Developments in Water Treatment 2 by W.M.Lewis PDF

DEVELOPMENTS IN WATER
TREATMENT—2
THE DEVELOPMENTS SERIES
Developments in many fields of science and technology occur at such a pace that
frequently there is a long delay before information about them becomes available and
usually it is inconveniently scattered among several journals.
Developments Series books overcome these disadvantages by bringing together within
one cover papers dealing with the latest trends and developments in a specific field of
study and publishing them within six months of their being written.
Many subjects are covered by the series including food science and technology,
polymer science, civil and public health engineering, pressure vessels, composite
materials, concrete, building science, petroleum technology, geology, etc.
Information on other titles in the series will gladly be sent on application to the
publisher.
DEVELOPMENTS IN WATER
TREATMENT—2
Edited by
W.M.LEWIS
M.Chem.A., C.Chem., F.R.I.C., F.I.W.E.S.
WHO Consultant EURO, Environmental
Health—Drinking Water Quality,
Copenhagen, Denmark
Managing Director, Coventry Chemical
Consultancy Ltd, Coventry, UK
APPLIED SCIENCE PUBLISHERS LTD

LONDON
APPLIED SCIENCE PUBLISHERS LTD RIPPLE ROAD, BARKING, ESSEX, ENGLAND
This edition published in the Taylor & Francis e-Library, 2005.
“To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of
thousands of eBooks please go to http://www.ebookstore.tandf.co.uk/.”
British Library Cataloguing in Publication Data
Developments in water treatment.—(Developments series). 2 1. Water—Purification I. Lewis,
W.M. II. Series 628.1′6 TD430
ISBN 0-203-97492-1 Master e-book ISBN
ISBN 0-85334-903-7 (Print Edition)

WITH 24 TABLES AND 54 ILLUSTRATIONS
© APPLIED SCIENCE PUBLISHERS LTD 1980
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or
otherwise, without the prior written permission of the publishers, Applied Science Publishers Ltd,
Ripple Road, Barking, Essex, England
PREFACE
‘Surveillance of Drinking Water Quality’ published by WHO in 1976 in its introduction

stated that ‘Public health protection of drinking-water supplies should assure that each
component of the system—source, treatment, storage and distribution—functions without
risk of failure’. Drinking water is perhaps, together with the air we breathe, a unique
commodity in that the general population is normally permitted no freedom of choice, so
the assurance that the water available for drinking is of the highest quality is of
paramount importance.
Since the universal introduction of disinfecting agents in water supplies in developed
countries, risks to health from microbially contaminated drinking water have been
dramatically reduced. Today the problem confronting personnel responsible for ensuring
the public’s water supply is perhaps of a much more subtle character brought about, in
part at least, by the rapid progress in analytical chemistry and the environmental
awareness of the general public resulting in the demand for the creation of standards of
quality for drinking water.
Chemicals present in raw water supply range from simple ions extracted from soil and
minerals in the watershed to (in some instances) unidentified waste products from the
chemical industry, the length of the list being limited only by the capabilities of the
analytical chemists and their instruments.
Some medical researchers proclaim that the presence, or absence, of a certain
substance in drinking water is directly associated with the differences in death rates from
specific diseases or the incidence of morbidity. The problem is compounded due to the
fact that the toxic effects of many identified chemicals are insufficiently understood.
Thus the responsibility devolving upon the shoulders of personnel responsible for
‘Treatment’ is to provide a process (or combination of processes) which will ensure, as
far as is practicable, that the water supply is not only aesthetically acceptable, but also of
the best chemical standard and in a condition which will not damage the integrity of the
distribution system which could result in subsequent and additional contamination. With
rivers of the calibre of the Danube, Trent and Rhine, to mention but three, the resulting
treatment, to provide drinking water whose quality is beyond suspicion, needs to be very
sophisticated.
The ‘treatment process’ is not a single identifiable parameter but is dependent upon
the nature and quality of the raw material and may for example involve only simple
filtration or filtration plus disinfection. On the other hand if the quality of the supply
water is from a lowland river, such as the three previously mentioned, then it follows that
a combination of individual processes, commencing with coagulation for the removal of
suspended matter, etc., and employing perhaps the majority of the techniques described,
will be essential to provide the required quality of drinking water.
Within this series will be found the various important facets of treatment each written
by an author, expert in the particular field, who has introduced his topic with a brief
historical background before providing the reader with the most up-to-date information
available on the subject.
It is a salutary thought that in 1975 (latest information available) some 78% of the
world’s rural population and even 22% of the urban population were without an adequate
water supply. Of the urban population of the world having access to a piped water supply
(77%), 57% only had house connections and 54% of the population served by public
piped supply received it only on an intermittent flow basis.
Conscious of the urgent need to rectify these shortcomings, the UN Water
Conference—Mar del Plata, March 1977—urged the adoption of ‘The International
Drinking-Water Supply and Sanitation Decade, 1981–1990’. The aim of the latter is to
encourage and assist all countries of the world to adopt programmes with realistic
standards for both quality and quantity and to provide water to all people by 1990, if
possible.
It is unfortunate, but nevertheless true that at present, and for how long into the future
we know not, many countries—not only the developing ones—are experiencing financial
constraints of varying magnitude. To achieve the above objectives will therefore strain
the ingenuity and professional expertise of all concerned with the task of supplying the
community with drinking water.
It is consequently singularly appropriate that these first two volumes on
‘Developments in Water Treatment’ should be available at this time, for within their
pages will be found that information on ‘Treatment’, appropriate to the needs, to enable
people to overcome the financial constraints laid upon them.
W.M.LEWIS
CONTENTS
Preface v
List of Contributors x
1. Filtration 1
T.H.Y.TEBBUTT
2. Removal of Organic Compounds 22
C.S.SHORT
3. Removal of Nitrogen Compounds 53
R.B.GAUNTLETT
4. Desalination 79
M.J.BURLEY and J.D.MELBOURNE
5. Disinfection 122
A.T.PALIN
6. Sludge Treatment and Disposal 142
M.A.HILSON
7. Water Quality Monitoring 162
P.J.MORLEY and J.COPE
Index 189
LIST OF CONTRIBUTORS
M.J.BURLEY
Consultant, Sir M.Mac Donald & Partners, Demeter House, Station Road, Cambridge,
CB1 2RS, UK.
J.COPE
Scientific Officer, Headquarters Staff, Severn-Trent Water Authority, Tame House,
Newhall Street, Birmingham, B3 1SE, UK.
R.B.GAUNTLETT
Treatment Division, Water Research Centre, Medmenham, Marlow, Bucks, SL7 2HD,
UK.
M.A.HILSON
Principal Scientist, Water Treatment and Supply, North West Water Authority, Dawson
House, Great Sankey, Warrington, WA5 3LW, UK.
J.D.MELBOURNE
Managing Director, Melcon Water International Ltd, 165 Reading Road, Henley-on-
Thames, Oxon., RG9 1DP, UK.
P.J.MORLEY
Principal Scientist, Avon Division, Severn-Trent Water Authority, Avon House, De
Montford Way, Cannon Park, Coventry, CV4 7EJ, UK.
A.T.PALIN
Consulting Chemist, 7 Montagu Court, Montagu Avenue, Newcastle upon Tyne, NE3 4JL,
UK.
C.S.SHORT
Yorkshire Water Authority, Olympia House, Gelderd Road, Leeds, LS12 6DD, UK.
T.H.Y.TEBBUTT
Senior Lecturer, Department of Civil Engineering, The University of Birmingham, P.O.
Box 363, Birmingham, B15 2TT, UK.
Chapter 1
FILTRATION
T.H.Y.TEBBUTT, B.Sc., S.M., Ph.D., M.I.C.E., M.I.W.E.S.
Senior Lecturer, Department of Civil Engineering, The University of
Birmingham, Birmingham, UK
SUMMARY
Developments in water filtration have enabled a more efficient use of the process and
have provided a better understanding of the mechanisms involved. Good filtrate quality
can be achieved at higher loadings than would have been considered possible some years
ago and the advent of dual and multi media beds has brought economic benefits. A
consequence of more effective use of bed capacity is that the process is more likely to be
influenced by the nature of the suspended solids in the feed. A consequence of obtaining
deeper penetration of suspended matter into a filter bed is the need to ensure adequate
cleaning arrangements. The complexity of the filtration process is such that mathematical
models cannot as yet provide a universal solution for any situation but their existence is
of value in helping to develop more efficient filtration plants.
1.1. INTRODUCTION
Some form of filtration is almost certain to be incorporated in the treatment used to

produce potable water from most surface water sources. Indeed, the principle of filtering
rain water through a layer of porous material to remove suspended matter has been used
for centuries although the first recorded purpose-built sand filters for a public water
supply were not installed until 1804 in Paisley. The early installations were of the slow
filter type and their considerable purification ability in no small way contributed to the
great improvement in public health which occurred during the second half of last century.
By 1870 the rapid pressure filter concept made its appearance, to be shortly followed by
the rapid gravity filter whose basic design has not changed greatly to the present day.
However, since the introduction of filtration as a water treatment process a considerable
amount of knowledge has been accumulated about the behaviour and performance of the
process. The continuing pressures to require more efficient treatment, in operational and
economic terms, and the demands for higher quality filtrates have meant that a variety of
developments to and modifications of the conventional filtration process have been
introduced. Coupled with these developments has been an intensive study of the actual
mechanisms involved in filtration with the aim of providing a better understanding of the
process to aid in future design. Filtration through porous media is however a process of
great complexity and it is not yet possible to produce a completely satisfactory theoretical
model.
Developments in water treatment—2 2
1.2. APPLICATIONS OF FILTRATION
Except in rare cases all surface waters will contain, at least at times, sufficient suspended
matter to given turbidities in excess of 5 units which is the highest desirable level
stipulated in the World Health Organisation International Drinking Water Standards.
EEC standards set out a guide level of 5 FTU with a maximum admissible concentration
of 10 FTU. In the USA the National Interim Drinking Water Standards lay down a
mandatory turbidity limit of 1 unit for most surface-water-derived supplies. The
American Water Works Association has adopted a turbidity goal of 0.1 units and it has
been suggested that this may in the future be reduced to 0.05 units. The rationale behind
the adoption of such low turbidity levels is not that the turbidity itself is likely to be
harmful but that its presence may inhibit disinfection processes. With such limits for
turbidity being required, even high quality impounded waters would need turbidity
removal for much of the time and all lowland river derived supplies would necessitate
comprehensive treatment for turbidity removal. With high turbidity raw waters, chemical
coagulation and sedimentation are commonly employed to remove the bulk of the
turbidity with filtration providing the final polishing stage to bring turbidity down to
around 1 unit. With coloured upland sources, turbidity in the raw water may not be a
serious problem but chemical coagulation is frequently used to remove the colour by
what is in effect a precipitation process so that here again filtration is necessary following
sedimentation. With low turbidity raw waters direct filtration using either gravity or
pressure rapid filters aided by chemical coagulation is widely used and in parts of the UK
and Europe slow sand filters are used either as a single stage of treatment or as a second
stage following rapid filtration used as a roughing or preliminary stage of treatment.
Whilst all these types of filtration should be able to produce filtrates with turbidity less
than 1 unit the reliable production of turbidities less than 0·1 or 0·05 units is another
matter. It is important to appreciate that most treatment processes, including filtration,
follow a law of diminishing returns in that the unit cost of successive increments of
purification increases, sometimes almost exponentially. Before setting stringent turbidity
levels it should therefore be considered whether such levels are achievable by normal
means of treatment and whether the cost of such treatment is justifiable in respect of the
marginal improvement in water quality which they would produce.
It is of interest to note that developments in wastewater purification have resulted in
the adoption of filtration as a tertiary treatment stage in situations where the effluent is
discharged to a receiving water with little dilution or which provides a raw water source
some distance downstream of the outfall. In such circumstances sand filtration, usually in
rapid gravity units, removes a considerable amount of the suspended solids which escape
from final settling tanks together with the organic matter associated with these suspended
solids. Thus a normal secondary effluent of 30 mg/litre suspended solids and 20 mg/litre
biochemical oxygen demand can be polished by filtration to give a final effluent of
around 10 mg/litre suspended solids and biochemical oxygen demand. The adoption of
tertiary filters has encouraged a considerable amount of research into design and
operational factors which has enhanced knowledge of the filtration process.
Filtration 3
1.3. CONVENTIONAL FILTRATION
Before discussing recent developments in the filtration of water it is useful to briefly

outline what may be considered as conventional usage which had not changed greatly
until the early 1970s and which is summarised in Table 1.
1.3.1. Slow Filters

As indicated earlier, the first form of sand filter was the slow type, so called because of
its relatively low hydraulic loading of 1–4m3/m2d. Because of this low hydraulic loading
there is only superficial penetration of suspended matter into the bed and filter runs of
several weeks or even months are
TABLE 1 CHARACTERISTICS OF
CONVENTIONAL FILTERS
Characteristic Slow filter Rapid filter
Filtration rate 1–4 m3/m2d 100–150
m3/m2d
Size of bed About 2000 About 100
m2 m2
Depth of bed 1m 0·5–0·8 m
unstratified stratified
Sand size 0·5–1·0 mm 0·5–1·5 mm
Uniformity 2·0–2·5 1·2–1·5
coefficient
Head loss Up to 1 m Up to 2·5 m
Length of run 20–90 days 1–5 days
Cleaning water 0·2–0·6% of 1–6 % of
consumption output output
possible before excessive head loss terminates operation. The usual depth of water above
the sand surface is 1·0–1·50 m so that when the head loss approaches this figure filtration
must be stopped to prevent the possibility of sub-atmospheric pressures being created
within the bed with consequent deterioration in performance. The bed can be restored to
operation by scraping off the top few cm of sand which can be washed and later used to
replenish the bed. The long interval possible between cleaning procedures enables
biological activity to become established in the bed particularly in the surface layers
where a biological slime, the schmutzdecke, contributes significantly to the removal of
fine suspended matter. In addition the biological growth utilises organic constituents in
the water as a food source oxidising them to inorganic end-products and often preventing
or reducing possible taste and odour problems in the finished water. Nitrification of
ammonia is also usually a feature of slow sand filtration and this can make the ensuing
disinfection stage more reliable. With small levels of ammonia, which would be all that
would be likely to be present in a potable water source, the small addition of nitrate to the
finished water is unlikely to be objectionable although of course levels of nitrate nitrogen

of more than a few mg/litre are to be viewed with suspicion. Because of their large size
slow filters are labour intensive as regards cleaning operations and although a number of
plants have been equipped with mechanical scraping devices, the cleaning of slow filters
is a costly operation. For this reason it is imperative that they are only used for relatively
low turbidity (10–20 units) raw waters. Because of this limitation, slow filters are not
normally suitable for sources such as lowland rivers with high turbidities which usually
require chemical coagulation since the additional carry-over of floc from settling tanks
would be sufficient to rapidly clog the filter surface. The increased dependence on
lowland sources has resulted in a decline in the use of slow sand filters in many parts of
the world. However in the UK much of London’s water is treated by slow filters which
follow primary rapid filters. In the Netherlands slow filters have found considerable
application as the final treatment stage when purifying heavily polluted river waters such
as those from the Rhine.
1.3.2. Rapid Filters

The introduction of rapid sand filters no doubt arose partly from the seemingly inevitable
trend to increase the capacity of treatment units and the need to handle high turbidity raw
waters. The land area required for slow filters became a factor when constructing
treatment plants for large communities so that considerable pressures were exerted on
designers to develop more compact units. The result was the rapid filter with a hydraulic
loading of around 100 m3/m2d. At this rate of flow, penetration of solids into the bed is
considerable and clogging to a maximum allowable head loss takes only a day or two
depending upon feed quality. Alternatively, penetration of turbidity may be complete
resulting in breakthrough of suspended matter in the filtrate which will terminate the run
even though the head loss limit may not have been reached. The deep penetration of
solids and frequent need for cleaning would of course rule out the type of cleaning
operation used for slow filters. Fortunately, however, the incorporation of backwashing
systems was able to provide a practical and economical means of cleaning rapid filters.
When using the normal bed of graded media, backwashing results in a stratified bed
which is not desirable from the point of view of optimum utilisation of bed capacity but
which has been generally accepted as a characteristic of rapid filtration.
The bulk of filtration capacity worldwide is in the form of rapid gravity filters
preceded, where appropriate, by chemical coagulation and sedimentation units. In the
UK, rapid pressure filters with direct injection of coagulants find favour in hilly areas
where their hydraulic characteristics prove useful when siting treatment plants but in
other parts of the world pressure filters tend to be mainly used for industrial water
supplies. For low turbidity raw waters rapid filtration alone, possibly with the addition to
the raw water of a small dose of coagulant, can provide a filtrate of satisfactory quality.
In most cases, however, the bulk of the turbidity is removed by preliminary coagulation
and sedimentation processes which prolong filter runs. As mentioned in the previous
section, rapid gravity filters are also used as the primary stage in a double filtration
process. In the case of pressure filters separate coagulation and sedimentation tanks are
not feasible so that the coagulant is dosed directly into the raw water and flocculation
takes place largely within the filter bed. This does of course mean that a much larger
Filtration 5
amount of suspended matter is trapped by the filters and thus filter runs are much shorter
than if the same water were treated by conventional gravity filters preceded by
coagulation and sedimentation units. The omission of these units does however give a
cost reduction to offset the shorter filter runs and more frequent washing required by
pressure filters.
An important feature of rapid gravity filters is the arrangement made for backwashing
and in particular the design of the underdrain and filter floor. It is important that the
active filter medium is supported in such a way that it is contained within the filter shell
whilst preventing any short circuiting of flow during filtration or backwashing. The
commonest system in the UK is to employ a concrete floor with plastic collection and
distribution nozzles covered by a layer of graded gravel but there are a considerable
number of proprietary filter bottoms. Some of these are of complex and costly design and
it is by no means certain that any advantages they may have are cost effective. There are
also variations in the form of filter control and backwash procedures adopted by different
manufacturers and designers.
A variation of the pressure filter, used quite widely in the UK for swimming pool
water treatment, employs diatomaceous earth or cellulose powder deposited on rigid
formers as the filtering medium. These units can give excellent turbidity removal and are
found suitable for small potable supplies in some parts of the world, notably in the USA.
In operation it may be beneficial to dose the raw water with a small concentration of the
filtering medium. At the end of a run the medium together with the trapped suspended
matter is discharged to waste and the filter element recoated with powder. The units are
relatively small in size and can provide an economic form of treatment for low turbidity
supplies.
1.4. THE FILTRATION PROCESS
At first sight the manner in which a bed of porous medium removes suspended particles
from a fluid passing through the bed might appear to be easily explained as some form of
straining action. Clearly, if the suspended particles are larger than the voids in the bed
they will be removed in the same way as a sieve prevents the passage of particles larger
than the mesh size. Such a removal mechanism would mean that no penetration of solids
into the bed could occur and that there would therefore be no need for a deep bed of
medium. Examination of the solids removal performance of a granular bed shows,
however, that particles very much smaller than the voids in the bed are effectively
removed. Thus in a sand bed with grain size of 0·5–1·0 mm the voids are likely to be of
the order of 100–200 µm. In such a bed, removal of fine colloidal particles such as silt
and bacteria with sizes of about 1 µm are readily demonstrated. It is evident, therefore,
that straining cannot be the only mechanism operating in a bed of granular medium and
indeed in most situations straining plays a relatively insignificant part in the overall
removal of turbidity. Removal of suspended particles is controlled by a number of
transport and attachment mechanisms. The transport mechanisms move particles into the
vicinity of grains of medium in the bed when the attachment mechanisms can operate to
trap the particle.
1.4.1. Transport Mechanisms

In most filters laminar flow conditions prevail and thus the velocity in a void within the
bed varies from zero at the grain surface to a maximum at the centre of the void.
Transport mechanisms thus have to be able to move suspended particles out of the flow
streamlines into lower velocities near the bed grains.
The important transport mechanisms in water filtration are:
1. Interception: streamlines may pass close enough to bed grains so that suspended
particles actually come into contact with the grain surface.
2. Diffusion: colloidal particles are influenced by random Brownian movements which
may bring them into the vicinity of the grain surface. This mechanism is only
important for very small particles (<1 µm).
3. Sedimentation: gravitational forces acting on particles can move them across
streamlines and bring them to rest on upward-facing grain surfaces.
4. Hydrodynamic: particles in a velocity gradient tend to rotate and are thus subjected to
lateral forces which can move them across streamlines.
Because of the relation between viscosity of water and temperature, changes in
temperature can significantly alter the effectiveness of transport mechanisms.
1.4.2. Attachment Mechanisms

Once suspended particles have been brought into close proximity to grain surfaces or
existing deposits, attachment can take place under the influence of physico-chemical and
molecular forces similar to those which operate in the coagulation process. Thus the
concepts of bridging and double layer formation in coagulation can be used to explain the
entrapment of particles in a granular bed. It is postulated that attachment is due to the
interactions between the approaching particles and the bed grain as a result of the
molecular attractive forces and the electrokinetic forces. In practice, molecular forces are
only likely to be important when the particle has approached very close to the grain
surface, probably within 0·1 µm. Thus the forces between the charged particles are
mainly responsible for controlling the progress of the attachment process. Variation of
the electrokinetic force depends upon the sign and magnitude of the surface charge and
the thickness of the double layer. With most natural waters containing significant
dissolved solids the charges on suspended particles and filter grains are probably fairly
small so that only weak repulsive forces are produced. These can be modified by
alterations in the ionic strength of the system resulting from the addition of coagulants or
pH control. In addition the use of polyelectrolytes can introduce bridging mechanisms
which can bind particles to the grain surfaces. The nature of the voids in a granular bed
provides continual changes in direction and velocity of flow so that particles may grow in
size owing to flocculation as they penetrate into the bed. Thus particles which are not
successful in reaching a grain surface in the upper layers of a bed may so change in
characteristics that they become trapped lower down in the bed.
Filtration 7
1.4.3. Hydraulic Considerations

A basic understanding of the hydraulics of filtration can be made by assuming a single-
size medium in the filter shown schematically in Fig. 1.
D’Arcy’s relationship for head loss per unit depth is
(1)
where h=loss of head in a bed of depth l with face velocity υ, and k =coefficient of
permeability.
A more comprehensive relationship was developed by Carman1 based on the Kozeny
equation
(2)
FIG. 1. Schematic arrangement of

gravity filter.
where E=150[(1−f)/R]+1·75; R=Reynolds Number (υd/v); f=bed porosity;
d=characteristic dimension of bed grain; ψ=particle shape factor=(surface area of sphere
volume V/surface area of bed grain volume V); v=kinematic viscosity; and g=acceleration
due to gravity. The Carman-Kozeny equation predicts the head loss characteristics of a
clean bed with single-size medium. Filter beds are normally composed of a graded
medium so that eqn (2) must be solved as a stepwise integration between adjacent sieve
sizes.
When a filter receives water containing suspended matter some or all of these solids
will be trapped in the bed thus reducing the void space and increasing the head loss (Fig.
2). In these circumstances the total head loss across a bed is the composite of the ‘clean’
head loss plus an additional head loss due to deposits within the bed. Ives and Gregory2
have shown that this additional head loss is a linear function of the specific deposit, σ, in
the bed.
FIG. 2. Head loss during filtration.

Specific deposit is defined as the volume occupied by deposit in the bed per unit bed
volume. Thus the total head loss, H, across a bed containing deposited particles is given
by
(3)
where H0=Carman-Kozeny head loss, and K=a constant.

If the bed retains the majority of the particles in the feed Ives and Gregory2 showed
that an approximate solution of eqn (3) for a unisized medium is
(4)
where C0=feed concentration to bed, and t=time.

For graded medium the expression is basically similar but replaces K with another
factor to express the effect of size grading.
1.4.4. Solids Removal

In many respects it is the performance of a filter in removing suspended matter which is
of primary importance and this is an area which has received considerable attention from
research workers. Most considerations are based on the proposition, originally formulated
by Iwasaki3 for slow filters, that the removal of suspended particles with depth in a filter
bed is proportional to the concentration of particles in suspension, i.e.
(5)
where c=concentration of particles at depth/in bed, and λ=a constant termed the filter
coefficient.
Filtration 9
Ives4 has developed this concept to permit modelling the filtration process in three
stages. An initial spherical grain format changes to a situation analogous to flow through
capillary tubes as deposits accumulate in the voids. In the final stage of a run velocities in
the voids increase to the point where previously deposited material is scoured. The
mathematical expression of this concept can be developed to give the relationship
between the filter coefficient at some stage of the run and the initial filter coefficient (λ0)
at the start of the run
(6)
where σu=ultimate specific deposit, and β, x, y, z are empirical constants. Several other
workers, notably Minz,5 Deb,6 Camp,7 Mohanka8 have produced variations on the theme
of a physical model of the filtration process but in spite of intensive study the complex
nature of filtration has meant that it is not yet possible to obtain a reliable predictive
model for universal application. Nevertheless the models which are available can be
useful in helping to produce more economical design and operating procedures.
1.4.5. Filter Cleaning

In all filters the accumulation of deposits in the bed will eventually terminate the run
either owing to a maximum allowable head loss being reached or owing to breakthrough
of suspended matter with consequent deterioration in filtrate quality.
With slow filters breakthrough is unlikely so that the run will be terminated on a head
loss criterion. 11 is customary to remove about 10 mm of the sand which contains
virtually all of the deposit and this material is washed and replaced on the bed when
necessary.
Backwashing of rapid filters is achieved by introducing filtered water into the bottom
of the bed at relatively high rates of flow to give superficial velocities of 0·3–0·5 m/min.
In hot climates higher upflow velocities are used because the reduced viscosity at
elevated temperatures hinders expansion. The upward flow of water produces an
expansion of the bed to give a fluidised state in which the voids are increased in size and
a considerable scouring action occurs with resultant release of deposits from the bed
grains. As the backwash starts the bed begins to expand and the head loss increases until
the whole bed is just suspended in the upflow. At this point the upward force of water
balances the gravitational force on the grains suspended in water. Further increase in
backwash flow will increase the expansion of the bed but will not necessarily give better
cleaning action since the scouring effect may be reduced at large expansions.
At the point of fluidisation
hegρw=le(ρs−ρw)g(1−fe)
(7)
where he=head loss in expanded depth le; ρw=mass density of water; ρs=mass density of
bed grains; and fe=porosity of expanded bed.
The bed grains are kept in suspension because of the drag force exerted on them by the
upflow of water. This drag force is a function of the velocity onto the bed and the
expanded porosity. An empirical relationship for the expansion of a unisize grain bed of
normal depth l given by Fair et al.9 is
(8)
where υe=wash rate as superficial velocity, and υs=bed grain settling velocity.
With a conventional graded bed each layer is carried into suspension in turn as the
wash rate increases and an arithmetic integration of eqn. (8) between sieve sizes is
necessary to determine the overall expansion produced by a particular wash rate.
UK practice is to operate with bed expansions during backwashing of around 20%
which when supplemented by air scour, prior to and sometimes together with the water
wash, give efficient cleaning. In the USA, air scour was not common in the past and it
was customary to use much higher wash rates to give bed expansions of up to 50% often
supplemented by surface wash arrangements. There seems little justification for such
excessive expansions which must be costly to provide both from capital and operational
aspects.
1.5. TRENDS AND DEVELOPMENTS IN FILTRATION
The preceding sections have set out some of the basic factors governing the process of
filtration as it is conventionally used for the production of potable water. It will be
apparent that the process is one of considerable complexity and as such the performance
of a filter is likely to be influenced by many parameters. Over the last few years there
have been a considerable number of developments in configuration, design and operation
of filters aimed at improving their efficiency and cost effectiveness whilst maintaining or
even improving filtrate quality. Most of these developments have been related to rapid
filters since the nature of the slow filter gives less scope for useful change.
1.5.1. Effective Use of Bed Capacity

The inevitable consequence of backwashing a bed of graded medium is that the grains
settle to give a stratified bed with the finest particles at the top and the coarsest at the
bottom. This does not give efficient use of the voids in that a limiting terminal head loss,
largely due to clogging of the relatively small voids in the upper layers of the bed, will be
reached long before the majority of the voids have been utilised. A number of
modifications to the conventional filter could provide a means of increasing the
penetration of solids into the bed and hence providing fuller utilisation of the void
capacity. The use of larger filter media allows increased penetration and the size of
individual voids will be greater. The total volume of voids will however remain the same
if the grain shape and packing are unaltered. The specific surface of the larger grains will
be less so that particles of suspended matter are less likely to be trapped in any given
void. The use of larger media in itself does not alter the basic characteristic of a
downflow bed of graded material in that the bed becomes less efficient in terms of solids
Filtration 11
removal in the direction of flow. There are also clearly limits as to the amount by which
grain size can be increased before breakthrough of solids becomes a problem. The
essential need is a way of either increasing the specific surface in the direction of flow or
of reducing the velocity in the direction of flow. These aims can be achieved in several
ways. Increased specific surface in the direction of flow can be achieved by using dual or
multi media downflow beds or by using a graded single medium in an upflow
configuration, both of which techniques are now well established. Reduction of velocity
in the direction of flow can be obtained in radial flow filtration which to date has had
only limited application.
Dual and Multi Media Beds

By using more than one medium in a bed it is possible, by selection of appropriate
materials and gradings, to improve utilisation of bed capacity. Table 2 gives examples of
types of media which have been used in water filtration for this purpose. By far the most
popular combination of media is anthracite and sand usually aranged as about 0·3 m of
1·25–2·50 mm anthracite above a similar depth of 0·50–1·00 mm sand. With such an
arrangement good separation of the layers can be maintained even after vigorous
backwashing. In operation the bulk of suspended particles should
TABLE 2 TYPES OF FILTER MEDIA
Material Specific Typical size
gravity range (mm)
Polystyrene 1·04 1·5–3·5
beads
Anthracite 1·4–1·7 1·0–2·5
Sand 2·5–2·6 0·5–1·5
Garnet 3·8 0·3–0·7
Ilmenite 4·8 0·3–0·7
be trapped in the anthracite layer with its relatively low head loss per unit depth, the sand
layer in effect serving as a second stage filter. Figure 3 shows the advantage of a dual
media filter over a single layer unit in terms of run length for a specified terminal head
loss. Provided the characteristics of the suspended particles are suitable in respect of size,
floc strength and physicochemical properties, the filtrate quality from a dual media bed is
usually the
FIG. 3. Head loss relationships for

single and dual media beds of same
overall depth.
same as that produced by a single layer bed of the same overall depth using the same
sand. The lower initial head loss and more efficient usage of the voids in a dual media
bed permit longer filter runs than with a single medium bed at the same filtration rate, or
alternatively permit operation of the dual media bed at higher filtration rate for the same
length of run. This latter feature can be of particular value in uprating the capacity of an
existing filter installation by removing half the original sand and replacing it with a layer
of anthracite. In this way it may be possible to double the capacity of a filtration plant at a
cost much less than the construction of duplicate single medium filters. The influence of
filtration rate on removal of suspended matter does not appear to be great at least when
relatively large floc particles are concerned, Tebbutt.10 However it should be appreciated
that at high filtration rates more of the particles may penetrate into the sand layer causing
a more rapid increase in head loss and/or early breakthrough. A comprehensive review of
the use of anthracite in water filtration has been produced by the Water Research
Association.11
A natural extension of the dual media principle is to add a further layer or layers to the
bed in an effort to produce the desired decrease in void space with increasing depth into
the bed. This can be accomplished by adding a coarse polystyrene bead layer to the top of
an anthracite-sand bed or placing a layer of fine garnet or ilmenite at the bottom of an
anthracite-sand bed. The latter approach has achieved some popularity in the USA in the
form of a proprietary filter incorporating carefully sized layers of anthracite, sand and
garnet in such a way that there is a degree of intermixing between layers and sharp
Filtration 13
interfaces are thus avoided. Conley and Hsuing12 have described the design of such beds
which are stated to be more efficient than dual media installations. Multi media beds have
received little interest in the UK partly owing to the increased cost and also to the law of
diminishing returns which affects all treatment processes. Some pilot scale studies have
recently been undertaken in the UK and have been reported by Lekkas et al.13 In this
work the performance of anthracite-sand and polystyrene-anthracite-sand beds were
compared when treating low turbidity (usually about 3 FTU) stored water. The three layer
bed gave up to twice the length of run achieved by the dual layer bed for the same filtrate
quality (<0.5 FTU) owing largely to prevention of surface clogging which took place in
the dual layer unit. The studies were carried out over a range of filtration rates, 180–360
m3/m2d, and it was noted that differences in performance between the two beds were not
apparent at the higher rates because then more suspended solids were carried through to
the sand layer in both units.
In practice it would appear that dual media beds are the most cost effective way of
improving the utilisation of a downflow filter but it should be realised that such a bed
may be more sensitive to changes in loading and nature of suspended solids than a
conventional single-medium bed which has, in effect, a fairly high inbuilt factor of safety.
Upflow Filters
To overcome the undesirable effect of stratification produced by backwashing of a graded
bed an obvious solution is to alter the direction of filtration from downflow to upflow so
that the small voids on the top of the bed will only receive particles not trapped in the
lower regions of the bed. The principle of upflow filtration has been known for a
considerable time and indeed some units were in operation in the eighteenth century. It
was however not until the advent of the Immedium unit in the 1950s that upflow filtration
became more than a curiosity. Since then a considerable number of Immedium filters
have been installed for both potable water treatment and tertiary treatment of effluents.
The basic problem in upflow filtration is that of preventing the upward flow of water
from expanding the bed as occurs during backwashing. It is thus necessary to limit the
filtration rate to avoid this effect or to provide some means of holding the bed in place to
prevent expansion during filtration. The Immedium filter achieves this end by the
insertion of a coarse grid in the top of the bed, Fig. 4(a), against which the bed grains
arch under filtration. At the higher backwash rates the arched grains are no longer able to
resist the upward force so that expansion can proceed normally.
FIG. 4. (a) Immedium type upflow

filter. (b) AKX or biflow filter.
Studies by Gregory et al.14 show that upward flow filters do have a high bed capacity
thus allowing long filter runs. There may however be need for care in operation to
prevent sudden breakthrough and variable quality raw waters could cause problems in
this respect. Because of the high bed capacity and full solids penetration efficient
backwashing is particularly important for upflow filters. In the investigations, which were
carried out on a pilot scale Immedium type unit, flow rates of up to 350 m3/m2 d were
used successfully and good filtrate turbidities of less than 1 FTU were obtained from a
variety of raw waters. It was however felt that because of the nature of the upflow filter
its use as the final filtration stage in place of rapid gravity units would not be really
suitable.
An American upflow unit, L’Eau Claire, has been described by Haney and Steimle.15
This uses a dual-layer sand bed, 0·9 m of 1·8 mm and 0·9 m of 0·95 mm, with retaining
grids above both layers. Pilot studies showed that a filtrate of less than 0·1 FTU could be
produced from polyelectrolyte-dosed Mississippi water at filtration rates of up to 500
m3/m2d.
A rather different approach to upflow filtration has been developed in the USSR as the
biflow or AKX filter, Fig. 4(b). Here a portion of the flow is fed into the top of the filter
as in a normal downflow unit and serves to prevent expansion of the remainder of the bed
under the influence of the upflow. Ray16 reports that by using dual media biflow units,
filtration rates of up to 430 m3/m2 d have been achieved in the USSR. The downflow
portion of the bed is usually about a quarter of the total bed depth and it is claimed that
the biflow process gives savings in capital and operating costs of 15–30% as compared
with conventional downflow filters.
Radial Filters
In a radial filter with inlet at the centre and outlet round the periphery there is a decrease
in velocity of flow as the water moves outward through the bed. This velocity distribution
Filtration 15
encourages greater penetration of suspended solids into the bed and thus should improve
the utilisation of bed capacity. Radial flow filters in the form of fixed beds have received
some experimental attention on a laboratory basis. Studies by Ives4 suggest that the
performance of a radial flow bed would compare favourably with a
FIG. 5. Typical radial flow filter with

fluidised bed and continuous
desludging by air lift.
conventional vertical flow bed. There are however operational problems with a fixed-bed
radial flow system in that conventional backwashing cannot be carried out and thus some
other form of cleaning would have to be devised. Because of the cleaning problems
associated with fixed beds, commercial development of the radial flow filter has been
directed towards the use of fluidised bed units with continuous washing and recycling
facilities. The Tenten unit is an example of this concept which is shown diagrammatically
in Fig. 5. Ray16 quotes the use of fluidised-bed radial flow units for removing inorganic
turbidity from a groundwater to give a finished water of about 50% lower turbidity than
the feed which averaged 1·2 units. The commercial units are factory built and occupy
relatively small floor area having a typical surface loading of around 300 m3/m2 d
depending upon type and concentration of suspended solids.
1.5.2. Process Modifications

In efforts to produce less costly forms of water treatment some interest has developed in
the use of high rates of filtration. A study in the USA by King et al.17 reported that many
regulatory authorities viewed filtration rates in excess of about 175 m3/m2d as being high
rate but that some authorities were prepared to approve filtration rates of up to 570
m3/m2d. In the UK most studies of high rate filtration have been carried out for tertiary
treatment purposes (Tebbutt,10 West et al.18) where at filtration rates of up to 800 m3/m2d
little correlation was found between filtration rate and suspended solids removal.
Considerable interest has appeared recently in the USA in the direct filtration process,
i.e. the filtration of raw water usually with the addition of coagulant and polyelectrolyte.
The process is not of course new in that pressure filters must operate in this manner and
direct filtration of good quality raw waters on rapid gravity filters has been employed for
years. The more recent studies have been concerned with determining the potential of
direct filtration when aided by the modern armoury of polymers. Logsdon19 pointed out
that the process was normally only suitable for relatively low turbidity raw waters;
average turbidity for US plants was 25 FTU with average colour 20°H. Of 16 plants
surveyed only one had average filtrate turbidity greater than 1 FTU although five
exceeded 1 FTU at least once a year. Culp,20 however, suggested that direct filtration
could satisfactorily handle poorer quality raw waters and indicated practicable limits of
100–200 FTU, 100°H, with turbidity and colour together not exceeding 25 units each.
Stump and Novak21 have emphasised the importance of the correct polymer to obtain
optimum performance in direct filtration and found cationic high molecular weight
compounds to be most satisfactory. The importance of the correct type of floc for input to
a direct filter has been emphasised by Treweek22 who found that a Gt value of 4·2×104
was adequate to produce a filtrate turbidity of less than 0·1 FTU. Increasing Gt values of
up to 2·7×105 gave no further improvement in filtrate quality although the floc particles
so formed were visibly larger. Recent UK pilot studies on direct filtration of stored water
of about 3 FTU turbidity have been reported by Lekkas et al.13 There is clearly need for
comparative studies to be undertaken on the performance of conventional coagulation,
sedimentation and filtration plants with those using coagulation and direct filtration.
Economic advantages could result from the abandonment of the sedimentation stage with
at least certain types of water currently using conventional treatment. Pilot scale studies
by Vianna23 have shown that when treating the same raw water a direct filter may be able
to produce a better quality filtrate than a conventional plant, and this work is continuing.
A revival of interest in slow filtration has resulted from WHO-sponsored studies in a
number of developing countries. Here the simplicity of construction and operation may
counterbalance the high capital cost. Paramasivam and Sundaresan24 have shown from
studies in India that slow filters operating at rates of between 2·4 and 7·2 m3/m2 d will
produce a filtrate of less than 1 FTU from an input of up to 30 FTU. Work in Thailand by
Komolrit et al.25 has shown that input turbidities of up to 50 FTU can be handled by slow
filters but that capital costs are still high. Frankel26 has described a two stage filtration
system capable of dealing with raw waters of 100–400 FTU without the need for
chemical coagulation. To reduce capital cost local media such as shredded coconut husks
or burned rice husks have satisfactorily replaced sand. The first roughing filter operates at
30 m3/m2d and the second polishing filter can operate at 2–30 m3/m2d. Final filtrate
quality of less than 0·3 FTU was achieved and the inexpensive media could be replaced
about once a month in the roughing filter with the polishing filter only requiring removal
and replacement of the superficial layer about 30 mm deep at the same frequency. It
seems likely that the water supply needs of developing countries can in many instances
Filtration 17
be best satisfied by the adoption of such forms of appropriate technology using local
materials and skills wherever possible.
1.5.3. Filter Control

Because of the long runs achieved by slow filters the hydraulic characteristics do not
change significantly from day to day and thus manual adjustment of the throughput is
satisfactory. In the case of rapid filters virtually continuous adjustment of control valves
is necessary to maintain constant filtration rate. This feature of rapid filtration has
normally been catered for by provision of some form of automatic flow controller to
ensure constant output. Various proprietary devices are available which provide a high
head loss at the start of a run when the filter head loss is low. As the run proceeds and
filter head loss increases the hydraulic control is gradually relaxed so that the overall
head loss through filter bed and controller remains constant. These flow control units are
relatively expensive and are not always trouble free so that simpler forms of filter control
are becoming more popular. Constant rate filtration can be achieved by using a deeper
filter shell and allowing the top water level to increase as the head loss in the bed builds
up. Flow to the filter is set at a constant rate with free discharge over a weir into the shell.
The additional capital cost of the extended shell depth is a disadvantage of this type of
operation. In declining rate filtration using a battery of filters a fixed top water level is
used so that as the bed clogs the output declines. Provided the backwashing of beds is
staggered, a reasonably constant output will be maintained from a battery of several
filters and the process is claimed to offer a number of advantages over constant rate
operation. In particular a decline in filtrate quality towards the end of a run does not
occur, filter runs are longer and capital and operating costs are lower. In spite of these
claims relatively few declining rate installations have been built in the UK although they
are more popular in the USA and Japan.
Arboleda27 has discussed the various forms of hydraulic control techniques used in
filtration and concluded that declining rate filtration did offer advantages over constant
rate operation. It seems likely that this type of control would certainly be more
appropriate than the conventional constant-rate method for developing country situations.
At the other end of the scale the automation of filtration plants is becoming increasingly
common with large installations being equipped with process control and monitoring
facilities to handle both filtration and backwashing stages of operation.
1.5.4. Backwashing
The development of more effective filtration techniques with greater penetration of
suspended solids into the bed leads inevitably to the need to ensure that the bed is
efficiently cleaned. The growing use of granular beds for the filtration of wastewater
effluents has also placed increased demands on cleaning arrangements since the organic
nature of the deposits can result in difficulties in their removal from filter media.
In a basic study by Amirtharajah28 it has been stated that the main cleaning mechanism
in a backwashing situation is hydrodynamic shear and that abrasion between grains is
negligible. The optimum porosity for cleaning was found to be about 0·70 which for a
graded sand would be likely to occur at 40–50% expansion of the top layers.
Backwashing with water alone was found to be an inherently weak process and air scour
or surface wash were felt to be indispensable for effective cleaning. Cleasby and
Lorence29 carried out a long term study of backwashing techniques for tertiary effluent
filters. Techniques studied were: water wash only, air scour followed by water, surface
and subsurface wash before and during main wash, simultaneous air scour and
subfluidisation water wash. The last method was found to be most effective for relatively
coarse sand beds although there was some danger of loss of media because of the
violence of the air and water action. Gould and Patterson30 studied three backwash
methods on a laboratory scale using an iron floc to provide the deposit. Procedures used
were: air scour at 0·45 m/min for 3 min followed by water at 0·32 m/min for 5 min, water
at 0·8 m/min for 5 min, and air at 0·9 m/min with water at 0·32 m/min for 5 min with the
water continuing alone for a further 5 min. The first and third techniques gave better
cleaning than the water only wash even though a high velocity wash was employed. It
was felt that the simultaneous air and water wash was likely to give the best long term
performance. It is of interest to note that this simultaneous technique is popular in Europe
although for it to be applied successfully care must be taken with the design of the
underdrains, filter bottom and gravel layer.
Amirtharajah and Cleasby31 have described a modified model of backwashing which
predicts the required wash rate from knowledge of media characteristics and the density
viscosity of the backwash water.
1.5.5. Design Methods

Because of the complex nature of the filtration process many parameters can affect
performance. These include: feed water characteristics, required filtrate quality, rate of
flow, available head, media type, grading and arrangement, operating temperature range.
The cost of a filtration plant will be influenced by such factors as: number of units, length
of run, required output per cycle, backwash arrangements, costs of construction for filter
shells and for pipework, valves, controls, etc.
In terms of optimised design it is important to appreciate that in general there will be a
gradual rise in filtrate turbidity as a run proceeds together with a rise in head loss across
the bed. Both filtrate turbidity and head loss have limiting values the exceeding of which
will necessitate termination of a run. Optimum operational performance will occur when
both of these limits are reached simultaneously. By using pilot-scale filters or
mathematical models it is possible to produce plots of the form shown in Fig. 6 which
relates the depth of bed and the length of run to satisfy the filtrate quality and head loss
criteria. The intersection of the two curves
Filtration 19
FIG. 6. Determination of optimum bed

depth and filter run based on head loss
and filtrate quality criteria. (Fixed
filtration rate and single-size medium.)
provides the optimum solution for a given filtration rate and medium. Additional points
are obtained for different filtration rates and different media with the end-product being
capable of expression in the form of a response surface which shows the operational
optimum conditions (Fig. 7). Ives32 has recently given a full description of this procedure
and has considered in some detail the economic optimum design of filters. Adin33 has
described a technique for predicting filter performance in an optimum design using the
similarity which exists between turbidity breakthrough in a filter and the breakthrough
pattern in an adsorption bed.
Since the various mathematical models of filtration which have been produced tend to
be restricted in application to special circumstances it is generally agreed that information
for use in optimising models is best obtained from pilot studies using the actual media
and water under consideration. The design of such studies has been detailed by Ives.34
A problem which arises in filtration work is that of characterising the
FIG. 7. Response surface for optimum

filter design.
nature of the suspension to be treated. This factor was appreciated during the
development of microstrainers with the introduction of the filterability index. This
concept can also be used to relate the filterability of a suspension to filter media taking
into account clarification clogging and flow rate as described by Ives.35
REFERENCES
1. CARMAN, P.C., Trans.Instn Chem. Engrs, 1937, 15, 150.

2. IVES, K.J. and GREGORY, J., Proc. Soc. Water Treat. Exam., 1967, 16, 147.
3. IWASAKI, T., J.Amer. Water Works Assoc.. 1937, 29, 1591.
4. IVES, K.J., Proc. 8th Congr. Int. Water Supply Assoc., Vienna, 1969, 1, K1.
5. MINZ, D.M., Proc. 6th Congr. Int. Water Supply Assoc., Stockholm, 1964, 1, E1.
6. DEB, A.K., J. Sanit. Eng. Div., Am. Soc. Civil Engrs, 1969, 95, 399.
7. CAMP, T.R., J. Sanit. Eng. Div., Am. Soc. Civil Engrs, 1964, 90, 1.
8. MOHANKA, S., J. Sanit. Eng. Div., Am. Soc. Civil Engrs, 1969, 95, 1079.
9. FAIR, G.M., GEYER, J.C. and OKUN, D.A., Waste Water Engineering, 2. Water Purification
and Waste-Water Treatment and Disposal, 1967, John Wiley, New York.
10. TEBBUTT, T.H.Y., Water Res., 1971, 5, 81.
11. Water Research Association, Anthracite Sand Filtration, Conference Papers, 1972, WRA,
Medmenham, UK.
12. CONLEY, W.R. and HSUING, K., J. Amer. Water Works Assoc., 1969, 61, 97.
Filtration 21
13. LEKKAS, T.D., Fox, G.T.J. and MCNAUGHTON, J.G., J. Instn Water Engrs & Scientists,
1978, 32, 239.
14. GREGORY, R., MCNAUGHTON, J.G. and MILLER, D.G., Experiences with Upflow
Filtration, TP 102, 1973, WRA, Medmenham, UK.
15. HANEY, B.J. and STEIMLE, S.S., J. Environ. Eng. Div., Am. Soc. Civil Engrs, 1975, 101, 489.
16. RAY, W.J.F., Proc. 10th Congr. Int. Water Supply Assoc., Brighton, 1974, 1, B1.
17. KING, P.H., Johnson, R.L., RANDALL, C.W. and REHBERGER, G.W., J. Environ. Eng. Div.,
Am. Soc. Civil Engrs, 1975, 101, 479.
18. WEST, J., RACHWAL, A.J. and Cox, G.C., J. Instn Water Engrs & Scientists, 1979, 33, 45.
19. LOGSDON, G.S., Civil Eng, 1978, 48 (July), 68.
20. CULP, R.L., J. Amer. Water Works Assoc., 1977, 69, 375.
21. STUMP, V.L. and NOVAK, J.T., J. Amer. Water Works Assoc., 1979, 71, 339.
22. TREWEEK, G.P., J. Amer. Water Works Assoc., 1979, 71, 96.
23. VIANNA, M.D.B., Performance Relationships in Water Treatment, M.Sc. Thesis, 1979,
University of Birmingham.
24. PARAMASIVAM, R. and SUNDARESAN, B.B., Aqua, 1979, (3), 13.
25. KOMOLRIT, K., CHAINARONG, L. and BUASEEMUONG, S., Aqua, 1979, (4), 12.
26. FRANKEL, R.J., J. Amer. Water Works Assoc., 1974, 66, 124.
27. ARBOLEDA, J., J. Amer. Water Works Assoc., 1974, 66, 87.
28. AMIRTHARAJAH, A., J. Environ. Eng. Div., Am. Soc. Civil Engrs, 1978, 104, 917.
29. CLEASBY, J.L. and LORENCE, J.C., J. Environ. Eng. Div., Am. Soc. Civil Engrs, 1978, 104,
749.
30. GOULD, M.H.and PATTERSON, P., Water Services, 1979, 83, 573.
31. AMIRTHARAJAH, A. and CLEASBY, J.L., J. Amer. Water Works Assoc., 1972, 64, 52.
32. IVES, K.J., in Mathematical Models in Water Pollution Control, A.J.James (Ed.), 1978, John
Wiley and Sons, Chichester, 339.
33. ADIN, A., Filtrn. Separn., 1978, 15, 55.
34. IVES, K.J., Effl. Water Treat. J., 1966, 6, 552, 591.
35. IVES, K.J., Prog. Water Tech., 1978, 10, 123. A comprehensive review of filtration is
presented in The Scientific Basis of Filtration, K.J.Ives (Ed.), 1975, Noordhoff, Leyden.
Chapter 2
REMOVAL OF ORGANIC
COMPOUNDS
C.S.SHORT B.Sc., C.Eng., M.I.Chem.E., M.I.W.E.S.
Yorkshire Water Authority, Leeds, UK
SUMMARY
The use of activated carbon for municipal water treatment continues to grow as water
quality standards are raised and polluted sources are increasingly drawn on. Minor
developments in technique have been introduced in the 1970s, principally in regeneration
furnace design. New resinous adsorbents have been developed for removing a wide range
of organics, but these are not at present competitive with carbon for potable water
treatment,particularly because of the problems associated with liquid regenerants. The
high organics removal performance of reverse osmosis has been demonstrated but it
remains prohibitively expensive, while the related process of ultrafiltration is cheaper but
effective only for removing the larger organic species. It is concluded that activated
carbon adsorption is the process of choice for many duties from taste and odour removal
to the general reduction of synthetic organics. Ion exchange resins will continue to be
used for the removal of naturally occurring organic acids in the production of deionised
water.
2.1. ORGANIC COMPOUNDS IN WATER
Organic compounds are present in most surface waters, but are rarely found in significant
quantity in groundwaters. The most common naturally occurring organic compounds are
humic and fulvic acids derived from vegetation. In addition, rivers receiving sewage and
industrial effluents and run-off from agricultural and urban catchments may contain a
very wide range of synthetic organic compounds. Total levels of organic matter can be
determined in terms of organic carbon, or indirectly in terms of u.v. absorption. Many
individual substances have been detected, generally at very low concentrations. However,
the most advanced current analytical techniques cannot identify more than 10 to 20% of
the individual organic compounds present in polluted surface waters. Development of
high performance liquid chromatography in conjunction with mass spectrometry will
allow the detailed investigation of the remaining fraction.1
Removal of organic compounds 23
2.2. OBJECTIVES
The objectives of organics removal depend on the application. Table 1 indicates the
organic compounds of concern in a number of applications. For some uses organics in
general are undesirable, whereas for others specific compounds or groups of compounds
must be removed. However, currently available treatment methods can be made specific
for given types of compound only to a limited extent, as will be discussed in later
sections.
Knowledge of the health effects of organic compounds at the low levels found in
water supplies is very limited. However, where a water source is known to contain
significant amounts of industrial effluent, this has given rise to concern. Furthermore,
naturally occurring organic compounds have been found to react with chlorine, during the
disinfection process, to form a variety of chlorinated organics some of which are thought
to be potentially hazardous. As organics removal can be expected to add considerably to
treatment costs, there is an urgent need to improve our knowledge of the relationships
between organic compounds and health. In the meantime, if the current (1979)
requirements of the United States Environmental Protection Agency (EPA) and the EEC
are to be met, a considerable expenditure on organics removal for health reasons may be
anticipated in the next decade.
2.3. THE PERFORMANCE OF CONVENTIONAL AND

OXIDATIVE PROCESSES
Both colloidal and dissolved organic matter are removed to some extent by conventional
processes, but a limited amount of information is available on their performance except in
respect of colour removal.
Slow sand filtration may be expected to remove colloidal material very
TABLE 1 OBJECTIVES OF ORGANICS
REMOVAL
Application Organic Reason for Maximum
compounds of concern level
concern recommended
Potable water Organics arising Possible health Reduce to
supply from industrial effects lowest
effluents practical level
(EPA) PV 5
mg/litre (EEC)
Carbon
chloroform
extract
0.1mg/litre
(EEC)
Organochlorine Possible health 0·1 µg/litre

pesticides effects each, 0·5
µg/litre total|
(EEC)
Endrin 0·2 µg/litre
(EPA)
Lindane 4 µg/litre
(EPA)
Methoxychlor 100 µg/litre
(EPA)
Toxaphene 5 µg/litre
(EPA)
2,4-D 100 µg/litre
(EPA)
2,4,5-TP 10 µg/litre
(EPA)
Polynuclear Possible health 0·2 µg/litre
aromatic effects (EEC)
hydrocarbons
Trihalomethanes Possible health 0·1 mg/litre
effects (EPA)
Compounds Consumer
causing tastes reaction
and odours
Compounds Consumer 15 Hazen units
caus-ing colour reaction (EPA)
Detergents Taste, 200 µg/litre
appearance (EEC) 500
µg/litre (EPA)
Biodegradable Substrate for Total Organic
organics microbiological Carbon (TOC)
growth in 0·5 mg/litre
mains
Boiler Weak organic Increase in Total
feedwater acids conductivity conductivity
0·1 µS/cm for
high pressure
boilers
Pretreatment Weak organic Cause 0·4 kg PV per
for ion acids irreversible m3 resin per
exchange and fouling of cycle, or
electrodialysis resins or PV/TDS 0·002
membrane
Electronic Colloidal Affects –
components material performance of
manufacture components
Soft drinks Polysaccharides Form –
manufacture (products of precipitate at
algal decay) low pH
Preparation of Pyrogens Cause –

injections and pyrogenic
pharmaceuticals reaction in
patients
effectively, while about 30% reduction of dissolved humic matter is achieved probably
by biological oxidation. Reductions of Permanganate Value (PV) reported by Houghton2
apparently show the adverse effect of low temperature on a biological process (Table 2).
Slow sand filters are effective in reducing some tastes and odours, but are also known
to cause them under certain circumstances.3
TABLE 2 PV REMOVAL, CHIGWELL Row
SLOW SAND FILTERS, ESSEX WATER Co.
(From Houghton, Reference 2)
Nov. 1965– April-Oct.
March 1966 1966
Raw water PV mg 2·0 2·2
O2/litre
Filtered water PV 1·5 1·3
mg O2/litre
Removal (%) 25 40
Storage of raw water in open reservoirs may change the organic content, for instance as a
result of the growth and decay of algae. It also appears that volatile organic substances
may be lost to the atmosphere by desorption, as reported by Rook;4 during storage of
approximately 20 days, the Threshold Taste Number was reduced by up to 90% in
summer, 75% in winter.
Coagulation, sedimentation and rapid filtration reduce organic content by precipitation
and adsorption onto coagulant flocs. Over 80% removal of ‘apparent’ colour (i.e. soluble
and colloidal) may be expected. The removal of ‘true’ colour (e.g. that passing through a
0·45 µm membrane filter) depends largely on the coagulant dose. Virtually complete
removal is achievable from those raw waters containing moderate amounts of humic
colour. However, results from the River Trent water treatment study5 indicated that, even
with pre-chlorination, a significant amount of colour remained after coagulation
treatment of this water containing industrial effluents.
Again from the Trent study, total dissolved organic carbon was removed in varying
degree, but up to 40% reduction could be expected. Similar results were obtained when
precipitation softening was practised.
Over 90% removal of polynuclear aromatic hydrocarbons (PAH) may be expected as
these are very insoluble. Residuals of less than 10 ng/litre were achieved at several
treatment plants examined by Lewis.6
The removal of pesticides by coagulation treatment is reported to be negligible.7
The organics remaining after coagulation and clarification are sufficient to form
measurable amounts of haloforms on chlorination.8 Ozonation at 8 mg/litre, with 2 min
contact, suppressed the formation of haloforms by 50%. At 2 mg/litre, 8 min contact, the
reduction was about 65%. However, if a longer contact time occurred between ozonation
and chlorination, the haloform levels were not reduced.
Ozone, though a powerful oxidant, appears to be capable of only partial oxidation of
many organic compounds. Some results are given by Gomella.9 Ozone was found to
render various river water odours less unpleasant, while not sufficiently reducing odour
intensity.10 From the point of view of health, the products of ozonation may be no more
acceptable than the original compounds. However, they may be more readily bio-
degraded in subsequent activated carbon beds; this is discussed in a later section. Recent
studies11,12 indicate that the action of ozone and u.v. radiation together readily oxidises
some organics to carbon dioxide and water. It is unlikely, however, that this would be
economically acceptable on a practical scale for municipal water treatment.
As with ozone, other oxidising agents such as chlorine, chlorine dioxide and
potassium permanganate break down organic molecules incompletely and therefore do
not necessarily reduce any health risk. Indeed, chlorine at normal disinfection doses
forms possibly undesirable substitution products with organic substances, as discussed
above. Chlorine dioxide and potassium permanganate are reported to reduce various
natural organic odours but at rather high doses.10 Chlorine dioxide is used for disinfection
as an alternative to chlorine when phenols are present in the raw water, as it does not
form chlorophenol tastes but partially oxidises the phenols.13
To summarise, the reduction in the level of organic compounds achieved by
conventional processes is generally insufficient to meet the latest requirements for
industrial water treatment, as indicated in Table 1, or for potable water when polluted
sources are used. Nevertheless, such reduction as is achieved incidentally to the normal
clarification and colour removal role of these processes may significantly reduce the cost
of subsequent organics removal processes. Oxidants modify rather than remove organic
matter and their application is limited to the removal or avoidance of tastes and odours
except for their possible use to enhance the performance of activated carbon beds.
2.4. ACTIVATED CARBON
2.4.1. Introduction
The use of activated carbon for municipal and industrial water treatment is well
established. In the municipal field, carbon has been applied almost exclusively to the
removal of odorous substances, for which purpose it is often the most, or only, effective
treatment. There are therefore many examples of its use throughout the world on
reservoir and river abstractions where odour problems are common. Because odours often
occur intermittently and are associated with very low concentrations of the odorous
substance, it has been economical to use powdered carbon in the majority of these plants.
There are a growing number of granular carbon installations, notably on river
abstractions suffering continuously from odours originating in industrial and municipal
effluents. Very few carbon plants have been installed specifically for the removal of
organics to reduce the possible health hazard. However, in 1978 the US Environmental
Protection Agency (EPA) proposed that all municipal water treatment systems in the
USA serving populations greater than 75 000 should use granular carbon for this
purpose.14
The plan of this section is to cover the theory, design and performance of powdered
and granular carbon systems with emphasis on practical applications and recent
developments.
There is a very extensive literature on the use of carbon for water treatment. However,
much of the existing knowledge was brought together at a Water Research Association
conference in 1973, and frequent reference will be made here to the proceedings.15 The
Carbon Adsorption Handbook16 is another encyclopaedic reference.
2.4.2. Theory
Adsorption of solute molecules onto activated carbon is considered to result from a
combination of forces: electrical (ionic attraction), physical (van der Waals) and chemical
combination.15a Major factors in the degree and rate of adsorption of individual solutes
are the polarity of the solute and the molecule size. Adsorption decreases with increasing
polarity and therefore with increasing water solubility. The principal properties of the
activated carbon affecting adsorption are pore size distribution and surface area; surface
chemistry and charge play a smaller part. Carbons activated at high temperature usually
have a net negative (basic) charge. Details of the preparation and chemistry of activated
carbons are given by Cookson.16a
At the low concentrations of solute normally found in water supply applications,
adsorption of many solutes from single component solutions in laboratory batch tests is
found to follow the Freundlich isotherm, which can be expressed as the equilibrium
equation
where Cc is the concentration of solute in the carbon at equilibrium, Cw is the

concentration of solute in the water at equilibrium, and K and n are constants for each
solute/carbon system.
n is less than one for most organic solutes, implying that a greater proportion of the
solute will be adsorbed from a dilute solution than from a concentrated one. On the other
hand, the uptake of solute per unit weight of carbon increases with increasing
concentration of the solute in the solution. It follows that activated carbon is well suited
to the treatment of dilute solutions, but for maximum use of carbon it should achieve
equilibrium with the solute at its most concentrated, i.e. at the feed concentration.
Rate of adsorption depends on diffusion both to and within a carbon particle. The
solute molecular weight, carbon particle diameter and pore size distribution all affect the
rate of intraparticle diffusion, which is generally the rate-cootrolling step. However,
transport of solute to the carbon particles can be appreciably slowed by the occluding
effect of coagulant flocs as in the case where powdered carbon is dosed upstream of floc
blanket tanks or where granular carbon is used in the dual role of adsorber and filter.
Adsorption rates may be determined by stirred batch tests or by granular carbon column
tests. For the case where intraparticle diffusion is rate-controlling, an empirical equation
of the form
dy/dt=−KC ln y
may be used to fit the data, where C is the concentration of solute in the solution at any
instant; y is the concentration of solute in the carbon at the same instant, expressed as a
fraction of the equilibrium concentration in contact with a solution containing the solute
at concentration C (i.e. y can be regarded as the fractional saturation of the carbon); and K
is the rate constant.
These concepts of adsorption rate and capacity are valuable for the comparison of
different carbons and also suggest ways in which the efficiency of use of activated carbon
could be maximised. For instance, it can be seen that granular carbon beds must be used
to achieve maximum carbon use. The upstream layers of the bed will eventually become
saturated with solute at the feed concentration; whereas powdered carbon added before
clarification is well mixed with the water and becomes saturated with solute at the treated
water concentration. However it is not always cheaper to install granular rather than
powdered carbon as capital costs of granular systems are higher. Selection and
optimisation of carbon systems are discussed later in this section.
The relationships in the above equations are empirical, and batch and/or column tests
are invariably required to determine the equilibrium and rate constants for a particular
carbon/water/solute system. Granular carbon plants should not be designed from the
results of batch tests alone as these do not take into account the effects of microbiological
activity, occlusion by suspended solids, and consequent backwashing all of which occur
in granular carbon beds.
2.4.3. Powdered Activated Carbon
The Material
Powdered activated carbon has been widely used for many years and is available in a
range of qualities. It is typically a powder of bulk density around 700 kg/m3 and particle
size less than 75 µm The performance depends on the raw material and the method of
activation. Recent developments include a water treatment carbon comprising a mixture
of coal-based and peat-based material to obtain a wide spread of pore sizes; and a carbon
based on anthracite, activated by steam at 950°C in a fluidised bed furnace.
Point of Addition
Powdered carbon may be added before sedimentation or immediately before rapid
filtration. In either case the carbon must have a sufficiently long contact time to allow
equilibrium to be achieved with the solutes of concern. Removal by flotation would be
rapid and the contact time practically limited to that provided in the flocculation stage.
Performance is adversely affected by the concurrent addition of softening chemicals,
coagulants and oxidants including chlorine.16a,17 In some Japanese plants contact time
before flocculation is provided.18
Gomella19 has shown the economies that may be expected if powdered carbon is
added at two or more positions, provided that equilibrium is approached at each stage.
Lettinga et al.20 proposed a fluidised bed system which would permit greatly improved
powdered carbon utilisation if the fluid bed could be divided into several separate stages.
However, unlike the usual methods of applying powdered carbon, this would require an
expensive special vessel.
Regeneration
Powdered activated carbon is at present regarded as non-regenerable, and it is therefore
used once only before being discarded as part of the sludge arising from clarification.
However, the powder, though considerably cheaper than granular carbon, is expensive
enough to have prompted investigations of possible methods of regeneration and re-use.
Separation of the carbon from coagulant sludge would be a considerable problem.
Successful regeneration of an activated sludge/powdered carbon slurry by wet oxidation
has been reported,16b but the technique is unlikely to be economical at the dose levels
normally associated with potable water treatment.
Engineering of Dosing Systems

A comprehensive review of design considerations and systems for dosing powdered
carbon was made by Aldrich.15b The choice of equipment for delivery, storage and dosing
of the carbon depends to some degree on the size of the installation. Figure 1 gives some
indication of the relationship between size, preferred systems and costs. In larger
installations a dry feeding system will often be required either for dosing direct to the
process or for making up a slurry of known concentration. Dry feeders may be batch or
continuous and deliver a measured volume or a measured weight of powder. Either
method is subject to inaccuracies resulting from variations in bulk density in the case of
volumetric feeders and from variations in moisture content in the case of gravimetric
feeders.
Rigorous dust control is required to protect operators and equipment,
FIG. 1. Costs of powdered carbon

dosing systems (from Burley and
Short, Reference 15c).
but powdered carbon is not thought to represent a significant fire or explosion hazard in
British practice.
A recently installed system for dosing up to 350 kg/h is illustrated in Fig. 2. This
comprises bulk delivery to storage silos, gravimetric dry feeding to produce a slurry of
known concentration, and metered pumping of the slurry into a stream of carriage water
which conveys the carbon to the point of application at a suitable velocity to prevent
blockage or erosion of the pipeline. Slurry preparation and dosing is fully automatic.
Tubular screw conveyor casings are used to prevent the free flow of powder which has
caused ‘flooding’ in some installations. The mid 1979 cost of the mechanical and
electrical equipment was about £65000. Applications and Performance
Powdered activated carbon is used in many countries for taste and odour control, with the
notable exception of the Federal Republic of Germany where only granular carbon beds
have been employed. Doses are typically in the range 5 to 50 mg/litre. It must be
appreciated that carbons are not highly specific for odorous substances and part of their
capacity will therefore be used up in adsorption of non-odorous organics.
The choice between powdered and granular carbon for a given application is often an
economic one and a cost comparison was made by Burley and Short.15c
Sumner21 described the continuous use of powdered carbon at the River Dee treatment
plants. At one plant a minimum dose is added of 12 mg/litre in summer and 8 mg/litre in
winter, to afford a degree of protection against occasional pollutions as well as for taste
and odour reduction. Kennett22 reported that at times doses of up to 80 mg/litre could not
sufficiently reduce tastes in the Derbyshire R.Derwent. Indeed, the persistent tastes in this
source eventually led to the installation of granular carbon beds. The superiority of
powdered carbon to oxidants for treating a range of surface water odours was shown by
Gauntlett.10 Gardiner15d reported that the filter run length of rapid sand filters became
unacceptably short at carbon doses above 25 mg/litre. However, anthracite/sand filters
could cope with doses of 60 mg/litre with no significant increase in head loss
development. Gauntlett17 demonstrated the adverse effects on phenol adsorption of humic
and fulvic acids and alkyl benzene sulphonate. This implied that powdered carbon should
if possible be added at a late stage of treatment after organic material has been
substantially reduced by coagulation and sedimentation.
FIG. 2. Schematic of powdered carbon

dosing system, Elvington water
treatment plant (by courtesy of
Yorkshire Water Authority and
Portasilo Ltd). Key to details: 1.
Pneumatic delivery from bulk tanker.
2. Air vent incorporating automatically
cleaned bag filter. 3. Load cells. 4.
Aeration pads supplied with air at 35
kN/m2. 5. Slide plate isolating valve. 6.
Two-speed screw conveyor. 7.
Pneumatically operated butterfly valve.
8. Pneumatically operated swing valve.
9. Screw conveyor. 10. Vortex mixing
cone. 11. High speed mixer. 12. Low
speed agitator. 13. Duplex slurry
metering pump. 14. Transport water
pumps. 15. Duplicate 50 mm dia. PVC
pipes.
2.4.4. Granular Activated Carbon
The Material
Many grades are available to cover a wide variety of industrial uses. Those grades most
suited to water treatment are generally produced by high temperature steam activation.
The surface area of typical water treatment carbons is in the range 500–1500 m2/g as
measured by the standard BET method (based on the adsorption of nitrogen). Particle
sizes giving the best combination of adsorption kinetics and head loss characteristics are
in the range 0·5–2 mm. The shape is normally irregular, but one carbon is available as
uniformly sized cylindrical pellets 0·8 mm diameter, formed by extrusion.
Particle strength should be sufficient to allow regeneration of exhausted carbon with
attrition losses of not greater than 5%. However, it may sometimes be economical to use
a cheaper, softer grade on a throw-away basis. This is referred to as non-regenerable
carbon. It is possible that the exhausted material could be reprocessed into powdered
activated carbon.
Points of Application
Granular carbon adsorbers are preferably installed following clarification processes. This
arrangement has several advantages;
1. Some organic matter is removed by coagulation and clarification, reducing the load on
the carbon and extending its operating ‘life’ between regenerations.
2. The risk of occlusion of the carbon by particulate matter and coagulant floc is
minimised.
3. Backwashing frequency of the carbon units is minimised. It is inevitable that during
the upflow washing the arrangement of carbon particles in the bed is somewhat
altered. This tends to disrupt the desirable adsorption profile in which the relative
exhaustion of the carbon varies inversely with distance from the bed inlet. Adsorption
efficiency and the operating life of a bed could therefore be significantly reduced.
However, there are a number of plants where, for speed and cheapness of installation,
carbon has been substituted for sand in existing rapid gravity filters; some reported
results are reviewed below. Irregularly shaped granular carbon is an effective filtration
medium and its low density compared with sand makes it economical to backwash.
Where pre-chlorination is used, chlorine levels entering the carbon beds are normally
kept to a minimum. Chlorine coming into contact with the carbon will be reduced to
chloride. The corresponding oxidation of the carbon surface may have an adverse effect
on the adsorption of many organics of concern.16a Also, the cost of restoring the lost
chlorine after the carbon beds must be considered.
Adsorber System Design

A number of predictive mathematical models of granular carbon adsorbers have been
developed, and have shown reasonable agreement with the observed adsorption of single
and binary solute systems.23 However, it is likely that some important factors in carbon
bed performance will remain unquantifiable. Adsorber designs are therefore generally
based on the results of pilot column tests in which samples of water are taken from a
number of depths within the bed at regular intervals of time. The concentrations of the
solutes of interest are measured in each sample and the data can be presented as a plot of
concentration against either time at each depth (‘breakthrough curve’) or depth at each
interval of time (‘adsorption front’). Using the latter (Fig. 3), the minimum depth of bed
to give the required treated water concentration is readily determined, and the progressive
saturation of the carbon is illustrated.
FIG. 3. Idealised adsorption front

curves.
These data are used to evaluate the effectiveness of alternative adsorber designs in
respect of carbon utilisation. Carbon removed for regeneration or disposal should as far
as possible be saturated with respect to the organic solutes of concern. Three systems
which ensure a high efficiency of carbon utilisation are illustrated in Fig. 4.
The carbon system must be designed to accommodate the worst anticipated organic
loads. Gomella19 suggests using carbon beds for base-load operation with powdered
carbon dosing to reduce peak concentrations.
The effect of water velocity (surface loading) on minimum contact time and carbon
utilisation appears to vary according to the particular solutes/sorbents in reported
studies.16c,23,24 Pilot column tests should therefore include parallel trials at different
velocities in the suggested range 2 to 25 m/h.
Head loss and backwash data are provided by suppliers for each grade of carbon.
However, measurements should be made during column tests to
FIG. 4. Granular carbon adsorber

configurations.
observe the effect of feedwater turbidity and biological growths on head loss
development, and the variation of backwash efficiency with time. Bossy and Sanchez25
attributed such variations to changes in carbon density and particle size. However,
washwater temperature is probably the most significant factor affecting bed expansion
during backwashing.
As noted above, backwashing can adversely affect carbon utilisation as it tends to
disrupt the adsorption front. The effect is mitigated by using two or more beds in series,
as demonstrated by the results of Hyde.24
Given the required data, the choice between systems and surface loadings is
principally economic, taking into account capital costs (number and size of adsorbers and
pumps), carbon utilisation, and power costs (related to pressure drop and backwash
requirements). Burley and Short15c developed a computer program permitting the rapid
evaluation of many alternative systems for a given application.
Adsorber Installations
A small number of pulsed-bed adsorbers have been installed for wastewater treatment;
engineering details are given by Culp.26 However, no examples are known in potable or
industrial water treatment applications. Nor have any installations of beds in series been
reported, though their efficiency has been demonstrated in a number of pilot column
investigations. Also, a cost comparison of gravity adsorbers used singly or in reversible
two-bed series indicated that the series combination could be considerably cheaper when
long contact times are required.15c
The full-scale installations for municipal water treatment which have been identified
all employ adsorbers in parallel. Flow velocities range from 4 to 24 m/h and contact times
from 3 to 15 min. There are a number of examples of combined filtration/adsorption in
rapid gravity filters, for instance at Göteburg27 and several USA plants.15e Some of the
latter have carbon on top of a thin layer (150–300 mm) of sand. Plants at Zürich and
Düsseldorf16d use ozonation followed by pressure filters/adsorbers. Pressure vessels are
used to contain carbon beds preceded by coagulation, sedimentation and sand filtration at
Nottingham,28 Turin27 and Buckingham.15f A similar treatment sequence is employed at
Mulheim and Rouen, with the addition of pre-ozonation.16d
Provision is usually made for the removal of exhausted carbon from the beds by
eduction into a stream of water; techniques are described by Janecek.29 Vessels for
receiving exhausted and fresh carbon may also be provided. The requirements of a
complete adsorption/regeneration system are exemplified by the Church Wilne plant at
Nottingham (Fig. 5).
FIG. 5. Granular activated carbon

adsorption/regeneration system,
Church Wilne water treatment plant
(by courtesy of Severn–Trent Water
Authority and Humphreys and
Glasgow Ltd).
One carbon supplier offers prefabricated adsorption systems for hire, including
periodic replacement of the carbon.30 This is intended for industrial wastewater treatment
but could also find an application in the provision of temporary water supplies from low
quality sources.
Performance of Granular Activated Carbon Adsorbers

It is difficult to compare and summarise the reported results of batch, column and full-
scale operation, as carbon utilisation rates are strongly influenced by feedwater quality
and the desired treated water quality. Utilisation rates between 2 and 200 g carbon per m3
water treated have been reported for applications varying from taste and odour removal to
general organics reduction. Table 3 indicates where detailed operational and experimental
results can be found for various organic parameters. General conclusions are:
(a) Earthy/musty and pesticide tastes and odours are readily removed and have a low
carbon utilisation rate even in the presence of a considerable amount of humic matter.
(b) Total organic material is readily reduced but has a comparatively high carbon
utilisation rate, and there is usually a significant fraction which is not adsorbed by the
carbon however long the contact time.
(c) Most soluble polyaromatic hydrocarbons, non-ionic pesticides, organochlorine and
organophosphorus pesticides are well adsorbed, though their carbon utilisation rates
may differ considerably.
(d) Colour, phenols and anionic detergents are well adsorbed despite their polar character.
(e) Trihalomethanes, particularly chloroform, are among the compounds less readily
adsorbed by common water treatment grades of activated carbon.
The performance of carbon adsorbers with and without prior sand filtration was
investigated in full-scale trials at Vigneux-sur-Seine.15g The performance of the adsorber
preceded by sand filtration was markedly better than that of the dual-purpose beds in
terms of the removal of both general and specific organics. For instance, the useful life
for taste removal was almost doubled.
If biodegradable organic material is adsorbed in a granular carbon bed, the bed
becomes an ideal growth medium for micro-organisms. For example, when pre-
chlorination was applied to River Trent water, the low
TABLE 3 KEY TO REPORTED DATA ON
ORGANICS REMOVAL BY GRANULAR
CARBON
Reference System Organics
investigated
15 Foxcote TOC, PV, Chemical
waterworks. Oxygen Demand
Carbon beds (COD), Taste
following
coagulation,
sedimentation
and sand
filtration
28 Church Wilne TOC, threshold odour
waterworks. number, detergents,
Carbon beds haloforms
after
coagulation,
chlorination,
sedimentation
and sand
filtration
16d Lengg COD u.v. absorption
waterworks,
Zürich.
Carbon beds
after
ozonation
16d Dohne COD u.v. absorption
waterworks,
Mülheim.
Carbon beds
after
coagulation,
ozonation,
sedimentation,
sand filtration
15 Vigneux-sur- Threshold taste value,
Seine full- PV, phenols, anionic
scale detergents, pesticides
experiment
with and
without sand
filtration
5 Pilot columns, TOC, u.v. absorption,
R.Trent colour, phenols,
anionic and non-ionic
detergents, pesticides
31 Pilot columns, TOC, u.v. absorption,
R.Trent colour, threshold odour
number, monohydric
phenols, PAH,
pesticides
24 Pilot columns, u.v. absorption, 1, 2-
R.Thames dichlorobenzene,
chloroform,
bromodichloromethane,
α-BHC, γ-BHC,
Dieldrin
16d Pilot columns, PV, u.v. absorption,
Amsterdam colour
7 Pilot columns Endrin, Lindane,
Toxaphene, Silvex, 2,
4, 5-T ester
8 Pilot columns TOC, COD, haloforms
23 Pilot columns Haloforms, carbon
tetrachloride,
nitromethane, methyl
ethyl ketone, n-butanol,
1, 4-dioxane
32 Small Geosmin, 2-methyl
columns isoborneol in presence
of humic acids
33 Carbon Organochlorine
extraction pesticides
apparatus Organophosphorus
(minisampler) pesticides
numbers of surviving bacteria entering a carbon bed were considerably increased in the
outlet.5 Conversely if high levels of bacteria were presented to the carbon bed these were
attenuated, presumably because in this case the filtration effect of the carbon was greater
than the growth effect.
Biologically active carbon beds can support the development not only of chlorine-
resistant bacteria which may lead to odour problems in distribution, but also of algae and
invertebrates,15f,28 In some installations therefore—e.g. Nottingham and Göteburg—high
levels of chlorine are continuously applied so as to penetrate the carbon beds and prevent
microbiological activity. The chlorine demand of the beds at Nottingham is about 0·7
mg/litre, or 0·3 g per m3 carbon per hour.
However, in a number of European water treatment plants (e.g. Mülheim, Zürich,
Düsseldorf, Rouen) microbiological activity within carbon beds is deliberately promoted
by ozonating the water upstream of the adsorbers at doses of 1 to 5 mg/litre. This appears
to enhance organics removal by a combination of the following factors:
(i) Non-biodegradable substances may be broken down into biodegradable components
by ozone.
(ii) Residual ozone may provide additional oxygen for bacterial respiration and biological
oxidation of organics.
(iii) The biological activity thus encouraged oxidises biodegradable organic substances
and effects a continuous partial regeneration of the adsorptive potential of the carbon,
thus extending the operating life of a bed.
(iv) Ozone does not form trihalomethanes as does chlorine, but may indeed reduce the
concentration of trihalomethanes if these are present in the raw water. Therefore if
trihalomethanes are an important criterion of treated water quality, and it is necessary
to disinfect the water at an early stage of treatment, ozone must be preferred to
chlorine for disinfection. Thus, changing from pre-chlorination to pre-ozonation
extended the useful life of activated carbon beds at Mülheim from 1 to at least 20
months.16d
Regeneration System Design and Operation

Spent granular carbon can be returned to the manufacturer for regeneration but in some
circumstances—particularly where large quantities of carbon are involved—it may be
economical to provide on-site regeneration facilities. A comprehensive review of
regeneration methods and systems34 suggests that only high temperature thermal
reactivation is practicable for water treatment applications. Several stages have been
identified in this process:
1. Drying—up to 100°C.
2. Loss of volatile organics—100 to 300°C.
3. Pyrolysis of non-volatile organics—200 to 650°C.
4. Gasification of pyrolysis products—800 to 1000°C.
The two middle steps occur rapidly but adequate time must be allowed for drying and
gasification. Juhola15h showed that slow drying improved the final activity of the carbon,
but increased capital costs may make this uneconomical. Heating is normally achieved by
burning gas or oil in a stream of air which then passes over or through the carbon. A
furnace heated by infrared radiation is also being developed. The degree of reactivation is
often enhanced by injecting steam at the gasification stage; it is also improved by
increasing the gasification temperature. Very exacting control is required to minimise the
loss by overheating of the carbon itself and to preserve the refractory lining of the
furnace.
Multiple hearth and fluidised bed furnaces have been installed at a small number of
water treatment plants. Some performance data are given in Table 4. Rotary kilns are also
attractive in terms of first cost and reliability, but higher carbon losses and fuel
requirements have been reported.
The multiple hearth type has an efficient counter-current flow of hot gas and carbon
such that the incoming carbon is dried by the gas leaving the reactivation hearths. This
type can also have considerable flexibility of operating temperature profile. The provision
of an afterburner to destroy possibly noxious components in the waste gas adds
appreciably to the cost per kg carbon regenerated unless the resulting heat in the stack gas
can be reclaimed for use. Carbon after six regenerations in the multiple hearth furnace at
Church Wilne28 was equal to fresh carbon in bulk density, methylene blue adsorption and
taste/odour removal performance. Iodine number (representing small diameter pores) was
reduced by 20%, but could probably be restored to the fresh carbon value at the expense
of slightly higher carbon losses by raising the oxygen level in the furnace gas. A detailed
account of design and operational factors and an economic study of multiple hearth
furnace regeneration will be found in references 16e and 15c respectively.
The principal advantage associated with the use of a fluidised bed furnace is the lack
of mechanical rotating equipment with consequently lower maintenance and downtime
and less attrition of the carbon. The system
TABLE 4 PERFORMANCE OF
REGENERATION FURNACES
Plant Type Re- % losses Approx. Reference
activation per re- heat input,
temperature generation 106 kJ/t
carbon
regenerated
Church Multiple 950–975 4·5–5 20 (10·5 at 28 Private
Wilne hearth 100% load communication
waterworks, with steam and with air
Nottingham, injection cooling)
UK and
afterburner
Alelyckan Multiple 900–1000 6 8·5 Private
waterworks, hearth (excluding communication
Göteburg, with steam steam)
Sweden injection
and
afterburner
South Tahoe Multiple 900–925 Av. 5% 6–12 26
advanced hearth
wastewater with steam
treatment injection
plant, USA and
occasional
afterburner
Zürich Fluidised 900 3·5–5 – 35
waterworks, bed with
Switzerland steam
injection,
afterburner
and
vacuum
drier
Industrial Two-stage 815–980 6–8 5 36
wastewater fluidised
treatment bed with
plant, steam
Mississippi, injection
USA and waste
gas
recycling
Hypothetical Any type – 7 8·6 14
(EPA cost with
estimates) afterburner
and
scrubber
installed at Zürich waterworks35 includes a vacuum drier prior to the reactivation furnace;
heat extracted from the waste gases is used to raise steam for injection and to assist
drying. Another system using two fluidised beds in a vertical series, plus partial
recirculation of waste gases to the gasification stage, is reported to have comparatively
low fuel requirements.36
Granular Activated Carbon Applications

Activated carbon is an effective flexible and economical method for the removal of a
wide range of dissolved organic substances, from naturally occurring tastes and odours to
industrial pollutants. Its flexibility is illustrated by the many ways in which it can be
integrated into a sequence of water treatment processes, even if required fulfilling the
dual role of adsorbent and filter medium. It is likely that granular carbon beds will be
applied to heavier and more stringent organics removal duties in the production of
potable water, and that series beds or pulsed beds will be used for greater economy of
carbon use. On-site carbon regeneration will also be attractive for such duties. Thermal
regeneration has the important advantage of completely destroying the organics removed
from the water. Developments in furnace design may considerably reduce operating
costs.
2.5. ION EXCHANGE AND POLYMERIC ADSORBENTS
2.5.1. Introduction
The use of ion exchange resins for organics removal derives from observations of the
fouling of synthetic anion exchange resins in boiler feedwater treatment systems. It was
found that organic substances—in particular, humic acids—were strongly adsorbed, and
insufficiently desorbed during regeneration. With the development of high pressure
boilers the problems caused by fouling became more acute. At the same time, practieally
complete removal of the organic acids was necessary as a significant residual
concentration would raise the conductivity of the treated water to an unacceptable level.
Two new types of anion resins were developed to overcome the fouling problem:
macroporous, which have a large, open pore structure, and isoporous, which have a more
regularly cross-linked structure than conventional gel type resins. Both macroporous and
isoporous resins allow organic solutes to be adsorbed, and subsequently removed during
regeneration.
Experience has shown that for demineralisation of boiler feedwater, the isoporous
strong base anion exchange resins are more economical than macroporous in terms of
both capital and operating costs. However, in some circumstances macroporous resins
may be used to reduce the organic load on following gel type strong base anion exchange
resins. Regenerated by caustic brine, strong base macroporous resins are effective in
removing the organic acids responsible for fouling, but make no contribution to the
deionisation as anions in the feedwater are exchanged for chloride. Macroporous resins
used in this way are commonly known as organic scavengers.
Weakly basic macroporous resins have been developed for their ability to regenerate
effectively with sodium hydroxide; they can economically combine the removal of
organic acids and of anions of strong acids, and may be used to precede strong base
resins in either separate or stratified beds.
A detailed account of the causes, effects and control of organic fouling in
demineralisation systems is provided by Tilsley;37 reference may also be made to
McWilliam.38 A pilot-scale evaluation of isoporous and macroporous resins for the
deionisation of R.Trent water is described by Brown and Ray.39
2.5.2. Resins for Organics Removal

Synthetic polymeric resins of all the above-mentioned types have been evaluated in
recent years specifically as a means of removing organic compounds from water. To
some extent their properties—in particular, polarity and pore size—can be varied to
improve performance with respect to specific groups of compounds. For instance,
hydrophobic (non-polar) resins are available which are designed to adsorb non-polar
solutes.
Two additional types of resin have been developed specifically for the removal of
organic substances from liquid and gas streams; carbonaceous and cellulosic.
Carbonaceous adsorbents are said to have a structure between those of polymeric resins
and activated carbons. One such adsorbent is described as being produced by partial
pyrolysis of beads of a macroporous synthetic polymer with specially designed surface
functional groups.23 Carbonaceous adsorbents are available in a variety of polarities and

pore size distributions. A non-polar variety is suggested for the removal of halogenated
organic compounds, e.g. haloforms and pesticides. More polar molecules such as phenols
require a hydrophilic (polar) variety.
Cellulosic ion exchange resins are based on a cellulose matrix rather than the more
expensive polystyrene or acrylic ester on which conventional ion exchange resins are
based. They are reported to have a high capacity for specific organic and inorganic
components. They have extensive macropores and are particularly effective for the
removal of very large molecules, for instance proteins and enzymes, from aqueous
solutions. The development and preparation of cellulose based resins are described by
Grant.40 Various ionic forms corresponding to the main types of conventional resins are
available. For the removal of humic acids from water, strong or weak base varieties are
indicated.
2.5.3. Theory
It will be noted from the foregoing that removal of organic solutes takes place by a
combination of ion exchange and adsorption, the mechanism depending on the polarity of
the solute and of the functional groups in the resin. When the removal of a solute is by
ion exchange predominantly, resin performance will be sensitive to the pH of the solution
as this affects the degree of ionisation of the solute.
The capacity of a resin for a given solute depends on the surface area presented to the
solute, which in turn depends on the size of the solute molecule, the porosity of the resin
and the distribution of pore sizes.
These concepts are sufficient to indicate the resins most likely to provide effective
removal of given organic substances. More detailed predictive models and associated
data are not available for the design of systems. Resin selection and plant design
therefore require experimental data from laboratory evaluations.
2.5.4. Methods of Evaluation

Rapid screening of a range of resins may be carried out by contacting known quantities of
resin and solution in batch tests and measuring the equilibrium concentration of the
solutes of interest in the solution. Gauntlett41 noted that fresh resins imparted organic
matter to the test solution and described a procedure for extracting this prior to
equilibrium tests.
Alternative regenerants could be similarly screened by contacting with exhausted
resins.
It is essential to evaluate promising resins by column tests over several
exhaustion/regeneration cycles to obtain information on the capacity of fresh and
regenerated resins for the solutes of interest. A number of resins may be tested in parallel.
The regenerants of choice for strong and weak base anion exchange resins will normally
be caustic brine or sodium hydroxide solution. The regeneration of non-polar resins may
require the use of organic solvents, e.g. acetone or methanol. The importance of thorough
rinsing in potable water applications can readily be appreciated, and rinse volumes should
be determined in the column tests.
25 mm diameter glass or perspex columns are suitable for laboratory tests. Superficial
velocities of 5 to 25 m/h may be used, giving superficial residence times of 5 to 1 min in
beds 400 mm deep. Further details of laboratory test procedures are given by Brown and
Ray39 and Chriswell et al.42
As with granular activated carbon studies as described in the preceding section, it is
important when evaluating resin adsorbents to obtain breakthrough data for design
purposes. Large column tests may be required when investigating the removal of certain
compounds for which potable water standards have been set at the µg/litre level (see
Table 1), in order to obtain sufficient treated water for analytical purposes.
2.5.5. Test Results

Table 5 is a key to recently published test results for a variety of resins and solutes. The
salient conclusions are as follows:
All types of macroporous and isoporous weak and strong base resins are effective for
the removal of humic acids, phenols and anionic detergents.
However, anionic detergents are so strongly adsorbed by basic resins that they are
hardly desorbed at all from strong base resins by conventional regeneration. The
desorption from weak base resins is very low but according to Oehme and Martinola47
levels of up to 0·3 mg anionic detergents per litre can be tolerated if sodium hydroxide
regeneration is applied.
The uptake of phenol by a weak base phenol-formaldehyde resin was found to be
highly pH dependent, with the best performance at pH 5.
Ammonia was found to be as effective as sodium hydroxide for regenerating one weak
base resin, but somewhat inferior for another.
Non-polar polyacrylic and polystyrene adsorbents (‘white carbons’) were found to be
inferior to activated carbon in respect of a variety of solutes.
The performance of a non-polar carbonaceous resin was comparable to that of
activated carbon.
2.5.6. Applications
Macroporous and isoporous weak base and strong base resins are well established for the
removal of weak organic acids, in combination with, or preceding, demineralisation of
boiler feedwater. When treating river water or highly coloured moorland water, the resins
would normally be preceded by coagulation and clarification. However, Grant has
suggested40 that the low costs and high organics capacity of cellulosic ion exchange
resins make it possible to consider these as an alternative to coagulation, clarification and
scavenger resin for the treatment of highly coloured, low turbidity waters.
For potable water treatment it appears that resins are generally not competitive with
activated carbon. The advantage of in-situ regeneration is outweighed in most
circumstances by the difficulty and cost of disposing of the spent regenerant solution. The
use of ammonia as a recoverable
TABLE 5 SUMMARY OF REPORTED DATA

ON ORGANICS REMOVAL BY ION
EXCHANGE AND ADSORBENT RESINS
Refe Type Trade Test Solutes Rege Column test
rence name /Parameters nerants conditions
Velocity Contact
(m/h) time
(min)
42 M np XAD-2 Batch/column 100 various – – –
pD (A) hybrid
31 M sb IRA Large column TOC Caustic 18 2·7
pD 904 (A) brine
50°C
40 M – Column Total Caustic – –
wb organics brine
ce
43 M an WRL Column Colour Caustic 10 1·5
ce 200A brine
44 M Duolite Equilibrium p-nitrophenol – 2·7 4
wb S37(D) and column
pf
23 M np XAD-2 Column n-butanol – – 2·1
pD (A)
M XAD-7 Column methyl ethyl – – 2·1
mp (A) ketone
ac
M np XE-340 Column nitromethane – – 2·1
ca (A) 1,4-dioxane
41 M sb MP 500 Equilibrium – – –
u.v.absorption
pD (L) and column
M MP 62 Equilibrium – – –
wb (L) and column
pD
M sb IRA Equilibrium – – –
pD 904 (A) and column
M IRA 93 Equilibrium – – –
wb (A) and column
pD
I sb XE 258 Equilibrium – – –
ac (A) and column
I wb XE 236 Equilibrium – – –
ac (A) and column
M Duolite Equilibrium – – –
wb S37(D) and column
pf
I sb FF/IP Equilibrium – – –
po (P) and column
I wb H/IP Equilibrium – – –
po (P) and column
I wb M/IP Equilibrium – – –
po (P) and column
M np XAD-2 Equilibrium – – –
pD (A) only
M np XAD-7 Equilibrium – – –
ac (A) only
M sb MP 500 Extended Caustic 2.4 12
pD (L) column test brine
40°C
M MP62 Extended Sodium 2·4 12
wb (L) column test hydroxide
pD 40°C
I sb XE 258 Extended Caustic 2·4 12
u.v.absorption
ac (A) column test brine
TOC
40°C
DBS
I sb TR (P) Extended 2,4-D Caustic – –
po column test γ-D brine
40°C
M MP 62 Extended Ammonia 2·4 12
wb (L) column 20°C
pd test
M Duolite Extended NaOH 2·4 12
wb S37(D) column 20°C
pf test
45 M sb MP500 Column PV
pD (L) test
Organic Caustic 67·5 1·1
carbon brine (re-
used)
46 M Grantex Column PV
wb (E) test
Organic Caustic 6 10
carbon brine
Key to abbreviations
M macroporous I isoporous np non-polar mp medium polar an anionic wb weak
base sb strong base
po polystyrene pD polystyrene-DVB acrylic ce cellulosic (viscose) pf phenol-
formaldehyde ca carbonaceous
A Amberlite (Rohm and Haas) D Diamond Shamrock L Lewatit (Bayer) P Permutit
E Ecotech
regenerant is of interest, but a concentrated waste liquor would remain to be disposed of.
However, if a very high standard of organics removal is required the series combination
of activated carbon and a macroporous anion exchange resin would be a powerful one.
Such a system has been used for the recovery of a wide range of organics from water
samples prior to analysis.
2.6. REVERSE OSMOSIS AND ULTRAFILTRATION
2.6.1. Introduction
The process of reverse osmosis, as developed for desalination of water, is also effective
in removing organic compounds from water. Chapter 4 describes the application of
reverse osmosis (RO) to desalination and covers the theory, design and operation of RO
plant in depth. It is necessary here to make only brief reference to the fundamental
concept of RO and to define certain terms.
Water is forced through a membrane which retains (rejects) dissolved inorganic and
organic compounds as well as bacteria, viruses and larger particulate matter. The rejected
species are therefore concentrated into a waste stream as illustrated in Fig. 6.
FIG. 6. Reverse osmosis schematic.

The feedwater has to be applied to the membrane at a high pressure to overcome the
natural osmotic pressure (which increases with the salinity of the feed) and provide an
adequate flux (rate of permeate production per unit area of membrane).
2.6.2. Removal Mechanisms

Organic species may be rejected either because they have a low solubility in the
membrane material or because, whatever their solubility, their size hinders their passage
through the membrane. In either case the organic molecules permeate the membrane at a
much lower rate than do the water molecules, and they are therefore concentrated in the
waste stream.
Negative rejection is also possible; that is, certain organic substances permeate the
membrane more readily than water and their concentration is therefore higher in the
permeate than in the feed.
It has also been observed in laboratory tests that certain organic compounds, notably
aromatics including polynuclear aromatic hydrocarbons (PAH), are adsorbed onto the
membrane and are thereby attenuated in the permeate. This has implications for the use
of laboratory batch tests for the prediction of RO performance. In continuous processing,
however, adsorbed solutes should rapidly saturate the membrane and thereafter be
rejected according to their solubility and size. Adsorption of large quantities of some
organics might be expected to improve the rejection of other species, but also to reduce
the permeate flux rate.
Klein et al.48 describe the use of solubility parameters to predict the rejection of
organic compounds by cellulose triacetate and ethyl cellulose membranes. The theory
predicts that in general aldehydes, ketones, ethers and esters will be rejected less
effectively than will ionised compounds or highly hydrogen-bonding compounds. To
remove the former groups the theory indicates that an aliphatic nylon membrane would
be effective. This type of membrane can also be operated at extremes of pH which would
rapidly hydrolyse cellulose acetate. A series of RO modules with different membranes,
perhaps operated at diiferent pH values, could therefore be used to obtain a very
comprehensive concentrate of organic materials from water for analytical investigation.
The implications for full-scale treatment are that, if particular compounds or groups are to
be removed, the selectivity of various RO systems could be considered.
2.6.3. Experimental Results

The rejection of many organic compounds has been studied mainly in laboratory units
using the common cellulose acetate membrane material. Results for the removal of the
pesticides Lindane and 2,4-D are summarised by Love;7 concentrations and/or pressures
used in the reported studies do not however appear to accord with expected full-scale
conditions.
About 80% of organic matter (as TOC) was rejected from Cincinatti tap water by
cellulose acetate membranes.49 The concentrate included hydrocarbons, phthalates,
chlorinated compounds, polyethers and long chain fatty acids.
Cellulose acetate membranes were also used in continuous operation on River Trent
water.31 This gave valuable experience on the practical operation of both tubular and
‘spaghetti’ type RO modules. On the basis of product flux recovery and product quality,
it was concluded that daily flushing with Calgon solution was the most effective of the
methods tried. This regime permitted a flux of about 0·013 m3/m2 atm to be maintained at
28 atm feed pressure with either tube or spaghetti modules; the product to feed flow ratio
was 75%. It should be noted that treatment of the river water prior to RO was limited to
rapid sand filtration and pH adjustment to about pH 6 using sulphuric acid.
The rejection of total dissolved solids under these conditions was 95% or higher. The
removal of organic materials is illustrated by Table 6. Product quality is satisfactory
compared with the potable water quality requirements of Table 1 except in respect of
threshold odour number. It appears that odorous compounds present in the Trent water
were less readily rejected than the bulk of the organic matter present, as would be
predicted from their aromatic nature. TOC was reduced by 97% on average by the tubular
modules.
TABLE 6 PERFORMANCE OF TUBULAR RO
ON RlVER TRENT WATER. After Melboume,31
by courtesy of Water Research Centre
Parameter Concentration Concentration
in feedwater in product
Colour 0·048–0·093 0–0·01
(optical
density at 400
nm)
Optical 0·24–0·87 0–0·3
density at 300
nm
Total organic 4·27–21·4 0·1–0·64
carbon
(mg/litre)
Anionic 0·13–0·54 0–0·1
detergents
(mg/litre)
Monohydric 0–0·004 0–0·002
phenols
(mg/litre)
Threshold 20–100 5–24
odour number
Pesticides α- 0·008–0·172 0–0·12
BHC (µg/litre)
γ-BHC 0·033–0·21 0–0·14
(µg/litre)
2.6.4. Membrane Configurations

Besides the tubular and ‘spaghetti’ systems already mentioned, three other configurations
are available: flat plate, spiral wound and the very high specific surface area hollow fibre
system using aromatic polyamide material. Hollow fibre units are susceptible to organic
fouling and are not guaranteed for use on feedwaters with more than 5 Hazen units of
colour.50
2.6.5. Applications
The high cost of RO derives from the sophisticated technology, the high pressure
required, and the difficulty of disposing of the reject stream. It is unlikely that RO would
be the process of choice specifically for organics removal in the production of potable
water. However, it could be competitive if the removal of inorganic salts was also
required. There is a possible application in the preparation of samples for organic
analysis, as mentioned above. Other applications which have been suggested include:
The removal of pyrogens in the preparation of pharmaceuticals and

injections.
As an intermediate stage in boiler feedwater treatment, to remove
organics and reduce the inorganic ion load prior to a final ion exchange
stage.
The removal of algal products (polysaccharides) which can cause
problems in soft drink manufacture.51
The concentration of regenerant liquors from organic scavenger resins.
2.6.6. Ultrafiltration
Ultrafiltration (UF) is closely related to reverse osmosis. Compared with RO, the UF
membrane is more ‘open’ so that only large molecules are rejected. Inorganic salts are
not rejected so there is little osmotic pressure to be overcome and operating pressures are
normally below 7 atm, whereas RO of waters containing up to 10 000 mg/litre salinity
requires pressures of 20 to 40 atm. Product flux rate is considerably higher in UF. The
mechanism may be thought of as sieving.
UF has been used to concentrate large organic species, e.g. proteins, viruses and
polypeptides. Dissolved solids of molecular weight greater than 1000 were found to be
removed in the study of Hardt et al.52 In water treatment, humic acids would be removed
and one possible application would be the reduction of humic colour prior to RO of the
hollow fibre type.
2.7. CONCLUSIONS
The processes of coagulation, disinfection and chemical oxidation remove organic matter
to some extent, but insufficiently to meet the latest stringent requirements for industrial
and municipal water quality. Adsorbent resins, though widely used in boiler feedwater
treatment, have not yet found practical application in potable water supply. Reverse
osmosis is extremely effective for the removal of a wide range of organic compounds but
is prohibitively expensive except for small specialised duties or where its deionisation
capability is also required. Ultrafiltration is considerably cheaper but effective only for
the removal of large organic molecules. Resins and membrane processes suffer from the
disadvantage of a concentrated liquid waste which must be disposed of, often with
considerable difficulty at high cost.
Activated carbon is generally the most cost-effective treatment for the removal of
specific organic compounds and groups of compounds including taste- and odour-
producing substances. There are many installations of powdered carbon for the removal
of tastes from potable water. For heavier or more stringent organics removal duties,
granular activated carbon beds offer considerable economies and the opportunity to
employ on-site regeneration of exhausted carbon in high temperature furnaces. Thermal
regeneration has the advantage of completely destroying the organic matter removed
from the water.
REFERENCES
1. FIELDING, M. and PACKHAM,R. F., J. Instn Water Engrs & Scientists, 1977, 31, 353.
2. HOUGHTON, G.U., in Proc. Symposium Water Treatment in the Seventies, Soc. Water Treat.
Exam./Water Research Assoc., Jan. 1970, 79–99.
3. HUISMAN, L. and WOOD, W.E., Slow Sand Filtration, 1974, World Health Organisation,
Geneva.
4. ROOK, J.J., Water Treat. Exam., 1972, 21, 259.
5. MILLER, D.G. and SHORT, C.S., The Trent Research Programme, Vol. 5, Costs of River Water
Treatment, 1972, Water Resources Board, Reading.
6. LEWIS, W.M., Water Treat. Exam., 1975, 24, 243.
7. LOVE, O.T., in Manual of Treatment Techniques for meeting the Interim Primary Drinking
Water Regulations, 1977, US Environmental Protection Agency, Office of Research and
Development, Municipal Environmental Health Laboratory, Water Supply Research Division,
Cincinnati, Ohio.
8. ROOK, J.J., J. Amer, Water Works Assoc., 1976, 68, 168.
9. GOMELLA, C., J. Amer. Water Works Assoc., 1972, 64, 39.
10. GAUNTLETT, R.B., TP79, ‘A Comparative Evaluation of Odour Treatment Methods’, 1970,
Water Research Assoc., Medmenham, Marlow, Bucks.
11. NAKAYAMA, S. and ESAKI, K., Water Purification & Liquid Wastes Treat. (Japan), 1978,
19, 241.
12. PRENGLE, H.W. and MAUK, C.E., in Proc. Conf. Ozone/Chlorine Dioxide Oxidation
Products of Organic Materials, R.G.Rice and J.A.Cotruvo (Eds.), 1978, Int. Ozone Inst., Inc.,
Cleveland, Ohio.
13. DOWLING, L.T., Water Treat. Exam., 1974, 23, 190.
14. US ENVIRONMENTAL PROTECTION AGENCY, ‘Interim Primary Drinking Water
Regulations. Control of Organic Chemical Contaminants in Drinking Water’, Federal Register,
Feb. 9, 1978, 43, No. 28 Part II.
15. WATER RESEARCH ASSOCIATION, Proc. Conf. Activated Carbon in Water Treatment,
1974, Water Research Assoc., Medmenham, Marlow, Bucks.
15a. WEBER, W.J.JR, in above, 53–71.
15b. ALDRIDGE, J.L., in above, 145–158.
15c. BURLEY, M.J. and SHORT, C.S., in above, 203–239.
15d. GARDINER, E.R., in above, 241–261.
15e. LOVE, O.T.JR., ROBECK, G.G., SYMONS, J.M. and BUELOW, R.W., in above, 279–312.
15f. FORD, D.B., in above, 263–278.
15g. RICHARD, Y., in above, 313–345.
15h. JUHOLA, A.J., in above, 177–202.
16. CHEREMISINOFF, P.N. and ELLERBUSCH, F. (Eds.), Carbon Adsorption Handbook, 1978,
Ann Arbor Science Publishers Inc., Ann Arbor, Mich.
16a. COOKSON, J.T.JR, Ch. 7 in above.
16b. KNOPP, P.V., GITCHEL, W.B., MEIDL, J.A. and BERNDT, C.L., Ch. 15 in above.
16c. SCHULIGER, W.G., Ch.2 in above.
16d. RICE, R.G., MILLER, G.W., ROBSON, C.M. and KÜHN, W., Ch. 14 in above.
16e. LOMBANA, L.A. and HALABY, D.; VON DREUSCHE, C.JR; Chapters 25 and 26 in above.
17. GAUNTLETT, R.B., TP93, ‘Powdered Carbon in Water Treatment’, 1972, Water Research
Assoc., Medmenham, Marlow, Bucks.
18. TOMONO, K., J. Amer. Water Works Assoc., 1977, 69, 166.
19. GOMELLA, C., Techniques et Science Municipales—l’Eau, 1973 (April), p. 151.
20. LETTINGA,G., BEVERLOO, W.A. and VAN LIER, W.C., Prog. Water Tech., 1978, 10, 537.
21. SUMNER, A.M., paper presented to the Institution of Water Engineers and Scientists, Northern
Section. J. Instn. Water Engrs. & Scientists, 1979, 33, 455.
22. ADAMS, R.W., ROBINSON, R.D. and KENNETT, C.A., J. Instn Water Engrs, 1973, 27, 15.
23. MCGUIRE, M.J. and SUFFET, I.H., J. Amer. Water Works Assoc., 1978, 70, 621.
24. HYDE, R.A., paper presented to American Chemical Soc. Symposium, Miami Beach, Sept.
1978, in Activated Carbon Adsorption of Organics from the Aqueous Phase, Vol. 2, M.J.
McGuire and I.H. Suffet (Eds.), 1980, Ann Arbor Science Publishers Inc., Ann Arbor, Mich. (In
press).
25. BOSSY, G. and SANCHEZ, Y., La Technique de l’Eau, July/Aug. 1972, p. 15.
26. CULP, R.L. and CULP, G.L., Advanced Wastewater Treatment, 1971, Van Nostrand Rheinhold
Co., New York.
27. WATER RESEARCH CENTRE, Water Purification in the EEC, published for the Commission
of the European Communities, 1977, Pergamon Press, Oxford.
28. OSBORNE, D.J. and KENNETT, C.A., papers presented to US Environmental Protection
Agency Conf. Adsorption Techniques—Practical Application of Adsorption Techniques in
Drinking Water, Reston, Va., May 1979.
29. JANECEK, K.F., J. Amer. Water Works Assoc., 1978, 70, 581.
30. ANON., Effl. Water Treat. J., 1975, 15, 516.
31. MELBOURNE, J.D., TR 74, ‘River Trent Treatment’, 1978, Water Research Centre,
Medmenham, Marlow, Bucks.
32. HERZING, D.R., SNOEYINK, V.L. and WOOD, N.F., J. Amer. Water Works Assoc., 1977, 69,
223.
33. EICHELBERGER, J.W. and LICHTENBERG, J.J., J. Amer. Water Works Assoc., 1971, 63, 25.
34. LOVEN, A.W., Chem. Engng Prog., 1973, 69, (11), p. 56.
35. SCHALEKAMP, M. and BAKKER, S.P., Effl. Water Treat. J., 1978, 18, 28.
36. ANON., Chem. Engng, June 18, 1979, p. 93.
37. TILSLEY, G.M., Chemy Ind., March 3, 1979, p. 142.
38. MCWILLIAM, J.D., Chem. Engng, May 22, 1978, p. 80.
39. BROWN, J. and RAY, N.J., Effl. Water Treat. J., 1974, 14, 417.
40. GRANT, R.A., in New Processes of Waste Water Treatment and Recovery, G. Mattock (Ed.),
1978, Ellis Horwood Ltd, Chichester.
41. GAUNTLETT, R.B., TR 10,‘A Comparison between lon-Exchange Resins and Activated
Carbon for the Removal of Organics from Water’, 1975, Water Research Centre, Medmenham,
Marlow, Bucks.
42. CHRISWELL, C.D., ERICSON, R.L., JUNK, G.A., LEE, K.W., FRITZ, J.S. and SVEC, H.J.,
J. Amer. Water Works Assoc., 1977, 69, 669.
43. JØRGENSEN, S.E., Water Supply & Management, 1978, 2, 475.
44. KIM, B.R., SNOEYINK, V.L. and SAUNDERS, F.M., J. Water Pollut. Control Fed., 1976, 48,
120.
45. KOLLE, W., Water & Sewage Works, 1979, 126, 68.
46. ROWE, M.C., Effl. Water. Treat. J., 1975, 15, 519.
47. OEHME, C. and MARTINOLA, F., Chemy Ind., Sep. 1, 1973, p. 823.
48. KLEIN, E., EICHELBERGER, J., EYER, C. and SMITH, J., Water Res., 1975, 9, 807.
49. DIENZER, M., MELTON, R. and MITCHELL, D., Water Res., 1975, 9, 799.
50. ROBERTSON, W.J., Water Services, 1975, 79, 16.
51. TREANOR, A.I., Chemy Ind., June 4, 1977, p. 431.
52. HARDT, F.W., CLESCERI, L.S., NEMEROW, N.L. and WASHINGTON, D.R., J. Water
Pollut. Control Fed., 1970, 42, 2135.
Chapter 3
REMOVAL OF NITROGEN
COMPOUNDS
R.B.GAUNTLETT, B.A., M.I.W.E.S.
Water Research Centre, Medmenham, Bucks., UK
SUMMARY
Improvements in sewage and water treatment processes are reducing the need for
removal of high concentrations of ammonia for potable waters. Traditionally, ammonia
removal has been carried out by breakpoint chlorination. There is now, however, an
increasing acceptance of biological methods for ammonia removal because of
trihalomethane formation resulting from the application of high chlorine doses to surface
waters.
Nitrate levels in both surface and underground water supplies are tending to rise due
to the use of agricultural fertilisers. At present ion exchange is sometimes used for nitrate
removal from underground waters before they enter supply. Biological methods are
however being developed and these are expected to find their principal application in the
treatment of surface waters. The best method to use will depend on the particular
circumstances at the treatment site; and the relative advantages and disadvantages of ion
exchange and biological treatment are reviewed in the chapter.
3.1. INTRODUCTION
The two forms of nitrogen that may require removal during potable water treatment are
ammonia and nitrate. Ammonia can interfere with the disinfection of water by
chlorination, while high nitrate levels can cause methaemoglobinaemia in infants and
there has been some suggestion that they may also give rise to the formation of
carcinogenic nitrosamines in the stomach. Nitrite may also occasionally be found in
water, but usually at levels too low to be of consequence.
Ammonia occurs in surface waters as the result of pollution. The excretions of wild or
farm animals may be one source; while in populated areas, incompletely treated sewage
or possibly industrial effluent may be another. Ammonia is sometimes found in
groundwaters as a result of biological denitrification or from the breakdown of
proteinaceous organic material by saprophytic bacteria and fungi. No direct hazard to
health has been demonstrated due to ammonia in drinking water. Work in the Netherlands
has shown, however, that levels of ammonia in excess of 0·3 mg/litre can lead to
aftergrowths in the distribution system with associated taste problems. The World Health
Organisation (WHO) European Standards recommend a level of not more than 0.05
mg/litre in supply. Ammonia in its undissociated form is toxic to fish at quite low levels
(<0.5 mg/litre) and its removal can be important for fish farming, particularly where a
high proportion of the water is recycled.
The occurrence of nitrate in both surface and groundwaters is associated with arable
farming practice and the application of nitrogenous fertilisers. Most nitrate is released
after the soil has been ploughed, allowing organically bound nitrogen to become oxidised
to nitrate which is then washed into the watercourse or aquifer by rain. In highly
populated areas, nitrified sewage effluent has in a few cases been found to contribute a
significant amount of nitrate to the river it discharges into, particularly at times of low
flow. The proposed EEC maximum admissible concentration for drinking water is 11·3
mg/litre of nitrate-nitrogen, which corresponds to the maximum ‘recommended’ WHO
level. Levels above 22·6 mg/litre nitrate-nitrogen are ‘not recommended’ by WHO; and
give rise to concern in relation to methaemoglobinaemia in infants.
The current trends in the ammonia and nitrate levels found in sources used for potable
supply are in opposite directions: while river pollution becomes subject to closer control
and sewage treatment becomes more efficient, ammonia levels in rivers have tended to
fall. Nitrate levels on the other hand have been rising in the United Kingdom as a result
of an increase in the use of fertilisers and in the proportion of land used for arable
cultivation.
Ammonia levels in relatively unpolluted rivers such as the Severn or Thames are
under 0·5 mg/litre on average. More polluted rivers such as the Trent or Rhine now have
an average of 1–1·5 mg/litre ammonia. At particular locations or times the level may be
at least twice the average. In winter, for example, the natural nitrification processes
taking place in rivers slow down and sewage works may also have difficulty in producing
a nitrified effluent, thus causing ammonia levels in rivers to rise. Ammonia is seldom a
problem in groundwaters in the United Kingdom. On the Continent, however, levels
above 2 mg/litre are not uncommon and boreholes containing as much as 8 mg/litre are
used for potable supply after treatment for ammonia removal.
High nitrate levels in some underground waters have already become a problem in the
United Kingdom and treatment is being practised at two borehole sites in East Anglia.
Nitrate-nitrogen levels above 11·3 mg/litre have been experienced at a number of
boreholes; but treatment can often be avoided where water from a nearby low nitrate
source can be used for blending prior to putting the water into supply. Nitrate levels in
several rivers such as the Thames and Great Ouse now regularly exceed 11·3 mg/litre for
periods during the winter months. Where raw water storage reservoirs are used, it may be
possible to cease abstraction from the river during periods of high river nitrate levels.
Some reduction in nitrate levels is also often found on storage, due to bacterial or algal
action in the reservoir.
The treatment methods available for the removal of ammonia and nitrate will be
described individually in more detail later, but in each case they can be divided into three
main categories: chemical, physico-chemical and biological.
Ammonia is the most reduced form of nitrogen and requires a 7·6:1 weight ratio of
chlorine: ammonia-nitrogen for oxidation to nitrogen. Nitrate on the other hand, being the
most highly oxidised form of nitrogen, has been found to require, for example, at least
seven times its own weight of ferrous iron to reduce it to nitrous oxide and ammonia.
Thus the amounts of oxidant or reductant required tend to be large; and there is always
the possibility of undesirable end-products appearing, such as nitrogen trichloride from
Removal of nitrogen compounds 55
ammonia oxidation or ammonia from nitrate reduction. Reaction times can also be long,
depending on conditions. These disadvantages are most marked in the case of nitrate
reduction.
Physico-chemical methods such as ion exchange or reverse osmosis are usually able to
treat the water in a shorter time, but produce waste streams that are often difficult to
dispose of in an acceptable manner. Reverse osmosis is unlikely to be an economic
choice in most cases, whereas ion exchange often merits consideration, particularly for
nitrate removal. Air-stripping seems an obvious possibility for ammonia removal but,
apart from its high cost, it can suffer from freezing problems at low air temperatures and
from calcium carbonate precipitation if the water is hard.
Biological treatment methods rely on processes already occurring to some extent in
nature. Indeed, in the absence of further pollution, water in a river or reservoir can be
expected to lose ammonia, and ultimately also nitrate, by biological action:
The nitrifying bacteria derive their energy from oxidising ammonia to nitrate, via nitrite,
using the oxygen dissolved in the water to do this. The denitrifying bacteria also derive
their energy from oxidation, and require a suitable source of oxidisable organic carbon.
They carry out the oxidation, first using the free oxygen available, and then, when
conditions have become anaerobic, the bound oxygen in the nitrate. Thus the two
processes, which may occur at different locations in a river or reservoir, can ultimately
convert inorganic nitrogen compounds to nitrogen gas. Algal activity can also result in a
significant reduction in nitrate levels in raw water storage reservoirs.
Where only ammonia removal is needed, the nitrification step alone will usually
suffice. If however the nitrate so produced results in an unacceptably high nitrate
concentration, denitrification may also be required.
Biological treatment methods are often given little consideration because they are
thought to be slow and unreliable. In fact they can usually be made to operate much faster
than chemical methods. The basic requirement for achieving this is the attainment of as
high an interfacial surface area as possible, between the water and the bacteria, per unit
volume of treatment vessel. This involves the use of high concentrations of bacteria in the
form of small flocs or as a thin film covering a finely divided medium such as sand. For
denitrification, a simple easily oxidised organic compound, such as methanol or ethanol,
is added to give good reaction rates. Biological methods also have the advantage of
producing few, if any, unwanted side products and no environmentally undesirable
wastes.
As regards reliability, denitrifying bacteria grow easily and adhere well to solid
surfaces, so that there is no difficulty in maintaining the bacterial mass or film. In
practice, their performance has been found to be almost unaffected by the variations in
the quality of a lowland river such as the Thames. In contrast, the growth of nitrifiers is
slow and the cell yield is small. They do not adhere as readily as denitrifiers to surfaces
and are more susceptible to poisons. Nevertheless, large nitrification plants have been in
operation for many years in France and Holland treating groundwaters whose
composition can be relied upon to remain reasonably constant. With river waters, on the
other hand, experience has shown that seasonal variations can be expected in the bacterial
population of a nitrifying unit; for instance, a decline is often observed in the autumn.
The reasons for this are not clear and there is scope for more research into ways of
improving the reliability of river water nitrification.
Biological methods compare very favourably in cost with other methods. In view of
this and their other advantages they will be discussed in some detail in the following
sections dealing separately with ammonia and nitrate removal.
3.2. AMMONIA REMOVAL
3.2.1. Breakpoint Chlorination

This is the most frequently used method for ammonia removal. It is usually applied in the
treatment of surface waters; and has the advantage of being cheap to install and simple to
operate. The basic chemistry of the process is now well known1 and will not be described
here. In essence, a sufficient chlorine dose is applied prior to clarification to react with all
the ammonia present to give predominantly nitrogen and a minimum of chloramines and
nitrogen trichloride. If no organic nitrogen compounds are present, this occurs at the Cl:N
weight ratio of 7·6:1 (the breakpoint) that varies slightly with pH. The time spent in the
clarification stages of treatment should be sufficient for completion of the chlorination
reactions prior to final chlorination of the water before it enters supply. Sometimes excess
chlorine is applied initially so that dechlorination (e.g. with sulphur dioxide) is used to
regulate the residual entering supply. It may not always be easy, however, to control the
overall chlorination procedure so that an adequate chlorine residual is maintained in
supply for disinfection. This is particularly the case when organic nitrogen compounds, as
well as ammonia, are present thus increasing the variety and duration of the chlorination
reactions. Also, nitrogen trichloride or dichloramine can be formed under many
conditions, giving rise to taste and odour problems in supply.
The disadvantages of breakpoint chlorination become most evident when ammonia
levels are consistently high. The process then becomes much more expensive to operate
in comparison with the biological method (see Fig. 1). Breakpoint chlorination has then
been advocated2 as a back-up
FIG. 1. Estimated costs of ammonia

removal processes.
process to supplement biological nitrification where the latter cannot always be relied
upon to remove all the ammonia.
High ammonia levels also increase the likelihood of unacceptable amounts of nitrogen
trichloride being formed, especially when the chlorine dose exceeds that required to reach
the breakpoint and if the pH is low. In practice, the chlorine dose will almost certainly be
in excess of that required to reach the breakpoint, if only to guard against variations in
ammonia level. Where dechlorination is then applied, the excess free chlorine will be
removed before the dichloramine and nitrogen trichloride, which may remain. With
further dechlorinating agent, the chloramines and nitrogen trichloride are converted back
into ammonia. Thus dechlorination cannot entirely remedy the effects of chlorinating
beyond the breakpoint.
It is perhaps worth noting that activated carbon can be used for dechlorination.3 It can
only be used for total dechlorination, however, so that the required chlorine residual for
supply would have to be added after dechlorination by carbon. Active carbon has the
advantage over sulphur dioxide of being able to convert nitrogen trichloride and
dichloramine into nitrogen instead of ammonia,4 while monochloramine is converted to a
mixture of nitrogen and ammonia. Thus the possibility of ammonia reappearing during
dechlorination is substantially reduced.
Variation in the ammonia level of the raw water can of itself present problems when
breakpoint chlorination is practised, even if ammonia levels are quite low. Saunier5
maintains that chlorine doses beyond the breakpoint can give rise to objectionable
amounts of nitrogen trichloride at ammonia levels above about 0·2 mg/litre. If, on the
other hand, the initial chlorine dose is insufficient to reach the breakpoint, final
chlorination may result in a continuation of the chlorination reactions, and make it
difficult to regulate the free chlorine residual. The breakpoint chlorination method
therefore requires an accurate adjustment of the chlorine dose according to the variations
of ammonia level in the raw water.
Apart from accurate chlorine dosing an essential requirement for the success of
breakpoint chlorination is an adequate contact time with the chlorine before either
dechlorination or final chlorination. An inadequate contact time renders the result of the
initial chlorination uncertain and may result in ammonia or dichloramine remaining in the
water after dechlorination. Where dechlorination is not practised, the chlorination
reactions may continue in supply, thus rendering the chlorine residual impossible to
control.
Saunier5,6 has found from theoretical predictions and practical tests carried out at 15°C
that an effective contact time of 2 min was required to eliminate 20 mg/litre of NH3–N,
and 40 min to eliminate 1 mg/litre NH3–N, at pH 7·3. A lowering of temperature, or a pH
outside the range 7 to 8, resulted in longer contact times being needed. The effect of
temperature and pH was studied by Wei and Morris7 who found that a temperature drop
of 10°C almost doubled the time required for reaction. The combined effects of low
temperature, a pH below 7, and short circuiting in clarification or holding tanks can result
in a long actual residence time being required. This could well amount to two or three
hours.
High chlorine doses can have undesirable consequences quite unconnected with the
presence of ammonia, since it may react with the organic material in surface waters to
form trihalomethanes (THM). This has been found8,9 to take place with naturally
occurring humic and fulvic acids to a degree dependent on the chlorine dose. The results
of a national survey carried out on THM formation during water treatments in the USA
showed10 that there was a fair degree of correlation between the chlorine demand of a
water and its THM content after chlorination. Rook has also shown11 how the THM
formed in a surface water after 1 h increased with chlorine dose. Once formed, THMs are
not easily removed, since they are poorly adsorbed by active carbon, for instance.
Breakpoint chlorination may also adversely affect later treatment stages: Sontheimer
found12 that it lowered the overall efficiency of organics removal by biologically active
carbon as a result of the formation of organochloro compounds. The current trend on the
Continent, where active carbon is often used and ammonia levels in surface waters may
be high, is therefore away from breakpoint chlorination as the principal means of
ammonia removal. Instead, biological methods are preferred for removing the greater part
of it.
Chlorine is the only oxidant that has found practical use for ammonia removal. Ozone
reacts too slowly with ammonia under normal circumstances, although solid catalysts
have been found13,14 to increase the reaction rate. Ozone reacts only with free ammonia
forming nitrate and the rate of reaction therefore increases as the pH is raised through the
range 8 to 10. However, carbonate ions and organic solutes interfere strongly with the
reaction above pH 9, by competition with ammonia for the available ozone. It has been
estimated15 that for a typical surface water of pH 8 containing 1 mg/litre NH3–N, the
ozone would react at least an order of magnitude faster with the other solutes present.
3.2.2. Physico-chemical Methods

Physico-chemical methods of ammonia removal have found little application in potable
water treatment, but have been given more serious consideration for wastewater
treatment, where ammonia levels can be much higher. Considering first ion exchange, a
comparative study carried out by Jorgensen on several different types of cation
exchanger16 showed that clinoptilolite, a natural zeolite, had the highest ammonia
removal capacity. It was also less expensive than conventional synthetic cation
exchangers. A direct comparison of the selectivities of Hector clinoptilolite with a strong
acid polystyrene ion exchange resin has shown17 that the latter preferred calcium ions to
ammonium ions whereas for clinoptilolite the opposite was true. It is therefore possible to
regenerate clinoptilolite with relatively inexpensive lime and calcium chloride.
Pilot plant tests carried out on a wastewater containing 10 to 20 mg/litre NH3–N by the
US Environmental Protection Agency18 have shown that columns of Hector clinoptilolite
brought the ammonia level down to 0·5 to 1·0 mg/litre NH3–N. Further treatment by
breakpoint chlorination might therefore be required to produce a potable water. As much
as 99% ammonia removal has been demonstrated17 using two laboratory columns in
series. The regenerant used in the pilot plant study was a mixture of sodium and calcium
chlorides adjusted to pH 11 with lime. Regenerant recovery was also studied, using either
air stripping to blow off the ammonia, or electrolysis which generated chlorine to react
with the ammonia and so convert it to nitrogen. An advantage of the electrochemical
method of removing ammonia from spent regenerant was that neutral regenerant
solutions could be used, thus avoiding precipitation and scaling problems associated with
the use of lime.
Ion exchange using clinoptilolite was found to be a more expensive process than direct
air stripping. Most studies of ammonia removal by air stripping have been carried out on
wastewaters. Since ammonia is very soluble in water, an air: water volume ratio of about
3000:1 is required for 90% removal at 10°C in a counter-current air stripping tower. As
the temperature falls, the amount of air required increases. For air stripping to be
successful, the greater part of the ammonia must be converted into its free form by raising
the pH of the water. Several studies have shown that ammonia removal efficiency
increases considerably as the pH is raised to 10·5, but that little further improvement can
be expected beyond pH 11. The lime dose needed to achieve pH 11 depends on the
alkalinity of the water, being about 400 mg/litre Ca(OH)2 for an alkalinity of 200
mg/litre. Scaling can often cause problems when alkaline waters are treated.
A pilot plant study of air stripping for potable water treatment has been carried out19
using a feedwater containing 3 mg/litre NH3–N and a tower packed with serrated wood
slats. Water flowrates of up to 2 m/h were used and under the most favourable conditions
an HTU of 2·3 m was obtained at 15–20°C. This implies a tower packing height of at
least 7 m for 95% ammonia removal. The estimated cost of air stripping, compared to
other methods of ammonia removal, is shown in Fig. 1. The major cost item was
electrical power for the air blowers. Chemical costs for pH adjustment were not included
in the estimate; and could result in a doubling of the total cost for highly alkaline waters.
The results of lower removal efficiency (due to higher ammonia solubility) at low
temperature and the possibility of freezing need also to be taken into account when
deciding if air stripping would be a feasible process in practice.
3.2.3. Biological Nitrification

The cheapest process capable of reducing high ammonia levels (>0.3 mg/litre NH3–N) to
a point (<0·1 mg/litre NH3–N) where problems arising on chlorination are unlikely, is
biological nitrification. Nitrifying bacteria utilise about 4·6 mg/litre of dissolved oxygen
to oxidise 1 mg/litre NH3–N. Therefore if it is required to remove more than about 2
mg/litre NH3–N special measures must be taken to oxygenate the water during the
nitrification process. This is usually accomplished by forced aeration through fixed beds
of sand or granular material to which the nitrifying bacteria adhere.
Normally, nitrification can occur at almost any stage within an existing river water
treatment plant, provided the water has not already been chlorinated. Chedal2 has
described how at Mery-sur-Oise nitrification in the flocculators is encouraged by sludge
recirculation, so that after about two weeks operation 60% of the ammonia is removed.
Up to a further 1 mg/litre NH3–N is then removed by nitrification in the rapid sand filters,
depending on the backwash procedure employed. It was also found advantageous to
reduce the filtration rate from 6 m/h in summer (20°C) to 3 m/h in winter (4°C), to
compensate for the lower rate of bacterial activity.
Ammonia removal by bacterial nitrification on rapid sand filters has also been studied
by Taylor20 who found that residual chlorine present in excess of 0·2 mg/litre adversely
affected the process. His results showed that about 75% ammonia removal was achieved
on the filters over a period of a year, during which the NH3–N level in the raw water
varied between 0·1 and 4 mg/litre and the temperature fell as low as 1°C on one occasion.
Usually, however, there was a tendency for ammonia removal performance to deteriorate
when temperatures fell below 4°C.
The ammonia removal performance of a number of different types of filter has been
studied by Madsen21 using water from the River Weser, which contained ammonia levels
varying seasonally between 0·2 and 2·0 mg/litre. It was observed that during the
prolonged frost of the 1962/63 winter, nitrification in the rapid sand filter ceased and did
not resume until the temperature had risen again to 10°C. The performance of a slow sand
filter was similarly affected, although during several other winters over 80% removal was
maintained. Small-scale tests using granular active carbon or limestone of similar particle
size to rapid sand (~1·5 mm) showed better nitrification performance than rapid sand. A 3
m deep granular carbon bed run at a flowrate of 10 m/h was able to almost totally remove
the ammonia at temperatures down to 5°C. This was entirely due to bacterial action in the
bed since carbon cannot adsorb ammonia. Below 5°C the performance fell off rapidly.
The diminished nitrification performance of filters at low temperature is unfortunate,
since the ammonia levels in rivers also tend to be highest at this time. Madsen has
stressed21 the importance of maintaining the water as close as possible to 100% saturation
with oxygen to assist performance at low temperature. He also demonstrated with a
carbon filter that low temperature performance could be better maintained if the filter had
already been acclimatised to removing the higher winter ammonia levels before the
temperature dropped. This was done by dosing ammonia into the water while the
temperature was still above 10°C and before the ammonia level had risen in the river.
Granular active carbon is capable of removing both ammonia, by bacterial action on
its surface, and dissolved organic matter by a combination of bacterial action and
adsorption. This ability has been put into practice at Müllheim12 where filtration through
granular active carbon is now preceded by ozonation in place of breakpoint chlorination,
in treating water from the River Dohne. This has improved the removal of organic matter
by the carbon since poorly adsorbed chlorinated organic compounds have been replaced
by the readily biodegradable products of ozone oxidation. At the same time the ammonia
(~1 mg/litre) is now removed by bacterial action in the carbon filters, which are 2·5 m
deep and are operated at a flowrate of 10 m/h. A similar 2000 m3/h plant in France22
achieves over 80% ammonia removal on active carbon from ozonised water containing
1·5 mg/litre NH3–N.
The fall-off in nitrification performance at low temperatures that is observed in sand
filters can be largely overcome by the use of fluidised beds in place of fixed beds. Two
waterworks on the River Severn use fluidised beds for ammonia removal23,24 and the
process has also been studied on pilot plants treating water from the Thames25 and
Trent.26 The average ammonia levels are about 0·3 mg/litre in the Thames and Severn,
and 1 to 1·5 mg/litre NH3–N in the Trent. In all cases the fluidised beds were composed
of river silt which formed the support medium for the nitrifying bacteria. When starting
up the beds it has usually been found advantageous to charge the beds with fine (0·05–0·2
mm) sand, although this may not be necessary at times of high river turbidity. The time
taken to establish a fully active biological bed can vary from a few weeks to as much as
three months, depending on conditions.
The waterworks use inverted pyramidal tanks with upflow rates of 7 and 10 m/h at the
top, giving residence times of 14 and 19 min. Usually more than 90% of the ammonia is
removed, although sudden increases in the ammonia level of the river have resulted in
short periods of poorer performance, particularly when the temperature was low. Similar
upflow rates and residence times were used for the River Trent pilot plant and its
performance was also very similar despite the higher (0·5–2 mg/litre) NH3–N
concentration in this river.
The work done with the River Thames pilot plant25 indicated that upflow rates as high
as 24 m/h could be used in a tank with vertical sides. A combination of temperatures
below 8°C and upflow rates higher than 24 m/h caused excessive expansion of the
fluidised bed and a significant drop in the reaction rate. Effective operation of the bed
required a fluidised solids concentration of over 35% by settled volume, as may be seen
from Fig. 2.
FIG. 2. Effect of fluidised bed solids

concentration on nitrification
performance.
The rate of reaction was found to be approximately first-order with respect to ammonia
concentration, as indicated by the gradient of unity in Fig. 3. Thus:
where C is the concentration of NH3–N. The value of K, the rate constant, was found to
be between 0·5 and 1·5 min−1 depending on conditions such as temperature within the
range 5°C to 20°C. Using the lower figure of 0·5 min–1 for K, the calculated residence
times to achieve residual NH3–N concentrations of either 0·1 or 0·04 mg/litre are shown
in Fig. 4. The better performance attained by fluidised beds compared with fixed bed
filters is due to their higher bacterial mass per unit volume.
FIG. 3. Variation of nitrification rate

with ammonia concentration.
Thus far, only the performance of nitrifying systems relying on thedissolved oxygen
already present in the raw water has been considered. They are therefore only capable of
removing 2 to 2·5 mg/litre NH3–N at most. If it is desired to remove more than this
amount of ammonia, the dissolved oxygen content of the raw water may be increased to
40 mg/litre by equilibrating the water with pure oxygen gas at the maximum hydrostatic
pressure available at the base of the nitrification unit. This would enable up to 8 mg/litre
NH3–N to be removed. A fluidised bed system capable of removing 20 mg/litre NH3–N
from a wastewater at Bay Park, USA, has been described.27 It involved recycling just
over two-thirds of the nitrified effluent from the unit back to the inlet, where it re-entered
after being mixed at a ratio of 2·3:1 with the raw wastewater. Hence the combined
FIG. 4. Residence time required in

fluidised bed.
influent to the fluidised bed unit contained only 6 mg/litre NH3–N which was completely
removed in presence of 36 mg/litre of dissolved oxygen, obtained by equilibrating the
influent with oxygen gas. For most potable water treatment purposes, the use of either
oxygen equilibration or partial recycling with re-aeration would be sufficient.
To date, the removal of more than 2 mg/litre NH3–N from potable waters has been
accomplished using two types of aerated fixed bed. They have been employed mainly for
the treatment of groundwaters on the Continent where ammonia levels of 8 or even 16
mg/litre are not unknown. The treatment of these waters by nitrification is facilitated by
their comparatively constant ammonia level and temperature. The two types of aerated
fixed bed are the ‘dry’ sand filters favoured in Holland and the aerated pumice stone beds
used in France.
The operation of ‘dry’ sand filters has been described by Boorsma,28 who studied their
performance for the simultaneous removal of ammonia, iron and manganese. They
usually employ a 1 to 2 m bed depth of 1–2 mm sand, although sand graded down to 0·5
mm has been used. The beds are totally enclosed so that in service both air and water can
be passed downwards through them, using sprays to distribute the water and blowers to
introduce the air. The beds are backwashed periodically in such a way as not to
significantly impair their performance when brought back into service. Operated at
flowrates of from 4 to 9 m/h, the ‘dry’ filters at Hillegom were able to reduce NH3–N
levels of 3–3·5 mg/litre to less than 0·15 mg/litre. Kegel29 quotes two examples of ‘dry’
filters operated at 12 m/h and an air: water volume ratio of 2·5:1 which brought ammonia
levels of 1·4 and 4·5 mg/litre down to less than 0·05 mg/litre.
The best published performance was found by Rossner30 of Hamburg waterworks.
Using two ‘dry’ filters in series at flowrates of 10 m/h and an air: water volume ratio of
3:1, two borehole-derived waters containing 7 mg/litre NH3–N in one instance and 16
mg/litre in another were thereby treated to produce a water containing less than 0·125
mg/litre NH3–N.
Devillers has described in some detail31 the aerated pumice stone beds used at two
5000 m3/h waterworks near Paris. Both the air, supplied by compressors, and the water
are passed upwards through open 2 m deep beds of 8–15 mm pumice. The water then
overflows onto sand filters to remove the bacterial matter it carries over. The size of the
pumice medium was chosen to be small enough to present a large surface area per unit
volume for the bacteria to grow on without at the same time causing an unacceptably
large pressure drop through the bed. It was found necessary to add a trace of phosphate to
the raw water in order to maintain satisfactory bacterial performance.
Water flowrates of up to 10 m/h are used, while the air flowrate is adjusted to provide
sufficient oxygen to complete nitrification without producing a water that precipitates
chalk. The water has an alkalinity of about 300 mg/litre and aeration inevitably removes
some carbon dioxide, so raising the pH. However, since nitrification tends to lower the
pH:
it was found possible to balance the two opposing effects and thus produce a stable water.
One plant treats water containing 4 to 6·5 mg/litre NH3–N while the other treats water
with about 2·5 mg/litre. In the latter case, the aeration is only applied intermittently to
avoid expelling too much carbon dioxide.
Despite the successes that have been achieved with biological nitrification, the fact
remains that it is at present a less reliable (i.e. less predictable) process than biological
denitrification. The period required for start-up may be long and can best be shortened by
seeding the bed with bacteria from another similar plant already in operation. Before any
nitrification plant is built for treating water going into supply it is essential that pilot plant
trials be carried out, lasting several months for a groundwater or at least a year for a
surface water. There remains a need for fundamental research to better determine all the
factors affecting the process.
3.3. NITRATE REMOVAL
3.3.1. Possible Methods

Before considering treatment methods for nitrate removal alone, it may be noted that any
demineralisation process, removing both cations and anions from the water (e.g. reverse
osmosis or a two-bed ion exchange system), will greatly reduce or eliminate the nitrate
content. Economically, reverse osmosis only becomes competitive with ion exchange
when the total dissolved solids content of the water exceeds about 1000 mg/litre. Reverse
osmosis membranes do not reject nitrate preferentially to other common anions, whereas
anion exchange resins retain nitrate in preference to both chloride and bicarbonate. Thus
where the removal of nitrate only is required, ion exchange possesses this advantage over
reverse osmosis. In a short but comprehensive review of nitrate removal methods, Sorg32
states that neither electrodialysis nor reverse osmosis systems have yet been installed
solely for the removal of nitrate from water.
Disregarding demineralisation, three main methods of nitrate removal have received

attention in the past: chemical reduction, bacterial denitrification and ion exchange using
a strong base resin. These are illustrated schematically in Fig. 5 which shows in each case
the chemical requirements of the method entering at the top, the waste products leaving at
the bottom, and the constituents added to the water as a result of treatment at the right.
The possibility of reducing nitrate chemically using ferrous hydroxide precipitated in
situ has been investigated in some detail by Gunderloy33 and Buresh.34 This reductant
was chosen after carrying out an extensive literature survey followed by feasibility tests
on a number of reducing and deamination agents. The presence of cupric or silver ions
was found to catalyse the reduction by ferrous hydroxide. But the time still required for
the reaction, the variety of possible end-products and the voluminous sludge produced
made the method look unattractive as a practical treatment.
A possible future method for nitrate reduction could involve the use of immobilised
enzymes. Some work has been carried out to elucidate the enzyme mechanism used by
bacteria for denitrification; and the reduction of nitrate has been achieved in the
laboratory using simple molybdenum
FIG. 5. Nitrate removal methods.

complexes at both Manchester35 and Oakland, Michigan36 Universities. This work is still
however at an early research stage.
Biological denitrification by naturally occurring denitrifying bacteria has been studied
extensively for the purification of wastewaters and sewage effluents. In the presence of a
suitable organic substrate, these bacteria can reduce nitrate to nitrogen gas. The most
usual substrate used has been methanol since it is relatively cheap and easy to handle.
The only waste product is the excess bacterial growth which must be removed from the
unit for disposal. More recently, this process has been studied37 on a pilot plant for nitrate
removal from River Thames water.
In the UK nitrate levels in rivers tend to increase in late autumn and winter when the
rain washes it out of the soil. Treatment for nitrate removal may therefore not be needed
during the summer months. In any biological treatment method, however, the rate of
bacterial action declines at low temperatures, thus making necessary a longer residence
time in the treatment unit. The bacterial growth also requires time to become established
in the unit after start-up, and at least two weeks are usually needed to attain full
performance. The denitrified water is always anaerobic, may contain traces of residual
methanol and will also carry over some bacterial floc with it. Thus denitrification needs
to be followed by aeration and filtration. Any residual methanol may be removed by
evaporation in the aerator and, provided the water is not pre-chlorinated, by further
bacterial action in the clarifier or filters.
Although a denitrification plant may not be as convenient to operate as a single-bed
ion exchange plant, it is less expensive to install and operate. It has not however yet been
used in practice as has ion exchange. Figure 6 compares the estimated costs of the two
processes, not including nitrate or methanol monitoring costs. The intercept on the cost
axis represents the annual capital cost amortised at 10% over 15 years for ion exchange
and over 20 years for denitrification plant. Complete resin replacement has been assumed
every five years. The gradient of the lines is proportional to the operating costs which can
be high for ion exchange. As an alternative to treatment, Fig. 6 also includes the cost of
blending the high nitrate water with a low nitrate water piped in from a distant source
using 3:1 and 1:1 blend ratios to lower the nitrate level in supply.
Ion exchange has already been used for the removal of nitrate from groundwaters. A 2
mgd continuous ion exchange plant has been used at Long Island, USA;38 and a similar
plant of 0·37 mgdcapacity is operating in Lincolnshire.39 A 0·25 mgd fixed bed plant with
the capacity for treating at least double this flow has been undergoing trials in Suffolk.39
There is as yet no nitrate removal plant treating a river water. Organic fouling of the resin
could well be a problem if ion exchange were used for this application, so that biological
denitrification would be the more promising alternative. Apart from organic matter, iron
and to a lesser extent silica can cause resin fouling when present.40 Pretreatment for iron
removal may therefore be required with some waters.
Ion exchange has the advantage of being a process that is predictable in performance
and little affected by temperature change. Hence it is
FIG. 6. Costs for nitrate removal per 1

mgd (December 1976). (mls=miles.)
amenable to automation and should be capable of working at a borehole site with little
supervision. On the other hand a troublesome waste disposal problem is usually presented
by the exhausted regenerant and rinsewater which consist of a mixed solution of sodium
chloride, nitrate and sulphate, resulting from regeneration of the resin with brine. If this
waste has to be tankered away, to the coast for instance, the operating cost of the plant
could be at least doubled, as indicated in Fig. 6.
Since biological denitrification and ion exchange are the only two practical methods of
nitrate removal at the present time, these will now be considered in greater detail,
commencing with the longer established ion exchange method.
3.3.2. Ion Exchange

The aim of treatment will usually be to reduce the level of nitrate in the water put into
supply to below 10 mg/litre as N.Since ion exchange removes almost all the nitrate from
the water being treated, it may only be necessary to treat a proportion of the water. The
treated portion can then be blended with the remaining untreated portion to give the
required nitrate level in supply.
It has been found40 that commercially available strong and weak base resins show
greater selectivity for sulphate than for nitrate over the concentration range (<5000
mg/litre TDS) normally found in ground-waters. It is not, therefore, at present possible to
remove nitrate by ion exchange without also removing sulphate. The single-bed process,
using a strong base resin in the chloride form, replaces these anions by chloride ions. The
resin is then regenerated with sodium chloride which is a relatively cheap chemical.
However, this treatment may produce a water containing a high enough ratio of chloride
to bicarbonate for it to dezincify domestic brass fittings.
If an increase in chloride concentration is unacceptable or if calcium and magnesium
removal (i.e. softening) is also required, a two-bed ion exchange process which at least
partially demineralises the water may be used. This process employs a strong acid resin,
regenerated with mineral acid, followed by a weak base resin, regenerated with alkali.
Carbon dioxide gas is formed as the water passes through the strong acid resin bed so that
a degasser must be included at a later stage to remove it.
Clifford and Weber41 have compared the performance and cost of the single and two
bed processes and found the latter to be up to twice as costly to install and operate. They
also determined how the performance of the anion resin bed in each process was
influenced by the nitrate, chloride and sulphate selectivities of the resin, using water
containing equivalent amounts of each of these anions. A wide range of styrene-
divinylbenzene resins and some macroporous acrylic-amine and phenol-formaldehyde
resins were recommended as a result of their tests, and these all performed very similarly.
To reduce spent regenerant disposal costs for the two-bed process it has been
suggested41 that ammonium hydroxide and nitric acid be used as regenerants. The
combined spent regenerant could then be given away as fertiliser on account of its
ammonium nitrate content. Alternatively, the cost of regeneration could be lowered by
using lime slurry to regenerate the weak base resin, as in the ‘Ducol’ process.42 Also, the
regenerant requirement may be reduced by postponing regeneration until divalent metal
breakthrough occurs on the acid resin and nitrate breakthrough on the basic resin. Any
sodium chloride present in the raw water is then not retained by the resins, but is allowed
to break through before termination of the service cycle.
However, in practice first consideration is usually given to the single-bed process if
nitrate removal alone is required. The performance of some commercial strong base
resins treating two model hard waters having the same nitrate content (14 mg/litre as N)
but differing sulphate contents (20 and 120 mg/litre) has been studied43 using laboratory
columns. Figure 7 shows the nitrate, bicarbonate, and chloride levels found in the column
effluent during an exhaustion run. The sulphate was totally removed until well beyond
the point of nitrate breakthrough where a service run would normally be terminated and
regeneration carried out. The large variation in chloride level and also in alkalinity (and
hence pH) during the run would be unacceptable for a water going into supply. Several
resin beds run in parallel and regenerated in sequence, combined perhaps with a treated
water blending tank, are therefore required in practice to supply a water of uniform
composition.
The problem of variation in treated water quality can also be largely overcome by use
of a continuous ion exchange plant such as the Chemseps continuous loop.38 In one
section of the loop resin treats the water, while exhausted resin is regenerated and rinsed
in another. The resin is periodically (every 20 min or so) pulsed round the loop so that a
portion of exhausted resin leaves the top of the treatment section and is replaced by
freshly regenerated resin entering the bottom of the section. Thus the variations in treated
water quality, which are less marked than with fixed beds, occur over a period of a few
minutes instead of several hours.
To minimise regenerant consumption, fixed beds are usually partially regenerated in a
direction counterflow to the direction of the water flow when in service. Partial
regeneration involves using 0·8 to 1·0 bed volumes of 10% brine or twice this volume of
5% brine. The regeneration efficiencies that are then obtained, expressed as equivalents
of nitrate removed per equivalent of sodium chloride regenerant used, were found to be43
0·24 and 0·12 when treating waters containing 14 mg/litre NO3–N with 20 mg/litre SO4
and 120 mg/litre SO4 respectively. Similar regeneration efficiencies were found using a
continuous loop instead of columns.
To minimise spent regenerant disposal costs, it is important to keep the spent
regenerant and rinse water volumes to a minimum. In this respect, the continuous loop
has an advantage over the conventional fixed bed system. When treating a high sulphate
water (>120 mg/litre), partial regeneration of a fixed bed may be expected to give a
volume of waste for disposal in excess of 2% of the water treated, whereas a continuous
loop produces a waste volume of under 0·5% of the volume treated. This is because the
continuous loop can use saturated brine for regeneration and because the volume of rinse
water it uses can be kept small. Treating low sulphate water
FIG. 7. Treated water quality from an

ion exchange column.
(20 mg/litre) results in the volume of water treated per regeneration being doubled and
hence in halving the regenerant and waste disposal costs of either ion exchange system.
The effect that high raw water sulphate levels have in raising regeneration and waste
disposal costs could be overcome by the development of a resin that was more selective
for sulphate than for nitrate. Grinstead and Jones44 have achieved this objective with
liquid amidine resins for which they proposed ammonium hydroxide as regenerant. The
major problem that they anticipated was in the means by which a liquid resin could
actually be employed to treat water and in preventing an unacceptable concentration of

the resin itself appearing in the treated water. More recently45 the introduction of amidine
groups onto conventional acrylic resin beds has resulted in a resin with significantly
reduced sulphate selectivity and which can be regenerated with brine. However, it also
exhibits a low capacity and low ‘regenerability’ and it is therefore doubtful if it will
possess any net advantage over existing commercial resins.
3.3.3. Biological Denitrification

A comprehensive study has been carried out37 to assess the performance and applicability
of biological denitrification to the treatment of water for potable supply. The data
presented here are based on the results of this study in which methanol was used as the
source of organic carbon for the denitrifying bacteria. The water was derived from the
River Thames with provision for dosing extra nitrate when required. The treatment
objectives were to obtain design criteria for a plant capable of reducing the nitrate level
by 10 mg/litre (as N) at temperatures below 5°C, without necessarily removing the nitrate
completely. The methanol level in the treated water was however required to be kept very
low (<0·5 mg/litre), since it was anticipated that in practice strict limits would be placed
on the methanol level in the water entering supply after filtration, if not also on that in the
water leaving the denitrification unit itself. Since both temperature and methanol
concentration were rate-limiting, these conditions greatly affected the residence time
required in the denitrification unit.
Three types of denitrification unit were studied: fixed beds, suspended growth, and
fluidised sand units. One fixed bed contained 1 m depth of 10 mm gravel, and another 2
m depth of 25 to 40 mm gravel. The bacteria grew as a film on the gravel surface and the
excess growth had to be removed periodically by back-scouring to prevent the beds from
clogging. Water flowrates of 0·7 to 2·8 m/h were used during treatment.
The suspended growth unit was a tall tank in which the water flowed vertically
upwards. After initial seeding with fine sand or river silt as a basis for attachment, the
naturally occurring denitrifying bacteria formed grey gelatinous globules which remained
suspended as a fluidised layer in the tank. Most of the globules were 1 to 2 mm in
diameter, but some were as large as 6 mm. The suspended layer increased in depth until,
when it filled
FIG. 8. Concentration profiles

(fluidised sand bed).
the tank, the excess bacterial growth was carried over in the denitrified water. Upflow
rates of up to 24 m/h could be used, although 12 m/h was more usual. The lower rate was
necessary at low water temperatures to allow an adequate suspended solids concentration

to be maintained in the unit.
The fluidised sand unit was similar except that it was initially filled to one third of its
depth with 0·2 to 0·5 mm settled sand. The sand became fluidised to about twice its
settled depth when an upflow rate of 20 m/h was used. After a week or two of operation a
thin film of bacterial growth became established on the sand surface causing further
expansion of the fluidised layer. To prevent carry-over of sand by the denitrified water,
the top of the fluidised layer was scoured periodically with a high speed stirrer.
The methanol dose required depended on the dissolved oxygen content of the water
and on the nitrate removal required. The denitrifying bacteria first utilised the dissolved
oxygen (DO) to oxidise the methanol to carbon dioxide and water. Only when the oxygen
had been almost totally removed did the bacteria start to utilise the chemically bound
oxygen in nitrate for the same purpose. A typical set of profiles for methanol, DO and
nitrate through a denitrification unit is shown in Fig. 8 and the following relationship was
found to hold:
CH3OH consumed=1·1 DO+2·47 NO3–N removed (mg/litre).
The methanol requirement was about 30% above that stoichiometrically needed for its
oxidation since the bacteria utilised some methanol for cell and slime synthesis.
Unlike nitrate removal by ion exchange which takes place rapidly, denitrification rates
tend to be low and determine absolutely the residence times and hence size of unit
required. Table 1 shows the denitrification rates that were obtained in the different types
of unit, expressed as the weight of NO3–N removed per cubic metre of the unit per hour.
The volume of each type of unit required to perform a given denitrification duty is then
inversely proportional to the rate. It may be seen from Table 1 that the denitrification
rates were related to the bacterial film areas per unit volume, which in turn were
governed by the particle sizes of the media used.
The rates of denitrification were found to be independent of nitrate concentration
down to very low nitrate levels (less than 1 mg/litre as N). They were however dependent
on the methanol concentration, becoming approximately first-order with respect to
methanol at low (less than 5mg/litre) methanol concentrations. The curvature of the
methanol and nitrate profiles in Fig. 8 reflects this effect. The denitrification rates given
in
TABLE 1 DENITRIFICATION UNIT
CHARACTERISTICS
Fixed Fixed Suspended Fluidised
bed bed growth sand
Media dia. (mm) 25–40 10 1–2 0.2–0.5
Bacterial matter
concentration (4–8) (5– 5–20 10–15
(kg/m3) 10)
Bacterial film area
per unit volume 120 360 700 3000
(m2/m3)
Denitrification 2.5 12 60 160

10°CM
rate (gN/m3h) 5 24 120 310
20°CJ
Table 1 were average rates found for a methanol concentration drop from 20 mg/litre to 1
mg/litre. A lower final methanol concentration would result in a lower average
denitrification rate.
The effect of temperature is indicated in Table 1 and further illustrated in Fig. 9. A
10°C drop in temperature approximately halved the denitrification rates.
The coarsest media gave the lowest denitrification rates and vice versa. The fixed beds
therefore exhibited the lowest rates and would consequently need to be unacceptably
large in practice. They also had the disadvantage of giving impaired performance for a
time after each scouring, which was carried out every few weeks.
Considering therefore the requirements for a practical suspended growth or fluidised
bed denitrification unit: the incoming water, which must not be chlorinated, is dosed with
methanol, about 36 mg/litre being required for example to remove 10 mg/litre DO and 10
mg/litre NO3–N. The flow distribution arrangement at the base of the denitrification unit
must be adequate to totally fluidise the bed since unfluidised pockets can turn black and
emit hydrogen sulphide, particularly in warm conditions. The period required to start up a
bed and establish bacterial growth can be reduced if the water leaving the unit is initially
almost totally recirculated to the inlet, thus retaining within the system any fine bacterial
floc not yet adhering to the bed and preventing unutilised methanol from being
discharged.
An operational suspended growth or fluidised sand unit usually contained about 12.15
kg/m3 of bacterial matter (volatile suspended solids). In the case of the fluidised sand
unit, this bacterial concentration was controlled by adjusting the amount of sand held in
the bed. Some mechanical device was also needed to keep down the top of the fluidised
sand bed to a constant level by continuously or intermittently stripping excess bacterial
growth from the sand surface.
Table 2 shows the effective depths and residence times of the beds required to remove
10 mg/litre DO and 10 mg/litre NO3–N at 2°C, assuming that an effluent methanol
concentration of not more than 0·2 mg/litre is required (thus lowering the average
denitrification rates shown in Fig. 9 by a factor of 0·75).
FIG. 9. Effect of temperature on

performance of suspended growth
(SG) and fluidised sand (FS) biological
denitrification units.
TABLE 2 DENITRIFICATION UNIT
OPERATING REQUIREMENTS FOR
REMOVAL OF 10 mg/Htre NO3–N AT 2°C
Type of Upflow Bed Residence
unit rate (m/h) depth time (min)
(m)
Suspended 12 5·0 25
growth
Fluidised 20 3·6 11
sand
The fluidised sand bed is clearly the fastest type of denitrification unit. The unit used in
this study37 was a 0·3 m diameter column with a throughput of about 35 m3/day. Larger
fluidised sand pilot plants (150 to 300 m3/d) have been used for both ammonia and nitrate
removal from municipal wastewater;27 and their use has now been patented.46 Monitoring
requirements are likely to be more stringent for potable than for wastewater treatment
applications; and the equipment for this could more than double the capital cost of the
plant shown in Fig. 6. Even so, biological denitrification compares very favourably in
cost with ion exchange for nitrate removal.
REFERENCES
1. PALIN, A.T., Water & Water Engng, 1950, 54, 151, 189, 248.
2. CHEDAL, M., Techniques et Sciences Municipales, 1975, 70, 173.
3. HYNDSHAW,A.Y., Taste & Odour Control J., 1975, 41, No. 2, Westvaco Corp., Covington,
Va.
4. BAUER, R.C. and SNOEYINK, V.L., J. Water Pollut. Control Fed., 1973, 45, 2290.
5. SAUNIER, B., L’Eau et L’Industrie, 1976, 8, 71.
6. SAUNIER, B.M. and SELLECK, R.E., J. Amer. Water Works Assoc., 1979, 71, 164.
7. WEI, W.I. and MORRIS, J.C., Chemistry of Supply, Treatment and Distribution, 1974, Ch. 13,
Ann Arbor Science Publishers, Ann Arbor, USA.
8. BABCOCK, D.B. and SINGER, P. C., J. Amer. Water Works Assoc., 1979, 71, 149.
9. OLIVER, B.G. and LAWRENCE, J., J. Amer. Water Works Assoc., 1979, 71, 161.
10. SYMONS, J.L., BELLAR, T.A., CARSWELL, J.K., DE MARCO, J., KROPP, K.L.,
ROBECK, G.G., SEEGER, D.R., SLOCUM, C.J., SMITH, B.L. and STEVENS, A.A., J. Amer.
Water Works Assoc., 1975, 67, 634.
11. ROOK, J.J., J. Amer. Water Works Assoc., 1976, 68, 168.
12. SONTHEIMER, H., HEILKER, E., JEKEL, M.R., NOLTE, H. and VOLLMER, F.H., J. Amer.
Water Works Assoc., 1978, 70, 393.
13. GNIESER, J., Chemische Rundschau, 1977, 34.
14. GINOCCHIO, J., Wasserwirtschaft, 1978, 68, 180.
15. HOIGNÉ, J. and BADER, H., Envir. Sci. Technol., 1978, 12, 79.
16. JORGENSEN, S.E., Water Res., 1976, 10, 213.
17. MERCER, B.W.,AMES, L.L.,ToUHILL, C.J.,VANSLYKE, W.J. and DEAN, R.B., J. Water
Pollut. Control Fed., 1970, 42, R95.
18. Water Pollut. Control Research Series, Project 17010, ECZ, 1971, US EPA, Water Quality
Office, Washington.
19. SHORT, C.S., Tech. Paper No. 101, 1973, Water Research Centre, Medmenham, Bucks.
20. TAYLOR, E.W., Proc. Instn. Civ. Engrs, 1953, 3, 398.
21. MADSEN, S., Vom Wasser, 1975, 45, 103.
22. GOMELLA, C., L’Eau et L’Industrie, 1977, 16, 78.
23. PARKER, S.S., Water Treat. Exam., 1972, 21, 315.
24. MILLINER, R., BOWLES, D.A. and BRETT, R.W., Water Treat. Exam., 1972, 21, 318.
25. SHORT, C.S., Tech. Rep. No. 3, 1975, Water Research Centre, Medmenham, Bucks.
26. MELBOURNE, J.D., Tech. Rep. No. 74, 1978, Water Research Centre, Medmenham, Bucks.
27. JERIS, J.S., OWENS, R.W. and HICKEY, R., J. Water Pollut. Control Fed., 1977, 49, 816.
28. BOORSMA, H.J., H2O, 1976, 9, 363.
29. KEGEL, J., Gas- u. Wasserfach, 1962, 103, 396.
30. ROSSNER, F.X., Wasserfachliche Aussprachetagung des DVGW und BGW, 1971, Wiesbaden.
31. DEVILLERS, G., Techniques et Sciences Municipales, 1975, 60, 295.
32. SORG, T.J., J. Amer. Water Works Assoc., 1978, 70, 105.
33. GUNDERLOY, F.C., WAGNER, R.J. and DAYAN, V.M., Water Pollut. Control Research
Series, Project 17010, EEX, 1970, US EPA, Water Quality Office, Washington.
34. BURESH, K.J. and MORAGHAN, J.T., J. Environ. Qual., 1976, 5, 320.
35. GARNER, C.D., HYDE, M.R. and MABBS, F.E., Nature, 1975, 253, 623.
36. KETCHUM, P.A., TAYLOR, R.C. and YOUNG, D.C., Nature, 1976, 259, 202.
37. GAUNTLETT, R.B. and CRAFT, D., Tech. Rep. No. 98, 1979, Water Research Centre,
Medmenham, Bucks.
38. GREGG, J.D., Civ. Engng, 1973, 43, 45.
39. GREENE, L.A., Water Pollut. Control, 1978, 77, 478.

40. BUELOW, R.W., KROPP, K.L., WITHERED, J. and SYMONS, J.M., J. Amer. Water Works
Assoc., 1975, 67, 528.
41. CLIFFORD, D.A. and WEBER, J.W., Ind. Water Engng, 1978, 15, 18.
42. EVANS, S.J., Water Pollut. Control Fed., 1973, 45, 632.
43. GAUNTLETT, R.B., Water Treat. Exam., 1975, 24, 172.
44. GRINSTEAD, R.R. and JONES, K.C., Water Pollut. Control Research Series, Project 17010,
FSJ, 1971, US EPA, Water Quality Office, Washington.
45. Diaprosim Ltd, Contract Report, 1979, Water Research Centre, Medmenham, Bucks.
46. Brit. Pat. 1430410, 1973.
Chapter 4
DESALINATION
M.J.BURLEY, B.Sc., M.Sc., C.Eng., M.I.Chem.E., M.I.E.W.E.S.
Sir M.Mac Donald and Partners, Demeter House, Cambridge, UK
AND
J.D.MELBOURNE, B.E.(Chem) (Aust), Ph.D.(Cantab.)
Melcon Water International Ltd, Henley-on-Thames, Oxon., UK
SUMMARY
Desalination has progressed significantly since the simple distillation techniques first
used for producing small quantities of drinking water from sea or saline ground waters.
Distillation is normally now only applied to sea waters and the most successful, reliable
and widely applied technique is the multi-stage flash process, particularly in the Middle
East. Improved scale control and use of materials have been the major developments with
this process. A new technique using fluidised bed heat exchange may lead to cost
reductions, but this will not be fully commercial for several years. Reverse osmosis and
electrodialysis, both membrane techniques, are the two most commonly applied processes
for the desalination of brackish waters. Of these, reverse osmosis has been the more
successful and, although a relatively new process, is now considered to be technically
and commercially reliable. Developments with reverse osmosis membranes have also led
to a capability of desalting sea water and several small commercial plants have been
installed. It is possible that the slightly more attractive economics of sea water reverse
osmosis will result in a decrease in sea water distillation applications. Electrodialysis
also has the capability of treating sea water but this is not yet commercial. The other
techniques of freezing and ion exchange have had little commercial impact in the
desalination field.
4.1. INTRODUCTION
Distillation processes, in a simple form, have been employed for over a century as a
means of producing drinking water from the sea. However it was not until the late 1950s,
with the development of oil communities in the Middle East and the Caribbean that
demand arose for large-scale land-based plant. This market spurred several governments
to finance research and development in the field of desalination and foremost of these
programmes was that of the United States Office of Saline Water.
After the initial 4500 m3/day multi-stage flash (MSF) distillation plants were installed
in the late 1950s, the process was rapidly established as the foremost desalination
technique. Works of the 1970s involve units of up to 35000 m3/d capacity and countries
such as Kuwait and Saudi Arabia have total installed capacities of about 500000 m3/d.
The process remains relatively unchanged from the original concept but the better
understanding of aspects such as scale control, flashing brine characteristics and materials
has limited the ever rising costs and allowed more reliable operation. For the future,
techniques such as the use of fluidised bed heat exchange systems show promise for cost
reduction but the basic MSF process seems likely to remain the major sea water
distillation process.
Reverse osmosis has continued to develop rapidly since its beginnings in 1960. The
progress in the manufacture of membranes and modular support systems has led to the
establishment of the technique as a reliable and economic means of brackish water
treatment which is now being widely employed on a commercial basis. Two systems are
being used: one based upon spirally wound membranes and the other upon self-
supporting, fine, hollow fibres. During the 1970s the major changes which have occurred
are the gradual improvement of membrane properties, a better understanding of
membrane fouling and the measures required to eliminate the associated problems, and
the development of membranes capable of treating sea water in a single pass. Whilst sea
water treatment by reverse osmosis has not yet been widely used it is clear that the
process will develop and become an economic competitor of distillation for large-scale
sea water treatment.
Whilst the use of electrodialysis for the treatment of brackish waters with total
dissolved solids contents up to about 5000 mg/litre has continued to grow, the application
of the process has clearly suffered from the competition of the simpler reverse osmosis
process. The use of frequent current reversal to counteract polarisation effects and
consequent scaling problems is the most significant single development during the 1970s
since this reduces the need for pre-treatment and the use of chemicals, a factor of
economic importance in situations where the costs of imported chemicals are high.
Other processes under study in the 1960s included solar distillation, ion exchange and
freezing techniques. The development of these processes has slowed down, if not ceased,
during the 1970s. Despite the potential of lower energy consumptions it seems unlikely
that these desalination methods will be developed to a commercial stage for large-scale
use.
4.2. SEA WATER DISTILLATION
The boiling of saline water and condensing of steam to produce potable water has been
employed for centuries but it is only in the last 30 years that the factory-built plants to
produce large quantities of water have been developed. As early as 18651 the water
supply for General Napier’s expedition into Ethiopia was provided by two 30 ton/day
units installed at Alexandria. The process was basically similar to the multiple-effect
submerged tube distillation which was employed through to the 1950s.
In this type of plant feed steam is condensed on the inside of coils or tube banks
suspended below the surface of the feed sea water held in the first effect. The steam
produced from the vessel is passed as heating steam to the next effect. In general,
submerged tube plants consist of no more than three effects and therefore have a poor
efficiency in terms of energy use. The performance ratio R (mass of product water per
Desalination 81
unit mass of feed steam consumed) is typically only about two. The process has two other
serious defects:
(1) The rate of heat transfer between condensing steam and evaporating water is very low
owing to poor turbulence outside the tubes—the overall heat transfer coefficient is
typically 500 W/m2 °C (100 Btu/ft2 h °F). Thus large areas of expensive nonferrous
heat transfer surface are required.
(2) Scale which forms on the outside of the tube bundles is extremely difficult to remove.
The introduction of multi-stage flash (MSF) distillation in the late 1950s revolutionised
desalination and provided a practical, economic and relatively trouble-free means of
supplying drinking water. This development arose directly as a result of the large
demands for drinking water which were created by the post-war growth of the Arabian
Gulf Oil States and similar demands in the Caribbean and developing island communities
FIG. 1. 22750 m3/d MSF Evaporator

(one of four units constructed by Weis
Wertgerth Ltd for Quatar).
and has led to the construction of works similar to that shown in Fig. 1. The major
benefits of MSF distillation over the submerged tube process are as follows:
(a) Sea water passes at high velocity over heat transfer surfaces thereby allowing the heat
transfer coefficients to be increased by a factor of about five to 2800 W/m2 °C (500
Btu/ft2h °F).
(b) Sea water is not boiled at the heat transfer surface and, as a result, scale formation can
be more readily controlled.
(c) The plant design allows the use of a large number of evaporation stages which has the
effect of increasing the efficiency. Performance ratios of 10 became a reality
overnight.
These achievements were the direct result of the development work of R.S.Silver and
A.Frankel (Silver, private communication) who both came to almost identical
conclusions while working independently.
Whilst several other means of sea water evaporation have been developed, including
various forms of multiple effect (ME) and vapour compression (VC) processes, the multi-
stage flash distillation process is still used almost exclusively for the large land-based sea
water desalination plants (greater than 5000 m3/d capacity). Therefore MSF distillation
has been selected for detailed consideration.
4.2.1. Multi-Stage Flash Distillation

The multi-stage flash process can be illustrated as shown in Fig. 2. Cooling water is fed
to the condensers of the heat rejection section of the plant. On
FIG. 2. Simplified flow sheet MSF.

emerging, a portion of the cooling water is separated and, after chemical treatment, is
passed to the inlet of the recycle pump where it is mixed with flashing brine withdrawn
from the final stage (stage 7 in the works illustrated). After discharge of the concentrated
waste stream, MB, the recycle flow MR is passed through the condensers of the heat
recovery section where it is successively heated by condensing steam. On emerging from
the condensers of stage 1 the recycle stream is heated to its highest temperature, TFO, by
heat exchange with feed steam in the brine heater.
Desalination 83
The heated brine then enters the flash chamber of stage 1 which is held at such a
pressure that a portion of the brine stream flashes off thereby cooling the flashing stream
to its boiling point at the stage pressure. As the flashing brine stream passes from stage to
stage it is successively cooled with a portion of the flow flashing off, to be condensed as
distillate in each stage.
The temperature distribution is shown in Fig. 3. Mass and heat balances
FIG. 3. Temperature distribution MSF.

can readily show that the performance ratio R approximates to the following:2
(1)
Thus the energy consumption can be reduced by the following means:

(i) Increasing ∆TT the total flashing range.
Since the lower temperature TRI is fixed by the temperature of the sea, ∆TT can
only be increased by raising the top temperature TF0. Developments in scale
control have been aimed at increasing the top temperature whilst avoiding undue
corrosion.
(ii) Reducing the stage temperature drop ∆t.
This can only be achieved by increasing the number of stages employed and
hence increasing the capital cost of the plant.
(iii) Reducing the value of , the stage approach temperature (the temperature difference
between recycle brine leaving the stage and condensing steam).
This can be achieved by use of more heat transfer area in each stage or increasing
the heat transfer coefficients.
α is the temperature difference between flashing brine and condensing distillate and is the
sum of the boiling point elevation and the temperature difference caused by pressure
losses. Little can be done to reduce the value of α.
From points (ii) and (iii) above it is clear that plant design can be optimised to achieve
a lowest water cost by balancing capital with fuel costs (or balancing heat transfer area
and number of stages with performance ratio). This is shown by the following
approximate relationship:
(2)
where As, the specific heat transfer area, represents the heat transfer area per unit of
product water output; λ is the mean latent heat of evaporation over the flashing
temperature range; U is the mean heat transfer coefficient; ∆TT is the flashing
temperature range; n is the number of stages; and R is the performance ratio.
Using typical values, this relationship has been plotted as shown in Fig. 4.
FIG. 4. Heat transfer requirements

MSF.
Desalination 85
From eqn (2) and Fig. 4 it can be seen that costs of MSF distillation can be reduced by
the following means:
(1) Increasing the flashing range ∆TT.
(2) Increasing the heat transfer coefficient U.
(3) Use of cheaper heat transfer and shell materials.
(4) Use of improved designs to permit smaller flash chamber.
(5) Use of cheaper energy sources.
It is therefore of value to consider the developments in distillation in terms of their effects
upon these factors.
4.2.2. Scale Control

The flashing temperature range is limited by the maximum brine temperature at which
scale formation can be satisfactorily controlled. Alkaline scale formation, caused by the
breakdown of bicarbonates in sea water to deposit calcium carbonate and magnesium
hydroxide scales, has traditionally been controlled by one of two alternative means:
(a) The dosing of polyphosphates at low levels, 5 to 10 mg/litre, does not completely
eliminate scale formation but rather causes scales to form as sludges which do not
adhere strongly to the heat transfer tube surfaces. The use of these materials limits the
top temperature to about 90°C since at higher temperatures the phosphates break down
and become ineffective. Periodically it is necessary to acid flush the condenser tubing
in order to remove the sludge build-up which will otherwise cause a deterioration of
plant performance.
(b) Pretreatment of the feed water by dosing of acid, at a concentration of about 120
mg/litre, effectively removes the scale-forming bicarbonates with evolution of carbon
dioxide. The gas is removed by air or steam stripping prior to entry of feedwater to the
plant, otherwise severely corrosive conditions will be created within the evaporators.
Even so, the negligible buffering capacity of the recycle brine makes the control of pH
difficult and therefore the risk of corrosion will always be present when acid dosing is
used for scale control.
By the use of acid, top temperatures of 110−115°C can be achieved without
serious scale formation.
The benefit of this higher temperature can be illustrated as follows:
If the temperature of sea water controls the bottom flashing temperature to 38°C, the
total flashing range with acid treatment will be 77°C (=115 −38) as compared with 52°C
(=90−38) for operation with polyphosphates. Substitution of these values in eqn (2) will
show that the heat transfer area can be reduced by about 47% by the use of acid dosing
while still retaining the same number of stages and performance ratio.
A rigorous evaluation will show the difference in costs between polyphosphate and
acid dosed plants to be greater at high performance ratios. The approximate relationship
is shown in Fig. 5.
FIG. 5. Plant capital costs—

polyphosphate/acid.
This capital cost saving has led many users to select acid dosed plant in preference to the
more expensive polyphosphate dosed works. In a number of cases poor pH control and
the use of unsuitable materials has caused premature failure of heat exchange tubing and
steelwork, so much so that the cost of replacements and repairs has outweighed the initial
benefits. The Ministry of Electricity and Water, Kuwait, is typical of several authorities
who have deliberately avoided the use of acid and in so doing have maintained a high
standard of reliability.
Belgarde for Scale Control

One of the most significant improvements of recent years is the development of additives
which effectively control scale at high temperatures without the creation of corrosive
conditions. Whilst many of the chemicals developed have shown only marginal benefits
over the traditional polyphosphates the Belgarde range of chemicals, developed by Ciba-
Geigy Ltd have been found to be most effective. Top temperatures of 105–110°C have
been successfully achieved over many months of operation with insignificant rates of
fouling of heat transfer surfaces and with iron and copper dissolution rates typically 10%
and 30% of those experienced in acid dosing plants.
Trials with Belgarde at Qatar are typical of many uses of the chemical. In this case the
plant, 9000 m3/d capacity, had previously been operated with polyphosphates at a
maximum temperature of 85°C and intermittent acid cleaning at six-monthly intervals.
Dosing with 7·5 mg/litre of Belgarde over a five-week monitored test period showed no
increase in fouling factor with a top temperature of 110°C and an output of 11 800 m3/d,
an increase of 30% above that for polyphosphate operation. Extended operation of 150
days at a top temperature of 105°C and a Belgarde dose of 6 mg/litre showed the fouling
Desalination 87
factor to remain constant whilst the output was maintained at 10 900 m3/d, an increase of
21% over the production rate when polyphosphates were employed.
Without doubt the high maintenance cost often associated with high temperature plant
will be minimised in the future by the use of scale-control polymers such as Belgarde.
4.2.3. Fluidised Bed MSF

A second development which permits the use of high temperatures without resort to acid
dosing involves the use of a fluidised bed of ‘seeds’ within the recycle condenser tubes.3
In this process, the stages of a multi-stage flash evaporator are mounted vertically with
condensers located at the side of the flashing chambers (Fig. 6). The recycle brine is
pumped vertically up the condenser tubes at a velocity adequate to fluidise glass particles
typically of 2 to 4 mm diameter. The purpose of the fluidised bed is fourfold:
(1) To create sites for scale precipitation and thereby minimise deposition on heating
surfaces.
(2) To abrade the surface in order to remove what scale does deposit.
(3) To take advantage of the high film heat transfer coefficients experienced with
fluidised beds.
(4) To achieve scale control without incurring the costs associated with the use of
chemicals.
Whilst the development of the process is by no means complete the
FIG. 6. Fluidised bed evaporator MSF.

results of trials undertaken with a 50 m3/d experimental unit have been sufficiently
promising4 that a 500 m3/d trial unit has been constructed and started operation in 1979.
In addition to the benefits listed above, significant economies are likely owing to the
reduction in stage steelwork requirements in the use of a vertical layout. Wire mesh
demisters to reduce the carry-over of entrained brine droplets into the distillate have been
found to be unnecessary and a very close approach to flashing equilibrium has been
found possible. These effects could have the effect of reducing the value of α (eqn 1) by
about 0·6°C which, in high performance ratio plant, could reduce area requirements by
5%. The reduced steelwork costs could give a further 5–10% saving.
The potential high temperature operation could lead to a 10% saving and
improvements in heat transfer a further 10%.
Thus overall a capital cost reduction of some 40% may prove possible and since the
process, like the basic MSF process, is amenable to capital/operating cost optimisation,
this advantage can be shared between capital and fuel costs. It is anticipated that the
continuing development, jointly by Delft University, Esmil Ltd and the Dutch
Government, will lead to the use of this process in the later 1980s.
Desalination 89
4.2.4. Multiple Effect Distillation

Multiple effect distillation has been employed in several forms. A vertical falling film
process was employed at Freeport, Texas, for many years as a part of the US Office of
Saline Water test programme. Vertical tube climbing film evaporators have been tested
and horizontal tube multiple effect evaporators have been employed. All of these types
have a single common design feature—steam is condensed on one side of the heat
transfer surface and water is boiled on the other.
It has been shown in several cases5,6 that the multiple effect cycle is potentially more
economic than the MSF, particularly if the high heat transfer rates, that are theoretically
possible with falling film, spray film and fluted tube systems, can be reliably maintained
in practice. Unfortunately the thin films employed to achieve high heat transfer rates tend
to be unstable. Maldistribution of the film, created either by blockage of the sensitive
water distribution nozzles or by slight scale deposits formed at the sites of imperfections
in the heating surface, creates local points where the tube wall temperature rises and the
rate of scale formation increases. The laminar flow conditions within the film provide
insufficient turbulence to remove deposits and hence accelerating scaling develops. It is
unlikely that the consequent instability of heat transfer performance can be avoided
particularly since the feed sea water is always somewhat variable in quality and contains
suspended matter.
Despite the theoretical potential of multiple effect systems, particularly in combination
with vapour recompression cycles in situations when high performance ratios are called
for, multi-stage flash distillation is likely to remain the only serious contender for the
large-scale sea water evaporator market.
4.2.5. Materials for Evaporators7

The early MSF plants followed normal marine practice, with the use of copper-base
alloys for heat exchanger tubing and, in general, this practice has continued. Thus the
Kuwait MSF plants normally use 70/30 cupronickel tubing for the heat reject, brine
heater and high temperature stages with aluminium brass used elsewhere. Approximately
25% of the tubes are cupro-nickel and the remainder aluminium brass. The cupro-nickel
alloy used in the first plants was the 2% iron 2% manganese version. This practice has
continued and has spread to most other areas, apart from the USA, where standard 70/30
cupro-nickel has been specified. Tube plates for the plants are usually Naval brass or
90/10 cupro-nickel.
In contrast to the above practice for polyphosphate dosed plants, there has been a
tendency to use 90/10 cupro-nickel instead of aluminium brass in acid dosed plant, and
some units have been built with this alloy in all sections—an example of this is Jeddah 1.
Overall, therefore, there has been an increase in usage of 90/10 cupro-nickel at the
expense of aluminium brass. Some data8 show the overall picture of about 65%
aluminium brass in 1971 and about 35% in 1975.
Data from the AD 1972 Little Survey9 shows 90/10 cupro-nickel to perform
significantly better than aluminium brass in acid dosed plants so that this switch to 90/10
cupro-nickel is probably based on an increased usage of acid dosed plants during the
1971–75 period.
There has been no significant usage of titanium for tubing apart from Jubail I (still
under construction). However, as Jubail II has reverted to 90/10 cupro-nickel this does
not seem to represent a breakthrough for titanium. For ejector condensers, where vapour
side conditions are unfavourable to copper-base alloys titanium has often been used, but
as this represents less than 0·5% of the market it does not affect the overall trends.
Aluminium bronze is the usual tube plate material where titanium tubes have been used.
Water-Boxes
Experience has shown that bare carbon steel water-boxes, even in polyphosphate plants,
suffer severe corrosion and most plants now employ protection of some kind, the most
common being 90/10 cupro-nickel (either as solid or clad plate). Some linings of 70/30
cupro-nickel have been used but this is much less common and seems to be declining.
In some cases all water-boxes have been metal lined, but elsewhere only the raw sea
water and high temperature boxes have been protected. Rubber linings are sometimes
used for raw sea water but epoxy and similar coatings which, in general, have performed
poorly are no longer widely used.
Flash Chambers
Following the trend in water-boxes many acid dosed plants are now lined with corrosion-
resistant material. The two commonly used materials are 90/10 cupro-nickel and stainless
steel, usually type 316L. Both appear to perform satisfactorily but experience with
completely lined stainless steel chambers is sparse and time is needed to be sure that this
alloy can withstand the varying conditions and in particular oxygen levels to be expected
in MSF plant. In low oxygen conditions, stainless steel can be expected to perform well;
however, at high oxygen levels pitting and perhaps cracking may occur. 90/10 cupro-
nickel is not prone to these problems but is more expensive than stainless steel, hence the
use of both materials for this application. For smaller units, where the shell thickness is
low, cupro-nickel is the normal choice where an alloy material is specified, as stainless
steel would suffer problems externally.
For large plants clad steel plate is used for economic reasons and this circumvents
external corrosion problems where stainless steel is specified.
For polyphosphate plants alloy linings are not normally used apart from local areas
around weirs and nozzles. Elsewhere deposits of carbonate and hydroxides on the steel
surface seem to markedly reduce corrosion of the underlying metal. This trend towards
the use of alloy lined chambers is probably the most significant change in material usage
in desalination in recent years.
It can in general be concluded that over the last 10 years no new materials have been
employed in evaporators but rather the practice has been to upgrade materials used and
consequently fewer problems tend to be experienced.
4.2.6. Cost of Sea Water Evaporation
Capital Costs
The capital costs of sea water evaporator installations are comprised of three elements:
(1) The evaporators.
Desalination 91
(2) The feed steam system—self-contained boilers or, more commonly, the supply
system from combined power generation plant.
(3) Ancillary works—civil works for the evaporators, post treatment or blending
facilities, storage and distribution works.
Typical costs for high temperature, acid or Belgarde dosed evaporators are shown in Fig.
7. The costs are presented in terms of pounds sterling per unit of plant size and
performance ratio. It can be seen that a typical evaporator of 10000 m3/d capacity with a
performance ratio of 10 will cost about £5·8 million at 1979 prices. Clearly such costs
can only serve as a guide since site conditions and the state of the market as well as
specific design details will influence the precise cost in an actual case. Costs of
polyphosphate dosed plant are higher, as shown in Fig. 5.
FIG. 7. Capital costs MSF evaporators

(acid dosed) 1979.
The civil works directly associated with the evaporator typically add 15% to the cost
of the installation but if post treatment works to artificially harden the distillate prior to
distribution are employed higher additional costs will be incurred.
The total cost of self-contained, low pressure, water-tube boilers installed together
with auxiliary equipment has been taken as £30 per kg/h of steam raising capacity.
Thus typical capital costs for a 10000 m3/d installation are shown in the following
table. For the fuel costs assumed, the performance ratio for a single purpose plant is
optimised at 12 whilst that for corresponding dual plant is 10.
Operating Costs
The major items contributing to the running costs of evaporators are as follows:
Fuel for steam raising
Auxiliary power
Chemicals
Maintenance and operation
Steam supplies. The largest single operating cost is that of fuel for generating feed steam.
This cost is extremely variable, being dependent upon the source of energy. The highest
values are those associated with the raising of steam in self-contained boilers where the
effective cost of the steam is directly related to the world market rate for energy. Such a
case is shown in Table 1 with oil priced at $20/barrel and a boiler efficiency of 85%
assumed.
The lowest values quoted in the literature tend to be based upon the use of free or very
low priced gas which would otherwise be flared off. An authority may wish to set a low
price for such a fuel but this can only be considered as a subsidy and does not reflect a
realistic value. Similarly ‘waste heat’ from incinerators or the exhausts from gas turbines
or diesels cannot be realistically valued at zero cost.
Probably the lowest economic value which can be placed on feed steam is that part of
the costs associated with generation of steam which can be attributed to low pressure use
after expansion from high pressure in the generation of electricity. In such cases the
allocation of fuel costs between electricity and power is somewhat arbitrary. In the
second case quoted in Table 1 50% of steam raising costs have been allocated to the low
pressure steam employed for desalination.
TABLE 1 COSTS OF MSF DISTILLATION
(1979–70% LOAD FACTOR)
Single Dual
purpose purpose
Capital costs (£
million)
Opt. performance ratio (12) (10)
Distiller cost 6·90 5·80
Boiler cost 1·23 1·52
Civil cost 1·03 0·87
Total capital 9·16 8·19
Operating costs (£ £mpa p/m3 £mpa p/m3
3
million p.a and p/m )
1. Fixed charges 1·08 42·3 0·96 37·7
11.75% (10% interest,
20 years)
2. Fuel at $20 per barrel 0·83 32·5 0·50 19·5
(half cost for dual
purpose)
3. Chemicals Polymer 0·09 34 0·09 3·4
Desalination 93
£2000/tonne, 7 mg/litre
4. Auxiliary power (at 0·20 7·9 0·20 7·9
2p/kWh)
5. Maintenance 0·25 10·0 0·25 10·0
operation, staff and
materials
Total annual cost (£m) 2·45 – 2·00 –
Total water cost (p/m3) – 96·1 – 78·5
The variation in fuel price for the two cases would lead to the selection of difference
performance ratios. With a 10% interest rate and repayment over the 20 years life of the
plant and an average load factor of 70% of the optimum performance ratios become 12
for the single purpose plant and 10 for the dual purpose plant. Capital costs and capital
charge rates quoted reflect these different efficiencies.
Auxiliary power. Major power consumption includes the pumping of recycle brine, the
running of distillate, blowdown and condensate pumps and site instrumentation and
services. This, for a 10000 m3/d plant will approximate to 1600 kW, being slightly lower
for the lower performance ratio plants.
Chemicals. Prices for chemicals for scale control are extremely variable, dependent
upon the location of the works. Typical costs quoted by Wade10 are:
Polyphosphates £500/tonne
Polymer £2000/tonne
Acid £120/tonne
With typical doses of 7 mg/litre for polymer and 120 mg/litre for acid there is little
between the costs of these two chemicals for high temperature operation.
The polymer cost quoted in the table of operating costs is based upon the use of a
feedwater rate 2·4 times that of the product, thereby limiting the concentration of the
brine blowdown to 1·7 times that of the feed sea water.
Maintenance and operation. The annual cost of maintenance typically contributes
some 1·5% of the capital cost of the distiller whilst operation is approximately double this
value for a plant of 10000 m3/d. Thus the combined cost contributes about 10 p/m3 to the
cost of water.
Water costs. It can be seen that for the conditions assumed total water costs range
from 78·5 p/m3 for the dual purpose works to 96·5 p/m3 for the single purpose case. The
selection of steam turbine drives for major pumps for single purpose plant can save up to
about 4 p/m3 on the cost quoted by raising steam at an adequate pressure to drive the
turbines before being employed as feed to the brine heater.
4.3. REVERSE OSMOSIS
Reverse osmosis is a pressure driven, membrane desalination process which has

undergone the most rapid development of any desalination technique. To some extent this
has been due to substantial funding of research and development by the United States and
other international governments. The process, also known as hyper filtration, has now
been commercially available for the treatment of brackish waters for approximately 10
years. During this time membranes have been improved and developed to achieve
excellent performances in brackish water treatment and finally to be capable of desalting
water of up to sea salinity. Sea water systems have been commercially available to a
limited extent for the past three years but research and development is continuing in this
area. Engineering developments have occurred in the packaging of membrane and the
overall system designs but these have basically been refinements to the original designs.
The reverse osmosis process operates at ambient temperature by sufficient pressure
being applied to a saline water to overcome the osmotic pressure and to force fresh water
through a thin semipermeable membrane at a realistic rate (Fig. 8). The membranes are
ideally permeable only to water but in practice a small amount of salt transfer also occurs.
There are three basic types of membrane in use:
(1) Forms of cellulose acetate (CA).
(2) Polyamide (PA).
(3) Composite membranes employing a thin film of polyamide or similar desalting layer
formed upon a porous substrate of a material such as polysulphone (TFC).
FIG. 8. Basic principles of reverse

osmosis.
Desalination 95
The CA and TFC membranes have normally been employed in the form of a flat sheet
and the polyamide as thin hollow fibres. However, there is one hollow fibre system with
cellulose triacetate membranes.
The osmotic pressure of brackish waters would typically be in the range 140 to 350
kN/m2 (20 to 50 psi) and that of sea water 2450 kN/m2 (350 psi). In practice it is
necessary to apply a sufficient pressure above the osmotic pressure to provide satisfactory
product water rates. Hence operating pressures for brackish water systems are typically
2800 kN/m2 (400 psi) but can vary through the range 1400–4200 kN/m2 (200–600 psi)
dependent on the application. Pressures of 5600–7000 kN/m2 (800–1000 psi) are required
for sea water. The total salt rejection in brackish cases can be controlled to some extent
by the type of membrane applied but typically 95% plus would be expected in a single
pass system. For sea water, single pass systems can be employed using membranes which
give 99% salt rejection; however, lower rejection membranes have been employed in a
two pass system in which the product from the first pass is used as feed for the second
pass.
The total world installed capacity for brackish water RO systems can be estimated as
1550 Ml/d. This is based on the 1977 Desalting Plants Inventory of the US Department of
the Interior11 with allowance being made for small plant not included in that inventory
and the growth in the market to June 1979. The largest installed plant at this time is at
Salbukh in Saudi Arabia and has a capacity of 46 Ml/d. The largest installation proposed
is the 360 Ml/d plant for Yuma in Arizona, USA. This is to desalt saline drainage water
from the Wellton Mohawk irrigation and drainage area before it discharges into the
Colorado River. The scheme is to maintain a satisfactory quality in the Colorado before it
passes to Mexico.
The total world installed capacity of sea water RO systems to June 1979 is
approximately 19 Ml/d. The largest of these is a 12·5 Ml/d plant in Jeddah in Saudi
Arabia. However, a number of plants are under construction and the largest of these is a
13 Ml/d plant for treating the Caspian Sea in Russia. Within a year it is expected that the
total installed capacity will be approximately 45 Ml/d.
Ultrafiltration (UF) is also a pressure-driven membrane process worth noting, although
it will not be discussed in any detail in this chapter as it has had very little application in
the water treatment field. It is an extension and development of reverse osmosis using
more open high flux membranes (4–8m3/m2/d) which will only reject larger molecules,
and is applied to solutions with dissolved solutes with molecular weights above 1000.
Dissolved inorganic salts and small organic molecules will pass through the membrane
with the permeate fluid. The osmotic pressure for these systems is thus very small and
operating pressures are usually between 175 and 700 kN/m2 (25–100 psi). Its main
potential as a water treatment process is for the removal of colloids and high molecular
weight dissolved organic substances in the feed water to reverse osmosis or ion exchange
plant. In addition it has been applied in a few instances for the direct removal of colour,
i.e. fulvic and humic acids, from water. This may be a potential area of increased
application but unless costs are significantly reduced it will only be used where
performance is more important than cost. It is estimated that UF applications in Europe
may have a total installed membrane area of 25000 m2 mainly in the electropaint and
dairy industry. No doubt the installed area in the USA would be significantly higher than
this.
4.3.1. Membranes
The first commercial reverse osmosis desalination membrane was an asymmetric
cellulose acetate membrane which emerged from the work of Loeb and co-workers in
1962. Although it was anticipated that cellulose acetate would be rapidly superseded, the
‘Permasep’ polyamide hollow fibre of the DuPont Company, Plastic Products and Resins
Department, released in 1970, was the only real commercial alternative to cellulose
acetate or modified forms of it until 1977. At this time the first commercial thin film
composite membranes became available from Universal Oil Products (UOP), Fluid
Systems Division, following their CA membrane.
Both the cellulose acetate and polyamide membranes are asymmetric. This means that
the membrane is formed in a one stage process with an outer, thin dense active layer,
typically 0·1 to 1 µm thick for desalination, supported by a thicker porous layer of the
same material. The cellulose acetate membrane is cast onto a supporting fabric to provide
mechanical strength for handling. The polyamide hollow fibres of DuPont are no thicker
than a human hair and are self-supporting.
The thin film composite membranes consist of a very thin active layer of polymer
deposited onto a porous support of a different polymer and thus fabrication is a two stage
process. The construction details of all three membrane types are shown in Fig. 9.
The fabrication advantages of composite membranes include the independent selection
of polymers from which to optimise the properties and fabrication of each layer, plus the
ability to vary the layer thicknesses for different applications. They are dry processed and
can be wet-dry cycled with no effect on membrane performance. Hydranautics Water
Systems offer a form of this property with dry ship, dry store for their cellulose acetate
membranes.
FIG. 9. Membrane configurations.

Typical performance characteristics of the various membrane types are detailed in
Table 2.
It can be seen that the membranes have pH and temperature limitations which at times
can restrict the direct application of the process. CA membranes are subject to hydrolysis
Desalination 97
but within the range specified this is not excessive. Longer membrane life can be
achieved by operating between
TABLE 2 TYPICAL CHARACTRISTICS OF
OSMOSIS MEMBRANES (2800 kN/m2 (400 psi)
FOR BRACKISH WATER AND 6000 kN/m2 (850
psi) FOR SEA WATER)
Membrane pH Max. Inorganic Flat Allowable
type range Temp. salts sheet free Cl in
(°C) rejection flux feed
(%) (m3/m2d) (mg/litre)
Cellulose
acetate
(a) 3–8 40 98 0·73 0·5 above
Brackish 25°C
(several 95 0·85 1·0
types) below25°C
(b) Sea 3–8 35 98 0·6 0·5 above
25°C
1·0 below
25°C
Polyamide
hollow
fibre
(a) 4–11 35 95 0·1
Brackish
(b) Sea 5–9 35 98·5 0·1
TFC
(a) 2–12 45 98 0·7 Nil
Brackish
(b) Sea 2–12 45 98·5 0·7 Nil
Note: These data refer to plant performance of 75%,
recovery for brackish water and 30% recovery for sea
water.
pH 5 and 6 where minimum hydrolysis occurs, although the rate does increase with
higher operating temperatures. For hollow fibre polyamide membranes there is no
membrane life or performance advantage in operating at any particular pH within the
allowable range. TFC membranes can be seen to have the widest range of pH resistance.
Again there is no membrane advantage in operating at any particular pH within the range
but optimum rejection appears to occur between pH 5 and 7 for monovalent salts. The
rejection of divalent salts is largely independent of pH.
All membranes have temperature limitations with the TFC offering the most
flexibility. Owing to the upper limits it is necessary in the Middle East where
groundwaters can emerge at 50–55°C to cool the water before treatment. In addition the
temperature of the feedwater affects the productivity of the membranes although it has
little effect on product quality. Membrane product flux at 25°C is rated as 100% and over
the operating temperature range the flux varies almost linearly by approximately 2·8%
per degree Centigrade. This does vary marginally for the different types of membrane but
could be taken as a realistic average. The flux decline over the membrane life also varies
with operating temperature. The higher the operating temperature the greater the decline,
although the major part of the decline would occur in the first year of operation, as shown
in Fig. 10 for DuPont hollow fibre membranes. This occurs for all membrane types.
The overall salt rejection achieved by the membranes for brackish and sea waters can
be in excess of 98%. However this is not always necessary for brackishwater treatment of
up to 5000 mg/litreTotal Dissolved Solids (TDS) and membranes with a 95% rejection
are adequate. However the higher the
FIG. 10. Flux decline in hollow fibre

membranes (by courtesy of DuPont).
rejection the more raw water that can be blended to achieve a 500 mg/litre TDS potable
supply. For single pass sea water desalination with a product requirement of 500 mg/litre
TDS or less, a rejection of greater than 98% is necessary. Partial second stages will be
necessary with some membranes and for some high salinity sea waters.
Typical performance data from cellulose acetate membranes operating at 2800 kN/m2
(400 psi) and 50% recovery on a brackish water are shown in Table 3.
Desalination 99
TABLE 3 RO TREATMENT OF BRACKISH

WATER AT YUMA—USA BY SPIRAL WOUND
CELLULOSE ACETATE MEMBRANES12
Feed Product
(mg/litre) (mg/litre)
TDS 2900 85·0
Ca 113 0·87
Mg 70 0·54
Na 836 31·0
K 8 0·15
Total Fe <0·06 <0·06
Sr 2 0·2
Total Alk as 9 6·5
CaCO3
Cl 1036 44·0
SO4 826 2·0
pH 5·5 5·2
In addition it is known that such a membrane would have approximately an 86% rejection
of nitrate ands would largely reject organic compounds with a molecular weight down to
approximately 200. Hollow fibre membranes would have very similar rejection
capabilities. TFC membranes operating on a brackish water would also provide similar
rejections except that improved nitrate and organics rejection would be expected.13 A
rejection of up to 99% could be obtained for nitrate. The improved organics rejection
could be significant if chlorinated and other low molecular weight organic compounds
must be removed from drinking water. However more data need to be generated to
determine the full potential of organics removal with TFC membranes. It may also be
possible to tailor the membrane properties to reject different types of organic compounds.
RO membranes can also almost completely reject aluminium, heavy metals, bacteria and
pyrogens and hence overall a high quality water can be produced.
TABLE 4 RO TREATMENT OF SEAWATER—
JEDDAH, SAUDI ARABIA, WITH SPIRAL
WOUND TFC MEMBRANES14
Feed Product
Temp. (°C) 32 max.
PH 8 7.5
TDS 41200 ~800
Ca 520 20
Mg 1460 ?
HCO3(asCaCO3) 125 65
Cl 22000 600
SO4 2960 20
SiO2 0·4
Total Fe 0·2
For sea water treatment with systems operating at approximately 7000 kN/m2 (1000 psi)
and 30% recovery typical data are shown in Tables 4 and 5 and for single pass operation.
The flux per unit membrane area from flat sheet membrane can be seen in Table 2 to
be in the region of 0·5–0·75 m3/m2d for brackish water at 2800 kN/m2 (400 psi) or sea
water at 5600 kN/m2 (800 psi). However, this cannot be compared directly with hollow
fibre membrane which has much lower unit fluxes but much greater surface area to
achieve the required
TABLE 5 RO TREATMENT OF SEAWATER—
SOUTH CAICOS ISLAND, BRITISH WEST
INDIES, WITH POLYAMIDE HOLLOW FIBRE
Feed Product
pH 7·4 6·5
TDS 42000 435
Ca 600 8
Mg 1118 1
Na 16000 133
HCO3 (as 190 20
CaCO3)
Cl 25000 210
SO4 2900 1
Fe 3·8 0·2
output. The fluxes quoted are for new membrane but these do decrease over the
membrane life as seen in Fig. 10. This decrease is due to compaction or densification of
the polymer support layer which occurs from operation at high pressures. All membranes
are subject to compaction but this is relatively low below a 2800 kN/m2 (400 psi)
operating pressure. The TFC membranes have minimal compaction since it is possible to
optimise the support layer during fabrication. Significant progress is being made in the
development of TFC membranes which have the above flux and rejection characteristics
for brackish waters but which will operate at 1400 kN/m2 (200 psi). A membrane of this
type should be commercially available in the near future and ultimately similar benefits
are likely to be achieved for sea water membranes.
Biological activity can develop in RO modules and to prevent attack on cellulose
acetate membranes a free chlorine residual in the feed water is often recommended. The
chlorine also controls this type of activity in the other components of the overall plant.
The hollow fibre and TFC membranes are non-biodegradable and hence do not require
this form of protection. In fact the polyamide or similar polymers are very sensitive to
chlorine and can be readily oxidised. Hence very low or zero chlorine must be present in
their feedwaters. However, while non-cellulosic chlorine resistant membranes with their
wider range of properties are highly desirable, the sensitivity of the present membranes
Desalination 101
should not technically limit their applications. In fact polyamide hollow fibres have been
widely applied and for chlorinated feedwaters a dechlorination stage is readily included.
For a relatively new process such as RO the membrane life is a key factor in the
operating costs. For brackish waters most major manufacturers now have sufficient
experience and confidence to offer a three-year guarantee provided the feedwater
specifications have been met. In many cases they would expect the life to be substantially
longer. The more recent sea water and TFC membranes have yet to be fully proven in
large installations but again manufacturers are confident of a minimum three-year life.
4.3.2. Reverse Osmosis Modules

Not only must reverse osmosis membranes have satisfactory desalination properties but
they must be produced in such a form as to be capable of withstanding the high forces
created under operating conditions. Whilst the fine hollow fibres are mechanically self-
supporting all other types of membrane have to be provided with some form of porous
support. There are five different module designs which have been considered seriously,
namely tubular, plate and frame, spiral wound, hollow fibre and the rod type.
One form of the tubular module concept typically consists of 13 mm internal diameter
porous tubes up to 3660 mm long, connected in series in a repeating ‘S’ configuration
with up to 18 tubes per module. These tubes are the membrane support and must be
capable of withstanding the operating pressure of the system. Cellulose acetate type
membranes have been generally used and can be cast directly onto the inner surface of
the tube or more usually the membrane is cast onto the inner surface of a synthetic paper
tube. The paper membrane tube is then inserted into the support tube. The advantage of
this approach is that the modules can be very simply and economically remembraned
where necessary. Saline water under pressure is pumped at the required velocity through
these tubes in the modules. Fresh water passes through the membrane and porous support
and is collected by an outer shroud surrounding the tube bundle. Tubular systems have
well defined open flow channels and are very tolerant towards suspended solids and are
difficult to block. They are particularly amenable to cleaning by water flushing and
chemicals, and, when necessary, can be mechanically cleaned by foam swabs without
dismantling the equipment. However, the packaging of large areas of membrane in this
way is not economically competitive with the alternative spiral wound and hollow fibre
systems, and therefore tubular membranes are not often employed for municipal water
treatment. They do however find extensive use for special applications in the dairy
industry and are offered by Paterson Candy International Ltd in the UK. The plate and
frame concept, similar to a filter press, is a viable approach technically but again is not
applied in the water field as it also is not competitive economically.
The ‘rod’ system was developed by the United Kingdom Atomic Energy Authority,
Harwell,16 and in terms of packed membrane area per unit volume of module it falls
between the tubular and the higher packed density of the spiral wound and hollow fibre
configurations. In its latest concept the rod system is based upon a 3 mm diameter
microporous polypropylene tube onto which the membrane can be cast directly and
continuously. Bundles of 37 or 109 of these rods, 4 or more metres long with one end
sealed and the other mounted in a header block, are placed in tubular pressure vessels.
The feed liquid is pumped through the pressure vessel at an appropriate velocity and
pressure on the outside of the rods. Permeate passes through the membrane and porous
tube and is collected from the header block end. Membrane bundles can be quickly and
easily replaced and are disposable. However, the porous support tubes are not suitable at
this time for operating pressures greater than 600 psi which would limit the system to
brackish water applications.
The system is more tolerant to suspended solids than spiral wound and hollow fibre
modules but less tolerant than tubular designs. It also has the advantage that in the event
of severe malfunction the membrane rod bundle can very easily be removed for some
form of physical cleaning. However, although the system has significant potential it is
considered to be not commercially competitive for municipal water treatment by Paterson
Candy International Ltd who hold the marketing rights for the process. Hence at the
present time it has limited application.
The two most extensively used module types, the spiral wound and the hollow fibre,
differ significantly in designs. The spiral wound element was developed by General
Atomic of the Gulf Oil Corporation, now the Fluid Systems Division of Universal Oil
Products (UOP) Inc., under US Government sponsorship. It has since been adopted by
other US companies, typically Hydranautics and Envirogenics, as their basic
configuration. The membrane element usually consists of several membrane leaves
attached to a central product carrying tube. A basic leaf is shown in Fig. 11.
FIG. 11. Spiral wound membrane

module.
Two flat membrane sheets with their desalting skin facing outwards are separated by a
very fine, porous, polymer sheet less than 1 mm thick which supports the membranes
under pressure. Three sides of the ‘envelope’ are sealed with a suitable glue and the
fourth side is glued to a porous section of the central tube. A more open porous mesh
approximately 1 mm thick is placed on the outside of the membranes. Then this leaf and
others attached in the same way are rolled around the centre tube. The membrane
elements are usually 1 m long with diameters of 4 in or 8 in being generally used
although UOP can supply 12 in diameter elements. An 8 in diameter element 1 m long
would have an effective membrane area of approximately 320 ft2. Up to six elements can
be inserted in series into each fibreglass pressure vessel. The feedwater at pressure passes
Desalination 103
through the porous mesh channels parallel to the membrane and the fine mesh supporting
the membrane is the product water carrier to the centre tube.
The output capacity for 6×8 in diameter membrane elements, 1 m long, operating at
2800 kN/m2 (400 psi), 25°C and a total 60% recovery treating a brackish water would be
approximately 185 m3/day of permeate. The maximum back pressure from the product
water which this type of module can safely withstand is 120 kN/m2 (14 psi).
The Permasep polyamide, hollow fibre module was developed by the DuPont
Company and consists of millions of the hair-like fibres bundled together. These are
wound around a plastic mesh material approximately 1 mm thick and 450 mm long which
in turn is wound around a central porous tube. This continues until the overall diameter of
the fibre bundle is 4 in or 8 in, whichever is required. This is then wrapped with a fine
cloth to keep the fibres in place. Each end of the fibre bundle is then encapsulated in
epoxy resin. One of these ends is machined back to cut off one end of the fibre bundle to
leave fibres in a ‘U’ shape and open at the two exposed ends. This tube bundle is then
inserted into a fibreglass pressure vessel as shown in Fig. 12. End plates with ‘O’ ring
seals are fitted and on the product end a porous support block is also fitted. These are all
held in place by circlips. Saline feedwater at pressure is pumped into the central
distributor tube where it passes out radially between the fibres pressurising them from the
FIG. 12. Permasep permeator.

outside. The module basically operates under laminar flow conditions. Fresh water passes
through the fibre wall into the hollow centre where it moves towards the open ends of the
fibres. It is collected by the porous block and is removed via a central tube. The brine
outlet removes the remaining concentrate. Hence in this case each hollow fibre
membrane bundle has its own pressure vessel. The output from an 8 in diameter module
operating at 2800 kN/m2 (400 psi), 25°C and 75% conversion, treating a brackish water,
would be approximately 68 m3/day.
4.3.3. Pretreatment of Feedwater

For reverse osmosis equipment to operate satisfactorily there must of course be an
adequate engineering design. Accepting that, the most important operational factor is to
keep the membranes in a functional condition and therefore they must be protected from
fouling constituents in the feedwater. Adequate pre-treatment is thus essential to limit the
concentration of potential foulants in the feedwater. In addition the water must be treated
to satisfy other membrane limitations: pH control for cellulose acetate membranes and
dechlorination for polyamide hollow fibre and composite membranes may be necessary.
There are several types of fouling which can occur and some of these are important to
greater or lesser extents in the various membrane module configurations. Membrane
fouling from calcium carbonate and/or sulphate scale is a common consideration for all.
These constituents are present in the feedwater and, depending on their concentration, can
be concentrated by a factor of two to five times depending upon the plant conversion. If
the solubility products are exceeded fouling will occur owing to precipitation. Calcium
carbonate and calcium sulphate are the most common salts which can cause problems but
silica, strontium sulphate, barium sulphate and calcium fluoride can also cause scaling.
Scaling can be controlled by operating the systems at low conversions so that the
solubilities of the problem salts are not exceeded. However, this is usually not practical
for large plant and the scaling can be controlled by pretreatment to remove one of the
ions of the scale-forming compounds or by chemical dosing to control salt precipitation.
Softening, either chemically with lime/soda or by ion exchange, will largely remove
calcium and other divalent scale-forming cations. However this is not always viable
unless high conversions are required. To prevent calcium carbonate scale formation the
Langelier Index of the RO concentrate must be negative. To prevent calcium sulphate
scale without further treatment, the ionic product of the calcium and sulphate ions should
not exceed the solubility product Ksp for calcium sulphate. In pure water at 25°C this is
Ksp=2·4×10–4. However the solubility of calcium sulphate increases with ionic strength of
the solution and therefore this should be taken into account when determining the
maximum allowable calcium concentration in the RO concentrate for the particular plant
conversion.
It may be preferable to control calcium carbonate scale by acid dosing to achieve a
negative Langelier Index which would occur in the pH range 5–6. In these circumstances
calcium sulphate scale would be controlled by dosing a scale inhibitor such as sodium
hexa-metaphosphate. Doses of 10 mg/litre have allowed the concentrations of calcium
and sulphate ions in the concentrate to be raised to such levels that their product is as
high as 10–3 mol2/litre2 without calcium sulphate scaling becoming a problem.
It is important to achieve the correct feedwater quality and operating conditions in
order to avoid calcium sulphate scale formation as, once deposited, it is very difficult to
remove.
Problems due to silica scaling from the concentrate are also common to all module
types. Its solubility in pure water at 25°C is approximately 105 mg/litre as SiO2. This is
an increasing linear function with increasing temperature. It would appear that the silica
solubility is not affected by the solution ionic strength but it is dependent upon pH.
Solutions can become supersaturated with silica whereupon it can polymerise to form
colloidal silica or precipitate as calcium silicate if calcium is present. However, silica
scale does not precipitate rapidly and operating experience indicates that levels of 150
possibly up to 200 mg/litre can be tolerated in a continuously operating plant. If the plant
is shut down then the concentrate must be flushed from the RO modules to remove the
scale-forming, supersaturated silica solution. Silica can be removed from feedwaters by
Desalination 105
being adsorbed on precipitated magnesium hydroxide in the cold lime or cold lime-soda
softening process.
Fouling of RO membranes can occur by the oxidation of soluble species in the
feedwater which if not removed can then precipitate as insoluble hydroxides on the
membrane. Iron and manganese fouling can occur in this way. This type of fouling can
normally be avoided by removing the oxidisable species from the water or by keeping the
feedwater out of contact with air and so avoiding the oxidising conditions. Iron fouling is
the most common and this can be removed from groundwaters after oxidation by sand
filters before passing to the RO modules. Fouling could also occur from well oxidised
surface waters containing precipitated ferric hydroxide. Hence in this case the iron would
also be removed by filtration or by two stage treatment employing both clarification and
filtration if necessary. If there is no oxygen in the feedwater and oxygen can be excluded,
levels of 3–4 mg/litre of iron can be tolerated for hollow fibre membranes. Otherwise the
level of iron should be kept below 0·05 mg/litre in the RO feed. Cellulose acetate spiral
wound membranes are slightly more tolerant with levels of up to 5 mg/litre and 0·1
mg/litre being acceptable for oxygen-free and oxidising conditions respectively.
Hydrogen sulphide which is frequently found in groundwaters may also cause problems
if oxidised. Very fine colloidal sulphur could then irreversibly foul the membranes.
Hence if possible the feedwater should be pumped through the system without exposure
to air with the hydrogen sulphide then being removed from the permeate by degassing.
Cellulose acetate membranes are susceptible to biological fouling and attack and
hence to prevent this the membranes have to be frequently disinfected in a cleaning
sequence or protected by continuously dosing up to 0·5 mg/litre free chlorine to the
feedwater. The hollow fibre and TFC membranes however are completely free from
bacterial attack and hence no disinfection of the feedwater is required. These membranes
in fact can be readily oxidised by chlorine and hence any chlorine must be completely
removed from the feedwater. This would normally be done by passing through an
activated carbon bed where it would be absorbed or by dosing with sulphur dioxide.
A very important fouling consideration, particularly for hollow fibre membranes, is
the potential plugging of the membrane by coagulated ultrafine colloidal particles. Even
though a feedwater may have been clarified and sand filtered it may still contain
significant material with a size range of approximately 0·2 to 1·0 µm which can cause
severe problems. These particles are usually hydrophobic and may be inorganic or
organic in nature.
An empirical fouling index test has been devised by DuPont17 to determine whether
the colloidal content of the water is likely to cause a fouling problem for their hollow
fibre membranes. Variations of the test have been used by other manufacturers. Turbidity
measurements cannot indicate the presence of colloidal material. DuPont determine what
they call a ‘Silting Density Index’ (SDI) from the following equation when the feedwater
is passed through a 47 mm diameter 0·45 µm filter at 30 psig.
where ti=initial time (s) to obtain a sample of say 100 ml; tf=time (s) required to obtain a
100 ml sample after time T (min) of operation; and T =time (min) total test time, usually
15 min.
The colloidal content is considered to be low enough to avoid fouling problems if an
SDI of three or less is obtained. Groundwaters usually fall into this category. However
surface waters even after treatment can have an SDI in the range of 10 to over 100 which
can lead to severe fouling problems. Spiral wound modules are considered to be much
more tolerant to suspended solids and colloidal content in the feedwater. At the present
time a turbidity of less than 1·0 FTU is considered to be adequate to feed spiral wound
modules. However it is possible that the new TFC membranes with their polyamide thin
film desalting layer may need to have a feedwater which conforms to an SDI not
significantly greater than that required for polyamide hollow fibre membranes.
Fouling by colloids can be reduced by either removing them from the feedwater or by
increasing their stability. Coagulation and filtration techniques can be used to reduce the
SDI of a water. DuPont suggest that for an SDI above 50 either aluminium or ferric
sulphate coagulants can be used with sedimentation followed by rapid gravity filtration to
remove colloidal material. However there are cases where even this treatment is
insufficient and further filtration with polyelectrolyte addition is necessary. The
technique of dosing with polyelectrolyte followed by pressure filtration is the
recommended method of reducing the SDI to below three for feedwaters which have an
initial value below 50. It should be pointed out that clarification by dissolved air flotation
rather than by sedimentation has considerable potential when two stage treatment is
necessary. Flotation is a much higher rate process than sedimentation and therefore in a
more compact plant can be employed. In addition it is a process where solids are removed
by being floated to the surface by micro bubbles of air. This should be beneficial in
removing material such as colloids which typically will not settle.
DuPont also suggest that measurement of the zeta potential of a solution is an effective
guide to colloid stability. If, for example, the zeta potential can be increased from −10
mV to −30 mV colloidal fouling will be significantly reduced. The zeta potential can be
approximately doubled by softening to below 5 mg/litre as CaCO3 which increases the
stability of the colloids and reduces their tendency to coagulate. Thus if larger particles
are not formed they will not foul the membrane and the small colloidal particles will be
purged from the system with concentrate.
Fouling of membranes can also occur from severe organic contamination in the
feedwater. This is largely reversible for cellulose acetate membranes and to a lesser
degree for non-cellulosic membranes. If necessary some organics can be removed by
clarification and filtration but usually routine cleaning can cope with the problem. If
absolutely necessary activated carbon could be considered as a further pretreatment stage
following filtration.
Even though membranes may be fouled during operation, in many cases it is possible
to remove this with suitable cleaning techniques. Hydrated metal oxides can be removed
by types of ammoniated citric acid solution and calcium carbonate removed by a solution
containing weak hydrochloric acid both at pH 4. Calcium sulphate can be removed in part
by an ammoniated citric acid, or EDTA solution at pH 8 and organics can be removed by
detergent at pH 8 or higher depending upon the membrane. Inorganic colloids and
silicates can be removed by detergent or caustic solutions also at pH 8 or higher. Bacteria
Desalination 107
can be removed by a mixed detergent, formaldehyde solution. It is sound practice to build

routine cleaning cycles into the operation of any plant and the type of cleaning may vary
depending upon the feedwater quality.
4.3.4. Reverse Osmosis Systems Design

The basic individual components in reverse osmosis systems are all relatively simple
apart from the sophistication of the associated electronic controls. Ignoring pre- and post-
treatment considerations, a simple system consists of a 5–25 µm pre-filter to protect the
high pressure pump and module, the high pressure pump itself, the single desalting
membrane element in its pressure vessel and the reject flow control valve. In systems of
this size design considerations are limited but they become more important with
increasing size. The important factors will be briefly discussed for larger systems.
The micron filters are usually of the cartridge type where up to fifty cartridge filter
elements may be contained in the one pressure shell. On a large plant there may be tens
of these pressure shells to provide the capacity for the plant. The cartridge particle size
cut-off rating will vary depending upon the type of membrane element being used. For
hollow fibre permeators which are more sensitive to suspended material ratings of 5–10
µm are normal, while for the spiral wound 25 µm is common. With satisfactory pre-
treatment these filter elements have an extended life and hence filter changes are not too
frequent. However when they do require changing this operation is somewhat labour
intensive. Interest is currently being shown in continuous screen filters.18 These
potentially can provide a particle size cut-off down to 5 µm, can be automatically
backwashed on pressure drop or time and can pass large volumes per single unit. These
continuous filters with coarser screens might also replace the pre-treatment sand filters.
Several small units are being evaluated.
The high pressure RO feed pumps require sufficient feedwater at a small positive
pressure to operate properly. A low pressure cut-out control is normal to guard against
any deficiency. The selection of these high pressure pumps also depends upon the size
and duty of the plant. Typically for small plants, single stage, stainless steel, high speed
centrifugal pumps are used for pressures in the range 2800–4200 kN/m2 (400–600 psi).
They can develop up to 6300–7000 kN/m2 (900–1000 psi) if necessary. However the
efficiency of these pumps is only 40–50% resulting in a higher power consumption than
for more efficient pumps. For larger installations multistage turbine or centrifugal pumps
can be used to develop up to 4550 kN/m2 (650 psi) and these have higher efficiencies of
up to 75%. Although reciprocating pumps can have an efficiency in excess of 90% they
are not particularly popular for brackish water applications as they generally have a
higher maintenance cost. However for large sea water applications the reciprocating
pump has the advantage of being able to readily develop the required pressures whereas
the others cannot and hence it would usually be selected. An accumulator should be used
with reciprocating pumps to dampen out pulsations. In choosing a suitable high pressure
pump it is important to select one with a performance specification as close as possible to
the required duty. Pumps which have a higher pressure and volume specification than
required, and are merely throttled back, consume unnecessary energy. This also applies to
reciprocating pumps as although they cannot be throttled directly water can be
recirculated through the pump. In addition in determining the energy required to drive the
pump the normal viscosity of the water should be taken into account so as not to
unnecessarily overrate the motor. Both brackish and sea water pumps would typically
have materials of construction of stainless steel or aluminium bronze.
With the increasing need to conserve energy and reduce costs more consideration is
currently being given to energy recovery from high pressure brine, particularly on sea
water systems operating at 7000 kN/m2 (1000 psi) and only 30% conversion. Some pump
manufacturers are proposing an energy recovery turbine coupled directly to a double
ended shaft of a motor that drives the high pressure feed pump. In this way the net
electrical power input required by the motor to pressurise the water can be reduced by
possibly over 50% for sea water plants.
An important aspect in designing reverse osmosis systems is to optimise the
conversion to product water without creating scaling and concentration polarisation
conditions. Adequate pre-treatment allows higher conversions to be obtained without
scaling but there is a limit as 100% conversion is not possible. Concentration polarisation
is the condition where a much higher salt concentration is created at the membrane
surface than in the bulk stream. This significantly increases the osmotic pressure of the
system and decreases performance. It is caused by the removal of fresh water through the
membrane, increasing the surface salt concentration and reducing the flow to the point
where the velocity at the membrane surface is too low to cause adequate mixing between
the concentrated layer and the bulk flow. There is then also the possibility of precipitation
from the concentrated layer causing scale formation on the membrane surface, even
though the bulk flow may be unsaturated. To avoid these conditions on a large plant, the
pressure vessels are connected in arrays, with the concentrate from a bank of modules
operating in parallel being passed as feed to secondary banks containing fewer modules.
Thus the required minimum velocity conditions can be maintained in the membrane
elements and up to 90% conversion can be achieved. Popular arrangements are 2, 1 or 8,
4, 2 arrays as shown in Fig. 13.
FIG. 13. RO arrays.

RO systems are now more commonly designed to operate at variable pressures in order to
maintain a constant product output. However the systems can readily be operated at
Desalination 109
constant pressure. For constant output operation the systems are commonly designed with
automatic valves to control the feedwater and reject water flow rates.
Care should also be given to selecting suitable corrosion resistant materials of
construction for all process items and pipe work. As far as possible to minimise corrosion
it would be recommended that non-metal components be used whenever they are
practical and economical. This is not however possible for components exposed to high
pressures. Stainless steel is commonly employed for both feed and reject pipework.
Owing to the high concentrations of chlorides commonly found in these streams, care is
required in the selection of appropriate grades of stainless steel and the standards of
welding; pickling of weld areas with passivating solutions being often necessary. If
corrosion occurs not only is there an eventual failure of the component but, if it is prior to
the membrane elements, the corrosion products can foul the membranes.
In typical designs instrumentation is provided to monitor temperature, pressure, flow,
pH, conductivity and other parameters to ensure correct operation and control.
Instrumentation related to critical parameters is connected to alarms to allow abnormal
conditions to be promptly observed and remedied. The controls and indicating
instrumentation can be mounted on the equipment or in a local control panel.
The post-treatment of the product water can be an integral part of a system design. In
addition, since the product is usually of a higher quality than necessary, it is blended with
raw water to provide a higher volume with an acceptable quality. The post-treatment may
consist of degassing the product of some carbon dioxide produced from acid dosing the
feed and destroying the alkalinity. This degassed water would pass to a clear well where
it could be blended with raw water. The blended water would then have the pH adjusted
to approximately 8 by lime or soda ash dosing and be chlorinated before being pumped
into a distribution system. Some alkalinity may be formed from the lime and residual
carbon dioxide in the blended water which would help control corrosion in a distribution
system.
The last very important consideration in an RO system design is the disposal of the
brine concentrate which would normally vary from 10 to 25% of the total flow for
brackish water and up to 70% for sea water. In some circumstances disposal can be
simple by returning the brine to the sea. However in other situations it may be necessary
to collect the brine in ground level ponds and in hot climates evaporate it away. When
this occurs there is often concern that the brine may percolate through the ground strata
and contaminate any groundwater below. When significant calcium sulphate is present in
the brine the danger of this occurring is minimal as the calcium sulphate will precipitate
out on the floor of the pond in a very hard impervious layer. However, to alleviate all
fears of this, ponds with rubber liners have been proposed. When considering the costs of
the process for a specific case the disposal costs for the brine and any pre-treatment
residues should be included.
4.3.5. Reverse Osmosis Plant

As with any new technology, large process plants are not built until smaller plants have
been installed and the process performance and reliability established. Brackish water RO
has passed through this phase and large plants are now being ordered and installed,
mainly in response to the water demand in the Middle East.
The largest operating RO plant at the present time is that installed by Degremont at
Salbukh in Saudi Arabia. This uses DuPont hollow fibre modules to produce 46 Ml/d
from a 1500 mg/litre TDS brackish water. This has cooling, softening with silica
removal, filtration and acid and polyphosphate dosing as a pre-treatment before the RO
units. A marginally smaller plant of 45 Ml/d on a similar water has been installed at
Buwayb in Saudi Arabia by Ames Crosta Babcock and is currently being commissioned.
This plant uses spiral wound modules from UOP. There are four other RO plants so far in
the Riyadh water treatment complex with an average capacity of 32 Ml/d each which are
due to be commissioned in 1980.19
Hydranautics has several 15 Ml/d plants and Paterson Candy International Ltd,
Envirogenics and other companies have numbers of smaller plants installed on brackish
water in the Middle East. A typical RO plant would consist of multiple blocks of pressure
vessels with membrane elements to make up the required capacity plus pumps and
controls, as shown in Fig. 14.
The largest RO plant yet considered has been approved for installation at Yuma in
Arizona, USA, to treat 360 Ml/d of a 3000 mg/litre TDS brackish irrigation water before
it drains into the Colorado River. This will maintain the quality of the Colorado before it
passes to Mexico. Extensive pretreatment and desalination trials have been conducted for
several years. Pre-treatment of partial lime softening followed by multi media filtration
has now been chosen for the final design. Nine RO and electrodialysis manufacturers
conducted desalination trials and from a performance and economic standpoint two RO
spiral wound module manufacturers have been selected to supply the desalting
equipment. UOP Fluid Systems are to provide 276 Ml/d capacity and Hydranautics 84
Ml/d. The final design of the project is proceeding with the initial operation of the plant
scheduled for 1982. At January 1979 prices the total cost of the desalting complex has
been estimated to be US $190 million. The total investment cost per m3 of
Desalination 111
FIG. 14. Typical components for a

Hydranautics plant to produce 1 Ml/d
from brackish water.
FIG. 15. 1·7 Ml/d section of the UOP

Fluid Systems’ Jeddah sea water RO
plant.
product per day of installed capacity would be approximately US$0·53/m3.20

Sea water RO technology is more recent and the largest plant installed is for 12·5 Ml/d
at Jeddah in Saudi Arabia. This was supplied and installed on a turn key basis by UOP
Fluid Systems and uses their TFC spiral wound membrane elements operating at 6100
kN/m2 (870 psi) and 30% recovery. It is expected that product water of 800 to 900
mg/litre will be produced in a single stage for up to two years. However when the
specified limit of 1000 mg/litre is exceeded a partial second stage will be used. A section
of the plant is shown in Fig. 15. The pretreatment for the plant consists of dual media
filters only followed by acid and polyphosphate dosing before the final pre-treatment
cartridge filters.
A slightly larger plant for 13·0 Ml/d with DuPont B10 hollow fibre sea water modules
is being installed in Russia to treat the Caspian Sea and should be operating in 1980.
There, a number of much smaller sea water or highly brackish water plants installed with
the B10 modules have been successfully operating for several years. A typical plant
installed by Paterson Candy International Ltd in Bahrain to produce 2·3 Ml/d of fresh
water from a 10000 mg/litre TDS feed is shown in Fig. 16.
FIG. 16. A 2·3 Ml/d plant using

DuPont hollow fibre sea water
modules, installed by Paterson Candy
International Ltd.
Desalination 113
TABLE 6 TYPICAL REVERSE OSMOSIS

COSTS Q3 1979 FOR A 90% LOAD FACTOR
Brackish water Sea water
1500 mg/litre 35000 mg/litre
TDS feed 50 TDS feed 500
mg/litre TDS mg/litre TDS
prod. 2800 kN/m2 prod. 6020 kN/m2
(400 psi) 25°C, (860 psi) 25°C,
75% conversion 30% conversion
4 Ml/d 20 Ml/d 4 Ml/d 20 Ml/d
Capital costs (£ 0·56 2·4 1·73 8.28
million)
Operating costs £mpa £mpa £mpa £mpa
(£ million p.a p/m3 p/m3 p/m3 p/m3
3
and p/m )
1. Fixed charges 0·065 5·0 0·28 4·3 0·20 15·5 0·97 14·8
11·75% (10%
interest, 20 yr
plant life)
2. Power(at 3·2
p/kWh)
(a) Brackish 0·076 5·8 0·38 5·8
(1·8 kWh/m3)
(b) Sea (8·0 0·34 25·6 1·70 25·6
kWh/m3)
3. Chemicals
Acid
Polyphosphate
Chlorine
Detergent
Formaldehyde
(a) Brackish 0·033 2·5 0·17 2·5
(b) Sea 0·041 3·1 0·20 3·1
4. Membrane
replacement
3-year life
(a) Brackish 0·053 4·0 0·27 4·0
(b) Sea 0·32 24·4 1·53 24·4
5. Maintenance, 0·035 2·6 0·14 2·2 0·035 2·6 0·14 2·2
operation,
staff and
materials
Total annual 0·26 – 1·24 – 0·94 – 4·54 –
cost (£ million)
Total water cost – 19·9 – 18·1 – 71·2 – 70·1
(p/m3)
A significant growth in the application of sea water RO can be expected in the future.
In time it is probable that, owing to the apparently more favourable economics, this
technique will gradually replace distillation for many applications.
4.3.6. Costs of Reverse Osmosis

Typical estimated costs for installed reverse osmosis plant alone can be presented for
both brackish and sea water systems at third quarter (23) 1979 for small and intermediate
plant sizes (Table 6). Different bases and assumptions can be used for estimating if
desired.
These RO costs must be used with caution since they only constitute the desalination
element of the total water supply cost. If the costs of wells or an intake, pre-treatment
works, civil and building works, product post-treatment and storage, and waste disposal
are included, the total capital cost can be more than double the figures quoted. Similarly
if power has to be generated at site or product water has to be pumped to supply,
significant additional costs will be incurred.
The costs quoted in Table 6 should not be compared directly with the distillation costs
quoted earlier since the bases employed are not identical. This situation arises because
different cost elements need to be included or excluded depending upon the site and the
process being considered. For example, distillation plant will largely not need to be
housed in a building while housing will be essential for reverse osmosis. Hence it must be
decided what cost elements need to be considered and included for overall cost estimates.
For a particular application for sea water desalination, it will usually be found that the
estimated total costs are similar whether desalting is performed by distillation or reverse
osmosis, with the RO costs usually being the lower, particularly for small plant.
4.4. ELECTRODIALYSIS
Electrodialysis (ED) is a membrane process which has been used commercially since the
mid 1950s for the removal of inorganic salts from water, i.e. as a desalination process. It
has also been used in Japan as a brine concentration process. In water, salts dissolve to
form positively charged cations and negatively charged anions. If a d.c. electric field is
applied across the solution, cations migrate towards the negatively charged cathode and
anions towards the positively charged anode. With permselective ion exchange
membranes, one cation and one anion placed between the electrodes, the permselective
property ideally allows the cation membrane to only pass cations and the anion
membrane to only pass anions. If now a stack of membranes, alternately cation and
anion, is placed between the electrodes, the solution between one pair of membranes
becomes depleted in ions while the solutions on either side become enriched, as shown in
Fig. 17.
The basic cell pair can be seen to consist of a feed or diluate stream between a cation
and anion membrane plus a concentrate on the other side of each membrane. In a
commercial configuration there can be several hundred cell pairs per stack, clamped as in
a filter press, between the two electrodes. The diluate and concentrate streams are
pumped in co-current flow through their respective cells. The cell thicknesses are
Desalination 115
approximately 1 mm and a plastic mesh or spacer separates the membranes as well as

acting as a turbulence promoter within the flow channels. The ion exchange membranes
are cast as flat sheets typically 0·6 mm thick around a woven fabric to provide
mechanical strength.
The membrane electrical properties of high conductance and ion selectivity control the
process through the application of Ohm’s Law for stack resistance and Faraday’s Law for
salt passage. Thus desalting energy requirements are proportional to the quantity of salts
removed.
There are two types of electrodialysis systems commercially available. The first that of
Ionics of the USA who have probably supplied over 75% of the world’s installed
electrodialysis plants. They manufacture their own membranes and have a tortuous path
spacer design where the liquid follows a continuous compressed ‘S’ shape across the cell.
Mixing is accomplished by ‘bridges’ across the flow path. The fluid velocity along this
path is approximately 0·5 m/s. For the past 3–4 years they have almost exclusively
offered plant with polarity reversal as the means of controlling scale formation. With this
facility, chemical dosing is not necessary, although simple chemical cleaning every two
or three weeks may be required. For normal brackish water treatment the polarity reversal
occurs automatically three or four times an hour and by means of motorised valves
reverses the product water, concentrate and electrode streams. The reversal purges
membrane and electrode surfaces of scaling and fouling materials which may have
deposited during the cycle. With this mode of operation plant is normally operated with
approximately 80% recovery. Owing to the economy of the process being directly linked
to the amount of salt removed the upper economic limit for treating brackish waters is
7000 mg/litre TDS.
FIG. 17. Principles of electrodialysis.

The membranes for this duty have an upper temperature limitation of 43°C and can
tolerate a pH range of 1–12.
The second type of electrodialysis system is that of the Japanese, Russians and
Permutit-Boby of the UK. The Japanese and UK systems use Japanese membranes and
are based on a mesh type spacer to separate the membranes and induce turbulence. They
typically operate at 0·1 m/s open cell velocity across the width of the cell. Scale
formation and fouling are controlled by acid dosing of the feed and adjustment of the
blowdown rate to prevent excessive concentrations of salts which potentially precipitate
on the membrane surfaces. The blowdown typically varies between 5 and 20% dependent
upon the feedwater quality. These plants often have the capability of polarity reversal but
this is done at possibly monthly intervals and the change over is manual. The limitations
of these systems are very similar to those of Ionics.
When a voltage is applied across a stack the diluate streams are depleted in ions,
increasing the electrical resistance of these streams. It is possible for them to be depleted
to the point where the stack resistance is excessive and operation of the stack impractical.
For the Japanese/European systems the lower practical limit of salinity of the product
water which electrodialysis can produce would be in the range 2–300 mg/litre TDS.
However, with the polarity reversal system it should be possible to achieve a product
quality of 50–100 mg/litre TDS as the high resistance layers are frequently broken up by
the polarity/flow reversals. In general terms the voltage applied to brackish water
electrodialysis plant can be between 1 and 2 V/cell pair. The current which flows is
proportional to the salt removed and at ambient temperature from Faraday’s Law for each
1000 mg/litre TDS removed per m3 the actual desalting power requirement is
approximately 0·7 kWh/m3 of product. There is usually a current limit of 130–150 A
between electrodes and if this is to be exceeded a further stage or stack is required.
The cross-sectional area of large electrodialysis stacks has usually been limited by the
size of the membrane which can be fabricated and readily handled. The maximum
commercial membrane size is typically 0·5 ×1·5 m. Often for a given duty and with
satisfactory operating parameters the residence time is insufficient for adequate
desalination to occur. Hence multistage/stack plants are required. It is then common for
progressive stacks to operate at lower currents to reduce the formation of excessive
resistances in the diluate streams. However if two stages are operated with one common
electrode in a single stack the current will be constant. Fewer cell pairs in the second
stage allow the use of higher flow velocities which help reduce cell resistances. The use
of two or more stacks in series for electrodialysis is quite common. In addition to the
power for the actual desalination for tortuous path spacers a further 0·5–0·75 kWh/m3 of
product would be required for pumping the product, concentrate and electrode streams
through the stacks. The pressure drop is higher through the tortuous path spacer systems.
Electrodialysis systems are relatively sensitive to feed water quality. The basic
specification for all manufacturers would be as follows:
The feedwater should have a turbidity of less than 2·0 FTU to give a reasonable life to
the following 10 µm prefilter and contain less than 0·3 mg/litre iron and less than 0·1
mg/litre Mn. In addition hydrogen sulphide, chlorine, organics or other polymeric
material which may cause membrane oxidation or fouling must be removed to very low
levels. However one membrane manufacturer does report a membrane which has good
electrical and mechanical properties and excellent resistance against high temperature and
Desalination 117
chemicals.21 It is claimed that any foulant on the membranes can be completely removed
by a high concentration of acid and/or alkali cleaning.
The total world installed capacity of ED to June 1979 is approximately 300 Ml/d. The
largest polarity reversal plant is a 15 Ml/d plant on the island of Corfu. There is
considerable experience with smaller plants of this type and a typical performance on a
brackish water can be seen in Table 7. A small polarity reversal ED plant can be seen in
Fig. 18.
The process has not typically been applied to the desalination of sea
TABLE 7 PERFORMANCE OF POLARITY
REVERSAL ELECTRODIALYSIS ON A 3000
mg/litre TDS BRACKISH WATER22
Feed Product
Na 170 54
Ca 508 61
Mg 188 13
Cl 340 53
HCO3 (as CaCO3) 211 81
SO4 1550 170
F 2.4 0.95
TDS 2969 433
pH 7.25 7.05
Conductivity 3540 650
(µS/cm, 25°C)
FIG. 18. An Ionics 0·5 Ml/d polarity

reversal, brackish water, ED plant (by
courtesy of Ionics Inc.).
water since it has been uneconomic, However recent development work in the USA with
high temperature electrodialysis23 and in Japan with large plant24 at ambient temperatures
has led to cost estimates for large-scale desalination of sea water which appear to be
similar to those for sea water reverse osmosis. While small batch or development plants
have been installed for sea water, the process could not be considered to be commercial at
this time. However the Japanese developments originated from concentration of brine by
electrodialysis which is now a standard commercial technique. Hence commercial sea
water desalination plants can be expected in the near future.
4.4.1. Costs of Electrodialysis

Typical estimated costs for installed brackish water electrodialysis are presented in Table
8 for Q3 1979. As for reverse osmosis, they only represent the desalination element of the
total water supply costs and hence must be used with caution. The additional costs for
associated plant site works and buildings, dependent upon the feedwater quality and the
site, must be added to obtain total water supply capital costs.
Desalination 119
TABLE 8 TYPICAL ELECTRODIALYSIS

COSTS Q3 1979 FOR A 90% LOAD FACTOR
(1800 mg/litre feedwater; 200 mg/litre TDS
product; 21°C; 78% recovery)
4 Ml/d 20 Ml/d
Capital costs (£ million) 0·7 2·8
Operating costs (£ million £mpa p/m3 £mpa p/m3
p.a and p/m3)
1. Fixed charges 11·75% 0·082 6·3 0·33 5·0
2. Power (at 3·2 p/kWh) 0·076 5·8 0·38 5·8
Desalting 0·7 kWh/m3
per 1000 mg/litre TDS
removed, i.e. 1·12
kWh/m3 Pumping 0·7
kWh/m3 Total 1.82
kWh/m3
3. Chemicals Provided 0·005 0·4 0·026 0·4
polarity reversal used
4. Membrane replacement 0·050 3·8 0·25 3·8
7-year life
5. Maintenance, operation, 0·046 3·5 0·20 3·1
staff and materials
Total annual cost (£ 0·26 1·12
million)
Total water cost (p/m3) 19·8 18·1
These ED costs are very close to those for a similar RO feedwater with in general the
capital costs being higher than for RO. However the higher capital costs are partly offset
by a longer membrane life and a reduced cost per m3 for membrane replacement. If a
brackish feedwater of 4000–5000 mg/litre TDS was to be treated by ED then the capital
costs would increase as extra membrane area would be required. In addition the power
requirements would more than double. The costs for RO however on this feedwater
would not greatly change. Hence for low TDS waters RO and ED are competitive, but as
the feed salinity rises RO can become the more attractive process.
Few ED plants have been installed with capacities greater than 16 Ml/d while
significantly larger brackish water RO plants have been installed. Hence there is also
more confidence in RO for large-scale applications. Since sea water electrodialysis is not
yet commercial, cost estimates have not been presented for this.
4.5. OTHER DESALINATION PROCESSES
A number of other desalination processes have received attention but none has been
developed beyond the pilot-scale or very small full-scale capacity. Solar energy has been
employed to distil sea water. Although plastics have been tried as still covers the limited
life of the material has led to the preference for glass covers. Work by the Battelle
Memorial Institute in the United States, the Commonwealth Scientific and Industrial
Research Organisation in Australia and Portham Ltd in the UK led to the construction of
a number of solar stills for the supply of small communities in the 1965 to 1970 period.
However, the high construction costs and limitations in performance have restricted the
use of the process. The output of a still is typically five times greater in the summer than
in the winter. Since this pattern of output does not match demand, season to season
storage is required. The provision of this storage adds significantly to the high capital
costs of solar distillation works.
The Sirotherm ion exchange process was developed in Australia as a means of
desalination for brackish waters. This process relies upon the change in ion exchange
properties of synthetic resins when they are exposed to different temperatures and by this
means regeneration is achieved, without the use of chemicals, at a temperature of 90°C.
This process has not yet been employed to any significant extent although there may be
some requirement for small-scale units of this type in remote locations where chemicals
are not available.
In the United States, Israel and the UK, considerable effort was devoted to the
development of freezing desalination processes which rely upon the melting of salt-free
ice formed from saline solutions. The attraction of the processes has been the
theoretically low energy demand involved in ‘heat-pumping’ the latent heat of fusion.
The secondary refrigerant process employed an organic refrigerant such as butane. In the
freezing section of the plant butane liquid was vaporised whilst in the melter the
compressed butane vapour provided the heat source for the melting of ice to form water.
In both the United States and the UK initial optimism for the process was never
justified by results. Difficulties in handling ice, washing ice free of salt, and separation of
ice/salt solution/refrigerant mixtures led to plants becoming excessively complex and
unreliable in operation. For these reasons development of the process virtually ceased
during the first half of the 1970s.
The direct contact vapour compression freezing process involves the formation of ice
and simultaneous evaporation under vacuum conditions. The vapour produced after
compression was used as a direct contact heat source for melting washed ice. Whilst
reliable operation of the process was proven in Israel and the United States it was found
to be impossible to scale-up the vapour compressors owing to the massive volumes of
vapour to be handled, and consequently plant capacities were limited to about 1000 m3/d.
Whilst consideration has more recently been given to the use of absorption
refrigeration cycles as a means of overcoming the practical limitations of compression in
vacuum freezing, it is rather unlikely that the process will be developed to a stage where
it can compete with the alternative sea water desalination processes.
REFERENCES
1. MOORHEAD, A., The Blue Nile, 1967, Dell, New York.

2. BURLEY, M.J. and MAWER, P.A., Water Res. Assoc. TP50, 1967, Water Research Centre,
Marlow, Bucks.
3. VEENMAN, A.W. and BROEKENS, J.D., Proc. 6th Int. Symposium on Fresh Water from the
Sea, Las Palmas, 1978, 2, 53.
Desalination 121
4. KLAREN, D.G. and WINDT, J., Proc. 6th Int. Symposium on Fresh Water from the Sea, Las
Palmas, 1978, 2, 15.
5. BURLEY, M.J., Proc 2nd Int. Symposium on Fresh Water from the Sea. 2, paper 16, 10 pp,
Athens, May 1967.
6. WADE, N.M., Proc. 6th Int. Symposium on Fresh Water from the Sea, Las Palmas, 1978, 1, 327.
7. TODD, B., Private communication, International Nickel Co.., Birmingham, UK.
8. SATO, S., Bull. Japan. Inst. Metals, 1978, 575.
9. NEWTON, E.H., BIRKETT, J.D. and KETTERINGHAM, J.M., Report to U.S. Office of Saline
Water, Contract No. 14/30/2721, March 1972.
10. WADE, N.M., Desalination. 1979, 31, (2), 309.
11. U.S. Dept. of the Interior, Office of WaterResearch and Technology, Desalting Plants Inventory
No. 6, October 1977.
12. DESAI, A.M., Desalination. 1977, 23, 367.
13. RILEY, R.L., Fox, R.L., LYONS, C.R., MILSTEAD, C.E., SEROY, M.W. and TAGAMI, M.,
First Desalination Congress of The American Continent 1, 11–1, 1976, Elsevier Scientific
Publishing Co., Amsterdam.
14. AL-GHOLAIKAH, A., EL-RAMLY, N., JAMJOON, I. and SEATON, R., National Water
Supply Improvement Association J., Jan. 1979, NWSIA, USA.
15. CARMONA, J. and DE BUSSY, R.P., Presented at the Comision Federal de Electricidad,
Mexico City, June, 1976.
16. GROVER, J.R., GAYLER, R. and DELUE, M.H., Proc. 4th Int. Symposium on Fresh Water
from the Sea, Heidelberg, 1973, 4, 349.
17. DuPont Company, Plastic Products and Resins Dept., Permasep Products, Tech. Bull. 491,
1977.
18. The Plenty Group, Technical Bulletin on Self Cleaning Filters, Newbury, UK, 1978.
19. GHULAIGAH, H.E.A. and ERRICSSON, B., Proc. Int. Congress on Desalination and Water
Re-Use, Nice, 1979, 1, 3–301.
20. LOPEZ, M., Proc. Int. Congress on Desalination and Water Re-Use, Nice, 1979, 1, 1–15.
21. KISHI, M., SERIZAURA, W. and NAKARO, W., Desalination, 1977, 23, 203.
22. KALZ, W.E., Ionics Inc., Bulletin TP. 307, 1977.
23. McRAE, W.A., PARSI, E.J., GANZI, G., JHA, A. and O’DONOGHUE, K., Proc. 6th Int.
Symposium on Fresh Water from the Sea, Las Palmas, 1978, 3, 101.
24. KAWAHARA, T., ASAKA, T. and SUZAKI, K., Proc. 6th Int. Symposium on Fresh Water
from the Sea, Las Palmas, 1978, 3, 95.
Chapter 5
DISINFECTION
A.T.PALIN, O.B.E., B.Sc., Ph.D., F.R.I.C., F.I.W.E.S.
Consulting Chemist, Newcastle upon Tyne, UK
SUMMARY
The principal disinfecting agents used in water treatment are chlorine, chloramine,
chlorine dioxide and ozone. After a preliminary discussion of their relative importance,
these disinfectants are dealt with in turn with reference to such aspects as historical
development, chemical and bacteriological properties and practical application.
While chlorination remains the most widely used process at the present time, studies
of recent years have revealed problems associated with the possible formation in some
waters of traces of harmful carcinogenic by-products such as trihalomethanes.
Consideration is given to the influence of these findings on future developments.
Finally, an outline of the DPD analytical tests for treatment control is presented of
which a special advantage lies in their ability to determine mixtures of disinfectant
residuals in the treated water.
5.1. INTRODUCTION
The process of disinfection is applied to water for the purpose of destroying or

inactivating pathogenic bacteria, viruses and other disease-producing organisms. In the
vast majority of cases the disinfectant used at the present time is chlorine, a chemical first
introduced for regular water treatment at about the turn of the century. Treatment by
chloramine, a form of combined chlorine, was widely practised during the 1930 to 1940
period but gradually became less popular because of its much slower rate of disinfection
compared with that of chlorine and its inability to provide other desired quality
improvements, especially when dealing with the more polluted sources. Nevertheless
considerable use is still made of the chloramine process and it may, for certain types of
water, remain the preferred method.
The only other disinfecting agents used in water treatment on any appreciable scale are
chlorine dioxide and ozone. Demand for these is likely to increase in view of the attention
now being paid to the possibilities of using one or other in conjunction with or alternative
to chlorine. In addition, renewed interest is being taken in the chloramine process and in
those advantages it possesses compared with straight chlorination. These developments
arise from the findings of recent years that the chlorination of polluted waters, and even
those containing natural organic colouring matter, may produce undesirable traces of
chloroform, a recognised carcinogen, and similar potentially carcinogenic
Disinfection 123
trihalomethanes. It is this aspect of disinfection to which most research is directed at the

present time.
Other disinfectants such as ultraviolet radiation, bromine and iodine are not regarded
as practical for drinking water treatment except possibly on a very small scale. U.v. does
not provide an active disinfectant residual in the water, bromine is difficult to handle and
is expensive, and iodine, while satisfactory for emergency purposes, can impart taste,
odour and colour to the water and also is very expensive. It may be mentioned that for
emergency use on a small scale boiling the water is completely effective.
In considering developments in disinfection the present discussion is concerned only
with those agents having any significant large-scale application in the field of water
treatment, that is to say, chlorine and the related chloramines, chlorine dioxide and ozone.
The survey will cover their chemistry, mode of application, bactericidal efficiency, and
the necessary treatment control to ensure full compliance with chemical and
bacteriological standards of drinking water quality. Reference will also be made to such
modifications to treatment as might be required to minimise risks of harmful by-products
being produced in the finished water.
In the case of turbid waters the presence of suspended particulate matter may interfere
with disinfection by affording protection to embedded bacteria from the action of the
disinfectants. Clarification by filtration is therefore essential usually preceded by
coagulation and sedimentation. Such processes provide the further advantage of
physically removing a substantial proportion of the micro-organisms thus reducing the
pollution load at the disinfection stage and the general chlorine demand of the water.
5.2. CHLORINE
5.2.1. History
In the earliest applications of chlorine for water disinfection the chemical was applied in
the form of hypochlorite. From about 1915 increasing use was made of chlorine gas
contained under pressure as liquid in cylinders. Its excellent bactericidal properties
became firmly established and additional benefits, resulting from its oxidising power,
became evident such as removal of iron and manganese, and of tastes and odours. On the
other hand it was found that chlorination itself could impart to some waters undesirable
tastes variously described as medicinal, iodoform or chlorophenol. These were due to
reactions of the chlorine with trace amounts of phenols and similar compounds.
Furthermore the use of too much chlorine could itself give a chlorinous taste because of
the presence of excessive amounts of ‘residual chlorine’, that is the amount remaining in
the water after treatment.
From the public health point of view the benefits derived from the use of chlorine as a
disinfectant were unquestionable so that further investigation up to about 1935 was
directed more towards improvements and modifications of treatment aimed at
overcoming any possibly undesirable side-effects on general water quality. During this
period attention was paid to the use of chloramines, these being compounds formed in the
water by reaction between chlorine and ammonia. Ammonia is present in many raw water
sources, usually as a consequence of pollution, or it can be added if required in the form
of ammonia, from liquid ammonia under pressure in cylinders, or in smaller installations

as ammonium sulphate in solution form.
The chloramine or ammonia-chlorine process had the advantage, because of its lower
reactive and oxidising power, of greatly reducing the risks of producing undesirable
tastes and odours but at the same time its bactericidal activity, or its ‘rate of kill’ was
much lower than that of chlorine itself. It thus became necessary in practice to make
provision for an adequate time of contact to ensure complete disinfection of the water.
Other modifications of treatment to be developed at that time included ‘superchlorination
and dechlorination’ signifying the use of much higher chlorine doses than required
normally, possibly to compensate for short contact time, with subsequent dechlorination
to remove the excess residual chlorine.
These developments towards greater flexibility in methods of chlorination were
coupled with an awareness of the need for more fundamental research into the chemistry
of water chlorination. This research received much impetus from the reported discovery
in 1939 of so-called ‘breakpoint’ chlorination by Faber1 and Griffin,2 a phenomenon
which had in fact been described in the literature as long ago as 1914 by Ruys3 and again
by Holwerda in 1930.4 It was found that as the rate of application of chlorine to some
waters was increased the residual chlorine after a period of contact could behave in an
erratic manner. With polluted waters a high chlorine dose might well give a lower
residual than that obtained from a low dose. Furthermore, the removal of tastes and
odours was complete in such cases only when the residual chlorine had passed the
minimum point, referred to as the ‘breakpoint’, and had begun to increase more or less in
line with progressive increases in chlorine dose.
The indications were that the presence of ammonia was a principal factor contributing
to these unusual dose-residual relations. In the continuing work from 1940 onwards
several workers demonstrated the close relation between the presence of ammonia and
this characteristic type of chlorine dose-residual curve, the so-called ‘hump and dip’
curve. Ample confirmation was thus provided for the earlier work to which may be added
that of Gerstein5 who had also reported in 1931 that in the chlorination of water
containing ammonia a point was reached where with increasing chlorine dose the residual
chlorine began to break down and disappear more rapidly than if a lower dose were used.
Many theories were presented to account for the breakpoint in chlorination, often with
a minimum of firm supporting experimental evidence. A full explanation of the chemistry
of the chlorine-ammonia and similar reactions awaited the development of suitable
analytical techniques for the determination of both the type and the amount of the
residual chlorine compounds at these relatively minute concentrations in water. The
behaviour of such compounds, especially their ability to react both among themselves
and with free chlorine, thus provided a rewarding field for further exploration eventually
made possible by the advances made in analytical methods.6,7 Subsequent researches then
revealed that residual chlorine in water could exist in different chemical forms with
different oxidising and bactericidal powers. The results emphasised the need, from a
practical point of view, for suitable water chlorination control tests since it had become
clear that an indication of the chemical nature as well as the amount of the residual
chlorine in the treated water was essential. With these refinements in test procedures the
chlorination process could be applied to give optimum results under all conditions of raw
water quality.
Disinfection 125
5.2.2 Forms of Chlorine

In any comparison of strengths of chlorinating chemicals in terms of chlorine yield it is
standard practice to refer to their ‘available chlorine’ content. Similarly the concentration
of residual chlorine in water is always expressed in terms of available chlorine no matter
in what chemical form it might be present. While regarded by some as a misnomer, this
remains the standard form of expression for the strengths or capacities of the various
forms of chlorine as well as for the doses in which they are applied and for the residues
that remain in the water. Any chlorine that eventually appears as chloride, the final
reduction product, represents a complete loss of available chlorine.
The expression ‘available chlorine’ is defined in precise chemical terms and relates to
capacity rather than oxidising or disinfecting power. It is based upon the determination of
the equivalent amount of iodine liberated from iodide in acid solution by a known amount
of the chlorine compound. Since the available chlorine content of pure chlorine is 100%
the corresponding figures for other compounds may be obtained as a percentage by
comparing the amounts of iodine thus liberated with the amount liberated from the same
weight of chlorine.
To illustrate the application of the available chlorine concept the following four cases
may be considered using potassium iodide KI and hydrochloric acid HCl. (In practice
another acid, say acetic acid, would be used in the determinations since HCl itself may
contain traces of chlorine. For the present purpose it enables the reactions to be presented
more simply.)
By calculation from molecular weights it may be demonstrated that the amounts of iodine
liberated from 1 g of each of the above chemicals are 3·6 g, 4·84 g, 4·93 g and 9·41 g
respectively. In terms of percentages, taking chlorine as 100%, the available chlorine
contents become HOCl—134%, NH2Cl—137% and ClO2—261%. As previously noted it
would be incorrect to equate these figures to oxidising power and bactericidal efficiency.
Monochloramine is far less reactive than hypochlorous acid. Similarly chlorine dioxide
although first introduced as being two and a half times more powerful than chlorine is in
fact considerably weaker in some reactions.
A knowledge of the available chlorine content is of value in assessing the comparative
costs of chlorine from different commercial products and in checking for any gradual loss
on storage (Table 1).
TABLE 1 AVAILABLE CHLORINE CONTENT

OF DIFFERENT FORMS OF CHLORINE AND
THE QUANTITIES REQUIRED TO GIVE 1 lb
OF AVAILABLE CHLORINE
Form of chlorine Available Equivalent
chlorine quantities
Chlorine gas 100% 1 lb
High Test 65–70% 1·5 lb
Hypochlorite
(calcium
hypochlorite)
Bleaching Powder 30–35% 3·0 lb
(chloride of lime)
Sodium hypochlorite 15% 0·5 gal
Chlorine gas is produced commercially from sodium chloride (common salt) by an

electrolytic process and is supplied under pressure in cylinders and drums. In full
containers most of the chlorine is present in liquid form. For application the chlorine is
withdrawn as gas with the rate of flow being set to the desired level by suitable metering
equipment. The gas is then dissolved in a minor flow of water, usually by means of an
injector, before mixing with the main flow.
Chlorine solutions as supplied are usually prepared from chlorine and caustic soda, the
product being sodium hypochlorite. When fresh the available chlorine content is up to
15% by weight but gradual loss occurs on storage. In practice the solution may require
dilution to provide more convenient application with solution-feed dosing equipment.
Of the solid hypochlorites one of the best known is bleaching powder or chloride of
lime. The essential constituent is calcium oxychloride which is decomposed by water to
produce calcium hypochlorite. When fresh its available chlorine strength is about 33% by
weight but there is fairly rapid loss on storage unless conditions are cool and dry. A
superior form of solid hypochlorite is available under the name High Test Hypochlorite.
This consists of solid calcium hypochlorite in the form of free-flowing granules which
dissolve readily in water or as tablets for smaller-scale use. The available chlorine content
is up to 70% by weight and the shelf life is considerably longer than that of bleaching
powder.
5.2.3. Chemistry of Chlorination

When chlorine is dissolved in pure water two practically instantaneous reactions occur.
First, the chlorine hydrolyses to form hydrochloric acid and hypochlorous acid thus
Secondly, the hypochlorous acid partly dissociates to give hydrogen ions and
hypochlorite ions thus:
Disinfection 127
Three forms of available chlorine are involved in these reactions, namely molecular
chlorine Cl2, hypochlorous acid HOCl and hypochlorite ion OCl−. A balance exists
between these forms and at any given time the relative proportions are governed by water
temperature and pH of which by far the more important is pH. This relation is shown in
Fig. 1. The equilibrium is independent of the form in which the chlorine is initially added
assuming that such addition does not itself alter the pH. It may be seen that within the pH
ranges encountered normally in drinking water only two forms are present, namely HOCl
and OCl−. These together comprise what is defined as ‘free available chlorine’.
The small amount of hydrochloric acid produced in the above hydrolysis of chlorine is
neutralised by the natural alkalinity of the water. The resulting effect upon pH is
insignificant except for relatively soft waters low in buffering capacity (i.e. in-built
resistance to pH change).
Having considered the reactions that occur when chlorine is dissolved in pure water it
is necessary next to examine the effect of those impurities whose presence is to be
expected in the chlorination of natural and polluted waters. As already noted, the
presence of ammonia has a particularly marked influence upon the chemistry of water
chlorination. Its reactions with chlorine may be represented as follows:
FIG. 1. Effect of pH value on form of

free available chlorine in water.
The lower the pH of the water and the higher the ratio of chlorine to ammonia the greater
is the tendency to produce the more highly chlorinated derivatives. Within the normal pH
range of water, however, the product consists almost entirely of monochloramine and the
reaction is complete within one minute. For conversion of all the ammonia to
monochloramine about five times as much chlorine is required, that is the combining
weight ratio is 5 of Cl2 to 1 of NH3 calculated in terms of nitrogen.
If more chlorine is applied than is required for this rapid initial reaction, continuing
oxidation reactions occur at a slower rate eventually producing mainly nitrogen thus:
2NH2Cl+HOCl=N2+3HCl+H2O
Small amounts of dichloroamine and nitrogen trichloride may also appear during the
course of the breakpoint reactions especially in the zone corresponding to chlorine doses
Disinfection 129
above the 5:1 ratio, but the net result as the chlorine is further increased and adequate
contact period is allowed corresponds to the following overall equation:
2NH3+3Cl2=N2+6HCl
According to this equation the amount of chlorine required to oxidise one part by weight
of ammonia-nitrogen is 7·6 parts by weight. While nitrogen is the main end-product small
amounts of nitrate and possibly nitrogen trichloride may also remain with the result that
the observed ratio becomes somewhat higher at about 8·3:1. In natural waters there may
be additional chlorine absorption by other types of impurity giving a ratio about 10:1. For
polluted waters the ratio could be much higher.
The chlorine dose-residual curve may be plotted for a given sample of water by
treating a series of aliquots with increasing doses and determining the residual chlorine
values after a predetermined period of contact. If the water contains only a small amount
of ammonia the curve obtained will be similar to that of Fig. 2. On the other hand, where
appreciable amounts of ammonia are present the distinctive breakpoint curve will be
obtained as in Fig. 3.
FIG. 2. Breakpoint curve for water

with ammonia-nitrogen 0·06 mg/litre.
FIG. 3. Breakpoint curve for water

with ammonia-nitrogen 5·2 mg/litre.
FIG. 4. Chlorine dose-residual curve at

pH 6·0 after 1 day. Initial ammonia 0·5
mg/litre (N).
Disinfection 131

mg/litre (N).

mg/litre (N).
The effect of pH on the composition of the residual chlorine at different stages of the
breakpoint curve is shown in Figs. 4, 5 and 6. The ammonia was added as ammonium
chloride, the water used being otherwise of zero chlorine-demand. In these systems,
therefore, the reactions involved only chlorine and ammonia. The most important feature
of the curves is that before the breakpoint the residual chlorine is present in the form of
chloramines, that is combined chlorine. After the breakpoint it is present mostly as free
chlorine.
Recognition of these two forms is of vital importance in the control of water
chlorination since the bactericidal and virucidal properties of free chlorine are vastly
superior to those of combined chlorine. Therefore for maximum safety in the disinfection
of water it is generally desirable to chlorinate beyond the breakpoint, that is to the point
of establishing free chlorine.
A disadvantage of breakpoint chlorination lies in the possibility of producing traces of
nitrogen trichloride unless the pH is fairly high. This may impart to the water an
objectionable chlorinous-type odour for the eradication of which additional treatment
may be necessary, possibly involving complete dechlorination with a final stage of
ammonia-chlorine treatment.8
5.2.4. Modern Chlorination Practice

From the foregoing review of the chemistry of chlorination it is clear that the following
definitions are fundamental to modern practice.
Free available chlorine is residual chlorine in the form of hypochlorous acid and
hypochlorite ion.
Combined available chlorine is residual chlorine existing in combination with
ammonia or organic nitrogen compounds.
The chlorine demand of a water is the difference between the applied dose and the
residual chlorine. It is necessary to specify the conditions, that is whether the residual
produced is in the free or combined form, and the time of contact. The chlorine dose also
should be specified. In those cases where the chlorine demand is due mainly to non-
nitrogenous matter or oxidisable inorganic impurities the chlorine dose-residual curve
may not, as already noted, exhibit a breakpoint. Nevertheless after the demand has been
satisfied free residual chlorine will appear and continue to increase with increasing dose.
The modern approach to the classification of chlorination methods is based upon this
distinction between free and combined chlorine. The two main types of process therefore
are free residual chlorination and combined residual chlorination. Variations may be
introduced to meet particular requirements but the results obtained under any given
conditions will always depend upon the nature and amount of the residual chlorine
produced in the water. The production of residuals of known composition by suitable
control of chlorination has been made possible by the methods of chlorine residual
differentiation now available.
Where dechlorination is required it is usual to apply sulphur dioxide which reacts
rapidly to destroy the residual chlorine in an approximate 1:1 ratio. Sulphur dioxide is a
gas under normal pressure but may be compressed to liquid for supply in cylinders and
other containers. Dosing equipment is similar to that used for chlorine. Both free and
combined chlorine are destroyed thus:
SO2+H2O=H2SO3 (sulphurous acid)
H2SO3+HOCl=H2SO4+HCl
H2SO3+NH2Cl+H2O=NH4HSO4+HCl
Disinfection 133
Passage of the water through granular activated carbon filters provides another means of
dechlorination, the principal reaction being as follows:
C+2Cl2=2H2O=4HCl+CO2
For small-scale dechlorination use can be made of such chemicals as sodium thiosulphate
(‘hypo’), sodium sulphite and sodium bisulphite.
The bactericidal and virucidal power of free chlorine may be influenced by other
factors of which one of the most important is pH. The pH of the water governs the
relative proportions of hypochlorous acid and hypochlorite ion in the free chlorine
residual. With pH rising above 6·0 the proportion as HOCl declines from virtually 100%
down to almost zero at pH 9.0. Since the bactericidal activity of HOCl is something like
80 times more powerful than that of the OCl− ion it is evident that in free residual
chlorination the higher the pH the less active is the residual. At pH values of 7·0, 7·5 and
8·0 the HOCl proportions are very approximately 75%, 50% and 25%.
If in addition to the free chlorine reading the pH is known it is possible to calculate the
residual HOCl quite simply from the established dissociation factors. Where treatment is
concerned it may be necessary in practice to work to higher free chlorine residuals at the
higher pH values. There is evidence that the killing power of combined chlorine
diminishes also with rising pH.
The bactericidal power of both free chlorine and combined chlorine decreases as the
water temperature falls. Thus pH and temperature,
TABLE 2 MINIMUM CYSTICIDAL AND
BACTERICIDAL RESIDUALS (AFTER 30-
MINUTE CONTACT)
pH Free chlorine Combined
chlorine
Bactericidal Cysticidal Cysticidal Bactericidal
0–25°C 22–25°C 2–5°C 0–25°C
6·0 0·2 2·0 7·5 2·0
7·0 0·2 2·5 10·0 2·5
8·0 0·2 5·0 20·0 3·0
9·0 0·6 20·0 70·0 3·5
especially if the first is high and the second low, may together have an important bearing
on the period of contact required to achieve satisfactory disinfection. Under favourable
conditions free residual chlorination may require no more than a few minutes whereas
combined chlorine under similar conditions might require from 30 minutes to 2 hours.
Whatever the conditions the final test resides in microbiological examination of the
treated water.
From studies carried out originally by the US Public Health Service9,10 it was
established that within the same contact period about twenty-five times as much
chloramine as free chlorine was required for complete destruction of bacteria.
Furthermore, for residuals of the same amount about 100 times the exposure period was
required with chloramine compared with free chlorine.
In 1956 the National Research Council11 re-analysed the earlier data and submitted
revised recommendations for minimum safe residuals, based on 30-minute contact, with
additional recommendations regarding cysticidal chlorine residuals (Table 2).
In practice the performance of chloramine will be rather better than is indicated
because a short period of up to one minute is required for completion of the reaction
between chlorine and ammonia after their addition to the water. During this period some
unreacted free chlorine will be present thus giving a significant improvement in bacterial
kill.
5.3. CHLORINE DIOXIDE
5.3.1. History
Chlorine dioxide was first applied as a water treatment chemical in 1944 at a plant in
Niagara Falls, USA. It was found to be effective in controlling unpleasant tastes and
odours especially those of phenolic origin. It was subsequently investigated in the UK by
the author12 with particular reference to its decolorising effect on peaty moorland waters.
During this work its general chemical properties were examined and this led to the
discovery of its inertness to ammonia. Because of this chlorine dioxide does not give a
breakpoint type of dose-residual curve.
In addition to its use in the control of tastes and odours it now finds application in the
oxidation of iron and manganese and some forms of organic impurity. While originally
its effectiveness as a disinfectant was open to some doubt, and thus it was always used
with an excess of chlorine, it is now fully accepted as a disinfectant in its own right. It
can provide a longer-lasting residual than chlorine which is advantageous where it is
desired to maintain residual disinfectant throughout the water distribution system.
Although its use has expanded considerably, it still remains relatively small compared
with that of chlorine. This position may change in view of current researches into the
trihalomethane problem.
5.3.2. Generation
Chlorine dioxide is produced on site by mixing strong chlorine solution, as delivered
from the normal type of chlorinator, with a solution of sodium chlorite when the
following reaction occurs:
Cl2+2NaClO2=2NaCl+2ClO2
If the pH of the mixture is not low enough the reaction may not go to completion thus
leaving unreacted chlorine and chlorite in the discharge to the water. A system has been
devised, known as the CIFEC method, whereby the chlorine water is recirculated around
an enrichment loop to ensure almost 100% production of chlorine dioxide after admixture
with the sodium chlorite solution.13
For smaller installations other methods may be used in which the chlorine dioxide is
generated under controlled conditions from mixtures of sodium chlorite and either
hydrochloric or sulphuric acid or from similar mixtures with the addition of hypochlorite.
Disinfection 135
5.3.3. Chemical and Bactericidal Properties

As previously noted chlorine dioxide, unlike chlorine, does not react with ammonia to
form chloramines. In waters of high ammonia content this lack of reactivity is beneficial
compared with chlorine which, unless the dose is high enough, becomes converted to the
slow-acting chloramines.
In water treatment only a part of the full oxidising capacity of chlorine dioxide is used
because in most of its reactions it undergoes only a first-stage reduction corresponding to
the change from chlorine dioxide to chlorite. This is equivalent to no more than one-fifth
of its total available chlorine as expressed normally. Thus in the majority of reactions
chlorine is a better oxidant than chlorine dioxide except in special circumstances, such as
the presence of phenols, when the advantage lies with the dioxide.
As a general rule chlorine dioxide is as effective a disinfectant as chlorine and is less
sensitive to changes in the pH of the water. Thus with pH rising over 7 it is better able to
maintain its disinfecting power unlike chlorine which gradually becomes less effective
because of the shift from hypochlorous acid to the much less active hypochlorite ion.
5.4. OZONE
5.4.1. History
The earliest recorded use of ozone as a water disinfectant occurred in France in 1886. As
a result of further pilot studies a number of treatment plants adopted ozonation in the
early years of the present century including, for example, that at Nice where the process
has operated continuously from 1906. Thus ozonation has a long history of use but never
on so extensive a scale as that of chlorine. Apart from disinfection it may be applied for
other purposes such as taste and odour control, iron and manganese oxidation and
organics removal. The strong disinfecting and oxidising action of ozone is due to the
release of the very reactive third atom of oxygen with reversion to oxygen O2 thus:
2O3=3O2
There are today possibly more than one thousand plants using ozone.
5.4.2. Generation and Application

Ozone is generated on site from oxygen or, as is more usual in large-scale commercial
production, from the air. In the generation process the air, which must be clean and dry, is
passed between electrodes separated by an air gap across which a very high voltage is
maintained to produce a silent electric discharge.
The amount of ozone produced may be up to 30 g/m3 of air. This relatively low
concentration coupled with its low solubility requires very intimate mixing and
dispersion of the ozonised air throughout the water in order to achieve the highest
possible degree of transfer from the gas to the liquid phase. For this purpose the methods
used may take the form of injectors, porous diffusers or mechanical mixers. For
maximum efficiency the aim must be to produce as small a bubble size as possible. As
the dispersed bubbles gradually grow in size so the efficiency of transfer falls away with
the decreasing interfacial area of contact.
In calculating applied doses it is assumed that all the ozone is transferred from the air
to the water although in practice the proportion of ozone dissolved may range from 60%
to possibly better than 90%. Recommended dosage rates for drinking water are generally
within the range 0·2 to l·5 mg/litre. The ozone disappears very rapidly from the water and
it is therefore not possible to maintain an active residual as is the case with chlorination.
The absence of any lasting residual disinfecting action is a disadvantage of the process
and for this reason a supplementary stage of chlorination is usually applied.
Satisfactory disinfection by ozonation is assured normally by 0·4 mg/litre or more of
residual ozone acting for a minimum contact period of 4 to 5 min. In practice the
presence of ozone residuals of about 0·1 mg/litre at the outlet of the ozonation chamber is
generally effective.
5.5. RELATIVE COSTS
Unit costs for five disinfection processes have been developed by the U.S. Environmental
Protection Agency.14 The basis of these estimates includes capital, operational and
maintenance costs. In the case of chlorine, chlorine dioxide and ozone it is assumed that
the specific doses have equivalent disinfecting capability. This is not the case where
chloramine is concerned and it may be necessary to allow further for additional
disinfection unless high quality waters are being treated.
While the results varied with design capacity the general indication was that chlorine
dioxide, chloramine and ozone cost about 65%, 20% and 100%, respectively, more than
chlorination.
5.6. DISINFECTION BY-PRODUCTS
The problems associated with the formation of traces of chloroform and other
trihalomethanes as possible by-products of chlorination have led to a review of the entire
subject of drinking water disinfection. Such traces have been detected in polluted waters
after chlorination and also where chlorine has been applied with long contact period for
colour removal. As well as chemical pollutants the naturally occurring constituents of
peaty water, such as humic acid and similar organic matter, can act as precursors of the
trihalomethanes. The carcinogenicity of chloroform has now been acknowledged by the
U.S. Environmental Protection Agency who are urging that modified treatment practices
be adopted to minimise chloroform and similar trihalomethane formation from the
organic materials found in natural and polluted waters. The slow-acting chloramines do
not form these compounds.
While there is at present no acceptable substitute for chlorine for general use in water
disinfection there is no doubt that increased consideration will be given to ozone, chlorine
dioxide and even chloramine where conditions are suitable for its application. In addition,
mixed treatments such as, for example, chlorine plus chlorine dioxide and ozone plus
chlorine are likely to receive more attention. Selection of points of application will
Disinfection 137
require further thought. Shortening of pre-chlorination contact periods or applying the

chlorine at a later stage of treatment, for instance after coagulation and sedimentation,
can result in substantially reduced chloroform production.
There are two general approaches to the problem from a treatment point of view. First,
to reduce, by improved or modified pre-treatment prior to chlorination, the level of
organic pollutants acting as precursors of the trihalomethanes. Secondly, to use other
disinfectants in conjunction with chlorine so as to enable chlorine doses and contact times
to be reduced. A favoured approach in attempting to achieve the first objective is
adsorption by passage through granular activated carbon (GAC) filters. Improved results
are obtained by pre-ozonation which assists by breaking down those organic materials not
readily adsorbed into more readily adsorbable and biodegradable compounds. It is now
recognised that the carbon becomes biologically active. This process in which the use of
ozone is combined with biological activated carbon (BAC) is receiving much attention
with very promising results. BAC is an effective process for ammonia removal.
Chlorine dioxide treatment appears not to produce trihalomethanes when prepared free
of chlorine. As normally produced for water treatment by reacting sodium chlorite with
chlorine, some excess of chlorine remains. Results indicate that formation of
trihalomethanes can then occur with these mixed residuals but at lower concentration
than would be produced from an equivalent amount of chlorine alone. Trace amounts of
chlorite may appear as a by-product of chlorine dioxide treatment either because of
TABLE 3 OUTLINE OF DPD TESTS FOR FREE
CHLORINE AND OTHER DISINFECTANT
RESIDUALS
Test Reagents Notes
(in order
of use)
1. Free chlorine 1
2. Free chlorine 1
Combined 3
chlorine
3. Total available 4 Reagents 1 and 3
chlorine added together may
be used in place of
reagent 4.
4. Chlorine G Result gives one-
dioxide 1 fifth chlorine
dioxide. Multiply
by 5. This result is
in terms of available
chlorine. To express
as ClO2, multiply
instead by 1·9.
5. Chlorine 1 To obtain free
dioxide plus 3 chlorine from
free chlorine. reagent 1 reading,
Combined deduct one-fifth
chlorine chlorine dioxide as

given in test 4.
6. Chlorite A Deduct four-fifths
(continuation N chlorine dioxide in
of test 5) terms of available
chlorine as obtained
from test 4 to obtain
chlorite.
7. Ozone 4 Reagents 1 and 3
added together may
be used in place of
reagent 1.
8. Total chlorine 4 Provides for
plus ozone G separate
Total chlorine 4 determination.
only
9. Free chlorine 1 Ozone response is
plus ozone incomplete but
difference gives
combined chlorine.
Total chlorine 3 Combine with test 8
plus ozone for complete
differentiation.
Key to Reagents
1—DPD indicator+buffer
2—for differentiation of combined chlorine (not
included here)
3—potassium iodide
4—1 and 3 combined
G—glycine
A—acidifying agent
N—neutralising agent
incomplete conversion or because of reversion in the treated water, bearing in mind that
chlorite is the first reduction product of chlorine dioxide. The possible toxicity of chlorite
has not been ruled out and preliminary indications so far suggest that it is about the same
as that of nitrite.
In the control of these modified disinfection processes involving mixed treatment by
chlorine, chlorine dioxide and ozone a suitable residual control test capable of their
separate determination is essential. Only the DPD test,15,16 adopted as standard in many
countries for free chlorine and combined chlorine compounds, can meet these further
analytical requirements. An outline of the procedures is given in Table 3. Although
monochloramine, dichloramine and nitrogen trichloride may be determined separately, if
required, by using quite simple modifications of the method they are here included
together as combined chlorine.
The preparation of the various reagents is given in standard manuals of water analysis
and in the author’s published papers.17,18 Standardised tablets and powder reagents are
available commercially.19 These are used with colour comparators in test kits as shown in
Figs. 7 and 8.
Disinfection 139
FIG. 7. Tintometer Lovibond test kit

for residual chlorine, pH value and
alkalinity.
FIG. 8. LaMotte test kit for residual

chlorine and pH value.
REFERENCES
1. FABER, H.A., J. Amer. Water Works Assoc., 1939, 31, 1539.

2. GRIFFIN, A.E., J. Amer. Water Works Assoc., 1939, 31, 2121.
3. RUYS, J.D., Drinkwater-reiniging met Hypochlorieten, 1914.
4. HOLWERDA, K., Mededeelingen van den dienst der Volksgezondheid in Nederlandsch-Indie,
1930, 19, 325.
5. GERSTEIN, H.H., J. Amer. Water Works Assoc., 1931, 23, 1334.
6. PALIN, A.T., Ph.D. Thesis, University of London, England, 1949.
7. PALIN, A.T., J. Inst. Water Engrs, 1949, 3, 100.
8. WILLIAMS, D.B., J. Amer. Water Works Assoc., 1949, 41, 441.
9. BUTTERFIELD, C.T. and WATTIE, E., Public Health Reports, 1943, 58, 1837; 1946, 61, 157.
10. BUTTERFIELD, C.T., Public Health Reports, 1948, 63, 934.
11. SNOW, W.B., J. Amer. Water Works Assoc., 1956, 48, 1510.
13. CIFEC (Cie Industrielle de Filtration et d’Equipment Chimique S.A.), Chlorine Dioxide
Generator: The French Method with Enrichment Loop, Paris, Nov., 1976, Notice No. 167.
Disinfection 141
14. US Environmental Protection Agency, Ozone, Chlorine Dioxide and Chloramines as

Alternatives to Chlorine for Disinfection of Drinking Water, Revised April 1978, Cincinnati,
USA.
15. UK Department of the Environment, Analysis of Raw, Potable and Waste Waters, 1972,
London, England.
16. American Public Health Association, American Water Works Association and Water Pollution
Control Federation, Standard Methods for the Examination of Water and Wastewater, 14th edn,
1975, Washington, DC, USA.
17. PALIN, A.T., J. Amer. Water Works Assoc., 1957, 49, 873.
19. Wilkinson and Simpson, Ltd, Tablet Tests and Reagents for Water Analysis and Treatment
Control, 2nd Edn. 1980 (in preparation), Gateshead, England.
Chapter 6
SLUDGE TREATMENT AND DISPOSAL
M.A.HILSON, C.Chem., F.R.I.C., F.I.W.E.S.
Principal Scientist, Water Treatment and Supply, North West Water
Authority, Warrington, UK
SUMMARY
The various types of sludge arising from water treatment processes are considered and
this is followed by a consideration of disposal methods. The methods of water treatment
plant operation aimed at producing the minimum volume of sludge for subsequent
treatment are then discussed. The dewatering of sludge is considered as two separate
processes: initial thickening or increase in concentration to minimise the size of
treatment equipment required followed by a review of the various methods available for
further dewatering. Recently introduced water treatment processes are then considered
with particular reference to their impact on sludge treatment. Quantities of sludge
arising and operational procedures for sludge treatment installations are then dealt with.
Much of the material in this chapter is drawn from the experience of the author and
his close colleagues and as such is perhaps inclined towards the problems of the
treatment of sludge arising from peaty upland waters, but attempts have been made to
point out where differences may arise when dealing with sludges produced from waters
of various types.
6.1. INTRODUCTION
Sludge is the waste product produced in a water treatment process and as such is a
material of little or no intrinsic value. It is produced in large quantities and presents the
industry with a very significant disposal problem. Increasing interest in environmental
matters together with pressures on land use and concern for public safety have escalated
the problems of waste disposal.
The term sludge can be taken as describing any suspension of solid material in a
liquid. In the context of water treatment processes the liquid phase of the sludge is
invariably aqueous whilst the solid phase will consist of any materials derived from the
raw water together with the residues of any chemicals added in the treatment process.
It is apparent that in the past insufficient thought was given to the disposal of waste
products at the design stage of a water treatment plant, the problem being left in abeyance
pending operational experience. This philosophy has resulted in large areas of land in the
vicinity of water treatment plants being given over to sludge-holding lagoons. Such areas
ultimately become filled with gelatinous sludge which presents a safety hazard as well as
a visual eyesore and merely postpones the day when a viable sludge treatment and
Sludge treatment and disposal 143
disposal system has to be provided. Indeed in the long term such an interim scheme
compounds the disposal problems in that any new process for disposal must be capable
not only of dealing with the sludge produced from the treatment plant but also capable of
gradually dealing with the old sludge from the lagoons to enable such areas to be made
safe and put to a more productive use.
It is therefore recommended that sludge treatment, and where practicable wastewater
recycling, should be an integral part of the plant design. Since the costs of sludge
treatment can amount to up to 17% of the total costs of the water treatment process it is
apparent that sludge treatment should be worthy of rigorous attention to detail during the
design stage of a treatment plant.
The objective of a sludge treatment process should be to separate the liquid and solid
components in a manner in which both are capable of re-use or disposal. In other words
the liquid component must be of such a quality that it can be either recycled or discharged
to a watercourse and the solid component reduced to a form in which it is readily
transportable for disposal and which will not, as a result of such disposal, give rise to a
hazardous environment.
6.2. TYPES OF SLUDGE
The properties of different sludges will vary considerably dependent upon their origin
and the type of treatment applied to the raw water. Sludges can conveniently be grouped
into two types:
(1) Sludges produced without chemical treatment.
(2) Sludges produced from a chemical treatment process.
6.2.1. Sludges Produced Without Chemical Treatment
Primary Sedimentation Sludges

In general such sludges will consist of the coarser solid materials, such as sand and coarse
silt, settled from the raw water. As such they should be relatively free draining and
present few or no problems of separation into their two components. Sludges of this type
will generally accumulate in channels and sumps at the inlets to treatment works.
Washwater from Microstrainers and Filters used Without Coagulant

When a raw water is of an acceptable colour but contains small amounts of suspended
matter such as algae it is often possible to remove this suspended material by either
microstraining through fine wire mesh fabric or by filtration through sand filters without
the use of coagulant. Such processes are often used prior to slow sand filtration. The
washwater produced will contain algal remains and some larger particulate matter. The
washwater from a microstrainer installation will generally be of quite low solids content
since the backwashing of such installations is carried out on a continuous basis.
Washwater from rapid gravity sand filters used without coagulant can, however, contain
quite high concentrations of suspended solids since the backwashing will be carried out at
time intervals depending on filter performance and solids load in the raw water. The
batches of backwash water when they arise will therefore contain the accumulated
suspended solids from a considerable volume of throughput water.
Effluent from Slow Sand Filter, Sand-Washing Plants

When slow sand filters become partially blocked by solid matter in the upper layer of
sand, this sand is skimmed off and washed free from debris by some mechanical process.
The suspended solids concentration of the effluent from the sand-washing process can be
quite high and will normally consist of organic material from the removal of algae, the
hydrated oxides of certain metals present in the raw water together with fine sand
particles.
6.2.2. Sludges Produced from a Chemical Treatment Process
Sludges Produced by Coagulation

Sludges produced from water treatment processes utilising coagulation by means of
aluminium or iron salts can vary considerably depending on the chemicals used and the
type of treatment plant. The suspended solids concentration of backwash water from sand
filters employed after coagulation whether they be pressure filters in use as a single stage
treatment, or rapid gravity filters used after sedimentation, will usually be within the
range 100–500 mg/litre. The sludge obtained from desludging sedimentation tanks will
vary considerably in suspended solids content. Given efficient management of desludging
a concentration of 0·5% (5000 mg/litre) should be attainable from plants treating soft
upland water whilst higher concentrations may be obtained from plants dealing with
lowland river waters. Whatever the coagulant used there is a physical and chemical
bonding of water to the sludge particles which makes dewatering difficult.
Sludges Produced by Precipitation Softening

The quality of sludges will vary widely, ranging between pure calcium carbonate,
through mixtures of calcium carbonate and magnesium hydroxide, to those containing
suspended matter from the raw water mixed with the softening products together with the
residues of any coagulant used in the process. Of these sludges pure precipitated calcium
carbonate may have some commercial value to the pharmaceutical industry. Other
sludges in this category may find application for treating agricultural land in place of
ground limestone.
Softening sludges are normally of a higher solids concentration than coagulant sludges
and, being more particulate, they are more easily dewatered.
Sludges Produced by Iron and Manganese Removal Plants

Sludges derived from such plants generally have similar characteristics to coagulant
sludges.
6.3. CHOICE OF METHOD FOR DISPOSAL
Apart from the case of the softening sludges mentioned above, which form a very small
proportion of the total water treatment sludge production, the disposal method usually
applied is one of ‘throwing away’. This can be achieved either by discharge to sewer,
possibly after some pre-treatment or transport direct from the waterworks to a suitable tip
after local treatment.
Discharge to sewer is a convenient method of disposal if sewers of sufficient capacity
are available in the vicinity of the water works. This is often not the case as the majority
of water treatment works tend to be removed from centres of population and hence from
sewers of significant size. Consideration of sludge disposal should be one of the factors
taken into account at the design stage of a new waterworks and location of the works
relative to the sewerage system should receive due consideration at this stage. Provision
of a suitable sewer for sludge disposal should be the subject of an economic appraisal
against the costs of sludge treatment on the waterworks site followed by transport to a
suitable tip. Discharge of waterworks sludge to sewer will of course only be possible if
the treatment processes employed at the sewage works are amenable to the reception of
sludge of this type and the necessary hydraulic capacity is available. The subsequent
sludge treatment and disposal practice at the sewage works will also play a part in
determining whether or not waterworks sludge is acceptable. If the sewage works sludge
is disposed of, after suitable treatment, to agricultural land then it may be that the
introduction of a considerable inorganic component to the sludge, from the coagulant
residues, would render this method of disposal less attractive to the agricultural industry.
Much depends on the proportion of the sewage works capacity taken up by the
waterworks sludge. Another factor which must be taken into account when considering
this method of disposal is the avoidance of deposition of solids within the sewer. This
should not be a serious problem in the case of coagulant sludges since the density
difference between the suspended matter and the water carrier vehicle will be quite small.
It may also be necessary to avoid intermittent ‘slug’ disposal because of possible effects
on the processes at the receiving sewage works. If this is the case then the provision of
suitably sized holding or balancing tanks at the waterworks together with means of
maintaining the sludge in suspension will be another economic factor which must be
allowed for in the appraisal of sludge disposal strategy.
Discharge of waterworks sludge to sewer is not therefore an end in itself; it is merely
transferring the disposal problem from one place to another.
If the sludge has to be disposed of to tip after suitable treatment at the waterworks then
the problem resolves itself into selecting a suitable method of dewatering. The degree of
dewatering required is that which produces a solid component which is suitable for
tipping. The process should seek to strike a balance between disfigurement of the
environment and the economics of disposal. It is not economic to produce a solid of
lower moisture content than is necessary at a high cost, nor is it desirable to incur
transportation costs of large quantities of water. The objective should be to produce a
solid component which when placed on a tip will not revert to a gelatinous condition and
which has sufficiently high load bearing characteristics to enable the tipping area to be
reclaimed by the application of topsoil followed by reseeding.
As has been mentioned earlier, in the case of coagulant sludges, there is a physical and
chemical bonding of water to the sludge particles which makes dewatering difficult.
Indeed the ultimate dewatering that can be achieved for a sludge produced from a soft
upland coloured water treated with aluminium sulphate as coagulant is only about 25%
solids w/w. However at this concentration even with a 75% moisture content the sludge
cake is quite dry and hard to the touch and indeed its ‘feel’ resembles that of a piece of
softwood. The presence of increased amounts of suspended solids in the raw water, as
would be the case when treating water from lowland rivers, will enable sludge cakes of a
somewhat higher solids content to be produced. It has been found in practice that, for the
case of sludge produced from soft upland waters of negligible turbidity, if the solids
content of the sludge cake exceeds 18–20% then there will be no tendency for the sludge
to revert to a gelatinous nature on exposure to the atmosphere even after heavy rainfall.
Sludge cake of 20–25% solids content will, after a few weeks atmospheric exposure, lose
additional water by evaporation to produce a dry friable solid which is quite suitable as
land-filled material.
It is worth noting that in the United Kingdom a site licence may be required under the
Control of Pollution Act 1974 for the tipping of waterworks sludge.
6.4. DEWATERING OF WATERWORKS SLUDGE
The capital cost of any sludge dewatering process is influenced largely by the quantity of
sludge to be handled. There are great economic advantages if the sludge can be thickened
and hence the volume reduced as much as possible prior to subsequent processes. The
upper limit of concentration to which a given sludge can be thickened is governed by the
flow characteristics of the sludge. The sludge should be thickened to the maximum
concentration at which it is still capable of being pumped. This concentration will
normally be not less than 4% in the case of sludges produced from coloured upland
waters of low turbidity and could be higher for sludges produced from waters containing
appreciable levels of suspended solids. Dewatering of waterworks sludge can therefore be
considered as falling into two separate stages, thickening followed by further dewatering.
It is important that any sludge thickening process employed must not only result in the
production of a sludge of the desired concentration but must also produce a supernatant
water of a quality suitable for either recycling through the water treatment plant or
discharge either to drain or a local watercourse.
Before considering methods of thickening and dewatering it is worth considering the
way in which the sludge to be handled is produced from the water treatment plant and
what operational practices can be used to minimise the subsequent problems of sludge
dewatering. As has been mentioned earlier when considering the various types of sludge,
the sludge produced from sedimentation tanks is at a considerably higher concentration
than that arising from the backwashing of filters. Even when treating coloured upland
waters of low turbidity it should be possible to operate the sludge bleeds from modern
sedimentation tanks to yield a sludge of 0·5% solids concentration. If this is not being
achieved then attention should be given to the desludging operation to increase the sludge
bleed concentration to the maximum possible level. Whilst it is perfectly feasible to
separate the solids from filter backwash water and then treat these solids along with the
sedimentation tank sludge this can involve, in effect, the operation of two separate sludge
separation and thickening processes which can increase both capital and operational
costs. It is better where possible to use the existing structures of the water treatment plant
to achieve as much sludge thickening as possible. This can be done by arranging for the
dirty filter wash water to be collected in a holding tank and returned to the treatment plant
inlet at a uniform rate without any separation of the solid material. Filter wash water
return can assist the coagulation and flocculation of the incoming raw water by providing
additional solids within the clarifier, this being particularly useful when treating waters of
low turbidity. The advantage of this method of operation is that the only sludge arising
for treatment is that arising from the sludge bleeds of the clarifier and the subsequent
processes are simplified by only having to deal with what is in effect a single point source
of sludge of a reasonably consistent concentration. This method of operation can, of
course, only be used when the water treatment plant is based on a two-stage process of
sedimentation followed by filtration. It is also applicable to the operation of precipitation
softening plants although in this case, in view of the normally much larger quantities of
solid material involved, care must be taken that the desludging arrangements of the
sedimentation tanks are not overloaded by the return of the filter backwash water,
resulting in possible blockages. In the case of single-stage water treatment plants
consisting of filtration of coagulated raw water there is no option but to regard the
starting point for sludge treatment as being the much larger quantity, often 3–5% of the
plant throughput, of dirty filter wash water of low suspended solids concentration.
6.5. SLUDGE THICKENING
The most widely used method of sludge thickening is by slow stirring. Work carried out
over twenty years ago1 showed that the sludge produced from the treatment of a coloured
upland water could be thickened to about 2·5–3·0% solids by this method although the
quality of the supernatant water from such thickening was often poor. The introduction of
synthetic polyelectrolytes during recent years has had a great impact on sludge thickening
both in terms of the solids concentration attainable and the quality of the supernatant
water for disposal.
In the present state of knowledge of the theory of the action of polyelectrolytes it is
not possible to scientifically define an application and thereby select the best product for
the purpose. It is more a case of trying a range of products and selecting the best one of
that range for the job in hand. If it is intended to use a polyelectrolyte for sludge
thickening and the resultant supernatant water therefrom is to be discarded, then a wide
range of polyelectrolytes is available from which the ultimate selection can be made. If it
is intended to recover the supernatant water then the choice of products will be restricted
to those products which have been approved on toxicity grounds, and even then the dose
which can be applied to the sludge may be governed by the maximum recommended dose
which may be applied to the water being treated for potable purposes. This is particularly
so where a polyelectrolyte is also being used as part of the treatment process.
The efficiency of action of any polyelectrolyte depends on the rapid and complete
dispersion of the small dose applied into the bulk of the ‘dirty’ water or sludge being
treated. This is necessary because the molecular chains of the polyelectrolyte, particularly
those of high molecular weight, appear to have a voracious appetite for the solid sludge
particles. The rapid and thorough dispersion is necessary to ensure that all sludge
particles are equally exposed to the action of the polyelectrolyte. If the dispersion is not
rapid then those sludge particles which first encounter the polyelectrolyte molecules
become, in effect, overdosed whilst the remainder are underdosed. The nett result of this
is that the full action of the polyelectrolyte is not realised since overdosing is equally
detrimental to the effect being sought as is no dosing at all. In order to assist this rapid
mixing and dispersion the polyelectrolyte is usually dosed to the sludge as a very dilute
solution of about 0·1–0·2% strength.
The problems associated with and the benefits obtainable from the use of
polyelectrolytes can perhaps best be illustrated by a description of and comments on their
use for sludge thickening at two water treatment plants of the former Fylde Water Board
which is now part of the North West Water Authority.
The Fishmoor plant treats 32 Ml/d by means of sedimentation and rapid gravity
filtration. The dirty filter backwash water is returned to the plant inlet at a uniform rate
and the sludge bleeds from the sedimentation stage give about 0.1 Ml/d of sludge
containing 0·6% solids as the starting point for treatment. The Stocks plant is a pressure
filter installation which treats about 100 Ml/d and can produce up to 5 Ml/d of dirty filter
wash water containing 0·03–0·04% solids. Exhaustive laboratory tests showed that
treatment with the same anionic polyelectrolyte of high molecular weight would give
improved sludge thickening and supernatant water quality at both plants. Although the
same polyelectrolyte is in use at both Fishmoor and Stocks its method of application, in
order to achieve rapid dispersion, is quite different.
The original sludge treatment installation at Fishmoor consisted of two 9 m diameter
circular sludge thickening tanks equipped with picket fence stirrers. These two tanks
were arranged in series in order to perform a two-stage thickening process. The sludge
bleeds from the sedimentation tanks fed sludge on an intermittent basis into the upper
tank, the discharge being by means of a submerged weir. The supernatant water from this
tank was returned to the plant whilst the sludge from the bottom was transferred by
gravity flow through an| underground pipeline to the lower tank for further thickening.
Some sludge thickening occurred in the upper tank and the sludge passed to the lower
tank with a solids concentration of about 0·8–1·0% solids. The laboratory tests had shown
that the use of polyelectrolyte could be expected to improve the thickening to such an
extent that the capacity of the lower tank would be sufficient to achieve the degree of
thickening required and the upper tank could be used as a balancing tank to iron out
variations in flow and sludge solids concentration arising from the clarifier sludge bleeds.
The problem therefore became one of injecting polyelectrolyte into the sludge as it
passed between the two tanks. It was felt that injection of polyelectrolyte solution into the
pipeline between the two tanks would not give the required rapid dispersion and mixing;
experimental work was therefore undertaken which led to the system which has been in
use for the last 14 years. The sludge is dosed with polyelectrolyte on a batch basis in 455-
litre batches. Sludge is drawn from the upper tank by means of a pump on the time
control to deliver 455-litre quantities into a mixing tank. This mixing tank is a circular
tank of 1·2 m diameter with a conical bottom section leading to an outlet gate valve and
discharge pipe. The mixing tank is fitted with a 130 mm propeller-type stirrer which
operates at 960 rpm. The polyelectrolyte as a 0·15% solution is dosed into the sludge
from a header tank the outlet of which is controlled by a solenoid valve. The dose of
polyelectrolyte added is adjusted by varying the time for which the solenoid valve is
opened. When the 455-litre quantity of sludge is in the mixing tank the stirrer starts up
and after 5 s the solenoid valve opens to allow the preset volume of polyelectrolyte
solution to run into the stirred sludge. After the solenoid valve closes the stirring
continues for a further 5s, the stirrer then stops and the outlet gate valve opens, the dosed
sludge being discharged by gravity to the lower thickening tank; this discharge takes
place via submerged weir below the surface of the supernatant water in the lower
thickening tank.
In this process the rate and duration of stirring are critical. It is important to stir the
sludge prior to polyelectrolyte addition to ensure homogeneity and to continue stirring
only until the polyelectrolyte solution added is completely dispersed in the sludge. If the
stirring is continued for more than 5 s after the completion of dosing then the large floc
particles which are formed rapidly by the polyelectrolyte are broken down by the
shearing action of the stirring and only reform slowly once the sludge has been
discharged, thereby aflfecting the subsequent thickening. For similar reasons it is also
important that the dosed sludge be discharged below the water surface of the lower
thickening tank; any free fall discharge at this stage again adversely affects the
thickening. Experiments were carried out with lower stirring speeds in the mixing tank
but these did not give results as good as those obtained using the 960 rpm stirrer. It would
appear that a stirrer speed of 960 rpm is ideally suited to size of propeller and mixing
tank for the polyelectrolyte used and the sludge produced at the Fishmoor plant. It has
been found in this process that the solids content of the final sludge produced from the
bottom of the lower stirrer can be varied by varying the dose of polyelectrolyte applied.
The dose of polyelectrolyte required to yield a final sludge of a given solids content is
proportional to the solids content of the initial sludge being treated. In order to produce a
final sludge of 5% solids content it is necessary to inject polyelectrolyte at the rate of
1·5mg of polymer for each gram of dry solids in the initial sludge. The sludge thickening
by this process appears to be largely unaffected by the temperature of the sludge. The
sludge thickening and polyelectrolyte dosing arrangements at the Fishmoor plant are
shown diagrammatically in Fig. 1.
The above may appear at first sight to be a complicated method of achieving a rapid
and complete dispersion of the polyelectrolyte solution in the sludge to be treated;
nevertheless it has, over the years, been a method which has proved outstandingly
successful at the Fishmoor plant and particularly suited to the configuration of the
installed sludge thickeners.
On new treatment plants it is usual to design the sludge thickening facility so that
polyelectrolyte solution is injected into the sludge at a point of high turbulence, such as
the passage over a weir, in an effort to achieve the desired rapidity and thoroughness of
mixing thus avoiding the expense of the installation of a batch dosing system such as that
described.
At the Stocks plant the dirty filter backwash water is passed to two horizontal flow
sedimentation tanks. In this case the polyelectrolyte solution is injected into the main
carrying the dirty wash water to the sedimentation tanks at a dose of about 1·5–2·0
mg/litre. There is sufficient turbulence in this main and at the submerged bell-mouth inlet
to the sedimentation tanks to achieve the required degree of and rapidity of mixing. The
thickened sludge is withdrawn from the bottom of the primary section of these tanks. A
great deal of experimental work was carried out into sludge thickening by slow stirring at
this plant during the 1950s. This work and a description of the horizontal sedimentation
tanks has been described by Doe.1 These two sedimentation tanks had been installed
originally when the throughput of the plant was 32 Ml/d and proved capable of dealing
with the wash water arising from an increased throughput up to 79 Ml/d. By the latter
half of the 1960s the plant throughput had further increased and the wash-water
sedimentation tanks, even when operated in parallel, were becoming so overloaded that
the sludge thickening (to 2·5–3·0% solids) was being impaired and the supernatant water
quality passing to the river was outside the standard required by the local river authority.
The position was becoming so serious that consideration was being given to the
construction of a third and possibly even a fourth sedimentation tank. The advent of the
use of polyelectrolyte so changed the position that one of the original sedimentation tanks
is now able to cope with the wash water arising from a plant throughput of over 100 Ml/d
whilst achieving an improved sludge thickness (4·5% solids) and greatly improved
quality of supernatant water.
During the period September 1972 to October 1974 the then Water
FIG. 1. The sludge thickening and

polyelectrolyte dosing arrangements at
the Fishmoor water treatment plant.
Research Association carried out extensive pilot-scale investigations at the Stocks plant,
the results of which have been published by the Association’s successor, the Water
Research Centre.2 It is interesting to note that virtually all the conclusions arrived at
during the original investigations and experiments carried out at the plant over a period of
years were verified during the above pilot-scale experiments.
A seminar was held at the Water Research Centre early in 1978 on: Alum Sludge
Disposal, Current Practice and Trends. At this seminar four papers were discussed
including one on ‘Research in Alum Sludge Treatmen’ by the Water Research Centre
staff. The proceedings of this seminar have been published3 and the research at the Centre
into sludge treatment is continuing.
Whilst it must be realised that there will be differences in the thickening
characteristics of different sludges, it is interesting to note the variation in dose of the
polyelectrolyte required to produce a thickened sludge of similar solids content at the two
plants which both treat waters of an essentially similar type but using different mixing
techniques. To yield a final sludge of 4.5% solids the polymer dose used at Stocks is of
the order of 4·5 mg of polymer per g of dry sludge solids whilst at Fishmoor it is only 1·5
mg/g.
6.6. THEORETICAL BASIS OF SLUDGE DEWATERING
Before giving detailed consideration to the various methods available for further
dewatering of coagulant-based waterworks sludge it is necessary to consider the
mechanism of such dewatering which is applicable to most of the processes to be
considered. Most of these processes involve some form of filtration of the sludge and it is
only during the very early stages of such filtration that the filtering medium has a
controlling effect on the rate of dewatering. After a very short time the filter medium
becomes blinded by a layer of the sludge and it is this increasingly thick layer of solids
which governs the subsequent rate of dewatering. The rate of dewatering by filtration is
therefore governed by the properties of the sludge itself, and is virtually independent of
the characteristics of the filter medium provided such medium is capable of retaining a
layer of sludge which is of course necessary if the quality of the filtrate is to be
acceptable. The dewatering takes place according to the following equation, which is
derived from the Carman-Kozeny and D’Arcy equations:
where V=volume of filtrate; P=applied pressure; S=compressibility index; A=area of

filter; t=time; η=viscosity of filtrate; r0=specific resistance of sludge at unit pressure; and
C=initial concentration of solids in sludge.
It is apparent from this equation that the sludge properties which control the
dewatering process are S the compressibility index and r0 the specific resistance of the
sludge at unit pressure. These properties are readily determinable on a small scale in the
laboratory.
The apparatus used for this laboratory determination is shown in Fig. 2. As mentioned
above the grade of filter paper used does not matter, since it plays no part in the
dewatering except in the very early stages. Glass fibre filter paper is used because of its
wet strength.
The procedure used is as follows:
1. Place a filter paper in funnel G and make sure it is thoroughly bedded down by
drawing a little distilled water through.
2. Assemble the apparatus as shown in Fig. 2 with valve B and tap E both closed.
3. Turn on the water to vacuum pump A at a rate in excess of that required to obtain the
desired degree of vacuum.
4. Open valve B until the required vacuum shows on the gauge C (allow about 5 cm
higher vacuum than is desired for the experiment). Close valve B. If all connections
are a good fit, the vacuum should hold constant for the duration of the experiment. It is
undesirable to attempt to adjust the vacuum by manipulation of the valve B during the
experiment, since slight changes in vacuum produce a marked momentary effect on

the rate of filtration.
5. Place the sludge in the funnel G.
6. Open tap E and start a stop watch. (The vacuum will fall slightly.)
7. Note the volume of filtrate collected at minute intervals for 10 minutes or up to the 50
ml mark, whichever is the less. Whilst taking these readings keep the level of sludge in
the funnel G topped up to a reasonably constant level, in order to simulate a constant
feed of sludge to the filter.
The filtration rate is normally determined at four different pressures, i.e. levels of
vacuum. It is important that the units used are systematic and it is recommended that the
cgs system is employed, i.e. V=volume in ml; P =pressure in g/cm2; A=area of filter in
cm2; t=time in s; η=viscosity of filtrate in poise (see below); and C=initial concentration
of sludge in g dry solids per g sludge.
FIG. 2. Apparatus for determining the

compressibility index and the specific
resistance of sludge.
A=water-driven vacuum pump
B=Saunders valve
C=vacuum gauge
D=vacuum reservoir (500 ml)
E=glass stopcock
F=50 ml graduated measuring cylinder
G=Buchner funnel to take 5·5 cm filter paper.
The viscosity of the filtrate will vary with the temperature of the sludge.
Temperature of sludge Viscosity of filtrate
(°C) (poise)
0 1·79×10−2
5 1·52×10–2
10 1·31×10−2
15 1·14×10–2
20 1·01×10–2
The value of S and r0 are obtained from the experimental data by means of the following
treatment:
therefore
log(V2/t)=(1−S) log P+log(2A2/ηr0C).
A plot of V2. against t for each pressure examined gives gradient g. A plot of log g against
log P gives a straight line of gradient (1−S), hence S is determined.
This straight line gives the value of log(2A2/ηr0C) from its intercept on the y axis at
log P=0 (i.e. P=1), therefore r0 can be calculated.
The conditioning of a sludge prior to dewatering in order to increase the rate of
dewatering involves treatment by polyelectrolyte or other means to alter the values of S
and r0 and so the above experiment yields valuable predictive information on a laboratory
scale on the benefits likely to be obtained on a full scale.
Details of the above experiment have been described by the author4 along with other
information on the conditioning of sludge by means of polyelectrolyte. A simpler method
of predicting changes in the dewatering characteristics of sludge is by means of the
capillary suction time method developed at the then Water Pollution Research
Laboratory,5 but this method has the disadvantage that it is performed at a single pressure
equivalent to the capillary suction pressure of the paper being used and hence does not
take account of changes in the compressibility index S with pressure, which can be
important when the sludge is to be dewatered at elevated pressures as in a filter press.
6.7. METHODS OF FURTHER DEWATERING
There are six principal methods of further dewatering of coagulant based sludges, each of
which is examined below.
6.7.1. Drying Beds

A sludge drying bed is a development of a simple sludge lagoon in that it incorporates a
system of underdrains overlain by a permeable filtration medium to facilitate drainage
and dewatering of the sludge. Many different materials have been proposed and used as
the permeable filtration medium in drying beds, the main ones being sand, ‘no-fines’
concrete and specially woven wire fabric known as ‘wedge-wire’. Whichever permeable
medium is used it is essential that it should not become blocked by the sludge particles. If
blockage does take place then the drying bed merely acts as a lagoon and the only
dewatering achieved is that which takes place by evaporation. This system has been
investigated by Gauntlett and Packham6 who conclude that drying beds of suitable
construction provide an effective means of drying waterworks clarification sludge during
those months of the year when evaporation proceeds at a reasonable rate. They also stated
that drying proceeds at a very slow rate indeed during the winter months. From this it is
apparent that evaporative loss of water plays a significant part in the sludge drying and as
such it is not a process which is of much use in temperate climates such as the UK but
may be of use in the hotter and drier parts of the world.
6.7.2. Rotary Vacuum Filtration

The rotary vacuum filter consists of a large drum or cylinder the circumference of which
is covered with a suitably fine filtration fabric. The drum is mounted in a frame and
motor driven so that it rotates slowly about its cylindrical axis. A partial vacuum is
applied to the interior of the drum, the lower part of which passes through a trough into
which the sludge is fed. The partial vacuum causes a thin film of sludge to adhere to the
filtration fabric as it passes through the sludge in the trough. Water is drawn from this
film of sludge, by the vacuum, during the remainder of the rotational cycle. The water
drawn into the centre of the drum is collected in a drainage trough and hence to waste
whilst the dewatered sludge on the outside of the drum is scraped from the fabric by
means of a horizontally mounted knifeedge just before it re-enters the sludge trough.
This type of machine has been established for many years and has proved useful for
the dewatering of many types of sludge. It suffers however from the fact that the
maximum pressure which can be used in the dewatering process is limited to the level of
vacuum which can be maintained in the interior of the drum which is of course something
less than one atmosphere. Additionally the maximum time which can be allowed for
dewatering to take place is the time taken for that part of the revolution of the drum
between leaving the sludge trough and the sludge being scraped off the fabric, i.e. it is
governed by the rotational speed of the drum. Difficulties can also be encountered in the
selection of a suitable filtration fabric capable of supporting the sludge to be dewatered
without allowing an excessive quantity of fine sludge particles to pass through with the
filtrate. This can be overcome by the use of a precoat material, which in effect is a
material the particles of which are of a size and shape that are retained on the filter fabric
and then act as the support for the sludge particles. The precoat is applied by allowing the
circumference of the drum to pass through a trough containing a suspension of the
precoat material immediately prior to the sludge trough. The use of a precoat adds to both
the cost of dewatering and the quantity of sludge for ultimate disposal.
The rotary vacuum filter has found the widest application in the water supply industry
for the dewatering of sludges of the more particulate type such as those produced in water
softening processes. Because of the limitations mentioned above it has not found great
favour for the dewatering of coagulant-based sludges.
6.7.3. Belt Pressing

During recent years a number of pieces of sludge dewatering equipment have become
available which can best be described as belt presses. In some respects they resemble
vacuum filters but are provided with additional means of achieving a greater level of
dewatering than is possible on a normal vacuum filter.
They consist of a belt of fine filtration fabric of considerably greater length than the
circumference of a rotary vacuum filter. The belt passes through a trough of sludge or is
fed with sludge from above and passes over a drum or chamber in which a partial vacuum
is established. This results in a film of sludge adhering to the belt which is further
dewatered by passing between a series of rollers at gradually increasing pressures so that
further water is squeezed out of the sludge. The limitations of pressure and dewatering
time inherent in the rotary filter are thereby alleviated to some extent but the problem of
selecting a suitable filtration fabric for the belt remains, if the use of a precoat material is
to be avoided. Such belt pressing equipment has found more application in dealing with
the more fibrous type of sludges produced from sewage than with coagulant-based
sludges.
6.7.4. Filter Pressing

The filter press or plate press has been in use for many years in various process industries
as a means of dewatering suspensions of solids in liquids. It comprises a multiplicity of
recessed plates the surfaces of which are covered by a detachable filter cloth; each plate
has a hole of about 100 mm diameter in the centre which forms the sludge inlet system to
the press. The plates are mounted on a heavy cast-iron frame either slung from a top
beam or supported on side bars. The frame is of such a length that the plates can be
moved successively along their supporting beam to enable the pressed sludge cakes to be
removed, and is fitted with hydraulic or mechanical equipment to enable the plates to be
held tightly together during the dewatering cycle.
The principle of operation is that the press is closed up and sludge is pumped into the
central core and hence into each chamber formed between adjacent plates. The solids in
the sludge are retained in the chamber to form a sludge cake whilst the water passes
through the filter cloth and drains away through channels in the plate itself. Plates can be
obtained with different depths of cavity to yield sludge cakes of different thicknesses, the
normal thickness being 25–30 mm. Filter presses normally operate at pressures of up to
6·7 bar, but equipment working at even higher pressures is now available.
It does not necessarily follow that the dewatering is more rapid if a higher pressure is
used. This depends to a large extent on the compressibility index of the sludge in question
and its effect on the rate of filtrate production according to the Carman-Kozeny equation.
The cake thickness chosen for a filter press installation will again depend on the
dewatering characteristics of the sludge itself and should be selected to give the desired
degree of dewatering in a reasonable press cycle time.
Many coagulant-based sludges in the past did not prove amenable to filter pressing,
merely producing a sludge cake with a hard outer skin and a soft centre. To overcome
this problem the selected cake thickness would have had to be so small as to render the
number of plates required for a given sludge throughput excessive and beyond the bounds
of economic feasibility. The advent of polyelectrolytes with their ability to so condition
coagulant-based sludges as to vastly improve their dewatering characteristics has meant
that filter presses are now an economic proposition for even the most intractable of
sludges. There is no doubt that filter pressing, allowing as it does the optimisation of the
variables of cake thickness, applied pressure and cycle time, offers one of the most
economical and controllable methods of dewatering coagulant-based sludges produced by
water clarification processes. Several papers have been written on the development of
filter press systems for this purpose7,8 and the reader is recommended to refer to them for
further practical details.
It has been claimed in the past that the addition of lime to the sludge will considerably
improve the dewatering characteristics and enable shorter pressing cycles to be achieved.
Whilst this is undoubtedly true it aggravates the disposal problem since the addition of
lime increases the quantity of sludge to be disposed of to tip and the increased pH value
of the sludge can result in poorer filtrate quality.
6.7.5. Centrifuging
In this process the sludge is fed, after suitable conditioning with a polyelectrolyte, into
one end of a drum which is spinning at a rate of several thousand revolutions per
minute—the principle being that the solids and water are separated by centrifugal force,
the solids adhering to the wall of the drum and the water overflowing a ‘beach’ at the far
end of the drum. The solids are scraped from the wall by a scroll revolving at a slightly
slower speed than the drum itself and directed to an outlet chute. Several experiments and
full-scale operations have been reported, claiming various degrees of success.3,9 The
process has the advantage that it is a continuous rather than a batch process, but it suffers
from the fact that it depends for efficient separation of solid and liquid on difference in
density between the solid and liquid phases which is often very slight in the case of
coagulant-based waterworks sludges. This tends to limit the degree of dewatering
attainable to about 14–17% solids content and often produces an aqueous phase of a
quality unfit to discharge to a watercourse because of its high residual suspended solids.
In spite of these limitations centrifuging can prove an economically attractive process
for dewatering certain types of waterworks sludges.
6.7.6. Freezing
The dewatering characteristics of a gelatinous, coagulant-based water-works sludge can
be dramatically altered by freezing and thawing. If the sludge is frozen slowly to a solid
block and then allowed to thaw, the solids are changed to a granular nature which allows
the water to drain away freely when deposited on a tip provided with underdrains. The
solids so deposited have sufficiently high load bearing characteristics to enable the tip to
be reclaimed by topsoiling and seeding.
The freezing must proceed slowly if the required benefits of the process are to be
obtained; flash freezing does not produce the desired results, for some reason. If the
process is carried out correctly and completely the filtrate from the thawed sludge is of a
very good quality; if, however, the freezing is not complete and small pockets of sludge
remain unfrozen then the filtrate quality is greatly impaired.
A great attraction of the freeze-thaw process is that the sludge remains transferable by
pipeline through to the tip, where the final dewatering takes place, thus avoiding the
operational complexity of solids handling equipment. Apart from the use of a
polyelectrolyte for sludge thickening to minimise the size of freezing plant required, the
process does not depend on any sludge conditioning, working equally well on sludges
whether or not they contain a polyelectrolyte.
The freeze-thaw process has been developed to full-scale operational plant at three
waterworks locations in the UK, but owing to increasing running and maintenance costs
has been discontinued at two of those locations where the advent of polyelectrolytes for
sludge conditioning render filter pressing both feasible and more economically
attractive.10
6.8. OPTIMISATION OF WATER AND SLUDGE TREATMENT

PROCESSES
The foregoing has been mainly concerned with the handling and treatment of sludge from
existing or ‘conventional’ water treatment processes. There are however certain water
treatment processes which have been developed in recent years which, whilst not
necessarily reducing the quantity of sludge, do in fact minimise its handling problems by
producing the sludge from the water treatment process at a higher concentration and
thereby simplifying subsequent processing. The two most notable of such recent
processes are flotation and lamella sedimentation.
6.8.1. Flotation
The flotation process as applied in water treatment is normally the dissolved air flotation
process of solid-liquid separation. In this process the suspended particles produced by
coagulation and flocculation of the raw water are induced to rise to the surface by the
attachment of minute air bubbles. The flotation process is carried out in a tank equipped
with mechanical means of skimming the floating sludge layer from the surface of the
clarified water. The skimming process can be either on a continuous or an intermittent
basis. Continuous removal of the sludge from the surface of the tank, by whatever means,
does not provide any benefits in the way of simplifying the sludge treatment and disposal;
it may indeed worsen the problem since the sludge will be at a very low solids
concentration and a thickening process capable of dealing with quite a high throughput
will be required. If, however, the sludge skimming is intermittent and the layer of sludge
can be allowed to remain on the surface of the tank for a period of hours then a
considerable amount of dewatering of the sludge takes place by water draining from the
aerated sludge layer back into the tank.
If the sludge removal from the surface of the flotation tank can be carried out in this
intermittent manner without adversely affecting the quality of the clarified water passing
forward for filtration, then it is possible to obtain sludge concentrations in excess of 2%
solids direct from the flotation unit. The fact that sludge of this concentration from a
flotation unit contains considerable quantities of air does not appear to significantly affect
its subsequent handling and dewatering characteristics.
The operational aspects of the dissolved air flotation process, both as a method for the
clarification of raw water and its effect in subsequent sludge handling, have been the
subject of considerable investigation by the Water Research Centre.
At the present time there is little full-scale operational experience in this field;
however, the experiences of one operational plant incorporating dissolved air flotation as
part of the water treatment and sludge treatment by centrifuging have been described.3 At
least one other treatment plant incorporating dissolved air flotation followed by filter
pressing of the sludge is in the course of construction and details of its performance will
no doubt be published in due time.
In addition to the above, dissolved air flotation is being used at at least one treatment
plant in the UK for treating the backwash water arising from rapid gravity filters, used
without coagulation, as a single-stage treatment for a particular water. A similar process
is being incorporated at two other water treatment plants of a similar type which are at
present in the early stages of design and construction.
6.8.2. Lamella Sedimentation

The lamella sedimentation process is a continental development and has not been much
utilised in the UK. It is a high rate sedimentation process which takes place in tanks
equipped with inclined plates which form in effect a series of very shallow sedimentation
tanks. The clarified water rises between the plates whilst the sludge solids settle on the
plates and run down the plate surface into the bottom of the tank which is equipped with
what in effect is a totally submerged picket fence slow stirring sludge thickener. It is
claimed that sludge concentrations of the order of 4% can easily be obtained from such
equipment.
In addition to new processes which minimise sludge treatment there are two additional
developments which minimise sludge production and hence minimise the problems of
sludge treatment and disposal. They are coagulant replacement and coagulant recovery.
6.8.3. Coagulant Replacement

The use of a traditional inorganic hydrolysing coagulant in a water treatment process in
itself produces a considerable quantity of sludge in addition to the materials removed by
such coagulant from the raw water. An example of this is the use of aluminium sulphate
in which approximately 20% of the dose level as ‘aluminoferric’ or block alum
contributes to the sludge as aluminium hydroxide, i.e. a dose of 40 mg/litre of block alum
will result in 8 mg/litre of dry solids in the sludge. This can often mean that 50% of the
sludge produced in the treatment process arises from the chemicals introduced.
The partial or total replacement of the hydrolysing coagulant by a suitable cationic
polyelectrolyte is claimed to reduce sludge production considerably. This claim is
undoubtedly true if such a polyelectrolyte can be found which will produce results as
good as the traditional coagulants at a reasonably economic cost. Unfortunately the
development of such cationic polyelectrolytes is still in its infancy and the types currently
available are only applicable to certain types of raw waters. In addition to this the cost of
such materials is such as to render their use uneconomical as compared with the use of a
hydrolysing coagulant together with a polyelectrolyte as a coagulant aid.
Since the full-scale use of cationic polyelectrolytes as a replacement for the more
traditional materials has been very limited, there has not been the opportunity to study in
any great detail the handling and dewatering characteristics of the sludges produced.
6.8.4. Coagulant Recovery

When a hydrolysing coagulant such as a salt of aluminium or iron is used it is possible to
minimise the quantity of sludge for ultimate disposal by recovering for re-use as much as
possible of the coagulant from the sludge. This recovery normally consists of thickening
the sludge as much as possible and then extracting the used coagulant with sulphuric
acid. It is claimed that coagulant recovery levels as high as 70% can be obtained and the
quantity of sludge for subsequent disposal is greatly reduced. It is also claimed that the
dewatering characteristics of the residual solids after acid extraction are modified in a
beneficial manner. The operation of such a process requires careful control of sulphuric
acid dosage to avoid recovery of undesirable metals from the sludge along with the
aluminium or iron salts.
A coagulant recovery process is at present used at one waterworks in the UK where
the recovery is allied to a sludge freeze-thaw dewatering system. The pilot plant work on
which this plant design was based has been described in the literature.11 Coagulant
recovery is an expensive process since it involves the use of acid resistant materials in
many parts of the sludge handling equipment and its economic viability depends on the
relative costs of fresh coagulant, sulphuric acid and suitable dewatering equipment at the
site in question. In addition to the economic aspects of the recovery process there is
always the doubt about what other materials will be recovered inadvertently from the
sludge by acid extraction. It is indeed a philosophical question as to whether it is correct
to go to considerable expense to treat a water with a coagulant to remove undesirable
materials and then extract the sludge with acid and return the acid-soluble fraction to the
water undergoing treatment.
6.9. SLUDGE QUANTITIES
A dilemma facing the designers of new treatment plants in which it is desired to

incorporate sludge treatment at the outset is the estimation of the quantity of sludge
which is liable to arise from the water treatment. This is indeed a problem if data on the
quality of the raw water to be treated are limited and is often the reason why temporary
disposal methods by lagooning are adopted initially, with a view to formulating firm
proposals for sludge treatment and disposal after initial operational experience of the
water treatment plant has been obtained.
The lead-in time to the development of a new water supply scheme or new treatment
plant is normally sufficiently long to enable adequate data on raw water quality to be
obtained allowing the sludge treatment facilities to be incorporated at the outset. All that
is required is a knowledge of the suspended solids and colour levels in the raw water and
their variability with time along with estimates of the coagulant doses likely to be
required, the latter being required anyway to enable the chemical storage and dosing
requirements to be specified.
The amount of sludge produced in terms of dry solids can be calculated from the
suspended solids and colour in the raw water together with the solids arising from the
coagulant dose envisaged along with the residues from any other chemicals used in the
water treatment process such as coagulant aids and activated carbon. Equations have been
published for this quantification.12 Having ascertained the quantity of dry solids expected
to arise from the process it is a relatively simple matter to convert this into the volumes of
liquid sludge at various stages throughout the water treatment and sludge handling
processes and to design the necessary tanks and dewatering equipment accordingly.
6.10. OPERATIONAL ASPECTS OF SLUDGE TREATMENT
Whilst it is recommended that sludge treatment and disposal should be designed as an

integral part of the whole water treatment system, care must be taken to ensure that the
whole of the various processes involved do not become too interdependent. An obvious
example of this is a problem arising on the sludge thickening system which results in an
inability to desludge the sedimentation tanks at the required intervals and this in turn
results in a deterioration in settled water quality passing forward for filtration. Sufficient
capacity must be built into the system to allow for breakdowns on the sludge treatment
being rectified before their effect is transmitted back to the water treatment. Sludge
treatment must always be subservient to the primary purpose of producing a treated water
of the required quality.
Another example of the possible interaction of various processes is between the
dewatering and ultimate disposal of the sludge. If the sludge is dewatered by filter
pressing and the ultimate disposal is by road transport, the emptying of the filter presses
should not be dependent on the availability of road transport at the appropriate time.
Neither should the loading rate of road transport vehicles be dependent on the rate of
emptying of the filter presses. In such cases it is better if the filter presses are provided
with hoppers below them capable of holding one press load of dewatered sludge cake and
a means of loading the sludge from the hoppers onto the road transport at a reasonable
rate either by means of screw or belt conveyors or a combination of the two. The
provision of such a system of hoppers and conveyors can also make for savings in man-
power costs since the presses can be emptied and put back into the filtration mode
immediately and only after this has been done is there a need for the same personnel to
concern themselves with the actual handling and disposal of sludge solids.
6.11. CONCLUSION
From the foregoing it can be seen that several processes are available for the efficient and
economical treatment and disposal of waterworks sludge. It is important at the design
stage, whether the sludge treatment facility is to be part of a new or an existing water
treatment plant, to take account not only of the technical aspects of the design and
selection of the various components but to subject the proposed process to a degree of
operational and systems research in order to produce a workable system.
ACKNOWLEDGEMENT
Figures 1 and 2 in this chapter are drawn from a paper by the author, published in the
Journal of the Institution of Water Engineers, 1971, 25, 402, with the permission of the
successor to that Institution, the Institution of Water Engineers and Scientists.
REFERENCES
1. DOE, P.W., J. Instn Water Engrs, 1958, 12, 409.

2. Water Research Centre Tech. Rep. TR62, 1977.
3. Proc. WRC Seminar of Alum Sludge Disposal, 1978.
4. HILSON, M.A., J. Instn Water Engrs, 1971, 25, 402.
5. BASKERVILLE, R.C. and GALE, R.S., Water Pollution Control, 1968, 67, 233.
6. GAUNTLETT, R.B. and PACKHAM, R.F., J. Instn Water Engrs, 1972, 26, 185.
7. SANKEY, F.A., J. Instn Water Engrs, 1967, 21, 367.
8. BENN, D. and BRIDGES, L., J. Instn Water Engrs, 1971, 25, 417.
9. J. Instn Water Engrs & Scientists, 1979, 33, 80.
10. DOE, P.W., BENN, D. and BAYS, R.L., J. Instn Water Engrs, 1965, 19, 251.
11. WEBSTER, J.A., J. Instn Water Engrs, 1966, 20, 167.
12. J. Instn Water Engrs, 1973, 27, 399.
Chapter 7
WATER QUALITY MONITORING
P.J MORLEY, C.Chem., M.R.I.C.
Principal Scientist, Avon Division, Severn-Trent Water Authority,
Coventry, UK
AND
J.COPE, C.Chem., M.R.I.C., Ph.D.
Scientific Officer, Severn-Trent Water Authority, Birmingham, UK
SUMMARY
The firstpart of this chapter deals with the historical aspect of water quality monitoring
and the need for increased vigilance in recent years. This has taken two courses—the use
of continuous monitors, and modern automated laboratory analyses. The former offers
rapid answers, but needs a high capital outlay and suffers from doubtful reliability and a
lack of versatility. The latter can provide reliability and a variety of analyses, but has the
disadvantages of errors caused by transport of samples and the time delay before results
are available. The two approaches are discussed in detail together with their relative
merits and future development.
7.1. INTRODUCTION
Historically quality control of the water cycle was the responsibility of such agencies as
the Public Analyst and the River Authority. The analytical techniques involved frequently
required considerable skill and time from their staff. In certain areas, such as plant
control, methods were considerably simplified to provide results quickly and cheaply
from non-chemical personnel. Although this gave useful information, when used in
context of how it was gained, it led to many rough imprecise data being accumulated.
In the early 1960s there was an acceleration of interest in the environment and a
growing awareness that raw water supplies were at risk from a variety of sources.
Predictions of much higher water consumption created a need for information relating to
the quality of our rivers, canals, lakes, boreholes and springs. The data available were in
many areas unreliable and the acquisition of new records promised to be labour intensive
and consequently very expensive.
The approach to the problem has taken two courses—the use of modern laboratory
techniques and continuous monitoring on site. The need for a cheap instrumental monitor
to continuously gauge water quality at source unattended is obvious. The development
work so far, however, has produced equipment capable of measuring only a limited
number of parameters. Those first determined were dissolved oxygen, electrical
Water quality monitoring 163
conductivity, pH and temperature, all of which are best measured on site and such
instruments have found wide use. The later generations of monitors, which utilise such
sensors as the ammonia and fluoride specific ion electrodes, have also found considerable
areas of application. Their main advantage of instant response at the point of sampling
has to be balanced against a number of other factors. They are expensive to install and
service, at trace levels they can lack sensitivity and much remains to be done to produce
reliable sensors which maintain calibration between checks. Both the use of a continuous
monitor or a modern laboratory facility share the same serious problem. It is as difficult
to present a representative flow to an in situ sensor as it is to obtain a bottle of sample
which fairly reflects the state of a heterogeneous mass of water. The main advantage of
laboratory analysis is that the number of parameters analysed can be much greater,
although the cost of equipping and staffing such an establishment is high. The recent
trends are to utilise modern techniques and automation to minimise the cost and
continuous monitors are constantly being improved and their range becoming more
diverse.
At present these two methods of data collection complement one another and it is not
likely, in the foreseeable future, that either will present itself as the complete answer.
7.2. LABORATORY INSTRUMENTATION AND AUTOMATION
As previously described, the last twenty years has produced an ideal climate for
improvement in the laboratory. This has taken the course of the use of equipment to
automate manual techniques and the development of instrumentation capable of
measuring trace quantities previously unattainable. The approach to the handling of
results has also been considerably revised with the help of advancements in the fields of
data capture and computing.
Currently the instruments manufactured and suggested for the analysis of large
numbers of individual sample solutions can be classified into two areas—continuous flow
analysers and batch analysers. In continuous flow analysers the samples are pumped from
their containers into a glass or plastic tube through which they move until the reaction is
complete. The samples are thus part of a continuously moving stream with reagents being
added at controlled rates at specific points. The stream then flows to a measuring device
which is commonly a colorimeter or specific ion electrode. This technique, above all
others, is extremely versatile. The sample stream may be split for multiple analyses,
filtered, dialysed, heated, extracted with organic solvent and distilled. It is also very
flexible and, after 15 years of intense development in the water industry, still finds new
areas of application. The chief disadvantage is that each sample is liable to contamination
from the preceding one and this limits the rate of throughput.
The batch analyser treats each sample as a discrete unit and therefore does not suffer
the same degree of cross-contamination. Samples are transferred into a reaction vessel
together with, at pre-determined intervals, additions of reagent. The vessel can then be
heated for a fixed period of time and a portion removed to the colorimeter or other
detector. The discrete analyser exhibits rates of analysis of up to 300 samples per hour
compared with 40–60 per hour obtained from conventional air segmented continuous
flow analysis. It is, however, considerably less versatile, has more moving parts to
service and replace, and is much more expensive.
The analysis of ‘trace’ quantities is largely concentrated on toxic metal and organic
contaminants. The developments in atomic absorption, chromatography and mass
spectrometry have brought about immense progress in these fields and will be discussed
in more detail later in the chapter.
The production of large numbers of data from these analysers and other techniques
introduces a high clerical involvement. This can be divided into two areas—data
processing and laboratory organisation. The section dealing with laboratory data
collection and processing will review the work already carried out in water laboratories
and the possibilities offered by recent technological development. Because of the rapidly
changing nature of this field it should be read in conjunction with the more recent
literature available.
7.3. CONTINUOUS FLOW ANALYSERS
In the mid 1960s The Technicon Instruments Company Ltd introduced the Auto
Analyser, already well known in clinical biochemistry, into the field of water analysis.
The technique of air segmented continuous flow analysis1 is a simple way to automate
any colorimetric or turbidimetric measurement and is now a common sight in water
laboratories. A basic form of analyser is shown in Fig. 1. Sample, reagents and air are
pumped along fine-bore plastic tubing to glass mixing coils. These produce a
FIG. 1. A simple air segmented

continuous flow analyser.
homogeneous mixture uniformly segmented between air bubbles which may be

thermostatically controlled in oil or water baths to promote hydrolysis, reduction,
oxidation or colour formation. Before the stream enters the flow-through colorimeter the
air bubbles are removed in a glass T-piece. The required flow is pumped from the bottom
of this separator into the flow cell and finally to waste. The output from the colorimeter is
generally fed to a flat-bed chart recorder.
An important aspect of any colorimetric analysis is the separation of particulate matter
even though this still leaves problems with natural colour and turbidity. Early attempts at
continuous filtration utilised a roll of filter paper fed between two spools in a manner
similar to that of a tape recorder. The sample stream is allowed to flow onto the paper and
the filtrate collected underneath. The strip is driven continuously to provide a fresh
surface. A far more effective means of removing particulate matter, however, was found
by using the clinical technique of dialysis. This will remove turbidity and natural colour
as well as filtrable solids, and a typical unit is shown in Fig. 2. The plates are made of
rigid plastic and have a flow
FIG. 2. Dialysis unit.

channel, of hemispherical cross-section, machined on one surface. When two of these
plates are screwed together, separated by a polyethylene dialysis membrane, the unit is
complete. A stream of water to be analysed is passed over the top of the polyethylene and
a receiving stream flows, usually at the same speed, underneath. The nature of the
receiving stream is chosen to encourage the transfer of the relevant constituent from the
water sample. For example, a solution of lower pH than the donor stream will be used to
effect the transfer of the ammonium ion. All molecular or ionic species other than the
simple small variety will remain in the donor stream and flow to waste. It is thus possible
to separate the ammonium ion, chloride, phosphate, nitrate, nitrite and many others from
colour and particulate matter in the main stream.
The use of continuous flow analysis together with dialysis enables a variety of
determinations to be carried out at speeds up to 60 per hour (see Table 1). Replacement
of the polyethylene with a thin sheet of silicone rubber gives a membrane surface that
cannot be wetted by either of the streams. This enables the transfer of gas and the
determination of cyanide and sulphide has been carried out in this way. For example in
the analysis of waters for free and simple complexed cyanide the air-segmented sample is
mixed with acid and passed through a gas dialysis unit. The recipient stream is a solution
of caustic soda and the cyanide is then determined by the standard pyrazolone or
barbituric acid colorimetric techniques.
TABLE 1 CONTINUOUS FLOW METHODS
DEVELOPED FOR WATER ANALYSIS
Alkalinity
Aluminium
Ammonia
Anionic detergent
Boron
Calcium
Chloride
Cyanide
Fluoride
Hardness
Iron
Magnesium
Nitrite
Phenol
Phosphate (total soluble)
Silicate
Sulphate
Sulphide
Sulphite
Total oxidised nitrogen
Manganese
Organic nitrogen
Organic phosphorus
The manual determinations of phenol and total cyanide are good examples that can
require a preliminary distillation stage. The continuous distillation head has been
introduced with particular success for these two parameters and is illustrated in Fig. 3.
The sample and a 10% v/v aqueous solution of phosphoric acid are mixed together and
passed through a heated mixing coil at 165°C. The distillation head is connected to the
flow at the exit from the heating bath and all the volatile constituents are collected
FIG. 3. A continuous distillation head.

in the trap following the water condenser. The phosphoric acid and nonvolatile portion
flow to waste whilst the condensate is mixed with more air and the colour reagents. The
only disadvantage of this system is that a nonvolatile, thermally stable liquid such as
phosphoric acid must be present to prevent the coil from becoming blocked and the
conditions, therefore, tend to be very rigorous and limit the application.
It is even possible to introduce an acid digestion into the continuous flow system by
utilising a long glass helix. The sample, air, and digestion acid are mixed and dropped
into the bottom of the helix. Rotation of the tube causes movement of the liquid from one
end to the other across heater blocks. With optimum adjustment of acid conditions,
temperature and speed of helix rotation the solution emerging from the other end of the
tube will be completely digested. Total nutrient concentrations, such as nitrogen and
phosphorus, can be monitored in this way and the equipment can also be used to
automate the determination of chemical oxygen demand (COD). It should be noted here
that COD is an arbitrary value derived from a set of empirical oxidation conditions. The
use of the heated helix gives a different value to the standard manual method but it can be
effectively utilised to study trends in oxygen demand. A further illustration of the
versatility of continuous flow analysis is its application to solvent extraction systems and
the laborious manual method for estimating synthetic anionic detergent in waters was one
of the first to be automated. The extraction of the methylene blue-detergent complex is
carried out by introducing chloroform into the flow shortly before a mixing coil. The two
phases are sufficiently agitated to give an efficient extraction and the heavier solvent may
be separated from the aqueous layer and air using a glass T-piece.
The chief disadvantage of the continuous flow analyser is cross-contamination of
samples. Ruzicka and Hansen2 argued in 1975 that the theoretical maximum was
approximately 40 samples per hour but recent suggestions, mainly by manufacturers, are
that this is 60 per hour. Rates higher than these lead to significant carry-over effects and a
number of approaches have been used to circumvent the problem. They range from the
introduction of a calculated correction factor3 to computer regeneration4 of the recorded
curve but have achieved limited application, mainly owing to the loss of precision and
accuracy, or expense and expertise involved. The most promising advance of all has been
the development of flow injection analysis. The presence of the air bubble in continuous
flow systems is commonly believed to separate sample and alternated wash into well
defined ‘slugs’. The theory of flow within the tubes, however, has revealed that carry-
over between samples is caused by the extent of laminar flow prevailing. The presence of
an air bubble causes friction at the tube walls and gives a turbulent flow even at very low
pumping velocities.5 The net effect is a reduction in carry-over but there are also a
number of disadvantages. The chief ones are the compressibility of air, which causes the
stream to pulse rather than to flow regularly, the necessity of removing bubbles before
colour or turbidity is measured and the need to accurately control bubble patterns at fast
sampling rates. Ruzicka and Hansen2 exploited the fact that fast flows in sufficiently
small diameter tubing are predominantly turbulent. In flow injection analysis a small
volume of sample (approximately 0·5 ml) is injected, via a syringe and septum, into a
turbulently flowing carrier stream of reagent, the air bubble no longer being necessary.
Early work with simple colorimetric reactions, such as the determination of chloride
using iron (III) thiocyanate,6 suggests possible rates of 250–500 samples per hour. The
application to the water industry has so far been disappointingly small. The technique
suffers the disadvantage that it appears to be only suited to simple reagent addition and
that long heating times for reaction development are difficult to accommodate. It is,
however, a very good system for presenting a buffered sample to ion-selective electrodes
in that the response time of these sensors
FIG. 4. Automation of sample

presentation to an ion-selective
electrode using flow injection.
increases dramatically in fast moving streams. A diagram of the equipment required is

shown in Fig. 4.7 The flow injection stream of sample and buffer is allowed to impinge
on the electrode surface and then falls in a continuous stream to a waste container, which
houses the reference electrode. This stream maintains an electrical connection between
the two electrodes and increases of response time up to a factor of four can be obtained.
7.4. BATCH OR DISCRETE ANALYSERS
Discrete analysers have found wide application in the clinical field and have been
available to the water industry for some years. Early experience with them, however,
showed poor reliability and few properly developed water chemistry methods. Recent
improvements in the mechanics of these systems, particularly in the area of pneumatic
operation, have produced an extremely useful rapid laboratory analyser. There are at
present approximately 10 discrete analysers available and they fall into two main groups.
The most common simulate manual addition and mixing and subsequently present the
reacted solution to a spectrophotometer. The other design uses a disc into which a
number of radial troughs have been cut, the depth of the cut normally increasing towards
the centre. Sample and reagents are placed individually into these arms and the whole
disc is spun rapidly in a thermostatically controlled environment. The reaction mixtures
are formed at the circumference and their optical densities can be measured as they pass
through a light beam. At present an instrument of this design is available but its cost will
limit any application to the water industry. The manual simulation is considerably
cheaper and at £12000–£15000 (October 1979) it is possible to buy a single channel unit
which performs the functions of sample dilution, reagent addition, mixing, incubation and
transfer at speeds of up to 240 samples per hour.
One of the better instruments available at present is the Pye ACl system in which most
moving parts are pneumatically driven. A preset volume of sample is automatically
withdrawn by syringe from a tube and dispensed, together with the first reagent, into a
reaction vessel. This then moves along a rack, the temperature controlled by a variable
water bath, until it reaches
TABLE 2 A COMPARISON OF CONTINUOUS
FLOW AND DISCRETE SYSTEMS
Continuous flow Discrete
1. Maximum rate of 40– 240–300 samples per
60 samples per hour hour
2. Approx. £10000 for £12000–£15000 for
cheapest five-channel sequential analyser
unit
3. High reagent Low reagent
consumption consumption
4. Susceptible to Not applicable (no
interruptions in ‘settling down’
sampling programme period)
5. Susceptible to carry- Not applicable
over
6. Peaks must be Output available as a
‘picked’ from recorder digital display
output before data can
be processed
7. No limit to reagent Usually limited to
additions four
8. Possible to automate Not possible
filtration, dialysis,
distillation and solvent
extraction
9. Large number of water Limited number
chemistry methods available
developed
10. Can be used with Use usually restricted
detectors other than to a
colorimeters (e.g. ion spectrophotometer
specific electrodes)
11. Low maintenance cost Higher, yet not
restrictive, costs
the next reagent addition station which dispenses a fixed quantity of reagent and
mechanically mixes the solution. With a possibility of three such stations four reagents
can be added to the sample and, at a rate of 240 per hour, there is a total reaction time of
12 min. This speed can be reduced to increase the time. The final mixture is drawn into a
spectrophotometer for measurement of absorbance. The manufacturer has with this
instrument invested some time in providing simplicity of operation and rapid changeover
for a number of water chemistry methods.
It is interesting at this point to compare air-segmented continuous flow and discrete
analysers and to specify their areas of application. Table 2 illustrates the main points of
comparison of the two techniques. As a general rule a discrete analyser should be used in
a laboratory dealing with a large workload of ‘clean’ samples (e.g. potable and river
waters) and where simple reagent addition alone is required. The continuous flow
systems find application where workloads have a high proportion of samples containing
high suspended solids and organic matter and when the sample throughput is insufficient
to support a discrete analyser.
7.5. THE DETERMINATION OF METALS
The technique of atomic absorption spectrophotometry is particularly useful for the

monitoring of a variety of metallic contamination at trace levels. The analytical speed,
ease of operation and relative lack of interferences have made the method attractive for
routine water quality analysis. The first step towards automation was a turntable feeding
the samples to the instrument and a recorder or printout providing a continuous output of
absorbance. This allowed rates of 120 metals per hour to be achieved. More recently
improvements in nebuliser and atomiser design have made possible speeds in excess of
600 per hour. Microprocessor units, now integral in many instruments, control automatic
setting of machine conditions and process raw absorbance data to original concentrations.
One manufacturer has produced a microprocessor controlled sampler which has such
facilities as multiple sampling from individual tubes and recalibration on request.
Modern technological advancements have also made practicable the always promising
technique of flameless atomic absorption. In the most common design volumes of water
of up to 30 µl are dispensed into a carbon tube through which a beam of monochromatic
light shines. A potential is applied across the tube in three stages; the first is a drying
stage which involves a slow increase in temperature to approximately 100°C and this is
followed by an ashing stage to 600°C to burn off organic matter. The final step is a rapid
increase to volatilise the metal in its ground state (temperatures up to 2600°C). An
important advance with this technique has been the availability of a reliable automatic
sampler. This removes many of the precision problems previously encountered with the
method and enables simple application of the standard addition technique that is
necessary for accuracy. The latter is essential because of the change in electrical
characteristics of the tube with age.
Further flameless methods which have achieved popularity are those of hydride
generation and the cold vapour technique. Arsenic, selenium and antimony can be
estimated by generation of the respective hydride8,9 and this is then allowed to pass into
the flame. There is a considerable increase in sensitivity over the standard method. In the
cold vapour technique10 mercury ions in solution are reduced to the metal by the addition
of stannous chloride and the element is then flushed out, by the passage of air, into a flow
cell with silica end windows. Alignment of the cell in the light beam allows mercury to
be determined to 0·0001 mg/litre.
TABLE 3 COMPARATIVE DETECTION LIMITS
(µg/litre)
Element Inductively Flame Flameless
coupled (electrothermal)
plasma
Al 10 20 0·004
As 15 100 0·06
B 2 1000 –
Cd 1 1 0·008
Co 2 5 0·03
Cr 2 3 0·005
Cu 2 2 0·008
Fe 1 5 0·003
Mn 0·5 3 0·004
Mo 5 10 0·06
Ni 5 8 0·02
P 30 105 3·0
Pb 15 10 0·03
Pt 20 50 0·45
Se 15 100 0·10
Si 10 60 0·10
Ti 1 50 0·30
V 75 7000 –
Zn 1 0·6 0·0007
Emission spectroscopy has always found use in the determination of the alkaline earth
cations. It has also been used for a wide variety of metals in qualitative and semi-
quantitative investigations using a medium quartz d.c. spectrograph. The development of
a stable plasma source has enabled fast multi-element analysis to be seriously
contemplated and the high temperature of the plasma (10000°C) minimises the
interference effects normally encountered. An RF with a power of the order of 5 kW is
applied to an induction coil encircling an inert gas (usually argon). The sample is sprayed
into the resultant plasma and the ‘tail’ of the flame is viewed through a monochromator.
The multi-element analysis is carried out by using a number of pre-set photomultiplier
tubes or by allowing the grating to scan.
Table 3 shows the comparative detection limits obtained with a number of metals
using the different techniques.
7.6. MONITORING OF ORGANIC CONTAMINANTS
The organic fraction of raw water consists largely of naturally occurring compounds
originating from land drainage or domestic sewage discharge. The classical determinands
of suspended solids—biochemical oxygen demand, permanganate value and chemical
oxygen demand—are retained by most Water Authorities as measures of solid matter and
organic strength. This work represents the most labour intensive effort remaining in the
laboratory as the empirical nature of the tests resists full automation. The tasks can be
eased by the use of heated digester blocks, dissolved oxygen electrodes and sophisticated
electronic balances but they remain expensive.
An alternative measurement of organic strength is to measure total organic carbon
(TOC). This instrumental technique, capable of dealing with 15 samples per hour
unattended, consists of an oxidation furnace in which all the carbonaceous compounds
are converted to carbon dioxide. The quantity of this gas produced can then be measured
either by infrared detector or titrimetrically. A further sophistication is the instrument
which will reduce the carbon dioxide by mixing with hydrogen and passing over heated
Raney nickel. Measurement of the methane produced can be achieved with high
sensitivity using a flame ionisation detector. Typically the TOC values for waters range
from less than 1 mg/litre for some treated borehole supplies to greater than 5 mg/litre for
certain surface waters.
Approximately 20% of this organic fraction consists of man-made chemicals and these
form the greatest cause for concern. Evidence of their impact on the environment, either
alone or combined, has led to the necessity to monitor the individual identity and
quantities of the component compounds. Until recently effective separation and analysis
at these trace levels would have been impossible. The combining of capillary column gas
chromatography with mass spectrometry (GCMS) has produced an instrument with a
very high performance in terms of resolving power and selectivity. Modern commercial
instruments can produce GC peaks as narrow as two seconds and scan the eluant every
second using the mass spectrometer. The machine can be interfaced with a small desk-top
computer and disc storage to enable fragmentation patterns produced to be compared
against those associated with some thousands of known synthetic organic compounds.
Leahy and Purvis11 describe a technique of extracting the organic fraction onto a
macroreticular resin. After eluting with ether and further evaporating a 10000-fold
concentration can be achieved before submission to the GCMS. With some sacrifice of
the most volatile compounds a further ten-fold increase can be achieved. The authors
quote a number of compounds found in raw and finished waters including some which
appear to have been introduced by the treatment process. Commonly present in surface
waters are phthalates, organic phosphates and herbicides. In addition to these, finished
waters can contain haloforms, benzaldehyde and benzyl cyanide. The technique has been
found particularly useful in identifying sources likely to cause taste and odour problems
in areas of industrial pollution. GCMS will only separate and quantify those compounds
which are volatile or form volatile derivatives. Although most of the synthetic chemicals
fall into this category the remaining fraction could be identified with a high performance
liquid chromatograph again linked to a mass spectrometer. Because of their high cost the
number of these instruments available is small but their presence is essential unless the
industry is to ignore the impact of this ever-increasing complex organic mixture on our
population and environment.
7.7. LABORATORY DATA COLLECTION AND PROCESSING
The automation of a laboratory to accommodate the analysis of a large number of

samples produces a purely clerical problem. This includes sample reception, processing
of automatically produced and manually produced data, progress chasing and report
production. Morley, Musty and Cope12 describe a system which uses a desk-top computer
in combination with magnetic tape data recording. The sample can be booked in at the
computer keyboard by typing sample location, date, time, sampler’s name and analyses
required. Laboratory work sheets and an outstanding samples report can be produced at
this point. By connecting the instrument recorders (fitted with re-transmitting slidewires)
to a magnetic tape logger unit the continuous flow analyser output can be recorded and at
the end of the analysis run the tape is replayed to the desk-top computer which picks
peaks, ‘draws’ calibration graphs, corrects for base-line and sensitivity drift and
computes final concentrations. There is an edit facility to allow dilution factors to be
applied and for faulty data to be erased. Similarly instruments capable of producing a
digital output, e.g. Atomic Absorption Spectrophotometers, can be monitored at a
standard BCD (binary-coded-decimal) output and the data recorded and processed. Any
manually produced data such as suspended solids, chemical oxygen demand, etc., may be
keyed in by the analyst. A simple program sorts through the computer disc store and
brings together all data relating to an individual sample for report production. A
worthwhile by-product of this system has been to largely eliminate clerical, transcription
and arithmetic errors inherent in the manual equivalent. A common criticism by the
misinformed is that computer systems dehumanise the analyst and leave him without a
‘feel’ for the samples. They are merely a powerful aid and the system can be devised
which allows as much contact with the results as manual methods but eliminates the
drudgery and error. Much depends on how involved the analyst is with the design of the
computer procedures.
Because of the rapidly improving technology relating to this area the previously
described approach is already becoming outmoded. Most commercial instruments which
give a digital output are fitted with microprocessors to compute final sample
concentration. They are, however, not always very well programmed and it is advisable
to find an answer to the question, ‘How are the results calculated?’ With a satisfactory
instrument, however, there is no need for the complex software to compute results. Even
the continuous flow analysers can be fitted with data processors to give an output of
sample concentration. The introduction of interface systems such as the Hewlett Packard
Interface Bus, now allows the computer to act as a collator of information. The individual
instrument can plug into a ‘ring main’ circuit and automatically convey data, to the disc
store, as it is produced and when its priority rating in the system permits. A number of
instruments can be connected on-line to the computer in this way and the data collected
and sorted in the background of normal keyboard activity.
The presence of a computer in the laboratory has another distinct benefit. It is often
possible in such environments to lose sight of the original objective and produce the
analytical report for its own sake. With very little effort the analytical results can be
combined with other data such as flow to produce useful operational information. The
information can be combined with previous data and shown as a graphical trend or
compared with historical limits permitting conclusions to be drawn relating to its
significance.
This concludes the part of the chapter dealing predominantly with laboratory based
monitoring and the advantages and disadvantages of such systems have been highlighted.
In general its chief weakness is the delay in obtaining a result, whilst its main strength is
the reliability and versatility of analysis. The next section deals with automatic stations
and a combination of the two techniques should show a satisfactory approach to water
quality monitoring.
7.8. AUTOMATIC MONITORING OVERVIEW
The concept of a monitor that will continually survey raw water, and water entering
supply, is extremely appealing. Should pollution occur by natural phenomena, an
accident or wilful intent, then the system could interrupt the flow of water to prevent its
passage into supply, identify the pollutant and its concentration. If the monitor was one of
a chain of such devices then information as to the location of the pollution would also be
available. In addition to performing this role the monitoring system could be used for
routine historical water quality data gathering for use in water quality models and as
future design data.
Present-day monitoring systems are a long way from providing the user with
information that he really requires and the reliability of many systems would require
extreme care on behalf of the user before fully automatic control systems could operate
on a foolproof basis.
Broadly, the problems arise from the lack of suitable sensors for many of the
determinands that are required, together with sampling and fouling problems. Once a
transducer signal has been obtained electronic treatment and data processing of this signal
can now be carried out easily and with a reliability that will match and probably exceed
that of all the other components in the water system.
The present range of sensors is limited and the determinands available to the user are
often non-specific in identifying pollution, e.g. pH, conductivity, redox, etc., but even
allowing for more specific sensors to be developed it seems that it would be virtually
impossible to continuously monitor for all possible pollutants and be able to identify
these. Good laboratory analytical services are an essential part of the monitoring system
and will remain so for a considerable time.
7.9. BASIC TYPES OF MONITORS
There are two basic methods of deployment of automatic water quality monitors; that is,
they may either be land-based on a fixed site or they may be submersible and totally
immersed in the water they are measuring. Each type has advantages and drawbacks.
In addition to the distinction between land-based and submersible types, a further
division can be made between those monitors offered as a multi-determinand package by
a manufacturer and monitoring systems assembled by the user incorporating various
discrete analytical systems and possibly simpler packaged monitors.
7.10. DESIGN CONSIDERATIONS
When designing a monitoring system factors other than the determinands measured
should also be given serious consideration.
The siting of the station for security, ease of access and construction, the ease with
which samples may be taken and the relevance of the readings obtained are very
important. The design should be vandalproof but allow good access for users and allow
maintenance and repair work to be carried out.
The sampling system should be designed to take a representative sample in such a way
that it does not degrade it for subsequent measurements (e.g. lifting may affect dissolved
oxygen).
A burglar alarm system is recommended for unattended stations in remote sites.
7.11. SUBMERSIBLE MONITORS
A submersible monitor is a monitoring system which can be wholly immersed in the

water it is measuring. The data it measures can be recorded for subsequent treatment or
relayed to another site via cable or radio, or the device may act as a transponder.
It will be seen that such a device consists of a waterproof housing containing the
necessary electronics, power supply and data recording system with the sensors exposed
to the surrounding water. The sensors may be contained in an elaborate housing with a
flow impeller to obtain optimum performance.
If the unit can receive no external power (i.e. it is not connected to a cable or acting as
a transponder) then the design must optimise the power requirements of the unit to
maximise power supply life. Care must also be taken with the seals and external
connectors used on such a device, as a leak would probably mean that the device would
have to be written off.
The advantages of a submersible monitor are that it can theoretically be sited
anywhere in a river, lake or construction. It can be easily transported and located at the
point of interest. Locating monitors at different sites enables profiles to be built up.
The disadvantages seem to outweigh the advantages for most inland applications. (The
same is not of course true for oceanographic submersible monitors which are beyond the
scope of this work.) Unless the unit is equipped with a means of data transmission to the
outside world, which is often impracticable or undesirable, then the submersible monitor
can only record historical data. This is satisfactory if the data are for modelling or archive
work but of no use for pollution warning.
A submersible monitor is by its nature a sophisticated and therefore expensive device
and it is also vulnerable to vandalism, damage by boats or total loss during storms or
flood, and users of monitoring equipment must be prepared for some losses if using this
type of monitor on open sites.
The anchoring of a submersible monitor also presents substantial problems. The fixing
may not be difficult if carried out within the confines of a works but a land-based monitor
may be more convenient in this situation.
In an open lake or river there may be no convenient anchorage point and an anchor
may have to be fixed to the river bed or lake bottom. This may be extremely difficult and
relocation of the monitor may be hampered if it has moved. A radio beacon may help
here. If a buoy is fitted to identify the location then this invites vandalism and may hinder
navigation. Attachment to a bridge, pier or jetty may also be considered, but this has the
disadvantages that the monitor may strike the structure damaging itself, also such a unit
could easily be replaced by terrorists by a similar looking explosive device.
Calibration of submersible monitors may be a problem but designers should endeavour
to simplify this. The obvious requirements are for small amounts of calibration solutions
to be required and for the calibration electronics to be readily accessible whilst
calibration is being carried out. The idea of a universal calibrant for all determinands of a
submersible monitor is an appealing concept but in reality each determinand will
probably require its own calibration solutions. If no output to the external world is
provided a recording real time clock on this type of device is essential (see Section 7.26,
Data Logging).
7.12. SAMPLERS
As has been discussed, a monitoring system may have indicated a pollution or suspect
reading but it is highly unlikely that the precise cause of the pollution can be identified
from the data given by the monitor. It is also likely that changes in determinand levels
were occurring before an alarm state was signalled. In addition, by the time qualified
personnel arrive to investigate the pollution, the water may have returned to being
unpolluted.
To help identify pollutants with present sensor technology a useful method is to
automatically take samples of the water presented to the monitoring system and retain
these for a set period after measurement. If no pollution occurs these may be discarded
but when a pollution does occur the samples are retained together with samples at the
time of the indicated pollution and subsequent samples. These samples may then be
submitted to a laboratory for further investigation by more sophisticated and specific
techniques.
7.13. DISSOLVED OXYGEN
Dissolved oxygen is most commonly monitored using either a sensor of the Mackereth
type13 or a polarographic sensor of the Clark type.14 Both these sensors use gas-
permeable membranes which may be subject to fouling. The membranes may be cleaned
by mechanical action although care must be taken not to scratch or puncture them. If
fouling is a problem other sensors are available, a common one of which is the sacrificial
thallium electrode.
When monitoring dissolved oxygen it is extremely important not to carry out any
processes on the water which will affect the reading. Lift pumping will tend to lower the
dissolved oxygen level whereas turbulence entraining air will increase the level. Water
for dissolved oxygen monitoring should be measured in situ or pumped using
submersible pumps. The dissolved oxygen measurement should be taken before the
sample is subjected to any chemical treatment.
7.14. pH
pH is a measure of the acidity or alkalinity of an aqueous liquid. It is most commonly

determined by measuring the potential between a reference electrode and a specially
constructed glass electrode immersed in the water. A very high input impedance
measurement system is used and the measurement is amplified and scaled to give pH. A
more detailed explanation of the measurement system will be found in most physical
chemistry text books. The glass electrode is a delicate device and may encounter
problems if used in a monitoring environment. Present-day glass electrodes are relatively
robust but exposure to abrasion or vibration may damage them irreparably. An alternative
electrode for the determination of pH, where glass electrode failing is a problem, is the
antimony-antimony oxide electrode. This device often incorporates a mechanical scraper
to expose fresh electrode material and is a useful system for automatic monitoring.
Care should be exercised when measuring and interpreting pH data on low
conductivity waters as electrode contamination, gas absorption and the effects of the
measurement system may change the pH value significantly. In these circumstances the
value of measuring pH and its relevance should be borne in mind when interpreting these
data.
7.15. ION SELECTIVE ELECTRODES
Devices of this type have been available for many years, the most widely used example
of which is the glass electrode used in pH measurement.
The concept of an electrode which may be immersed in water and give a readout of a
specific determinand is very appealing. Unfortunately, very few of these devices can be
used without sample pre-treatment and complete specificity. Before installing detectors
of this type a user should ensure that provision is made for the correct sample pre-
treatment if required, and that the system being measured will not present any interfering
species to the electrode system.
Electrodes which find common use are the ammonia gas permeation electrode, the
fluoride electrode and the chloride electrode.16
Whilst manufacturers may list electrodes for a wide variety of determinands, users
must ensure that they can be successfully applied to the media being measured.
7.16. OXIDATION-REDUCTION POTENTIAL (ORP) OR REDOX

POTENTIAL
This is sometimes measured in pollution monitoring systems often using a platinum and a
calomel reference electrode. It is probably one of the most difficult determinands to
evaluate but indications as to changes in the system will often show a change in redox
potential and it can help in verifying trends shown by other sensors.
7.17. SUSPENDED SOLIDS
There are two ways of interpreting suspended solids and it is important to be aware of
this when considering automatic methods for their determination.
The laboratory test for suspended solids involves the filtering of a known volume of
sample through a pre-weighed filter paper, drying the paper and re-weighing to est ablish
the weight of solid the volume of sample contained. This yields a result in weight per unit
volume which is the way solids levels are most commonly expressed. The automation of
this technique for monitoring poses many serious technical problems and is not
practicable at present.
The opacity of liquids containing suspended matter is easily observed by eye and this
property is frequently used to assess suspended solids levels.17 The turbidity or light-
scattering properties of the liquid may be determined using techniques based on classic
optics and the result expressed in turbidimetric or nephelometric units. The interpretation
of these results requires care but does enable one set of results to be compared with
another (as do those from gravimetric determinations). Optical methods do not offer any
method of establishing the weight per unit volume of solids. If the nature of the solids,
e.g. colour, spatial scatter, etc., is consistent, then weight per volume figures can be
inferred from optical data, but in absolute terms there is no optical property which relates
to the weight of the suspended matter. Optical suspended solids meters which are scaled
in weight per volume units should not be believed blindly, but the user should ensure that
he has a clear knowledge of the operating principles of the meters and the nature of the
samples under investigation before placing reliance on indicated readings. Such systems
may also give results which are at variance with reality if ‘artificial’ standards such as
formazin or fullers’ earth are used as calibrants rather than the solids being measured.
7.18. CONDUCTIVITY
Conductivity is a determinand that is easy to measure and is thought to give an indication

of an ionic pollution.15 Care must be taken when applying temperature corrections to
conductivity measurements and it is preferable to record the absolute conductivity
together with the temperature of the sample. Conductivity is normally measured with an
a.c. electrical bridge system operating at either 1000 or 1592 Hz. D.c. measurements
should be avoided owing to electrolytic and polarisation phenomena. Electrode designs
vary but it is common to use ring-type electrodes for monitoring work. These are
arranged as rings on the inside of a tube and guard electrodes are often incorporated in
the design. These electrodes are easily cleaned or kept free of fouling by high flow rates.
7.19. BIOLOGICAL SENSORS
As has been previously discussed, it is impossible to provide physical and chemical

automatic sensors which would provide total coverage of pollutions hazardous to life,
especially human life.
In an attempt to assess water conditions which may endanger living species, sensor
systems have been designed which monitor the changes in behaviour of life forms when
exposed to the water under examination.18
One such method is to record the behaviour of fish living in the water to be monitored.
It is postulated that the fish have an established behaviour pattern in unpolluted water. If
the pollution level in the water rises then the fish behaviour will become more agitated
and this is monitored electronically and assessed by computer.
Biological sensors which monitor the metabolic rate of lower organisms are also used.
A pollution will cause the metabolic rate to drop and this can be measured relatively
simply and the metabolic rate displayed on a chart recorder.
A pollution detected by biological sensors of these types still has to be identified, and
this may prove a difficult task requiring substantial laboratory facilities and expertise. It
is essential that when a pollution has been indicated by a biological sensor, samples of the
suspect water are taken for subsequent laboratory investigation.
7.20. CONTINUOUS FLOW ANALYSERS
The air segmented continuous flow analyser has been adapted to perform a continuous
monitor function. The pump rates and therefore reagent usage are lower than in
conventional laboratory systems and the instrument receives sample from one source
only. The advantages of this unit are that continuous flow methods used in the laboratory
can be adapted to be used in a monitoring situation, and that it enables tests to be

performed in a monitoring station which are not possible by other methods.
7.21. TEMPERATURE MEASUREMENT
Automatic temperature measurement is usually performed by a device such as a

resistance thermometer, thermistor or thermocouple. Remote measurement techniques
from satellites and aircraft commonly use infrared thermography.
Temperature measurement serves two functions. It is an important determinand in its
own right and an indication of possible thermal pollution. It is also essential to the correct
interpretation of data from other sensors being used for water quality monitoring.
An example of the second role is in connection with measuring electrical conductivity
which is temperature-dependent. Most standard laboratory methods require conductivities
to be determined at the standard temperatures of either 20 or 25°C. When monitoring raw
or supply water it is impractical to adjust the temperature to that of standard laboratory
conditions. A common practice is to measure the conductivity at the ambient water
temperature and then to use a device known as an automatic temperature compensator to
‘adjust’ the conductivity to the value that it supposedly would be, had it been determined
at the standard temperature. The correction is difficult as the temperature coefficient for a
water at a specific time is seldom known accurately and large adjustments are made to
the reading which can commonly be of the order of 50%. If an automatic temperature
compensator is used it may be impossible to establish the absolute conductivity of a water
as measured from data so treated. If a record is made of absolute conductivity and
temperature, then a temperature compensation can be easily carried out if desired but the
integrity of the measured data is not destroyed.
Dissolved oxygen, pH and selective ion measurements and redox potential are all
affected by temperature variation but not as dramatically as is electrical conductivity. In
all cases where temperature-dependent measurements are made then a simultaneous
record of temperature enables correct interpretations to be made at a later date and helps
establish absolute integrity of data.
7.22. CLEANING SYSTEMS
Where fouling of a monitoring system presents a problem there are several methods
available which offer the user a means of substantially extending the time between
complete overhauls of the system. When an automatic cleaning system is used care
should be taken in its choice in that it should not damage the monitoring assembly or the
sensors and that, after cleaning, any causes of misreading of the sensors are adequately
removed, e.g. chemical agents or air. Some of the cleaning systems available for use on
automatic monitoring systems are given below.
7.22.1. Mechanical Cleaning

A mechanically actuated brush or wiper is used to clear fouling. The operation may be
continuous or intermittent. Care must be taken if sensors are cleaned this way to ensure
they are designed to undergo this treatment (e.g. some pH glass electrodes and certain
dissolved oxygen electrodes may be damaged by this treatment).
7.22.2. Ultrasonic Cleaning

This is a form of mechanical cleaning using ultrasonic transducers to generate ultrasound
in the liquid in the monitoring system to detach fouling. It may be used with the water
being measured by the system or used in conjunction with a chemical cleaning agent or
detergent. In practice it is found that this will reduce fouling but, if sufficient sonic power
is used to remove all fouling, damage to the system will result. It would be bad practice
to take sensor readings whilst an ultrasonic transducer was switched on.
7.22.3. Air Cleaning

If the routine measurement cycle is interrupted and a mixture of compressed air and water
(possible with chemical cleaning agents) passed through the system the agitation
produced will give an effective cleaning action. Again care should be taken that the
sensors used can take this treatment and that the air is fully removed from the sensors
after cleaning.
7.22.4. Chemical Agents

Substances such as sodium hypochlorite solution or detergents can be used to clean a
system but they will be much more efficient if used in conjunction with a mechanical
cleaning method. If the chemical agent used will interfere with a sensor reading then it
should be properly removed before the system is allowed to produce real measurements.
If a chemical agent is known to have no effect on a test being performed then a continual
dosing of the chemical into the water being measured is a very effective way of reducing
fouling. This works quite well with biocides.
7.22.5. Sterilisation Methods

If the water entering the monitoring system has been sterilised then fouling will be
substantially reduced unless the problem is one of silt. Various methods are available for
doing this and the chemical method has already been discussed. Treatment with
ultraviolet light (u.v.−C radiation) is one method, but by its nature may cause chemical
changes in the sample. The same is true of ozone together with potential damage to
components and the invalidation of several sensors, e.g. dissolved oxygen. Radiation
sterilisation offers perhaps the best way of achieving sterilisation, but it may be
undesirable to use this at an unattended station.
7.22.6. Station Design

A monitoring system can be designed in such a way as to reduce fouling. Attention to
pipework to have all smooth surfaces with no inaccessible corners is essential, as is a
design which enables easy manual cleaning. If liquid flow rates are kept high through the
unit it will be difficult for fouling to occur; liquid velocities should in any case be greater
than isokinetic to prevent suspended material being deposited in the system. Light will
often encourage the growth of algae and where possible systems should be protected
from it. Sample filtering may help to reduce fouling, but total automatic solids removal in
an unattended system is difficult to attain unless solids levels are comparatively low.
7.23. ATOMIC ABSORPTION SPECTROSCOPY AND PLASMA

EMISSION SPECTROSCOPY
At present atomic absorption spectroscopy is widely used for the laboratory

determination of metals in water.19 With the development of flameless systems and
instruments which can be fully computer-controlled it should be possible to incorporate
an atomic absorption instrument into an automatic monitoring system. The design of the
instrument would probably allow sophisticated control to be used. The advantage of this
instrument is that it would offer a means of monitoring specific metal levels in water.
An associated field is the determination of metals from their atomic emission lines
when excited in a high temperature plasma generated by radio-frequency power or d.c.
arc.20 At the time of writing, instruments are being announced which offer similar ease of
operation to atomic absorption systems. With increasingly low cost data processing and
optical developments such as fast Fourier transform systems and electronic image
dissectors the plasma spectrometer may offer a means of determining a large number of
elements on a continuous automatic basis.
7.24. REAL TIME
With any automatic data logging record there is a possibility that readings may have been
missed, power or electronic failures occur and if the data have been accumulated over a
period and the number of readings obtained is found to be incorrect then it may be
impossible to correctly assign times to the data, which would mean that the whole of the
data must be discarded. The recording of real time together with a day indication is
essential for data being logged from a monitoring station. The real time clock should be
independent of the station power supplies with standby batteries or should have the
facility of being linked to a digital time standard so that when power is resumed it can re-
establish the correct time.21 Such a time standard would probably be one of the radio
transmissions carrying digital time code information; for example, in the United
Kingdom the National Physical Laboratory transmits coded time information on the Post
Office transmitter at Rugby on 60 kHz, call sign M.S.F. Many other countries throughout
the world provide similar transmissions.
7.25. TELEMETRY
The data obtained from monitoring systems have maximum value if they can be relayed
immediately to enable correct plant operation, detection of pollution and other immediate
consequential actions.
Unless the monitor is actually on the works, data need to be relayed to interested
parties by a telemetric system. There are two methods of telemetry available and the user
must establish which is most cost effective for a particular application. The telemetry can
be continuously transmitted via dedicated lines (electrical or optical), radio or microwave
links. This has the advantage that current information is always available, but conversely
it is unlikely that a water system will undergo a dramatic change in condition. A
continuous link may prove expensive if the user cannot provide his own and has to hire or
rent such links and this cost may be difficult to justify. This system becomes more
important if security of communication is a high priority which ‘dial-up’ systems could
not give. Warning of failure of communication lines could also be given.
A lower cost alternative is to use a ‘dial-up’ or on demand system. This may use the
voice switched telephone network or a private communication system. Status reports
could be given at regular intervals or the equipment interrogated at will. In addition,
should alarm conditions occur an automatic call out would be initiated alerting the user to
the alarm. Devices of this latter type may be of very low cost giving bleeps to indicate
numerical values or incorporate a talk out device to tell the user the status. For increased
sophistication digital devices giving full status reports and possibly relaying data to a
central data store are available.
An extension of telemetry is to provide the receiver of data with a means of
retransmitting to the station, to effect changes or control plant, this being known as
telecontrol.
Telemetric systems will form an important part in the development of intelligent
monitoring systems.
7.26. DATA LOGGING*
Monitoring may simply be used to detect a pollution and activate an alarm, no further use
being made of the data, but this would be an extravagant waste of resources. Further use
can always be made of water quality data for modelling, archiving and establishing
average levels of determinands. The prime record in the monitoring station will probably
be on one or several recorder charts and data in this form are not suited to further analysis
other than by simple visual observation. The manual transcription of data from recorder
charts or even tally roll printers can be extremely tedious and prone to error.
If the monitor is equipped with an efficient means of data logging the analytical results
can be recorded onto some storage medium and when convenient these data can be
recovered and used with a data processing system to obtain graphs, tables, corrected
figures and archive data as required.
Various data recording mediums are available for use with monitoring systems and the
situation is one which may rapidly change in the light of modern electronic
developments.
Data logging on punched paper tape is now regarded as obsolescent. The relatively
high data error rate and the highly mechanical nature of a paper tape punch means that
they are not ideal in this application. The environment in which the paper would be kept
may not suit this type of data logging system. Another disadvantage is the high power
requirement and bulk of this system.
Magnetic media offer well established data recording systems and the two that could
be considered for water quality data logging are magnetic disk and magnetic tape. It is
difficult to see why anyone would wish to use magnetic cards for this task, but such items
are available.
Magnetic disks are normally used as medium speed random access devices on data
processing systems. They incorporate high precision mechanical devices and often
require reasonable amounts of power. Disk drives incorporate some housekeeping
electronics (which may itself be a computer) to ensure correct recording. The random
access nature of a disk is of little advantage when recording monitoring station data
unless there
*
There is some confusion over the term data logging. Here it refers to the recording of data on
some medium which can subsequently be utilised as a direct input to a data processing system. The
term often refers to a device scanning various sensors and outputting the data to a printer—this is
not what is referred to here.
was a pressing reason, say, to record its determinands in a different file, but this sort of
data manipulation is easily and more effectively achieved when recovering the logged
data with a computer system. The high writing (and reading) speed of a disk offers no
advantage on a system which is generating data at an incredibly low relative data rate.
The premium paid for this random access and relatively high speed is that the system is a
sophisticated high tolerance mechanical unit not ideally suited to monitoring
environments. Even with new low cost small floppy disk drives which are beginning to
be commonly available the disk has little to recommend it for this type of work.
Magnetic tape is a well established medium for both analogue and digital recording.
It offers probably the only means of recording true analogue data to enable this to be
replayed to a data processing system. It seems unlikely that few users would have a need
for true analogue records of their data and the cost of such a system may be high.
Digital tape systems probably offer the most effective way of logging data from
monitoring stations at the time of writing.22 It is obvious that this situation will soon
change dramatically and that solid state storage either onto bubble memory or random
access memory (RAM) will be readily available with cost and reliability benefits.
Magnetic tape is available in many forms which can and have been used for
monitoring stations.
7.27. CHART RECORDERS
Data as commonly presented on chart recorders are satisfactory for continual

surveillance, but often are not suitable to give a summary of results over, say, a week. A
common use of magnetic tape logged data is to turn this back into a chart with more
reasonable co-ordinates than the original record. It would be useful, therefore, if chart
recorder manufacturers could produce a recorder with charts of a convenient size (say
A4) with a grid marked in real time (and days if required) so that each week an easily
filable graphical output of results was obtained. Circular chart recorders approach this
idea, as do some flow and meteorological recorders, but there does not appear to be an
electrical input version with a rectangular format. Such a recorder may remove some
people’s requirement for data logging.
7.28. INTELLIGENT INSTRUMENTATION
The well publicised current growth in electronic technology is both increasing instrument
reliability and making available at low cost sophisticated data handling systems.23 These
developments will have significant effects on water quality monitoring instrumentation in
two respects.
First, it will be possible to have an intelligent device controlling and collating
individual sensor units and in overall control of the analytical system functioning. The
system would control routine operation and maintenance and such functions as sampling,
calibration, cleaning and data collection. The intelligence of the system would enable it to
examine data from its sensor units and evaluate this. It could query readings from sensors
and perform cleaning, zeroing and re-calibration steps before accepting the sensor
reading as valid and perhaps initiating an alarm or controlling the abstraction process.
Further, it could cross-correlate data from different sensors and detect possible sensor
malfunctions. An example might be a dramatic change in pH; this would be expected to
be accompanied by a change in conductivity but if this was not the case the system could
self test and if operating correctly alert the user to this observed state of affairs. A master
controller of this type could also undertake data processing and reduction tasks, perhaps
even producing hard copy summaries of measured data.
The other impact of increasingly intelligent devices will be that each individual
instrument or sensor system will contain a unit of this type. (This may also be a lower
cost option than current non-intelligent analogue or discrete logic technology.) The
instrument will be able to perform calibration, curve correction and data processing by
itself. It will then present a processed result which may be printed out or passed to some
central data system.
It follows that as instruments are becoming increasingly able to provide data for
collation by a central system a practical universal system of instrumentation
interconnection is developed and implemented by manufacturers and users.
7.29. INSTRUMENTATION BUSES
As instrumentation becomes increasingly intelligent and we move towards intelligently

controlled monitoring stations, there is a requirement for increasing ease of interchange
of data between instruments, controllers and data handling systems. The concept of the
data bus ideally suits itself to a monitoring station. All instruments and data recording
devices together with a station controller would ideally be plugged into identical sockets
on a common bus cable, similar in concept to a domestic ring main. The controller
manages the bus instructing readings to be taken, collecting and examining the data,
checking recalibration and cleaning and outputting the data to the recording devices. It
should be able to sense devices which have been plugged in or removed from the bus and
react accordingly.23,24
One such implementation of this concept is the IEEE 488 Interface bus (sometimes
referred to as GPIB or HPIB after the Hewlett Packard company who were responsible
for devising the system).
The task of interconnecting instrumentation for such purposes as monitoring stations
will be greatly facilitated by international standards being set and followed for the direct
plug to plug connection of instrumentation.
7.30. VIEWDATA
The current and future implementation of viewdata systems in many countries offers a
novel means of utilising data from water quality monitoring systems, together with other
relevant data including that obtained in the laboratory.25
The viewdata system enables a subscriber to access data stored on a computer facility
via the voice switched telephone network and a suitably modified domestic television
receiver. Keying or dialling a sequence of numbers a user gains access to the desired
information by a process of successive branching. Restriction of availability of data is
made possible by designing the system to require pass codes before restricted data are
made available.
Data can be recovered as alphameric characters or displayed graphically in
monochrome or various colours. Parts of the display can be made to flash and update to
show trends or historical data.
Data would be fed to the viewdata computer via telemetric input from automatic
monitoring system and via keyboard or automatic data input for laboratory data. A user
could interrogate data on the computer at will or if an alarm status was triggered an
automatic verbal telephone message would be sent to the user. On receipt of an alarm
message the user could investigate current data for the suspect site and call on previous
information to evaluate trends, etc. Data from other sites could also be called up to
establish trends or help identify the site of a pollution.
Future developments may allow the user telecontrol facilities in addition to simply
abstracting data.
7.31. CONCLUSION
Laboratory methods and automatic monitoring systems are both essential to water quality
monitoring. They complement each other and it is not likely that developments in the
immediate future will render the laboratory measurement side redundant.
Detailed elemental and cation and anion determinations are rarely carried out if at all
by automatic systems although instrumental developments will make monitoring of these
more feasible.
The identification of pollutants, especially if these are complex organic molecules,

may be made by a monitoring system but sophisticated laboratory techniques such as
mass spectrometry and spectroscopy together with experienced interpretation offer the
only way to obtain correct and reliable compound identification.
Technological developments will mean that automatic systems will be less prone to
failure and enable the integrity of the results to be higher. In both the laboratory and the
field instruments will become more reliable and capable of giving final corrected results
to analyses.
Developments in data processing and communication systems will enable easier
collation and handling of data, and developments in telemetry and viewdata systems will
enable better use to be made of these data.
New laboratory techniques such as plasma emission spectroscopy offer means of
performing analyses for a wide variety of species quickly and at relatively low cost, and
developments in fields such as mass spectrometry enable speedy identifications of
compounds to be carried out.
It is unfortunate that sensor development for automatic monitoring systems has not
made much progress in recent years. This is an area in which worthwhile research could
be made.
REFERENCES
1. SKEGGS, L.T., Amer. J. Clin. Pathol, 1957, 28, 311.

2. RUZICKA, J. and HANSEN, E.H., Analytica Chim. Acta, 1975, 78, 145–157.
3. STRICKLER, H.S.. STANCHAK, P.J. and MAYDAK, J.J., Anal Chem., 1970, 42, 1576.
4. WALKER, W.H. C., Clin. Chim. Acta, 1971, 32, 305.
5. GERKE, J.R. and FERRARI, A., Technicon Symp., 1967, 1, 531.
6. RUZICKA, J., STEWART, J.W.B. and ZAGATTO, E.A., Analytica Chim. Acta, 1976, 81, 387–
396.
7. RUZICKA, J., HANSEN, E.H. and ZAGATTO, E.A., Analytica Chim. Acta, 1977, 88, 1–16.
8. HOLAK, Anal. Chem., 1969, 41, 1712.
9. FERNANDEZ, Perkin Elmer A.A.Newsletter, 1973 (July-Aug.), 12, No. 4.
10. POLUEKTOV, VITKUN, ZELYUKOVA, Zh. Anal Khim., 1964, 19, 873.
11. LEAHY, J.S.and PURVIS, M., J. Instn Water Engrs & Scientists, 1979, 33, 311.
12. MORLEY, P.J., MUSTY, P.R. and COPE, J., J. Instn Water Engrs & Scientists (in press).
13. NATIONAL RESEARCH DEVELOPMENT CORPORATION and MACKERETH, F.J.H.,
British Patents 19294(1962) and 12882(1963); MACKERETH, F.J.H., J. Sci. Instrum., 1964,
41, 38.
14. CLARK, L.C., Trans. Amer. Soc. Artificial Internal Organs, 1956/7, 2, 41–47.
15. Her Majesty’s Stationery Office, Methods for the Examination of Waters and Associated
Materials, ‘The Measurement of Electrical Conductivity and the Laboratory Determination of
the pH Value of Natural, Treated and Waste Waters’, 1978.
16. MOODY, E.J.and THOMAS, J.D. R., Selective Ion Sensitive Electrodes, Merrow Publishing
Co. Ltd, Watford, 1971.
DURST, R.A. (Ed.), Ion Selective Electrodes, National Bureau of Standards, Washington, D.C.,
1969.
BAILEY, P.L., Analysis with Ion-Selective Electrodes, Heyden & Son Ltd, London, 1976.
WHITFIELD, M., Ion Selective Electrodes for the Analysis of Natural Waters,
Australian Marine Sciences Association, Sydney, Handbook No.2, 1971.
17. BRIGGS, R., J. Sci. Instrum., 1962, 39, 2–7.

18. HOLLAND, G.J., GREEN, A., STROUD, K.C.G. and JONES, D.B., Water Treatment &
Exam., 1975, 24, 81–119.
STENSEL, H.D., McDowELL, C.S. and RITTER, E.D., J, Water Pollut. Control Fed., 1976, 48,
2343–2350.
MARSHALL, W.C., ‘Non-contact Sensing of the Electrical Field surrounding Trout’, M.Sc. Thesis,
University of Wyoming, Laramie, Wyoming, 1972.
POELS, C.L.M., Water Treatment & Exam., 1975, 24, 46–56.
MILLER, W.F., ‘The Development of the Water Research Centre Water Quality Monitor using
Fish’, Water Research Centre Seminar on Practical Aspects of Water Quality Monitoring
Systems, Stevenage, 1977.
19. WALSH, A., ‘The Application of Atomic Absorption Spectra to Chemical Analysis’,
Australian Patent 23041/53.
WALSH, A., Spectrochimica Acta, 1955, 7, 108.
THOMPSON, K.C. and REYNOLDS, R.J., Atomic Absorption, Fluorescence and Flame Emission
Spectroscopy, Charles Griffin & Co. Ltd, London, 1978.
20. GREENFIELD, S., JONES, I.L. and BERRY, C.T., Analyst, 1964, 89, 713.
DICKINSON, G.W. and FASSEL, V.A., Anal. Chem., 1969, 41, 1021–1024.
21. Post Office, International and Maritime Telecommunications Region, Rugby Radio Station,
‘Standard Frequency and Time Broadcasts, Rugby Radio Station’.
CROSS, A.F., Wireless World, 1976 (Feb.) , 82, 30–35 et seq.
HELSBY, N.C., Wireless World, 1976 (Aug.), 82, 47–51 et seq.
22. European Computer Manufacturers Association, Standard Nos. 34 and 46.
23. McARTHUR, N., WINGFIELD, A.J. and WITTEN, I.H., Wireless World, 1979 (Dec.), 85, 46–
51.
24. Institute of Electrical and Electronics Engineers, Standard IEEE 488–1975.
American National Standards Institute (A.N.S.I.), Standard MC 1.1.
WITTEN, I.H., Wireless World, 1979 (Feb.), 85, 32–35 et seq.
COTTIS, R.A., New Electronics, 1979 (Dec.), 12, 70–74.
25. FEDIDA, S., ‘Viewdata—an Interactive Information Medium for the General Public using the
Telephone Networks’, 6th International Broadcasting Convention, 20–24 Sept. 1976.
In addition to these references the reader will find much valuable information in
manufacturers’ data and handbooks, in many of the standard texts on physics and
physical chemistry, and in current journals and periodicals which are essential in order to
keep in touch with the very rapid technological advances currently being made.
INDEX
Absorption refrigeration cycles, 139

Acrylic-amine resins, 78
Activated carbon, 29–46, 64
adsorption of solute molecules onto, 30–2
biological, 158
granular, 30, 35–46, 158
adsorber installation, 39–41
adsorber system design, 36–9
applications, 46
performance, 41
points of application, 36
regeneration system design and operation, 43–5
powdered, 32–5
applications and performance, 35
dosing systems, 33–4
point of addition, 32
regeneration, 32–3
Adsorbents, 46–51
Adsorption, 28, 30–2, 36–41
Aluminium sulphate, 184
Ammonia, 59–63
removal, 63–74
biological nitrification, 67–74
breakpoint chlorination, 63–6
ion exchange resins, 66–7
physico-chemical methods, 66–7
Ammonia-chlorine process, 143, 144
Ammonium hydroxide, 81
Anion exchange resins, 46
Anionic detergents, 49
Atomic absorption spectrophotometry, 199, 214
Automation, 191
Bactericidal residuals, 154

Batch analysers, 191, 197–9
BCD (binary-coded-decimal) output, 203
Belgarde range of chemicals, 98
Belt pressing, 180
BET method, 35
Biological activated carbon (BAC), 158
Biological denitrification, 75–7, 81–6
Biological methods, 62–3
Index 190
Biological nitrification, 67–74

Biological sensors, 210
Bleaching powder, 146–7
Boiler feedwater treatment, 46, 119–23
‘Breakpoint’ chlorination, 63–6, 144
Brine concentrate disposal, 126
Bromine, 142
Carbon
activated. See Activated carbon
total organic, 201
Carbonaceous adsorbents, 47
Carcinogens, 59, 158
Carman-Kozeny equation, 8–9, 175, 181
Cellulose acetate membranes, 52, 53, 107–9, 113, 121
Cellulosic ion exchange resins, 47
Centrifuging, 181–2
Chart recorders, 217
Chemical methods, 61
Chemical oxygen demand (COD), 196
Chemseps continuous loop, 79
Chloramines, 63, 141, 143, 154
Chloride determination, 197
Chlorination, 63–6, 143, 144, 152–4, 158
Chlorine, 65, 66, 141, 143–54, 158
available, 145, 146, 152
chemistry of, 147–52
dioxide, 29, 142, 145, 146, 154–6, 158, 160
chemical and bactericidal properties, 155–6
generation, 155
history, 154–5
dose-residual curve, 149
forms of, 145–7
history, 143
production, 146
residual, 152
solutions, 146
Chlorite, 158
Chloroform, 158
Clarification, 28, 142
Cleaning, 212–14
air, 213
automatic, 212
chemical, 213
mechanical, 212
ultrasonic, 212–13
Clinoptilolite, 66
Coagulant
recovery, 185
replacement, 184–5
Coagulation, 28, 142, 165–6
Index 191
Colloidal material, 26
Colloids, fouling by, 122
Colour removal, 26
Compressibility index, 176
Conductivity, 210
Continuous distillation head, 195
Continuous flow analysers, 191–7, 211
Control of Pollution Act 1974, 168
Cross-contamination, 196
Cyanide determination, 195
Cysticidal residuals, 154
2,4-D, 53
D’Arcy equation, 175
Data
collection, 202–4
logging, 214, 216–17
processing, 202–4
Demineralisation, 47, 74
Denitrification, 75–7, 81–6
Desalination, 89–140
miscellaneous processes, 138–9
multi-stage flash distillation, 90–106
reverse osmosis, 106–31
Dialysis unit, 193, 194
Dichloramine, 64, 65, 149, 160
Discrete analysers, 197–9
Disinfection, 141–62
agents used in, 141–2
by-products, 157–60
mixed treatments, 158
relative costs, 157
Distillation processes, 90–106
multiple effect, 92, 100
multi-stage flash. See Multi-stage flash (MSF) distillation
Dosing systems, 33
DPD test, 160
Drying beds, 178–9
‘Ducol’ process, 78
Dust control, 33
Electrodialysis, 131–8
capacity installed, 135
costs, 136–8
description of process, 131–6
polarity reversal, 135
principle of, 132
size, 134
systems available, 132
Emission spectroscopy, 201
Energy
Index 192
consumption, 94
recovery, 124
Enzyme mechanisms, 74
Ethyl cellulose membranes, 52
Evaporation, 179
Evaporator
costs, 102–6
materials, 100–2
Faraday’s Law, 132

Feed pumps, 124
Ferrous hydroxide, 74
Filterability index, 23
Filter (s), 158, 165
backwashing, 20–1, 169
cleaning, 11
continuous screen, 123
control, 19
conventional, 4
‘dry’ sand, 72–3
hydraulic control, 20
Immedium, 15–16
L’Eau Claire, 16–17
micron, 123
pilot-scale, 22
pressing, 180–1
radial flow, 17–18
rapid, 5–6
rotary vacuum, 179–80
slow, 3–5, 19
slow sand, 26, 28, 165
Tenten, 18
types of media, 13
upflow, 15–17
Filtration, 1–24, 142, 169–71, 176
applications, 2–3
attachment mechanisms, 8
conventional, 3
design methods, 21–3
dual and multi media beds, 13
effective use of bed capacity, 12–18
head loss
during, 9
relationships, 14
hydraulics, 8–10
Filtr ation—con td.
mathematical modelling, 10, 22
nature of suspension to be treated, 23
process
details, 6–12
modifications, 18–19
Index 193
solids removal, 10
transport mechanisms, 7
trends and developments, 12–23
Fish behaviour, 210
Fishmoor plant, 171
Flameless atomic absorption, 199
Flash chambers, 101–2
Flotation, 183–4
Fluidised bed
MSF, 98–100
system, 71
Fluidised sand units, 81–6
Fouling
colloids, by, 122
index test, 121
Freeze-thaw process, 182
Freezing processes, 91, 138–9, 182
Freundlich isotherm, 30
Gas chromatography with mass spectrometry (GCMS), 202

Granular carbon. See Activated carbon
Groundwaters, 122
Heat-pumping, 138
Hewlett Packard Interface Bus, 203
High Test Hypochlorite, 146
Humic acids, 49
Hydrochloric acid, 145, 147
Hypochlorites, 146, 147
Hypochlorous acid, 145, 147
Immedium filters, 15–16

Immobilised enzymes, 74
Instrumentation, 191, 218
buses, 218–19
Iodine, 142
Ion exchange resins
applications, 49
desalination, 91
evaluation methods, 48–9
nitrate removal, 75–81
nitrogen compounds, 61, 66–7
organics removal, 46–51
test results, 49
theory, 48
Ion selective electrodes, 208–9
Iron (III) thiocyanate, 197
Kozeny equation, 8
Index 194
Lamella sedimentation, 184

LaMotte test kit, 161
Langelier Index, 120
L’Eau Claire filter, 16–17
Lindane, 53
Membrane(s)
composite, 109
configuration, 54, 109
hollow fibre, 112
materials, 52–3
performance characteristics, 111
processes, 106, 131
product flux, 112
types, 107, 109–15
Metal
analysis, 199–201
detection limits, 201
Methaemoglobinaemia, 59
Methanol, 83, 84, 86
Microstrainers, 165
Monitoring, 189–222
automatic, 204–5
real time, 214–15
station design, 213
Monitors
basic types, 205
design considerations, 205
submersible, 205–7
Monochloramine, 145, 146, 160
Monochromator, 201
Multiple effect distillation, 92, 100
Multi-stage flash (MSF) distillation, 90–106
evaporator costs, 102–6
fluidised bed, 98–100
Nitrate, 59–61
removal, 74–86
biological denitrification, 75–7, 81–6
ion exchange, resins, 75–81
possible methods, 74–7
Nitrification, 67–74
Nitrogen, 65
compounds, 59–88
trichloride, 64, 65, 149, 160
Nitrosamines, 59
Odorous substances, 30
Ohm’s Law, 132
Organic compounds, 25–7
health effects, 26
Index 195
objectives or removal, 26
oxidative processes, 29
removal by conventional processes, 26
water, in, 25–6
Organic contaminant monitoring, 201–2
Oxidation-reduction potential (ORP), 209
Oxidative processes, 29
Oxidising agents, 29
Oxygen, dissolved, 207–8
Ozone, 29, 142, 156–8
application, 156–7
applied doses, 157
generation, 156–7
history, 156
Permanganate Value (PV), 28

Permasep polyamide, 118
Pesticides, 28, 53
pH effects
chlorination, 148, 153
chlorine dioxide disinfection, 156
desalination, 123
monitoring, 208
phenol uptake, 49
residual chlorine, 152
reverse osmosis, 52
Phenol
determination, 195
uptake, 49
Phosphoric acid, 195
Physico-chemical methods, 61, 66–7
Plasma emission spectroscopy, 214
Pollution, 207, 211
Polyamide, 107, 114, 118
Polyelectrolytes, 170, 172, 173, 181, 182, 185
Polymeric resins, 47
Polynuclear aromatic hydrocarbons (PAH), 28, 52
Polysulphone (TFC), 107, 111, 113, 115
Potassium
iodide, 145
permanganate, 29
Precipitation, 28
softening, 166
Rate constant, 70
Real time, 214–15
Redox potential, 209
Reference electrodes, 209
Reverse osmosis, 51–4
basic principles, 107
costs, 131
Index 196
desalination, 90, 106–31

disposal of brine concentrate, 126
feed pumps, 124
instrumentation, 126
nitrogen compounds, 61
organic compounds, 51–4
plant developments and installations, 127–31
post-treatment of product water, 126
systems design, 123–7
Ring-type electrodes, 210
Samplers, 207
Sand-washing plants, 165
Scale control, 96–8, 119
Sedimentation, 28, 142, 171, 184
Sensors, 204, 210
Silting Density Index (SDI), 121
Sirotherm ion exchange process, 138
Sludge, 163–88
coagulant-based, 181
conditioning, 178
dewatering, 168–70
capital costs, 168
further methods, 178–82
theory, 175–8
estimation of, 186
lime addition, 181
primary sedimentation, 165
produced by
coagulation, 165–6
iron and manganese removal plants, 166
precipitation softening, 166
produced from chemical treatment process, 165–6
produced without chemical treatment, 165
thickening, 170–5
treatment
operational aspects, 186–7
optimisation process, 183–6
types of, 164–5
use of term, 164
waterworks, 168–70
Sodium
bisulphite, 153
chloride, 78
sulphite, 153
thiosulphate, 153
Solar distillation, 91
Solar energy, 138
Specific resistance, 176
Sterilisation, 213
Stocks plant, 173
Index 197
Styrene-divinylbenzene resins, 78
Sulphur dioxide, 153
Suspended solids, 209–10
Telemetry, 215
Temperature measurement, 211–12
Tenten filter, 18
Tintometer Lovibond test kit, 160
Trace analysis, 191
Transport mechanisms, 7
Trihalomethanes (THM), 65–6
Ultrafiltration, 54
Ultraviolet radiation, 142
Vapour compression (VC) processes, 92

Viewdata, 219–20
Wastewater recycling, 164

Water-boxes, 101
Water Research Centre, 175
‘Wedge-wire’, 178
Zeta potential, 122

Developments in Water Treatment 2 by W.M.Lewis PDF

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Developments in Water Treatment 2 by W.M.Lewis PDF

Caricato da

Copyright:

Formati disponibili

DEVELOPMENTS IN WATER

APPLIED SCIENCE PUBLISHERS LTD

ISBN 0-203-97492-1 Master e-book ISBN

ISBN 0-85334-903-7 (Print Edition)

‘Surveillance of Drinking Water Quality’ published by WHO in 1976 in its introduction

Some form of filtration is almost certain to be incorporated in the treatment used to

1.2. APPLICATIONS OF FILTRATION

1.3. CONVENTIONAL FILTRATION

Before discussing recent developments in the filtration of water it is useful to briefly

1.3.1. Slow Filters

finished water is unlikely to be objectionable although of course levels of nitrate nitrogen

1.3.2. Rapid Filters

1.4. THE FILTRATION PROCESS

1.4.1. Transport Mechanisms

1.4.2. Attachment Mechanisms

1.4.3. Hydraulic Considerations

FIG. 1. Schematic arrangement of

FIG. 2. Head loss during filtration.

where H0=Carman-Kozeny head loss, and K=a constant.

where C0=feed concentration to bed, and t=time.

1.4.4. Solids Removal

1.4.5. Filter Cleaning

1.5. TRENDS AND DEVELOPMENTS IN FILTRATION

1.5.1. Effective Use of Bed Capacity

Dual and Multi Media Beds

FIG. 3. Head loss relationships for

FIG. 4. (a) Immedium type upflow

FIG. 5. Typical radial flow filter with

1.5.2. Process Modifications

1.5.3. Filter Control

1.5.5. Design Methods

FIG. 6. Determination of optimum bed

FIG. 7. Response surface for optimum

1. CARMAN, P.C., Trans.Instn Chem. Engrs, 1937, 15, 150.

2.1. ORGANIC COMPOUNDS IN WATER

2.3. THE PERFORMANCE OF CONVENTIONAL AND

Organochlorine Possible health 0·1 µg/litre

Preparation of Pyrogens Cause –

2.4. ACTIVATED CARBON

where Cc is the concentration of solute in the carbon at equilibrium, Cw is the

2.4.3. Powdered Activated Carbon

Engineering of Dosing Systems

FIG. 1. Costs of powdered carbon

FIG. 2. Schematic of powdered carbon

2.4.4. Granular Activated Carbon

Adsorber System Design

FIG. 3. Idealised adsorption front

FIG. 4. Granular carbon adsorber

FIG. 5. Granular activated carbon

Performance of Granular Activated Carbon Adsorbers

Regeneration System Design and Operation

Granular Activated Carbon Applications

2.5. ION EXCHANGE AND POLYMERIC ADSORBENTS

2.5.2. Resins for Organics Removal

functional groups.23 Carbonaceous adsorbents are available in a variety of polarities and

2.5.4. Methods of Evaluation

2.5.5. Test Results

TABLE 5 SUMMARY OF REPORTED DATA

2.6. REVERSE OSMOSIS AND ULTRAFILTRATION

FIG. 6. Reverse osmosis schematic.

2.6.2. Removal Mechanisms

2.6.3. Experimental Results

2.6.4. Membrane Configurations