxidation of VOCs are
shown in Figure 6 (26). The simplest (Fig. 5a) uses a direct contact open flame
to preheat the gas stream upstream of the catalyst. The second (Fig. 5b) involves
Vol. 2 EMISSION CONTROL, INDUSTRIAL 551

Process
exhaust

Clean Clean
Burner air Burner air
Preheat section Preheat section
Catalyst Catalyst

Heat Process
exchanger exhaust
(a) (b)

To dryer
Secondary heat
exchanger

Clean
Burner air
Preheat section

Catalyst bed
Process Air
Primary heat exhaust
exchanger

(c)

Fig. 6. Catalytic system designs (26) of (a) basic VOC catalytic converter containing a
preheater section, a reactor housing the catalyst, and essential controls, ducting, instru-
mentation, and other elements; (b) a heat exchanger using the cleaned air exiting the
reactor to raise the temperature of the incoming process exhaust; and (c) extracting
additional heat from the exit gases by a secondary heat exchanger.

only a catalyst bed over which the gas stream passes, usually after some indirect
preheating. The third (Fig. 5c) involves more extensive indirect preheating and
heat exchange. The difference in the three configurations is the method for pre-
heating the gas.
There are two general temperature policies: increasing the temperature
over time to compensate for loss of catalyst activity, or operating at the maxi-
mum allowable temperature. These temperature approaches tend to maximize
destruction, yet may also lead to loss of product selectivity. Selectivity typically
decreases with increasing temperature; faster deactivation; and increased costs
for reactor materials, fabrication, and temperature controls.
Reactor Design. The catalytic reactor is designed to be operated in the
masstransfer controlled catalytic region. The prime design parameter is the geo-
metric surface area. The honeycomb catalyst shows substantial advantage over
other forms because of the high geometric surface areas obtainable with low
pressure drop (35).
Catalyst Selection. The choice of catalyst is one of the most important
design decisions. Selection is usually based on activity, selectivity, stability,
552 EMISSION CONTROL, INDUSTRIAL Vol. 2

mechanical strength, and cost (46). Stability and mechanical strength, which
make for steady, long-term performance, are the key characteristics. The basic
strategy in process design is to minimize catalyst deactivation, while optimizing
pollutant destruction.
Both catalyst space velocity and bed geometry play a role. The gas hourly
space velocity (GHSV) is used to relate the volumetric flow rate to the catalyst
volume. GHSV has units of inverse hour and is defined as the volume flow
rate per catalyst volume.
The size of the catalyst bed depends mainly on the degree of VOC reduction
required (29). VOC destruction efficiencies up to 95% can usually be attained
using reasonable space velocities (29). However, the low GHSVs, and subse-
quently high catalyst volumes required to achieve extremely high (eg, 99%) con-
versions, can sometimes make catalytic oxidation uneconomical. Conventional
bed geometries may be found in the literature (29).
Process Conditions. To effectively design a catalytic control system, the
Manufacturers of Emissions Controls Association recommends the following data
be obtained (5): list of all VOCs present and range of concentration of each, flow
rate of exhaust and expected variability, oxygen concentration in exhaust and
expected variability, temperature of exhaust and expected variability, static
pressure, potential uses for heat recovery, particular performance criteria and/
or regulations to be met, capture efficiency, ie, fraction of all organic vapors
generated by the processes that are directed to the control device, presence of
hydrocarbon aerosols in the effluent exhaust, identity and quantity of all inor-
ganic and organic particulate, amount of noncombustibles, presence of possible
catalyst deactivators, and anticipated start-up/shutdown frequency of the
system.
Pilot Studies. Applications requiring the reduction of VOC emissions
have increased dramatically. On-site pilot tests are beneficial in providing useful
information regarding VOC emission reduction applications. Information that
can be obtained includes optimum catalyst operating conditions, the presence
of contaminants in the gas stream, and the effects of these contaminants.
Catalyst Inhibition. A number of potential applications for catalytic oxida-
tion of organic materials have resulted in serious odor or eye irritation, or visible
emission problems (23). Some of these failures are a result of fouling of the cat-
alyst surface. Others occur because materials such as halogens in the gas stream
interfere with or suppress the activity of the catalyst, or because the substances
react with the precious metals, rendering them permanently inactive. Finally, all
catalysts eventually deteriorate by aging or thermal processes (23).
Many of the exhaust streams that must be purified contain significant
amounts of halogenated organics, such as polychlorinated ethanes and ethylenes
vented in the manufacture of vinyl chloride monomer or released in usage sol-
vents (47). However, the catalysts used in the conventional catalytic oxidation
are severely inhibited by the halogen atoms in these compounds (47). Other
trace contaminants of concern in air streams may include phosphorus-, nitro-
gen-, and sulfur-containing compounds. Whereas gases containing chlorine, sul-
fur, and other atoms can deactivate supported noble metal catalysts such as
platinum, chlorinated VOC can be treated by certain supported metal oxide
catalysts (22).
Vol. 2 EMISSION CONTROL, INDUSTRIAL 553

Coating of
catalyst site Plugging of Inactive compound
catalyst pores Exhaust
Exhaust Active site
gas
gas

Wall of support

(a) (b)

Sintering of Sintering of
alumina precious metal

Oxide/oxide

Exhaust
gas Ceramic oxide

Roughened surface

Ceramic oxide
monolith Strong chemical and
physical bond
Thermal expansion

(C)
Corrosion at interface
Oxide/metal

Exhaust
gas Ceramic oxide

Smooth surface

Metal substrate
monolith Weaker bonds

Thermal expansion

(d)

Fig. 7. Catalyst inhibition mechanisms where (.) are active catalyst sites; , the cata-
lyst carrier; and the catalytic support: (a) masking of catalyst; (b) poisoning of catalyst;
(c) thermal aging of catalyst; and (d) attrition of ceramic oxide metal substrate monolith
system, which causes the loss of active catalytic material resulting in less catalyst in the
reactor unit and eventual loss in performance.

The four basic mechanisms of catalyst decay are shown in Figure 7 (5,33,39).
These are fouling or masking, poisoning, thermal degradation through aging or
sintering, and loss of catalyst material through formation and escape of vapors.
Poisoning and vapor transport are basically chemical phenomena, whereas fouling
is mechanical. Table 5 lists substances that inhibit catalyst activity (5).
554 EMISSION CONTROL, INDUSTRIAL Vol. 2

Table 5. Substances that Inhibit Catalyst Activitya


Type of
inhibitor Effect Examples Regeneration

fast-acting reduction of catalyst phosphorus, catalyst


inhibitors activity at rate bismuth, lead, regeneration is
depending on arsenic, sometimes difficult
concentration and antimony, or impossible
temperature mercury
slow-acting reduction of catalyst iron, tin, silicon catalyst
inhibitors activity; higher regeneration
concentrations than remains difficult or
those of fast-acting impossible
catalyst inhibitors
may be tolerated
reversible inhibitors/ surface coating of sulfur, halogens, regeneration is
maskers catalyst active area; silicon, zinc, possible
rate also dependent phosphorus
on concentration and
temperature
surface maskers surface coating of organic solids removed by
catalyst active areas increasing catalyst
temperature or by
acid and alkaline
washing
surface eroders surface coating of inert surface coating is
and maskers catalyst active area, particulates easily removed by
or erosion of catalyst washing
surface; both result
in loss of catalyst
activity; rate
dependent on
particle size, grain
loading, and gas
stream velocity
thermal degradation loss of catalyst surface higher regeneration
and sintering area because of temperature generally very
catalyst dispersion for extended difficult; best
and crystal growth, time, avoided by
or catalyst support temperature operating
collapse through excursions, hot in optimum
sintering spots in bed temperature range
and avoiding
temperature
excursions
vapor transport loss of catalytic CO, NO, must replace lost
and attrition material through hydrogen catalytic material;
formation of metal sulfide, vaporization
carbonyl oxides, halogens, and generally not a
sulfides, and halides, particulates factor; attrition
or surface shear particularly
effects resulting important in
from exhaust gas fluidized beds
velocity,
particulates, or
thermal shock
a
Refs. 5 and 37.
Vol. 2 EMISSION CONTROL, INDUSTRIAL 555

Masking or Fouling. Masking or fouling is a physical deposition of spe-


cies from the fluid phase onto the catalyst surface (Fig. 6a) that results in block-
age of reaction sites or pores (39). Masking or fouling is caused by a gradual
accumulation of noncombusted, solid material that mechanically coats the
catalyst’s surface and prevents or slows down the diffusion of reactants to the
catalyst.
Typical masking or fouling agents include (5,23,36,39) airborne dust or dirt;
metal oxides formed from materials in the process, such as silicon dioxide ash
remaining when silicone compounds are oxidized; aggregate compound form-
ation on the catalyst surface, ie, phosphorus for lubricating oils; corrosion products
from the duct system; and organic char or tars formation from incomplete com-
bustion products, often caused by too low a reactor operating temperature.
Low levels of particulates or potential poisons can sometimes be tolerated
without a dramatic decrease in performance. Generally, it has been recom-
mended that the maximum particulate concentration not exceed 115 mg/m2
and that the maximum poison concentration not exceed 25 ppm (29). In addition,
every effort should be made to avoid flow over a cold catalyst bed for any
extended period of time, as a process stream containing volatile organics may
condense on a cold catalyst bed (5).
Combustible masking materials such as organic char may be partially
or completely removed by periodic elevations of the catalyst bed temperature.
Noncombustible masking materials may be removed by air lancing or aqueous
washing generally with a leaching solution (35,36).
Poisons Halogens, sulfur dioxide [7446-09-5], SO2, nitrogen dioxide
[10102-44-0], NO2, and numbered other materials act as catalyst suppress-
ants for precious-metal oxidation catalysts. These compounds tend to adsorb
strongly on the catalytic surface, preventing the reactants from doing so. The
strength of adsorption is ordinarily such that the suppressant materials can
gradually be stripped off after there is no longer a concentration of suppressant
materials in the gas stream passing through the catalyst (23). In other cases,
the adsorption is irreversible. A poison blocks the catalytic sites, and may
also induce changes in the surface to result in formation of compounds (39).
Active precious-metal sites become inactive, reducing catalyst performance
(see Fig. 6b).
At low (>4508C) temperatures, the presence of these materials, particularly
the oxides, leads to simple masking or fouling. In some cases, a catalyst that
shows reduced activity believed to be from poisoning may simply be masked,
and activity can be rejuvenated by cleaning with aqueous leaching solutions (36).
Poisoning is operationally defined. Often catalysts believed to be perma-
nently poisoned can be regenerated (5) (see CATALYSTS, REGENERATION). A species
may be a poison in some reactions, but not in others, depending on its adsorption
strength relative to that of other species competing for catalytic sites (39), and
the temperature of the system. Catalysis poisons have been classified according
to chemical species, types of reactions poisoned, and selectivity for active catalyst
sites (39).
Groups 14 and 15 (VA and VIA) elements act as poisons. The interaction
depends on the poison’s oxidation state and chemical structure. For sulfur, the
order of decreasing poisoning activity is H2S > SO2 > SO2 4. Adsorption studies
556 EMISSION CONTROL, INDUSTRIAL Vol. 2

indicate that H2S adsorbs strongly and dissociates on nickel surfaces. The sulfur
adsorbs essentially irreversibly and over most of the catalyst–metal surface. It
has been observed that SO2 and SO3 also poison catalysts differently (28); SO2
selectively adsorbs on Pt or Pd in an oxidation catalyst, whereas SO3 reacts
with the Al2O3 carrier, forming Al2(SO4)3, which destroys the structure of the
catalyst. The latter can be prevented by using a more inert support such as
SiO2 or TiO2. The former requires a change in operating conditions such as a
higher temperature. When No. 2 fuel oil is used in some gas turbines, the sulfur
compound in the fuel oil can be converted to SO2 at levels of 40–150 ppm in the
exhaust. For such applications, the presence of 100–200 ppm SO2 can require
150–2008C higher temperature for the catalyst to give the same CO conversion
as without SO2.
Toxic heavy metals and ions, eg, Pb, Hg, Bi, Sn, Zn, Cd, Cu, and Fe, may
form alloys with catalytic metals (39). Materials such as metallic lead, zinc,
and arsenic react irreversibly with precious metals and make the surface una-
vailable for catalytic reactions. Poisoning by heavy metals ordinarily destroys
the activity of a precious-metal catalyst (23).
Molecules having unsaturated bonds, eg, CO, NO, HCN, and benzene, may
chemisorb through multiple bonds (39).
Catalysts having improved poison resistance have been developed. Cata-
lysts are available that can destroy chlorine-, fluorine-, or bromine-containing
organic compounds (5).
Thermal Degradation and Sintering. Thermally induced deactivation
of catalysts may result from redispersion, ie, loss of catalytic surface area
because of crystal growth in the catalyst phase (36,39,48) or from sintering, ie,
loss of catalyst-support area because of support collapse (39). Sintering processes
generally take place at high (>5008C) temperatures and are generally acceler-
ated by the presence of water vapor (see Fig. 6c). Another thermal effect is the
transformation of catalytic phases to noncatalytic ones, eg, the reaction of nickel
and alumina to form nickel aluminate (39). Each catalyst has a recommended
temperature window of operation. At temperatures above this window (usually
7608C), sintering can occur.
Loss of Catalyst by Vapor Transport. The direct volatilization of cata-
lytic metals is generally not a factor in catalytic processes, but catalytic metal
can be lost through formation of metal carbonyl oxides, sulfides, and halides in
environments containing CO, NO, O2, and H2S, and halogens (39).
The ceramic oxide carrier is bonded to the monolith by both chemical and
physical means. The bonding differs for a ceramic monolith and a metallic mono-
lith. Attrition is a physical loss of the carrier from the monolith from the surface
shear effects caused by the exhaust gas, a sudden start-up or shutdown causing a
thermal shock as a result of different coefficients of thermal expansion at the
boundary between the carrier and the monolith, physical vibration of the cata-
lyzed honeycomb, or abrasion from particulates in the exhaust air (36)
(see Fig. 6d).
Avoiding Catalyst Deactivation. Catalyst deactivation is more easily
prevented than cured. Poisoning by impurities may be prevented by removing
impurities from the reactants. Carbon deposition and coking may be prevented
by minimizing formation of precursors and manipulating mass-transfer regimes
Vol. 2 EMISSION CONTROL, INDUSTRIAL 557

so as to minimize the carbon’s or coke’s effect on activity. Most sintering is irre-


versible, or reversible only with great difficulty, so it is important to choose reac-
tion, ie, lower temperatures, that do not sinter the catalyst. Additionally, when
process upsets that could release inhibitors or cause small fluctuations in the
heating value of the oxidizer are highly probable a thermal system is favored
over a catalytic one (22).
Except for No. 2, fuel oil should not be considered as auxiliary fuel when
using a catalytic system because of the sulfur and vanadium the fuel oil may con-
tain (22). In some cases even the sulfur in No. 2 fuel oil can present a problem.
Galvanized metal should not be used in process ovens or ductwork because zinc is
a catalyst poison.
Proper system design and catalyst maintenance are key to minimizing
deactivation and providing long-term catalyst service. For example, control of
air dilution, use of temperature control loops, and use of catalysts having high
intrinsic thermal stability can provide necessary protection against high tem-
perature damage caused by reaction exotherms and from operational upsets (35).
Experimental Evaluation. Often the deactivation kinetics for a catalytic
oxidation system can be evaluated in a series of laboratory studies (39). Reactors
should be gradientless with respect to reactant poison concentration and
temperature. Heat- and mass-transfer effects should be avoided because these
disguise the intrinsic kinetics. Experiments should be designed to study
one deactivation process at a time, and accelerated targets must be representa-
tive of the process. Deactivation can be accelerated by using smaller amounts of
catalyst, operating at higher temperatures or different pressures, at greater resi-
dence times, or at different gas compositions.
Whereas changing catalyst volume or residence time rarely yields compli-
cations, changing temperature or pressure could introduce sintering. The
properties of the catalyst should be measured both before and after deactiva-
tion and inlet and outlet streams should be analyzed by chromatography (qv)
or spectrometry.
Around 1972, it was reasoned that the problem of catalyst deactivation
could not always be entirely eliminated, but that continuous replacement of a
portion of the catalyst bed during normal operation would allow continuing
operation at high efficiency even in the presence of poisoning agents. Hence,
the fluidized bed was born (23). In some applications, fluidized-bed oxidation pro-
cesses overcame poisoning, masking, and thermal aging. A process in which per-
formance depends on the continuous attrition of the external surface of the
catalyst particles, however, has many unattractive features, including the effort
required for trapping, collecting, and disposing of the fine particulate released
from the reactor (47).
At least one printer using catalytic oxidation has experienced relatively
rapid catalyst deactivation, requiring replacement after useful lives as short as
3–6 months, when producing high quality printed matter using some of the most
desirable lithographic printing plates and materials (49). It was observed that
the use of phosphorus additives caused rapid deactivation of the conventional
catalyst used to destroy the hydrocarbons in the solvent-laden air (SLA) dis-
charged from the press dryer. A precious-metal catalyst containing platinum
and palladium was being used in this application. It had replaced an earlier
558 EMISSION CONTROL, INDUSTRIAL Vol. 2

base metal catalyst, which showed rapid deactivation as a result of sulfur in the
SLA, presumably introduced in the fuel used to fire the heatset dryer. It was
found that the P concentration in the SLA might be as high as 0.16 ppm. The
phosphorus concentration on the deactivated catalyst was found to be 1.4% of
the catalyst. The printer was urged by the supplier to find and eliminate the
cause of the phosphorus contamination of the waste gas entering the catalyst
bed. However, after it was determined that use of phosphorus-containing addi-
tives was crucial to many of the high quality printing jobs, attention was directed
at the catalyst bed itself.
A catalyst with a substantially improved resistance to poisoning by
phosphorus in catalytic oxidation applications was developed. In part, the cata-
lyst in this program permitted printers to use photolighographic technology
without paying an unreasonable cost in terms of frequent replacement of oxida-
tion catalysts.
Catalyst Reactivation. Some catalytic systems are reported to have
operted continuously for >10 years with little or no loss in control efficiency
(5). In most processes catalysts inevitably lose activity, and when the activity
has declined to a critical level, the catalyst needs to be discarded or regenerated
(39). Regeneration is only possible when the deactivation is reversible by chemi-
cal washing or heat treatment or oxidation (36,39).
Thermal Treatment. A thermal treatment for catalyst regeneration
is usually effective when deactivation is a result of coking or masking of the
catalyst surface. Thermal treatment can usually be done on-site, by elevating
the temperature of the catalyst bed by 50–1008C above the normal operating
point and running at this oxidizing condition for a specified limited period of
time (35). The elevated temperature vaporizes or oxidizes the organic compounds
or char that may be masking the catalyst surface.
Physical Treatment. If inspection of the catalyst indicates deposits
again, or if an excessive pressure drop across the catalyst is noted, then the cat-
alytic bed may be lanced, on-site, using compressed air or water until the depos-
its are removed. Abrasion by contact with excessively high pressure from the
compressed air should be avoided (35). If this treatment is combined with heat-
ing, hot spots or overtemperatures that could further deactivate the catalyst
should be avoided (39). In many cases, periodic maintenance, removing the
catalyst bed and blowing or washing off residues, has restored catalyst to original
or near-original activity levels (5).
Chemical Treatment. The most involved regeneration technique is
chemical treatment (35) which often follows thermal or physical treatment,
after the char and particulate matter has been removed. Acid solution soaks,
glacial acetic acid, and oxalic acid are often used. The bed is then rinsed with
water, lanced with air, and dried in air. More involved is use of an alkaline solu-
tion such as potassium hydroxide, or the combination of acid washes and alkaline
washes. The most complex treatment is a combination of water, alkaline, and
acid washes followed by air lancing and drying. The catalyst should not be appre-
ciably degraded by the particular chemical treatment used.
Analyses of a catalyst used in a process involving cloaning products and pig-
ments and achieving a hydrocarbon destruction capacity of only 13% showed
deposition of P, Sn, Pb, and Na contaminants (35). Initial acid treatment
Vol. 2 EMISSION CONTROL, INDUSTRIAL 559

increased the hydrocarbon destruction capacity from 13 to 63%. Alkaline treat-


ment increased the capacity to 90% of that new.

6. Exhaust Control Technologies

In addition to VOCs, specific industrial exhaust control technologies are avail-


able for nitrogen oxides, NOx, carbon monoxide, CO, halogenated hydrocarbon,
and sulfur and sulfur oxides, SOx.
6.1. Nitrogen Oxides. Annual releases of nitrogen oxides (NOx) into the
atmosphere amounted to 550 106 t in the early 1990s. A number of states,
in addition to California, regulate NOx emissions (50). New atmospheric model-
ing analysis has indicated that in many cases, NOx has a significant role in the
formation of tropospheric ozone, a heavily regulated atomospheric pollutant
(51,52). The production of nitrogen oxides can be controlled to some degree by
reducing formation in the combustion system. The rate of NOx formation for
any given fuel and combuster design are controlled by the local oxygen concen-
tration, temperature, and time history of the combustion products. Techniques
employed to reduce NOx formation are collectively referred to as combustion con-
trols and U.S. power plants have shown that furnace modifications can be a cost-
effective approach to reducing NOx emissions. Combustion control technologies
include operational modifications, such as low excess air, biased firing, and
burners-out-of-service, which can achieve 20–30% NOx reduction; and equipment
modifications such as low NOx burners, overfire air, and reburning, which can
achieve 40–60% reduction (53,54) As of the early 1990s, 600 boilers having
10,000 MW of capacity use combustion modifications to comply with the New
Source Performance Standards (NSPS) for NOx emissions (55). Since then,
such modifications have become standard practice (52).
When NOx destruction efficiencies approaching 90% are required, some
form of postcombustion technology applied downstream of the combustion zone
is needed to reduce the NOx formed during the combustion process. Three post-
combustion NOx control technologies are utilized: selective catalytic reduction
(SCR); nonselective catalytic reduction (NSCR); and selective noncatalytic reduc-
tion (SNCR).
Selective Catalytic Reduction. Selective catalytic reduction (SCR) is
widely used in Japan and Europe to control NOx emissions (1). SCR converts
the NOx in an oxygen-containing exhaust stream to molecular N2 and H2O
using ammonia as the reducing agent in the presence of a catalyst. NOx
removals of 90% are achievable. The primary variable is temperature, which
depends on catalyst type (56). The principal components of an SCR system
include the catalyst, the SCR reactor, the ammonia injection grid (AIG), the
ammonia—air dilution system, the ammonia storage—vaporization system,
the ammonia addition control system, and a continuous emissions monitoring
system (57).
The AIG is used to uniformly inject diluted ammonia into the reactor. Uni-
form mixing of the ammonia into the flue gas is necessary to maintain catalyst
performance at its highest level and to minimize ammonia leakage (ammonia
slip) past the catalyst.
560 EMISSION CONTROL, INDUSTRIAL Vol. 2

The ammonia—air dilution system dilutes the vaporized ammonia by a


factor of 20–25 with air for better admixing through the AIG and to prevent
explosive ammonia—air mixtures. Once the catalyst volume is selected, the
NOx removal is set by the NH3/NOx mole ratio at the inlet of the SCR system (57).
SCR was first developed in the United States in the late 1950s, targeted at
nitric acid tail-gas exhausts, using precious-metal catalysts. In the mid-1970s,
SCR entered widespread commercial use in Japan using base metal catalysts.
The first SCR systems in Germany started up in 1986 (57), and German utilities
are committed to installing SCR systems on the majority of oil- and coal-fired boi-
lers to achieve between 60 and 90% NOx reductions (57). By the end of 1992,
there were 120 SCR plants in service in Germany alone (58). An increasing
amount of experience has been documented in the United States (59–61),
although commercial service began in 1985 and has continued to increase
(62,63) in those areas where NOx remains a concern.
Performance criteria for SCR are analogous to those for other catalytic oxi-
dation systems: NOx conversion, pressure drop, catalyst/system life, cost, and
minimum SO2 oxidations to SO3. An optimum SCR catalyst is one that meets
both the pressure drop and NOx conversion targets with the minimum catalyst
volume. Because of the interrelationship between cell density, pressure drop,
and catalyst volume, a wide range of optional catalyst cell densities are needed
for optimizing SCR system performance.
Reactions. The SCR process is termed selective because the ammonia
reacts selectively with NOx at temperatures >2328C in the presence of excess
oxygen (64). The optimum temperature range for the SCR catalyst is determined
by balancing the needs of the redox reactions.
SCR reactions

4 NO þ 4 NH3 þ O2 ! 4 N2 þ 6 H2 O ð10Þ
2 NO2 þ 4 NH3 þ O2 ! 3 N2 þ 6 H2 O ð11Þ

The NO reduction is the most important because NO2 accounts for only
5–10% of the NOx in most exhaust gases.
Ammonia oxidation reactions

4 NH3 þ 5 O2 ! 4 NO þ 6 H2 O ð12Þ
4 NH3 þ 3 O2 ! 4 N2 þ 6 H2 O ð13Þ

When sulfur dioxide is also present there are important side reactions in which
SO2 is oxidized to SO3. The main side reaction in the SCR catalyst is the conver-
sion of SO2 to SO3, thus facilitating the reaction above. The SO3 in turn reacts
with ammonia to form ammonium sulfates.

NH3 þ SO3 þ H2 O ! NH4 HSO4 ð14Þ


NH3 þ SO3 þ H2 O ! ðNH4 Þ2 SO4 ð15Þ

The formation of ammonium bisulfate is strongly temperature dependent.


Formation is favored at the lower temperatures. The temperature at which
Vol. 2 EMISSION CONTROL, INDUSTRIAL 561

ammonium bisulfate is not formed depends strongly on the SO3 concentration in


the exhaust gas. The temperature needed to minimize bisulfate formation has
been reported to increase by 158C (around 3508C) when the SO3 concentration
increases from 5 to 15 ppm (38). The formation of the bisulfate is reversible, ie, if
the temperature is raised to 208C above the minimum temperature, the reaction
is shifted to result in the decomposition of the bisulfate formed. When chlorides
are present, ammonium chlorides can be formed:

NH3 þ HCl ! NH4 Cl ð16Þ

When sulfuric acid is present, ammonium bisulfate can be formed:

NH3 þ H2 SO4 ! NH4 HSO4 ð17Þ

These various reactions should be minimized to avoid plugging the catalyst and
to prevent fouling of the downstream air preheaters, when these components
condense from the gas at the lower temperatures.
The SCR Process. The first step in the SCR reaction is the adsorption of
the ammonia on the catalyst. SCR catalysts can adsorb considerable amounts of
ammonia (65). However, the adsorption must be selective and high enough to
yield reasonable cycle times for typical industrial catalyst loadings, ie, uptakes
in excess of 0.1% by weight. The rate of adsorption must be comparable to the
rate of reaction to ensure that suitable fronts are formed. The rate of desorption
must be slow. Ideally the adsorption isotherm is rectangular. For optimum per-
formance, the reaction must be irreversible and free of side reactions.
It has been suggested that the first step, weak coadsorption of NO and O2 on
a reduced vanadium site, may represent the slow step in the mechanism. Subse-
quent formation of a N2O3-like intermediate could be a fast step, because it is
known that in the gas phase the equilibrium of NONO2 and N2O3 is established
within microseconds (52).
At low temperatures the SCR reactions dominate and nitrogen oxide con-
version increases with increasing temperature. But as temperature increases,
the ammonia oxidation reactions become relatively more important. As the tem-
perature increases further, the destruction of ammonia and generation of nitro-
gen oxides by the oxidation reactions causes overall nitrogen oxide conversion to
reach a plateau then decreases with increasing temperatures. Examples are
shown in Figure 8 (64).
In the SCR process, ammonia, usually diluted with air or steam, is injected
through a grid system into the flue/exhaust stream upstream of a catalyst bed
(55). The effectiveness of the SCR process is also dependent on the NH3 to NOx
ratio. The ammonia injection rate and distribution must be controlled to yield an
1:1 molar ratio. At a given temperature and space velocity, as the molar ratio
increases to approximately 1:1, the NOx reduction increases. At operations above
1:1, however, the amount of ammonia passing through the system increases (56).
This ammonia slip can be caused by catalyst deterioration, by poor velocity dis-
tribution, or inhomogeneous ammonia distribution in the bed.
Types of SCR Catalysts. The catalysts used in the SCR were initially
formed into spherical shapes that were placed either in fixed-bed reactors for
562 EMISSION CONTROL, INDUSTRIAL Vol. 2

100

80

NOx Conversion, %
60

40

20

0
100 200 300 400 500 600
Temperature, °C

Fig. 8. NOx reduction versus temperature for SCR catalysts: ( ) CAT LT-1, ( ) CAT
LT-2, ( ) CAT VNX, and ( ) CAT 2NX (59).

clean gas applications or moving-bed reactors where dust was present. The
moving-bed reactors added complexity to the design and in some applications
resulted in unacceptable catalyst abrasion. As of 1993 most SCR catalysts are
either supported on a ceramic or metallic honeycomb or are directly extruded
as a honeycomb (1). A typical honeycomb block has face dimensions of 150 by
150 mm and can be as long as one meter. The number of cells per block varies
from 20 by 20 up to 45 by 45 (57).
No SCR catalyst can operate economically over the whole temperature
range possible for combustion systems. As a result, three general classes of cat-
alysts have evolved for commercial SCR systems (44): precious-metal catalysts
for operation at low temperatures, base metals for operation at medium tempera-
tures, and zeolites for operation at higher temperatures.
The precious-metal platinum catalysts were primarily developed in the 1960s
for operation at temperatures between 200 and 3008C (1,56,64). However,
because of sensitivity to poisons, these catalysts are unsuitable for many combus-
tion applications. Variations in sulfur levels of as little as 0.4 ppm can shift the
catalyst required temperature window completely out of a system’s operating tem-
perature range (64). Additionally, operation with liquid fuels is further compli-
cated by the potential for deposition of ammonium sulfate salts within the pores
of the catalyst (64). These low temperature catalysts exhibit NOx conversion that
rises with increasing temperature, then rapidly drops off, as oxidation of ammonia
to nitrogen oxides begins to dominate the reaction (see Fig. 7).
The most popular SCR catalyst formulations are those that were developed
in Japan in the late 1970s comprised of base metal oxides such as vanadium
pentoxide [1314-62-1], V2O5, supported on titanium dioxide [13463-67-7], TiO2
(1). As for low temperature catalysts, NOx conversion rises with increasing tem-
peratures to a plateau and then falls as ammonia oxidation begins to dominate
the SCR reaction. However, peak conversion occurs in the temperature range
between 300 and 4508C, and the fall-off in NOx conversion is more gradual
than for low temperature catalysis (64).
A family of zeolite catalysts has been developed, and is being increasingly
used in the United States in SCR applications. Zeolites that can function at
Vol. 2 EMISSION CONTROL, INDUSTRIAL 563

higher temperatures than the conventional catalysts, are claimed to be effective


over the range of 300 to 6008C, having an optimum temperature range from
360 to 5808C (55,56). However, ammonia oxidation to NOx begins 4508C and
is predominant at temperatures in excess of 5008C. Zeolites suffer the same per-
formance and potential damage problems as conventional catalysts when used
outside the optimum temperature range. In particular, at 5508C the zeolite
structure may be irreversibly degraded because of loss of pore density. Zeolite
catalysts have not been continuously operated commercially at temperatures
>5008C (55).
Using zeolite catalysts, the NOx reduction takes place inside a molecular
sieve ceramic body rather than on the surface of a metallic catalyst (see MOLECULAR
SIEVES). This difference is reported to reduce the effect of particulates, soot, SO2/
SO3 conversions, heavy metals, etc, which poison, plug, and mask metal cata-
lysts. Zeolites have been in use in Europe since the mid-1980s and there are
approximately 100 installations on stream. Process applications range from
use of natural gas to coal as fuel. Typically, nitrogen oxide levels are reduced
80 to 90% (55).
Catalyst Selection. For an SCR application, catalyst selection depends
largely on the temperature of the flue gas being treated. A given catalyst exhibits
optimum performance within a temperature range of 30–508C. Below this opti-
mum temperature range, the catalyst activity is greatly reduced, allowing
unreacted ammonia to slip through. Above this range, ammonia begins to be
oxidized to form additional NOx. Operations having adequate temperature
controls are important, as are uniform flue gas temperatures (55,56).
Problems. A number of difficulties in utilizing SCR operations have been
identified. Problems in European installations are of particular interest because
SCR systems are subjected to conditions not experienced in Japan, but also
encountered in the United States. Difficulties include matching the NH3 injec-
tion pattern to the nonuniform flow of NOx in the ductwork ahead of the SCR
rector; inability to optimize the NH3 injection rate by feedback control of slip
ammonia for lack of a reliable NH3 monitor; erosion and plugging on units retro-
fitted to boilers that fire high ash coal; catalyst deactivation caused by arsenic
poisoning on wet-bottom units that recycle flyash; process control under load
swings; and spent catalyst disposal. For medium to high sulfur coals, the poten-
tial exists for accelerated catalyst deactivation caused by sulfur poisoning and
contamination by trace metals in flyash, and deposit buildup and corrosion of
the air heater.
For boilers, SCR DENOx plants can be installed at the exhaust exit just
before the air preheater. These are called high dust plants (17-g dust/m3) (38)
because the flue gas still contains volatile trace elements as well as flyash parti-
cles. When the SCR system is installed after the flue gas desulfurization (FGD)
system, it is called a low dust or tail-end plant. The high dust plant has the
advantages of not requiring any additional energy because of the high tempera-
tures present. The low dust plant requires a regenerative heat exchanger, but
requires less catalyst because (58): there is less dust to contaminate the bed,
potential catalyst poisons are removed from the flue gas by the FGD system,
and a high activity catalyst can be used when only low concentrations of poisons
such as SO2 remain after the flue gas system.
564 EMISSION CONTROL, INDUSTRIAL Vol. 2

Other problems that can be associated with the high dust plant can include
alkali deterioration from sodium or potassium in the stack gas deposition on the
bed, calcium deposition, when calcium in the flue gas reacts with sulfur trioxide,
or formation and deposition of ammonium bisulfate. In addition, plugging of the
air preheater as well as contamination of flyash and FGD wastewater discharges
by ammonia are avoided if the SCR system is located after the FGD (38).
A significant problem area for initial SCR systems has been the continuous
emission monitoring (CEM) systems. In power plants, all sites equipped with
CEM systems report the highest failure frequency. The CEM systems are the
most labor intensive component, requiring as much as full-time attention from
one technician. At one power plant CEM systems were responsible for 100% of
73 reported SCR system shutdowns (56). As CEM systems continue to improve,
these concerns may disappear.
Nonselective Catalytic Reduction. Hydrocarbons, hydrogen, or carbon
monoxide can be used as reducing agents for NOx in applications where the
exhaust oxygen concentration is low, as it is in fuel rich-burn reciprocating
engines, where it is <1%, and in nitric acid plants, when it is from 2 to 3%.
This approach is called nonselective catalytic reduction (NSCR). In some applica-
tions, the oxygen must be removed from the feed stream prior to the catalyst (52).
An oxygen sensor in the exhaust stream signals the air–fuel delivery system to
adjust the air/fuel ratio so it is just slightly fuel-rich, having enough reducing
agent present to react with all the oxygen and nitrogen oxides (1).
Nonselective catalytic reduction systems are often referred to as three-way
conversions. These systems reduce NOx, unburned hydrocarbon, and CO simul-
taneously. In the presence of the catalyst, the NOx are reduced by the CO result-
ing in N2 and CO2 (55). A mixture of platinum and rhodium has been generally
used to promote this reaction . It has also been reported that a catalyst using pal-
ladium has been used in this application (1). The catalyst operation temperature
limits are 350–8008C, and 425–6508C are the most desirable. Temperatures
>8008C result in catalyst sintering (55). Automotive exhaust control systems
are generally NSCR systems, often shortened to NCR.
Typically NOx conversion ranges from 80 to 95% and there are correspond-
ing decreases in CO and hydrocarbon concentrations. Potential problems asso-
ciated with NSCR applications include catalyst poisoning by oil additives, such
as phosphorus and zinc, and inadequate control systems (37).
6.2. Carbon Monoxide. Carbon monoxide is emitted by gas turbine
power plants, reciprocating engines, and coal-fired boilers and heaters. Carbon
monoxide can be controlled by a precious-metal oxidation catalyst on a ceramic
or metal honeycomb. The catalyst promotes reaction of the gas with oxygen to
form CO2 at efficiencies that can exceed 95%. Carbon monoxide oxidation cata-
lyst technology is broadening to applications requiring better catalyst durability,
such as the combustion of heavy oil, coal (qv), municipal solid waste (qv), and
wood (qv). Research is underway to help cope with particulates and contami-
nants, such as flyash and lubricating oil, in gases generated by these fuels (1,66).
Carbon monoxide conversion is a function of both temperature and catalyst
volume, and increases rapidly beginning at just under 1008C until it reaches a
plateau at 1508C. But, unlike NOx catalysts, >1508C there is little benefit to
further increasing the temperature (64). Above 1508C, the CO conversion is
Vol. 2 EMISSION CONTROL, INDUSTRIAL 565

controlled by the bulk phase gas mass transfer of CO to the honeycomb surface.
That is, the catalyst is highly active, and its intrinsic CO removal rate is exceedingly
greater than the actual gas transport rate (36). When the activity falls to such an
extent that the conversion is no longer controlled by gas mass transfer, a decline of
CO conversion occurs, and a suitable regeneration technique is needed (36).
It has been reported that below 3708C, sulfur oxides reversibly inhibit CO
conversion activity. This inhibition is greater at lower temperatures. CO conver-
sion activity returns to normal shortly after removal of the sulfur from the
exhaust (64). Above 3158C, sulfur oxides react with the high surface area oxides
to disperse the precious-metal catalytic agents and irreversibly poison CO
conversion activity.
Catalyst contamination from sources such as turbine lubricant and boiler
feed water additives is usually much more severe than deactivation by sulfur
compounds in the turbine exhaust. Catalyst formulation can be adjusted to
improve poison tolerance, but no catalyst is immune to a contaminant that
coats its surface and prevents access of CO to the active sites. Between 1986
and 1990 >25 commercial CO oxidation catalyst systems operated on gas turbine
cogeneration systems, meeting both CO conversion (40–90%) and pressure drop
requirements.
6.3. Halogenated Hydrocarbons. Destruction of halogenated hydro-
carbons presents unique challenges to a catalytic oxidation system (65–72).
The first step in any control strategy for halogenated hydrocarbons is recovery
and recycling (65). However, even upon full implementation of economic recovery
steps, significant halocarbon emissions can remain. In other cases, halogenated
hydrocarbons are present as impurities in exhaust streams (65). Impurity
sources are often intermittent and dispersed.
The principal advantage of a catalytic oxidation system for halogenated
hydrocarbons is in operating cost savings. Catalytically stabilized combusters im-
prove the incineration conditions, but still must employ very high temperatures
as compared to VOC combustors; eg, carbon tetrachloride [56-23-5], CCl4, has a
40-fold lower heat of combustion than a typical organic vapor such as toluene
[108-88-3], thus CCl4 requires much more supplemental fuel to burn than do
typical organics (65). Alternatively, the low temperature catalytic oxidation pro-
cess is typically designed for a maximum adiabatic temperature rise of only
2008C. This would correspond to only 1500 ppm of an organic compound in
the exhaust stream. But, with the lower heat of combustion, up to 40,000 ppm
of carbon tetrachloride could be treated in the same temperature rise, or with
less dilution air.
By-Product Formation. The presence of halogenated hydrocarbons dra-
matically increases the yield of aldehydes from the oxidation process (65). For
example, in the partial oxidation of methane on a PdO sponge catalyst (34),
methylene chloride, CH2Cl2, was added in pulses to the inlet gas, which also
contained the methane. Methane oxidation was strongly inhibited and formal-
dehyde was formed. The formaldehyde production continued even after the
methylene chloride addition was stopped, suggesting a strong interaction of
chlorine with the catalyst. However, pulses of pure CH 4 plus oxygen gradually
restored the original activity to the catalyst, indicating that the effect of this
interaction was reversible.
566 EMISSION CONTROL, INDUSTRIAL Vol. 2

Catalyst Deactivation. Catalyst deactivation (65) by halogen degradation


is a very difficult problem particularly for platinum (PGM) catalysts, which make
up 75% of the catalysts used for VOC destruction (25). The problem may well lie
with the catalyst carrier or washcoat. Alumina, eg, a common washcoat, can
react with a chlorinated hydrocarbon in a gas stream to form aluminum chloride
which can then interact with the metal. Fluid-bed reactors have been used to off-
set catalyst deactivation but these are large and costly (65).
Catalytic Reaction. The desired reaction of the chlorine group on a
chlorinated hydrocarbon is

RCl þ O2 ! CO2 þ HCl þ R0 ð18Þ

It is important to produce HCl rather than elemental chlorine, Cl2, because HCl
can be easily scrubbed out of the exhaust stream, whereas Cl2 is very difficult to
scrub from the reactor off-gas. If the halogenated hydrocarbon is deficient in
hydrogen relative to that needed to produce HCl, low levels of water vapor
may be needed in the entering stream (65) and an optional water injector may
be utilized. For example, trichloroethylene [79-01-6], C2HCl3, and carbon tetra-
chloride require some water vapor as a source of hydrogen (65).

C2 HCl3 þ H2 O þ 1:5 O2 ! 2 CO2 þ 3 HCl ð19Þ

Groundwater contaminated with chlorinated hydrocarbons is being remedi-


ated by a conventional air stripper or a rotary stripper, producing an air stream
containing the halogenated hydrocarbon vapors and saturated with water vapor
(65), which is then passed through a catalyst bed.
At least two catalytic processes have been used to purify halogenated
streams. Both utilize fluidized beds of probably nonnoble metal catalyst particles.
One has been estimated to oxidize >9000 t/yr of chlorinated wastes from a vinyl
chloride monomer plant (65). Several companies have commercialized catalysts
which are reported to resist deactivation from a wider range of halogens. These
newer catalysts may allow the required operating temperatures to be reduced,
and still convert >95% of the halocarbon, such as trichlorethylene, from an
exhaust stream. Conversions of C1 chlorocarbons utilizing an Englehardt HDC
catalyst are shown in Figure 9. For this system, as the number of chlorine atoms
increases, the temperatures required for destruction decreases.

7. Uses

Catalytic oxidation of exhaust streams is increasingly used in those industries


involved in the following (28,30,32):

Surface coatings: aerospace, automobile, auto refinishing, can coating, coil


coating, fabric coating, large appliances, marine vessels, metal furniture,
paper coating, plastic parts coating, wire coating and enameling, wood
furniture.
Printing inks: flexographic, lithographic, rotogravure, screen printing.
Vol. 2 EMISSION CONTROL, INDUSTRIAL 567

100
90

Conversion of chlorocarbons, %
80 CCl4 CHCl3 CH2Cl2
70
60
50
CH3Cl
40
30
20
10
0
100 150 200 250 300 350 400 450 500 550
Inlet temperature, °C

Fig. 9. Destruction of C1 chlorocarbons over HDC in the presence of 1.5% H2O in air at
15,000 h1 GHSV at STP. Chlorocarbon concentrations are CCl4, 900 ppmv; CHCl3,
500 ppmv; CH2Cl2, 800 ppmv; and CH3Cl, 600 ppmv.

Solvent usage: adhesives, disk manufacture, dry cleaning, fiber glass manu-
facture, food tobacco manufacture, metal cleaning, pharmaceutical, photo
finishing labs, semiconductor manufacture.
Chemical and petroleum processes: cumene manufacture, ethylene oxide
manufacture, acrylonitrile manufacture, caprolactam manufacture, maleic
anhydride manufacture, monomer venting, phthalic anhydride manufac-
ture, paint and ink manufacture, petroleum product refining, petroleum
marketing, resin manufacture, textile processing.
Industrial/commercial processes: aircraft manufacture, asceptic packaging,
asphalt blowing, automotive parts manufacture, breweries/wineries, carbon
fiber manufacture, catalyst regeneration, coffee roasting, commercial char-
broiling, electronics manufacture, film coating, filter paper processing, food
deep frying, gas purification, glove manufacture, hospital sterilizers, pea-
nut and coffee roasting, plywood manufacture, rubber processing, spray
painting, tire manufacture, wood treating.
Engines: diesel engines, lean burn internal combustion, natural gas compres-
sors, oil field steam generation, rich burn internal combustion, gas turbine
power generation.
Cross media transfer: air stripping, groundwater cleanup, soil remediation
(landfills), hazardous waste treatment, odor removal from sewage gases.

The most important factors affecting performance are operating temperature,


surface velocity, contaminant concentration and composition, catalyst proper-
ties, and the presence or absence of poisons or inhibitors.
7.1. Air Stripping of Groundwater. Treatment of exhaust streams
from the air stripping of contaminated groundwater is a particular challenge,
because the emissions from air stripping units may consist of a complex mixture
568 EMISSION CONTROL, INDUSTRIAL Vol. 2

of both fuel and solvent fractions (13). The catalytic oxidation of any given com-
pound is generally negatively impacted by the presence of others in mixtures,
and higher catalyst bed operating temperatures are required to achieve adequate
destruction.
Some catalysts exposed to air stripping off-gas were subject to deactivation.
However, using a catalytic oxidizer at a U.S. Coast Guard facility (Traverse City,
Mich.) for the destruction of benzene, toluene, and xylene stripped from the
groundwater, the catalytic oxidization unit operated at 260–3158C, and was
able to achieve 90% destruction efficiency.
7.2. Printing and Graphic Arts. In the graphic arts industry, the cata-
lyst in the oxidizer needs to be monitored regularly because it is susceptible
to contamination by phosphorus from fountain solutions, silica from silicone
gloss enhancer sprays, and chlorides from chlorinated solvents or blanket
wash solutions. Phosphorus and silica accumulate most rapidly on the leading
edge of the catalyst bed, deactivating the catalyst by masking the precious
metals. In a fluidized-bed configuration, the catalyst surface is continually
renewed by abrasion and the problem of masking the catalyst surface with sili-
cones is avoided.
7.3. Chemical Processing. Terephthalic Acid Production. The con-
trol of exhaust from production of pure terephthalic acid (PTA) has been a
challenge (see CARBOXYLIC ACIDS SURVEY) (73). Eight million metric tons of PTA
are produced annually worldwide for use primarily in high grade polyester
fiber production. Based on a para-xylene feedstock, vent gases from the process
contain such by-products as methyl acetate, organic acids, and often methyl bro-
mide. These exhausts have been estimated to total 34,000-m3 worldwide. Histori-
cally, the presence of the methyl bromide limited the use of fixed-bed catalytic
oxidation as a control technology using precious-metal catalysts. Thus base
metal catalysts in fluidized-bed reactors have been the primary catalytic technol-
ogy of choice. In this application, the continuous abrasion of the outer layer of the
catalyst particles exposes a fresh surface of unpoisoned material to the reactants,
allowing the catalyst to effectively treat the exhaust stream.
In the late 1980s, however, the discovery of a noble metal catalyst that
could tolerate and destroy halogenated hydrocarbons such as methyl bromide
in a fixed-bed system was reported (73,74). The products of the reaction were
water, carbon dioxide, hydrogen bromide, and bromine. Generally, a scrubber
would be needed to prevent downstream equipment corrosion. However, if the
focus of the control is the VOCs and the CO rather than the methyl bromide, a
modified catalyst formulation can be used that is able to tolerate the methyl bro-
mide, but not destroy it. In this case the methyl bromide passes through the bed
unaffected, and designing the system to avoid downstream effects is not neces-
sary. Destruction efficiencies of hydrocarbons and CO of better than 95% have
been reported, and methyl bromide destructions between 0 and 85% (73).
Latex Monomer Production. ARI Technologies, Inc. has introduced a cat-
alyst system which, it is claimed, can operate at an average bed temperature of
3708C while achieving conversion efficiency in excess of 99.99% on exhaust
streams from latex monomer production.
Acrylonitrile Manufacture. In the manufacture of acrylonitrile (qv), off-
gases containing from 1–3% of CO plus various hydrocarbons are emitted.
Vol. 2 EMISSION CONTROL, INDUSTRIAL 569

Catalytic beds of platinum-group metals are used to reduce the regulated com-
pounds to acceptable levels. Close attention to bed design is required to prevent
the formation of appreciable quantities of NOx caused by the fixation of combus-
tion-air nitrogen. Some NOx also is produced from fuel nitrogen by oxidation.
Because of the high thermal energy content of the off-gases, considerable heat
recovery is possible in abating acrylonitrile plant emissions (see HEAT-EXCHANGE
TECHNOLOGY).
Vinyl Monomer Manufacturing. Process vent gases containing small
quantities of halogenated hydrocarbons and substantial quantities of nonhaloge-
nated hydrocarbons have been successfully reduced to comply with regulatory
objectives in large-scale laboratory–pilot catalytic fume abaters having satisfac-
tory long-term catalyst performance. The design freedoms offered by precious
metals on ceramic honeycomb support catalysts have been demonstrated in
equipment that utilizes the heat energy resulting from the substantial exotherm
of the nonhalogenated hydrocarbon oxidation to preheat the exhaust gases. Fuel
consumption is thereby minimized.
7.4. Coatings Industries. Surface coating processes produce similar
air pollution problems in a number of different industries.
Can Manufacturing. An internal coating is necessary to protect the
purity and flavor of can contents for beverages or any edible product that
might react with the container metal. Both the exterior decorative and interior
sanitary coatings are applied to the metal surface by rolls or spray guns using a
solvent vehicle. Catalytic oxidation systems are used by the principal can
manufacturers to treat coatings exhaust streams. The can manufacturers’ indus-
try is estimated to utilize more catalysts than any of the other surface coating
industries.
A large number of diverse solvents are used in exterior and interior coatings
in plants for manufacturing both three-and two-piece cans. Most of the organic
solvents are found in the cure-oven exhausts at concentrations of 2–16% of the
lower explosive limit (LEL). The oven exhaust volumes are usually 1–35 m3/s.
When burned, these concentrations of combustibles provide an exotherm of
30–2208C. The heat that is released is used for preheating the incoming effluent
and/or heating the cure oven by recycling the hot, cleaned gases to the supply
blowers or by heating makeup air by heat exchange. A few plants use the heat
of the cleaned exhaust to produce hot water for the two-piece can line washers,
hot air for dry-off ovens, or building space heating. For example, one large can
company utilizes the heat energy contained in the stream, leaving some of
their catalytic fume abaters to supply all the heat energy required by the
oven’s heating zones, which have no burners. The fuel energy supplied to the catalytic
fume abater is less than would be needed to heat the oven if the solvent fumes were
exhausted directly to the atmosphere without use of the fume abater. The exhaust rate
of the oven is adjusted to maintain a solvent concentration of at least 8% of the LEL,
equivalent to a 1108C temperature differential.
The various reaction rate properties of the different solvents influence the
design of a catalytic reactor. For example, for a specific catalyst bed design, an
effluent stream containing a preponderance of monohydric alcohols, aromatic
hydrocarbons, or propylene requires a lower catalyst operating temperature
than that required for solvents such as isophorone and short-chain acetates.
570 EMISSION CONTROL, INDUSTRIAL Vol. 2

Design considerations and costs of the catalyst, hardware, and a fume


control system are directly proportional to the oven exhaust volume. The size
of the catalyst bed often ranges from 1.0 m3 at 08C and 101 kPa/ 1000 m3/min
of exhaust, to 2 m3 for 1000 m3/min of exhaust. Catalyst performance at a
number of can plant installations has been enhanced by proper maintenance.
Annual analytical measurements show reduction of solvent hydrocarbons
to be in excess of 90% for 3–6 years, the equivalent of 12,000–30,000 operating
hours. When propane was the only available fuel, the catalyst cost was recov-
ered by fuel savings (vs thermal incineration prior to the catalyst retrofit) in
two to three months. In numerous cases the fuel savings paid for the catalyst
in 6–12 months.
Can manufacturers often regenerate the catalyst beds on an annual or
biannual basis during a weekend downtime. Both air lancing and an aqueous
bath are utilized to remove noncombustible particulates that mask the active
sites. Frequently, condensed organic material on the catalyst is removed by
short-term (4–6 h) heating excursions to 370 or 4308C; the organic matter is
removed much like a self-cleaning oven. The gaseous and organic smoke,
which is usually evolved from the first few cm of catalyst bed depth, is oxidized
in the latter part of the bed. If allowed to operate too long at temperatures that
promote condensation, high boiling organic compounds, the subsequent carbon
char that is formed, may require temperatures of 480–5408C to convert the car-
bon to carbon monoxide for subsequent oxidation. The higher temperatures
required for burn-offs should be approached in small (0–308C) increments to
bring about slow evolution and partial oxidation; this prevents autogenous com-
bustion of local high concentrations of combustible material.
Day-to-day operating techniques that are employed by one large can man-
ufacturer and are intended to prevent organic condensation are dictated by the
use of a low cost, well-established, sanitary coating for beer and beverage three-
piece cans. Polybutadiene and other sanitary coatings may have volatile resin
monomers entrained in the oven atmosphere as a result of rapid
evaporation of solvent before polymerization takes place. A short (4–6 h) heating
excursion up to a catalyst inlet temperature of 4008C after use of the coating
usually burns off any condensed organic materials. It has become standard prac-
tice in some plants to turn the catalytic afterburner up to 3708C for these coat-
ings versus the normal 3158C operating temperature for vinyls, acrylics, etc.
Coil Coating. Coil coating is the prefinishing of many sheet metal items
with protective and decorative coatings that are applied by roll coating on one or
both sides of a fast-moving metal strip. The metal strip (from 13 mm to 1.7 m in
width) unwinding from a coil travels at rates of 30–150 m/min through the coat-
ing applicator rolls and bake ovens. It is rewound into a coil for transport to a
forming operation for products that are to be used in cans, appliances, industrial
and residential siding, shelving, cars, gutters, downspouts, etc. The source of
hydrocarbon emissions in coil coating is the coating application area and the
cure-oven exhaust. The coatings include primers, finishes, and metal protective
(5 mm) films or backers.
The increasing use of siliconized coatings for weather durability caused
severe masking problems for the all-metal, filter mesh-like catalyst elements
available in the 1970s. Interest in catalytic afterburners increased when
Vol. 2 EMISSION CONTROL, INDUSTRIAL 571

dispersed-phase precious metal-alumina-on-ceramic honeycomb catalysts offered


economically attractive results.
The hot (260–3708C) oven exhaust can be oxidized catalytically without pre-
heat, but when the coater-area exhaust (at room temperature) is combined with the
oven exhaust, a preheat burner becomes necessary. The greatest energy savings
potential having the least capital investment is obtained by recycling a portion of
the hot, cleaned exhaust to the oven. This principle has been demonstrated at a
number of can manufacturing plants and at least four coil coating facilities. One
operator preheats oven make-up air that has been taken from the oven cooler sec-
tion by means of a heat exchanger; whereas others recycle directly to the oven. In
one case, the heat energy for the dry-off oven is supplied by the catalytic incinerator
exhaust (4828C) remaining after supplying most of the heat energy to operate the
four zones in the oven. The concentration of solvents in the exhaust is 12% LEL
(1678C temperature differential). The net fuel energy consumption is 20% of that
required to fire the paint, bake, and dry-off ovens without fume control.
Coil coaters operate equipment continuously and, in most cases, operate cat-
alytic fume abaters 6000–7000 h/year. Under these conditions the anticipa-
ted catalyst life is years, with an annual aqueous solution cleaning. However,
the catalyst may last no more than two years if frequent maintenance is needed,
such as in-place air lancing every 60–90 days to remove noncombustible parti-
culates. Frequent maintenance may be needed if coatings such as siliconized
polyester (15–40% silicones) comprise 30% of the coatings put through the system.
Filter Paper Processing. In the fabrication of fuel oil and air filters for
vehicles such as motorcycles and diesel locomotives, heat processing of the filter
paper is required to cure the resin (usually phenolic) with which the paper (qv)
is impregnated. The cure-oven exhaust, which contains water vapor, alcohols,
and dimers and trimers of phenol, produces a typical blue haze aerosol having
a pungent odor. The concentration of organic substances in the exhaust is
usually rather low.
The paper-impregnation drying oven exhausts contain high concentrations
(10–20% LEL) of alcohols and some resin monomer. Vinyl resins and melamine
resins, which sometimes also contain organic phosphate fire retardants, may be
used for air filters. The organic phosphates could shorten catalyst life depending
on the mechanism of reduction of catalyst activity. Mild acid leaching removes
iron and phosphorus from partially deactivated catalyst and has restored activity
in at least one known case.
Catalysis is utilized in the majority of new paper filter cure ovens as part of
the oven recirculation/burner system that is designed to keep the oven interior
free of condensed resins and provide an exhaust without opacity or odor. The
application of catalytic fume control to the exhaust of paper-impregnation dryers
permits a net fuel saving by oxidation of easy-to-burn methyl or isopropyl alcohol,
or both, at adequate concentrations to achieve a 110–2208C exotherm.

8. Environmental Technology Verification

In 1995, the U. S. Environmental Protection Agency (EPA) established the


Environmental Technology Verification (ETV) Program. This program was
572 EMISSION CONTROL, INDUSTRIAL Vol. 2

established to evaluate the truth of a commercially ready environmental tech-


nology’s performance under specific, predetermined criteria and adequate
quality assurance (QA) procedures. As of 2003, this program has five opera-
tional centers to evaluate the technologies organized around the critical tech-
nologies being reviewed (including greenhouse gas technology, air
pollution control technology and pollution prevention processes). By mid
2003, >200 technologies were verified under the program with 140 applica-
tions for testing pending (75).
With active EPA participation, both vendors and users will be able to offer
or select control equipment that would best fit their applications with a high
degree of confidence.

BIBLIOGRAPHY

‘‘Exhaust Control, Industrial’’ in ECT 3rd ed., Vol. 9, pp. 511–541, by J. H. Conklin,
D. M. Sowards, and J. H. Kroehling, E. I. du Pont de Nemours & Co., Inc.; in ECT
4th ed., Vol. 9, pp. 1022–1060, by Ronald L. Berglund, The M. W. Kellogg Company;
‘‘Exhaust Control, Industrial’’ in ECT (online), posting date: December 4, 2000, by Ronald
L. Berglund, The M. W. Kellogg Company.

CITED PUBLICATIONS

1. R. J. Farrauto, R. M. Heck, and B. K. Speronnelo, C & EN, 34–44 (Sept. 7, 1992).


2. C. C. Lewis, CPI Purch. 29–33 (Aug. 1991).
3. J. C. Summers, J. E. Sawyer, and A. C. Frost, in R. G. Silver, J. E. Sawyer, and J. C.
Summers, eds., ‘‘The 1990 Clean Air Act and Catalytic Emission Control Technology
for Stationary Sources,’’ Catalytic Control of Air Pollution: Mobile and Stationary
Sources, ACS Symposium Series 495, 1992.
4. R. L. Berglund and C. T. Lawson, Chem. Eng. (Sept. 1991).
5. Catalytic Control of VOC Emissions, A Cost Effective Viable Technology for Indus-
trial, Commercial and Waste Processing Facilities, Manufacturers of Emission Con-
trols Association, Washington, D.C., 1992.
6. R. L. Berglund, D. R. DuRivage, and M. R. Hebert, Revisiting the LDAR Programs:
Preparing For the Long Haul, Paper 96-FA148.01, Proceedings of the 89th AWMA
Meeting, Nashville, TN, June 23–28, 1996.
7. Colman, Russel and Silvio Stojic Managing Fugitives—Capturing Value. Paper pre-
sented at CASANZ Conference—CASNO3, Newcastle, November 2003.
8. ‘‘Oil Refineries Fail to Report Millions of Pounds of Harmful Emissions’’, Prepared for
Representative Henry A. Waxman, Minority Staff, Special Investigations Division,
Committee on Government Reform, U.S. House of Representatives, November 10,
1999.
9. United States General Accounting Office, Air Pollution, EPA Should Imporve Over-
sight of emissions Reporting by Large Facilities. Report GAO-01-46, April 2001.
10. ‘‘Protocol for Equipment Leak Emission Estimates’’, U.S. Environmental Protection
Agency, EPA-453/R-95-017, November 1995.
11. ‘‘Preferred and Alternative Methods for Estimating Fugitive Emissions From Equip-
ment Leaks’’, Emission Inventory Improvement Program, Volume II, Chapter 4,
November 1996.
Vol. 2 EMISSION CONTROL, INDUSTRIAL 573

12. R. L., Berglund, Maintenance Operations and Pollution Prevention’’, Chapter 33 in


Industrial Pollution Prevention Handbook, H. F. Freeman, ed., 1995, McGraw-Hill,
Inc. New York.
13. M. Kosusko and C. M. Nunez, JAWMA, 40(2) (Feb. 1990).
14. Serageldin M., The Effect of Legislation and Operational Knowledge on the Bioreac-
tor Market, Paper #70319 presented at the 2003 Annual Meeting of the Air and Waste
Management Association, San Diego.
15. United States Environmental Protection Agency, Using Bioreactors to Control Air
Pollution, EPA-456/R-03-003, September 2003.
16. Biofiltration: Project Report and Scale-up and Design Guide, Center of Waste Reduc-
tion Technologies of the American Institute of Chemical Engineers, New York, 1990;
pp. 163–175.
17. M. Gueissaz-Teufel and P. Wolstenholme, Biofiltration Design, Operational Details,
and Case Studies, Paper #69677, presented at the 2003 Annual Meeting of the Air
and Waste Management Association, San Diego.
18. J. Berry, Engineered Bioreactors, Dependable and Affordable Oxidation, Paper
#69412, Presented at 2003 Annual Meeting of the Air and Waste Management Asso-
ciation, San Diego, CA.
19. Hsu, Shu-kang and co-workers, Operating and Maintaining Experiences of Several
Biofilters in Taiwan Paper #69556, presented at the 2003 Annual Meeting of the
Air and Waste Management Association, San Diego.
20. J. S. Devinny, ‘‘Clearing the Air Biologically’’, Civil Engineering Magazine, Septem-
ber 1998, pp. 46–49.
21. F. J. Martin and R. C. Loehr, J. Air Waste Manage. Assoc. 46, 539 (1996).
22. D. R. van der Vaart, W. M. Vatvuk, and A. H. Wehe, JAWMA 41(1), 92 (Jan. 1991).
23. L. C. Hardison and E. J. Dowd, Chem. Eng. Prog. 31–35 (Aug. 1977).
24. R. J. Martin, R. E. Smyth, and J. T. Schofield, ‘‘Elimination of Petroleum Industry Air
Toxic Emissions with a Flameless Thermal Oxidizer’’, Petro-Safe ’93, Houston, Tex.,
Jan. 29, 1993.
25. G. Parkinson, Chem. Eng. 37–43 (July, 1991).
26. A. F. Hodel, Chem Proc. 88–90 (June 1992).
27. R. Yarrington and L. Morris, Indust. Finish. (Mar. 1992).
28. J. Chen, R. M. Heck, and R. J. Farraoto Catal. Today 11, 517 (1992).
29. M. S. Jennings, N. E. Krohn, and R. S. Berry, Control of Industrial VOC Emissions by
Catalytic Incineration, Vol. 1, U.S. Environmental Protection Agency, Research
Triangle Park, N.C., July 1984.
30. E. J. Dowd, W. M. Sheffer, and G. E. Addison, ‘‘A Historical Perspective on the Future
of Catalytic Oxidation of VOCs’’, paper 92-109.03, 85th Annual Meeting of Air and
Waste Management Association, Kansas City, Mo., June 21–26, 1993.
31. D. M. VanBenshchoten, ‘‘On-Site Pilot Testing Demonstrates Catalytic Emission
Control Technology for New VOC Applications’’, paper 92-109.01, presented at 85th
Annual AWMA Meeting & Exhibition, Kansas City, Mo., June 21–26, 1992.
32. J. J. Spivey, Ind. Eng. Chem. Res. 26, 2165 (1987).
33. K. R. Bruns, ‘‘Use of Catalysts for VOC Control’’, presented at the New England
Environmental Expo, Boston, Mass., Apr. 10–12, 1990.
34. J. J. Spivey and J. B. Butt, Catal. Today 11, 465 (1992).
35. R. M. Heck, M. Durilla, A. G. Bouney, and J. M. Chen ‘‘Air Pollution Control—Ten
Years Operating Experience with Commercial Catalyst Regeneration’’, paper pre-
sented at the 81st APCA Annual Meeting & Exhibition, Dallas, Tex., June 19, 1988.
36. R. M. Heck, J. M. Chen, and M. F. Collins ‘‘Oxidation Catalyst for Cogeneration
Applications—Regeneration of Commercial Catalyst’’, paper 90-105.1, presented at
the 83rd Annual AWMA Meeting & Exhibition, Pittsburgh, Pa., June 24–29, 1990.
574 EMISSION CONTROL, INDUSTRIAL Vol. 2

37. R. E. Kenson, ‘‘Control of Volatile Organic Emissions’’, Bulletin 1015, Series 1000,
Met-Pro Corp., 1981.
38. W. L. Prins and Z. L. Nuninga, Catal. Today 16, 187–105 (1993).
39. C. H. Bartholomew, Chem. Eng. 96 (Nov. 12, 1984).
40. A. C. Frost and co-workers, Environ. Sci. Technol. 25(12), 2065 (1991).
41. D. Ciccilella and B. Holt Environ. Protec., 41 (Sept. 1992).
42. K. J. Herbert, ‘‘Catalysts for Volatile Organic Control in the 1990’s’’, presented at the
1990 Incineration Conference, San Diego, Calif., May 14–18, 1990.
43. M. A. Palazzolo, J. I. Steinmetz, D. L. Lewis, and J. F. Beltz ‘‘Parametric Evaluation
of VOC/HAP Destruction via Catalytic Incineration’’, U.S. Environmental Protection
Agency, Report no. EPA/600/S2-85/041, Research Triangle Park, N.C., July 1985.
44. S. K. Gangwal, M. E. Mulling, J. J. Spivey, and P. R. Caffrey, Appl. Catal. 36, 231 (1988).
45. M. A. Palazzolo, and C. L. Jamgonhian, ‘‘Destruction of Chlorinated Hydrocarbons by
Catalytic oxidation’’, U.S. Environmental Protection Agency, Report EPA/600/82-86/
079, Research Triangle Park, N.C., Jan. 1987.
46. Difford and Spensor, Chem. Eng. Prog. 71, 31 (1975).
47. G. R. Lester, ‘‘Catalytic Destruction of Hazardous Halogenated Organic Chemicals’’,
presented at the 82nd Annual AWMA Meeting & Exhibition, Anaheim, Calif., June
25–30, 1992.
48. S. E. Wanke and P. C. Flynn, Cat. Rev. Sci. Eng. 12, 93 (1975).
49. G. R. Lester and J. C. Summers ‘‘Poison-Resistant Catalyst for Purification of Web
Offset Press Exhaust’’, presented at the Air Pollution Control Association, 81st
Annual Meeting, Dallas, Tex., June 19–24, 1988.
50. R. G. Nicolli, ‘‘Overview of NOx Chemistry and Emissions Control Technology’’, pre-
sented at 16th AESF/EPA Pollution Prevention and Control Conference, Feb 1995.
51. U.S. Environmental Protection Agency, Alternative Control Techniques Document—
NOx Emisisons from Industrial/Commercial/Institutional (ICI) Boilers, Office of Air
Quality Planning and Standards, March 1994 (EPA-453/R-94-022).
52. F. Luck and J. Roiron, Catal. Today 4, 205 (1989).
53. E. Cichanowicz, Power Eng., 36 (Aug. 1988).
54. T. A. Melnick, R. Sommerlad, R. Payne, T. Sommer, ‘‘Low NOx Burner Modifications
for Cost Effective NOx Control’’, Forum 2000 on Cutting NOx Emissions, Institute of
Clean Air Companies, March 2000 (Arlington, VA).
55. L. M. Campbell, D. K. Stone, and G. S. Shareef, Sourcebook: NOx Control Technology
Data, EPA Report NO. EPA600/S2-91/029., Washington, D.C., Aug. 1991.
56. G. S. Shareef, D. K. Stone, K. R. Ferry, K. L. Johnson, and K. S. Locke ‘‘Selective
Catalytic Reduction NOx Control for Small Natural Gas-Fired Prime Movers’’, paper
92-136.06, in Ref. 16.
57. J. R. Donnelly and B. Brown ‘‘Joy/Kawasaki Selective Reduction DE-NOX Technol-
ogy’’, paper No 89-96B.6, in Ref. 32.
58. H. Gutberlet and B. Schallert, Catal. Today 16, 207 (1993).
59. R. D. Walloch, Oil Gas J., 39 (June 13, 1988).
60. R. Craig, G. Robinson, and P. Hatfield, ‘‘Performance of High Temperature SCR
Catalyst System at Unocal Science and Technology Division’’, paper 92-109-08, in
Ref. 15.
61. D. Borsare, R. Hanson, E. Angelini, R. Peterson, and J. Buschmann, ‘‘Recent Experi-
ences in SCR System Design’’, Forum 2000 on Cutting NOx Emissions, Institute of
Clean Air Companies, March, 2000 (Arlington, VA).
62. D. L. Champagne, Gas Turb. World, 20 (Nov.–Dec. 1987).
63. R. Rodenhausen, ‘‘Case Study: Choosing Selective Catalytic Reduction as a Preferred
Technology for the Destruction of NOx’’, Environmental Progress, Winter 1999,
pp. 260–266.
Vol. 2 EMISSION CONTROL, INDUSTRIAL 575

64. B. K. Speronello, J. M. Chen, and R. M. Heck ‘‘Family of Versatile Catalyst Technol-


ogies for NOx and CO Removal in Co-Generation’’, paper 92-109.06, in Ref. 16.
65. D. W. Agar and W. Ruppel, ‘‘Extended Reactor Concept for Dynamic DeNOx Design’’,
Chem. Eng. Sci. 43(8), (1988); J. R. Kittrell, C. W. Quinian, and J. W. Eldridge J. Air
Waste Manage. Assoc. 41(8), 1129 (Aug. 1991).
66. C. B. Sedman, ‘‘Controlling Emissions from Fuel and Waste Combustion’’, Chemical
Engineering, Vol. 106, No.1, January 1999, pp. 82–88.
67. T. D. Hylton, Environ. Prog. 11(1), 54 (Feb. 1992).
68. Y. Wang, H. Shaw, and R. J. Farrauto, ‘‘Catalytic Oxidation of Trace Concentrations
of Trichloroethylene over 1.5% Platinum on alpha-Alumina’’, in Ref. 3.
69. T. C. Yu, H. Shaw, and R. J. Farrauto, ‘‘Catalytic Oxidation of Trichloroethylene over
PdO Catalyst on Alpha Al2O3’’, in Ref. 3.
70. J. L. Lin and B. E. Bent, ‘‘Thermal Decomposition of Halogenated Hydrocarbons on a
Cu (111) Surface’’, in Ref. 3.
71. S. L. Hung and L. D. Pfefferie Environ. Sci. Technol. 23(9), 1085 (1989).
72. M. Mirghanbari, D. J. Muno, and J. A. Bacchetti, Chem. Proc. 45, 14 (Dec. 1982).
73. T. G. Otchy and K. J. Herbert ‘‘First Large Scale Catalytic Oxidation System for PTA
Plant CO and VOC Abatement’’, paper presented at the 85th Annual Meeting &
Exhibition, AWMA, June 1992, p. 2026.
74. K. J. Herbert, ‘‘Catalytic Oxidation of the Vent Gas from a PTA Plant’’, 1992.
75. G. Theodore, Environmental Technology Verification: A Vehicle for Independent,
Credible Performance Results on Commercially Ready Technologies, Paper #71006
presented at 2003 Annual Meeting of the Air and Waste Management Association,
San Diego, Calif.

RONALD L. BERGLUND
Terracon
PART III:
PETROCHEMICALS

PETROCHEMICAL
FEEDSTOCKS
1. Introduction

By definition, petrochemicals are either isolated from or derived from natural


gas or petroleum. Other sources of feedstocks, such as coal and agricultural
products, remain very small sources of petrochemicals. The choice of feedstock
for a given operator is an economic decision, but may be constrained by hard-
ware limitations on the part of the operator or the availability of indigenous
hydrocarbon supplies in a given geographical region. Because the primary
uses of natural gas and petroleum are as sources of energy, the costs of petro-
chemical feedstocks are closely related to the alternative energy values of the
various feedstocks. The petrochemical producer must buy feedstocks out of the
alternative energy markets. Hence, the structure and economics of the energy
industry in a given area affect the choice of petrochemical feedstocks. Seasonal
effects in the prices of energy products introduce variability in petrochemical
feedstock prices and affect the choice of preferred feedstock for a given opera-
tor at a point in time. International events such as the OPEC oil embargo
of 1973– 1974, the Iranian revolution of 1979, the Middle East war of
1990 –1991, or the war in Iraq during 2003 have dramatically affected the
costs and choice of petrochemical feedstocks for a period of time. Government
regulations, such as the U.S. Clean Air Act Amendments of 1990 and major
programs in the U.S. and Brazil to promote the use of ethanol as a vehicle

576

Kirk-Othmer Encyclopedia of Chemical Technology. Copyright John Wiley & Sons, Inc. All rights reserved.
10.1002/0471238961.1605201801210218.a01.pub2
Vol. 2 PETROCHEMICAL FEEDSTOCKS 577

fuel, also impact petrochemical feedstock availability, relative cost, and choice
of alternative feedstocks.
It is convenient to divide the petrochemical industry into two general sec-
tors: (1) olefins and (2) aromatics and their respective derivatives. Olefins are
straight- or branched-chain unsaturated hydrocarbons, the most important
being ethylene (qv), [74-85-1] propylene (qv) [115-07-1], and butadiene (qv)
[106-99-0]. Aromatics are cyclic unsaturated hydrocarbons, the most important
being benzene (qv) [71-43-2], toluene (qv) [108-88-3], p-xylene [106-42-3], and
o-xylene [95-47-5] (see XYLENES). There are two other large-volume petro-
chemicals that do not fall easily into either of these two categories: ammonia
[7664-41-7] and methanol [67-56-1]. These two products are derived primarily
from methane [74-82-8] (natural gas).
Olefins are produced primarily by thermal cracking of a hydrocarbon feed-
stock that takes place at low residence time in the presence of steam in the tubes
of a furnace. In the United States, natural gas liquids derived from natural gas
processing, primarily ethane [74-84-0] and propane [74-98-6], have been the
dominant feedstock for olefins plants, accounting for 60–70% of ethylene pro-
duction. Most of the remainder has been based on cracking naphtha or gas oil
hydrocarbon streams that are derived from crude oil. Naphtha is a hydrocarbon
fraction boiling between 40 and 1708C, whereas the gas oil fraction boils
between 310 and 4908C. These feedstocks, which have been used primarily
by producers with refinery affiliations, account for most of the remainder of
olefins production. In addition, a substantial amount of propylene and a small
amount of ethylene are recovered from gas streams produced in petroleum
refinery cracking units.
In most of the rest of the world, the olefins industry was originally based on
naphtha feedstocks. Naphtha is the dominant olefins feedstock in Europe and
Asia. In the mid 1980s, several large olefins complexes were built outside the
United States primarily based on ethane feedstocks, most notable in western
Canada, Saudi Arabia, and Scotland. During the 1990s, new large olefins com-
plexes based on ethane feedstocks were also built in Kuwait, Qatar, and Malaysia.
In each case, the driving force was the production of natural gas, perhaps asso-
ciated with crude oil production, which was in excess of energy demands.
Recently, several large olefins complexes have been designed based on the
use of condensates, a heavier natural gas liquid. Condensates are mixtures of
liquid hydrocarbons, typically mixtures of pentane (C-5) and heavier compo-
nents, which are recovered from natural gas either at the wellhead or in gas pro-
cessing plants. Condensates usually are classified as either ‘‘light’’ or ‘‘heavy,’’
according to their boiling range, and light condensates are sometimes classified
as ‘‘natural gasolines’’. The typical properties of light condensates— >65% par-
affins content and a boiling range of 30–1808C—make them desirable feedstocks
for steam cracking to produce olefins.
Heavy condensates have a boiling range between 30 and 4008C and have
characteristics of light crude oils. Like light crudes, heavy condensates contain
naphtha, kerosene, gas oil, and minor quantities of heavier fractions, which
makes them attractive feedstocks for refinery primary distillation, as well as
olefins and aromatics production. A large U.S. Gulf Coast olefins complex
based on naphtha feedstocks derived from heavy condensate streams started
578 PETROCHEMICAL FEEDSTOCKS Vol. 2

up in 2001 (1) and another world-scale olefins complex based on heavy conden-
sate feedstocks is now under construction in China (2).
Since the early 1980s many olefin plants in the United States were designed
to have substantial flexibility to consume a wide range of feedstocks. Most of the
flexibility to use various feedstocks is found in plants with associated refineries,
where integrated olefins plants can optimize feedstocks using either gas liquids
or heavier refinery streams. Companies whose primary business is the produc-
tion of ethylene derivatives, such as thermoplastics, tend to use ethane and pro-
pane feedstocks that minimize by-product streams and maximize ethylene
production for their derivative plants.
Flexibility allows the operator to pick and choose the most attractive feed-
stock available at a given point in time. The steam-cracking process produces not
only ethylene, but other products as well, such as propylene, butadiene, buty-
lenes (a mixture of monounsaturated C-4 hydrocarbons), aromatics, etc. With
ethane feedstock, only minimal quantities of other products are produced. As
the feedstocks become heavier (ie, as measured by higher molecular weights
and boiling points), increasing quantities of other products are produced. The
values of these other coproduced products affect the economic attractiveness,
and hence the choice of feedstock.
Aromatics are produced primarily from two sources. The most important in
the United States is refinery catalytic reformer operations. Catalytic reformers
are high temperature catalytic dehydrogenation processes that convert
naphthenes into aromatics in order to increase the octane level, and hence the
gasoline blending quality of the stream being processed. Reformers produce
large quantities of the primary aromatic chemicals. Benzene, toluene, and a
mixed-xylene stream are subsequently recovered by extraction or extractive
distillation using a solvent (see BTX PROCESSING). Recovery of p-xylene from a
mixed-xylene stream requires a further processing step of crystallization and fil-
tration or adsorption followed by desorption on beds of molecular sieves.
The second source of aromatics is pyrolysis gasoline from olefins plants.
Pyrolysis gasoline is one of the by-product streams produced in olefin plants
when cracking heavier feedstocks such as naphtha or gas oil. Pyrolysis gasoline
contains hydrocarbon fractions from C-5 olefins through C-9 aromatics. Recovery
of aromatics from this stream is by extraction and fractionation, just as in recov-
ery from reformate. Pyrolysis gasoline is a more important aromatics feedstock
in Europe and Asia than in the United States. A reason is that naphtha is the
dominant feedstock for olefins production in these regions of the world, and
less gasoline is produced in the refineries of these regions; hence, there is less
reforming capacity. Otherwise there are only minimal differences in aromatics
production in various regions of the world.

2. Olefin Feedstocks

Olefins are produced primarily by steam cracking of hydrocarbon feedstock.


Steam cracking is a thermal cracking process in the presence of steam that
takes place at low residence times in the tubes of a furnace. The thermal cracking
process was originated in the early 1920s and has been the dominant route to
Vol. 2 PETROCHEMICAL FEEDSTOCKS 579

olefins since that time (3). Olefin plants are complex and costly, with both com-
plexity and cost increasing as the design feedstock increases in heaviness (mole-
cular weight) from the lightest feed (ethane) to the heaviest feed (gas oil).
Heavier feeds produce a wider array and larger quantities of products that
require more complex separation and recovery processes. A new world-scale
ethylene plant with a capacity of 1000 106 kg/year based on naphtha feedstock
will cost $750 million (4). The total plant cost, including off-site facilities, is
$1.1 billion. Originally olefins plants were designed to consume a specific feed-
stock, such as ethane or naphtha, or a narrow range of feedstocks, most com-
monly ethane–propane mixtures. Starting 1980, producers in the United
States that were based on heavier feedstocks such as naphtha or gas oil began
to introduce feedstock flexibility to consume a wide range of feedstocks in their
olefin plants. This was done to allow the operator to pick and choose the most
attractive feedstock at a given point in time in order to maximize profitability.
Olefin plants in the United States have incorporated more feed flexibility than
those in Europe, and are therefore able to take advantage of a wider array of
potential feedstocks.
Most of the flexibility to use various feedstocks is found in plants with asso-
ciated refineries, where integrated olefins plants can optimize feedstocks using
either gas liquids or heavier refinery streams. Figure 1 illustrates the many lin-
kages between an olefin plant and a refinery that is typical of integrated com-
plexes in the United States.
Companies whose primary business is the production of ethylene derivatives,
such as thermoplastics, tend to use ethane and propane feedstocks that minimize
by-product streams and maximize ethylene production for their derivative plants.
Table 1 provides a summary of the 2003 production quantity and value of primary
olefins petrochemicals and olefin feedstocks in the United States.

Refinery–olefin plant
Cat product linkages
cracker

Distillation Off-gases Ethylene

Propylene

Kerosene Butadiene
Crude Naphtha
Benzene
Gas oil Olefin plant
Condensate–NGL Fuels

Gasoline

Purchased Pitch
feeds
Condensate Heavy Gas oil Jet fuel
NGL’s gas oil Hydrocracker

No. 2 fuel oil

Refinery return

Fig. 1. Typical refinery–olefin plant complex. Courtesy of Shell Oil Co.


580 PETROCHEMICAL FEEDSTOCKS Vol. 2

Table 1. Olefin Feedstocks and Olefins Production, 2003a


Material Quantity, million kg Cost, million $
Feedstocks
ethane 13,813 4,043
propane 9,577 2,824
butane 1,917 576
heavier than C-4s 17,880 4,931
Primary olefins
ethylene 22,966 14,407
propylene 13,916 7,031
butadiene 1,901 1,247
a
See Refs. 5 and 6. Courtesy of Nexant ChemSystems.

Table 2. Yieldsa from Typical Ethylene Feedstocks,b wt %


Feedstock Ethane Propane Naphthac Naphthad Gas oil
density, g/L 683.0 706.3 890.5
ASTM ibp, 8Ce 37 37 310
ASTM ebp, 8Cf 141 167 488
Products
hydrogen 3.64 1.41 1.03 0.92 0.58
methane 3.13 23.64 17.50 15.80 9.40
acetylene 0.33 0.49 1.05 0.82 0.30
ethylene 50.04 37.15 35.40 31.70 20.50
ethane 40.00 4.23 3.85 3.80 2.90
propadiene 0.04 0.48 1.11 0.90 0.51
propylene 0.75 13.91 14.90 14.40 13.80
propane 0.16 8.00 0.35 0.30 0.27
butadiene 0.85 3.64 4.60 4.60 4.45
butylenes 0.19 0.98 3.90 3.90 4.10
butane 0.23 0.11 0.23 0.20 0.10
C-5–C-8 0.39 1.83 3.40 4.90 4.75
naphtha
benzene 0.20 2.26 5.50 5.80 6.50
toluene 0.05 0.52 1.60 2.80 2.25
C-8 aromatics 0.00 0.08 0.20 1.40 1.70
styrene 0.00 0.41 0.70 0.75 0.70
C-9-205 FBP 0.00 0.75 1.08 2.05 3.00
naphtha
fuel oil 0.00 0.11 3.60 4.96 24.19
total 100.00 100.00 100.00 100.00 100.00
a
Single pass.
b
Courtesy of ABB Lummus Global Inc.
c
Light naphtha.
d
Full-range naphtha.
e
Initial boiling point (bp).
f
End bp.
Vol. 2 PETROCHEMICAL FEEDSTOCKS 581

Table 3. Yields at High and Low Severity for Full-Range


Naphtha Feedstocka,b
High severity Low severity
Product yield, wt% yield, wt%
hydrogen 0.92 0.74
methane 15.80 12.60
acetylene 0.82 0.35
ethylene 31.70 26.00
ethane 3.80 4.40
propadiene 0.90 0.60
propylene 14.40 17.50
propane 0.30 0.50
butadiene 4.60 4.40
butylenes 3.90 6.80
butane 0.20 0.85
C-5–C-8 naphtha 4.90 11.70
benzene 5.80 3.10
toluene 2.80 2.75
C-8 aromatics 1.40 2.50
styrene 0.75 0.57
C-9-205 FBP naphtha 2.05 2.30
fuel oil 4.96 2.34
total 100.00 100.00
a
Courtesy of ABB Lummus Global Inc.
b
Density ¼ 706.3 g/L; ASTM ibp ¼ 378C; ASTM ebp ¼ 1678C.

The cracking process produces not only ethylene, but other products as well.
With ethane feedstock, only minimal quantities of other products are produced.
As the feedstocks become heavier, increasing quantities of other products are
produced. Most operators characterize these other products as either coproducts
or by-products. In this article, coproducts are defined as other primary olefins
(propylene and butadiene), and primary aromatics (benzene, toluene, and
xylenes). By-products are everything else ranging from C-4 monoolefins to
tars. The principal distinction between coproducts and by-products is that copro-
ducts are the more desired and valuable products where operators often adjust
feedstocks and operating conditions to maximize or minimize production,
depending on the economics at the point in time. The values of these coproducts
and by-products affect the economic attractiveness, and hence the choice of feed-
stock. Table 2 presents typical yields for a new plant employing the latest tech-
nology from various feedstocks as well as a characterization of the feedstock
density and boiling range. Yields from older existing plants differ somewhat
from these yields, generally being lower in ethylene production (see ETHYLENE).
It can be seen that in the case of ethane, the yields of ethylene are the great-
est and only minimal quantities of other products are produced. In the case of gas
oil, the yields of ethylene are the lowest and large quantities of heavier copro-
ducts and by-products are produced. When cracking heavier feedstocks, eg,
naphtha, the severity of the cracking can be varied by adjusting the cracking
temperature. Yields of ethylene can be maximized at the expense of propylene
582 PETROCHEMICAL FEEDSTOCKS Vol. 2

and butylene by increasing severity, or ethylene yields can be minimized and


thus propylene and butylene yields maximized by lowering severity. Table 3
illustrates the effect of severity on yields for full-range naphtha cracking at
high and low severity.
The selection of feedstock and severity of the cracking process are economic
choices, given that the specific plant has flexibility to accommodate alternative
feedstocks. The feedstock prices are driven primarily by energy markets and
secondarily by supply and demand conditions in the olefins feedstock markets.
The prices of individual feedstocks vary widely from time to time as shown in
Figure 2, which presents quarterly prices of the various feedstocks in the United
States from 1985 through 2003 in dollars per metric ton (1000 kg) (6).
It can be seen that the feedstock prices are not only volatile, but also vary
with respect to each other. In 1985, high crude oil prices and natural gas price
controls that tended to restrain the prices of natural gas liquids made gas liquids
feeds much more attractive than heavier naphtha feeds. This situation encour-
aged olefin operators, whose plants were designed for heavy feeds, to modify
their plants to increase their choice of feedstocks. On the other hand, at the
end of 1991, temporary supply shortages of gas liquids caused ethane and pro-
pane prices to rise relative to heavier feeds. It can be seen that prices for all feed-
stocks have generally converged since 1991, with a key exception being during
1995 when depressed natural gas prices resulted in lower ethane prices.
The relative cost of various feedstocks can be more easily understood by
examining the ratios of one to another. In Figure 3, the ratios of the various
alternative feeds to ethane are shown. The high ratio of naphtha to ethane dur-
ing most of the 1985–1991 period and in late 1995 clearly shows the relative dis-
advantage of naphtha feedstock during these periods.

400

350
CURRENT DOLLARS PER TON

300

250

200

150

100

50

0
1985 1987 1989 1991 1993 1995 1997 1999 2001 2003
ETHANE
PROPANE
LIGHT NAPHTHA

Fig. 2. Quarterly olefin feedstock prices, 1985–2003, for (*) ethane (^) propane, and
(~) light naphtha.
Vol. 2 PETROCHEMICAL FEEDSTOCKS 583

1.8

1.6
PRICE RATIO BY WEIGHT

1.4

1.2

0.8

0.6

0.4

0.2

0
1985 1987 1989 1991 1993 1995 1997 1999 2001 2003
PROPANE
LIGHT NAPHTHA

Fig. 3. Price ratio of various feeds to ethane, 1985–2003; (^) propane and, (~) light
naphtha.

Simply looking at the feedstock prices or price ratios is insufficient to accu-


rately identify the most attractive feedstock because the values of all of the
coproducts and by-products must also be taken into account. This is usually
accomplished by calculating the cost to produce ethylene with all other coproduct
and by-product yields credited against the cost of ethylene. An example of the
cost of ethylene is presented in Table 4. The cash costs of ethylene from various
feedstocks are compared for the first two quarters of 2003. Cash costs reflect all
plant manufacturing costs except depreciation and are a measure of the out-of-
pocket cash costs generated by the operation.
The data for the January–March period are generally typical of ethylene
costs when natural gas and natural gas liquids prices are weak relative to prices
for naphtha, and for crude oil and gasoline whose prices heavily influence the
price of naphtha. During this period, crude oil and gasoline prices were particu-
larly strong due to an oil workers’ strike in Venezuela and tensions in the Middle

Table 4. Cash Costsa of Ethylene from Various Feedsb


Ethane Propane Naphtha

Jan.– April– Jan.– April– Jan.– April–


Cost March June March June March June
feedstock 38.7 36.7 73.9 62.8 109.8 83.4
by-product credits 14.8 13.8 50.5 49.3 86.5 81.5
catalyst and chemicals 0.2 0.2 0.2 0.2 0.4 0.4
utilities 9.4 8.8 13.1 12.2 21.0 19.6
operating 2.1 2.1 2.2 2.2 2.5 2.6
plant overhead 2.5 2.6 2.7 2.8 3.2 3.3
total cash costs 38.2 36.6 41.6 31.0 50.4 27.7
a
2003.
b
Courtesy of Nexant ChemSystems.
584 PETROCHEMICAL FEEDSTOCKS Vol. 2

Table 5. Historical U.S. Ethylene Production by Feedstock (million kg of ethylene


production)a
Feedstock 1999 2000 2001 2002 2003
ethane 11,523 11,328 11,098 10,798 10,109
propane 5,023 3,481 3,984 5,077 4,072
n-butane 810 912 1,116 900 778
naphthas 4,279 5,988 2,813 3,430 3,991
gas oil 2,539 2,054 2,288 2,904 3,001
refinery offgas 977 1,021 1,034 1,076 1,010
recovery
total 25,152 24,785 22,332 24,185 22,961
Percent from heavy liquids (naphtha and gas oil)
27.1 32.5 22.8 26.2 30.5
a
Courtesy of Nexant ChemSystems.

East leading up to the war in Iraq. This situation more typically occurs during
summer months when natural gas prices are weaker due to lower seasonal
demand and prices for gas liquids are seasonally lower. Conversely, gasoline
demand and prices are seasonally higher in the summer due to the peak driving
season, and the cost of naphtha reflects this strength.
During the January–March 2003 period, the cash costs of producing ethy-
lene from naphtha were 12¢ per kg of ethylene higher than the cash costs from
ethane and about nine cents per kg of ethylene higher than the cash costs from
propane. Therefore, gas liquids were the most economical source of feedstock
during that period. Table 4 also presents the cash costs of ethylene from the
same feedstocks during the April–June 2003 period when ethane became the
least economical feedstock and naphtha was the preferred feedstock. Very low
inventories of natural gas resulted in sharply higher natural gas prices, also for-
cing up ethane prices in order to provide an incentive to recover it from natural
gas. Thus an ehtylene producer with feedstock flexibility would vary the feed-
slate over time in order to achieve the lowest possible costs. The historical feed-
stock sources for ethylene production in the United States are presented in
Table 5. The shift to heavier feedstocks due to the high natural gas prices during
2003 is also shown.
The corresponding historical feedslate for ethylene production in the United
States is presented in Table 6.

Table 6. Historical U.S. Ethylene Feedslate (million kg feedstock)a


Feedstock
ethane 14,058 13,821 13,540 13,174 12,333
propane 10,549 7,310 8,366 10,662 8,551
n-butane 1,782 2,005 2,454 1,980 1,712
naphthas 12,968 18,146 8,523 10,394 12,094
gas oil 10,233 8,279 9,221 11,703 12,094
refinery offgas recovery 977 1,021 1,034 1,076 1,010
total 50,567 50,583 43,138 48,988 47,795
a
Courtesy of Nexant ChemSystems.
Vol. 2 PETROCHEMICAL FEEDSTOCKS 585

Most new plants built in the United States during the 1990s were based pri-
marily on ethane and propane feeds because the companies constructing these
facilities are predominantly thermoplastic producers requiring ethylene, but
most industry observers expect that olefin feedslates in the United States will
become progressively heavier in the future. Additional supplies of ethane will be
limited, barring significant new discoveries of natural gas, and higher natural gas
prices relative to prices for liquid fuels are expected to reduce the future attractive-
ness of ethane as a feedstock. Some additional propane will probably be imported
from foreign sources. Adequate supplies of naphtha and gas oil are available to
meet the remainder of growth demands for ethylene. On a worldwide basis, gas
liquids accounted for 40% of ethylene production in 2003 and are expected to con-
tinue at that level in the future because of additional recovery of gas liquids asso-
ciated with crude oil and natural gas production outside of the United States. As
the United States olefins feedslate becomes somewhat heavier, the feedslate in the
rest of the world will become somewhat lighter.
Refinery gas streams are a significant source of propylene and a minor source
of ethylene in the United States. About one-half of the propylene produced in the
United States is recovered directly from refinery gas streams, which are largely
produced in fluid catalytic cracking (FCC) units (FLUID CATALYTIC CRACKING (FCC)
UNITS, REGENERATION). FCC units are designed to convert heavy gas oil fractions
into gasoline blend stocks and are used more extensively in the United States
than in Europe and Asia. When used for producing a fuel, this propylene stream
would be converted into heavier gasoline blend stocks by alkylation or dimerization
processes or blended into liquefied petroleum gas (LPG). Refinery-grade propylene
is usually in the 60–70% purity range, with the principal impurity being propane.
Propylene prices are normally such that it is more attractive for refiners in close
proximity to propylene consumers to sell or transfer the refinery-grade propylene
to operators for the purpose of purification to chemical-grade quality (minimum
92% propylene) or polymer-grade quality (99þ% propylene). Other refiners also
have facilities to perform this purification and thus are able to sell one of the higher
purity grades of propylene. Purification consists of fractionation and removal of
trace impurities. There are a few chemical processes that can utilize refinery-
grade propylene without purification; cumene [98-82-8], 2-propanol [67-63-0], and
higher olefins such as nonene (C-9 monoolefins) and tetramer (C-12 monoolefins).

3. Aromatics Feedstocks

Aromatics are produced primarily from two sources, refinery catalytic reformer
operations and recovery from the pyrolysis gasoline fraction from olefin plants.
The most important source in the United States is refinery catalytic reformer
operations. Catalytic reformers are high temperature catalytic dehydrogenation
processes that convert naphthenes contained in virgin naphthas into aromatics
in order to increase the octane value, and hence the gasoline blend value of
the stream being processed. Reformers produce large quantities of the primary
aromatics; benzene, toluene, and isomers of xylene (see BTX PROCESSING).
The preferred feedstocks for aromatics production are naphthas containing
high concentrations of the naphthene precursors to benzene, toluene, and xylenes.
586 PETROCHEMICAL FEEDSTOCKS Vol. 2

Naphthas are characterized according to their naphtheneþaromatic (NþA) values


and boiling range. Naphthas with NþA values in the range of 40–45 or above are
preferred reformer feedstocks and command premium prices. Naphthas with lower
NþA values are poor reformer feedstocks and have lower values, but happily these
low NþA naphthas are preferred for olefins plant feedstocks. The production of aro-
matic chemicals in the reforming process can be maximized by adjusting the distil-
lation cut points on the naphtha feedstock to include the C-6 to C-8 cycloparaffins
and aromatics. The resulting reformate contains from 4 to 8% benzene, with the
higher level preferred. A typical composition (vol%) of the C-6 to C-8 aromatics frac-
tion from a reformer operating at 100 Research Octane Number (RON) severity is
benzene, 12%; toluene, 43%; and xylenes fraction, 45%. The xylene fraction is 25%
o-xylene, 20% p-xylene, 40% m-xylene, and 15% ethylbenzene.
Benzene, toluene, and a mixed-xylene stream are subsequently recovered by
extraction or extractive distillation using a solvent. Recovery of p-xylene from a
mixed-xylene stream requires a further process step of either crystallization and fil-
tration or adsorption on molecular sieves. o-Xylene can be recovered from the raffi-
nate by fractionation. In p-xylene production, it is common to isomerize the m-xylene
in order to maximize the production of p-xylene and o-xylene. Additional benzene can
also be produced by the hydrodealkylation of toluene to benzene to balance supply
and demand. Less common is the hydrodealkylation of xylenes to produce benzene
and the disproportionation of toluene to produce xylenes and benzene, but the latter
process has been more widely applied, particularly in Asia, during the past 10 years.
The second source of aromatics feedstock is pyrolysis gasoline from olefins
plants. Pyrolysis gasoline is one of the by-product streams produced in olefin
plants when cracking heavier feedstocks such as naphtha or gas oil. Pyrolysis
gasoline contains hydrocarbon fractions from C-5 olefins through C-9 aromatics.
Recovery of aromatics from this stream is by extraction and fractionation, just as
in the recovery from reformate, following treatment to saturate diolefins. Pyro-
lysis gasoline is a more important aromatics feedstock in Europe and Asia than
in the United States. This is because naphtha feedstocks are used more for
olefins production in these regions of the world and less gasoline is produced
in these regions; hence there is less reforming in refineries. A typical composition
of C-6–C-8 aromatics stream from pyrolysis gasoline is benzene, 54%; toluene,
31%; and xylenes fraction, 15%. The xylene fraction is 17% o-xylene, 23%
p-xylene, 37% m-xylene, and 23% ethylbenzene.
Pyrolysis gasoline is richer in benzene than reformate, 31% vs 7%. Essen-
tially all of the pyrolysis gasoline in the United States is processed to recover aro-
matic chemicals because of the higher benzene content and problems associated
with blending pyrolysis gasoline into motor gasoline. Pyrolysis gasoline contains
higher molecular unsaturated compounds that tend to form gum materials in
gasoline unless hydrogen treated.
The U.S. Clean Air Act Amendments of 1990 limited the amount of benzene in
reformulated gasoline in the United States to 1% (7). Initially there was some con-
cern that this would disrupt the benzene supply and demand balance in the che-
mical industry because at that time gasoline contained benzene >1%. If it had
been necessary for refiners to extract all of the benzene >1%, substantial addi-
tional benzene could have been produced. However, only modest increases in
the quantity of benzene produced from catalytic reformer sources occurred
when reformulated gasoline became required in 1995 as most refiners adjusted
Vol. 2 PETROCHEMICAL FEEDSTOCKS 587

the composition of reformer feed and reformer severity to produce less benzene. In
addition, reformulated gasoline accounts for about one-third of U.S. sales since its
use is only required in metropolitan areas with the most severe summertime
ozone levels. Therefore, benzene levels >1% are still permissible in the other
two-thirds of gasoline sold in the United States, making it possible for refiners
to preferentially use low benzene streams for reformulated gasoline blends.

4. Other Feedstocks

The only other petrochemical feedstock of significant commercial use is methane


(natural gas) that is used primarily to produce ammonia and methanol. Con-
sumption factors are 28 and 31 GJ/metric ton, respectively (58,300 and
64,700 BTU/lb) (8). Approximately 339 106 GJ (321 1012 BTU) of methane
was processed into 8,770 106 kg of ammonia (9) and 3000 106 kg of metha-
nol in the United States in 2003. Natural gas prices generally follow crude oil
prices in the United States because they compete in energy markets and tradi-
tionally have been lower in cost on a fuel basis. Historical natural gas and crude
oil prices are shown in Figure 4. In contrast, the price of natural gas is decoupled
from crude oil prices in countries with extensive natural gas reserves and limited
domestic markets for natural gas. Global ammonia and methanol production is
increasingly being concentrated in these locations and U.S. production of these
products has declined as U.S. natural gas prices have increased since 2000.
The relationship between natural gas and crude oil prices can be more
easily understood by examining the ratios of one to another. In Figure 5, the
ratio between the U.S. benchmark prices for natural gas and crude oil on a
fuel basis is shown. The ratio was well below 1.0 during the entire period from
1985 through late 2000 when U.S. natural gas supplies were plentiful. Due to
increasing U.S. natural gas demand and limited increases in supplies, the
ratio has risen >1.0 several times since 2000 and a ratio near 1.0 is forecast to
continue in the future.

5
DOLLARS PER GJ

0
1985 1987 1989 1991 1993 1995 1997 1999 2001 2003
CRUDE OIL
NATURAL GAS

Fig. 4. Prices of natural gas (~) and crude oil (^), 1985–2003. To convert ¢/GJ to
¢/BTU, multiply by 1.055 106.
588 PETROCHEMICAL FEEDSTOCKS Vol. 2

1.4
NATURAL GAS TO CRUDE OIL RATIO
1.2

0.8

0.6

0.4

0.2

0
1985 1987 1989 1991 1993 1995 1997 1999 2001 2003

Fig. 5. Fuel price ratio of natural gas at Henry Hub to WTI crude oil, 1985–2003.

5. Alternative Feedstocks

Alternative feedstocks for petrochemicals have been the subject of much research
and study over the past several decades, but have not yet become economically
attractive. Chemical producers are expected to continue to use fossil fuels for
energy and feedstock needs for many decades in the future. The most promising
sources that have received attention include coal, tar sands, oil shale, and
biomass. Near-term advances in coal-gasification technology offer the potential
to replace oil- and gas-based feedstocks in selected applications (10) (see FEED-
STOCK, COAL CHEMICALS). Processes for the production of olefins from alternative
feedstocks are in the development phase with no commercial plants operating.
Several processes using methanol as a feedstock are considered to be the most
promising and are economically attractive since the methanol could be produced
from natural gas in locations having cheap, abundant natural gas reserves.
Lurgi’s MTP technology has been demonstrated in relatively small scale.
This technology converts methanol into propylene at high selectivity. The
UOP/HYDRO MTO process converts crude methanol into an olefinic gas rich
in ethylene and propylene. The ratio of propylene to ethylene produced can be
adjusted within the MTO process. The process yield performance for the UOP/
HYDRO MTO process as presented in public domain literature is presented in
Table 7 (10). The effect of operational severity enables the ethylene/propylene
ratio to be changed. UOP has reported that this ratio can be varied from 1.5 to
0.64 by changing reactor conditions.
Because oil and gas are not renewable resources, at some point in time alter-
native feedstocks will become attractive; and as natural gas and petroleum prices
rise, the timetable for commercialization of alternative feedstocks will accelerate.
Of the alternatives, only agricultural-based materials are a renewable resource
(see BIOMASS ENERGY). Currently, the only chemical produced from a natural pro-
duct in large-volume commercial quantities is ethanol via fermentation of natural
sugar and starches. The vast majority of fermentation ethanol is used for fuel
Vol. 2 PETROCHEMICAL FEEDSTOCKS 589

Table 7. UOP/HYDRO MTO Process Product Yields (Weight


Percent)a
Maximum ethylene Maximum propylene
Product mode mode
ethylene 48 34
propylene 31 45
butenes 9 12
other products 12 9
Ethylene/propylene ratio
1.55 0.76
a
Ref. 10.

applications and this is likely to grow significantly as U.S. production of ethanol


for use as a gasoline blendstock increases. Ethanol (qv) can be converted into a
number of petrochemical derivatives and could become a significant source of che-
micals, especially if new technology for converting the cellulose and hemicellulose
in low cost biomass, such as corn stalks, sugar cane residue and rice straw, to etha-
nol is successfully developed. Several start-up companies in the United States and
Canada are in the throes of such development efforts.
Other notable developments for converting biobased materials to industrial
chemicals include DuPont’s process for taking glucose to a new monomer, 1,3-
propanediol (PDO). This monomer, when condensed with terephthalic acid,
makes a polyester, trimethylene terephthalate (PTT), that is an ideal polymer for
making carpeting. PTT combines the inherent stain resistance characteristic of
polyesters with the excellent resiliency of nylon. DuPont has formed a joint venture
with Tate & Lyle to create products from bioderived PDO. In addition to carpeting,
PDO based polymers may find use in textiles, clothing, and interior applications. In
the late 1990s, Shell developed a petrochemical based process for making PDO and
is also developing the market for PTT. It will be interesting to see which route, the
bio based or petrochemical-based process, is the ultimate winner.
Cargill and Dow have teamed up to develop fermentation technology to con-
vert corn starch to lactic acid at rates that will allow the resulting polymer, poly-
lactic acid (PLA) to compete with large volume petrochemical based plastics such
as nylon and PET. In addition to developing efficacious biocatalysts to speed up
reaction rates, Cargill Dow is increasing the plant size as a means to lower costs.
The capacity of Cargill Dow’s first PLA plant, which came on stream in 2002, is
140,000 metric tons per year and they are already planning additional capacity.
Cargill Dow believes the market potential for PLA could reach 500,000 metric
tons by 2010. Products based on PLA offer many good properties and will find
application in consumer goods, food packaging, bedding, clothing and other
applications. PLA, in addition to being made from renewable resources, also
has the benefit of being biodegradable or easily recycled back to its starting
monomer. Lactic acid esters, such as ethyl or buty lactate, are also being
explored as environmentally friendly substitutes for petrochemical-based sol-
vents such as methylethyl Ketone (MEK) or isopropanol.
590 PETROCHEMICAL FEEDSTOCKS Vol. 2

Cargill also has other programs in place to produce industrial chemicals from
renewable agricultural resources. Some of these projects include conversion of glu-
cose to 3-hydroxypropionic acid (3-HP). 3-HP, using conventional industrial chem-
istry, can be converted to acrylic acid, PDO, and other specialty polyesters;
production of urethane polyols from vegetable oils and using metathesis chemistry
to cleave and modify unsaturated natural oils to olefins and difunctional olefins.
Another promising bioconversion process that has the potential to compete
with petrochemical processes is the conversion of glucose to succinic acid. Several
U.S. Government sponsored laboratories (Argonne National Laboratory, Pacific
Northwest National Laboratory and National Renewable Energy Laboratory)
have collaborated to develop fermentation technology to convert sugars to succinic
acid. Succinic acid can undergo hydrogenolysis to 1,4-butanediol (BDO). In 2003,
the global market for BDO is estimated to be over 800,000 metric tons per year.
Besides the U.S. based development in this area, Mitsubishi Chemical, working
together with Ajinomoto in Japan, is also investigating a similar approach.

BIBLIOGRAPHY

‘‘Petroleum Products’’ in ECT 1st ed., Vol. 10, pp. 161–177, by V. L. Shipp, Socony-
Vacuum Oil Co., Inc.; ‘‘Petroleum Products’’ under ‘‘Petroleum’’ in ECT 2nd ed., Vol. 15,
pp. 77–92, by G. B. Gibbs and H. L. Hoffman, Hydrocarbon Processing; ‘‘Petroleum
(Products)’’ in ECT 3rd ed., Vol. 17, pp. 257–271, by H. L. Hoffman, Gulf Publishing
Co.; ‘‘Feedstocks’’ in ECT 3rd ed., Vol. 9, pp. 831–845, by F. A. M. Buck, King, Buck, &
Associates, and M. G. Marbach, Shell Chemical Co.; ‘‘Feedstocks, Petrochemicals’’ in
ECT 4th ed., Vol. 10, pp. 349–360, by Darryl C. Aubrey, Sacred Heart University Chem
Systems, Inc.; ‘‘Feedstocks, Petrochemicals’’ in ECT (online), posting date: December 4,
2000, by Darryl C. Aubrey, Sacred Heart University Chem Systems, Inc.

CITED PUBLICATIONS

1. Oil Gas J. 99(42), 56 (2001).


2. Asian Chemical News 8(375), 22 (2002).
3. P. Spitz, Petrochemicals, The Rise of an Industry, Wiley-Interscience, 1988, pp.
77–78.
4. Hydrocarbon Process. 82(3) 97 (2003).
5. NPRA Petrochemical Surveys, National Petrochemical & Refiners Association,
Washington, D.C., Fourth Quarter 2003.
6. Platts Oilgram Price Reports.
7. Sections 211(k)(1) and 211(k)(3) of 1990 Clean Air Act Amendments.
8. Hydrocarbon Process. 70(3), 134, 164 (1991).
9. Ammonia Statistics 2003, International Fertilizer Industry Association, Paris,
France, April 2004.
10. UOP Flysheet for UOP/HYDRO Process.

JOHN T. BOEPPLE
Nexant ChemSystems
Vol. 2 HYDROCARBONS 591

HYDROCARBONS
1. Saturated Hydrocarbons

The class of saturated hydrocarbons comprises a myriad of individual com-


pounds. A large number of the theoretically possible saturated hydrocarbons is
known (1), although only a limited number of individual saturated hydrocarbons
are used as raw materials in the chemical industry.
1.1. Physical Properties. Saturated hydrocarbons are colorless nonpo-
lar substances, immiscible with polar solvents, but miscible with many nonpolar
organic solvents. Some physical properties are given in Table 1. The boiling
points and, starting with C3, melting points of n-alkanes increase with increas-
ing molecular mass. At 20 8C and atmospheric pressure, C1 to C4 n-alkanes are
gases, C5 to C16 are liquid, and from C17 solid. i-Alkanes do not display a definite
correlation between the number of carbon atoms and their boiling- and melting
points. The boiling points of i-alkanes all are lower than those of the correspond-
ing n-alkanes; this is also true for many of the melting points. Furthermore, the
boiling points of isomeric alkanes decrease with increasing degree of branching.
The boiling points and melting points of cycloalkanes are generally higher than
those of n-alkanes having the same number of carbon atoms (Table 1).
The density of the saturated hydrocarbons in the liquid state at 20 8C is
<1 g/cm3 ; it varies from 0.6 g/cm3 for compounds with low carbon numbers to
0.8 g/cm3 for compounds with high carbon numbers (Table 1). Additional physical
properties are compiled in Tables 2-5.
1.2. Chemical Properties. Alkanes and cycloalkanes are saturated,
nonpolar, and lack functional groups; such hydrocarbons can undergo reaction
only after cleavage of CH or CC bonds. Consequently, the scope of primary
reaction steps is essentially limited to dehydrogenation, substitution, and
chain- or ring cleavage. Saturated hydrocarbons cannot undergo addition reac-
tions, which represent a versatile synthetic tool for unsaturated hydrocarbons.
Most industrial reactions involving saturated hydrocarbons are radical reac-
tions, e.g, thermal cracking, oxidation, sulfoxidation, halogenation, sulfochlori-
nation, and nitration. Industrial ionic reactions of saturated hydrocarbons are
restricted to acid-catalyzed processes with strong acids. Such reactions are
employed mainly in the processing of petroleum by catalytic cracking, isomeriza-
tion, and alkylation; they yield complex mixtures of products. Other ionic reac-
tions of saturated hydrocarbons can be carried out with super acids (e.g,
FSO3H – SbF5, HF – SbF5): the alkylation of alkanes by alkanes (7), alkylation
of benzene by alkanes (7), and ionic chlorination (8) and bromination (8) of
alkanes have been explored, but are not used industrially.
Reactions of saturated hydrocarbons, because of the lack of functional
groups, are nonselective, with respect to both the region of attack (regioselectiv-
ity) and the number of reaction sites (degree of substitution), unless the molecule
possesses specific structural features, e.g, tertiary hydrogen atoms. Such reac-
tions frequently afford mixtures of isomeric or structurally analogous compounds
which can only be separated with difficulty or not at all.
On the basis of free enthalpy of formation (see Table 2) most of the saturated
hydrocarbons are thermodynamically unstable with respect to the elements

ß2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim


10.1002/14356007.a13_227
592 HYDROCARBONS Vol. 2

Table 1. Physical Properties of Saturated Hydrocarbons (2)


CAS
Empirical registry Z(20 8C), %;(20 8C),
Compound formula number mp, 8C bp, 8C mPa  s g/cm3 n20
D
Methane CH4 [74-82-8] 182.5a * 161.5
Ethane C2H6 [74-84-0] 183.27a * 88.6
Propane C3H8 [74-98-6] 187.69a * 42.1 0.5005b
Cyclopropane C3H6 [75-19-4] 127.4 32.8
n-Butane C4H10 [106-97-8] 138.4 0.5 0.5788b 1.3326b
Cyclobutane C4H8 [287-23-0] 90.7 12.5 0.6943b 1.365a
2-Methylpropane C4H10 [75-28-5] 159.6 11.7 0.5572b
n-Pentane C5H12 [109-66-0] 129.7 36.1 0.234 0.62624 1.35748
Cyclopentane C5H10 [287-92-3] 93.9 49.3 0.438 0.74538 1.40645
2-Methylbutane C5H12 [78-78-4] 159.9 27.9 0.224 0.61967 1.35373
2,2-Dimethylpropane C5H12 [463-82-1] 16.6 9.5 0.5910b 1.342b
n-Hexane C6H14 [110-54-3] 95.3 68.7 0.3117 0.65937 1.37486
2-Methylpentane C6H14 [107-83-5] 153.7 60.3 0.65315 1.37145
3-Methylpentane C6H14 [96-14-0] 63.3 0.66431 1.37652
2,2-Dimethylbutane C6H14 [75-83-2] 99.9 49.7 0.64916 1.36876
2,3-Dimethylbutane C6H14 [79-29-8] 128.5 58.0 0.66164 1.37495
n-Heptane C7H16 [142-82-5] 90.6 98.4 0.4169 0.68376 1.38764
2-Methylhexane C7H16 [591-76-4] 118.3 90.1 0.67859 1.38485
3-Methylhexane C7H16 [589-34-4] 91.9 0.68713 1.38864
3-Ethylpentane C7H16 [617-78-7] 118.6 93.5 0.69816 1.39339
2,2-Dimethylpentane C7H16 [590-35-2] 123.8 79.2 0.67385 1.38215
2,3-Dimethylpentane C7H16 [565-59-3] 89.8 0.69508 1.39196
2,4-Dimethylpentane C7H16 [108-08-7] 119.2 80.5 0.67270 1.38145
3,3-Dimethylpentane C7H16 [562-49-2] 134.46d 86.1 0.69327 1.39092
2,2,3-Trimethylbutane C7H16 [464-06-2] 24.9 80.9 0.69011 1.38944
n-Octane C8H18 [111-65-9] 56.8 125.7 0.5450 0.70252 1.39743
2-Methylheptane C8H18 [592-27-8] 109.0 117.6 0.69792 1.39494
3-Methylheptane C8H18 [589-81-1] 120.5 118.9 0.70582 1.39848
4-Methylheptane C8H18 [589-53-7] 121.0 117.7 0.70463 1.39792
2,2,3-Trimethylpentane C8H18 [564-02-3] 112.3 109.8 0.71602 1.40295
2,2,4-Trimethylpentane C8H18 [540-84-1] 107.4 99.2 0.69193 1.39145
2,3,3-Trimethylpentane C8H18 [564-02-3] 100.7 114.8 0.72619 1.40750
2,3,4-Trimethylpentane C8H18 [565-75-3] 109.2 113.5 0.71906 1.40422
2,2,3,3-Tetramethyl- C8H18 [594-82-1] 100.7 106.5
butane
n-Nonane C9H20 [111-84-2] 53.5 150.8 0.7139 0.71763 1.40542
n-Decane C10H22 [124-18-5] 29.7 174.1 0.9256 0.73005 1.41189
n-Undecane C11H24 [1120-21-4] 25.6 195.9 1.185 0.74024 1.41725
n-Dodecane C12H26 [112-40-3] 9.6 216.3 1.503 0.74869 1.42160
n-Tridecane C13H28 [629-50-5] 5.4 235.4 1.880 0.75622 1.42560
n-Tetradecane C14H30 [629-59-4] 5.9 253.5 2.335 0.76275 1.42892
n-Pentadecane C15H32 [629-62-9] 9.9 270.6 2.863 0.76830 1.43188
n-Hexadecane C16H34 [544-76-3] 18.2 286.8 3.474 0.77344 1.43453
n-Heptadecane C17H36 [629-78-7] 21.98 302.2 4.196c 0.7779c 1.4368c
n-Octadecane C18H38 [593-45-3] 28.2 316.7 0.7818c 1.4389c
n-Nonadecane C19H40 [629-92-5] 31.9 330.6 0.7854c 1.4408c
n-Icosane C20H42 [112-95-8] 36.4 343.8 0.7886c 1.4425c
n-Triacontane C30H62 [638-68-6] 65.8 449.7 0.8096c 1.4535c
n-Tetracontane C40H82 [4181-95-7] 81.5 525 0.8204c 1.4592c
a
For saturation pressure (triple point).
b
At saturation pressure.
c
For the super-cooled liquid below the normal melting point.
d
Value for the melting point of the metastable crystalline form.
Vol. 2 HYDROCARBONS 593

Table 2. Molar Enthalpies of Fusion, Vaporization and Combustion (2)


Molar enthalpy, kJ/mol
Fusion Vaporization (25 8C) Combustiona
Methane 0.942 802.861
Ethane 2.86 5.02 1 428.787
Propane 3.53 15.1 2 045.377
Cyclopropane 5.44 1 960.637
n-Butane 4.66 21.1 2 658.827
Cyclobutane 5.78 23.7 2 569.523
2-Methylpropane 4.54 19.1 2 650.454
n-Pentane 8.41 26.8 3 274.287
Cyclopentane 4.89 28.5 3 101.707
2-Methylbutane 5.15 24.8 3 266.248
2,2-Dimethylpropane 3.15 21.8 3 254.735
n-Hexane 13.1 31.7 3 886.81
2-Methylpentane 6.27 30.0 3 879.07
3-Methylpentane 30.4 3 881.71
2,2-Dimethylbutane 0.579 27.8 3 869.78
2,3-Dimethylbutane 0.80 29.2 3 877.86
n-Heptane 14.1 36.6 4 501.44
n-Octane 20.8 41.5 5 115.57
n-Dodecane 35.9 61.3 7 580.076
n-Hexadecane 51.9 81.1 10 040.616
n-Icosane 69.9 100.9b 12 501.198
a
For the reaction: Hydrocarbon (g) ! CO2 (g) þ H2O (g).
b
For the supercooled liquid, below the normal melting point.

Table 3. Enthalpy of Formation, Entropy, and free Energy of Formation for Saturated
Hydrocarbons as Gases (25 8C) (3)
DH0, kJ/mol S0, J mol1 K1 DG0, kJ/mol
Methane 74.898 186.313 50.828
Ethane 84.724 229.646 32.908
Propane 103.916 270.090 23.505
Cyclopropane þ53.382 237.810 104.503
n-Butane 126.232 310.326 17.166
Cyclobutane þ27.256 265.569 110.741
2-Methylpropane 134.606 294.834 20.934
n-Pentane 146.538 349.179 8.374
Cyclopentane 77.079 293.076 38.895
2-Methylbutane 154.577 343.820 14.654
2,2-Dimethylpropane 166.090 306.599 15.240
n-Hexane 167.305 388.661 0.293
2-Methylpentane 174.422 380.789 5.024
3-Methylpentane 171.743 380.036 2.135
2,2-Dimethylbutane 185.685 358.474 9.923
2,3-Dimethylbutane 177.897 366.010 4.103
n-Heptane 187.945 428.058 þ8.039
n-Octane 208.586 467.038 16.412
n-Dodecane 291.066 622.912 50.074
n-Hexadecane 373.588 778.829 83.820
n-Icosane 456.068 934.745 117.398
594 HYDROCARBONS Vol. 2

Table 4. Critical Data and Molar heat Capacities of Saturated Hydrocarbons (2)
Compound Tc, 8C pc, MPa cp, * J mol1 K1
Methane 82.6 4.60 35.74
Ethane 32.28 4.88 52.67
Propane 96.7 4.25 73.56
Cyclopropane 125.1 5.57 56.27
n-Butane 152.0 3.80 97.51
Cyclobutane 187 4.99 72.26
2-Methylpropane 135.0 3.65 96.88
n-Pentane 196.5 3.37 120.3
Cyclopentane 238.5 4.50 82.98
2-Methylbutane 187.2 3.38 118.9
2,2-Dimethylpropane 160.6 3.20 121.7
n-Hexane 234.2 3.01 143.2
2-Methylpentane 224.3 3.01 144.3
3-Methylpentane 231.2 3.12 143.2
2,2-Dimethylbutane 215.6 3.08 142.0
2,3-Dimethylbutane 226.8 3.13 140.6
n-Heptane 267.0 2.74 166.1
n-Octane 295.6 2.49 189.0
n-Nonane 321.4 2.29 211.8
n-Decane 344.4 2.10 234.7
n-Undecane 365.6 1.97 257.6
n-Dodecane 385.1 1.82 280.5
n-Tridecane 402.6 1.72 303.4
n-Tetradecane 418.7 1.62 326.3
n-Pentadecane 433.6 1.52 349.2
n-Hexadecane 447.4 1.42 372.0
n-Heptadecane 460.2 1.32 394.9
n-Octadecane 472.1 1.22 417.8
n-Nonadecane 483 1.12 440.7
n-Icosane 494 1.12 463.6
*Based on 25 8C and ideal gas state.

carbon and hydrogen. They are, however, kinetically stable at ambient tempera-
ture. Thermal decomposition of saturated hydrocarbons proceeds stepwise by
loss of hydrogen or hydrocarbon fragments with concomitant formation of indust-
rially useful unsaturated cracking products, such as acetylene, olefins, or aromatic
hydrocarbons.
On ignition, mixtures of saturated hydrocarbons and oxygen or air may
lead to combustion or explosion, depending on the ratio of hydrocarbon and
oxygen. Such reactions can be initiated either by imposed ignition or, if the
ignition temperature of the mixture is exceeded, by self-ignition. These reac-
tions are the basis for the use of hydrocarbons as heating- and engine fuel.
Pertinent fuel properties of individual saturated hydrocarbons are summarized
in Table 4.
1.3. Production. By far the largest amount of saturated hydrocarbons is
obtained from the natural sources natural gas and petroleum, either by isolation
or by suitable conversion reactions. Additional sources include various products
derived from coal processing. A number of saturated hydrocarbons, unavailable
from natural sources, are produced by special synthesis or by conversion
processes.
Vol. 2 HYDROCARBONS 595

Table 5. Fuel Properties of Saturated Hydrocarbons


Explosion
limits (20 8C), b Ignition Octane
temperature,a number,c
Flash point,a 8C lower upper 8C RON

Methane 5.00 15.00 595


Ethane 3.00 12.50 515
Propane 2.12 9.35 470
Cyclopropane 2.40 10.40 495
n-Butane 1.86 8.41 365 93.8
Cyclobutane 1.8a
2-Methylpropane 1.8a 8.5a (460)d
n-Pentane <20 1.40 7.80 285 61.7
Cyclopentane <20 380 102.5
2-Methylbutane <20 1.32 7.6a 420 92.3
2,2-Dimethylpropane 1.38 7.50 (450)d 85.5
n-Hexane <20 1.18 7.40 240 24.8
i-Hexane <20 ca. 1d ca. 7.4d ca. 260
2,2-Dimethylbutane <20 1.2a 7.0a 435 91.8
2,3-Dimethylbutane <20 1.2a 7.0a 415 103.5
n-Heptane 4 1.10 6.70 215 0
i-Heptane <4 ca. 1c ca. 7c ca. 220
2,3-Dimethylpentane <0 1.12 6.75 330 91.9
2,2,3-Trimethylbutane <0 450 112.1
2,3-Dimethylhexane <12 0.95 6.5a 210
n-Octane 12 0.95 6.5a 71.3
i-Octane 12 1.0 6.0 410 100
2,2,3-Trimethylpentane <21 430 109.5
2,3,3-Trimethylpentane <21 425 106
n-Nonane 31 0.83d 5.6a;d 205
n-Decane 46 0.77d 5.35d 205
n-Dodecane 74 0.6a;d 200
n-Hexadecane >100 205
n-Icosane >100
a
Ref. (4)
b
Ref. (4)
c
Ref. (5)
d
Approximate values.
e
Values are valid for t > 208C.

From Natural Gas and Petroleum. Natural Gas contains methane as


the single major component (see also Natural Gas). Depending on the particular
source, natural gas may also contain acyclic saturated hydrocarbons up to C5 in
proportions that permit their isolation (9). The isolation of individual compounds
from natural gas (see Figure 1) can be performed either by absorption or by par-
tial condensation at low temperature, followed by distillation.
Petroleum is the most abundant source of saturated hydrocarbons, both
with respect to the absolute amount and to the variety of individual compounds.
Petroleum is separated into individual fractions by distillation in a refinery.
Such fractions contain several (liquefied petroleum gas, LPG; see Liquefied
Petroleum Gas) or many individual hydrocarbons (all other fractions); these frac-
tions can be used as such or processed further for use as heating- or engine fuels,
as lubricants, as raw materials for the production of petrochemicals (synthesis
596 HYDROCARBONS Vol. 2

Fig. 1. Saturated hydrocarbons from natural gas and from petroleum.

gas see, Gas Production, acetylene, olefins, aromatics), or for the recovery of
industrially important alkanes and cycloalkanes.
From the liquefied petroleum gas fraction, propane, n-butane, and i-butane
are isolated by distillation. n-Pentane and i-pentane can be isolated from the
liquid components of natural gas or from light gasoline (naphtha) either by mole-
cular sieve separation or by superfractionation (10). Steam-cracking of light
gasoline provides, among other products, cyclopentadiene (see Cyclopentadiene
and Decyclopentaden), benzene (see Benzene), and 1,3-butadiene (see Buta-
diene); these can, in turn, be converted into the C5-, C6-, C8-, and C12-cycloalk-
anes. Catalytic reforming of heavy gasoline provides C6- to C8-aromatics,
which can subsequently be converted into cyclohexane (see Cyclohexane), e.g,
the C7- and C8-aromatics are hydrodealkylated, and the ensuing benzene subse-
quently hydrogenated. From the higher boiling petroleum, gas-oil, and wax dis-
tillate fractions, mixtures of homologous n-alkanes are isolated by techniques,
such as molecular sieve separation or urea extractive crystallization. Paraffin
waxes can also be isolated by precipitation with suitable solvents.
Higher boiling petroleum distillates and distillation residues can be con-
verted into mixtures of lower molecular mass hydrocarbons by hydrocracking
(11,12). From such mixtures, saturated C3 – C5-hydrocarbons and a broad
spectrum of higher n-alkanes can be recovered.
Vol. 2 HYDROCARBONS 597

Although petroleum and the various petroleum distillates contain a large


variety of other saturated hydrocarbons in addition to those already mentioned,
the recovery and subsequent industrial use of individual compounds is largely
confined to those depicted in Figure 1; the increasing number of structural iso-
mers with increasing molecular mass poses severe limitations on the economical
separation of individual compounds from such distillate mixtures on an indus-
trial scale.
By Synthesis and by Conversion of other Hydrocarbons. The major
processing techniques are cyclization, hydrogenation, and isomerization, forming
the basis of the industrial production of most cycloalkanes. Other techniques
include alkylation of alkenes with alkanes to provide i-alkanes, and hydrocrack-
ing to give predominantly low molecular mass i- and n-alkanes, including
methane.
Cyclization and Hydrogenation. Among the cycloalkanes only those
containing a 5- or a 6-membered ring are directly available from natural
sources, i.e, from petroleum or coal-derived products. Steamcracking of
naphtha or gas oil produces cyclopentadiene as a byproduct, which can be con-
verted into cyclopentane by catalytic hydrogenation. Similarly, cyclohexane is
produced by catalytic hydrogenation of benzene, which in turn is formed by
steam cracking of liquid hydrocarbons, by catalytic reforming of heavy gasoline,
or by coking of coal.
The skeletons of all other industrially produced cycloalkanes are synthe-
sized from acyclic precursors. Although there is a wide variety of synthetic pro-
cedures for such reactions, only a limited number are used industrially. Of
prime importance are cycloaddition reactions of acyclic unsaturated com-
pounds followed by hydrogenation of the resulting unsaturated cyclic products:
Thus, [4 þ 2]-cycloaddition (Diels – Alder reaction) of conjugated dienes and
monoolefins yields saturated 6-membered ring compounds via cyclohexene
intermediates, e.g, the synthesis of cyclohexane from 1,3-butadiene and
ethylene (13):

the catalyzed [4 + 4] -cycloaddition of 1,3-butadiene to give cyclooctane via


1,5-cyclooctadiene;

and the catalyzed [4 + 4 + 4] -cyclotrimerization of 1,3-butadiene to form cyclo-


dodecane via 1,5,9-cyclododecatriene:
598 HYDROCARBONS Vol. 2

Cyclizations that lead to small (C3-, C4-) rings are not industrially impor-
tant, although various laboratory methods are available (14). Examples include
the coupling of terminal carbon atoms by elimination of suitable a,o-substituents
from acyclic compounds, e.g, a,o-dihalides, o-halogenated ketones, esters, or
nitriles (Eq. 1), addition of carbenes, to olefins to give cyclopropanes (Eq. 2) and
[2 + 2] -cycloadditions of suitable unsaturated compounds, such as olefins,
allenes, or acetylenes, to give cyclobutanes (Eq. 3) (15):

ð1Þ

ð2Þ

ð3Þ

Most of these synthetic methods yield substituted or unsaturated cyclic


compounds in the cyclization step. These primary products can be subsequently
converted into the corresponding cycloalkanes by suitable measures.
Isomerization and Alkylation. On account of their specific properties
(antiknock activity, possibility for selective reactions), some i-alkanes are in
higher demand than can be met by natural sources. i-Alkanes are produced either
by acid-catalyzed isomerization of n-alkanes or by acid-catalyzed alkylation of ole-
fins with i-alkanes. The most important isomerizations are the conversion of n-
butane to i-butane, n-pentane to i-pentane, and n-hexane to i-hexane (16,17).
Industrially important alkylations include the reaction of i-butane with propylene
or with butenes to give highly branched C7- or C8-alkanes, respectively (see also
! Oil Refining). The latter exhibit high octane numbers (17,18).
Hydrocracking. A range of acyclic saturated hydrocarbons having low
carbon numbers is accessible by catalyzed hydrocracking of heavy distillates or
distillation residues. Hydrocracking of distillates can be optimized by adjusting
the reaction conditions, so that the products are predominantly in the range of
liquefied petroleum gases, including i-butane (12), or in the range of middle dis-
tillates (19). The cracking of petroleum distillates with steam over a nickel cat-
alyst produces methane (20,21).
1.4. Uses. The major use of saturated hydrocarbons is as a mixture,
sometimes with unsaturated or aromatic compounds, as heating fuels, motor
fuels and lubricants. Saturated hydrocarbons also serve as raw materials for
the production of carbon black. In the chemical industry, the following processes
are important (Fig. 2).
Thermal Cracking. Thermal cracking of individual or mixed saturated
hydrocarbons is the basis for the petrochemical production of acetylene, olefins,
diolefins, and to some extent BTX aromatics: high temperature pyrolysis at
>1000 8C affords acetylene (see Acetylene) and, depending on the hydrocarbon
feed and the particular process applied, ethylene. The range of hydrocarbon
Vol. 2 HYDROCARBONS 599

Fig. 2. Industrial use of saturated hydrocarbons.

feed materials can vary from methane to crude oil. Medium temperature
pyrolysis at 750–900 8C in the presence of inert diluents such as steam
(steam-cracking) or hydrogen (hydropyrolysis) yields ethylene (see Ethylene).
A wide range of hydrocarbons, e.g, ethane, liquefied petroleum gas, naphtha,
and gas oil may serve as raw material, and increasing proportions of C3- to
C5-olefins, C4- and C5-dienes, and C6- to C8-aromatics are coproduced with
increasing molecular mass of the feed material. Pyrolysis of petroleum waxes
at 500–600 8C affords mixtures of homologous olefins with predominantly
terminal double bonds.
Catalytic Dehydrogenation. Catalytic dehydrogenation of acyclic satu-
rated hydrocarbon atoms yields mono- and/or diolefins with the same number of
carbon atoms. Industrial processes include the production of butenes and buta-
diene from n-butane, of isoprene from i-pentane, and C6- to C19-monoolefins,
mainly with internal double bonds, from the corresponding n-alkanes (22).
More recently, improved processes for the production of ethylene, propene, and
600 HYDROCARBONS Vol. 2

i-butene by catalytic dehydrogenation of the corresponding saturated hydrocar-


bons have been developed (23). Catalytic dehydrogenation of cyclohexane and
methyl cyclohexanes yields benzene and methylbenzenes, respectively.
Catalytic dehydrocyclooligomerization of propane and n-butane over modi-
fied zeolite catalysts represents a new method for the production of BTX-aro-
matics (24). The process is apparently ready for industrial application.
Partial Oxidation. Partial oxidation of saturated hydrocarbons either by
catalyzed reaction with steam (steam reforming) or by noncatalyzed reaction
with deficient amounts of oxygen affords a mixture of carbon monoxide and
hydrogen, which is used as town gas or as synthesis gas (! Gas Production).
Desulfurized saturated hydrocarbons from methane to compounds with bp ca.
200 8C are suitable feed materials for catalytic processes; for noncatalytic pro-
cesses all hydrocarbons may be used from methane to distillation residues.
Selective Oxidation. Selective oxidation represents the most important
tool for the introduction of oxygen-containing functional groups into saturated
hydrocarbons: Liquid-phase autoxidation of i-butane with air yields tert-butylhy-
droperoxide; the latter is used for the selective oxidation of propene to propylene
oxide (25). Liquid-phase oxidation of n-alkanes or cycloalkanes in the presence of
boric acid affords the corresponding secondary alcohols. The reaction is used
industrially for the oxidation of C10- to C20-n-alkanes, providing raw materials
for detergents and for the oxidation of cyclododecane to cyclododecanol as an
intermediate for the production of nylon 12. Catalyzed liquid-phase oxidation
of n-alkanes and cycloalkanes in the absence of boric acid leads to mixtures of
secondary alcohols and ketones or to carboxylic acids, depending on the reaction
conditions: oxidation of cyclohexane leads to a mixture of cyclohexanol and cyclo-
hexanone; oxidation of n-butane gives acetic acid; oxidation of longer chain
n-alkanes produces mixtures of fatty acids; and oxidation of cyclohexane gives
adipic acid. Liquid phase oxidation of bicyclo [4.4.0] decane yields the correspond-
ing tertiary hydroperoxide, which is eventually converted into sebacic acid by a
sequence of reaction steps (26).
Heterogeneously catalyzed gas-phase oxidation of n-butane has been used
industrially as an attractive alternative for the production of maleic anhydride
(27). Selective oxidation of methane to methanol (28) or formaldehyde (29) and of
i-butane to methacrylic acid (30) has been examined, but is not yet used industrially.
Oxidative Couplings. Oxidative coupling of methane either with itself
to give ethane and ethylene (31) (Eq. 4), or with compounds having activated
methyl groups (Eq. 4) to give terminal olefinic compounds (32) has been investi-
gated, but is not yet industrially used.

O2
CH4 ! CH3 CH3 þ CH2 ¼ CH2 ð4Þ
H2 O
O2
CH4 þ CH3  X ! CH2 ¼ CH  X ð5Þ
H2 O

X ¼ CH ¼ CH2 ;CðCH3 Þ ¼ CH2 ; C6 H5 ; C  N

Three-Component Oxidation. Three-component oxidation allows


introduction of hetero atoms or groups of hetero atoms other than oxygen into
Vol. 2 HYDROCARBONS 601

saturated hydrocarbons. The radical-induced action of both oxygen and sulfur


dioxide (sulfoxidation) on alkanes yields the corresponding alkanesulfonic
acids. The latter, produced from C14 – to C18 – n-alkanes, are useful as raw mate-
rials for detergents. The heterogeneously catalyzed reaction of oxygen and
ammonia with methane (ammoxidation) affords hydrogen cyanide. The latter
reaction can be also carried out in the absence of oxygen as a two-component
codehydrogenation. The reaction of oxygen and ammonia with propane to give
acrylonitrile has recently been explored.
Chlorinations. Chlorination of saturated hydrocarbons yields mono- or
polychlorinated hydrocarbons, depending on conditions. The following reactions
are used industrially: Chlorination of methane to give the four chloromethanes;
the chlorination of n-alkanes to give mono- or polychlorinated alkanes having the
same number of carbon atoms and chlorinolysis of low molecular mass alkanes or
chloroalkanes to give tetrachloroethene and tetrachloromethane. Reaction of
chlorine and sulfur dioxide together with alkanes yields alkanesulfonyl chlor-
ides, which in turn can be converted into the corresponding sulfonic acids, sulfo-
nates, sulfonamides, or sulfonic esters.
Photonitrosation. Photonitrosation of saturated alkanes with nitrosyl
chloride yields oximes via labile secondary nitroso compounds. The method has
been used for the production of caprolactam from cyclohexane.
Nitrosation. Nitrosation of saturated hydrocarbons with nitric acid has
been carried out with low molecular mass alkanes, particularly with propane
in the gas phase. The reaction gave mixtures of nitro compounds with the
same number of and fewer carbon atoms than those of the starting alkanes.
Such mixtures are useful as specialty solvents, e.g, for polymers.
Reaction with Sulfur. Reaction of methane with sulfur gives carbon
disulfide in high yields with hydrogen sulfide as byproduct. The reaction can
be performed by thermal or catalytic methods.
Fermentation. Fermentation of certain saturated hydrocarbons by
microorganisms (yeast, bacteria) yields single cell proteins (33–37). The microor-
ganisms oxidize the hydrocarbons in the presence of oxygen and nutrients with
concomitant formation of protein-containing cells. Certain microorganisms are
specific for the fermentation of n-alkanes. Industrial processes have therefore
been developed both for the fermentation of isolated n-alkanes and for n-alkanes
in admixture with branched or cyclic hydrocarbons (e.g, diesel oil). The products
obtained are used as animal feed.
1.5. Individual Saturated Hydrocarbons Some industrially used
saturated hydrocarbons are discussed below.
Ethane. Ethane is present in many natural gas sources and in refinery
gases; it can be recovered for industrial use. Furthermore, ethane is formed in
thermal- and hydrocracking of hydrocarbons and in the liquefaction of coal.
The single most important industrial use for ethane is the production of ethylene
by steam-cracking (see Ethylene – Commercial Cracking Yields). Thermally-
induced chlorination of ethane yields predominantly monochloro-, 1,1-dichloro-,
or 1,1,1-trichloroethane, depending on the conditions used. Combined chlorina-
tion and oxichlorination of ethane in a salt melt affords vinyl chloride (38).
Reaction of ethane with nitric acid in the gas phase yields nitroethane and
nitromethane.
602 HYDROCARBONS Vol. 2

Propane. Propane is a component of liquefied petroleum gas (LPG),


which is derived from natural gas or petroleum. Propane can be recovered
from LPG by distillation. The major industrial use of propane is the production
of ethylene and propene by the steam cracking process (see Ethylene–Commer-
cial Cracking Yields). Chlorinolysis of propane at elevated temperature yields a
mixture of tetrachloroethene and tetrachloromethane. The conversion of propane
into BTX aromatics by zeolite-catalyzed dehydrocyclooligomerization has been
developed recently and may find industrial application. Liquefied propane is
used for the deasphalting of petroleum residues.
n-Butane. n-Butane can be recovered from LPG by distillation; it has a
variety of industrial uses: steamcracking yields ethylene and propene, catalytic
dehydrogenation yields butadiene (see Butadiene), and acid-catalyzed isomeri-
zation provides i-butane. Oxidation of n-butane in the gas phase with heteroge-
neous catalysis is a modern process for the production of maleic anhydride (27).
Noncatalyzed oxidation of n-butane in the liquid or gas phase has been used
industrially to produce acetic acid (39).
i-Butane. i-Butane can be obtained either by recovery from LPG or by
isomerization of n-butane. Traditional industrial uses include the production of
high octane engine fuel by the alkylation of olefins (17,18) and liquid-phase auto-
xidation of i-butane to give tert-butylhydroperoxide. Recent developments com-
prise improved processes for the catalytic dehydrogenation of i-butane (3) to
satisfy the increasing demand for i-butene as a pecursor for methyl tert-butyl
ether (MTBE) (see Methyl Tert-Butyl Ether), and the production of methacrylic
acid from i-butane by oxidation (30).
n-Pentane and i-Pentane. Both isomers are present in light gasoline
fractions and can be recovered individually by superfractionation (10,40), possi-
bly in combination with molecular sieve separation. n-Pentane is used as a sol-
vent and for the production of i-pentane by acid-catalyzed isomerization (13). i-
Pentane is used as a blending component for high octane gasoline and for the
production of isoprene by catalytic dehydrogenation (3,40).
n-Hexane. n-Hexane can be isolated from suitable sources (e.g, light
gasoline or BTX raffinates) by superfractionation, or by molecular sieve separa-
tion. It is used for the extraction of vegetable oils (e.g, from soybeans), as a
solvent in chemical reactions (e.g, for coordination complex catalyzed polymeri-
zation of olefins) and in adhesive formulations.
Higher n-Alkanes. n-Alkanes having more than six carbon atoms are
not isolated individually but as mixtures of several homologues. Suitable
sources are the appropriate petroleum distillate fractions, from which the n-
paraffins can be isolated in high isomeric purity ( 95% linearity) by selective
separation (e.g, molecular sieve separation or urea extractive crystallization).
They are used mainly in applications for which i-alkanes are not acceptable for
biological reasons, e.g, the production of detergents or proteins. Catalyzed gas-
phase dehydrogenation of n-alkanes over noble metal catalysts and at low con-
version (ca. 10%) yields the corresponding n-alkenes with predominantly inter-
nal double bonds. The alkenes can be isolated in high purity by selective
molecular sieve processes (22,41). n-Alkenes in the range C7 to C10 are used
as raw materials for the production of flexibilizers, by oxo-synthesis with
Vol. 2 HYDROCARBONS 603

hydrogen and carbon monoxide. n-Alkenes in the range C10 to C19 are used
for the production of biodegradable detergents: acid-catalyzed reaction of
C10- to C13-n-alkenes with benzene, followed by sulfonation and neutralization
of the ensuing alkylbenzene sulfonic acids, affords linear alkylbenzenesulfo-
nate (LABS) detergents. Oxo-reaction of C10- to C19-n-alkenes yields C11- to
C20-alcohols, which are converted into nonionic detergents by reaction
with ethylene oxide, or into anionic detergents by reaction with sulfur
trioxide or chlorosulfonic acid. Liquid-phase oxidation of n-alkanes with
oxygen in the presence of boric acid yields predominantly secondary alcohols.
The latter can be dehydrated to the corresponding n-alkenes. The C12- to
C19-alcohols are also used for the production of both ionic and nonionic
detergents.
Chlorination of n-alkanes yields mono- or polychlorinated alkanes, depend-
ing on the conditions. Mixtures of C10- to C13-monochloro-alkanes are used for
the production of linear alkylbenzenes as raw materials for LABS-detergents.
For this, the chloroalkanes are either directly reacted with benzene or they are
dehydrochlorinated to the corresponding n-alkenes, which are then used to
alkylate benzene. Polychlorinated n-alkanes are used in a number of special
applications.
n-Alkanes in the range of C10 to C21 are suitable substrates for the pro-
duction of single cell proteins by fermentation. The microorganisms used selec-
tively attack n-alkanes, so that industrial processes have been developed in
which either isolated n-alkanes or the appropriate petroleum distillates
containing n-alkanes (petroleum-, gas oil-, or lube oil fractions) are used as
starting materials.
Solid n-alkanes (paraffin waxes) are used in a variety of applications, both
as such and as feed materials in cracking, oxidation and chlorination.
Cycloalkanes. Of the various accessible cycloalkanes, only cyclohexane
and cyclododecane are used industrially on a large scale; cyclooctane is used
on a relatively small scale. The strained C3- and C4-cycloalkanes are not
produced industrially. Cyclopentane is present in natural hydrocarbon sources
and is formed as a byproduct in the processing of cracking products, e.g,
by hydrogenation of cyclopentadiene or cyclopentene in the C5-fraction
of steamcracked products. No important industrial use exists for cyclopentane.
Cyclooctane is accessible by catalytic hydrogenation of 1,5-cyclooctadiene.
Liquid-phase oxidation of the former with air yields 1,8-octanedicarboxylic acid,
which is used on a moderate scale for the production of nylon 68.
Cyclododecane is produced by liquid-phase hydrogenation of 1,5,9-cyclo-
dodecatriene over nickel catalysts. Liquid-phase oxidation of cyclododecane with
air in the presence of boric acid yields mixtures of cyclododecanol and cyclodode-
canone in high selectivity. Further oxidation of such alcohol – ketone mixtures
with nitric acid gives 1,12-dodecanedicarboxylic acid, used for the production
of polyamides, polyesters, and synthetic lubricating oils. Catalyzed dehydrogena-
tion of alcohol-ketone mixtures in the liquid phase yields cyclododecanone, which
is then converted into laurinlactam and nylon 12.
Bicyclo [4.4.0] decane (decalin) is obtained by catalytic hydrogenation
of naphthalene. Liquid-phase oxidation of decalin with oxygen gives the
604 HYDROCARBONS Vol. 2

corresponding tertiary hydroperoxide, which is eventually converted into


1,10-decanedicarboxylic acid (sebacic acid) by a four-step sequence (26).

2. Olefins

Olefins are aliphatic hydrocarbons containing at least one carboncarbon dou-


ble bond. Monoolefins (alkenes) contain a single C – C double bond and form a
homologous series with the empirical formula CnH2n. Hydrocarbons with two
double bonds are differentiated according to the relative positions of the
double bonds: in cumulated dienes (cumulenes, allenes) they are immediately
adjacent to each other; in conjugated dienes they are separated by a
single bond; and in other diolefins, the double bonds are isolated from each
other.
More highly unsaturated polyenes are known, e.g, trienes, tetraenes. The
name olefin is derived from the property of these compounds to form oily liquids
on reaction with halogens (gaz ole´fiant, oil-forming gas).
The industrial importance of olefins started in the 1950s when the lower
olefins became widely available from thermal cracking of wet natural gas and
petroleum fractions, displacing acetylene (ethyne) as the dominant commodity
chemical (see Acetylene). In the time since then, ethene (ethylene), propene,
1,3-butadiene, and isoprene (2-methyl-1,3-butadiene) have found wide applica-
tion, especially for synthetic polymers (see Butadiene; Ethylene; Isoprene).
2.1. Monoolefins. The low molecular mass monoolefins ethylene,
propene, styrene, cyclopentene, and the butenes are separate keywords:
(see Cyclopentadiene and Dicyclopentadiene; Ethylene). Among the higher ole-
fins (C6 – C18), the linear a-olefins are of particular industrial interest.
Properties. Physical Properties. Under normal conditions, the olefins
from ethylene to the butenes are gases; from the pentenes to 1-octadecene
liquids; and from 1-icosene onwards solids. These thermal phase transitions do
not differ significantly from those of the corresponding saturated hydrocarbons
(Chap. 1). The density of olefins ranges from 0.63 to 0.79 g/cm3, only a few percent
higher than those of the corresponding alkanes.
The heats of combustion of alkenes and the corresponding alkanes are
also nearly identical. Olefins have low solubility in water or are insoluble; they
Vol. 2 HYDROCARBONS 605

dissolve well in most organic solvents, e.g, alcohols, ethers, and aromatic
hydrocarbons. Some physical properties of important monoolefins are listed in
Table 6.
Chemical Properties. The chemistry of monoolefins is dominated by the
reactive double bond. Some important reactions of olefins are summarized in
Table 7.
Most addition reactions proceed by an electrophilic mechanism via carbo-
nium ions. Chlorination and bromination yields the trans adducts. Hydrohalo-
genation generally follows Markovnikov’s rule to give a product containing
halogen linked to secondary carbon. Addition of sulfuric acid to olefins affords
alkyl sulfates, which are readily hydrolyzed to alcohols; the reaction corresponds
to the acid-catalyzed addition of water. Ethanol, isopropylalcohol, and tert-butyl
alcohol are produced industrially by this method from ethylene, propene, and iso-
butene, respectively.
Hydrogenation of olefins to alkanes requires a catalyst: Raney nickel is
most frequently used industrially, although platinum-group metals are also
used.
Epoxidation of ethylene may be carried out with oxygen in the presence of a
silver catalyst. Higher olefins are converted to epoxides using hydroperoxides or
peracids. A related reaction is the hydroxylation of olefins with hydrogen perox-
ide in glacial acetic acid: an epoxide is initially formed, which subsequently
reacts with water in the acidic medium to form a glycol.
Another olefin oxidation is the Wacker oxidation of ethylene to acetalde-
hyde. The stoichiometric reaction of ethylene with water and palladium(II) chlor-
ide yields acetaldehyde and palladium metal. Addition of copper(II) chloride
converts the stoichiometric reaction into a catalytic process, wherein the
copper(II) chloride reoxidizes the palladium to palladium(II) chloride the result-
ing copper(I) chloride is reoxidized to copper(II) chloride by atmospheric oxygen
in hydrochloric acid solution.
Ozonolysis is used mainly for structure determination of olefins, but may
also be used preparatively. Reaction of olefins with carbon monoxide and addi-
tional reagents (e.g, hydrogen, water, alcohols) leads to a plethora of valuable
products (e.g, aldehydes, alcohols, carboxylic acids, esters). These reactions pro-
ceed under catalysis by transition metals, e.g, hydroformylations (Oxo process)
with rhodium- or cobalt carbonyls.
Sulfonation of olefins leads to alkenesulfonic acids, which are used as deter-
gents. Hydroboration is used to generate primary alcohols via trialkylboranes. In
contrast to the direct addition of water, this pathway is regioselective to yield
the anti-Markovnikov products.
Olefins may undergo a number of reactions resulting in new C-C bonds.
With aromatic compounds alkylaromatics are formed by the Friedel – Crafts
reaction; with isoalkanes chain-lengthened, branched hydrocarbons, which are
used in motor fuels, are formed. Olefins may also react with themselves or
each other to form oligomers and polymers.
The industrial use of the higher monoolefins is discussed in Section uses of
Higher Olefins.
Production of Higher Olefins. Olefins are synthesized on a laboratory
scale by a variety of routes: e.g, selective hydrogenation of alkynes; dehydration
Table 6. Physical Properties of Monoolefins
CAS registry
Olefin number mp, 8C bp, 13 8C d20 n20
Ethylene [74-85-1] 169.5 103.8 0.5699 1.363
(103.8 8C) (100 8C)
Propene [115-07-1] 185.3 47.7 0.6095 1.3640
(47.4 8C) (50 8C)
0.5139
1-Butene [106-98-9] 185.3 6.26 0.6356 1.3791
(15.1 8C) (25 8C)
cis-2-Butene [590-18-1] 138.9 3.7 0.6447 (0 8C) 1.3783
(0 8C)
trans-2-Butene [624-64-6] 105.5 0.88 0.6269 (0 8C) 1.3696
(20 8C)
Isobutene [115-11-7] 140.4 6.9 0.6398 (20 8C) 1.3776
1-Pentene [109-67-1] 167.2 29.97 0.6405 1.3718
cis-2-Pentene [627-20-3] 151.3 37.1 0.6554 1.3828
trans-2-Pentene [646-04-8] 140.2 36.4 0.6482 1.3798
2-Methyl-1-butene [58718-78-8] 137.5 31.2 0.6504 1.3778
3-Methyl-1-butene 168.5 20.1 0.6332 1.3640
2-Methyl-2-butene [26760-64-5] 133.6 38.6 0.6625 1.3874
1-Hexene [592-41-6] 140.0 63.5 0.6732 1.3878
cis-2-Hexene [7688-21-3] 141.8 68.8 0.6969 1.3976
trans-2-Hexene [4050-45-7] 133.5 67.5 0.6780 1.3935
cis-3-Hexene [7642-09-3] 137.8 66.5 0.6794 1.3951
trans-3-Hexene [13269-52-8] 113.4 68.0 0.6778 1.3937
1-Heptene [592-76-7] 119.6 93.8 0.6993 1.3998
cis,trans-2-Heptene 109.5 98.2 0.7034 1.4041
cis-3-Heptene [7642-10-6] 136.6 95.8 0.7030 1.4059
trans-3-Heptene [14686-14-7] 136.3 95.7 0.6981 1.4043
1-Octene [111-66-0] 102.7 121.3 0.7160 1.4086
cis-2-Octene [7642-04-8] 100.5 125.6 0.7243 1.4146
trans-2-Octene [13389-42-9] 87.8 124.5 0.7184 1.4132
cis-3-Octene [14850-22-7] 126 122.7 0.7223 1.4140
trans-3-Octene [14919-01-8] 110.0 123.3 0.7149 1.4129
cis-4-Octene [7642-15-1] 118.1 122.5 0.7212 1.4150
trans-4-Octene [14850-23-8] 93.8 122.3 0.7136 1.4120
1-Nonene [124-11-8] 81.6 146.8 0.7292 1.4157
2-Nonene [2216-38-8] 149.4 0.7407 1.4201
3-Nonene [20063-77-8] 148.2 (102 kPa) 0.7334 1.4191
4-Nonene [2198-23-4] 148.0 0.7313 1.4200
1-Decene [872-05-9] 66.6 170.6 0.7408 1.4215
2-Decene [6816-17-7]
3-Decene [19348-37-9] 173–173.5 0.7447 1.4221
4-Decene [19689-18-0] 170.6 0.7404 1.4243
cis-5-Decene [7433-78-5] 112 169.5 (98.5 kPa) 0.7445 1.4252
trans-5-Decene [7433-56-9] 73 170.2 0.7412 1.4235
1-Undecene [821-95-4] 49.1 192.67 0.75033 1.4261
5-Undecene [4941-53-1] 192.2 0.7511 1.4289
1-Dodecene [25378-22-7] 33.6 213.3 0.7583 1.4300
6-Dodecene [7206-29-3] 212.9 (99.3 kPa) 0.7597 1.4328
1-Tridecene [2437-56-1] 22.2 104 (1.33 kPa) 0.7654 1.4335
1-Tetradecene [1120-36-1] 12 125 (1.85 kPa) 0.7726 1.4367
1-Pentadecene [13360-61-7] 3.8 133.5 (1.33 kPa) 0.7763 1.4389
1-Hexadecene [629-73-2] 4.0 155 (2.0 kPa) 0.7811 1.4412
1-Heptadecene [6765-39-5] 10.7 159 (1.33 kPa) 0.7860 1.4436
1-Octadecene [112-88-9] 18.0 176 – 178 (1.85 kPa) 0.7888 1.4450
1-Icosene [3452-07-1] 28.5 151 (0.2 kPa) 1.4440
(30 8C)
1-Heneicosene [1599-68-4] 35.5 134 (0.005 kPa) 0.7985 1.4510
1-Triacontene [18435-53-5] 62–63 218 (0.07 kPa) 0.7620 (100 8C)
a
At 101.3 kPa, unless otherwise noted.
Vol. 2 HYDROCARBONS 607

Table 7. Important Reactions of Olefins


Reaction Reagents Products
Halogenation Cl2, Br2 dihaloalkanes
Hydro- HCl, HBr alkylhalides
halogenation
Sulfation H2SO4 alkyl sulfates
Hydration H2O; catalyst: H+ alcohols
Hydrogenation H2; catalyst: Ni, Pt alkanes
Epoxidation a) O2 catalyst: Ag ethylene
oxide
b) RCOOOH, ROOH epoxides
Hydroxylation H2O2; catalyst: H+ glycols
Wacker oxidation H2O; catalyst: acetaldehyde,
Pd – Cu ketones
Ozonolysis O3 aldehydes,
ketones
Hydro- CO, H2; catalyst: Co, aldehydes
formylation Rh
Hydrocar- CO, H2O; catalyst: carboxylic
boxylation Ni, Co acids
Sulfonation SO3, NaOH alkene-
sulfonates
Hydroboration B2H6 trialkyl-
boranes
Alkylation a) of aromatics, alkyl-
with a Friedel – Crafts aromatics
catalyst
b) of isoalkanes, branched
catalyst: H+ alkanes
Polymerization Catalyst: radicals, polymers
metal complexes
Oligomerization Catalyst: metal oligomers
complexes, BF3

of alcohols; pyrolysis of alkyl lithium compounds; the Wittig reaction; the


Peterson reaction; and deoxygenation of vicinal diols. Only processes starting
with cheap raw materials have gained industrial importance for the pro-
duction of higher olefins. The following discussion considers processes that
start either with long-chain alkanes (see Section Production from Paraffins)
or those that build up longer chain olefins from their lower molecular mass
homologues.
Production from Paraffins. Thermal Cracking of Waxes. Thermal
cracking of long chain alkanes yields mainly a-olefins.

R1  CH2 CH2 CH2  R2 !R1  CH ¼ CH2 þ R2  CH ¼ CH2 þ H2

Thermal cracking involves a radical mechanism. Cleavage of a C-C bond leads to


carbon radicals, which are converted into olefins by loss of a hydrogen atom. The
latter predominantly combine to form hydrogen; a small fraction reacts with
carbon radicals to form alkanes.
608 HYDROCARBONS Vol. 2

To generate linear a-olefins, linear alkanes (waxes) must be used. Suita-


ble sources are the petroleum, diesel oil, and lubricant fractions of paraffin-
based crude oils. One possibility of enriching the linear paraffin content of
such fractions is to dilute with a suitable solvent, followed by chilling to crys-
tallize a mixture of paraffins. Linear paraffins can be separated from their
branched and cyclic analogs by subsequent treatment with molecular sieves.
Another method for the separation of n-paraffins, especially suitable for long-
chain (>C20) species, is extractive crystallization with urea. This process uses
the property of urea to form inclusion compounds with linear paraffins, but not
with branched or cyclic alkanes, or aromatics. During crystallization the urea
molecules assemble to form cavities having a diameter of 0.53 nm, in which
only linear alkanes can be accommodated. The alkanes are then liberated by
subsequent decomposition of the complex, e.g, with water at 75 8C. Industrial
processes use solid urea (Nurex process) or its saturated aqueous solution
(Edeleanu process).
The long-chain paraffins are then cracked by first heating the vapor to
400 8C within a few seconds, followed by thermolysis at 500 – 600 8C in the pre-
sence of steam. Rapid quenching of the reaction mixture reduces side reactions
such as isomerization or cyclization.
Using C18 – C36 alkanes as starting material, a-olefins with even and odd
numbers of carbon atoms in the range C6 – C20 are obtained; only small amounts
of paraffins are present. Byproducts include nonterminal olefins, conjugated and
nonconjugated dienes, and traces of aromatics. Typical composition of the olefin
fractions from paraffin cracking are given in Table 8.
The dependence of product quality on the starting material and process
parameters is discussed in a review (42); catalytic cracking and cracking in the
presence of oxygen are also reviewed. The latter processes have not gained accep-
tance because of the presence of numerous byproducts, e.g, branched olefins and
oxidation products.
The first industrial paraffin cracking plant was brought into service in
1941 by Shell at Stanlow (United Kingdom). Other plants were built at Pernis
(The Netherlands) and Berre (France). A U.S. paraffin cracking plant owned
by Chevron is located in Richmond, California. Future expansion of paraffin
cracking processes is unlikely because of the limited flexibility of paraffin
cracking plants with respect to carbon number distribution, the limited supply
of paraffin-containing crudes, the high percentage ( 10%) of byproducts in the
a-olefin cut, and because of more efficient competitive processes.

Table 8. Composition of Fractions (wt %) Obtained in a Paraffin Cracking Process


(Chevron Process)
C6 – C7 C8 – C10 C11 – C12 C13 – C14 C15 – C18 C18 – C20
a-Olefins 83 83 88 88 89 86
Internal olefins 12 11 5 5 3 1
Diolefins 3 4 5 5 6 4
Paraffins 2 2 2 2 2 9
Aromatics 0.4 0.2 0 0 0 0
Vol. 2 HYDROCARBONS 609

Catalytic Dehydrogenation. Catalytic dehydrogenation of paraffins


leads to olefins with the same number of carbon atoms and with random location
of the double bond along the chain. Thermally, this reaction cannot be carried out
economically because the energy required to cleave a C-H bond (365 kJ/mol)
significantly exceeds that for breaking a C-C bond (245 kJ/mol). Cracking of
longer chain hydrocarbons is, therefore, favored over dehydrogenation; however,
the use of a catalyst facilitates the dehydrogenation of linear paraffins to linear
olefins having predominantly the same number of carbon atoms.
The most important industrial process is the Pacol (paraffin catalyst ole-
fin) process of Universal Oil Products (UOP). The reaction is carried out at
450 – 510 8C and 0.3 MPa using a hydrogen : carbon ratio of 9 : 1. A supported
platinum catalyst, ca. 0.8 wt% Pt, on alumina, activated by addition of lithium
and arsenic or germanium, is used. To minimize the formation of byproducts,
especially aromatic hydrocarbons, which reduce catalyst activity, relatively low
conversion rates of 10 – 15% are used. This results in paraffin – olefin mixtures
that are separated by means of the UOP Olex (olefin extraction) process. The
Olex process employs molecular sieves, which absorb olefins more strongly
than alkanes. The combination of both processes is known as the Pacol –
Olex process (43). In 1983, there were four Pacol – Olex plants worldwide, as
well as 18 other plants where the Pacol process was combined with the alkyla-
tion of aromatics (44). A simplified flow diagram of a Pacol benzene alkylation
unit is shown in Figure 3.
Fresh n-alkane is added to the circulating paraffin, heated via heat exchan-
gers, and dehydrogenated in the Pacol fixed-bed reactor (a) in the presence of
hydrogen and at high throughput velocity. The products are separated into
hydrogen and hydrocarbons in the separator (b) and volatile cracking products
are removed in the stripper (c); these are then used chiefly as fuel. The stripper
sump contains the higher linear monoolefins, which are reacted with benzene
under acid catalysis to yield alkylbenzenes. The unreacted n-alkanes are
returned to the Pacol reactor (45).

Fig. 3. Simplified flow diagram for a combination of the UOP PACOL process and
benzene alkylation. a) Pacol reactor; b) Separator; c) Stripper; d) Alkylation reactor.
610 HYDROCARBONS Vol. 2

Chlorination – Dehydrochlorination. Another method for generating


linear olefins from linear paraffins consists of chlorinating the paraffins to
chloroalkanes, followed by a catalyzed second step in which hydrogen chloride
is eliminated. Chlorination is carried out continuously in the liquid phase
at 120 8C; the conversion is limited to a maximum of 30%, to avoid excessive
formation of dichlorides. A random mixture of linear olefins results from the
dehydrohalogenation because the chlorine substituents are almost randomly dis-
tributed along the alkyl chain. The dehydrochlorination of chlorinated paraffins
occurs in the presence of iron or iron alloy catalysts at 250 8C. The lower boiling
olefins and paraffins are drawn off at the still head; the unreacted, higher boiling
chlorinated paraffins remain in the reactor. Separating products in this manner
requires that rather narrow cuts be taken, because the boiling points of the
n-olefins and the chloroparaffins are quite close together (46,47).
Oligomerization of Lower Olefins. All the methods described so far for
the production of olefins from paraffins suffer from the disadvantage that crude
oil containing linear paraffins with the required distribution of carbon atoms is
in short supply. Several companies have, therefore, developed alternative routes
to olefins that are based on oligomerization of olefins, especially of ethylene. Such
starting materials are readily available from the pyrolysis of liquefied natural
gas, naphtha, and gas oil fractions. Ethylene supplies appear assured for the for-
seeable future because it can be generated by the Mobil process from methanol,
which is derived from coal-based synthesis gas.
The methods for buildup of longer chain olefins vary greatly. In the follow-
ing sections syntheses via organoaluminum compounds, transition metal cata-
lyzed oligomerization, and acid-catalyzed processes will be discussed.
Ethylene Oligomerization in the Presence of Organoaluminum
Compounds. The first process for the industrial oligomerization of ethylene
was discovered by Ziegler in the early 1950s (48). The reaction proceeds in two
stages: first a growth step followed by an elimination step.

AlEt3 þ3nCH2 ¼ CH2 ! Al½ðCH2 CH2 Þn Et 3

Al½ðCH2 CH2 Þn Et 3 þ 3CH2 ¼ CH2 ! AlEt3 þ 3CH2 ¼ CHðCH2 CH2 Þn1 Et

The growth step occurs at ca. 100 8C under 10 MPa ethylene pressure. In
the second stage, the high temperature displacement reaction, the alpha-olefins
are displaced by ethylene at ca. 300 8C and 1 MPa. The composition of the product
mixture corresponds to a Poisson distribution. This process suffers from the dis-
advantage of requiring stoichiometric quantities of reaction components, i.e,
industrial plants would require large quantities of aluminum alkyls. Conse-
quently, Gulf Oil and Ethyl Corporation developed two process variants that
require less triethylaluminum.
The Gulf Oil Process. In the Gulf process, only catalytic quantities of
triethylaluminum are used and growth step and elimination reactions occur
simultaneously in the same reactor (49). The reaction is carried out at <200 8C
and ca. 25 MPa. The product mixture consists of a Schulz – Flory distribution of
linear olefins. A typical composition of individual fractions from C6 – C18 is listed
in Table 9.
Vol. 2 HYDROCARBONS 611

Table 9. Composition of Fractions (wt %) Obtained From the Gulf Process *


C6 C8 C10 C12 C14 C16 C18
Linear a-olefins 97.0 96.0 95.0 94.0 93.0 92.0 91.0
b-Branched a-olefins ** 1.4 2.5 3.4 4.6 5.6 6.6 7.8
* n-Alkane content 1.4 %.
** Including traces of olefins with internal double bonds.

The Gulf process produces markedly purer a-olefins than can be obtained
with paraffin cracking (see Table 8). The major impurities in the a-olefins
(trade name: Gulftenes) obtained by the Gulf process are ca. 1.4% paraffins;
with increasing carbon number, an increasing amount of b-branched a-olefin is
also formed. The occurrence of these vinylidene and 2-ethyl compounds may be
rationalized by consecutive reactions of a-olefins with the aluminum alkyls, as
follows:

The Gulf Oil Co. uses the Gulf process to produce linear a-olefins in the
C4 – C30 range at their plant in Cedar Bajou, Texas. Mitsubishi Chemical Indus-
tries uses the same process at Mizushima.
Ethyl Process. The Ethyl process combines a stoichiometric and a catalytic
stage (50). This modified Ziegler process is capable of controlling the chain length
of the resulting a-olefins much more precisely, because the shorter chain olefins
may be subjected to further chain growth.
The Ethyl process is illustrated schematically in Figure 4. Ethylene is first
oligomerized using catalytic quantities of triethylaluminum in analogy to the
Gulf process. The resulting products are fractionated (b) in Fig. 4) yielding frac-
tions containing C4, C6 – C10, and C12 – C18. The higher boiling fractions may be
used directly, whereas the shorter a-olefins, especially C4, are subjected to trans-
alkylation with long chain aluminum alkyls (c) in Fig. 4). This reaction releases
the desired long-chain a-olefins; short-chain trialkylaluminum compounds are
simultaneously formed. The aluminum trialkyls are separated in the second dis-
tillation step (Fig. 4, (d)) and transformed in another reactor (e) in Fig. 4) into
longer chain alkyls by stoichiometric reaction with ethylene. The latter are
then recycled to the transalkylation.
The Ethyl Corporation has operated a plant in Pasadena, Texas using this
process since 1970; the initial capacity was 110 000 t/a. The process permits 95%
conversion of ethylene into higher a-olefins, a conversion efficiency not achiev-
able by the single-step catalytic Gulf Oil process. This advantage is however,
counterbalanced by a much lower product quality of the long-chain a-olefins.
The C16 – C18 cut consists of only 63% linear a-olefins (Table 10), the remainder
being mainly b-branched a-olefins and internal olefins.
612 HYDROCARBONS Vol. 2

Fig. 4. Typical flow diagram of Ethyl process. A) Catalytic step; B) Stoichiometric step;.
a) Catalytic reactor; b) First distillation; c) Transalkylation reactor; d) Second distillation;
e) Stoichiometricreactor.

Ethylene Oligomerization with Transition-Metal Catalysis. In the


presence of nickel, cobalt, titanium, or zirconium catalysts, ethylene may be con-
verted into oligomers (51). Commercial importance has been attained exclusively
by the Shell higher olefin process (SHOP).
Shell Higher Olefin Process. The SHOP process is the most recent devel-
opment in a-olefin synthesis. It works in the liquid phase using a nickel catalyst
and yields a-olefins of unusually high purity (52–54). Monoolefins are formed
almost exclusively; analytical data indicate only trace amounts of dienes, aro-
matics, and alkanes. The content of a-olefin for the carbon numbers is in the
range 96 – 97%. A typical product composition is given in Table 11.
The SHOP process involves a combination of three different reactions: oli-
gomerization, isomerization, and metathesis. This integration of chain lengthen-
ing and shortening steps makes the SHOP process very flexible. Whereas all
other a-olefin processes permit only limited control over product distribution,
with the SHOP process it is possible to achieve almost any olefin cut desired.
The SHOP process is shown schematically in Figure 5 (54–57).
Oligomerization is carried out in a polar solvent such as ethylene glycol or
1,4-butanediol. The catalyst is produced in situ from a nickel salt, e.g, nickel
chloride, sodium borohydride, and a chelating ligand. Suitable ligands are
compounds of the general formula RR1PCH2COR2, e.g, diphenyl phosphinoa-
cetic acid, dicyclohexylphosphinoacetic acid, or 9-(carboxymethyl)-9-phosphabi-
cyclo [3.3.1]- nonane. Ethylene is oligomerized to a-olefins with a Schulz – Flory

Table 10. Composition of Fractions (wt%) Obtained From the Ethyl Process
C6 C8 C10 C12 C12 – C14 C14 – C16 C16 – C18
Linear a-olefins 97.5 96.5 96.2 93.5 87.0 76.0 62.7
Internal olefins 0.6 1.2 1.6 1.5 4.2 5.0 8.2
b-Branched a-olefins 1.9 2.3 2.2 5.0 8.8 19.0 29.1
Alkanes 0.1 0.6 0.3 0.4 0.4 0.4 0.8
Vol. 2 HYDROCARBONS 613

Table 11. Composition of a-Olefin Fractions (wt%) Obtained From the SHOP Process
C6 C8 C10 C12 C14 C16 C18
All monoolefins 99.9 99.9 99.9 99.9 99.9 99.9 99.9
Linear a-olefins 97.0 96.5 97.5 96.5 96.0 96.0 96.0
Branched olefins 1.0 1.0 1.0 2.0 2.5 2.5 2.5
Internal olefins 2.0 2.4 1.0 1.5 1.5 1.5 1.5
Dienes <0.1 <0.1 <0.1 <0.05 <0.05 <0.05 <0.05
Aromatics <0.1 <0.1 <0.1 <0.05 <0.05 <0.05 <0.05
Alkanes <0.1 <0.1 <0.1 <0.05 <0.05 <0.05 <0.05

distribution at 80 – 120 8C and 7 – 14 MPa. The olefins formed are immiscible


with the polar solvent; product and catalyst phases are thereby readily separated
so that the catalyst can be recycled repeatedly. Oligomerization is accomplished
in a series of reactors interspersed with heat exchangers to remove the heat of
reaction. The reaction rate can be regulated by the amount of catalyst. A high
partial pressure of ethylene is required for high product linearity and a suitable
rate of reaction. The chain length of the a-olefins is determined by the geometric
factor K of molar growth:

nðCnþ2  olefinsÞ

nðCn  olefinsÞ

where n is the number of moles.


The mass distribution of various a-olefins relative to the growth factor K is
shown in Figure 6.
Controlling the growth factor is extremely important for the overall process
because K determines both the product distribution and the median chain length
of the olefins obtained. It is, therefore, advantageous to be able to vary the
K-factor by the composition of the catalyst.
After separation of the catalyst and product phase, the latter is washed with
fresh solvent to remove the last traces of catalyst. In a subsequent distillation

Fig. 5. Typical flow diagram of SHOP process. a) Oligomerization reactor; b) Phase


separator; c) Distillation; d) Isomerization reactor; e) Metathesis reactor.
614 HYDROCARBONS Vol. 2

Fig. 6. Product distribution as a function of growth factor K.

(see Fig. 5) the mixture is separated into the desired C12 – C18 a-olefins and into low
(C4 – C10) and high-boiling (>C20) fractions. The C4 – C10 fraction can then be
further fractionated, if desired, into individual compounds, for example, those
used as comonomers for the manufacture of linear low density polyethylene
(LLDPE). However, the C4 – C10 fraction and the C 20 fractions are generally
combined for isomerization and subsequent metathesis. These two steps require
only moderate reaction conditions: 80 – 140 8C, 0.3 – 2 MPa. The heat of reaction
of both steps is quite low; because in isomerization only intramolecular migration
of double bonds occurs, in metathesis carbon – carbon bonds are simultaneously
cleaved and reformed. Isomerization takes place in the liquid phase in the presence
of magnesium oxide catalyst (58). About 90% of the terminal olefins are converted
to internal olefins in this reaction.

The subsequent joint metathesis of the internal, low, and high molecular
mass olefins, mainly over heterogeneous rhenium or molybdenum catalysts,
yields a mixture of internal olefins having a new chain length distribution, i.e,
a random mixture of olefins of even and odd numbers of carbon atoms.

The products of metathesis are distilled and the C11 – C14 fraction, repre-
senting ca. 10 – 15% of the mixture, is separated for production of alcohols
used in the detergent industry (see Section Uses of Higher Olefins).
The composition of the internal C11 – C14 olefins produced by the SHOP pro-
cess is given in Table 12; more than 99.5% is monoolefins, with only small
amounts of dienes, aromatics, and paraffins.
Although the double bonds of the olefins at the beginning of metathesis are
distributed almost at random over the carbon backbone, they are shifted in the
Vol. 2 HYDROCARBONS 615

Table 12. Composition of Internal Olefins (wt%) Obtained


From the SHOP Process
C11 – C12 C13 – C14
All monoolefins >99.5 >99.5
Linear olefins >96 >96
Branched olefins 3 3
Dienes <0.1 <0.1
Aromatics <0.1 <0.1
Paraffins <0.5 <0.5

end product toward the chain end, a result of the high concentration of
short-chain olefins (2-butene, 2-hexene, 3-hexene) in the metathesis feed.
Distillation of the metathesis products also yields low- and high-boiling
components in addition to the C11 – C14 fraction. One of the great advantages
of the SHOP process is that these fractions do not have to be discarded but
can be recycled. The fraction consisting of compounds with less than 11 carbon
atoms is returned directly to the metathesis reactor; the fraction containing ole-
fins with more than 14 carbon atoms is again subjected to isomerization before it
reaches the metathesis stage. By combining oligomerization, isomerization, and
metathesis, the SHOP process is capable of converting almost all the ethylene
feed into olefins having the desired number of carbon atoms.
An interesting alternative of the SHOP process described above is the
metathesis of the isomerized C 20 fraction together with a ten-fold molar excess
of ethylene (58). Olefins, predominantly a-olefins, within the desired carbon
range may be prepared in the presence of a rhenium catalyst (20% Re2O7 on alu-
minum oxide).

The first SHOP processing plant was built in 1977 by Shell Oil Corporation
at Geismar, Louisiana.
The mechanism of the highly selective ethylene oligomerization with nickel
chelate catalysts has been largely elucidated by the research group of W. KEIM
(52,59–63). A chelated nickel hydride complex is formed initially. This complex
reacts with ethylene (Fig. 7) to form alkyl nickel complexes. Linear a-olefins
are eliminated from the alkyl nickel complexes; the nickel hydride species is
formed simultaneously to reenter the catalytic cycle.
The SHOP process is by far the most important method for oligomerization
of ethylene involving transition-metal catalysts. To complete the picture, two
other processes will be described below, although they have not yet reached com-
mercial importance.
Exxon Process. The Exxon Corporation has developed a process in
which ethylene is converted to high molecular mass olefins in the presence of a
catalyst consisting of a transition-metal component and a soluble alkyl alumi-
num chloride (64). Titanium chloride or an alkoxy titanium chloride are favored
616 HYDROCARBONS Vol. 2

Fig. 7. Mechanism of ethylene oligomerization in the presence of nickel PO-chelate


complex catalyst (for PO ligand see text).

as the transition metal components. The catalyst is soluble in saturated hydro-


carbons and active at 25 – 70 8C. This process yields linear a-olefins in the range
C4 – C100 or higher; it appears to be especially useful for the production of
high-melting waxes, i.e, olefins higher than C20.
Alphabutol Process. In the Alphabutol process, developed by the Insti-
tut Français du Pétrole (IFP), ethylene is selectively dimerized to 1-butene (65)
using a homogeneous titanium catalyst. In contrast to the SHOP process, the
mechanism does not involve metal hydrides but a titanium(IV) cyclopentane
species that decomposes under the reaction conditions (50 – 60 8C, slight positive
pressure) with b-hydrogen transfer to 1-butene:

The 1-butene, which contains only traces of impurities (hexenes, 2-butene,


butanes) may be used as comonomer in the production of LLDPE.
Oligomerization of Propene and Butenes with Transition-Metal
Catalysis. Ziegler catalysts consisting of nickel compounds and aluminum
trialkyls can be used to dimerize or codimerize propene and butene to give pre-
dominantly branched olefins. This process was also developed by the Institut
Français du Pétrole and is referred to as the Dimersol process (66–71). IFP
research began as early as the mid 1960s. The first commercial Dimersol-G
plant was brought into production in 1977 by Total in Alma, Michigan. In the
Dimersol G process, propene (in admixture with propane) is dimerized mainly
(80%) to isohexenes; small amounts of propene trimers (18%) and tetramers
(2%) are also formed:
Vol. 2 HYDROCARBONS 617

The reaction takes place in the liquid phase without solvent. The pressure
in the reactor is adjusted so that the C3 compounds remain liquid at the operat-
ing temperature. The product mixture leaving the reactor head is fed to a vessel
where the catalyst is neutralized with alkali. The product (trade name: Dimate)
is then washed with water and freed of any remaining propane and propene on a
stabilizing column.
The Dimersol X process is similar, although the starting materials are
butenes or a mixture of propene and butenes. The products formed are, accord-
ingly, isooctenes or mixtures of C6 – C8 olefins. The first Dimersol X plant was
started in 1980 by Nissan Chem. in Kashima, Japan. A typical starting material
for the Dimersol X process is the raffinate II obtained from the C4 cut of the
steam cracker, after removal of butadiene and isobutene. The resulting C8 olefins
can be converted by Oxo-synthesis into nonanols, which are used for the synth-
esis of plasticizers (dinonyl phthalate).
A more recent development is the Dimersol E process, which employs the
exhaust gases of fluid catalytic crackers (FCC) that contain ethylene and pro-
pene. A gasoline fraction can be obtained from this product after hydrogenation.
The first experiments with cracker exhaust gases from the Elf cracker near Fey-
zin (France) began in 1984.
By 1982, IFP had licensed 16 Dimersol plants, eight of which were in pro-
duction. In addition to IFP, other industrial and academic groups have investi-
gated the transition metal-catalyzed oligomerization of lower olefins (72–80). The
observation that in the propene dimerization with nickel-based Ziegler catalysts
the addition of phosphines may significantly affect the isohexene product distri-
bution is of interest. With the catalyst [NiBr(Z3-C3H5)(PCy3)] – EtAlCl2 (Cy = cy-
clohexyl) a remarkable turnover of 800 000 mol propene per mol nickel h1 could
be achieved. A turnover of 60 000 s1 was calculated for operation at room tem-
perature. This value is of an order of magnitude usually achieved only by enzyme
reactions.
Oligomerization of Propene and Butene by Acid Catalysis. Mineral
acids such as phosphoric or sulfuric acid can also be used as catalysts for the
oligomerization of olefins; the major products are branched olefins. Alkylben-
zenes were once prepared from tetramers of propenes (tetrapropylene). This
reaction is now rarely used in Europe because of the low biodegradability of
the alkylbenzenesulfonates obtained.
Oligomerization of isobutene leads to the formation of diisobutene, triisobu-
tene, and tetraisobutene; after hydrogenation, these may be added to gasoline to
improve the octane rating (polymer gasoline, oligomer gasoline). In the cold acid
process, isobutene-containing C4 fractions are treated with 60 – 70% sulfuric acid
at room temperature. Isobutene dissolves in the sulfuric acid as tert-butyl sul-
fate. This solution is then oligomerized at 100 8C. The products, mainly diisobu-
tenes, separate as a second phase. In the hot acid process, the n-butenes of the C4
fraction are also converted, forming i-C4 – n-C4 cooligomers.
The oligomerization can also be performed with poly(styrene sulfonic
acid) resins as catalysts. This process is used by Erdölchemie – Dormagen
using the raffinate I (C4 fraction after extraction of butadiene). The reaction
is performed in loop reactors at 100 8C and 1.5 – 2.0 MPa with residence
times of 20 – 60 min. The product is composed of 36% diisobutene, 22% C8
618 HYDROCARBONS Vol. 2

codimers, 38% triisobutene, and 4% tetraisobutene. The product distribution


may be changed by recycling specific fractions.
Uses of Higher Olefins. The most important uses of linear olefins are
described below.
Synthesis of Oxo-Alcohols. Linear a-olefins can be converted to linear
primary alcohols, which are used in the synthesis of plasticizers (<C11) or deter-
gents (>C11) by hydroformylation (oxo synthesis). The reaction can also be used
to give aldehydes as the major products.

The yield of linear products can be increased to >90% if phosphine-modified


cobalt- or rhodium catalysts are used.
The described synthesis route for long-chain alcohols has strong competi-
tion, however, from the natural fatty alcohols and from the Alfoles, generated
from ethylene by an organo aluminum synthesis with subsequent oxidation.
In the United States Monsanto has produced oxo alcohols from C6 – C10
a-olefins since 1971. Exxon has used the oxo process since 1983 to produce plas-
ticizer alcohols in the C7 – C11 range at Baton Rouge, Louisiana. Shell also con-
verts olefins to alcohols. In recent years, olefins derived from paraffin crackers as
starting materials have been replaced by internal olefins from the SHOP process.
The possibility of converting olefin-derived aldehydes to short-chain
synthetic fatty acids by oxidation is also notable:

In 1980 Celanese (now Hoechst-Celanese) built a plant in Bay City, Texas


with an annual capacity of 18,000 t to produce heptanoic and nonanoic acids from
1-hexene and 1-octene, respectively, which are used in the production of lubri-
cants. Again, there is strong competition from the natural products, which are
generally less expensive. In addition, lubricant production requires highly linear
acids. Fatty acids from oxo aldehydes have a linearity of >95%; natural fatty
acids have a linearity of almost 100%.
Synthesis of Linear Alkylbenzenes. Currently the most important
class of detergents for domestic use is still the linear alkylbenzenesulfonates
(LAS or LABS). They are prepared by Friedel – Crafts reaction of benzene with
linear olefins followed by sulfonation:

In Europe, Asia, South Africa, and South America linear alkylbenzenes are
produced from C10 – C15 alpha-olefins. In 1982, 420 000 t of LAB were synthe-
sized in Western Europe from ca. 80 000 t of linear olefins.
Vol. 2 HYDROCARBONS 619

Synthesis of a-Alkenesulfonates. a-Alkene sulfonates (a-olefin sulfo-


nates, AOS or OS) are obtained by direct sulfonation of C14 – C18 a-olefins. An
ionic mechanism results in a product mixture of 40% isomeric alkenesulfonic
acids (with the double bond predominantly in the b- or the g-position) and
60% 1,3- or 1,4-alkanesulfones. Following hydrolysis with sodium hydroxide, a
mixture of alkenesulfonates and hydroxyalkanesulfonates is obtained:

The double bonds and the hydroxyl groups contribute to the excellent water
solubility of the AOS compounds. They are effective laundry agents even in hard
water and at low concentration and are used increasingly in liquid soaps and cos-
metics. In order to be used in such products the AOS must be of high quality,
especially with regard to color; this is only possible with a-olefins that contain
no more than traces of dienes.
Although AOS compounds have been known since the 1930s, they gained
commercial importance only in the late 1960s. In AOS represents an interesting
alternative to alkyl sulfates and alkyl ether sulfates in cosmetic products.
Synthesis of Bromoalkanes and Derived Products. An important
reaction of a-olefins is radical hydrobromation to give primary bromoalkanes.
These compounds are valuable intermediates for, e.g, thiols, amines, amine
oxides, and ammonium compounds:

Thiols are produced by this route in Europe and in the United States by
Pennwalt Corp. and Phillips Petroleum Co. They are used in the synthesis of her-
bicides, pesticides, pharmaceuticals, in the textile industry, or as polymer regu-
lators.
In the U.S. alkyldimethylamine (ADA) from linear a-olefins were directly
converted to fatty amine oxides (FAO) or quaternary ammonium salts (Quats).
Fatty amine oxides are nonionic surfactants that, because they are biodegrad-
able and do not irritate the skin, are incorporated into numerous products,
e.g, into rinsing agents. Quats are produced by alkylation of amines, e.g, with
620 HYDROCARBONS Vol. 2

benzylchloride. Benzyl-Quats possess biocidal and antimicrobial properties and


are incorporated, for example, into disinfectants and deodorants.
Production of Synthetic Lubricants Oligomerization of a-olefins in
the middle C-number range, in particular 1-decene, to isoparaffins (mostly tri-
mers with a few tetramers and pentamers) leads to lubricants known as
poly(a-olefins) (PAO) or synthetic hydrocarbons (SHC). They are used indust-
rially, and in the automobile and aircraft sectors (81,82). They are in many
instances superior to natural mineral oils, especially at lower temperature,
because of their favorable viscosity, low volatility, high flash point, and high
thermal stability. Oligomerization of the a-olefins can be performed with, e.g,
BF3, to which protonic cocatalysts, e.g, water or alcohols, are added.
Production of Copolymers. Linear a-olefins, in particular 1-butene,
1-hexene, and 1-octene, are used as comonomers in the production of high
density polyethylene (HDPE) and linear low density polyethylene (LLDPE). By
adding a-olefins the density and other properties of the polymers may be signifi-
cantly modified.
HDPE copolymers contain only relatively small quantities of a-olefins,
generally 0.5 – 3%. The density of the HDPE homopolymer (0.965 – 0.955 g/
cm3) is reduced to 0.959 – 0.938 g/cm3 for the HDPE copolymer. The most fre-
quently employed comonomer is 1-hexene, followed by 1-butene.
On the other hand, 4 – 12% a-olefin is generally added to LLDPE to reduce
the density to 0.935 – 0.915. The more a-olefins added to the polymer, the greater
the number of short-chain branches and the lower the density. Addition of
1-butene is preferred in many of the gas-phase processes, e.g, in the Unipol pro-
cess introduced by Union Carbide in 1977 (83), whereas 1-octene and 1-hexene
are preferred in liquid-phase processes.
Because Shell began marketing SHOP olefins in Western Europe, the use of
HDPE and LLDPE copolymers has increased in these countries. Olefins derived
from paraffin cracking plants are unsuitable as comonomers for polyethylene
because of the impurities they contain.
Additional Reactions of a-Olefins. The reaction of a-olefins with
peracid to form epoxides is an interesting route to bifunctional derivatives
(84). Some examples are given below.
Vol. 2 HYDROCARBONS 621

a-Olefin epoxides have also been used in polymer chemistry, e.g, for the
modification of epoxy resins. United States producers of a-olefin epoxides
include: Union Carbide, Viking, and Dow Chemical.
Another use of a-olefins is conversion into secondary monoalkylsulfates,
which are readily degradable surfactants useful in wetting agents and household
detergents. They are synthesized by reaction with concentrated sulfuric acid at
low temperature (10 – 20 8C) and with short reaction times (5 min):

Hydrocarboxylation of a-olefins leads to odd-numbered carboxylic acids.


This reaction is carried out with carbon monoxide and water at 200 8C and
20 MPa. The formation of linear carboxylic acids is favored in the presence of
cobalt carbonyl – pyridine catalysts. Both the free acids and their esters are
used as lubricant additives.

The alkenylsuccinic anhydrides (ASA) should also be mentioned. They are


obtained by heating maleic anhydride with a- or internal olefins, and are used as
lubricant additives, detergents, and in leather and paper production.
2.2. Dienes and Polyenes. Low Molecular Mass 1,3-Dienes. In-
dustrially, the most important 1,3-dienes are butadiene, isoprene, and cyclopen-
tadiene (see BUTADIENE; CYCLOPENTENE AND DICYCLOPENTADIENE). The conjugated
dienes listed below are of lesser importance.
Piperylene. (1,3-pentadiene) is present at a level of >10% in the C5
steam cracker fraction; it is expensive to separate from the C5 cut. Polymers
and copolymers with, for example, butadiene have been produced but have not
yet found a technical application.
2,3-Dimethylbutadiene was the first monomer to be converted on an
industrial scale to a synthetic rubber, referred to as methyl rubber. It was devel-
oped by HOFFMANN in 1910 and produced during 1914 – 1918. It suffered, however,
from several disadvantages (excessive hardness, sensitivity to oxidation) and later
lost its importance. The early synthesis of the monomer involved conversion of
acetone to pinacol, which is then dehydrogenated to the diene in the presence
of aluminum oxide as catalyst. Modern synthesis starts with propene, which is
dimerized to 2,3-dimethylbutene (79,80) and subsequently dehydrogenated:
622 HYDROCARBONS Vol. 2

Chloroprene, CH2=CClCH=CH2, 2-chloro-1,3-butadiene, is used to form


polychloroprene (neoprene), which is extremely resistant to oil, abrasion, and
ageing. As with all polymers containing a large proportion of halogen, these pro-
ducts are resistant to ignition. Vulcanized polychloroprene is used in, for exam-
ple, conveyer belts, cable shielding, and protective clothing. Nonvulcanized
polymers are used as contact adhesives in plastics manufacturing.
Chloroprene can be produced by two routes. In the older process, acetylene
is first dimerized to vinyl acetylene and hydrogen chloride then added to the tri-
ple bond; both steps take place in the presence of CuCl as catalyst. The second
process involves chlorination of butadiene, at 300 8C with chlorine to yield a
mixture of dichlorobutenes. The 1,2-adduct is subsequently dehydrochlorinated
to yield chloroprene. The 1,4-adduct can be isomerized to the 1,2-adduct:

Synthesis of Dienes and Polyenes by Oligomerization. A multitude of


linear and cyclic dienes and polyenes can be synthesized by transition metal-cat-
alyzed oligomerization or by cooligomerization (51). Only a few examples will be
described below, some of which are used commercially.
Butadiene can be converted into six cyclooligomers in a nickel-catalyzed
reaction (85). The synthesis can be steered in the desired direction by a suitable
choice of catalyst.

1,5-Cyclooctadiene is produced by Shell and by Hüls and is used, for


example, as starting material for suberic acid and poly(octenamer).
1,5,9-Cyclododecatriene is produced by DuPont, Shell, and Hüls using
nickel or titanium as catalyst; it is used in the synthesis of nylon 12, dodecandioic
acid, and hexabromocyclododecane.
Vol. 2 HYDROCARBONS 623

4-Vinylcyclohex-1-ene is also accessible from butadiene but is, as yet, of lit-


tle commercial importance. The possibility of using it as a precursor in styrene
synthesis has been discussed.
An example of catalytic cooligomerization is the reaction between butadiene
and ethylene (86).

Whereas the cyclic 1 : 1 product vinylcyclobutane and the C10 oligomers cyclode-
cadiene and 1,4,9-decatriene are not yet used commercially, the linear oligomer,
1,4-hexadiene, is produced as a monomer for the production of ethylene – pro-
pene – diene (EPDM) elastomers. Homogeneous catalysis with rhodium or palla-
dium complexes leads to trans-1,4-hexadiene; cobalt and iron catalysts yield
predominantly the cis-1,4-product. Rhodium catalysts are used in commercial
production.
Ethylidenenorbornene, also a comonomer for EPDM polymers, is also acces-
sible via cooligomerization. One of the routes employs the Diels – Alder reaction
starting with butadiene and cyclopentadiene, the other couples norbornadiene
and ethylene in the presence of a nickel catalyst. Vinylnorbornene is formed as
an intermediate in both routes.

Synthesis of Dienes and Polyenes by Metathesis. Metathesis offers a sim-


ple path to a,o-diolefins that are not easily obtained by other routes (87). Starting
materials are cycloolefins, which can be prepared, for example, by cyclooligomer-
ization of butadiene (see Section Syntheis of Dienes and Polyenes by Oligomer-
ization) and subsequent partial hydrogenation. Metathetic conversion with
ethylene opens the cycloolefin to an a,o-diene. In this manner, 1,5-cyclooctadiene
gives two molecules of 1,5-hexadiene; cyclooctene forms 1,9-decadiene; and
624 HYDROCARBONS Vol. 2

cyclododecene leads to 1,13-tetradecadiene.

Heterogeneous rhenium catalysts are preferred because they allow the


reaction to proceed under very mild conditions (5 – 20 8C, 0.1 – 0.2 MPa ethy-
lene). A plant with an annual capacity of 3000 t using this FEAST (Further
Exploitation of Advanced Shell Technology) process was brought on stream
by Shell in Berre (France) in 1987. The products are suitable for the synthesis
of numerous fine chemicals, e.g, production of long-chain diketones, for
dibromides, diepoxides, dialdehydes, dithiols, and diamines.
4-Vinylcyclo-1-hexene undergoes self metathesis. Elimination of ethylene
yields 1,2-bis(3-cyclohexenyl)ethylene, which has been used as a starting
material for fire retardants.

3. Alkylbenzenes

See separate articles for Toluene and Xylenes.


3.1. Trimethylbenzenes. Properties. The three isomers of trimethyl-
benzene are colorless liquids. Some of their physical properties are listed in
Table 13.

The methyl groups of trimethylbenzenes are converted to carboxyl groups


on oxidation. Dilute nitric acid oxidizes mesitylene to the corresponding
Vol. 2 HYDROCARBONS 625

tricarboxylic acid, trimesic acid [554-95-0]. In pseudocumene the preferred site of


reaction for halogenation, sulfonation, and nitration is the 5-position; disubstitu-
tion leads to 3,5-derivatives. Mild oxidation of pseudocumene or mesitylene, with
manganese- or lead dioxide, or electrolytically, leads to 2,4- or 3,5-dimethylben-
zaldehyde, respectively.
Production. Hemimellitene, pseudocumene, and mesitylene are found in
coal tar oil. The compounds can be isolated by distillation and purified by the dif-
ferential hydrolysis of their sulfonic acids (89). More important industrially is
the occurrence of trimethylbenzenes in some mineral oils, and their formation
during processing of crude oil, especially catalytic cracking and reforming. Pseu-
documene is obtained in 99% purity by superfractionation from the C9 cut of
crude oil refining (90).
Pseudocumene and hemimellitene can be obtained by extractive crystalliza-
tion of the C9 cut with urea (91). Trimethylbenzenes can be produced by
Friedel – Crafts methylation of toluene or xylene with chloromethane in the
presence of aluminum chloride. Trimethylbenzenes, especially pseudocumene
and mesitylene, are also formed by liquid-phase disproportionation of xylene in
the presence of aluminum chloride. These reactions can be carried out in the gas
phase at high temperature on aluminum silicate catalysts.
A more recent method for the methylation of xylene with methanol or
dimethyl ether on fluorine-containing crystalline borosilicate, aluminosilicate,
or boroaluminosilicate (ZSM-5) at 300 8C and atmospheric pressure yields pseu-
documene and durene almost exclusively, both of which may be isolated at >99%
purity (92). Trimethylbenzenes are also generated in the reaction of toluene with
synthesis gas over a dual catalyst consisting of a metal oxide component and an
aluminum silicate, at 200 – 400 8C and up to 20 MPa (93). Pseudocumene is also
generated from methanol as a byproduct of the Mobil MTG (methanol to gaso-
line) process (see Section Tetramethylbenzenes). Fusel oils and other fermenta-
tion products may be reacted on the zeolite catalyst HZSM-5 to form aromatic
compounds, including trimethylbenzenes (94).
Uses. Mesitylene is used in the synthesis of dyes and antioxidants, and as
a solvent. Pseudocumene is a starting material for dyes, pharmaceuticals and,
especially, for trimellitic anhydride [552-30-7], which is then converted to heat-
resistant polyamideimides and polyesterimides. Hemimellitene is an intermedi-
ate in the production of fragrances.
3.2. Tetramethylbenzenes. Properties. Some physical properties of
the three tetramethylbenzene isomers are listed in Table 13. Prehnitene and
isodurene are colorless liquids; durene forms colorless, monoclinic crystals
that sublime slowly at room temperature and that have an odor reminiscent
of camphor.
626 HYDROCARBONS Vol. 2

Table 13. Physical Properties of Alkylbenzenes (79)


CAS
registry
3
Compound number Mr mp, 8C bp, 8C d20 20
4 , g/cm nD
1,2,3-Trimethylbenzene, [526-73-8] 120.19 25.4 176.1 0.8944 1.5139
hemimellitene
1,2,4-Trimethylbenzene, [95-63-6] 120.19 43.8 169.3 0.8758 1.5048
pseudocumene
1,3,5-Trimethylbenzene, [108-67-8] 120.19 44.7 164.7 0.8652 1.4994
mesitylene
1,2,3,4-Tetramethylbenzene, [488-23-3] 134.22  6.2 205 0.9052 1.5203
prehnitene
1,2,3,5-Tetramethylbenzene, [527-53-7] 134.22 23.7 198 0.8903 1.5130
isodurene
1,2,4,5-Tetramethylbenzene, [95-93-2] 134.22 79.2 196–198 0.8380a 1.4790a
durene
Pentamethylbenzene [700-12-9] 148.25 54.5 232 0.917 1.527
Hexamethylbenzene, [87-85-4] 162.28 166–167 265 1.0630b
mellitene
1,2-Diethylbenzene [135-01-3] 134.22 31.2 183.5 0.8800 1.5035
1,3-Diethylbenzene [141-93-5] 134.22 83.9 181.1 0.8602 1.4955
1,4-Diethylbenzene [105-05-5] 134.22 42.8 183.8 0.8620 1.4967
1,2,3-Triethylbenzene [42205-08-3] 162.27
1,2,4-Triethylbenzene [877-44-1] 162.27 217.5c 0.8738 1.5024
1,3,5-Triethylbenzene [102-25-0] 162.27 66.5 215.9 0.8631 1.4969
1,2,3,4-Tetraethylbenzene [642-32-0] 190.33 11.8 251d 0.8875 1.5125
1,2,3,5-Tetraethylbenzene [38842-05-6] 190.33 21 247d 0.8799 1.5059
1,2,4,5-Tetraethylbenzene [635-81-4] 190.33 10 250 0.8788 1.5054
Pentaethylbenzene [605-01-6] 218.38 <20 277 0.8985e 1.5127
Hexaethylbenzene [604-88-6] 246.44 129 298 0.8305f 1.4736f
1-Ethyl-2-methylbenzene, [611-14-3] 120.19 80.8 165.2 0.8807 1.5046
2-ethyltoluene
1-Ethyl-3-methylbenzene, [620-14-4] 120.19 95.5 161.3 0.8645 1.4966
3-ethyltoluene
1-Ethyl-4-methylbenzene, [622-96-8] 120.19 62.3 162 0.8614 1.4959
4-ethyltoluene
a
At 818C.
b
At 258C.
c
100.4 kPa.
d
97.6 kPa.
e
At 19 8C.
f
At 130 8C.

Tetramethylbenzenes are readily soluble in aromatic hydrocarbons,


acetone, ether, and alcohol. Durene is highly reactive; nitration, chlorination,
sulfonation, and chloromethylation occur almost as readily as with phenol. The
methyl groups may be successively oxidized to carboxylic acids; the presence of
the corresponding mono-, di-, tri-, and tetracarboxylic acids can be demon-
strated during liquid-phase oxidation of durene in the presence of cobalt
catalyst (95).
Production. The tetramethylbenzenes also occur in coal tar from anthra-
cite and lignite, and in the reformed fraction of oil refineries. The isolation of
Vol. 2 HYDROCARBONS 627

durene by crystallization is aided by its relatively high melting point; isodurene


becomes enriched in the mother liquor. Methylation of xylene leads to an iso-
meric mixture, from which durene, prehnitene, and isodurene can be obtained
(89). Durene is produced by liquid-phase methylation of pseudocumene with
chloromethane in the presence of aluminum chloride (96). In more recent pro-
cesses durene is produced by reaction of xylene or pseudocumene with methanol
in the gas phase over aluminum silicates; the most successful process uses the
zeolites ZSM-5 and TSZ (97,98). In the MTG process, commercialized in New
Zealand since 1986, the production of automotive fuels from methanol over
ZSM-5 yields an unacceptable quantity of durene, which has to be removed
from the fuel to avoid carburetor clogging by the high melting point impurity.
The durene is currently removed by catalytic isomerization, disproportionation,
and reductive demethylation. The possibility of isolating durene from the MTG
process is being investigated (99).
Uses. Oxidation of durene leads to the commercially important pyromel-
litic dianhydride [89-32-7] used as a starting material for heat-resistant polyi-
mides and as a hardener for epoxy resins. Pyromellitic esters serve as
migration-resistant plasticizers for poly(vinyl chloride).
3.3. Penta- and Hexamethylbenzene Properties. Some physical
properties are listed in Table 13. Both compounds are crystalline at room tem-
perature; they dissolve easily in benzene and ethanol. The methyl groups are
readily oxidized to carboxyl groups. Irradiation with light converts hexamethyl-
benzene into its valence isomer hexamethyl Dewar benzene (1,2,3,4,5,6-hexam-
ethylbicyclo [2.2.0] hexa-2,5-diene, [7641-77-2]). Electrophilic substitution on the
free ring site of pentamethylbenzene proceeds readily.
Production. Penta- and hexamethylbenzene are generated by liquid-
phase Friedel – Crafts alkylation of xylene or pseudocumene with chloromethane
in the presence of aluminum chloride (96). If the reaction is carried out at 140 –
180 8C in o-dichlorobenzene, hexamethylbenzene with a purity of 98% can be
crystallized directly from the reaction mixture (100). In a more recent method,
penta- and hexamethylbenzene can be produced simultaneously by reaction of
phenol and methanol in toluene in the presence of a zinc aluminate catalyst at
330 – 360 8C (101). The products are then separated by fractional crystallization.
Hexamethylbenzene is formed by thermal rearrangement of hexamethyl Dewar
benzene, which is readily obtained from 2-butyne (102).
Uses. Pentamethylbenzene is of little commercial importance. Hexa-
methylbenzene is an intermediate in the fragrance industry. It is of scientific
interest in studies of charge transfer complexes and exciplexes, in which, like
other methylbenzenes, it may function as donor (103).
3.4. Diethylbenzenes. Properties. For some physical properties see
Table 13.
628 HYDROCARBONS Vol. 2

Production. The three isomeric diethylbenzenes (DEB) occur as bypro-


ducts in the production of ethylbenzene from benzene and ethylene but are,
generally, recycled to undergo transalkylation (see Ethylbenzene). In the
liquid-phase Friedel – Crafts process with aluminum chloride an increase in
the feedstock ratio of ethylene to benzene may raise the DEB concentration
in the output to 35%. The DEB fraction consists of ca. 65% 1,3-DEB, 30% 1,4-
DEB, and only 5% 1,2-DEB. A similar isomer distribution is found in the pro-
duct mixture from gas-phase alkylation of ethylbenzene with ethylene on an
unmodified HZSM-5 zeolite catalyst. By modifying the catalyst, the zeolite
pore may be narrowed to such an extent that, based on geometry alone, only
1,4-diethylbenzene is formed, rather than the bulkier 1,2- and 1,3-isomers.
The Mobil gas-phase alkylation, which is based on such a shape-selective zeo-
lite, converts ethylbenzene to DEB at 400 8C and 0.7 MPa with 88% selectivity;
more than 99% of the DEB is the 1,4-isomer (104). Nearly quantitative forma-
tion of 1,3-DEB occurs at low temperature in the liquid phase by disproportio-
nation of ethylbenzene in the presence of at least 10 vol% HF and 0.3–0.5 mol
BF3 per mol ethylbenzene.
Uses. Diethylbenzenes may be converted to divinyl benzenes by catalytic
dehydrogenation. The latter are useful in the production of cross-linked poly-
styrenes. DEB is contained in the product of the MTG process (see Section
TETRAMETHYLBENZENES) and improves the properties of the gasoline.
3.5. Triethylbenzenes and more highly Ethylated Benzenes.
Properties. For some physical properties see Table 13.
Production. The more highly ethylated benzenes are produced in the
liquid phase in a similar manner as that described for diethylbenzene. 1,3,5-
Triethylbenzene may be generated in excellent yield by disproportionation of
diethylbenzene with BF3 and HF at 75 8C. The same result is obtained with
AlCl3 –HCl as catalyst. 1,3,5-Triethylbenzene is also formed by cyclocondensa-
tion of 2-butanone in the presence of sulfuric acid. Hexaethylbenzene is formed
in nearly quantitative yield when ethylene is passed into the AlCl3 complex of
aromatics until a mass of crystals is formed, which is then worked up with
water.
3.6. Ethylmethylbenzenes (Ethyltoluenes). Properties. Some phy-
sical properties are collected in Table 13.
Production. Ethyltoluene is produced in similar fashion to ethylbenzene,
i.e, by ethylation of toluene. Alkylation of toluene with ethylene in the presence
of AlCl3 and HCl gives all three isomers of ethyltoluene and more highly ethy-
lated derivatives. A typical reaction mixture contains 48% unreacted toluene,
about 12% 4-ethyltoluene, 19% 3-ethyltoluene, 4% 2-ethyltoluene, 14% higher
aromatics, and barely 1% tar (104). According to a Dow process, subsequent dis-
tillation gives a mixture of 3- and 4-ethyltoluenes (105). 2-Ethyltoluene, toluene,
and the more highly alkylated toluenes are then recycled. The gas-phase ethyla-
tion of toluene on unmodified HZSM-5 zeolite catalyst at 3508C gives 23%
4-ethyltoluene, 53% 3-ethyltoluene, 8% 2-ethyltoluene, and 10% higher
aromatics (106). Formation of the 2- and 3-isomers in the ethylation of toluene
can be almost completely suppressed by the use of a shape-selective ZSM-5
zeolite that has been modified by addition of phosphorus, boron, or other ele-
ments, and by suitable physical treatment. In a Mobil process, 4-ethyltoluene
Vol. 2 HYDROCARBONS 629

of 97% purity is generated in this manner at 350–400 8C (106). A patent


granted to Universal Oil Products claims a process for the production of an
alkylbenzene mixture, which also contains ethyltoluene, from toluene and
synthesis gas (93).
Uses. The mixture of 3- and 4-ethyltoluene obtained by the Dow process is
catalytically dehydrogenated to the corresponding mixture of methylstyrenes,
known commercially as vinyltoluene [25013-15-4], and then polymerized to
‘‘poly(vinyltoluene).’’ Dehydrogenation of the practically pure 4-ethyltoluene
from the Mobil process to 4-methylstyrene [622-97-9] occurs analogously on
the route to ‘‘poly( p-methylstyrene)’’. A plant for the alkylation of toluene to
4-ethyltoluene by the Mobil process was opened in 1982 by Hoechst (now
Hoechst-Celanese) in Baton Rouge, Louisiana (107); the plant has an annual
capacity of 16 000 t of 4-methylstyrene and was purchased by Deltech in 1989.
The output of the two products vinyltoluene and p-methylstyrene is estimated
at 25 000 t/a (108).
3.7. Cumene Properties. Cumene [98-82-8], 2-phenylpropane, isopro-
pylbenzene, C9H12, Mr120.19, is a colorless, mobile liquid with a characteristic,
aromatic odor.

The effect of cumene in increasing the octane rating of motor fuels


(see OCTANE ENHANCERS) has long been known. Large plants for production of
cumene, which was used as a component of aviation fuel, were constructed in
the United States during the early 1940s. Some physical properties of cumene
are listed below (109,110):

mp, 8C 96.04
bp, 8C 152.39
d20. g/cm3 0.86179
n20
D 1.49145
Specific heat (25 8C), J g1 K1 1.96
Heat of vaporization, kJ/mol
at 25 8C 44.23
at 152.4 8C 37.74
Tcrit, K 631
pcrit, MPa 3.21
Vcrit, cm3/mol 428
Flash point, 8C 43
Ignition temperature, 8C 425

The variation of vapor pressure with temperature is given below:


t, 8C 2.9 38.3 66.1 88.1 129.2
Vapor pressure, kPa 0.13 1.33 5.33 15.33 53.33

For more thermodynamic data, see (111). Explosion limits for the system
cumene vapor–oxygen–nitrogen at standard pressure and at 0.56 MPa are
630 HYDROCARBONS Vol. 2

given in (112). For cumene–air mixtures the explosion limits are 0.8–6.0 vol%
cumene, corresponding to a cumene concentration of 40–300 g/m3 (113). Cumene
forms azeotropes with water, aliphatic carboxylic acids, ethylene glycol ethers,
cyclohexanol, and cyclohexanone.
Production. Cumene is produced from benzene [71-43-2] and propene
[115-07-1] in the presence of an acidic catalyst:

Two processes are employed industrially (see also Acylation and Alkylation).
In the UOP process a fixed bed consisting of polyphosphoric acid on a silica sup-
port is used as the catalyst (114). The reaction takes place mainly in the gas
phase at 200 8C and 2 MPa. Pure propene or propene with a propane content
5% is used as raw material, together with benzene (min. purity 99.5%) in up to
12-fold molar excess; this leads to high distillation costs in the product work-up.
The yield of cumene is 97% based on benzene and 90% based on propene (115).
The losses are caused by overalkylation and oligomerization of the propene. The
high-boiling residue contains 50–60% diisopropylbenzenes (o- : m- : p- ratio
1 : 2 : 2–3) and numerous other byproducts. It is not possible to recycle these pro-
ducts to the alkylation reactor, because polyphosphoric acid has no transalkyla-
tion properties.
The second industrial process works in the liquid phase at 50–80 8C; alumi-
num chloride or the liquid double compound formed from aluminum chloride,
(isopropyl-)benzene, and hydrogen chloride is employed as the catalyst. This
complex is separated from the product stream, as a heavy phase and is returned
to the alkylation reactor. Except for a small purge stream higher alkylated
byproducts are recycled to the alkylation step, whereby essentially complete
use of both benzene and propene bound in di- and triisopropylbenzenes is
achieved by transalkylation. The yield of cumene, referred to benzene and pro-
pene, is 99% and 98%, respectively. The UOP process is used predominantly.
Higher selectivity and the lower energy requirements, however, are now making
the aluminum chloride process more attractive (115).
Uses. The production capacity for cumene in Western Europe, the United
States, and Japan amounts to 2 106, 2 106, and 4 105 t/a, respectively. Cumene
is used almost exclusively for the production of phenol [108-95-2] and acetone
[67-64-1]. This process is based on the discovery in 1944 that cumene hydroper-
oxide [80-15-9], obtained by oxidation of cumene with oxygen, can be cleaved
under the catalytic action of strong acids:

Both reactions are exothermic; the enthalpy of reaction amounts to 121 and
255 kJ/mol, respectively.
Pure a-methylstyrene can be obtained as a byproduct of the phenol process.
Cumene hydroperoxide is used as a radical initiator for the copolymerization of
styrene with butadiene and acrylates, and also for the radical cross-linking of unsa-
turated polyester resins. Addition of cumene hydroperoxide to a-methylstyrene
Vol. 2 HYDROCARBONS 631

Table 14. Physical Properties of Diisopropylbenzene Isomers (96)


Property m-DIPB o-DIPB p-DIPB
CAS registry number [99-62-7] [577-55-9] [100-18-5]
mp, 8C 63.13 56.68 17.07
bp, 8C 203.18 203.75 210.35
d20, g/cm3 0.8559 0.8779 0.8568
h20
D 1.4883 1.4960 1.4893

gives dicumyl peroxide, used for radical cross-linking of polyolefins. Hydrogenation


of the aromatic nucleus in cumene gives isopropylcyclohexane (hydrocumene), a
cycloaliphatic solvent with a high boiling point (154.5 8C) and a low freezing
point (90 8C).
3.8. Diisopropylbenzenes. 1,4-Diisopropylbenzene [100-18-5] p-diiso-
propylbenzene, p-DIPB, C12H18, Mr162.26, 1,3-diisopropylbenzene [99-62-7],
m-diisopropylbenzene, m-DIPB, and mixtures of these isomers, are industrially
important. Some physical properties of diisopropylbenzenes are given in
Table 14.

The flash point of DIPB mixtures is 77 8C; the ignition temperature


is 449 8C. Inhalation of DIPB vapor can lead to the appearance of narcotic
effects.
Production. Diisopropylbenzenes are present in the high boiling frac-
tions, in which they may account for 50–70 wt%, of cumene plants, together
with numerous other byproducts. Processes for the production of m- and
p-DIPB from these fractions have been suggested (116,117). The high-boiling
fraction of the aluminum chloride process frequently contains 1,1,3-trimethylin-
dan [2613-76-5], bp 204.8–204.9 8C, which cannot be separated from the m- and
p-DIPB at a justifiable cost.
For production from benzene and propene alkylation catalysts that
possess high isomerization and transalkylation activities are employed.
Examples include aluminum chloride–hydrogen chloride (118,119), silica–
alumina (120,121), boron trifluoride-treated alumina (122), and zeolithes
(123). The process is controlled in such a way that cumene formed at a later
stage is alkylated with propene. The desired DIPB isomer is isolated by distil-
lation and the undesired DIPB isomer, together with triisopropylbenzenes,
are converted to cumene by reaction with benzene in a transalkylation
step. The aim in both the alkylation and the transalkylation steps is to
reach thermodynamic equilibrium, at which the o-diisopropylbenzene is pre-
sent at only ca. 0.1%.
632 HYDROCARBONS Vol. 2

Uses. 1,3-Diisopropyl- and 1,4-diisopropylbenzene are the starting mate-


rials for new synthetic routes to resorcinol and hydroquinone, respectively (124),
which are used industrially in Japan and the United States (121). Furthermore,
a process for the production of phloroglucin, 1,3,5-trihydroxybenzene, from 1,3,5-
triisopropylbenzene has been developed (125). The monohydroperoxides of 1,3-
and 1,4-diisopropylbenzene are used as initiators for the thermal cross-linking
of unsaturated polyester resins, and for the emulsion polymerization of dienes,
styrene, and acrylates. Oxidation gives a,a,a0 ,a0 -tetramethylphenylenebiscarbi-
nols. Dehydrogenation gives diisopropenylbenzenes, which by reaction with
phenol or aniline give bisphenols or bisanilines, respectively, which are used
as building blocks for polycarbonates, polyamides, polyesteramides, and polyur-
ethanes. Addition of isocyanic acid to diisopropenylbenzenes yields diisocya-
nates, from which light-stable polyurethanes can be produced (126). Bis(tert-
butylperoxyisopropyl)benzene, preferably prepared from the mixture of the
corresponding biscarbinols of 1,3- and 1,4-DIPB, is used as a cross-linking
agent for polyolefins.
3.9. Cymenes; C4- and C5-Alkylaromatic Compounds. The com-
pounds listed in Table 15 are industrially important. Mixtures of 3-isopropylto-
luene (m-cymene) and 4-isopropyltoluene (p-cymene) can be obtained by
propylation of toluene in the presence of an isomerization-active Friedel–Crafts
catalyst. These two isomers can be separated by distillation only with difficulty.
The mixture is used for the production of m- and p-cresol, (127). Treatment of
(methyl-)aromatic compounds with isobutene yields tert-butyl-substituted com-
pounds, which are used as solvents (tert-butylbenzene) and for the production
of fragrances, pharmaceuticals, and herbicides.

The introduction of a tert-pentyl group can be achieved by the use of tert-


pentyl chloride as the alkylating agent and aluminum chloride, iron(III) chloride,

Table 15. Physical Properties of Some Alkylbenzenes


CAS
registry
Compound number mp, 8C bp, 8C h20
D
3-Isopropyltoluene [535-77-3] 63.75 175.05 1.4929
4-Isopropyltoluene [99-87-6] 67.94 177.10 1.4909
2-Isopropyltoluene [527-84-4] 71.54 178.15 1.5005
Isobutylbenzene [538-93-2] 51.48 172.76 1.4864
tert-Butylbenzene [98-06-6] 57.85 169.12 1.4926
4-tert-Butyltoluene [98-51-1] 53 192.7 1.4919
5-tert-Butyl-m-xylene [98-19-1] 18.3 209 1.4956
tert-Pentylbenzene [2049-95-8] 192.38 1.4958
Vol. 2 HYDROCARBONS 633

or zirconium tetrachloride as the catalyst. 2-tert-pentylanthraquinone,


obtained from tert-pentylbenzene and phthalic anhydride, is employed in
the industrial production of hydrogen peroxide. Isobutylbenzene, required for
the synthesis of pharmaceuticals and fragrances, is produced in industrial
quantities by treatment of toluene with propene over catalysts containing
alkali metals, such as potassium–graphite inclusion compounds or sodium–
potassium on a support (128).
3.10. Monoalkylbenzenes with Alkyl Groups >C10. Monoalkylben-
zenes with secondary alkyl groups C10 to ca. C14 are used as starting materials
for the production of alkylbenzenesulfonates (ABS). Only secondary alkylbenze-
nesulfonates with unbranched (linear) side chains (LABS) are of practical impor-
tance as raw materials for detergents, as they are extensively degraded (>90%)
by microorganisms in sewage plants after a relatively short period of time.
Before the introduction of LABS, which took place in the Federal Republic
of Germany in 1962, alkylbenzenesulfonates with highly branched C12 side
chains were used. These compounds, tetrapropylene benzenesulfonates, were
produced from the products of alkylation of benzene with propene tetramer, itself
obtained by the tetramerization of propene by contact with phosphoric acid.
These sulfonates possess excellent detergent properties, but their biological
degradeability is insufficient.
Production. Linear secondary alkylbenzenes are obtained industrially
by reaction of benzene with secondary alkyl chlorides (chloroparaffins) or with
olefins under the influence of a Friedel–Craft catalyst such as aluminum chlor-
ide, hydrofluoric acid, or sulfuric acid. The following methods have been
described for the production of the required alkyl chlorides and olefins (129):

1. Dehydrogenation of C10 –C13-n-alkanes, according to the Pacol process of


UOP (see Section Production from Paraffins), on a fixed bed at 400–
500 8C, with a conversion of 10–15%; the ensuing mixture of internal olefins
and unreacted alkanes is used for the alkylation of benzene. The n-C10 –C13-
alkanes used in the dehydrogenation are obtained from the kerosene
fraction of crude oil by adsorption on molecular sieves, according to well
known processes such as IsoSiv of Union Carbide, or Molex of UOP.
2. By chlorination of C10 –C13-alkanes at ca. 100 8C to a conversion of ca. 30%,
and use of the resulting mixture of alkanes and secondary monochloroalk-
anes for alkylation of benzene.
3. By chlorination of C10 –C13-n-paraffins as in 2) above, followed by dehydro-
chlorination (elimination of hydrogen chloride) of the monochloroalkanes at
200–300 8C over an iron catalyst to give linear alkenes with internal double
bonds; these alkenes, in admixture with unreacted paraffins, are then used
to alkylate benzene.
4. Metathesis of higher (>C18) with lower (<C10) internal olefins from the
Shell SHOP process (see Section Oligomerization of Lower Olefins) to
give olefins of intermediate molecular mass.

A C10 –C13-a-olefin fraction suitable for use in detergents can also be


obtained by thermal cleavage of C20 –C40-n-paraffins at 500–600 8C, according
634 HYDROCARBONS Vol. 2

Fig. 8. Production of alkylbenzenes, process of Hüls AG (121).

to the older wax-cracking process of Shell (see Section Production from


Paraffins).
The alkylation of benzene with chloroparaffin–paraffin mixtures takes
place in the presence of aluminum chloride at ca. 50 8C. A relatively high propor-
tion of 2-phenylalkanes is characteristic of secondary alkylbenzenes obtained in
this manner (130,131). Dichloroalkanes, always present in the mixture of chlor-
oparaffins even at low levels of chlorination, can lead to the formation of
diphenylalkanes, 1,4-dialkyltetralins, and 1,3-dialkylindans. For formation and
analytical detection see (132). The removal of these compounds from monoalkyl-
benzenes is difficult. Formation of these byproducts can be largely avoided by
carrying out the alkylation of benzene with olefins from the dehydrochlorination,
where dichloroalkanes are dehydrochlorinated to give tar-like products that are
easily separated. In industrial production the reaction of C10 –C13-olefins with
benezene is carried out with a 1 : 10 molar ratio in the presence of anhydrous
hydrofluoric acid at 10 8C, a process that dates back to 1949 (133). The industrial
process using the steps paraffin chlorination–dehydrochlorination–alkylation in
the presence of hydrofluoric acid was first realized by Chemische Werk Hüls in
1962 (131). This process is shown in Figure 8. In this way a quality standard for
alkylbenzenes for use in detergents was created; in ecological terms, this stan-
dard is still valid.
3.11. Diphenylmethane. Diphenylmethane [101-81-5] benzylbenzene,
C13H12, Mr168.24, was first prepared in 1871 by ZINCKE from benzyl chloride
and benzene in the presence of zinc dust or copper (I) oxide.
Vol. 2 HYDROCARBONS 635

Physical Properties. Diphenylmethane forms colorless prismatic nee-


dles with a harsh herbaceous odor reminiscent of geranium leaves. It is insoluble
in water, soluble in alcohol, ether, chloroform, and benzene.

mp 26–27 8C
bp
101.3 kPa 261–262 8C
1.33 kPa 125.5 8C
d20
4 1.006
Flash point ca. 130 8C
Enthalpy of formation
(room temperature) 109.6 kJ/mol
Enthalpy of combustion
(20 8C) 41.2 103 kJ/kg

Chemical Properties. As an araliphatic hydrocarbon, diphenylmethane


displays the chemical properties of both aromatic and aliphatic compounds. It is
difficult to limit substitution in the phenyl radicals to one of the rings; therefore,
controlled syntheses of particular unsymmetrically substituted diphenyl-
methanes must be carried out by different paths. Under suitable reaction condi-
tions the methylene group can also be substituted, e.g, by halogen or nitro
groups. Benzophenone can be produced by oxidation, for example, with oxygen
over various catalysts, chromic acid or dilute nitric acid. Hydrogenation in etha-
nol over a nickel catalyst leads to dicyclohexylmethane. When diphenylmethane
is passed over platinum–charcoal at 300 8C, cyclization to fluorene takes place.
Production. Of the many known methods to synthesize diphenyl-
methane only a few are used commercially.
The Friedel–Craft reaction of benzyl chloride with benzene in the presence
of aluminum chloride is an important method for production of diphenyl-
methane:

Other known catalysts of this process are elemental aluminum, iron(III)


chloride, antimony pentachloride, zinc chloride–hydrogen chloride and boron
fluoride dihydrate. When sulfuric acid is used as a condensation agent diphenyl-
methane yields are 83% (135) and with polyphosphoric acid yields are over 90%
(136).
Condensation of benzene with formaldehyde in presence of 85% sulfuric
acid yields 79% diphenylmethane (137).

Quality Specifications. Commercial diphenylmethane should be at


least 97% pure (GC). It must be free of halogens and have a minimum melting
point of 24 8C.
Uses. Diphenylmethane is used in the fragrance industry both as a fixa-
tive and in scenting soaps. It can serve as a synergist for pyrethrin in pesticides
636 HYDROCARBONS Vol. 2

and insecticides (138,139). Diphenylmethane is recommended as a plasticizer to


improve the dyeing properties (140), as a solvent for dyes (141), and as a dye car-
rier for printing with disperse dyes (142). The addition of diphenylmethane to
saturated, linear polyesters improves their thermal stability (143), and its addi-
tion to jet fuels increases their stability and lubricating properties (144).
Substituted diphenylmethanes are used as solvents for pressure-sensitive
recording materials (145).

4. Biphenyls and Polyphenyls

In the series of aromatic hydrocarbons with polyphenyls or polyaryls of the gen-


eral formula C6H5(C6H4)nC6H5, only the lowest members (n = 0, 1) are economic-
ally important. Higher polyphenyls (n > 2) are also known and may be isolated
from coal tar or from aryl halides by the Ullmann reaction.
4.1. Biphenyl. Biphenyl [92-52-4], phenylbenzene, diphenyl, C12H10,
Mr154.2, was discovered in 1862 by FITTIG by the reaction of bromobenzene
with sodium metal. BERTHELOT obtained the compound in 1867 by passing ben-
zene vapor through a heated tube. In 1875 BÜCHNER demonstrated the presence
of biphenyl in coal tar (146).

Physical Properties. Pure biphenyl is a white solid; when only slightly


impure it has a yellow tint. It crystallizes from solution results as glossy plates or
monoclinic prisms. Biphenyl is almost insoluble in water, but is easily soluble in
organic solvents such as ethanol, diethyl ether, and benzene. The dipole moment
and X-ray measurements demonstrate that the two benzene rings are coplanar
in the solid state. In the melt, in solution, and in the vapor free rotation about the
central CC bond can take place; in substituted biphenyls, however, this rota-
tion may be severely restricted, e.g, in dinitrodiphenic acid, leading to the exis-
tence of optically active isomers. The most important physical properties are
listed below:
mp 69.2 8C (147)
bp (101.3 kPa) 255.2 8C (147)
d20
4 (solid) 1.041
d77
4 (liquid) 0.9896 (148)
n77
D 1.5873
Tcrit 515.7 8C (149)
pcrit 4.05 mPa
Flash point 113 8C (150)
Ignition temperature 570 8C
Heat of combustion 6243.2 kJ/mol
Heat of vaporization 53.9 kJ/mol
Heat of fusion 18.60 kJ/mol
Vol. 2 HYDROCARBONS 637

Table 16. Temperature dependence of some physical properties of biphenyl (148)


t, 8C 100 200 300 350
Vapor pressure, kPa 25.43 246.77 558.06
Liquid density, g/cm3 0.97 0.889 0.801 0.751
Heat capacity, J/g 1.786 2.129 2.468 2.640

The variation of some physical properties with temperature is given in


Table 16.
Chemical Properties. Biphenyl is a very stable organic compound; in
an inert atmosphere it remains unchanged even at a temperature >300 8C.
The compound sublimes, distills with steam, and undergoes various substitu-
tion reactions; the reactivity is similar to, but less than that of benzene.
Substitution reactions, e.g, chlorination, nitration, and sulfonation, occur
at the 2- and 4-positions, the latter being the preferred site of attack in
Friedel–Crafts reactions.
Production. Hydrodealkylation of Benzene. Biphenyl is currently
obtained mainly as a byproduct in the production of benzene by thermal or cata-
lytic hydrodealkylation of toluene (! Benzene–Hydrodealkylation) (151–154).
The reaction is carried out in a hot tube reactor at 700 8C with a hydrogen pres-
sure of 4 MPa; the ratio of hydrogen to toluene is 4 : 1. Toluene conversion ranges
from 35 to 85%, with methane being formed at 2.5 mol% above the stoichiometric.

Biphenyl is obtained as the pot residue, following removal of gas and distil-
lation of benzene and toluene; about 1 kg of biphenyl is obtained per 100 kg ben-
zene. The biphenyl can be enriched to a purity of 93–97% by distillation.
Thermal Dehydrocondensation of Benzene. Until the early 1970s, biphenyl
was produced exclusively by specific thermal dehydrocondensation of benzene
(155). This process yields a higher quality product than the toluene route, e.g,
distilled material with a purity >99.5%; varying quantities of terphenyl isomers
and several higher homologues are also obtained.

The slightly endothermic dehydrocondensation (DH ¼ þ 8 kJ/mol (156)) of


benzene to biphenyl and terphenyl has been investigated intensively (157,158).
The formation of biphenyl from benzene starts at ca. 480 8C and increases with
increasing temperature and residence time.
Industrially, the reaction is carried out in a multitube reactor, heated elec-
trically (159) or by heat exchanger, at 700–850 8C using residence times of the
638 HYDROCARBONS Vol. 2

order of seconds. An optimal tube length of 45–75 m and diameter of 100–


120 mm have been reported (160–163). None of the published procedures work
in equilibrium because carbon black formation increases with increasing tem-
perature and longer residence time.
The preferred residence time is ca. 10 s, which results in a benzene con-
version of ca. 10% (161,164). By recycling the benzene, 1 kg of starting material
may yield 0.8–0.85 kg biphenyl, 0.07–0.12 kg terphenyl, and some higher
polyphenyls. The major technical problem in the production of biphenyl by
dehydrocondensation of benzene is carbonization at the high temperature
required (165). The carbon deposit on the reactor tube walls and the release
of large carbon flakes at bends or narrow passages within the tubes may
lead to blockages. Countermeasures include addition of about 0.1% oxygen or
sulfur compounds, such as methanol, ethanol, acetone, or carbon disulfide
(166–168) to the benzene feedstock, the use of special alloys (169,170), uniform
heating of the reactor by means of circulating gases, and high feedstock velo-
city, and the use of turbulent flow (161). A proposal for further improving the
process is to carry out the reaction in fluidized beds. Quartz sand (171) or sin-
tered corundum (172) is used as energy carrier in this process; these materials
are recycled after the carbon deposits have been removed by burning them
outside of the reactor.
Other Synthetic Routes. In addition to these two industrial processes
for the production of biphenyl, other synthetic routes have been described:

1. Ullmann reaction: heating halogenobenzenes with copper powder (173).


2. Reduction of phenyldiazonium chloride solution with copper and zinc (174).
3. Heating phenylmagnesium halides with copper(II) salts (175).
4. Oxidative dehydrocondensation of benzene (176–181). Benzene is oxidized
to biphenyl by platinum- or palladium salts in glacial acetic acid at 120 8C;
the salts are reduced to the pure metals, and stoichiometric quantities of
the salts are required.
5. Dehydrocondensation of benzene in the presence of catalytic quantities of
noble metal salts requires reoxidation of the metals during the course of the
reaction. This may be done with pure oxygen, but a partial pressure of
6 MPa is required (182). Additives such as b-diketones (183), heteropolya-
cids (184), or organic copper salts (185) are incapable of significantly low-
ering the required oxygen pressure, so that a potentially explosive regime
cannot be avoided (184). Yields of biphenyl are ca. 260 mol%, calculated on
the palladium salt charge.
6. Hydrodimerization of benzene to phenylcyclohexane, followed by dehydro-
genation to biphenyl (186–190):

Suitable catalysts for this dimerization are noble metals on molecular


sieve, alkali metals on alumina or carbon, and transition metals such as nickel,
Vol. 2 HYDROCARBONS 639

tungsten, or zinc on the zeolites faujasite or mordenite. A selectivity of up to 87%


has been claimed. Subsequent dehydrogenation is carried out at about 400 8C
over Pt–Al2O3 (190), Cr2O3 – Na2O – Al2O3 (189,191), Cr2O3 – K2CO3 – Fe2O3
(192), or Cr2O3 –MgO–Al2O3 (193) giving yields of up to 99%. Data concerning
residence times over the catalyst are not available, either for the hydrodimeriza-
tion or the dehydrogenation processes.
Analysis. Gas chromatography is used to analyze biphenyl and its
derivatives.
Storage and Transportation. Biphenyl is marketed as a solid in the
form of plates, flakes, or pellets, containing <10 ppm benzene and with a biphe-
nyl content of >99.5%; it is transported in bags or fiber drums; in the molten
state (ca. 120 8C) in tank cars. It is not a regulated material for transportation,
on account of its toxicity; there are no restrictions for transportation by sea, land,
or air. The greatest hazards in handling biphenyl are the risk of dust explosions
and the ignition of biphenyl vapor–air mixtures over the molten product.
Uses. Biphenyl is an important heat transfer agent, because of its high
thermal stability. It is used as a composite with diphenyl ether, especially as a
melting point depressant. The eutectic mixture of 26.5% biphenyl and 73.5%
diphenyl ether is marketed under the trade names Diphyl (Bayer), Dowtherm
A (Dow), Thermic (ICI), Gilotherm (Rhône-Poulenc), Therm S 300 (Nippon
Steel), Santotherm VP and Therminol VP-1 (both Monsanto) for use at up to
400 8C. The eutectic begins to boil at 256 8C; the vapor pressure at the maximum
operating temperature (400 8C) is ca. 1.1 MPa.
A further large quantity of biphenyl is used as a dye carrier (194,195). Until
a few years ago, polychlorinated biphenyls (PCBs), generated by chlorination of
biphenyl, were produced in large quantities for use as nonflammable hydraulic
fluids and as transformer dielectrics. Production has largely ceased on ecological
considerations. Among the hydroxy derivatives, o- and p-hydroxybiphenyl and
p,p0 -dihydroxybiphenyl are used industrially. o-Hydroxybiphenyl is used as
a preservative and fungicide. It is synthesized mainly from cyclohexanone.
p-Hydroxybiphenyl and p,p0 -dihydroxybiphenyl are generated by sulfonation of
biphenyl, followed by fusion with alkali.
4.2. Terphenyls. Physical Properties. The three isomeric terphe-
nyls, C18H14, Mr230.29, are colorless to pale yellow crystalline solids
(196–201). Some physical properties are listed in Table 17.
640 HYDROCARBONS Vol. 2

Table 17. Physical Properties of Terphenyls


o-Terphenyl m-Terphenyl p-Terphenyl
Property [84-15-1] [92-06-8] [92-94-4] Reference
mp, 8C 59 87 212 (199)
bp (101.3 kPa), 8C 332 365 376 (200)
Flash point, 8C 171 206 210 (202)
Tcrit, 8C 613 603 621 (203)
Pcrit, MPa 3.903 3.503 3.330 (204)
Heat of vaporization (at bp),
kJ/mol 58.3 64.2 62.6 (200)
Vapor pressure, kPa
93 8C 0.01172 0.00165 (199,205)
204 8C 2.834 0.8274 0.789
315.6 8C 64.4 27.3 24.56
426.7 8C 439.9 240.6 174.4

Chemical Properties. Terphenyls, like biphenyls, are thermally extre-


mely stable organic compounds, again offering a potential for use as heat trans-
fer media. Terphenyls are sublimable, steam volatile, and show the typical
substitution reactions of aromatic hydrocarbons, such as bromination and nitra-
tion. The substitution pattern for the preferred positions of the three terphenyls
is shown in Table 18.
Production. Terphenyls are byproducts in the production of biphenyl by
dehydrocondensation of benzene; they are found in the high-boiling fraction of
the pyrolysis products. The pot residue has the following approximate composi-
tion (206):
3–8% o-Terphenyl
44% m-Terphenyl
24% p-Terphenyl
1.5% Triphenylene
22–27% Higher polyphenylenes and tar.

When, in the early 1960s, terphenyl was proposed to be used as coolant and
moderator in nuclear reactors (207), numerous reports were published of
attempts to enhance the yield of terphenyl in dehydrocondensation reactions.
Increasing the temperature or residence time was ruled out because of increased
carbonization. These procedures also enhanced the biphenyl fraction, so that the
latter had to be recycled to the reactor (162,166), (208). Recycling a concentration
of up to 30% biphenyl has been described (162). A practical level appears to be
using benzene with 10% biphenyl (177). Attempts have also been made to recycle
the extremely high melting p-terphenyl, after isolation from the reaction

Table 18. Terphenyl Substitution Pattern


Position for substitution
First Second Third Fourth
00 0
o-Terphenyl 4- 4 - 4- 50 -
m-Terphenyl 40 - 4- 400 -
p-Terphenyl 4- 400 - 20 -
Vol. 2 HYDROCARBONS 641

mixture, to obtain a lower melting terphenyl mixture (208). The o-derivative is


readily separated from the terphenyl isomer mixture by distillation; m- and p-
terphenyl distill together and the pure isomers can be obtained by zone refining
(211). Terphenyl isomers, including hydrogenated terphenyl, can be analyzed by
gas chromatography.
Storage and Transportation. Solid terphenyl is shipped as flakes in
laminated bags or in fiber drums, liquid, hydrogenated terphenyl in tank cars,
steel drums, or barrels. Terphenyls and hydrogenated terphenyls possess a low
toxicity, and are therefore not regarded as dangerous goods for the purpose of
transportation.
Uses. Terphenyl is used as a heat transfer agent because of its excellent
thermal stability; a mixture with a composition of 2–10% o-terphenyl, 45–49%
m-terphenyl, 25–35% p-terphenyl, and 2–18% higher polyphenyls is used.
This composition approximates the pot residue from the distillation of biphenyl.
This mixture has bp ca. 360 8C; a disadvantage is the high mp, 145 8C. The lar-
gest fraction is partially hydrogenated to yield a clear oil, miscible with hydrocar-
bons and chlorinated hydrocarbons. Hydrogenation lowers the setting point and
the viscosity but unfortunately also reduces the thermal stability. Partially
hydrogenated terphenyl (212) is used mainly as a dye carrier for pressure sensi-
tive recording materials and copy paper, and as a heat transfer oil. Heat transfer
oils consisting of partially hydrogenated terphenyl isomers are marketed under
the following trade names:
Santotherm 66 or 88, Therminol (Monsanto, United States), Gilotherm
(Rhône-Poulenc, France), and Therm S 900 (Nippon Steel, Japan). Their usage
range extends to 340 8C.
4.3. Polyphenyls. Polyphenyls with four or more phenyl residues have
generally been synthesized as p-linked isomers and are used as scintillators
(213). They are economically unimportant.
A symmetrical hydrocarbon within this class is 1,3,5-triphenylbenzene,
C24H18, mp 174 8C, prepared by condensation of three molecules of acetophenone.

5. Toxicology and Occupational Health

5.1. Alkanes. Methane is toxicologically virtually inert. In very high


concentrations (80–90 vol%) it causes respiratory arrest (214). The bulk of an
inhaled dose is exhaled unchanged (215).
Ethane easily causes dyspnea in laboratory animals, beginning at concen-
trations of approximately 2–5 vol% in the air inhaled (216). At high concentra-
tion respiratory arrest occurs (217). A slight sensitization of the myocard to
catecholamines has been described at a concentration of 15–19 vol% (218).
Propane. has narcotic properties (217). It is neither a skin nor a mucosal
irritant. At very high concentration it also causes respiratory arrest (219). In
dogs adverse effects on circulatory function were found with inspiratory concen-
trations of 1% or greater. A negative inotropic effect, a drop in the mean aortic
pressure, and an increase in pulmonary vascular resistance were found (220).

MAK (1987): 1000 ppm.


642 HYDROCARBONS Vol. 2

Butane causes slight drowsiness beginning at a concentration of 10 000 ppm


in the air inhaled (221). In the mouse the inhalatory LC50 is 680 g/m3 (222). It has
been proven in the dog that butane also sensitizes the myocardium to the effects
of catecholamines (220).
MAK (1987): 1000 ppm.
n-Pentane has a less pronounced narcotic effect than the C1 – C4 hydrocar-
bons. In humans the lowest lethal dose after acute inhalatory exposure is appar-
ently 130 000 ppm, the lowest toxic dose 90 000 ppm (223). Pentane apparently
possesses a neurotoxic effect, although it is not very pronounced (224,225).
n-Hexane possesses a marked neurotoxic effect that distinguishes it from
the other alkanes. The causal factor for peripheral neuropathies is the oxidation
product (cytochrome P 450-dependent oxidases) 2,5-hexadione. Toxicity mani-
fests itself clinically in polyneuropathies (226,227). Hexane is also a skin irritant.
The LC50 (inhalation mice) is 120 g/m3 (228). In rats the acute oral LD50 is 24–
49 mL/kg (229). Hexane is absorbed through the skin. Dermal exposure can lead
to poisoning (LD dermal in rabbits ca. 5 mL/kg) (230). In rats subchronic expo-
sure to a concentration of 400–600 ppm leads to neuropathies (231), which are
characterized by degeneration of both myelin sheaths and axons (232).
MAK (1987): 50 ppm.
n-Heptane has narcotic properties. In humans 0.1% in the inhaled air
leads to dizziness and 0.5% to equilibrial disturbances with loss of motor coordi-
nation (221). Within 3 min 4.8 vol% in the air inhaled leads to asphyxia (235).
Like hexane, heptane is a skin and mucosal irritant (221). The biotransformation
of n-heptane, as with hexane, is by oxidation. Heptane apparently also possesses
neurotoxic properties (236). Myocardial sensitization to catecholamines has also
been found (237).
MAK (1987): 500 ppm.
n-Octane is similar to n-heptane in regard to its narcotic effect; however,
it does not seem to cause any other effects on the nervous system (228). Octane is
also metabolized by oxidation (238).
MAK (1987): 500 ppm
Only limited toxicological data are available on the higher molecular mass
alkanes.
5.2. Alkenes. The alkenes under consideration are toxicologically not
very active. Higher molecular mass compounds possess narcotic properties.
The alkenes are not neurotoxic. The a-olefins generally seem to be more reactive
and toxic than the b-isomers.
5.3. Alkylbenzenes. Trimethylbenzenes. The toxicological data
available on trimethylbenzenes is mainly rather old. The acute inhalatory
toxic dose for the 1,2,4- and the 1,3,5-isomers is in the range 7000–9000 ppm
(239,240).
As a result of their occurrence in automotive fuel and heating oil, low con-
centrations of trimethylbenzenes can be demonstrated in air and water. For
example, measurements in 1974 in a tunnel in Rotterdam indicated a level of
Vol. 2 HYDROCARBONS 643

0.015 ppm pseudocumene; see (241) for more details. An analysis of the drinking
water of Cincinnati, Ohio, in 1980, revealed a concentration of 45 ng/L hemimel-
litene, 127 ng/L pseudocumene, and 36 ng/L mesitylene (242). Trimethylben-
zenes are toxic to aquatic organisms; LC50 values between 10 and 100 ppm
have been reported (243).
The TLV (TWA) level for trimethylbenzene is 25 ppm (244). The MAK for Tri-
methylbenzenes (all isomers (1,2,3–;1,2,4–;1,3,5–) has been set at 20 ppm. (245).
The United States toxicity assigment ranges from low to moderately toxic for pseu-
documene, to highly toxic for the isomer mixture; disturbances in the central ner-
vous system and abnormal blood pictures have been reported (243).
Tetramethylbenzenes. The oral toxicity of tetramethylbenzenes is low;
LD50 >5000 mg/kg (rat, oral). Durene is additionally classified as intravenously
highly toxic; LD50 180 mg/kg (i.v., mice). Tetramethylbenzenes cause mild skin
reactions (243). Two values are cited for the odor threshold concentration of dur-
ene: 0.083 and 0.087 mg/m3 (241).
Hexamethylbenzene. Hexamethylbenzene has low oral toxicity; LDLo
5000 mg/kg (rat). It is, however, suspected of causing neoplastic effects.
Addition of nitromethane to hexamethylbenzene can lead to an explosive
reaction (243).
Diethylbenzenes. Diethylbenzene is more toxic than monoethylbenzene
(246). DEB is emitted into the environment from engine fuel and heating oil
(242). Diethylbenzenes are toxic to aquatic organisms; the LC50 is between 10
and 100 mg/L (243). In the cell multiplication inhibition test with protozoa, the
toxicity threshold is <10 mg/L (241). In the Federal Republic of Germany DEB
is classified as an aquatic hazard; aquatic hazard class 2. DEB causes mild to
moderate irritation of the eye and mucous membranes. The oral toxicity is low:
LDLo 5000 mg/kg (rat) (243). The vapor possesses an anesthetic action and may
cause headache, vertigo, or vomiting; the odor threshold concentration is
<10 ppm (247).
Triethylbenzene. The LC50 of triethylbenzene (aquatic organisms) is
between 100 and 1000 mg/L. The oral toxicity is low: LDLo 5000 mg/kg (rat).
Implant experiments with hexaethylbenzene in mice have raised the suspicion
that the compound may be carcinogenic (243).
Ethyltoluene. Ethyltoluene can, as with aromatic hydrocarbons, be
detected in the environment (241). The oral toxicity of 2- and 4-ethyltoluene is
low: LDLo 5000 mg/kg (rat) (243).
Cumene. Cumene appears to be orally as nontoxic as propylbenzene, it is
even less so by inhalation (239,248,249). Cumene has a prolonged depressant
effect on the central nervous system (250).
The MAK for cumene has been set at 250 mg/m3, corresponding to a cumene
vapor concentration in air of 50 mL/m3. The vapor pressure of cumene at 20 8C is
0.5 kPa. There is a risk of cumene being absorbed through the skin (251). Inhalation
of cumene vapor leads to the delayed appearance of a long lasting narcotic effect (252).
5.4. Biphenyls and Polyphenyls. Biphenyls. Biphenyl dust or
vapor is irritating to the eye and mucous membrane at a concentration as low
as 3–4 ppm (246), and to the skin after extended exposure. A concentration of
>5 mg/m3 for long periods is considered a health hazard; systemic toxic effects
were elicited in humans by a concentration with inhalatory exposure maxima
644 HYDROCARBONS Vol. 2

of 128 ppm (253,254). The olfactory threshold is ca. 0.06–0.3 mg/m3. Some toxi-
city data are listed below (255–257):

LD50 (rat, oral) 3280 mg/kg


LD50 (rabbit, oral) 2400 mg/kg
LD50 (rabbit, dermal) 2500 mg/kg
TLm (fathead minnow, 96 h) 1.5 mg/L

Triphenyls are only minimally toxic (258).


Terphenyls. Toxicologically, terphenyl isomers should be treated like
biphenyl. The oral LD50 is ca. 4.6–4.7 g/kg (259). The toxicity of partially (40%)
hydrogenated terphenyl (LD50 17.5 g/kg) is much less than that of the fully
aromatic terphenyls (260–263)

REFERENCES

1. F. D. Rossini, B. J. Mair, Adv. Chem. Ser. 5 (1952) 341.


2. Selected values of properties of hydrocarbons and related compounds, Am. Petrol.
Inst. Research Project 44, Supplementary vol. A-72, 1975.
3. B. V. Vora, T. Imai, Hydrocarbon Process. 61 (1982) no. 4, 171.
4. K. Nabert, G. Schön, Sicherheitstechnische Kennzahlen brennbarer Gase und
Dämpfe, Deutscher Eichverlag, Berlin 1965.
5. R. C. Weast, ed., Handbook of Chemistry and Physics, 55th ed., CRC Press, Cleve-
land 1974.
6. C. Zerbe, Mineralöle und Verwandte Produkte, Springer, Berlin 1969.
7. G. A. Olah et al., J. Am. Chem. Soc. 91 (1969) 3261.
8. G. A. Olah et al., J. Am. Chem. Soc. 95 (1973) 7680.
9. Deutsche Shell AG, Erdöl-Nachrichten, March 1976, p. 4.
10. F. Schmerling, H. Heneka, Erdöl Kohle Erdgas Petrochem. 21 (1968) 705.
11. K. Hedden, J. Weitkamp, Chem. Ing. Tech. 47 (1975) 505.
12. J. W. Ward, Hydrocarbon Process. 54 (1975) no. 9, 101.
13. P. D. Bartlett, K. E. Schueller, J. Am. Chem. Soc. 90 (1968) 6071.
14. Houben-Weyl, 4th ed., vol. IV, ‘‘1parts 3 and 4’’.
15. J. H. Lucas, F. Baardman, A. P. Kouwenhoven, Angew. Chem. 87 (1975) 740, 88
(1976) 412.
16. W. C. van Ziill Langhout, Proc. World Pet. Congr. 1975, vol. 5, 197.
17. W. L. Lafferty Jr., W. R. Stokeld, Adv. Chem. Ser. 103 (1971) no. 7, 130.
18. L. F. Albright, Chem. Eng. 73 (1966) no. 14, 119; 73 (1966) no. 19, 205; 73 (1966)
no. 21, 209.
19. L. Alcock, R. Martin, K. H. Bourne, W. A. Peet, J. Winsor, Oil Gas J. 72 (1974) no. 27,
102.
20. D. P. Thornton, D. J. Ward, R. A. Erickson, Hydrocarbon Process. 51 (1972) no. 8, 81.
21. H. Jockel, B. E. Triebskorn, Hydrocarbon Process. 52 (1973) no. 1, 93.
22. Hydrocarbon Process. 52 (1973) no. 11, 145.
23. P. R. Pujado, B. V. Vora, Energy Prog. 4 (1984) no. 3, 186.
24. M. S. Scurrell, Appl. Catal. 32 (1987) 1.
25. P. Leprince, A. Chauvel, L. Castex: Proce´de´s de Pe´trochimie, Editions Technip, Paris
1971, p. 269.
26. K. Ito, I. Dogane, K. Tanaka, Hydrocarbon Process. 64 (1985) no. 10, 83.
27. S. C. Arnold, G. D. Sucui, L. Verde, A. Neri, Hydrocarbon Process. 64 (1985) no. 9, 123.
Vol. 2 HYDROCARBONS 645

28. Japan. Chemical Reporter, 1981, October, p. 2.


29. F. Solymosi, I. Tombácz, G. Kutsan, J. Chem. Soc. Chem. Commun. 1985, 1455.
30. R. V. Porcelli, B. Juran, Hydrocarbon Process. 65 (1986) no. 3, 37.
31. W. Hinsen, M. Baerns, Chem. Zt. 107 (1982) 9.
32. Kh. E. Khcheyan, O. M. Revenko, A. N. Shatalova: ‘‘Oxidative Methylation, a new
route to petrochemicals through alternative feedstock,’’ Special Paper No. 16, World
Petroleum Congress 11th, 1983 (London).
33. G. de Pontanel: Proteins from Hydrocarbons, Academic Press, London 1972.
34. S. Gutcho: Proteins from Hydrocarbons, Noyes Data Corp., Park Ridge, N.J. 1973.
35. P. J. Rockwell: Single Cell Proteins from Cellulose and Hydrocarbons, Noyes Data
Corp., Park Ridge, N.J. 1976.
36. G. Voss, Erdöl Kohle Erdgas Petrochem. Brennst. Chem. 26 (1973) 249.
37. W. Dimmling, R. Seipenbusch, Hydrocarbon Process. 54 (1975) no. 9, 169.
38. H. Riegel, H. D. Schindler, M. C. Sze, Erdöl Kohle Erdgas Petrochem. Brennst.
Chem. 26 (1973) 704.
39. R. P. Lowry, A. Aguilo, Hydrocarbon Process. 53 (1974) no. 11, 103.
40. L. B. Phillips, Hydrocarbon Process. 42 (1963) no. 6, 159.
41. Universal Oil Products, US 3 649 177, 1972 (D. H. Rosback).
42. K. G. Mittal, J. Chem. Technol. Biotechnol. 36 (1986) 291–299.
43. D. B. Broughton, R. C. Berg, Chem. Eng. 77 (1970) 86.
44. B. V. Vora, P. R. Pujado, J. B. Spinner, T. Imai, Proc. World Tenside Congr. 1984
(Munich) vol. II, 16–26.
45. R. C. Berg, B. V. Vora in McKetta, Cunningham (ed.): Encyclopedia of Chemical Pro-
cessing and Design, vol. 15, Marcel Dekker, New York 1982, 266–284.
46. H. D. Wulff, Fette Seifen Anstrichm. 69 (1967) 32.
47. R. Stöbele, Chem. Ing. Tech. 36 (1964) 858.
48. K. Ziegler et al., Angew. Chem. 64 (1952) 323, 67 (1955) 542.
49. Gulf Res. Dev. Co., DE 1 443 927, 1961.
50. Ethyl Corp., US 3 906 053, 1973.
51. W. Keim, A. Behr, M. Röper in G. Wilkinson (ed.): Comprehensive Organometallic
Chemistry, vol. 5, Pergamon Press, Oxford 1982, 371.
52. W. Keim, Chem. Ing. Tech. 56 (1984) 850.
53. Shell Development Comp., US 3 635 937, 1972; 3 637 636, 1972; 3 644 564, 1972;
3 644 563, 1972; 3 647 914, 1972; 3 647 915, 1972; 3 661 803, 1972; 3 686 159, 1972
(W. Keim et al.).
54. E. R. Freitas, R. C. Gum, Chem. Eng. Progress. 75 (1979) no. 1, 73.
55. E. L. T. M. Spitzer, Seifen öle Fette Wachse 107 (1981) 141–144.
56. M. Sherwood, Chem. Ind. (London) (1982) 994–995.
57. W. Keim, F. H. Kowaldt, Erdöl Kohle Erdgas Petrochem. Compendium (1978 /79)
453–462.
58. Shell Oil Comp., US 3 647 906, 1972 (F. F. Farley).
59. M. Peuckert, W. Keim, Organometallics 2 (1983) 594.
60. W. Keim, F. H. Kowaldt, R. Goddard, C. Krüger, Angew. Chem. 90 (1978) 493.
61. W. Keim, A. Behr, B. Limbäcker, C. Krüger, Angew. Chem. 95 (1983) 505.
62. W. Keim et al., J. Mol. Catal. 6 (1979) 79.
63. W. Keim et al., Organometallics 5 (1986) 2356.
64. Chem. Eng. News (1972), April 10, 16–17.
65. C. Commereuc, Y. Chauvin, J. Gaillard, J. Léonard, J. Andrews, Hydrocarbon
Process. 63 (1984) no. 11, 118–120.
66. W. J. Benedek, J.-L. Mauleon, Hydrocarbon Process. 59 (1980) no. 5, 143.
67. J. Léonard, J. F. Gaillard, Hydrocarbon Process. 60 (1981) no. 3, 99.
68. J. F. Boucher, G. Follain, D. Fulop, J. Gaillard, Oil Gas J. 80 (1982) 84–86.
646 HYDROCARBONS Vol. 2

69. Y. Chauvin, J. Gaillard, J. Léonard, P. Bonnifay, J. W. Andrews, Hydrocarbon


Process. 61 (1982) no. 5, 110–112.
70. D. Commereux, Y. Chauvin, G. Leger, J. Gaillard, Rev. Inst. Fr. Pet. 37 (1982) 639–649.
71. J. Gaillard, Pet. Tech. 314 (1985) 20–27.
72. B. Bogdanović, Adv. Organomet. Chem. 17 (1979) 105.
73. W. Keim, A. Behr, G. Kraus, J. Organomet. Chem. 251 (1983) 377.
74. A. Behr, V. Falbe, U. Freudenberg, W. Keim, Isr. J. Chem. 27 (1986) 277–279.
75. W. Keim, Ann. N.Y. Acad. Sci. 415 (1983) 191–200.
76. B. Bogdanović, B. Spliethoff, G. Wilke, Angew. Chem. Int. Ed. Engl. 19 (1980) 622.
77. G. Speier, J. Organomet. Chem. 97 (1975) 109.
78. M. G. Barlow, M. J. Bryant, R. N. Haszeldine, A. G. Mackie, J. Organomet. Chem. 21
(1970) 215.
79. S. Datta, M. B. Fisher, S. S. Wreford, J. Organomet. Chem. 188 (1980) 353.
80. R. Schrock, S. McLain, J. Sancho, Pure Appl. Chem. 52 (1980) 729.
81. J. A. Brennan, Ind. Eng. Chem. Prod. Res. Dev. 19 (1980) 2–6.
82. R. L. Shubkin, M. S. Baylerlan, A. R. Maler, Ind. Eng. Chem. Prod. Res. Dev. 19
(1980) 15–19.
83. F. J. Karol, CHEMTECH 1983 (April) 222.
84. B. W. Werdelmann, Fette Seifen Anstrichm. 76 (1974) 1.
85. G. Wilke, Angew. Chem. Int. Ed. Engl. 2 (1963) 105.
86. A. C. L. Su, Adv. Organomet. Chem. 17 (1979) 269.
87. S. Warwel, Erdöl Erdgas Kohle 103 (1987) 238.
88. R. C. Weast, M. J. Astle (eds.): CRC Handbook of Data on Organic Compounds, CRC
Press, Boca Raton, Florida 1985.
89. S. Coffey (ed.): Rodd’s Chemistry of Carbon Compounds, III A, Elsevier, Amsterdam
1971, pp. 228–231.
90. A. D. Rudkovskii et al., Chem. Technol. Fuels Oils (Engl. Transl.) 19 (1983) 387–391.
91. F. P. McCandless, Ind. Eng. Chem. Prod. Res. Dev. 19 (1980) 612–616; 21 (1982)
483–488.
92. Idemitsu Kosan Co. Ltd., EP 0 104 607, 1983 (H. Sato).
93. Universal Oil Products, US 4 487 984, 1984 (T. Imai).
94. O. A. Anunziata et al., Appl. Catal. 15 (1985) 235–245.
95. J. Hanotier, M. Hanotier-Bridoux, J. Mol. Catal. 12 (1981) 133–147.
96. L. I. Smith, Org. Synth., coll. vol. 2 (1948) 248–253.
97. Toyo Soda Manufacturing, DE-OS 3 330 912, 1983 (T. Sakamoto, T. Hironaka, K. Se-
kizawa, Y. Tsutsumi).
98. Toa Nenryo, EP 0 101 651, 1983 (T. Maejima, N. Tagaya, S. Sakurada).
99. J. Haggin, Chem. Eng. News, (1987) June 1, 22–28; (1987) June 22, 22–25.
100. International Flavors and Fragrances, US 3 869 524, 1974 (K. K. Light, J. B.
Anglim).
101. Phillips Petroleum, US 4 568 784, 1985 (R. L. Cobb).
102. S. A. Shama, C. C. Wamser, Org. Synth. 61 (1983) 62–64.
103. H. F. Davis, S. K. Chattopadhyay, P. K. Das, J. Phys. Chem. 88 (1984) 2798–2803.
104. W. W. Kaeding, G. C. Barile, M. M. Wu, Catal. Rev. Sci. Eng. 26 (1984) 597–612.
105. Dow Chemical, US 2 763 702, 1956 (J. L. Amos, K. E. Coulter).
106. W. W. Kaeding, L. B. Young, C.-C. Chu, J. Catal. 89 (1984) 267–273.
107. W. W. Kaeding, G. C. Barile, Polym. Sci. Technol. (Plenum) 25 (1984) 223–241.
108. Eur. Chem. News, Oct. 17 (1988) p. 4.
109. Physical Constants of Hydrocarbons C1 to C10, ‘‘ASTM Spec. Techn. Publ. STP’’ 106
(1963)
110. Handbook of Chemistry and Physics, R. C. Weast (ed.), 63rd ed., CRC Press, Boca
Raton 1982–1983, p. D-213.
Vol. 2 HYDROCARBONS 647

111. D. R. Stull, E. F. Westrum, G. C. Sinke: The Chemical Thermodynamics of Organic


Compounds, J. Wiley & Sons, New York 1969, p. 370.
112. J. C. Butler, W. P. Webb, Chem. Eng. Data Series 2 (1957) no. 1, 42–46.
113. G. Sorbe: Sicherheitstechnische Kenndaten chemischer Stoffe, vol. 2, ecomed-
Verlagsgesellschaft mbH, Landsberg 1983, pp. 15–16.
114. P. R. Pujado, J. R. Salazar, C. V. Berger, Hydrocarbon Process. 55 (1976) no. 3, 91.
115. R. C. Canfield, T. C. Unruh, Chem. Eng. NY 90 (1983) no. 3, 32.
116. Toyo Soda Manufacturing Co., DE-AS 2 418 038, 1974 (T. Yanagihera et al.).
117. Hitachi Chemical Co., JP-Kokai 77 139 030, 1974 (E. Kudo et al.); Chem. Abstr. 88
(1978) 120757 r.
118. BASF, DE-OS 2 141 491, 1971 (F. Neumayr).
119. Mitsui Petrochemical Ind., EP 0 046 678, 1981 (M. Hisaya).
120. J. P. Fortuin, M. J. Waale, R. P. Oosten, Pet. Refiner 38 (1959) no. 6, 189.
121. A. H. Olzinger, Chem. Eng. NY 82 (1975) no. 6, 50.
122. Universal Oil Products, US 3 763 259, 1973 (G. L. Hervert).
123. Mobil Oil Corp., EP-A 0 149 508, 1985 (W. W. Kaeding).
124. J. Ewers, H. W. Voges, G. Maleck, Erdöl Kohle Erdgas Petrochem. Compendium
74/75 (1975) 487.
125. K. Ito, I. Dogane, K. Tanaka, Hydrocarbon Process. 65 (1986) no. 11, 89.
126. H. A. Colvin, J. Muse, CHEMTECH 1986, no. 500.
127. K. Ito, Hydrocarbon Process. 52 (1973) no. 8, 89.
128. Phillips Petroleum Co., US 3 291 847, 1966 (P. F. Warner); US 3 316 315, 1967
(W. A. Jones).
129. K. Weissermel, H.-J. Arpe: Industrielle Organische Chemie, 3rd ed., VCH Verlagsge-
sellschaft, Weinheim, Germany 1987, pp. 72–79.
130. A. C. Olson, Ind. Eng. Chem. 52 (1960) no. 10, 833.
131. R. Ströbele, Chem. Ing. Tech. 36 (1964) 858.
132. H.-D. Wulf, T. Böhm-Gößl, L. Rohrschneider, Fette Seifen Anstrichm. 69 (1967) 32.
133. California Research Corp., US 2 477 382, 1949 (A. H. Lewis, A. C. Ettling); Chem.
Abstr. 44 (1950) 1137 e.
134. H. Hartig, Chem. Ztg. 99 (1975) 175.
135. K. Ichikawa, H. Shingu, Kogyo Kagaku Zasshi 54 (1951) 183–184; Chem. Abstr. 47
(1953) 8662 b.
136. S. Murohashi, Nagaoka Kogyo Koto Senmon Gakko Kenkyu Kiyo 9 (1973) 103–106;
Chem. Abstr. 80 (1974) 132917 r.
137. J. G. Matveev, D. A. Drapkina, R. L. Globus, Tr. IREA 21 (1956) 83–89, Chem.
Abstr. 52 (1958) 15474 e.
138. A. D. Harford, H. W. Vernon, US 2 897 112, 1956.
139. British Petroleum Co., Ltd., DE-AS 1 094 035, 1957 (A. Douglas, H. W. Vernon).
140. N. K. Moshinskaya, V. S. Olifer, L. I. Zhuoiev, SU 137 993, 1960.
141. E. Siggel, US 2 710 849, 1952.
142. BASF, DAS 1 142 338, 1960 (W. Braun, E. Hartwig, R. Krallmann).
143. FMC Corp., US 3 539 640, 1967 (M. I. Stewart, O. K. Carlson).
144. Ashland Oil & Refining Co., US 3 529 944, 1967 (A. M. Leas).
145. Japan Petrochemicals Co., Ltd., DE-OS 2 210 133, 1972 (A. Sato, Y. Aida, I. Shimizi).
146. H. Winkler: Der Steinkohlenteer, Verlag Glückauf, Essen 1951, p. 143.
147. J. Chipman, S. B. Peltier, Ind. Eng. Chem. 21 (1929) , 1106.
148. M. McEwen: Organic Coolant Handbook, sponsored by U.S. Atomic Energy Comm.,
publ. by Monsanto Co., St. Louis, Mo. 1958.
149. J. M. Cork, H. Mandel, N. Ewbank (North American Aviation Inc.), U.S. At. Energy
Comm. Sci. Rept. 5129 (1960) 1–28. Recalculated by A. N. Syverud, The Dow
Chemical Co. Thermal Laboratory, Midland, Mich.
648 HYDROCARBONS Vol. 2

150. M. McEwen, E. Wiederhold, Nucl. Eng. Part IV Symp. Ser. 5 (1959) no. 22, 9–15.
151. M. J. Foole, Chem. Eng. Prog. 58 (1962) no. 4, 37–40.
152. Hydrocarbon Process. 46 (1967) no. 11, 187.
153. S. Feigelman, C. B. O’Connor, Hydrocarbon Process. 45 (1966) no. 5, 140–144.
154. Hydrocarbon Process. Pet. Refin. 42 (1963) no. 11, 204.
155. Bayer, BIOS Final Rep. Nr. 1787.
156. R. Egbert, Trans. AICHE vol. 34 (1938) 435–488.
157. R. Dasgupta, B. R. Maiti, Ind. Eng. Chem. 25 (1986) 381–386.
158. Otkrytiya/Itobret, SU 1 165 677, 1985.
159. BIOS Final Rep. Nr. 893.
160. Soc. Progil, US 3 227 525, 1964 (M. E. Degeorges, M. Jaymond).
161. Soc. Progil, US 3 112 349, 1961 (M. E. Degeorges, M. Jaymond).
162. Soc. Progil, DE-AS 1 249 244, 1961 (M. Jaymond).
163. Dow Chemical Co., US 1 938 609, 1933 (J. H. Reilly).
164. Monsanto Co., US 3 228 994, 1962 (W. D. Robinson, W. Groves).
165. Hawker Siddeley Nuclear Power Co., FR 1 272 068, 1961.
166. Montecatini, BE 586 826, GB 865 302, IT Anm. 1144/59, 1960, IT Prior. 1959 (C.
Ferry, B. Ponzini).
167. Monsanto Co., US 2 143 509, 1939 (C. Conover, A. E. Huff).
168. Montecatini, IT 586 113, 1962.
169. Kambaror YU G; SU 6 79 565, 1977, Chem. Abstr. 92 (1980) 58407 d.
170. Monsanto Co., US 3 228 994, 1962 (W. D. Robinson).
171. N. M. Indyukov, R. I. Gasanova, Khim. prom-st. (Moscow) 1975, no. 7, 555.
172. Farbenfabr. Bayer AG, DE-AS 1 012 595, 1955 (H. Holzrichter, O. Tegtmeyer).
173. F. Ullmann, Justus Liebigs Ann. Chem. 332 (1904) 48.
174. L. Gattermann, R. Ehrhardt, Ber. Dtsch. Chem. Ges. 23 (1890) 1226.
175. I. H. Gardner, P. Bergstrom, J. Am. Chem. Soc. 51 (1962) 3376.
176. Asahi Kasei Kogyo K. K., DE-AS 2 039 904, 1970.
177. Shell Oil Co., US 3 145 237, 1962 (R. van Helden, B. Balder).
178. R. van Helden, G. Verberg, Rec. Trav. Chim. Pays-Bas 84 (1965) 1263.
179. Gulf Research & Developm. Co., US 3 401 207, 1966 (C. M. Selwitz).
180. E. S. Rudakov, A. I. Lutsyk, R. I. Rudakova, Kinet. Katal. 18 (1977) 525; Kinet.
Catal. (Engl. Transl.) 18 (1977) 441.
181. M. K. Starchevskii, M. N. Vargaftik, I. I. Moiseev, Izv. Akad. Nauk SSSR 1979,
242;(Engl. Transl.) 227.
182. Teijin Ltd., US 3 963 787, 1973 (Y. Ichikawa, T. Yamaji).
183. Ube Industries, JP 55 079 324, 1978.
184. L. N. Rachkovskaya et al., Kinet. Katal. 18 (1977) 792; Kinet. Catal. (Engl. Transl.)
18 (1977) 660.
185. Ube Industries, JP 80 141 417, 1979, Chem. Abstr. 94 (1981) 156 561 k.
186. British Petroleum, DE-OS 1 909 045, 1969 (D. Fiolka, C. R. Jentsch).
187. L. H. Slaugh, Tetrahedron 24 (1968) 4523.
188. L. H. Slaugh, J. A. Leonard, J. Catal. 13 (1969) 385.
189. G. N. Koshel et al., Dokl. Akad. Nauk SSSR 237 (1977) no. 1, 164 (Chem. Tech.);
Chem. Abstr. 88 (1978) 50 394 u.
190. Phillips Petroleum, US 4 123 470, 1977 (T. P. Murths).
191. Farberov M I, SU 416 339, 1972; Chem. Abstr. 88 (1978) 62 124 u.
192. Standard Oil, US 4 218 572, 1978 (S. R. Dolhyj, L. J. Velenyi).
193. Y. Yamazaki, T. Kawai, A. Yagishita, Sekiyu Gakkaishi 21 (1978) no. 4, 229.
194. Koppers Co., US 3 059 987 1962 (H. P. Baumann).
195. N. Hendrich, Melliand Textilber. 43 (1962) 258.
196. W. E. Bachmann, H. T. Carke, J. Am. Chem. Soc. 49 (1927) 2089–98.
Vol. 2 HYDROCARBONS 649

197. L. W. Pickett, Proc. R. Soc. London A 142 (1933) 333–346.


198. H. G. Römer, E. Hertel, Z. Phys. Chem. Abt. B 21 (1933) 292–296.
199. R. J. Good, E. E. Hardy, A. M. Ellenburg, H. B. Richards Jr., J. Am. Chem. Soc. 75
(1953) 436.
200. J. A. Ellard, W. H. Yanko: Thermodynamic Properties of Biphenyl and the Isomeric
Terphenyls, Final Report IDO-11,008, Monsanto Research Corp., U.S. At. Energy
Comm. Contract AT (10–1)-1088, Oct. 31, 1963.
201. R. O. Bolt, R. O. Johnson, R. S. Kirk: Organic Materials for Use as Reactor Coolant
Moderators, California Research Corp., June 13, 1956.
202. P. L. Geiringer: Handbook of Heat Transfer Media, Reinhold Publ., New York 1962,
p. 173.
203. W. H. Hedley, M. V. Mines, J. Chem. Eng. Data 15 (1970) 122.
204. H. Mandel, N. Eubank: Critical Constants of Diphenyl and Terphenyl, Report No.
NAA-SR 5129, Atomics International, Dec. 1960, p. 18.
205. N. Andrews, R. Ubbelohde, Proc. R. Soc. London 228 A (1955) 435–37.
206. E. Pajda, J. Rusin, Chemik Wkladka 1972, 422.
207. R. Balent, Proc. Am. Power Conf. 21 (1959) 120.
208. Soc. Progil, DE-AS 1 272 281, 1962 (M. Jaymond).
209. Farbenfabr. Bayer AG, GB 957 065 = FR 1 297 408, 1961.
210. Hawker Siddeley Nuclear Power Co., FR 1 272 068, 1961.
211. D. B. Orechkin, SU 570 584, 1977.
212. Monsanto Co., DE-OS 2 306 454, 1973 (J. K. Sears).
213. M. He, Huaxue Shiji 8 (1986) no. 1, 60–61; Chem. Abstr. 105 (1986) 208 532.
214. W. R. von Oettingen: ‘‘Toxicity and Potential Dangers of Aliphatic and Aromatic
Hydrocarbons,’’ Public Health Bull. (1940) no. 255.
215. E. A. Wahrenbrock, E. I. Eger, R. G. Laravuso, G. Maruschak, Anesthesiology 40
(1974) no. 1, 19.
216. A. H. Nuckolls, Underwriters Laboratory Report (Nov. 13, 1933) no. 2375.
217. Y. Henderson, H. W. Haggard: Noxious Gases, 2nd ed., Reinhold Publ. Co.,
New York 1943.
218. J. C. Krantz Jr., C. J. Carr, J. F. Vitcha, J. Pharmacol. Exp. Therap. 94 (1948) 315.
219. T. Ikoma, Nichidai Igaku Zasshi 31 (1972) no. 2, 71.
220. D. M. Aviado, S. Zakhari, T. Watanabe: Non-fluorinated Propellants and Solvents for
Aerosols, CRC Press, Cleveland, Ohio 1977, pp. 49–81.
221. F. A. Patty, W. P. Yant: ‘‘Odor Intensity and Symptoms Produced by Commercial
Propane, Butane, Pentane, Hexane and Heptane Vapor,’’ U.S. Bur. Mines, Rep.
Invest. 1929 no. 2979.
222. B. B. Shugaev, Arch. Environ. Health 18 (1969) 878.
223. Documentation of the Threshold Limit Values for Substances in Workroom Air, 3rd
ed., American Conference of Governmental Industrial Hygiene, Cincinnati 1971.
224. M. G. Rumsby, J. B. Finean, J. Neurochem. 13 (1966) 1513.
225. D. A. Haydon, B. M. Hendry, S. R. Levinson, J. Requena, Biochim. Biophys. Acta 470
(1977) 17.
226. N. Battistini et al., Riv. Patol. Nerv. Ment. 95 (1974) 871.
227. G. W. Paulson, G. W. Waylonis, Arch. Intern. Med. 136 (1976) 880.
228. N. W. Lazarew, Arch. Exp. Pathol. Pharmakol. 143 (1929) 223.
229. E. T. Kimura, D. M. Elery, P. W. Dodge, Toxicol. Appl. Pharmacol. 19 (1971) 699.
230. C. H. Hine, H. H. Zuidema, Ind. Med. 39 (1970) no. 5, 215.
231. H. H. Schaumburg, P. S. Spencer, Brain 99 (1976) 182.
232. K. Kurita, Jpn. J. Ind. Health Exp. 9 (1974) 672.
233. A. K. Ashbury, S. L. Nielsen, R. Telfer, J. Neuropathol. Exp. Neurol. 33 (1974) 191.
234. E. G. Gonzales, J. A. Downey, Arch. Phys. Med. Rehabil. 53 (1972) 333.
650 HYDROCARBONS Vol. 2

235. H. E. Swann Jr., B. K. Kwon, G. H. Hogan, W. M. Snellings, Am. Ind. Hyg. Assoc.
J. 35 (1974) no. 9, 311.
236. W. R. F. Notten, P. W. Henderson, Biochem. Pharm. 24 (1975) 1093.
237. C. F. Reinhardt et al., Arch. Environ. Health 22 (1971) 265.
238. A. Y. H. Lu, H. W. Strobel, M. J. Coon, Mol. Pharamcol. 6 (1970) 213.
239. N. W. Lazarew, Arch. Exp. Pathol. Pharmakol. 143 (1929) 223.
240. Criteria for a Recommended Standard–Occupational Exposure to Xylene, U.S. Dept.
Health, Education and Welfare, Natl. Inst. Occup. Safety and Health, Washington
D.C. (1975).
241. O. A. Anunziata et al., Appl. Catal. 15 (1985) 235–245. K. Verschueren: Handbook
of Environmental Data on Organic Chemicals, 2nd ed., Van Nostrand Reinhold, New
York 1983.
242. W. E. Coleman et al., Arch. Environ. Contam. Toxicol. 13 (1984) 171–178.
243. N. I. Sax: Dangerous Properties of Industrial Materials, 6th ed., Van Nostrand Re-
inhold, New York 1984.
244. Threshold Limit Values and Biological Exposure Indices for 1987–1988, American
Conference of Governmental Industrial Hygienists. Cincinnati, Ohio 1987.
245. DFG, MAK- und BAT-Werte Liste 2001, Mitteilung 37, Wiley-VCH, Weinheim,
Germany 2001.
246. H. W. Gerarde: Toxicology and Biochemistry of Aromatic Hydrocarbons, Elsevier
Publishing Co., London 1960.
247. G. Hommel: Handbuch der gefährlichen Güter, Springer Verlag, Berlin Heidelberg
1987.
248. H. W. Gerarde, Am. Med. Assoc. Arch. Ind. Health 19 (1959) 403.
249. M. A. Wolf et al., Am. Med. Assoc. Arch. Ind. Health 14 (1956) 387.
250. H. W. Werner, R. C. Dunn, W. F. von Oettingen, Ind. Health Toxicol. 26 (1974) 264.
251. Maximale Arbeitsplatzkonzentrationen und biologische Arbeitsstofftoleranzwerte,
Dt. Forschungsgemeinschaft, Verlag Chemie, Weinheim 1984.
252. N. I. Sax: Dangerous Properties of Industrial Materials, 6th ed., van Nostrand Rein-
hold. New York 1984, p. 819.
253. I. Hakkinen et al., Arch. Environ. Health 26 (1973) 70.
254. A. M. Seppalainen, I. Hakkinen, J. Neurol. Neurosurg. Psychiatry 38 (1975) 248.
255. H. E. Christensen, T. T. Luginbyhl: Registry of Toxic Effects of Chemical Substances,
U.S. Dept. of H. E. W., U.S. Government Printing Office, Washington, D.C. 1975.
256. J. Opdyke, Food Cosmet. Toxicol. 12 (1974) 707.
257. J. M. Haas, H. W. Earhart, A. S. Todd: Book of Papers, 1974 AATCC National Tech-
nical Conference, American Association of Textile Chemists and Colorists, Research
Triangle Park, N. C., p. 427.
258. Patty’s Industrial Hygiene and Toxicology, vol. 2 B, Toxicology, 3rd ed.
259. Z. F. Khromenko, Gig. Tr. Prof. Zabol. 1976, 44.
260. Y. R. Adamson, J. L. Weeks, Arch. Environ. Health 27 (1973) no. 8, 68.
261. Y. R. Adamson, J. M. Furlong, Arch. Environ. Health 28 (1974) no. 3, 155.
262. R. J. Hawkins, J. L. Weeks, At. Energy Can. Ltd. (Rep.) 4439 (1973); Chem. Abstr.
80 (1974) 65 963.
263. J. S. Henderson, J. L. Weeks, Ind. Med. Surg. 40 (1973) 10.

GENERAL REFERENCES

F. Asinger: Die Petrolchemische Industrie, vol. 1 (Gewinnung der Olefine), vol. 2


(Verwendung der Olefine) Akademie-Verlag, Berlin 1971.
J. Falbe (ed.): Methodium Chimicum, vol. 4, Thieme Verlag, Stuttgart 1980, pp. 35, 111.
Vol. 2 HYDROCARBONS 651

Houben-Weyl, 4th ed., vol. 5/1 b, Thieme Verlag, Stuttgart 1972.


S. Patai, J. Zabicky: The chemistry of alkenes, vol. 1 (1964), vol. 2 (1970), Interscience
Publishers, New York
B. Fell, Tenside Deterg. 12 (1975) 3.
H. Isa, CEER Chem. Econ. Eng. Rev., 9 (1977) 26.
A. H. Turner, J. Am. Oil Chem. Soc. 60 (1983) 623.
A. Behr, W. Keim, Arabian J. Sci. Eng. 10 (1985) 377.
H.-G. Franck, G. Collin: Steinkohlenteer, Springer Verlag, Berlin–Heidelberg–New York
1968.
H.-G. Franck, J. W. Stadelhofer: Industrielle Aromatenchemie, Springer Verlag,
Berlin–Heidelberg–New York–London –Paris–Tokyo 1987.
Beilstein, 5, 586; 5 (1), 274; 5 (2), 494; 5 (3), 1776; 5 (4), 1834.
Beilstein, 5, 515; 5 (1), 248; 5 (2), 410; 5 (3), 1355; 5 (4), 1532.
Beilstein, 5, 625; 5 (1), 300; 5 (2), 531; 5 (3), 1936; 5 (4), 2142.
Beilstein, 5, 685; 5 (1), 344; 5 (2), 609; 5 (3), 2276; 5 (4), 2463.
Beilstein, 5, 667; 5 (1), 327; 5 (2), 579; 5 (3), 2136; 5 (4), 2293.
Beilstein, 5, 693; 5 (1), 343; 5 (2), 609; 5 (3), 2279; 5 (4), 2467.

KARL GRIESBAUM
Universität Karlsruhe (TH) Karlsruhe
Federal Republic of Germany
ARNO BEHR
Henkel KGaA, Düsseldorf Federal Republic
of Germany
DIETER BIEDENKAPP
BASF Aktiengesellschaft Ludwigshafen,
Federal Republic of Germany
HEINZ-WERNER VOGES
Hüls Aktiengesellschaft, Marl, Federal
Republic of Germany
DOROTHEA GARBE
Haarmann & Reimer GmbH Holzminden
Federal Republic of Germany
CHRISTIAN PAETZ
Bayer AG Leverkusen, Federal Republic of
Germany
GERD COLLIN
Rüttgerswerke AG, Duisburg, Federal
Republic of Germany
DIETER MAYER
Hoechst Aktiengesellschaft, Frankfurt
Federal Republic of Germany
HARTMUT HÖKE
Rüttgerswerke AG, Castrop-Rauxel,
Federal Republic of Germany
652 ACETYLENE Vol. 2

ACETYLENE
1. Introduction

Acetylene [74-86-2] is the simplest hydrocarbon with a triple bond. In the days
before oil gained widespread acceptance as the main feedstock of chemical
industry, acetylene was the predominant building block of industrial organic
chemistry. The calcium carbide process was the sole route for acetylene produc-
tion until 1940, when thermal cracking processes using methane and other
hydrocarbons were introduced. At first, these processes used an electric
arc; then, in the 1950s, partial oxidation and regenerative processes were
developed.
However, along with the expansion of the petroleum industry there was a
changeover from coal chemistry to petrochemistry, in the 1940s in the United
States and in the 1950s in Europe. As a consequence, acetylene lost its compe-
titive position to the much cheaper and more readily available naphtha-derived
ethylene and other olefins. This competition between acetylene and ethylene as
feedstocks for chemical industry has been much discussed over the last
20 years (1,2). The few hopes, such as BASF’s contribution to the submerged
flame process, Hoechst’s crude oil cracking (HTP), or Hüls’ plasma process,
have not halted the clear trend toward ethylene as a basic chemical. With
the first oil price explosion in 1973, the development of crude cracking pro-
cesses suffered a setback, and the new processes, such as the Kureha/Union
Carbide process, DOW’s PCC process (PCC ¼ partial combustion cracking), or
the Kureha/Chiyoda/Union Carbide ACR process (ACR ¼ advanced cracking
reactor), raise little hope for a comeback of acetylene chemistry. Acetylene pro-
duction peaked in the United States at 480000 t in the 1960s, and in Germany
at 350000 t in the early 1970s (3). Since then, acetylene production has
decreased steadily. In both countries the losses were principally in carbide-
derived acetylene; in fact, Germany has produced acetylene for chemical
purposes almost exclusively from natural gas and petrochemical sources
since 1975.
All acetylene processes, including carbide processes, are high-temperature
processes, requiring a large amount of energy. They differ essentially only in
the manner in which the necessary energy is generated and transferred. They
can be classified into three groups: partial combustion processes, electrothermic
processes, and processes using heat carriers. Finally, the use of byproduct acet-
ylene from olefin plants is economically viable in many cases. For each group of
acetylene processes several variants have been developed using various feed-
stocks and techniques. Today, only three processes remain for the commercial
production of acetylene: the calcium carbide route, in which the carbide is pro-
duced electrically, the arc process, and the partial oxidation of natural gas.
Other once popular processes have become uneconomical as the price of naphtha
has increased.
Some processes were shelved in the experimental or pilot-plant stage as the
importance of acetylene declined. However, other new processes involving the
use of coal, sulfur-containing crude oil, or residues as feedstocks for acetylene
production are in the pilot-plant stage.

ß2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim


10.1002/14356007.a01_097
Vol. 2 ACETYLENE 653

However, the position of acetylene in chemical industry may improve


because of the variety of valuable products to which acetylene can be converted
with known technology and high yields.

2. Physical Properties

Due to the carbon – carbon triple bond and the high positive energy of formation,
acetylene is an unstable, highly reactive unsaturated hydrocarbon. The C – C
triple bond and C – H s bond lengths are 0.1205 and 0.1059 nm, respectively.
For the electronic structure of acetylene and a molecular orbital description,
see (6). The acidity of acetylene (pKa ¼ 25) permits the formation of acetylides
(see Section 3.2).
Under normal conditions acetylene is a colorless, nontoxic but narcotic gas;
it is slightly lighter than air. The main physical properties are listed in Table 1.
The critical temperature and pressure are 308.32 K and 6.139 MPa. The triple
point at 128.3 kPa is 192.4 K. The vapor pressure curve for acetylene is shown
in Figure 1. The formation of acetylene is strongly endothermic (DHf ¼ þ 227.5
kJ/mol at 298.15 K).
Self-decomposition can be initiated when certain pressure limits above
atmospheric pressure are exceeded (for details see Section 5.1).
The crystalline structure of solid acetylene changes at  140.15 8C from a
cubic to an orthorhombic phase. The heat of reaction for this phase change is
2.54 kJ/mol (8); two different values for the enthalpy of fusion are reported in
the literature (6,7). Figure 2 shows the density of liquid and gaseous acetylene.
Details about flame properties, decomposition, and safety measures are
given in Section 5.

Table 1. Physical Properties of Acetylene (6–8)


Molecular mass 26.0379
Critical temperature 308.32 K (35.17 8C)
Critical pressure 6.139 MPa
Critical volume 0.113 m3/kmol
Triple point 192.4 K ( 80.75 8C)
Triple point pressure 128.3 kPa
Normal sublimation point and normal boiling point 189.15 K ( 84.0 8C)
Crystal transition point 133.0 K ( 140.15 8C)
Enthalpy of transition 2.54 kJ/mol
Density 760.2 kg/m3 (131 K)
764.3 kg/m3 (141 K)
Density (liquid C2H2) 465.2 kg/m3 (273.15 K)
Enthalpy of vaporization 10.65 kJ/mol (273.15 K)
(calculated)
Enthalpy of sublimation 21.168 kJ/mol (5.55 K)
Enthalpy of formation 227.5  1.0 kJ/mol (298.15 K)
Gibbs free energy of formation 209.2  1.0 kJ/mol (298.15 K)
Entropy of formation 200.8 J mol1 K1 (298.15 K)
Enthalpy of combustion 1255.6 kJ/mol (298.15 K)
Vapor pressure 2.6633 MPa (273.15 K)
Heat capacity (ideal gas state) 43.990 J mol1 K1 (298.15 K)
654 ACETYLENE Vol. 2

Fig. 1. Vapor pressure of acetylene (6,7).

Fig. 2. Density of acetylene vapor (at 1.013 bar) and liquid.


Vol. 2 ACETYLENE 655

Table 2. Solubility Coefficients of C2H2 in Various Solvents (in mol kg1 bar1) (9)
Solvent C2H2 pressure, bar  208C 258C
Methanol 0.98 1.979 0.569
Ethanol 0.98 0.851 0.318
n-Butanol 0.245 – 0.657 0.237
1,2-Dichlorethane 0.4–1.05 0.569 0.218
Carbon tetrachloride 0.98 0.164 0.075
n-Hexane 6.90 0.523 0.264
n-Octane 0.196 – 14.71 0.205 0.146 (08C)
Benzene 0.98 0.225
Toluene 0.98 0.619 0.214
Xylene (tech.) 0.98 0.528 0.189
4-Methyl-1,3-dioxalan-2-one 0.98 1.137 0.350
Tri-n-butylphosphate 0–0.4 2.366 0.614
Methyl acetate 0.98 2.912 0.878
Triethylene glycol 0.98 0.205
Acetone 0.98 4.231 1.069
N-Methyl-2-pyrrolidone 0.98 5.687 1.319
N,N-Dimethylformamide 0.98 5.096 1.501
Dimethyl sulfoxide 0.98 1.001
Ammonia 0.98 7.052 2.229

Solubility coefficients of acetylene in organic solvents are listed in Table 2


(9). Further solubility data are available as Bunsen absorption coefficients a
(208C, m3 (STP) m3 atm1), as solubilities (g/kg of solvent), and for different
pressures (see (10)). The solubilities of acetylene at infinite dilution are shown
in Figure 3 for water, methanol, DMF, and N-methyl-2-pyrrolidone (NMP)
[872-50-4]. Figure 4 shows the solubility of acetylene in acetone for various par-
tial pressures and temperatures. The heat of solution depends on the concentra-
tion of acetylene in the solvent: dissolving 0.5 kg of acetylene in 1 kg of solvent
generates 293 kJ for acetone and 335 kJ for DMF. For details on the influence
of water, of partial pressure, and deviations from Henry’s law, see (9,10). The
temperature dependence of the solubility of acetylene in DMF at infinite dilution
is compared with those of ethylene and ethane in Figure 5 (see also (9,11) for
selectivities).
The solubility of acetylene in water at 258C is 0.042 mol L1 bar1. Under
pressure of acetylene (e.g, >0.5 MPa at 08C) and at temperatures between 268
and 283 K, waxy hydrates of the composition C2H2  (H2O)5.8 are formed (9,10).
The hydrates can block equipment; shock waves may initiate self-decomposition.
Liquid oxygen dissolves only traces of acetylene (5.5 ppm at 90 K (12)); the
solubilities of ethylene and ethane in oxygen are much higher (factor of 350 and
2280, respectively). The prepurification of the process air in air separation plants
with molecular sieves removes acetylene to <1 ppb provided there is no break-
through of carbon dioxide. This fact guarantees a safe operation of the down-
stream equipment (13).
Typical adsorption isotherms of acetylene are shown in Figure 6 for molecu-
lar sieves, activated carbon and silica gel at 258C (14); additional information for
activated carbon is summarized in (15).
656 ACETYLENE Vol. 2

Fig. 3. Solubility of acetylene in various solvents at infinite dilution.

3. Chemical Properties

Because of its strongly unsaturated character and high positive free energy of for-
mation, acetylene reacts readily with many elements and compounds. As a result
acetylene is used as raw material for a great variety of substances. Important are
addition reactions, hydrogen replacements, polymerization, and cyclization.
Acetylene is more susceptible to nucleophilic attack than, for instance, ethy-
lene. In addition, the polarized C-H bond makes acetylene acidic (pKa ¼ 25) (16).

Fig. 4. Solubility of acetylene in acetone (9).


Vol. 2 ACETYLENE 657

Fig. 5. Solubility of C2 hydrocarbons in DMF at infinite dilution.

Because of this acidity, acetylene is very soluble in basic solvents (17,18), forming
hydrogen bonds with them (19). Therefore, the vapor pressures of such solutions
cannot be described by Raoult’s law (20).
The development of the acetylene pressure reactions by W. REPPE (1892 –
1969), BASF Ludwigshafen (Federal Republic of Germany) (21–23) began mod-
ern acetylene chemistry. The most interesting groups of reactions are vinylation,
ethynylation, carbonylation, and cyclic and linear polymerization.

Fig. 6. Adsorption isotherms for acetylene on 4A and 5A molecular sieves, activated


carbon, and silica gel at 25 8C (14).
658 ACETYLENE Vol. 2

3.1. Industrially Important Reactions. Vinylation Reactions and


Products(24). Vinylation is the addition of compounds with a mobile hydrogen
atom, such as water, alcohols, thiols, amines, and organic and inorganic acids, to
acetylene to form vinyl compounds chiefly used for polymerization.
The two types of vinylation reactions are heterovinylation and the less usual C
vinylation. In the former, the hydrogen atom originates from the heteroatoms O, S,
and N, whereas C vinylation occurs when the mobile hydrogen atom is directly
bound to a carbon atom. Examples of C vinylation are dimerization and trimeriza-
tion of acetylene, the synthesis of acrylonitrile from acetylene and hydrogen cya-
nide, and the addition of acetylene to unsaturated hydrocarbons with activated
hydrogen atoms, such as cyclopentadiene, indene, fluorene, and anthracene.
The first industrial vinylation products were acetaldehyde, vinyl chloride,
and vinyl acetate. Many other products followed.
Some examples of industrial vinylation processes are given below:
Acetaldehyde [75-07-0]:

HC  CH þ H2 O ! CH3 CHO

Catalyst: acidic solutions of mercury salts, such as HgSO4 in H2SO4 .


Liquid-phase reaction at 928C.
Vinyl chloride [75-01-4]. (! Chlorinated Hydrocarbons):

HC  CH þ HCl ! CH2 ¼ CHCl

Catalyst: HgCl2 on coal. Gas-phase reaction at 150 – 1808C.


Vinyl acetate [108-05-04] (! Vinyl Esters):

HC  CH þ CH3 COOH ! CH2 ¼ CHOOCCH3

Catalyst: cadmium, zinc, or mercury salts on coal. Gas-phase reaction at


180–2008C.
Vinyl ethers conjectured reaction steps:

where R is an alkyl group. Reaction temperature of 120–1508C; pressure


high enough to avoid boiling the alcohol used, e.g, 2 MPa with methanol to pro-
duce methyl vinyl ether (acetylene pressure reaction).
Vinyl phenyl ether [766-94-9], vinylation with KOH catalyst:

Vinyl sulfides, KOH catalyst:

HC  CH þ RSH ! CH2 ¼ CH  S  R
Vol. 2 ACETYLENE 659

Vinyl esters of higher carboxylic acids:

HC  CH þ R  COOH ! RCOO  CH ¼ CH2

Catalyst: zinc or cadmium salts. Liquid-phase reaction.


Vinyl amines, vinylation with zinc or cadmium compounds as catalyst:

R1 R2 NH þ HC  CH ! R1 R2 N  CH ¼ CH2

where R1 and R2 are alkyl groups.


N-Vinylcarbazole [1484-13-5], vinylation of carbazole in a solvent, e.g,
N-methylpyrrolidone, at 1808C.
Vinylation of ammonia,complex Co and Ni salts as catalysts, reaction tem-
perature of 958C:

Vinylation of acid amides, potassium salt of the amide as catalyst:

HC  CH þ RCO  NH2 ! RCO  NH  CH ¼ CH2

N-Vinyl-2-pyrrolidone [88-12-0], vinylation of 2-pyrrolidone with the potas-


sium salt of the pyrrolidone as catalyst.
Acrylonitrile [107-13-1], C-vinylation of HCN in aqueous hydrochloric acid
with CuCl and NH4Cl catalyst:

HC  CH þ HCN H2 C ¼ CH  CN

Ethynylation Reactions and Products (25). Ethynylation is the addi-


tion of carbonyl compounds to acetylene with the triple bond remaining intact.
REPPE found that heavy metal acetylides (see Section 3.2), especially the copper(I)
acetylide of composition Cu2C2  2 H2O  2 C2H2 , are suitable catalysts for the
reaction of aldehydes with acetylene. Alkaline catalysts are more effective
than copper acetylide for the ethynylation of ketones. The generalized reaction
scheme for ethynylation is:

HC  CH þ RCOR0 ! HC  C  CðOHÞRR0

where R and R0 are alkyl groups or H.


The most important products from ethynylation are propargyl alcohol and
butynediol.
Propargyl alcohol, 2-propyn-1-ol [107-19-7]:

CH  CH þ HCHO ! HC  CCH2 OH

Catalyst: Cu2C2  2 H2O  2 C2H2 .


660 ACETYLENE Vol. 2

Butynediol, 2-butyne-1,4-diol [110-65-6]:

HC  CH þ 2HCHO HOCH2 C  CCH2 OH

Catalyst: Cu2C2  2 H2O  2 C2H2.


Other examples of ethynylation are the reactions of aminoalkanol and
secondary amines with acetylene:

Carbonylation Reactions and Products (26). Carbonylation is the rea-


ction of acetylene and carbon monoxide with a compound having a mobile hydrogen
atom, such as water, alcohols, thiols, or amines. These reactions are catalyzed
by metal carbonyls, e.g, nickel carbonyl, Ni(CO)4 [13463-39-3]. Instead of metal
carbonyls, the halides of metals that can form carbonyls can also be used.
Acrylic acid [79-10-7]:

HC  CH þ CO þ H2 O ! CH2 ¼ CH  COOH

The reaction of acetylene with water or alcohols and carbon monoxide using
Ni(CO)4 catalyst was first reported by W. REPPE (26). If water is replaced by
thiols, amines, or carboxylic acids, then thioesters of acrylic acid, acrylic amides,
or carboxylic acid anhydrides are obtained.
Ethyl acrylate [140-88-5]:

C2 H2 þ 4C2 H5 OH þ NiðCOÞ4 þ 2HCl


ð1Þ
! 4CH2 ¼ CHCOOC2 H5 þ H2 þ NiCl2

C2 H2 þ C2 H5 OH þ CO ! CH2 ¼ CHCOOC2 H5 ð2Þ

Catalyst: nickel salts. Reaction temperature: 30–508C. The process starts


with the stoichiometric reaction (1); afterwards, most of the acrylate is formed
by the catalytic reaction (2). The nickel chloride formed in the stoichiometric
reaction (1) is recovered and recycled for carbonyl synthesis.
Hydroquinone [123-31-9] is formed in a suitable solvent, e.g, dioxane, at
1708C and 70 MPa (27). The catalyst is Fe(CO)5 :
Vol. 2 ACETYLENE 661

Hydroquinone is formed at 0–1008C and 5–35 MPa if a ruthenium carbonyl


compound is used as catalyst (28):

Bifurandiones: The reaction of acetylene and CO in the presence of octa-


carbonyldicobalt, (CO)3Co–(CO)2 – Co(CO)3 [10210-68-1], forms a cis–trans mix-
ture of bifurandione. The reaction is carried out under pressure (20–100 MPa) at
temperatures of about 1008C (29):

New aspects of such CO insertion reactions have been reported (30).


Cyclization and Polymerization of Acetylene. In the presence of sui-
table catalysts, acetylene can react with itself to form cyclic and linear polymers.
Cyclization was first observed by BERTHELOT, who polymerized acetylene to a
mixture of aromatic compounds including benzene and naphthalene. In 1940,
REPPE synthesized 1,3,5,7-cyclooctatetraene [629-20-9] with a 70% yield at an
only slightly elevated pressure:

Reaction temperature of 65–1158C, pressure of 1.5–2.5 MPa, Ni(CN)2 catalyst.


The reaction is carried out in anhydrous tetrahydrofuran. The byproducts
are mostly benzene (about 15%), chain oligomers of acetylene of the empirical for-
mulas C10H10 and C12H12 , and a black insoluble mass, called niprene after the
nickel catalyst.
If dicarbonylbis(triphenylphosphine)nickel [13007-90-4], Ni(CO)2[(C6H5)3P]2 ,
is used as catalyst, the cyclization products are benzene (88% yield) and styrene
(12% yield). The reaction is carried out in benzene at 65–758C and 1.5 MPa (31,32).
Linear polymerization of acetylene occurs in the presence of a copper (I) salt
such as CuCl in hydrochloric acid. Reaction products are vinylacetylene, diviny-
lacetylene, etc. (33):

HC  CH þ HC  CH ! H2 C ¼ CH  C  CH
662 ACETYLENE Vol. 2

A particular polymerization product, known as cuprene, is formed when


acetylene is heated to 2258C in contact with copper sponge. Cuprene is chemi-
cally inert, corklike in texture, and yellow to dark brown.
Polyacetylene (34,35) is formed with Ziegler–Natta catalysts, e.g, a mixture
of triethylaluminum, Al(C2H5)3 , and titanium tetrabutoxide, Ti(n-OC4H9)4 , at
102 to 1 MPa:

Polymerization can be carried out in an auxiliary inert liquid, such as an aliphatic


oil or petroleum ether. The monomer can also be copolymerized in the gas
phase.
Polyacetylene is a low-density sponge-like material consisting of fibrils
with diameters of 20–50 nm. The ratio cis- to trans-polyacetylene depends on
the reaction temperature.
Polyacetylene doped with electron acceptors (I2 , AsF5), electron donors (Na, K),
or protonic dopants (HClO4 , H2SO4) is highly conductive and has the properties of a
one-dimensional metal (35).
3.2. Other Reactions; Derivatives. Metal Acetylides (36). The hydro-
gen atoms of the acetylene molecule can be replaced by metal atoms (M) to
yield metal acetylides. Alkali and alkaline-earth acetylides can be prepared via
the metal amide in anhydrous liquid ammonia:

C2 H2 þ MNH2 MC2 H þ NH3

The direct reaction of the acetylene with a molten metal, such as sodium, or
with a finely divided metal in an inert solvent, such as xylene, tetrahydrofuran,
or dioxane, at a temperature of about 408C, is also possible:

2M þ C2 H2 ! M2 C2 þ H2

The very explosive copper acetylides, e.g, Cu2C2  H2O, can be obtained by
reaction of copper(I) salts with acetylene in liquid ammonia or by reaction of
copper(II) salts with acetylene in basic solution in the presence of a reducing
agent such as hydroxylamine. Copper acetylides can also form from copper oxides
and other copper salts. For this reason copper plumbing should be avoided in
acetylene systems.
Vol. 2 ACETYLENE 663

Silver, gold, and mercury acetylides, which can be prepared in a similar


manner, are also explosive.
In sharp contrast to the highly explosive Cu2C2  H2O, the catalyst used for
the synthesis of butynediol, Cu2C2  2 H2O  2 C2H2 , is not as sensitive to shock or
ignition.
Halogenation. The addition of chlorine to acetylene in the presence of FeCl3
yields 1,1,2,2-tetrachloroethane [79-34-5], an intermediate in the production of
the solvents 1,2-dichloroethylene [540-59-0], trichloroethylene [79-01-6], and
perchloroethylene [127-18-4].
Bromine and iodine can also be added to acetylene. The addition of iodine to
acetylene stops with formation of 1,2-diiodoethylene.
Hydrogenation. Acetylene can be hydrogenated, partly or completely, in the
presence of Pt, Pd, or Ni catalysts, giving ethylene or ethane.
Organic Silicon Compounds (37,38). The addition of silanes, such as
HSiCl3 , can be carried out in the liquid phase using platinum or platinum com-
pounds as catalysts:

HC  CH þ HSiCl3 ! CH2 ¼ CH  SiCl3

Oxidation. At ambient temperature acetylene is not attacked by oxygen;


however, it can form explosive mixtures with air or oxygen (see Chap. 5).
The explosions are initiated by heat or ignition. With oxidizing agents such
as ozone or chromic acid, acetylene gives formic acid, carbon dioxide, and
other oxidation products. The reaction of acetylene with dilute ozone yields
glyoxal.
Hydrates. At temperatures below ca. 158C, under pressure, hydrates of the
composition C2H2  6 H2O are formed (see Section 2).
Chloroacetylenes (39). Monochloroacetylene, HC:CCl, Mr 60.49, bp
32 to 308C, a gas with nauseating odor that irritates the mucous membranes,
is obtained by reaction of 1,2-dichloroethylene with alcoholic NaOH in the pre-
sence of Hg(CN)2. It ignites in the presence of traces of oxygen. In air it explodes
violently. Chloroacetylene is very poisonous.
Dichloroacetylene, ClC:CCl, Mr 94.93, mp  66 to  64.28C, a colorless oil
of unpleasant odor, explodes in the presence of air or on heating. It is obtained
from acetylene in strongly alkaline potassium hypochlorite solution (40) or by
reaction of trichloroethylene vapor with caustic alkali.

4. Production

4.1. Thermodynamic and Kinetic Aspects. The production of acety-


lene from hydrocarbons, e.g,

2CH4 ! C2 H2 þ 3H2 DHð298KÞ ¼ 376:4kJ=mol

requires very high temperatures and very short reaction times. The
main reasons for the extreme conditions are the temperature dependence of
the thermodynamic properties (molar enthalpy of formation, DHf , and molar
664 ACETYLENE Vol. 2

Table 3. Standard Molar Enthalpies of Formation and Gibbs Free Energy of Formation at
298 K
DHf (kJ/mol) DGf (kJ/mol)
C (s) 0 0
H2 (g) 0 0
CH4 (g)  74.81  50.82
C2H2 (g) +226.90 +209.30
C2H4 (g) + 52.30 + 68.15
C2H6 (g)  84.64  32.90
C3H6 (g) + 20.43 + 62.75
C3H8 (g) 103.90  23.48
n-C4H10 (g) 126.11  17.10

free energy of formation, DGf) of the hydrocarbons; the position of the chemical
equilibria under the reaction conditions; and the kinetics of the reaction.
Thermodynamic data relevant to the hydrocarbon–acetylene system are
shown in Table 3 and Figure 7. These data show clearly that at normal tempera-
tures acetylene is highly unstable compared to the other hydrocarbons. However,
Figure 7 also shows that the free energy of acetylene decreases as temperature
increases, whereas the free energies of the other hydrocarbons increase. Above
about 12308C, acetylene is more stable than the other hydrocarbons. The tem-
perature at which the acetylene line intersects an other line in Figure 7 is higher
the shorter the chain length of the hydrocarbons. Acetylene production from
methane requires higher reaction temperatures than production from heavier
hydrocarbons.

Fig. 7. Gibbs free energy of formation per carbon atom of several hydrocarbons as a
function of temperature.
Vol. 2 ACETYLENE 665

Fig. 8. Equilibrium curve for the methane cracking reaction, 2 CH4 Ð C2H2 þ 3 H2.

The equilibrium curve for the methane reaction as a function of tempera-


ture (Fig. 8) shows that acetylene formation only becomes apparent above 1000
K (7308C). Therefore, a very large energy input, applied at high temperature, is
required.
However, even at these high temperatures acetylene is still less stable than
its component elements, carbon und hydrogen (see Fig. 7). In fact, the large dif-
ference in free energy between acetylene and its component elements favors the
decomposition of acetylene to carbon and hydrogen up to temperatures of about
4200 K.

C2 H2 ! 2CðsÞ þ H2 ðgÞ  DGf ð298KÞ ¼ 209:3 kJ=mol

Thus cracking and recombination of the hydrocarbons and decomposition of


acetylene compete. To achieve reasonable acetylene yields and to avoid the ther-
modynamically favorable decomposition into the elements, rapid quenching of
acetylene produced in the cracking reaction is necessary. In practice, the resi-
dence time at high temperature is between 0.1 and 10 ms.
Higher temperatures also increase the rate of conversion of acetylene
to byproducts. Again, the residence time must be sufficiently short to prevent
this.
In the case of cracking by partial oxidation, the combustion reaction of the
hydrocarbon supplies the energy necessary for the production of acetylene from
the other part of the hydrocarbon feed:

CH4 þ O2 ! CO þ H2 þ H2 O DHð298KÞ ¼ 277:53kJ=mol

CO þ H2 O ! CO2 þ H2 DHð298KÞ ¼ 41:19kJ=mol

From these reaction enthalpies, the amount of oxygen needed to produce


the high reaction temperature can be calculated. Therefore, in addition to the
short residence time, the correct methane : oxygen ratio, which also determines
the reaction temperature, is essential to obtain good acetylene yields.
666 ACETYLENE Vol. 2

4.2. Partial Combustion Processes. In this group of processes, part of


the feed is burnt to reach the reaction temperature and supply the heat of
reaction. The necessary energy is produced where it is needed. Almost all car-
bon-containing raw materials can be used as feedstocks: methane, ethane, nat-
ural gas liquids (NGL), liquefied petroleum gas (LPG), naphtha, vacuum gas oil,
residues, and even coal or coke. Natural gas is especially suitable because it is
available in many parts of the world and because its only other uses are for heat-
ing and for the production of synthesis gas. Only under the conditions of acety-
lene synthesis can methane be transformed into another hydrocarbon in a single
process step, and this is the essential reason for using the thermodynamically
unfavorable acetylene synthesis.
The partial combustion processes for light hydrocarbons, from methane to
naphtha, all follow similar schemes. The feed and a certain amount of oxygen
are preheated separately and introduced into a burner. There they pass through
a mixing zone and a burner block into the reaction zone, where they are ignited. On
leaving the reaction zone the product mixture is cooled rapidly, either by water or
oil. Cooling by water is easier, and more common, but it is thermally less efficient
than cooling by oil. Alternatively, the gases can be cooled with light hydrocarbon
liquids, which leads to additional acetylene and ethylene formation between
1500 and 8008C. These processes are usually called two-step processes.
Burner design is very important for all partial combustion processes. The
residence time of the gas in the reaction zone must be very short, on the order
of a few milliseconds, and it should be as uniform as possible for all parts of
the gas. Flow velocity within the reaction zone is fixed within narrow limits by
the requirements of high yield and the avoidance of preignition, flame separation
from the burner block, and coke depositions. A survey of the processes operating
according to these principles is given in (8,41). Only the BASF process is
described here in detail, because it is the most widely used process for the partial
combustion of natural gas.
The submerged flame process, SFP, was developed by BASF with the aim of
producing acetylene from crude oil or its heavy fractions, and thus to be indepen-
dent of the more expensive refined oil products used in olefin chemistry. One unit
of this kind was built in Italy, but it became uneconomic and was shut down after a
year of operation (42). Nevertheless, the process is described in some detail below
because of its simple cracking section, because of the simultaneous formation of
acetylene and ethylene, and because of its high thermal efficiency and its high
degree of carbon conversion (perhaps of even greater importance in the future).
The partial combustion carbide process, also developed by BASF, uses coke,
oxygen, and lime as feed. It was developed in the 1950s to reestablish the com-
petitive position of carbide in the face of the new acetylene processes on a petro-
chemical–natural gas basis. Some attention is given here to the basics of this
process, although it has never gone beyond the pilot-plant stage. When petro-
chemical feedstocks become scarce, this process may have a place in a future
coalbased chemistry because it has a higher degree of carbon conversion and a
higher thermal efficiency than the electric carbide process.
All these acetylene processes based on partial combustion yield a number
of byproducts, such as hydrogen and/or carbon monoxide, which may cause
problems if acetylene is the only product desired. Within a complex chemical
Vol. 2 ACETYLENE 667

plant, however, these may be converted to synthesis gas, pure hydrogen, and
pure CO and can actually improve the economics of acetylene production.
BASF Process (Sachsse-Bartholome). The BASF process for the pro-
duction of acetylene from natural gas has been known since 1950 (43). World-
wide, some 13 plants used this process in 1983, a total capacity of about
400000 t/a. All use a water quench, except the plant in Ludwigshafen (Federal
Republic of Germany) operated with an oil quench (44).
The basic idea of partial combustion involves a flame reaction on a premixed
feed of hydrocarbon and oxygen. In this way the rate of hydrocarbon conversion
is made independent of the gas-mixing rate, which is governed by diffusion. Only
then can the residence time in the reaction zone be made much smaller than the
average decay time of acetylene. The separate preheating of the reactants to the
highest temperature possible before introduction into the burner reduces the
consumption of oxygen and the hydrocarbon within the burner. It also causes
a higher flame propagation speed and therefore a higher mass flow within the
acetylene burner.
The smallest, but most important, part of a partial oxidation acetylene
plant is the burner, Figure 9. Its design is nearly identical in the two process var-
iants (i.e, oil and water quench).
At the top of the burner, the preheated reactants, (6008C in the case of
methane) must be mixed (c) so rapidly that there are no domains with a high oxy-
gen concentration. Such domains cause preignition before the reactants are
introduced into the reaction zone (g). In fact, the reaction mixture ignites after
an induction time depending on the hydrocarbon used as feed and on the preheat
temperature, on the order of a few tenths of a second. The maximum preheat
temperature is lower for higher hydrocarbons than for methane. Backmixing
of the gas between the mixing and the reaction zones is avoided by the diffuser
(e), a tube which connects the mixing zone and the burner block (f). Because of its
smooth surface and the small opening angle the reaction feed is decelerated
gently and backmixing does not occur.
The burner block (f) consists of a water-cooled steel plate with a large number
of small channels. The flow velocity through these channels is substantially higher
than the flame propagation speed, so that the flame below the burner block cannot
backfire into the diffuser. The lower side of the burner block has small openings
between the channels through which additional oxygen is fed into the reaction
mixture. At these openings small flames form and initiate the flame reaction.
The strong turbulence below the burner block stabilizes the flame.
Under unfavorable conditions the flame may appear above the burner
block. In this case the oxygen feed must be shut off immediately and replaced
by nitrogen. This extinguishes the preignition before it can cause any damage
to the equipment. Such preignitions can result from a momentary shift in the
oxygen : hydrocarbon ratio or the entrainment of small particles of pyrophoric
iron formed from rust in the preheaters.
As mentioned above, the hot gas leaves the reaction chamber within a few
milliseconds and passes through sprays of water or oil, which cool the gas almost
instantaneously, to about 808C in the case of water or 200–2508C in the case of
oil. The quench system consists of a set of nozzles that are fed by three annular
tubes below the reaction chamber.
668 ACETYLENE Vol. 2

Fig. 9. BASF acetylene burner. A) The burner: a) Oxygen: b) Hydrocarbon; c) Mixer;


d) Concrete lining; e) Diffuser; f) Burner block; g) Reaction chamber; h) Rupture disk;
i) Quench-medium inlet; j) Quench rings; k) Quench chamber; l) Manual scraper; m)
Cracked-gas outlet; n) Quench-medium outlet. B) The burner block.

The concentrations of the major constituents of the cracked gas depend on


the oxygen : hydrocarbon ratio in the feed as shown in Figure 10. As the oxygen
supply is increased, the acetylene concentration increases until it passes through
a smooth maximum. At the same time there is an increase in the volume of the
cracked gas. Thus maximum acetylene production is attained when a little more
oxygen is used than the amount required for maximum acetylene concentration
in the cracked gas. This is clear from the consumption of natural gas per ton
of acetylene produced and the reduction in unconverted methane. When the
oxygen : hydrocarbon ratio is too low, the reaction time is insufficient for
complete conversion of oxygen, and the cracked gas contains free oxygen. Free
oxygen can be tolerated only up to a certain concentration. When the oxygen :
Vol. 2 ACETYLENE 669

Fig. 10. Burner characteristics. a) Burner block; b) Reaction chamber; c) Flame


front; d) Quench-medium inlet.

hydrocarbon ratio is too high, the increased velocity of flame propagation


exceeds the flow velocity in the channels of the burner block, leading to
preignitions.
Coke deposits in the reaction chamber have to be removed from time to time
with a manual or an automatic scraper. Normally, a burner produces 25 t of acet-
ylene per day from natural gas and 30 t per day from liquid feedstocks.
Acetylene Water Quench Process (AWP), Soot Removal (Figure 11).
After quenching with water the cracked gas leaves the burner (b) at 80–908C. A
certain amount of soot is formed in the reaction chamber in spite of the very short
reaction time. When natural gas is used as a feedstock, the soot is 50 kg per ton of
acetylene, with LPG feedstock it is 250 kg, and with naphtha it is 350 kg. The
soot is partly removed from the gas by the quench, then by washing with recir-
culated water in a cooling column (c), and by passing the gas through an electro-
filter (d). After cooling and soot removal, the gas has a pressure slightly above
atmospheric, a temperature of about 308C, and a soot content of about 1 mg/
m3. The water effluents from the quench system, the cooling column, and the
electrofilter carry the washed-out soot. Some gas remains attached to the soot,
causing it to float when the soot-containing water flows slowly through basin
decanters (e). The upper soot layer, which contains 4–8 wt% of carbon, depend-
ing on the feedstock, is scraped off the water surface and incinerated.
Acetylene Oil Quench Process (AOP), Soot Removal (Figure 12). In
this process the cracked gas is quenched with oil sprays and leaves the burner at
670 ACETYLENE Vol. 2

Fig. 11. Acetylene water quench process (AWP). a) Preheaters; b) Acetylene burner;
c) Cooling column; d) Electrofilter; e) Soot decanter; f ) Cooling tower.

200–2508C. The oil absorbs the heat from the gas and then passes through waste
heat boilers before returning to the quench. The sensible heat of the cracked gas
represents more than 15% of the heating value of the feedstock. The pressure of
the generated steam depends on the process configuration and can reach 15 bar
(1.5 MPa).
Unlike the water quench process, where the scraped coke deposits sink to
the bottom of the quench chamber and are easily removed, in the oil quench the
coke deposits do not settle immediately. In order to prevent plugs in the quench

Fig. 12. Acetylene oil quench process (AOP). a) Preheaters; b) Acetylene burner;
c) Burner column; d) Mill pump; e) Coker; f) Decanter; g) Final cooler.
Vol. 2 ACETYLENE 671

nozzles a mill pump (d) is installed immediately underneath the burner col-
umn. the coke and soot content in the quench circuit is kept near 25% by send-
ing a fraction of the coke-containing oil to externally heated, stirred kettles
(coker (e)). In the kettles the volatile matter evaporates very quickly, leading
to fluidization of the coke bed. The vapor is returned to the burner column,
while the soot is agglomerated. A fine-grained coke is withdrawn from the
bottom of the coker.
Because of the cracking losses in the quench a certain amount of quench oil
has to be added continuously to the process. This makeup oil is at least 0.15 to
0.3 t per ton of acetylene, depending on the stability of the oil used. When resi-
dual oil from steam crackers is used, it can be desirable to add up to 1 t of oil per
ton of acetylene, because the excess oil is partially converted to light aromatic
hydrocarbons.
The cracked gas leaving the quench is cooled in a burner column (c), where
there are additional oil circuits for the production of 3-bar steam and for boiler
feedwater preheat. At the top of the column a small amount of a low-boiling oil
(BTX ¼ benzene, toluene, and xylene) is added to prevent deposit-forming aro-
matics (mainly naphthalene) from passing downstream into other parts of the
plant. The cracked gas, which has to be compressed before separation, is cooled
further (g) by water. At this stage most of the BTX condenses and is separated
from the water in a large decanter (f).
Table 4 shows the cracked gas compositions for the BASF acetylene oil
quench process when natural gas, liquid petroleum gas (LPG), or naphtha is
used as feedstock. The water quench process gives very similar compositions.
The relative amounts of hydrogen and carbon monoxide formed depend on the
hydrogen : carbon ratio of the feedstock used. Even when naphtha is used, almost
no ethylene forms. This is because the reaction takes place above 12008C where
the formation of ethylene is thermodynamically impossible. Only a prequench
with additional naphtha or LPG produces additional acetylene and ethylene at
intermediate temperatures, as in the case of two-step processes. The higher
hydrocarbons require a somewhat lower reaction temperature than methane

Table 4. BASF Acetylene Oil Quench Process, Cracked Gas Composition (vol %)
Raw material (DH, kJ/mol)
Methane LPG Naphtha
Component * (400) (325) (230)
H2 56.5 46.4 42.7
CH4 5.2 5.0 4.9
C2H4 0.3 0.4 0.5
C2H2 7.5 8.2 8.8
C3þ** 0.5 0.6 0.7
CO 25.8 35.0 37.9
CO2 3.2 3.4 3.5
O2 0.2 0.2 0.2
Inerts balance
*Dry gas, water, and aromatic compounds condensed out.
**Hydrocarbons with three or more carbon atoms.
672 ACETYLENE Vol. 2

and have a less endothermic heat of reaction: oxygen consumption per ton of
acetylene is lower for the higher hydrocarbons in spite of the lower preheating
temperature.
Comparison of Oil Quench and Water Quench Processes. The
advantage of the oil quench process is obvious: the heat recovery in the form
of steam makes the overall thermal efficiency in relation to primary energy
input rather high. If the thermal efficiency for the production of electricity is
33%, over 70% of the net heating value of the overall primary energy input is
recovered in the form of products and steam. A comparison between the oil
quench and water quench (see Table 6) shows that the oil quench requires a
net heating value input of 300–330 GJ per ton of acetylene, of which 82 GJ
(27–25%) is lost, whereas the water quench requires a 288 GJ input, of which
113 GJ (39%) is lost.
Acetylene Recovery. Liquid acetylene is a dangerous product, even at
low temperatures. Separation of the cracked gas by cryogenic processes such
as those used in olefin production is clearly ruled out. One exception to this
rule is the acetylene recovery unit of the submerged flame process (Section
4.2.3) (45), in which all hydrocarbons except methane are condensed
at –1658C. Otherwise, acetylene is recovered by selective absorption into a sol-
vent. This procedure is economical only when the cracked gas is compressed.
The upper limit for the pressure is determined by the danger of explosions,
and as a rule the partial pressure of acetylene should be kept below 1.4 bar
(0.14 MPa).
The solubility of acetylene in the solvents used is between 15 and 35 m3
(STP) per m3 of solvent under process conditions. The dissolved gas is recov-
ered by depressurizing the solvent and by vapor stripping at higher tempera-
tures. All solvents used commercially, N-methylpyrrolidone (NMP), methanol,
ammonia, and dimethylformamide (DMF), are miscible with water. They are
recovered from the gas streams leaving the plant by water scrubbing and
distillation.
The kinetics of acetylene formation always lead to the formation of higher
homologues of acetylene as byproducts (46), mainly diacetylene, but also methy-
lacetylene, vinylacetylene, and others. These compounds polymerize very easily
and must be removed from the cracked gas as soon as possible. Because they are
much more soluble in the solvents than acetylene, scrubbing the cracked gas
with a small amount of solvent before it enters the acetylene recovery stages is
sufficient.
Absorption Section (Fig. 13). Acetylene recovery is illustrated here by the
BASF process. N-Methylpyrrolidone is used to separate the cracked gas into
three streams:

1. Higher homologues of acetylene and aromatics, the most soluble part of the
cracked gas. (This is a small stream of gas, which is diluted with crude
synthesis gas for safety reasons and is used as fuel.)
2. Product acetylene, less soluble than the higher acetylenes, but much more
soluble than the remainder of the gas
3. Crude synthesis gas (off-gas), mainly hydrogen and carbon monoxide
Vol. 2 ACETYLENE 673

Fig. 13. BASF acetylene process — N-methylpyrrolidone absorption section. a) Com-


pressor; b) Prescrubber; c) Acetylene stripper; d) Main scrubber; e) Stripper; f) Vacuum
column; g) Vacuum stripper; h) Side column; i) Condenser; j) Vacuum pumps.

In the prescrubber (b) the cracked gas is brought into contact with a small
amount of solvent for removal of nearly all the aromatic compounds and C4 and
higher acetylenes except vinylacetylene. This is done after the compression of the
gas if screw compressors are used but before compression if turbo compressors
are used because turbo compressors cannot tolerate deposits on their rotors. In
the main scrubber (d) the gas is brought into contact with a much larger amount
of N-methylpyrrolidone (NMP), which dissolves all the acetylene, the remaining
homologues, and some carbon dioxide. Crude synthesis gas (off-gas) leaves at the
top of the column.
The NMP solution is degassed in several steps in which the pressure is
reduced and the temperature increased. The stripper (e) operates at pressures
and temperatures slightly above ambient. In this tower, the solution is put in
contact with a countercurrent gas stream from the subsequent degassing step
(f). This leads to the evolution of carbon dioxide, the least soluble of the dis-
solved gases, at the top of the stripper. The carbon dioxide is recycled to the suc-
tion side of the compression and thereby is shifted into the crude synthesis gas.
The acetylene product is withdrawn as a side stream from the stripper. The N-
methylpyrrolidone solution is then completely degassed (f) in two further steps
at 110–1208C, first at atmospheric, then at reduced pressure. Vinylacetylene,
methylacetylene, and excess process water are withdrawn as bleed streams
from the vacuum column (f). The water content of the solvent is controlled by
the reboiling rate in the vacuum column. At the bottom of the vacuum column,
degassing is completed, and the solvent is cooled and returned to the main
scrubber (d).
The small amount of solvent from the prescrubber (b) is stripped with crude
synthesis gas for recovery of the dissolved acetylene, the overhead gas being
674 ACETYLENE Vol. 2

Table 5. Purity of the Acetylene from the BASF Process


Component Crude acetylene, vol% Purified acetylene, vol%
Acetylene ca. 98.42 99.70
Propadiene 0.43 0.016
Propyne 0.75 traces
Vinylacetylene 0.05 0
1,3-Butadiene 0.05 0
Pentanes 0.01 0.01
Carbon dioxide ca. 0.10 0
Nitrogen ca. 0.30 0.30

recycled to the suction side of the compressor. The solvent is then degassed com-
pletely in the vacuum stripper (g), a column which also accepts the bleed stream
from the vacuum column (f) containing the excess process water together with
some higher acetylenes. The overhead vapor of the vacuum stripper contains
the higher acetylenes, water, and some NMP vapor. In a side column (h) the
NMP is recovered by scrubbing with a small amount of water, which is recycled
to the main solvent stream. The gas is cooled (i) by direct contact with water
from a cooling circuit to condense most of the water vapor. The higher
acetylenes are diluted with crude synthesis gas before they enter and after
they leave the vacuum pump (j). The diluted higher acetylenes, which are now
at a pressure slightly above atmospheric, can be used as fuel gas, e.g, for soot
incineration.
In order to minimize the polymer content of the solvent, about 2% of the cir-
culating flow is withdrawn continuously from the vacuum stripper circuit and
distilled under reduced pressure, leaving the polymers as a practically dry
cake for disposal.
The acetylene product from the process as described above has a purity of
about 98.4%, the remainder consisting mainly of propadiene, methylacetylene,
and nitrogen. For most applications the purity is increased to 99.7% by scrubbing
with sulfuric acid and sodium hydroxide solutions. Table 5 compares the compo-
sitions of crude and purified acetylene. Table 6 compares the consumption and
product yields per ton of acetylene for the oil quench process with those for the
water quench process.
Other Partial Combustion Processes. The main features of the BASF
process described in detail above are common to all partial oxidation processes.
Therefore only the differences between the BASF acetylene burner and burners
used in the Montecatini and the SBA processes (41,47) are described. These two
processes have also attained some importance. The details of the acetylene recov-
ery process depend on the properties of the solvent, but here too the basic prin-
ciples are the same for all processes.
Montecatini Process. The Montecatini burner (48) has the same main
components as the BASF burner: mixing unit, gas distributor, reaction chamber,
and quench. The essential difference is the pressure for acetylene synthesis,
which can be as high as several bar. This saves compression energy, improves
heat recovery from the quench water, which is obtained at 1258C, and is claimed
Vol. 2 ACETYLENE 675

Table 6. BASF Acetylene Process, Consumption and Product Yields Per Ton
of Acetylene
Consumption and product yields Oil quench Water quench
Feed and energy requirements
Natural gas, 36000 kJ/m3(STP) 5833 m3 = 210 GJ 5694 m3 = 205 GJ
(LHV) *
Oxygen, 0.55 kWh/m3(STP) ** 3400 m3 = 20.4 GJ 3400 m3 = 20.4 GJ
Fuel gas = 12.0 GJ = 18.0 GJ
Residue oil minimum (surplus) 0.3 (1.0) t = 12.0 (40.0) GJ
Sulfuric acid 160 kg 160 kg
Sodium hydroxide 5 kg 5 kg
N-Methylpyrrolidone 5 kg 5 kg
Electric energy ** 3200 kWh = 34.9 GJ 3100 kWh = 33.8 GJ
Steam, 4 bar 5.0 t = 11.7 GJ 4.5 t = 10.5 GJ

Energy input 301.0 (329.0) GJ 287.7 GJ


Product yields
Acetylene, 48650 kJ/kg (LHV) 1.0 t = 48.6 GJ 1.0 t = 48.6 GJ
Crude sythesis gas, 10600 m3 = 128.3 GJ 10150 m3 = 122.8 GJ
12100 kJ/m3(STP) (LHV)
Coke (with residue surplus), 0.3 (0.46) t = 10.7 (16.3) GJ –
35500 kJ/kg
BTX (with residue surplus), 0.05 (0.12) t = 2.0 (4.8) GJ –
40250 kJ/kg (LHV)
Naphthalenes (with residue 0.0 (0.41) t = – (15.9) GJ –
surplus), 38770 kJ/kg (LHV)
Steam (up to 15 bar) 13.0 (14.0) t = 30.3 (32.6) GJ 1.5 t = 3.5 GJ

Energy output 219.9 (246.5) GJ 174.9 GJ

Thermal efficiency 73.0 (74.9) % 60.8 %

Energy losses, absolute per ton 81.1 (82.5) GJ 112.8 GJ


acetylene
*If the natural gas contains inerts and higher hydrocarbons, the required input will remain approxi-
mately the same on a heating value basis (LHV = low heating value), but the cracked gas analyses and
the crude synthesis gas analyses will differ slightly.
**Thermal efficiency of electricity production is assumed to be 33%.

to make soot removal easier because the cracked gas is scrubbed with water
above 1008C. Although it is well known (8) that acetylene decomposition is accel-
erated under pressure at high temperatures (>10008C), the acetylene yield is
comparable to that obtained at atmospheric pressure because of the short resi-
dence time in the reactor. Methanol is used at cryogenic temperatures for acet-
ylene recovery. The main steps of the gas separation are absorption of higher
acetylenes and of aromatics, absorption of acetylene, stripping of coabsorbed
impurities, and desorption of acetylene.
SBA Process (of the Société Belge de l’Azote). The SBA burner (49)
has the same main components as the other processes. However, it has a tele-
scope-like reaction chamber and a device for shifting the quench up and down.
Thus it is possible to adjust the length of the reaction zone for optimum
676 ACETYLENE Vol. 2

residence time at any throughput. The walls of the reaction chamber are
sprayed with demineralized water to prevent coke deposits. This eliminates the
need to scrape the reaction chamber periodically. Acetylene recovery is carried
out with several scrubbing liquids — kerosene, aqueous ammonia, caustic soda,
and liquids ammonia, each with its own circuit. After soot is separated from the
gas in an electrofilter, higher hydrocarbons are absorbed in kerosene or gas oil.
Carbon dioxide is scrubbed in two steps, first with aqueous ammonia and then
with caustic soda solution. The acetylene product is absorbed into anhydrous
ammonia and must be scrubbed with water after desorption. All the ammonia–
water mixtures are separated in a common distillation unit. This recovery scheme
leads to exact separation of the various cracked gas components.
Additional Remarks. The Montecatini and SBA processes can also be
operated with two-stage burners. A prequench with light hydrocarbons cools
the cracked gas to about 8008C. After a residence time at this intermediate tem-
perature the gas is cooled down with water. In this way the heat content of the
hot gases is used for further cracking of hydrocarbons to yield extra acetylene
and olefins. The presence of additional components in the cracked gas requires
more process steps in the gas separation units.
Submerged Flame Process. The submerged flame process (SFP) of
BASF attracted considerable interest up to 1973 as a partial combustion pro-
cess for the production of acetylene, ethylene, C3, and C4 hydrocarbons, and
synthesis gas from feedstock of crude oil and residues, such as Bunker C oil
and vacuum residue (44,45). Although it was abandoned at the end of 1973,
the need to make the most economic use of raw materials has renewed interest
in this process (50).
Oxygen compressed to 16 bar (1.6 MPa) feeds a flame that is submerged in
the oil. The oil surrounding the flame is partially burnt to obtain the necessary
reaction temperature and also acts as the quenching medium. This process dif-
fers from the partial oxidation processes using natural gas and lighter hydrocar-
bons in five main respects:

1. Crude oil can be gasified without the formation of residues, and the process
can be operated under certain conditions with heavy fuel oil.
2. All the soot formed is consumed when crude oil feedstock is used, eliminat-
ing all the problems associated with the storage, disposal, or utilization of
acetylene soot.
3. The heat of reaction is removed by steam generation at 8 bar (0.8 MPa).
4. The process is operated at 9 bar (0.9 MPa) so that the oxygen is the only
compressed stream. The cracked gas is formed at a pressure sufficient for
economic separation.
5. The design of the cracking unit is greatly simplified because the reaction
feed, fuel, and quenching medium are identical.

The process is described in detail in the literature cited; therefore, only gen-
eral overviews of the cracking unit (Fig. 14) and the separation unit (Fig. 15) are
shown here. The capacities of a submerged flame burner for acetylene and ethy-
lene are 1 t/h and 1.15 t/h, respectively. To produce these quantities, 5000 m3
Vol. 2 ACETYLENE 677

Fig. 14. Submerged flame process (SFP) — cracking unit. a) Reactor; b) Oil cooler;
c) Steam generator; d) Oil recycle pump; e) Scrubber; f) Naphtha cooler; g) Naphtha se-
parator; h) Naphtha pump; i) Spray cooler; j) Separating vessel; k) Recycle-water pump;
l) Recycle-water cooler.

(STP) of oxygen and 8–10 t of oil are required per hour. The cracked gas shows
the following average composition (vol%, the components grouped as streams
leaving the separation unit):

Main products
Acetylene 6.2
Ethylene 6.5
Crude synthesis gas
Carbon monoxide 42.0
Hydrogen 29.0
Methane 4.0
Inerts 0.6
Other hydrocarbons
Ethane 0.5
Propane 0.1
Propene 1.2
Propadiene, propyne 0.7
1,3-Butadiene 0.5
Other C4 and C5+ 14 hydrocarbons 1.5
Remainder
Carbon dioxide 7.0
Hydrogen sulfide 0.05–0.5
Carbon oxide sulfide 0.03–0.3
*C5+ , five or more carbons

Unlike all other processes the submerged flame process uses low tempera-
tures (–1658C) to separate the off-gas, consisting of carbon monoxide, hydro-
gen, and methane, from the C2 and higher hydrocarbons. On account of the
acetylene in the condensed phase, extensive decomposition tests have been
678 ACETYLENE Vol. 2

Fig. 15. Submerged flame process — purification unit

carried out. Whereas the cracking unit (Fig. 14) and the amine scrubbing unit
have been tested by Soc. Ital. Serie Acetica Sintetica, Milan, on a commercial
scale, the remaining purification units (Fig. 15) have not. However, the experi-
ence obtained with a pilot plant indicates that major difficulties are not to be
expected.
The submerged flame process may become competitive because of its ability
to use crude oil and especially residues and because of its low losses on the pri-
mary energy input.
Partial Combustion Carbide Process. Calcium carbide production from
lime and coal requires a large high-temperature heat input (see Section 4.3.4).
In the thermal process some of the coal must be burnt to attain the necessary
reaction temperature and supply the heat of reaction. The thermal carbide pro-
cess was developed by BASF (8,51) from 1950 to 1958 to eliminate the input of
electrical energy necessary in the classic carbide process. Starting in 1954, a
large pilot plant, with a nominal carbide capacity of 70 t/d, was operated, but
in 1958 the more economical petrochemical acetylene production halted further
development. Carbide production is just one way of converting coal chemically;
other methods include pyrolysis, hydrogenation, and gasification. The question
arises as to the conditions under which a thermal carbide process using oxygen
can compete with the electric carbide process. The biggest drawback of carbide
production in a shaft furnace (Fig. 16) compared to the electric carbide process
is the lack of commercial-scale operational experience. Specific disadvantages
are greater susceptibility to disruption because of plugging of the furnace feed,
Vol. 2 ACETYLENE 679

Fig. 16. Partial combustion carbide process. a) Carbide furnace; b) Refractory brick lin-
ing; c) Charging hopper; d) Gas outlet; e) Oxygen jet; f) Tapping burner; g) Tapping chute;
h) Bogey; i) Cyclone; j) Washing column; k) Desintegrator; l) Compressor.

more stringent specifications for the raw materials, more handling of solids, and
the large amount of byproduct. There are two main advantages:

1. A thermal efficiency of about 50% versus about 30% for the electrothermal
process if the thermal efficiency of electricity production is 33%
2. Carbon monoxide production, which is desirable because carbon monoxide
can be converted to synthesis gas by the water-gas shift reaction (!Gas
Production)

If the carbon monoxide is converted to synthesis gas and the electrical


energy is produced from fossil fuels, production costs are about one third lower
for the thermal process than for the electrical process (4) based on the pilot-plant
consumption data (Table 7).
4.3. Electrothermic Processes. Because calcium carbide is produced
electrothermally, the production of acetylene from this material also is discussed
in this group of processes (Section 4.3.4).

Table 7. Partial Combustion Carbide Process, Consumption and Product Yields Per Ton
Acetylene
Raw materials
Coke, dry (88% C) 5700 kg
Lime (92% CaO) 3140 kg
Oxygen (98%) 3560 m3 (STP) 5090 kg
Total consumption 13930 kg
Products
Carbide (80.5%) 2850 kg ¼
^ 1000 kg acetylene
Carbon monoxide 7980 m3 (STP) 9975 kg
(CO 95.5, H2 2.0, N2 2.0, CO2 0.5 vol %)
Dust 900 kg
Losses 205 kg
Total products 13930 kg
680 ACETYLENE Vol. 2

Electrothermic processes have the following advantages over partial oxidation:

The energy requirement for the formation of acetylene can be made indepen-
dent of the hydrocarbons used as feedstock.
Hydrocarbon consumption can be reduced by 50%.
Provided that electrical energy is available under favorable conditions
(nuclear power, hydroelectric power, cheap coal) and/or the availability of
hydrocarbons is limited, electrothermic processes are more economical.

In the case of acetylene formation, the electric-arc process offers optimal


conditions for the endothermic reaction at high temperatures.
The development of the electric-arc process for cracking light hydrocarbons
to acetylene began in 1925 in Germany. The acetylene was to be used as feed-
stock for butadiene production. In 1940, the first commercial plant was put
on stream at Chemische Werke Hüls in Marl, Germany. The Hüls process
has since been improved , and the capacity raised, but it retaired the original
principles (52).
Feedstock for electric-arc processes may be gaseous or liquid hydrocarbons
or even solids such as coal. The design of the arc furnace and the purification sec-
tion for the cracked products have to be adapted to the different feedstocks. For
gaseous or gasified hydrocarbons the classical one-step process is used: the arc
burns directly in the gas being cracked. For liquid and solid feeds, a one- or
two-step process may be used. In the two-step process hydrogen is first heated
in the arc furnace, and then liquid or solid feed is injected into the hydrogen
plasma (53). Figure 17 shows both types of arc furnaces. Because of hydrogen for-
mation during the cracking reaction, the arc burns in a hydrogen atmosphere in
both processes. The conductivity and the high rate of ion–electron recombination
for hydrogen mean that arcs above a certain length cannot be operated with
alternating current at normal frequency and high voltage. All commercial plants
therefore run on direct current.
Production from Gaseous and/or Gasified Hydrocarbons (Hüls Arc
Process). The plant for the Hüls arc process includes the arc furnace section
itself (Fig. 17 A), which is operated at a pressure of 1.2 bar, and a low and high
pressure purification system.
Arc Furnace. A cathode, a vortex chamber, and an anode make up the arc
furnace. Cathode and anode are water-jacketed tubes of carbon steel 0.8 m and 1.5
m long, respectively, and with inner diameters of 150 and 100 mm, respectively.
The arc burns between cathode and anode with a length of about 1.2 m and with a
current of 1 200 A. The cathode is connected to the high-voltage side of the rectifier
(7.1 kV) and electrically isolated from the other parts of the furnace. Between cath-
ode and anode is the vortex chamber. The gas is injected into it tangentially at a
specific velocity to stabilize the arc by creating a vortex. The arc burns in the dead
zone, and the striking points of the arc on the electrodes are forced into a rapid
rotation so that they only burn for fractions of a millisecond at one point, which
gives the electrodes a lifetime up to 1000 h. Temperatures reach 200008C in the
center of the arc. Because of the tangential flow of the gas, the arc is surrounded
by a sharply decreasing coaxial temperature field, and the temperatures at the
Vol. 2 ACETYLENE 681

Fig. 17. Hüls electric-arc furnaces for gaseous, liquid, and solid feed. A) One-step
process; B) Two-step process.

wall of the electrode are only 6008C. Thermal losses are therefore limited to less
than 10% of the electrical power input of 8.5 MW.
The residence time of the gas in the arc furnace is a few milliseconds. In this
interval, the hydrocarbons are cracked, mainly into acetylene, ethylene, hydro-
gen, and soot. At the end of the arc furnace, the gases are still at a temperature of
about 18008C. The high heat content of this gas can be exploited for additional
ethylene production by means of a prequench with liquid hydrocarbons. This
lowers the temperature to about 12008C. Because acetylene rapidly decomposes
into soot and hydrogen at these temperatures, the gases must be quenched
immediately with water to about 2008C, i.e, a quench rate of 1068C/s must be
achieved.
The specific energy requirement (SER) and the acetylene yield depend on
the geometry and dimensions of the furnace and electrodes, the velocity distribu-
tion of the gas, and the kind of hydrocarbon to be cracked. Once the furnace has
been designed, only the hydrocarbons can be varied.
Process Without Prequench. Figure 18 shows acetylene and ethylene
yields and the specific energy requirement (SER) of various saturated hydrocar-
bons under constant conditions without prequench. Methane shows the highest
SER and acetylene yield, but the lowest ethylene yield. As the chain length is
increased, both acetylene yield and SER decline, corresponding to the declining
heat of acetylene formation from the various hydrocarbons.
Normally, pure hydrocarbons are not available. The results obtained from
mixtures of hydrocarbons can be expressed as a function of the carbon number,
682 ACETYLENE Vol. 2

Fig. 18. Acetylene yield, ethylene yield, and energy consumption for various hydrocar-
bons in the Hüls arc process.

which is the number of moles of carbon atoms bound in hydrocarbons per mole of
the gaseous mixture. Figure 19 shows specific amounts of acetylene, ethylene,
and hydrogen formed and of hydrocarbon consumed as a function of carbon num-
ber. This function enables the Hüls process to be optimized within certain limits,
for example, for hydrogen output in relation to acetylene production.
Process with Prequench. Cracking in the prequench section is essen-
tially an ultrasevere steam cracking process. The kind and amount of hydrocar-
bons used for the prequench can be varied. Figure 20 shows the specific product
yield for different prequench rates for feeding methane to the arc furnace and
propane to the prequench. Acetylene and hydrogen yields are unaffected,

Fig. 19. Specific values for acetylene and hydrogen formation.


Vol. 2 ACETYLENE 683

Fig. 20. Specific product yield for different prequench rates.

whereas ethylene shows a slight maximum and declines when the temperature is
not sufficient at a given residence time. Propene shows a steady increase, and the
C3 : C2 ratio is below 0.25. The relative ethylene yield from various hydrocarbons
is as follows: ethane 100, propane 75, n-butane 72, isobutane 24, 1-butene 53.
Oil quench. Because the gas temperature of the furnace gas after pre-
quench is on the order of 12008C, an oil quench system has been developed to
regain about 80% of the sensible heat content of the furnace gas as steam by
heat exchange. The soot–oil mixture formed can be upgraded to a sulfur- and
ash-free high-grade petroleum coke. Figure 21 shows the newly designed Hüls
system with oil quench.
The Purification System. The process of purification depends on the
type of the quench system. In the case of water quenching, 80% of the carbon
black is removed by cyclones as dry carbon black, the remaining 20% as soot

Fig. 21. Process with oil quench system. a) Heat recovery; b) Arc furnace; c) Oil recovery;
d) Separation of medium-boiling compounds; e) Separation of low-boiling compounds;
f) Oil regeneration.
684 ACETYLENE Vol. 2

Fig. 22. Principal separation and purification steps for the furnace gas of the Hüls arc
process.

in water-operated spray towers. In a combined oil–water scrubbing system, aro-


matic compounds are removed and benzene, toluene, and xylene (BTX) are recov-
ered in a distillation process.
Figure 22 shows the principle separation and purification steps for the fur-
nace gas. The gas leaves the first three purification sections with a carbon black
content of 3 mg/m3 and is compressed by four-stage reciprocating compressors
to 19 bar (1.9 MPa). The gas is washed in towers with water in a countercurrent
flow. At the bottom of the tower, the water is saturated with acetylene, whereas
the overhead gas contains less than 0.05 vol% acetylene. The acetylene–water
solution is decompressed in four stages. Gas from the first decompression stage
returns to the compressor to improve selectivity. The last two stages operate
at 0.2 and 0.05 bar (20 and 5 kPa). The gas still contains about 10 vol% of
higher acetylenes, which are removed by a cryogenic process. The higher
acetylenes are liquefied, diluted with flux oil, stripped, and returned to the
arc furnace together with spent hydrocarbon. A more selective solvent such
as N-methylpyrrolidone or dimethylformamide is preferred to the water wash.
Linde and Hüls have designed an appropriate purification system. Hydrogen
and ethylene are separated by well-known technology, such as the cryogenic
process or pressure-swing adsorption.
Process Data. Hüls operates its plant with a mixture of natural gas,
refinery gas, and liquefied petroleum gas. The carbon number varies between
one and two. Table 8 shows a typical analysis of feed gas and cracked gas.
The Hüls plant has 19 arc furnaces, the number operated depending on the
electricity supply. The arc furnaces can be started up and shut down immedi-
ately. Two large gas holders provide a storage volume of 350000 m3 so that the
purification section operates on permanent load and there is a dependable supply
of products, even if the arc furnace section is operated at higher or lower load.
Vol. 2 ACETYLENE 685

Table 8. Typical Analysis of Feed and Cracked Gas


Feed gas, vol% Cracked gas, vol%
C2H2 0.4 15.5
C3H4 1.4 0.4
C4H2 1.2 0.3
C4H4 1.7 0.4
C2H4 0.8 6.9
C3H6 3.6 1.0
Allene 0.4 0.2
C4H8 1.0 0.2
C4H6 0.9 0.2
C5H6 0.6 0.2
C6H6 0.5 0.5
CH4 64.6 13.8
C2H6 7.5 0.4
C3H8 3.6 0.3
C4H10 4.6 1.0
C5H12 0.5 0.1
H2 4.5 57.6
CO 0.5 0.6
O2 0.1 0.0
N2 1.6 0.4

The plant has an annual capacity of 120000 t acetylene, 50000 t ethylene,


400 106 m3 (STP) hydrogen, 54000 t carbon black and soot, and 9600 t aromatic
compounds. The energy consumption is 1.5 106 MW h/a.
Specific data for consumption of hydrocarbons and energy and the produc-
tion of byproducts per ton of acetylene produced are as follows:

Hydrocarbons to the arc furnace 1.8 t


Hydrocarbons for prequench 0.7 t
Energy for the arc furnace 9800 kW h
Energy for gas purification 2500 kW h
Ethylene 0.42 t
Hydrogen 3300 m3 (STP)
Carbon black and soot 0.45 t
Aromatics 0.08 t
Residue 0.12 t
Heating gas 0.12 t

Production from Liquid Hydrocarbons (Plasma Arc Process). Two dif-


ferent plasma furnaces, each with the appropriate reactor for the cracking of
liquid hydrocarbons, were developed by Hoechst and Chemische Werke Hüls
in close coorperation. Both units were tested on an industrial scale at a power
level of 8–10 MW (54). However, neither process has actually come into use for
acetylene production on account of the economics.
The scheme of the plasma generator used by Hüls is shown in Figure 17 B.
The unit consists of three parts: the arc furnace, the reactor, and the quench
system. The arc burns over a length of 1.6 m at 7 kV d.c. and 1.2 kA, resulting
686 ACETYLENE Vol. 2

Fig. 23. Acetylene and ethylene yields as a function of the low-boiling components.

in a power input of 8.5 MW. It is stabilized by hydrogen injected tangentially


through the vortex chamber. The thermal efficiency of the furnace is ca. 88% of
the electrical power input. The hydrogen plasma jet passing through the anode
nozzle has an energy density of 3.5 kW h/m3(STP), corresponding to an average
temperature of 3500 K. The liquid hydrocarbons (e.g, crude oil) to be cracked are
injected into the cylindrical reactor to achieve good mixing with the plasma jet and
to avoid the formation of carbonaceous deposits on the wall. Within several milli-
seconds the hydrocarbons are heated and cracked to acetylene, ethylene, hydro-
gen, soot, and other byproducts before the mixture is quenched with oil to
3008C. The acetylene ratio can be adjusted by varying the residence time. By
operation of an oil quench with the high-boiling residue of the crude oil, 80% of
the sensible heat content of the cracked gas can be recovered as steam. The soot
is taken up by the quench oil and is removed from the system as an oil–soot dis-
persion having 20% soot concentration. The unconverted vaporized fractions of the
oil are condensed in oil scrubbers at a lower temperature, simultaneously cleaning
the gas of the aromatic components and fine soot. These oil fractions are recycled to
the reactor and the quench system, respectively.
Tests were carried out with a variety of hydrocarbons from propane to
naphtha, but mainly with crude oil and residue oils. The cracking results depend
on the chemical nature of the feed. Consumption figures and yields for various
feedstocks are given in (54). For high-boiling petroleum fractions, the acetylene
and ethylene yields increase with the content of low-boiling components in the
feed (see Fig. 23). Consumption and byproduct yields per ton of acetylene for a
Libyan crude oil are summarized below:

Consumption
Crude oil consumed 3.5 t
Power (d.c.) 10500 kW h
Byproducts
Ethylene 0.46 t
Hydrogen (99.9%) 1100 m3 (STP)
Fuel gas 0.74 t
Soot–oil mixture (20% soot) 1.2 t
Vol. 2 ACETYLENE 687

Table 9. Consumption and Yield Per Ton of Acetylene for the Hoechst Arc Process and
Naphtha Feed for Different Reactor Designs
Low ethylene yield High ethylene yield
Consumption
Naphtha 1.92 t 2.50 t
Quench oil 0.53 t 0.63 t
Energy(2-phase a.c.) 9300 kW h 10500 kW h
Byproducts
Ethylene 0.5 t 0.95 t
Hydrogen 1450 m3 (STP) 1500 m3 (STP)
Soot – oilmixture(20% soot) 0.75 t 1.00 t

Hoechst used a high-intensity three-phase a.c. arc furnace at 1.4 kV and 4.2 kA,
giving a power input of 10 MW (54). The thermal efficiency was 90%. Because of
the high amperage the graphite electrodes had to be replenished continually. The
generator was lined with graphite. Different reactor designs for ethylene: acety-
lene ratios of 0.5 and 1.0 were developed by varying the mixing intensity of the
hydrogen plasma jet with the liquid hydrocarbon. The tests were carried out with
naphtha feed (see Table 9). The cracked gas was quenched with residue oil, in a
manner similar to that described in the Hüls process.
The acetylene concentration in the Hüls process and the Hoechst process
was ca. 14 vol% so that in principle the same acetylene separation process can
be used as described above for the arc process.
Production from Coal (Arc Coal Process). Numerous laboratory tests for
the conversion of coal to acetylene using the arc or plasma processes have been
carried out since the early 1960s (55). The results can be summarized as
follows:

Acetylene yields up to 30% can be obtained.


Because of the rapid heating of the coal in the plasma jet, a higher total gas
yield can be achieved than is indicated by the volatiles of the coal measured
under standard conditions.
Hydrogen (instead of argon) plasma gas considerably increases the acetylene
yield.

The AVCO arc furnace (Fig. 24) consists of a water-cooled tungsten-tip cath-
ode and a water-cooled anode (56). The arc is stabilized by a magnetic field sur-
rounding the anode, forcing the anode striking point of the arc to rotate rapidly
and so avoiding burnthrough. The dried and finely ground coal is injected by
means of a hydrogen gas flow around the cathode. Additional gas without coal
is introduced around the cathode and at the anode as a sheath. On passing the
arc zone the coal particles are heated up rapidly. The volatiles are released and
are cracked to acetylene and byproducts, leaving a residue of fine coke particles
covered with soot. After a residence time of some milliseconds the gas–coke mix-
ture is quenched rapidly with water or gases. The use of a prequench system
688 ACETYLENE Vol. 2

Fig. 24. Principal scheme of the AVCO plasma furnace for the pyrolysis of coal.

similar to that of the Hüls arc process was also tested. The system pressure can
be varied between 0.2 and 1.0 bar (20 and 100 kPa).
Hüls’ pilot plant uses the same plasma furnace as for the crude oil cracking,
but with 500 kW of power (57). The dried and ground coal is injected into the
plasma jet, and the coal is cracked to acetylene and byproducts in the reactor.
The reactor effluent can be prequenched with hydrocarbons for ethylene produc-
tion or is directly quenched with water or oil. Char and higher boiling compo-
nents are separated by cyclones and scrubbers, respectively. The problem in
the reactor design is to achieve thorough and rapid mixing of the coal with the
plasma jet and to avoid forming carbonaceous deposits on the wall. Smaller
amounts of deposits can be removed by periodic wash cycles with water. Opera-
tion times of 2.5 h by AVCO and 5 h by Hüls have been reported.
Experiments published by Hüls and AVCO show that at the optimal resi-
dence time the energy density of the plasma jet, the specific power, and the pres-
sure all greatly affect the acetylene yield (Fig. 25 and Fig. 26). Other parameters
affecting the yield are the amounts of volatiles in the coal and the particle size.
The lowest figures for the specific energy consumption published by AVCO are of
the order of 27–37% based on water-free coal.
In addition to acetylene, the exit gas contains considerable amounts of CO,
depending on the oxygen content of the coal. Because nitrogen and sulfur are
present in the coal, other byproducts are HCN, CS2, COS, and mercaptans.
The gas separation system is therefore designed accordingly (58). Depending
on the hydrogen content of the coal, the process is either self-sufficient in hydro-
gen or has a slight surplus. The total gas yield of the coal based on a volatile con-
tent in the coal of 33% is up to 50%. Thus 50% of the coal remains as char. Tests
with a view to using this char in the rubber industry have been unsuccessful so
far. Thus the char can be used only for gasification or as a fuel.
Vol. 2 ACETYLENE 689

Fig. 25. Effect of the energy density of the plasma on the cracking of coal (AVCO).

In all the processes under development, the production of ethylene from


coal requires several process steps (Fig. 27), resulting in a high capital demand
for a production plant. In contrast, the acetylene production from coal arc
pyrolysis is straightforward, leading to lower investment costs. Demonstration
units on a higher power level are therefore scheduled by both AVCO and
Hüls.
Production from Calcium Carbide. At present, the generation of
acetylene from calcium carbide is of primary importance for welding and for
the production of carbon for batteries. The particular raw-material situation

Fig. 26. Acetylene yield and specific energy requirement as a function of pressure
(Hüls).
690 ACETYLENE Vol. 2

Fig. 27. Alternative routes from coal to ethylene and acetylene.

and the use of special processes are two common reasons for continuing to use
acetylene generated from carbide in the chemical industry.
The reaction of calcium carbide and water to form acetylene and calcium
hydroxide is highly exothermic:

CaC2 þ 2H2 O ! C2 H2 þ CaðOHÞ2 DH ¼ 129kJ=mol

The acetylene generator used for commercial production must therefore be


designed to allow dissipation of the heat of reaction. In the event of inadequate
heat dissipation, for example, when gasification proceeds with insufficient water,
the carbide may become red-hot. Under certain circumstances (including increas-
ing pressure), this may cause the thermodynamically unstable acetylene to
decompose into carbon and hydrogen. (For safety precautions see Section 5.) Car-
bide for the production of acetylene is used in the following grain sizes (mm):
2–4, 4–7, 7–15, 15–25, 25–50, 50–80. This classification is virtually identical
in most countries: DIN 53922 (Federal Republic of Germany); BS 642:1965 (Uni-
ted Kingdom); JIS K 1901–1978 (Japan); Federal Specification 06–101 b/GEN
CHG NOT 3 (United States). In addition, grain size 0–3 is used for the dry
generation of acetylene.
Vol. 2 ACETYLENE 691

Pure calcium carbide has a yield number of 372.66. This means that
the gasification of 1 kg of carbide yields 372.66 L acetylene at 158C and
1013 mbar (101.3 kPa). Commercially available carbide has a yield number of
260–300.
A distinction is made between two groups of acetylene generators (with con-
tinuous rates of production greater than 10 m3 acetylene per hour): the wet type
and the dry type.
In wet generators, the acetylene is converted with a large water excess. In
most cases, a lime slurry containing 10–20 wt% calcium hydroxide is obtained.
The heat of reaction increases the temperature of the generator water and is
removed from the reactor with the lime slurry.
In dry generators, the water mixed with the carbide is just sufficient for che-
mical reaction and for dissipating the heat of reaction. The calcium hydroxide is
obtained in the form of a dry, easily pourable powder having a residual moisture
content of 1–6%. The heat of reaction is dissipated by evaporation of part of the
generator water.
Generators are classified according to their working pressure as
either low or medium pressure. This classification is governed by the regula-
tions concerning acetylene plants and calcium carbide storage facilities
(Acetylenverordnung) (59) issued by Deutscher Acetylenausschuß (German
acetylene committee) and the associated technical rules for acetylene plants
and calcium carbide storage facilities TRAC (Technische Regeln für Acetyle-
nanlagen und Calciumcarbidlager) (60). The Acetylenverordnung and TRAC
constitute a comprehensive set of rules for handling acetylene. Recommenda-
tions in other countries (e.g, United States) deal only with some aspects,
such as safety precautions (61).
Low-pressure generators are designed for a maximum allowable working
pressure of 0.2 bar. They must be rated for an internal pressure of at least
1 bar. Lower pressure ratings are possible if proof is given in each particular
case that the generator can withstand the expected stress (maximum working
pressure, water filling, agitator, etc.; TRAC 201).
Medium-pressure generators have a maximum allowable working pressure
of 1.5 bar. They must be rated for an internal pressure of 24 bar. A design pres-
sure of 5 bar suffices whenever the generators are equipped with rupture disks of
a defined size and specified response pressure (3–4.5 bar, TRAC 201).

Wet Generators. Wet generators are used primarily for the production of
small amounts of acetylene, e.g, for welding purposes. Wet generators work by
one of three different principles (62):

1. The carbide-to-water principle, where the carbide is mixed with a large


excess of water at a rate corresponding to the gas withdrawal rate. Most
generators work by this principle.
2. The water-to-carbide principle (drawer type generators), where water is
added at a controlled rate to the carbide, which is held in a replaceable con-
tainer (drawer).
3. The contact principle (basket generators), where the carbide, which is held
in a basket, is immersed into the generator water. This type is designed so
692 ACETYLENE Vol. 2

Fig. 28. Medium-pressure wet generator. a) Carbide skip; b) Hopper; c) Feeding system;
d) Gasification chamber; e) Agitator; f) Slurry valve, g) Safety device.

that the water drifts off the basket as a result of the gas pressure at low gas
withdrawal rates and, conversely, returns to the basket when gas withdra-
wal rates increase.

Medium-Pressure Generators. The Messer Griesheim MF 1009 is a typical


carbide-to-water generator (Fig. 28).
The carbide skip (a) is filled with carbide of the 4–7 grain size. The skip is
connected by gas-tight gates to the hopper (b) and is purged of air by nitrogen or
acetylene. The carbide drops into the hopper (b) and is fed continuously by the
feeding system (c) to the gasification chamber (d). The gasification chamber con-
tains water up to a level defined by the generator capacity and is equipped with
an agitator (e) for whirling the lime slurry. The heat evolved in gasification heats
the generator water. For continuous operation the water temperature must not
exceed 908C; therefore, fresh water is admitted continuously to the gasification
chamber. If the defined water level is exceeded, the slurry valve (f), controlled by
a float, opens, allowing the excess water and the lime slurry to be discharged
from the generator.
The acetylene generated collects above the water and is withdrawn. The
feeding system (c) is controlled by the gas pressure, i.e, the rate at which
carbide drops into the gasification chamber varies directly with the rate of
gas withdrawal.
The carbide stock in the hopper (b) is sufficient for about one hour, but
the skip (a) can be refilled with carbide and replaced on the hopper so that
Vol. 2 ACETYLENE 693

continuous operation is possible. The wet generator described has a continuous


hourly output of 75 m3 of acetylene. The skip holds 1000 kg of carbide.
Low-Pressure Generators. The working principle of the low-pressure
carbide-to-water generators is very similar to that of the medium-pressure
carbide-to-water generator described above. In most cases, a downstream acety-
lene holder, normally of the floating gas bell type, is provided. In contrast to the
medium-pressure generator, in which the carbide feed rate is controlled by the
acetylene gas pressure, the feed rate in the low-pressure generator is controlled
by the position of the bell in the acetylene holder, i.e, by the gas quantity.
Products. The acetylene generated in the wet generator can be used for
welding, often without further purification. In certain cases, coke- or gravel-filled
purifiers or a wet scrubber are connected downstream from the generator for
separating solid or liquid particles. Before it is fed to a synthesis unit, the acet-
ylene must be purified chemically (see page 31).
The lime slurry formed is fed into pits. Here, the calcium hydroxide settles
in the form of a lime dough containing 35–75 wt% water (wet lime, carbide lime
dough). This dough is used as carbide lime.
Dry Generators. Dry generators are mainly used for the production of
large quantities of acetylene for chemical synthesis.
Compared to the wet generator, the primary advantage of the dry generator
is that the dry calcium hydroxide formed as a byproduct can be used in other pro-
cesses more easily, more cheaply, and in a more diversified way than the lime
slurry obtained in the wet generator (63). Moreover, lime recycling into the car-
bide production process is only possible with dry calcium hydroxide.
A high gasification rate and the elimination of the risk of overheating were
originally the most important criteria for the design of dry generators. Early
designs of dry generators worked by continuous renewal of the reaction surfaces
of coarse carbide with the aid of rotating drums, blades, vibrating screens, and
similar equipment.
Typical examples are the early generator of Shawinigan Chemicals (64) and
the Piesteritz dry generator (65).
Although a number of factors affect the gasification rate of carbide, e.g, den-
sity, porosity, and crystalline structure, above all it is the specific surface that
affects the carbide gasification rate the most. Hence, dry generators work with
finely ground carbide (0–3 mm), which gasifies in a fraction of the time needed
for coarse carbide. The result is a high space–time yield.
A typical application of this principle is the large-scale Knapsack dry gen-
erator, which was developed at the Knapsack works of Hoechst. This type of gen-
erator is used worldwide and is described in more detail below (Fig. 29).
Carbide of the grain size 0–3 falls from the chain conveyor (a) into the sub-
divided feed bin (b). The chain conveyor is loaded with material from the carbide
bin. Because of the recirculating stream of carbide the feed bin (b) is full at all
times. The carbide layer in the feed bin (b) acts as a gas seal between the genera-
tor and the carbide conveying system.
The carbide is fed to the generator (e) via the star wheel (c) and the carbide
feed screw (d). The largest generator of this kind built to date has a diameter of
3.5 m and an overall height of approximately 8.0 m. The generator has up to 13
circular trays. These are so designed as to leave alternate annular gaps on the
694 ACETYLENE Vol. 2

Fig. 29. Knapsack dry generator. a) Chain conveyor; b) Feed bin; c) Star wheel; d) Car-
bide feed screw; e) Generator; f) Lime lockhopper; g) Lime discharge screw; h) Lime scra-
per; i) First scrubbing tower; k) Second scrubbing tower; l) Dip seal.

shell side and at the central agitator shaft. The agitator shaft moves stirrer pad-
dles across the trays.
The carbide first reaches the uppermost tray where the generator water is
also admitted. The reaction mixture consisting of carbide, water, and calcium
hydroxide is pushed by the stirrer paddles towards the outer edge, drops on to
the second tray, returns towards the center, etc. When it reaches the last tray,
the carbide has been fully gasified. The calcium hydroxide, which still contains
up to 6% water, drops into the lime lockhopper (f). Here, a lime layer two meters
deep serves as the gas closure between generator and lime conveying system.
The lime is withdrawn continuously. the gas leaving the generator through the
lime scraper consists of 25% acetylene and ca. 75% water vapor. The water vapor
is the result of dissipating the major portion of the reaction heat. Depending on
the generator load, up to several hundred kilograms of lime hydrate dust are car-
ried along with the acetylene. The lime scraper (h) retains the major portion of
this dust and returns it to the generator. The remainder is sent together with the
gas into the first scrubbing tower (i). Here, lime slurry is sprayed into the hot
acetylene gas (ca. 908C) to scrub out the lime dust; part of the water vapor con-
denses because of the simultaneous cooling. In the second scrubbing tower (k),
the acetylene is sprayed with atomized water to cool the gas below 408C; addi-
tional water vapor condenses here. Any ammonia still present in the gas is
also removed. the acetylene leaves the generator via the dip seal (l). It still con-
tains certain impurities in the form of sulfur and phosphorus compounds. the
Knapsack dry generator is suitable for a carbide throughput of 15 t/h, corre-
sponding to an acetylene quantity of 3750 m3/h. During this process, about
Vol. 2 ACETYLENE 695

17.5 t of calcium hydroxide per hour are obtained. The pressure in this low-pres-
sure generator amounts to approximately 1.15 bar (115 kPa).
The dry generator of Shawinigan Chemicals, Montreal (66), also processes
finely ground carbide and has a variety of applications. It consists of several
superimposed troughs. Carbide and water are fed into the uppermost trough.
The reacting mixture, which is constantly kept in motion by blades, flows over a
weir onto the trough below, etc. At the uppermost trough, water is admitted at
such a rate that carbide-free calcium hydroxide can be withdrawn at the lowermost
trough. The generated acetylene is purified in two scrubbing towers and cooled.
The calcium hydroxide formed (carbide lime) has a wide range of applications,
e.g, in the building industry (for preparing mortar, cement, etc.), in the chemical
industry (for neutralization and for recycling to the carbide furnace), in agriculture
(as fertilizer), and for water purification and waste water treatment (63).
Acetylene Purification. During the gasification of carbide with water,
gaseous compounds become mixed with the acetylene, and these must be
removed because they have a harmful effect on the downstream chemical synth-
esis processes. The impurities are mainly sulfur and phosphorus compounds.
They can be removed by one of the following purification processes.
In the first process, dilute chlorinated water is used as the oxidizing agent.
The chlorine concentration of the water is limited to 1.5 g/L to prevent the forma-
tion of unstable chlorine compounds, which present an explosion hazard. The
chlorine scrubbing step is followed by a caustic soda scrubber to remove the hydro-
gen chloride formed during the oxidation process. The disadvantage of this purifi-
cation process is that considerable quantities of scrubbing water are produced.
The second process uses 98% sulfuric acid as the oxidizing agent (67).
Because very small quantities of sulfuric acid are admitted, it is difficult to dis-
sipate the heats of absorption and reaction. Heating the acetylene results in
increased formation and settling of polymerization products in the purification
stage. For this reason the gas requires additional cooling in the event that the
acetylene contains appreciable quantities of impurities. Moreover, it is recom-
mended that a second scrubbing tower be kept on standby if a high onstream fac-
tor is desired (e.g, 91% ¼ ^ 8000 h/a operating time).
The sulfuric acid scrubber is followed by a caustic soda scrubber, in which
the sulfur dioxide formed during oxidation is removed.
The main advantage of this purification method is that virtually no waste
water is obtained. The small amount of polluted, highly concentrated sulfuric
acid can be used, for example, in fertilizer plants.
These two purification processes yield the following acetylene purities
(by volume):

Acetylene > 99.5%


Sulfur, as H2S <10 ppm
Phosphorus, as H3P <10 ppm

As a result of the extremely good sorption properties of the concentrated


sulfuric acid, very pure acetylene can be expected.
4.4. Other Cracking Processes. Thermal Cracking By Heat Carriers.
Well-known processes using heat carriers, such as the Wulff and Hoechst
696 ACETYLENE Vol. 2

high-temperature pyrolysis (HTP) processes, are no longer used because they


require refined petrochemical feedstocks such as naphtha and liquid petroleum
gas. The Wulff process uses refractory material as the heat carrier, whereas the
Hoechst HTP process uses hot combustion gases.
Newer processes, which are able to convert crude and heavy distillates into
olefins and considerable amounts of acetylene, are still in the pilot-plant stage.
These processes include the advanced cracking reactor process developed by Kur-
eha, Chiyoda, and Union Carbide, using high-temperature superheated steam,
and Dow’s partial combustion cracking process, using hot combustion gases pro-
duced from oxygen and fuel oil as the heat carrier.
Wulff process (68), (69, p. 58). This process is based on indirect heat
transfer, an approach fundamentally different from the partial-oxidation and
electric-arc processes. The hydrocarbon feed is cracked in refractory ovens pre-
viously heated by combustion gas. After cracking, the products are quenched out-
side the reactor. Soot formation is a serious problem because the feed cannot be
heated as rapidly as in the partial-oxidation or arc processes. This problem can
be diminished by using a feed with a higher hydrogen:carbon ratio. However,
methane is not suitable because of the high temperature and high heat of
reaction required, resulting in a low conversion rate. Thus the best feed for the
Wulff process is ethane or propane.
Hoechst High-Temperature Pyrolysis (HTP) Process (Fig. 30) (1),
(69, p. 55), (70). This is a two-stage process. In the first stage, heat is produced

Fig. 30. High-temperature pyrolysis (HTP) process.


Vol. 2 ACETYLENE 697

in the burner by the combustion of residual cracked gas from the acetylene recov-
ery section (CO, H2, CH4) with oxygen. Immediately after combustion, the tem-
perature is about 27008C; this is moderated to about 23008C by the injection of
steam before the reactor is entered. In the second stage, the feedstock naphtha is
injected, and the adiabatic cracking reaction takes place. A final temperature of
about 13008C is reached: This determines the cracked gas composition. By vary-
ing the feed rate of naphtha the acetylene–ethylene ratio can be altered from
30 : 70 to 70 : 30. However, thermodynamic and economic considerations show
that the optimum ratio is 40 : 60.
After a reaction time of a few milliseconds the cracked gas is quenched to
approximately 2508C by the injection of cracked oil from the process. The oil
absorbs heat from the cracked gas and is passed through waste heat boilers, rais-
ing the steam pressure. No soot is formed in this process, even when crude oil is
used as feed because of the high steam content of the carrier gas.
After the oil crisis of 1973 the process became uneconomical in spite of its
high thermal efficiency, and in 1976 it was shut down after 15 years of operation.
However, one unit is still running in Czechoslovakia.
Kureha, Chiyoda, Union Carbide Advanced Cracking Reactor
(ACR) Process (71). To avoid dependence on oil refineries or gas processors
for the supply of feedstocks, processes for directly cracking crude oil have been
developed by various companies for the production of olefins. Some of these pro-
cesses operate at reaction temperatures intermediate between those of the usual
crack processes for olefins and those for acetylene. The ACR process (Fig. 31) uses
a multi-port burner to produce a heat carrier gas of 20008C by the combustion of
H2 – CH4 mixtures with oxygen in the presence of steam preheated to 8008C.
The oil to be cracked is introduced through nozzles into the stream of car-
rier gas and passes into an advanced cracking reactor, where the reaction takes
place adiabatically at 5 bar (0.5 MPa). The initial temperature is 16008C; the

Fig. 31. Advanced cracking reactor process (ACR). a) Crude distillation column; b) Bur-
ner; c) Advanced cracking reactor; d) Ozaki quench cooler; e) Oil gasoline fractionator;
f) Compressor; g) Acid gas removal column; h) Gas separator.
698 ACETYLENE Vol. 2

Fig. 32. Ozaki quench cooler.

final temperature at the exit of the reactor is 700–9008C after a residence time of
10–30 ms. The cracked gas is quenched by oil in an Ozaki quench cooler (Fig. 32),
where steam production up to 120 bar (12 MPa) is possible. This particular boiler
design was developed for a high heat transfer rate without coke formation on the
exchanger surfaces. Yields reported for Arabian light crude oil are 11.2 wt%
hydrogen and methane, 40.7 wt% olefins, and 4.2 wt% acetylene. The acetylene
yield is about ten times higher than in usual olefin processes.
Dow Partial Combustion Cracking (PCC) Process (71). The basic
idea of this process is to reduce coking and soot formation considerably when
heavy feeds are cracked and when hydrogen is present in the reaction mixture.
The PCC process (Fig. 33), which accepts crude oil and heavy residue as feed-
stock, attains a high partial pressure of hydrogen in the reaction zone by recy-
cling the quench oil (produced in the process) to the burner where it is partially
oxidized to yield synthesis gas. Thus there is no need to find a use for the quench
oil as in the case of the ACR process. Starting from residual oil boiling above 3438C,
yields are given as 12.4 wt% methane, about 38 wt% alkenes, and 2.5 wt% acety-
lene. This is seven to eight times more acetylene than that obtained from a steam
cracker, but less than the acetylene yield of the ACR process, because of a resi-
dence time in the reaction zone which is three to ten times longer.
Acetylene as a Byproduct of Steam Cracking. In a steam cracker satu-
rated hydrocarbons are converted to olefinic products such as ethylene and pro-
pylene. Besides these desired components, acetylene and many other products
are formed in the cracking process (Fig. 36). The concentration of acetylene
depends on the type of feed, the residence time, and temperature (cracking
Vol. 2 ACETYLENE 699

Fig. 33. Dow partial combustion cracking process (PCC). a) Reactor; b) Quench boiler;
c) Quench column; d) Stripper; e) Decanter.

severity: expressed as conversion or propene/ethylene ratio P/E). Typical data


are given in Table 10. The acetylene concentration in the off-gas from the furnace
varries between 0.25 and 1.2 wt%; the corresponding content of acetylene in the
C2 fraction is about 0.4–2.5 wt%. An ethylene plant producing 400 000 t/a ethy-
lene produces 4500–11 000 t/a acetylene. The acetylene is removed by catalytic
selective hydrogenation or solvent extraction.
Acetylene Hydrogenation. Most ethylene plants are equipped with
a hydrogenation unit. Acetylene is converted selectively to ethylene on a
Pd-doped catalyst. Typical process conditions are temperatures of about
40–120 8C, pressures of 15–40 bar and space velocities of 1000–
120 000 kg L1 h1. Depending on the type of feed and the plant, there are
two process options:

1. Front-end hydrogenation (C2 stream containing H2, CO, methane, C2H2,


C2H4 and C2H6)
2. Tail-end hydrogenation (pure C2 stream containing C2H2, C2H4, and C2H6;
separate addition of an equimolar amount of hydrogen)

General aspects of the process and the catalyst requirements are reviewed
in (73,74).

Table 10. Yields of Unsaturated Components (wt%) in Raw Gas From Steam Cracking
Feedstock Cracking severity Acetylene Propyne Propadiene
Ethane 65% convers. 0.4–0.50 0.04 0.02
LPG 90% convers. 0.65–1.20 0.63 0.35
Full-range naphtha P/E: 0.4 0.9–1.05 0.81 0.54
Full-range naphtha P/E: 0.53 0.5–0.70 0.68 0.50
Full-range naphtha P/E: 0.65 0.25–0.42 0.46 0.38
Atmospheric gas oil P/E: 0.55 0.40 0.34 0.29
Hydrocracker residue P/E: 0.55 0.50 0.36 0.31
700 ACETYLENE Vol. 2

Fig. 34. Acetylene recovery process (75).

Acetylene Recovery. Acetylene is extracted from the C2 fraction of the


steam cracker. The solvent must fulfil the following criteria:

Melting point lower than the dew point of the feed gas
High solubility of acetylene at a temperature near the dew point of the C2
fraction
High acetylene selectivity (9,11)
High chemical and thermal stability
No foaming tendency due to traces of hydrocarbons
Low toxicity
Low vapor pressure at the operating temperature

After testing many solvents, including DMF, N-methyl-2-pyrrolidone (NMP),


and acetone, the most suitable solvent for such a process proved to be DMF.
The solubility of acetylene as a function of temperature is shown in Figure 5.
The process for the recovery of high-purity acetylene is shown in Figure 34
(75). The gaseous C2 mixture, consisting of ethylene, ethane, and acetylene, is fed
to the acetylene absorber; the gas stream is contacted with counterflowing lean
DMF at a pressure of 0.8–3.0 MPa. The process is suitable for the full pressure
range prevailing in any of the known ethylene processes. The entire acetylene
and some of the ethylene and ethane are dissolved by the solvent. Entrainment
of DMF at the top of the column is avoided by a reflux stream. The purified C2
fraction, containing <1 ppm of acetylene, is fed to the C2 splitter. The rich solvent
stream is sent to the ethylene stripper, which operates slightly above atmo-
spheric pressure. Ethylene and ethane are stripped off and recycled to the first
stage of the gas compressor for the cracked gas. Any acetylene entrained with the
overhead gas is recovered by washing with cold solvent at the top of the stripper.
In the acetylene stripper, pure acetylene is isolated from the top of the column.
Vol. 2 ACETYLENE 701

Table 11. Material Balance (mol %) for an Acetylene Recovery Process Operating on the
C2 Fraction from a Plant Producing 400 000 t/a C2H4 (Linde) (76)
Gas to absorber Purified gas Recycle gas Product C2H2
Methane trace trace
Acetylene 2 1 ppm 4 99.8
Ethylene 82 83.5 85.7 0.2
Ethane 16 16.5 10 trace
C3 trace trace 0.3 trace
DMF 1 ppm trace trace
Temperature, K 252 249 255 258
Pressure, MPa 2 1.98 0.11 0.12
Flow rate, kmol/h 2186 2126 17.5 52.5

After cooling and heat recovery, the acetylene-free solvent is recycled to the
absorber and ethylene stripper. The acetylene product has a purity of > 99.8%
and a DMF content of less than 50 ppm and is available at 10 kPa and ambient
temperature.
The material balance and the utilities consumption of an acetylene recovery
unit are listed in Tables 11 and 12. At present, more than 112 000 t/a of petro-
chemical acetylene from twelve olefin plants worldwide is recovered by this tech-
nology; three other facilities are under construction (1998). Figure 35 shows an
industrial plant with a design capacity of 14 400 t/a of high-purity acetylene.
A material balance for ethylene plant outputs including acetylene extrac-
tion or hydrogenation is shown in Figure 36 (72). The economic evaluation
shows that petrochemical acetylene remains attractive even if the price of ethy-
lene is doubled. It is economical to retrofit acetylene absorption in an existing
olefin plant equipped with a catalytic hydrogenation.
A similar process is available for propyne (see Chap. 7)

5. Safety Precautions, Transportation, and Storage

General literature is given in (10,77,78).


5.1. General Safety Factors and Safety Measures. Decomposition
and Combustion. Acetylene is thermodynamically unstable under normal
conditions. Decomposition into carbon and hydrogen can achieve temperatures

Table 12. Consumption of Utilities for an Acetylene Recovery Process Operating on the
C2 Fraction from a Plant Producing 400 000 t/a C2H4 (Linde) (76)
DMF, kg/h 1.3
Heating steam, t/h 3.9
Cooling water, m3/h 100
Electrical energy, kW 125
Refrigerant, GJ/h 6.3
Quench water, GJ/h 3.1
Plot area, m m 15 40
702 ACETYLENE Vol. 2

Fig. 35. Acetylene recovery plant (name plate capacity: 14 400 t/a of high-purity
acetylene).

of about 31008C, but due to formation of other products, the temperature reached
adiabatically is 2800–29008C. The decomposition can be initiated by heat of reac-
tion, by contact with a hot body, by an electrostatic spark, by compression heat-
ing, or by a shock wave. The decomposition induced by heating the wall of the
container or pipe is very sensitive to the pressure, the size and shape of the
container or the diameter of the pipe, the material of the container, and traces
of impurities or other components. Solid particles such as rust, charcoal,
alumina, and silica can lower the ignition temperature compared to clean
steel pipe.
Decomposition gives rise to different scenarios:

Working range I (deflagration): a flame produced by decomposition or combus-


tion and propagates at a velocity below the velocity of sound into the uncon-
verted gas (pressure rises simultaneously in front of and behind the flame
front)
Vol. 2 ACETYLENE 703

Fig. 36. Material balances of a 300 000 t/a ethylene plant equipped with either C2 hydro-
genation or acetylene extraction (all rates in kg/h, the numbers in parentheses are for the
solvent extraction process) (72). *Chemical grade.

Working range III (detonation): the flame propagates at ultrasonic velocity


into the unconverted gas (shock wave between low pressure in the uncon-
verted gas and high pressure in the converted gas)
Working range II (intermediate between I and III): often the propagation ve-
locity of a deflagration is not constant (it increases with increasing density,
temperature, and turbulence), and therefore a change from deflagration to
detonation is observed

As consequence, design criteria for piping and other components are


proposed for the different working ranges and depend on the diameter of
the pipe. Limit lines for deflagration and detonation are given in (79) on
the basis of the work of Sargent (80). An extended Sargent diagram is shown
in Figure 37.
The limits are influenced by the method of ignition (e.g, melting wire or a
detonator cap). Changing the method of ignition from a melting Pt wire (Reppe)
to the exploding wire ignitor (ignition energy ca. 70 J) used by BAM resulted in
lower stability pressures for acetylene mixtures (81) and pure acetylene (82)
(Table 13).
Mixtures of acetylene with methane have higher stability pressures than
those with nitrogen or hydrogen. Further information on the effect of additional
gases on acetylene decomposition is given in (83). The dependence of stability
pressure on the energy of the ignition source has led to ongoing discussion
about its relevance for industrial design and operations (84).
Figure 38 shows the decomposition pressure of acetylene, propyne, and
propadiene as function of ignition energy. This relationship is the basis for the
safe design of processes for the recovery of acetylenic components (see Sections 4
and 7). Additional investigations have been published on the dependence
704 ACETYLENE Vol. 2

Fig. 37. Detonability limits of acetylene (80,85). A) Deflagration limit; B) Detonation


limit Detonation limits: a) Thermal ignition in a plain pipe (a1 melting wire, 20 – 80 J;
a2 detonator cup, &2400 J); b) Thermal ignition plus orifice; c) Ignition by chemical reac-
tion in a shock wave; d) Range of possible quasi-detonation depending on ignition energy
of shock wave; x) and y) Limiting ignition pressure for thermal ignition with melting Pt
wire and with detonator cup, respectively.

of deflagration pressure on the flow in pipes (85) and the decomposition of high-
pressure acetylene in branched piping (86). Solid acetylene is not critical with
regard to decomposition, provided it is the only material involved (8). In liquid
oxygen, solid acetylene can readily ignite on mechanical impact and react vio-
lently (87). Recommendations for equipment used in gas welding and cutting
technology, such as rubber hoses, safety devices, and flame arresters, are
given in (88,89,90).
Combustion of Acetylene in Oxygen (Air). The reaction of acetylene
and oxygen at 258C and 1 bar to form water and CO2 generates 1255.6 kJ/mol.

Table 13. Stability Pressure (bar) of Acetylene and Acetylene Mixtures for Two Methods
of Ignition and Pressure Increase (81,82)
Ignition method pex/pa
Mixture Reppe BAM (BAM)
100% C2H2 1.4 0.8
90% C2H2/10% N2 1.8 1.0 8.6
90% C2H2/10% CH4 2.1 1.0 7.2
90% C2H2/10% H2 1.6 0.9 6.9
50 % C2H2/50% N2 9.0 3.6 6.3
50% C2H2/50% CH4 14.7 12.9
50% C2H2/50% H2 4.7 2.5 5.5
*pex/pa: ratio of maximum pressure to pressure before ignition.
Vol. 2 ACETYLENE 705

Fig. 38. Decomposition pressure versus ignition energy for unsaturated hydrocarbons.

Temperatures of around 31008C can be reached. Figure 39 shows flame tempera-


tures and flame front velocities for mixtures of oxygen with hydrocarbons (91).
Acetylene allows the highest temperatures and flame front velocities to be
attained. The maximum temperature is very sensitive to the mixing ratio,
which also determines whether a reducing, neutral, or oxidizing flame exists
(Fig. 40).
Fundamental safety data for acetylene–air and acetylene–oxygen mixtures
are listed in Table 14. At atmospheric pressure and 258C mixtures of 2.4–
93.0 vol% acetylene in oxygen are explosive; the possibility of self-decomposition
at high acetylene must also be taken into account.
Handling of Acetylene. For pure acetylene the prescribed safety
instructions, for example, the Technische Regeln für Acetylenanlagen und Cal-
ciumcarbidlager (Technical regulations for acetylene plants and calcium carbide
depots), TRAC, (92), have to be strictly followed. However, it is not possible to

Fig. 39. Flame temperatures and ignition velocities of acetylene – oxygen mixtures and
mixtures of other hydrocarbons with oxygen.
706 ACETYLENE Vol. 2

Fig. 40. Chemical composition of an oxygen – acetylene flame at its tip versus mixing
ratio.

formulate general safety instructions for the great variety of chemical processes
with acetylene as reaction component under diverse reaction conditions.
Both handling acetylene and experiments with it necessitate critical exam-
ination of sources of possible danger. The literature cited can only serve as an aid
to decisions on precautions. The safety regulations mentioned above have been
determined in experiments with well-defined apparatus dimensions (length, dia-
meter, geometry). For other dimensions they can only serve as an indication of
explosive behavior and should not be considered as rigid limits. The development
of economical chemical processes involving acetylene at elevated pressures or
under other hazardous conditions calls for decomposition tests for the crucial
stages where decomposition could occur. This must be done in close cooperation
with official testing institutions, such as the Bundesanstalt für Materialprüfung
(BAM) in the Federal Republic of Germany.
In general, the following rules should be observed in handling acetylene:

Temperature and pressure must be selected so as to avoid liquefaction of


acetylene.

Table 14. Fundamental Safety Data for Acetylene – Air and Acetylene – Oxygen Mixtures
Air Oxygen
Lower flammability limit, vol% 2.5 2.4
Upper flammability limit, vol% 82 93
Flame temperature*, K 2863 3343
Flame front velocity, m/s 1.46 7.6
Increase of pressure (deflagration) 11 50
Detonation velocity, m/s 2300 2900
*Stoichometric mixture.
Vol. 2 ACETYLENE 707

Reactions of acetylene in solvents or with liquid reaction components must be


carried out at such acetylene concentrations that explosive decomposition
of the acetylene in the liquid phase cannot occur. In many cases this condi-
tion is fulfilled at an acetylene loading below 100 m3 (STP) per cubic meter
of solvent. Higher loadings are only permitted if additional precautions are
taken, such as filling the volumes containing the liquid with steel packings.
The formation of a separate gas phase has to be avoided.
The technical rules (such as TRAC (92)) are valid for pure gaseous acetylene.
If an inert gas, such as nitrogen, is added to the acetylene, higher acetylene
partial pressures are permitted.
In the design of apparatus the partial pressure of acetylene should be selected
so that the minimum distance to the decomposition limit is about 20%. The
apparatus should be designed to withstand pressures (1) 12-fold the initial
pressure for pure acetylene systems or (2) the initial pressure plus 12-fold
the acetylene partial pressure for mixtures and solutions.
Formation of hydrates (see Chap. 2) under pressure must be avoided because
this leads to obstructions in the apparatus and pipelines. The melting point
of these compounds is in the range 0–138C; therefore, pressurized acety-
lene containing water has to be kept above 158C.

In addition to the measures for building construction, electrical installa-


tions, fire protection, purging, and leak detection, acetylene plants and distribu-
tion systems are provided with flame traps and flashback and release valves and
locks (93,94). Flame traps consist either of tubes immersed into water-filled
cylinders (wet trap) or cylinders filled with a packing of high surface area to
decelerate the decomposition. A wet arrester, which is used for an acetylene dis-
tributing line, is shown in Figure 41. Suitable materials for dry-trap packings are
sinter metals, ceramic beads (e.g, Raschig rings), bundles of small tubes, and cor-
rugated metal foils (95,96).
Tapping points for acetylene distribution units which meet the German
TRAC rules include a nonreturn valve to avoid intrusion of air from downstream,
a sinter-metal flame trap, and a thermo- or pressure-sensitive spring lock. The
last closes if a flame is stopped by the trap but still burns outside of the flame
trap. Detailed information is given in (93).
Transportation in Pipelines. Acetylene is occasionally transported in
pipelines. Figure 42 shows the safety components of an acetylene pipeline
between Burghausen and Gendorf, Federal Republic of Germany. The pipeline
was operated until 1976 without incident. Its length is 8 km, and the pipes
are 300 mm in diameter. Design pressure was 100 bar, although operation pres-
sure was only 2 bar at the inlet and 1.25 bar at the outlet. The pipeline was
provided with rupture disks, which open to atmosphere in case of decomposition.
Quick closing valves are initiated simultaneously to protect both upstream and
downstream equipment. At each end, part of the pipeline is filled with tube
bundles to stop propagation of any acetylene decomposition. The flame traps
consist of 600-mm-diameter U’s filled with Raschig rings.
A report (97) is available on an acetylene-decomposition event in a
pipeline system, which demonstrates the need for safety measures. Instead of
708 ACETYLENE Vol. 2

Fig. 41. Hydraulic flame trap for acetylene lines (Union Carbide), (97).

transportation of pure acetylene, pipeline transportation of acetylene solutions


in acetone was proposed as safer (98). In the United States, transportation of
acetylene solutions in liquid ammonia was considered for existing ammonia dis-
tribution systems (99).
Hazardous Acetylene Traces in Low-Temperature Processes. Ace-
tylene is the most dangerous component in gas mixtures processed in low-tem-
perature plants. In air separators, for example, acetylene can be suspended in
liquid oxygen as a solid or as a segregated liquid phase that is quite unstable

Fig. 42. The 8 km acetylene pipeline from the Marathon refinery, Burghausen, to Farb-
werke Hoechst, Gendorf, Federal Republic of Germany (98). a) Compressor; b) Control
points; c) Automatic quick-closing valves; d) Rupture disks to atmosphere; e) Flame traps;
f) Pipeline.
Vol. 2 ACETYLENE 709

and tends to uncontrollable and violent decomposition. (The solubility of acety-


lene in liquid oxygen is low, see Chap. 2) Therefore much attention must be
given to checking for and removing acetylene in low-temperature separation
plants.
Normally air contains some acetylene, up to 0.3 mL/m3. In industrial areas,
especially in the proximity of petrochemical plants, higher concentrations (up to
1 mL/m3) can occur. Without any measures, acetylene in the feed air of an air
separator would be enriched in the cold section of the unit.
In modern air separators alternating molecular sieve adsorbers are used.
The adsorbers obey the following breakthrough sequence:

To avoid acetylene breakthrough the adsorber is operated for sufficient CO2


removal and is regenerated when the CO2 concentration at the adsorber outlet
starts to increase. Further details are given in (100-104).
Another possible way to remove acetylene and other combustible air con-
taminants is catalytic oxidation prior to the separation (102,105). This method,
being expensive, is rarely used.
All air separators are provided with routine analysis systems for
acetylene. Routine analysis concentrates on the liquid oxygen of the main con-
denser (106).
The removal of acetylene in cracked gas separation is treated in Section 4.4.
The processing of other acetylene-containing gases, e.g, coke-oven gas, by low-
temperature separation is described in (107).
5.2. Acetylene Storage in Cylinders. Because of its tendency to defla-
grate or to detonate, acetylene cannot be compressed and stored in gas cylinders
like other gases.
For desensitizing, acetylene stored in gas cylinders is dissolved in a solvent
in which the acetylene is very soluble. This solvent is dispersed in a porous solid
that completely fills the gas cylinder. As well as giving better solvent distribution
the porous material arrests any local acetylene decomposition induced, for
instance, by flashback.
Acetone and dimethylformamide are the preferred solvents for acetylene in
cylinders. An advantage of dimethylformamide is its lower vapor pressure,
resulting in lower solvent losses during acetylene discharge. A disadvantage is
its higher toxicity. The total pressure of an acetone-containing acetylene cylinder
depends on the acetylene : acetone ratio and on temperature as is shown in
Figure 43. Deviations from the plotted curves resulting in higher pressures are
caused by the porous filling of the gas cylinder, which absorbs acetone, changing
the effective acetylene : acetone ratio (108).
Impurities in the acetylene decrease the dissolving capacity of the acetone.
Figure 44 shows the effect of moisture on acetylene solubility. As a result, acet-
ylene produced from calcium carbide and water has to be dried.
710 ACETYLENE Vol. 2

Fig. 43. Total pressure of acetylene solution in acetone as a function of acetylene :


acetone ratio and temperature (8).

Calcium carbide-based acetylene contains further impurities that have to


be scrubbed out to avoid decreased solubility in acetone. Examples are divinyl
sulfide and phosphine: 1 wt% divinyl sulfide in the acetone reduces the acetylene
solubility from 35 g/kg to 31 g/kg at 208C and an acetylene pressure of 0.1 MPa.
Further values, also for phosphine, are given in (109,110). The impurities have
to be scrubbed out to residual concentrations of 0.5 g of phosphorus and 0.1 g of
sulfur per m3 of acetylene.

Fig. 44. Solubility of acetylene in water-containing acetone at 25 8C and pC2 H2 ¼ 1 bar


(0.1 MPa) (109).
Vol. 2 ACETYLENE 711

Table 15. Permitted Acetylene and Acetone Filling of Seamless Gas Cylinders
(Satisfying German Standards and Safety Rules) (112)
Gas cylinder Acetone filling, kg

Vol. of gas Outer Maximum


cylinder, diameter, Length,
L mm mm Minimum General Exceptional Acetylene filling, kg
3 140 300 0.789 0.8625 0.9375 0.4725
5 140 460 1.315 1.4375 1.5625 0.7875
10 140 850 2.630 2.8750 3.125 1.575
20 204 810 5.260 5.750 6.250 3.150
27 204 1040 7.101 7.7625 8.4375 4.2525
40 204 500 10.520 11.50 12.50 6.30
40 229 1210 10.520 11.50 12.50 6.30

During acetylene production from calcium carbide, disperse calcium hydro-


xide (0.1–1.0 mm) is produced. This contaminates the product gas. The calcium
hydroxide present in acetylene filled into acetone-containing gas cylinders
catalyzes aldol condensation of the solvent and reduces the solubility for
acetylene:

2ðCH3 Þ2 CO ! ðCH3 Þ2 COHCH2 COCH3 diacetone alcohol

Therefore, the solids content of calcium carbide-based acetylene for filling


acetone-containing gas cylinders must be kept below 0.1 mg/m3 (109,111).
The gas cylinders have to be filled with definite amounts of acetylene and
solvent. Commercial-grade, seamless gas cylinders which meet specified stan-
dards (in the Federal Republic of Germany, DIN 4664) may be filled with the
amounts listed in Table 15.
The amounts are fixed by regulations for handling pressurized gases (113).
Corresponding regulations in the United States have been issued by the Depart-
ment of Transportation (114).
To determine maximum acetylene filling of gas cylinders, extensive igni-
tion, impact, and heating tests have been worked out (115,116).
The porous material in the acetylene gas cylinders must satisfy the follow-
ing requirements: no interaction with the cylinder material, acetylene, or acet-
one and suitable mechanical properties, such as sufficient impact resistance.
Suitable materials include pumiceous compounds, silica, charcoal, asbestos
fiber, and alkaline carbonates. The porosity of these materials varies between
70 and 80% (109,117). Modern monolithic materials are made preferably from
silica, lime, and glass fiber. The mixtures are suspended in water to obtain a
pasty material which is filled into the gas cylinders. The material is hardened
at about 2008C and subsequently dried and activated at 350–4008C. A porosity
of about 90% is obtained.
Any porous material to be used for acetylene cylinders has to be examined
and approved by competent authorities. The examination includes the determi-
nation of the maximum acetylene and solvent filling, the maximum filling
712 ACETYLENE Vol. 2

Table 16. Examples of Approved Silica-Based Porous Materials in German Acetylene


Cylinders (117) (for Seamless 40-L Cylinders Satisfying DIN 4664, Blatt 2)
Material Approved filling
a b
Type Origin Acetylene , Acetonec , Acetylene : Maximum
kg kg acetone, kg/kg pressured , bar
Linde M1 Linde, Munich 8.0 12.7 0.63 19
AGA 2 AGA, Hamburg 8.0 12.4 0.645 19
SIAD 2 SIAD, Sabbio 8.0 12.4 0.645 19
a
Approval only when porous filling is prepared at place of origin;
b
Maximum;
c
Desired value;
d
Gage, at 15 8C.

pressure, and ignition and impact testing. Table 16 lists three porous materials
approved for use in Germany. Figure 45 shows a photograph of ‘‘Linde M1’’ mag-
nified 1 : 10000, clearly revealing the porous structure of such materials.
Methods for examining the materials have been standardized by CPI
(Commission Permanente Internat. de l’Acétylène, de la Soudure Autogène et
des Industries qui s’y rattachent, Paris) and ISO (International Organization
for Standardization) (117).
Discharging acetylene from a gas cylinder leads to acetone losses because
the partial pressure of acetone at 158C ranges from 0.14 bar at 15 bar total

Fig. 45. Porous silica material Linde M1 for acetylene cylinders (magnification
1 : 10000).
Vol. 2 ACETYLENE 713

Fig. 46. Temperature profile of an oxyacetylene flame.

pressure to 0.18 bar at 1 bar total pressure. Solvent loss has to be replaced when
an acetylene cylinder is reloaded.

6. Uses and Economic Aspects

6.1. Use in Metal Processing. Acetylene has many applications in the


processing of metals and other materials. This is because of the high flame tem-
perature and propagation velocity resulting in high energy densities and rapid
heat transfer to the piece being worked. The properties of an oxyacetylene
flame given here supplement those in Section 5.1.
The temperature profile of an oxyacetylene flame consists of a hotter pri-
mary flame and a scattered secondary flame. The highest flame temperature is
at the tip of the primary flame (Fig. 46). For material processing the primary
flame is the more important.
The heating efficiency of the primary flame is the product of the volume-
based heat released by the primary flame and the propagation velocity. This is
plotted in Figure 47 for the oxyacetylene flame and some other flames. The
heat transferred in welding is generated by radiation, convection, and thermal
conduction (see Table 17). The heat transfer is promoted by a high temperature
gradient between flame and workpiece.

Fig. 47. Heating efficiency of acetylene – oxygen mixtures and mixtures of other hydro-
carbons with oxygen. *based on area of primary flame cone
714 ACETYLENE Vol. 2

Table 17. Heat Transfer in Welding (118)


Welding temperature, K
Gas 800 1200 1600
temperature,
K QS QK QS QK QS QK
1000 2.3 13.4
2000 3.8 55.4 3.4 37.0 2.3 18.5
3000 4.0 83.2 3.8 68.0 3.6 50.4
QS, heat transfer by radiation (kJ cm2 h1);
QK, heat transfer by convection (kJ cm2h1) at a gas velocity of 50 m/s

Oxidizing, neutral, or reducing (carburizing) flames can be obtained by


varying the oxygen : acetylene ratio (Fig. 40). For steel, alumina, and copper
welding usually neutral or slightly reducing flames are used, whereas oxidizing
flames are preferred for brass welding, cutting, pickling, and surface hardening
(119). Acetylene is burned with oxygen in single torches or in bundles of torches,
the chief components of which are the connections for acetylene and oxygen,
regulating valves, a mixing chamber (usually of the injection type), a flashback
protection element, and a nozzle adapted to the specific applications (120,121)
Oxyacetylene flames are used in welding, cutting, brazing, soldering, sur-
facing, flame spraying, heating, hardening, straightening, cleaning, pickling,
rust removal, and decarbonizing.
Acetylene–air flames are occasionally used for tin brazing, hot air welding
of thermoplasts, glassworking, and paint removal (120), although the conveni-
ence and safety of fuels such as propane or butane has displaced acetylene in
those applications. Soft and hard soldering, flame hardening, and flame temper-
ing are important applications for the softer acetylene–air flame. For acetylene–
air mixtures, self-aspirating Bunsen-type and acetylene–compressed air burners
are used.
The different uses of oxyacetylene and acetylene–air flames in metal work-
ing, the procedures, and the equipment are comprehensivelydescribed in (8) and
(122); other sources of information for oxyacetylene flame properties in welding
are (118,119), and (123).
6.2. Use as Raw Material in Chemical Industry. Because of the diver-
sity of acetylene chemistry (see Section 3.1), acetylene has been used as a start-
ing material for a great variety of industrially important products. These are
summarized, together with their applications, in Figure 48.
Between 1960 and 1970, when worldwide acetylene production peaked,
most of the products listed in Figure 48 were produced via acetylene. During
the last 15 years the competition between acetylene and the olefins (see Sec-
tion 6.3) has resulted in substitution of ethylene and propene for acetylene, espe-
cially in the production of acetaldehyde and acrylonitrile. At present, acetylene is
used mainly for the production of vinyl chloride, vinyl acetate, and other vinyl
esters; acrylic acid; acetylene black; and acetylenic chemicals such as 1,4-buty-
nediol and acetylenic alcohols. For the acetylenic chemicals the acetylene
route is either the only commercial production process available or the predo-
minant process. Vinyl chloride, vinyl acetate, and acrylic acid, formerly the
Vol. 2 ACETYLENE 715

Fig. 48. Acetylene as a starting material for industrial products.

main products from acetylene, are produced today mainly from ethylene and
propene (3).
6.3. Competitive Position of Acetylene as Chemical Feedstock.
Today, acetylene plays an important role only in the production of the acetylenic
chemicals. The fact that acetylene production has not decreased further seems to
indicate that the competition from the olefins is no longer as strong as it was. The
main reason for this is that European olefin chemistry depends on refinery pro-
ducts, which have become more expensive than natural gas, the main feedstock
for acetylene. Another contributing factor is that acetylene is produced only in
old plants, which have low capital costs.
In addition, process improvements, such as an increase in thermal effi-
ciency and optimum use of byproducts by other plants, can make acetylene
more competitive. The position of acetylene in chemical industry may be
advanced because of the variety of valuable products to which it can be converted
in high yields with known technology. Acetylene must compete with ethylene for
the production of vinyl chloride and vinyl acetate, and for the production of
acrylic acid and its esters it must compete with propene. The Stanford Research
Institute (3) has investigated this question in detail. The results, which take into
account both capital investment (25% return) and specific consumption figures
for the various processes, show that the prices at which the alternate routes
are competitive can be expressed by the following equations:

for vinyl chloride A ¼ 1.10 E þ 0.42


for vinyl acetate A ¼ 1.23 E þ 0.40
for acrylic acid A ¼ 1.74 P þ 0.23
716 ACETYLENE Vol. 2

Fig. 49. Acetylene – methanol plant.

where A is the acetylene price, E is the ethylene price, and P is the propene price,
all in $/kg. For example, if ethylene costs 0.65 $/kg, acetylene can cost 1.15 $/kg
for the production of vinyl chloride and 1.21 $/kg for the production of vinyl
acetate. If propene costs 0.49 $/kg, acrylic acid can be profitably produced from
acetylene if it costs 1.08 $/kg or less.
Acetylene prices of 1.08 to 1.21 $/kg or less can be reached in a new plant
only with optimal integration of energy and byproducts within an integrated
chemical plant. Figure 49 shows the flow sheet of a 100000 t/a gas-based partial
oxidation acetylene plant in a chemical complex. The required oxygen facility
and methanol plant based on acetylene synthesis gas (off-gas) are included.
The main products of the complex are acetylene and methanol. The acetylene
process is operated with improved quenching technology, allowing a high
proportion of energy to be regained in the form of steam. The aromatic
residue oil from a steam cracker is converted into high-purity coke and light
aromatics.
A production cost estimate based on power, feedstock, and product prices
roughly corresponding to market prices in 1982 shows that low production
costs for acetylene are possible under the conditions described, in spite of the
relatively high natural gas price of 5.5 $/106 BTU (0.021 $/kW h), which in heat-
ing value terms comes very close to that of crude oil (0.022 $/kW h). The differ-
ence in production costs between acetylene from natural gas and ethylene
from naphtha (price 333 $/t, corresponding to 0.028 $/kW h) is so slight that acet-
ylene from new plants can once again compete with ethylene for certain synth-
eses, provided that there is a difference between the costs of natural gas and
naphtha. This was nearly 0.007 $/kW h in 1982 in Germany. A similar calcula-
tion for the acetylene production by the Hüls arc process based on an ethylene
price of 600 $/t and a hydrogen price 40% above the heating value results in
the same acetylene product value at 0.038 $/kW h for electrical power. Lower
electrical energy costs favor the arc process.
Vol. 2 ACETYLENE 717

7. Propyne

Propyne [74-99-7], methylacetylene, is obtained in cracking processes mostly as a


byproduct together with its isomer propadiene [463-49-0], allene. Typical concen-
trations depend on the feedstock and the cracking conditions (Table 10) and
vary between 0.3 and 0.8 wt%. The corresponding figures for propadiene are
0.3–0.55 wt%.
Pure propyne is a colorless, nontoxic, flammable gas. The important physi-
cal properties of propyne are listed in Table 18 (6,7,124). Vapor pressure curves
of propyne and propadiene including melting and critical points are shown in
Figure 50 (124). Table 19 summarizes solubility coefficients of propyne for var-
ious solvents (9); further data are available in (10). The solubility of propyne
in various solvents is shown in Figure 51; the solubilities of the C3 hydrocarbons
in DMF are plotted in Figure 52 for infinite dilution.
The equilibrium between the two isomers of C3H4 is reached in the presence
of catalysts (for example Al2O3/3, SiO2, actived carbon, and g-Al2O3/Na2CO3)
(125,126). At 2708C the equilibrium mixture contains 82% propyne, and at
58C, 91.1% propyne. For calculated data, see (127). This equilibrium is important
for the industrial propyne recovery process.
The decomposition pressure as a function of the ignition energy for propyne
(and propadiene) is plotted in Figure 38. The lower and upper flammability limits
of propyne in air are about 2.3 and 16.8 vol%. For propadiene only ranges are
available: 1.7–2.5 and 12–17 vol% (128).
Production. Propyne and propadiene can be recovered from cracked gas
by solvent extraction. The process is outlined in Figure 53 (72,129) and is similar
to the recovery of acetylene (see Section 4.4). An important step is the catalytic
isomerization of propadiene to propyne in the liquid phase on a catalyst (126).

Table 18. Physical Properties of Propyne (6,7,124)


Molecular mass 40.065
Critical temperature 402.39 K (129.24 8C)
Critical pressure 5.626 MPa
Critical volume 0.1635 m3/kmol
Melting point 170.45 K ( 102.7 8C)
Dipole moment 2.61 1030 C  m
Density (liquid) 638.92 kg/m3 (273.15 K)
Normal boiling point 249.97 K ( 23.18 8C)
Enthalpy of vaporization (273.15 K) 20.765 kJ/mol (273.15 K)
Molar volume 0.05962 m3/kmol (249.91 K)
Enthalpy of formation 185.5  1.0 kJ/mol (298.15 K)
Gibbs free energy of formation 193.8  1.0 kJ/mol (298.15 K)
Entropy of formation 248.4 J mol1 K1 (298.15 K)
Enthalpy of combustion 1938.943 kJ/mol (298.15 K)
Heat capacity constant pressure 59.842 J mol1 K1 (273.15 K)
Viscosity (liquid) 1.7500 104 Pa  s (273.15 K)
Viscosity (gaseous) 8.3300 106 Pa  s (293.15 K)
Thermal conductivity (liquid) * 0.14560 W m1 K1 (233.45 K)
Thermal conductivity (gas) * 0.014310 W m1 K1 (273.15 K)
Surface tension * 1.47 102 N/m (273.15 K)
* Predicted or estimated.
718 ACETYLENE Vol. 2

Fig. 50. Vapor pressure curves of propyne and propadiene (124).

The bottom product from propene/propane splitter is routed to depropanizer


II for removal of traces of C4+ hydrocarbons originating from the feed and the iso-
merization reactor effluent. The overhead is sent to the absorption column. The
propane (and some propene) is routed back to the cracking furnace. Propyne
(MA) and propadiene (PD) are stripped off in the corresponding columns. The
propadiene fraction is converted to propyne in the isomerization reactor and
recycled to the feed. Traces of solvent in the pure propyne are removed by cool-
ing. The product can be sent directly to methyl methacrylate (MMA) unit, where

Table 19. Solubility Coefficients of Propyne in Various Solvents (mol kg1 bar1)(9)
Solvent C3H4 pressure, bar  20 8C 25 8C
Methanol 0.1 9.099 1.865
1,2-Dichlorethane 0.098 – 0.196 3.276 (0 8C) 1.546
Carbon tetrachloride 0.25 4.732 0.842
n-Octane 0.098 3.412
Toluene 0.25/1.0 8.644 2.047
Xylene (technical). 0.49/0.98 9.782 1.683
4-Methyl-1,3-dioxalan-2-one 0.49/1.0 8.644 1.183
Triethylene glycol 1.0 0.400 (30 8C)
Acetone 0.6/1.0 35.03 4.186
N-Methyl-2-pyrrolidone  0.78 6.597 (0 8C) 2.502
DMF  0.15 12.512 3.003
Water 1.0 0.071
Ammonia  0.04 20.018 4.436
Vol. 2 ACETYLENE 719

Fig. 51. Solubility of propyne in various solvents at infinite dilution. a) Water;


b) Methanol; c) N-methylpyrrolidone; d) DMF.

Fig. 52. Solubility of the C3 hydrocarbons in DMF at infinite dilution. a) Propyne;


b) Propadiene; c) Propene; d) Propane.
720 ACETYLENE Vol. 2

Fig. 53. Propyne recovery process (Linde, Shell (75)). MA: methylacetylene; PD:
propadiene.

MMA is manufactured by reaction of propyne with carbon monoxide and an


alcohol in the presence of a Pd-based carboxylation catalyst (126,130).
Usually propyne and propadiene are undesireable byproducts of the steam
cracking process and they are removed by selective hydrogenation. The reaction
is carried out in the liquid or gas phase. Typical conditions are pressures of
1.5–5 MPa and inlet temperatures of 20–1008C. The vapor-phase processes are
declining in importance; liquid-phase (or composite phase) processes have some
advantages:

Operation at lower temperatures (higher selectivity)


Lower operating costs (no vaporization of the feed and recondensation of the
product)
Regeneration of reactor in situ (removal of polymers)
Lower frequency of catalyst regeneration
For high concentration of C3H4 in the feed, two reactors are often sufficient;
the gas-phase process requires more reactors
Control of conversion by injection of hydrogen (no excess hydrogen) reduces
investment cost
Control of temperature by use of vaporization minimizes risk of runaway

The selectivity in case of liquid-phase hydrogenation to propylene is about


60–70% and depends on the concentration of C3H4 in the C3 feed. For better con-
trol of reaction heat, some product is normally recycled to the feed. The cycle time
of the catalyst is very sensitive to contaminants (and to byproduct formation in
gas-phase processes), especially in the first reactor. A reactivation procedure is
necessary when the catalyst loses its activity.
Stabilized Propyne–Propadiene Mixtures. As a replacement for acet-
ylene, stabilized propyne–propadiene mixtures are available commercially.
Vol. 2 ACETYLENE 721

Trade names are Tetrene or MAPP gas. These mixtures are stabilized by propa-
ne,propene, and/or butane and are used for metal cutting, welding, hardening,
and brazing. The flame properties are closer to those of propane–propene mix-
tures. Therefore stabilized C3H4 mixtures have not yet won a large-scale market.
Further information is available in (131).

8. Toxicology and Occupational Health

Pure acetylene is a simple asphyxiant. When generated from calcium carbide,


acetylene is frequently contaminated with arsine, hydrogen sulfide, or phos-
phine, and exposure to this impure acetylene has often resulted in serious con-
sequences. Commercial acetylene no longer contains these impurities and is
therefore less harmful (132).
The lowest published lethal concentration for rats is 9 vol% (133). Dogs are
less sensitive: 80 vol% acetylene in the air is necessary to produce a narcosis
accompanied by an increased blood pressure and a decreased pulse frequency
(stimulation of vasomotor and vagus centers) (134). In humans, the inhalation
of air containing 10 vol% acetylene has a slight intoxicating effect, marked intox-
ication occurs at 20 vol%, incoordination at 30 vol%, and unconsciousness within
5 min on exposure to 35 vol%. Inhaling 35 vol% for 5–10 min or 10 vol% for
30–60 min is lethal. Symptoms of intoxication are excitement, coma, cyanosis,
weak and irregular pulse, and memory failure (135–137).
There is no evidence that repeated exposure to tolerable levels of acetylene
has effects deleterious to health (138). Inhalation of air with 33 vol% of acetylene
by humans led to unconciousness within 6 min, but when the experiment was
repeated within the week the susceptibility to acetylene decreased: 9 min were
required on the second exposure and more than 33 min on the third exposure
to produce unconsciousness (134).
Acetylene does not irritate the mucous membranes (132). Neither threshold
limit value (TLV) nor a MAK has been established. The standard air concentra-
tion allowed for OSHA and NIOSH is 2500 ppm (136).

REFERENCES

1. Acetylene or Ethylene as Feedstocks for the Chemical Industry, Conference proceed-


ings Dechema and SCI, Frankfurt/Main, 27–29 March 1968, SCI, London 1968.
E. Schenk: ‘‘Acetylen oder Äthylen als Rohstoffe der chemischen Industrie,’’
DECHEMA, Frankfurt/Main 1968.
2. O. Horn, Erdöl Kohle Erdgas Petrochem. 26 (1973) no. 3, 129.
3. Yen-Chen Yen, Acetylene, Report Nr. 16, Supplement A, Stanford Research Inst.,
Menlo Park, Calif., Nov. 1981, and Information of the VCI, Frankfurt/Main.
4. J. Schulze, M. Homann, ‘‘Die mögliche Stellung des Acetylens in der zukünftigen
Kohlechemie,’’ Erdöl Kohle Erdgas Petrochem. 36 (1983) no. 5, 224.
5. D. A. Plattner, Y. Li, K. N. Houk: ‘‘Modern Computational and Theoretical Aspects of
Acetylene Chemistry’’, Chapter 1.2. in P. J. Stang, F. Diederichs (eds.): Modern Acet-
ylene Chemistry, VCH Verlagsgesellschaft, Weinheim, Germany 1995.
722 ACETYLENE Vol. 2

6. DIPPR Database, STN International, Design Institute for Physical Property Data
c/o DIPPR Project Staff, Pennsylvania State University 167 Fenske Lab, University
Park, Pa 16802 U.S.A American Institute of Chemical Engineers, 3345 E. 47th
Street, New York, NY 10017, U.S.A., basis February 1998.
7. Beilstein Database, STN International, Beilstein Informationssysteme Carl-Bosch-
Haus, Varrentrappstr. 40–42, D-60486 Frankfurt am Main, Germany, basis Febru-
ary 1998.
8. W. Wiechmann, Amts- und Mitteilungsblatt der Bundesanstalt für Material-
forschung und -prüfung (BAM) (1987) 3, 505.
9. A. Kruis: ‘‘Gleichgewicht der Absorption von Gasen in Flüssigkeiten,’’ in H. Hausen
(ed.): Landolt-Börnstein Zahlenwerte und Funktionen, 4. Teil, Bestandteil c,
Springer Verlag, Berlin–Heidelberg–New York 1976.
10. S. A. Miller: Acetylene, Its Properties, Manufacture and Uses, Ernest Benn, London
1965.
11. F. Rottmayr, H. Reimann, U. Lorber, Linde Ber. Tech. Wiss. 30 (1971) 3.
12. H. Schmidt, D. Forney: Oxygen Technology Survey, vol. 9, NASA SP-3090, ASRDI,
1975.
13. E. Lassmann, Linde Reports on Science and Technology 57 (1996) 16.
14. UCC isotherm data sheet No. 44, Union Carbide Corp., New York 1980.
15. D. P. Velenuela, A. L. Myers: Adsorption Equilibrium Data Handbook, Department
of Chem. Engineering, University of Pennsylvania, Philadelphia, Prentice Hall,
Englewood Cliffs, NJ 1989.
16. T. F. Rutledge: Acetylenic Compounds–Preparation and Substitution Reactions,
Reinhold Publ. Co., New York 1968.
17. A. C. Mc Kinnis, Ind. Eng. Chem. 27 (1962) 2928.
18. S. A. Miller: Acetylene, Its Properties, Manufacture and Uses, Ernest Benn, London
1965.
19. R. C. West, C. S. Kraihanzel, J. Am. Chem. Soc. 83 (1961) 765.
20. H. J. Copley, C. E. Holley Jr., J. Am. Chem. Soc. 61 (1939) 1599.
21. W. Reppe: Neue Entwicklungen auf dem Gebiet der Chemie des Acetylens und
Kohlenoxids, Springer Verlag, Berlin-Göttingen-Heidelberg 1949.
22. W. Reppe: Chemie und Technik der Acetylen-Druckreaktionen, 2nd ed., Verlag
Chemie, Weinheim, Germany 1952.
23. N. von Kutepow in: Ullmann, 4th ed., vol. 7, p. 44.
24. W. Reppe, Justus Liebigs Ann. Chem. 601 (1956) 81.
25. W. Reppe et al., Justus Liebigs Ann. Chem. 569 (1955) 1.
26. W. Reppe et al., Justus Liebigs Ann. Chem. 582 (1953) 1.
27. W. Reppe, N. von Kutepow, A. Magin, Angew. Chem. 81 (1969) 717.
28. Lonza, DE 1251329, 1964 (P. Pino, G. Braca, G. Sbrana).
29. BASF, DE-AS 1071077, 1965 (W. Reppe, A. Magin). Du Pont, DE 1054086, 1955
(J. C. Sauer). J. C. Sauer, R. D. Cramer, V. A. Engelhardt, T. A. Ford, H. E. Holmquist,
B. W. Howk, J. Am. Chem. Soc. 81 (1959) 3677. G. Albanesi, M. Toraglieri, Chim. Ind.
(Milan) 41 (1959) 189.
30. G. Palyi, G. Varadi, J. T. Horvath, J. Mol. Catal. 13 (1981) 61.
31. W. Reppe et al., Justus Liebigs Ann. Chem. 560 (1948) 1.
32. G. Schröder: Cyclooctatetraen, Verlag Chemie, Weinheim, Germany 1965.
33. J. A. Nieuwland, W. S. Calcott, F. B. Downing, A. S. Carter, J. Am. Chem. Soc.
53 (1931) 4197.
34. J. C. W. Chien, Polym. News 6 (1979) 52. D. Bloor, New Sci. 93 (1982) no. 1295, 577.
H. Shirakawa, Kotai Butsuri 16 (1981) no. 7, 402; Chem. Abstr. 96 (1982) 172568.
A. J. Heger, A. G. Macdiarmid, Int. J. Quantum Chem. Quantum Chem. Symp. 15
(1981) 243. H. Shirakawa, Kagaku (Kyoto) 37 (1982) no. 3, 181; Chem. Abstr. 96
Vol. 2 ACETYLENE 723

(1982) 153398. H. Shirakawa, Kagaku, Zokan (Kyoto) 1980, no. 87, 165; Chem.
Abstr. 94 (1981) 139105 x.
35. G. Wegner, Angew. Chem. 93 (1981) 352.
36. T. Mole, J. R. Suertes, Chem. Ind. (London) 1963, 1727.
37. P. N. Rylander, ‘‘Platinum Metal Catalysts in Organosilicon Chemistry,’’ Engelhard
Ind. Tech. Bull. 10 (1970) no. 4, 130.
38. J. L. Speier, Adv. Organomet. Chem. 17 (1979) 407.
39. E. Ott et al., Ber. Dtsch. Chem. Ges. 76 (1943) 80–91.
40. I. G. Farbenind., DE 495787, 1927.
41. Chem. Week 98 (1966) Apr. 16, 90.
42. Eur. Chem. News 25 (1974) Feb. 1, 5.
43. H. Sachsse, Chem. Ing. Tech. 26 (1954) 245. E. Bartholome, Chem Ing. Tech.
26 (1954) 253. T. P. Forbath, B. I. Gaffney, Pet. Refiner 33 (1954) 160.
44. H. Friz, Chem. Ing. Tech. 40 (1968) 999.
45. K. G. Baur, K. Taglieber, Chem. Ing. Tech. 47 (1975) 385.
46. M. J. Zundel: Proble`mes de Fabrication d’Acetylène et d’Ethylène à partir d’Hydro-
carbures, Societe de Chimie Industrielle, Paris 1962.
47. Hydrocarbon Process. Pet. Refiner 44 (1965) no. 11, 163.
48. Montecatini, GB 1000480, 1962.
49. J. L. Petton et al., Pet. Refiner 37 (1958) 180.
50. Hydrocarbon Process. 54 (1975) no. 11, 104. Stanford Research Report no. 109,
September 1976. L. Verde, R. Riccardi, S. Moreno, Hydrocarbon Process. 57 (1978)
no. 1, 159.
51. G. Hamprecht, M. Gettert: ‘‘Sauerstoff-Thermisches Calziumcarbid,’’ Festschrift für
Carl Wurster, BASF, Ludwigshafen 1960, p. 43.
52. H. Gladisch, Hydrocarbon Process. Pet. Refiner 41 (1962) 159–164.
53. H. Gladisch, Chem. Ing. Tech 61 (1969) 204–208.
54. K. Gehrmann, H. Schmidt, Water Air Conserv. Pet. Ind, 1971, 379.
55. R. L. Bond, W. R. Ladner, G. I. McCommet, Fuel 45 (1966) 381–395. R. E. Gannon,
V. J. Krukonis, Th. Schoenberg, Ind. Eng. Chem. Prod. Res. Dev. 9 (1970) 343–347.
R. E. Gannon, S. K. Ubhayakar, Fuel 56 (1977) 281. AVCO Corp., US 179144,
1980 (Ch. Kim). D. Bittner, H. Baumann, C. Peuckert, J. Klein, H. Jüntgen, Erdöl
Kohle Erdgas Petrochem. 34 (1981) no. 6, 237–242. R. Müller, C. Peuckert,
Int. Symp. on Plasma Chem., 5th, Edinburgh 1981, p. 197–202.
56. Eur. Chem. News 36 (1981) Feb. 2, 19.
57. R. Müller, World Hydrogen Energy Conf., Conf. Proc., 4th, Pasadena, 1982,
p. 885–900.
58. GAF, US 219756, 1980 (M. Katz, F. Carluccio).
59. Verordnung über Acetylenanlagen und Calcium-carbidlager (Acetylenverordnung–
Acet V) vom 27. Februar 1980, Carl Heymanns Verlag, Köln 1980.
60. TRAC 201–Acetylenentwickler, Carl Heymanns Verlag, Köln.
61. Acetylene Transmission for Chemical Synthesis (Recommended Minimum
Safe Practices for Piping Systems) International Acetylene Association, New York
1980.
62. C. Hase, W. Reitze: Fachkunde des Autogenschweißens, 7th ed., Girardet, Essen
1965.
63. Carbidkalk, Hinweise für seine Verwendung, Fachbuchreihe Schweißtechnik,
9th ed., Deutscher Verlag für Schweißtechnik (DVS), Düsseldorf 1968, p. 16.
64. Shawinigan Chemicals, US 1872741, 1931 (R. S. Jane).
65. Bayerische Stickstoff-Werke, DE 714 323, 1938 (R. Wendlandt, R. Neubner).
66. Shawinigan Chemicals, US 1343185, 1942 (A. C. Holm, E. Poirier).
67. Linde, DE 906 005, 1951 (F. Rottmayr). Linde, DE 2 549 399, 1975 (E. Laßmann).
724 ACETYLENE Vol. 2

68. Hydrocarbon Process. 46 (1967) no. 11, 139. T. Wett, Oil Gas J. 70 (1972)
101–110.
69. Ullmann, 4th ed., vol. 7.
70. H. K. Kamptner, Erdöl Kohle Erdgas Petrochem. 16 (1963) 547. H. K. Kamptner,
W. R. Krause, H. P. Schilken, Hydrocarbon Process. Pet. Refiner 45 (1966) no. 4,
187. K. Lissa, Chem. Anlagen Verfahren 1970, no. 6, 83.
71. Y. C. Hu, Hydrocarbon Process. 61 (1982) no. 11, 109.
72. D. Sohns, Linde Reports Science and Technology 30 (1979) 21.
73. F. Mey, H. D. Neubauer, R. Schubert, Petrochemicals and Gas Processing Petroleum
Technology quaterly (1997) Autumn, 119.
74. K. J. Sasaki, Petrochemicals and Gas Processing Petroleum Technology quaterly
(1997) Autumn, 113.
75. P. Cl. Haehn, Dr. E. Haidegger, Dr. N. Schödel, Hydrocarbon Engineering (1997)
January/February, 41.
76. Linde München folder C/3.3.e/93
77. G. Marcks, Schadenprisma 3 (1982) 37. D. Lietze, Amts- und Mitteilungsblatt der
Bundesanstalt für Materialprüfung (BAM) 16 (1986) 1, 23.
78. W. Reppe, Chem. Ing. Tech. 22 (1950) 273.
79. CEN/TC 121/SC 7/WG3 German Proposal: Pipelines in Acetylene Systems,November
1996, NAS-DIN Deutsches Institut für Normung e.V., D 10772 Berlin or ‘‘Empfeh-
lungen für Acetylenleitungen auf der Grundlage von Arbeitsbereichen’’, Industrial
Gases Committee IGC Document 9/78/D, F 75880 Paris, CEDEX 18
80. H. B. Sargent, Chem. Eng. 64 (1957) 250.
81. Th. Schendler, H.-P. Schulze, Chem.-Ing.-Tech. 62 (1990) 1, 41. W. Reppe: Chemie
und Technik der Acetylen-Druck-Reaktionen, Verlag Chemie, Weinheim, Germany
1951.
82. D. Lietze, H. Pinkofsky, T. Schendler, H.-P. Schulze, Chem.-Ing.-Tech. 61 (1989) 9,
736. H. Große-Wortmann, N. Kalkert, H.-G. Schecker, Chem.-Ing.-Tech. 53 (1981)
461.
83. A. Williams, D. B. Smith, Chem. Rev. 70 (1970) 267. B. A. Ivanov, S. M. Kogarko,
Int. Chem. Eng. 4 (1964) 4, 670. C. M. Detz, Combust. Flame 34 (1979) 187. M. A.
Glikin et al., Sov. Chem. Ind. (Engl. Transl.) 7 (1975) 1373. A. Baumeier, D.
Conrad, S. Dietlen, W. Pezold, T. Schendler und H.-P. Schulze, Chem.-Ing.-Tech.
64 (1992) 3, 260.
84. R. Grätz, M. Wagenknecht, vfdb, Zeitschrift für Forschung, Technik und Manage-
ment im Brandschutz (1994) 3, 103.
85. D. Lietze, Chem. Ing.-Tech. 63 (1991) 11, 1148.
86. D. Lietze, Chem.-Ing.-Tech. 62 (1990) 3, 238.
87. E. Karwat, Chem Eng. Progr. 54 (1958) 10, 96.
88. D. Lietze, Journal of Hazardous Materials. 54 (1997) 227.
89. D. Lietze, J. Loss Prev. Process Ind. 8 (1995) 6, 319.
90. D. Lietze, J. Loss. Process Ind. 8 (1995) 6, 325.
91. A. D. Hewitt: Technology of Oxy-Fuel Gas Processes, Welding and Metal Fabrica-
tion, Part 2, 1972 (November) 382. H. Meinass, H. Manhard, J. Schlander, L. Fruh-
storfer, Linde Reports on Science and Technology 24 (1976) 32. L. Kögel, Linde
Reports on Science and Technology 32 (1981) 36.
92. Technische Regeln für Acetylenanlagen und Calciumcarbidlager, TRAC 207, 208,
Deutscher Acetylenausschuß, Bundesanstalt für Arbeitsschutz und Unfall-
forschung, Dortmund 1969.
93. K. H. Roch, Amts Mitteilungsbl. Bundesanst. Materialprüf. Berlin 12 (1982) 283.
94. K. H. Roch, Schweissen + Schneiden 25 (1973) 94.
95. D. Lietze, Amts Mitteilungsbl. Bundesanst. Materialprüf. Berlin 2 (1972) 9.
Vol. 2 ACETYLENE 725

96. D. Lietze, Berufsgenossenschaft 1976, 435.


97. M. E. Sutherland, M. W. Wegert, Chem. Eng. Prog. 69 (1973) no. 4, 48.
98. C. Isting, Erdöl Kohle Erdgas Petrochem. 23 (1970) 29.
99. Chem. Eng. (N.Y.) 76 (1969) no. 1, 89.
100. E. Karwat, Chem. Eng. Prog. 53 (1957) no. 4, 27. E. Karwat, Linde Ber. Tech. Wiss.
13 (1962) 12; ‘‘Safety in Air and Ammonia Plants,’’ Chem Eng. Prog. Tech. Man.
5 (1963) 43. G. Klein in: ‘‘Luftzerlegungsanlagen,’’ Linde-Arbeitstagung, München
1975.
101. L. W. Coleman: ‘‘Safety in Air and Ammonia Plants,’’ Chem. Eng. Prog. Tech. Man.
4 (1962) 26.
102. E. Karwat, Chem. Eng. Prog. 57 (1961) no. 4, 5.
103. F. G. Kerry, Chem. Eng. Prog. 52 (1956) no. 11, 3.
104. J. Reyhing, Linde Rep. Sci. Technol. 36 (1983) 14.
105. Engelhard Ind., DE-AS 1283805, 1968 (J. G. Cohn A. J. Haley, Jr.).
106. G. Klein, Linde Ber. Tech. Wiss. 17 (1964) 24. Cryogenics Safety Manual, Part II,
British Cryogenics Council, London 1970. H. H. Hofmaier: ‘‘Safety in Air and Ammo-
nia plants’’, Chem Eng. Prog. Tech. Man. 5 (1963) 22.
107. E. Karwat, Chem. Eng. Prog. Tech. Man. 2 (1960) A-18.
108. G. Drewes, M. Ermscher, Chem. Tech. (Leipzig) 35 (1983) 57.
109. K. H. Möller, C. Stöber, K. Schulze, Arbeitsschutz 23 (1972) no. 1, 18.
110. P. Hölemann, R. Hasselmann, Forschungsberichte des Landes Nordrhein-Westfalen,
no. 765, Westdeutscher Verlag, Köln-Opladen 1959.
111. P. Hölemann, Forschungsberichte des Landes Nordrhein-Westfalen, no. 888 and
1151, Westdeutscher Verlag, Köln-Opladen 1960 and 1963.
112. K. H. Möller, Arbeitsschutz 22 (1971) no. 1, 6.
113. ‘‘Technische Grundsätze für ortsbewegliche Druckgasbehälter,’’ Ziffer 29 and 31, Ar-
beitsschutz 21 (1970) no. 3.‘‘Allgem. Verwaltungsvorschrift zu 14, 17–19 der Verord-
nung über ortsbewegliche Behälter und über Füllanlagen für Druckgase,’’
20. 6. 1968.
114. Department of Transportation, Office of Hazardous Materials Regulations, Code of
Federal Regulations 49.
115. K. H. Möller, Arbeitsschutz 23 (1972) no. 2, 30.
116. International Organization for Standardization– ISO/TC 58/WG 1.
117. K. H. Möller, Berufsgenossenschaft 1972, 375, 422.
118. H. Springmann, Linde Rep. Sci. Technol. 34 (1982) 54.
119. E. Zorn, Mitt. BEFA 14 (1963) no. 6, 2.
120. Safety in the Production and Use of Acetylene, Commission permanente internatio-
nale de l’acetylène, Paris 1968.
121. A. D. Hewitt: Welding and Metal Fabrication, part 1, Oct. 1972, 347; part 2, Nov.
1972, 382; part 3, Dec. 1972, 416, IPC Science and Tech. Press, Guildford, England.
122. F. Houldcroft: Welding Processes, Oxford University Press, London 1975. J. Ruge:
Handbuch der Schweißtechnik, Springer Verlag, Berlin 1980. Bibliographies, Amer-
ican Welding Soc., New York (appears annually).
123. H. Weiler, Schweißen + Schneiden 26 (1974) 220. L. Kögel, Linde Ber. Tech. Wiss.
48 (1980) 36.
124. Engineering Sciences Data Unit, ‘‘London Vapour Pressures and Critical Points of
Liquids’’, Part 2C: alkadienes and alkynes, Number 86001 (1986)
125. J. F. Cordes, H. Günzler, Chem. Ber. 92 (1959) 1055. J. F. Cordes, H. Günzler,
Z. Naturforschung 15b (1960) 682. C. P. Khulbe, R. S. Mann, Prep.-Can. Symp.
Catal., 5th (1977) 384. P. Kos, I. Kiricsi, K. Varga, P. Fejes, Acta Phys. et. Chem.
Szeged 33 (1987) 109. F.-D. Zeiseler, G. Zimmermann, Journal für Prakt. Chemie
319 (1977) 4, 655.
726 ACETYLENE Vol. 2

126. Shell Internationale, EP 0 392 601, 1990 (M. J. Doyle, J. Van Gogh, J. Van Ravens-
waay Claasen)
127. D. A. Frank-Kamenetzki, V. G. Markovich, Acta Physicochim, (U.R.S.S.) 17 (1942)
308. H. Zeise (ed.): Thermodynamik, vol. III/1, Verlag Hirzel, Leipzig 1954.
128. T. Redeker, G. Schön, Sicherheitstechnische Kennzahlen brennbarer Gase und
Dämpfe, 6. Nachtrag Deutscher Eichverlag GmbH, Braunschweig 1990.
129. Linde AG, US 4 701 190, 1986 (P. C. Haehn), EP 0 224 748, 1986 (P. C. Haehn).
130. J. Happel, S. Umemura, Y. Sakakibara, H. Blanck, S. Kunichika, Ind. Eng. Chem.
Process Des. Develop. 14 (1975) 1, 44. E. Drent, P. Arnoldy, P. H. M. Budzelaar,
Journal of Organometallic Chemistry, 455 (1993) 247. Shell Internationale, EP
0 539 628, 1993 (J. Hengeveld, P. B. de Blank).
131. R. F. Huston, C. A. Barrios, R. A. Holleman, J. Chem. Eng. Data 15 (1970) 168. R. D.
Green, Met. Prog. 108 (1975) 2, 71. A. Farwer, Gas Aktuell (1982 Nov.) 24.
132. L. T. Fairhall: Industrial Toxicology, 2nd ed., The Williams and Wilkins Co.,
Baltimore 1957, p. 270.
133. The International Technical Information Institute: Toxic and Hazardous Industrial
Chemicals Safety Manual, Tokyo, 1975, p. 10.
134. T. Sollmann: A Manual of Pharmacology and its Application to Therapeutics, 7th
ed., W. B. Saunders Co., Philadelphia-London 1949, p. 662.
135. F. Flury, F. Zernik: Schädliche Gase, Springer Verlag, Berlin 1931, p. 270.
136. NIOSH, U.S. Dept. of Health and Human Services: Registry of Toxic Effects of
Chemical Substances, vol. 1, Washington, DC, 1980, p. 70.
137. E. R. Plunkett: Handbook of Industrial Toxicology, Heyden, Barberton, Ohio, 1976,
p. 8.
138. F. A. Patty: Industrial Hygiene and Toxicology, 2nd ed., Interscience Publ.,
New York 1962, p. 1205.

PETER PÄSSLER
WERNER HEFNER
BASF Aktiengesellschaft, Ludwigshafen,
Federal Republic of Germany
KLAUS BUCKL
HELMUT MEINASS
HANS-JÜRGEN WERNICKE
Linde AG, Höllriegelskreuth, Federal
Republic of Germany
GÜNTER EBERSBERG
RICHARD MÜLLER
Degussa-Hüls AG, Marl, Federal Republic
of Germany
JÜRGEN BÄSSLER
Uhde GmbH, Dortmund, Federal Republic
of Germany
HARTMUT BEHRINGER
DIETER MAYER
Hoechst Aktiengesellschaft, Pharma,
Forschung, Toxikologie, Frankfurt,
Federal Republic of Germany
Vol. 2 ETHYLENE 727

ETHYLENE
1. Introduction

Ethylene [74-85-1], ethene, H2C CH2, Mr 28.52, is the largest-volume


petrochemical produced worldwide. Ethylene, however, has no direct end uses,
being used almost exclusively as a chemical building block. It has been recovered
from coke-oven gas and other sources in Europe since 1930 (1). Ethylene emerged
as a large-volume intermediate in the 1940s when U.S. oil and chemical compa-
nies began separating it from refinery waste gas and producing it from ethane
obtained from refinery byproduct streams and from natural gas. Since then,
ethylene has almost completely replaced acetylene for many syntheses. Ethylene
is produced mainly by thermal cracking of hydrocarbons in the presence of
steam, and by recovery from refinery cracked gas.
In 1996 total worldwide ethylene production capacity was 79.3  106 t, with
an actual demand of ca. 71  106 t/a (2), which has growth projections of 4.5% per
year worldwide for the period of 1996 to 2005 (3–5).

2. Physical Properties

Ethylene is a colorless flammable gas with a sweet odor. The physical properties
of ethylene are as follows:
mp 169.158C
bp 103.718C
Critical temperature, Tc 9.908C
Critical pressure, Pc 5.117 MPa
Critical density 0.21 g/cm3
Density
at bp 0.57 g/cm3
at 08C 0.34 g/cm3
Gas density at STP 1.2603 g/L
Density relative to air 0.9686
Molar volume at STP 22.258 L
Surface tension
at bp 16.5 mN/m
at 08C 1.1 mN/m
Heat of fusion 119.5 kJ/kg
Heat of combustion 47.183 MJ/kg
Heat of vaporization
at bp 488 kJ/kg
at 08C 191 kJ/kg
Specific heat
of liquid at bp 2.63 kJ kg1 K1
of gas at Tc 1.55 kJ kg1 K1
Enthalpy of formation 52.32 kJ/mol
Entropy 0.220 kJ mol1 K1
Thermal conductivity
at 08C 177104 W m1 K1
at 1008C 294104 W m1 K1
at 4008C 805104 W m1 K1

ß2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim


10.1002/14356007.a10_045
728 ETHYLENE Vol. 2
Viscosity of liquid
at mp 0.73 mPa  s
at bp 0.17 mPa  s
at 08C 0.07 mPa  s
of gas
at mp 36104 m Pa  s
at 08C 93104 m Pa  s
at 1508C 143104 m Pa  s
Vapor pressure
at 1508C 0.002 MPa
at bp 0.102 MPa
at 508C 1.10 MPa
at 08 4.27 MPa
Explosive limits in air at
0.1 MPa and 208C
lower (LEL) 2.75 vol% or 34.6 g/cm3
upper (UEL) 28.6 vol% or 360.1 g/cm3
Ignition temperature 425–5278C

3. Chemical Properties

The chemical properties of ethylene result from the carbon – carbon double
bond, with a bond length of 0.134 nm and a planar structure. Ethylene is
a very reactive intermediate, which can undergo all typical reactions of
a short-chain olefin. Due to its reactivity ethylene gained importance as a
chemical building block. The complex product mixtures that have to be
separated during the production of ethylene are also due to the reactivity of
ethylene.
Ethylene can be converted to saturated hydrocarbons, oligomers, polymers,
and derivatives thereof. Chemical reactions of ethylene with commercial impor-
tance are: addition, alkylation, halogenation, hydroformylation, hydration, oligo-
merization, oxidation, and polymerization.
The following industrial processes are listed in order of their 1993
worldwide ethylene consumption (6):

1. Polymerization to low-density polyethylene (LDPE) and linear low-density


polyethylene (LLDPE)
2. Polymerization to high-density polyethylene (HDPE)
3. Addition of chlorine to form 1,2-dichloroethane
4. Oxidation to oxirane [75-21-8] (ethylene oxide) over a silver catalyst
5. Reaction with benzene to form ethylbenzene [100-41-4], which is dehydro-
genated to styrene [100-42-5]
6. Oxidation to acetaldehyde
7. Hydration to ethanol
8. Reaction with acetic acid and oxygen to form vinyl acetate
9. Other uses, including production of linear alcohols, linear olefins, and
ethyIchloride [75-00-3], and copolymerization with propene to make
ethylene – propylene (EP) and ethylene – propylene – diene (EPDM) rubber
Vol. 2 ETHYLENE 729

4. Raw Materials

Table 1 lists the percentage of ethylene produced worldwide from various feed-
stocks for 1981 and 1992 (7). In Western Europe and Japan, over 80% of ethylene
is produced from naphthas – the principal ethylene raw materials.
A shift in feedstocks occurred for the period from 1980 to 1991. In the
United States and Europe larger amounts of light feedstocks (LPG: propane þ
butane) and NGL (ethane, propane, butane) are used for ethylene production,
whereas in Japan more naphtha was used in 1991 compared to 1981. The use
of gas oils for ethylene production decreased slightly during the 1980s.
Ethane [74-84-0] is obtained from wet natural gases and refinery waste
gases. It may be cracked alone or as a mixture with propane. Propane [74-98-6]
is obtained from wet natural gases, natural gasolines, and refinery waste gases.
Butanes are obtained from natural gasolines and refinery waste gases. A mixture
of light hydrocarbons such as propane, isobutane [75-28-5], and n-butane [106-97-81],
commonly called liquefied petroleum gas (LPG) and obtained from natural gasolines
and refinery gases, is also used as a feedstock.
Naphthas, which are the most important feedstocks for ethylene pro-
duction, are mixtures of hydrocarbons in the boiling range of 30 –2008C.
Processing of light naphthas (boiling range 30 –908C, full range naphthas
(30 –2008C) and special cuts (C6 – C8 raffinates) as feedstocks is typical for
naphtha crackers.
A natural-cut full-range naphtha contains more than 100 individual compo-
nents, which can be detected individually by gas chromatography (GC). Depend-
ing on the origin naphtha quality can vary over a wide range, which necessitates
quality control of the complex feed mixtures. Characterization is typically based
on boiling range; density; and content of paraffins (n-alkanes), isoalkanes, ole-
fins, naphthenes, and aromatics (PIONA analysis) by carbon number. This char-
acterization can be carried out by GC analysis or by a newly developed infrared
method (8). Full characterization of feedstocks is even more important when pro-
duction is based on varying feedstocks, eg, feedstocks of different origins pur-
chased on spot markets.
The quality of a feedstock is depending on the potential to produce the
target products (ethylene and propylene). Simple yield correlations for these
products can be used to express the quality of a feedstock in a simple figure,

Table 1. Raw Materials for Ethylene Production (as a Percentage of Total Ethylene
produced)
USA W. Europe Japan World
Raw
materials 1979 1991 1981 1991 1981 1991 1981 1991
Refinery 1 3 2 17
gas
LPG, NGL 65 73 4* 14 10* 2* 31* 27
Naphtha 14 18 80 72 90 98 58 48
Gas oil 20 6 16 12 0 0 11 8
*Including refinery gas.
730 ETHYLENE Vol. 2

the quality factor, which indicates wether yields of the target products are high
or low, with aromatic feedstocks being poor and saturated feedstocks being good
feedstocks.
Quality characterization factors for naphthas have been developed, which
indicate the aromatics content by empirical correlation. Since aromatics contri-
bute little to ethylene yields in naphtha cracking, a rough quality estimate can be
made for naphthas with a typical weight ratio of n- to isoparaffins of 1–1.1. The
K factor is defined as (9):

ð1:8Tk Þ1=3

d

where Tk is the molal average boiling point in K. Naphthas with a K factor of 12


or higher are considered saturated; those below 12 are considered naphthenic or
aromatic. The K factor does not differentiate between iso- and n-alkanes. The
U.S. Bureau of Mines Correlation Index (BMCI) (10) can also be used as a
rough quality measure of naphthas:

BMCI ¼ 48640=T þ 473:7d  456:8

where T is the molal average boiling point in K and d is the relative density
d15:6
15:6 . A high value of BMCI indicates a highly aromatic naphtha; a low value,
a highly saturated naphtha. Gas oils are feedstocks that are gaining import-
ance in several areas of the world. Gas oils used for ethylene production are
crude oil fractions in the boiling range of 180 –3508C (atmospheric gas oils,
AGO) and 350– 6008C (vacuum gas oils, VGO). In contrast to naphtha and
lighter gas feeds, these feedstocks can not be characterized by individual
components.
Gas chromatography coupled with mass spectrometry (GC – MS) or high
performance liquid chromatography (HPLC) allow the analysis of structural
groups, ie, the percentage of paraffins, naphthenes, olefins, monoaromatics,
and polyaromatics in the gas oil, and can be used to determine the quality of
the hydrocarbon fraction. If this information is used together with data such
as hydrogen content, boiling range, refractive index, etc., the quality can be
determined quite accurately. A rough estimate of feed quality can be made by
using the BMCI or the calculated cetane number of a gas oil. The cetane number,
normally used to calculate the performance of diesel fuels, is an excellent quality
measure, since it is very sensitive to the n-paraffin content, which is one of the
key parameters for the ethylene yield. The cetane number CN is calculated as
follows (11):

CN ¼ 12:822 þ 0:1164 CI þ 0:012976 CI2

where CI ¼ 0.9187 (T50/10)1.26687 (n20


D =100)
1.44227
, where T50 is the volume
20
average boiling point in8C and nD the refractive index at 208C
Vol. 2 ETHYLENE 731

5. Production

5.1. Ethylene from Pyrolysis of Hydrocarbons. The bulk of the


worldwide annual commercial production of ethylene is based on thermal
cracking of petroleum hydrocarbons with steam; the process is commonly called
pyrolysis or steam cracking. The principal arrangement of such a cracking reac-
tor is shown in Figure 1.
A hydrocarbon stream is heated by heat exchange against flue gas in the
convection section, mixed with steam, and further heated to incipient cracking
temperature (500–6808C, depending on the feedstock). The stream then enters
a fired tubular reactor (radiant tube or radiant coil) where, under controlled resi-
dence time, temperature profile, and partial pressure, it is heated from 500–650
to 750–8758C for 0.1–0.5 s. During this short reaction time hydrocarbons in the
feedstock are cracked into smaller molecules; ethylene, other olefins, and diole-
fins are the major products. Since the conversion of saturated hydrocarbons to
olefins in the radiant tube is highly endothermic, high energy input rates are
needed. The reaction products leaving the radiant tube at 800–8508C are cooled
to 550–6508C within 0.02–0.1 s to prevent degradation of the highly reactive pro-
ducts by secondary reactions.
The resulting product mixtures, which can vary widely, depending on feed-
stock and severity of the cracking operation, are then separated into the desired

Fig. 1. Principal arrangement of a cracking furnace.


732 ETHYLENE Vol. 2

products by using a complex sequence of separation and chemical-treatment


steps.
The cooling of the cracked gas in the transfer-line exchanger is carried out
by vaporization of high-pressure boiler feed water (BFW, p ¼ 6–12 MPa), which
is separated in the steam drum and subsequently superheated in the convection
section to high-pressure superheated steam (HPSS, 6–12 MPa).
Cracking Conditions. Commercial pyrolysis of hydrocarbons to ethylene
is performed almost exclusively in fired tubular reactors, as shown schemati-
cally in Figure 1. These furnaces can be used for all feedstocks from ethane
to gas oil, with a limitation in the end point of the feedstock of 6008C. Higher
boiling materials can not be vaporized under the operating condition of a
cracking furnace.
Increasing availability of heavy gas oil fractions, due to a shift in demand to
lighter fractions, offers cost advantages for processing heavy feedstocks in some
areas of the world. Furthermore, the availability of large quantities of residual
oil have led some companies to investigate crude oil and residual oils as ethylene
sources. Such feedstocks cannot be cracked in conventional tubular reactors.
Various techniques employing fluidized beds, molten salts, recuperators, and
high-temperature steam have been investigated, but none of these have attained
commercial significance (12).
Pyrolysis of hydrocarbons has been studied for years. Much effort has been
devoted to mathematical models of pyrolysis reactions for use in designing fur-
naces and predicting the products obtained from various feedstocks under differ-
ent furnace conditions. Three major types of model are used: empirical or
regression, molecular, and mechanistic models (13).
Today, mechanistic computer models, which are available from various
companies, are used for design, optimization and operation of modern olefin
plants. Sophisticated regression models are also used, mainly by operators,
and offer the advantage of a much lower computer performance requirements
than mechanistic models.
The regression models are based on a data set, which can consist of histor-
ical data or calculated data. Depending on the quality of the data base the
empirical regression models can be of sufficient accuracy for most operating pro-
blems, within the range of the data field. These models can be run on small com-
puters and are well suited for process computer control and optimization.
Molecular kinetic models that use only apparent global molecular reactions
and thus describe the main products as a function of feedstock consumption have
been applied with some success to the pyrolysis of simple compounds such as
ethane, propane, and butanes.
For example, cracking of propane can be described as

C3 H8 ! a H2 þ b CH4 þ c C2 H4 þ d C3 H6 þ e C4 H8 þ f C5þ

where a, b, c, d, e, f are empirical factors depending on the conversion of


propane.
Gross oversimplification is required if these models are applied to complex
mixtures such as naphthas or gas oils, but some success has been attained even
with these materials.
Vol. 2 ETHYLENE 733

Advances have been made in mechanistic modeling of pyrolysis, facilitated


by the availability of more accurate thermochemical kinetic and pyrolysis data
and of high-speed computers. The major breakthrough in this area, however,
has been the development of methods to integrate large systems of differential
equations (14–16).
Mechanistic models need less experimental data and can be extrapolated.
The accuracy of these models is very good for most components, but they require
permanent tuning of the kinetic parameters, especially for computing the
cracked-gas composition for ultrashort residence times. The main application
for mechanistic models is the design of cracking furnaces and complete ethylene
plants. The accuracy of the models has been improved, driven by the competition
between the contractors for ethylene plants. A number of mechanistic models are
used today in the ethylene industry, describing the very complex kinetics with
hundreds of kinetic equations (17–19).
To demonstrate the complexity of the chemical reactions, the cracking of
ethane to ethylene is discussed here in detail. A simple reaction equation for
ethane cracking is:

C2 H6 ! C2 H4 þ H2 (1)

If this were the only reaction, the product at 100% conversion would consist
solely of ethylene and hydrogen; at lower conversion, ethylene, hydrogen and
ethane would be present. In fact, the cracked gas also contains methane, acety-
lene, propene, propane, butanes, butenes, benzene, toluene, and heavier compo-
nents. This reaction (Eq. 1) is clearly not the only reaction occurring.
In the 1930s, the free-radical mechanism for the decomposition of hydrocar-
bons was established (20). Although the free-radical treatment does not explain
the complete product distribution, even for a compound as simple as ethane, it
has been extremely useful. Ethane cracking represents the simplest application
of the free-radical mechanism. Ethane is split into two methyl radicals in the
chain initiation step (Eq. 2). The methyl radical reacts with an ethane molecule
to produce an ethyl radical (Eq. 3), which decomposes to ethylene and a hydrogen
atom (Eq. 4). The hydrogen atom reacts with another ethane molecule to give a
molecule of hydrogen and a new ethyl radical (Eq. 5).

Initiation

C2 H6 ! CH 3 þ CH 3 (2)

Propagation

CH 3 þ C2 H6 ! CH4 þ C2 H 5 (3)

C2 H 5 ! C2 H4 þ H (4)

H þ C2 H6 ! H2 þ C2 H 5 (5)

If reactions (4) and (5) proceed uninterrupted, the molecular reaction in


Equation (1) results. If only reactions (3)–(5) occurred, the cracked gas would
734 ETHYLENE Vol. 2

contain traces of methane (Eq. 3) and equimolar quantities of ethylene and


hydrogen with unreacted ethane. This is not observed.
Reactions (3) and (4) terminate if either an ethyl radical or a hydrogen atom
reacts with another radical or atom by reactions such as:
Termination

H þ H ! H2 (6)

CH 3 þ H ! CH4 (7)

H þ C2 H 5 ! C2 H6 (8)

C2 H 5 þ CH 3 ! C3 H8 (9)

C2 H 5 þ C2 H 5 ! C4 H10 (10)

On termination of chain propagation, new methyl or ethyl radicals or a new


hydrogen atom must be generated (Eqs. 2–4) to start a new chain. Thus, every
time a new chain is initiated, a molecule of methane is formed (Eq. 3) and a mole-
cule of ethylene is produced (Eq. 4). Other normal and branched-chain alkanes
decompose by a similar, but more complex, free-radical mechanism (21). The
number of possible free radicals and reactions increases rapidly as chain length
increases.
The free-radical mechanism is generally accepted to explain hydrocarbon
pyrolysis at low conversion (20). As conversion and concentrations of olefins
and other products increase, secondary reactions become more significant.
Partial pressures of olefins and diolefins increase, favoring condensation reac-
tions to produce cyclodiolefins and aromatics. The cracking of heavy feed, such
as naphthas or gas oils, often proceeds far enough to exhaust most of the crack-
able material in the feedstock.
The reaction scheme with heavier feeds is much more complex than with
gaseous feedstocks, due to the fact the hundreds of reactants (feed components)
react in parallel and some of those components are formed as reaction products
during the reaction. Since the radicals involved are relatively short lived, their
concentration in the reaction products is rather low.
Radical decomposition is one of the most important types of reaction and it
directly produces ethylene according to the following scheme:
Radical decomposition

RCH2 CH2 CH 2 ! RCH 2 þ C2 H4 (11)

This b-scission reaction produces a shorter radical (RCH 2 ) and ethylene. Radi-
cals normally decompose in the b-position, where the C C bond is weaker due
to electronic effects. Large radicals are more stable than smaller ones and can
therefore undergo isomerization.

Radical isomerization

RCH2 CH2 CH 2 ! RCH2 CHCH


_ 3 (12)
Vol. 2 ETHYLENE 735

The free-radical decomposition of n-butane (Eqs. 12–14) results in the molecular


equation (Eq. 15):

n-C4 H10 þ H ! n-C4 H 9 þ H2 (14)

n-C4 H 9 ! C2 H4 þ C2 H 5 (15)

C2 H 5 ! C2 H4 þ H (16)

n-C4 H10 ! 2 C2 H4 þ H2 (17)

Reactions like (1) and (15) are highly endothermic. Reported values of DH at
8278C are þ144.53 kJ/mol for Equation (1) and þ232.244 kJ/mol for Equation (15).
The mathematical description of these complex systems requires special
integration algorithms (22). Based on the pseudo steady state approximation,
the chemical reactions can be integrated and the concentration of all components
at each location of the reactor (cracking coil) can be computed (23,24).
In a generalized and very simplified form the complex kinetics of cracking of
hydrocarbons (ethane to gas oil) in steam crackers can be summarized as follows:

Primary reactions Secondary reactions


Feedstock/steam ! ethylene ! C4 products
propylene C5 products
acetylene C6 products
hydrogen aromatics
methane C7 products
etc. heavier products

The fundamentals of furnace design and the main influences of the different
parameters can be understood even with this simplified mechanism:

– Residence time: From the above scheme it is clear that a long residence time
favors the secondary reactions, whereas a short residence time increases
the yields of the primary products, such as ethylene and propylene.
– Partial pressure: Since most of the secondary products result from reactions
in which the number of molecules decreases, increasing the pressure favors
the secondary products. One function of the steam present in the system is
to reduce the hydrocarbon partial pressure and thus favor the formation of
primary products.
– Temperature and temperature profiles: The oligomerization reactions
involved in the formation of secondary products are favored by lower
temperatures; therefore, special temperature profiles are applied along the
cracking coil to avoid long residence times at low temperatures.

Most commercial pyrolysis to produce ethylene is carried out in fired tubu-


lar reactors in which the temperature of the reactant increases continuously
from the inlet to the outlet. Typical inlet temperatures are 500–6808C, depend-
ing on the material being processed. Typical outlet temperatures are 775–8758C.
Modern cracking furnaces are designed for rapid heating at the radiant coil
inlet, where reaction rate constants are low because of the low temperature. Most
736 ETHYLENE Vol. 2

Fig. 2. Process gas temperatures along radiant coils.

of the heat transferred simply raises the reactant from the inlet temperature to
the necessary reaction temperature. In the middle of the coil, the rate of tem-
perature rise is lower, but cracking rates are appreciable. In this section, the
endothermic reaction absorbs most of the heat transferred to the mixture. At
the coil outlet, the rate of temperature rise again increases but never becomes
as rapid as at the inlet.
The designers of cracking coils try to optimize the temperature and pres-
sure profiles along the radiant coils to maximize the yield of valuable products
yields by special coil design that allows rapid temperature increase in the inlet
section and low pressure drops in the outlet section of the cracking coils.
Typical process gas temperature profiles along the radiant coil of modern
ethylene furnaces are shown in Figure 2 for ethane, propane, butane, and
naphtha cracking.
The quantity of steam used, generally expressed as steam ratio (kilograms
of steam per kilogram of hydrocarbon), varies with feedstock, cracking severity,
and design of the cracking coil. Typical steam ratios used at a coil outlet pressure
of 165–225 kPa (1.65–2.25 bar) for various feedstocks are:
Ethane 0.25–0.35
Propane 0.30–0.40
Naphthas 0.4–0.50
Atmospheric gas oils (cut: 180–3508C) 0.60–0.70
Hydrocracker bottoms (cut: 350–6008C) 0.70–0.85

Steam dilution lowers the hydrocarbon partial pressure, thereby enhancing


olefin yield. It also reduces the partial pressure of high-boiling, high-molecular-
mass aromatics and heavy tarry materials, reducing their tendency to deposit
and form coke in the radiant tubes and to foul quench-exchanger tubes. Steam
reduces the fouling of the radiant tubes by reacting with deposited coke to
form carbon monoxide, carbon dioxide, and hydrogen in a water gas reaction,
but as can be seen from the concentration of carbon monoxide in the cracked
gas (normally ca. 100 ppm), only to a limited extent.
Vol. 2 ETHYLENE 737

The sulfur content of the feed is important, since sulfur passivates active Ni
sites of the cracking coil material by forming nickel sulfides, which do not cata-
lyze coke gasification, in contrast to nickel metal and nickel oxides. This explains
why CO production is high (up to 1% in the cracked gas) if the feedstock is free of
sulfur. To prevent this effect, ca. 20 ppm of sulfur (eg, as dimethyl sulfide) are
added to sulfur-free feedstocks.
Feedstock composition is another important consideration in commercial
production of ethylene. Ethylene-plant feedstocks generally contain straight-
and branched-chain alkanes, olefins, naphthenes, and aromatics. Ethylene and
other olefins are formed primarily from alkanes and naphthenes in the feed.
n-Alkanes are the preferred component for high ethylene yields, and those con-
taining an even number of carbon atoms give slightly better ethylene yields than
odd-numbered ones. n-Alkanes also produce propene, whose yield decreases with
increasing chain length.
Isoalkanes, in general, produce much smaller yields of ethylene and pro-
pene than n-alkanes. They give higher yields of hydrogen and methane, and of
C4 and higher olefins. Isoalkanes tend to produce more propene than ethylene
compared to n-alkanes.
Olefin and diolefin yields from cyclopentane, methylcyclopentane, and
cyclohexane have been reported (21); methylcyclopentane and cyclohexane give
substantial amounts of butadiene.
Simple and condensed ring aromatics produce no ethylene. Benzene
is stable under normal cracking conditions. It is formed during cracking
and remains unchanged in the feed. Other aromatics yield primarily higher
molecular mass components. Aromatics with long side chains, such as
dodecylbenzene, yield ethylene and other olefins by side-chain cracking.
The aromatic nucleus still, however, generally produces pyrolysis fuel oil
and tar.
In summary, maximum ethylene production requires:

– A highly saturated feedstock


– High coil outlet temperature
– Low hydrocarbon partial pressure
– Short residence time in the radiant coil
– Rapid quenching of the cracked gases

These conditions maximize the yield of olefins and minimize the yield of
methane and high molecular mass aromatic components.
Cracking conditions and yield structure are often optimized for economic
reasons in plant design and operation; even small changes in cracked gas compo-
sition influence the overall plant economics dramatically.
The dilution steam ratio is in most cases at the minimum of the range for
the specified feedstock, due to the contribution of the dilution steam to the pro-
duction costs.
Heat Requirements for Hydrocarbon Pyrolysis. Primary hydrocarbon
pyrolysis reactions are endothermic; heat requirements are determined by feed-
stock and cracking conditions. The temperature required to attain a given reac-
tion rate decreases as the carbon chain length in the feedstock increases. Thus,
738 ETHYLENE Vol. 2

the enthalpy required to reach cracking temperature depends on feedstock and


cracking conditions.
Heat requirements for hydrocarbon pyrolysis can be divided into three
components:

1. The enthalpy required to heat the feedstock, including the latent heat of
vaporization of liquids
2. The endothermic heat of cracking
3. The enthalpy required to heat the cracked gas from the radiant-coil inlet
temperature to the radiant-coil outlet temperature

The first is accomplished in the convection section of the furnace; the last two, in
the radiant section. Enthalpies (1) and (3) can be calculated from standard heat-
capacity data; they represent 25–60% of total process heat input in the furnace
as the feedstock changes from ethane to gas oils (21).
A typical heat balance for a modern naphtha furnace including the TLE is
shown in Figure 3.

Fig. 3. Heat balance of a naphtha cracking furnace.


Vol. 2 ETHYLENE 739

The endothermic heat of cracking is most simply calculated from heats of


formation by the following equation:

DHr ¼ DHp  DHf


where DHr ¼ endothermic heat of cracking, kJ/kg; DHp ¼ heat of formation of the
cracked products, kJ/kg; and DHf ¼ heat of formation of the feed, kJ/kg. Heats of
formation for some alkanes, olefins, naphthenes, and aromatics are as follows
(in kJ/kg) (25):

Hydrogen 0.0
Methane 5646.3
Ethane 3548.1
Ethylene 1345.3
Acetylene 8540.7
Propane 2949.9
Propene 17.9
Methylacetylene 4317.1
Propadiene 4496.8
n-Butane 2694.4
Isobutane 2827.5
1-Butene 459.4
cis-2-Butene 649.5
trans-2-Butene 668.2
Isobutylene 746.5
Ethylacetylene 2720.4
1,3-Butadiene 1721.8
Benzene 783.6
Toluene 243.4
Cyclopentane 1611.3
Methylcyclopentane 1722.6
Cyclohexane 1846.4
Methylcyclohexane 1890.3

As the alkane chain length increases, the heat of formation approaches


1800 kJ/kg asymptotically. Ethane, the most refractory alkane besides methane,
has the most endothermic heat of cracking, þ4893 kJ/kg. The corresponding
figure for propane is þ4295 kJ/kg and for a high-molecular-mass n-alkane,
ca. 1364 kJ/kg.
Endothermic heat of cracking increases with degree of saturation of the
feed; as conversion increases, secondary condensation products increase. These
products have less negative heats of formation than alkanes, or positive heats
of formation.
Commercial Cracking Yields. Determining the complete yield patterns
(gas and liquid products) for a commercial cracking furnace requires a standar-
dized procedure, especially if the feedstock is changed from ethane to naphtha
and gas oil. A homogeneous sample is withdrawn from the furnace effluent,
quenched, and separated into gaseous and liquid fractions. These are analyzed
separately and then combined to give a complete cracked-gas analysis.
The material balance is obtained by matching the hydrogen, carbon, and
sulfur content of the feedstock with that of the cracked gas. This type of analysis
is time-consuming. In a faster system, gas – liquid separation is eliminated (26)
740 ETHYLENE Vol. 2

or only a gas-phase analysis is taken. Due to many sources of errors, series of


samples have to be taken and analyzed. Statistical methods are used to check
the quality of the results.
For gas feedstocks, conversion is used as a measure of the severity of cracking:

Conversion ¼ ðCn;in  Cn;out =Cn;in Þ

where Cn,in ¼ quantity of component n at inlet in kg/h and Cn,out ¼ quantity of


component n at outlet in kg/h. For liquid feedstocks the severity is defined as the
ratio of propylene to ethylene (on weight basis) P/E or as ratio of methane to
propylene M/P. Both ratios P/E and M/P are a linear function of temperature
over a wide range and can be measured with much higher accuracy than the
apparent bulk coil outlet temperature of a cracking furnace. The use of P/E and
M/P ratios for severity definition is accepted in industry.
Ethane. Ethane is cracked commercially in all types of furnaces, from
long residence time to short residence time. However, since ethane is a very
stable paraffin, cracking temperatures at the coil inlet typically are higher
than for other hydrocarbons. For ethane the coil inlet temperatures are in the
range of 650–6808C. Typical conversion ranges of commercial furnaces are 60–
70%, with 67% conversion being a typical figure for modern designs.
Table 2 lists yields for ethane cracking at various residence times. There is a
slight increase in yield at short residence times. However, as the run length of

Table 2. Yields from Ethane Cracking with Various Residence Times


Conversion, kg/kg 65.01 64.97 65.01 65.01
Steam dilution, kg/kg 0.3 0.3 0.3 0.3

Residence time, s 0.4607 0.3451 0.186 0.1133

H2 wt% 4.04 4.05 4.09 4.12


CO wt% 0.04 0.04 0.03 0.03
CO2 wt% 0.01 0.01 0.01 0.01
H2S wt% 0.01 0.01 0.01 0.01
CH4 wt% 3.75 3.52 3.19 2.84
C2H2 wt% 0.44 0.47 0.54 0.75
C2H4 wt% 51.88 52.31 52.85 53.43
C2H6 wt% 34.99 35.03 34.99 34.99
C3H4 wt% 0.02 0.02 0.02 0.02
C3H6 wt% 1.22 1.13 1.06 0.97
C3H8 wt% 0.12 0.12 0.12 0.13
C4H4 wt% 0.05 0.05 0.05 0.06
C4H6 wt% 1.80 1.80 1.79 1.65
C4H8 wt% 0.19 0.19 0.18 0.16
C4H10 wt% 0.21 0.21 0.21 0.22
Benzene wt% 0.55 0.47 0.38 0.26
Toluene wt% 0.08 0.07 0.06 0.04
Xylenes wt% 0.00 0.00 0.00 0.00
Ethylbenzene wt% 0.01 0.00 0.00 0.00
Styrene wt% 0.03 0.02 0.02 0.01
Pyrolysis gasoline wt% 0.35 0.32 0.29 0.24
Pyrolysis fuel oil wt% 0.21 0.16 0.11 0.06
Sum wt% 100.00 100.00 100.00 100.00
Vol. 2 ETHYLENE 741

Fig. 4. Run length versus residence time for ethane cracking.

short-residence-time ethane furnaces is very low (Fig. 4) and yield improvement


is moderate, application of short-residence-time furnaces for ethane cracking is
not feasible.
In naphtha plants the ethane contained in the cracked gas is separated and
recycled to a segregated ethane-cracking furnace, in which cracking is carried
out under typical ethane-cracking conditions. In case the capacity of the ethane
furnace is not sufficient to crack all the ethane, cocracking of ethane with
naphtha is performed, with ethane conversions of 40–50%. Due to the build up
of large recycle streams in such operations, cocracking should be avoided wher-
ever possible.
Propane is cracked in all types of furnaces at typical conversions of
90–93%. The application of short-residence-time furnaces for propane cracking
has to be evaluated carefully, due to the dramatic reduction in run length
between successive decokings. In many naphtha plants with propane recycle,
propane is co-cracked together with naphtha, without a dramatic drop in pro-
pane conversion. Typical yields for propane cracking for various residence
times are listed in Table 3. For very mild propane cracking conditions (70%
conversion) yields of propylene show a maximum at 18 – 19 wt % based on
propane feed.
Butane. All types of furnaces can be used for butane cracking, which
typically is performed at 94– 96% conversion. Yields for n-butane and iso-
butane differ widely. For n-butane, ethylene yields in the range of 40% are
typical (Table 4), whereas for isobutane only 15 wt % ethylene yield is
achieved. Due to the poor ethylene yield from isobutane, other applications
such as dehydrogenation, oligomerization, and alkylation seem to be more fea-
sible than cracking isobutane to produce ethylene. However, mild cracking at
70% conversion and high pressures has been claimed to yield isobutene as
primary product (27).
Naphtha. Naphtha, the refinery hydrocarbon fraction with the boil-
ing range of 35– 1808C, can vary in composition and boiling range, depend-
ing on source and refinery conditions. Today naphtha cuts from 35 to 908C
(light naphtha), 90–1808C (heavy naphtha) and 35– 1808C (full-range
naphtha) are processed. In Europe and Asia/Pacific naphtha is the primary
raw marerial for ethylene cracking, with full range naphthas being processed
most.
Table 3. Yields from Propane Cracking with Various Residence Times
Conversion, kg/kg 90.020 90.035 89.926 89.983
Steam dilution, kg/kg 0.3 0.3 0.3 0.3
Residence time, s 0.4450 0.3337 0.1761 0.1099
H2 wt% 1.51 1.55 1.61 1.68
CO wt% 0.04 0.04 0.03 0.04
CO2 wt% 0.01 0.01 0.01 0.01
H2S wt% 0.01 0.01 0.01 0.01
CH4 wt% 23.43 23.27 22.82 22.40
C2H2 wt% 0.46 0.51 0.59 0.82
C2H4 wt% 37.15 37.51 38.05 38.59
C2H6 wt% 3.06 2.80 2.37 1.96
C3H4 wt% 0.52 0.57 0.65 0.89
C3H6 wt% 14.81 14.82 15.01 15.27
C3H8 wt% 9.97 9.96 10.07 10.01
C4H4 wt% 0.08 0.08 0.09 0.11
C4H6 wt% 2.85 2.9 2.98 2.99
C4H8 wt% 1.00 1 1.02 1.09
C4H10 wt% 0.04 0.04 0.05 0.05
Benzene wt% 2.15 2.12 2.02 1.80
Toluene wt% 0.43 0.4 0.36 0.28
Xylenes wt% 0.05 0.05 0.04 0.03
Ethylbenzene wt% 0.01 0.01 0.01 0.00
Styrene wt% 0.21 0.2 0.18 0.15
Pyrolysis gasoline wt% 1.27 1.26 1.27 1.24
Pyrolysis fuel oil wt% 0.94 0.89 0.76 0.58
Sum wt% 100.00 100.00 100.00 100.00

Table 4. Yields from Butane Cracking with Various Residence Times


Conversion, kg/kg 93.95 93.43 93.44 93.92
Steam dilution, kg/kg 0.35 0.35 0.35 0.35
Residence time, s 0.4775 0.3440 0.1781 0.1117
H2 wt% 1.05 1.09 1.16 1.22
CO wt% 0.04 0.04 0.03 0.04
CO2 wt% 0.01 0.01 0.01 0.01
H2S wt% 0.01 0.01 0.01 0.01
CH4 wt% 20.48 20.29 19.85 19.33
C2H2 wt% 0.37 0.42 0.51 0.71
C2H4 wt% 35.03 35.81 37.25 38.36
C2H6 wt% 4.57 4.16 3.47 3.02
C3H4 wt% 0.87 0.93 1.04 1.32
C3H6 wt% 17.37 17.24 17.03 16.99
C3H8 wt% 0.37 0.35 0.33 0.33
C4H4 wt% 0.09 0.1 0.11 0.13
C4H6 wt% 3.98 4.08 4.27 4.38
C4H8 wt% 2.85 2.83 2.78 2.85
C4H10 wt% 6.05 6.07 6.06 6.08
Benzene wt% 2.78 2.69 2.52 2.18
Toluene wt% 0.83 0.77 0.66 0.49
Xylenes wt% 0.12 0.11 0.09 0.06
Ethylbenzene wt% 0.01 0.01 0.01 0.01
Styrene wt% 0.25 0.24 0.22 0.18
Pyrolysis gasoline wt% 1.85 1.81 1.80 1.72
Pyrolysis fuel oil wt% 1.02 0.94 0.79 0.58
Sum wt% 100.00 100.00 100.00 100.00

742
Vol. 2 ETHYLENE 743

Table 5. Yields from Low-severity Naphtha Cracking


P/E *, kg/kg 0.65 0.65 0.65 0.65
Steam dilution, kg/kg 0.4 0.4 0.4 0.4
Residence time, s 0.4836 0.3526 0.1784 0.1096
H2 wt% 0.82 0.83 0.85 0.86
CO wt% 0.03 0.02 0.02 0.02
CO2 wt% 0.00 0.00 0.00 0.00
H2S wt% 0.00 0.00 0.00 0.00
CH4 wt% 13.56 13.2 12.44 11.83
C2H2 wt% 0.32 0.34 0.36 0.45
C2H4 wt% 25.23 25.6 26.08 26.43
C2H6 wt% 4.76 4.36 3.62 3.16
C3H4 wt% 0.59 0.63 0.70 0.92
C3H6 wt% 16.42 16.63 16.97 17.18
C3H8 wt% 0.68 0.65 0.61 0.60
C4H4 wt% 0.09 0.09 0.10 0.12
C4H6 wt% 4.63 4.86 5.27 5.68
C4H8 wt% 5.76 5.94 6.34 6.69
C4H10 wt% 0.72 0.75 0.80 0.85
Benzene wt% 6.37 5.98 5.18 4.47
Toluene wt% 3.08 2.9 2.60 2.26
Xylenes wt% 1.17 1.15 1.09 0.96
Ethylbenzene wt% 0.79 0.8 0.83 0.85
Styrene wt% 1.08 1.01 0.87 0.77
Pyrolysis gasoline wt% 11.25 11.81 13.19 14.09
Pyrolysis fuel oil wt% 2.65 2.45 2.08 1.81
Sum wt% 100.00 100.00 100.00 100.00
* Propylene to ethylene ratio, a measure of severity.

The properties of the naphtha feed stock used to generate the yield patterns
given in Tables 5, 6, and 7 are as follows:

Density at 208C 0.692 g/mL


S content 55 mg/kg
H content 15.17 wt%
C content 84.80 wt%
Average molar mass 92 g/mol
Boiling curve (ASTM D 86)
initial boiling point 268C
5 vol% 498C
10 vol% 538C
20 vol% 588C
30 vol% 648C
40 vol% 708C
50 vol% 778C
60 vol% 868C
70 vol% 998C
80 vol% 1168C
90 vol% 1388C
95 vol% 1528C
final boiling point 1838C
PIONA analysis
n-Paraffins 36.13 wt%
Table 6. Yields from Medium-severity Naphtha Cracking
P/E *, kg/kg 0.55 0.55 0.55 0.55
Steam dilution, kg/kg 0.45 0.45 0.45 0.45
Residence time, s 0.4840 0.3572 0.1828 0.1132
H2 wt% 0.90 0.92 0.94 0.96
CO wt% 0.04 0.04 0.03 0.04
CO2 wt% 0.00 0.00 0.00 0.00
H2 S wt% 0.00 0.00 0.00 0.00
CH4 wt% 15.23 14.91 14.31 13.82
C2H2 wt% 0.44 0.47 0.53 0.71
C2H4 wt% 27.95 28.45 29.24 29.87
C2H6 wt% 4.59 4.23 3.57 3.14
C3H4 wt% 0.71 0.76 0.87 1.21
C3H6 wt% 15.38 15.64 16.09 16.43
C3H8 wt% 0.53 0.51 0.48 0.47
C4H4 wt% 0.12 0.13 0.16 0.20
C4H6 wt% 4.54 4.79 5.28 5.79
C4H8 wt% 4.41 4.52 4.75 4.95
C4H10 wt% 0.42 0.44 0.47 0.49
Benzene wt% 7.45 7.12 6.5 5.85
Toluene wt% 3.26 3.10 2.82 2.46
Xylenes wt% 1.18 1.16 1.12 0.97
Ethylbenzene wt% 0.62 0.62 0.64 0.63
Styrene wt% 1.21 1.14 1.00 0.88
Pyrolysis gasoline wt% 7.70 7.96 8.52 8.79
Pyrolysis fuel oil wt% 3.32 3.09 2.68 2.34
Sum wt% 100.00 100.00 100.00 100.00
* Propylene to ethylene ratio, a measure of severity.

Table 7. Yields from High-severity Naphtha Cracking


P/E *, kg/kg 0.45 0.45 0.45 0.45
Steam dilution, kg/kg 0.5 0.5 0.5 0.5
Residence time, s 0.4930 0.3640 0.1897 0.1170
H2 wt% 0.9 1.00 1.03 1.06
CO wt% 0.06 0.06 0.05 0.06
CO2 wt% 0.00 0.00 0.00 0.00
H2S wt% 0.00 0.00 0.00 0.00
CH4 wt% 16.9 16.61 16.10 15.74
C2H2 wt% 0.64 0.69 0.80 1.14
C2H4 wt% 30.25 30.81 31.74 32.38
C2H6 wt% 4.32 4.00 3.43 3.00
C3H4 wt% 0.81 0.88 1.02 1.48
C3H6 wt% 13.63 13.88 14.29 14.55
C3H8 wt% 0.37 0.35 0.33 0.32
C4H4 wt% 0.18 0.2 0.24 0.32
C4H6 wt% 4.20 4.45 4.90 5.39
C4H8 wt% 3.11 3.17 3.26 3.28
C4H10 wt% 0.17 0.18 0.19 0.18
Benzene wt% 8.32 8.06 7.63 7.15
Toluene wt% 3.40 3.26 3.00 2.63
Xylenes wt% 1.14 1.12 1.09 0.95
Ethylbenzene wt% 0.44 0.44 0.44 0.40
Styrene wt% 1.44 1.37 1.24 1.12
Pyrolysis gasoline wt% 5.25 5.35 5.55 5.57
Pyrolysis fuel oil wt% 4.38 4.12 3.67 3.28
Sum wt% 100.00 100.00 100.00 100.00
* Propylene to ethylene ratio, a measure of severity.
744
Vol. 2 ETHYLENE 745
Isoparaffins 36.62 wt%
Olefins 0.21 wt%
Napthenes 21.06 wt%
Aromatics 5.98 wt%

The yield increases with decreasing residence time. The maximum benefit
of a selective cracking coil is achieved in high-severity cracking, in which a 5%
feedstock saving can be achieved by using short residence times. However, care-
ful analysis is required since, similar to gas cracking, shorter run lengths are
typical for short-residence-time cracking. Maximum annual furnace productivity
is achieved with a combination of short residence time and long run length.
Gas Oil (AGO; Boiling Range 180–3508C). Compositions and boiling
ranges of gas oils vary greatly with source and refinery conditions. Saturated
gas oils may have BMCI values as low as 20, whereas aromatic gas-oil BMCI
values can exceed 45. Typical yield distributions are shown in Table 8 for an
AGO with properties as listed in Table 9.
Gas oils produce much more C5+ material than the above feedstocks because
they contain large amounts of condensed polynuclear aromatics. These compo-
nents resist cracking and may remain unchanged or condense to higher molecu-
lar mass materials.

Table 8. Yields for AGO and HCR Cracking


AGO HCR
P/E *, kg/kg 0.54 0.53
Steam dilution, kg/kg 0.8 1

Residence time, s 0.32 0.3

H2 wt% 0.71 0.68


CO wt% 0.01 0.01
CO2 wt% 0.01 0.01
H2S wt% 0.00 0.00
CH4 wt% 10.58 11.08
C2H2 wt% 0.38 0.60
C2H4 wt% 25.93 25.98
C2H6 wt% 2.82 2.81
C3H4 wt% 0.63 0.97
C3H6 wt% 14.07 16.02
C3H8 wt% 0.37 0.42
C4H4 wt% 0.12 0.04
C4H6 wt% 5.73 8.03
C4H8 wt% 3.61 4.14
C4H10 wt% 0.05 0.06
Benzene wt% 5.44 5.64
Toluene wt% 3.49 2.73
Xylenes wt% 0.92 0.46
Ethylbenzene wt% 0.37 0.34
Styrene wt% 1.18 1.12
Pyrolysis gasoline wt% 7.28 6.91
Pyrolysis fuel oil wt% 16.30 7.95
Sum wt% 100 100
* Propylene to ethylene ratio, a measure of severity.
746 ETHYLENE Vol. 2

Table 9. Properties of AGO and HCR Used for Yield Calculations


in Table **
AGO HCR
Density at 20 8C, g/mL 0.8174 0.8280
C, wt% 86.20 85.28
H, wt% 13.64 14.34
S, wt% 0.18 0.003
BMCI 24.8 9.37
Paraffins þ naphthalenes, wt% 80.38 89.4
Monoaromatics, wt% 12.45 7.3
Polyaroamtics, wt% 7.17 3.3
Boiling point analysis, 8C
ibp 170 313
10% 203 351
30% 227 393
50% 248 405
70% 273 422
90% 316 451
95% 340 460
fbp 352 475

Hydrocracker Residue (HVGO or HCR; Boiling Range 350–


6008C). Hydrocracker residues are the unconverted product fraction from
hydrocrackers. Due to the severe operating conditions of the hydrocrackers,
HCR is a highly saturated product, with only a limited content of aromatics
and low contents of polyaromatics. HCR or HVGO is a feedstock that can result
in ethylene yields as high as in naphtha cracking. Typical yields for a HCR are
shown in Table 8 for an AGO with properties as listed in Table 9.
A comparison of the feedstock requirements as a function of residence time
for a 500 000 t/a plant for ethane, propane, butane and naphtha is shown in
Figure 5. It can be seen that a naphtha cracker has about three times the feed-
stock throughput of an ethane cracker, and that naphtha cracking shows some
sensitivity to short residence times, whereas ethane crackers are not very sensi-
tive with respect to feedstock savings as short residence times are applied. A
careful economic analysis is required to select the optimum residence time
range for a given feedstock, since shorter residence times lead to reduced avail-
ability due to shorter run length, as shown for ethane cracking in Figure 4.
Commercial Cracking Furnaces. Furnace Design. Modern cracking
furnaces typically have one or two rectangular fireboxes with vertical radiant
coils located centrally between two radiant refractory walls. Firebox heights of
up to 15 m and firebox widths of 2–3 m are standard design practice in the indus-
try. Firing can be performed with wall- or floor-mounted radiant burners, or a
combination of both, which use gaseous or combined gaseous and liquid fuels.
Fireboxes are under slight negative pressures with upward flow of flue gas.
Flue gas flow and draft are established by induced-draft fans. In some furnace
designs, burners are mounted on terraced walls.
Firebox length is determined by the total ethylene production rate desired
from each furnace and the residence time of the cracking operation. This also
determines the number of individual radiant coils needed, with short residence
Vol. 2 ETHYLENE 747

Fig. 5. Feedstock requirements for a 500 000 t/a ethylene plant.

times requiring many more individual coils than longer residence times for the
same production capacity. This is due to the shorter lengths of the short-resi-
dence-time coils, which can be as short as 10–16 m per coil. Long-residence-
time coils can have lengths of 60–100 m per coil. The number of coils required
for a given ethylene capacity is determined by the radiant coil surface, which
is in the range of 10–15 m2 per tonne of feedstock for liquid feedstocks.
Production rate for each coil is determined by its length, diameter, and
charge rate, which translates into a certain heat flux on the radiant coil. Values
of 85 kW/m2 for the average heat flux of a coil should be the maximum.
Radiant coils are usually hung in a single plane down the center of the
firebox. They have also been nested in a single plane or placed parallel in a stag-
gered, double-row tube arrangement. Staggered tubes have nonuniform heat-
flux distributions because of the shadowing effect of one coil on its companion.
If maximum capacity and full utilization of the firebox are desired, tube stagger-
ing can be adjusted so that the shadowing effect is only marginally greater than
with a single-row arrangement.
748 ETHYLENE Vol. 2

Heat transfer to the radiant tubes occurs largely by radiation, with only
a small contribution from convection. Firebox temperature is typically
1000–12008C.
The radiant coil of an ethylene furnace is a fired tubular chemical reactor.
Inlet conditions must meet the required temperature, pressure, and flow rate.
Hydrocarbon feed is heated from its convection-section entry temperature to
the temperature needed for entry into the radiant coils. Dilution steam is intro-
duced into the hydrocarbon stream in the convection section. The combined
hydrocarbon and steam stream is then superheated to the target temperature
at the radiant coil inlet (crossover temperature, XOT). Gaseous feeds require
only the energy necessary to heat the feed and steam to the radiant coil inlet tem-
perature. Liquid feeds require additional energy to heat the liquid to its vapor-
ization temperature plus its latent heat of vaporization.
Energy is saved in the convection section by preheating the cracking
charge; by using utility preheating, ie, boiler-feedwater preheating, dilution-
steam superheating, high-pressure steam superheating or by preheating the
combustion air.
A typical arrangement of furnace elements is shown in Figure 6. In this
case, the convection section contains six separate zones. As an example, the
feed could be preheated in the upper zone, the boiler feedwater in the next,
and mixed feed and dilution steam in the third zone. The process steam is super-
heated in the fourth zone, the HP steam is superheated in the fifth zone and the
final feed and dilution steam are preheated in the bottom zone. The number of
zones and their functions can differ according to furnace design. For example,
the upper zone could be used to recover heat for air preheating.
With the trend to higher capacities per individual furnace a new furnace
set-up has been introduced to the industry with great commercial success. In
this type two fireboxes are connected to one convection section. This design
allows liquid feedstock cracking furnaces with capacities of 130 000–150 000 t/a
to be installed. This twin radiant cell concept has been applied more than 60
industrial furnaces since the late 1980s.
The induced-draft fans allow complex convection sections to be designed
with very high gas velocities to give optimum heat transfer in the convection sec-
tion banks. Modern designs include integrated catalyst systems for NOx removal
by SCR (selective catalytic reduction) technology, whereby the NOx present in
the flue gas reacts with ammonia:

4 NOx þ 4 NH3 þ O2 ! 4 N2 þ 6 H2 O

The temperature of the cracked gas leaving the radiant coils can range from 750
to 9008C. Rapid reduction of gas temperature to 500–6508C, depending on the
feedstock, is necessary to avoid losses of valuable products by secondary reac-
tions. This is accomplished by a transfer-line exchanger (TLE) (see Fig. 6),
which cools the cracked gases and recovers much of the heat contained in the
cracked gas as high-pressure steam.
In all modern furnaces the radiant coils exit the firebox at the top; the
TLEs are top-mounted. In older designs coils also leave the bottom of the firebox,
with TLEs mounted vertically along the side of the furnace or horizontally
Vol. 2 ETHYLENE 749

Fig. 6. Typical arrangement of furnace elements.

underneath it. Radiant-coil designs are generally considered proprietary


information. All modern designs strive for short residence time, high tempera-
ture, and low hydrocarbon partial pressure. Most coil configurations have two
features in common: coils are hung vertically, usually suspended on spring or
constant hangers above the top of the radiant firebox, and they are fired from
both sides of the radiant coil. Horizontal tubes are no longer used in ethylene
furnaces.
Coils can range from a single, small-diameter tube with low feed rate and
many coils per furnace, to long, large-diameter tubes, with high feed rate and few
coils per furnace. Longer coils consist of lengths of tubing connected with return
bends. Individual tubes can be the same diameter or swaged to larger diameters
750 ETHYLENE Vol. 2

Fig. 7. Short-residence-time coil arrangement.

one or more times along the complete coil; two or more individual tubes may be
combined in parallel.
An example of a so-called split radiant coil arrangement is shown in
Figure 7; here parallel small-diameter coils are combined into a larger diameter
outlet coil. This arrangement allows advantageous temperature profiles to be
applied, with a rapid temperature increase in the inlet section, and offers slightly
higher yields than a uniform diameter coil of the same residence time.
Coil lengths and diameters vary widely as can be seen from the following
design characteristics:

Ethylene capacity (per furnace) (50–130)  106 t/a


Number of coils (per furnace) 4–200
Tube d

Potrebbero piacerti anche