Developments in Water Treatment 2 (Development Series)

DEVELOPMENTS IN WATER TREATMENT—2

THE DEVELOPMENTS SERIES Developments in many fields of science and technology occur at such a pace that frequently there is a long delay before information about them becomes available and usually it is inconveniently scattered among several journals. Developments Series books overcome these disadvantages by bringing together within one cover papers dealing with the latest trends and developments in a specific field of study and publishing them within six months of their being written. Many subjects are covered by the series including food science and technology, polymer science, civil and public health engineering, pressure vessels, composite materials, concrete, building science, petroleum technology, geology, etc. Information on other titles in the series will gladly be sent on application to the publisher.

DEVELOPMENTS IN WATER TREATMENT—2 Edited by

W.M.LEWIS M.Chem.A., C.Chem., F.R.I.C., F.I.W.E.S. WHO Consultant EURO, Environmental Health—Drinking Water Quality, Copenhagen, Denmark Managing Director, Coventry Chemical Consultancy Ltd, Coventry, UK

APPLIED SCIENCE PUBLISHERS LTD LONDON

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

ISBN 0-85334-903-7 (Print Edition) WITH 24 TABLES AND 54 ILLUSTRATIONS © APPLIED SCIENCE PUBLISHERS LTD 1980 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of the publishers, Applied Science Publishers Ltd, Ripple Road, Barking, Essex, England

PREFACE ‘Surveillance of Drinking Water Quality’ published by WHO in 1976 in its introduction stated that ‘Public health protection of drinking-water supplies should assure that each component of the system—source, treatment, storage and distribution—functions without risk of failure’. Drinking water is perhaps, together with the air we breathe, a unique commodity in that the general population is normally permitted no freedom of choice, so the assurance that the water available for drinking is of the highest quality is of paramount importance. Since the universal introduction of disinfecting agents in water supplies in developed countries, risks to health from microbially contaminated drinking water have been dramatically reduced. Today the problem confronting personnel responsible for ensuring the public’s water supply is perhaps of a much more subtle character brought about, in part at least, by the rapid progress in analytical chemistry and the environmental awareness of the general public resulting in the demand for the creation of standards of quality for drinking water. Chemicals present in raw water supply range from simple ions extracted from soil and minerals in the watershed to (in some instances) unidentified waste products from the chemical industry, the length of the list being limited only by the capabilities of the analytical chemists and their instruments. Some medical researchers proclaim that the presence, or absence, of a certain substance in drinking water is directly associated with the differences in death rates from specific diseases or the incidence of morbidity. The problem is compounded due to the fact that the toxic effects of many identified chemicals are insufficiently understood. Thus the responsibility devolving upon the shoulders of personnel responsible for ‘Treatment’ is to provide a process (or combination of processes) which will ensure, as far as is practicable, that the water supply is not only aesthetically acceptable, but also of the best chemical standard and in a condition which will not damage the integrity of the distribution system which could result in subsequent and additional contamination. With rivers of the calibre of the Danube, Trent and Rhine, to mention but three, the resulting treatment, to provide drinking water whose quality is beyond suspicion, needs to be very sophisticated. The ‘treatment process’ is not a single identifiable parameter but is dependent upon the nature and quality of the raw material and may for example involve only simple filtration or filtration plus disinfection. On the other hand if the quality of the supply water is from a lowland river, such as the three previously mentioned, then it follows that a combination of individual processes, commencing with coagulation for the removal of suspended matter, etc., and employing perhaps the majority of the techniques described, will be essential to provide the required quality of drinking water. Within this series will be found the various important facets of treatment each written by an author, expert in the particular field, who has introduced his topic with a brief

historical background before providing the reader with the most up-to-date information available on the subject. It is a salutary thought that in 1975 (latest information available) some 78% of the world’s rural population and even 22% of the urban population were without an adequate water supply. Of the urban population of the world having access to a piped water supply (77%), 57% only had house connections and 54% of the population served by public piped supply received it only on an intermittent flow basis. Conscious of the urgent need to rectify these shortcomings, the UN Water Conference—Mar del Plata, March 1977—urged the adoption of ‘The International Drinking-Water Supply and Sanitation Decade, 1981–1990’. The aim of the latter is to encourage and assist all countries of the world to adopt programmes with realistic standards for both quality and quantity and to provide water to all people by 1990, if possible. It is unfortunate, but nevertheless true that at present, and for how long into the future we know not, many countries—not only the developing ones—are experiencing financial constraints of varying magnitude. To achieve the above objectives will therefore strain the ingenuity and professional expertise of all concerned with the task of supplying the community with drinking water. It is consequently singularly appropriate that these first two volumes on ‘Developments in Water Treatment’ should be available at this time, for within their pages will be found that information on ‘Treatment’, appropriate to the needs, to enable people to overcome the financial constraints laid upon them. W.M.LEWIS

CONTENTS Preface

v

List of Contributors

x

1. Filtration T.H.Y.TEBBUTT 2. Removal of Organic Compounds C.S.SHORT 3. Removal of Nitrogen Compounds R.B.GAUNTLETT 4. Desalination M.J.BURLEY and J.D.MELBOURNE 5. Disinfection A.T.PALIN 6. Sludge Treatment and Disposal M.A.HILSON 7. Water Quality Monitoring P.J.MORLEY and J.COPE Index

1 22 53 79 122 142 162

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-onThames, Oxon., RG9 1DP, UK. P.J.MORLEY Principal Scientist, Avon Division, Severn-Trent Water Authority, Avon House, De Montford Way, Cannon Park, Coventry, CV4 7EJ, UK. A.T.PALIN Consulting Chemist, 7 Montagu Court, Montagu Avenue, Newcastle upon Tyne, NE3 4JL, UK. C.S.SHORT Yorkshire Water Authority, Olympia House, Gelderd Road, Leeds, LS12 6DD, UK. T.H.Y.TEBBUTT Senior Lecturer, Department of Civil Engineering, The University of Birmingham, P.O. Box 363, Birmingham, B15 2TT, UK.

Chapter 1 FILTRATION T.H.Y.TEBBUTT, B.Sc., S.M., Ph.D., M.I.C.E., M.I.W.E.S. Senior Lecturer, Department of Civil Engineering, The University of Birmingham, Birmingham, UK SUMMARY Developments in water filtration have enabled a more efficient use of the process and have provided a better understanding of the mechanisms involved. Good filtrate quality can be achieved at higher loadings than would have been considered possible some years ago and the advent of dual and multi media beds has brought economic benefits. A consequence of more effective use of bed capacity is that the process is more likely to be influenced by the nature of the suspended solids in the feed. A consequence of obtaining deeper penetration of suspended matter into a filter bed is the need to ensure adequate cleaning arrangements. The complexity of the filtration process is such that mathematical models cannot as yet provide a universal solution for any situation but their existence is of value in helping to develop more efficient filtration plants.

1.1. INTRODUCTION Some form of filtration is almost certain to be incorporated in the treatment used to produce potable water from most surface water sources. Indeed, the principle of filtering rain water through a layer of porous material to remove suspended matter has been used for centuries although the first recorded purpose-built sand filters for a public water supply were not installed until 1804 in Paisley. The early installations were of the slow filter type and their considerable purification ability in no small way contributed to the great improvement in public health which occurred during the second half of last century. By 1870 the rapid pressure filter concept made its appearance, to be shortly followed by the rapid gravity filter whose basic design has not changed greatly to the present day. However, since the introduction of filtration as a water treatment process a considerable amount of knowledge has been accumulated about the behaviour and performance of the process. The continuing pressures to require more efficient treatment, in operational and economic terms, and the demands for higher quality filtrates have meant that a variety of developments to and modifications of the conventional filtration process have been introduced. Coupled with these developments has been an intensive study of the actual mechanisms involved in filtration with the aim of providing a better understanding of the process to aid in future design. Filtration through porous media is however a process of great complexity and it is not yet possible to produce a completely satisfactory theoretical model.

Developments in water treatment—2

2

1.2. APPLICATIONS OF FILTRATION Except in rare cases all surface waters will contain, at least at times, sufficient suspended matter to given turbidities in excess of 5 units which is the highest desirable level stipulated in the World Health Organisation International Drinking Water Standards. EEC standards set out a guide level of 5 FTU with a maximum admissible concentration of 10 FTU. In the USA the National Interim Drinking Water Standards lay down a mandatory turbidity limit of 1 unit for most surface-water-derived supplies. The American Water Works Association has adopted a turbidity goal of 0.1 units and it has been suggested that this may in the future be reduced to 0.05 units. The rationale behind the adoption of such low turbidity levels is not that the turbidity itself is likely to be harmful but that its presence may inhibit disinfection processes. With such limits for turbidity being required, even high quality impounded waters would need turbidity removal for much of the time and all lowland river derived supplies would necessitate comprehensive treatment for turbidity removal. With high turbidity raw waters, chemical coagulation and sedimentation are commonly employed to remove the bulk of the turbidity with filtration providing the final polishing stage to bring turbidity down to around 1 unit. With coloured upland sources, turbidity in the raw water may not be a serious problem but chemical coagulation is frequently used to remove the colour by what is in effect a precipitation process so that here again filtration is necessary following sedimentation. With low turbidity raw waters direct filtration using either gravity or pressure rapid filters aided by chemical coagulation is widely used and in parts of the UK and Europe slow sand filters are used either as a single stage of treatment or as a second stage following rapid filtration used as a roughing or preliminary stage of treatment. Whilst all these types of filtration should be able to produce filtrates with turbidity less than 1 unit the reliable production of turbidities less than 0·1 or 0·05 units is another matter. It is important to appreciate that most treatment processes, including filtration, follow a law of diminishing returns in that the unit cost of successive increments of purification increases, sometimes almost exponentially. Before setting stringent turbidity levels it should therefore be considered whether such levels are achievable by normal means of treatment and whether the cost of such treatment is justifiable in respect of the marginal improvement in water quality which they would produce. It is of interest to note that developments in wastewater purification have resulted in the adoption of filtration as a tertiary treatment stage in situations where the effluent is discharged to a receiving water with little dilution or which provides a raw water source some distance downstream of the outfall. In such circumstances sand filtration, usually in rapid gravity units, removes a considerable amount of the suspended solids which escape from final settling tanks together with the organic matter associated with these suspended solids. Thus a normal secondary effluent of 30 mg/litre suspended solids and 20 mg/litre biochemical oxygen demand can be polished by filtration to give a final effluent of around 10 mg/litre suspended solids and biochemical oxygen demand. The adoption of tertiary filters has encouraged a considerable amount of research into design and operational factors which has enhanced knowledge of the filtration process.

Filtration

3

1.3. CONVENTIONAL FILTRATION Before discussing recent developments in the filtration of water it is useful to briefly outline what may be considered as conventional usage which had not changed greatly until the early 1970s and which is summarised in Table 1. 1.3.1. Slow Filters As indicated earlier, the first form of sand filter was the slow type, so called because of its relatively low hydraulic loading of 1–4m3/m2d. Because of this low hydraulic loading there is only superficial penetration of suspended matter into the bed and filter runs of several weeks or even months are

TABLE 1 CHARACTERISTICS OF CONVENTIONAL FILTERS Characteristic

Slow filter Rapid filter

Filtration rate

1–4 m3/m2d

Size of bed

About 2000 m2 1m unstratified 0·5–1·0 mm 2·0–2·5

100–150 m3/m2d About 100 m2 0·5–0·8 m stratified 0·5–1·5 mm 1·2–1·5

Up to 1 m 20–90 days 0·2–0·6% of output

Up to 2·5 m 1–5 days 1–6 % of output

Depth of bed Sand size Uniformity coefficient Head loss Length of run Cleaning water consumption

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

finished water is unlikely to be objectionable although of course levels of nitrate nitrogen of more than a few mg/litre are to be viewed with suspicion. Because of their large size slow filters are labour intensive as regards cleaning operations and although a number of plants have been equipped with mechanical scraping devices, the cleaning of slow filters is a costly operation. For this reason it is imperative that they are only used for relatively low turbidity (10–20 units) raw waters. Because of this limitation, slow filters are not normally suitable for sources such as lowland rivers with high turbidities which usually require chemical coagulation since the additional carry-over of floc from settling tanks would be sufficient to rapidly clog the filter surface. The increased dependence on lowland sources has resulted in a decline in the use of slow sand filters in many parts of the world. However in the UK much of London’s water is treated by slow filters which follow primary rapid filters. In the Netherlands slow filters have found considerable application as the final treatment stage when purifying heavily polluted river waters such as those from the Rhine. 1.3.2. Rapid Filters The introduction of rapid sand filters no doubt arose partly from the seemingly inevitable trend to increase the capacity of treatment units and the need to handle high turbidity raw waters. The land area required for slow filters became a factor when constructing treatment plants for large communities so that considerable pressures were exerted on designers to develop more compact units. The result was the rapid filter with a hydraulic loading of around 100 m3/m2d. At this rate of flow, penetration of solids into the bed is considerable and clogging to a maximum allowable head loss takes only a day or two depending upon feed quality. Alternatively, penetration of turbidity may be complete resulting in breakthrough of suspended matter in the filtrate which will terminate the run even though the head loss limit may not have been reached. The deep penetration of solids and frequent need for cleaning would of course rule out the type of cleaning operation used for slow filters. Fortunately, however, the incorporation of backwashing systems was able to provide a practical and economical means of cleaning rapid filters. When using the normal bed of graded media, backwashing results in a stratified bed which is not desirable from the point of view of optimum utilisation of bed capacity but which has been generally accepted as a characteristic of rapid filtration. The bulk of filtration capacity worldwide is in the form of rapid gravity filters preceded, where appropriate, by chemical coagulation and sedimentation units. In the UK, rapid pressure filters with direct injection of coagulants find favour in hilly areas where their hydraulic characteristics prove useful when siting treatment plants but in other parts of the world pressure filters tend to be mainly used for industrial water supplies. For low turbidity raw waters rapid filtration alone, possibly with the addition to the raw water of a small dose of coagulant, can provide a filtrate of satisfactory quality. In most cases, however, the bulk of the turbidity is removed by preliminary coagulation and sedimentation processes which prolong filter runs. As mentioned in the previous section, rapid gravity filters are also used as the primary stage in a double filtration process. In the case of pressure filters separate coagulation and sedimentation tanks are not feasible so that the coagulant is dosed directly into the raw water and flocculation takes place largely within the filter bed. This does of course mean that a much larger

Filtration

5

amount of suspended matter is trapped by the filters and thus filter runs are much shorter than if the same water were treated by conventional gravity filters preceded by coagulation and sedimentation units. The omission of these units does however give a cost reduction to offset the shorter filter runs and more frequent washing required by pressure filters. An important feature of rapid gravity filters is the arrangement made for backwashing and in particular the design of the underdrain and filter floor. It is important that the active filter medium is supported in such a way that it is contained within the filter shell whilst preventing any short circuiting of flow during filtration or backwashing. The commonest system in the UK is to employ a concrete floor with plastic collection and distribution nozzles covered by a layer of graded gravel but there are a considerable number of proprietary filter bottoms. Some of these are of complex and costly design and it is by no means certain that any advantages they may have are cost effective. There are also variations in the form of filter control and backwash procedures adopted by different manufacturers and designers. A variation of the pressure filter, used quite widely in the UK for swimming pool water treatment, employs diatomaceous earth or cellulose powder deposited on rigid formers as the filtering medium. These units can give excellent turbidity removal and are found suitable for small potable supplies in some parts of the world, notably in the USA. In operation it may be beneficial to dose the raw water with a small concentration of the filtering medium. At the end of a run the medium together with the trapped suspended matter is discharged to waste and the filter element recoated with powder. The units are relatively small in size and can provide an economic form of treatment for low turbidity supplies.

1.4. THE FILTRATION PROCESS At first sight the manner in which a bed of porous medium removes suspended particles from a fluid passing through the bed might appear to be easily explained as some form of straining action. Clearly, if the suspended particles are larger than the voids in the bed they will be removed in the same way as a sieve prevents the passage of particles larger than the mesh size. Such a removal mechanism would mean that no penetration of solids into the bed could occur and that there would therefore be no need for a deep bed of medium. Examination of the solids removal performance of a granular bed shows, however, that particles very much smaller than the voids in the bed are effectively removed. Thus in a sand bed with grain size of 0·5–1·0 mm the voids are likely to be of the order of 100–200 µm. In such a bed, removal of fine colloidal particles such as silt and bacteria with sizes of about 1 µm are readily demonstrated. It is evident, therefore, that straining cannot be the only mechanism operating in a bed of granular medium and indeed in most situations straining plays a relatively insignificant part in the overall removal of turbidity. Removal of suspended particles is controlled by a number of transport and attachment mechanisms. The transport mechanisms move particles into the vicinity of grains of medium in the bed when the attachment mechanisms can operate to trap the particle.

Developments in water treatment—2

6

1.4.1. Transport Mechanisms In most filters laminar flow conditions prevail and thus the velocity in a void within the bed varies from zero at the grain surface to a maximum at the centre of the void. Transport mechanisms thus have to be able to move suspended particles out of the flow streamlines into lower velocities near the bed grains. The important transport mechanisms in water filtration are: 1. Interception: streamlines may pass close enough to bed grains so that suspended particles actually come into contact with the grain surface. 2. Diffusion: colloidal particles are influenced by random Brownian movements which may bring them into the vicinity of the grain surface. This mechanism is only important for very small particles (<1 µm). 3. Sedimentation: gravitational forces acting on particles can move them across streamlines and bring them to rest on upward-facing grain surfaces. 4. Hydrodynamic: particles in a velocity gradient tend to rotate and are thus subjected to lateral forces which can move them across streamlines. Because of the relation between viscosity of water and temperature, changes in temperature can significantly alter the effectiveness of transport mechanisms. 1.4.2. Attachment Mechanisms Once suspended particles have been brought into close proximity to grain surfaces or existing deposits, attachment can take place under the influence of physico-chemical and molecular forces similar to those which operate in the coagulation process. Thus the concepts of bridging and double layer formation in coagulation can be used to explain the entrapment of particles in a granular bed. It is postulated that attachment is due to the interactions between the approaching particles and the bed grain as a result of the molecular attractive forces and the electrokinetic forces. In practice, molecular forces are only likely to be important when the particle has approached very close to the grain surface, probably within 0·1 µm. Thus the forces between the charged particles are mainly responsible for controlling the progress of the attachment process. Variation of the electrokinetic force depends upon the sign and magnitude of the surface charge and the thickness of the double layer. With most natural waters containing significant dissolved solids the charges on suspended particles and filter grains are probably fairly small so that only weak repulsive forces are produced. These can be modified by alterations in the ionic strength of the system resulting from the addition of coagulants or pH control. In addition the use of polyelectrolytes can introduce bridging mechanisms which can bind particles to the grain surfaces. The nature of the voids in a granular bed provides continual changes in direction and velocity of flow so that particles may grow in size owing to flocculation as they penetrate into the bed. Thus particles which are not successful in reaching a grain surface in the upper layers of a bed may so change in characteristics that they become trapped lower down in the bed.

Filtration

7

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

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

Developments in water treatment—2

8

FIG. 2. Head loss during filtration. Specific deposit is defined as the volume occupied by deposit in the bed per unit bed volume. Thus the total head loss, H, across a bed containing deposited particles is given by (3) where H0=Carman-Kozeny head loss, and K=a constant. If the bed retains the majority of the particles in the feed Ives and Gregory2 showed that an approximate solution of eqn (3) for a unisized medium is (4) where C0=feed concentration to bed, and t=time. For graded medium the expression is basically similar but replaces K with another factor to express the effect of size grading. 1.4.4. Solids Removal In many respects it is the performance of a filter in removing suspended matter which is of primary importance and this is an area which has received considerable attention from research workers. Most considerations are based on the proposition, originally formulated by Iwasaki3 for slow filters, that the removal of suspended particles with depth in a filter bed is proportional to the concentration of particles in suspension, i.e. (5) where c=concentration of particles at depth/in bed, and λ=a constant termed the filter coefficient.

Filtration

9

Ives4 has developed this concept to permit modelling the filtration process in three stages. An initial spherical grain format changes to a situation analogous to flow through capillary tubes as deposits accumulate in the voids. In the final stage of a run velocities in the voids increase to the point where previously deposited material is scoured. The mathematical expression of this concept can be developed to give the relationship between the filter coefficient at some stage of the run and the initial filter coefficient (λ0) at the start of the run (6) where σu=ultimate specific deposit, and β, x, y, z are empirical constants. Several other workers, notably Minz,5 Deb,6 Camp,7 Mohanka8 have produced variations on the theme of a physical model of the filtration process but in spite of intensive study the complex nature of filtration has meant that it is not yet possible to obtain a reliable predictive model for universal application. Nevertheless the models which are available can be useful in helping to produce more economical design and operating procedures. 1.4.5. Filter Cleaning In all filters the accumulation of deposits in the bed will eventually terminate the run either owing to a maximum allowable head loss being reached or owing to breakthrough of suspended matter with consequent deterioration in filtrate quality. With slow filters breakthrough is unlikely so that the run will be terminated on a head loss criterion. 11 is customary to remove about 10 mm of the sand which contains virtually all of the deposit and this material is washed and replaced on the bed when necessary. Backwashing of rapid filters is achieved by introducing filtered water into the bottom of the bed at relatively high rates of flow to give superficial velocities of 0·3–0·5 m/min. In hot climates higher upflow velocities are used because the reduced viscosity at elevated temperatures hinders expansion. The upward flow of water produces an expansion of the bed to give a fluidised state in which the voids are increased in size and a considerable scouring action occurs with resultant release of deposits from the bed grains. As the backwash starts the bed begins to expand and the head loss increases until the whole bed is just suspended in the upflow. At this point the upward force of water balances the gravitational force on the grains suspended in water. Further increase in backwash flow will increase the expansion of the bed but will not necessarily give better cleaning action since the scouring effect may be reduced at large expansions. At the point of fluidisation hegρw=le(ρs−ρw)g(1−fe) (7) where he=head loss in expanded depth le; ρw=mass density of water; ρs=mass density of bed grains; and fe=porosity of expanded bed. The bed grains are kept in suspension because of the drag force exerted on them by the upflow of water. This drag force is a function of the velocity onto the bed and the

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.

1.5. TRENDS AND DEVELOPMENTS IN FILTRATION The preceding sections have set out some of the basic factors governing the process of filtration as it is conventionally used for the production of potable water. It will be apparent that the process is one of considerable complexity and as such the performance of a filter is likely to be influenced by many parameters. Over the last few years there have been a considerable number of developments in configuration, design and operation of filters aimed at improving their efficiency and cost effectiveness whilst maintaining or even improving filtrate quality. Most of these developments have been related to rapid filters since the nature of the slow filter gives less scope for useful change. 1.5.1. Effective Use of Bed Capacity The inevitable consequence of backwashing a bed of graded medium is that the grains settle to give a stratified bed with the finest particles at the top and the coarsest at the bottom. This does not give efficient use of the voids in that a limiting terminal head loss, largely due to clogging of the relatively small voids in the upper layers of the bed, will be reached long before the majority of the voids have been utilised. A number of modifications to the conventional filter could provide a means of increasing the penetration of solids into the bed and hence providing fuller utilisation of the void capacity. The use of larger filter media allows increased penetration and the size of individual voids will be greater. The total volume of voids will however remain the same if the grain shape and packing are unaltered. The specific surface of the larger grains will be less so that particles of suspended matter are less likely to be trapped in any given void. The use of larger media in itself does not alter the basic characteristic of a downflow bed of graded material in that the bed becomes less efficient in terms of solids

Filtration

11

removal in the direction of flow. There are also clearly limits as to the amount by which grain size can be increased before breakthrough of solids becomes a problem. The essential need is a way of either increasing the specific surface in the direction of flow or of reducing the velocity in the direction of flow. These aims can be achieved in several ways. Increased specific surface in the direction of flow can be achieved by using dual or multi media downflow beds or by using a graded single medium in an upflow configuration, both of which techniques are now well established. Reduction of velocity in the direction of flow can be obtained in radial flow filtration which to date has had only limited application. Dual and Multi Media Beds By using more than one medium in a bed it is possible, by selection of appropriate materials and gradings, to improve utilisation of bed capacity. Table 2 gives examples of types of media which have been used in water filtration for this purpose. By far the most popular combination of media is anthracite and sand usually aranged as about 0·3 m of 1·25–2·50 mm anthracite above a similar depth of 0·50–1·00 mm sand. With such an arrangement good separation of the layers can be maintained even after vigorous backwashing. In operation the bulk of suspended particles should

TABLE 2 TYPES OF FILTER MEDIA Material Polystyrene beads Anthracite Sand Garnet Ilmenite

Specific gravity

Typical size range (mm)

1·04

1·5–3·5

1·4–1·7 2·5–2·6 3·8 4·8

1·0–2·5 0·5–1·5 0·3–0·7 0·3–0·7

be trapped in the anthracite layer with its relatively low head loss per unit depth, the sand layer in effect serving as a second stage filter. Figure 3 shows the advantage of a dual media filter over a single layer unit in terms of run length for a specified terminal head loss. Provided the characteristics of the suspended particles are suitable in respect of size, floc strength and physicochemical properties, the filtrate quality from a dual media bed is usually the

Developments in water treatment—2

12

FIG. 3. Head loss relationships for single and dual media beds of same overall depth. same as that produced by a single layer bed of the same overall depth using the same sand. The lower initial head loss and more efficient usage of the voids in a dual media bed permit longer filter runs than with a single medium bed at the same filtration rate, or alternatively permit operation of the dual media bed at higher filtration rate for the same length of run. This latter feature can be of particular value in uprating the capacity of an existing filter installation by removing half the original sand and replacing it with a layer of anthracite. In this way it may be possible to double the capacity of a filtration plant at a cost much less than the construction of duplicate single medium filters. The influence of filtration rate on removal of suspended matter does not appear to be great at least when relatively large floc particles are concerned, Tebbutt.10 However it should be appreciated that at high filtration rates more of the particles may penetrate into the sand layer causing a more rapid increase in head loss and/or early breakthrough. A comprehensive review of the use of anthracite in water filtration has been produced by the Water Research Association.11 A natural extension of the dual media principle is to add a further layer or layers to the bed in an effort to produce the desired decrease in void space with increasing depth into the bed. This can be accomplished by adding a coarse polystyrene bead layer to the top of an anthracite-sand bed or placing a layer of fine garnet or ilmenite at the bottom of an anthracite-sand bed. The latter approach has achieved some popularity in the USA in the form of a proprietary filter incorporating carefully sized layers of anthracite, sand and garnet in such a way that there is a degree of intermixing between layers and sharp

Filtration

13

interfaces are thus avoided. Conley and Hsuing12 have described the design of such beds which are stated to be more efficient than dual media installations. Multi media beds have received little interest in the UK partly owing to the increased cost and also to the law of diminishing returns which affects all treatment processes. Some pilot scale studies have recently been undertaken in the UK and have been reported by Lekkas et al.13 In this work the performance of anthracite-sand and polystyrene-anthracite-sand beds were compared when treating low turbidity (usually about 3 FTU) stored water. The three layer bed gave up to twice the length of run achieved by the dual layer bed for the same filtrate quality (<0.5 FTU) owing largely to prevention of surface clogging which took place in the dual layer unit. The studies were carried out over a range of filtration rates, 180–360 m3/m2d, and it was noted that differences in performance between the two beds were not apparent at the higher rates because then more suspended solids were carried through to the sand layer in both units. In practice it would appear that dual media beds are the most cost effective way of improving the utilisation of a downflow filter but it should be realised that such a bed may be more sensitive to changes in loading and nature of suspended solids than a conventional single-medium bed which has, in effect, a fairly high inbuilt factor of safety. Upflow Filters To overcome the undesirable effect of stratification produced by backwashing of a graded bed an obvious solution is to alter the direction of filtration from downflow to upflow so that the small voids on the top of the bed will only receive particles not trapped in the lower regions of the bed. The principle of upflow filtration has been known for a considerable time and indeed some units were in operation in the eighteenth century. It was however not until the advent of the Immedium unit in the 1950s that upflow filtration became more than a curiosity. Since then a considerable number of Immedium filters have been installed for both potable water treatment and tertiary treatment of effluents. The basic problem in upflow filtration is that of preventing the upward flow of water from expanding the bed as occurs during backwashing. It is thus necessary to limit the filtration rate to avoid this effect or to provide some means of holding the bed in place to prevent expansion during filtration. The Immedium filter achieves this end by the insertion of a coarse grid in the top of the bed, Fig. 4(a), against which the bed grains arch under filtration. At the higher backwash rates the arched grains are no longer able to resist the upward force so that expansion can proceed normally.

Developments in water treatment—2

14

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

Filtration

15

encourages greater penetration of suspended solids into the bed and thus should improve the utilisation of bed capacity. Radial flow filters in the form of fixed beds have received some experimental attention on a laboratory basis. Studies by Ives4 suggest that the performance of a radial flow bed would compare favourably with a

FIG. 5. Typical radial flow filter with fluidised bed and continuous desludging by air lift. conventional vertical flow bed. There are however operational problems with a fixed-bed radial flow system in that conventional backwashing cannot be carried out and thus some other form of cleaning would have to be devised. Because of the cleaning problems associated with fixed beds, commercial development of the radial flow filter has been directed towards the use of fluidised bed units with continuous washing and recycling facilities. The Tenten unit is an example of this concept which is shown diagrammatically in Fig. 5. Ray16 quotes the use of fluidised-bed radial flow units for removing inorganic turbidity from a groundwater to give a finished water of about 50% lower turbidity than the feed which averaged 1·2 units. The commercial units are factory built and occupy relatively small floor area having a typical surface loading of around 300 m3/m2 d depending upon type and concentration of suspended solids. 1.5.2. Process Modifications In efforts to produce less costly forms of water treatment some interest has developed in the use of high rates of filtration. A study in the USA by King et al.17 reported that many

Developments in water treatment—2

16

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

Filtration

17

be best satisfied by the adoption of such forms of appropriate technology using local materials and skills wherever possible. 1.5.3. Filter Control Because of the long runs achieved by slow filters the hydraulic characteristics do not change significantly from day to day and thus manual adjustment of the throughput is satisfactory. In the case of rapid filters virtually continuous adjustment of control valves is necessary to maintain constant filtration rate. This feature of rapid filtration has normally been catered for by provision of some form of automatic flow controller to ensure constant output. Various proprietary devices are available which provide a high head loss at the start of a run when the filter head loss is low. As the run proceeds and filter head loss increases the hydraulic control is gradually relaxed so that the overall head loss through filter bed and controller remains constant. These flow control units are relatively expensive and are not always trouble free so that simpler forms of filter control are becoming more popular. Constant rate filtration can be achieved by using a deeper filter shell and allowing the top water level to increase as the head loss in the bed builds up. Flow to the filter is set at a constant rate with free discharge over a weir into the shell. The additional capital cost of the extended shell depth is a disadvantage of this type of operation. In declining rate filtration using a battery of filters a fixed top water level is used so that as the bed clogs the output declines. Provided the backwashing of beds is staggered, a reasonably constant output will be maintained from a battery of several filters and the process is claimed to offer a number of advantages over constant rate operation. In particular a decline in filtrate quality towards the end of a run does not occur, filter runs are longer and capital and operating costs are lower. In spite of these claims relatively few declining rate installations have been built in the UK although they are more popular in the USA and Japan. Arboleda27 has discussed the various forms of hydraulic control techniques used in filtration and concluded that declining rate filtration did offer advantages over constant rate operation. It seems likely that this type of control would certainly be more appropriate than the conventional constant-rate method for developing country situations. At the other end of the scale the automation of filtration plants is becoming increasingly common with large installations being equipped with process control and monitoring facilities to handle both filtration and backwashing stages of operation. 1.5.4. Backwashing The development of more effective filtration techniques with greater penetration of suspended solids into the bed leads inevitably to the need to ensure that the bed is efficiently cleaned. The growing use of granular beds for the filtration of wastewater effluents has also placed increased demands on cleaning arrangements since the organic nature of the deposits can result in difficulties in their removal from filter media. In a basic study by Amirtharajah28 it has been stated that the main cleaning mechanism in a backwashing situation is hydrodynamic shear and that abrasion between grains is negligible. The optimum porosity for cleaning was found to be about 0·70 which for a graded sand would be likely to occur at 40–50% expansion of the top layers.

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

Filtration

19

FIG. 6. Determination of optimum bed depth and filter run based on head loss and filtrate quality criteria. (Fixed filtration rate and single-size medium.) provides the optimum solution for a given filtration rate and medium. Additional points are obtained for different filtration rates and different media with the end-product being capable of expression in the form of a response surface which shows the operational optimum conditions (Fig. 7). Ives32 has recently given a full description of this procedure and has considered in some detail the economic optimum design of filters. Adin33 has described a technique for predicting filter performance in an optimum design using the similarity which exists between turbidity breakthrough in a filter and the breakthrough pattern in an adsorption bed. Since the various mathematical models of filtration which have been produced tend to be restricted in application to special circumstances it is generally agreed that information for use in optimising models is best obtained from pilot studies using the actual media and water under consideration. The design of such studies has been detailed by Ives.34 A problem which arises in filtration work is that of characterising the

Developments in water treatment—2

20

FIG. 7. Response surface for optimum filter design. nature of the suspension to be treated. This factor was appreciated during the development of microstrainers with the introduction of the filterability index. This concept can also be used to relate the filterability of a suspension to filter media taking into account clarification clogging and flow rate as described by Ives.35

REFERENCES 1. CARMAN, P.C., Trans.Instn Chem. Engrs, 1937, 15, 150. 2. IVES, K.J. and GREGORY, J., Proc. Soc. Water Treat. Exam., 1967, 16, 147. 3. IWASAKI, T., J.Amer. Water Works Assoc.. 1937, 29, 1591. 4. IVES, K.J., Proc. 8th Congr. Int. Water Supply Assoc., Vienna, 1969, 1, K1. 5. MINZ, D.M., Proc. 6th Congr. Int. Water Supply Assoc., Stockholm, 1964, 1, E1. 6. DEB, A.K., J. Sanit. Eng. Div., Am. Soc. Civil Engrs, 1969, 95, 399. 7. CAMP, T.R., J. Sanit. Eng. Div., Am. Soc. Civil Engrs, 1964, 90, 1. 8. MOHANKA, S., J. Sanit. Eng. Div., Am. Soc. Civil Engrs, 1969, 95, 1079. 9. FAIR, G.M., GEYER, J.C. and OKUN, D.A., Waste Water Engineering, 2. Water Purification and Waste-Water Treatment and Disposal, 1967, John Wiley, New York. 10. TEBBUTT, T.H.Y., Water Res., 1971, 5, 81. 11. Water Research Association, Anthracite Sand Filtration, Conference Papers, 1972, WRA, Medmenham, UK. 12. CONLEY, W.R. and HSUING, K., J. Amer. Water Works Assoc., 1969, 61, 97.

Filtration

21

13. LEKKAS, T.D., Fox, G.T.J. and MCNAUGHTON, J.G., J. Instn Water Engrs & Scientists, 1978, 32, 239. 14. GREGORY, R., MCNAUGHTON, J.G. and MILLER, D.G., Experiences with Upflow Filtration, TP 102, 1973, WRA, Medmenham, UK. 15. HANEY, B.J. and STEIMLE, S.S., J. Environ. Eng. Div., Am. Soc. Civil Engrs, 1975, 101, 489. 16. RAY, W.J.F., Proc. 10th Congr. Int. Water Supply Assoc., Brighton, 1974, 1, B1. 17. KING, P.H., Johnson, R.L., RANDALL, C.W. and REHBERGER, G.W., J. Environ. Eng. Div., Am. Soc. Civil Engrs, 1975, 101, 479. 18. WEST, J., RACHWAL, A.J. and Cox, G.C., J. Instn Water Engrs & Scientists, 1979, 33, 45. 19. LOGSDON, G.S., Civil Eng, 1978, 48 (July), 68. 20. CULP, R.L., J. Amer. Water Works Assoc., 1977, 69, 375. 21. STUMP, V.L. and NOVAK, J.T., J. Amer. Water Works Assoc., 1979, 71, 339. 22. TREWEEK, G.P., J. Amer. Water Works Assoc., 1979, 71, 96. 23. VIANNA, M.D.B., Performance Relationships in Water Treatment, M.Sc. Thesis, 1979, University of Birmingham. 24. PARAMASIVAM, R. and SUNDARESAN, B.B., Aqua, 1979, (3), 13. 25. KOMOLRIT, K., CHAINARONG, L. and BUASEEMUONG, S., Aqua, 1979, (4), 12. 26. FRANKEL, R.J., J. Amer. Water Works Assoc., 1974, 66, 124. 27. ARBOLEDA, J., J. Amer. Water Works Assoc., 1974, 66, 87. 28. AMIRTHARAJAH, A., J. Environ. Eng. Div., Am. Soc. Civil Engrs, 1978, 104, 917. 29. CLEASBY, J.L. and LORENCE, J.C., J. Environ. Eng. Div., Am. Soc. Civil Engrs, 1978, 104, 749. 30. GOULD, M.H.and PATTERSON, P., Water Services, 1979, 83, 573. 31. AMIRTHARAJAH, A. and CLEASBY, J.L., J. Amer. Water Works Assoc., 1972, 64, 52. 32. IVES, K.J., in Mathematical Models in Water Pollution Control, A.J.James (Ed.), 1978, John Wiley and Sons, Chichester, 339. 33. ADIN, A., Filtrn. Separn., 1978, 15, 55. 34. IVES, K.J., Effl. Water Treat. J., 1966, 6, 552, 591. 35. IVES, K.J., Prog. Water Tech., 1978, 10, 123. A comprehensive review of filtration is presented in The Scientific Basis of Filtration, K.J.Ives (Ed.), 1975, Noordhoff, Leyden.

Chapter 2 REMOVAL OF ORGANIC COMPOUNDS C.S.SHORT B.Sc., C.Eng., M.I.Chem.E., M.I.W.E.S. Yorkshire Water Authority, Leeds, UK SUMMARY The use of activated carbon for municipal water treatment continues to grow as water quality standards are raised and polluted sources are increasingly drawn on. Minor developments in technique have been introduced in the 1970s, principally in regeneration furnace design. New resinous adsorbents have been developed for removing a wide range of organics, but these are not at present competitive with carbon for potable water treatment,particularly because of the problems associated with liquid regenerants. The high organics removal performance of reverse osmosis has been demonstrated but it remains prohibitively expensive, while the related process of ultrafiltration is cheaper but effective only for removing the larger organic species. It is concluded that activated carbon adsorption is the process of choice for many duties from taste and odour removal to the general reduction of synthetic organics. Ion exchange resins will continue to be used for the removal of naturally occurring organic acids in the production of deionised water.

2.1. ORGANIC COMPOUNDS IN WATER Organic compounds are present in most surface waters, but are rarely found in significant quantity in groundwaters. The most common naturally occurring organic compounds are humic and fulvic acids derived from vegetation. In addition, rivers receiving sewage and industrial effluents and run-off from agricultural and urban catchments may contain a very wide range of synthetic organic compounds. Total levels of organic matter can be determined in terms of organic carbon, or indirectly in terms of u.v. absorption. Many individual substances have been detected, generally at very low concentrations. However, the most advanced current analytical techniques cannot identify more than 10 to 20% of the individual organic compounds present in polluted surface waters. Development of high performance liquid chromatography in conjunction with mass spectrometry will allow the detailed investigation of the remaining fraction.1

Removal of organic compounds

23

2.2. OBJECTIVES The objectives of organics removal depend on the application. Table 1 indicates the organic compounds of concern in a number of applications. For some uses organics in general are undesirable, whereas for others specific compounds or groups of compounds must be removed. However, currently available treatment methods can be made specific for given types of compound only to a limited extent, as will be discussed in later sections. Knowledge of the health effects of organic compounds at the low levels found in water supplies is very limited. However, where a water source is known to contain significant amounts of industrial effluent, this has given rise to concern. Furthermore, naturally occurring organic compounds have been found to react with chlorine, during the disinfection process, to form a variety of chlorinated organics some of which are thought to be potentially hazardous. As organics removal can be expected to add considerably to treatment costs, there is an urgent need to improve our knowledge of the relationships between organic compounds and health. In the meantime, if the current (1979) requirements of the United States Environmental Protection Agency (EPA) and the EEC are to be met, a considerable expenditure on organics removal for health reasons may be anticipated in the next decade.

2.3. THE PERFORMANCE OF CONVENTIONAL AND OXIDATIVE PROCESSES Both colloidal and dissolved organic matter are removed to some extent by conventional processes, but a limited amount of information is available on their performance except in respect of colour removal. Slow sand filtration may be expected to remove colloidal material very

TABLE 1 OBJECTIVES OF ORGANICS REMOVAL Application Potable water supply

Organic compounds of concern

Reason for concern

Maximum level recommended

Organics arising Possible health Reduce to 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

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

Removal of organic compounds

Preparation of Pyrogens injections and pharmaceuticals

25

Cause pyrogenic reaction in patients

–

effectively, while about 30% reduction of dissolved humic matter is achieved probably by biological oxidation. Reductions of Permanganate Value (PV) reported by Houghton2 apparently show the adverse effect of low temperature on a biological process (Table 2). Slow sand filters are effective in reducing some tastes and odours, but are also known to cause them under certain circumstances.3

TABLE 2 PV REMOVAL, CHIGWELL Row SLOW SAND FILTERS, ESSEX WATER Co. (From Houghton, Reference 2) Nov. 1965– April-Oct. March 1966 1966 Raw water PV mg O2/litre Filtered water PV mg O2/litre Removal (%)

2·0

2·2

1·5

1·3

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 biodegraded in subsequent activated carbon beds; this is discussed in a later section. Recent studies11,12 indicate that the action of ozone and u.v. radiation together readily oxidises some organics to carbon dioxide and water. It is unlikely, however, that this would be economically acceptable on a practical scale for municipal water treatment. As with ozone, other oxidising agents such as chlorine, chlorine dioxide and potassium permanganate break down organic molecules incompletely and therefore do not necessarily reduce any health risk. Indeed, chlorine at normal disinfection doses forms possibly undesirable substitution products with organic substances, as discussed above. Chlorine dioxide and potassium permanganate are reported to reduce various natural organic odours but at rather high doses.10 Chlorine dioxide is used for disinfection as an alternative to chlorine when phenols are present in the raw water, as it does not form chlorophenol tastes but partially oxidises the phenols.13 To summarise, the reduction in the level of organic compounds achieved by conventional processes is generally insufficient to meet the latest requirements for industrial water treatment, as indicated in Table 1, or for potable water when polluted sources are used. Nevertheless, such reduction as is achieved incidentally to the normal clarification and colour removal role of these processes may significantly reduce the cost of subsequent organics removal processes. Oxidants modify rather than remove organic matter and their application is limited to the removal or avoidance of tastes and odours except for their possible use to enhance the performance of activated carbon beds.

2.4. ACTIVATED CARBON 2.4.1. Introduction The use of activated carbon for municipal and industrial water treatment is well established. In the municipal field, carbon has been applied almost exclusively to the removal of odorous substances, for which purpose it is often the most, or only, effective treatment. There are therefore many examples of its use throughout the world on reservoir and river abstractions where odour problems are common. Because odours often occur intermittently and are associated with very low concentrations of the odorous substance, it has been economical to use powdered carbon in the majority of these plants. There are a growing number of granular carbon installations, notably on river abstractions suffering continuously from odours originating in industrial and municipal effluents. Very few carbon plants have been installed specifically for the removal of organics to reduce the possible health hazard. However, in 1978 the US Environmental Protection Agency (EPA) proposed that all municipal water treatment systems in the

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

where Cc is the concentration of solute in the carbon at equilibrium, Cw is the concentration of solute in the water at equilibrium, and K and n are constants for each solute/carbon system. n is less than one for most organic solutes, implying that a greater proportion of the solute will be adsorbed from a dilute solution than from a concentrated one. On the other hand, the uptake of solute per unit weight of carbon increases with increasing concentration of the solute in the solution. It follows that activated carbon is well suited to the treatment of dilute solutions, but for maximum use of carbon it should achieve equilibrium with the solute at its most concentrated, i.e. at the feed concentration. Rate of adsorption depends on diffusion both to and within a carbon particle. The solute molecular weight, carbon particle diameter and pore size distribution all affect the rate of intraparticle diffusion, which is generally the rate-cootrolling step. However, transport of solute to the carbon particles can be appreciably slowed by the occluding effect of coagulant flocs as in the case where powdered carbon is dosed upstream of floc blanket tanks or where granular carbon is used in the dual role of adsorber and filter. Adsorption rates may be determined by stirred batch tests or by granular carbon column tests. For the case where intraparticle diffusion is rate-controlling, an empirical equation of the form dy/dt=−KC ln y

Developments in water treatment—2

28

may be used to fit the data, where C is the concentration of solute in the solution at any instant; y is the concentration of solute in the carbon at the same instant, expressed as a fraction of the equilibrium concentration in contact with a solution containing the solute at concentration C (i.e. y can be regarded as the fractional saturation of the carbon); and K is the rate constant. These concepts of adsorption rate and capacity are valuable for the comparison of different carbons and also suggest ways in which the efficiency of use of activated carbon could be maximised. For instance, it can be seen that granular carbon beds must be used to achieve maximum carbon use. The upstream layers of the bed will eventually become saturated with solute at the feed concentration; whereas powdered carbon added before clarification is well mixed with the water and becomes saturated with solute at the treated water concentration. However it is not always cheaper to install granular rather than powdered carbon as capital costs of granular systems are higher. Selection and optimisation of carbon systems are discussed later in this section. The relationships in the above equations are empirical, and batch and/or column tests are invariably required to determine the equilibrium and rate constants for a particular carbon/water/solute system. Granular carbon plants should not be designed from the results of batch tests alone as these do not take into account the effects of microbiological activity, occlusion by suspended solids, and consequent backwashing all of which occur in granular carbon beds. 2.4.3. Powdered Activated Carbon The Material Powdered activated carbon has been widely used for many years and is available in a range of qualities. It is typically a powder of bulk density around 700 kg/m3 and particle size less than 75 µm The performance depends on the raw material and the method of activation. Recent developments include a water treatment carbon comprising a mixture of coal-based and peat-based material to obtain a wide spread of pore sizes; and a carbon based on anthracite, activated by steam at 950°C in a fluidised bed furnace. Point of Addition Powdered carbon may be added before sedimentation or immediately before rapid filtration. In either case the carbon must have a sufficiently long contact time to allow equilibrium to be achieved with the solutes of concern. Removal by flotation would be rapid and the contact time practically limited to that provided in the flocculation stage. Performance is adversely affected by the concurrent addition of softening chemicals, coagulants and oxidants including chlorine.16a,17 In some Japanese plants contact time before flocculation is provided.18 Gomella19 has shown the economies that may be expected if powdered carbon is added at two or more positions, provided that equilibrium is approached at each stage. Lettinga et al.20 proposed a fluidised bed system which would permit greatly improved powdered carbon utilisation if the fluid bed could be divided into several separate stages. However, unlike the usual methods of applying powdered carbon, this would require an expensive special vessel.

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

FIG. 1. Costs of powdered carbon dosing systems (from Burley and Short, Reference 15c).

Developments in water treatment—2

30

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

FIG. 2. Schematic of powdered carbon dosing system, Elvington water treatment plant (by courtesy of Yorkshire Water Authority and Portasilo Ltd). Key to details: 1. Pneumatic delivery from bulk tanker. 2. Air vent incorporating automatically cleaned bag filter. 3. Load cells. 4. Aeration pads supplied with air at 35 kN/m2. 5. Slide plate isolating valve. 6. Two-speed screw conveyor. 7. Pneumatically operated butterfly valve. 8. Pneumatically operated swing valve. 9. Screw conveyor. 10. Vortex mixing cone. 11. High speed mixer. 12. Low speed agitator. 13. Duplex slurry metering pump. 14. Transport water pumps. 15. Duplicate 50 mm dia. PVC pipes.

Developments in water treatment—2

32

2.4.4. Granular Activated Carbon The Material Many grades are available to cover a wide variety of industrial uses. Those grades most suited to water treatment are generally produced by high temperature steam activation. The surface area of typical water treatment carbons is in the range 500–1500 m2/g as measured by the standard BET method (based on the adsorption of nitrogen). Particle sizes giving the best combination of adsorption kinetics and head loss characteristics are in the range 0·5–2 mm. The shape is normally irregular, but one carbon is available as uniformly sized cylindrical pellets 0·8 mm diameter, formed by extrusion. Particle strength should be sufficient to allow regeneration of exhausted carbon with attrition losses of not greater than 5%. However, it may sometimes be economical to use a cheaper, softer grade on a throw-away basis. This is referred to as non-regenerable carbon. It is possible that the exhausted material could be reprocessed into powdered activated carbon. Points of Application Granular carbon adsorbers are preferably installed following clarification processes. This arrangement has several advantages; 1. Some organic matter is removed by coagulation and clarification, reducing the load on the carbon and extending its operating ‘life’ between regenerations. 2. The risk of occlusion of the carbon by particulate matter and coagulant floc is minimised. 3. Backwashing frequency of the carbon units is minimised. It is inevitable that during the upflow washing the arrangement of carbon particles in the bed is somewhat altered. This tends to disrupt the desirable adsorption profile in which the relative exhaustion of the carbon varies inversely with distance from the bed inlet. Adsorption efficiency and the operating life of a bed could therefore be significantly reduced. However, there are a number of plants where, for speed and cheapness of installation, carbon has been substituted for sand in existing rapid gravity filters; some reported results are reviewed below. Irregularly shaped granular carbon is an effective filtration medium and its low density compared with sand makes it economical to backwash. Where pre-chlorination is used, chlorine levels entering the carbon beds are normally kept to a minimum. Chlorine coming into contact with the carbon will be reduced to chloride. The corresponding oxidation of the carbon surface may have an adverse effect on the adsorption of many organics of concern.16a Also, the cost of restoring the lost chlorine after the carbon beds must be considered. Adsorber System Design A number of predictive mathematical models of granular carbon adsorbers have been developed, and have shown reasonable agreement with the observed adsorption of single and binary solute systems.23 However, it is likely that some important factors in carbon bed performance will remain unquantifiable. Adsorber designs are therefore generally based on the results of pilot column tests in which samples of water are taken from a

Removal of organic compounds

33

number of depths within the bed at regular intervals of time. The concentrations of the solutes of interest are measured in each sample and the data can be presented as a plot of concentration against either time at each depth (‘breakthrough curve’) or depth at each interval of time (‘adsorption front’). Using the latter (Fig. 3), the minimum depth of bed to give the required treated water concentration is readily determined, and the progressive saturation of the carbon is illustrated.

FIG. 3. Idealised adsorption front curves. These data are used to evaluate the effectiveness of alternative adsorber designs in respect of carbon utilisation. Carbon removed for regeneration or disposal should as far as possible be saturated with respect to the organic solutes of concern. Three systems which ensure a high efficiency of carbon utilisation are illustrated in Fig. 4. The carbon system must be designed to accommodate the worst anticipated organic loads. Gomella19 suggests using carbon beds for base-load operation with powdered carbon dosing to reduce peak concentrations. The effect of water velocity (surface loading) on minimum contact time and carbon utilisation appears to vary according to the particular solutes/sorbents in reported studies.16c,23,24 Pilot column tests should therefore include parallel trials at different velocities in the suggested range 2 to 25 m/h. Head loss and backwash data are provided by suppliers for each grade of carbon. However, measurements should be made during column tests to

Developments in water treatment—2

34

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

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).

FIG. 5. Granular activated carbon adsorption/regeneration system, Church Wilne water treatment plant (by courtesy of Severn–Trent Water Authority and Humphreys and Glasgow Ltd). One carbon supplier offers prefabricated adsorption systems for hire, including periodic replacement of the carbon.30 This is intended for industrial wastewater treatment but could also find an application in the provision of temporary water supplies from low quality sources. Performance of Granular Activated Carbon Adsorbers It is difficult to compare and summarise the reported results of batch, column and fullscale operation, as carbon utilisation rates are strongly influenced by feedwater quality and the desired treated water quality. Utilisation rates between 2 and 200 g carbon per m3 water treated have been reported for applications varying from taste and odour removal to

Developments in water treatment—2

36

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 prechlorination was applied to River Trent water, the low

TABLE 3 KEY TO REPORTED DATA ON ORGANICS REMOVAL BY GRANULAR CARBON Reference 15

28

16d

System Foxcote waterworks. Carbon beds following coagulation, sedimentation and sand filtration Church Wilne waterworks. Carbon beds after coagulation, chlorination, sedimentation and sand filtration Lengg waterworks,

Organics investigated TOC, PV, Chemical Oxygen Demand (COD), Taste

TOC, threshold odour number, detergents, haloforms

COD u.v. absorption

Removal of organic compounds

16d

15

5

31

24

16d 7

8 23

32

33

37

Zürich. Carbon beds after ozonation Dohne COD u.v. absorption waterworks, Mülheim. Carbon beds after coagulation, ozonation, sedimentation, sand filtration Vigneux-sur- Threshold taste value, Seine fullPV, phenols, anionic scale detergents, pesticides experiment with and without sand filtration Pilot columns, TOC, u.v. absorption, R.Trent colour, phenols, anionic and non-ionic detergents, pesticides Pilot columns, TOC, u.v. absorption, R.Trent colour, threshold odour number, monohydric phenols, PAH, pesticides Pilot columns, u.v. absorption, 1, 2R.Thames dichlorobenzene, chloroform, bromodichloromethane, α-BHC, γ-BHC, Dieldrin Pilot columns, PV, u.v. absorption, Amsterdam colour Pilot columns Endrin, Lindane, Toxaphene, Silvex, 2, 4, 5-T ester Pilot columns TOC, COD, haloforms Pilot columns Haloforms, carbon tetrachloride, nitromethane, methyl ethyl ketone, n-butanol, 1, 4-dioxane Small Geosmin, 2-methyl columns isoborneol in presence of humic acids 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 chlorineresistant bacteria which may lead to odour problems in distribution, but also of algae and invertebrates,15f,28 In some installations therefore—e.g. Nottingham and Göteburg—high levels of chlorine are continuously applied so as to penetrate the carbon beds and prevent microbiological activity. The chlorine demand of the beds at Nottingham is about 0·7 mg/litre, or 0·3 g per m3 carbon per hour. However, in a number of European water treatment plants (e.g. Mülheim, Zürich, Düsseldorf, Rouen) microbiological activity within carbon beds is deliberately promoted by ozonating the water upstream of the adsorbers at doses of 1 to 5 mg/litre. This appears to enhance organics removal by a combination of the following factors: (i) Non-biodegradable substances may be broken down into biodegradable components by ozone. (ii) Residual ozone may provide additional oxygen for bacterial respiration and biological oxidation of organics. (iii) The biological activity thus encouraged oxidises biodegradable organic substances and effects a continuous partial regeneration of the adsorptive potential of the carbon, thus extending the operating life of a bed. (iv) Ozone does not form trihalomethanes as does chlorine, but may indeed reduce the concentration of trihalomethanes if these are present in the raw water. Therefore if trihalomethanes are an important criterion of treated water quality, and it is necessary to disinfect the water at an early stage of treatment, ozone must be preferred to chlorine for disinfection. Thus, changing from pre-chlorination to pre-ozonation extended the useful life of activated carbon beds at Mülheim from 1 to at least 20 months.16d Regeneration System Design and Operation Spent granular carbon can be returned to the manufacturer for regeneration but in some circumstances—particularly where large quantities of carbon are involved—it may be economical to provide on-site regeneration facilities. A comprehensive review of regeneration methods and systems34 suggests that only high temperature thermal reactivation is practicable for water treatment applications. Several stages have been identified in this process: 1. Drying—up to 100°C. 2. Loss of volatile organics—100 to 300°C. 3. Pyrolysis of non-volatile organics—200 to 650°C. 4. Gasification of pyrolysis products—800 to 1000°C.

Removal of organic compounds

39

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

Church Wilne waterworks, Nottingham, UK

Type

Re% losses Approx. Reference activation per re- heat input, temperature generation 106 kJ/t carbon regenerated

Multiple 950–975 hearth with steam injection and afterburner Alelyckan Multiple 900–1000 waterworks, hearth Göteburg, with steam Sweden injection and

4·5–5

20 (10·5 at 28 Private 100% load communication and with air cooling)

6

8·5 Private (excluding communication steam)

Developments in water treatment—2

afterburner South Tahoe Multiple advanced hearth wastewater with steam treatment injection plant, USA and occasional afterburner Zürich Fluidised waterworks, bed with Switzerland steam injection, afterburner and vacuum drier Industrial Two-stage wastewater fluidised treatment bed with plant, steam Mississippi, injection USA and waste gas recycling Hypothetical Any type (EPA cost with estimates) afterburner and scrubber

40

900–925

Av. 5%

6–12

26

900

3·5–5

–

35

815–980

6–8

5

36

–

7

8·6

14

installed at Zürich waterworks35 includes a vacuum drier prior to the reactivation furnace; heat extracted from the waste gases is used to raise steam for injection and to assist drying. Another system using two fluidised beds in a vertical series, plus partial recirculation of waste gases to the gasification stage, is reported to have comparatively low fuel requirements.36 Granular Activated Carbon Applications Activated carbon is an effective flexible and economical method for the removal of a wide range of dissolved organic substances, from naturally occurring tastes and odours to industrial pollutants. Its flexibility is illustrated by the many ways in which it can be integrated into a sequence of water treatment processes, even if required fulfilling the dual role of adsorbent and filter medium. It is likely that granular carbon beds will be applied to heavier and more stringent organics removal duties in the production of potable water, and that series beds or pulsed beds will be used for greater economy of carbon use. On-site carbon regeneration will also be attractive for such duties. Thermal regeneration has the important advantage of completely destroying the organics removed from the water. Developments in furnace design may considerably reduce operating costs.

Removal of organic compounds

41

2.5. ION EXCHANGE AND POLYMERIC ADSORBENTS 2.5.1. Introduction The use of ion exchange resins for organics removal derives from observations of the fouling of synthetic anion exchange resins in boiler feedwater treatment systems. It was found that organic substances—in particular, humic acids—were strongly adsorbed, and insufficiently desorbed during regeneration. With the development of high pressure boilers the problems caused by fouling became more acute. At the same time, practieally complete removal of the organic acids was necessary as a significant residual concentration would raise the conductivity of the treated water to an unacceptable level. Two new types of anion resins were developed to overcome the fouling problem: macroporous, which have a large, open pore structure, and isoporous, which have a more regularly cross-linked structure than conventional gel type resins. Both macroporous and isoporous resins allow organic solutes to be adsorbed, and subsequently removed during regeneration. Experience has shown that for demineralisation of boiler feedwater, the isoporous strong base anion exchange resins are more economical than macroporous in terms of both capital and operating costs. However, in some circumstances macroporous resins may be used to reduce the organic load on following gel type strong base anion exchange resins. Regenerated by caustic brine, strong base macroporous resins are effective in removing the organic acids responsible for fouling, but make no contribution to the deionisation as anions in the feedwater are exchanged for chloride. Macroporous resins used in this way are commonly known as organic scavengers. Weakly basic macroporous resins have been developed for their ability to regenerate effectively with sodium hydroxide; they can economically combine the removal of organic acids and of anions of strong acids, and may be used to precede strong base resins in either separate or stratified beds. A detailed account of the causes, effects and control of organic fouling in demineralisation systems is provided by Tilsley;37 reference may also be made to McWilliam.38 A pilot-scale evaluation of isoporous and macroporous resins for the deionisation of R.Trent water is described by Brown and Ray.39 2.5.2. Resins for Organics Removal Synthetic polymeric resins of all the above-mentioned types have been evaluated in recent years specifically as a means of removing organic compounds from water. To some extent their properties—in particular, polarity and pore size—can be varied to improve performance with respect to specific groups of compounds. For instance, hydrophobic (non-polar) resins are available which are designed to adsorb non-polar solutes. Two additional types of resin have been developed specifically for the removal of organic substances from liquid and gas streams; carbonaceous and cellulosic. Carbonaceous adsorbents are said to have a structure between those of polymeric resins and activated carbons. One such adsorbent is described as being produced by partial pyrolysis of beads of a macroporous synthetic polymer with specially designed surface

Developments in water treatment—2

42

functional groups.23 Carbonaceous adsorbents are available in a variety of polarities and pore size distributions. A non-polar variety is suggested for the removal of halogenated organic compounds, e.g. haloforms and pesticides. More polar molecules such as phenols require a hydrophilic (polar) variety. Cellulosic ion exchange resins are based on a cellulose matrix rather than the more expensive polystyrene or acrylic ester on which conventional ion exchange resins are based. They are reported to have a high capacity for specific organic and inorganic components. They have extensive macropores and are particularly effective for the removal of very large molecules, for instance proteins and enzymes, from aqueous solutions. The development and preparation of cellulose based resins are described by Grant.40 Various ionic forms corresponding to the main types of conventional resins are available. For the removal of humic acids from water, strong or weak base varieties are indicated. 2.5.3. Theory It will be noted from the foregoing that removal of organic solutes takes place by a combination of ion exchange and adsorption, the mechanism depending on the polarity of the solute and of the functional groups in the resin. When the removal of a solute is by ion exchange predominantly, resin performance will be sensitive to the pH of the solution as this affects the degree of ionisation of the solute. The capacity of a resin for a given solute depends on the surface area presented to the solute, which in turn depends on the size of the solute molecule, the porosity of the resin and the distribution of pore sizes. These concepts are sufficient to indicate the resins most likely to provide effective removal of given organic substances. More detailed predictive models and associated data are not available for the design of systems. Resin selection and plant design therefore require experimental data from laboratory evaluations. 2.5.4. Methods of Evaluation Rapid screening of a range of resins may be carried out by contacting known quantities of resin and solution in batch tests and measuring the equilibrium concentration of the solutes of interest in the solution. Gauntlett41 noted that fresh resins imparted organic matter to the test solution and described a procedure for extracting this prior to equilibrium tests. Alternative regenerants could be similarly screened by contacting with exhausted resins. It is essential to evaluate promising resins by column tests over several exhaustion/regeneration cycles to obtain information on the capacity of fresh and regenerated resins for the solutes of interest. A number of resins may be tested in parallel. The regenerants of choice for strong and weak base anion exchange resins will normally be caustic brine or sodium hydroxide solution. The regeneration of non-polar resins may require the use of organic solvents, e.g. acetone or methanol. The importance of thorough rinsing in potable water applications can readily be appreciated, and rinse volumes should be determined in the column tests.

Removal of organic compounds

43

25 mm diameter glass or perspex columns are suitable for laboratory tests. Superficial velocities of 5 to 25 m/h may be used, giving superficial residence times of 5 to 1 min in beds 400 mm deep. Further details of laboratory test procedures are given by Brown and Ray39 and Chriswell et al.42 As with granular activated carbon studies as described in the preceding section, it is important when evaluating resin adsorbents to obtain breakthrough data for design purposes. Large column tests may be required when investigating the removal of certain compounds for which potable water standards have been set at the µg/litre level (see Table 1), in order to obtain sufficient treated water for analytical purposes. 2.5.5. Test Results Table 5 is a key to recently published test results for a variety of resins and solutes. The salient conclusions are as follows: All types of macroporous and isoporous weak and strong base resins are effective for the removal of humic acids, phenols and anionic detergents. However, anionic detergents are so strongly adsorbed by basic resins that they are hardly desorbed at all from strong base resins by conventional regeneration. The desorption from weak base resins is very low but according to Oehme and Martinola47 levels of up to 0·3 mg anionic detergents per litre can be tolerated if sodium hydroxide regeneration is applied. The uptake of phenol by a weak base phenol-formaldehyde resin was found to be highly pH dependent, with the best performance at pH 5. Ammonia was found to be as effective as sodium hydroxide for regenerating one weak base resin, but somewhat inferior for another. Non-polar polyacrylic and polystyrene adsorbents (‘white carbons’) were found to be inferior to activated carbon in respect of a variety of solutes. The performance of a non-polar carbonaceous resin was comparable to that of activated carbon. 2.5.6. Applications Macroporous and isoporous weak base and strong base resins are well established for the removal of weak organic acids, in combination with, or preceding, demineralisation of boiler feedwater. When treating river water or highly coloured moorland water, the resins would normally be preceded by coagulation and clarification. However, Grant has suggested40 that the low costs and high organics capacity of cellulosic ion exchange resins make it possible to consider these as an alternative to coagulation, clarification and scavenger resin for the treatment of highly coloured, low turbidity waters. For potable water treatment it appears that resins are generally not competitive with activated carbon. The advantage of in-situ regeneration is outweighed in most circumstances by the difficulty and cost of disposing of the spent regenerant solution. The use of ammonia as a recoverable

Developments in water treatment—2

44

TABLE 5 SUMMARY OF REPORTED DATA ON ORGANICS REMOVAL BY ION EXCHANGE AND ADSORBENT RESINS Refe Type Trade rence name

42 31

40

43 44

23

41

Test

M np XAD-2 Batch/column pD (A) hybrid M sb IRA Large column pD 904 (A) M – Column wb ce M an WRL Column ce 200A M Duolite Equilibrium wb S37(D) and column pf M np XAD-2 Column pD (A) M XAD-7 Column mp (A) ac M np XE-340 Column ca (A) M sb MP 500 Equilibrium pD (L) and column M MP 62 Equilibrium wb (L) and column pD M sb IRA Equilibrium pD 904 (A) and column M IRA 93 Equilibrium wb (A) and column pD I sb XE 258 Equilibrium ac (A) and column I wb XE 236 Equilibrium ac (A) and column M Duolite Equilibrium wb S37(D) and column pf I sb FF/IP Equilibrium po (P) and column

Solutes /Parameters

Rege nerants

Column test conditions Velocity Contact (m/h) time (min)

100 various

–

–

–

TOC

Caustic brine 50°C Caustic brine

18

2·7

–

–

10

1·5

p-nitrophenol

Caustic brine –

2·7

4

n-butanol

–

–

2·1

methyl ethyl ketone

–

–

2·1

nitromethane 1,4-dioxane

–

–

2·1

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

–

Total organics Colour

u.v.absorption

Removal of organic compounds

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 pD (L) column test M wb pD I sb ac

MP62 Extended (L) column test XE 258 Extended (A) column test

I sb TR (P) Extended po column test

45

M MP 62 wb (L) pd M Duolite wb S37(D) pf M sb MP500 pD (L)

Extended column test Extended column test Column test

u.v.absorption TOC DBS 2,4-D γ-D

M wb

Grantex Column (E) test

–

–

–

–

–

–

–

–

–

–

–

–

Caustic brine 40°C Sodium hydroxide 40°C Caustic brine 40°C Caustic brine 40°C Ammonia 20°C

2.4

12

2·4

12

2·4

12

–

–

2·4

12

NaOH 20°C

2·4

12

Caustic brine (reused)

67·5

1·1

6

10

PV Organic carbon

46

45

PV Organic carbon

Caustic 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 phenolformaldehyde 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. REVERSE OSMOSIS AND ULTRAFILTRATION 2.6.1. Introduction The process of reverse osmosis, as developed for desalination of water, is also effective in removing organic compounds from water. Chapter 4 describes the application of reverse osmosis (RO) to desalination and covers the theory, design and operation of RO plant in depth. It is necessary here to make only brief reference to the fundamental concept of RO and to define certain terms. Water is forced through a membrane which retains (rejects) dissolved inorganic and organic compounds as well as bacteria, viruses and larger particulate matter. The rejected species are therefore concentrated into a waste stream as illustrated in Fig. 6.

FIG. 6. Reverse osmosis schematic. The feedwater has to be applied to the membrane at a high pressure to overcome the natural osmotic pressure (which increases with the salinity of the feed) and provide an adequate flux (rate of permeate production per unit area of membrane). 2.6.2. Removal Mechanisms Organic species may be rejected either because they have a low solubility in the membrane material or because, whatever their solubility, their size hinders their passage through the membrane. In either case the organic molecules permeate the membrane at a much lower rate than do the water molecules, and they are therefore concentrated in the waste stream.

Removal of organic compounds

47

Negative rejection is also possible; that is, certain organic substances permeate the membrane more readily than water and their concentration is therefore higher in the permeate than in the feed. It has also been observed in laboratory tests that certain organic compounds, notably aromatics including polynuclear aromatic hydrocarbons (PAH), are adsorbed onto the membrane and are thereby attenuated in the permeate. This has implications for the use of laboratory batch tests for the prediction of RO performance. In continuous processing, however, adsorbed solutes should rapidly saturate the membrane and thereafter be rejected according to their solubility and size. Adsorption of large quantities of some organics might be expected to improve the rejection of other species, but also to reduce the permeate flux rate. Klein et al.48 describe the use of solubility parameters to predict the rejection of organic compounds by cellulose triacetate and ethyl cellulose membranes. The theory predicts that in general aldehydes, ketones, ethers and esters will be rejected less effectively than will ionised compounds or highly hydrogen-bonding compounds. To remove the former groups the theory indicates that an aliphatic nylon membrane would be effective. This type of membrane can also be operated at extremes of pH which would rapidly hydrolyse cellulose acetate. A series of RO modules with different membranes, perhaps operated at diiferent pH values, could therefore be used to obtain a very comprehensive concentrate of organic materials from water for analytical investigation. The implications for full-scale treatment are that, if particular compounds or groups are to be removed, the selectivity of various RO systems could be considered. 2.6.3. Experimental Results The rejection of many organic compounds has been studied mainly in laboratory units using the common cellulose acetate membrane material. Results for the removal of the pesticides Lindane and 2,4-D are summarised by Love;7 concentrations and/or pressures used in the reported studies do not however appear to accord with expected full-scale conditions. About 80% of organic matter (as TOC) was rejected from Cincinatti tap water by cellulose acetate membranes.49 The concentrate included hydrocarbons, phthalates, chlorinated compounds, polyethers and long chain fatty acids. Cellulose acetate membranes were also used in continuous operation on River Trent water.31 This gave valuable experience on the practical operation of both tubular and ‘spaghetti’ type RO modules. On the basis of product flux recovery and product quality, it was concluded that daily flushing with Calgon solution was the most effective of the methods tried. This regime permitted a flux of about 0·013 m3/m2 atm to be maintained at 28 atm feed pressure with either tube or spaghetti modules; the product to feed flow ratio was 75%. It should be noted that treatment of the river water prior to RO was limited to rapid sand filtration and pH adjustment to about pH 6 using sulphuric acid. The rejection of total dissolved solids under these conditions was 95% or higher. The removal of organic materials is illustrated by Table 6. Product quality is satisfactory compared with the potable water quality requirements of Table 1 except in respect of threshold odour number. It appears that odorous compounds present in the Trent water were less readily rejected than the bulk of the organic matter present, as would be

Developments in water treatment—2

48

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 (optical density at 400 nm) Optical density at 300 nm Total organic carbon (mg/litre) Anionic detergents (mg/litre) Monohydric phenols (mg/litre) Threshold odour number Pesticides αBHC (µg/litre) γ-BHC (µg/litre)

0·048–0·093

0–0·01

0·24–0·87

0–0·3

4·27–21·4

0·1–0·64

0·13–0·54

0–0·1

0–0·004

0–0·002

20–100

5–24

0·008–0·172

0–0·12

0·033–0·21

0–0·14

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

Removal of organic compounds

49

The removal of pyrogens in the preparation of pharmaceuticals and injections. As an intermediate stage in boiler feedwater treatment, to remove organics and reduce the inorganic ion load prior to a final ion exchange stage. The removal of algal products (polysaccharides) which can cause problems in soft drink manufacture.51 The concentration of regenerant liquors from organic scavenger resins. 2.6.6. Ultrafiltration Ultrafiltration (UF) is closely related to reverse osmosis. Compared with RO, the UF membrane is more ‘open’ so that only large molecules are rejected. Inorganic salts are not rejected so there is little osmotic pressure to be overcome and operating pressures are normally below 7 atm, whereas RO of waters containing up to 10 000 mg/litre salinity requires pressures of 20 to 40 atm. Product flux rate is considerably higher in UF. The mechanism may be thought of as sieving. UF has been used to concentrate large organic species, e.g. proteins, viruses and polypeptides. Dissolved solids of molecular weight greater than 1000 were found to be removed in the study of Hardt et al.52 In water treatment, humic acids would be removed and one possible application would be the reduction of humic colour prior to RO of the hollow fibre type.

2.7. CONCLUSIONS The processes of coagulation, disinfection and chemical oxidation remove organic matter to some extent, but insufficiently to meet the latest stringent requirements for industrial and municipal water quality. Adsorbent resins, though widely used in boiler feedwater treatment, have not yet found practical application in potable water supply. Reverse osmosis is extremely effective for the removal of a wide range of organic compounds but is prohibitively expensive except for small specialised duties or where its deionisation capability is also required. Ultrafiltration is considerably cheaper but effective only for the removal of large organic molecules. Resins and membrane processes suffer from the disadvantage of a concentrated liquid waste which must be disposed of, often with considerable difficulty at high cost. Activated carbon is generally the most cost-effective treatment for the removal of specific organic compounds and groups of compounds including taste- and odourproducing 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.

3.2. AMMONIA REMOVAL 3.2.1. Breakpoint Chlorination This is the most frequently used method for ammonia removal. It is usually applied in the treatment of surface waters; and has the advantage of being cheap to install and simple to operate. The basic chemistry of the process is now well known1 and will not be described here. In essence, a sufficient chlorine dose is applied prior to clarification to react with all the ammonia present to give predominantly nitrogen and a minimum of chloramines and nitrogen trichloride. If no organic nitrogen compounds are present, this occurs at the Cl:N weight ratio of 7·6:1 (the breakpoint) that varies slightly with pH. The time spent in the clarification stages of treatment should be sufficient for completion of the chlorination reactions prior to final chlorination of the water before it enters supply. Sometimes excess chlorine is applied initially so that dechlorination (e.g. with sulphur dioxide) is used to regulate the residual entering supply. It may not always be easy, however, to control the overall chlorination procedure so that an adequate chlorine residual is maintained in supply for disinfection. This is particularly the case when organic nitrogen compounds, as well as ammonia, are present thus increasing the variety and duration of the chlorination reactions. Also, nitrogen trichloride or dichloramine can be formed under many conditions, giving rise to taste and odour problems in supply. The disadvantages of breakpoint chlorination become most evident when ammonia levels are consistently high. The process then becomes much more expensive to operate in comparison with the biological method (see Fig. 1). Breakpoint chlorination has then been advocated2 as a back-up

Removal of nitrogen compounds

57

FIG. 1. Estimated costs of ammonia removal processes. process to supplement biological nitrification where the latter cannot always be relied upon to remove all the ammonia. High ammonia levels also increase the likelihood of unacceptable amounts of nitrogen trichloride being formed, especially when the chlorine dose exceeds that required to reach the breakpoint and if the pH is low. In practice, the chlorine dose will almost certainly be in excess of that required to reach the breakpoint, if only to guard against variations in ammonia level. Where dechlorination is then applied, the excess free chlorine will be removed before the dichloramine and nitrogen trichloride, which may remain. With further dechlorinating agent, the chloramines and nitrogen trichloride are converted back into ammonia. Thus dechlorination cannot entirely remedy the effects of chlorinating beyond the breakpoint. It is perhaps worth noting that activated carbon can be used for dechlorination.3 It can only be used for total dechlorination, however, so that the required chlorine residual for supply would have to be added after dechlorination by carbon. Active carbon has the advantage over sulphur dioxide of being able to convert nitrogen trichloride and dichloramine into nitrogen instead of ammonia,4 while monochloramine is converted to a

Developments in water treatment—2

58

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. 3.2.2. Physico-chemical Methods Physico-chemical methods of ammonia removal have found little application in potable water treatment, but have been given more serious consideration for wastewater treatment, where ammonia levels can be much higher. Considering first ion exchange, a comparative study carried out by Jorgensen on several different types of cation exchanger16 showed that clinoptilolite, a natural zeolite, had the highest ammonia removal capacity. It was also less expensive than conventional synthetic cation exchangers. A direct comparison of the selectivities of Hector clinoptilolite with a strong acid polystyrene ion exchange resin has shown17 that the latter preferred calcium ions to ammonium ions whereas for clinoptilolite the opposite was true. It is therefore possible to regenerate clinoptilolite with relatively inexpensive lime and calcium chloride. Pilot plant tests carried out on a wastewater containing 10 to 20 mg/litre NH3–N by the US Environmental Protection Agency18 have shown that columns of Hector clinoptilolite brought the ammonia level down to 0·5 to 1·0 mg/litre NH3–N. Further treatment by breakpoint chlorination might therefore be required to produce a potable water. As much as 99% ammonia removal has been demonstrated17 using two laboratory columns in series. The regenerant used in the pilot plant study was a mixture of sodium and calcium chlorides adjusted to pH 11 with lime. Regenerant recovery was also studied, using either air stripping to blow off the ammonia, or electrolysis which generated chlorine to react with the ammonia and so convert it to nitrogen. An advantage of the electrochemical method of removing ammonia from spent regenerant was that neutral regenerant solutions could be used, thus avoiding precipitation and scaling problems associated with the use of lime. Ion exchange using clinoptilolite was found to be a more expensive process than direct air stripping. Most studies of ammonia removal by air stripping have been carried out on wastewaters. Since ammonia is very soluble in water, an air: water volume ratio of about 3000:1 is required for 90% removal at 10°C in a counter-current air stripping tower. As the temperature falls, the amount of air required increases. For air stripping to be successful, the greater part of the ammonia must be converted into its free form by raising the pH of the water. Several studies have shown that ammonia removal efficiency increases considerably as the pH is raised to 10·5, but that little further improvement can be expected beyond pH 11. The lime dose needed to achieve pH 11 depends on the alkalinity of the water, being about 400 mg/litre Ca(OH)2 for an alkalinity of 200 mg/litre. Scaling can often cause problems when alkaline waters are treated. A pilot plant study of air stripping for potable water treatment has been carried out19 using a feedwater containing 3 mg/litre NH3–N and a tower packed with serrated wood slats. Water flowrates of up to 2 m/h were used and under the most favourable conditions an HTU of 2·3 m was obtained at 15–20°C. This implies a tower packing height of at least 7 m for 95% ammonia removal. The estimated cost of air stripping, compared to

Developments in water treatment—2

60

other methods of ammonia removal, is shown in Fig. 1. The major cost item was electrical power for the air blowers. Chemical costs for pH adjustment were not included in the estimate; and could result in a doubling of the total cost for highly alkaline waters. The results of lower removal efficiency (due to higher ammonia solubility) at low temperature and the possibility of freezing need also to be taken into account when deciding if air stripping would be a feasible process in practice. 3.2.3. Biological Nitrification The cheapest process capable of reducing high ammonia levels (>0.3 mg/litre NH3–N) to a point (<0·1 mg/litre NH3–N) where problems arising on chlorination are unlikely, is biological nitrification. Nitrifying bacteria utilise about 4·6 mg/litre of dissolved oxygen to oxidise 1 mg/litre NH3–N. Therefore if it is required to remove more than about 2 mg/litre NH3–N special measures must be taken to oxygenate the water during the nitrification process. This is usually accomplished by forced aeration through fixed beds of sand or granular material to which the nitrifying bacteria adhere. Normally, nitrification can occur at almost any stage within an existing river water treatment plant, provided the water has not already been chlorinated. Chedal2 has described how at Mery-sur-Oise nitrification in the flocculators is encouraged by sludge recirculation, so that after about two weeks operation 60% of the ammonia is removed. Up to a further 1 mg/litre NH3–N is then removed by nitrification in the rapid sand filters, depending on the backwash procedure employed. It was also found advantageous to reduce the filtration rate from 6 m/h in summer (20°C) to 3 m/h in winter (4°C), to compensate for the lower rate of bacterial activity. Ammonia removal by bacterial nitrification on rapid sand filters has also been studied by Taylor20 who found that residual chlorine present in excess of 0·2 mg/litre adversely affected the process. His results showed that about 75% ammonia removal was achieved on the filters over a period of a year, during which the NH3–N level in the raw water varied between 0·1 and 4 mg/litre and the temperature fell as low as 1°C on one occasion. Usually, however, there was a tendency for ammonia removal performance to deteriorate when temperatures fell below 4°C. The ammonia removal performance of a number of different types of filter has been studied by Madsen21 using water from the River Weser, which contained ammonia levels varying seasonally between 0·2 and 2·0 mg/litre. It was observed that during the prolonged frost of the 1962/63 winter, nitrification in the rapid sand filter ceased and did not resume until the temperature had risen again to 10°C. The performance of a slow sand filter was similarly affected, although during several other winters over 80% removal was maintained. Small-scale tests using granular active carbon or limestone of similar particle size to rapid sand (~1·5 mm) showed better nitrification performance than rapid sand. A 3 m deep granular carbon bed run at a flowrate of 10 m/h was able to almost totally remove the ammonia at temperatures down to 5°C. This was entirely due to bacterial action in the bed since carbon cannot adsorb ammonia. Below 5°C the performance fell off rapidly. The diminished nitrification performance of filters at low temperature is unfortunate, since the ammonia levels in rivers also tend to be highest at this time. Madsen has stressed21 the importance of maintaining the water as close as possible to 100% saturation with oxygen to assist performance at low temperature. He also demonstrated with a

Removal of nitrogen compounds

61

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

FIG. 2. Effect of fluidised bed solids concentration on nitrification performance. The rate of reaction was found to be approximately first-order with respect to ammonia concentration, as indicated by the gradient of unity in Fig. 3. Thus:

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

Removal of nitrogen compounds

63

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

Developments in water treatment—2

64

FIG. 4. Residence time required in fluidised bed. influent to the fluidised bed unit contained only 6 mg/litre NH3–N which was completely removed in presence of 36 mg/litre of dissolved oxygen, obtained by equilibrating the influent with oxygen gas. For most potable water treatment purposes, the use of either oxygen equilibration or partial recycling with re-aeration would be sufficient. To date, the removal of more than 2 mg/litre NH3–N from potable waters has been accomplished using two types of aerated fixed bed. They have been employed mainly for the treatment of groundwaters on the Continent where ammonia levels of 8 or even 16 mg/litre are not unknown. The treatment of these waters by nitrification is facilitated by their comparatively constant ammonia level and temperature. The two types of aerated fixed bed are the ‘dry’ sand filters favoured in Holland and the aerated pumice stone beds used in France. The operation of ‘dry’ sand filters has been described by Boorsma,28 who studied their performance for the simultaneous removal of ammonia, iron and manganese. They usually employ a 1 to 2 m bed depth of 1–2 mm sand, although sand graded down to 0·5 mm has been used. The beds are totally enclosed so that in service both air and water can be passed downwards through them, using sprays to distribute the water and blowers to introduce the air. The beds are backwashed periodically in such a way as not to significantly impair their performance when brought back into service. Operated at flowrates of from 4 to 9 m/h, the ‘dry’ filters at Hillegom were able to reduce NH3–N levels of 3–3·5 mg/litre to less than 0·15 mg/litre. Kegel29 quotes two examples of ‘dry’ filters operated at 12 m/h and an air: water volume ratio of 2·5:1 which brought ammonia levels of 1·4 and 4·5 mg/litre down to less than 0·05 mg/litre. The best published performance was found by Rossner30 of Hamburg waterworks. Using two ‘dry’ filters in series at flowrates of 10 m/h and an air: water volume ratio of 3:1, two borehole-derived waters containing 7 mg/litre NH3–N in one instance and 16

Removal of nitrogen compounds

65

mg/litre in another were thereby treated to produce a water containing less than 0·125 mg/litre NH3–N. Devillers has described in some detail31 the aerated pumice stone beds used at two 5000 m3/h waterworks near Paris. Both the air, supplied by compressors, and the water are passed upwards through open 2 m deep beds of 8–15 mm pumice. The water then overflows onto sand filters to remove the bacterial matter it carries over. The size of the pumice medium was chosen to be small enough to present a large surface area per unit volume for the bacteria to grow on without at the same time causing an unacceptably large pressure drop through the bed. It was found necessary to add a trace of phosphate to the raw water in order to maintain satisfactory bacterial performance. Water flowrates of up to 10 m/h are used, while the air flowrate is adjusted to provide sufficient oxygen to complete nitrification without producing a water that precipitates chalk. The water has an alkalinity of about 300 mg/litre and aeration inevitably removes some carbon dioxide, so raising the pH. However, since nitrification tends to lower the pH:

it was found possible to balance the two opposing effects and thus produce a stable water. One plant treats water containing 4 to 6·5 mg/litre NH3–N while the other treats water with about 2·5 mg/litre. In the latter case, the aeration is only applied intermittently to avoid expelling too much carbon dioxide. Despite the successes that have been achieved with biological nitrification, the fact remains that it is at present a less reliable (i.e. less predictable) process than biological denitrification. The period required for start-up may be long and can best be shortened by seeding the bed with bacteria from another similar plant already in operation. Before any nitrification plant is built for treating water going into supply it is essential that pilot plant trials be carried out, lasting several months for a groundwater or at least a year for a surface water. There remains a need for fundamental research to better determine all the factors affecting the process.

3.3. NITRATE REMOVAL 3.3.1. Possible Methods Before considering treatment methods for nitrate removal alone, it may be noted that any demineralisation process, removing both cations and anions from the water (e.g. reverse osmosis or a two-bed ion exchange system), will greatly reduce or eliminate the nitrate content. Economically, reverse osmosis only becomes competitive with ion exchange when the total dissolved solids content of the water exceeds about 1000 mg/litre. Reverse osmosis membranes do not reject nitrate preferentially to other common anions, whereas anion exchange resins retain nitrate in preference to both chloride and bicarbonate. Thus where the removal of nitrate only is required, ion exchange possesses this advantage over reverse osmosis. In a short but comprehensive review of nitrate removal methods, Sorg32 states that neither electrodialysis nor reverse osmosis systems have yet been installed solely for the removal of nitrate from water.

Developments in water treatment—2

66

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

Removal of nitrogen compounds

67

FIG. 5. Nitrate removal methods. complexes at both Manchester35 and Oakland, Michigan36 Universities. This work is still however at an early research stage. Biological denitrification by naturally occurring denitrifying bacteria has been studied extensively for the purification of wastewaters and sewage effluents. In the presence of a suitable organic substrate, these bacteria can reduce nitrate to nitrogen gas. The most usual substrate used has been methanol since it is relatively cheap and easy to handle. The only waste product is the excess bacterial growth which must be removed from the

Developments in water treatment—2

68

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

FIG. 6. Costs for nitrate removal per 1 mgd (December 1976). (mls=miles.) amenable to automation and should be capable of working at a borehole site with little supervision. On the other hand a troublesome waste disposal problem is usually presented by the exhausted regenerant and rinsewater which consist of a mixed solution of sodium chloride, nitrate and sulphate, resulting from regeneration of the resin with brine. If this waste has to be tankered away, to the coast for instance, the operating cost of the plant could be at least doubled, as indicated in Fig. 6. Since biological denitrification and ion exchange are the only two practical methods of nitrate removal at the present time, these will now be considered in greater detail, commencing with the longer established ion exchange method. 3.3.2. Ion Exchange The aim of treatment will usually be to reduce the level of nitrate in the water put into supply to below 10 mg/litre as N.Since ion exchange removes almost all the nitrate from the water being treated, it may only be necessary to treat a proportion of the water. The treated portion can then be blended with the remaining untreated portion to give the required nitrate level in supply. It has been found40 that commercially available strong and weak base resins show greater selectivity for sulphate than for nitrate over the concentration range (<5000 mg/litre TDS) normally found in ground-waters. It is not, therefore, at present possible to

Developments in water treatment—2

70

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 styrenedivinylbenzene 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

FIG. 7. Treated water quality from an ion exchange column. (20 mg/litre) results in the volume of water treated per regeneration being doubled and hence in halving the regenerant and waste disposal costs of either ion exchange system. The effect that high raw water sulphate levels have in raising regeneration and waste disposal costs could be overcome by the development of a resin that was more selective for sulphate than for nitrate. Grinstead and Jones44 have achieved this objective with liquid amidine resins for which they proposed ammonium hydroxide as regenerant. The major problem that they anticipated was in the means by which a liquid resin could

Developments in water treatment—2

72

actually be employed to treat water and in preventing an unacceptable concentration of the resin itself appearing in the treated water. More recently45 the introduction of amidine groups onto conventional acrylic resin beds has resulted in a resin with significantly reduced sulphate selectivity and which can be regenerated with brine. However, it also exhibits a low capacity and low ‘regenerability’ and it is therefore doubtful if it will possess any net advantage over existing commercial resins. 3.3.3. Biological Denitrification A comprehensive study has been carried out37 to assess the performance and applicability of biological denitrification to the treatment of water for potable supply. The data presented here are based on the results of this study in which methanol was used as the source of organic carbon for the denitrifying bacteria. The water was derived from the River Thames with provision for dosing extra nitrate when required. The treatment objectives were to obtain design criteria for a plant capable of reducing the nitrate level by 10 mg/litre (as N) at temperatures below 5°C, without necessarily removing the nitrate completely. The methanol level in the treated water was however required to be kept very low (<0·5 mg/litre), since it was anticipated that in practice strict limits would be placed on the methanol level in the water entering supply after filtration, if not also on that in the water leaving the denitrification unit itself. Since both temperature and methanol concentration were rate-limiting, these conditions greatly affected the residence time required in the denitrification unit. Three types of denitrification unit were studied: fixed beds, suspended growth, and fluidised sand units. One fixed bed contained 1 m depth of 10 mm gravel, and another 2 m depth of 25 to 40 mm gravel. The bacteria grew as a film on the gravel surface and the excess growth had to be removed periodically by back-scouring to prevent the beds from clogging. Water flowrates of 0·7 to 2·8 m/h were used during treatment. The suspended growth unit was a tall tank in which the water flowed vertically upwards. After initial seeding with fine sand or river silt as a basis for attachment, the naturally occurring denitrifying bacteria formed grey gelatinous globules which remained suspended as a fluidised layer in the tank. Most of the globules were 1 to 2 mm in diameter, but some were as large as 6 mm. The suspended layer increased in depth until, when it filled

Removal of nitrogen compounds

73

FIG. 8. Concentration profiles (fluidised sand bed). the tank, the excess bacterial growth was carried over in the denitrified water. Upflow rates of up to 24 m/h could be used, although 12 m/h was more usual. The lower rate was

Developments in water treatment—2

74

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

TABLE 1 DENITRIFICATION UNIT CHARACTERISTICS Fixed Fixed Suspended Fluidised bed bed growth sand Media dia. (mm) 25–40 Bacterial matter concentration (4–8) (kg/m3) Bacterial film area 120 per unit volume (m2/m3)

10

1–2

0.2–0.5

(5– 10)

5–20

10–15

360

700

3000

Removal of nitrogen compounds

Denitrification 10°CM rate (gN/m3h) 20°CJ

75

2.5

12

60

160

5

24

120

310

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

FIG. 9. Effect of temperature on performance of suspended growth (SG) and fluidised sand (FS) biological denitrification units. TABLE 2 DENITRIFICATION UNIT OPERATING REQUIREMENTS FOR REMOVAL OF 10 mg/Htre NO3–N AT 2°C Type of unit Suspended growth Fluidised sand

Upflow Bed rate (m/h) depth (m)

Residence time (min)

12

5·0

25

20

3·6

11

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

39. GREENE, L.A., Water Pollut. Control, 1978, 77, 478. 40. BUELOW, R.W., KROPP, K.L., WITHERED, J. and SYMONS, J.M., J. Amer. Water Works Assoc., 1975, 67, 528. 41. CLIFFORD, D.A. and WEBER, J.W., Ind. Water Engng, 1978, 15, 18. 42. EVANS, S.J., Water Pollut. Control Fed., 1973, 45, 632. 43. GAUNTLETT, R.B., Water Treat. Exam., 1975, 24, 172. 44. GRINSTEAD, R.R. and JONES, K.C., Water Pollut. Control Research Series, Project 17010, FSJ, 1971, US EPA, Water Quality Office, Washington. 45. Diaprosim Ltd, Contract Report, 1979, Water Research Centre, Medmenham, Bucks. 46. Brit. Pat. 1430410, 1973.

Chapter 4 DESALINATION M.J.BURLEY, B.Sc., M.Sc., C.Eng., M.I.Chem.E., M.I.E.W.E.S. Sir M.Mac Donald and Partners, Demeter House, Cambridge, UK AND J.D.MELBOURNE, B.E.(Chem) (Aust), Ph.D.(Cantab.) Melcon Water International Ltd, Henley-on-Thames, Oxon., UK SUMMARY Desalination has progressed significantly since the simple distillation techniques first used for producing small quantities of drinking water from sea or saline ground waters. Distillation is normally now only applied to sea waters and the most successful, reliable and widely applied technique is the multi-stage flash process, particularly in the Middle East. Improved scale control and use of materials have been the major developments with this process. A new technique using fluidised bed heat exchange may lead to cost reductions, but this will not be fully commercial for several years. Reverse osmosis and electrodialysis, both membrane techniques, are the two most commonly applied processes for the desalination of brackish waters. Of these, reverse osmosis has been the more successful and, although a relatively new process, is now considered to be technically and commercially reliable. Developments with reverse osmosis membranes have also led to a capability of desalting sea water and several small commercial plants have been installed. It is possible that the slightly more attractive economics of sea water reverse osmosis will result in a decrease in sea water distillation applications. Electrodialysis also has the capability of treating sea water but this is not yet commercial. The other techniques of freezing and ion exchange have had little commercial impact in the desalination field.

4.1. INTRODUCTION Distillation processes, in a simple form, have been employed for over a century as a means of producing drinking water from the sea. However it was not until the late 1950s, with the development of oil communities in the Middle East and the Caribbean that demand arose for large-scale land-based plant. This market spurred several governments to finance research and development in the field of desalination and foremost of these programmes was that of the United States Office of Saline Water. After the initial 4500 m3/day multi-stage flash (MSF) distillation plants were installed in the late 1950s, the process was rapidly established as the foremost desalination technique. Works of the 1970s involve units of up to 35000 m3/d capacity and countries

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 selfsupporting, fine, hollow fibres. During the 1970s the major changes which have occurred are the gradual improvement of membrane properties, a better understanding of membrane fouling and the measures required to eliminate the associated problems, and the development of membranes capable of treating sea water in a single pass. Whilst sea water treatment by reverse osmosis has not yet been widely used it is clear that the process will develop and become an economic competitor of distillation for large-scale sea water treatment. Whilst the use of electrodialysis for the treatment of brackish waters with total dissolved solids contents up to about 5000 mg/litre has continued to grow, the application of the process has clearly suffered from the competition of the simpler reverse osmosis process. The use of frequent current reversal to counteract polarisation effects and consequent scaling problems is the most significant single development during the 1970s since this reduces the need for pre-treatment and the use of chemicals, a factor of economic importance in situations where the costs of imported chemicals are high. Other processes under study in the 1960s included solar distillation, ion exchange and freezing techniques. The development of these processes has slowed down, if not ceased, during the 1970s. Despite the potential of lower energy consumptions it seems unlikely that these desalination methods will be developed to a commercial stage for large-scale use.

4.2. SEA WATER DISTILLATION The boiling of saline water and condensing of steam to produce potable water has been employed for centuries but it is only in the last 30 years that the factory-built plants to produce large quantities of water have been developed. As early as 18651 the water supply for General Napier’s expedition into Ethiopia was provided by two 30 ton/day units installed at Alexandria. The process was basically similar to the multiple-effect submerged tube distillation which was employed through to the 1950s. In this type of plant feed steam is condensed on the inside of coils or tube banks suspended below the surface of the feed sea water held in the first effect. The steam produced from the vessel is passed as heating steam to the next effect. In general, submerged tube plants consist of no more than three effects and therefore have a poor efficiency in terms of energy use. The performance ratio R (mass of product water per

Desalination

81

unit mass of feed steam consumed) is typically only about two. The process has two other serious defects: (1) The rate of heat transfer between condensing steam and evaporating water is very low owing to poor turbulence outside the tubes—the overall heat transfer coefficient is typically 500 W/m2 °C (100 Btu/ft2 h °F). Thus large areas of expensive nonferrous heat transfer surface are required. (2) Scale which forms on the outside of the tube bundles is extremely difficult to remove. The introduction of multi-stage flash (MSF) distillation in the late 1950s revolutionised desalination and provided a practical, economic and relatively trouble-free means of supplying drinking water. This development arose directly as a result of the large demands for drinking water which were created by the post-war growth of the Arabian Gulf Oil States and similar demands in the Caribbean and developing island communities

FIG. 1. 22750 m3/d MSF Evaporator (one of four units constructed by Weis Wertgerth Ltd for Quatar). and has led to the construction of works similar to that shown in Fig. 1. The major benefits of MSF distillation over the submerged tube process are as follows: (a) Sea water passes at high velocity over heat transfer surfaces thereby allowing the heat transfer coefficients to be increased by a factor of about five to 2800 W/m2 °C (500 Btu/ft2h °F).

Developments in water treatment—2

82

(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 multistage flash distillation process is still used almost exclusively for the large land-based sea water desalination plants (greater than 5000 m3/d capacity). Therefore MSF distillation has been selected for detailed consideration. 4.2.1. Multi-Stage Flash Distillation The multi-stage flash process can be illustrated as shown in Fig. 2. Cooling water is fed to the condensers of the heat rejection section of the plant. On

FIG. 2. Simplified flow sheet MSF. emerging, a portion of the cooling water is separated and, after chemical treatment, is passed to the inlet of the recycle pump where it is mixed with flashing brine withdrawn from the final stage (stage 7 in the works illustrated). After discharge of the concentrated waste stream, MB, the recycle flow MR is passed through the condensers of the heat recovery section where it is successively heated by condensing steam. On emerging from the condensers of stage 1 the recycle stream is heated to its highest temperature, TFO, by heat exchange with feed steam in the brine heater.

Desalination

83

The heated brine then enters the flash chamber of stage 1 which is held at such a pressure that a portion of the brine stream flashes off thereby cooling the flashing stream to its boiling point at the stage pressure. As the flashing brine stream passes from stage to stage it is successively cooled with a portion of the flow flashing off, to be condensed as distillate in each stage. The temperature distribution is shown in Fig. 3. Mass and heat balances

FIG. 3. Temperature distribution MSF. can readily show that the performance ratio R approximates to the following:2 (1)

Thus the energy consumption can be reduced by the following means: (i) Increasing ∆TT the total flashing range. Since the lower temperature TRI is fixed by the temperature of the sea, ∆TT can only be increased by raising the top temperature TF0. Developments in scale control have been aimed at increasing the top temperature whilst avoiding undue corrosion. (ii) Reducing the stage temperature drop ∆t. This can only be achieved by increasing the number of stages employed and hence increasing the capital cost of the plant.

Developments in water treatment—2

84

(iii) Reducing the value of , the stage approach temperature (the temperature difference between recycle brine leaving the stage and condensing steam). This can be achieved by use of more heat transfer area in each stage or increasing the heat transfer coefficients. α is the temperature difference between flashing brine and condensing distillate and is the sum of the boiling point elevation and the temperature difference caused by pressure losses. Little can be done to reduce the value of α. From points (ii) and (iii) above it is clear that plant design can be optimised to achieve a lowest water cost by balancing capital with fuel costs (or balancing heat transfer area and number of stages with performance ratio). This is shown by the following approximate relationship: (2) where As, the specific heat transfer area, represents the heat transfer area per unit of product water output; λ is the mean latent heat of evaporation over the flashing temperature range; U is the mean heat transfer coefficient; ∆TT is the flashing temperature range; n is the number of stages; and R is the performance ratio. Using typical values, this relationship has been plotted as shown in Fig. 4.

FIG. 4. Heat transfer requirements MSF.

Desalination

85

From eqn (2) and Fig. 4 it can be seen that costs of MSF distillation can be reduced by the following means: (1) Increasing the flashing range ∆TT. (2) Increasing the heat transfer coefficient U. (3) Use of cheaper heat transfer and shell materials. (4) Use of improved designs to permit smaller flash chamber. (5) Use of cheaper energy sources. It is therefore of value to consider the developments in distillation in terms of their effects upon these factors. 4.2.2. Scale Control The flashing temperature range is limited by the maximum brine temperature at which scale formation can be satisfactorily controlled. Alkaline scale formation, caused by the breakdown of bicarbonates in sea water to deposit calcium carbonate and magnesium hydroxide scales, has traditionally been controlled by one of two alternative means: (a) The dosing of polyphosphates at low levels, 5 to 10 mg/litre, does not completely eliminate scale formation but rather causes scales to form as sludges which do not adhere strongly to the heat transfer tube surfaces. The use of these materials limits the top temperature to about 90°C since at higher temperatures the phosphates break down and become ineffective. Periodically it is necessary to acid flush the condenser tubing in order to remove the sludge build-up which will otherwise cause a deterioration of plant performance. (b) Pretreatment of the feed water by dosing of acid, at a concentration of about 120 mg/litre, effectively removes the scale-forming bicarbonates with evolution of carbon dioxide. The gas is removed by air or steam stripping prior to entry of feedwater to the plant, otherwise severely corrosive conditions will be created within the evaporators. Even so, the negligible buffering capacity of the recycle brine makes the control of pH difficult and therefore the risk of corrosion will always be present when acid dosing is used for scale control. By the use of acid, top temperatures of 110−115°C can be achieved without serious scale formation. The benefit of this higher temperature can be illustrated as follows: If the temperature of sea water controls the bottom flashing temperature to 38°C, the total flashing range with acid treatment will be 77°C (=115 −38) as compared with 52°C (=90−38) for operation with polyphosphates. Substitution of these values in eqn (2) will show that the heat transfer area can be reduced by about 47% by the use of acid dosing while still retaining the same number of stages and performance ratio. A rigorous evaluation will show the difference in costs between polyphosphate and acid dosed plants to be greater at high performance ratios. The approximate relationship is shown in Fig. 5.

Developments in water treatment—2

86

FIG. 5. Plant capital costs— polyphosphate/acid. This capital cost saving has led many users to select acid dosed plant in preference to the more expensive polyphosphate dosed works. In a number of cases poor pH control and the use of unsuitable materials has caused premature failure of heat exchange tubing and steelwork, so much so that the cost of replacements and repairs has outweighed the initial benefits. The Ministry of Electricity and Water, Kuwait, is typical of several authorities who have deliberately avoided the use of acid and in so doing have maintained a high standard of reliability. Belgarde for Scale Control One of the most significant improvements of recent years is the development of additives which effectively control scale at high temperatures without the creation of corrosive conditions. Whilst many of the chemicals developed have shown only marginal benefits over the traditional polyphosphates the Belgarde range of chemicals, developed by CibaGeigy Ltd have been found to be most effective. Top temperatures of 105–110°C have been successfully achieved over many months of operation with insignificant rates of fouling of heat transfer surfaces and with iron and copper dissolution rates typically 10% and 30% of those experienced in acid dosing plants. Trials with Belgarde at Qatar are typical of many uses of the chemical. In this case the plant, 9000 m3/d capacity, had previously been operated with polyphosphates at a maximum temperature of 85°C and intermittent acid cleaning at six-monthly intervals. Dosing with 7·5 mg/litre of Belgarde over a five-week monitored test period showed no increase in fouling factor with a top temperature of 110°C and an output of 11 800 m3/d, an increase of 30% above that for polyphosphate operation. Extended operation of 150 days at a top temperature of 105°C and a Belgarde dose of 6 mg/litre showed the fouling

Desalination

87

factor to remain constant whilst the output was maintained at 10 900 m3/d, an increase of 21% over the production rate when polyphosphates were employed. Without doubt the high maintenance cost often associated with high temperature plant will be minimised in the future by the use of scale-control polymers such as Belgarde. 4.2.3. Fluidised Bed MSF A second development which permits the use of high temperatures without resort to acid dosing involves the use of a fluidised bed of ‘seeds’ within the recycle condenser tubes.3 In this process, the stages of a multi-stage flash evaporator are mounted vertically with condensers located at the side of the flashing chambers (Fig. 6). The recycle brine is pumped vertically up the condenser tubes at a velocity adequate to fluidise glass particles typically of 2 to 4 mm diameter. The purpose of the fluidised bed is fourfold: (1) To create sites for scale precipitation and thereby minimise deposition on heating surfaces. (2) To abrade the surface in order to remove what scale does deposit. (3) To take advantage of the high film heat transfer coefficients experienced with fluidised beds. (4) To achieve scale control without incurring the costs associated with the use of chemicals. Whilst the development of the process is by no means complete the

Developments in water treatment—2

88

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

Desalination

89

4.2.4. Multiple Effect Distillation Multiple effect distillation has been employed in several forms. A vertical falling film process was employed at Freeport, Texas, for many years as a part of the US Office of Saline Water test programme. Vertical tube climbing film evaporators have been tested and horizontal tube multiple effect evaporators have been employed. All of these types have a single common design feature—steam is condensed on one side of the heat transfer surface and water is boiled on the other. It has been shown in several cases5,6 that the multiple effect cycle is potentially more economic than the MSF, particularly if the high heat transfer rates, that are theoretically possible with falling film, spray film and fluted tube systems, can be reliably maintained in practice. Unfortunately the thin films employed to achieve high heat transfer rates tend to be unstable. Maldistribution of the film, created either by blockage of the sensitive water distribution nozzles or by slight scale deposits formed at the sites of imperfections in the heating surface, creates local points where the tube wall temperature rises and the rate of scale formation increases. The laminar flow conditions within the film provide insufficient turbulence to remove deposits and hence accelerating scaling develops. It is unlikely that the consequent instability of heat transfer performance can be avoided particularly since the feed sea water is always somewhat variable in quality and contains suspended matter. Despite the theoretical potential of multiple effect systems, particularly in combination with vapour recompression cycles in situations when high performance ratios are called for, multi-stage flash distillation is likely to remain the only serious contender for the large-scale sea water evaporator market. 4.2.5. Materials for Evaporators7 The early MSF plants followed normal marine practice, with the use of copper-base alloys for heat exchanger tubing and, in general, this practice has continued. Thus the Kuwait MSF plants normally use 70/30 cupronickel tubing for the heat reject, brine heater and high temperature stages with aluminium brass used elsewhere. Approximately 25% of the tubes are cupro-nickel and the remainder aluminium brass. The cupro-nickel alloy used in the first plants was the 2% iron 2% manganese version. This practice has continued and has spread to most other areas, apart from the USA, where standard 70/30 cupro-nickel has been specified. Tube plates for the plants are usually Naval brass or 90/10 cupro-nickel. In contrast to the above practice for polyphosphate dosed plants, there has been a tendency to use 90/10 cupro-nickel instead of aluminium brass in acid dosed plant, and some units have been built with this alloy in all sections—an example of this is Jeddah 1. Overall, therefore, there has been an increase in usage of 90/10 cupro-nickel at the expense of aluminium brass. Some data8 show the overall picture of about 65% aluminium brass in 1971 and about 35% in 1975. Data from the AD 1972 Little Survey9 shows 90/10 cupro-nickel to perform significantly better than aluminium brass in acid dosed plants so that this switch to 90/10 cupro-nickel is probably based on an increased usage of acid dosed plants during the 1971–75 period.

Developments in water treatment—2

90

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 corrosionresistant 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 cupronickel is not prone to these problems but is more expensive than stainless steel, hence the use of both materials for this application. For smaller units, where the shell thickness is low, cupro-nickel is the normal choice where an alloy material is specified, as stainless steel would suffer problems externally. For large plants clad steel plate is used for economic reasons and this circumvents external corrosion problems where stainless steel is specified. For polyphosphate plants alloy linings are not normally used apart from local areas around weirs and nozzles. Elsewhere deposits of carbonate and hydroxides on the steel surface seem to markedly reduce corrosion of the underlying metal. This trend towards the use of alloy lined chambers is probably the most significant change in material usage in desalination in recent years. It can in general be concluded that over the last 10 years no new materials have been employed in evaporators but rather the practice has been to upgrade materials used and consequently fewer problems tend to be experienced. 4.2.6. Cost of Sea Water Evaporation Capital Costs The capital costs of sea water evaporator installations are comprised of three elements: (1) The evaporators.

Desalination

91

(2) The feed steam system—self-contained boilers or, more commonly, the supply system from combined power generation plant. (3) Ancillary works—civil works for the evaporators, post treatment or blending facilities, storage and distribution works. Typical costs for high temperature, acid or Belgarde dosed evaporators are shown in Fig. 7. The costs are presented in terms of pounds sterling per unit of plant size and performance ratio. It can be seen that a typical evaporator of 10000 m3/d capacity with a performance ratio of 10 will cost about £5·8 million at 1979 prices. Clearly such costs can only serve as a guide since site conditions and the state of the market as well as specific design details will influence the precise cost in an actual case. Costs of polyphosphate dosed plant are higher, as shown in Fig. 5.

FIG. 7. Capital costs MSF evaporators (acid dosed) 1979. The civil works directly associated with the evaporator typically add 15% to the cost of the installation but if post treatment works to artificially harden the distillate prior to distribution are employed higher additional costs will be incurred. The total cost of self-contained, low pressure, water-tube boilers installed together with auxiliary equipment has been taken as £30 per kg/h of steam raising capacity. Thus typical capital costs for a 10000 m3/d installation are shown in the following table. For the fuel costs assumed, the performance ratio for a single purpose plant is optimised at 12 whilst that for corresponding dual plant is 10. Operating Costs

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 purpose

Dual purpose

Capital costs (£ million) Opt. performance ratio (12) Distiller cost 6·90 Boiler cost 1·23 Civil cost 1·03 Total capital 9·16 Operating costs (£ £mpa p/m3 £mpa 3 million p.a and p/m ) 1. Fixed charges 1·08 42·3 0·96 11.75% (10% interest, 20 years) 2. Fuel at $20 per barrel 0·83 32·5 0·50 (half cost for dual purpose) 3. Chemicals Polymer 0·09 34 0·09

(10) 5·80 1·52 0·87 8·19 p/m3 37·7

19·5

3·4

Desalination

£2000/tonne, 7 mg/litre 4. Auxiliary power (at 2p/kWh) 5. Maintenance operation, staff and materials Total annual cost (£m) Total water cost (p/m3)

0·20

93

7·9 0·20

7·9

0·25 10·0 0·25 10·0

2·45 – 2·00 – – 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 Polymer Acid

£500/tonne £2000/tonne £120/tonne

With typical doses of 7 mg/litre for polymer and 120 mg/litre for acid there is little between the costs of these two chemicals for high temperature operation. The polymer cost quoted in the table of operating costs is based upon the use of a feedwater rate 2·4 times that of the product, thereby limiting the concentration of the brine blowdown to 1·7 times that of the feed sea water. Maintenance and operation. The annual cost of maintenance typically contributes some 1·5% of the capital cost of the distiller whilst operation is approximately double this value for a plant of 10000 m3/d. Thus the combined cost contributes about 10 p/m3 to the cost of water. Water costs. It can be seen that for the conditions assumed total water costs range from 78·5 p/m3 for the dual purpose works to 96·5 p/m3 for the single purpose case. The selection of steam turbine drives for major pumps for single purpose plant can save up to about 4 p/m3 on the cost quoted by raising steam at an adequate pressure to drive the turbines before being employed as feed to the brine heater.

4.3. REVERSE OSMOSIS Reverse osmosis is a pressure driven, membrane desalination process which has undergone the most rapid development of any desalination technique. To some extent this has been due to substantial funding of research and development by the United States and other international governments. The process, also known as hyper filtration, has now

Developments in water treatment—2

94

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

FIG. 8. Basic principles of reverse osmosis.

Desalination

95

The CA and TFC membranes have normally been employed in the form of a flat sheet and the polyamide as thin hollow fibres. However, there is one hollow fibre system with cellulose triacetate membranes. The osmotic pressure of brackish waters would typically be in the range 140 to 350 kN/m2 (20 to 50 psi) and that of sea water 2450 kN/m2 (350 psi). In practice it is necessary to apply a sufficient pressure above the osmotic pressure to provide satisfactory product water rates. Hence operating pressures for brackish water systems are typically 2800 kN/m2 (400 psi) but can vary through the range 1400–4200 kN/m2 (200–600 psi) dependent on the application. Pressures of 5600–7000 kN/m2 (800–1000 psi) are required for sea water. The total salt rejection in brackish cases can be controlled to some extent by the type of membrane applied but typically 95% plus would be expected in a single pass system. For sea water, single pass systems can be employed using membranes which give 99% salt rejection; however, lower rejection membranes have been employed in a two pass system in which the product from the first pass is used as feed for the second pass. The total world installed capacity for brackish water RO systems can be estimated as 1550 Ml/d. This is based on the 1977 Desalting Plants Inventory of the US Department of the Interior11 with allowance being made for small plant not included in that inventory and the growth in the market to June 1979. The largest installed plant at this time is at Salbukh in Saudi Arabia and has a capacity of 46 Ml/d. The largest installation proposed is the 360 Ml/d plant for Yuma in Arizona, USA. This is to desalt saline drainage water from the Wellton Mohawk irrigation and drainage area before it discharges into the Colorado River. The scheme is to maintain a satisfactory quality in the Colorado before it passes to Mexico. The total world installed capacity of sea water RO systems to June 1979 is approximately 19 Ml/d. The largest of these is a 12·5 Ml/d plant in Jeddah in Saudi Arabia. However, a number of plants are under construction and the largest of these is a 13 Ml/d plant for treating the Caspian Sea in Russia. Within a year it is expected that the total installed capacity will be approximately 45 Ml/d. Ultrafiltration (UF) is also a pressure-driven membrane process worth noting, although it will not be discussed in any detail in this chapter as it has had very little application in the water treatment field. It is an extension and development of reverse osmosis using more open high flux membranes (4–8m3/m2/d) which will only reject larger molecules, and is applied to solutions with dissolved solutes with molecular weights above 1000. Dissolved inorganic salts and small organic molecules will pass through the membrane with the permeate fluid. The osmotic pressure for these systems is thus very small and operating pressures are usually between 175 and 700 kN/m2 (25–100 psi). Its main potential as a water treatment process is for the removal of colloids and high molecular weight dissolved organic substances in the feed water to reverse osmosis or ion exchange plant. In addition it has been applied in a few instances for the direct removal of colour, i.e. fulvic and humic acids, from water. This may be a potential area of increased application but unless costs are significantly reduced it will only be used where performance is more important than cost. It is estimated that UF applications in Europe may have a total installed membrane area of 25000 m2 mainly in the electropaint and dairy industry. No doubt the installed area in the USA would be significantly higher than this.

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.

FIG. 9. Membrane configurations. Typical performance characteristics of the various membrane types are detailed in Table 2. It can be seen that the membranes have pH and temperature limitations which at times can restrict the direct application of the process. CA membranes are subject to hydrolysis

Desalination

97

but within the range specified this is not excessive. Longer membrane life can be achieved by operating between

TABLE 2 TYPICAL CHARACTRISTICS OF OSMOSIS MEMBRANES (2800 kN/m2 (400 psi) FOR BRACKISH WATER AND 6000 kN/m2 (850 psi) FOR SEA WATER) Membrane pH Max. Inorganic Flat Allowable type range Temp. salts sheet free Cl in (°C) rejection flux feed (%) (m3/m2d) (mg/litre) Cellulose acetate (a) Brackish (several types) (b) Sea

3–8

3–8

40

35

98

0·73

95

0·85

98

0·6

0·5 above 25°C 1·0 below25°C 0·5 above 25°C 1·0 below 25°C

Polyamide hollow fibre (a) 4–11 35 95 Brackish (b) Sea 5–9 35 98·5 TFC (a) 2–12 45 98 0·7 Brackish (b) Sea 2–12 45 98·5 0·7 Note: These data refer to plant performance of 75%, recovery for brackish water and 30% recovery for sea water.

0·1 0·1 Nil Nil

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

FIG. 10. Flux decline in hollow fibre membranes (by courtesy of DuPont). rejection the more raw water that can be blended to achieve a 500 mg/litre TDS potable supply. For single pass sea water desalination with a product requirement of 500 mg/litre TDS or less, a rejection of greater than 98% is necessary. Partial second stages will be necessary with some membranes and for some high salinity sea waters. Typical performance data from cellulose acetate membranes operating at 2800 kN/m2 (400 psi) and 50% recovery on a brackish water are shown in Table 3.

Desalination

99

TABLE 3 RO TREATMENT OF BRACKISH WATER AT YUMA—USA BY SPIRAL WOUND CELLULOSE ACETATE MEMBRANES12 Feed (mg/litre) TDS Ca Mg Na K Total Fe Sr Total Alk as CaCO3 Cl SO4 pH

Product (mg/litre)

2900 113 70 836 8 <0·06 2 9

85·0 0·87 0·54 31·0 0·15 <0·06 0·2 6·5

1036 826 5·5

44·0 2·0 5·2

In addition it is known that such a membrane would have approximately an 86% rejection of nitrate ands would largely reject organic compounds with a molecular weight down to approximately 200. Hollow fibre membranes would have very similar rejection capabilities. TFC membranes operating on a brackish water would also provide similar rejections except that improved nitrate and organics rejection would be expected.13 A rejection of up to 99% could be obtained for nitrate. The improved organics rejection could be significant if chlorinated and other low molecular weight organic compounds must be removed from drinking water. However more data need to be generated to determine the full potential of organics removal with TFC membranes. It may also be possible to tailor the membrane properties to reject different types of organic compounds. RO membranes can also almost completely reject aluminium, heavy metals, bacteria and pyrogens and hence overall a high quality water can be produced.

TABLE 4 RO TREATMENT OF SEAWATER— JEDDAH, SAUDI ARABIA, WITH SPIRAL WOUND TFC MEMBRANES14 Feed (mg/litre) Temp. (°C) PH TDS Ca Mg HCO3(asCaCO3) Cl SO4

32 max. 8 41200 520 1460 125 22000 2960

Product (mg/litre) 7.5 ~800 20 ? 65 600 20

Developments in water treatment—2

SiO2 Total Fe

100

0·4 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 (mg/litre) pH TDS Ca Mg Na HCO3 (as CaCO3) Cl SO4 Fe

Product (mg/litre)

7·4 42000 600 1118 16000 190

6·5 435 8 1 133 20

25000 2900 3·8

210 1 0·2

output. The fluxes quoted are for new membrane but these do decrease over the membrane life as seen in Fig. 10. This decrease is due to compaction or densification of the polymer support layer which occurs from operation at high pressures. All membranes are subject to compaction but this is relatively low below a 2800 kN/m2 (400 psi) operating pressure. The TFC membranes have minimal compaction since it is possible to optimise the support layer during fabrication. Significant progress is being made in the development of TFC membranes which have the above flux and rejection characteristics for brackish waters but which will operate at 1400 kN/m2 (200 psi). A membrane of this type should be commercially available in the near future and ultimately similar benefits are likely to be achieved for sea water membranes. Biological activity can develop in RO modules and to prevent attack on cellulose acetate membranes a free chlorine residual in the feed water is often recommended. The chlorine also controls this type of activity in the other components of the overall plant. The hollow fibre and TFC membranes are non-biodegradable and hence do not require this form of protection. In fact the polyamide or similar polymers are very sensitive to chlorine and can be readily oxidised. Hence very low or zero chlorine must be present in their feedwaters. However, while non-cellulosic chlorine resistant membranes with their wider range of properties are highly desirable, the sensitivity of the present membranes

Desalination

101

should not technically limit their applications. In fact polyamide hollow fibres have been widely applied and for chlorinated feedwaters a dechlorination stage is readily included. For a relatively new process such as RO the membrane life is a key factor in the operating costs. For brackish waters most major manufacturers now have sufficient experience and confidence to offer a three-year guarantee provided the feedwater specifications have been met. In many cases they would expect the life to be substantially longer. The more recent sea water and TFC membranes have yet to be fully proven in large installations but again manufacturers are confident of a minimum three-year life. 4.3.2. Reverse Osmosis Modules Not only must reverse osmosis membranes have satisfactory desalination properties but they must be produced in such a form as to be capable of withstanding the high forces created under operating conditions. Whilst the fine hollow fibres are mechanically selfsupporting all other types of membrane have to be provided with some form of porous support. There are five different module designs which have been considered seriously, namely tubular, plate and frame, spiral wound, hollow fibre and the rod type. One form of the tubular module concept typically consists of 13 mm internal diameter porous tubes up to 3660 mm long, connected in series in a repeating ‘S’ configuration with up to 18 tubes per module. These tubes are the membrane support and must be capable of withstanding the operating pressure of the system. Cellulose acetate type membranes have been generally used and can be cast directly onto the inner surface of the tube or more usually the membrane is cast onto the inner surface of a synthetic paper tube. The paper membrane tube is then inserted into the support tube. The advantage of this approach is that the modules can be very simply and economically remembraned where necessary. Saline water under pressure is pumped at the required velocity through these tubes in the modules. Fresh water passes through the membrane and porous support and is collected by an outer shroud surrounding the tube bundle. Tubular systems have well defined open flow channels and are very tolerant towards suspended solids and are difficult to block. They are particularly amenable to cleaning by water flushing and chemicals, and, when necessary, can be mechanically cleaned by foam swabs without dismantling the equipment. However, the packaging of large areas of membrane in this way is not economically competitive with the alternative spiral wound and hollow fibre systems, and therefore tubular membranes are not often employed for municipal water treatment. They do however find extensive use for special applications in the dairy industry and are offered by Paterson Candy International Ltd in the UK. The plate and frame concept, similar to a filter press, is a viable approach technically but again is not applied in the water field as it also is not competitive economically. The ‘rod’ system was developed by the United Kingdom Atomic Energy Authority, Harwell,16 and in terms of packed membrane area per unit volume of module it falls between the tubular and the higher packed density of the spiral wound and hollow fibre configurations. In its latest concept the rod system is based upon a 3 mm diameter microporous polypropylene tube onto which the membrane can be cast directly and continuously. Bundles of 37 or 109 of these rods, 4 or more metres long with one end sealed and the other mounted in a header block, are placed in tubular pressure vessels. The feed liquid is pumped through the pressure vessel at an appropriate velocity and

Developments in water treatment—2

102

pressure on the outside of the rods. Permeate passes through the membrane and porous tube and is collected from the header block end. Membrane bundles can be quickly and easily replaced and are disposable. However, the porous support tubes are not suitable at this time for operating pressures greater than 600 psi which would limit the system to brackish water applications. The system is more tolerant to suspended solids than spiral wound and hollow fibre modules but less tolerant than tubular designs. It also has the advantage that in the event of severe malfunction the membrane rod bundle can very easily be removed for some form of physical cleaning. However, although the system has significant potential it is considered to be not commercially competitive for municipal water treatment by Paterson Candy International Ltd who hold the marketing rights for the process. Hence at the present time it has limited application. The two most extensively used module types, the spiral wound and the hollow fibre, differ significantly in designs. The spiral wound element was developed by General Atomic of the Gulf Oil Corporation, now the Fluid Systems Division of Universal Oil Products (UOP) Inc., under US Government sponsorship. It has since been adopted by other US companies, typically Hydranautics and Envirogenics, as their basic configuration. The membrane element usually consists of several membrane leaves attached to a central product carrying tube. A basic leaf is shown in Fig. 11.

FIG. 11. Spiral wound membrane module. Two flat membrane sheets with their desalting skin facing outwards are separated by a very fine, porous, polymer sheet less than 1 mm thick which supports the membranes under pressure. Three sides of the ‘envelope’ are sealed with a suitable glue and the fourth side is glued to a porous section of the central tube. A more open porous mesh approximately 1 mm thick is placed on the outside of the membranes. Then this leaf and others attached in the same way are rolled around the centre tube. The membrane elements are usually 1 m long with diameters of 4 in or 8 in being generally used although UOP can supply 12 in diameter elements. An 8 in diameter element 1 m long would have an effective membrane area of approximately 320 ft2. Up to six elements can be inserted in series into each fibreglass pressure vessel. The feedwater at pressure passes

Desalination

103

through the porous mesh channels parallel to the membrane and the fine mesh supporting the membrane is the product water carrier to the centre tube. The output capacity for 6×8 in diameter membrane elements, 1 m long, operating at 2800 kN/m2 (400 psi), 25°C and a total 60% recovery treating a brackish water would be approximately 185 m3/day of permeate. The maximum back pressure from the product water which this type of module can safely withstand is 120 kN/m2 (14 psi). The Permasep polyamide, hollow fibre module was developed by the DuPont Company and consists of millions of the hair-like fibres bundled together. These are wound around a plastic mesh material approximately 1 mm thick and 450 mm long which in turn is wound around a central porous tube. This continues until the overall diameter of the fibre bundle is 4 in or 8 in, whichever is required. This is then wrapped with a fine cloth to keep the fibres in place. Each end of the fibre bundle is then encapsulated in epoxy resin. One of these ends is machined back to cut off one end of the fibre bundle to leave fibres in a ‘U’ shape and open at the two exposed ends. This tube bundle is then inserted into a fibreglass pressure vessel as shown in Fig. 12. End plates with ‘O’ ring seals are fitted and on the product end a porous support block is also fitted. These are all held in place by circlips. Saline feedwater at pressure is pumped into the central distributor tube where it passes out radially between the fibres pressurising them from the

FIG. 12. Permasep permeator. outside. The module basically operates under laminar flow conditions. Fresh water passes through the fibre wall into the hollow centre where it moves towards the open ends of the fibres. It is collected by the porous block and is removed via a central tube. The brine outlet removes the remaining concentrate. Hence in this case each hollow fibre membrane bundle has its own pressure vessel. The output from an 8 in diameter module operating at 2800 kN/m2 (400 psi), 25°C and 75% conversion, treating a brackish water, would be approximately 68 m3/day. 4.3.3. Pretreatment of Feedwater For reverse osmosis equipment to operate satisfactorily there must of course be an adequate engineering design. Accepting that, the most important operational factor is to keep the membranes in a functional condition and therefore they must be protected from

Developments in water treatment—2

104

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

can be removed by a mixed detergent, formaldehyde solution. It is sound practice to build routine cleaning cycles into the operation of any plant and the type of cleaning may vary depending upon the feedwater quality. 4.3.4. Reverse Osmosis Systems Design The basic individual components in reverse osmosis systems are all relatively simple apart from the sophistication of the associated electronic controls. Ignoring pre- and posttreatment considerations, a simple system consists of a 5–25 µm pre-filter to protect the high pressure pump and module, the high pressure pump itself, the single desalting membrane element in its pressure vessel and the reject flow control valve. In systems of this size design considerations are limited but they become more important with increasing size. The important factors will be briefly discussed for larger systems. The micron filters are usually of the cartridge type where up to fifty cartridge filter elements may be contained in the one pressure shell. On a large plant there may be tens of these pressure shells to provide the capacity for the plant. The cartridge particle size cut-off rating will vary depending upon the type of membrane element being used. For hollow fibre permeators which are more sensitive to suspended material ratings of 5–10 µm are normal, while for the spiral wound 25 µm is common. With satisfactory pretreatment these filter elements have an extended life and hence filter changes are not too frequent. However when they do require changing this operation is somewhat labour intensive. Interest is currently being shown in continuous screen filters.18 These potentially can provide a particle size cut-off down to 5 µm, can be automatically backwashed on pressure drop or time and can pass large volumes per single unit. These continuous filters with coarser screens might also replace the pre-treatment sand filters. Several small units are being evaluated. The high pressure RO feed pumps require sufficient feedwater at a small positive pressure to operate properly. A low pressure cut-out control is normal to guard against any deficiency. The selection of these high pressure pumps also depends upon the size and duty of the plant. Typically for small plants, single stage, stainless steel, high speed centrifugal pumps are used for pressures in the range 2800–4200 kN/m2 (400–600 psi). They can develop up to 6300–7000 kN/m2 (900–1000 psi) if necessary. However the efficiency of these pumps is only 40–50% resulting in a higher power consumption than for more efficient pumps. For larger installations multistage turbine or centrifugal pumps can be used to develop up to 4550 kN/m2 (650 psi) and these have higher efficiencies of up to 75%. Although reciprocating pumps can have an efficiency in excess of 90% they are not particularly popular for brackish water applications as they generally have a higher maintenance cost. However for large sea water applications the reciprocating pump has the advantage of being able to readily develop the required pressures whereas the others cannot and hence it would usually be selected. An accumulator should be used with reciprocating pumps to dampen out pulsations. In choosing a suitable high pressure pump it is important to select one with a performance specification as close as possible to the required duty. Pumps which have a higher pressure and volume specification than required, and are merely throttled back, consume unnecessary energy. This also applies to reciprocating pumps as although they cannot be throttled directly water can be recirculated through the pump. In addition in determining the energy required to drive the

Developments in water treatment—2

108

pump the normal viscosity of the water should be taken into account so as not to unnecessarily overrate the motor. Both brackish and sea water pumps would typically have materials of construction of stainless steel or aluminium bronze. With the increasing need to conserve energy and reduce costs more consideration is currently being given to energy recovery from high pressure brine, particularly on sea water systems operating at 7000 kN/m2 (1000 psi) and only 30% conversion. Some pump manufacturers are proposing an energy recovery turbine coupled directly to a double ended shaft of a motor that drives the high pressure feed pump. In this way the net electrical power input required by the motor to pressurise the water can be reduced by possibly over 50% for sea water plants. An important aspect in designing reverse osmosis systems is to optimise the conversion to product water without creating scaling and concentration polarisation conditions. Adequate pre-treatment allows higher conversions to be obtained without scaling but there is a limit as 100% conversion is not possible. Concentration polarisation is the condition where a much higher salt concentration is created at the membrane surface than in the bulk stream. This significantly increases the osmotic pressure of the system and decreases performance. It is caused by the removal of fresh water through the membrane, increasing the surface salt concentration and reducing the flow to the point where the velocity at the membrane surface is too low to cause adequate mixing between the concentrated layer and the bulk flow. There is then also the possibility of precipitation from the concentrated layer causing scale formation on the membrane surface, even though the bulk flow may be unsaturated. To avoid these conditions on a large plant, the pressure vessels are connected in arrays, with the concentrate from a bank of modules operating in parallel being passed as feed to secondary banks containing fewer modules. Thus the required minimum velocity conditions can be maintained in the membrane elements and up to 90% conversion can be achieved. Popular arrangements are 2, 1 or 8, 4, 2 arrays as shown in Fig. 13.

FIG. 13. RO arrays. RO systems are now more commonly designed to operate at variable pressures in order to maintain a constant product output. However the systems can readily be operated at

Desalination

109

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

Developments in water treatment—2

110

The largest operating RO plant at the present time is that installed by Degremont at Salbukh in Saudi Arabia. This uses DuPont hollow fibre modules to produce 46 Ml/d from a 1500 mg/litre TDS brackish water. This has cooling, softening with silica removal, filtration and acid and polyphosphate dosing as a pre-treatment before the RO units. A marginally smaller plant of 45 Ml/d on a similar water has been installed at Buwayb in Saudi Arabia by Ames Crosta Babcock and is currently being commissioned. This plant uses spiral wound modules from UOP. There are four other RO plants so far in the Riyadh water treatment complex with an average capacity of 32 Ml/d each which are due to be commissioned in 1980.19 Hydranautics has several 15 Ml/d plants and Paterson Candy International Ltd, Envirogenics and other companies have numbers of smaller plants installed on brackish water in the Middle East. A typical RO plant would consist of multiple blocks of pressure vessels with membrane elements to make up the required capacity plus pumps and controls, as shown in Fig. 14. The largest RO plant yet considered has been approved for installation at Yuma in Arizona, USA, to treat 360 Ml/d of a 3000 mg/litre TDS brackish irrigation water before it drains into the Colorado River. This will maintain the quality of the Colorado before it passes to Mexico. Extensive pretreatment and desalination trials have been conducted for several years. Pre-treatment of partial lime softening followed by multi media filtration has now been chosen for the final design. Nine RO and electrodialysis manufacturers conducted desalination trials and from a performance and economic standpoint two RO spiral wound module manufacturers have been selected to supply the desalting equipment. UOP Fluid Systems are to provide 276 Ml/d capacity and Hydranautics 84 Ml/d. The final design of the project is proceeding with the initial operation of the plant scheduled for 1982. At January 1979 prices the total cost of the desalting complex has been estimated to be US $190 million. The total investment cost per m3 of

Desalination

111

FIG. 14. Typical components for a Hydranautics plant to produce 1 Ml/d from brackish water.

FIG. 15. 1·7 Ml/d section of the UOP Fluid Systems’ Jeddah sea water RO plant.

Developments in water treatment—2

112

product per day of installed capacity would be approximately US$0·53/m3.20 Sea water RO technology is more recent and the largest plant installed is for 12·5 Ml/d at Jeddah in Saudi Arabia. This was supplied and installed on a turn key basis by UOP Fluid Systems and uses their TFC spiral wound membrane elements operating at 6100 kN/m2 (870 psi) and 30% recovery. It is expected that product water of 800 to 900 mg/litre will be produced in a single stage for up to two years. However when the specified limit of 1000 mg/litre is exceeded a partial second stage will be used. A section of the plant is shown in Fig. 15. The pretreatment for the plant consists of dual media filters only followed by acid and polyphosphate dosing before the final pre-treatment cartridge filters. A slightly larger plant for 13·0 Ml/d with DuPont B10 hollow fibre sea water modules is being installed in Russia to treat the Caspian Sea and should be operating in 1980. There, a number of much smaller sea water or highly brackish water plants installed with the B10 modules have been successfully operating for several years. A typical plant installed by Paterson Candy International Ltd in Bahrain to produce 2·3 Ml/d of fresh water from a 10000 mg/litre TDS feed is shown in Fig. 16.

FIG. 16. A 2·3 Ml/d plant using DuPont hollow fibre sea water modules, installed by Paterson Candy International Ltd.

Desalination

113

TABLE 6 TYPICAL REVERSE OSMOSIS COSTS Q3 1979 FOR A 90% LOAD FACTOR Brackish water Sea water 1500 mg/litre 35000 mg/litre TDS feed 50 TDS feed 500 mg/litre TDS mg/litre TDS prod. 2800 kN/m2 prod. 6020 kN/m2 (400 psi) 25°C, (860 psi) 25°C, 75% conversion 30% conversion 4 Ml/d 20 Ml/d 4 Ml/d 20 Ml/d Capital costs (£ 0·56 2·4 1·73 8.28 million) £mpa £mpa £mpa £mpa Operating costs p/m3 p/m3 p/m3 p/m3 (£ million p.a 3 and p/m ) 1. Fixed charges 0·065 5·0 0·28 4·3 0·20 15·5 0·97 14·8 11·75% (10% interest, 20 yr plant life) 2. Power(at 3·2 p/kWh) (a) Brackish 0·076 5·8 0·38 5·8 (1·8 kWh/m3) 0·34 25·6 1·70 25·6 (b) Sea (8·0 kWh/m3) 3. Chemicals Acid Polyphosphate Chlorine Detergent Formaldehyde (a) Brackish 0·033 2·5 0·17 2·5 (b) Sea 0·041 3·1 0·20 3·1 4. Membrane replacement 3-year life (a) Brackish 0·053 4·0 0·27 4·0 (b) Sea 0·32 24·4 1·53 24·4 5. Maintenance, 0·035 2·6 0·14 2·2 0·035 2·6 0·14 2·2 operation, staff and materials Total annual 0·26 – 1·24 – 0·94 – 4·54 – cost (£ million) Total water cost – 19·9 – 18·1 – 71·2 – 70·1 (p/m3)

Developments in water treatment—2

114

A significant growth in the application of sea water RO can be expected in the future. In time it is probable that, owing to the apparently more favourable economics, this technique will gradually replace distillation for many applications. 4.3.6. Costs of Reverse Osmosis Typical estimated costs for installed reverse osmosis plant alone can be presented for both brackish and sea water systems at third quarter (23) 1979 for small and intermediate plant sizes (Table 6). Different bases and assumptions can be used for estimating if desired. These RO costs must be used with caution since they only constitute the desalination element of the total water supply cost. If the costs of wells or an intake, pre-treatment works, civil and building works, product post-treatment and storage, and waste disposal are included, the total capital cost can be more than double the figures quoted. Similarly if power has to be generated at site or product water has to be pumped to supply, significant additional costs will be incurred. The costs quoted in Table 6 should not be compared directly with the distillation costs quoted earlier since the bases employed are not identical. This situation arises because different cost elements need to be included or excluded depending upon the site and the process being considered. For example, distillation plant will largely not need to be housed in a building while housing will be essential for reverse osmosis. Hence it must be decided what cost elements need to be considered and included for overall cost estimates. For a particular application for sea water desalination, it will usually be found that the estimated total costs are similar whether desalting is performed by distillation or reverse osmosis, with the RO costs usually being the lower, particularly for small plant.

4.4. ELECTRODIALYSIS Electrodialysis (ED) is a membrane process which has been used commercially since the mid 1950s for the removal of inorganic salts from water, i.e. as a desalination process. It has also been used in Japan as a brine concentration process. In water, salts dissolve to form positively charged cations and negatively charged anions. If a d.c. electric field is applied across the solution, cations migrate towards the negatively charged cathode and anions towards the positively charged anode. With permselective ion exchange membranes, one cation and one anion placed between the electrodes, the permselective property ideally allows the cation membrane to only pass cations and the anion membrane to only pass anions. If now a stack of membranes, alternately cation and anion, is placed between the electrodes, the solution between one pair of membranes becomes depleted in ions while the solutions on either side become enriched, as shown in Fig. 17. The basic cell pair can be seen to consist of a feed or diluate stream between a cation and anion membrane plus a concentrate on the other side of each membrane. In a commercial configuration there can be several hundred cell pairs per stack, clamped as in a filter press, between the two electrodes. The diluate and concentrate streams are pumped in co-current flow through their respective cells. The cell thicknesses are

Desalination

115

approximately 1 mm and a plastic mesh or spacer separates the membranes as well as acting as a turbulence promoter within the flow channels. The ion exchange membranes are cast as flat sheets typically 0·6 mm thick around a woven fabric to provide mechanical strength. The membrane electrical properties of high conductance and ion selectivity control the process through the application of Ohm’s Law for stack resistance and Faraday’s Law for salt passage. Thus desalting energy requirements are proportional to the quantity of salts removed. There are two types of electrodialysis systems commercially available. The first that of Ionics of the USA who have probably supplied over 75% of the world’s installed electrodialysis plants. They manufacture their own membranes and have a tortuous path spacer design where the liquid follows a continuous compressed ‘S’ shape across the cell. Mixing is accomplished by ‘bridges’ across the flow path. The fluid velocity along this path is approximately 0·5 m/s. For the past 3–4 years they have almost exclusively offered plant with polarity reversal as the means of controlling scale formation. With this facility, chemical dosing is not necessary, although simple chemical cleaning every two or three weeks may be required. For normal brackish water treatment the polarity reversal occurs automatically three or four times an hour and by means of motorised valves reverses the product water, concentrate and electrode streams. The reversal purges membrane and electrode surfaces of scaling and fouling materials which may have deposited during the cycle. With this mode of operation plant is normally operated with approximately 80% recovery. Owing to the economy of the process being directly linked to the amount of salt removed the upper economic limit for treating brackish waters is 7000 mg/litre TDS.

FIG. 17. Principles of electrodialysis.

Developments in water treatment—2

116

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 (mg/litre) Na Ca Mg Cl HCO3 (as CaCO3) SO4 F TDS pH Conductivity (µS/cm, 25°C)

170 508 188 340 211 1550 2.4 2969 7.25 3540

Product (mg/litre) 54 61 13 53 81 170 0.95 433 7.05 650

Developments in water treatment—2

118

FIG. 18. An Ionics 0·5 Ml/d polarity reversal, brackish water, ED plant (by courtesy of Ionics Inc.). water since it has been uneconomic, However recent development work in the USA with high temperature electrodialysis23 and in Japan with large plant24 at ambient temperatures has led to cost estimates for large-scale desalination of sea water which appear to be similar to those for sea water reverse osmosis. While small batch or development plants have been installed for sea water, the process could not be considered to be commercial at this time. However the Japanese developments originated from concentration of brine by electrodialysis which is now a standard commercial technique. Hence commercial sea water desalination plants can be expected in the near future. 4.4.1. Costs of Electrodialysis Typical estimated costs for installed brackish water electrodialysis are presented in Table 8 for Q3 1979. As for reverse osmosis, they only represent the desalination element of the total water supply costs and hence must be used with caution. The additional costs for associated plant site works and buildings, dependent upon the feedwater quality and the site, must be added to obtain total water supply capital costs.

Desalination

119

TABLE 8 TYPICAL ELECTRODIALYSIS COSTS Q3 1979 FOR A 90% LOAD FACTOR (1800 mg/litre feedwater; 200 mg/litre TDS product; 21°C; 78% recovery) 4 Ml/d Capital costs (£ million) Operating costs (£ million p.a and p/m3) 1. Fixed charges 11·75% 2. Power (at 3·2 p/kWh) Desalting 0·7 kWh/m3 per 1000 mg/litre TDS removed, i.e. 1·12 kWh/m3 Pumping 0·7 kWh/m3 Total 1.82 kWh/m3 3. Chemicals Provided polarity reversal used 4. Membrane replacement 7-year life 5. Maintenance, operation, staff and materials Total annual cost (£ million) Total water cost (p/m3)

20 Ml/d

0·7 2·8 £mpa p/m3 £mpa p/m3 0·082 6·3 0·33 0·076 5·8 0·38

5·0 5·8

0·005 0·4 0·026 0·4 0·050 3·8 0·25

3·8

0·046 3·5 0·20

3·1

0·26

1·12 19·8

18·1

These ED costs are very close to those for a similar RO feedwater with in general the capital costs being higher than for RO. However the higher capital costs are partly offset by a longer membrane life and a reduced cost per m3 for membrane replacement. If a brackish feedwater of 4000–5000 mg/litre TDS was to be treated by ED then the capital costs would increase as extra membrane area would be required. In addition the power requirements would more than double. The costs for RO however on this feedwater would not greatly change. Hence for low TDS waters RO and ED are competitive, but as the feed salinity rises RO can become the more attractive process. Few ED plants have been installed with capacities greater than 16 Ml/d while significantly larger brackish water RO plants have been installed. Hence there is also more confidence in RO for large-scale applications. Since sea water electrodialysis is not yet commercial, cost estimates have not been presented for this.

4.5. OTHER DESALINATION PROCESSES A number of other desalination processes have received attention but none has been developed beyond the pilot-scale or very small full-scale capacity. Solar energy has been employed to distil sea water. Although plastics have been tried as still covers the limited

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 1. MOORHEAD, A., The Blue Nile, 1967, Dell, New York. 2. BURLEY, M.J. and MAWER, P.A., Water Res. Assoc. TP50, 1967, Water Research Centre, Marlow, Bucks. 3. VEENMAN, A.W. and BROEKENS, J.D., Proc. 6th Int. Symposium on Fresh Water from the Sea, Las Palmas, 1978, 2, 53.

Desalination

121

4. KLAREN, D.G. and WINDT, J., Proc. 6th Int. Symposium on Fresh Water from the Sea, Las Palmas, 1978, 2, 15. 5. BURLEY, M.J., Proc 2nd Int. Symposium on Fresh Water from the Sea. 2, paper 16, 10 pp, Athens, May 1967. 6. WADE, N.M., Proc. 6th Int. Symposium on Fresh Water from the Sea, Las Palmas, 1978, 1, 327. 7. TODD, B., Private communication, International Nickel Co.., Birmingham, UK. 8. SATO, S., Bull. Japan. Inst. Metals, 1978, 575. 9. NEWTON, E.H., BIRKETT, J.D. and KETTERINGHAM, J.M., Report to U.S. Office of Saline Water, Contract No. 14/30/2721, March 1972. 10. WADE, N.M., Desalination. 1979, 31, (2), 309. 11. U.S. Dept. of the Interior, Office of WaterResearch and Technology, Desalting Plants Inventory No. 6, October 1977. 12. DESAI, A.M., Desalination. 1977, 23, 367. 13. RILEY, R.L., Fox, R.L., LYONS, C.R., MILSTEAD, C.E., SEROY, M.W. and TAGAMI, M., First Desalination Congress of The American Continent 1, 11–1, 1976, Elsevier Scientific Publishing Co., Amsterdam. 14. AL-GHOLAIKAH, A., EL-RAMLY, N., JAMJOON, I. and SEATON, R., National Water Supply Improvement Association J., Jan. 1979, NWSIA, USA. 15. CARMONA, J. and DE BUSSY, R.P., Presented at the Comision Federal de Electricidad, Mexico City, June, 1976. 16. GROVER, J.R., GAYLER, R. and DELUE, M.H., Proc. 4th Int. Symposium on Fresh Water from the Sea, Heidelberg, 1973, 4, 349. 17. DuPont Company, Plastic Products and Resins Dept., Permasep Products, Tech. Bull. 491, 1977. 18. The Plenty Group, Technical Bulletin on Self Cleaning Filters, Newbury, UK, 1978. 19. GHULAIGAH, H.E.A. and ERRICSSON, B., Proc. Int. Congress on Desalination and Water Re-Use, Nice, 1979, 1, 3–301. 20. LOPEZ, M., Proc. Int. Congress on Desalination and Water Re-Use, Nice, 1979, 1, 1–15. 21. KISHI, M., SERIZAURA, W. and NAKARO, W., Desalination, 1977, 23, 203. 22. KALZ, W.E., Ionics Inc., Bulletin TP. 307, 1977. 23. McRAE, W.A., PARSI, E.J., GANZI, G., JHA, A. and O’DONOGHUE, K., Proc. 6th Int. Symposium on Fresh Water from the Sea, Las Palmas, 1978, 3, 101. 24. KAWAHARA, T., ASAKA, T. and SUZAKI, K., Proc. 6th Int. Symposium on Fresh Water from the Sea, Las Palmas, 1978, 3, 95.

Chapter 5 DISINFECTION A.T.PALIN, O.B.E., B.Sc., Ph.D., F.R.I.C., F.I.W.E.S. Consulting Chemist, Newcastle upon Tyne, UK SUMMARY The principal disinfecting agents used in water treatment are chlorine, chloramine, chlorine dioxide and ozone. After a preliminary discussion of their relative importance, these disinfectants are dealt with in turn with reference to such aspects as historical development, chemical and bacteriological properties and practical application. While chlorination remains the most widely used process at the present time, studies of recent years have revealed problems associated with the possible formation in some waters of traces of harmful carcinogenic by-products such as trihalomethanes. Consideration is given to the influence of these findings on future developments. Finally, an outline of the DPD analytical tests for treatment control is presented of which a special advantage lies in their ability to determine mixtures of disinfectant residuals in the treated water.

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

Disinfection

123

trihalomethanes. It is this aspect of disinfection to which most research is directed at the present time. Other disinfectants such as ultraviolet radiation, bromine and iodine are not regarded as practical for drinking water treatment except possibly on a very small scale. U.v. does not provide an active disinfectant residual in the water, bromine is difficult to handle and is expensive, and iodine, while satisfactory for emergency purposes, can impart taste, odour and colour to the water and also is very expensive. It may be mentioned that for emergency use on a small scale boiling the water is completely effective. In considering developments in disinfection the present discussion is concerned only with those agents having any significant large-scale application in the field of water treatment, that is to say, chlorine and the related chloramines, chlorine dioxide and ozone. The survey will cover their chemistry, mode of application, bactericidal efficiency, and the necessary treatment control to ensure full compliance with chemical and bacteriological standards of drinking water quality. Reference will also be made to such modifications to treatment as might be required to minimise risks of harmful by-products being produced in the finished water. In the case of turbid waters the presence of suspended particulate matter may interfere with disinfection by affording protection to embedded bacteria from the action of the disinfectants. Clarification by filtration is therefore essential usually preceded by coagulation and sedimentation. Such processes provide the further advantage of physically removing a substantial proportion of the micro-organisms thus reducing the pollution load at the disinfection stage and the general chlorine demand of the water.

5.2. CHLORINE 5.2.1. History In the earliest applications of chlorine for water disinfection the chemical was applied in the form of hypochlorite. From about 1915 increasing use was made of chlorine gas contained under pressure as liquid in cylinders. Its excellent bactericidal properties became firmly established and additional benefits, resulting from its oxidising power, became evident such as removal of iron and manganese, and of tastes and odours. On the other hand it was found that chlorination itself could impart to some waters undesirable tastes variously described as medicinal, iodoform or chlorophenol. These were due to reactions of the chlorine with trace amounts of phenols and similar compounds. Furthermore the use of too much chlorine could itself give a chlorinous taste because of the presence of excessive amounts of ‘residual chlorine’, that is the amount remaining in the water after treatment. From the public health point of view the benefits derived from the use of chlorine as a disinfectant were unquestionable so that further investigation up to about 1935 was directed more towards improvements and modifications of treatment aimed at overcoming any possibly undesirable side-effects on general water quality. During this period attention was paid to the use of chloramines, these being compounds formed in the water by reaction between chlorine and ammonia. Ammonia is present in many raw water sources, usually as a consequence of pollution, or it can be added if required in the form

Developments in water treatment—2

124

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

Disinfection

125

5.2.2 Forms of Chlorine In any comparison of strengths of chlorinating chemicals in terms of chlorine yield it is standard practice to refer to their ‘available chlorine’ content. Similarly the concentration of residual chlorine in water is always expressed in terms of available chlorine no matter in what chemical form it might be present. While regarded by some as a misnomer, this remains the standard form of expression for the strengths or capacities of the various forms of chlorine as well as for the doses in which they are applied and for the residues that remain in the water. Any chlorine that eventually appears as chloride, the final reduction product, represents a complete loss of available chlorine. The expression ‘available chlorine’ is defined in precise chemical terms and relates to capacity rather than oxidising or disinfecting power. It is based upon the determination of the equivalent amount of iodine liberated from iodide in acid solution by a known amount of the chlorine compound. Since the available chlorine content of pure chlorine is 100% the corresponding figures for other compounds may be obtained as a percentage by comparing the amounts of iodine thus liberated with the amount liberated from the same weight of chlorine. To illustrate the application of the available chlorine concept the following four cases may be considered using potassium iodide KI and hydrochloric acid HCl. (In practice another acid, say acetic acid, would be used in the determinations since HCl itself may contain traces of chlorine. For the present purpose it enables the reactions to be presented more simply.)

By calculation from molecular weights it may be demonstrated that the amounts of iodine liberated from 1 g of each of the above chemicals are 3·6 g, 4·84 g, 4·93 g and 9·41 g respectively. In terms of percentages, taking chlorine as 100%, the available chlorine contents become HOCl—134%, NH2Cl—137% and ClO2—261%. As previously noted it would be incorrect to equate these figures to oxidising power and bactericidal efficiency. Monochloramine is far less reactive than hypochlorous acid. Similarly chlorine dioxide although first introduced as being two and a half times more powerful than chlorine is in fact considerably weaker in some reactions. A knowledge of the available chlorine content is of value in assessing the comparative costs of chlorine from different commercial products and in checking for any gradual loss on storage (Table 1).

Developments in water treatment—2

126

TABLE 1 AVAILABLE CHLORINE CONTENT OF DIFFERENT FORMS OF CHLORINE AND THE QUANTITIES REQUIRED TO GIVE 1 lb OF AVAILABLE CHLORINE Form of chlorine Chlorine gas High Test Hypochlorite (calcium hypochlorite) Bleaching Powder (chloride of lime) Sodium hypochlorite Sodium hypochlorite Sodium hypochlorite

Available chlorine

Equivalent quantities

100% 65–70%

1 lb 1·5 lb

30–35%

3·0 lb

15% 10% 5%

0·5 gal 0·8 gal 1·6 gal

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

Secondly, the hypochlorous acid partly dissociates to give hydrogen ions and hypochlorite ions thus:

Disinfection

127

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

Developments in water treatment—2

128

FIG. 1. Effect of pH value on form of free available chlorine in water. The lower the pH of the water and the higher the ratio of chlorine to ammonia the greater is the tendency to produce the more highly chlorinated derivatives. Within the normal pH range of water, however, the product consists almost entirely of monochloramine and the reaction is complete within one minute. For conversion of all the ammonia to monochloramine about five times as much chlorine is required, that is the combining weight ratio is 5 of Cl2 to 1 of NH3 calculated in terms of nitrogen. If more chlorine is applied than is required for this rapid initial reaction, continuing oxidation reactions occur at a slower rate eventually producing mainly nitrogen thus: 2NH2Cl+HOCl=N2+3HCl+H2O Small amounts of dichloroamine and nitrogen trichloride may also appear during the course of the breakpoint reactions especially in the zone corresponding to chlorine doses

Disinfection

129

above the 5:1 ratio, but the net result as the chlorine is further increased and adequate contact period is allowed corresponds to the following overall equation: 2NH3+3Cl2=N2+6HCl According to this equation the amount of chlorine required to oxidise one part by weight of ammonia-nitrogen is 7·6 parts by weight. While nitrogen is the main end-product small amounts of nitrate and possibly nitrogen trichloride may also remain with the result that the observed ratio becomes somewhat higher at about 8·3:1. In natural waters there may be additional chlorine absorption by other types of impurity giving a ratio about 10:1. For polluted waters the ratio could be much higher. The chlorine dose-residual curve may be plotted for a given sample of water by treating a series of aliquots with increasing doses and determining the residual chlorine values after a predetermined period of contact. If the water contains only a small amount of ammonia the curve obtained will be similar to that of Fig. 2. On the other hand, where appreciable amounts of ammonia are present the distinctive breakpoint curve will be obtained as in Fig. 3.

FIG. 2. Breakpoint curve for water with ammonia-nitrogen 0·06 mg/litre.

Developments in water treatment—2

130

FIG. 3. Breakpoint curve for water with ammonia-nitrogen 5·2 mg/litre.

FIG. 4. Chlorine dose-residual curve at pH 6·0 after 1 day. Initial ammonia 0·5 mg/litre (N).

Disinfection

131

FIG. 5. Chlorine dose-residual curve at pH 7·0 after 1 day. Initial ammonia 0·5 mg/litre (N).

FIG. 6. Chlorine dose-residual curve at pH 8·0 after 1 day. Initial ammonia 0·5 mg/litre (N). The effect of pH on the composition of the residual chlorine at different stages of the breakpoint curve is shown in Figs. 4, 5 and 6. The ammonia was added as ammonium chloride, the water used being otherwise of zero chlorine-demand. In these systems,

Developments in water treatment—2

132

therefore, the reactions involved only chlorine and ammonia. The most important feature of the curves is that before the breakpoint the residual chlorine is present in the form of chloramines, that is combined chlorine. After the breakpoint it is present mostly as free chlorine. Recognition of these two forms is of vital importance in the control of water chlorination since the bactericidal and virucidal properties of free chlorine are vastly superior to those of combined chlorine. Therefore for maximum safety in the disinfection of water it is generally desirable to chlorinate beyond the breakpoint, that is to the point of establishing free chlorine. A disadvantage of breakpoint chlorination lies in the possibility of producing traces of nitrogen trichloride unless the pH is fairly high. This may impart to the water an objectionable chlorinous-type odour for the eradication of which additional treatment may be necessary, possibly involving complete dechlorination with a final stage of ammonia-chlorine treatment.8 5.2.4. Modern Chlorination Practice From the foregoing review of the chemistry of chlorination it is clear that the following definitions are fundamental to modern practice. Free available chlorine is residual chlorine in the form of hypochlorous acid and hypochlorite ion. Combined available chlorine is residual chlorine existing in combination with ammonia or organic nitrogen compounds. The chlorine demand of a water is the difference between the applied dose and the residual chlorine. It is necessary to specify the conditions, that is whether the residual produced is in the free or combined form, and the time of contact. The chlorine dose also should be specified. In those cases where the chlorine demand is due mainly to nonnitrogenous matter or oxidisable inorganic impurities the chlorine dose-residual curve may not, as already noted, exhibit a breakpoint. Nevertheless after the demand has been satisfied free residual chlorine will appear and continue to increase with increasing dose. The modern approach to the classification of chlorination methods is based upon this distinction between free and combined chlorine. The two main types of process therefore are free residual chlorination and combined residual chlorination. Variations may be introduced to meet particular requirements but the results obtained under any given conditions will always depend upon the nature and amount of the residual chlorine produced in the water. The production of residuals of known composition by suitable control of chlorination has been made possible by the methods of chlorine residual differentiation now available. Where dechlorination is required it is usual to apply sulphur dioxide which reacts rapidly to destroy the residual chlorine in an approximate 1:1 ratio. Sulphur dioxide is a gas under normal pressure but may be compressed to liquid for supply in cylinders and other containers. Dosing equipment is similar to that used for chlorine. Both free and combined chlorine are destroyed thus: SO2+H2O=H2SO3 (sulphurous acid) H2SO3+HOCl=H2SO4+HCl H2SO3+NH2Cl+H2O=NH4HSO4+HCl

Disinfection

133

Passage of the water through granular activated carbon filters provides another means of dechlorination, the principal reaction being as follows: C+2Cl2=2H2O=4HCl+CO2 For small-scale dechlorination use can be made of such chemicals as sodium thiosulphate (‘hypo’), sodium sulphite and sodium bisulphite. The bactericidal and virucidal power of free chlorine may be influenced by other factors of which one of the most important is pH. The pH of the water governs the relative proportions of hypochlorous acid and hypochlorite ion in the free chlorine residual. With pH rising above 6·0 the proportion as HOCl declines from virtually 100% down to almost zero at pH 9.0. Since the bactericidal activity of HOCl is something like 80 times more powerful than that of the OCl− ion it is evident that in free residual chlorination the higher the pH the less active is the residual. At pH values of 7·0, 7·5 and 8·0 the HOCl proportions are very approximately 75%, 50% and 25%. If in addition to the free chlorine reading the pH is known it is possible to calculate the residual HOCl quite simply from the established dissociation factors. Where treatment is concerned it may be necessary in practice to work to higher free chlorine residuals at the higher pH values. There is evidence that the killing power of combined chlorine diminishes also with rising pH. The bactericidal power of both free chlorine and combined chlorine decreases as the water temperature falls. Thus pH and temperature,

TABLE 2 MINIMUM CYSTICIDAL AND BACTERICIDAL RESIDUALS (AFTER 30MINUTE CONTACT) pH

6·0 7·0 8·0 9·0

Free chlorine

Combined chlorine Bactericidal Cysticidal Cysticidal Bactericidal 0–25°C 22–25°C 2–5°C 0–25°C 0·2 0·2 0·2 0·6

2·0 2·5 5·0 20·0

7·5 10·0 20·0 70·0

2·0 2·5 3·0 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. CHLORINE DIOXIDE 5.3.1. History Chlorine dioxide was first applied as a water treatment chemical in 1944 at a plant in Niagara Falls, USA. It was found to be effective in controlling unpleasant tastes and odours especially those of phenolic origin. It was subsequently investigated in the UK by the author12 with particular reference to its decolorising effect on peaty moorland waters. During this work its general chemical properties were examined and this led to the discovery of its inertness to ammonia. Because of this chlorine dioxide does not give a breakpoint type of dose-residual curve. In addition to its use in the control of tastes and odours it now finds application in the oxidation of iron and manganese and some forms of organic impurity. While originally its effectiveness as a disinfectant was open to some doubt, and thus it was always used with an excess of chlorine, it is now fully accepted as a disinfectant in its own right. It can provide a longer-lasting residual than chlorine which is advantageous where it is desired to maintain residual disinfectant throughout the water distribution system. Although its use has expanded considerably, it still remains relatively small compared with that of chlorine. This position may change in view of current researches into the trihalomethane problem. 5.3.2. Generation Chlorine dioxide is produced on site by mixing strong chlorine solution, as delivered from the normal type of chlorinator, with a solution of sodium chlorite when the following reaction occurs: Cl2+2NaClO2=2NaCl+2ClO2 If the pH of the mixture is not low enough the reaction may not go to completion thus leaving unreacted chlorine and chlorite in the discharge to the water. A system has been devised, known as the CIFEC method, whereby the chlorine water is recirculated around an enrichment loop to ensure almost 100% production of chlorine dioxide after admixture with the sodium chlorite solution.13 For smaller installations other methods may be used in which the chlorine dioxide is generated under controlled conditions from mixtures of sodium chlorite and either hydrochloric or sulphuric acid or from similar mixtures with the addition of hypochlorite.

Disinfection

135

5.3.3. Chemical and Bactericidal Properties As previously noted chlorine dioxide, unlike chlorine, does not react with ammonia to form chloramines. In waters of high ammonia content this lack of reactivity is beneficial compared with chlorine which, unless the dose is high enough, becomes converted to the slow-acting chloramines. In water treatment only a part of the full oxidising capacity of chlorine dioxide is used because in most of its reactions it undergoes only a first-stage reduction corresponding to the change from chlorine dioxide to chlorite. This is equivalent to no more than one-fifth of its total available chlorine as expressed normally. Thus in the majority of reactions chlorine is a better oxidant than chlorine dioxide except in special circumstances, such as the presence of phenols, when the advantage lies with the dioxide. As a general rule chlorine dioxide is as effective a disinfectant as chlorine and is less sensitive to changes in the pH of the water. Thus with pH rising over 7 it is better able to maintain its disinfecting power unlike chlorine which gradually becomes less effective because of the shift from hypochlorous acid to the much less active hypochlorite ion.

5.4. OZONE 5.4.1. History The earliest recorded use of ozone as a water disinfectant occurred in France in 1886. As a result of further pilot studies a number of treatment plants adopted ozonation in the early years of the present century including, for example, that at Nice where the process has operated continuously from 1906. Thus ozonation has a long history of use but never on so extensive a scale as that of chlorine. Apart from disinfection it may be applied for other purposes such as taste and odour control, iron and manganese oxidation and organics removal. The strong disinfecting and oxidising action of ozone is due to the release of the very reactive third atom of oxygen with reversion to oxygen O2 thus: 2O3=3O2 There are today possibly more than one thousand plants using ozone. 5.4.2. Generation and Application Ozone is generated on site from oxygen or, as is more usual in large-scale commercial production, from the air. In the generation process the air, which must be clean and dry, is passed between electrodes separated by an air gap across which a very high voltage is maintained to produce a silent electric discharge. The amount of ozone produced may be up to 30 g/m3 of air. This relatively low concentration coupled with its low solubility requires very intimate mixing and dispersion of the ozonised air throughout the water in order to achieve the highest possible degree of transfer from the gas to the liquid phase. For this purpose the methods used may take the form of injectors, porous diffusers or mechanical mixers. For maximum efficiency the aim must be to produce as small a bubble size as possible. As

Developments in water treatment—2

136

the dispersed bubbles gradually grow in size so the efficiency of transfer falls away with the decreasing interfacial area of contact. In calculating applied doses it is assumed that all the ozone is transferred from the air to the water although in practice the proportion of ozone dissolved may range from 60% to possibly better than 90%. Recommended dosage rates for drinking water are generally within the range 0·2 to l·5 mg/litre. The ozone disappears very rapidly from the water and it is therefore not possible to maintain an active residual as is the case with chlorination. The absence of any lasting residual disinfecting action is a disadvantage of the process and for this reason a supplementary stage of chlorination is usually applied. Satisfactory disinfection by ozonation is assured normally by 0·4 mg/litre or more of residual ozone acting for a minimum contact period of 4 to 5 min. In practice the presence of ozone residuals of about 0·1 mg/litre at the outlet of the ozonation chamber is generally effective.

5.5. RELATIVE COSTS Unit costs for five disinfection processes have been developed by the U.S. Environmental Protection Agency.14 The basis of these estimates includes capital, operational and maintenance costs. In the case of chlorine, chlorine dioxide and ozone it is assumed that the specific doses have equivalent disinfecting capability. This is not the case where chloramine is concerned and it may be necessary to allow further for additional disinfection unless high quality waters are being treated. While the results varied with design capacity the general indication was that chlorine dioxide, chloramine and ozone cost about 65%, 20% and 100%, respectively, more than chlorination.

5.6. DISINFECTION BY-PRODUCTS The problems associated with the formation of traces of chloroform and other trihalomethanes as possible by-products of chlorination have led to a review of the entire subject of drinking water disinfection. Such traces have been detected in polluted waters after chlorination and also where chlorine has been applied with long contact period for colour removal. As well as chemical pollutants the naturally occurring constituents of peaty water, such as humic acid and similar organic matter, can act as precursors of the trihalomethanes. The carcinogenicity of chloroform has now been acknowledged by the U.S. Environmental Protection Agency who are urging that modified treatment practices be adopted to minimise chloroform and similar trihalomethane formation from the organic materials found in natural and polluted waters. The slow-acting chloramines do not form these compounds. While there is at present no acceptable substitute for chlorine for general use in water disinfection there is no doubt that increased consideration will be given to ozone, chlorine dioxide and even chloramine where conditions are suitable for its application. In addition, mixed treatments such as, for example, chlorine plus chlorine dioxide and ozone plus chlorine are likely to receive more attention. Selection of points of application will

Disinfection

137

require further thought. Shortening of pre-chlorination contact periods or applying the chlorine at a later stage of treatment, for instance after coagulation and sedimentation, can result in substantially reduced chloroform production. There are two general approaches to the problem from a treatment point of view. First, to reduce, by improved or modified pre-treatment prior to chlorination, the level of organic pollutants acting as precursors of the trihalomethanes. Secondly, to use other disinfectants in conjunction with chlorine so as to enable chlorine doses and contact times to be reduced. A favoured approach in attempting to achieve the first objective is adsorption by passage through granular activated carbon (GAC) filters. Improved results are obtained by pre-ozonation which assists by breaking down those organic materials not readily adsorbed into more readily adsorbable and biodegradable compounds. It is now recognised that the carbon becomes biologically active. This process in which the use of ozone is combined with biological activated carbon (BAC) is receiving much attention with very promising results. BAC is an effective process for ammonia removal. Chlorine dioxide treatment appears not to produce trihalomethanes when prepared free of chlorine. As normally produced for water treatment by reacting sodium chlorite with chlorine, some excess of chlorine remains. Results indicate that formation of trihalomethanes can then occur with these mixed residuals but at lower concentration than would be produced from an equivalent amount of chlorine alone. Trace amounts of chlorite may appear as a by-product of chlorine dioxide treatment either because of

TABLE 3 OUTLINE OF DPD TESTS FOR FREE CHLORINE AND OTHER DISINFECTANT RESIDUALS Test

Reagents (in order of use)

1. Free chlorine 2. Free chlorine Combined chlorine 3. Total available chlorine

1 1 3

4. Chlorine dioxide

G 1

5. Chlorine dioxide plus free chlorine. Combined

1 3

4

Notes

Reagents 1 and 3 added together may be used in place of reagent 4. Result gives onefifth chlorine dioxide. Multiply by 5. This result is in terms of available chlorine. To express as ClO2, multiply instead by 1·9. To obtain free chlorine from reagent 1 reading, deduct one-fifth

Developments in water treatment—2

chlorine 6. Chlorite (continuation of test 5)

A N

7. Ozone

4

8. Total chlorine plus ozone Total chlorine only 9. Free chlorine plus ozone

4 G 4

Total chlorine plus ozone

3

1

138

chlorine dioxide as given in test 4. Deduct four-fifths chlorine dioxide in terms of available chlorine as obtained from test 4 to obtain chlorite. Reagents 1 and 3 added together may be used in place of reagent 1. Provides for separate determination. Ozone response is incomplete but difference gives combined chlorine. Combine with test 8 for complete differentiation.

Key to Reagents 1—DPD indicator+buffer 2—for differentiation of combined chlorine (not included here) 3—potassium iodide 4—1 and 3 combined G—glycine A—acidifying agent N—neutralising agent

incomplete conversion or because of reversion in the treated water, bearing in mind that chlorite is the first reduction product of chlorine dioxide. The possible toxicity of chlorite has not been ruled out and preliminary indications so far suggest that it is about the same as that of nitrite. In the control of these modified disinfection processes involving mixed treatment by chlorine, chlorine dioxide and ozone a suitable residual control test capable of their separate determination is essential. Only the DPD test,15,16 adopted as standard in many countries for free chlorine and combined chlorine compounds, can meet these further analytical requirements. An outline of the procedures is given in Table 3. Although monochloramine, dichloramine and nitrogen trichloride may be determined separately, if required, by using quite simple modifications of the method they are here included together as combined chlorine. The preparation of the various reagents is given in standard manuals of water analysis and in the author’s published papers.17,18 Standardised tablets and powder reagents are available commercially.19 These are used with colour comparators in test kits as shown in Figs. 7 and 8.

Disinfection

139

FIG. 7. Tintometer Lovibond test kit for residual chlorine, pH value and alkalinity.

Developments in water treatment—2

140

FIG. 8. LaMotte test kit for residual chlorine and pH value. REFERENCES 1. FABER, H.A., J. Amer. Water Works Assoc., 1939, 31, 1539. 2. GRIFFIN, A.E., J. Amer. Water Works Assoc., 1939, 31, 2121. 3. RUYS, J.D., Drinkwater-reiniging met Hypochlorieten, 1914. 4. HOLWERDA, K., Mededeelingen van den dienst der Volksgezondheid in Nederlandsch-Indie, 1930, 19, 325. 5. GERSTEIN, H.H., J. Amer. Water Works Assoc., 1931, 23, 1334. 6. PALIN, A.T., Ph.D. Thesis, University of London, England, 1949. 7. PALIN, A.T., J. Inst. Water Engrs, 1949, 3, 100. 8. WILLIAMS, D.B., J. Amer. Water Works Assoc., 1949, 41, 441. 9. BUTTERFIELD, C.T. and WATTIE, E., Public Health Reports, 1943, 58, 1837; 1946, 61, 157. 10. BUTTERFIELD, C.T., Public Health Reports, 1948, 63, 934. 11. SNOW, W.B., J. Amer. Water Works Assoc., 1956, 48, 1510. 12. PALIN, A.T., J. Inst. Water Engrs, 1948, 2, 61. 13. CIFEC (Cie Industrielle de Filtration et d’Equipment Chimique S.A.), Chlorine Dioxide Generator: The French Method with Enrichment Loop, Paris, Nov., 1976, Notice No. 167.

Disinfection

141

14. US Environmental Protection Agency, Ozone, Chlorine Dioxide and Chloramines as Alternatives to Chlorine for Disinfection of Drinking Water, Revised April 1978, Cincinnati, USA. 15. UK Department of the Environment, Analysis of Raw, Potable and Waste Waters, 1972, London, England. 16. American Public Health Association, American Water Works Association and Water Pollution Control Federation, Standard Methods for the Examination of Water and Wastewater, 14th edn, 1975, Washington, DC, USA. 17. PALIN, A.T., J. Amer. Water Works Assoc., 1957, 49, 873. 18. PALIN, A.T., J. Inst. Water Engrs, 1974, 28, 139. 19. Wilkinson and Simpson, Ltd, Tablet Tests and Reagents for Water Analysis and Treatment Control, 2nd Edn. 1980 (in preparation), Gateshead, England.

Chapter 6 SLUDGE TREATMENT AND DISPOSAL M.A.HILSON, C.Chem., F.R.I.C., F.I.W.E.S. Principal Scientist, Water Treatment and Supply, North West Water Authority, Warrington, UK SUMMARY The various types of sludge arising from water treatment processes are considered and this is followed by a consideration of disposal methods. The methods of water treatment plant operation aimed at producing the minimum volume of sludge for subsequent treatment are then discussed. The dewatering of sludge is considered as two separate processes: initial thickening or increase in concentration to minimise the size of treatment equipment required followed by a review of the various methods available for further dewatering. Recently introduced water treatment processes are then considered with particular reference to their impact on sludge treatment. Quantities of sludge arising and operational procedures for sludge treatment installations are then dealt with. Much of the material in this chapter is drawn from the experience of the author and his close colleagues and as such is perhaps inclined towards the problems of the treatment of sludge arising from peaty upland waters, but attempts have been made to point out where differences may arise when dealing with sludges produced from waters of various types.

6.1. INTRODUCTION Sludge is the waste product produced in a water treatment process and as such is a material of little or no intrinsic value. It is produced in large quantities and presents the industry with a very significant disposal problem. Increasing interest in environmental matters together with pressures on land use and concern for public safety have escalated the problems of waste disposal. The term sludge can be taken as describing any suspension of solid material in a liquid. In the context of water treatment processes the liquid phase of the sludge is invariably aqueous whilst the solid phase will consist of any materials derived from the raw water together with the residues of any chemicals added in the treatment process. It is apparent that in the past insufficient thought was given to the disposal of waste products at the design stage of a water treatment plant, the problem being left in abeyance pending operational experience. This philosophy has resulted in large areas of land in the vicinity of water treatment plants being given over to sludge-holding lagoons. Such areas ultimately become filled with gelatinous sludge which presents a safety hazard as well as a visual eyesore and merely postpones the day when a viable sludge treatment and

Sludge treatment and disposal

143

disposal system has to be provided. Indeed in the long term such an interim scheme compounds the disposal problems in that any new process for disposal must be capable not only of dealing with the sludge produced from the treatment plant but also capable of gradually dealing with the old sludge from the lagoons to enable such areas to be made safe and put to a more productive use. It is therefore recommended that sludge treatment, and where practicable wastewater recycling, should be an integral part of the plant design. Since the costs of sludge treatment can amount to up to 17% of the total costs of the water treatment process it is apparent that sludge treatment should be worthy of rigorous attention to detail during the design stage of a treatment plant. The objective of a sludge treatment process should be to separate the liquid and solid components in a manner in which both are capable of re-use or disposal. In other words the liquid component must be of such a quality that it can be either recycled or discharged to a watercourse and the solid component reduced to a form in which it is readily transportable for disposal and which will not, as a result of such disposal, give rise to a hazardous environment.

6.2. TYPES OF SLUDGE The properties of different sludges will vary considerably dependent upon their origin and the type of treatment applied to the raw water. Sludges can conveniently be grouped into two types: (1) Sludges produced without chemical treatment. (2) Sludges produced from a chemical treatment process. 6.2.1. Sludges Produced Without Chemical Treatment Primary Sedimentation Sludges In general such sludges will consist of the coarser solid materials, such as sand and coarse silt, settled from the raw water. As such they should be relatively free draining and present few or no problems of separation into their two components. Sludges of this type will generally accumulate in channels and sumps at the inlets to treatment works. Washwater from Microstrainers and Filters used Without Coagulant When a raw water is of an acceptable colour but contains small amounts of suspended matter such as algae it is often possible to remove this suspended material by either microstraining through fine wire mesh fabric or by filtration through sand filters without the use of coagulant. Such processes are often used prior to slow sand filtration. The washwater produced will contain algal remains and some larger particulate matter. The washwater from a microstrainer installation will generally be of quite low solids content since the backwashing of such installations is carried out on a continuous basis. Washwater from rapid gravity sand filters used without coagulant can, however, contain quite high concentrations of suspended solids since the backwashing will be carried out at time intervals depending on filter performance and solids load in the raw water. The

Developments in water treatment—2

144

batches of backwash water when they arise will therefore contain the accumulated suspended solids from a considerable volume of throughput water. Effluent from Slow Sand Filter, Sand-Washing Plants When slow sand filters become partially blocked by solid matter in the upper layer of sand, this sand is skimmed off and washed free from debris by some mechanical process. The suspended solids concentration of the effluent from the sand-washing process can be quite high and will normally consist of organic material from the removal of algae, the hydrated oxides of certain metals present in the raw water together with fine sand particles. 6.2.2. Sludges Produced from a Chemical Treatment Process Sludges Produced by Coagulation Sludges produced from water treatment processes utilising coagulation by means of aluminium or iron salts can vary considerably depending on the chemicals used and the type of treatment plant. The suspended solids concentration of backwash water from sand filters employed after coagulation whether they be pressure filters in use as a single stage treatment, or rapid gravity filters used after sedimentation, will usually be within the range 100–500 mg/litre. The sludge obtained from desludging sedimentation tanks will vary considerably in suspended solids content. Given efficient management of desludging a concentration of 0·5% (5000 mg/litre) should be attainable from plants treating soft upland water whilst higher concentrations may be obtained from plants dealing with lowland river waters. Whatever the coagulant used there is a physical and chemical bonding of water to the sludge particles which makes dewatering difficult. Sludges Produced by Precipitation Softening The quality of sludges will vary widely, ranging between pure calcium carbonate, through mixtures of calcium carbonate and magnesium hydroxide, to those containing suspended matter from the raw water mixed with the softening products together with the residues of any coagulant used in the process. Of these sludges pure precipitated calcium carbonate may have some commercial value to the pharmaceutical industry. Other sludges in this category may find application for treating agricultural land in place of ground limestone. Softening sludges are normally of a higher solids concentration than coagulant sludges and, being more particulate, they are more easily dewatered. Sludges Produced by Iron and Manganese Removal Plants Sludges derived from such plants generally have similar characteristics to coagulant sludges.

6.3. CHOICE OF METHOD FOR DISPOSAL Apart from the case of the softening sludges mentioned above, which form a very small proportion of the total water treatment sludge production, the disposal method usually

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.

6.4. DEWATERING OF WATERWORKS SLUDGE The capital cost of any sludge dewatering process is influenced largely by the quantity of sludge to be handled. There are great economic advantages if the sludge can be thickened and hence the volume reduced as much as possible prior to subsequent processes. The upper limit of concentration to which a given sludge can be thickened is governed by the flow characteristics of the sludge. The sludge should be thickened to the maximum concentration at which it is still capable of being pumped. This concentration will normally be not less than 4% in the case of sludges produced from coloured upland waters of low turbidity and could be higher for sludges produced from waters containing appreciable levels of suspended solids. Dewatering of waterworks sludge can therefore be considered as falling into two separate stages, thickening followed by further dewatering. It is important that any sludge thickening process employed must not only result in the production of a sludge of the desired concentration but must also produce a supernatant water of a quality suitable for either recycling through the water treatment plant or discharge either to drain or a local watercourse. Before considering methods of thickening and dewatering it is worth considering the way in which the sludge to be handled is produced from the water treatment plant and what operational practices can be used to minimise the subsequent problems of sludge dewatering. As has been mentioned earlier when considering the various types of sludge, the sludge produced from sedimentation tanks is at a considerably higher concentration than that arising from the backwashing of filters. Even when treating coloured upland waters of low turbidity it should be possible to operate the sludge bleeds from modern sedimentation tanks to yield a sludge of 0·5% solids concentration. If this is not being achieved then attention should be given to the desludging operation to increase the sludge bleed concentration to the maximum possible level. Whilst it is perfectly feasible to separate the solids from filter backwash water and then treat these solids along with the sedimentation tank sludge this can involve, in effect, the operation of two separate sludge separation and thickening processes which can increase both capital and operational costs. It is better where possible to use the existing structures of the water treatment plant to achieve as much sludge thickening as possible. This can be done by arranging for the dirty filter wash water to be collected in a holding tank and returned to the treatment plant

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.

6.5. SLUDGE THICKENING The most widely used method of sludge thickening is by slow stirring. Work carried out over twenty years ago1 showed that the sludge produced from the treatment of a coloured upland water could be thickened to about 2·5–3·0% solids by this method although the quality of the supernatant water from such thickening was often poor. The introduction of synthetic polyelectrolytes during recent years has had a great impact on sludge thickening both in terms of the solids concentration attainable and the quality of the supernatant water for disposal. In the present state of knowledge of the theory of the action of polyelectrolytes it is not possible to scientifically define an application and thereby select the best product for the purpose. It is more a case of trying a range of products and selecting the best one of that range for the job in hand. If it is intended to use a polyelectrolyte for sludge thickening and the resultant supernatant water therefrom is to be discarded, then a wide range of polyelectrolytes is available from which the ultimate selection can be made. If it is intended to recover the supernatant water then the choice of products will be restricted to those products which have been approved on toxicity grounds, and even then the dose which can be applied to the sludge may be governed by the maximum recommended dose which may be applied to the water being treated for potable purposes. This is particularly so where a polyelectrolyte is also being used as part of the treatment process. The efficiency of action of any polyelectrolyte depends on the rapid and complete dispersion of the small dose applied into the bulk of the ‘dirty’ water or sludge being treated. This is necessary because the molecular chains of the polyelectrolyte, particularly those of high molecular weight, appear to have a voracious appetite for the solid sludge particles. The rapid and thorough dispersion is necessary to ensure that all sludge particles are equally exposed to the action of the polyelectrolyte. If the dispersion is not rapid then those sludge particles which first encounter the polyelectrolyte molecules become, in effect, overdosed whilst the remainder are underdosed. The nett result of this

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 455litre 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

FIG. 1. The sludge thickening and polyelectrolyte dosing arrangements at the Fishmoor water treatment plant. Research Association carried out extensive pilot-scale investigations at the Stocks plant, the results of which have been published by the Association’s successor, the Water Research Centre.2 It is interesting to note that virtually all the conclusions arrived at during the original investigations and experiments carried out at the plant over a period of years were verified during the above pilot-scale experiments. A seminar was held at the Water Research Centre early in 1978 on: Alum Sludge Disposal, Current Practice and Trends. At this seminar four papers were discussed including one on ‘Research in Alum Sludge Treatmen’ by the Water Research Centre staff. The proceedings of this seminar have been published3 and the research at the Centre into sludge treatment is continuing. Whilst it must be realised that there will be differences in the thickening characteristics of different sludges, it is interesting to note the variation in dose of the polyelectrolyte required to produce a thickened sludge of similar solids content at the two

Sludge treatment and disposal

151

plants which both treat waters of an essentially similar type but using different mixing techniques. To yield a final sludge of 4.5% solids the polymer dose used at Stocks is of the order of 4·5 mg of polymer per g of dry sludge solids whilst at Fishmoor it is only 1·5 mg/g.

6.6. THEORETICAL BASIS OF SLUDGE DEWATERING Before giving detailed consideration to the various methods available for further dewatering of coagulant-based waterworks sludge it is necessary to consider the mechanism of such dewatering which is applicable to most of the processes to be considered. Most of these processes involve some form of filtration of the sludge and it is only during the very early stages of such filtration that the filtering medium has a controlling effect on the rate of dewatering. After a very short time the filter medium becomes blinded by a layer of the sludge and it is this increasingly thick layer of solids which governs the subsequent rate of dewatering. The rate of dewatering by filtration is therefore governed by the properties of the sludge itself, and is virtually independent of the characteristics of the filter medium provided such medium is capable of retaining a layer of sludge which is of course necessary if the quality of the filtrate is to be acceptable. The dewatering takes place according to the following equation, which is derived from the Carman-Kozeny and D’Arcy equations:

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

Developments in water treatment—2

152

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

FIG. 2. Apparatus for determining the compressibility index and the specific resistance of sludge. A=water-driven vacuum pump B=Saunders valve C=vacuum gauge D=vacuum reservoir (500 ml) E=glass stopcock F=50 ml graduated measuring cylinder G=Buchner funnel to take 5·5 cm filter paper. The viscosity of the filtrate will vary with the temperature of the sludge. Temperature of sludge Viscosity of filtrate (°C) (poise) 0 5 10

1·79×10−2 1·52×10–2 1·31×10−2

Sludge treatment and disposal

15 20

153

1·14×10–2 1·01×10–2

The value of S and r0 are obtained from the experimental data by means of the following treatment:

therefore log(V2/t)=(1−S) log P+log(2A2/ηr0C). A plot of V2. against t for each pressure examined gives gradient g. A plot of log g against log P gives a straight line of gradient (1−S), hence S is determined. This straight line gives the value of log(2A2/ηr0C) from its intercept on the y axis at log P=0 (i.e. P=1), therefore r0 can be calculated. The conditioning of a sludge prior to dewatering in order to increase the rate of dewatering involves treatment by polyelectrolyte or other means to alter the values of S and r0 and so the above experiment yields valuable predictive information on a laboratory scale on the benefits likely to be obtained on a full scale. Details of the above experiment have been described by the author4 along with other information on the conditioning of sludge by means of polyelectrolyte. A simpler method of predicting changes in the dewatering characteristics of sludge is by means of the capillary suction time method developed at the then Water Pollution Research Laboratory,5 but this method has the disadvantage that it is performed at a single pressure equivalent to the capillary suction pressure of the paper being used and hence does not take account of changes in the compressibility index S with pressure, which can be important when the sludge is to be dewatered at elevated pressures as in a filter press.

6.7. METHODS OF FURTHER DEWATERING There are six principal methods of further dewatering of coagulant based sludges, each of which is examined below. 6.7.1. Drying Beds A sludge drying bed is a development of a simple sludge lagoon in that it incorporates a system of underdrains overlain by a permeable filtration medium to facilitate drainage and dewatering of the sludge. Many different materials have been proposed and used as the permeable filtration medium in drying beds, the main ones being sand, ‘no-fines’ concrete and specially woven wire fabric known as ‘wedge-wire’. Whichever permeable medium is used it is essential that it should not become blocked by the sludge particles. If

Developments in water treatment—2

154

blockage does take place then the drying bed merely acts as a lagoon and the only dewatering achieved is that which takes place by evaporation. This system has been investigated by Gauntlett and Packham6 who conclude that drying beds of suitable construction provide an effective means of drying waterworks clarification sludge during those months of the year when evaporation proceeds at a reasonable rate. They also stated that drying proceeds at a very slow rate indeed during the winter months. From this it is apparent that evaporative loss of water plays a significant part in the sludge drying and as such it is not a process which is of much use in temperate climates such as the UK but may be of use in the hotter and drier parts of the world. 6.7.2. Rotary Vacuum Filtration The rotary vacuum filter consists of a large drum or cylinder the circumference of which is covered with a suitably fine filtration fabric. The drum is mounted in a frame and motor driven so that it rotates slowly about its cylindrical axis. A partial vacuum is applied to the interior of the drum, the lower part of which passes through a trough into which the sludge is fed. The partial vacuum causes a thin film of sludge to adhere to the filtration fabric as it passes through the sludge in the trough. Water is drawn from this film of sludge, by the vacuum, during the remainder of the rotational cycle. The water drawn into the centre of the drum is collected in a drainage trough and hence to waste whilst the dewatered sludge on the outside of the drum is scraped from the fabric by means of a horizontally mounted knifeedge just before it re-enters the sludge trough. This type of machine has been established for many years and has proved useful for the dewatering of many types of sludge. It suffers however from the fact that the maximum pressure which can be used in the dewatering process is limited to the level of vacuum which can be maintained in the interior of the drum which is of course something less than one atmosphere. Additionally the maximum time which can be allowed for dewatering to take place is the time taken for that part of the revolution of the drum between leaving the sludge trough and the sludge being scraped off the fabric, i.e. it is governed by the rotational speed of the drum. Difficulties can also be encountered in the selection of a suitable filtration fabric capable of supporting the sludge to be dewatered without allowing an excessive quantity of fine sludge particles to pass through with the filtrate. This can be overcome by the use of a precoat material, which in effect is a material the particles of which are of a size and shape that are retained on the filter fabric and then act as the support for the sludge particles. The precoat is applied by allowing the circumference of the drum to pass through a trough containing a suspension of the precoat material immediately prior to the sludge trough. The use of a precoat adds to both the cost of dewatering and the quantity of sludge for ultimate disposal. The rotary vacuum filter has found the widest application in the water supply industry for the dewatering of sludges of the more particulate type such as those produced in water softening processes. Because of the limitations mentioned above it has not found great favour for the dewatering of coagulant-based sludges.

Sludge treatment and disposal

155

6.7.3. Belt Pressing During recent years a number of pieces of sludge dewatering equipment have become available which can best be described as belt presses. In some respects they resemble vacuum filters but are provided with additional means of achieving a greater level of dewatering than is possible on a normal vacuum filter. They consist of a belt of fine filtration fabric of considerably greater length than the circumference of a rotary vacuum filter. The belt passes through a trough of sludge or is fed with sludge from above and passes over a drum or chamber in which a partial vacuum is established. This results in a film of sludge adhering to the belt which is further dewatered by passing between a series of rollers at gradually increasing pressures so that further water is squeezed out of the sludge. The limitations of pressure and dewatering time inherent in the rotary filter are thereby alleviated to some extent but the problem of selecting a suitable filtration fabric for the belt remains, if the use of a precoat material is to be avoided. Such belt pressing equipment has found more application in dealing with the more fibrous type of sludges produced from sewage than with coagulant-based sludges. 6.7.4. Filter Pressing The filter press or plate press has been in use for many years in various process industries as a means of dewatering suspensions of solids in liquids. It comprises a multiplicity of recessed plates the surfaces of which are covered by a detachable filter cloth; each plate has a hole of about 100 mm diameter in the centre which forms the sludge inlet system to the press. The plates are mounted on a heavy cast-iron frame either slung from a top beam or supported on side bars. The frame is of such a length that the plates can be moved successively along their supporting beam to enable the pressed sludge cakes to be removed, and is fitted with hydraulic or mechanical equipment to enable the plates to be held tightly together during the dewatering cycle. The principle of operation is that the press is closed up and sludge is pumped into the central core and hence into each chamber formed between adjacent plates. The solids in the sludge are retained in the chamber to form a sludge cake whilst the water passes through the filter cloth and drains away through channels in the plate itself. Plates can be obtained with different depths of cavity to yield sludge cakes of different thicknesses, the normal thickness being 25–30 mm. Filter presses normally operate at pressures of up to 6·7 bar, but equipment working at even higher pressures is now available. It does not necessarily follow that the dewatering is more rapid if a higher pressure is used. This depends to a large extent on the compressibility index of the sludge in question and its effect on the rate of filtrate production according to the Carman-Kozeny equation. The cake thickness chosen for a filter press installation will again depend on the dewatering characteristics of the sludge itself and should be selected to give the desired degree of dewatering in a reasonable press cycle time. Many coagulant-based sludges in the past did not prove amenable to filter pressing, merely producing a sludge cake with a hard outer skin and a soft centre. To overcome this problem the selected cake thickness would have had to be so small as to render the

Developments in water treatment—2

156

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

6.8. OPTIMISATION OF WATER AND SLUDGE TREATMENT PROCESSES The foregoing has been mainly concerned with the handling and treatment of sludge from existing or ‘conventional’ water treatment processes. There are however certain water treatment processes which have been developed in recent years which, whilst not necessarily reducing the quantity of sludge, do in fact minimise its handling problems by producing the sludge from the water treatment process at a higher concentration and thereby simplifying subsequent processing. The two most notable of such recent processes are flotation and lamella sedimentation. 6.8.1. Flotation The flotation process as applied in water treatment is normally the dissolved air flotation process of solid-liquid separation. In this process the suspended particles produced by coagulation and flocculation of the raw water are induced to rise to the surface by the attachment of minute air bubbles. The flotation process is carried out in a tank equipped with mechanical means of skimming the floating sludge layer from the surface of the clarified water. The skimming process can be either on a continuous or an intermittent basis. Continuous removal of the sludge from the surface of the tank, by whatever means, does not provide any benefits in the way of simplifying the sludge treatment and disposal; it may indeed worsen the problem since the sludge will be at a very low solids concentration and a thickening process capable of dealing with quite a high throughput will be required. If, however, the sludge skimming is intermittent and the layer of sludge can be allowed to remain on the surface of the tank for a period of hours then a considerable amount of dewatering of the sludge takes place by water draining from the aerated sludge layer back into the tank. If the sludge removal from the surface of the flotation tank can be carried out in this intermittent manner without adversely affecting the quality of the clarified water passing forward for filtration, then it is possible to obtain sludge concentrations in excess of 2% solids direct from the flotation unit. The fact that sludge of this concentration from a

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. 6.8.2. Lamella Sedimentation The lamella sedimentation process is a continental development and has not been much utilised in the UK. It is a high rate sedimentation process which takes place in tanks equipped with inclined plates which form in effect a series of very shallow sedimentation tanks. The clarified water rises between the plates whilst the sludge solids settle on the plates and run down the plate surface into the bottom of the tank which is equipped with what in effect is a totally submerged picket fence slow stirring sludge thickener. It is claimed that sludge concentrations of the order of 4% can easily be obtained from such equipment. In addition to new processes which minimise sludge treatment there are two additional developments which minimise sludge production and hence minimise the problems of sludge treatment and disposal. They are coagulant replacement and coagulant recovery. 6.8.3. Coagulant Replacement The use of a traditional inorganic hydrolysing coagulant in a water treatment process in itself produces a considerable quantity of sludge in addition to the materials removed by such coagulant from the raw water. An example of this is the use of aluminium sulphate in which approximately 20% of the dose level as ‘aluminoferric’ or block alum contributes to the sludge as aluminium hydroxide, i.e. a dose of 40 mg/litre of block alum will result in 8 mg/litre of dry solids in the sludge. This can often mean that 50% of the sludge produced in the treatment process arises from the chemicals introduced. The partial or total replacement of the hydrolysing coagulant by a suitable cationic polyelectrolyte is claimed to reduce sludge production considerably. This claim is undoubtedly true if such a polyelectrolyte can be found which will produce results as good as the traditional coagulants at a reasonably economic cost. Unfortunately the development of such cationic polyelectrolytes is still in its infancy and the types currently available are only applicable to certain types of raw waters. In addition to this the cost of

Sludge treatment and disposal

159

such materials is such as to render their use uneconomical as compared with the use of a hydrolysing coagulant together with a polyelectrolyte as a coagulant aid. Since the full-scale use of cationic polyelectrolytes as a replacement for the more traditional materials has been very limited, there has not been the opportunity to study in any great detail the handling and dewatering characteristics of the sludges produced. 6.8.4. Coagulant Recovery When a hydrolysing coagulant such as a salt of aluminium or iron is used it is possible to minimise the quantity of sludge for ultimate disposal by recovering for re-use as much as possible of the coagulant from the sludge. This recovery normally consists of thickening the sludge as much as possible and then extracting the used coagulant with sulphuric acid. It is claimed that coagulant recovery levels as high as 70% can be obtained and the quantity of sludge for subsequent disposal is greatly reduced. It is also claimed that the dewatering characteristics of the residual solids after acid extraction are modified in a beneficial manner. The operation of such a process requires careful control of sulphuric acid dosage to avoid recovery of undesirable metals from the sludge along with the aluminium or iron salts. A coagulant recovery process is at present used at one waterworks in the UK where the recovery is allied to a sludge freeze-thaw dewatering system. The pilot plant work on which this plant design was based has been described in the literature.11 Coagulant recovery is an expensive process since it involves the use of acid resistant materials in many parts of the sludge handling equipment and its economic viability depends on the relative costs of fresh coagulant, sulphuric acid and suitable dewatering equipment at the site in question. In addition to the economic aspects of the recovery process there is always the doubt about what other materials will be recovered inadvertently from the sludge by acid extraction. It is indeed a philosophical question as to whether it is correct to go to considerable expense to treat a water with a coagulant to remove undesirable materials and then extract the sludge with acid and return the acid-soluble fraction to the water undergoing treatment.

6.9. SLUDGE QUANTITIES A dilemma facing the designers of new treatment plants in which it is desired to incorporate sludge treatment at the outset is the estimation of the quantity of sludge which is liable to arise from the water treatment. This is indeed a problem if data on the quality of the raw water to be treated are limited and is often the reason why temporary disposal methods by lagooning are adopted initially, with a view to formulating firm proposals for sludge treatment and disposal after initial operational experience of the water treatment plant has been obtained. The lead-in time to the development of a new water supply scheme or new treatment plant is normally sufficiently long to enable adequate data on raw water quality to be obtained allowing the sludge treatment facilities to be incorporated at the outset. All that is required is a knowledge of the suspended solids and colour levels in the raw water and their variability with time along with estimates of the coagulant doses likely to be

Developments in water treatment—2

160

required, the latter being required anyway to enable the chemical storage and dosing requirements to be specified. The amount of sludge produced in terms of dry solids can be calculated from the suspended solids and colour in the raw water together with the solids arising from the coagulant dose envisaged along with the residues from any other chemicals used in the water treatment process such as coagulant aids and activated carbon. Equations have been published for this quantification.12 Having ascertained the quantity of dry solids expected to arise from the process it is a relatively simple matter to convert this into the volumes of liquid sludge at various stages throughout the water treatment and sludge handling processes and to design the necessary tanks and dewatering equipment accordingly.

6.10. OPERATIONAL ASPECTS OF SLUDGE TREATMENT Whilst it is recommended that sludge treatment and disposal should be designed as an integral part of the whole water treatment system, care must be taken to ensure that the whole of the various processes involved do not become too interdependent. An obvious example of this is a problem arising on the sludge thickening system which results in an inability to desludge the sedimentation tanks at the required intervals and this in turn results in a deterioration in settled water quality passing forward for filtration. Sufficient capacity must be built into the system to allow for breakdowns on the sludge treatment being rectified before their effect is transmitted back to the water treatment. Sludge treatment must always be subservient to the primary purpose of producing a treated water of the required quality. Another example of the possible interaction of various processes is between the dewatering and ultimate disposal of the sludge. If the sludge is dewatered by filter pressing and the ultimate disposal is by road transport, the emptying of the filter presses should not be dependent on the availability of road transport at the appropriate time. Neither should the loading rate of road transport vehicles be dependent on the rate of emptying of the filter presses. In such cases it is better if the filter presses are provided with hoppers below them capable of holding one press load of dewatered sludge cake and a means of loading the sludge from the hoppers onto the road transport at a reasonable rate either by means of screw or belt conveyors or a combination of the two. The provision of such a system of hoppers and conveyors can also make for savings in manpower costs since the presses can be emptied and put back into the filtration mode immediately and only after this has been done is there a need for the same personnel to concern themselves with the actual handling and disposal of sludge solids.

6.11. CONCLUSION From the foregoing it can be seen that several processes are available for the efficient and economical treatment and disposal of waterworks sludge. It is important at the design stage, whether the sludge treatment facility is to be part of a new or an existing water treatment plant, to take account not only of the technical aspects of the design and

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 1. DOE, P.W., J. Instn Water Engrs, 1958, 12, 409. 2. Water Research Centre Tech. Rep. TR62, 1977. 3. Proc. WRC Seminar of Alum Sludge Disposal, 1978. 4. HILSON, M.A., J. Instn Water Engrs, 1971, 25, 402. 5. BASKERVILLE, R.C. and GALE, R.S., Water Pollution Control, 1968, 67, 233. 6. GAUNTLETT, R.B. and PACKHAM, R.F., J. Instn Water Engrs, 1972, 26, 185. 7. SANKEY, F.A., J. Instn Water Engrs, 1967, 21, 367. 8. BENN, D. and BRIDGES, L., J. Instn Water Engrs, 1971, 25, 417. 9. J. Instn Water Engrs & Scientists, 1979, 33, 80. 10. DOE, P.W., BENN, D. and BAYS, R.L., J. Instn Water Engrs, 1965, 19, 251. 11. WEBSTER, J.A., J. Instn Water Engrs, 1966, 20, 167. 12. J. Instn Water Engrs, 1973, 27, 399.

Chapter 7 WATER QUALITY MONITORING P.J MORLEY, C.Chem., M.R.I.C. Principal Scientist, Avon Division, Severn-Trent Water Authority, Coventry, UK AND J.COPE, C.Chem., M.R.I.C., Ph.D. Scientific Officer, Severn-Trent Water Authority, Birmingham, UK SUMMARY The firstpart of this chapter deals with the historical aspect of water quality monitoring and the need for increased vigilance in recent years. This has taken two courses—the use of continuous monitors, and modern automated laboratory analyses. The former offers rapid answers, but needs a high capital outlay and suffers from doubtful reliability and a lack of versatility. The latter can provide reliability and a variety of analyses, but has the disadvantages of errors caused by transport of samples and the time delay before results are available. The two approaches are discussed in detail together with their relative merits and future development.

7.1. INTRODUCTION Historically quality control of the water cycle was the responsibility of such agencies as the Public Analyst and the River Authority. The analytical techniques involved frequently required considerable skill and time from their staff. In certain areas, such as plant control, methods were considerably simplified to provide results quickly and cheaply from non-chemical personnel. Although this gave useful information, when used in context of how it was gained, it led to many rough imprecise data being accumulated. In the early 1960s there was an acceleration of interest in the environment and a growing awareness that raw water supplies were at risk from a variety of sources. Predictions of much higher water consumption created a need for information relating to the quality of our rivers, canals, lakes, boreholes and springs. The data available were in many areas unreliable and the acquisition of new records promised to be labour intensive and consequently very expensive. The approach to the problem has taken two courses—the use of modern laboratory techniques and continuous monitoring on site. The need for a cheap instrumental monitor to continuously gauge water quality at source unattended is obvious. The development work so far, however, has produced equipment capable of measuring only a limited number of parameters. Those first determined were dissolved oxygen, electrical

Water quality monitoring

163

conductivity, pH and temperature, all of which are best measured on site and such instruments have found wide use. The later generations of monitors, which utilise such sensors as the ammonia and fluoride specific ion electrodes, have also found considerable areas of application. Their main advantage of instant response at the point of sampling has to be balanced against a number of other factors. They are expensive to install and service, at trace levels they can lack sensitivity and much remains to be done to produce reliable sensors which maintain calibration between checks. Both the use of a continuous monitor or a modern laboratory facility share the same serious problem. It is as difficult to present a representative flow to an in situ sensor as it is to obtain a bottle of sample which fairly reflects the state of a heterogeneous mass of water. The main advantage of laboratory analysis is that the number of parameters analysed can be much greater, although the cost of equipping and staffing such an establishment is high. The recent trends are to utilise modern techniques and automation to minimise the cost and continuous monitors are constantly being improved and their range becoming more diverse. At present these two methods of data collection complement one another and it is not likely, in the foreseeable future, that either will present itself as the complete answer.

7.2. LABORATORY INSTRUMENTATION AND AUTOMATION As previously described, the last twenty years has produced an ideal climate for improvement in the laboratory. This has taken the course of the use of equipment to automate manual techniques and the development of instrumentation capable of measuring trace quantities previously unattainable. The approach to the handling of results has also been considerably revised with the help of advancements in the fields of data capture and computing. Currently the instruments manufactured and suggested for the analysis of large numbers of individual sample solutions can be classified into two areas—continuous flow analysers and batch analysers. In continuous flow analysers the samples are pumped from their containers into a glass or plastic tube through which they move until the reaction is complete. The samples are thus part of a continuously moving stream with reagents being added at controlled rates at specific points. The stream then flows to a measuring device which is commonly a colorimeter or specific ion electrode. This technique, above all others, is extremely versatile. The sample stream may be split for multiple analyses, filtered, dialysed, heated, extracted with organic solvent and distilled. It is also very flexible and, after 15 years of intense development in the water industry, still finds new areas of application. The chief disadvantage is that each sample is liable to contamination from the preceding one and this limits the rate of throughput. The batch analyser treats each sample as a discrete unit and therefore does not suffer the same degree of cross-contamination. Samples are transferred into a reaction vessel together with, at pre-determined intervals, additions of reagent. The vessel can then be heated for a fixed period of time and a portion removed to the colorimeter or other detector. The discrete analyser exhibits rates of analysis of up to 300 samples per hour compared with 40–60 per hour obtained from conventional air segmented continuous

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.

7.3. CONTINUOUS FLOW ANALYSERS In the mid 1960s The Technicon Instruments Company Ltd introduced the Auto Analyser, already well known in clinical biochemistry, into the field of water analysis. The technique of air segmented continuous flow analysis1 is a simple way to automate any colorimetric or turbidimetric measurement and is now a common sight in water laboratories. A basic form of analyser is shown in Fig. 1. Sample, reagents and air are pumped along fine-bore plastic tubing to glass mixing coils. These produce a

FIG. 1. A simple air segmented continuous flow analyser.

Water quality monitoring

165

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

FIG. 2. Dialysis unit. channel, of hemispherical cross-section, machined on one surface. When two of these plates are screwed together, separated by a polyethylene dialysis membrane, the unit is complete. A stream of water to be analysed is passed over the top of the polyethylene and a receiving stream flows, usually at the same speed, underneath. The nature of the receiving stream is chosen to encourage the transfer of the relevant constituent from the water sample. For example, a solution of lower pH than the donor stream will be used to

Developments in water treatment—2

166

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

FIG. 3. A continuous distillation head. in the trap following the water condenser. The phosphoric acid and nonvolatile portion flow to waste whilst the condensate is mixed with more air and the colour reagents. The only disadvantage of this system is that a nonvolatile, thermally stable liquid such as phosphoric acid must be present to prevent the coil from becoming blocked and the conditions, therefore, tend to be very rigorous and limit the application. It is even possible to introduce an acid digestion into the continuous flow system by utilising a long glass helix. The sample, air, and digestion acid are mixed and dropped into the bottom of the helix. Rotation of the tube causes movement of the liquid from one end to the other across heater blocks. With optimum adjustment of acid conditions, temperature and speed of helix rotation the solution emerging from the other end of the tube will be completely digested. Total nutrient concentrations, such as nitrogen and phosphorus, can be monitored in this way and the equipment can also be used to automate the determination of chemical oxygen demand (COD). It should be noted here that COD is an arbitrary value derived from a set of empirical oxidation conditions. The use of the heated helix gives a different value to the standard manual method but it can be effectively utilised to study trends in oxygen demand. A further illustration of the versatility of continuous flow analysis is its application to solvent extraction systems and the laborious manual method for estimating synthetic anionic detergent in waters was one of the first to be automated. The extraction of the methylene blue-detergent complex is

Developments in water treatment—2

168

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 carryover between samples is caused by the extent of laminar flow prevailing. The presence of an air bubble causes friction at the tube walls and gives a turbulent flow even at very low pumping velocities.5 The net effect is a reduction in carry-over but there are also a number of disadvantages. The chief ones are the compressibility of air, which causes the stream to pulse rather than to flow regularly, the necessity of removing bubbles before colour or turbidity is measured and the need to accurately control bubble patterns at fast sampling rates. Ruzicka and Hansen2 exploited the fact that fast flows in sufficiently small diameter tubing are predominantly turbulent. In flow injection analysis a small volume of sample (approximately 0·5 ml) is injected, via a syringe and septum, into a turbulently flowing carrier stream of reagent, the air bubble no longer being necessary. Early work with simple colorimetric reactions, such as the determination of chloride using iron (III) thiocyanate,6 suggests possible rates of 250–500 samples per hour. The application to the water industry has so far been disappointingly small. The technique suffers the disadvantage that it appears to be only suited to simple reagent addition and that long heating times for reaction development are difficult to accommodate. It is, however, a very good system for presenting a buffered sample to ion-selective electrodes in that the response time of these sensors

FIG. 4. Automation of sample presentation to an ion-selective electrode using flow injection.

Water quality monitoring

169

increases dramatically in fast moving streams. A diagram of the equipment required is shown in Fig. 4.7 The flow injection stream of sample and buffer is allowed to impinge on the electrode surface and then falls in a continuous stream to a waste container, which houses the reference electrode. This stream maintains an electrical connection between the two electrodes and increases of response time up to a factor of four can be obtained.

7.4. BATCH OR DISCRETE ANALYSERS Discrete analysers have found wide application in the clinical field and have been available to the water industry for some years. Early experience with them, however, showed poor reliability and few properly developed water chemistry methods. Recent improvements in the mechanics of these systems, particularly in the area of pneumatic operation, have produced an extremely useful rapid laboratory analyser. There are at present approximately 10 discrete analysers available and they fall into two main groups. The most common simulate manual addition and mixing and subsequently present the reacted solution to a spectrophotometer. The other design uses a disc into which a number of radial troughs have been cut, the depth of the cut normally increasing towards the centre. Sample and reagents are placed individually into these arms and the whole disc is spun rapidly in a thermostatically controlled environment. The reaction mixtures are formed at the circumference and their optical densities can be measured as they pass through a light beam. At present an instrument of this design is available but its cost will limit any application to the water industry. The manual simulation is considerably cheaper and at £12000–£15000 (October 1979) it is possible to buy a single channel unit which performs the functions of sample dilution, reagent addition, mixing, incubation and transfer at speeds of up to 240 samples per hour. One of the better instruments available at present is the Pye ACl system in which most moving parts are pneumatically driven. A preset volume of sample is automatically withdrawn by syringe from a tube and dispensed, together with the first reagent, into a reaction vessel. This then moves along a rack, the temperature controlled by a variable water bath, until it reaches

TABLE 2 A COMPARISON OF CONTINUOUS FLOW AND DISCRETE SYSTEMS Continuous flow 1. Maximum rate of 40– 60 samples per hour 2. Approx. £10000 for cheapest five-channel unit 3. High reagent consumption 4. Susceptible to interruptions in sampling programme 5. Susceptible to carry-

Discrete 240–300 samples per hour £12000–£15000 for sequential analyser Low reagent consumption Not applicable (no ‘settling down’ period) 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.

7.5. THE DETERMINATION OF METALS The technique of atomic absorption spectrophotometry is particularly useful for the monitoring of a variety of metallic contamination at trace levels. The analytical speed, ease of operation and relative lack of interferences have made the method attractive for routine water quality analysis. The first step towards automation was a turntable feeding the samples to the instrument and a recorder or printout providing a continuous output of absorbance. This allowed rates of 120 metals per hour to be achieved. More recently improvements in nebuliser and atomiser design have made possible speeds in excess of

Water quality monitoring

171

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 As B Cd Co Cr Cu Fe Mn Mo Ni P Pb Pt Se

10 15 2 1 2 2 2 1 0·5 5 5 30 15 20 15

20 100 1000 1 5 3 2 5 3 10 8 105 10 50 100

0·004 0·06 – 0·008 0·03 0·005 0·008 0·003 0·004 0·06 0·02 3·0 0·03 0·45 0·10

Developments in water treatment—2

172

Si Ti V Zn

0·10 0·30 – 0·0007

10 1 75 1

60 50 7000 0·6

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 semiquantitative investigations using a medium quartz d.c. spectrograph. The development of a stable plasma source has enabled fast multi-element analysis to be seriously contemplated and the high temperature of the plasma (10000°C) minimises the interference effects normally encountered. An RF with a power of the order of 5 kW is applied to an induction coil encircling an inert gas (usually argon). The sample is sprayed into the resultant plasma and the ‘tail’ of the flame is viewed through a monochromator. The multi-element analysis is carried out by using a number of pre-set photomultiplier tubes or by allowing the grating to scan. Table 3 shows the comparative detection limits obtained with a number of metals using the different techniques.

7.6. MONITORING OF ORGANIC CONTAMINANTS The organic fraction of raw water consists largely of naturally occurring compounds originating from land drainage or domestic sewage discharge. The classical determinands of suspended solids—biochemical oxygen demand, permanganate value and chemical oxygen demand—are retained by most Water Authorities as measures of solid matter and organic strength. This work represents the most labour intensive effort remaining in the laboratory as the empirical nature of the tests resists full automation. The tasks can be eased by the use of heated digester blocks, dissolved oxygen electrodes and sophisticated electronic balances but they remain expensive. An alternative measurement of organic strength is to measure total organic carbon (TOC). This instrumental technique, capable of dealing with 15 samples per hour unattended, consists of an oxidation furnace in which all the carbonaceous compounds are converted to carbon dioxide. The quantity of this gas produced can then be measured either by infrared detector or titrimetrically. A further sophistication is the instrument which will reduce the carbon dioxide by mixing with hydrogen and passing over heated Raney nickel. Measurement of the methane produced can be achieved with high sensitivity using a flame ionisation detector. Typically the TOC values for waters range from less than 1 mg/litre for some treated borehole supplies to greater than 5 mg/litre for certain surface waters. Approximately 20% of this organic fraction consists of man-made chemicals and these form the greatest cause for concern. Evidence of their impact on the environment, either alone or combined, has led to the necessity to monitor the individual identity and quantities of the component compounds. Until recently effective separation and analysis at these trace levels would have been impossible. The combining of capillary column gas chromatography with mass spectrometry (GCMS) has produced an instrument with a very high performance in terms of resolving power and selectivity. Modern commercial

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.

7.7. LABORATORY DATA COLLECTION AND PROCESSING The automation of a laboratory to accommodate the analysis of a large number of samples produces a purely clerical problem. This includes sample reception, processing of automatically produced and manually produced data, progress chasing and report production. Morley, Musty and Cope12 describe a system which uses a desk-top computer in combination with magnetic tape data recording. The sample can be booked in at the computer keyboard by typing sample location, date, time, sampler’s name and analyses required. Laboratory work sheets and an outstanding samples report can be produced at this point. By connecting the instrument recorders (fitted with re-transmitting slidewires) to a magnetic tape logger unit the continuous flow analyser output can be recorded and at the end of the analysis run the tape is replayed to the desk-top computer which picks peaks, ‘draws’ calibration graphs, corrects for base-line and sensitivity drift and computes final concentrations. There is an edit facility to allow dilution factors to be applied and for faulty data to be erased. Similarly instruments capable of producing a digital output, e.g. Atomic Absorption Spectrophotometers, can be monitored at a standard BCD (binary-coded-decimal) output and the data recorded and processed. Any manually produced data such as suspended solids, chemical oxygen demand, etc., may be keyed in by the analyst. A simple program sorts through the computer disc store and brings together all data relating to an individual sample for report production. A worthwhile by-product of this system has been to largely eliminate clerical, transcription and arithmetic errors inherent in the manual equivalent. A common criticism by the misinformed is that computer systems dehumanise the analyst and leave him without a ‘feel’ for the samples. They are merely a powerful aid and the system can be devised

Developments in water treatment—2

174

which allows as much contact with the results as manual methods but eliminates the drudgery and error. Much depends on how involved the analyst is with the design of the computer procedures. Because of the rapidly improving technology relating to this area the previously described approach is already becoming outmoded. Most commercial instruments which give a digital output are fitted with microprocessors to compute final sample concentration. They are, however, not always very well programmed and it is advisable to find an answer to the question, ‘How are the results calculated?’ With a satisfactory instrument, however, there is no need for the complex software to compute results. Even the continuous flow analysers can be fitted with data processors to give an output of sample concentration. The introduction of interface systems such as the Hewlett Packard Interface Bus, now allows the computer to act as a collator of information. The individual instrument can plug into a ‘ring main’ circuit and automatically convey data, to the disc store, as it is produced and when its priority rating in the system permits. A number of instruments can be connected on-line to the computer in this way and the data collected and sorted in the background of normal keyboard activity. The presence of a computer in the laboratory has another distinct benefit. It is often possible in such environments to lose sight of the original objective and produce the analytical report for its own sake. With very little effort the analytical results can be combined with other data such as flow to produce useful operational information. The information can be combined with previous data and shown as a graphical trend or compared with historical limits permitting conclusions to be drawn relating to its significance. This concludes the part of the chapter dealing predominantly with laboratory based monitoring and the advantages and disadvantages of such systems have been highlighted. In general its chief weakness is the delay in obtaining a result, whilst its main strength is the reliability and versatility of analysis. The next section deals with automatic stations and a combination of the two techniques should show a satisfactory approach to water quality monitoring.

7.8. AUTOMATIC MONITORING OVERVIEW The concept of a monitor that will continually survey raw water, and water entering supply, is extremely appealing. Should pollution occur by natural phenomena, an accident or wilful intent, then the system could interrupt the flow of water to prevent its passage into supply, identify the pollutant and its concentration. If the monitor was one of a chain of such devices then information as to the location of the pollution would also be available. In addition to performing this role the monitoring system could be used for routine historical water quality data gathering for use in water quality models and as future design data. Present-day monitoring systems are a long way from providing the user with information that he really requires and the reliability of many systems would require extreme care on behalf of the user before fully automatic control systems could operate on a foolproof basis.

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.

7.9. BASIC TYPES OF MONITORS There are two basic methods of deployment of automatic water quality monitors; that is, they may either be land-based on a fixed site or they may be submersible and totally immersed in the water they are measuring. Each type has advantages and drawbacks. In addition to the distinction between land-based and submersible types, a further division can be made between those monitors offered as a multi-determinand package by a manufacturer and monitoring systems assembled by the user incorporating various discrete analytical systems and possibly simpler packaged monitors.

7.10. DESIGN CONSIDERATIONS When designing a monitoring system factors other than the determinands measured should also be given serious consideration. The siting of the station for security, ease of access and construction, the ease with which samples may be taken and the relevance of the readings obtained are very important. The design should be vandalproof but allow good access for users and allow maintenance and repair work to be carried out. The sampling system should be designed to take a representative sample in such a way that it does not degrade it for subsequent measurements (e.g. lifting may affect dissolved oxygen). A burglar alarm system is recommended for unattended stations in remote sites.

7.11. SUBMERSIBLE MONITORS A submersible monitor is a monitoring system which can be wholly immersed in the water it is measuring. The data it measures can be recorded for subsequent treatment or relayed to another site via cable or radio, or the device may act as a transponder. It will be seen that such a device consists of a waterproof housing containing the necessary electronics, power supply and data recording system with the sensors exposed

Developments in water treatment—2

176

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.

7.13. DISSOLVED OXYGEN Dissolved oxygen is most commonly monitored using either a sensor of the Mackereth type13 or a polarographic sensor of the Clark type.14 Both these sensors use gaspermeable membranes which may be subject to fouling. The membranes may be cleaned by mechanical action although care must be taken not to scratch or puncture them. If fouling is a problem other sensors are available, a common one of which is the sacrificial thallium electrode. When monitoring dissolved oxygen it is extremely important not to carry out any processes on the water which will affect the reading. Lift pumping will tend to lower the dissolved oxygen level whereas turbulence entraining air will increase the level. Water for dissolved oxygen monitoring should be measured in situ or pumped using submersible pumps. The dissolved oxygen measurement should be taken before the sample is subjected to any chemical treatment.

7.14. pH pH is a measure of the acidity or alkalinity of an aqueous liquid. It is most commonly determined by measuring the potential between a reference electrode and a specially constructed glass electrode immersed in the water. A very high input impedance measurement system is used and the measurement is amplified and scaled to give pH. A more detailed explanation of the measurement system will be found in most physical chemistry text books. The glass electrode is a delicate device and may encounter problems if used in a monitoring environment. Present-day glass electrodes are relatively robust but exposure to abrasion or vibration may damage them irreparably. An alternative electrode for the determination of pH, where glass electrode failing is a problem, is the antimony-antimony oxide electrode. This device often incorporates a mechanical scraper to expose fresh electrode material and is a useful system for automatic monitoring. Care should be exercised when measuring and interpreting pH data on low conductivity waters as electrode contamination, gas absorption and the effects of the measurement system may change the pH value significantly. In these circumstances the value of measuring pH and its relevance should be borne in mind when interpreting these data.

Developments in water treatment—2

178

7.15. ION SELECTIVE ELECTRODES Devices of this type have been available for many years, the most widely used example of which is the glass electrode used in pH measurement. The concept of an electrode which may be immersed in water and give a readout of a specific determinand is very appealing. Unfortunately, very few of these devices can be used without sample pre-treatment and complete specificity. Before installing detectors of this type a user should ensure that provision is made for the correct sample pretreatment if required, and that the system being measured will not present any interfering species to the electrode system. Electrodes which find common use are the ammonia gas permeation electrode, the fluoride electrode and the chloride electrode.16 Whilst manufacturers may list electrodes for a wide variety of determinands, users must ensure that they can be successfully applied to the media being measured.

7.16. OXIDATION-REDUCTION POTENTIAL (ORP) OR REDOX POTENTIAL This is sometimes measured in pollution monitoring systems often using a platinum and a calomel reference electrode. It is probably one of the most difficult determinands to evaluate but indications as to changes in the system will often show a change in redox potential and it can help in verifying trends shown by other sensors.

7.17. SUSPENDED SOLIDS There are two ways of interpreting suspended solids and it is important to be aware of this when considering automatic methods for their determination. The laboratory test for suspended solids involves the filtering of a known volume of sample through a pre-weighed filter paper, drying the paper and re-weighing to est ablish the weight of solid the volume of sample contained. This yields a result in weight per unit volume which is the way solids levels are most commonly expressed. The automation of this technique for monitoring poses many serious technical problems and is not practicable at present. The opacity of liquids containing suspended matter is easily observed by eye and this property is frequently used to assess suspended solids levels.17 The turbidity or lightscattering 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 Conductivity is a determinand that is easy to measure and is thought to give an indication of an ionic pollution.15 Care must be taken when applying temperature corrections to conductivity measurements and it is preferable to record the absolute conductivity together with the temperature of the sample. Conductivity is normally measured with an a.c. electrical bridge system operating at either 1000 or 1592 Hz. D.c. measurements should be avoided owing to electrolytic and polarisation phenomena. Electrode designs vary but it is common to use ring-type electrodes for monitoring work. These are arranged as rings on the inside of a tube and guard electrodes are often incorporated in the design. These electrodes are easily cleaned or kept free of fouling by high flow rates.

7.19. BIOLOGICAL SENSORS As has been previously discussed, it is impossible to provide physical and chemical automatic sensors which would provide total coverage of pollutions hazardous to life, especially human life. In an attempt to assess water conditions which may endanger living species, sensor systems have been designed which monitor the changes in behaviour of life forms when exposed to the water under examination.18 One such method is to record the behaviour of fish living in the water to be monitored. It is postulated that the fish have an established behaviour pattern in unpolluted water. If the pollution level in the water rises then the fish behaviour will become more agitated and this is monitored electronically and assessed by computer. Biological sensors which monitor the metabolic rate of lower organisms are also used. A pollution will cause the metabolic rate to drop and this can be measured relatively simply and the metabolic rate displayed on a chart recorder. A pollution detected by biological sensors of these types still has to be identified, and this may prove a difficult task requiring substantial laboratory facilities and expertise. It is essential that when a pollution has been indicated by a biological sensor, samples of the suspect water are taken for subsequent laboratory investigation.

7.20. CONTINUOUS FLOW ANALYSERS The air segmented continuous flow analyser has been adapted to perform a continuous monitor function. The pump rates and therefore reagent usage are lower than in conventional laboratory systems and the instrument receives sample from one source only. The advantages of this unit are that continuous flow methods used in the laboratory

Developments in water treatment—2

180

can be adapted to be used in a monitoring situation, and that it enables tests to be performed in a monitoring station which are not possible by other methods.

7.21. TEMPERATURE MEASUREMENT Automatic temperature measurement is usually performed by a device such as a resistance thermometer, thermistor or thermocouple. Remote measurement techniques from satellites and aircraft commonly use infrared thermography. Temperature measurement serves two functions. It is an important determinand in its own right and an indication of possible thermal pollution. It is also essential to the correct interpretation of data from other sensors being used for water quality monitoring. An example of the second role is in connection with measuring electrical conductivity which is temperature-dependent. Most standard laboratory methods require conductivities to be determined at the standard temperatures of either 20 or 25°C. When monitoring raw or supply water it is impractical to adjust the temperature to that of standard laboratory conditions. A common practice is to measure the conductivity at the ambient water temperature and then to use a device known as an automatic temperature compensator to ‘adjust’ the conductivity to the value that it supposedly would be, had it been determined at the standard temperature. The correction is difficult as the temperature coefficient for a water at a specific time is seldom known accurately and large adjustments are made to the reading which can commonly be of the order of 50%. If an automatic temperature compensator is used it may be impossible to establish the absolute conductivity of a water as measured from data so treated. If a record is made of absolute conductivity and temperature, then a temperature compensation can be easily carried out if desired but the integrity of the measured data is not destroyed. Dissolved oxygen, pH and selective ion measurements and redox potential are all affected by temperature variation but not as dramatically as is electrical conductivity. In all cases where temperature-dependent measurements are made then a simultaneous record of temperature enables correct interpretations to be made at a later date and helps establish absolute integrity of data.

7.22. CLEANING SYSTEMS Where fouling of a monitoring system presents a problem there are several methods available which offer the user a means of substantially extending the time between complete overhauls of the system. When an automatic cleaning system is used care should be taken in its choice in that it should not damage the monitoring assembly or the sensors and that, after cleaning, any causes of misreading of the sensors are adequately removed, e.g. chemical agents or air. Some of the cleaning systems available for use on automatic monitoring systems are given below.

Water quality monitoring

181

7.22.1. Mechanical Cleaning A mechanically actuated brush or wiper is used to clear fouling. The operation may be continuous or intermittent. Care must be taken if sensors are cleaned this way to ensure they are designed to undergo this treatment (e.g. some pH glass electrodes and certain dissolved oxygen electrodes may be damaged by this treatment). 7.22.2. Ultrasonic Cleaning This is a form of mechanical cleaning using ultrasonic transducers to generate ultrasound in the liquid in the monitoring system to detach fouling. It may be used with the water being measured by the system or used in conjunction with a chemical cleaning agent or detergent. In practice it is found that this will reduce fouling but, if sufficient sonic power is used to remove all fouling, damage to the system will result. It would be bad practice to take sensor readings whilst an ultrasonic transducer was switched on. 7.22.3. Air Cleaning If the routine measurement cycle is interrupted and a mixture of compressed air and water (possible with chemical cleaning agents) passed through the system the agitation produced will give an effective cleaning action. Again care should be taken that the sensors used can take this treatment and that the air is fully removed from the sensors after cleaning. 7.22.4. Chemical Agents Substances such as sodium hypochlorite solution or detergents can be used to clean a system but they will be much more efficient if used in conjunction with a mechanical cleaning method. If the chemical agent used will interfere with a sensor reading then it should be properly removed before the system is allowed to produce real measurements. If a chemical agent is known to have no effect on a test being performed then a continual dosing of the chemical into the water being measured is a very effective way of reducing fouling. This works quite well with biocides. 7.22.5. Sterilisation Methods If the water entering the monitoring system has been sterilised then fouling will be substantially reduced unless the problem is one of silt. Various methods are available for doing this and the chemical method has already been discussed. Treatment with ultraviolet light (u.v.−C radiation) is one method, but by its nature may cause chemical changes in the sample. The same is true of ozone together with potential damage to components and the invalidation of several sensors, e.g. dissolved oxygen. Radiation sterilisation offers perhaps the best way of achieving sterilisation, but it may be undesirable to use this at an unattended station.

Developments in water treatment—2

182

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

7.23. ATOMIC ABSORPTION SPECTROSCOPY AND PLASMA EMISSION SPECTROSCOPY At present atomic absorption spectroscopy is widely used for the laboratory determination of metals in water.19 With the development of flameless systems and instruments which can be fully computer-controlled it should be possible to incorporate an atomic absorption instrument into an automatic monitoring system. The design of the instrument would probably allow sophisticated control to be used. The advantage of this instrument is that it would offer a means of monitoring specific metal levels in water. An associated field is the determination of metals from their atomic emission lines when excited in a high temperature plasma generated by radio-frequency power or d.c. arc.20 At the time of writing, instruments are being announced which offer similar ease of operation to atomic absorption systems. With increasingly low cost data processing and optical developments such as fast Fourier transform systems and electronic image dissectors the plasma spectrometer may offer a means of determining a large number of elements on a continuous automatic basis.

7.24. REAL TIME With any automatic data logging record there is a possibility that readings may have been missed, power or electronic failures occur and if the data have been accumulated over a period and the number of readings obtained is found to be incorrect then it may be impossible to correctly assign times to the data, which would mean that the whole of the data must be discarded. The recording of real time together with a day indication is essential for data being logged from a monitoring station. The real time clock should be independent of the station power supplies with standby batteries or should have the facility of being linked to a digital time standard so that when power is resumed it can reestablish 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. 7.26. DATA LOGGING* Monitoring may simply be used to detect a pollution and activate an alarm, no further use being made of the data, but this would be an extravagant waste of resources. Further use can always be made of water quality data for modelling, archiving and establishing average levels of determinands. The prime record in the monitoring station will probably be on one or several recorder charts and data in this form are not suited to further analysis other than by simple visual observation. The manual transcription of data from recorder charts or even tally roll printers can be extremely tedious and prone to error. If the monitor is equipped with an efficient means of data logging the analytical results can be recorded onto some storage medium and when convenient these data can be recovered and used with a data processing system to obtain graphs, tables, corrected figures and archive data as required. Various data recording mediums are available for use with monitoring systems and the situation is one which may rapidly change in the light of modern electronic developments.

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.

7.27. CHART RECORDERS Data as commonly presented on chart recorders are satisfactory for continual surveillance, but often are not suitable to give a summary of results over, say, a week. A common use of magnetic tape logged data is to turn this back into a chart with more reasonable co-ordinates than the original record. It would be useful, therefore, if chart

Water quality monitoring

185

recorder manufacturers could produce a recorder with charts of a convenient size (say A4) with a grid marked in real time (and days if required) so that each week an easily filable graphical output of results was obtained. Circular chart recorders approach this idea, as do some flow and meteorological recorders, but there does not appear to be an electrical input version with a rectangular format. Such a recorder may remove some people’s requirement for data logging.

7.28. INTELLIGENT INSTRUMENTATION The well publicised current growth in electronic technology is both increasing instrument reliability and making available at low cost sophisticated data handling systems.23 These developments will have significant effects on water quality monitoring instrumentation in two respects. First, it will be possible to have an intelligent device controlling and collating individual sensor units and in overall control of the analytical system functioning. The system would control routine operation and maintenance and such functions as sampling, calibration, cleaning and data collection. The intelligence of the system would enable it to examine data from its sensor units and evaluate this. It could query readings from sensors and perform cleaning, zeroing and re-calibration steps before accepting the sensor reading as valid and perhaps initiating an alarm or controlling the abstraction process. Further, it could cross-correlate data from different sensors and detect possible sensor malfunctions. An example might be a dramatic change in pH; this would be expected to be accompanied by a change in conductivity but if this was not the case the system could self test and if operating correctly alert the user to this observed state of affairs. A master controller of this type could also undertake data processing and reduction tasks, perhaps even producing hard copy summaries of measured data. The other impact of increasingly intelligent devices will be that each individual instrument or sensor system will contain a unit of this type. (This may also be a lower cost option than current non-intelligent analogue or discrete logic technology.) The instrument will be able to perform calibration, curve correction and data processing by itself. It will then present a processed result which may be printed out or passed to some central data system. It follows that as instruments are becoming increasingly able to provide data for collation by a central system a practical universal system of instrumentation interconnection is developed and implemented by manufacturers and users.

7.29. INSTRUMENTATION BUSES As instrumentation becomes increasingly intelligent and we move towards intelligently controlled monitoring stations, there is a requirement for increasing ease of interchange of data between instruments, controllers and data handling systems. The concept of the data bus ideally suits itself to a monitoring station. All instruments and data recording devices together with a station controller would ideally be plugged into identical sockets on a common bus cable, similar in concept to a domestic ring main. The controller

Developments in water treatment—2

186

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

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

REFERENCES 1. SKEGGS, L.T., Amer. J. Clin. Pathol, 1957, 28, 311. 2. RUZICKA, J. and HANSEN, E.H., Analytica Chim. Acta, 1975, 78, 145–157. 3. STRICKLER, H.S.. STANCHAK, P.J. and MAYDAK, J.J., Anal Chem., 1970, 42, 1576. 4. WALKER, W.H. C., Clin. Chim. Acta, 1971, 32, 305. 5. GERKE, J.R. and FERRARI, A., Technicon Symp., 1967, 1, 531. 6. RUZICKA, J., STEWART, J.W.B. and ZAGATTO, E.A., Analytica Chim. Acta, 1976, 81, 387– 396. 7. RUZICKA, J., HANSEN, E.H. and ZAGATTO, E.A., Analytica Chim. Acta, 1977, 88, 1–16. 8. HOLAK, Anal. Chem., 1969, 41, 1712. 9. FERNANDEZ, Perkin Elmer A.A.Newsletter, 1973 (July-Aug.), 12, No. 4. 10. POLUEKTOV, VITKUN, ZELYUKOVA, Zh. Anal Khim., 1964, 19, 873. 11. LEAHY, J.S.and PURVIS, M., J. Instn Water Engrs & Scientists, 1979, 33, 311. 12. MORLEY, P.J., MUSTY, P.R. and COPE, J., J. Instn Water Engrs & Scientists (in press). 13. NATIONAL RESEARCH DEVELOPMENT CORPORATION and MACKERETH, F.J.H., British Patents 19294(1962) and 12882(1963); MACKERETH, F.J.H., J. Sci. Instrum., 1964, 41, 38. 14. CLARK, L.C., Trans. Amer. Soc. Artificial Internal Organs, 1956/7, 2, 41–47. 15. Her Majesty’s Stationery Office, Methods for the Examination of Waters and Associated Materials, ‘The Measurement of Electrical Conductivity and the Laboratory Determination of the pH Value of Natural, Treated and Waste Waters’, 1978. 16. MOODY, E.J.and THOMAS, J.D. R., Selective Ion Sensitive Electrodes, Merrow Publishing Co. Ltd, Watford, 1971. DURST, R.A. (Ed.), Ion Selective Electrodes, National Bureau of Standards, Washington, D.C., 1969. BAILEY, P.L., Analysis with Ion-Selective Electrodes, Heyden & Son Ltd, London, 1976. WHITFIELD, M., Ion Selective Electrodes for the Analysis of Natural Waters, Australian Marine Sciences Association, Sydney, Handbook No.2, 1971.

Developments in water treatment—2

188

17. BRIGGS, R., J. Sci. Instrum., 1962, 39, 2–7. 18. HOLLAND, G.J., GREEN, A., STROUD, K.C.G. and JONES, D.B., Water Treatment & Exam., 1975, 24, 81–119. STENSEL, H.D., McDowELL, C.S. and RITTER, E.D., J, Water Pollut. Control Fed., 1976, 48, 2343–2350. MARSHALL, W.C., ‘Non-contact Sensing of the Electrical Field surrounding Trout’, M.Sc. Thesis, University of Wyoming, Laramie, Wyoming, 1972. POELS, C.L.M., Water Treatment & Exam., 1975, 24, 46–56. MILLER, W.F., ‘The Development of the Water Research Centre Water Quality Monitor using Fish’, Water Research Centre Seminar on Practical Aspects of Water Quality Monitoring Systems, Stevenage, 1977. 19. WALSH, A., ‘The Application of Atomic Absorption Spectra to Chemical Analysis’, Australian Patent 23041/53. WALSH, A., Spectrochimica Acta, 1955, 7, 108. THOMPSON, K.C. and REYNOLDS, R.J., Atomic Absorption, Fluorescence and Flame Emission Spectroscopy, Charles Griffin & Co. Ltd, London, 1978. 20. GREENFIELD, S., JONES, I.L. and BERRY, C.T., Analyst, 1964, 89, 713. DICKINSON, G.W. and FASSEL, V.A., Anal. Chem., 1969, 41, 1021–1024. 21. Post Office, International and Maritime Telecommunications Region, Rugby Radio Station, ‘Standard Frequency and Time Broadcasts, Rugby Radio Station’. CROSS, A.F., Wireless World, 1976 (Feb.) , 82, 30–35 et seq. HELSBY, N.C., Wireless World, 1976 (Aug.), 82, 47–51 et seq. 22. European Computer Manufacturers Association, Standard Nos. 34 and 46. 23. McARTHUR, N., WINGFIELD, A.J. and WITTEN, I.H., Wireless World, 1979 (Dec.), 85, 46– 51. 24. Institute of Electrical and Electronics Engineers, Standard IEEE 488–1975. American National Standards Institute (A.N.S.I.), Standard MC 1.1. WITTEN, I.H., Wireless World, 1979 (Feb.), 85, 32–35 et seq. COTTIS, R.A., New Electronics, 1979 (Dec.), 12, 70–74. 25. FEDIDA, S., ‘Viewdata—an Interactive Information Medium for the General Public using the Telephone Networks’, 6th International Broadcasting Convention, 20–24 Sept. 1976.

In addition to these references the reader will find much valuable information in manufacturers’ data and handbooks, in many of the standard texts on physics and physical chemistry, and in current journals and periodicals which are essential in order to keep in touch with the very rapid technological advances currently being made.

INDEX Absorption refrigeration cycles, 139 Acrylic-amine resins, 78 Activated carbon, 29–46, 64 adsorption of solute molecules onto, 30–2 biological, 158 granular, 30, 35–46, 158 adsorber installation, 39–41 adsorber system design, 36–9 applications, 46 performance, 41 points of application, 36 regeneration system design and operation, 43–5 powdered, 32–5 applications and performance, 35 dosing systems, 33–4 point of addition, 32 regeneration, 32–3 Adsorbents, 46–51 Adsorption, 28, 30–2, 36–41 Aluminium sulphate, 184 Ammonia, 59–63 removal, 63–74 biological nitrification, 67–74 breakpoint chlorination, 63–6 ion exchange resins, 66–7 physico-chemical methods, 66–7 Ammonia-chlorine process, 143, 144 Ammonium hydroxide, 81 Anion exchange resins, 46 Anionic detergents, 49 Atomic absorption spectrophotometry, 199, 214 Automation, 191 Bactericidal residuals, 154 Batch analysers, 191, 197–9 BCD (binary-coded-decimal) output, 203 Belgarde range of chemicals, 98 Belt pressing, 180 BET method, 35 Biological activated carbon (BAC), 158 Biological denitrification, 75–7, 81–6 Biological methods, 62–3

Index

190

Biological nitrification, 67–74 Biological sensors, 210 Bleaching powder, 146–7 Boiler feedwater treatment, 46, 119–23 ‘Breakpoint’ chlorination, 63–6, 144 Brine concentrate disposal, 126 Bromine, 142 Carbon activated. See Activated carbon total organic, 201 Carbonaceous adsorbents, 47 Carcinogens, 59, 158 Carman-Kozeny equation, 8–9, 175, 181 Cellulose acetate membranes, 52, 53, 107–9, 113, 121 Cellulosic ion exchange resins, 47 Centrifuging, 181–2 Chart recorders, 217 Chemical methods, 61 Chemical oxygen demand (COD), 196 Chemseps continuous loop, 79 Chloramines, 63, 141, 143, 154 Chloride determination, 197 Chlorination, 63–6, 143, 144, 152–4, 158 Chlorine, 65, 66, 141, 143–54, 158 available, 145, 146, 152 chemistry of, 147–52 dioxide, 29, 142, 145, 146, 154–6, 158, 160 chemical and bactericidal properties, 155–6 generation, 155 history, 154–5 dose-residual curve, 149 forms of, 145–7 history, 143 production, 146 residual, 152 solutions, 146 Chlorite, 158 Chloroform, 158 Clarification, 28, 142 Cleaning, 212–14 air, 213 automatic, 212 chemical, 213 mechanical, 212 ultrasonic, 212–13 Clinoptilolite, 66 Coagulant recovery, 185 replacement, 184–5 Coagulation, 28, 142, 165–6

Index

191

Colloidal material, 26 Colloids, fouling by, 122 Colour removal, 26 Compressibility index, 176 Conductivity, 210 Continuous distillation head, 195 Continuous flow analysers, 191–7, 211 Control of Pollution Act 1974, 168 Cross-contamination, 196 Cyanide determination, 195 Cysticidal residuals, 154 2,4-D, 53 D’Arcy equation, 175 Data collection, 202–4 logging, 214, 216–17 processing, 202–4 Demineralisation, 47, 74 Denitrification, 75–7, 81–6 Desalination, 89–140 miscellaneous processes, 138–9 multi-stage flash distillation, 90–106 reverse osmosis, 106–31 Dialysis unit, 193, 194 Dichloramine, 64, 65, 149, 160 Discrete analysers, 197–9 Disinfection, 141–62 agents used in, 141–2 by-products, 157–60 mixed treatments, 158 relative costs, 157 Distillation processes, 90–106 multiple effect, 92, 100 multi-stage flash. See Multi-stage flash (MSF) distillation Dosing systems, 33 DPD test, 160 Drying beds, 178–9 ‘Ducol’ process, 78 Dust control, 33 Electrodialysis, 131–8 capacity installed, 135 costs, 136–8 description of process, 131–6 polarity reversal, 135 principle of, 132 size, 134 systems available, 132 Emission spectroscopy, 201 Energy

Index

consumption, 94 recovery, 124 Enzyme mechanisms, 74 Ethyl cellulose membranes, 52 Evaporation, 179 Evaporator costs, 102–6 materials, 100–2 Faraday’s Law, 132 Feed pumps, 124 Ferrous hydroxide, 74 Filterability index, 23 Filter (s), 158, 165 backwashing, 20–1, 169 cleaning, 11 continuous screen, 123 control, 19 conventional, 4 ‘dry’ sand, 72–3 hydraulic control, 20 Immedium, 15–16 L’Eau Claire, 16–17 micron, 123 pilot-scale, 22 pressing, 180–1 radial flow, 17–18 rapid, 5–6 rotary vacuum, 179–80 slow, 3–5, 19 slow sand, 26, 28, 165 Tenten, 18 types of media, 13 upflow, 15–17 Filtration, 1–24, 142, 169–71, 176 applications, 2–3 attachment mechanisms, 8 conventional, 3 design methods, 21–3 dual and multi media beds, 13 effective use of bed capacity, 12–18 head loss during, 9 relationships, 14 hydraulics, 8–10 Filtr ation—con td. mathematical modelling, 10, 22 nature of suspension to be treated, 23 process details, 6–12 modifications, 18–19

192

Index

193

solids removal, 10 transport mechanisms, 7 trends and developments, 12–23 Fish behaviour, 210 Fishmoor plant, 171 Flameless atomic absorption, 199 Flash chambers, 101–2 Flotation, 183–4 Fluidised bed MSF, 98–100 system, 71 Fluidised sand units, 81–6 Fouling colloids, by, 122 index test, 121 Freeze-thaw process, 182 Freezing processes, 91, 138–9, 182 Freundlich isotherm, 30 Gas chromatography with mass spectrometry (GCMS), 202 Granular carbon. See Activated carbon Groundwaters, 122 Heat-pumping, 138 Hewlett Packard Interface Bus, 203 High Test Hypochlorite, 146 Humic acids, 49 Hydrochloric acid, 145, 147 Hypochlorites, 146, 147 Hypochlorous acid, 145, 147 Immedium filters, 15–16 Immobilised enzymes, 74 Instrumentation, 191, 218 buses, 218–19 Iodine, 142 Ion exchange resins applications, 49 desalination, 91 evaluation methods, 48–9 nitrate removal, 75–81 nitrogen compounds, 61, 66–7 organics removal, 46–51 test results, 49 theory, 48 Ion selective electrodes, 208–9 Iron (III) thiocyanate, 197 Kozeny equation, 8

Index

Lamella sedimentation, 184 LaMotte test kit, 161 Langelier Index, 120 L’Eau Claire filter, 16–17 Lindane, 53 Membrane(s) composite, 109 configuration, 54, 109 hollow fibre, 112 materials, 52–3 performance characteristics, 111 processes, 106, 131 product flux, 112 types, 107, 109–15 Metal analysis, 199–201 detection limits, 201 Methaemoglobinaemia, 59 Methanol, 83, 84, 86 Microstrainers, 165 Monitoring, 189–222 automatic, 204–5 real time, 214–15 station design, 213 Monitors basic types, 205 design considerations, 205 submersible, 205–7 Monochloramine, 145, 146, 160 Monochromator, 201 Multiple effect distillation, 92, 100 Multi-stage flash (MSF) distillation, 90–106 evaporator costs, 102–6 fluidised bed, 98–100 Nitrate, 59–61 removal, 74–86 biological denitrification, 75–7, 81–6 ion exchange, resins, 75–81 possible methods, 74–7 Nitrification, 67–74 Nitrogen, 65 compounds, 59–88 trichloride, 64, 65, 149, 160 Nitrosamines, 59 Odorous substances, 30 Ohm’s Law, 132 Organic compounds, 25–7 health effects, 26

194

Index

objectives or removal, 26 oxidative processes, 29 removal by conventional processes, 26 water, in, 25–6 Organic contaminant monitoring, 201–2 Oxidation-reduction potential (ORP), 209 Oxidative processes, 29 Oxidising agents, 29 Oxygen, dissolved, 207–8 Ozone, 29, 142, 156–8 application, 156–7 applied doses, 157 generation, 156–7 history, 156 Permanganate Value (PV), 28 Permasep polyamide, 118 Pesticides, 28, 53 pH effects chlorination, 148, 153 chlorine dioxide disinfection, 156 desalination, 123 monitoring, 208 phenol uptake, 49 residual chlorine, 152 reverse osmosis, 52 Phenol determination, 195 uptake, 49 Phosphoric acid, 195 Physico-chemical methods, 61, 66–7 Plasma emission spectroscopy, 214 Pollution, 207, 211 Polyamide, 107, 114, 118 Polyelectrolytes, 170, 172, 173, 181, 182, 185 Polymeric resins, 47 Polynuclear aromatic hydrocarbons (PAH), 28, 52 Polysulphone (TFC), 107, 111, 113, 115 Potassium iodide, 145 permanganate, 29 Precipitation, 28 softening, 166 Rate constant, 70 Real time, 214–15 Redox potential, 209 Reference electrodes, 209 Reverse osmosis, 51–4 basic principles, 107 costs, 131

195

Index

196

desalination, 90, 106–31 disposal of brine concentrate, 126 feed pumps, 124 instrumentation, 126 nitrogen compounds, 61 organic compounds, 51–4 plant developments and installations, 127–31 post-treatment of product water, 126 systems design, 123–7 Ring-type electrodes, 210 Samplers, 207 Sand-washing plants, 165 Scale control, 96–8, 119 Sedimentation, 28, 142, 171, 184 Sensors, 204, 210 Silting Density Index (SDI), 121 Sirotherm ion exchange process, 138 Sludge, 163–88 coagulant-based, 181 conditioning, 178 dewatering, 168–70 capital costs, 168 further methods, 178–82 theory, 175–8 estimation of, 186 lime addition, 181 primary sedimentation, 165 produced by coagulation, 165–6 iron and manganese removal plants, 166 precipitation softening, 166 produced from chemical treatment process, 165–6 produced without chemical treatment, 165 thickening, 170–5 treatment operational aspects, 186–7 optimisation process, 183–6 types of, 164–5 use of term, 164 waterworks, 168–70 Sodium bisulphite, 153 chloride, 78 sulphite, 153 thiosulphate, 153 Solar distillation, 91 Solar energy, 138 Specific resistance, 176 Sterilisation, 213 Stocks plant, 173

Index

Styrene-divinylbenzene resins, 78 Sulphur dioxide, 153 Suspended solids, 209–10 Telemetry, 215 Temperature measurement, 211–12 Tenten filter, 18 Tintometer Lovibond test kit, 160 Trace analysis, 191 Transport mechanisms, 7 Trihalomethanes (THM), 65–6 Ultrafiltration, 54 Ultraviolet radiation, 142 Vapour compression (VC) processes, 92 Viewdata, 219–20 Wastewater recycling, 164 Water-boxes, 101 Water Research Centre, 175 ‘Wedge-wire’, 178 Zeta potential, 122

197