June, 2013 Vol.12, No.

Journal of
Pipeline Engineering
incorporating
The Journal of Pipeline Integrity

Great Southern Press Clarion Technical Publishers

 Journal of Pipeline Engineering

Editorial Board - 2013

Dr Husain Al-Muslim, Pipeline Engineer, Consulting Services Department, Saudi Aramco,

Dhahran, Saudi Arabia
Mohd Nazmi Ali Napiah, Pipeline Engineer, Petronas Gas, Segamat, Malaysia
Dr-Ing Michael Beller, Landolt Steuer & Unternehmensberatung AG, Luzern, Switzerland
Jorge Bonnetto, Operations Director TGS (retired), TGS, Buenos Aires, Argentina
Dr Andrew Cosham, Atkins Boreas, Newcastle upon Tyne, UK
Dr Sreekanta Das, Associate Professor, Department of Civil and Environmental Engineering,
University of Windsor, ON, Canada
Leigh Fletcher, Welding and Pipeline Integrity, Bright, Australia
Daniel Hamburger, Pipeline Maintenance Manager, Kinder Morgan, Birmingham, AL, USA
Dr Stijn Hertele, Universiteit Gent Laboratory Soete, Gent, Belgium
Prof. Phil Hopkins, Executive Director, Penspen Ltd, Newcastle upon Tyne, UK
Michael Istre, Chief Engineer, Project Consulting Services,
Houston, TX, USA
Dr Shawn Kenny, Memorial University of Newfoundland Faculty of Engineering and Applied
Science, St Johns, Canada
Dr Gerhard Knauf, Salzgitter Mannesmann Forschung GmbH, Duisburg, Germany
Prof. Andrew Palmer, Dept of Civil Engineering National University of Singapore, Singapore
Prof. Dimitri Pavlou, Professor of Mechanical Engineering,
Technological Institute of Halkida , Halkida, Greece
Dr Julia Race, School of Marine Sciences University of Newcastle,
Newcastle upon Tyne, UK
Dr John Smart, John Smart & Associates, Houston, TX, USA
Jan Spiekhout, Kema Gas Consulting & Services, Groningen, Netherlands
Prof. Sviatoslav Timashev, Russian Academy of Sciences Science
& Engineering Centre, Ekaterinburg, Russia
Patrick Vieth, President, Dynamic Risk, The Woodlands, TX, USA
Dr Joe Zhou, Technology Leader, TransCanada PipeLines Ltd, Calgary, Canada
Dr Xian-Kui Zhu, Senior Research Scientist, Battelle Pipeline Technology Center, Columbus, OH,
USA

2nd Quarter, 2013 69

The Journal of
Pipeline Engineering
incorporating
The Journal of Pipeline Integrity

Volume 12, No 2 Second Quarter, 2013

Contents
by Dr Charles Fernandez, Laurent Bourgouin, Frederic Riegert, and Alain Pecker
The induction of vibrations in transmission pipelines by the fall of a heavy structure nearby:
modelling the safety distances.......................................................................................................................................... 75

by Nima Ghazi and Dr Julia M Race

Techno-economic modelling and analysis of CO2 pipelines.............................................................................................. 83

by Prof. Jos L F Freire, Ronaldo D Vieira, Pablo M Fontes, Adilson C Benjamin, Luis S Murillo C,
and Antonio C Miranda
The critical-path method for assessment of pipelines with metal-loss defects..................................................................... 93

by Dr Andrew Francis
10-6 and all that: what do failure probabilities mean? A response.............................................................................. 109

by Chia Chor Yew and Asle Venas

SliPIPE: a new concept to deal with pipeline expansion................................................................................................ 113

by Dr Yong-Yi Wang, Dr Jing Ma, and Satish S Kulkarni

Assessment of vintage girth welds and challenges to ILI tools.........................................................................................117

by Randy Vaughn
Threat and the probability of failure............................................................................................................................. 127

by Nelson Tonui, Tom DeLong, and Adam Lind

Unpiggable pipelines: Kinder Morgan Canadas experience with various inspection tools and technologies.................. 131

OUR COVER PICTURE, courtesy of A Hak Industrial Services of Tricht,

Netherlands, shows a test loop replicating in full size a typical length of
difficult-to-inspect pipeline that the company has set up in order to test
The Journal of Pipeline Engineering
its tethered inline inspection equipment for a particular project. A recent has been accepted by the Scopus
conference in Houston discussed a range of issues surrounding unpiggable Content Selection & Advisory
Board (CSAB) to be part of the
pipelines, and two papers from this event are published in this issue, starting SciVerse Scopus database and
on pages 125 and 129. index.
70 The Journal of Pipeline Engineering

T HE Journal of Pipeline Engineering (incorporating the Journal of Pipeline Integrity) is an independent, international,
quarterly journal, devoted to the subject of promoting the science of pipeline engineering and maintaining
and improving pipeline integrity for oil, gas, and products pipelines. The editorial content is original papers
on all aspects of the subject. Papers sent to the Journal should not be submitted elsewhere while under
editorial consideration.

Authors wishing to submit papers should do so online at www.j-pipeng.com. The Journal of Pipeline Engineering
now uses the Aires Editorial Manager manuscript management system for accepting and processing manuscripts,
peer-reviewing, and informing authors of comments and manuscript acceptance. Please follow the link shown
on the Journals site to submit your paper into this system: the necessary instructions can be found on the
User Tutorials page where there is an Author's Quick Start Guide. Manuscript files can be uploaded in text
or PDF format, with graphics either embedded or separate.

Please contact the editor (see below) if you require any assistance.

The Journal of Pipeline Engineering aims to publish papers of quality within six months of manuscript acceptance.

Notes
1. Disclaimer: While every effort is made to check
the accuracy of the contributions published in The
Journal of Pipeline Engineering, Great Southern Press
Ltd and Clarion Technical Publishers do not accept
responsibility for the views expressed which, although
made in good faith, are those of the authors alone.
4. Back issues: Single issues from current and past
volumes are available for US$87.50 per copy.
5. Publisher: The Journal of Pipeline Engineering
is published by Great Southern Press Ltd (UK and
Australia) and Clarion Technical Publishers (USA):

2. Copyright and photocopying: 2013 Great Great Southern Press, PO Box 21, Beaconsfield
Southern Press Ltd and Clarion Technical Publishers. HP9 1NS, UK
All rights reserved. No part of this publication may tel: +44 (0)1494 675139
be reproduced, stored or transmitted in any form or fax: +44 (0)1494 670155
by any means without the prior permission in writing email: jtiratsoo@gs-press.com
from the copyright holder. Authorization to photocopy web: www.j-pipe-eng.com
items for internal and personal use is granted by the www.pipelinesinternational.com
copyright holder for libraries and other users registered
with their local reproduction rights organization. This Editor: John Tiratsoo
consent does not extend to other kinds of copying such email: jtiratsoo@gs-press.com
as copying for general distribution, for advertising and
promotional purposes, for creating new collective works, Clarion Technical Publishers, 3401 Louisiana,
or for resale. Special requests should be addressed to Suite 110, Houston TX 77002, USA
Great Southern Press Ltd, PO Box 21, Beaconsfield tel: +1 713 521 5929
HP9 1NS, UK, or to the editor. fax: +1 713 521 9255
web: www.clarion.org
3. Information for subscribers: The Journal of Pipeline
Engineering (incorporating the Journal of Pipeline Associate publisher: BJ Lowe
Integrity) is published four times each year. The email: bjlowe@clarion.org
subscription price for 2013 is US$350 per year (inc.
airmail postage). Members of the Professional Institute 6. ISSN 1753 2116
of Pipeline Engineers can subscribe for the special
rate of US$175/year (inc. airmail postage). Subscribers
receive free on-line access to all issues of the Journal
during the period of their subscription.
v v v

www.j-pipe-eng.com
is available for subscribers
2nd Quarter, 2013
Editorial
Unpiggableor not?
A t the 2013 Unpiggable Pipeline Solutions
Forum, recently held in Houston, semantics had
a surprising relevance as the debate moved towards
dropping the word unpiggable, and replacing it with
challenging, difficult-to-inspect, not-easily-piggable, and
similar epithets. Those who might be considered the
traditionalists won the day, however, and the consensus
view supported continued use of the original adjective.
While this might seem a trivial issue, the fact that
there was any sort of debate about how to describe
such pipelines is significant, as it demonstrates how
far the inspection industry has moved in recent years.
With technologies ranging from internally-mobilized free-
swimming tools, tethered tools, and robotic tools, to a
number of external-inspection solutions, the inspection
industry has at least 16 separate techniques available
to suit both on- and offshore applications, which are
summarized in a recently-published online e-Guide1.
Using one or more of these options means that almost
all pipelines previously considered as unpiggable can
be inspected with a degree of accuracy commensurate
with that for piggable lines.
Most of the difficulties posed by imprecise geometry
have been resolved, or at least are well-known enough
to be soluble, as a consequence of the enormous efforts
that many pipeline operators have made to remove
obstructions, smooth-out awkward bends, and replace
heavy-wall or unsuitably designed valves. Quantifying
the lengths of unpiggable versus piggable pipelines is
currently not possible; however, a recent survey presented
by INGAA (the Interstate Natural Gas Association of
America) shows that the proportion of its members
pipelines that are piggable has increased from 20% in
2002 to 64% in 2012, which gives a clear indication
of how much effort has been expended.
In the recent past, unpiggability was ascribed to
pipelines with geometry issues such as tight bends,
mitred bends, or varying diameters, or pipelines whose
location made them inaccessible, such as those at river
and road crossings. While these types of problem no
longer cause the same levels of concern that they
did only a few years ago, other issues which may
be considered to be more advanced are coming
to the fore, which present a different, non-geometric,
type of challenge. These include flow (too much, or
1 The ultimate guide to unpiggable pipelines is published by Pipelines International,
and is a free download from its site at: http://pipelinesinternational.com/
shop/the_ultimate_guide_to_unpiggable_pipelines/081249/
71
not enough, or even none), the inability to take a
line out of service, no launch or receive facilities,
and lack of knowledge about the lines condition
(not to mention where it is actually located). Quite
often lines that are challenging in this regard are
short and of small diameter maybe infield or
gathering although this does not detract from their
importance and the need to ensure their integrity. In
the past, such lines may have fallen outside the scope
of government regulations, although in a number of
jurisdictions this also is changing.
The dream is for a tool that can enter a previously-
unpigged line and fully inspect it to the same standard
as a piggable line, despite the fact that the line has
not been cleaned and its geometry is unknown. This
is certainly only a pipe-dream at present; however,
as the spotlight of risk-reduction, zero tolerance of
leakage, and maintenance of total integrity, focuses
more directly on unpiggable pipelines, the inspection
industry as represented by the companies reviewed in
the e-Guide will undoubtedly respond and, through
investment and lateral thinking, develop solutions
that really will remove unpiggable from the lexicon.
Summer reading
The Journal has received information on a number of
recently-published books of relevance to the pipeline
industry. Although we have not had the opportunity
of preparing full reviews, we include below outlines
of these books contents and other comments. While
three cover technical and design aspects of pipelines,
two cover the interesting field of law and finances for
cross-border pipelines (of which there is an increasing
number). These latter, mentioned first, are both the
results of work carried out by their authors at the
Centre for Energy, Petroleum, and Mineral Law and
Policy (CEPMLP) at the University of Dundee, an
internationally renowned graduate school in the field
of international business transactions and energy law
and policy.
Cross border pipeline arrangement: what would a single
regulatory framework look like?, by Dr Chowdhury Ishrak
Ahmed Siddiky
This book deals with the problems which occur when
one or more parties in a pipeline do not abide by
the obligations agreed between them at the beginning
of the project. Such problems are most serious when
geopolitical, legal, or economic developments lead
72
governments to intervene, resulting in the breach of
a legitimate expectation of the stakeholders involved.
Using regime theory as an analytical tool, the author
explores participant behaviour in seven specific case
studies that demonstrate different levels of intervention.
In the final analysis, Dr Siddiky proposes the creation
of an autonomous unifying agency with strong regime
credentials. He shows how such a body would reduce
the level of intervention by governments or other
parties in the pipeline regime, without interfering
in the sovereignty of any particular country, and
he outlines the process through which the agency
would use its enforcement capabilities. As the nature
of relations among energy exporting, importing,
and transit countries becomes more critical, this
book provides a useful approach to this very
important subject.
Characterisation and taxation of cross-border pipelines, by
Dr Knut Olsen
A cross-border pipeline is probably one of the
most complex assets to be found, as it can cross
international borders and boundaries, as well as being
laid across transit states, and such a unique asset
creates very intricate tax issues. This study offers
an in-depth discussion of the principal questions:
how is a cross-border pipeline currently characterized
according to the OECD model tax treaty and its
commentaries?; how is income from a cross-border
pipeline allocated between the states involved?; and
how does the OECDs authorized functional separate
entity approach apply to income from pipelines?
Dr Olsen points out that the issues relating to
the characterization and taxation of cross-border
pipelines are uncertain, unresolved, inconsistent, and
unpredictable. No common international approaches
for the characterization and taxation of pipelines
exist. Moreover, the fact that the OECD convention
and its commentaries sometimes suggest different
definitions of an income or asset without giving
clear recommendations does not make the situation
any clearer. This lack of uniformity potentially
creates significant risks for governments and the
petroleum industry.
To help answering these, and other questions, Dr
Olsens book provides a comprehensive analysis of
the issues, and offers solutions to the various tax
issues that a cross-border pipeline might raise. The
book concludes by recommending changes to the
OECD tax treaty and its commentaries to reduce
uncertainty, avoid either double taxation or less than
a single taxation, and establish a more common
international approach to cross-border pipelines and
their allocation.
The Journal of Pipeline Engineering
Hydrocarbon liquid transmission pipeline and storage systems:
design and operation, by Dr Mo Mohitpour, Dr Mike Youn,
and Jim Russell
This book, dedicated to its primary author Dr Mo
Mohitpour who passed away as it was about to
be published in August, 2012, brings together a
broad spectrum of hydraulics, design, and operating
requirements for pipeline transportation and storage of
hydrocarbon liquids. Forming part of an important series
published by ASME since 2000, the book combines
a professional reference with a comprehensive training
tool, and covers the range of subjects required to know
how most reliably to design and operate pipelines
and storage facilities with the least environmental
impact and energy transportation disruption. The
authors discuss aspects of steady-state and dynamic
hydraulics, mechanical design, and operational issues
relating to pipelines transporting liquid hydrocarbons
and petroleum products. In addition, liquid properties
important to the design of long-distance pipelines
requiring multiple pump stations and storage facilities
are also included.
No better summary of the books importance can be
given than by Stephen Wuori, president of liquids
pipelines for Enbridge in Canada, who writes in
his foreword: This book will well serve the need
for a single source of learning for new entrants to
the business and veterans, alike, and I can envision
the book becoming a vital teaching tool in pipeline
company engineering departments.and anywhere
that a deeper understanding of how pipelines really
work is sought.
Piping and pipeline calculations manual, by J Phillip Ellenberger
The authors premise for writing this useful text is
that lack of commentary, or historical perspective,
regarding the code and standard requirements for
piping design and construction is an obstacle to
those who want to provide a safe and economic
piping system. Intended to be an intensive manual,
this book combines wide-ranging text with many
calculations and examples based on the 40+ years
of experience of the author as both an engineer
and instructor. Each example demonstrates how the
relevant code or standard has been correctly and
incorrectly applied.
Designed, in the authors words, as a no-nonsense
guide to the principle intentions of the relevant
code or standard, the book also provides advice on
compliance. It is hoped that the reader should achieve
a clear understanding of how piping systems fail and
what is required to prevent such failures. The focus of
the book is to enhance understanding and application
of the spirit of each code or standard, and form a
2nd Quarter, 2013
plan for compliance; the major codes covered include:
ASME B16.4, B31.3, B31.4, B31.8, and B31.8S; and
API Spec 6D, 526, 527, 594, and 598.
Composite materials in piping applications, by Prof. Dimitrios
G Pavlou
Applying materials science theory and engineering to
an important infrastructure use, this book explains
the design, analysis, and performance of composite
materials in oil, gas, water, and wastewater piping.
The text furnishes design information for pipe
and its supports, damage-analysis, and corrosion
prediction, as well as in-service temperature and
pressure gradients, to allow loading calculations to
be performed. Optimization methods are presented
for cost analysis, and pre- and in-service quality
control and maintenance are discussed. The book
is accompanied by a CD containing algorithms for
pipe design and analysis using Mathematica software,
incorporating equations for calculating joint design,
hanger widths, expansion loops, as well as safe depths
for pipes under highways and railways.
Prof. Pavlou says in his Preface that the high
maintenance costs due to the ageing of steel pipelines,
as well as the reduction in the unit price of composite
materials, have led to a reconsideration of the optimum
material cost for pipeline applications. As he points
out, since fibre-reinforced polymeric (FRP) materials
have a much lower density and much higher strength
than carbon steel, the final cost of such materials
is currently comparable with that of carbon steel.
Moreover, he continues, the lower maintenance
costs of composite pipelines, which results from their
excellent resistance to corrosion and fatigue, leads to
the conclusion that the use of composite materials for
pipeline applications has now become advantageous
when compared to the use of carbon steel.
Such a premise clearly raises many interesting questions.
The Journal is aware of at least one major high-
pressure pipeline operator who is currently examining
the applicability of composite materials for particular
pipeline projects, and it may well be that the industry
is not far from seeing such a line in operation. Prof.
Pavlous comprehensive book, therefore, is timely, and
provides both a comprehensive reference and a useful
introduction to this emerging science.
Special issue on fracture toughness
September
To coincide with the 2013 Pipeline Technology
Conference being held in Ostend, Belgium, in early
October, the Journal is pleased to announce that
the September issue will be a special issue. Guest
edited by Dr Xian-Kui Zhu of the Battelle Memorial
73
Institute, the issue will bring together a number
of significant papers covering the latest thinking
on fracture-toughness testing and evaluation for
pipeline steels.
The contents are currently planned to include the
following original works:
Fracture toughness testing use curved wide plates,
by Prof. Rudi Denys, Lab. Soete, University of
Gent, Gent, Belgium
Evolving CTOA testing to achieve field-
comparable crack propagations, by Elizabeth S
Drexler et al., NIST, Boulder, CO, USA
Drop-weight tear-test application to natural gas
pipeline fracture control, by Robert Eiber, Robert
J Eiber Consultant Inc., Columbus, OH, USA
Charpy and DWTT testing in China, by C Y
Hou, TGRC, China
The Charpy impact test and its applications,
by Dr Brian N Leis, B N Leis, Consultant,
Inc, Worthington, OH, USA
CTOD and pipelines: the past, present, and
future, by Dr Philippa Moore and Dr Henryk
Pisarski, TWI Ltd, Abington, Cambridge, UK
Fracture resistance testing of pipeline girth welds
using bend and tensile fracture specimens, by
Claudio Ruggieri and Leonardo L S Mathias,
Department of Naval Architecture and Ocean
Engineering, University of Sao Paulo, Sao
Paulo, Brazil
Low-constraint toughness testing, by Dr William
R Tyson, Guowu Shen, Dr Dong-Yeob Park, and
James Gianetto, CanmetMATERIALS, Natural
Resources Canada, Ottawa and Hamilton ON,
and Calgary, AB, Canada
Testing for resistance to fast ductile fracture:
measurement of CTOA, by Dr Su Xu, Dr
William R Tyson, and Dr C H M Simha,
CanmetMATERIALS, Natural Resources Canada,
Hamilton and Ottawa, ON, Canada
Review of standard fracture-toughness (K, J)
testing, by Dr Xian-Kui Zhu, Battelle Memorial
Institute, Columbus, OH, USA
Apology
The Journal of Pipeline Engineering sincerely apologizes
to Dr Ted Anderson and its readers for errors that
were published in Dr Andersons paper in the March
issue, Vol 12, No 1. Due to an unforgivable editorial
mix-up, incorrect figures and unassociated captions
were used in the paper, rendering Dr Andersons work
somewhat incomprehensible. While it is regrettably
impossible to exchange all the incorrectly-produced
copies of the Journal, at least a corrected version of
the paper can be put online on the Journals site at
www.j-pipe-eng.com, and this has been done, along
74 The Journal of Pipeline Engineering

with an updated PDF of the complete issue. While engineering and technical publications, this is one
the Journals editorial team is a sincere supporter instance where the benefit of easily-amended electronic
of the benefit of hard copy vs electronic copy for copy is manifest.
2nd Quarter, 2013 75

The induction of vibrations in

transmission pipelines by the
fall of a heavy structure nearby:
modelling the safety distances
by Dr Charles Fernandez*1, Laurent Bourgouin1, Frederic Riegert2, and
Alain Pecker3
1 GDF Suez, CRIGEN, Innovation and Research Division, Saint-Denis, France
2 GDF Suez, GRTgaz, Bois-Colombes Cedex, France
3 Geodynamique et Structure, Bagneux, France

A RECENT PROBLEM concerning wind turbines has emerged: their hypothetical fall may induce damaging
vibrations to buried transmission pipelines. CRIGEN, the GDF Suez research centre for gas and new
energy, has performed a study in order to protect buried transmission pipelines against such external
influences. The common threshold used by the industry, and established by the American Gas Association,
is stated in peak particular velocity (PPV) at PPV 50 mm/s. In this article, a model for the propagation of
these vibrations is presented. It has been validated by extensive field measurements coupled with a non-
linear 2-D finite-element model for the soil. An experimental soil characterization through MASW tests
coupled with vibration measurements was performed in a representative soil. GRTgaz, the main French
transmission pipeline operator, now uses the safety distances between wind turbines and pipelines proposed
in the RAMCES software in which the model is implemented and which was developed by CRIGEN.

T HIS ARTICLE tackles damage to buried steel pipelines

from impact-induced vibrations. The impact can be
due to a hypothetical fall of a heavy and tall structure,
such as a wind turbine, but could also be a crane.
Over the last decades, this kind of vibration problem has
been dealt with in different ways. One could evaluate the
stress induced by the propagating waves into the ground or
the deformation in the soil, and then apply this directly
or partially to the pipe using soil/structure interaction
laws. Alternatively, one could use empirical tests directly
to assess the pipe integrity. This last approach was chosen
by GDF Suez, and was based on using particle velocity,
an approach that is largely accepted in the industry. The
resulting criteria were defined using the AGA reports on
buried explosive detonations ([2], [3]) and the thresholds
for diameters, steel grades, and pressures; the frequency
was set to 50 mm/s (1.97 in/s) for the soil particles close
to the pipe ([4]). This level of vibration is relatively high
This article is based on an article published at the IPC 2012: Modelling of
windturbine fall-induced vibrations near buried steel transmission pipelines
an updated RAMCES software extension [1].
*Corresponding author:
tel: +33 (1) 4922 5262
email: charles.fernandez@gdfsuez.com
compared to the thresholds of vibration usually allowed
near aerial structures; however, buried pipelines are robust
structures regarding vibration (robust and elastic material,
long linear structure, electric arc welds, etc.). In addition,
the particle velocity is a very easy-to-use practical-oriented
criterion, which is a useful advantage when applying it
in the field. It should be emphasized that this threshold
is for the operating limit state.
GRTgaz, the GDF Suez gas transmission operator, along
with other transmission pipeline operators, use the RAMCES
software to evaluate the risk induced by a hypothetical
wind turbine impact on the ground. This software has
been developed by, and is maintained by, CRIGEN1. In
the previous RAMCES releases [5], the particle velocity
evaluation was performed using empirical data based on
dynamic compaction ([6], [7], and [8]), and was considered
a relatively similar physical phenomenon to an explosive
detonation. However, the energy available during the fall
of a wind turbine to the ground is much higher than
that in the dynamic compaction case. Therefore, a very
conservative relationship was used when extrapolating the
levels of vibration induced by a falling wind turbine in
the Mayne [6] or Menard equations [7].
1. CRIGEN: Center for Research and Innovation on Gas and New Energies.
76
Over the past few years, extensive construction of
wind turbines has taken place all around the world in
areas where many steel pipelines are already buried in
the ground. Consequently, there have been more and
more critical locations which needed specific attention.
For that reason, it was decided to update the model
to get a more accurate and less-conservative release of
RAMCES so as to reduce the safety distances between
wind turbines and pipelines, which also applies to
cranes and other heavy and tall structures.
An attempt to develop a theoretical model was made using
basic hypotheses such as propagation in an elastic soil with
a Heaviside excitation to model the crash on the ground
[9]-[11]. This model was updated with measurements of
vibrations and complex seismic tests (MASW tests, see
[12]) in a real operational field. At the same time, a
non-linear 2-D finite-element model for the soil was used
to optimize the parameters for the calibration [13]-[16],
and to take into account a realistic impact model for
the falling mass (a craterization phenomenon).
The first section below deals with the state-of-the-art
relationships used to estimate the particle velocity in
the dynamic compaction scenario. The second section
presents the results of a simplified analytical model for
the particle velocity induced by an elastic wave following
an impact in an elastic soil. The third section describes
the field tests performed to calibrate the computational
model presented in the final section.
Overly conservative estimation of
the particle velocity
Until recently, the estimation of the peak particle
velocity (PPV) was made using the dynamic compaction
Mayne equation [6]:
The Journal of Pipeline Engineering
Fig.1. Illustration of the physical
problem (based on [1]).
1
MH 4
PPVMayne 70 (1)
r
where PPVMayne is in mm/s, M is the falling mass in
kg3, H is the height in m, and r the distance in
metres between the falling mass and the point where
the particle velocity is estimated. This relationship
has been established for values of M x H which do not
exceed 1000 tm (where 1 t = 103 kg, the product M x
H is equal to the energy divided by g = 9.81 m/s2, the
gravity). The Mayne particle velocity can be compared to
the Menard particle velocity, PPVMenard [7] and [8] which
is derived from field observation, and is written:
1.1
1
PPVMenard 340
r (2)
for values of M x H between 200 and 300 tm.
For wind turbines, the M x H energies are far greater
than 1,000 tm (and are closer to 24,000 tm). The energy
dissipation at the impact greatly increases as both the soil
yields and a larger crater appears for the heavier mass
compared to the masses used in dynamic compaction.
In general, the dissipation of energy is greater for larger
impact surfaces than for smaller ones; the phenomenon
is highly non-linear, and thus a finite-element model had
to be used to account for it. When extrapolating these
relationships to wind turbines, the energy dissipation is
underestimated and thus the model may be over-conservative,
as demonstrated by the results presented in Table 2.
Particle velocity analytical evaluation
In this section, the hypothesis and the attempted approach
to build an analytical model for the evaluation of the
particle velocity are presented.
2nd Quarter, 2013 77

Fig.2. Normalized surface

vertical displacement due to
Rayleigh waves as a function
of normalized time, showing
compression waves (P), shear
waves (S), and Rayleigh waves (R).

Fig.3. Illustration of the

two test campaigns:
a (top) first campaign;
b (bottom) second campaign.

The physical problem to model is the impact of a heavy
mass in the ground and the induced wave propagations
(see Fig.1). To begin with, the soil is assumed to be
an elastic half-space and that the excitation is a point
source modelled by a Heaviside function. The solution
of the differential equation for the displacements of
the surface in that case was solved by Pekeris ([9]).
It can then be shown that the Rayleigh waves are
dominant compared to other kinds of wave2, their
amplitude inducing greater disturbance in the soil in
terms of displacement, as the graph of the surface
vertical displacements from Pekeris shows in Fig.2.
2. Compression waves and other shear waves, such as Loeve waves, for example.
78
A solution with Rayleigh waves for this problem is given
by Chao et al. [10], although it is only valid for z/r <<
1 and for times close to the Rayleigh wave arrival time.
The solution for a point-source excitation, a pulse, and
then any source of force can also be calculated.
The impact force due to a fall of a wind turbine can
be modelled as a moving circular foundation. The
foundation is modelled as an impedance with stiffness
and damping factors. The solution for an impedance
set on the surface of an isotropic homogeneous half-
space has been given by Gazetas [11]; values of the
impedance parameters can be found in that reference.
Using the values in Table values and the Gazetas solution
for the impedance factors, the absolute value of vertical
and horizontal velocities are given, respectively, by:
cT2 mvz is similar to the relationships given
u
0 .
18 (3a) Menard (Eqns 1 and 2, respectively).
rz 5
The Journal of Pipeline Engineering
Fig.4. Illustration of the MASW tests.
Fig.5. Right: computational model; the
red arrow on the upper left indicates
the impact point. Left: example of the
measured velocity profile.
cT2 mvz
w 0.53 (3b)
rz 5
where vz 2 gH , g is the gravity, m is the Lams
coefficient, r is the distance between the impact point
and the pipe, z is the depth from the ground surface,
and cT is the shear wave velocity.
The direct application of Equns 3a and 3b does not
give an accurate prediction, because the general trend
is an overestimation of the particle velocity. However,
one can see that the analytical model gives a particle
velocity with the following shape:
k
PPV = (4)
rn
where n and k are problem-dependant constants. This
by Mayne and
2nd Quarter, 2013 79

Fig.6.The four applications of the

CHUTOSOL module: (a) fixed crane;
(b) mobile crane; (c) wind turbine;
(d) other structure.

Fig.7. A completed calculation for a wind

turbine using RAMCES/CHOUOSOL.
80
Data
Equivalent radius for the impact surface
Nacelle mass
Fall height
Soil shear wave velocity 150 < cT < 1000 m/s
The velocities estimated with Equn 3 are too conservative
because of:
the elastic impact;
the elastic soil;
the Rayleigh waves are dominant (this is only valid
for z/r << 1, and thus this relationship cannot
be used for short pipe-impact point distances).
Therefore, in the following, a non-linear computational
model with plasticity is implemented and calibrated with
field tests to update a velocity described in Equn 4.
Field tests: vibrations and
multichannel analysis of surface waves
Two types of field test were undertaken:
particle velocity measurements
multichannel analysis of surface waves (MASW)
tests
The first campaign (illustrated in Fig.3) was aimed
at measuring the surface ground-particle velocity at
distances of 3 m, 8 m, 15 m, 30 m, 50 m, and 80
m of a 1.5-t mass falling from heights of 3 m, 5 m,
10 m, and 15 m. During these tests, the effect of the
soil yielding was evaluated.
The second campaign was aimed at measuring the
soils geotechnical data using the MASW method [12],
a layout of which is shown in Fig.4. The mass used
weighted 80 N (8 kg); there were 24 sensors.
The tests showed that:
the frequencies of interest are in the [0-30 Hz]
range;
the depth of the mass imprint increases with
the height from 300 mm to 1 m;
the particle velocities were dispersed along the
sensors;
some sensors were saturated during the tests.
The Journal of Pipeline Engineering
Value
0.5 < r0 < 5 m
25 < M < 70 t
50 < H < 100 m
Table 1. Example of data for wind
turbines (based on [1]).
These tests allowed the particle velocity to be fitted to
a law such as that proposed in Equn 4.
In these tests, the dilatational wave velocity Vp, the shear
wave velocity Vs, the Youngs Modulus E, the Poissons
coefficient , the shear modulus G, and particle velocities
were measured and calculated. These parameters were
then used in the computational model.
Simulation of a heavy structure based
on a computational model
As shown above, wind turbines are much higher and heavier
than the masses used in standard dynamic compaction
and other tests. Therefore, a computational evaluation
was required in order to establish a predictive model
for the wind turbine configuration. As a first step, the
computational model was built and updated with the field
data and empirical relations, and in the second step, the
updated model was used to evaluate the wind-turbine data.
The axisymmetric finite-element model [13] was based on
a realistic constitutive law [14, 15], and the computation
was dynamic and non-linear. The mesh was composed
of ten elements per wavelength, corresponding to one
element every 50 mm. There are 14,523 elements
and 29,048 degrees of freedom (DoF); the boundary
conditions are lateral nodes only moving vertically, and
the base is fixed. The mesh with the different soil layers
is displayed in Fig.5.
The law which is proposed in Equn 4 has been updated
by the field and computational tests, and can therefore
be written as:
MVz k MVz
(5)
PPV
rn
where PPV is the particle velocity in mm/s, Vz 2 gH
is the falling mass velocity in m/s, r is the soil density
in 106 kg/m3, M is the mass in 106 kg, and r is the
distance between the impact point and the pipe.
2nd Quarter, 2013 81

Wind turbine Mrotor+nacelle (t) Mtower (t) H (m) DVelocity_Mayne (m) DRamcesIII (m)

V52 32 110 86 196 105*

V82 95 127 78 219 100*

V80 104 200 100 281 130

V90 111 235 105 302 138

AW-1500-70 67.5 95 60 165 79

AW-1500-82 67.5 135 80 212 101

AW-3000-100 154 840 100 404 170

Table 2. Safety
distances. AW-3000-116 154 1,100 120 489 209

*The apparent contradiction between these two distances (the V82 wind turbine is heavier than the V52) comes from the height
of the wind turbine V52.The height H takes into account the whole length of the blades and the computation for the fall only
takes into account the rotor height without the blades length.

RAMCES: software for buried and would have to be calculated separately, it can be
transmission pipelines subjected to seen that installation of a wind turbine at a distance
external loads from the pipe greater than D1 is a safe installation.

The RAMCES software has been developed and
maintained by the CRIGEN for more than a decade,
and is aimed at calculating the stresses induced by
the most common external loads on steel transmission
pipelines. It uses nine modules covering a wide variety
of different loading situations, only one of which
(CHUTOSOL3) will be described in this article.
The CHUTSOL module is made up of four tabs, each
one of which allows for the computation of a safety
distance between a buried pipe and a heavy and tall
structure.
The types of structure concerned are shown in Fig.6
a-d: in each of these, the updated Equn 5 is used
to evaluate the vibration-specific risk for each kind of
structure.
As an example, in the wind-turbine case, the software
gives two circular zones of safety (see Fig.7) in which:
Table 2 gives numerical values for Vestas and Acciona
wind turbines, data for which have been taken from
the companies web sites (dated 2010), [17, 18].
Mitigation measures
When the wind turbine is too close to the pipeline,
i.e. its fall will produce damaging vibrations on the
pipeline, mitigation measures have to be introduced.
Open trenches have been used for temporary works in
such cases, and are suitable when explosives are being
used, or for work with heavy cranes nearby, etc., but
this is not really adaptable to more permanent activities,
such as wind turbines.
To meet the need of protecting a pipeline against
vibration from the impact of a falling wind turbine,
a special trench has been designed which combines
both the advantages of:

the yellow circle (the smaller one) matches the
zone where the risk is the combination of the
structures fall (i.e. a direct impact by the structure)
and the fall-induced vibration of the impact;
the blue circle matches the zone where only the
fall-induced vibrations cause the damage (and the
height of the structure is no longer a problem).
With the exception of the separate risk if impact from
a turbine blade, which is not considered in this version
3. CHUTOSOL literally means fall to the ground.
reducing the vibrations; and
presenting no danger that someone or something
might fall in it.
The trench is filled with a mix of sand and recycled
tyre chips, a system that is already being used in other
application to dampen vibrations, such as in childrens
play areas and under tramways. This new mitigation
measure is currently under study, but theory and tests
have already been performed which show useful and
interesting results.
82
Conclusion
In this paper, a more-accurate and less-conservative
model has been presented that can be used in the
GDF Suez RAMCES software in order to predict safety
distances between buried transmission pipelines and
heavy and tall structures, particularly wind turbines.
The safety distance is calculated by taking into account
the height and mass of the wind turbines components:
a first safety circle represents the risk of a direct hit
from the structures height (with the exception of the
separate risk from the blade), and a second safety circle
represents the risk due to vibrations from impact to
the ground. The considered vibrations threshold is
50 mm/s for any pipe configuration. As an example,
the reduction in proximity for a Vestas V90 wind
turbine is about 164 m (from 302 m to 138 m) a
calculated by the previous and the updated versions of
RAMCES. Mitigation measures have been proposed for
both temporary and permanent installations.
Acknowledgments
The authors thank the GRTgaz Engineering Center
in Lille, and particularly Marc Rifaut who found the
right place to perform the seismic tests, and also Jean-
Luc Mattiuzzo from INNOGEO who performed the
field tests.
References
1. C.Fernandez, L.Bourgouin, F.Riegert, and A.Pecker,
2012. Modelling of wind turbine fall-induced
vibrations near buried steel transmission pipelines:
an updated RAMCES software extension. Proc. IPC
2012, Calgary, Canada.
2. E.D.Esparza, P.S.Westine, and A.B.Wenzel, 1981.
Pipeline response to buried explosive detonations,
Vols I & II, SRI. Final report of the American
Gas Association Project PR-15-109 for the Pipeline
Research Committee.
3. E.D.Esparza, 1991. Pipeline response to blasting in
rocks. Southwest Research Institute, American Gas
Association.
The Journal of Pipeline Engineering
4. P.A.Bordonne, S.Akel, and K.Dang Van, 1991. Tenue
mcanique des canalisations enterres soumises
des vibrations dorigine ponctuelle. Gaz de France
report no. CRSTA-BO/VH 91173.
5. GDF Suez Center for Research and Innovation, Gas
and New Energies. RAMCES software - theoretical
references. GDF Suez report.
6. P.W.Mayne, J.S.Jones, and J.C.Dumas, 1984. Ground
response to dynamic compaction. J. Geotechnical
Engineering, 109, pp757-774.
7. Menard Sol Traitement, 2001. Vitesse particulaire
resultante en function de la distance au point
dimpact. Private conversation.
8. J.F.Semblat and A. Pecker, 2010. Waves and vibrations
in soils: earthquakes, traffic shocks, construction
works. IUSS Press.
9. C.L.Pekeris, 1955. The seismic surface pulse. Proc.
National Academy of Sciences, 41, pp 469-480.
10. C.C.Chao, H.H.Bleich, and J.Sackman, 1961. Surface
waves in an elastic half space. J. Applied Mechanics,
28, pp 300-301.
11. G.Gazetas, 1983. Analysis of machine foundations
vibrations state of the art. Int. J. of Soil Dynamics
and Earthquake Engineering, 2, pp 2-42.
12. C.B.Park, R.D.Miller, and H.Miura, 2002. Optimum
field parameters of an MASW survey [Exp. Abs.].
SEG-J, Tokyo.
13. J.H.Prevost, 2002. DYNAFLOW A finite element
analysis program for the static and transient response
of linear and non linear two and three-dimensional
systems. Dept of Civil Engineering, Princeton Univ.,
VersionV02.
14. Idem, 1978. Plasticity theory for soil stress-strain
behaviour. J. Engineering Mechanics, 104, (EM5),
ASCE.
15. Idem, 1985. A simple plasticity theory for frictional
cohesionless soils. Soil Dynamics and Earthquake
Engineering, 4, 1, pp 9-17.
16. T.J.R.Hughes, 1983. Analysis of transient algorithms
with particular reference to stability behaviour.
Computational Methods for Transient Analysis,
Elsevier.
17. www.vestas.com (last consultation, 2010).
18. www.acciona-energia.com (last consultation, 2010).
2nd Quarter, 2013 83

Techno-economic modelling and

analysis of CO2 pipelines
by Nima Ghazi*1 and Dr Julia M Race2
1 Integrated Pipeline Projects Canada Ltd, Calgary, AB, Canada
2 Newcastle University, Newcastle-upon-Tyne, UK

T HE MAIN FOCUS of this paper is on techno-economic modelling and analysis of CO2 pipelines, as
it strives to develop a thorough understanding of the essential fluid-mechanics variables involved
in modelling and analysis of such pipelines. The authors investigate and analyse the reasons behind the
variations in the techno-economic results generated from seven different techno-economic models which
are commonly used for construction and operation of CO2 pipelines. Such variations often translate into
tens or, at times, hundreds of millions of dollars in terms of initial financial estimates at the pre-FEED
(front-end engineering design) or FEED stages for carbon capture and storage (CCS) projects. Variations
of this magnitude can easily bring much unwanted uncertainty to the feasibility of a CO2 pipeline project,
and they can potentially cause a major over- or under-estimation of the projects true costs. The summary
of a detailed analysis and assessment for these seven existing techno-economic models for CO2 pipeline
transport is presented: the analysis conducted indicates that some of these models are essentially identical
and are rooted in similar fluid-mechanics theories and assumptions. This type of analysis assists with
explaining and narrowing-down the variability of the models results. Based on these analyses, a refined and
more-accurate model was established and the development process was explained.The refined model uses
the Reynolds number, Colebrook-White equation using the Darcy friction factor, and the Darcy-Weisbach
pressure drop equation to establish the most accurate measure for the pipes diameter. To assess the CO2
pipelines total capital cost, total annual cost, and the levelised transport cost, a statistical regression analysis
approach was suggested and the adjusted-r2 measure was proposed to assess the goodness-of-the-fit of the
generated cost function. The accuracy of the new techno-economic model was validated with the figures
from a proposed CO2 infrastructure project in the UK, and also through hydraulic modelling.

G LOBAL ENERGY CONSUMPTION is increasing

significantly and is expected to grow by 49%, or
1.4%/year, from 495 quadrillion Btu in 2007 to 739
quadrillion Btu in 2035 [1]. This rise is, in part, due to
increases in the existing level of energy consumption and
also due to the energy required to fuel global economic
growth, particularly in emerging economies such as
Brazil, Russia, India, and China. In order to sustainably
meet the growing energy demands and challenges of the
future, reliance cannot be placed on the methods and
technologies of the past.
According to the Intergovernmental Panel on Climate
Change (IPCC), there is new and stronger evidence that
most of the warming observed over the last 50 years is
due to human activities [2]. Burning coal, oil, and natural
gas will continue to increase the CO2 concentrations in
the atmosphere which directly contributes to anthropogenic
global warming. The power industry accounts for about
*Corresponding author:
tel: +1 403 767 9990
email: nima.ghazi@ippeng.com
a quarter of the global greenhouse gas (GHG) emissions
with the majority of emissions attributable to fossil fuel
and, especially, coal-fired power stations (IHS, 2011).
Currently coal is the most significant fuel source for
power generation. It is estimated that coal-fired generation
accounts for 40% of the 20,000 terawatt-hours (TWh)
of electricity generated worldwide but also accounts for
three-quarters of all the CO2 emitted by the global
power sector [3]. While renewable technologies are under
much-needed development, currently the largest portion
of the global energy demand has to be met through
hydrocarbon sources. Therefore, it is essential, from an
environmental standpoint, to generate the demanded
energy with fewer emissions.
One of the technologies that have been identified to
produce a reduction in the levels of anthropogenic
emissions from fossil-fuelled energy generation is CCS.
The CCS chain of technologies involves the capture of
CO2 at a power plant or a large stationary industrial
source and transport it, generally via a pipeline, to
a geological storage site where the CO2 can either be
84
stored securely or used for enhanced oil recovery (EOR).
At the heart of determining the viability and applicability
of CCS lies the analysis of the technical and economic
components of the entire CCS chain, part of which is
the CO2 pipeline system. This paper reviews the techno-
economic modelling and analysis of CO2 pipelines for
CCS projects, and proposes a refined model, derived from
the perceived best practice in the existing models, that
can be used at the feasibility stages of a CCS project
to estimate the cost of the pipeline.
Review of existing techno-economic
models for CO2 pipelines
There are existing techno-economic models which provide
analytical tools for determining the CO2 pipeline diameter
based on hydraulic calculations, whilst also estimating the
costs associated with the construction and maintenance
of the pipeline. Predominantly, these models provide an
estimate of the cost of transporting a unit mass of CO2
(the levelised cost) over a certain distance.
In the current work, seven CO2 pipelines techno-economic
models have been reviewed, including MIT [4], Ecofys
[5], McCoy & Rubin [6], Ogden (Ogden et al., 2004),
IEA GHG PH4/6[8], IEA GHG 2005/2 [9], and IEA
GHG 2005/3 [10] models. These models have different
underlying assumptions and various methods of assessment.
The different approaches are discussed in the following
sections and summarized in Annex A. It is highlighted
that most of these models were developed for onshore
CO2 pipelines and there is less published analytical work
for offshore CO2 pipelines.
For pipeline diameter calculations, Annex A illustrates that
these models, based on their hydraulic analysis approach,
fall into three main categories: those that use the Darcy-
Weisbach method (coloured green), those based on a
mechanical energy balance calculation (coloured orange),
and those based on a mass flow rate calculation or a
rule of thumb calculation (coloured red).
Darcy-Weisbach-based models
The MIT, Ecofys, and the IEA GHG PH4/6 models all
apply the Darcy-Weisbach [11] fluid-mechanics principle.
Consequently, the pipeline diameters calculated by these
models are similar. The Darcy-Weisbach equation expresses
the pressure drop, h[m], as a function of the friction
factor, f; this relationship is further discussed in the
following sections. The calculation approach for these
models allows the pipeline diameter to be determined
based on robust fluid-mechanics relationships, thereby
reducing the number of set-value assumptions required.
Consequently, it is reasonable to expect that the calculated
diameter values would be more accurate using this
approach than those of other methods in which more
general assumptions are made.
The Journal of Pipeline Engineering
The strength of the MIT model lies in its simple iterative
fluid-mechanics approach; however, the model assumes
that the absolute pipeline roughness (e) of the pipe is
a constant figure (0.00015 ft). This is considered to be
one of this models simplifying assumptions because,
in reality, this value varies for different linepipes. The
model also assumes that the annual operation and
maintenance cost of the pipeline is only a function of
the length and not the diameter of the pipeline: this
is not an accurate assumption either for the cost of
operation or for the cost of maintenance. For example,
assuming similar pressure levels, a larger-diameter pipeline
often requires a more powerful booster station than a
smaller-diameter pipeline. This will automatically require
more power to operate the larger pipeline, which will
increase the operation cost. In addition, for maintenance
and integrity management of pipelines inline inspection
and cleaning tools, such as pigs, are often used. The
pipeline diameter determines the size of the inspection
or cleaning tool. Assuming similar pig technologies,
the larger the tool, the heavier it is to transport and
more costly it is to operate.
The Ecofys model accounts for various types of terrains
through the incorporation of a terrain factor, FT, in
its capital cost equation. This is an advantageous
consideration as it allows the equation to be adapted
to pipeline construction for different terrains. The
IEA GHG PH4/6 model also distinguishes between
various terrains using terrain factors but, in addition, it
acknowledges the importance of accounting for various
economic locations by incorporating location factors,
FL, in the calculations. However, it is considered that
the proposed values for location factors in this model
could be improved to account for different international
and regional economic drivers.
Mechanical energy balance models
For the calculation of the pipeline diameter, the McCoy
& Rubin and Ogden models apply equations that are
based on the mechanical energy balance, which essentially
assumes that the change in the mechanical energy of
an isothermal fluid flow is caused by friction. It is
applicable to systems with a single input and a single
output. The models assume that an accurate average
compressibility value, an accurate average temperature,
and an accurate average pressure can be found for any
given CO2 pipeline system. This may or may not be
the case, and it can potentially introduce inaccuracy in
the calculated diameter size of the pipeline.
The McCoy & Rubin model was developed in
the United States and divides the country into six
geographical regions of Midwest, North East, South
East, Central, South West, and West. This is due to
the fact that the cost of material, labour and business
operation are different in each of these regions, and
2nd Quarter, 2013
Fig.1. CO2 phase diagram.
Fig.2. Density of pure CO2 at various
temperatures and pressures.
the models structure accounts for these regional cost
differences. Ogdens model suggests various ranges for
average compressibility, average pressure, and average
temperature of the CO2 stream to be used with its
proposed equation. Finding the exact value for these
variables can be challenging.
Mass flow rate calculation models
The IEA GHG 2005/2 model is based on a mass flow
rate calculation in which the diameter is calculated
using the following equation:
D2 4m
(1)
m AV i V Di
4 V
where m kg/s is the mass flow rate, V m/s is the average
velocity of the fluid, A m2 is the internal cross-sectional
area of the pipe, Di m is the internal diameter of
the pipe, and r kg/m3 is the fluid density. A major
simplification in the IEA GHG 2005/2 model is that
neither the friction factor nor the pipe roughness
is taken into account for the diameter calculation
process. The impact of this assumption is that the
accuracy of the model will decrease as the length of
the pipeline increases.
85
Rule-of-thumb models
The IEA GHG 2005/3 model utilizes a rule-of-thumb
approach to determine the pipeline diameter. The
assumption is that the volumetric flow rate of CO2
should be equal to a fixed value of 0.65*106 scf/day/
in2 of the internal area of the pipe. In addition, as
with the IEA GHG 2005/2 model, the model does
not take into account either the friction factor or the
pipes roughness.
Consideration of CO2 physical
properties
Two of the inputs into the pipeline diameter calculations
process, in some of the reviewed models, are values
of the CO2 density and viscosity, measured at specific
temperature and pressure levels. It is important, therefore,
to understand how these properties can vary with
temperature and pressure.
A phase diagram for pure CO2 is presented in Fig.1.
This diagram exhibits two distinct features: a triple
point (5.2bar, -56C); and a critical point (73.8bar,
31C). In the vicinity of the triple point, CO2 can
exist as one of three phases: solid, liquid, or gas. At
pressures and temperatures above the critical point,
86
CO2 no longer exists in distinct gaseous and liquid
phases, but as a supercritical phase with the density
of a liquid but the viscosity of a gas. At pressures
above, but temperatures below, the critical point, the
CO2 exists as a liquid whose density increases with
decreasing temperature; this phenomenon is illustrated
in Fig.2. Based on Fig.2, it is evident that a change
of temperature, at a constant pressure, has a more
significant impact on the density of CO2 than a change
of pressure, at a constant temperature. The viscosity
of the CO2 stream is also a function of pressure and
temperature, as shown in Fig.3.
Consideration of impurities
CO2 that is captured from a power plant is not pure,
and the amount and type of impurities in the CO2
stream are dependent on the capture technology. There
are three main process routes for capturing CO2 from
power plants: post-combustion capture, pre-combustion
capture, and oxyfuel. In relation to the phase behaviour,
the impact of impurities is to raise the critical point
of the fluid and to lower the density and viscosity,
and this has important implications on the hydraulic
behaviour of the CO2 [13]. Although it is recognized
that impurities will impact the hydraulic component of
any model, all of the models reviewed, and indeed the
currently proposed model, do not explicitly account for
the effect of impurities.
Capital cost comparison of the
reviewed models
It is important to consider that the accuracy of a techno-
economic model for calculating a pipelines capital cost
1. The McCoy & Rubin model generates various graphs for different regional
areas of the United States, and these graphs are not depicted in Annex B.
The Journal of Pipeline Engineering
Fig.3.Variation of CO2 viscosity as a
function of temperature and pressure
(reproduced with permission [12].
is directly impacted by the accuracy of the models
methodology for calculating the pipelines diameter. In other
words, if the calculated pipeline diameter is only roughly
estimated (for example, calculated by a rule-of-thumb or
an over-simplified calculation), then the calculated costs
can be significantly over- or under-estimated. Annex B
illustrates a pipelines capital costs values as a function
of the mass flow rate, for a 100 km pipeline, for six of
the analysed models1.
Annex B illustrates that the level of variation in the
calculated costs can be as large as 200%. Such variations
are often equivalent to tens or, at times, hundreds of
millions of dollars in terms of economic uncertainty for
the financial feasibility estimates of a CO2 pipeline or a
CCS project. Variations of this magnitude bring much
unwanted uncertainty to the feasibility of the project and
can potentially cause a major over- or under-estimation
of the projects true costs. For example, Annex B shows
that the IEA GHG 2005/3 (which calculates the pipeline
diameter based on a rule-of-thumb) generates the highest
estimate of costs compared to IEA GHG PH4/6, which
adopts a more sophisticated approach for calculating the
pipeline diameter.
Proposed techno-economic model for
CO2 pipeline transportation
The aim of this current work was to develop a techno-
economic model that could be used with confidence,
particularly for CCS projects in the UK and North
America. Such a model requires two components: a
fluid-mechanics model for the calculation of the pipeline
diameter, and a robust cost model for the calculation
of construction, material, operation, and maintenance
costs. The review of the existing models, as outlined
in the previous sections, has highlighted the different
approaches that have been adopted for the diameter and
2nd Quarter, 2013 87

cost components of the models and the strengths and roughness is not specified for a pipe then e = 4.57
weaknesses of these models. As a result, this work x 10-5 m is a reasonable approximation [6].
proposes an alternative techno-economic model based
on the observed best practice in the current models. 2.51
1/ f 2 log10 ( )
3.7 Di Re f (7)
Diameter calculation process
As Equn 7 contains the friction factor (f) on both
The recommendation for the calculation of the diameter sides of the equation, an easier way to calculate the
of the pipeline is based on the premise that application Darcy friction factor is via calculation of the Fanning
of the model should require as few set-value assumptions friction factor (fF) and utilizing the relationship:
as possible; in addition, the model has to be based on
robust fluid-mechanics principles. These premises are to (8)
improve the models results accuracy. Consequently the
diameter-calculation methods using the Darcy-Weisbach The Fanning friction factor can be expressed [14] as:
fluid-mechanics principles are recommended as this
approach requires fewer initial assumptions to be made 1 / Di 5.02 / Di 5.02 / Di 13 (9)
2.0 log log log
(i.e. those coloured green in Annex A). 2 fF 3 .7 Re 3 .7 Re 3.7 Re

The Darcy-Weisbach equation expresses the loss of where Re is the Reynolds number and is given by the
pressure head caused by friction, h[m], as a function equation:
of the friction factor. The Darcy-Weisbach equation is
given by Menon [11] as:
VDi 4m
Re
= = (10)
/ Di
(2)

where L[m] is the pipeline length, f is the Darcy friction where Pa.s is the absolute dynamic viscosity of
factor, and g m/s2 is the acceleration due to gravity dense-phase CO2 which can be assumed to be roughly
(all other terms having being defined previously). From 6.06 x 10-5 Pa.s [4].
basic hydraulics it is known that:
It is observed from Equn 10 that the internal diameter
f ( L / Di )
P gh P V2 (3) is required as an input into the calculation of Re.
2
Therefore, the pipeline diameter has to be calculated by
and the average velocity can be obtained from Equn an iterative process, in which an initial estimate of the
1 as: diameter is required. As a rule of thumb, the values of
internal diameter tend to converge within a 10-6 range
4m
V = (4) in less than five iterations [6].
Di 2

Once the internal diameter has been calculated, to further

Therefore, substitution of Equns 2 and 4 into Equn specify the pipes geometry, the wall thickness can be
3 enables the pressure drop to be represented as: determined using:

2 (11)
f ( L / Di ) m f ( L / Di ) m2 8 fLm2
P (5)
2 A 2 D4 2 Di5
2 2 i
16
In this equation, E is the longitudinal joint factor and is
which, rearranged, enables an expression for internal assumed to be equal to unity unless otherwise specified; Do
diameter to be developed: is the external diameter, SMYS is the specified minimum
yield stress of the pipe material, and PMOP is the maximum
operating pressure. Evidently, the outer diameter can be
8 fm2 fLm2 fL Q 2
Di5 0.81 0.81 (6) calculated by adding the internal diameter with twice the
P P P
2 wall thickness of the pipe. DF is the design factor, and
L
its value is established based on the governing code and
where Q m3/s is the volumetric flow rate. operator specifications for the CO2 pipeline. To establish
the design factor value, the code will normally take into
The Darcy friction factor (f), can be calculated from account the fluid category to which dense-phase CO2
the Colebrook-White equation (Equn 7). In order to belongs and the associated population density based on
do this, the pipes relative roughness, e/D, is required. the route and location of the pipeline. Building proximity
e [m] is the pipes absolute roughness. If absolute distances are then established.
88
Cost calculation process
Although the diameter-calculation methodology can be
universally applied to various CO2 pipeline projects in
different locations around the world, finding a single
capital cost or maintenance cost equation accurately
and universally to evaluate the pipeline costs in all of
the various locations and regulatory structures around
the world is not practical. This is due to the various
location-specific cost elements and also due to the cost
implications of different regulations that impact the final
cost of each specific pipeline project.
In a simplified approach, the capital costs (CAPEX) for
the construction of a pipeline project can be estimated
by a function of the pipelines external diameter (Do) and
its length (L). Such estimations can be made through
regression-modelling analysis of the pipeline assets capital
costs sourced from databases such as the Oil and Gas
Journal. A more-detailed approach should consider the
specific pipelines wall-thickness. The same simplified
relationship can also be assumed for the operation and
maintenance costs, sometimes referred to as OPEX.
Indeed, the annual operation and maintenance costs of
a pipeline are usually estimated to be a specific fraction
of the CAPEX. In some models, the OPEX is even
further simplified and is expressed only as a function of
the pipelines length. This further simplification is not
recommended, particularly when considering operation
and maintenance costs of pipelines with large diameters.
Such a simplification can underestimate the actual
operation and maintenance costs involved and could
potentially cause operational cash-flow shortages. Based
on the models analysed above, 2%-5% of the CAPEX
is considered to be a realistic range for the operation
and maintenance costs.
Although there are limited cost data available for the
construction of CO2 pipelines, it is considered that the
available capital cost data for natural gas pipelines can
be used as the cost of construction should be largely
independent of the fluid. It is then recommended that
the data would be normalized to one reference year,
preferably to the current year. In order to undertake this
cost normalization accurately, for each pipeline project,
the projects value for return-on-equity (ROE) rate,
projects value for return-on-debt (ROD) rate, and each
corresponding years inflation rate, have to be specified
and accounted for in the calculation. Through application
of regression analysis it is then possible to derive an
equation that provides a best-fit to the available cost data,
i.e. the equation with the highest adjusted-r2 value. This
methodology has been used to derive the cost formulae
used in the IEA GHG PH4/6 and IEA GHG 2005/2
models. It is, therefore, these cost models that have
been adopted for the current analysis. The corresponding
capital cost formulae for onshore pipelines, based on the
IEA GHG 2005/2, model is:
The Journal of Pipeline Engineering
Pipeline capital Costonshore () =
FL x FT x 106 x [(0.057Lonshore + 1.8663) + (0.00129Lonshore) x
Do + (0.000486Lonshore - 0.000007) x Do2] (12)
The only adjustment made to IEA GHG 2005/2 for
generating the above equation is the inclusion of the
location factor, FL. The suggested location factor values for
Equn 12 are the same as those used for the IEA GHG
PH4/6 model. The location factor for USA, Canada,
and Europe is assumed to be 1, while it is set at
1.2 for the UK. The terrain factor, FT, is assumed to
be 1.10 for cultivated land, 1.00 for grassland, 1.05
for woodland, 1.10 for jungle, 1.10 for stony desert,
1.30 for mountainous (slopes < 20%), and 1.50 for
mountainous (slopes > 50%).
Similarly, the adjusted capital cost formulae for offshore
pipelines model is:
Pipeline capital Costoffshore () =
FL x 106 x [(0.4048Loffshore + 4.6946) -
0.00153Loffshore + 0.0113) x Do + (0.000511Loffshore +
0.00024) x Do2] (13)
Lonshore km refers to the length portion of the pipeline
which is situated onshore, while Loffshore km refers to the
length portion of the pipeline that is situated offshore.
The annual pipeline operation and maintenance costs
are estimated to be about 3% of the total pipeline
capital costs in the IEA GHG 2005/2 and these have
also been assumed in the current analysis.
It is highlighted that, for the estimation of the capital
construction costs for CO2 pipelines in the United States,
the McCoy and Rubin [6] models are recommended, as
the utilized database of cost data is comprehensive and
the analysis can be tailored for individual cost regions
of the country. In practice, however, determining the
capital costs involved with a high level of accuracy,
i.e. with less than 5% or even 8% deviation from
actual figures, is a much more detailed and project-
specific process. The models proposed here can only
provide a rough estimate of the actual costs: these
rough estimates are nevertheless useful for the feasibility
and FEED studies of the projects.
Case study and validation
Once the proposed techno-economic model had been
assembled, it was important that it was validated
against detailed hydraulic analysis and the cost analysis
of a potential CCS project. The case study that was
chosen for this analysis was the proposed Yorkshire
and Humber network of CO2 pipelines in the UK,
for which published data were available [15, 16]. The
proposed CCS scheme in the Yorkshire and Humber
region aims to capture 60 million tons of the regions
2nd Quarter, 2013 89

Annex A.Techno-economic models comparison.

Annex B. Comparison of capital cost estimates for the models reviewed.

90 million ton emissions and transport it via an on- where explicit data were not available, estimations were
and offshore pipeline network to a storage site in the made or values were assumed.
Southern North Sea.
Provided values:
The techno-economic model developed in this paper
was used to estimate and validate the diameter size of design life: 40 years
the networks trunk line and the associated costs of Pin: 125 bar
building the trunk line with those in the Yorkshire Pout = Pmin = 100 bar
Forward study. The following are the input values for Tmax: 30C
the projects variables used in this validation process: M: 1616.7 kg/s
90
Estimated values and assumptions:
Lonshore: 109 km
Loffshore: 135 km
Ltotal: 244 km
Tave: 20C
Plant capacity factor: 0.85
r: 800 kgm-3 (suggested density value in Ecofys,
IEA GHG PH4/6 and IEA GHG 2005/2 models)
: 6.06 x 10-5Pa.s (suggested viscosity value for the
MIT model)
DF: 0.72
The onshore and offshore pipeline lengths are estimated
values generated from a scaled map published by CO2Sense
Yorkshire [17]. Based on available literature [18, 19], it
is considered that 800kgm-3, assuming the 20C average
temperature, is a reasonable estimate of the density. For
diameter calculations, the on- and offshore trunk lines will
be considered as a single 244-km long pipeline; however,
for cost calculations, the onshore and offshore capital
costs will be calculated separately (using Equns 12 and 13)
and combined together to produce the total capital cost.
The only unknown for calculating the diameter is now
the friction factor, f, which is a function of the pipes
relative roughness, e/Di. Considering the very large mass
flow rate of this pipeline project, i.e. 1616.7kg/s, a large-
diameter trunk line will most likely be required. Hence, it
is expected that the friction factor value for this pipe will
fall within the low spectrum of friction factor values, i.e.
between 0.0100 0.0120, due to the relationship in Equn
9. A value of 0.0104 was selected as an initial estimate of
friction factor to reduce the number of iterations.
Based on the available data, and using the iterative approach
described for calculating the diameter, the internal diameter
for the pipeline is calculated to be 1.2139 m or 47.79 in.
As the final diameter iterations were within 0.18 in, a
diameter of 48 in (1.219 m) is the closest available API
diameter specification. Assuming that X60 pipe (i.e. SMYS
= 413.8MPa), manufactured as per API 5L specifications,
is a suitable linepipe option, the wall thickness can be
calculated using Equn 11 to be 25.57mm.
The diameter size calculated by this method was also cross-
checked and re-validated using fluid-mechanics simulation
of the pipeline with the ASPEN HYSYS software [20]. The
pipeline was simulated for pure CO2 using the Beggs and
Brill flow equation [21] and the Peng Robinson equation
of state [22]. The diameter calculated using this method
was 1.2140 m, or 47.80 in.
Knowing the diameter and the length values of the pipeline,
onshore and offshore capital costs were calculated using
Equns 12 and 13:
Pipeline capital cost onshore () = 181,210,642.6
Pipeline capital cost offshore () = 249,630,827.1
The Journal of Pipeline Engineering
Adding the on- and offshore costs will result in a total
cost of 430,841,469.7. It is important to notice that
this value is calculated based on 2005 cost figures and
considers a 10% annual discount rate. Therefore, the
equivalent Euro value for 2010 would be 693,874,495.4
(580,716,147).
Based on the projects estimated 40-year design life, the
annual capital cost (equivalent annuity value) would be
59,383,693. The annual operation and maintenance
cost is calculated based on 3% of this value, which is
1,781,511. Therefore, the total annual cost to build
and operate this pipeline will be 61,165,204, and the
levelized CO2 transport cost is:
(14)
The Yorkshire Forward report [15] estimates a levelized
transport cost for a number of different transport
scenarios depending on the amount of CO2 captured,
the availability of CO2 sources, and the year in which
CCS is adopted. In this respect, the analysis is more
sophisticated than the simple model outlined here,
particularly with respect to economic drivers affecting
the adoption and deployment of CCS. It is therefore
highlighted that the intention of the calculation in
this paper was validation for the methodology of the
current model and not to provide any comment on the
accuracy of the approach adopted by Yorkshire Forward.
However, it is deemed that the estimated costs of
1.20/ton are within the range of the costs calculated
in the Yorkshire Forward study, and consequently it
is considered that the model presented in this paper
could be used to provide feasibility study cost analyses
for CCS transport projects.
Conclusion
It is considered that CCS is one of the primary modern
technologies that can significantly contribute to reducing
anthropogenic CO2 emissions produced by a range of
industrial facilities and enable global emissions targets
for CO2 to be met. Pipelines are currently the most
common and economic means of transporting large
volumes of CO2 over long distances, and they allow
for safe and reliable transportation of CO2 from a
capture facility to a sequestration or an enhanced oil
recovery site.
A great effort is therefore underway to analyse and
support the viability of CCS projects, part of which is
the calculation of the cost of the pipeline transportation
of CO2. However, it was observed, through detailed
analyses, that the current pipeline techno-economic
models could produce differences in calculated costs
for transportation of up to 200%.
2nd Quarter, 2013
In this paper seven techno-economic models for CO2
pipeline transport have been reviewed and the main
similarities and differences of these models have been
highlighted. This analysis allowed a refined techno-
economic model to be developed based on the best-
practices of the seven models studied. All of the techno-
economic models involve a hydraulic calculation, to
determine the pipe diameter, and a cost calculation, to
determine CAPEX and OPEX. It is therefore important
that as few assumptions as possible are made in these
calculations in order to increase the accuracy and
relevance of the costs calculated. To establish a more-
accurate measure for the pipes diameter, the proposed
model applies the Reynolds number, the Colebrook-White
equation for calculating the Darcy friction factor, and
the Darcy-Weisbach pressure-drop equation. Following
the calculation of the pipeline diameter, and knowing
the length of the pipeline, the associated costs of the
pipeline project are estimated. To assess the pipelines
total capital cost, total annual cost, and the levelized
transport cost, a statistical regression analysis approach
was proposed using the adjusted-r2 measure to assess
the goodness of the fit of the generated cost function.
The accuracy of the proposed techno-economic model
was then validated with the figures of the trunk line
for a proposed CCS project in the UK. The models
estimates were also cross-checked through an ASPEN
HYSYS fluid-mechanics simulation. It was considered
that the models results were within accepted margins
of the published project report figures given the
assumptions made in the current analysis.
The refined techno-economic model developed in
this work can be used to support the infrastructural
development of CCS projects and to address some of
the important techno-economic questions facing CO2
pipeline transport projects. It can, however, only provide
an estimate of the actual costs. It is important to note
that such estimates are only useful for the feasibility
and sometimes FEED stages of the projects and are
not recommended for detailed engineering design stages
of such CO2 pipeline projects.
References
1. Energy Information Administration, 2010. US Department
of Energy: International Energy Outlook. Available at:
http://www.eia.doe.gov/oiaf/ieo/world.html (Accessed:
10 April 2011).
2. B.Metz, O.Davidson, H.de Coninck, M.Loos, and
L.Meyer (Eds), 2005. IPCC. Cambridge University
Press, UK. pp 431.
3. IHS, 2011. Advanced carbon capture: SRI Consulting
process economics program report. Available at: http://
press.ihs.com/press-release/energy-power/advanced-carbon-
capture-technologies-coal-fired-emissions-improved-still
(Accessed: 12 April 2011).
91
4. G.Heddle, H.Herzog, and M.Klett, 2003. The
economics of CO2 storage. MIT LFEE 2003-003 RP.
p115. Available at: http://sequestration.mit.edu/pdf/
LFEE_2003-003_RP.pdf (Accessed: 21 March 2010).
5. N.Hendriks, T.Wildenborg, P.Feron, T.Graus,
and R.Brandsma, 2003. EC-case carbon dioxide
sequestration. M70066, Ecofys.
6. S.McCoy and E.Rubin, 2007. An engineering-economic
model of pipeline transport of CO2 with application
to carbon capture and storage. Department of
Engineering and Public Policy: Carnegie Mellon
University, Pittsburgh.
7. J.Ogden, C.Yang, N.Johnson, J.Ni, and J.Johnson,
2004. Conceptual design of optimized fossil energy
systems with capture and sequestration of carbon
dioxide. Report to the US Department of Energy,
National Energy Technology Laboratory.
8. IEA, 2002. Transmission of CO2 and energy. IEA
Greenhouse Gas R&D Programme, Report no.
PH4/6.
9. IEA, 2005a. Building the cost curves for CO2
storage: European sector. IEA Greenhouse Gas R&D
Programme, Report no. 2005/2.
10. IEA, 2005b. Building the cost curves for CO2
storage: North America. IEA Greenhouse Gas R&D
Programme, Report no. 2005/3.
11. E.Menon, 2004. Liquid pipeline hydraulics. Marcel
Dekker, Inc., New York.
12. S.Bachu, 2011. Private e-mail communication with
N.Ghazi, 6 May, 2011.
13. P.Seevam, J.Race, M.J.Downie, and P.Hopkins, 2008.
Transporting the next generation of CO2 for carbon
capture and storage: the impact of impurities on
supercritical CO2 pipelines. 7th International Pipeline
Conference, Calgary, Canada.
14. D.J.Zigrang and N.D.Sylvester, 1982. Explicit
approximations to the solution of Colebrook friction
factor equation. AIChE Journal, 28, 3, pp514515.
15. Yorkshire Forward, 2009. A carbon capture and
storage network for Yorkshire and Humber. Prepared
by AMEC, Darlington, UK. Available at: www.
yorkshire-forward.com (Accessed: 21 April 2010).
16. A.Rennie, 2010. Understanding the case for shipping
versus pipeline transportation for CCS clusters by
AMEC. CO2 Shipping Conference. London, 6 May.
Carbon Capture Journal. Available at: http://www.
carboncapturejournal.com (Accessed: 15 June 2010).
17. CO2 Sense Yorkshire, 2009. CCS network to
the future report. CO2Sense Yorkshire. Available
at: http://www.co2sense.org.uk/uploads/public/
CCS%20Brochure.pdf (Accessed: 20 April 2010).
18. M.Mohitpour, H.Golshan, and A.Murray,(2007.
Pipeline design & construction: a practical approach.
3rd edn. ASME Press, New York.
19. D.McCollum and J.Ogden, 2006. Techno-economic
models for carbon dioxide compression, transport, and
storage & Correlations for estimating carbon dioxide
density and viscosity. Institute of Transportation
92
Studies: University of California, Davis UCD
ITSRR06-14.
20. HYSYS, 2004. ASPEN HYSYS 2004.2 aspenONE,
AspenTECH, Available at: www.aspentech.com.
21. H.D.Beggs and J.P.Brill, 1973. A study of two phase
flow in inclined pipes. J.Petroleum Technology, 25, pp607617.
22. D.Y.Peng and D.B.Robinson, 1976. A new two-constant
equation of state. Industrial and Engineering Chemistry:
Fundamentals, 15, pp5964.
The Journal of Pipeline Engineering
Bibliography
J.Gale and J.Davidson, 2004. Transmission of CO2-
safety and economic consideration. Energy Progress,
6, 4, p219.
Kinder Morgan Energy Partners, 2002. KMP Annual
Report No 10-K405SEC.
J.Watt, 2010. Transport for captured CO2. First International
CO2 Pipeline Forum. Gateshead, 1-2 July. AMEC, UK.
2nd Quarter, 2013 93

The critical-path method for

assessment of pipelines with
metal-loss defects
by Prof. Jos L F Freire*1, Ronaldo D Vieira1, Pablo M Fontes1, Adilson C
Benjamin2, Luis S Murillo C1, and Antonio C Miranda3
1 Pontifical Catholic University of Rio de Janeiro, Rio de Janeiro, RJ, Brazil
2 Petrobras R&D Center, Rio de Janeiro, RJ, Brazil
3 UnB, Universidade de Braslia, Braslia, DF, Brazil

T HE CRITICAL-PATH method (CPM) proposes a set of rules allowing the drawing of failure lines
that represent adjacent areas positioned along selected circumferential and longitudinal directions of
pipelines that contain colonies of corrosion defects. Failure pressures are calculated for each of those lines
to determine the most critical one.This selected line is considered as the most probable rupture path, and
it corresponds to the minimum calculated internal pressure to take the pipeline to fracture.The proposed
method was checked against 12 burst-pressure tests performed on pipeline tubular specimens. Three
specimens were labelled as control specimens one was a pipe without a defect, and the other two had
single small base defects of different depths. Nine of the specimens contained interacting corrosion defects,
which were composed of the combinations of two or more base defects. Comparisons were made of the
measured burst pressures with those predicted by the CPM, one by the recently proposed method called
MTI, version 1, or MTI V1, and four by other Level-1 or Level-2 assessment methods, namely the American
Society of Mechanical Engineers (ASME) B31G method, the Det Norske Veritas (DNV) RP-F101 for single
and for complex and interacting defects, and the RSTRENG effective-area method. The CPM and MTI V1
methods predicted the failure pressures closest to the actual test failure pressures, with the CPM presenting
a suitably small mean error of evaluation as well as very low standard deviation error for its predictions.

C ORROSION DEFECTS IN pipelines are detected

and sized through metal-loss inspection tools that
run inside the pipelines and contain arrays of non-
cluster are known (maximum cluster depth dclus, and
overall length Lclus and overall width Wclus of the cluster).

destructive evaluation transducers based on ultrasound
(US) or magnetic-flux leakage (MFL) sensors. The
post-run data processing comprises the boxing and
clustering processes. At the end of the boxing process,
the maximum depth (d), length (L), and width (w) of
each individual defect, and the spacing between them,
are known. The longitudinal and circumferential distances
between defects are respectively represented by s and c.
During the clustering process, an interaction rule is
applied to identify the adjacent defects that should be
grouped and considered as a cluster, and examples of
interaction rules are given in Table 1 of Ref.1. At the
end of the clustering process, the dimensions of each
This paper was presented at the International Pipeline Conference held in Calgary
in September, 2012, and organized by the Pipeline Systems Division of ASME.
*Corresponding author:
tel: +55 21 3527 1642
email: jlfreire@puc-rio.br
The Level-1 and Level-2 structural assessment methods
used to determine the allowable pressure capacity
of pipelines with corrosion defects assume that the
longitudinal area of metal loss A is the primary defect
geometry parameter. Level-1 assessment methods represent
A for an isolated defect on the basis of the maximum
defect depth d and the overall defect length L. The
ASME B31G method [2] and the DNV RP-F101 method
[3] for single defects are a couple of the currently
available Level-1 assessment methods.
Level-2 assessment methods were developed for the
assessment of isolated complex-shaped defects, as for
example the RSTRENG effective-area method [4, 5].
This can be used to calculate the failure pressure of a
group of defects by including the length of full wall-
thickness pipe that separates the adjacent defects as
part of the longitudinally projected depth profile of the
complex-shaped defect. Level-2 methods developed for
94
Fig.1. Examples of simple colonies for calculation cases using
the MTI V1 and CP methods.
Fig.2. Free body diagram of case (a) in Fig.1.
the assessment of groups of defects, as for example the
DNV RP-F101, take into account in their formulation
the length of full wall-thickness pipe that separates each
pair of longitudinally spaced defects.
The consequence of projecting the individual profiles of
closely spaced defects onto the longitudinal plane is the
loss of data on the circumferential spacing between the
pairs of defects and therefore the circumferential length
of full wall-thickness pipe existing between each pair
of defects within the colony is not taken into account.
The Journal of Pipeline Engineering
An approach that takes care of the thicker longitudinal
and circumferential spaces existing between the metal-
loss defects was recently proposed and is called the
mixed-type interaction version 1 (MTI V1) method [6].
In the present paper it will be shown that sometimes
the MTI V1 approach can lead to non-conservative
results as a result of taking into account some extra
non-corroded areas in the pipeline remaining strength
calculation. In order to overcome this difficulty, a new
method is proposed in this paper.
The critical-path method (CPM) [7, 8] proposes a set
of rules allowing the drawing of lines across groups
of adjacent metal-loss defects. These lines run along
selected circumferential and longitudinal directions of
the pipeline. Failure pressures are calculated for each
of those lines to determine which is the most critical;
this selected line is considered as the most probable
path of rupture, and corresponds to the minimum
calculated internal pressure to take the pipeline to
fracture. The present paper shows comparisons of
burst pressures calculated by the CPM with results
determined from 12 burst-pressure tests performed on
pipeline specimens. Comparisons were also made of the
CPM-determined burst pressures with those predicted
by finite-element numerical models [8], by the MTI V1
[1, 9, 10] method, and by four other Level-1 or Level-2
assessment methods. These included the American
Society of Mechanical Engineers (ASME) B31G method
[2], the Det Norske Veritas (DNV) RP-F101 for single
and for complex and interacting defects [3], and the
RSTRENG effective-area method [4, 5].
The MTI V1 method
Before presenting the CPM, a brief description is given
of the MTI V1 method. The MTI V1 is based on
geometric data that accurately describes the geometric
distribution of the defects as seen in the top view of
a colony of defects such as those presented in Figs
1-3. For this reason, this method was the only one
among the currently available assessment methods
that could take into account the length of full wall-
thickness pipe that exists between each pair of defects
within a colony of arbitrary shape (a mixed type of
interaction). The four basic assumptions of the MTI
V1 method are:
The failure pressure of a colony of closely
spaced corrosion defects is smaller than or equal
to the smallest failure pressure of the defects
when considered individually.
The failure pressure of a colony of closely
spaced corrosion defects may be smaller than
or equal to the failure pressure of a group of
defects within the colony.
The failure pressure of a group of defects (named
by the superscript k) within the colony can be
2nd Quarter, 2013 95

Groups of defects interacting in the cluster shown in Fig.1

Calculation case Parameter MTI V1 CP

Defect 1 Defect 2
Defects 1 & 2 Defects 1 & 2

Wgroup W1 W2

Lgroup L1 L2
Fig.1a

dgroup d1 d2

Wgroup W1 W2

Fig.1b
Lgroup L1 L2

dgroup d1 d2

Wgroup W W W

Fig.1c Lgroup L1 L2

dgroup d1 d2

Wgroup W1 W2 W2

Fig.1d Lgroup L1 L2

dgroup d1 d2

Table 1. Calculation of parameters using equations of methods MTI V1 and CP for the cases shown in Fig.1.

calculated using the failure pressure equation of
a Level-1 method (the DNV RP-F101 equation
for a single defect is used herein). It considers
that the dimensions of the combined defect are
the overall length Lkgroup of the group k and the
effective depth dkgroup of the group k.
The effective depth dkgroup of a group k of defects
within the colony can be calculated using the
total volume of metal loss of all (i) defects within
the group (k), LiWi di divided by the total area
i
delimited by the box that circumscribes all the
defects inside the group Lkgroup Wkgroup.
The MTI V1 uses the following Equns 1-4 to calculate
the failure pressures of all isolated defects (groups of
only one defect), and all possible groups generated
by the combinations of adjacent defects within the
colony; and to determine the smallest among all the
calculated pressures to be the failure pressure of the
colony of defects.
96 The Journal of Pipeline Engineering

Fig.3. Free-body diagram for a group

of three defects in order to calculate
the equivalent metal-loss depth using
equilibrium of forces.

Fig.4. Possible critical path (CP) lines

for a group of three defects where
the metal-loss depth of defect 2 is
smaller than the metal-loss depths of
defects 1 and 3.
2nd Quarter, 2013 97

Groups of defects interacting in the cluster shown in Fig 3

Parameter Method Individual Defects Defects Defects

defects 1 & 2 1 & 2 & 3 2 & 3 Defects 1 & 3
i = 1,2,3

MTI V1 Wi W1 + W2- + c12 W1 + W2 + W3 + c12+ c23 W2 + W3 + c23

Wgroup
- - - - L 2 + L '3
CP

MTI V1 No feasible interaction

L1 + L2 + s12- L1 + L2 + L3 + s12 L2 + L'3 L2 + L'3
Lgroup Li in the present cluster
CP

MTI V1 di
dgroup
CP

Table 2. Calculation of parameters using equations of methods MTI V1 and CP for the cases shown in Fig.3.

Fig.5. Schematic of dimensions given

for the groups of defects.The defects
inside a group have the same metal
loss depth d, same length L, and same
width W. Longitudinal distances s and
circumferential distances c may have
negative values if the defects overlap .The
upper defect, where the label length L is
placed, is centred in relation to the two
defects positioned below.

dgroup
k

1 As mentioned above, the MTI V1 method was presented,

2t t
pburst group Sflow
k

D t dgroup
k (1) explained, and tested in other previous articles [1, 6,
1 9, 10]. For completeness of the present paper, and to
tM k
group
help the presentation of the CPM, the MTI V1 method
will be applied to two simple examples represented by
2
Lkgroup (2) the colonies of defects depicted in Figs 1-3.
M group
k
1 0.31
Dt

Three groups of defects were analysed for the colonies
depicted
in Figs 1a to 1d: the two groups formed by
Vgroup k LW d i i i
(3)
dgroup
k
i each of the two isolated defects and the group formed
L
k
group .W k
group Lk
group .W k
group by both defects 1 and 2. The results obtained for
the group length Lgroup and for the group equivalent
metal loss depth dgroup, to be used in Equns 1-4,
pburst min( pburst group )
k
(4) are given in the fifth column of Table 1, under the
title MTI V1. For each calculation presented in Table
98 The Journal of Pipeline Engineering

Defect depth Sy Yield Sy Ultimate Burst Normalized

Specimen Group of defects to pipe strength strength Pressure burst
configuration (2) thickness ratio (MPa) + (MPa) + 3% (MPa) + pressure
d/t 3% 0.5%

IDTS 1 No defect 0 556 698 26.6 1.0

IDTS 2 0.665 22.7 0.868 (3)

IDTS 3 0.657 20.3 0.896

IDTS 4 0.694 21.1 0.932

IDTS 5 0.669 601 684 20.9 0.920

(4)

IDTS 6 0.665 18.7 0.823

IDTS 7 0.649 18.8 0.828

IDTS 8 0.469 24.2 0.867 (3)

IDTS 9 0.481 23.1 0.953

IDTS 10 0.476 23.2 0.960

589 731
(5)

IDTS 11 0.476 21.3 0.879

IDTS 12 0.460 20.2 0.833

1. Nominal pipe dimensions: D = 457mm, t = 7.9mm. Actual pipe and defect dimensions are
given in Table 4
2. Rectangles with shadow localize the base defects that failed
Notes 3. Normalized with respect to IDTS 1 result, corrected by the ratio of the tube's ultimate
strength
4. Normalized with respect to single defect - IDTS 2 result
5. Normalized with respect to single defect - IDTS 8 result

Table 3. Summary of geometric and material data, and burst pressures of the colonies of defects tested.

1, three burst pressures corresponding to three defect
groups (defect 1, defect 2, and defects 1+2) may be
determined. The failure burst pressure of each case
is the minimum pressure of the three burst pressures
calculated for each of the three groups.
The calculated parameters corresponding to the pairs
of defects shown in Figs 1a to 1d, and the way these
defects were displaced for each case, help in the
discussion of some of the features of the MTI V1
method. Comparison of cases 1a and 1c shows that
2nd Quarter, 2013 99

Diameter t* d (1) L W s (2) c (2) dclus Lclus Wclus dclus

Specimen and (mm) (mm) (mm) (mm) (mm) (mm) (mm) (mm) (mm)
thickness t

IDTS 1 457 No defect included

7.93

IDTS 2 2.71 5.39 39.6 31.9 - - 5.39 39.6 31.9 0.665 0.42

IDTS 3 2.78 5.32 39.6 31.9 20.5 -31.9 5.32 99.7 31.9 0.657 2.67

IDTS 4 458.8 2.48 5.62 39.6 32.0 -39.6 9.9 5.62 39.6 73.9 0.694 0.42
8.1
IDTS 5 2.68 5.42 39.5 32.1 -9.5 10.0 5.42 69.5 74.2 0.669 1.30

IDTS 6 2.71 5.39 39.6 32.2 20.5 9.6 5.39 99.7 115.8 0.665 2.67

IDTS 7 2.84 5.26 39.5 31.9 20.4 10.0 5.26 99.4 73.8 0.649 2.66

IDTS 8 4.25 3.75 40.05 32.00 - - 3.75 40.05 32.00 0.469 0.44

IDTS 9 4.15 3.85 40.05 32.15 -9.88 9.88 3.85 100.39 116.21 0.481 2.74
459.4
IDTS 10 8.0 4.19 3.81 40.00 32.08 19.98 10.01 3.81 99.98 116.26 0.476 2.72

IDTS 11 4.19 3.81 40.06 32.11 19.84 10.03 3.81 160.02 116.39 0.476 6.97

IDTS 12 4.32 3.68 40.04 32.14 20.06 9.99 3.68 280.44 116.40 0.460 21.40

Notes (1) d = t t* (2) Negative values for s or c mean that the projection of the defects
overlap.

Table 4. Actual dimensions of the tubular specimens and of the machined defects.

group 1+2 of case 1c is more critical than group 1+2 by Benjamin and Cunha [11], during the MTI JIP
of case 1a. The reason for that is the existence of (mixed-type interaction joint-industry project). The MTI
the circumferential space c between the defects 1 and V2 method will be published after the confidentiality
2 in case 1a. period of the JIP.

Comparisons between cases 1b, 1c, and 1d show possible
non-conservative results that the MTI V1 method will
furnish. The reason for that is the amount of extra
material (corresponding to the shaded areas in Fig.1)
that does not suffer metal loss but is included in the
calculation of cases 1b and 1d. In fact, in what concerns
the three cases (1b, 1c, and 1d) of interactive defects
1+2, the MTI V1 method will determine increasing
burst pressures in the sequence given by 1c, 1d, 1b.
Nevertheless it is expected that all three cases give the
same result (equal to the one calculated for 1c) due to
the fact that both defects are crossed by an expected
common-longitudinal-straight fracture line. In summary,
it should not be expected that the actual pressure to
failure would increase if defect 2 became wider. It
should be noted that an improved version of the MTI
method, named MTI Version 2 method, was developed
The critical-path method
The CP method was developed with the purpose of
overcoming the difficulties of the MTI V1 method cited
in the above paragraphs. As for the MTI V1 method,
the CPM looks to the pipe surface from above and
considers the circumferential distance c between two
defects. If necessary, the CPM also considers the defect
remaining thickness if the defect lies in the fracture
critical path.
The equations and rules of the CP method are very
similar to the MTI V1 method, with the difference that
the former calculates the equivalent metal-loss depth
using the equilibrium of forces in the circumferential
direction of the thin pipe. Equations 1, 2, and 4 are
kept the same for both methods, but calculation of the
 100 The Journal of Pipeline Engineering

Table 5. Material data for the finite-element analysis of the API 5L-X80 steel pipe.

equivalent metal-loss depth for a given critical path uses and therefore, for a long defect M (6a)
Equn 5. This equation is derived using the equilibrium
of forces in the circumferential direction. Derivation 1& 2 2t L1d1 + L2 d2 1
ct
pburst group Su . 1
2

of equilibrium for the critical path established for the D t L 1 L2 s .t
case of a group of two defects (1+2) interacting such (7)
d 1&2
2t
Su . 1
group
as showed in Fig.2 (free body of Fig.1a) leads to:
D t t


1 .L1 . t d1 s .s.t c .c.t + 2 .L2 . t d2
(5) Thus, for the case of a group formed by two defects:
c .( L1 L2 s).t
L1d1 + L2 d2 - 12 ct (8)
( d ) group =
1& 2

Considering a thin-thickness pipe and a factor f = 0.5 L 1 +L2 + s

to take into account the Tresca criterion, the burst
pressure and the pipe wall stresses along the critical It can be seen from the above that only the dimension
path at the failure instant will be: c (and not the defect widths W1 and W2) of the
circumferential gap, influences the strength of the pipe
1& 2 2t
pburst group c . in the case of Fig.1.
D t

1 s 2 S flow S u
(6) The results for dgroup calculated using the CPM for the other
S flow Su
f.S flow cases of Fig.1 are also given in Table 1, resulting in the
c 2 2
same value of dgroup for cases 1b and 1c. Thus, the influence
2nd Quarter, 2013
Specimen
IDTS 2
Table 6. Results for the single-
defect specimens (MPa). IDTS 8
of the extra amount of material, included by the MTI V1
method in those cases, is not felt when the CPM is applied.
Calculation of the equivalent metal-loss depth of a critical
path of a group of three interacting defects 1+2+3, as
depicted in Fig.3, shows a case where the width of a
defect that lies along the critical path helps to increase
the failure pressure (by decreasing the equivalent metal-
loss depth). Assuming that the metal loss of defect 2
(d2) is larger than the metal loss of defect 3 (d3), the
equilibrium of forces in the circumferential direction for
the 1+2+3 interacting group of defects gives:
L1d1 L2 d2 L '3 d3 12 c12t + c23t + W2 1 2 t CP method are very similar to the rules used to
(9)
1 & 2 & 3
d group
L 1 L2 s12 L '3
where L3 is the length of the critical path along defect
3. In this case it was taken into consideration that part
of the longitudinal projections of defects 2 and 3 overlap
and that defect 2 is deeper than defect 3. Therefore L3
= L3 L23, where L23 is the ovelapping length.
The application of the CPM for the simple cases
shown in Figs 1, 2, and 3 are summarized in Tables
1 and 2. Table 2 shows, besides the values developed
for the application of the MTI V1 method, the groups
of defects needed for the complete analysis to assess
and to find the minimum pressure that will represent
the burst pressure for both the MTI V1 and the CP
methods when three defects are present. At this point
it is advisable that a computer program be developed
so that all possible groups of defects inside a colony
(with a large number of defects) are considered in the
calculations of their corresponding critical paths.
The general form of the equation used to determine the
equivalent metal-loss depth for a given critical path is:
d defects that overlap in the circumferential
L d + L'
i i d j 12 cmt + Wn 1 n
j
d group
k i j (10)
i r j
L
i
r
s
j
L '
where:
k = the defect group and its critical path line
i = the defects that lay along the path and have
the full length considered
101
Test ASME DNV RP-F101 Finite CPM
MTI V1
result B31G (single) elements
22.7 21.1 22.0 22.4 22.0 22.0
24.2 21.2 24.6 24.5 24.6 24.6
j = the defects along the path that overlap
with others onto the longitudinal projection
and have smaller metal-loss depths
m = the circumferential spaces of full thickness
between adjacent defects along the circumferential
portion of the path
n = the defects that are crossed circumferentially
(along their widths) by the considered groups
critical path
r = the longitudinal spaces between adjacent
defects that lay along the critical path
Summarizing, the set of rules or steps to apply the
d
t
apply the MTI V1 method, with the only modification
being in the equation used to calculate the equivalent
metal-loss depth. The rules are:
Establishment of the colony of defects to be
assessed.
Prediction of the failure pressures of individual
defects using the single defect DNV RP F-101
Equn 1, where dgroup and Lgroup respectively
correspond to the metal-loss depth and to the
length of each isolated defect belonging to the
colony being analysed.
Selection of all combinations of possible groups of
adjacent defects and creation of one critical path
for each group. In order to create the possible
critical path of a group of defects, three simple
rules are given below and are illustrated in Fig.4.
The rules are:
the critical path (CP) lines have minimum
possible lengths;
the CP lines give preference to the defects
with the largest metal-loss depth in case these
defects would overlap in the longitudinal
projection; and
t t
direction are crossed by a straight CP line.
calculation of the equivalent metal-loss depth
using Equn 10 for of all possible CP lines.
prediction of the failure pressures of all possible
CP lines using Equn 1.
choice of the smallest among all the calculated
pressures to be the failure pressure of the
colony of defects using Equn (4).
102 The Journal of Pipeline Engineering

Fig.6.The actual geometric arrangements of defects before and after the burst tests.The most probable fracture critical path
(smallest calculated burst pressure) indicated by the CPM is represented by the dashed lines. Finite-element solutions at the
numerical burst pressure are represented by contour plots of the von Mises equivalent stresses.
2nd Quarter, 2013
Fig.7. Plot of the ratios predicted
to actual failure pressures.
Tables 1 and 2 present and summarize data sets of the
12 laboratory burst tests, nine of them with interacting
corrosion defects, and compare the test results with the
CPM, the MTI V1 method, and the traditional Level
1 and 2 approaches.
Experimental methods and results
The actual dimensions and mechanical properties of
the steel pipes used, as well as the geometric details of
the defects arrangements, are shown and summarized
in Tables 3 and 4, and in Fig.5. References 1, 9,
and 10 present more details of the experimental tests
and results. The raw material used in this research
was composed of three longitudinally welded tubes
made of API 5L X80 steel [12]. The length of these
pipes was approximately 12 m. The nominal outside
diameter and the nominal wall thicknesses of the
pipes were, respectively, 457.2 mm (18 in) and 7.93
mm (0.312 in).
Charpy V-notch impact tests at a temperature of
0oC for 2/3 thickness specimens cut from the tubes
used led to a mean impact energy equivalent for full
thickness specimens above 82 J.
Twelve tubular specimens 1.7 to 1.9 m long were
manufactured. The actual specimens outside diameters
D and wall thicknesses t are given in Table 4.
A summary of the defect configurations introduced
into the tubular specimens and the nominal basic
defect dimensions are given in the second and
third columns of Table 3, in Table 4, and in Fig
5. Specimen IDTS 1 was a defect-free pipe. All the
defects were machined using spark erosion and were
smooth rectangular defects, i.e. the shape of the area
of metal loss was rectangular with smooth edges, which
were made with a small radius of 3.2 to 3.5 mm.
The groups of defects were machined opposite the
103
tube seam weld at a centralized position in relation
to the ends of the specimens.
The defects belonging to the same group of defects
have the same remaining wall thickness t*, the same
maximum depth d, the same length L, and the same
width w. The cluster maximum depth dclus is equal to
the maximum depth of the deepest defect pertaining
to the group of defects; the cluster length Lclus is
the overall length of the group of defects; and the
cluster width wclus is the overall width of the group of
defects. For each group of defects, distances s and c
are presented in Table 4 and are depicted in Fig 5.
The ratios of cluster maximum depth dclus to wall
thickness t are presented in column 12 of Table 4.
From the values of these ratios it can be seen that all
the clusters have a maximum depth close to 50-70%
of the wall thickness. Ratios ( Lclus )2 / ( D.t ) are presented
in the last column of Table 4. By the values of these
ratios, it can be seen that all the clusters are short
according to the ASME B31G method (defects in
which L 20 D. t ), except the cluster pertaining to
Specimen IDTS 12.
Finite-element calculations
All specimens were also numerically tested using the
finite-element method (FEM), a detailed description
of which may be found in Ref. 8. The FE analysis
used four layers of elements across the pipe model
thickness along with the SOLID 95 element of the
ANSYS software [13].
Solutions used the calculated true material properties
given in Table 5. After yielding, very small increments
of pressure were ascribed to the FE analysis. Close
to fracture, defined as the first element to reach the
true ultimate material strength, increments of pressure
were small as 10-5 of the total applied pressure.
104 The Journal of Pipeline Engineering

DNV RP- RSTRENG DNV RP-

ASME Finite
Specimen F101 effective F101 MTI V1 CPM
B31G elements
(single) area (interacting)

IDTS 3 -14.2 -19.2 -16.6 -5.8 -3,3 -5.8 -5.8

IDTS 4 -1.1 2.5 -8.8 - 1.7 2.5 2.5

IDTS 5 -9.9 -10.4 -24.2 - 1.8 5.2 -4.3

IDTS 6 -7.1 -13.2 -27.4 - 1.0 2.0 -2.0

IDTS 7 -6.6 -11.6 -25.7 - 1.2 2.6 0.0

IDTS 9 -19.0 -9.9 -26.8 - 2.3 6.2 0.1

IDTS 10 -19.3 -10.1 -26.4 - -2.5 -4.1 -3.8

IDTS 11 -17.3 -12.9 -28.5 - 0.8 4.8 -1.8

IDTS 12 -39.6 -16.8 -29.6 - -1.4 0.4 -1.8

Mean
-14.9 -11 -24 -5.8 0.2 1.5 -1.9
deviation*

Standard
11 6.1 6.8 - 2.0 4.1 2.5
deviation*

Table 7. Deviation* of the failure pressure predictions (%).

* Deviation(%) = 100% (calculated experimental ) / ( experimental )

Mean deviation = ( deviation ) / number of data

Pressure-test results Performance of the assessment

Details of the monotonic hydrostatic internal pressure
tests are given in Refs 1, 9, and 10. Figure 6 presents
the photographs of the specimens before and after
failure and shows the defects that failed individually
or as a group. Some symmetric behaviour of necking
inside the defects can be noticed. Table 3 (in the last
two columns) presents the burst-test pressures measured
during the laboratory tests of the tubular specimens.
From the burst-pressure values it can be concluded
that all specimens with more than one base defect
presented interactive failures. The reason for this is
the lower burst-pressure tests measured consistently for
these specimens when compared to the burst pressures
of the specimens with only one base defect. The last
column of Table 3 presents the ratios between the
pressures for these specimens.
methods
Table 6 presents the comparison of results of burst
pressures determined by the experimental tests of
specimens IDTS 2 and IDTS 8 against their finite-
element solutions, and predictions of the commonly
used single-defect MTI V1 and CP methods. It is
clear that agreement among the results is quite good.
It should be noticed that, in these cases, the actual
materials strengths were used in the calculations
instead of the API 5L minimum specified strengths.
In Table 7 and Fig.7, the failure pressures measured in
the laboratory tests are compared with those predicted
by six assessments methods, namely the ASME B31G
method, the DNV RP-F101 method for single defects
(Part B), the RSTRENG effective-area method, the
2nd Quarter, 2013
DNV RP-F101 method for interacting defects (Part B),
the MTI V1 method, and the newly presented CPM.
Seven of the nine failure pressures predicted by the
MTI V1 method were above the test failure pressure.
However, the maximum values of the ratios of the
predicted pressures to the actual failure pressures were
1.05 for specimen IDTS 5 and 1.06 for specimen IDTS
9. This slight non-conservatism was created by the MTI
V1 calculation of a minimum pressure related with one
isolated defect. On the other hand, the CPM correctly
predicted the interaction of the two defects calculating
the lowest burst pressure for the group that contains
the two defects instead of the group that contains one
isolated defect. Calculations using the CPM lead to
adequate results such as a conservative result of -4.1%
for specimen IDTS 5 and a slight non-conservative
result of 0.15% for specimen IDTS 9.
The prediction of the fracture critical path made by
the CPM shows quite good results as can be seen
in Fig.6. This figure presents the actual geometric
arrangements of defects before and after the burst
tests. The most probable fracture critical path (smallest
calculated burst pressure) indicated by the CPM is
represented by the dashed lines. In only one case
(IDTS 10) did the predicted critical path not coincide
with the actual one, although in this case the CPM
still predicted a conservative value (3.8%). It also must
be commented that the calculated pressures for the
most critical paths of a colony may give results that
may be very close. For example, in the case of IDTS
10, the difference of pressure results between the
predicted case (22.342 MPa) and the one calculated
for the actual failure path (22.279 MPa) is insignificant
and smaller than the resolution and accuracy of the
experiment and data used.
Table 7 presents the deviations of the failure pressure
predictions with respect to the measured burst pressures
for the nine tubular specimens. The best accuracy
that one method could achieve would be a deviation
equal to 0%. However, predictions between 5% can
be considered quite accurate because of the scatter
that always exists in any set of experimental results.
A note on expected uncertainty of the experimental
results and predicted results must be given at this
point. Experimental burst pressures indicated by the
manometer are expected to be 0.1MPa accurate, or
about 0.5% of the measured burst pressure. Uncertainty
analysis performed with the CPM equations indicated
total deviations less than 5%. The uncertainty analysis
took into consideration the total geometric uncertainty
equal to 1% (most of it concentrated on the defect
thickness measurements) and uncertainty of the
ultimate strength for measurements made for each
tubular specimen, given by a total deviation of 3%
in the present case.
105
Finite-element results presented a very good agreement
with the test results, with the mean deviation and
standard deviation respectively equal to only 0.2% and
2.0%. Of course, well-designed and -executed FE solutions
are nowadays very reliable, but they still are expensive
and limited to be applied to a small number of cases.
Considering the Level-1 methods, the DNV RP-F101
method for single defects predicted the failure pressures
closest to the actual failure pressures. The DNV RP-F101
method presented the smallest mean deviation (11.8%),
followed by the B31G method (14.9%). Regarding the
Level-2 methods, the CPM and the MTI V1 method
predicted the failure pressures closest to the actual failure
pressures. The MTI V1 method presented the smallest
mean deviation (1.5%), followed by the CPM with -1.9%,
DNV RP-F101 method for interacting defects (-5.8%)
and the effective-area method (-24%), the positive and
negative signs representing respectively non-conservative and
conservative mean deviations. The CPM was the method
that was more effective in terms of standard deviation.
Conclusions
This paper presented and evaluated a new method to be
used for predicting the failure pressure of pipelines that
contain colonies of corrosion defects, called the critical-
path method, or CPM. The paper proposes a set of rules
allowing the drawing of lines that represent adjacent areas
along selected circumferential and longitudinal directions
of the pipeline. Failure pressures are calculated for each
of those lines to determine the most critical one. This
selected line is considered as the most probable path of
rupture, and it corresponds to the minimum calculated
internal pressure to take the pipeline to fracture. The
proposed method was checked against 12 burst-pressure
tests performed on pipeline tubular specimens. Nine of
the specimens contained interacting corrosion defects,
which were composed of the combinations of two or
more base defects. Comparisons were made of the
measured burst pressures with those predicted by the
CPM, by the MTI V1 method, and by four other
Level-1 or Level-2 assessment methods, which comprised
the American Society of Mechanical Engineers (ASME)
B31G method, the Det Norske Veritas (DNV) RP-F101
for single and for complex and interacting defects, and
the RSTRENG effective-area method. The CP and MTI
V1 methods predicted the failure pressures closest to
the actual failure pressures, and the CPM presented
suitable small conservative mean error of evaluation
(1.9%) as well as a very low standard deviation error
for its predictions (2.5%).
References
1. A.C.Benjamin, J.L.F.Freire, and R.D.Vieira, 2007.
Part 6: Analysis of pipelines containing interacting
corrosion defects. Experimental Techniques, 31, 3, May-
June, pp 74-82.
106
2. ASME, 1991, 2009. ASME-B31G: Manual for
determining the remaining strength of corroded
pipelines: a supplement to ANSI/ASME B31 Code for
pressure piping. The American Society of Mechanical
Engineers, New York.
3. DNV, 1999. Corroded pipelines: recommended
practice RP-F101. Det Norske Veritas, Hvik, Norway.
4. J.F.Kiefner and P.H.Vieth, 1989. A modified criterion
for evaluating the remaining strength of corroded
pipe. Final report on project PR 3-805, Pipeline
Research Committee, American Gas Association,
Washington, DC.
5. Idem, 1990. Evaluating pipe conclusion: PC program
speeds new criterion for evaluating corroded pipe.
Oil & Gas Journal, 88, 34, pp 91-93, 20 August.
6. A.C.Benjamin and D.J.S.Cunha, 2006. New method
for the assessment of colonies of corrosion defects.
The Journal of Pipeline Integrity, September, 145-161.
7. P.M.Fontes, 2011. A new method to predict the
rupture of pipes with corrosion defects. Engineering
Project Monograph, Mechanical Engineering
Department, Pontifical Catholic University of Rio
de Janeiro, in Portuguese.
The Journal of Pipeline Engineering
8. C.L.S.Murillo, 2011. Evaluation of a new method
for predicting the burst pressure of pipes containing
colonies of corrosion. MSc Dissertation, Mechanical
Engineering Department, Pontifical Catholic
University of Rio de Janeiro, in Portuguese.
9. A.C.Benjamin, J.L.F.Freire, R.D.Vieira, J.L.C.Diniz,
and E.Q.Andrade, 2005. Burst tests on pipeline
containing interacting corrosion defects. 24th Int.
Conf. Offshore Mechanics and Arctic Engineering,
OMAE 2005. The American Society for Mechanical
Engineers, New York, June.
10. A.C.Benjamin, J.L.F.Freire, R.D.Vieira, and
E.Q.Andrade, 2006. Burst tests on pipeline containing
closely spaced corrosion defects. 25th Int. Conf.
on Offshore Mechanics and Arctic Engineering,
OMAE 2006. The American Society for Mechanical
Engineers, New York, June.
11. A.C.Benjamin and D.J.SCunha, 2009. MTI Method
Version 2. Private Communication, 19 June.
12. API, 2004. Specification 5L: specification for line
pipe. 43rd Edn, the American Petroleum Institute,
Washington, DC, USA.
13. Software and user manual, v12, Ansys Inc.
DONT MISS
AN ISSUE!
SUBSCRIBE TODAY

Purchase your Pipelines International subscription online at

pipelinesinternational.com/subscribe
 Isnt it time you were in the

Pipeloop is the global directory of pipeline professionals. Getting into the

Pipeloop is the quickest and most effective way of promoting your presence,
your skills, your product, your service in the pipeline sector.
Like other networking sites, users create an account and build a profile
outlining their skills as well as their experience. They can also search for other
users using over 160 combinations of specialties and materials. The chief
benefit of Pipeloop is its unique focus on the pipeline sector.
Launched on Valentines Day this year, Pipeloop will bring the pipeline
community together like never before.
So, if youre looking for expert welding in Wyoming, want to sell inspection
tools in India, need a recruiter in Rio, or might be hiring graduates in Gabon
its time you got in the Pipeloop, the global directory of pipeline professionals.

SIGN UP TODAY FOR A FEE LISTING AT

WWW.PIPELOOP.COM

Pipeloop, brought to you by

2nd Quarter, 2013
10-6 and all that: what do failure
probabilities mean? A response
by Dr Andrew Francis
AFAA Ltd, Ripley, UK
I REFER TO the article [1] entitled 10-6 and all that:
what do failure probabilities mean? which appeared in
the December 2012 edition of the Journal of Pipeline
Engineering, and which was subsequently also published
in Pipelines International [2].
The article begins with the assertion that pipeline
failure probabilities, determined using pipeline reliability
analysis, are nominal and are unrelated to real failure
probabilities.
Based on the above statement and with the concession
that it might be politically unacceptable to dismiss the
ideas of reliability analysts completely a suggestion
appears to be offered in the article that the tags of
the offending numbers should, perhaps, be prefixed
with qualifiers in the form of weasel words such as
nominal or pseudo. However, it also appears to be
mooted that, whether or not such a stance is taken, it
does not really matter because the numbers produced
by reliability analysts are never likely to be used as a
basis for any decision. The article goes on to qualify
this statement by drawing attention to a further
assertion that most design code users would not even
trouble themselves to undertake such an analysis, let
alone use the numbers as a basis for any decision and,
moreover, that the most unusual and conscientious
user who might take up the challenge would possibly
abandon his efforts when he discovered there was no
honest way forward.
The sentiments conveyed thus appear to be that whilst
the numbers produced by reliability analysts should be
treated with some caution, and perhaps carry a health
warning, we need not concern ourselves too much with
this, since the numbers are never likely to be used as
a basis for any decision.
I choose to respond to this article for two reasons:
firstly it struck me that decisions might have already
been made based on such analyses; and secondly I
was not overly convinced by the justification of the
assertion which the article provides.
Authors contact details:
tel: +44 1773 513 366
email: andrew.francis@afaa.co.uk
109
Evidence
The evidence provided in the article appears to take two
forms: a citation of a case in which an organization,
presumably of some standing, appeared to make a
mistake, and a brief expos of a view of the validity
of the statistical analyses that are undertaken. I choose
only to challenge the latter.
The article shows a plot of a probability density
function of a strength parameter with annotation
illustrating the argument that sufficient data will never
be available to justify the behaviour of the distribution
in the lower extreme tail. The assertion is then made
that [the argument] can be trivially extended to multiple
components of strength.
It was not entirely clear to me what was meant by the
latter but I could share some empathy with the idea
when applied to a single component. However, in my
opinion, this alone is not sufficient to support the
assertion being made.
Challenge
To establish some common ground, I considered the
properties of X60 steel for which a mean of 448 MPa
and a standard deviation of 17 MPa could, perhaps, be
regarded as representative statistics. Of course, different
estimates would be obtained from different data but
(although a relevant matter) this is beyond the scope
of this note. My chosen values indicate that the SMYS
(414 MPa) is two standard deviations below the mean.
Assuming that a normal distribution can be considered
valid over a domain that extends at least a couple of
standard deviations on either side of the mean, these
figures indicate that there is roughly a 2% chance
that the actual yield strength, of a randomly selected
specimen of the material, will be less than the SMYS.
If we choose to believe that the distribution is valid
over a wider domain, then we will infer, from our
tables, that there is a 1 x 10-5 probability of a value
of (<376 MPa), a 1 x 10-6 probability of a value of
(<367 MPa), a 1 x 10-7 probability of a value of (<360
Mpa), and so on.
110 The Journal of Pipeline Engineering

It is not clear to me whether such values can simply Suppose that the wall thickness and yield strength are
be dismissed as meaningless, but if hard proof were normally distributed quantities, we can then write:
required that a value corresponding with a probability
of 10-6, for example, had been observed previously, (w w )2
1
p(w ) =
exp
it would not be available. If such a low value was 2 w
(2) 2 w
found during a set of 100 or so tests, say, it is most
likely that a reason would be sought to justify why
it was not representative of the population. (This and
does raise issues for the reliability analyst but these
are way beyond the scope of this note and hence 1 ( )2 (3)
p( ) = exp
are discussed no further here.) 2 2

It would therefore make little sense to offer a counter
argument to that being made in the article, and
I would concur that assuming the distribution to
be valid over a domain that extends beyond three
standard deviations or so on either side of the mean,
is somewhat futile.
which are assumed to be valid expressions over the domains
( w 3 w , w + 3 w ) and ( 3 , + 3 ) , respectively.
Here, m and denote the mean and variance, and the
subscripts identify the distributed variable to which
they apply. (Well throw the chopped-off bits away
because we dont believe they mean anything anyway.)

There is thus no reason to disagree with the article Since the resistance is equal to the product of the
when talking about a single distribution, so lets wall thickness and yield strength, it follows that the
impose a restriction that the domain of any individual valid domain is given by (Rmin, Rmax) where:
distribution, for which valid data are required as
justification, must not extend beyond three standard = ( w 3 w )( 3 )
Rmin (4)
deviations on either side of the mean.
and
We now need a case study.
( w 3 w )( 3 )
Rmax (5)
Pipeline integrity in its simplest
form (Ive just multiplied the two valid lower extremes and
the two valid upper extremes together).
In very basic terms, pipeline integrity is achieved by
ensuring that the inequality: We also know that the mean and variance of the
resistance are given by:
PD
f (1)
2w
R w (6)

is always satisfied, where f is the design factor. and

The upshot of this is that, at the design stage,
the pressure, P, and diameter, D, are determined R w w2 2 w
(7)
from operational requirements (and constraints), and
the wall thickness, w, and material strength, d, We can now determine the domain of R in terms
are selected to ensure that the pipeline will resist of multiples (n) of standard deviations on either side
the imposed load. It is thus not unreasonable to of the mean, i.e. ( R nlower R , R + nupper R ) , where:
(loosely) refer to the product PD as a load, and the
product, dw, as a resistance. Use of Equn 1 normally w + w 3 w

requires the use of suitably conservative values of nlower 3
(8)
w w2 2 w
yield strength and wall thickness (SMYS and wmin),
but were talking probabilities here so well look
at distributed quantities. (Note: Im using a simple and
limit-state function to make a point; Im not saying
that Equn 1 is all that is necessary.) w w 3 w
nupper 3
(9)
w w2 2 w
For our case study, we shall determine the distribution
of the resistance subject to our imposed restriction We immediately deduce that the valid domain of
on the extent of the domains of the underlying distribution of the resistance is not symmetrical about
distributions. the mean.
2nd Quarter, 2013
Fig.1. Combination of two valid distributions.
For completeness, the distribution of resistance can
be computed using:
w + 3
1 w
1 ( R / w )2 +
p( R ) = exp w dw
2 w w 2 w
w 3 w
which is valid for values of belonging to the domain
(Rmin, Rmax). The situation is illustrated pictorially in
Fig.1.
Suppose that wall thickness has a mean of 12.7 mm
and a standard deviation of 0.25 mm. Using the
mean and standard deviation of the yield strength
given above, it follows from Equns 8 and 9, that
nlower = 3.89, and nupper = 4.20.
It is immediately obvious that although we have
restricted the domains of the underlying distributions
(the ones supported by data) to three standard deviations
either side of the mean, the domain of distribution
of interest is significantly wider. The probability that
resistance exceeds the upper limit is about 2 x 10-5,
and the probability that it is less than the lower limit
is about 3 x 10-5. It thus follows that, in rough-and-
ready terms, tails corresponding with probabilities
of the order of 10-5 can be justified. Basically, this
represents a reduction by two orders of magnitude of
the probabilities that can be justified by the parent
distributions.
I chose to use the product simply to illustrate my
point. In reality, a more complex algebraic relationship
involving several distributed variables would arise from
111
a particular limit-state function. In these situations it
naturally follows that, with more variables, probabilities
of the order of 10-6, and lower, can be easily justified.
(w w )2
(10)
Summing up
Quite clearly the above is a very simple illustration,
but the point being made is that extremes of tails of
relevant distributions and very low failure probabilities
can be justified. This, of course, does not mean that a
sensible and justifiable answer will always be obtained;
theres still plenty of scope to get things wrong. It
does however, make provision for the possibility that
justifiable failure probabilities can be computed and
that they might have been in the past. The corollary is
that any decisions that may have been made previously,
based on use of the technique, need not necessarily
be the subject of concern.
Despite, me seeing fit to challenge the tail-sensitivity
issue, in my opinion, Prof. Palmers article is very
timely as it draws attention to the possibility that
misuse and misinterpretation could be damaging the
image of a very powerful technique and, if that was
the case, I think it would only be to the good of
the industry if the situation was to change.
References
1. A.C.Palmer, 2012. 10-6 and all that: what do failure
probabilities mean?. J.Pipeline Engineering, 11, 4.
2. Idem, 2013. Pipelines International, 15, March.
 Held under the Patronage of His Excellency Shaikh Ahmed bin Mohamed Al Khalifa, Minister of Finance,
Minister in Charge of Oil and Gas Affairs, Chairman of National Oil & Gas Authority, Kingdom of Bahrain

2023 October 2013, Bahrain

GULF CONVENTION CENTRE, BAHRAIN
ORGANIZERS

Global
W


Webb

SILVER SPONSORS
PLATINUM ELITE SPONSOR

Join leaders in the international pipeline industry as they converge for the Best Practice in Pipeline
Operations and Integrity Management Conference and Exhibition in Bahrain.

CONFERENCE EXHIBITION
Technical streams presented by industry leaders covering a wide A comprehensive exhibition will be part of the event, allowing
range of subjects will run over the two and a half day event. companies from around the world to showcase their products and
Some of the subjects to be discussed; services. Visit our website to book your space.
Planning, design, construction and materials
Operations and maintenance NETWORKING
Asset integrity management Throughout the event there will be ample opportunities to network
Inspection and cathodic protection with participants to further your business relationships. Meet with
Repair and rehabilitation industry leaders from around the world.
Automation and control
Leak detection
Paper abstracts are now being accepted.

T E R N OW
REGIS

www.pipelineconf.com
2nd Quarter, 2013 113

SliPIPE: a new concept to deal

with pipeline expansion
by Chia Chor Yew and Asle Venas*
Det Norske Veritas, Hvik, Norway

F INDING EASY and conventional sources of hydrocarbons has become harder, while global demand for
hydrocarbon products continues to grow. Oil and gas operators have turned to new geographic areas
to tap new resources. These areas are challenging, and can be remote, in harsh environments or in deep
water, where high-pressure and/or high-temperature (HPHT) reservoirs are often found. Transporting the
oil and gas by flowlines and pipelines from these HPHT reservoirs is a major challenge. SliPIPE is a new
concept developed to deal with the end expansion of a rigid pipeline subject to HPHT.

A PIPELINE laid on, or buried in, the seabed

responds to high pressure and high temperature
by expanding against the frictional resistance from the
soil and other restrictions, resulting in axial forces,
axial displacement (also known as end expansion),
lateral buckling, upheaval buckling, or a combination
of these, depending on whether the pipeline is partly
or fully restrained, or unrestrained.
In some cases, pipeline walking may occur after the
pipeline in operation is cooled down, for example in
a shut-down, heated-up for operation, and then the
thermally cycled repeatedly.
These pipeline movements and forces can cause failures
in the midline or at the tie-ins connected to the pipeline
end, and are critical to the integrity of a pipeline.
When a pipeline is subject to high pressure and high
temperature, its ends expand longitudinally and exert
forces and bending moments onto adjacent tied-in
structures or pipes to which it is connected. The tied-
in structures and pipes must be designed to withstand
these expansions and loads. Dumping rocks along the
pipeline has conventionally been adopted to reduce end
expansion, and a giant spool installed at the pipeline
end is another alternative, and these are often used
in combination to eliminate end expansion when it is
very large. However, such post-lay intervention work is
costly and requires long offshore time to accomplish.
Thus, a major challenge is to improve on the ways
the pipeline end movements can be controlled ways
which are simple, safe, and cost-effective. SliPIPE offers
a new possibility for doing this.
*Corresponding author:
tel: +47 9360 5054
email: asle.venas@dnv.com
The concept
SliPIPE works to reduce the axial force exerted at
a tie-in by absorbing the end expansion through
sliding within itself, and simultaneously reducing or
eliminating the effective axial compressive force in
the pipeline.
The new system consists of a sliding mechanism that
absorbs the temperature expansion as well as a pressure
chamber that, together with the sliding mechanism,
absorbs pressure-induced expansion such that the
expansion forces can in principle be eliminated.
The mechanism consists of an outer pipe connected
alongside a pressure chamber and an inner pipe that
can slide within them. Seals are placed at the contacts
between the pressure chamber and the inner pipe, and
the inner pipe slides in or out of the outer pipe in
response to an axial stress that can either be more or
less than a certain value. This value is predetermined
in the SliPIPE design and causes an axial tension in
the pipe wall to develop, which opposes the effective
axial compressive force component arising from the
inner fluid pressure.
The axial tensile pipe-wall force is produced by
allowing fluid pressure into one side of the pressure
chamber through holes in the inner pipe, separated
from the other side of the pressure chamber by an
annular partition wall. As the pressure in that side
of the chamber freely builds up, it pushes against
the partition wall and the pressurized end of the
chamber in opposite directions to one another until
an equilibrium is reached. This, in turn, develops a
tensile force in the pipe wall which can be scaled to
a desired value by pre-sizing the cross-sectional area
of the pressure chamber.
114
Fig.1.The main characteristics of the SliPIPE.
Between the outer pipe/pressure chamber and the inner
pipe of the SliPIPE are two main seals, a partition
wall seal, an environmental seal, and a scraper seal.
Each main seal consists of a pair of chevron seals
and T-seals with back-up rings, capable of preventing
a single failure from causing the loss of both barriers.
Other equivalent double barrier seals may be used.
Around the rim of the annular partition which moves
within the pressure chamber is a set of double T-seals;
seal is reinforced with back-up rings on either side,
and these provide efficient resistance to extrusion of
the seals. The seals are made of materials that allow
them to function at high temperatures up to 150oC
and pressures between 100 and 400 bars. Chevron
seals are made from thermoplastic material, and the
T-seals from elastomer.
The environmental seals and scraper seals remove
marine growth and other contamination on the
surface of the inner pipe before it makes contact
with the main seal.
Before use, all seals must be first qualified for HPHT
conditions and long-term reliability of the seals to
function under frequent two-directional sliding of the
surfaces that come in contact with them.
The Journal of Pipeline Engineering
Several practical issues that will influence the operation
of the SliPIPE were studied and feasible ways to
overcome them looked into, including:
The double seals at the annular partition wall
are relied upon to keep the differential pressure
between the non-pressurized compartment and
the pressured compartment of the chamber.
As a safeguard against the pressures on either
sides of the partition wall equalizing over time,
a one-way relief valve, suitable for underwater
application, can be installed at the far corner of
the non-pressurized compartment through which
any built-up pressure in the compartment can be
vented into the sea. Alternatively, a small pipe
connecting the non-pressurized compartment to
a faraway existing flare-off facility, if available,
will produce the same effect.
Assembling the components and seals together to
create the SliPIPE is feasible by casting, forging,
and welding and assembling the components
in a certain production sequence. Avoiding
damage to the seals from the heat generated by
welding parts together is achievable by careful
selection of the locations for welding.
Inspection of the seals after final factory
acceptance test can be undertaken by modifying
2nd Quarter, 2013
the free end of the pressure chamber to
become a pair of flanges with a metal seal
between them, one flange connected to the
chamber body and the other connected to the
chamber end. The flanges can be unbolted
to disassemble the components for inspection
of the seals. The metal seal located between
the flanges is then replaced with a new one
before the SliPIPE is reassembled.
The reliability of the environmental and scraper
seal to clean the surface of the inner pipe
that comes into contact with the main seal is
crucial for the main seal to hold against leak.
This can be improved by extending the free
end of the pressure chamber with an external
tubular housing with a tight-fit end and long
enough to shield the contactable inner pipe
surface from fouling.
Installation
The new system used for absorbing end expansion can
be pre-installed on a PLET which is then transported
and installed offshore to the end of the pipeline,
lowered onto the seabed, and connected to a manifold
or a riser via a short tie-in spool. A misalignment
flange may be included.
Alternatively, a direct tie-in (without a PLET and
short tie-in spool) is also feasible with the use of a
suitable installation guide currently available on the
market, such as a subsea installation guide (SIG).
The SIG is placed on the subsea structure close to
the connection point, and guides the pipeline end
towards a hub on the subsea structure until they are
separated by a small gap. The SliPIPE is then allowed
to slide until the small gap is completely closed and
the connectors clamped together. No post-installation
metrology, fabrication of the short spool, additional
spool installation, or subsea tie-in need be performed.
In the direct tie-in method, SliPIPEs has to be locked
115
to restrict any uncontrolled movement, and the lock
released before tie-in.
SliPIPE must be designed to have at least the same
capacity as the adjacent linepipe, which itself has already
been designed to resist the maximum tensile forces
and bending moments.
Application
The new concept is well-suited for installing tie-ins
between a submerged rigid pipeline and a subsea well,
subsea structure, or riser, typically from 10.75 in to 24
in (273 mm to 610 mm) in diameter, with operating
temperatures up to 150oC and a pressure range from
100 to 400 bars
In comparison to a giant tie-in spool, SliPIPE is a relatively
simple yet effective alternative means to eliminate the
effects of end expansion on tie-in structures. The key
advantages of the new technology are that it:
avoids the fabrication and complicated installation
associated with giant spools;
minimizes costly post-installation subsea
intervention work;
is space-efficient, ideal in areas congested with
many subsea facilities, as are often encountered in
brownfield modification work, where safeguarding
their integrity during intervention work can be
formidable.
Currently, SliPIPE is conceptual and will require
refinement and engineering through basic and detailed
design before it can be implemented on an actual
project. A global team of experienced engineers, headed
by DNV in Singapore, has developed the concept. It
has also taken into account comments received from
the offshore pipeline industry, in particular from two
major installation contractors, a seal company, and a
university professor.
2nd Quarter, 2013 117

Assessment of vintage girth

welds and challenges to ILI tools
by Dr Yong-Yi Wang*1, Dr Jing Ma1, and Satish S Kulkarni2
1 Center for Reliable Energy Systems, Dublin, OH, USA
2 Consultant, USA

A ssessment of vintage girth welds on the basis of fitness-for-service (FFS) principles is the best rational
approach to develop integrity-management practice.Although general procedures for the FFS assessment
of welds are well established, enhancement of these procedures may be necessary to take into account
the unique features of vintage girth welds. In addition, the values of necessary input parameters for the FFS
assessment, such as applied stress, flaw size, and material properties are often not available.

Elements necessary for the enhancement of FFS procedures are explored in this paper. Challenges of
estimating input parameters are examined. The limits and potentials of various ILI tools in detecting and
characterizing girth weld flaws are presented based on the fundamental principles of tools and published data.

A LARGE PORTION of pipelines in the United

States was installed prior to the enactment of federal
regulations in 1970s. The girth welds of these pipelines
were not 100% non-destructively inspected at the time
of construction. These welds may contain flaws that
could lead to occasional failures when additional stresses
beyond those stresses under normal operation conditions
are present. Recent accidents associated with vintage
pipelines elevate the public attention and highlight the
needs of integrity management of these pipelines. Due
to the low frequency and high consequence of those
failures, an effective management plan is necessary to
direct the limited resources to the welds of the highest
potential risk.
Many vintage pipelines were constructed without the
benefits of non-destructive testing (NDT) and the stringent
workmanship requirements of today. Welds made prior
to 1970s were not inspected 100% by NDT methods.
Consequently, some weld flaws were undetected and
left in the completed pipelines. These welds can also
have low toughness.
The operational history of these vintage pipelines has
demonstrated that the vast majority of the vintage
girth welds are safe for the intended service. This safe
This paper was presented at the Pipeline Pigging and Integrity Management
conference held in Houston in February 2013, and organized by Clarion
Technical Conferences and Tiratsoo Technical.
*Corresponding author:
tel: +1 614 808 4872
email: ywang@cres-americas.com
operation history is, in a major part, attributable to
the low longitudinal stresses experienced by the girth
welds of typical buried pipelines. In the absence of
ground-movement hazards or other upsetting conditions,
such as compressor station by-pass or unusual weather
conditions, the highest stresses on the girth welds for
onshore pipelines are typically in the pipe lowering-in
stage. After passing the construction phase, the undetected
flaws can remain dormant for decades without any
obvious negative consequences.
Failures of girth welds are usually associated with
additional stresses imposed on the welds beyond the
historically nominal level of stresses. The disturbance to
the pipe-support conditions would subject the welds to
longitudinal loads or bending moments which results in
high longitudinal tensile stresses. The disturbance of the
support conditions may come from ground settlement,
landslide, soil creep, or nearby construction activities.
The other source of additional stresses is the thermal
stress when the pipeline experiences colder-than-usual
temperatures due to compressor-station by-pass, unusually
cold temperature, etc. Identifying conditions of potential
additional stresses is a major goal of integrity management.
Significance of integrity management
Despite the overwhelmingly good safety records of the
pipeline industry, a pipeline is only as good as the
weakest point. There could be thousands of sound
welds, while one bad weld is sufficient to disrupt the
entire system. In some cases, one failure could prompt
regulatory actions and/or public outcry.
118
Vintage pipelines account for a large portion of
transmission pipelines in the US. More than 50%
of overall mileage was built prior to 1950 and more
than 75% of overall mileage was built prior to 1970
[1]. It is clearly in the interest of all stakeholders to
identify and mitigate high-risk factors before a failure
could occur.
Principles of fitness-for-service
assessment
General procedures to conduct fitness-for-service (FFS)
assessment of engineering structures are well-established.
In performing FFS assessments, fracture-mechanics
principles are used to correlate three major parameters
in the state of a postulated failure event: (1) applied
stress, (2) material toughness, and (3) flaw size. When
any of the two parameters are known, the critical value
of the third one can be determined. In most cases,
a safety factor is applied to the output to cover the
uncertainty of the input parameters.
For in-service pipelines, the FFS procedures may be
applied in one of two ways. The first one is comparing
the critical flaw size (the maximum flaw size at the
point of incipient failure) with the estimated flaw
dimensions (that can come from ILI indications or
other inspection methods). If the estimated flaw size is
smaller than the critical size, the girth weld is deemed
safe. The second approach is comparing the maximum
permissible stress/strain at the point of incipient failure
with the anticipated applied stress/strain on the weld. If
the anticipated stress/strain is lower than the maximum
permissible stress/strain, the weld is deemed safe for
continued operation.
The Journal of Pipeline Engineering
Fig.1. Charpy transition curves of
a SMAW girth weld made in early
1950s on an X52 pipe. Each data
point represents the averaged value
of three individual test specimens.
Stresses on pipeline girth welds
Stresses in pipes
Stresses in a pipe or pipeline may come from multiple
sources. The stresses in a pipe start at the pipe-
forming stage, and pipes may experience stresses during
transportation. There could also be residual stresses
from field bending. The pipe-forming and field-bending
stresses are self-balanced and should not affect girth
welds, while the stresses during transportation are typically
within the elastic range. These stresses return to their
original state at the completion of the transportation.
Stresses affecting girth welds in the construction phase
Stresses affecting girth welds start at the girth-welding
phase. Due to the shrinkage of deposited weld metal
during cooling of the weld, there is a local bending
created by the welding residual stress. For pipes of large
D/t ratio, the longitudinal residual stress (transverse
to the welding direction) is in tension on the ID side
of the pipe and in compression on the OD side. The
hoop residual stress is in tension.
After the completion of mainline welding, the next
phase is usually pipe lowering-in and, for most onshore
pipelines, girth welds experience the highest longitudinal
stress during this phase of operation. The lowering-in
stress is typically in the range of 40% to 90% SMYS for
an X70 pipe. The lowering-in stress is primarily driven
by the contour of the pipe profile, i.e. lift height and
spacing between the sidebooms [2]. The magnitude of
the stress is not related to pipe grade as the material
behaviour is in the elastic range.
2nd Quarter, 2013 119

Task Inspection tool Inspection principle Comment

Measures local diameter ,dents, buckles and

Geometry Caliper tool Mechanical ovalities; detects girth welds, wall-thickness
inspection (deformation tool) deflection changes, and installations (such s mainline
valves, tees, etc.)

Inertial navigation,
Measures angles and velocity changes in X, Y,
3-D position Mapping tool gyroscopes and
and Z coordinates between reference locations
accelerometers

Standard-resolution
(SR) MFL
Magnetic-flux
Liquid and gas liens; indirect measurement
leakage
Metal-loss High-resolution (HR)
inspection MFL

Compression wave
Ultrasonic tool for wall-thickness Requires liquid couplant; direct measurement
measurement

Magnetic-flux
Transverse MFL SCC detection
leakage

Crack inspection Requires liquid couplant; both axial and

45o shear wave
circumferential cracks
Ultrasonic tool
EMAT guided
Field-testing stage
waves

Table 1. Overview of ILI tools.

During a tie-in welding, the ends of pipes may have
to be moved for alignment. Tie-in welds are often
made near crossings and bends, and the varying pipe
stiffness may cause stresses to be concentrated at the
tie-in welds and, in addition, there could be differential
settlement near these locations. Similarly, the bedding
or support at tie-in welds could be different from that
for the mainline welds; all these conditions can lead
to high stresses at tie-in welds.
After backfill, stresses on the girth welds may come
from spanning the pipeline support or bedding, and
uneven support can lead to bending moment along
the pipeline. During hydrostatic testing, a longitudinal
tensile stress is generated due to the Poissons effect.
At the maximum pressure corresponding to a hoop
stress of 100% SMYS, the longitudinal stress from
the hydrostatic pressure is 30% SMYS. Consequently
hydrostatic pressure alone is not a good way to stress
girth welds sufficiently to expose their flaws. This stress
level is based on the assumption that the pipe at the
time of hydrostatic testing has the same temperature as
that at the time of pipe lowering-in. Any temperature
differential between hydrostatic test temperature and
lowering-in temperature can introduce additional stresses.
Stresses during pipeline service
Stresses during normal pipeline service may come from
a number of sources:
1. Lock-in stresses from the pipeline construction.
2. Internal pressure: a tensile longitudinal stress is
generated from the Poissons effect.
3. Thermal stress: it is generally assumed that a
pipe is in a near-stress-free state when it is laid
at the bottom of a trench. The pipe material
would generally be at a different temperature
during service. The temperature difference
between creates a longitudinal thermal stress.
If the service temperature is lower than the
construction temperature, a tensile longitudinal
stress is generated.
4. Ground movement: the ground movement includes
all forms of movements that tend to impose
certain motion on a pipeline, including soil
creep, landslide, settlement, earthquake, etc. The
magnitude of the stresses and strains imposed
by the ground movements can vary greatly. In
contrast, the magnitude of the stresses from
the first three sources listed above has limited
120
ranges. Consequently, ground movement can be
a major determining factor that can generate
sufficiently high stress to rupture a girth weld.
5. Unsupported pipe span or insufficient negative
buoyancy: an exposed pipe section or a section
in a body of water that is not sufficiently
weighed down can move laterally or vertically.
Such movement can produce both static and
cyclic stresses on the girth welds.
Determination of stresses of in-
service pipelines using ILI tools
Currently the inertial navigation systems integrated
with global-positioning system (GPS) technology has
revolutionized pipeline centreline mapping. The inertial
measurement unit (IMU) can provide pipe centreline
coordinates, bending strain [3], position on the maps
integrated with geographical-information system (GIS) [4],
and the pipe movement between runs [5].
The major advantage of IMU is that it can provide the
total bending strain accumulated since construction for
the entire line when the reference state at the time of
construction is known. Pipeline displacements calculated
from the successive runs provide a direct monitoring of
the pipeline movement and the influence of unstable
geotechnical conditions such as land slide, subsidence,
soil creep, river crossing, etc. [6]. Traditional strain
measurements such as strain gauge, slope indicator, etc.,
only give the differential strain since their installation
and usually only cover a portion of pipeline with known
threats [7].
There are some uncertainties with IMU data interpretation
related to the reference state which usually assumes
the pipeline is straight at the time of construction.
Engineering judgment must be applied to differentiate
the post-construction stress from the construction stress.
The Journal of Pipeline Engineering
Fig.2. Charpy transition curves
of a mechanized GMAW girth
weld made in late 2000s on an
X65 pipe. Each data point is
from one single specimen test.
If a reference curvature can be estimated, for example
through an as-built IMU run right after the construction,
the strain in the pipe developed during the service can be
inferred by taking out the construction-strain component.
It is also worth noting that the stress from bending
moment is only a part of aggregate stress acting on
the girth welds. Some other stresses coming from the
uniform longitudinal tension or compression, such as those
from internal pressure and/or temperature differential,
are not captured by using curvature data from IMU.
Similarly if a uniform centreline strain is applied due
to soil movement along the pipe longitudinal direction,
such strain cannot be captured through the IMU runs.
Material properties
The principal material properties of interest for assessing
the girth welds of in-service pipelines are the tensile and
toughness properties. The tensile properties include the
yield strength and UTS of pipe and weld metal. The
weld-metal property is relevant to the determination of
weld-strength mismatch. In many cases, the weld-metal
property is not available and a hardness traverse may
be used to estimate the relative strength between the
pipe and the weld metal.
For vintage girth welds, the most readily available
toughness parameter is the Charpy impact energy. Both
ductile-to-brittle transition temperature and the energy
level at the pipeline service temperature are critical
indicators of materials toughness. From the viewpoint
of weld integrity, the transition temperature is just as
important as the energy level at a given temperature.
However, most weld-property specifications focus more
on specific energy level at a given temperature. The
Charpy transition curves of a SMAW weld made in
the early 1950s and a mechanized GMAW weld made
in the late 2000s are shown in Figs 1 and 2 [8]. The
2nd Quarter, 2013
transition temperature of the mechanized GMAW weld
is lower than that of SMAW weld. The upper-shelf
energy of the SMAW weld is in the range of 50-60
ft-lbf; however, the same energy level would put the
GMAW weld on the lower transition. The upper-shelf
energy of the GMAW weld is in the range of 160-180
ft-lbf. This comparison clearly demonstrates the need to
understand the transition behaviour in addition to the
energy level at a given test temperature.
Detection of girth weld anomalies
by ILI tools
Autonomous in-line inspection (ILI) tools, commonly
referred to as intelligent pigs, are extensively used to
examine a large portion of long-distance transmission
pipelines without stopping the flow of products.
Through implementing the electromagnetic and ultrasonic
modalities, these inspection tools are capable of locating
and characterizing the anomalies of pipelines such as
mechanical damage, corrosion, cracks, gouges, wrinkle
bends, etc. The possibility to find a particular anomaly
depends on a number of factors, including the anomaly
size and orientation, pipeline condition such as material
variation, proximity to pipeline features, construction
practices, and debris and deposits, etc. [9, 10]. It is
important to realize that there are no universal intelligent
pigs able to detect and size all laws. A given type of
inspection tool is designed to look for a specific type
of defects based on particular technology (see Table
1) [11-13].
Magnetic-flux leakage (MFL) is the most common
inspection technology for in-service pipelines. The MFL
tool is the first generation of metal-loss survey tool,
with its high-resolution version becoming available in
1980s. When the pipeline steel becomes saturated
with magnetism, some flux leaks out through a metal
loss; the strength and shape of the leaking flux give
indications of the features dimensions. In addition
to detecting metal loss, advanced high-resolution MFL
tools are claimed to be able to define features such as
dents, buckles, wrinkles, and mechanical damage [13, 14].
Although designed for corrosion-like flaws, traditional
MFL tools have shown to be able to find some large
circumferential planar flaws by some vendors, provided
that the opening of the planar flaws is sufficiently large.
Transverse MFL tools were introduced in 1999 to detect
anomalies oriented in the pipes longitudinal direction,
and are able to detect and quantify longitudinally oriented
defects such as SCC, seam-weld defects, mechanical
damage, and groove corrosion [13, 15].
Ultrasonic technology (UT) has been practiced in
wall-thickness measurement since the mid-1980s. The
compression sound pulse produced by a piezoelectric
transducer propagates through the pipe wall and bounces
back individually from the inside and outside surfaces.
121
Fig.3. A girth weld with high-low misalignment and surface-
breaking weld centreline flaw from the ID side.
This time-of-flight interval directly gives the measure of
the wall thickness. UT angle-beam pigs were developed
for crack-like flaw detection in 1994 [16]. Compared to
MFL, UT provides more sensitive and stable capability
for crack inspection, especially for flaws with small
crack openings. However, a liquid couplant between
the piezoelectric transducer and the pipe wall is needed
to assist the ultrasound transmission. The coupling
becomes difficult to achieve at high pig velocity. UT
tools cannot be used for inspection of gas pipeline
unless the tools are operated in a liquid batch.
More recently, electro-magnetic acoustic transducer
(EMAT) technology has been introduced and overcomes
the limitations in UT by generating sound waves
directly inside the pipe wall. The concept of EMAT
has been considered as an inspection technology since
1970s. It works on a distinct set of physical principles
compared to MFL and UT [17]. There are two basic
components in an EMAT transducer: a static magnet
and an electric coil. The coil, placed near the surface
of the pipe wall, is driven by an alternating current
at a desired ultrasonic frequency. Ultrasound waves,
such as horizontal-shear (SH) waves and Lamb waves,
are generated through the Lorenz force experienced by
the eddy-current or the magnetostriction effect. The
couplant-free feature provides EMAT an ideal transducer
for gas pipeline inspection. Additionally, due to the
versatility of the magnet and coil combination, the
EMAT tool can be designed to excite complex wave
patterns and polarizations which would be difficult to
realize with fluid-coupled piezoelectric probes. EMAT
can be used for thickness measurement, flaw detection,
and material property characterization [18].
Historically, the inefficiency of the electromagnetic
induction and the inadequate understanding of guided-
wave propagation hindered the EMAT development.
Gao et al. provided a good summary of the EMAT
evolution and its current performance [19]. Recently
TransCanada PipeLines conducted case studies of two
20-in natural gas pipelines which showed that EMAT has
122
become a reliable and accurate method for detection,
identification and sizing of SCC cracks [20]. To the best
knowledge of the authors, EMAT tools specifically for
girth-weld-crack detection are still under development.
FFS assessment of vintage girth
welds
Inputs to FFS assessment
The determination of flaw size for in-service pipelines
can be quite challenging. For girth welds, the most
detrimental defects are surface-breaking planar flaws.
The tight surfaces of a planar flaw make it very hard
to be detected through traditional pigging facilities such
as MFL tools. The lack of radiographic and ultrasonic
records during the original construction hamper the
estimation of flaw-size distribution of vintage pipelines.
Effects of the weld cap
Other factors that can affect FFS assessment are the
weld-strength mismatch and the weld reinforcement. The
flaw depth of vintage girth welds found in the field
can vary greatly. There can be root flaws extending
well over 50% of the pipe-wall thickness with the
weld remaining intact, in a large part due to the large
weld-cap reinforcement. The ability of a weld surviving
a relatively deep and finite-length flaw can at least be
partially explained by the example below.
A girth weld with a relatively large weld cap, approximately
20% of the pipe wall thickness, is shown in Fig.3. A
finite-length surface-breaking flaw from the ID side is
located at the weld centreline. An elastic finite-element
analysis was conducted by imposing a longitudinal
The Journal of Pipeline Engineering
Fig.4. Normalized stress-
intensity factor as a function
of the relative flaw height for
flaws of finite length.
tension on the pipe. The stress-intensity factor, KI,
was computed at the deepest point of the flaw. The
primary variable is the flaw depth (which is the same
as flaw height, as this is a surface-breaking flaw).
The stress-intensity factor as a function of the relative
flaw depth (flaw depth / wall thickness) is shown in
Fig.4. As the relative flaw depth increases, the stress-
intensity factor increases first: after the flaw depth
reaches 60% of the wall thickness, the stress-intensity
factor starts to decrease for the shorter flaw (with
length / wall thickness = 3.9); for the longer flaw, the
peak of the stress-intensity factor is reached when the
flaw depth is approximately 80% of the wall thickness.
The most important feature is that the stress-intensity
factor does not increase monotonically with the flaw
depth as one might expect. The underlying mechanism
for the reduction in the stress-intensity factor after
a certain flaw depth is the pattern of load or stress
transfer from one side of the flaw to the other. When
the flaw depth exceeds a certain level, the deepest
point of the flaw starts to move away from the path
of the load/stress transfer. Indeed, when the deepest
point of the flaw moves into the weld-cap area, the
driving force at the crack tip is very low as the weld
cap is not an effective load/stress transfer path. This
is only true when the flaw length is finite. The peak
of the stress-intensity factor is reached at a small flaw
depth when the flaw length is short, as shown in Fig.4.
The trend shown in Fig.4 can translate to a reduced
flaw growth rate in a fatigue-loading environment
after the flaw depth has attained a certain value.
A flaw may stop growing when it reaches the weld
cap when the stress-intensity factor has reached a
sufficiently low level.
2nd Quarter, 2013
Fig.5. Flaw-acceptance curves
at two levels of applied
longitudinal stress with an
assumed surface-breaking
flaw depth of 3 mm.
Examples of FFS assessment
The principles of fitness-for-service can be applied
to determine the acceptance of girth-weld anomalies
in vintage girth welds. There could be a variety of
anomalies, from the relatively benign buried volumetric
imperfections to the most detrimental surface-breaking
planar imperfections, such as hydrogen cracks. The
precise identification of the nature of all the possible
girth weld anomalies is generally beyond the capabilities
of current ILI tools. Consequently, the conservative
approach is to treat all anomalies without proper
identification as surface-breaking planar flaws.
The tools available for detecting girth-weld anomalies
can be either MFL- or UT-based. In general, MFL
tools are incapable of providing flaw depth at a
sufficiently high accuracy for case-specific FFS analysis.
UT-based tools in principle are capable of providing
the length and depth of planar flaws.
A few examples are shown here to illustrate the
possible application of FFS assessment procedures
to vintage girth welds. These are purely sample
problems, and should not be used for actual field
applications.
The examples are divided into two groups. In the
first group, it is assumed that the flaw depth cannot
be determined on a case-by-case basis, such as in the
case using MFL. A reasonable assumption of flaw
depth is used to develop the flaw-acceptance criteria
in terms of flaw length. In the second group, it
is assumed that the flaw depth and length can be
determined on a case-by-case basis; the flaw-acceptance
criteria are therefore given as a continuous function
of flaw depth and flaw length.
123
The basic pipe criteria for the sample problems are
as follows:
diameter: 30 in
wall thickness: 0.330 in
grade: X52
girth weld reinforcement (cap) height: 0.0625
in (1/16 in)
The procedures described in Option 2 of API 1104
Appendix A [21] are used to generate the flaw-acceptance
criteria. Since the CTOD toughness is generally not
available for vintage girth welds, the CTOD toughness
values used in the Appendix A calculation are estimated
from established Charpy to CTOD conversion relations [22].
The critical flaw lengths as a function of Charpy impact
energy at two levels of applied longitudinal stress are
shown in Fig.5. The flaw depth is assumed to be 3
mm, or approximately one weld bead height. Option
2 of API 1104 Appendix A requires a safety factor
of 1.5 on the flaw length for new-construction welds.
In the current example, no safety factor is applied
because the safety factor for vintage girth welds can
be different, and the decision on an appropriate value
can be made at a later time. Without such a safety
factor, the critical flaw length is 211 mm, or 8.8%
of the pipe circumference at the applied longitudinal
stress level of 85% SMYS for a weld with a Charpy
impact energy of 50 ft-lbf.
The critical flaw lengths as a function of Charpy impact
energy at three levels of applied longitudinal stress are
shown in Fig.6, in which a flaw depth of 6 mm, or
72% of wall thickness, is assumed. The critical flaw
length is 60 mm, or 2.5% of the pipe circumference
124
at the applied longitudinal stress level of 85% SMYS
for a weld with Charpy impact energy of 50 ft-lbf.
When both flaw depth and length can be determined,
the critical flaw dimensions may be presented in a
format similar to that of Fig.7. For a given applied
longitudinal stress level, 70% SMYS in this case, the
critical flaw size is given as an envelope between flaw
depth and flaw length. Similar to Figs 5 and 6, no
safety factor is applied in the critical flaw size shown
in Fig.7.
The flaw-acceptance criteria may be used to determine
the acceptance of a particular flaw found in a weld.
If the size of the flaw is smaller than the acceptable
flaw size for a given set of material properties and
applied longitudinal stress, the flaw is considered
acceptable for the intended service condition. If the
size of the flaw is greater than the acceptable flaw
size, remedial action has to be taken, and these may
include one of the following:
relieving the applied stress on the girth weld,
such as excavating and re-routeing the pipeline;
installation of pressure-containing reinforcement
sleeves; or
cut out the weld and make new welds that can
safely sustain the expected stress on the welds.
Effectiveness of hydrostatic testing in exposing
poor-quality girth welds
If an X52 pipeline operating at 72% SMYS is
hydrostatically tested to 90% SMYS, the additional
longitudinal tensile stress imposed on the girth weld
is 2.8 ksi (0.18 52 0.3 = 2.8), or 5.4% of SMYS.
By contrast, every 5C (9F) drop of pipe temperature
would produce an additional longitudinal tensile stress
The Journal of Pipeline Engineering
Fig.6. Flaw-acceptance curves
at three levels of applied
longitudinal stress with an
assumed surface-breaking
flaw depth of 6 mm.
of 2.0 ksi, or 3.8% SMYS. A temperature drop of
7C (12.6F) therefore can be seen to produce the
same magnitude of additional longitudinal tensile
stress as the hydrostatic pressure test. Clearly, a large
temperature drop in a pipe can produce a much
greater longitudinal tensile stress than a hydrostatic
pressure test.
Concluding remarks
Vintage girth welds, referring to welds made prior
to the early 1970s, appear in a large portion of
transmission pipelines in the United States. These welds
have an overwhelmingly good safety record, although
occasional failures have occurred over the years at a
very low frequency. The challenge to pipeline owners
and operators is to identify the very small number of
welds that may lead to failures, and intervene before
such failures can occur.
This paper focuses on the capabilities of ILI tools in
detecting and sizing girth-weld flaws and establishing
acceptable flaw sizes using fitness-for-service principles.
A framework for the assessment and acceptance of
vintage girth-weld flaws is shown. There are a number
of practical challenges in the application of such
methodology, including the capability of ILI tools,
availability of material property data, and the ability
to estimate the longitudinal stresses on girth welds of
in-service pipelines. These challenges notwithstanding,
there is clearly a pathway to develop and implement a
rational approach that is technically justified, operationally
feasible, and financially not overly burdensome.
Acknowledgment
The authors are grateful to the insightful discussions
with Steve Rapp and William Webster of Spectra
2nd Quarter, 2013
Fig.7. Flaw-acceptance curves
at two levels of Charpy
impact energy at an applied
longitudinal stress level of
70% SMYS.
Energy. We are very encouraged by the willingness
of ILI vendors to share their understanding of the
capability of their tools, and by their enthusiasm in
developing new capabilities to address the needs of
assessing vintage girth weld integrity.
References
1. B.Leis, 2009. Vintage pipelines. PHMSA R&D
Forum, June.
2. M.Liu, Y.-Y.Wang, and G.Rogers, 2008. Stress analysis
of pipe lowering-in process during construction. Proc.
7th International Pipeline Conference, IPC2008-64630,
Calgary, AB, Canada.
3. P.Michailides and T.Deis, 1998. NPS 8 Geopig: inertial
measurement and mechanical caliper technology.
ASME International Pipeline Conference, Calgary,
AB, Canada.
4. J.A.Czyz and C.Pettigrew, 2000. Multi-pipeline
geographical information system based on high
accuracy inertial surveys. Proc. International Pipeline
Conference, Calgary, AB, Canada.
5. A.Lockey and A.Young, 2012. Predicting future
pipeline integrity in landslides using ILI mapping data.
Proc. 24th Pipeline Pigging & Integrity Management
Conference, Houston, TX, USA.
6. J.A.Czyz and S.E.Waiselboin, 2003. Monitoring
pipeline movement and its effect on pipe integrity
using inertial/caliper in-line inspection. Rio Pipeline
Conference & Exposition.
7. J.A.Czyz and E.McClarty, 2004. Prevention of
pipeline failures in geotechnically unstable areas by
monitoring with inertial and caliper in-line inspection.
Corrosion, NACE, New Orleans, LA, USA.
8. Y.-Y.Wang, M.Liu, W.Tyson, J.Gianetto, and D.Horsley,
2010. Toughness considerations for strain-based
design. Proc. 8th International Pipeline Conference,
IPC2010-31385, Calgary, AB, Canada.
125
9. M.Beller, K.Reber, and U.Schneider, 2002. Tools,
vendors, services a review of current in-line
inspection technologies. Pigging Products and Services
Association.
10. Pipeline Operators Forum, 2009. Specifications
and requirements for intelligent pig inspection
of pipelines.
11. H.Willems, B.Jaskolla, T.Sickinger, O.A.Barbian,
and F.Niese, 2010. Advanced possibilities for
corrosion inspection of gas pipelines using EMAT
technology. NDT.net.
12. J.B.Nestleroth, 2011. Nondestructive pipeline
inspection using pigs 101. In-Line Inspection
Symposium.
13. NACE, 2010. SP0102 (formerly RP0102): Standard
practice in-line inspection of pipelines. NACE
International.
14. R.Rempel and C.Hallam, 2006. The evolution of
MFL. World Pipelines.
15. J.B.Nestleroth, 2003. Circumferential MFL in-
line inspection for cracks in pipelines. Grant/
Cooperative Agreement No. DE-FC26-01NT41159,
prepared for the Department of Energy.
16. K.Reber and M.Beller, 2003. Ultrasonic in-line
inspection tools to inspect older pipelines for
cracks in girth and long-seam welds. Pigging
Products and Services Association.
17. H.Al-Qahtani, T.Beuker, and J.Damaschke,
2008. In-line inspection with high resolution
EMAT technology crack detection and coating
disbondment. 20th Pipeline Pigging, Integrity
Assessment, and Repair conference, Houston, TX,
USA.
18. R.E.Green, 2000. Advances in nondestructive
materials characterization. NDT.net, 5, 6.
19. M.Gao and S.T.Krishnamurthy, 2011. Evaluation of
EMAT tool performance by monitoring industry
experience. 18th Joint Technical Meeting on Pipeline
126
Research, PRCI, San Francisco, CA, USA.
20. J.Marr, R.Kania, G.Rosca, R.Ruda, E.S.J.Riverol,
R.Weber, S.Klein, N.Jansing, T.Beuker, and
N.D.Ronsky, 2012. Combining EMAT ILI and
multiple datasets for crack detection in natural
gas pipelines to reduce validation costs. Proc.
The Journal of Pipeline Engineering
24th Pipeline Pigging & Integrity Management
Conference, Houston, TX, USA.
21. API, 2007. Standard 1104, 20th Edition, Errata.
22. A.C.Bannister, 1998. Determination of fracture
toughness from Charpy impact energy: procedure
and validation. SINTAP Sub-Task 3.3 Report.
2nd Quarter, 2013 127

Threat and the probability

of failure
by Randy Vaughn
IM Programme Lead, Rail Road Commission of Texas, Pipeline Safety Division, Houston, TX, USA

I N THE formula for determining a pipeline segments total risk value (total risk value = probability of failure
x consequence of failure), every effort needs to be made to identify what threats are likely before even
assuming a value for a probability of failure. This is also true before deciding on an integrity-assessment
method to be used for any given pipeline system or segment.The Texas integrity-management programme
(IMP) rules have been in place since April, 2001, with each pipeline having a baseline assessment and, for
most, at least one integrity reassessment completed by now. Certainly each pipeline system should by now
have a well-defined risk value with identified threats, along with the preventative and mitigative measures
identified and in place to diminish those threats and reduce the pipelines overall risk, and all should be right
with the pipeline world. However, pipeline incidents are still occurring, including a few recent incidents of
pretty good magnitude. It would seem that some of us still do not completely understand how a good IMP
should be developed and implemented.

Possible reasons for continued A good foundation provides support

incidents
One of several possible reasons for the continuing
incidents could be attributed to having some operators
include the incidents root-cause-analysis conclusions
as a big component for determining a pipelines risk
(for those operators who continue to have incidents)
and adding it to the final risk-analysis results as a
known threat that had not been considered before.
And unfortunately the flip-side of this is that some
operators that are having incidents have actually
somehow determined that the incident was an isolated
situation, thus concluding that there is no need for
any further integrity concerns or considerations for
this newly-identified threat found to have initially
caused the incident. This is most evident in the fact
that some operators still feel a basic pressure test
is a valid integrity-assessment method no matter
what the risk or threat. It does not help that the
rule tends to support this until the operator is
reminded that only for some threats is this is a good
assessment method, but even at that, it is seldom
the best method.
throughout the entire programme
The foundation of any good IMP is knowing a
pipelines threats and properly addressing them. If an
operator does not identify the threats associated with
a particular pipeline system, and merely performs an
integrity assessment just as a matter of routine, that
could cause a real, existing, threat to go unnoticed
and unchecked with that operator continuing to
blissfully think that system is taken care of. It is
known that some assessment methods are best suited
for a particular integrity threat, and although use of
a particular method does find indications of other
possible threats, the indications generally are not
distinctive enough to alert the operator to the actual
severity of that threat. Inline inspection (ILI) tool
technology has developed very quickly of late, almost
to the point that a person with just a few years of
analytical experience cannot analyse the data, and it
will no doubt continue to get even more sophisticated
in the near future, with special tools designed primarily
to look for particular threat indications, and to find
them with very good accuracy.

Heres an example of why it is necessary to begin

with a good risk and threat analysis and then decide
This paper was presented at the Unpiggable Pipeline Solutions Forum held
in Houston in May, 2013, and organized by Clarion Technical Conferences on the appropriate method for assessment. Suppose
and Tiratsoo Technical. an operator determined that the highest threat to a
Authors contact details:
pipelines integrity was external corrosion due to poor
tel: +1 713 869 8425 coating. This operator would then decide to use a
email: randy.vaughn@rrc.state.tx.us high-resolution magnetic-flux-leakage (MFL) tool for
128
assessing the severity of external corrosion. But the
tool also has the capabilities of finding what appears
to be hard spots as well but only finding them, and
not providing much more information beyond that.
Any hard-spot indications identified by this particular
tool would be of the quality that they could only be
assumed to have occurred during the manufacture of
the pipe, by dents caused during initial construction,
or by excavation or other intrusive damage. In any
case, not enough information can be drawn from
the indications to make any definitive determinations.
In this situation, the assessment method is able
to find indications, but not able to provide much
more information for further decision-making. Had
the operator known that, for example, excavation
damage was going to be a possible integrity threat,
a more-sophisticated tool, or a combination of tools,
designed to locate and provide more-exact data for that
threat would have put the operator way out front in
the effort of maintaining the pipeline system to its
designed mechanical integrity. Also, if an analyst is
made aware beforehand to be looking for a particular
known threat, he or she can then evaluate the data
of an ILI tool in a more-exacting manner to better
locate and determine the severity of any indications
found by the tool.
Historical data considerations
It should be remembered that since the early years
of pipelining, the risk of failure has always been a
consideration, even during the pipelines design phase.
These considerations have become more sophisticated
over the years, mostly through experience. Most of the
earlier pipelines were laid without coating, for example,
and not much consideration was given to burial depth
concerns. Of course there was good reason for this;
not too many good below-ground coatings existed then,
and excavation damage was not invented yet, or at
least not considered much of a threat to pipelines in
those days. At one point early on, we even believed
that casing a road or railroad crossing would offer
enhanced protection against outside forces a kind
of a second wrap around the arm, as it were. And
before cathodic-protection systems became commonplace,
it seemed that corrosion was just something that we
would have to live with and in trying to live with
it, corrosion allowances were a part of determining
a pipelines wall thickness. The point to be made
here is that most of what we know today is the
direct result of experience later refined by research
and development, and this learning process continues
today. Just a few years ago we were introduced to
stress-corrosion cracking (SCC) and we are still trying
to get more familiar with that. And today we have
a new term erosion corrosion bought to us by
the transportation of bio-fuels, or more particularly,
ethanol. Once we have a better understanding of these
The Journal of Pipeline Engineering
new threats, their mechanics, and what causes them,
we can, as we have in the past, address them in our
design of new pipeline systems, and also do a better
job of addressing them in our existing systems using
the current IMP regimes.
An initial risk analysis has to rely on the design
and historical data thats available, and then fine-
tuned later as the IMP continues to develop and
more data are collected. At first, some of the data
may seem irrelevant, but it should be considered
because it generally adds insight to data that will be
collected later. Well save due-diligence for another
discussion, but it should be noted that it has not
been performed well, for the most part, as far as
collecting the data needed to determine a newly
acquired pipeline systems risk value, or threats
associated with the pipeline. The following is a list
of data expected to be available during a pipeline
system acquisition, and which should be available, at
the least in general terms:
leak history
incident-investigation reports
analysis of pipe removed
analysis of coating when exposed
operating history
general day-to-day maintenance records
input from subject matter experts (SME)
And some assumptions need to be made based
on initial design considerations and the effects
that time may have played in magnifying any
potential risks.
Designed safeguards for probable risks
The following is a short list of the more-common
design safeguards for probable risks or threats:
coating type
rock shield
weight coating
burial depth
additional wall thickness
earthquake or ground movement (designed
expansion loops or bends)
laying warning tape above the pipeline
cathodic-protection test-station installations
The following is a list some of the more-common
design consequence of failure safeguards:
avoiding highly populated areas
avoiding areas of human gatherings
adding wall thickness
installing a SCADA system
installing a leak-detection system
2nd Quarter, 2013
Fig.1. Comparative pipeline incident statistics for the US and Texas, 2030-2011.
avoiding bodies of water and other environmental
areas
increasing burial depth
considering additional mainline valves with actuators
Maintenance safeguards for probable risks
The following is a list of some of the more-common
maintenance safeguards:
monitoring and maintaining a cathodic-protection
system
installing damage barriers
taking security measures (fences, locks, etc.)
reinforcing banks of water crossings
Even when all considerations are made and accounted
for during design and operation, risk still exists. Looking
back, many recent incidents appear to have been
completely unexpected, and until the root-cause analysis
was completed, their occurrence made no sense at all.
Unpiggable
In the world of pipelining, the word unpiggable seems
to have a very general definition, but at the same time,
most dictionaries do not have a definition for it. As
far as the dictionaries are concerned, its as if the word
does not exist at all, and most word-processor software
think it is a misspelling. (But in all fairness, in the
world of pipelining, there are a lot of words used that
are not found in any dictionary.) This term does not
mean that the pipeline will not allow the running of
maintenance pigs; unpiggable, as we all know, simply
means that a particular pipeline was not originally
designed and constructed to allow an ILI tool to be
run from start to finish. And unfortunately, it further
means to some of us that the only integrity-assessment
129
method that is now available, since the pipeline will
not allow for an ILI tool run, is pressure testing.
Sometimes that means a decision has been made without
any regard to the threats that may actually exist. And
now, since there is only one integrity-assessment method
available, then the other IMP processes have become
irrelevant or at least their importance is somehow
lessened. Well, not really. Although the IMP requirement
is basically written as a performance-based rule, it does
state in so many words right up front that a risk analysis
is to be done with the threats identified, and that the
threats shall dictate the integrity-assessment method to
be used. Although the rule does recognize pressure
testing as an integrity-assessment method, it does not
say that if the pipeline is found to be unpiggable, all
bets are off and the operator may just pressure test and
remediate as necessary. For determining reassessment
information repeat steps one and two.
Summary
It is true that, when the IMP rules were first implemented,
a lot of significance was given to the development of
the baseline assessment plan through risk ranking of all
pipeline systems to better schedule the baseline integrity
assessment of each system. Consequently, it was easy to
give the risk-ranking determinations more importance
as a result of the risk-analysis process than to give too
much importance to the threats that made up the final
risk determinations.
As time goes on it becomes more and more apparent
that knowing the threats to a pipeline system is interlaced
with the total IMP process. Without knowing the threats
associated with a pipelines mechanical integrity, it
becomes very difficult to maintain and improve on that
mechanical integrity. Without knowing what threatens
130
a pipelines mechanical integrity, it is very difficult to
determine the correct method of assessment to use, the
most direct preventative and mitigative measures to take,
and to show any positive IMP performance. The same
can be said for using a pressure test as an assessment
method. If a pressure test is completed without incident,
The Journal of Pipeline Engineering
what information about the pipelines mechanical
integrity can be used in determining any preventative
and mitigative measures? And what information about
the IMPs progress can be derived when the pressure test
results are compared with the results of the previous,
or next pressure test?
2nd Quarter, 2013 131

Unpiggable pipelines: Kinder

Morgan Canadas experience
with various inspection tools
and technologies
by Nelson Tonui*,Tom DeLong, and Adam Lind
Kinder Morgan Canada, Calgary, AB, Canada

K INDER Morgan Canada (KMC) operates crude oil and refined products pipeline systems in Canada
and the United States. Like most North America pipeline infrastructure, there is an increasing focus
on the integrity of these assets by both the regulators and the public. KMCs mainlines are 100% piggable
using standard uni-directional in-line inspection tools and are equipped with proper launch/receive trap
facilities. Most of the piping at the pump stations, tank farms, and terminal facilities, however, does not have
permanent launch/receive facilities and access is restricted at one or both ends. Ensuring the integrity of
these difficult-to-inspect pipe segments is important for continued safe operation of the systems.

KMC has, over the years, tried and deployed various inspection technologies and tools to inspect facility
piping. This paper presents the companys experiences with some of the technologies and tools used,
including guided-wave technology, external magnetic-flux-leakage (MFL) tools, bi-directional in-line inspection
(ILI) tools, tethered ILI tools, and free-swimming ILI tools. It outlines the successes and challenges of these
inspection technologies and provides some comparison between the data reported by the tools and those
from validation digs and inspections.

Introduction This piping had been considered unpiggable mainly

OVER THE LAST few years, Kinder Morgan Canada
(KMC) has been trying to assess the condition of
difficult-to-inspect piping in facilities (terminals, tank
farms, and pump stations) along the companys pipeline
system (Fig.1). Although this piping represent less than
1% of the companys entire pipeline system, since it
is susceptible to many of the same integrity threats as
the mainline piping, a formal inspection programme is
warranted. This is particularly important for facilities
which are located in or have the potential to affect
high-consequence areas (HCAs). Since KMC has
experienced very few integrity issues with these lines
in the past, these inspections are mostly driven by the
desire to know their condition.
This paper was presented at the Unpiggable Pipeline Solutions Forum held
in Houston in May, 2013, and organized by Clarion Technical Conferences
and Tiratsoo Technical.
*Corresponding author:
tel: +1 403 514 6661
email: nelson_tonui@kindermorgan.com
because of a combination of factors that pose a lot of
difficulty to conventional inspection methods. These
factors, among other things, include lack of accessible
locations for ILI tool insertion and/or retrieval, multiple
diameters, unbarred tees and branches, and bends of
unknown radii. With the emergence of technologies
that can favourably be used to inspect these lines, KMC
has conducted several pilot assessments with varying
degrees of success. In this paper the experiences with
the tools and technologies that were used between 2010
and 2012 are presented.
Tethered, self-propelled, and free-
swimming ILI tools
Tethered ILI tools
Tethered ILI tools were chosen for inspection because
they only need a single access point and, unlike some
other methods, a complete set of inspection data
comparable to conventional ILI can be collected. At
KMCs Burnaby Terminal, located within an HCA,
132
Fig.2a. Gauge plate after retrieval from the first line.
tethered ILI was attempted in three different lines in
2010. For all three lines, a tethered gauge tool was
first run before the ILI tool. In all cases the lines
were drained of product by gravity from the manifold/
tethered-tool launch location.
The first attempt was on a 2000-ft (650-m) NPS-20 tank
line connected to a tank at a higher base elevation
than the launch point at the manifold. As with most
tank lines, as-built survey information was not available
which could be used to confirm the elevation profile,
type and number of bends, and potential for presence
of low points where product could remain even after the
gravity drain was completed. A tethered gauge tool was
to provide vital information on the geometric condition
of the line. After the line was drained down, the tool
was inserted and was moved along by injecting nitrogen
behind the tool. At the start, a nitrogen gauge pressure
of only 4 psig (0.275 bar) was needed to move the
tool, but the pressure required began to increase soon
The Journal of Pipeline Engineering
Fig.1. Kinder Morgan Canada system map.
Fig.2b. Gauge plate after retrieval from the first line.
after the launch. The tool was eventually stopped at a
location approximately 200 ft (66 m) from the launch
point when the pressure approached 30 psig (2.07 bar).
Prior to pulling the tool back, the nitrogen pressure
was bled off. It was evident that the tool had moved
backwards as the pressure was bled off since there
was little or no tension on the wireline for the first
part of the retraction. Uneven tension and intermittent
jerking caused by retraction of the tool through an
unconfirmed number of bends resulted in the wireline
severing before the tool was completely retracted. To
free the tool, it had to be pushed from the opposite
end from the launch point (at a high-point vent located
near the tank) with nitrogen. The tool could only be
pushed to the beginning of the manifold and therefore
final extraction was completed using a long fabricated
mechanical pig puller attached to a knuckle-boom crane.
Once out, the gauge tool had significant damage to
the blades (Figs 2a and b).
2nd Quarter, 2013 133

A review of the inspection operation was done to

determine the cause of wireline failure. The tool
was good for 2-D bends, so it was speculated that
there could be some significant dents in the line.
However, this turned out not to be the case. Looking
at the line profile, there was a mid-section that was
of relatively level topography in between portions of
incline at the start and towards the end of the line.
Therefore, there was a possibility that pockets of liquid
remained in the low points of the line after gravity
drain-down. These pockets of liquid eventually led to
an accumulation of liquid as the gauge tool neared
the end of the line (up the incline). It is believed
that the pressure from this column of liquid pushed
the tool backwards when the nitrogen pressure was
bled off, resulting in the tangling of the wireline
around the disks and scrapers of the sizing tool. Fig.3. Coiled-up wireline behind the tool before removal from
This was supported by the fact that a vacuum truck the second line.
connected to the end of the line (near the tank),
collected fluid equivalent to about 60 ft (19.7 m) of
linefill before the gauge tool was stopped. Based on
the experiences with the gauge tool in this tank line,
a decision was made not to attempt the inspection
with the tethered MFL tool.

A second attempt was made on a second NPS-20 line

approximately 1,500 ft (492 m) in length. Problems
similar to those experienced in the previous attempt
were expected, therefore the gauge tool was stopped
when the nitrogen pressure required to push it
approached 15 psig (1 bar approx.). At that point
the tool was about 700 ft (230 m) into the line.
Once the pressure was bled-off, the gauge tool moved
backwards slightly. As a result, the wireline was tangled
(Fig.3) but it did not break, and could still be used
to pull the tool out. The deflection in the gauge
blades was only in the backward direction, which Fig.4. Broken wireline after the tool was pulled into the third
line, due to vacuum.
was an indication that the damage was done when
the tool was pulled back against a coiled wireline.
Inspection of the line with the tethered ILI tool vent located near the tank inlet was opened to
proceeded. Given the uncertainty of bend radii and atmosphere for the pressure equalization. When the
the damage to the wireline and the gauge plate after bar holding the tool in place was removed, the tool
running the scraper pig 700 ft (230 m) into the was sucked into the line with enough force to sever
line, the ILI tool was only pushed 150ft (49.2 m) the wireline from the cable head (Fig.4). A near-miss
up the line. This ended as a partial success, since was recorded when the severed wireline kicked-back
only 10% of the entire line was inspected. and struck the vacuum truck operator in the back.

A third and last attempt was on a third NPS-20 tank
line approximately 400 ft (131 m) in length. This line
was a prime candidate for inspection because it was
short, had no lateral bends, and would have little or
no liquid remaining in the line after a drain-down
due to its assumed profile. Problems experienced
earlier as a result of liquid column accumulation were
not expected. After drain-down, a small vacuum was
applied to the low-point drain to draw vapour away
from the work crew as the tool was loaded into the
line. The vacuum was then isolated and the high-point
It was later determined that there was a residual
vacuum in the line. Evidently, the high-point vent
located near the tank was either too far away or
too small to be effective in maintaining atmospheric
pressure inside the line, resulting in a small vacuum
being imparted on the test section. The tool was
eventually retrieved from the line, but the wireline
measuring head had been damaged during the incident
(Fig.5). Waiting for the necessary parts would have
caused further delay; therefore, it was decided not
to proceed with any further wireline testing.
134
Fig.5. Damage to the wireline measuring head.
Lessons learned
Pushing a tethered ILI tool up a tank line which
cannot be verified as completely drained may
cause problems due to build-up of backpressure
ahead of the pig. This is of course subject to
the profile of the line being inspected, the
worst case being a line with an end point at
a higher elevation than the launch. As was
the case in this job, this can lead to the pig
moving backwards when the pushing pressure
is removed, tangling up the wireline before the
retraction process has even started. Without an
accurate line profile, it is difficult to estimate
how much liquid would remain in each tank
line after a drain-down. This information is
critical in determining how much pressure to
keep behind the pig during retraction.
Applying a vacuum, while good for removing
vapour and products, can lead to unexpected
pig movements. As was the case in this job,
there are cases where residual vacuum remains in
the line leading to unpredictable pig movement.
Ensuring adequate venting is necessary, and the
pressure inside the pipe must be ascertained to
be atmospheric before releasing the pig.
The lines need to be clean (free from debris)
as much as practicable before commencing an
inspection with a tethered tool. The amount
of debris affects the tool movement and hence
the amount of data collected.
Having learned from this experience, one year later,
in 2011, a self-propelled ILI tool run on two of the
lines (the first and second) was attempted.
Self-propelled ILI tool
A self-propelled ILI tool was used to inspect two
tank lines (previously attempted using tethered ILI)
The Journal of Pipeline Engineering
in KMCs Burnaby Terminal. The tool was launched at
the tank manifold and stopped a few feet from the tank
inlet. Access to the line was achieved by removing a
blind flange located at the dead-leg section of the line.
A prefabricated insertion sleeve was used to insert the
tool into the line: the tool was loaded into the sleeve
(Fig.6), and then both the tool and the sleeve were
pushed into the line.
Once the tool was inside the line and the temporary
flange bolted up, the line was re-flooded. Thereafter the
ILI tool was energized to move it out of the insertion
sleeve. Tool calibration was performed after the tool
had transitioned from the tank-manifold piping (with a
higher wall thickness) to the tank-line piping. The tool
commenced inspection going all the way to a reducer
upstream of the tank-inlet piping and back, by propelling
itself using onboard pumps. Continuous monitoring of the
location of the tool was done using markers strategically
placed at points of interest along the tank-line route and
from the computer in a nearby office trailer. Once the
tool was back at the tank manifold, the umbilical was
used to winch it back into the insertion sleeve. The
tool was de-energized and the by-pass opened to allow
for a final drain-down.
The line was drained-down again and the tool, together
with the insertion sleeve, was removed. A final bolt-up
was done followed by a second re-flood and the line was
ready to resume normal operations.
Results
The run was successful on both lines. However, there were
instances of data loss especially at the 6 oclock position,
which was mainly attributed to the debris accumulation
in the line.
An earlier attempt to run the tool in one tank line in
the Edmonton Terminal was, however, not successful. The
scaling/corrosion and sediment in the line damaged the
tool so significantly that it had to be re-built and cleaned
before use at Burnaby. The product in the line, isooctane,
also affected the performance of the seals and bearings.
Lessons learned
The self-propelled ILI tool is very susceptible to
line cleanliness. A small amount of scale in the
bottom of the pipe can gum-up the unit due to
the small clearances between the pump unit and
the pipe walls.
This technology appeared to be less robust and
was relatively expensive compared to other similar
inspection methods.
A well-thought-out tool-insertion technique can
significantly reduce preparation time and increase
the chances of success, as was the case at Burnaby.
2nd Quarter, 2013
Fig.6a. Positioning the insertion sleeve.
Free-swimming ILI tool
A free-swimming MFL tool was used to inspect two tank
lines at KMCs Burnaby Terminal. This technology was
chosen for the following reasons:
the potential to obtain a full set of inspection data
the accessibility of the two lines: it was possible
to access the lines at both ends
the topography of the lines: it proved easier to
move the tool from the tanks, located at a higher
elevation than the receiving point at the manifold.
Figure 7 shows a schematic of the tool run: the tool
was inserted at a higher point near the tank inlet by
removing a small spool from the line (Fig.8). Once the
tool was in the line and the spool back in place, a
transfer pump was connected to provide pressure and
move the tool to the receiving point. Since the location
of removing the tool was at a relatively high elevation,
a temporary access platform was set up. A blind flange
was then dropped and the tool pulled out.
This approach was successful in inspecting two lines,
and resulted in full set of data being collected. This
inspection method is a good candidate when there are
two access points on a line.
135
In another inspection at KMCs Edmonton Terminal,
the use of a free-swimming ILI tool with a similar
set-up was not successful as it turned out that the
movement of the tool was negatively affected by the
type of product in the line. The product in the line
during inspection was isooctane: this is much lighter,
less viscous, and has far less lubricity than the crude
oil used in previous successful attempts. As a result,
the isooctane found ways to by-pass the cups of the
tool, particularly in the tight bends, and this made it
difficult for the tool to be pushed through the line.
The tool got stuck three times and, after days of
attempts to free it, a decision was made to cut the
line in order to free the tool.
Lessons learned
Know the physical properties of the product in
the pipeline: different products affect a tool run
differently. This knowledge helps in determining
the appropriate pumping pressure and flow
rate, speed of the tool, design of the tool (cup
spacing, etc.) and bend-negotiating capabilities.
Detailed tool tracking is essential. It is worth
investing in additional tracking technology so
that the exact position of the tool is known
at all times.
136 The Journal of Pipeline Engineering

Fig.6b. Loading the tool into the insertion sleeve before inserting into the line.

Fig.7. Free-swimming tool set-up.

Run a cleaning pig before running the ILI

tool is highly recommended. Although this will
mean two mobilizations, it is better than trying
to run the tool in a dirty line. The chances
of a successful tool run are increased if the
general condition/cleanliness of the line can
be verified in advance. Extensive planning by
engineering and operations departments, and
a team approach to tool runs, are obviously
beneficial. Tool runs in unpiggable pipelines are
generally different from runs in conventional
lines, and may involve retrofits and modifications.
Therefore, adequate preparation and assessments
of hazards is essential.
Fig.8.The spool where the tool was launched.
2nd Quarter, 2013 137

Guided-wave results Direct examination

1 Category 3, high priority, follow-up area Through-wall corrosion

2 Category 2, medium priority, follow-up area Minor mechanical damage

3 Category 2, low priority, follow-up area Minor corrosion < 10%

Table 1. Comparison of results from 4 Category 2, low priority, follow-up area Surface slivers
guided-wave and direct assessments.

Other inspection technologies

Guide-wave inspection

Kinder Morgan has been using guided-wave inspection

technology for the last 10 years. This technology has
been widely used for various applications such as a
screening tool for air-to-soil interfaces and underground
piping where some sections have been exposed. From
experience, guided-wave provides a lot of advantages
including:

inspection of a large area from one access point

it is relatively easy to perform
in-service inspection can be performed (with no
service interruption) Fig.9. Guided-wave inspection.

However, guided-wave technology provides some

limitations as well, including:

it cannot differentiate between external and

internal wall loss
it is affected by the coating type and the soil type

These factors affects the attenuation of sound waves

and hence the length of pipe that can be inspected.
However, despite these limitations, successes have been
experienced with the use of guided-wave inspection
technology, and in 2011 it was used to inspect several
underground lines at KMCs Kamloops Terminal.

Preparation and inspection Fig.10. External MFL tool mounted on an above-ground pipe.

The underground sections of the lines were exposed

to attach the guided-wave tool collar, Fig.9, and to
perform the inspection. To inspect the entire intended
sections, several bell holes were excavated along the
length of the pipe. Although it was not done in this
particular job, a direct-current voltage-gradient (DCVG)
survey could be used to locate areas of coating failures:
the pipe would then be exposed in those areas so as
to repair the coating while at the same time providing
access to perform the inspection.
Results
A complete inspection of the intended sections was
achieved. Direct assessment was also done in some
exposed sections of the lines and in some locations where
the guided-wave results warranted further investigations.
Table 1 shows a few comparison between the guided-
wave inspection results and what was found by direct
examination.
138
External MFL tool
External MFL tools have been used primarily to inspect
above-ground piping. In 2011, this tool was used to
inspect several above-ground portions of lines KMCs
Edmonton Terminal.
Preparation
There was little preparation needed in order for this
tool to be used. The tool requires an almost direct
contact with the pipe which was readily available,
since the above-ground section was only painted,
Fig.10. Because the tool is external, the inspection was
completed while the lines were in service (and there
was no service interruption).
The Journal of Pipeline Engineering
Fig.11. A unity graph showing
the comparison of anomaly
depth measurements from
the external MFL tool
and the conventional UT
measurements.
Fig.12. Depth of anomalies
as measured by external MFL
and UT compared to the
length of the anomaly.
Inspection
Owing to the ease of mounting and removing the tool, a
high production rate was realized, which in turn reduced
the overall cost of the inspection. Several hundred feet of
pipe could be inspected in a single day. The downside
of this inspection method is the need to move the tool
around pipe supports, bends, and other obstacles; this
results in discontinuous data collection and poses the
possibility of some vital features being missed.
Results
More than 75% of the pipe was successfully inspected,
and for these particular pipe segments, all the features
that the tool identified warranted no further action.
2nd Quarter, 2013
Prior to using the external MFL tool, KMC performed
independent validation measurements of the tool. The tool
was used to measure wall loss in several above-ground pip-
ing segments at four different facilities, and the anomalies
reported were then measured manually using conventional
ultrasonic technology. Figure 11 shows a comparison between
the depths of the anomalies as reported by the external
MFL and as measured by conventional UT. The results
show little or no discrepancies between the two sets of data.
Although the depth measurements by the MFL tool sometimes
were higher than the UT measurements, they were mostly
within the 10% threshold. Figure 12 shows the relationship
between the measurements and the length of the anomalies:
it appears that the accuracy of the tool is independent of
the length of the anomaly. The tool appeared to be sizing
all defects (regardless of the length) with the same accuracy.
139
Conclusion
Under the banner of its Facility piping inspection
programme, Kinder Morgan Canada continues to explore
ways of inspecting all the unpiggable pipe segments
in its system. There have been some successes and
failures in past inspections. Most importantly, valuable
lessons have been learnt from those experiences that
KMC will leverage to enable more successful runs in
the future while ensuring the safety of the employees
and contractors performing the inspections. In its
pursuit of protecting the public and the environment,
KMC aims to complete the baseline assessments of
all high-risk-facility piping in the near future, and
looks forward to utilizing new tools and technologies
as they become available.
 Houston, February 1013, 2014

Cour s es . C on f er en ce . Exhibition
its 26 year, the PPIM Conference is recogniz
th
Now entering ed as
re m o s t in t e r n a tional forum for sharing and learning ab
the fo out best
e s in l if e t im e m aintenance and condition-monitoring te
practic chnology
for natural gas, crude oil and product pipelines.
www.clarion.org

The international gathering of the global pigging industry!

Organized by
 ARE YOU UP TO SPEED?
0 0 0 2 0 1 2

Training courses 2013

TRAINING
16-20 September Practical pigging course (Rio de Janeiro, Brazil)
SEP 2013 23-26 September Rio Pipeline Conference (Rio de Janeiro, Brazil)

1-4 October Subsea production systems (Houston)

6-9 October 6th Pipeline Technology Conference (Ostend, Belgium)

OCT 2013 20-23 October 2nd Best Practices in Pipeline Operations and Integrity Management Conference
(Bahrain)
28-30 October Defect assessment in pipelines (Kuala Lumpur, Malaysia)
28 Oct-1 Nov Onshore pipeline engineering (Kuala Lumpur, Malaysia)

DOT Pipeline Safety Regulations - Overview and Guidelines for Compliance

11-12 November (Houston)

2013
11-12 November Pipeline Repair Methods, Hot Tapping, and In-Service Welding (Houston)
11-13 November Pipeline Integrity Management (Houston)
11-15 November Onshore Pipeline Engineering (Houston)
11-15 November Pipeline Integrity Courses, Houston - Nov 2013 (Houston)
NOV 2013 11-15 November Subsea Production Systems Engineering (Aberdeen)
11-15 November Advanced Pipeline Risk Management (Houston)
11-15 November Stress Corrosion Cracking in Pipelines (Houston)
18-20 November Defect Assessment in Pipelines (Houston)
18-22 November Subsea Pipeline Engineering (Houston)
20-22 November Defect Assessment Calculations Workshop (Houston)

2-3 December Pigging & In-line Inspection (Calgary)

2-4 December Defect Assessment in Pipelines (Calgary)
2-6 December Pipeline Integrity Courses - Calgary (Calgary)
DEC 2013 4-6 December Pipeline Integrity Management (Calgary)
4-6 December Defect Assessment Calculations Workshop (Calgary)
5-6 December Geohazard Management for Pipeline Engineers (Calgary)

Working with a faculty of 38 leading industry experts, Clarion and Tiratsoo Technical are privileged to provide some of the best
available industry based technical training courses for those working in the oil and gas pipeline industry, both onshore and oshore.

Complete syllabus and registration details for each course are available at:
www.clarion.org