Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
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
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
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
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
Developments in water treatment—2 4
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.
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.
Developments in water treatment—2 6
(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)
(3)
(4)
(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.
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
Developments in water treatment—2 10
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.
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.
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.
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
Developments in water treatment—2 12
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.
Developments in water treatment—2 14
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
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.
1.5.4. 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.
Developments in water treatment—2 18
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.
REFERENCES
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.
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.
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)
Developments in water treatment—2 24
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
Developments in water treatment—2 26
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.
2.4.1. 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
Removal of organic compounds 27
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
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.
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.
Removal of organic compounds 29
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.
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.
Removal of organic compounds 31
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.
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.
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
Removal of organic compounds 35
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).
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,
Removal of organic compounds 37
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
Developments in water treatment—2 38
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
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
Developments in water treatment—2 40
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
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
2.5.3. 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.
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.
2.5.6. 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
Developments in water treatment—2 44
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.
Developments in water treatment—2 46
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.
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.
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)
2.6.5. 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:
Removal of organic compounds 49
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.
Developments in water treatment—2 50
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.
Removal of organic compounds 51
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
Developments in water treatment—2 54
(<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
Developments in water treatment—2 56
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.
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
Removal of nitrogen compounds 59
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.
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.
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.
Developments in water treatment—2 62
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.
Removal of nitrogen compounds 63
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.
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
Removal of nitrogen compounds 69
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
Removal of nitrogen compounds 71
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
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)
Removal of nitrogen compounds 75
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).
Developments in water treatment—2 76
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.
Removal of nitrogen compounds 77
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.
Developments in water treatment—2 78
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
Developments in water treatment—2 80
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.
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
(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.
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
(1)
(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.
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.
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.
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.
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.
Operating Costs
Developments in water treatment—2 92
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.
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).
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.
Developments in water treatment—2 96
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.
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
Developments in water treatment—2 98
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
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
(mg/litre) (mg/litre)
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
Developments in water treatment—2 100
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
(mg/litre) (mg/litre)
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.
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.
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
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.
Developments in water treatment—2 106
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
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.
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.
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
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.
4.4. 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
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
(mg/litre) (mg/litre)
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)
Developments in water treatment—2 118
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.
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
Developments in water treatment—2 120
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
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
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
Developments in water treatment—2 124
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).
Developments in water treatment—2 126
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:
Developments in water treatment—2 128
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.
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
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.
Developments in water treatment—2 134
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.
5.3.1. 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
5.4. 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.
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.
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.
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
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
REFERENCES
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.
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.
batches of backwash water when they arise will therefore contain the accumulated
suspended solids from a considerable volume of throughput water.
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
Sludge treatment and disposal 145
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
Developments in water treatment—2 146
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.
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
Sludge treatment and disposal 147
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.
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
Developments in water treatment—2 148
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
Sludge treatment and disposal 149
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
Developments in water treatment—2 150
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
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.
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:
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.
There are six principal methods of further dewatering of coagulant based sludges, each of
which is examined below.
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.
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
Sludge treatment and disposal 157
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
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
Developments in water treatment—2 158
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.
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.
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.
6.11. 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
Sludge treatment and disposal 161
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
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.
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
Developments in water treatment—2 164
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.
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
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
Water quality monitoring 167
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
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
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
Developments in water treatment—2 170
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.
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
Developments in water treatment—2 172
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.
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
Water quality monitoring 173
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.
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.
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.
Water quality monitoring 175
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.
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.
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.
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
Water quality monitoring 177
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.
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
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.
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.
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
Water quality monitoring 179
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
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
Developments in water treatment—2 180
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.
Water quality monitoring 181
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.
Water quality monitoring 183
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.
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.
Developments in water treatment—2 184
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.
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.
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.
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.
Water quality monitoring 187
REFERENCES
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
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
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
Kozeny equation, 8
Index 194
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
Rate constant, 70
Real time, 214–15
Redox potential, 209
Reference electrodes, 209
Reverse osmosis, 51–4
basic principles, 107
costs, 131
Index 196
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