Water, Wastewater,

and
Sludge
Filtration

Editor

Saravanamuthu Vigneswaran, D.Eng., Dr.Sc.

Associate Professor
Environmental Engineering Division
Asian Institute of Technology
Bangkok
Thailand

Co-Editor

Roger Ben Aim, Dr.Sc.

Professor and Director
Institut du Genie des Procedes Agro-Alimentaires
Agen
France

CRC Press, Inc.

Boca Raton, Florida
Library of Congress Cataloging-in-Publication Data

Water, wastewater, and sludge filtration / editor, Saravanamuthu

Vigneswaran ; co-editor, Roger Ben Aim.
p. cm.
Includes bibliographies and index.
ISBN 0-8493-6983-5
1. Water—Purification—Filtration. 2. Sewage—Purification—
Filtration. 3. Filters and filtration. I. Vigneswaran,
Saravanamuthu, 1952- . II. Ben-Aim, Roger.
TD441.W37 1989 88-9505
628.3— del 9 CIP

This book represents information obtained from authentic and highly regarded sources. Reprinted material is
quoted with permission, and sources are indicated. A wide variety of references are listed. Every reasonable effort
has been made to give reliable data and information, but the author and the publisher cannot assume responsibility
for the validity of all materials or for the consequences of their use.

All rights reserved. This book, or any parts thereof, may not be reproduced in any form without written consent
from the publisher.

Direct all inquiries to CRC Press, Inc., 2000 Corporate Blvd., N.W., Boca Raton, Florida, 33431.

© 1989 by CRC Press, Inc.

International Standard Book Number 0-8493-6983-5

Library of Congress Card Number 88-9505

Printed in the United States
 PREFACE

This work gives the readers some information on the various solid/liquid separation
processes available for water and wastewater treatment. It will be especially useful to
practicing engineers and technologists involved in treating water and wastewater in that they
will find this book to be a comprehensive review of the various treatment processes, with
their corresponding general/specialized application in water and wastewater treatment.
Since there are a number of books written by eminent authors on process fundamentals
and mechanisms, this treatise concentrates more on the influence of the process variables
on the treatment efficiency and on the various types and field of application of these processes.
In order to obtain a high level of expertise in each of the subject areas covered, we invited
experts in the field to contribute chapters focusing on their area of interest. Each chapter
has been written by an author who has actively contributed to the present state of the art of
the process discussed in the chapter.
This book consists of 14 chapters arranged in a sequence that reflects the current stage
of development. That is to say, deep bed filtration — a method evolved at an earlier period
to treat water — is dealt with in the first four chapters. Chapters 1 and 2 give a brief review
of the modifications which have occurred over the years in conventional deep bed filtration,
along with theoretical approaches. Whereas the major breakthrough in conventional deep
bed filtration — direct filtration with its specific applications, is described in Chapter 3,
Chapter 4 will be of special interest to wastewater treatment plant engineers, in that it deals
with the application of deep bed filtration to treatment of different types of wastewater. It
is a comparatively new phenomenon and should be a valuable addition to the technical data
base on the above-mentioned subject.
Other treatment processes, such as microstraining, cartridge filtration, and precoat filtra­
tion, which were developed at a later stage and used for pretreatment and for specific treatment
purposes, are considered in the next three chapters.
Membrane filtration processes, a gift from chemical engineering to environmental engi­
neering, which can be used to remove a wide range of dissolved particles (molecule to
submicron level) are being currently used to recover valuable metals and to obtain a high-
quality water. Chapters 8 through 11 discuss all membrane processes, namely, reverse
osmosis, electrodialysis, ultrafiltration, and microfiltration, with its specific applications in
water and wastewater treatment.
Chapters 12 through 14 discuss three different sludge dewatering methods commonly
used, namely, vacuum filtration, pressure filtration, and centrifugation. Though sludge
treatment is usually the last operation in any water or wastewater treatment plant, never­
theless, it is a very important operation. Its importance can be gauged, when one considers
that all the contaminants removed from the treated water/wastewater are accumulated as the
sludge and have to be treated and safely disposed of. Widely discussed sludge dewatering
methods such as vacuum filtration and pressure filtration are treated in a general sense, while
centrifugation is detailed in a rather comprehensive manner.
We are indebted to Professor R. Gimbel of the University of Duisburg, S. Ripperger of
AKZO, Wuppertal, West Germany; Dr. L. Coccagna of Culligan Italiana S.p.A., Italy;
Professor K. Fujita of the University of Tokyo, Japan; Mr. P. A. Jackson of E. Beaudrey
and Co., France; Dr. R. Illner of Manville de France, S.A.; Professor R. Audinos of the
Universite Paul Sabatier, Toulouse, France; and Professor C. Alt of the University of
Stuttgart, West Germany, who have contributed to this book. We thank Professor A. Rushton
of the University of Manchester Institute of Science and Technology, Manchester, U.K.,
for his kind review of the chapter on vacuum filtration. We also thank those who have
kindly consented to the reproduction of their figures and tables in this work.
Special thanks are to Mr. S. Bhuvendralingam, Mr. F. Rahman, Mr. S. Kugaprasatham,
and Mr. V. Balakrishnan for their kind help in completing this book. Mrs. R. Siengsukon
and Mrs. P. Sthapitanonda contributed much to the completing of the book through their
typing of the manuscript.

S. Vigneswaran
R. Ben Aim
 THE EDITORS

Dr. Saravanamuthu Vigneswaran is presently an Associate Professor in the Division

of Environmental Engineering of the Asian Institute of Technology, Bangkok, Thailand.
Dr. Saravanamuthu Vigneswaran was graduated in 1975 from the University of Sri Lanka,
Peradeniya Campus with a B.Sc. honors degree in chemistry and obtained the M.Sc. degree
in environmental engineering in 1978 from the Asian Istitute of Technology. He was awarded
the degree of Docteur Ingenieur in 1980 and Docteur es Sciences degree in 1987 in the field
of chemical engineering from the Universite de Montpellier, France and Institut National
de Poly technique de Toulouse, France, respectively.
He has published more than 60 research papers. His current interests are solid-liquid
separation techniques in water and wastewater treatment, such as deep bed filtration and
membrane processes. He also works in the area of pollutant transport in subsurface envi­
ronment.

Dr. Roger Ben Aim is Professor of Chemical Engineering at the Institut National
Polytechnique de Toulouse, France. Dr. Roger Ben Aim obtained his engineering degree
and degree of Docteur es Sciences from the Ecole Nationale Superieure des Industries
Chimiques, Nancy, France, in 1961 and 1970, respectively.
In 1972, Professor Ben Aim participated in the creation at the University of Montpellier
of a water engineering department and of a research laboratory on chemical engineering
applied to water and wastewater treatment. He was head of this laboratory untill 1980.
Professor Roger Ben Aim also established two technical research centers in Agen, France,
namely, the Institut de la Filration et des Techniques Separatives (IFTS) and the Institut du
Genie des Procedes Agro-Alimentaires (IGEPA). He was also the Director of IFTS from
1981 to 1985 before assuming the position of director in IGEPA, which is affiliated with
the Institut National Poly technique de Toulouse.
Professor Ben Aim has published widely in the field of solid-liquid separation techniques
in water and wastewater treatment and presented numerous lectures at national and inter­
national meetings at universities and institutes.
 CONTRIBUTORS

Christian Alt, Dr.-Ing. Rolf Gimbel, Dr.-Ing. habil.

Professor in Ordinary Emeritus Institute of Water Technology
Department of Chemical Engineering Univesity of Duisburg
University of Stuttgart Duisburg, West Germany
Stuttgart, West Germany
R. Illner
Remy Audinos, Dr. Sc.
Group Plant Manager
Professor
Manville de France
Department of Chemical Engineering
Saint Cloud, France
Paul Sabatier University
Toulouse, France
E. P. Jackson, Engenieur ECP
Roger Ben Aim, Dr.Sc. General Manager
Professor and Director E. Beaudrey & Cie
Institut du Genie des Procedes Agro- Paris, France
Alimentaires
Agen, France
Siegfried Ripperger, Dr.-Ing.
Manager of Development and Application
Luciano Coccagna
Technical Membranes
R & D Manager
AKZO, Enka AG
Culligan Italiana, S. p. A.
Wuppertal, West Germany
Bologna, Italy

Kenji Fujita, Ph.D. Saravanamuthu Vigneswaran, D.Eng.,

Professor Dr. Sc.
Department of Urban and Sanitary Associate Professor
Engineering Environmental Engineering Division
University of Tokyo Asian Institute of Technology
Tokyo, Japan Bangkok, Thailand
 TABLE OF CONTENTS

Chapter 1
Overview of Deep Bed Filtration: Different Types and Mathematical M odels................... 1
Saravanamuthu Vigneswaran and Roger Ben Aim

Chapter 2
Theoretical Approach of Deep Bed Filtration...........................................................................17
Rolf Gimbel

Chapter 3
Direct Filtration............................................................................................................................57
Luciano Coccagna

Chapter 4
Applications of Deep Bed Filtration in Wastewater Treatm ent............................................77
Kenji Fujita

Chapter 5
Microstraining..............................................................................................................................101
E. P. Jackson

Chapter 6
Precoat Filtration........................................................................................................................ 117
R. Illner

Chapter 7
Cartridge Filtration..................................................................................................................... 129
Saravanamuthu Vigneswaran

Chapter 8
Reverse O sm osis.........................................................................................................................139
Saravanamuthu Vigneswaran

Chapter 9
Ultrafiltration............................................................................................................................... 159
Saravanamuthu Vigneswaran

Chapter 10
Microfiltration..............................................................................................................................173
Siegfried Ripperger

Chapter 11
Electrodialysis..............................................................................................................................191
Remy Audinos and Saravanamuthu Vigneswaran

Chapter 12
Vacuum Filtration...................................................................................................................... 225
Saravanamuthu Vigneswaran
Chapter 13
Pressure Filtration...................................................................................................................... 237
Saravanamuthu Vigneswaran

Chapter 14
Centrifuges for Sludge Treatm ent........................................................................................... 249
Christian Alt

Index.............................................................................................................................................275
 1

Chapter 1

OVERVIEW OF DEEP BED FILTRATION: DIFFERENT TYPES AND

MATHEMATICAL MODELS

Saravanamuthu Vigneswaran and Roger Ben Aim

TABLE OF CONTENTS

I. Introduction.......................................................................................................................... 2

II. Improvements on Rapid F ilte r.......................................................................................... 2

A. Improvements on Filter M e d ia...........................................................................2
B. Improvements on Flow Rate................................................................................2
C. Elimination of Some Operationsfrom Conventional Water
Treatm ent.............................................................................................................. 3

III. Filter Backwash M ethods.................................................................................................. 4

A. Choice of Backwash M ethod............................................................................... 4
B. Backwashing with Effluent from Other Filter U n its........................................5

IV. Optimization of Filter Design...........................................................................................7

V. Mathematical M odels........................................................................................................ 9
A. Filtrate Q uality....................................................................................................... 9
B. Headloss................................................................................................................. 11
1. Clean B e d .................................................................................................11
2. Clogged B ed............................................................................................. 12

References 14
2 Water, Wastewater, and Sludge Filtration

I. INTRODUCTION

Filtration technologies are classified under two major categories, depending mainly on
the mode of filtration: slow sand filtration and rapid sand filtration. Slow sand filter, which
includes biological activity in addition to physical and chemical mechanisms for removing
impurities from the raw water, is especially suitable for small community water supplies,
because of its large areal requirement. Numerous documents are available on this technol-
ogy . 1' 7
Rapid filter, on the other hand, due to its lower areal requirement (25 to 150 times less
than slow sand filter) is used widely as a final clarification unit in municipal water treatment
plants. It is becoming increasingly important in wastewater treatment, particularly when
water reuse is envisaged. The applications of filtration in wastewater are discussed in detail
in Chapter 4.

II. IMPROVEMENTS ON RAPID FILTER

A. Improvements on Filter Media

The conventional rapid filter generally uses sand with an effective size of 0.6 mm and a
uniformity coefficient of 1.5 to 2. This results in stratification of filter medium after the
backwash, in which the finer medium remains at the top and the coarser medium at the
bottom of the filter bed. To overcome this problem, two alternatives have been proposed:
dual or multimedia filtration; or coarse size, narrowly graded media filtration.
In dual-media filtration, the size and specific gravity are carefully selected to minimize
intermixing. The commonly used media are anthracite coal and sand. Various research
workers have given optimum size ratios to avoid intermixing (Table 1). Extensive research
has been carried out to study the advantages and disadvantages of intermixing of filter media
in dual media filtration . 1315 While a group of researchers feel that the grain size of course
anthracite and fine sand should be chosen in such a way that the intermixing at the interface
is minimized, others believe that controlled mixing among filter media is beneficial. A
detailed design of media is discussed in the literature . 16,17
Another alternative is to replace a graded single medium with a coarse, narrowly graded
medium of larger depth. This arrangement results in deeper penetration of the suspended
solids and thus higher storage capacities which would lead to a longer run. The selection
of size depends on the raw water and required effluent qualities. The commonly used size
range is 0.9 to 1.1 mm. This arrangement can meet an increase in demand in existing units
because it can be operated at a higher filtration rate. However, deeper penetration of the
solids would entail a higher backwashing requirement. The rate of air and water used for
backwashing depends on the size of medium. For example, sand of 2 mm effective size
requires a washing rate of 90 to 110 m 3/m2*h of air and 19 to 24 m 3/m2*h of water.

B. Improvements on Flow Rate

Conventional rapid filtration operates at a constant rate of approximately 5 m 3/m 2’h.
Research work on the variation of flow rate has indicated that high-rate filtration and
declining-rate filtration are advantageous for most cases. If one achieves the desired filtrate
quality with a higher filter rate at an operational and maintenance cost comparable to that
of a conventional rapid filter, then one could achieve a significant capital saving by using
a high-rate filter. Similarly, research has indicated that declining-rate filtration produces a
better effluent quality than the conventional process. Therefore, if one could obtain the same
amount of filtered water with equal capital investments, then declining-rate filtration would
have a definite advantage over conventional rapid filtration. Another advantage of using
declining-rate filtration is that it does not require automatic rate control.
 3

Table 1
CALCULATED SIZE RATIOS TO
AVOID MIXING

Calculated effective
size ratio
(anthracite: sand) Comment Ref.

2.38 : 1 Neglects settling, 8

based on laminar
flow backwash
2.63 : 1 Neglects backwash- 9
ing, considers hind­
ered settling
2.73 : 1 Neglects backwash- 10
ing, considers hind­
ered settling
3.00 : 1 Based on experience 11
2.07 : 1 Based on pilot-scale 12
study

The concept of declining-rate filtration is not new. Basically, no rate controller is used
in this system, and instead it is replaced by a fixed orifice. The filtration rate in this system
is allowed to decline from a maximum value at the beginning of the run, when the filter is
clean, to a minimum value at the end of the filter run, when the filter is in need of
backwashing. In practice, several (a minimum of four) filters are used in series, and the
water level is maintained essentially at the same level in all operating filters at all times.
This is achieved by providing a relatively large influent header pipe or channel common to
all the filters, with a relatively large influent valve or gate to each individual filter. Details
on the declining-rate filter operational principles, design criteria, and plant operations in
developing countries can be found elsewhere . 17 21
High-rate filtration in which the filtration rate is about 10 to 20 m 3/m 2*h, as compared to
the rate of conventional rapid sand filtration which is of the order of 5.0 m 3/m 2*h, is useful
in upgrading existing plants. Such high filtration rates are possible, thanks to the development
of (1) dual-media or multimedia and (2) control of flocculation by polyelectrolytes. This
process uses a dual or mixed-media bed, while maintaining the required effluent quality
with the entire bed being used efficiently for effective filtration action.
High-rate filters with dual or coarse-medium arrangements have been successfully used
in the West, and the accumulated experience, as well as specific data, supports the concept
that further application of these processes is warranted . 22 This could be one of the economic
solutions for the expansion of existing water treatment plants and for the construction of
new plants. In order to fully exploit the economic benefits, filters should be designed in
order to operate at the highest practical rate, being economical at the same time, though
washing must be done more frequently . 23 Here, more attention should be paid to the selection
of the filter media and filtration rate so that the filtrate quality meets the required standard.

C. Elimination of Some Operations from Conventional Water Treatment

Conventional water treatment plants generally use the following unit operations: rapid
mixing, flocculation, sedimentation, filtration, and disinfection. Depending on the quality
of the water, one or more unit operations can be eliminated, thereby achieving a cost-
effective water treatment. Direct filtration falls into the above category. Filters used in direct
filtration thus differ little in construction from those for conventional treatment. The primary
difference in the operation of the two systems is related to solids storage capacity and
backwashing requirements.
4 Water, Wastewater, and Sludge Filtration

Direct filtration was first explored during the early 1900s, but these attempts were not
successful, due to the rapid clogging of the sand beds. The development of coarse-sand
filters has made it possible to store greater amounts of floe within the filter bed without
excessive headloss, and has thus increased the feasibility of the direct filtration process.
Further advances in filter design and the availability of a wider selection of chemical co­
agulants have resulted in a variety of filtration systems being designed in which coagulating
chemicals are employed. The flocculation basin is either eliminated or reduced in size, and
the sedimentation basin is not utilized. Such processes thus have only screening, rapid
mixing, coagulation, and flocculation prior to filtration. All suspended solids and floes
formed are deposited in the filter, which is usually a multi-media, granular bed containing
coal, sand, and perhaps other constituent media.
The American Water Works Association (AWWA) Filtration Committee’s report24 on a
worldwide survey of 70 operating and pilot plants has indicated that waters with less than
40 units of color, turbidity consistently below 5 formazine turbidity units (FTU), iron and
manganese concentrations less than 0.3 mg/€ and 0.05 mg/€, respectively, and algal counts
of up to 2 0 0 0 per m€ (measured in absorption units at 1 0 0 0 nm) appear to be perfect
candidates for direct filtration. Turbidity and color removals are consistently attained in this
process. By efficient postchlorination, bacteria and virus removal problems can be elimi­
nated. Most of the literature favors the use of dual or mixed media for direct filtration . 25,26
Direct filtration can be successfully used for low -turbid waters, because of its lower capital
and operational cost. It does not require any sophisticated equipment, although skilled
operators are needed in order to monitor the filters. Attention should be paid to the possibility
of poor bacteriological quality of the filters due to badly polluted raw water. Details on
direct filtration with its applications are discussed in Chapter 3.

III. FILTER BACKWASH METHODS

In a filter operation, suspended solids become clogged in the filter bed. This phenomenon
leads to development of headloss in the filter unit. When the headloss reaches the maximum
allowable limit, or when the effluent quality deteriorates below the required quality level,
the filter run should be stopped and the bed should be cleaned. This filter-cleaning operation
is done by the filter backwash method.
Various methods of backwashing exist. They include: (1) high-rate backwash with water
alone; (2) low-rate backwash with water alone; (3) water backwash with surface-wash
auxiliary; (4) water backwash with air auxiliary; and (5) backwashing with the effluent from
other units. An additional method is to take washwater from the high-pressure distribution
systems. This method wastes energy, but results in low installation costs. A pressure-reducing
valve is normally required so as not to blow out the filter.

A. Choice of Backwash Method

The past experiences in the filter backwashing show that the choice of a backwash method
partly depends on the type of filter medium used. The different factors influencing the
backwash effectiveness in a filter unit are

1. Media size — coarse media will behave differently, depending on the backwashing
method employed.
2. Media shape — rounded grains are generally thought to be easier to clean than angular
or flat grains.
3. Media density — denser material needs higher velocities to suspend it in the upflow.
4. Water quality — different waters behave differently in mud-ball formation and in the
attachment of particles to the grains.
 5

5. Coagulant used — the amount and type of coagulant used: metallic coagulants or
poly electrolytes change the adhesiveness of the film formed around the grains. Weak
and strong floes will also behave differently with regard to ease of backwashing.

The first four backwashing methods listed above have been used extensively. The last
one looks more promising, due to its low capital and operational costs. Therefore, the first
four methods are only summarized in Table 2, which indicates their design criteria and
applicability, whereas the last method is discussed in detail.

B. Backwashing with Effluent from Other Filter Units

In this method of backwash, which results in “ inter-filter backwashing” filters, the filter
units of a treatment plant have to be interconnected as shown in Figure l . 36 The effluent
from the interconnected units is collected by a single drainage channel, and the filter effluent
outlet is located at a higher level than that of the washwater trough of the individual units.
So, one cleans a particular unit by closing the inlet and opening the drainage outlet of this
particular unit. The water level in the unit is thus lowered, and a positive head (Hb) is created
in this unit, which reverses the direction of the flow through the filter bed. The backwashing
will start with the effluent from the adjoining units that provide the upward flow. Once the
filter-cleaning operation is finished, the outlet (drainage) is closed, and the inlet of this unit
is opened to resume the filtration process.
Schulz and Okun37 briefly describe the design principles as presented below.
The head available for backwashing (Hb) is the difference in elevation between the effluent
weir and the gullet lip in the filter. To obtain sufficient head for backwashing, the depth of
the filter box must be substantially greater than for conventional filters. The filter box for
the plant in Cali, Colombia has a depth of 6 m, compared with a depth of about 3 m in
conventional filters. Also, filter bottoms must be designed with much lower headlosses than
conventional filters — only 20 to 30 cm of headloss compared with about 1 to 1.4 m for
conventional filters.
The filter bottom can be produced from concrete beams with plastic tube orifices inserted
along the beam. The spacing of the orifices dictates the headloss in the system. As long as
the underdrain systems are interconnected, the backwash velocities will be low enough in
the plenum so that the wash water distribution will be fairly uniform. Therefore, by increasing
the depth of the water over the filter beds to about 1.5 to 2.5 m, limiting the headloss in
the underdrain system to about 20 to 30 cm, interconnecting the underdrain system, and
using dual-media filter beds, the backwashing headloss can be made sufficient to produce
the desired expansion of the filter media.
Interfilter washing filtration units can be designed either with unrestricted or restricted
declining flow rate. The following precautions must be taken into account in designing this
mode of backwash:27

• For one filter to be washed with the flow of the others, the total production of the
plant must be at least equal to the wash-water flow needed to clean one filter.
• The filter units must supply enough water for the required backwash rates. A minimum
of four filter units, capable of working at a rate one third higher, is necessary to
minimize the peak flow produced when one unit is out of service for washing.
• The filters must be so designed that one may be taken out of service for repairs without
interruption of the normal operation of the others.
• The underdrain must be specially designed to produce low headloss. This is feasible
because the filters are completely open at the bottom, and the wash-water flow rate
is, therefore, very low.
 Table 2
RECOMMENDED DESIGN VALUES OF VARIOUS BACKWASH METHODS

Low-rate water backwash Water backwash with air auxiliary

Simultaneous air
and low-rate Simultaneous air
water backwash, and low-rate water
Air scour followed followed by Air scour followed backwash, followed Water backwash with
High-rate water by low-rate water low-rate water by high-rate water by high-rate surface-wash
Parameters backwash272’9 backwash2730 backwash2730 33 backwash27 30 34 backwash33 auxiliary82
’ 7 35

Backwash rate >37.5 m3/m2*h 18 m3/m2*h 15— 18 m3/m2*h >18 m3/m2*h 15— 18 m3/m2h-
Pressured backwash 2.5 — 5 kg/cm2 2.5—5 kg/cm2 2.5—5 kg/cm2 2.5— 5 kg/cm2 2.5— 5 kg/cm2 0.25— 0.5 kg/cm2
water
Air scour rate 27 m3/m2h
- 18— 27 m3/m2*h 27 m3/m2h
- 36— 46 m3/m2'h
Surface wash rate 10— 12 m3/m2*h
Pressure of surface 1.5— 4 kg/cm2
scour water
Porosity range during 0.68— 0.7 This type of filter back-
Water, Wastewater, and Sludge Filtration

expansion washing is used when

mud-ball formation oc­
curs on the top of filter
bed
Expansion of medium 80— 100% Low Low High High
Time of washing 3— 6 min 3— 6 min 2— 3 min 3— 4 min 2— 3 min
Time of air scour 3— 6 min 2— 3 min 3— 4 min 2— 3min
application
Amount of washwater High Low Low High High High
needed
Efficiency of cleaning Low Fair Good Good Good
action
Applicability Single and multi- Single media only Single media only Single and multi- Single and multi- Single media filters
media filters media filters media filters
 7

■DRAIN VALVE
FALSE BOTTOM

FIGURE 1. Backwashing of one filter with the flow of the others.

This system can be used for both single- and multimedia filters, but it requires four or
more filter units in order to operate effectively. The advantages of this system are as follows:

1. This filter is easier to build than conventional filters. Only two valves are needed for
filter control; the entire system can be designed with concrete channels or box conduits;
and it is possible to eliminate the elaborate piping, valves, and control systems common
to conventional filtration schemes. 37 There is no need for headloss gauges (since the
headloss is evident to the operator, who can observe the water level in the filters),
flow -rate controllers, washing equipment, or pipe gallery. Capital and maintenance
costs can, therefore, be considerably reduced.
2. There is a minimum of mechanization. As a result, the system is simple in design,
operation, and maintenance.
3. The backwash water is applied to the bed, using the head development in the unit, so
one does not need to pump water into the bed. This leads to a reduction in the capital
and operational costs.
4. When one filter is taken out of service for backwashing, the filtration rate variations
are slow and smooth. Once the headloss is fixed, the washing starts very slowly, and,
therefore, a sudden expansion of the bed is prevented.

However, for a proper cleaning operation one requires higher headloss (55 to 80 cm). To
create this headloss in the unit, the height of the freeboard of the filter unit has to be
increased. This leads to an increase in the construction cost of the filter unit.
Filter backwashing with this method has been practiced in Australia for a long time, and
has been successfully used in more than 100 installations in the U .S . 36 Filters of this kind
have also been operating satisfactorily in large plants in Latin America, including those
serving the cities of Mexico City (24 m3/s); Monterrey, Mexico (24 m 3/s); Rio Grande,
Brazil ( 6 m 3/s); and Cali, Colombia (4 m 3/s), as well as in Peru, Bolivia, and the Dominican
Republic. 3637

IV. OPTIMIZATION OF FILTER DESIGN

An optimum is achieved when the filter design and operation cause the filter to reach its
headloss limit at the same time as the filtrate quality deteriorates to an unacceptable value
(Figure 2 ) . 38 This is achieved when the line relating time of run to filter depth for a given
headloss limit intersects a similar line for a given filtrate quality (Figure 3). The shapes of
Water, Wastewater, and Sludge Filtration

CONCENTRATION C
FIGURE 2. Optimum operation concept.

DEPTH L

FIGURE 3. Optimum filter run and depth for given conditions.

 9

these lines depend on which mathematical models are used, but the principle is valid for all
models. The mathematical models are used with experimental data obtained with laboratory -
or pilot-scale filters at specified conditions for optimization of filter design.

V. MATHEMATICAL MODELS

Rapid filtration is a dynamic and complex process where particle retention is a function
of filter depth and filtration time. The filter performance is generally characterized by the
filtrate quality and headloss across the filter bed. These two factors depend on:

Size and nature of filter medium

Size distribution and nature of particles in suspension
Depth and porosity of bed
Filtration rate
Concentration of suspension

The dependence of filter performance on chemical and surface characteristics of particles,

filter medium, and flocculent makes the process difficult to model. This is the reason that
most of the models are semiempirical in nature. Numerous mathematical models have been
put forward to explain filtration behavior at various stages, and they can, in general, be
classified into two groups: microscopic and macroscopic. In this chapter, only a brief
summary of selected macroscopic models used to predict concentration and headloss profiles
is presented. The mathematical models in detail are discussed in Chapter 2.

A. Filtrate Quality
The first mathematical model is based on material balance and kinetic equations . 39,40 The
first-order kinetics assumed (Equation 1) is then verified from experimental results.

— = -A.C (1)
dZ

where C is the local concentration of suspension (v/v), Z is the bed depth (L), and X is the
filtration coefficient (L - 1)*
The general form of material balance of particles in an element of filter of depth (AZ)
and a cross-sectional area of H at time t can be written as :41
Accumulation rate = rate of variation in flow of particles

[Wo- + eC)] + ^ - HD = 0 (2)

where H a AZ is the volume of particles retained, fleCAZ is the volume of particles in motion
entrained by liquid, UmC is the particle flow entrained by the fluid, — is the diffusional
flux of particles, e is the porosity of clogged bed (e = e0 - (3a), e0 is the clean bed porosity,
actual volume of deposit 1 .
[ ------------- — --------­ —-----­ r , and a is the specific
compacted volume of deposit J
deposit (v/v).
The above equation can be simplified by using the following hypotheses, which are
applicable in the case of deep-bed filtration.
• Velocity of suspension is constant throughout the filter run.
10 Water, Wastewater, and Sludge Filtration

Table 3
k VS. a RELATIONSHIPS

Type of
suspension Model Ref.

Discrete (D) X = X0 + c'a 27

D X = Xn — c'a 30

dC aq
Flocculant (F) — = be - 31
dZ c

D X = X0(l - a /a u) 32

D (i) X = X0 + Q a - c()a2/(e0 - a) 28,33

(ii) X = X0 + Q a — ba2

(iii)
<'+W -3’0-%
D X = X0 + C'a 34

F C'a 35

dC
= k i( €o “ °0c - k2a 36
dZ

29

The diffusion of particles is negligible for particles of size greater than 1 |xm.
The concentration of suspension is as low as 0.1% in water filtration, which makes
the term Pa negligible compared to e0 (i.e., e = €0).
The term eC is negligible compared to a.

Using these hypotheses, Equation 2 can be simplified as:

- + u .F l = 0 (3)
at "L azJ,

Combining Equations 1 and 3,

— = XUmC (4)
dt

In order to predict the concentration profile, one should know the value of X, and to this
end, several research workers have put forward empirical and semiempirical equations
relating X and a (Table 3).
The most general equation is the one proposed by Ives, who has taken into account the
 11

variations of specific surface of medium by the accumulation of deposit, porosity, and

interstitial velocity by the retention of particles .40 It is interesting to note that the coefficients
appearing in these equations are functions of suspension, filter-medium characteristics, and
operating parameters. Therefore, once these coefficients are calculated from pilot-scale or
laboratory-scale experiments with the given suspension and medium, one can simulate the
concentration profile for the same suspension for different operating parameters using Equa­
tions 1 , 3, and the relationship between X and o\
Herzig 41 put forward a rational approach for describing deep bed filtration by comple­
menting the existing two equations (Equations 1 and 3) with another equation (Equation 5)
which is in fact a material balance taken in an element of thickness, dZ, within a time
interval of dt. His approach leads to a simplified determination of the spatial variation of
specific deposit.
d<j K
^ = - 01 * ( 5)
where K = \U m.

From Equation 1, one can write

J_ dC _ _ _K
CdZ ~ U
(6 )

From Equation 5, one can write

I Hz ~ _ A (7)
a az ~ “ um

The subscript i denotes the entering layer (i.e., Z = 0)

C a
~r = ~ (8 )
i i

This relationship facilitates the calculation of cr at different depths, if C at different depths

and are known. The value of a l can be estimated experimentally.

B. Headloss
In the filtration process, the headloss development is as important as the effluent quality.

7. Clean Bed
The headloss through the clean bed can be easily calculated using Kozeny’s equation
(which describes the laminar flow of liquid through the granular bed, as is the case in deep
bed filtration). This equation is developed by assuming that the filter pores consist of identical
cylindrical tubes (Equation 9).

AP0 , Um (a,)’
= hkP- r

where hk is Kozeny’s constant, which takes a value of 5 (approximately) for a spherical­

shaped filter medium, (a c ) 0 is the specific surface of filter bed [ = (1 — e0) (ag)0], (ag) 0 is
the specific surface of filter grain [ = (6 /dc) for spherical-shaped filter grains], and p is the
viscosity.
12 Water, Wastewater, and Sludge Filtration

SATURATED ZONE
2. (WHERE Z. INCREASES WITH T IM E )

CLEAN BED

Z2 (Z2 = Z ­ V

FIGURE 4. Representation of filter bed into zones.

But as filtration proceeds, due to the particle deposition within the filter bed, the geometry
of filter pores will undergo a change. Consequently, the headloss will increase with the
particle deposition, and the Kozeny’s equation cannot be applied any longer.

2. Clogged Bed
Since the flow remains laminar, a modified form of Kozeny’s equation can be used to
compute the headloss of clogged bed (AP) as follows:

( 10)

Here the porosity of clogged bed (e) decreases with the filtration time and can be computed
from the following equation:

e = e0 — a ( 11)
On the other hand, the increase in specific surface with filtration time can be calculated
from the following equation:

ag = X(ag)j + (1 - X)(ag)0 ( 12)

where ag is the specific surface of a collector (i.e., a filter grain and associated particles on
the particular grain), (ag)j and (ag)0 are the specific surfaces of a particle and a filter grain,
respectively; and X is the volumetric ratio of particles to the total solid volume (x =
v
2— ). Here vp is the volume of retained particles, and vg is the volume of filter grains.
vp vg
It should also be noted that the particle deposition is not uniform throughout the
filter bed, and headloss varies from layer to layer. As a simple illustration, the filter bed
can be considered to consist of two zones, the top saturated zone and bottom clean zone
(Figure 4).
This representation leads to the following headloss equation for clogged bed.

As time increases, the value of Z2 increases while Z2 decreases.

Several research workers have considered the varying geometry of particle deposition in
 13

Table 4
DIFFERENT HEADLOSS MODELS

Model Ref.

AP = AP0(1- <r/e0) - 3 49

AP = AP0(1 + a a /e 0)3 (1 - <T/e0r 3 50

AP = AP0(1+ Kcr) 41

= (1 - e0 + o' ) 2 (1 - CT/e0) " 3 (1 - e0) “ 3 46

Z\r0

order to develop equations to describe headloss of clogged bed. As an example, if one

considers that the porous medium is made up of capillaries, and the particles are deposited
inside the capillary walls uniformly, then the following equation can be written:

= (« /.„ )- = • , ... ,14,

(1 - e0)(as)0 (1 - (Pcr/«)2}

If one assumes that particle deposition is uniform throughout the surface of filter grain, then
the above equation can be modified as

(1
(1 - e)ag .
- €0)(a ) 0 [■+A]
, . . , (15)

In the same way, different research workers have put forward different equations for headloss
computation based on the pattern of clogging (Table 4).
In any case, none of the above equations can satisfactorily be used to compute the headloss.
Due to this difficulty, semiempirical equations have been put forward. The common form
used is

AP _ 1

AP0 ~~ (1 - a ,a ) a2 (16)

where ax and a2 are constants.

Since the value of cr is relatively small, the above equation can be simplified as follows:

AP
- = 1 + ,17)

where m is a constant.
Although the aboveequation has no theoretical background, it can easily be used to
describe the temporal variationof headloss. The value of m, which isspecific to the filtration
system, has to be determined from laboratory-scale or pilot-scale experimental results,
maintaining a similar operating system. The average value of m obtained for different
filtration conditions is given in Table 5 as an example.
14 Water, Wastewater, and Sludge Filtration

Table 5
VALUE OF m AT DIFFERENT FILTRATION CONDITIONS

APo/L
Particles in um (AP0 in cm Filter dc
suspension cm/s of water) medium (pm) m Ref.

Ferric hydroxide 0.14 0.130 Calcium 1330 57.0 50

(dP = 10 |xm) 0.28 0.215 carbonate 1330 28.4
Algae (dP = 5 pm) 0.14 0.650 Sand 540 61.5 40
0.14 0.440 700 432.0
Secondary clarifier 0.20—0.40 Glass 2000—6000 3610 (av value) 51
effluent of munici­ beads
pal wastewater
Effluent from sedi­ 0.28—0.56 Sand 1000— 1190 753— 2343 52
mentation tank 1190— 1410 185— 306
(water treatment)
Kaolin 0.14—0.56 Sand 1000 130— 194 53

REFERENCES

1. Huisman, L., Developments of village-scale slow sand filtration, Prog. Water Technol., 11, 159, 1978.
2. Huisman, L. and Wood, W. E., Slow Sand Filtration, World Health Organization, Geneva, 1974.
3. IRC/CWSS, Slow Sand Filtration for Community Water Supply in Developing Countries, Bull. Ser. No.
9, International Reference Centre for Community Water Supply and Sanitation, the Hague, the Netherlands,
1977.
4. IRC/CWSS, Slow Sand Filtration for Community Water Supply in Developing Countries, A Selected and
Annotated Bibliography, International Reference Centre for Community Water Supply and Sanitation, the
Hague, the Netherlands, 1977.
5. IRC/CWSS, Slow Sand Filtration for Community Water Supply in Developing Countries, Bull. Ser. No.
16, International Reference Centre for Community Water Supply and Sanitation, the Hague, the Netherlands,
1981.
6. IRC/CWSS, Guidelines for Operation and Maintenance of Slow Sand Filtration Plants in Rural Areas of
Developing Countries, International Reference Centre for Community Water Supply and Sanitation, the
Hague, the Netherlands, 1983.
7. Vigneswaran, S., Tam, D. M., and Visvanathan, C., Filtration technologies for developing countries,
Environmental Sanitation Review, Environmental Sanitation Information Center, Asian Institute of Tech­
nology, Bangkok, Thailand, 1983, 12.
8. Fair, G. M., Geyer, J. C., and Okun, D. A., Water and Wastewater Engineering, Vol. 2, John Wiley
& Sons, New York, 1968.
9. Conley, W. R. and Hsiung, K., Design and application of multimedia filters, J. AWWA, 61, 97, 1969.
10. McCabe, W. L. and Smith, J. C., Unit Operations of Chemical Engineering, 2nd Ed., McGraw Hill,
New York, 1967.
11. Camp, T. R., Discussion, Experience with anthracite-sand filters, J. AWWA, 53, 1478, 1961.
12. Mazumdar, B., A Study of the Phenomenon of Intermixing in Dual-Media Filters and It ’s Effects on Filter
Performance, AIT Research Study, Asian Institute of Technology, Bangkok, 1984.
13. Deb, A. K., Theory of sand filtration, J. Sanit. Eng. Div.— Proc. ASCE, 95, 399, 1969.
14. Dostal, K. A. and Robeek, G. G., Studies of modifications in treatment of Lake Erie waters, J. AWWA,
58, 1489, 1966.
15. Hudson, H. E., Functional design of radial sand filters, J. Sanit. Eng. Div.— Proc. ASCE, 89, 17, 1963.
16. Kawamura, S., Design and operation of high rate filters, I, J. AWWA, 67, 535, 1975.
17. Amirtharajah, A., Design of granular-media filter units, in Water Treatment Plant Design, Sanks, R. L.,
Ed., Ann Arbor Science, Ann Arbor, Mich., 1978.
 15

18. Cleasby, J. L., Filter rate control without rate controllers, J. AWWA, 61, 181, 1969.
19. Cleasby, J. L., Declining rate filtration, J. AWWA, 73, 484, 1981.
20. Cleasby, J. L. and Dibernardo, L., Hydraulic considerations in declining - rate filtration plants, J. Environ.
Eng. Div.— Proc. ASCE, 106, 1043, 1980.
21. Arboleda, J., Hydraulic control systems of constant and declining flow rate in filtration, J. AWWA, 66,
87, 1974.
22. King, P. H., Johnson, R. L., Randall, C. W., and Rohberger, W., High-rate water treatment, The
state-of-the-art, J. Environ. Eng. Div.—Proc. ASCE, 101, 479, 1975.
23. Wagner, E. G., Simplifying design of water treatment plant for developing countries, J. AWWA, 75, 220,
1983.
24. American Water Works Association Filtration Committee, The status of direct filtration, J. AWWA, 72,
405, 1980.
25. Gadkari, S. K., Raman, V., and Gadkari, A. S., Studies of direct filtration of raw water, Indian J.
Environ. Health, 22, 57, 1980.
26. King, P. H. and Amy, W. T., The potential of direct filtration for water treatment in underdeveloped
countries, unpublished manuscript, 1979, 16 pp.
27. Clasby, J. L., Arboleda, J., Burns, D. E., Prendiville, P. W., and Savage,E. S.,Backwashing of
granular filters, J. AWWA, 69, 115, 1977.
28. Weber, W. J., Jr., Physico-chemical Processes for Water Quality Control, John Wiley & Sons,New
York, 1972.
29. Amirtharajah, A., Optimum backwashing of sand filters, J. Environ. Eng. Div.— Proc. ASCE, 104, 917,
1977.
30. Barnes, D., Bliss, P. J., Gould, B. W., and Vallentine, W. R., Water and Wastewater Engineering
Systems, Pitman Press, London, 1981.
31. Simonds, M. A., Public water supply of Queensland, Australia, J. AWWA, 55, 1044, 1963.
32. Jung, H. and Savage, E. S., Deep bed filtration, J. AWWA, 66, 73, 1974.
33. Degremont Water Treatment Handbook, Halstead Press, New York, 1979.
34. Denholm, E. W., Automatic operation of Broadside filter plant, Scotland, J. AWWA, 48,597, 1956.
35. Baylis, J. R., Nature and effect of filter backwashing, J. AWWA, 51, 126, 1959.
36. Valencia, J. A., Some basic ideas on establishing a water treatment technology adapted to developing
countries, in Proc. International Training Seminar on Community Water Supply in Developing Countries,
Bull. Ser. No. 10, International Reference Centre for Community Water Supply and Sanitation, the Hague,
the Netherlands, 1977.
37. Schulz, C. R. and Okun, D. A., Treating surface waters for communities in developing countries, J.
AWWA, 75, 212, 1983.
38. Mints, D. M., Modem Theory of Filtration, Spec. Subjects Rep. No. 10, International Water Supply
Congress and Exhibition, Barcelona, October 3 to 7, 1966.
39. Iwasaki, T., Some notes on sand filtration, J. AWWA, 29, 1591, 1935.
40. Ives, K. J., Theory of Filtration, Spec. Lect. No. 7, International Water Supply Congress and Exhibition,
Vienna, September 1 to 5, 1969.
41. Herzig, J. P., Leclerc, D. M., and Le Goff, P., Flow of suspensions through porous media — application
to deep bed filtration, Ind. Eng. Chem., 62, 8, 1970.
42. Ornatsaki, T., Theory of Clogging of a Sand Bed When a Clay Suspension is Filteredthrough It,Ph.D.
thesis, Moscow University, Moscow, 1955.
43. Mints, D. M. and Krishtul, V. P., J. Appl. Chem., 33, 302, 1970.
44. Maroudas, A., Clarification of Suspension: a Study of Particle Deposition in Granular Filter Media, Ph.D.
thesis, University of London, 1961.
45. Ives, K. J., Rational design of filters, Proc. Inst. Civ. Eng., 16, 163, 1960.
46. Sakthivadivel, R., Thanikachalam, V., and Seetharaman, S., Headloss theories on filtration, J. AWWA,
64, 233, 1972.
47. Heertjes, P. M. and Lerk, C. F., The functioning of deep bed filters, Trans. Inst. Chem. Eng., 45, 129,
1967.
48. Litwiniszyn, J., On some mathematical models of suspension flow in porous medium, Chem. Eng. Sci.,
22, 1315, 1967.
49. Shektman, Yu. M., Filtration of Suspensions of Low Concentration, Institute of Mechanics of U.S.S.R
Academy of Science, Moscow, 1961.
50. Mackrle, V., L ’Etude du Phenomene d ’Adherence: Colmatage dans le Milieu Poreux, Doctoral thesis,
University of Grenoble, Grenoble, France, 1960.
51. A1 Alousi, M., La Filtration dans la Masse en Epuration Tertiaire d ’eaux Residuaires, Doctoral thesis,
University of Montpellier, Montpellier, France, 1979.
16 Water, Wastewater, and Sludge Filtration

52. Yossarin, B., Comparison of Declining Rate Coarse Media Filtration with Constant Rate Dual Media
Filtration at Bangkaen Water Treatment Plant, M.Sc. thesis, Asian Institute of Technology, Bangkok,
Thailand, 1985.
53. Lo Chiu Hsiu, Upgrading of Deep Bed Filtration Performances, M.Eng. thesis, Asian Institute of Tech­
nology, Bangkok, Thailand, 1984.
 17

Chapter 2

THEORETICAL APPROACH TO DEEP BED FILTRATION

Rolf Gimbel

TABLE OF CONTENTS

I. Introduction....................................................................................................................... 18

II. Description of Macroscopic Filter Behavior................................................................. 18

A. Mass Balance........................................................................................................ 20
B. Kinetic Approach..................................................................................................20
1. Empirical Approaches for \ 0 or y 0 ....................................................... 23
2. Empirical Approaches for fx ................................................................. 25
a. Neglecting Particle D etachm ent.............................................. 25
b. Considering Particle Detachment.............................................26
C. Approach for the Pressure D rop........................................................................ 28
D. Limitations of the Macroscopic Description.................................................... 30

III. Microscopic Description of Particle Deposition Mechanisms.................................... 30

A. Modeling the Flow through a Clean Filter B ed...............................................30
B. Determination of Filter Efficiency in the InitialPhase.....................................32
1. Forces Acting on Particles in Liquid Deep Bed Filtration............... 33
2. The Elementary Filter Efficiency for BrownianParticles
(dp ^ 1 |Jim)........................................................................................... 36
3. The Elementary Filter Efficiency for Non-Brownian
Particles (dp ^ 1 |xm)............................................................................ 36
a. Transport Efficiency...................................................................42
b. Adhesion Probability................................................................. 46
C. Filter Efficiency in the Dynamic Filtration Phase............................................49

Acknowledgment..........................................................................................................................49

References......................................................................................................................................52

List of Sym bols............................................................................................................................50

18 Water, Wastewater, and Sludge Filtration

I. INTRODUCTION

Deep bed filtration is a process in water and wastewater treatment widely used to remove
suspended solids. It is applied particularly when high flow rates of water with relatively low
contents of suspended solids (by rule of thumb less than 0.05% in volume concentration)
have to be treated.
The principle of a deep bed filter is shown in Figure 1. It has the characteristic feature
that the suspended solids are deposited within a relatively high porous filter layer, leading
to more or less complete particle removal from the water passing the filter. In most cases,
the filter layer consists of a packing of filter grains (in the size range of mm), which can
be backwashed after a certain filter running time. During this time, the capacity of the filter
bed for the uptake of turbidities has been exhausted, leading either to a turbidity breakthrough
or to an intolerable increase of pressure drop. Besides this type of deep bed filter, which,
for example, is well represented by the so-called rapid sand filters, there also exist other
types, such as cartridge or precoat filters, but these may also represent some kind of com ­
bination of deep bed and surface or cake filters.
In a true deep bed filter, the deposition of particles requires two successive steps:

1. Transport of the particles inside the filter bed to the surface of the filter material or
to the surface of previously deposited solids
2. Adhesion of the particles after contact with the inner filter surface

Both steps are the result of several forces and torques acting on the particles, such as drag
and gravity forces, stochastic forces by Brownian motion, and physicochemical surface
forces. Thus, there exist extremely complicated situations, especially under practical con ­
ditions of water and wastewater treatment. For this reason, there are still many open questions
concerning the particle removal process, and so a comprehensive fundamental theory of
deep bed filtration is still lacking. But the understanding of this process up to now, which
will be summarized in this chapter, is a very important basis for the optimization of the
design and operation of deep bed filters, including the pretreatment of the raw water. This
is true despite the fact that, in many cases, there is still a need for experimental filtration
data to be used for modeling.
For understanding, one has to distinguish between the so-called macroscopic description
of the filter behavior and the microscopic filtration theory. The first describes, for practical
means only, the relevant parameters, like filtrate quality and pressure drop, as a function
of time on an extensive experimental basis. The second is strictly based on fundamental
physicochemical mechanisms, thus needing relatively little experimental information.

II. DESCRIPTION OF MACROSCOPIC FILTER BEHAVIOR

For optimal design and control of a deep bed filter, one should be able to determine the
particle volume concentration cv and the pressure p as functions of filter bed depth z and
filter running time t for the given operating conditions (see Figure 2). Since the particle
deposition processes and the pressure drop over the filter bed are both influenced by the
amount of deposit a v, the latter is to be considered as an additional variable to completely
describe the macroscopic filter behavior.
To describe cv, crv, and p as functions of z and t, the following points have been considered:

• A mass balance for a differential filter bed element such as is shown in Figure 2
• A kinetic approach describing mass transfer from the flowing liquid to the stationary
filter media
 19

Deep Bed F iltra tio n

suspension

filter layer with

deposited solids

filtrate
FIGURE 1. Schematic representation of a deep bed filter.

FIGURE 2. Schematic representation of a filter bed

with a differential filter element.
20 Water, Wastewater, and Sludge Filtration

• An approach describing the increase of pressure drop with increasing deposit

First, these fundamental equations will be discussed in detail.

A. Mass Balance
It is generally assumed that on a macroscopic scale the particle concentration over the
filter cross section is homogeneous, and the flow pattern is a one-dimensional plug flow.
Thus, the total mass balance for a differential filter element is

(change of the amount of particles with time) = (particle flux in) — (particle flux out)

d(av + ecv) = _ a[vfcv - Dd(dcv/dz)]

at dz
( 1)
with constant vf and constant Dd leading to:

dcrv d(ecv) dcv d 2cv

— + —r ~ + vf • Dd • — = 0 (2)
at at dz dz2

Equation 2 is commonly used in a simplified form by neglecting the term with the axial
dispersion coefficient Dd. Furthermore, if one defines a new time variable t by

t = t - ^ (3)

and neglecting the change of porosity with time (e = e0), Equation 2 reduces to1

+ vf ~ = 0 (4)
dr dz

Equation 4 is the common form of the mass balance. Here, in most cases, the filter running
time t is used instead of the corrected time t. This is justified for long ultimate filter running
times (e.g., up to a couple of days in rapid sand filtration), whereas the mean retention time
of the liquid in the filter bed is of the order of a few minutes. But the differences between
t and t can be of importance when interpreting experimental data obtained after relatively
short filter running times.

B. Kinetic Approach
The increase of the filter deposit according to Equation 4 is described by a general kinetic
approach like2 3

Y * = Ucy, <rv, Y) (5)

dT “

Y is a vector containing all parameters influencing particle deposition and the morphology
of the deposit within the filter. These are, for example, the initial porosity of the filter bed,
the filtration rate as well as a series of physicochemical parameters characterizing the prop­
erties of the particles, the filter grains, and the solution.
If fCTis described by empirical correlations, Y may also contain the fitting constants for
these correlations.
 21

In the practice of water and wastewater treatment, the turbid matter consists of particles
with different properties; thus:

Cv = S c, (6)

where c{ describes the volume concentration of particles of type i. Accordingly:

CTv = 2 ^ (7)

Thus d a jd t is the sum of all d v jd i. The latter may be strongly different from each other
and therefore may require, in general, the consideration of several types of particles in the
kinetic approach. To simplify nomenclature, the indices will be neglected in the following
discussion: thus, c is always the volume concentration and a the specific deposit of a distinct
type of particles.
In practical applications, the turbid matter is approximately described by one representative
type of particle. This simplification has often been justified when controlling the filter
efficiency by integral methods such as turbidity measurement.
To solve Equations 4 and 5, the kinetic approach has to be formulated in more detail.
Iwasaki4 suggested:

d(J
— = X • vf • c (8 )
dr

or in combination with Equation 4:

f = -\-c (9)
dz

According to Equation 8 , the increase in deposit is directly proportional to the amount of

particles entering a filter element per unit time. Thereby, it is assumed that there are no
interactions among the dispersed particles, which is true for the commonly filtered dilute
suspensions in water and wastewater treatment. The filter coefficient X is a measure of the
filter efficiency. From X, the so-called elementary filter efficiency y may be derived. It can
be approximated by:

y = X • dK for y < 1 (10)

where y describes the probability of distinct type of particles being deposited in an elementary
filter layer of height Az = dK (see Figure 2).

7 = £] £i±i for z.+) _ z. = dK (11)

X and 7 can generally be written as follows:

X = X0(Y) • fx(c, o’, Y) ( 12)

22 Water, Wastewater, and Sludge FUtration

specific de pos it oi/or

FIGURE 3. Change of the correction function fx with increasing filter
deposit.

7 = 7 0(Y) • fx(c, cr, Y) (13)

X0 and y 0 indicate the efficiency within the initial filtration period, during which the deposit
in the filter bed does not influence the particle deposition. The correction function fx de ­
scribing the change of X and 7 with a is therefore:

fx = 1 for (i = 0

For the initial filtration period, Equation 9 can easily be integrated for a homogeneous filter
bed (X0 = const). Thus, the concentration profile in the bed becomes:

c = c0 • exp( - \ 0 • z) = c0 • ex p ( - 7 o • z/dK) (14)

Equation 14 has been experimentally confirmed in many cases .5

Figure 3 qualitatively shows the observed changes of the filter efficiency with increasing
deposit in the form of correction function fx. There are cases where the efficiency decreases
steadily until it reaches zero at the ultimate deposit crs. In contrast to that, filters with a
ripening period show a temporary increase of filter efficiency.
To describe c and a as functions of t and z, the coupled partial differential Equations 4
and 8 have to be solved. The solution also requires an equation of type 12. In general, this
has to be done by numerical methods. The method of characteristics has been proved reli­
able . 6' 8 In special cases, where fx has a very simple form, less complex procedures or even
analytical solutions are applicable. This will be discussed later.
 23

Table 1
EXPERIMENTALLY DETERMINED
DEPENDENCIES OF THE
7 0-VALUES ON DIFFERENT
PARAMETERS IN WATER
TREATMENT9

Parameter Range of variation

Particle diameter dp
dp > 1 |xm 7o ~ dp0 to y0 ~ dp2
dp < 1 |xm 7 o ~ ^ 2/3 to 7o ~ dp0
Viscosity r\ 7 o ~ i) ~l to 70 ~ i f
Filter grain diameter dK 7 o ~ dK- 2 to 7o ~ dK°
Filtration rate vf 7 o ~ v r 1 to 7o ~ vf°

1. Empirical Approaches for A0 or y0

The dependence of the initial filter efficiency on various parameters in Y was observed
to be quite different. This is shown in Table 1. The order of magnitude of the y0-value is
indicated in Table 2. Here, under typical conditions of rapid filtration (vf = 5 to 10 m/h),
these values are commonly smaller than 1%. The strong variation of the results given in
Tables 1 and 2 shows that single parameters may have opposite influences on y0, depending
on the actual conditions.
To describe \ 0 and y 0 as functions of different parameters, several empirical and
semiempirical approaches have been suggested . 1,10 14 The semiempirical approaches result
either from very simplified physical filtration models (based on a spherical collector or on
a tube model for the filter bed) or from a dimensional analysis. Hereby, the so-called classical
mechanisms of deep bed filtration are mainly considered for the transport of particles to the
surface of the filter material. Under typical conditions of water treatment with rapid filters,
three mechanisms are of special importance. Figure 4 shows a schematic representation of
these three mechanisms based on laminar flow through the filter bed.

1. Deposition of particles by constricting the streamlines (see Figure 4a). Assuming a

particle moves with its center along such a streamline, contact between particle and
filter grain surface occurs if the distance between streamline and grain surface becomes
smaller than the particle radius. This is called the interception mechanism.
2. The gravity force (see Figure 4b). Turbid matter may leave the streamlines due to its
sedimentation behavior. This mechanism is important for particles with a density
significantly higher than that of the liquid and with a diameter larger than 1 pm.
3. The stochastic force due to thermal movement (see Figure 4c). The smaller the particles
are, the more they become influenced by Brownian motion, a mechanism that becomes
important for particles smaller than 1 pm.

Under typical conditions in water treatment, inertia forces are negligible in their action as
deposition mechanisms in deep bed filters. This is in contrast to aerosol filtration.
For example, Ives 15 combined several parameters influencing y 0 into the following
semiempirical equation:

(15)
24 Water, Wastewater, and Sludge Filtration

Table 2
EXPERIMENTALLY DETERMINED y 0- VALUES1

Suspension Filter material

Type dp in p,m Type dK in mm vf in m/h 70 in %

Polyvinyl-chloride 1.3 Glassbeads 0.46 2.2 0.11

(PVC) 0.15
0.32
5.0 0.09
7.2 0.63
Anthracite 2.2 0.31
5.0 1.38
7.2 0.92
Sand 0.42 2.2 0.47
0.60 5.0 0.27
Escherichia coli ~1 Sand 0.23 0.13 0.22
0.21 0.22
0.29 0.18
0.10 0.13 0.54
0.21 0.50
0.29 0.47
Ferric hydroxide 6— 20 Sand 0.46 5.0 0.92
9.7 0.70
4— 25 0.70 0.40
Algae ~ 15 0.54 5.0 0.61
0.70 1.00
Quartz 2— 22 Anthracite 0.77 5.0 0.49
1.02 0.75
Algae ~ 15 Sand 0.83 0.13 4.55
0.29 4.30
0.23 0.13 4.50
0.29 3.18
Ferric hydroxide ~ 10 Calcium­ 1.33 5.0 1.30
carbonate 10.0 0.59
Quartz < 20 Glassbeads 2.0 9.0 0.34
13.0 0.20
28.8 0.08

a ) Interception b) Sedimentation c)D iffusion

FIGURE 4. Schematic representation of the classical particle deposition mechanisms in deep bed filtration of
liquids.
 25

with: NPe = Peclet number to describe particle diffusion

= vf • dK/Dp

Nr = interception number
= dp/dK

Ng = gravity number
= (PP - Pf) g dp2/18T!Vf

Nx = inertia number
= (Pp - Pf) Vf dp2/TidK

a,ej = empirical constants

Dp = particle diffusion coefficient

= kT/37TT]dp

Such semiempirical approaches are usually based on very simplified assumptions. For
example, Equation 15 does not account for the hydrodynamic interaction between the filter
grain surface and an approaching particle, or for other effects due to London-van der Waals’
and electrostatic surface forces, or for overall particle adhesion behavior. Nevertheless, by
fitting empirical constants, all these simplifications may be approximately compensated.
Therefore, a satisfactory description of a set of experimentally obtained y0-values is possible.

2. Empirical Approaches for f k

There are two groups of approaches for the correction function fx (based on a homogeneous
suspension containing one representative type of particle):

fx = fx(cr, Y) (16)

In this case fx does not depend on particle concentration c. Such an approach results from
the simplified models, where particles, once having been deposited, are considered to stay
there for the rest of the filter run. The amount a of deposited particles, however, has an
influence on the subsequent deposition processes.
In the second case,

fx = fx(c, a , Y) (17)

Here an additional dependency on c is given. Such an approach results from models that
consider the possibility of detachment of deposited particles.

a. Neglecting Particle Detachment

Correction functions fx according to Equation 16 are mainly based on models that assume
changes of the filter bed geometry by particle deposition . 1 ,3 1 0 1 6 18 As a typical example, the
“ general equation” of Ives16 may be written as:
26 Water, Wastewater, and Sludge Filtration

with: crs = ultimate specific deposit in the filter; a, = fitting constants.

The first term in Equation 18 describes for a spherical collector model the increase of
surface area by a uniform coating of deposit. The second term describes, in analogy to the
first one, a decrease of surface area with a when using a capillary model for the filter bed.
The third term considers the increase of the mean interstitial velocity with <x. Depending on
the choice of the fitting constants, the different trends of fx are shown in Figure 3.
The approaches made according to Equation 16 do not consider the detachment of de­
posited particles by impinging particles or by the change of the interstitial velocity due to
the increase of deposit. Such detachment behavior has been observed in quite a few ex ­
periments , 1’5 1 2 14,19 22 but about its importance to the overall filtration process, there exists
a controversy in the literature . 1,13,14 23 As a first approximation, detachment may be considered
to occur by a reduced particle deposition rate in the kinetic approach. Hence, in many cases,
approaches such as Equation 18 have been proved sufficiently correct for the description of
empirical data.
Due to the independence of fx on the particle concentration c, the two coupled partial
differential equations for the mass balance (Equation 4) and for the kinetic approach (Equation
8 or 9) can be transferred to two ordinary differential equations to describe c ( z , t ) and a ( z , r ) .
This procedure is discussed in detail by Herzig et al . 1 In general, approaches according to
Equation 16 lead to a constant pattern of distribution of the deposit in the filter bed. After
some time (when the deposit in the inlet layer of the filter has reached its ultimate value
a s), this leads to a constant pattern of the concentration profile in the pore water. A qualitative
representation is shown in Figure 5.
For very simplified forms of fx, quite a few analytical solutions for c ( z , t ) and a ( z ,T ) are
given in the literature. 1,5,24' 28 As an example, with the correction function resulting from
Equation 18 (for e5 = e6 = 0 and e 7 = 1) one gets:

c exp (c0 vf X0 t a s ')

(19)
exp (c0 vf X0 t <xs ‘) + exp(X0 z) - 1

and

exp (c0 vf X0 t cts ') - 1

o\ exp (c0 vf X0 t <ts ') + exp(X0 z) - 1
(20)

For example, with Equation 19, the required filter bed height z may be easily determined
for a given filter running time t (or t) and for a maximum value of c/c0 ratio. To do so, \ 0
and a s have to be determined (from experiments or from available correlations). This is a
useful procedure for the optimal design of rapid sand filters.

b. Considering Particle Detachment

To consider a reversible particle deposition, the approach for the correction function fx
must contain the particle concentration term (see Equation 17). A typical example is the
kinetic approach of Mints. 13 It is based on the assumption that the particle deposition rate
is independent of the amount of deposit a and that there is at the same time an increased
detachment of already deposited particles with increasing cr.
This results in:

a* • <t

= deposition — detachment (2 1 )
= X0 • V f • c • fx(c, cr, Y)
 27

filter bed height in m

FIGURE 5. Concentration and deposit profiles as functions of z and t for constant pattern.

It follows for fx (with a* being an empirical constant describing detachment):

fx = 1 — a* • (j/ \ 0 • vf * c (2 2 )

With thiskinetic approach, the solution for c(z,t) is a specialcase of theThomas Equa­
tion . 2930 It maybe obtainedby means of Laplace-transformation. Thus, c/c0 becomes:

c(z, t)/c 0 = J(X0 • z, a* • t) (23)

= 1 - e ~ a*'T | ° e _£ • I0(2 V a* • t • £)d£

JO
28 Water, Wastewater, and Sludge Filtration

with J(a,P) = J-function; and IqO ') = modified Bessel-function of the first kind and order
zero.
Another approach according to fx(c,a,Y) has been suggested by Adin et al.31,32 They also
defined the temporal change in deposit as the difference between particle deposition and
detachment rates. Compared to the kinetic approach of Mints,13 their deposition term ad­
ditionally contains a free deposit capacity (as — a) as driving force. Furthermore, a pressure
drop gradient is considered within the detachment term, since this is a measure for the shear
forces acting on the deposits and therefore influencing the detachment probability. Using
the method of characteristics and fitting the appropriate empirical constants, a good descrip­
tion of the filter behavior was possible for suspensions made of kaolin, iron, or alum floes.

C. Approach for the Pressure Drop

Besides changes in concentration and deposit within a filter bed, the development of
pressure drop or headloss is the third important parameter describing the behavior of deep
bed filters. Knowledge of the pressure development is important for the design and operation
of a filter with respect to the determination of the maximal pressure drop and the avoidance
of negative head within the filter layer. At points with negative head, air compounds may
degasify that would be dissolved under normal conditions. This leads to an undesirable
additional blocking of filter pores.
Figure 6 shows qualitatively the pressure drop development vs. bed depth z for an open
single-medium filter. This type of diagram (Michau diagram) considers hydrostatic pressure
and pressure drop as a result of the flow resistance. To describe this, the gradient of the
pressure drop will be defined as follows:1-3

(24)
 29

where \ = pressure drop gradient in the clean filter bed (free of deposit), which can
\dz/o
be estimated for example by the Carman-Kozeny equation or by Happel’s-cell model. 33
The correction function fp describes the increase of the pressure drop gradient with in­
creasing deposit o\ Here, those influencing parameters of Y, determining the deposit mor­
phology, have to be considered. By its morphology, the deposit can influence more or less
the increase in pressure drop depending on where the particles preferentially will be removed
(for example, either in constricted zones of the filter pores or uniformly on the filter grain
surface).
If fp and ct( z , t ) are known, the total pressure drop across the filter bed can be determined
by integrating Equation 24 from z = 0 to z = L. For the description of fp, quite a few
simplified geometrical models have been developed, which are all based on the Carman-
Kozeny equation. These approaches are comprehensively discussed elsewhere. 1 318 34 Here,
only two examples differing in a characteristic way are mentioned.
Assuming a uniform distribution of particle deposit in an elementary filter layer, according
to Ives : 16

/ a * c r \ 2e5 / c r \ 2e6 - 3
f- = ( l + v ) ■( ‘ - tJ (25)
For e5 = e6 = 1 and developing fp into a Taylor series, the above relation will be simplified
as:

fD = 1 + ----------­ •cr for a e0 (26)

^0

As Ives hasfurther shown, using Equation 26 and assuming that fxdecreases linearly with
the deposit (i.e., fx = 1- ct/<ts), integration of Equation 24 approximatelyleads to a linear
increase of the total pressure drop Apv with the filter running time. Thus:

(fe ) . r dz + * ± i . /a p A . | \ dz
4 p - “ (V 'd tz a/ ol ' do
i dz + e0 V dz/o • doI. » dz <27)

s Apv,o + b • t

b = (2 a + 1) • vf • c0 _ /dp,
crs \dzJo

This relationship holds true for common filter operation, i.e., as long as c(z = L) < < c0,
and has been experimentally confirmed in many cases . 1’5 24’261 1 3 118 Equation 27 has been
derived assuming a very simplified special case for fx. As is shown in Reference 5, these
restrictions can be omitted and the validity of a linear increase of Apv with time can be
generalized, as long as the conditions for a constant pattern of the particle concentration
profile in the filter bed are fulfilled.
Tien et al . 35 distinguished two phases for the pressure drop in their model. In the first
one, the increase in Apv is described by an increase in the surface of the filter grains (assumed
to be spherical collectors). Here, uniform coating with particles and a corresponding decrease
of the filter-bed porosity is assumed. In the second phase, after reaching a certain filter
deposit, the model also considers the total hydraulic blocking of filter pores. By this, the
experimentally determined increase in pressure drop, which in some cases differs significantly
from a linear relationship, could be quite well described.
30 Water, Wastewater, and Sludge Filtration

D. Limitations of the Macroscopic Description

The above-discussed methods are the basis for an optimal filter design, which is especially
important for multimedia filters, because the design possibilities here are many. To determine
the fitting constants appearing in the above equations, filtration experiments have to be
carried out under specific conditions. Because all approaches in the macroscopic description
are based on strongly simplified models, an interpolation between experimentally realized
operation conditions is possible only within a small range. It is also quite problematic to
use the fitting constants for applications which are not very similar.
The detailed knowledge of the microscopic mechanisms of the particle deposition in deep
bed filters and their inclusion in realistic models, however, would allow an interpolation
between realized operation conditions, which are significantly different. Such knowledge
also allows certain extrapolations, so that a suitable analysis of the turbid matter without
filtration experiments may be enough for an optimal design and operation of a deep bed
filter. Besides, by knowing the microscopic deposition mechanisms, the number of required
filtration experiments and operation conditions to be tested can be reduced and carried out
in a much more sophisticated way. Thus, the microscopic approach of deep bed filtration
will be summarized in the following sections.

HI. MICROSCOPIC DESCRIPTION OF PARTICLE DEPOSITION MECHANISMS

The knowledge of the microscopic mechanisms in a deep bed filter allows, in the ideal
case, the correct calculation of the behavior of particles inside the filter bed. To describe
the total deposition process, where particles are transported from bulk solution in the filter
pores to the surface of the filter material to achieve a stable deposition there, all forces and
torques affecting the particles have to be considered. Thereby, it is useful to distinguish
between transport and adhesion mechanisms, which are characterized by different kinds of
dominating forces and torques, depending on the distance of the particle from the filter grain
surface. The result of these two types of mechanisms is given by the particle transport
efficiency and the adhesion probability. They compose the elementary filter efficiency.
A basis for a good computation of particle behavior is the realistic description of the
structure of the filter bed. As this is a very difficult process for practical conditions, suitable
models are used to describe mathematically the geometric structure of the clean filter bed,
at least approximately, with regard to the properties influencing the particle deposition. The
elementary filter layers are hereby replaced by collectors of various geometries.
To describe the deposition process in the initial filtration phase with which most of the
fundamental works are concerned, the interactions of the particles with the collector surface
have to be considered in addition to the behavior of a “ free” particle in the pore space. In
the subsequent dynamic filtration phase, where the deposit increasingly influences the dep ­
osition mechanisms, interactions between depositing particles and deposits on the collector
surface also have to be considered for model calculations. Besides, in this dynamic period,
a detachment of previously deposited particles may occur. However, it is not required to
consider interactions between the suspended particles themselves, since in common appli­
cations of deep bed filtration, the volume concentration of turbid matter is very low. Nearly
all fundamental theories up to now are limited to the deposition of spherical homogeneous
particles.

A. Modeling the Flow through a Clean Filter Bed

To describe the flow through granular filter beds, there exist fundamentally two different
models (see Figure 7):

• Internal flow models, whereby the elementary filter layer is represented by capillaries
of various geometries36' 42
 internal flow models external flow models

\ i
capillaries with constant single spherical
c
cross section collector
E
.2
a>
61
capillaries with constricting cell model for
cross section spherical collectors

I I
FIGURE 7. Schematic representation of different models for elementary filter layers.

• External flow models, whereby the elementary filter layer is represented by spherical
collectors being in a flow field of distinct boundary conditions2,33,43' 46

In all cases, laminar axially symmetrical flow conditions are considered, which allows
the introduction of a stream function to simplify the description of the flow fields of interest.
As a typical example, Happel’s cell model will be discussed shortly. Here, a representative
spherical collector of the filter bed is considered to be surrounded by a liquid shell, the flow
through which is under creeping conditions. The volume of the shell related to the total
volume of the cell (consisting of the collector with radius aK and the liquid shell) corresponds
to the void fraction of the filter bed. Thus, the outer radius b of the spherical cell can be
written as:

b = aK/p (28)
P = (1 - eo) '/3

The stream function 'P has been obtained assuming that the outer surface of the liquid shell
behaves like a free surface, resulting in:
32 Water, Wastewater, Sludge Filtration

where:

r/aK

1
A = —
w
B = - ( 3 + 2p5)/w
C = (2 + 3p5)/w
D = - p 5/w
w = 2- 3p + 3p5 - 2p6 (29)

Developing 'P in a Taylor series, one obtains approximately (the terms up to the second
order are considered47):

'P = ^ As • vf • (r — aK)2 •sin20 (30)

By use of the stream function,the axially symmetrical flowfield v is obtained as follows:

v = vr *er + ve - e 6 (31)

1 dV
vr = -
r2sin0 00
1 d'P
v« =
rsin 0 dr

B. Determination of Filter Efficiency in the Initial Phase

In the initial filtration phase, the removal efficiency of one representative collector may
be defined as:

7^ = — r (32)
TT • a2 • Cj • vf

where J = flux of a certain type of particle, which is deposited on one collector; and Cj =
particle concentrationon the upstream side of the collector in the j-thelementary filterlayer.
Considering an elementary filter layer of height dK (see Figure 2) one obtains for J:

j = F - v f - ( Cj - cj+l) (33)
nK

= 6FdK • (1 - e0)/Trd^

nK = mean number of collectors in an elementary filter layer.

Using the definition of elementary filter efficiency (see Equation 11), it follows from
Equations 32 and 33
 33

7o = X0 • dK = ^ (1 - e0) • y 0 (34)*

To determine y*0 or y 0 in terms of the various physicochemical parameters contained in

Y (see Equations 5 and 13), one has to distinguish:

1. Transport of the particles to the collector surface until they come into contact with it.
This is due to different transport mechanisms whose effects are described by the
transport efficiency y 0 x, a parameter which is defined as the fraction of particles (of
one special type) entering an elementary filter layer and coming into contact with the
collectors in this layer.
2. Adhesion of the particles after contact. This is due to different adhesion mechanisms
whose effects are described by the adhesion probability y 0 H, a parameter which is
defined as the fraction of particles which come into contact with the collector surface
and achieve a stable deposition there.

Regarding transport and adhesion of the particles as two subsequent and independent
processes, it follows:

7o = 7 o,t ‘ 7 o,h (3 5 )
or

7o = 7 o,t * 7 o,h

with: 7o , t = transport efficiency for one collector

A strict distinction between transport and adhesion on the basis of fundamental physi­
cochemical mechanisms has been made in very few cases up to now . 5 This distinction can
be characterized by the occurrence of a contact between particle and collector surface. As
there exists no clear physical definition of a contact between two immersed solid surfaces,
it seems suitable to consider first of all the forces acting on a suspended particle.

1. Forces Acting on Particles in Liquid Deep Bed Filtration

These forces can be qualitatively summarized as:

a. Stochastic force due to thermal motion of the liquid molecules. This is especially
significant for particles smaller than 1 |xm.
b. Hydrodynamic forces and torques under consideration of wall effects — Due to the
small particle -Reynolds numbers (Rep< < l ) commonly encountered in liquid deep bed
filtration, the hydrodynamic forces and torques acting on a particle can be calculated
on the basis of the Stokes equation. Under such creeping conditions, the behavior of
a spherical particle in the proximity of a smooth, rigid collector surface can be reduced
to a linear superposition of several special cases :45

* In the papers of Rajagopalan and Tien248 the removal efficiency of one collector (for Happel’s cell model) is
defined by using the total cross section of the cell. Thus

7o = ^(i - to)‘,5-in

Tj = J/tt • b2 • vf • Cj
 Water, Wastewater, and Sludge Filtration

• Translation of the particle in a stationary fluid

• Rotation of the particle in a stationary fluid
• A stationary particle in various flow fields

These special cases have been treated by several authors. Approximating the collector
surface to be a smooth plane wall, which may be justified by NR = dp/dK ^ 10 2,
the resulting hydrodynamic forces and torques acting on a freely moving particle can
be calculated. This procedure is discussed in more detail elsewhere. 2 5
Gravity force
London-van der Waals’ force — This force is the result of the interaction between
fluctuating dipoles of atoms, due to the rotation of their electrons .49 It is usually an
attractive force between solid surfaces. In colloid chemistry as well as for particle
aggregation and deposition, this force is a very important “ surface” -force (being
effective between surface distances up to several 1 0 0 nm).
There exist two different theories, the microscopic50' 52 and the macroscopic model,53,54
to describe the van der Waals’ interactions between solids. For the theoretical treatment
of the above-mentioned processes, the microscopic theory is commonly used, as it is
discussed in more detail elsewhere .5 Here, the indication may be sufficient that the
microscopic theory allows a distinction of two terms, the first one describing the
geometric situation, and the second one describing the properties of the interacting
materials. Thus, to calculate the van der Waals’ force between a spherical particle and
a smooth collector surface, one may use the following correlation proposed by
Gregory :55

(36)

with H 132 = Hamaker constant for two interacting solids (1 and 2) immersed in a
liquid (3). For particles in water, H 132 commonly ranges from 10 -21J to 10 - 19J ;56 and
X = characteristic electromagnetic wave-length (about 1 0 0 nm).
For very small surface distances ( 8 ^ 3 nm) the calculated van der Waals’ force
(according to Equation 36) begins to deviate significantly from measured values. This
is partly due to structural effects of the liquid and to the graininess of real surfaces
(in atomic dimensions). Thus, for “ clean” surfaces (i.e., those free of any adsorbed
macromolecules) a molecular contact distance 8 0 of about 0.4 to 1 nm is usually
considered as a minimum value for 8 .
Electrical double layer force — The solids dispersed in water commonly bear an
electrical surface charge arising, for example, from the dissociation of surface groups
or from the specific adsorption of ions. The resulting double layer force of two
interacting solid surfaces is one of the important aspects in colloidal stability. There
exists a lot of theoretical work to describe this force on the basis of the Poisson-
Boltzmann equation. As a good approximation for a sphere interacting with a plane
(representing the collector), one may use :56

fDL = 128^nkTap #
Y, • Y 2 - e xp [- K( 5 - S0)] (37)
K
 35

with
Yj = tanh(ze^s j/4kT)

Y 2 = tanh(ze'4/'s 2/4kT)

i/k=1 fa eT• k • T
e V 2 ( « , - z?)
i

= Debye-Hiickel-length.

As one can see from Equation 37, f°L is strongly dependent on the surface potential
of the solids: ^ s l or 2 (which may be changed by adsorbed polyelectrolytes) and
on the ionic strength of the salts dissolved in water. It should be mentioned that the
same sign of charge or potential on both surfaces leads to a repulsive force.
A more detailed discussion of double layer forces with respect to deep bed filtration
in water technology is given in Reference 5.
f. Structural forces of liquid layers close to solid surfaces — For extremely small surface
distances between immersed solids ( 8 ^ 3 nm), and due to the structure, repulsive
forces can arise which may occur in the first liquid layers adjacent to the surfaces. 57 58
In water, these forces are called hydration forces. In a complicated way, they depend
on the properties of the solid surfaces and of the surrounding water (e.g., with respect
to ionic content and pH -value). A comprehensive investigation of these forces was
started some years ago. Thus, a general quantitative description is not yet possible,
especially for real systems that are important in water technology .5
g. Steric forces and other effects of adsorbed molecules — Molecules adsorbed on the
collector or turbidity surface may influence directly or indirectly the interacting surface
forces. With adsorbed macromolecules, these effects may be extremely pronounced.
In an indirect way, adsorbed layers can change the van der Waals’ or the double-layer-
interaction forces . 59' 64 In a direct way, adsorbed macromolecules can create steric
repulsion forces, which are mainly based on the decrease of entropy of an interacting
system by the overlapping of adsorbed layers . 63’65,66
Another direct effect of adsorbed macromolecules or polymers is the so-called
bridging between approaching surfaces of solids .67' 71 Besides, it has been shown that,
for anionic polymers, there may be even a gel formation between the dispersed sur­
faces . 72 Thus, the interaction forces due to adsorbed molecules are quite manyfold,
and it is not yet possible to specify them in a general way.
h. Contact forces — These forces act on particles in atomic contact with a collector
surface. The contact may be caused by immediate touching of the clean surfaces of
the solids or by an intermediate compact layer of adsorbed molecules. 73 For such small
surface distances ( 8 ^ 1 nm), the correlations to calculate van der Waals’ and double
layer forces as discussed under (d) and (e) can be regarded only as a rough approxi­
mation .74 The reason for this may be, for example, strong electrostatic fields of atomic
dimensions , 74' 76 Ca2+-bridges between negatively charged surface groups , 77 or hydro­
gen bonds .78
These attractive forces are finally in a static equilibrium with the Bom repulsion
forces that arise from the impermeability of the electron clouds. This equilibrium leads
to a certain elastic and/or plastic deformation of the surfaces of the solids depending
on their properties.79' 81

A critical consideration of the various forces mentioned above leads to the following
conclusion (inertia forces can usually be neglected in liquid deep bed filtration):
36 Water, Wastewater, and Sludge Filtration

One should regard a particle to be in contact with the collector surface only if typical
short range forces, as they are discussed under (f), (g), and (h), are no more negligible
against the ones under (a) to (e). Most of the fundamental work on the microscopic particle
deposition mechanisms takes into account the forces under (a) to (e). Thus, all these should
be understood as theories to describe the particle transport behavior leading to y^T-

2. The Elementary Filter Efficiency fo r Brownian Particles (dp ^ 1 pm)

For this case, Levich’s solution of the convective diffusion equation 82 was applied to the
flow field according to Happel’s cell model. This leads to :83,84

y;.T = 3.98 • Ai' 3 • N Pe2/3 (38)

Due to the assumed boundary condition (c = 0 at the collector surface), an irreversible

particle adhesion is implied (corresponding to y 0 H = 100%). Furthermore, no surface forces
or hydrodynamic wall effects are considered in the derivation of Equation 38. Such effects
are taken into account, for example, by Kim and Rajagopalan . 85 Approximating the numerical
results for the solution of the extended convective diffusion equation, they obtained the
following correlation:

_ _________________ 4
^ • A 1/3 • N Pe_____________________
" 2/3
7 o,t ~ 1 + 0.5 • NR • B • exp(0.96Vmax/kT)

with

B - NLo- ° 22 + 0.05 N Lo2 7

NLo - H 132/ 6 kT
= dimensionless Hamaker-constant
Vmax = maximum of the potential barrier, due to the superposition of van der Waals’
and double-layer interactions

Due to the small size of the Brownian particles, the drag forces and torques acting on it
at the collector surface are usually quite small. Thus, in many practical cases the adhesion
probability may be approximated by 1 0 0 % . 5

3. The Elementary^ Filter Efficiency fo r Non-Brownian Particles (dp ^ 1 pm)

a. Transport Efficiency
Particles, which are not significantly influenced by the thermal motion of liquid molecules
(dp ^ 1 jxm), are moving on deterministic trajectories around a collector. Using an appropriate
flow model for the filter bed, these trajectories can be calculated by balancing the various
forces and torques acting on a particle. From these particle trajectories the transport efficiency
for one collector yj x can be obtained by applying the principle of the “ limiting trajectory’’,
as it is shown schematically in Figure 8 . Hereby, one has to distinguish between two types
of limiting trajectories.
A certain kind of particle moving on the trajectory of type I comes into contact with the
collector surface at 0p = it 12. Such trajectories are obtained by neglecting any interactions
between particle and collector surface. All particles of the same kind approaching the collector
closer to the flow axis as the one on the limiting trajectory will come into contact with the
collector at 0p < tt/2. Particles (again of the same kind) approaching the collector outside
the limiting trajectory will not touch it. Considering surface interactions (van der Waals’
forces, etc.), limiting trajectories of type II are obtained, which come to an end at 0p = it.
In both cases, yojX results from the particle flux Jx entering the body of revolution that is
 37

FIGURE 8. Schematic representation of different limiting trajectories around a

spherical collector.

generated by the limiting trajectory. Particles far away from the collector are not yet influ­
enced by the collector surface and therefore their trajectories coincide with fluid streamlines.
Thus, one obtains for JT:

Jx = it • (AG) 2 • vf • Cj (40)

with Equation 32, yo,T results in:

(AG) 2
7 o ,t = — 7^ (41)

The limiting trajectory is a special solution of a differential equation describing the particle
trajectories around the collector. This equation is obtained by balancing the forces and
torques acting on a particle, which finally leads to a nonlinear ordinary differential equation
(of first order) in the following general dimensionless form:

dr*
—E = f(0p, r*, various dimensionless numbers) (42)
dOp
38 Water, Wastewater, and Sludge Filtration

with rp* = rp/ap.

Generally, Equation 42 must be solved using numerical methods. The implied charac­
teristic dimensionless numbers depend on the forces considered in the trajectory calculations.
From the solution of Equation 42 with the appropriate boundary conditions for the limiting
trajectories, AG (and consequently y^T) ls obtained.
Without going into details (these are discussed elsewhere2,5), a typical example for the
results obtained with the discussed procedure is given by the correlation of Rajagopalan and
Tien .48 They considered Happel’s model for the flow field around a spherical smooth collector
and took into account drag forces and torques as well as gravity and van der Waals’ forces.
The approximation of their numerical results is given by:

7 S,t = (1 - e0) - 2/3 • AJ0.72 • NJ& • N >5/8

+ 2.4 • 10 ~ 3 • N6/5 • Nr 2/5] (43)

with Nvdw = H 132/9'TTTiap2 vf.

In various papers, the effect of electrical double layer forces has been discussed in more
detail. 86' 88 From these one can conclude that, under common conditions in water treatment,
these forces either have only a minor influence on y \ T or> if theyare strongly repulsive,
they almostcompletely prevent particle deposition. According to Spielmanand Cukor,87 the
double layer forces may be neglected (using Happel’s cell model) as long as the following
semiempirical relationship is satisfied:

T V , • Ka,, -|
V Nn vdw
exnl — ­
N„ NvdW • k 2 • expL ~ N Rr ^ .-4 .4I J
NJ • A, ' r v l ^ ...............
• Ka "I< 1 <44)

with = 3 e0 • er • ¥ Sfl • 2 • H132.

Comparing the theoretical results calculated from the limiting trajectories with experi­
mental values, partly good agreements and partly significant discrepancies have been ob ­
tained . 5,46,48,86,89 94 An example is illustrated in Figure 9.
In this graph, the averaged values of experimentally determined elementary filter effi­
ciencies y 0 are compared with theoretical results (according to Equations 43, 35, and 34).
In these experiments, the initially negative surface charge of the turbid matter (quartz
particles) has been compensated by adsorbing a cationic polyelectrolyte (Praestol 222K)
such that the electrophoretic mobility of the particles was about zero. Because this corre­
sponds to tySjl = 0, Relationship 44 is satisfied. In the theoretical results, an adhesion
probability y 0 H = 100% was assumed. This is in line with the original work ,48 where an
infinitely high van der Waals’ force is at least formally implied in the theory according to
the limiting case of diminishing surface distances ( 8 = 0 in Equation 36).
For relatively large particles, the experimental results are significantly smaller than the
theoretical ones. As will be discussed in more detail later, this behavior can be explained
by an insufficient particle adhesion, which is in contrast to the 1 0 0 % assumed for y 0 Hbefore.
However, the too high experimental filter efficiencies for smaller particles cannot be ex ­
plained in the same way. Thus, it is obvious that there must exist additional transport
mechanisms which have not been considered in the theory so far.
Looking at calculated limiting trajectories in more detail, one can show that a particle
moving on such a trajectory along the collector surface has an extremely small surface
 39

FIGURE 9. Theoretical (according to Equation 43) and experimental results for the filtration
of quartz particles in quartz sand filters under addition of a cationic polyelectrolyte
(16 fxg/€ Praestol 222K).

distance to it. In water treatment, this distance is commonly smaller than 1 |xm, when the
particle has left the proximity of the flow axis (i.e., 0p ^ 10° ) . 5 Thus, one can conclude
that roughness on the collector surface (which may reach heights up to several 10 |xm) must
increase the probability for a particle to come into contact, leading to an increase of the
transport efficiency.
To consider such effects theoretically, a model has been developed, in which a so-called
“ contact shell” around a spherical collector surface is introduced .512,95 The thickness RK
of this shell is governed by the mean height of the surface protrusions. A particle entering
this shell is considered to come into contact with the collector.
Using Happel’s cell model and taking the limiting trajectories of type I as a basis (see
Figure 8 ), the following correlation has been obtained from the analytical solution of Equation
42:

7o,T = [ I ■As • N 2 • (Yp 2 + NG] • exp( —2K2) (45)

40 Water, Wastewater, and Sludge Filtration

FIGURE 10. Theoretical (according to Equation 45) and experimental results for the filtration of quartz particles
in quartz sand filters under addition of a cationic poly electrolyte (16|xg/€ Praestol 222K). (In theory, the roughness
of the filter grain surface and the particle adhesion probability have been taken into account.)

2y ; • n r \
with k 2 = V 2 IV 3 A S • ( ir/2 — arctg
V 8N g/3As/

= Rk + ap
and P

Figure 10 shows the same experimental results as Figure 9, but now they are compared with
the theoretical values according to Equation 45. In the theory, a thickness of the contact
shell of Rk = 10 |xm is considered. This value represents an average height of the surface
protrusions on the used quartz filter grains, as it was determined from SEM investigations.
Furthermore, a particle adhesion probability y 0 H < 100% has been taken into account, as
will be discussed later in more detail. Generally, one can notice from Figure 10 that the
experimental results for small particles as well as for larger ones are now in a fairly good
agreement with the theoretical results.
The experimental verification of the positive effect of surface roughness on the filter
efficiency was achieved by using spherical glass collectors (dK = 1.5 to 2.0 mm) on which
protrusions of different heights have been generated. This has been done by sintering quartz
particles on the collector surfaces. The different portions of these particles had mean diameters
of about 2 |xm (OR 2), 10 pirn (OR 10), and 40 |xm (OR 40). The measured improvements
of the filter efficiencies are shown in Figure 11. As one can see, due to the surface roughness,
the increase in filter efficiency is as much as 300%.
 YO.OR/ Yo,smooth
particle diameter in \im particle diameter in \im

FIGURE 11. Improvement of the initial efficiency of deep bed filters by various surface protrusions on spherical glass collectors
(A) vf = 10 m/h; (B) vf = 20 m/h).
42 Water, Wastewater, and Sludge Filtration

In addition to the effects of roughness, one can also expect advantages of cavities in the
collector surface in relation to transport efficiency. In such cavities, a favored particle
deposition may take place due to the relatively low hydrodynamic forces acting on the
particles. This is discussed in more detail elsewhere .5 96 98 Here, only one example is shown
in Figure 12, where the improvement of the initial filter efficiency for differently structured
filter grains is plotted vs. the diameter of the filtered quartz particles. It is obvious that one
can achieve a remarkable increase in filter efficiency in relation to relatively smooth glass
spheres using strongly structured filter materials such as pumice or filter coke. These materials
have distinct roughness as well as pronounced cavities in their surface.

b. Adhesion Probability
The significance of the particle adhesion process in deep bed filtration has been dis­
cussed for years in the literature . 1219’22,99101 An indication for a particle adhesion probability
< 100% is also given by the results shown in Figure 9.
The basis for quantifying the adhesion probability is a balance of the forces and torques
acting on a stationary particle which is in contact with the collector surface. Due to the
commonly small NR-values (NR ^ 10“ 2) this surface can be regarded as a plane wall. A
simplified model to estimate y0H is shown in Figure 13. A spherical particle with nano­
roughness on its surface is regarded in a linear flow field of shear gradient k . A characteristic
surface protrusion of height RHhinders the particle from slipping or rolling along the surface.
For a stable position of the particle, the net adhesion force frres must be at least as large as
the net detachment force frmin, which results from a torque balance around the point of
rotation. Thus, the adhesion condition can be formulated as follows:

fres > fm,„ = (tD)m + (ap - r h) • [ ( f ? r + ,46.

r " r (Rh ' 2 ap - R’ )'*

Due to the difficulties of quantifying the net adhesion force, which is the result of various
short-range and long-range surface forces (as discussed above), it is not yet possible to
calculate frres only by use of theoretical approaches. Therefore, these forces have to be
determined with special experiments under conditions relevant for water treatment. For
example, this hasbeen done for the adhesion of quartz particles andglass microspheres on
quartz plates, whichled to the results given in Figure 14. From such adhesion force distri­
butions, one obtains an estimation for y 0 H by setting the calculated value for frmin equal to
the applied separation force in the model system. With this procedure, the y 0 Hvalues implied
in the theoretical results shown in Figure 10 have been obtained. It is discussed in detail
elsewhere . 5’22,95
A more realistic theoretical treatment of adhesion behavior should take into account the
general possibility for multiple contacts of a particle on the collector surface. As is shown
schematically in Figure 15, depending on the point of incidence, each particle that does not
come to a stable deposition at this point may have more-or-less additional adhesion possi­
bilities on its way along the collector surface. Thereby, the adhesion probability for each
situation decreases with increasing 0 (as long as 0 < tt/ 2 ), which is due to the increase of
the wall shear forces. Thus, a particle approaching the collector exactly on the flow axis
and coming into contact at 0 = 0 has an adhesion probability of 100%. On the other hand,
a particle approaching the collector on a limiting trajectory of type I (i.e., coming into
contact with the collector surface at 0 = tt / 2 for the first time) has a minimum adhesion
probability at this position.
To determine the mean overall adhesion probability for a large number of particles con ­
tacting the collector (mathematically, the expectation value for y 0 ,h ) > one has to integrate
the adhesion probabilities for particles with different points of incidence over the total
 spheres
glass
Yo/ Yo
10 15 20 25 30 35
particle diam eter in |xm particle diam eter in p.m

FIGURE 12. Effect of differently structured filter materials on the initial filter efficiency: (A) without polymer addition; (B) with
16 |xg/€ 222K.
44 Water, Wastewater, and Sludge Filtration

®Q
(fgr*f

collector surface
yr-
point of rotation
res

FIGURE 13. Model to describe the adhesion probability of a stationary particle in contact with the collector
surface.

separation force (10~10 N)

FIGURE 14. Adhesion force distribution of quartz particles and micro glass spheres on quartz
plates in tap water (normal and tangential separation forces, dp = 30 to 35 |±m)
 ----------------- d k ------------------ ---

FIGURE 15. Model to describe the particle adhesion behavior under

consideration of multiple contacts on the collector surface.

collector surface. This has to be done by taking into account the varying probabilities for
the realization of each point of incidence. For such a treatment, one has to consider the so-
called dynamic adhesion behavior, demonstrated as an example in Figure 16. As one can
see, there exists a strong dependency of the adhesion probability (corresponding to the
amount of adhering particles) from the time of contact between particle and quartz plate.
Here, the separation force acting on the particles after contact was the gravity force normal
to the plate surface. Furthermore, it may be seen that the addition of a cationic polyelectrolyte
(100 |xg/€ of Praestol 222K) considerably improves the adhesion probability even for very
small contact times.
The consideration of the dynamic adhesion behavior in the theoretical treatment of deep
bed filtration in water treatment is discussed in more detail else where. 5,22,99 Despite a certain
progress which has been achieved for this fundamental problem, there still remains a lot of
research work to be done in this field.
 46 Water, Wastewater, and Sludge Filtration

FIGURE 16. Dependence of the adhesion probability from the time of contact between particle and plate with and
without polyelectrolyte addition (100 |xg/€ 222K, adsorption time for the cationic polyelectrolyte: 2 h).

C. The Filter Efficiency in the Dynamic Filtration Phase

In the dynamic filtration phase, the efficiency of a deep bed filter is increasingly influenced
by the amount of deposit collected in the filter bed. Generally, to calculate the particle
deposition in this phase on a microscopic basis, one has to consider the same relations for
the relevant parameters as has been already discussed for the initial phase. Additionally,
one has to take into account the following effects:

• Change of the geometry of the filter bed by the deposit of turbid matter. This modifies
the hydrodynamic forces and torques influencing the particle deposition mechanisms.
• The surface forces are no more restricted to particle-collector interactions; however,
additional particle-particle interactions have to be considered.
• By modified flow conditions, as well as by impinging effects due to approaching
particles, detachment of already deposited single particles or particle agglomerates
may take place.

Most of the theoretical work related to the dynamic filtration phase is restricted to the
macroscopic description of the filter behavior. This is because a comprehensive microscopic
descripton of the particle removal processes is not yet available for the initial filtration phase
and because such a treatment is even more complicated by the effects of the deposits in the
filter. Only for very simplifying assumptions do there exist some microscopic approaches
to describe the dynamic filtration phase. However, these are not yet applicable to practical
problems.
Considering a filter bed consisting of initially smooth collectors, one can distinguish
between the following typical cases:
 47

• Formation of a homogeneous layer of deposit (with constant thickness) around the

collector surface. Such behavior has been partly observed for the removal of hydroxide-
flocs in model filters . 19102103
• Formation of a homogeneous layer of deposit mainly on the upstream side of the
collector. Such morphologies, resulting in a relatively low increase of pressure drop,
have been observed with particles of high density, such as quartz, kaolin, and lime­
stone . 5,102’104
• Formation of a quite heterogeneous morphology of deposit by favored agglomeration
at the collector surface. Such agglomerates may be detached by drag forces and
subsequently held back in constrictions of filter pores . 105 This can block single pores
or even whole regions of the filter bed.

For the case of a homogeneous layer of deposit with constant thickness Wnek et al . 106
considered both a mass balance and an electrical charge balance and their effect on the
deposition mechanisms. For this, they made some very simplifying assumptions and obtained
only partial agreement between theory and experiment. 107 This also applies for the results
of Rajagopalan and Chu , 108 where a good agreement between theory and experiment was
only possible by introducing a fitting constant in the kinetic approach (to account for particle
detachment), which corresponds to the macroscopic theory of Mints. 13
Tien et al . 35 proposed a model that distinguishes between two phases of the dynamic
behavior of a deep bed filter. In the first phase, they consider a spherical collector model,
assuming the deposit to be distributed in a homogeneous shell around the collector and
describing the flow field according to Happel’s model. Assuming further negligible double
layer forces and an adhesion probability of 1 0 0 %, it was possible to derive the correction
function fx (cr, Y) in an explicit manner on the basis of Equation 43. In a similar way,
fp (cr, Y) has been derived on the basis of the Carman-Kozeny equation. For the second
phase they considered a model of diverging and converging capillaries to describe the
geometry of the filter bed. In the correction functions fx and fp, the increasing portions of
blocked capillaries with increasing deposit were taken into account. Despite the very much
simplified assumptions, quite a good agreement between theory and experiment has partly
been obtained, at least in their tendency. But, indeed, the determination of a certain amount
of deposit “ crtrans” characterizing the transition from phase 1 to phase 2 represents some
kind of fitting process.
To model the dynamic filter behavior without any assumptions concerning the deposit
morphology, Pendse and Tien 109,110 simulated the formation of the morphology by the
subsequent deposition of single particles, again under the assumption of an irreversible
particle adhesion. But the comparison of theoretical with experimental results showed a too
high increase of the filter efficiency in theory with increasing deposit.
In later works , 18,111,112 a better agreement between such theoretical simulations and ex ­
periments was obtained by considering a particle adhesion probability < 100%. But there
are still remarkable discrepancies. The reason for these may be seen in the strongly simplified
particle trajectory calculations in these simulations on the basis of single particle behavior,
as well as in particle adhesion behavior which is not realistic enough. For example, the
latter has been considered without taking into account the dynamic particle adhesion process
as it has been discussed above. Thus, to simulate the dynamic phase of deep bed filtration
on a very fundamental basis, there still remains a lot of work to be done in the future.
Due to some qualitative considerations, 5 one can expect for this phase of filtration that
the structure of the filter material should have a strong influence on filter behavior, as it
has been already discussed for the initial phase. An example is demonstrated in Figure 17.
Besides the positive effects of strongly structured filter materials in the initial phase, there
is also a remarkable enhancement of the ultimate specific deposit crs for materials like filter
 dK = 1,5 - 2,0mm
100mg/l Sikron H200
100mg/l Sikron H200
tap water
water dp = 5 pm

filter coke grains

filter efficiency
coke grains

2 3 U 5
deposit o • 103
B

FIGURE 17. (A) Change of the elementary filter efficiency y for 25 |xm-quartz particles with increasing deposit a by using differently structured filter materials
(L = 15 cm, vf = 10 m/h). (B) Change of the elementary filter efficiency y for 5 (xm-quartz particles with increasing deposit a by using differently structured filter
materials (L = 15 cm, vf = 10 m/h)
 49

o
c

?►
-
Q)
0t_)
O)
0)
*o
o>
o
E
0)

I>
—N
o
c
0)
E
0)

specific deposit 6 in 10

FIGURE 18. Change of the elementary filter efficiency with increasing deposit for different additions of a cationic
polyelectrolyte.

coke in comparison to relatively smooth glass spheres. These effects may be partly explained
by the relatively low shear forces acting on particles deposited in the cavities of the filter
grain surface.
The importance of the particle adhesion behavior under conditions of larger deposits is
also demonstrated in Figure 18. Here, the elementary filter efficiency is plotted vs. the
specific deposit collected in the filter bed for different amounts of one cationic polyelectrolyte
(222K) added to the model suspension. As one can see, by the addition of 40 jxg/€ as well
as of 300 |xg/€, in both cases there was a remarkable improvement of y0 and of a s. But the
lower polyelectrolyte addition leads to a higher initial filter efficiency in comparison to the
results with 300 (xg/€ of 222K. On the other hand, the higher polyelectrolyte addition causes
a higher ultimate deposit. This can be qualitatively explained by the fact that, with increasing
deposit, the particle adhesion process determines more and more the overall filter behavior.
In the same way, particle adhesion is improved by higher amounts of adsorbed poly elec­
trolytes; however, it was observed that smaller amounts mainly enhance the particle transport
mechanisms." But such effects, which may be quite important (especially for practical
conditions), are not yet predictable on the basis of a fundamental theory for the dynamic
phase of deep bed filtration.

ACKNOWLEDGMENT

The author gratefully acknowledges the assistance of Mrs. Dagmar Schimer and Mrs.
Dipl.-Ing. Pia Lipp, M.Sc. in preparing the manuscript.
50 Water, Wastewater, and Sludge Filtration

LIST OF SYMBOLS

Symbol Dimension

a Fitting constant
aK Collector radius m
ap Particle radius m
a* Empirical constant describing particle detachment
A Constant
As Porosity function (Equation 30)
b Outer cell radius according to Happel’s model m
B Function of (Equation 39)
c Volume concentration of one type of turbidity
Ci Volume concentration of particles of type i
ci Concentration of turbidity in the elementary filter layer j
Cv Total volume concentration of turbidities
Co Volume concentration of one type of turbidity in the filter inlet
dK Grain diameter m
Particle diameter m
Dd Axial dispersion coefficient m2/s
DP Particle diffusion coefficient (Equation 15) m2/s
e Elementary charge As
Fitting constants
e Unit vector 1
Correction function for the course of the pressure within 1
the filter bed
P min
r Adhesion force at least necessary for stable particle deposition N
P res
Net adhesion force normal to the collector surface N
PC 0 -Component of the gravity force N
0

(feD)m 0 -Component of the drag force (parallel to the collector sur­ N

face)
fx Correction function for the filter coefficient 1
f. General function of a
F Cross section of the filter bed m2
g Acceleration of gravity ms-2
H Hamaker constant Nm
Io Modified Bessel-function of first kind and order zero 1
J J-function 1
J Particle flux m 3s _ 1
Jx Flux of particles coming into contact m3s -1
k Boltzmann constant NmK-1
L Total filter bed depth m
n Number of particles 1
n Ion concentration m-3
nK Number of collectors in an elementary filter layer
NG Gravity number (Equation 15)
N, Inertia number (Equation 15)
NLo Dimensionless Hamaker constant (Equation 39)
Npe Peclet number (Equation 15)
Nr Interception number (Equation 15)
 51

LIST OF SYMBOLS (continued)

Symbol Dimension

NydW Number for the influence of van der Waals’ forces (Equation 1
43)
N* Number for the influence of the electrical double layer forces 1
(Equation 44)
OR Artificial roughness on the grain surface m
P Porosity function 1
P Pressure Nm - 2
Pv Pressure drop Nm “ 2
Po Pressure in the filter inlet Nm ” 2
r Radial coordinate m
rP Radial coordinate of the center of the particle m
r* Dimensionless radial coordinate (Equation 29) 1

rP Dimensionless radial coordinate of particle center (Equation 42) 1

Rh Characteristic height of surface protrusion for particle adhesion m

Rk Thickness of the “ contact shell” around a collector m
t Filter running time s
T Absolute temperature K
vf Filtration rate m s- 1
vr Radial component of v m s- 1
v0 0 -component of v m s- 1
V Fluid velocity (vector) m s-1
w Porosity function (Equation 29) —
y1 1 /2 Function of the surface potential (Equation 37) 1
Y Parameter vector —
z Filter bed depth m
z Valence of counterions —
z i Valence of ions of type i —
(tD)m Torque acting on a stationary particle in a shear field Nm
8 Distance between particle and collector surface m
7 Elementary filter efficiency 1
7 Mean y-value for the total filter bed height 1
7o Elementary filter efficiency in the initial phase 1
7o Particle removal efficiency of one collector in the initial 1
phase
7 o ,h Particle adhesion probability in the initial phase 1
7 o ,T Particle transport efficiency in the initial phase 1

7 o ,T Transport efficiency of one collector in the initial phase 1

AG Distance from the flow axis m

Apv Pressure drop in a filter bed Nm - 2
Apv,o Pressure drop in a clean filter bed Nm ~ 2
e Porosity of the filter bed 1
Relative dielectric constant 1
eo Absolute dielectric constant As V _1m
^0 Porosity of the clean filter bed 1
i Zetapotential V
£p Zetapotential of particle V
Dynamic viscosity kg m _ 1s~
52 Water, Wastewater, and Sludge Filtration

LIST OF SYMBOLS (continued)

Symbol Dimension

^1 Particle removal efficiency of one collector 1

0 Angular coordinate —
0p
Angular coordinate of the particle center —
K Wall shear gradient s _1
K Reciprocal thickness of the double layer (reciprocal Debye- m- 1
Hiickel length)
X Filter coefficient m- 1
X Characteristic electromagnetic wavelength m
^■0 Filter coefficient for the initial phase m- 1
V Kinematic viscosity m 2s - 1
pf Liquid density kg m 3
Pp Particle density kg m ~ 3
a Specific deposit of a distinct type of particle 1
a Mean o -value for the total filter bed height 1
Specific deposit of particles of type, i 1
Ultimate specific deposit 1
o v Total specific deposit 1
T Corrected time variable (Equation 3) s
\Jr Stream function m3s ~ 1
V Electrical potential V
% Stem - or surface-potential V

REFERENCES

1. Herzig, J. P., Leclerc, D. M., and Le Goff, P., Flow of suspensions through porous media, Ind. Eng.
Chem., 62, 8, 1970.
2. Rajagopalan, R. and Tien, C., The theory of deep bed filtration, in Progress in Filtration and Separation,
Vol. 1, Wakeman, R. J., Ed., Elsevier, Amsterdam, 1979, 179.
3. Tien, C. and Payatakes A. C., Advances in deep bed filtration, AIChE J., 25, 737, 1979.
4. Iwasaki, T., Some notes on sand filtration, J. AWWA, 29, 1591, 1937.
5. Gimbel, R., Abscheidung von Triibstoffen aus Fliissigkeiten in Tiefenfiltem, in Veroffentlichungen des
Bereichs und des Lehrstuhls fur Wasserchemie am Engler-Bunte-lnstitut der Universitat Karlsruhe, Vol.
25, Bereich und Lehrstuhl fur Wasserchemie am Engler-Bunte-Institute der Universitat Karlsruhe, Karlsruhe,
1, 1984.
6. Lister, M., The numerical solution of hyperbolic partial differential equation by the methodof character­
istics, in Mathematical Methods for Digital Computers, Ralston A. and Wilf, H. S., Eds., John Wiley &
Sons, New York, 1960, 165.
7. Lapidus, L., Digital Computation for Chemical Engineers, McGraw-Hill, New York, 1962.
8. Aris, R. and Amundson, N. R., Mathematical Methods in Chemical Engineering, Vol. 2, Prentice Hall,
Englewood Cliffs, N.Y., 1973.
9. Filtration in der Wasseraufbereitung, I. Grundlagen, DVGW-Arbeitsblatt W 210, ZfGW-Verlag, Frankfurt,
1983.
10. Ives, K. J., Rapid filtration, Water Res., 4, 201, 1970.
11. Rolke, D., Vergleichende Untersuchungen an Trocken- und Uberstaufiltem zum Mechanismus der Parti-
kelablagerung in Kiesbettfiltem, Dissertation, University of Karlsruhe, Karlsruhe, 1973.
12. Gimbel, R., Untersuchungen zur Partikelabscheidung in Schnellfiltem, Dissertation, University of Karlsruhe
(TH), Karlsruhe, 1978.
13. Mints, D. M., Modem Theory of Filtration, Int. Water Supply Assoc., London, 1966, PI.
14. Ives, K. J., Theory of Filtration, Int. Water Supply Assoc., London, Vol. 1, 1969, K3.
 53

15. Ives, K. J., Capture mechanisms in filtration, in The Scientific Basis of Filtration, Ives, K. J., Ed.,
Noordhoff, Leyden, 1975, 183.
16. Ives, K. J., Mathematical models of deep bed filtration, in The Scientific Basis of Filtration, Ives, K. J.,
Ed., Noordhoff, Leyden, 1975, 203.
17. Spindler, P., Modelle und Beschreibungen von Filtrationsvorgangen: Konzentrationsverlauf, in Ver-
offentlichungen des Bereichs und des Lehrstuhls fiir Wasserchemie am Engler-Bunte-lnstitut der Universitat
Karlsruhe, Vol. 5, Bereich und Lehrstuhl fiir Wasserchemie am Engler-Bunte-Institute der Universitat
Karlsruhe, Karlsruhe, 1971, 92.
18. Tien, C. and Gimbel, R., On the development of a comprehensive model of deep bed filtration, Symp.
Preprints Water Filtration, Koninklijke Vlaamse Ingenieursvereniging, Antwerp, 1982, 1.1.
19. Stein, P. C., A Study of the Theory of Rapid Filtration of Water through Sand, D. Sc. thesis, Massachusetts
Institute of Technology, Cambridge, Mass., 1940.
20. Feben, D., Theory of flow in filter media, J. AWWA, 52, 940, 1960.
21. Adin, A. and Rebhun, M., Components of deep-bed filtration-mathematical model, Symp. Preprints Water
Filtration, Koninklijke Vlaamse Ingenieursvereniging, Antwerp, 1982, 1.15.
22. Zitzmann, W., Sehn, P., and Gimbel, R., Zur Bedeutung der Partikelhaftung bei der Tiefenfiltration, in
Verdffentlichungen des Bereichs und des Lehrstuhls fiir Wasserchemie am Engler-Bunte-lnstitut der Uni­
versitat Karlsruhe, Volume 20, 1982, 561.
23. Anon., Water Filtration — The Mints-Ives controversy 1970 — 73, Filtr. Sep., 13, 131, 1976.
24. Heertjes, P. M. and Lerk, C. F., The functioning of deep bed filters, Trans. Inst. Chem. Eng., 45, T129,
1967.
25. Spindler, P., Beitrag zur Beschreibung des Konzentrationsverlaufes in Schnellfiltem der Wasseraufbe-
reitung, Dissertation, University of Karlsruhe, Karlsruhe, 1973.
26. Ives, K. J., Simplified rational analysis of filter behaviour, Proc. Inst. Civ. Eng., 25, 345, 1963.
27. Shekhtman, Y. M., Filtration of suspensions of low concentration, Institute of Mechanics of U.S.S.R.
Academy of Science, Moscow, 1961.
28. Maroudas, A. and Eisenklam, P., Clarification of suspensions: a study of particle deposition in granular
media. II: A theory of clarification, Chem. Eng. Sci., 20, 875, 1965.
29. Sherwood, Th. K., Pigford, R. L., and Wilke, C. R., Mass Transfer, McGraw Hill, New York, 1975.
30. Thomas, H. C., Heterogeneous ion exchange in a flowing system, J. Am. Chem. Soc., 66, 1664, 1944.
31. Adin, A. and Rebhun, M., A model to predict concentration and head-loss profiles in filtration, J. AWWA,
69, 444, 1977.
32. Adin, A., Solution of granular bed filtration equation, J. Environ. Eng. Div., 104 EE3, 471, 1978.
33. Happel, J., Viscous flow in multiparticle systems: slow motion of fluids relative to beds of spherical
particles, AIChE J., 4, 197, 1958.
34. Albert, G., Modelle und Beschreibungen von Filtrationsvorgangen — Filterwiderstand, in Verdffent­
lichungen des Bereichs und des Lehrstuhls Fiir Wasserchemie am Engler-Bunte-lnstitut der Universitat
Karlsruhe, Vol. 5, 1971, 74.
35. Tien, C., Turian, R. M., and Pendse, H., Simulation of the dynamic behaviour of deep bed filters,
AIChE J., 25, 385, 1979.
36. Scheidegger, A. E., The Physics of Flow through Porous Media, University of Toronto Press, Toronto,
Canada, 1974.
37. Payatakes, A. C., Tien, C., and Turian, R. M., A new model for granular porous media, Part I. Model
formulation; Part II. Numerical solution of steady state incompressible newtonian flow through periodically
constricted tubes, AIChE J., 19, 58, 1973.
38. Chow, J. C. F. and Soda, K., Laminar flow in tubes with constriction, Phys. Fluids, 15, 1700, 1972.
39. Fedkiw, P. and Newman, J., Mass transfer at high Peclet numbers for creeping flow in a packed-bed
reactor, AIChE J., 23, 255, 1977.
40. Payatakes, A. C. and Neira, M. A., Model of the constricted unit cell type for isotropicgranular porous
media, AIChE J., 23, 922, 1977.
41. Neira, M. and Payatakes, A. C., Collocation solution of creeping Newtonian flow through periodically
constricted tubes with piecewise continuous wall profile, AIChE J., 24, 43, 1978.
42. Neira, M. and Payatakes, A. C., Collocation solution of creeping Newtonian flow through sinusoidal
tubes, AIChE J., 25, 725, 1979.
43. Brinkman, H. C., A calculation of the viscous force exerted by a flowing fluid on adense swarm of
particles, Appl. Sci. Res., A 1, 24, 1947.
44. Kuwabara, S., The forces experienced by randomly distributed parallel circular cylindersor spheres in a
viscous flow at small Reynolds numbers, J. Phys. Soc. Jpn., 14, 527, 1959.
45. Happel, J. and Brenner, H., Low Reynolds Number Hydrodynamics with Special Application to Particulate
Media, Noordhoff, Leyden, 1973.
46. Fitzpatrick, J. A., Mechanisms of Particle Capture in Water Filtration, Ph. D. thesis, Harvard University,
Cambridge, Mass., 1972.
54 Water, Wastewater, and Sludge Filtration

47. Spielman, L. A. and Fitzpatrick, J. A., Theory for particle collection under London and gravity forces,
J. Colloid Interface Sci., 42, 607, 1973.
48. Rajagopalan, R. and Tien, C., Trajectory analysis of deep bed filtration with the sphere-in-cell porous
media model, AIChE J., 22, 523, 1976.
49. Chu, B., Intermolecular Forces, Interscience, New York, 1966.
50. Hamaker, H. C., The London-van der Waals attraction between spherical particles, Physica, 4, 1058,
1937.
51. Gregory, J., The calculation of Hamaker constants, Adv. Colloid Interface Sci., 2, 396, 1969.
52. Kruyt, H. R., Colloid Science, Vol. 1., Elsevier, Amsterdam, 1952.
53. Lifshitz, E. M., The theory of molecular attractive forces between solids, J. Exper. Theoret. Phys. U.S.S.R.,
29, 94, 1955.
54. Dzyaloshinskii, J. E., Lifshitz E. M., and Pitaevskii, L. P., Van der Waals forces in liquid films, J.
Exper. Theoret. Phys. U.S.S.R., 37, 229, 1959.
55. Gregory, J., Approximate expressions for retarded van der Waals interaction, J. Colloid Interface Sci.,
83, 138, 1981.
56. Gregory, J., Interfacial phenomena, in The Scientific Basis of Filtration, Noordhoff, Leyden, 1975, 53.
57. Israelachvili, J. N., Forces between interfaces in liquids, Adv. Colloid Interface Sci., 16, 31, 1982.
58. Ninham, B. W., Surface forces — the last 30 A, Pure Appl. Chern., 53, 2135, 1981.
59. Visser, J., Adhesion of colloidal particles, Surf. Colloid Sci., 8, 3, 1976.
60. Langbein, D., Theory of van der Waals attraction, in Springer Tracts in Modern Physics, No. 72, Springer-
Verlag, Berlin, 1974.
61. Osmond, D. J. W., Vincent, B., and Waite, F. A., The van der Waals attraction between colloid particles
having adsorbed layers, I. A reappraisal of the “ Void effect” ; II. Calculation of interaction curves, J.
Colloid Interface Sci., 42, 262, 1973.
62. Everett, D. H., The effect of adsorption on the interaction between solid particles, Pure Appl. Chem., 48,
419, 1976.
63. Sato, R. and Ruch, R., Stabilization of Colloidal Dispersions by Polymer Adsorption, Dekker, New York,
1980.
64. Vincent, B., Bijsterbosch, B. H., and Lyklema, J., Competitive adsorption of ions and neutral molecules
in the Stem layer on silver iodide and its effect on colloid stability, J. Colloid Interface Sci., 37, 171,
1971.
65. Lyklema, J., Principles of the stability of lyophobic colloidal dispersions in non-aqueous media, Adv.
Colloid Interface Sci., 2, 65, 1968.
66. Napper, D. H., Polymeric stabilization, Colloidal Dispersions, 43, 99, 1981.
67. Riihrwein, R. A. and Ward, D. W., Mechanism of clay aggregation by polyelectrolytes, Soil Sci., 73,
485, 1952.
68 La Mer, V. K. and Healy, T. W., The role of filtration in investigating the flocculation and redispersion
of colloidal dispersions, J. Phys. Chem., 67, 2417, 1963.
69. Kitchener, J. A., Principles of action of polymeric flocculants, Br. Polymer J., 4, 217, 1972.
70. Scheutjens, J. M. H. M. and Fleer, G. J., Effect of polymer adsorption and depletion on the interaction
between two parallel surfaces, Adv. Colloid Interface Sci., 16, 361, 1982.
71. Di Marzio, E. A. and Rubin, R. J., Adsorption of a chain polymer between two plates J. Chem. Phys.,
55, 4318, 1971.
72. Sontheimer, H. and Albert, G., Untersuchungen iiber die Wirkung makromolekularer Flockungsmittel,
Chem. Ing. Techn., 46, 487, 1974.
73. Israelachvili, J. N., Forces between interfaces in liquids, Adv. Colloid Interface Sci., 16, 31, 1982.
74. Tabor, D., Attractive surface forces, Colloidal Dispersions, 43, 23, 1981.
75. Keesom, W. H., Die van der Waals’schen Kohasionskrafte, Phys. Z., 22, 129, 1921.
76. Debye, P., Die van der Waals’schen Kohasionskrafte, Phys. Z., 21, 178, 1920.
77. Iler, R. K., The Chemistry of Silica, John Wiley & Sons, New York, 1979.
78. v. d. Tempel, M., Interaction forces between condensed bodies in contact, Adv. Colloid Interface Sci., 3,
137, 1972.
79. Krupp, H., Particle adhesion — theory and experiment, Adv. Colloid Interface Sci., 1, 111, 1967.
80. Johnson, K. L., Kendall, K., and Roberts, A. D., Proc. R. Soc. London, Ser. A., 324, 1971, 301.
81a. Muller, V. M., Yushehenko, V. S., and Derjaguin, B. V., On the influence of molecular forces on the
deformation of an elastic sphere and its sticking to a rigid plane, J. Colloid Interface Sci., 77, 91, 1980.
81b. Muller, V. M., Yushehenko, V. S., and Derjaguin, B. V., General theoretical consideration of the
influence of surface forces on contact deformations and the reciprocal adhesion of elastic spherical particles,
J. Colloid Interface Sci., 1, 92, 1983.
81c. Muller, V. M., Derjaguin, B. V., and Toporov, Y. B., On two methods of calculation of the force of
sticking of an elastic sphere to a rigid plane, Colloids Surfaces, 7, 251, 1983.
82. Levich, V. G., Physicochemical Hydrodynamics, Prentice-Hall, Englewood Cliffs, N.J., 1962.
 55

83. Pfeffer, R. and Happel, J., An analytical study of heat and mass transfer in multiparticle systems at low
Reynolds numbers, AIChE J., 10, 605, 1964.
84. Cookson, J. Th., Removal of submicron particles in packed beds, Environ. Sci. Technol., 4, 128, 1970.
85. Kim, J. S. and Rajagopalan, R., A Comprehensive equation for the rate of adsorption of colloidal particles
and for stability ratios, Colloids Surfaces, 4, 17, 1982.
86. Rajagopalan, R., Stochastic Modelling and Experimental Analysis of Particle Transport in Water Filtration,
Ph.D. diss., Syracuse University, Syracuse, N.Y., 1974.
87. Spielman, L. A. and Cukor, P. U., Deposition of non-Brownian particles under colloidal forces., J.
Colloid Interface Sci., 43, 51, 1973.
88. Onorato, F. J., The Effect of Surface Interactions on Particle Deposition in Aqueous Media — Single
Collector Study, Dissertation, Syracuse University, Syracuse, N.Y., 1980.
89. Rajagopalan, R. and Tien, C., Single collector analysis of collection mechanism in water filtration, Can.
J. Chem. Eng., 55, 246, 1977.
90. Payatakes, A. C., Rajagopalan, R., and Tien, C., Application of porous media models to the study of
deep bed filtration, Can. J. Chem. Eng., 52, 722, 1974.
91. Payatakes, A. C., Tien, C., and Turian, R. U., Trajectory calculation of particle deposition in deep bed
filtration, I. Model formulation; II. Case study of the effect of the dimensionless groups and comparison
with experimental data, AIChE J., 20, 889, 1974.
92. Fitzpatrick, J. A. and Spielman, L. A., Filtration of aqueous latex suspensions through beds of glass
spheres, J. Colloid Interface Sci., 43, 350, 1973.
93. Ghosh, M. M., Jordan, T. A., and Porter, R. L., Physicochemical approach to water and wastewater
filtration, J. Environ. Eng. Div., 101, 71, 1975.
94. Rajagopalan, R. and Tien, C., Experimental analysis of particle deposition on single collectors, Can. J.
Chem. Eng., 55, 256, 1977.
95. Gimbel, R. and Sontheimer, H., Recent results on particle deposition in sand filters, Symp. Pap. Deposition
and Filtration of Particles from Gases and Liquids, London Society of Chemical Industry, Loughborough,
Sept. 6 to 8, 1978.
96. Gimbel, R. and Sontheimer, H., EinfluB der Oberflachenstruktur von Filtermaterialien auf die Partike-
labscheidung in Tiefenfiltem, Vom Wasser, 55, 131, 1980.
97. Gimbel, R., EinfluB der Filterkomstruktur auf das Verhalten von Tiefenfiltem, gwf-wasser/abwasser, 123,
220, 1982.
98. Gimbel, R., Influence of the filter grain surface structure on the transport and adhesion mechanisms in
deep-bed filters, Proc. Symp. on Water Filtration, Elsevier, Antwerp, 1982.
99. Sehn, P. and Gimbel, R., Effect of polymers on particle adhesion mechanisms in deep bed filtration, in
Solid-Liquid Separation, Gregory, J., Ed., Ellis Horwood, Chichester, 1984, 315.
100. Gimbel, R. and Tien, C., Particle adhesion on collector surfaces — The interplay between hydrodynamics
and surface conditions of particles, Symp. on Recent Developments in Interfacial Phenomena Related to
the Environment, AIChE, preprint, Washington, D.C., 1983.
101. Gimbel, R. and Sontheimer, H., Untersuchungen zur Wirksamkeit von kationischen Polyelektrolyten bei
der Triibstoffentfemung in Schnellfiltem, Vom Wasser, 51, 65, 1978.
102. Cleasby, J. L. and Baumann, E. R., Selection of optimum filtration rates for sand filters, Bulletin 198,
Iowa State Univ. Sci. Technol. Eng. Exp. Stn. Bull., LX No. 34, 1962.
103. Gregory, J. and Wishart, A. J., Deposition of latex particles on alumina fibers, Colloids Surfaces, 1,
313, 1980.
104. Ison, C. R. and Ives, K. J., Removal mechanisms in deep-bed filtration, Chem. Eng. Science, 24, 717,
1969.
105. Payatakes, A. C., Park, H. Y., and Petrie, J., A visual study of particle deposition and reentrainment
during depth filtration of hydrosols with a polyelectrolyte, Chem. Eng. Sci., 36, 1319, 1981.
106. Wnek, W. J., Gidaspow, D., and Wasan, D. T., The role of colloid chemistry in modeling deep-bed
liquid filtration, Chem. Eng. Sci., 30, 1035, 1975.
107. Wnek, W. J., The Role of Surface Phenomena and Colloid Chemistry in Deep Bed Liquid Filtration,
Ph.D. dissertation, Illinois Institute of Technology, Chicago, 1973.
108. Rajagopalan, R. and Chu, R. Q., Dynamics of adsorption of colloidal particles in packed beds, J. Colloid
Interface Sci., 86, 299, 1982.
109. Pendse, H. and Tien, C., A simulation model of aerosol collection in granular media, J. Colloid Interface
Sci., 87, 225, 1982.
110. Pendse, H., A Study of Certain Problems Concerning Deep Bed Filtration, Ph.D. dissertation, Syracuse
University, Syracuse, N.Y., 1979.
111. Chiang, H. -H., Transient Behaviour of Deep Bed Filtration, Ph.D. dissertation, Syracuse University,
Syracuse, N.Y., 1983.
112. Chiang, H. W., and Tien, C., Transient behaviour of deep bed filters, in Solid-Liquid Separation, Gregory,
J., Ed., Ellis Horwood, Chichester, 1984.
56 Water, Wastewater, Sludge Filtration

113. Bradke, H. J., Vergleichende Filterversuche und Betriebserfahrungen mit der Filter-Konditionierung
(Flockungsfiltration) bei der Aufbereitung von Oberflachenwassem, in Veroffentlichungen des Bereichs und
des Lehrstuhls fiir Wasserchemie der Universitat Karlsruhe, Vol. 5, 1971, 236.
114. Edzwald, J. K., Becher, W. C., and Tambini, S. J., Aspects of direct filtration in treatment of low
turbidity humic waters, Symp. Preprints Water Filtration, Koninklijke Vlaamse Ingenieursvereniging, Ant­
werp, 1982, 4.39.
115. Rebhun, M., Fuhrer, Z., and Adin, A., Contact flocculation-filtration of organic colloids, Symp. Preprints
Water Filtration, Koninklijke Vlaamse Ingenieursvereniging, Antwerp, 1982, 4.25.
116. Hsiung, K. and Cleasby, J. L., Prediction of filter performance, J. Sanit. Eng. Div., 94 SA6, 1043,
1968.
117. Sontheimer, H., Flockungsfiltration, in Veroffentlichungen des Bereichs und des Lehrstuhls fiir Wasser­
chemie am Engler-Bunte-Institut der Universitat Karlsruhe, Vol. 5, 1971, 159.
118. Grohmann, A., Horstmann, J., and Sollfrank, U., Direct filtration tests with pipe flocculation, Symp.
Preprints Water Filtration, Koninklijke Vlaamse Ingenieursvereniging, Antwerp, 1982, 1.35.
 57

Chapter 3

DIRECT FILTRATION

Luciano Coccagna

TABLE OF CONTENTS

I. Direct Filtration in Water Treatment.............................................................................58

II. Application of Direct Filtration..................................................................................... 60

III. Chemical Conditioning.................................................................................................... 61

A. Disinfecting A g en ts.............................................................................................61
B. pH Adjustment..................................................................................................... 62
C. Metal Coagulants..................................................................................................62
D. Filtration A ids.......................................................................................................63

IV. Filter Design..................................................................................................................... 63

V. Limits to Direct Filtration and Advanced Systems of Filtration...............................65

A. Consideration of Limits to Direct Filtration.................................................... 65
B. Example of Improvements in Direct Filtration: OFSY®
In-Series Filtration...............................................................................................6 8
VI. Application of In-Series Direct Filtration.................................................................... 69
A. Turbidity R em oval.............................................................................................. 69
B. Algae Removal..................................................................................................... 70
C. Color R em oval..................................................................................................... 71
D. Clarification of Effluents from Biological TreatmentUnits of
Sewage W ater..................................................................................................... 72
E. Upgraded Removal of Phosphorus from SewageW a te r................................. 72
F. Arsenic Rem oval..................................................................................................72

VII. Information on Costs........................................................................................................73

References 75
58 Water, Wastewater, and Sludge Filtration

I. DIRECT FILTRATION IN WATER TREATMENT

Throughout history, man has always sought the clearest water possible for his drinking
requirements, or that water which has already undergone natural filtration by passing through
the ground. By way of example, although the Romans could have drawn water from the
nearby Tevere River, they constructed superb waterworks to derive water from springs located
at a considerable distance from the city. These waterworks are still employed in the municipal
water supplies of Rome . 1
Many cities have experienced a great increase in population, thus forcing man to seek
new sources of water which, often enough, are not as clear as desirable. History is studded
with rudimentary examples of filtering systems, although the true era of modem filters dates
from the beginning of the 19th century. At the same time, filters were the only possible
means for treating water at extremely low filtering velocity, e.g ., 0 . 1 m 3/m 2*h, which entailed
longer operating cycles before requiring a manual cleaning of the filter. This manual operation
consisted of both the removal of the surface layer of sludge and the renewal of the filtering
bed by adding the minerals which had inevitably been lost. The flare-up of epidemics, the
search of their cause, and, above all, the acquired knowledge that water was the primary
vehicle for their spreading, contributed to the importance of filtration or the capacity exerted
by filters for retaining various kinds of microorganisms.
Up to now, this concept has not been considered carefully. As a matter of fact, the process
of disinfection still relies mainly on chemicals, whereas, in actual fact, the filter is a true
disinfecting system and above all a system which makes disinfection more effective. At the
beginning of the 2 0 th century, the importance of coagulation-flocculation was realized in
the removal of colloidal and dispersed particles. The extraordinary properties of chlorine,
as well as its derivatives, as a disinfecting agent were thus acknowledged. This introduced
a substantial change into water treatment technology. The filter no longer plays a unique
role but becomes a complementary part to that treatment hereinafter referred to as conven­
tional treatment of coagulation-flocculation-sedimentation-filtration.
Ever since then, conventional treatments have undergone substantial improvements and
filters, as well, have been modified so as to achieve a better operation from a mechanical
and hydraulic standpoint. However, direct filtration or water treatment carried out by filters
alone was brought back at a later stage, that is at the end of World War II, and it was in
the 1960s that numerous studies were carried out on the filtration mechanism.
Filtration is undoubtedly the most complex system of water treatment. Even the various
attempts at mathematical modeling were not successful in pin-pointing performance exactly
without resorting to experimental tests. Further studies2 are still required. Moreover, any
mathematical model that was found to be successful for specific conditions would in practice
be scarcely useful owing to the wide variability of natural waters. As a matter of fact, when
dealing with filtration, the following five factors must be considered:

1. Filter design (physicochemical, physical, and geometric characteristics of filtering

material)
2. Water and related variables, i.e., temperature, saline content and therefore viscosity,
density, and ionic strength
3. Suspended solids (size, shape, density, electric charge density) and those solids that
can be rendered insoluble during treatment
4. Nature, quantity, and instructions on how to employ coagulant and flocculant chemicals
5. Operating conditions (filtering velocities and variations, pressure, backwashing, etc.)

There is no single way of defining direct filtration. It is generally meant that in this process
there is no other treatment aiming at the removal of suspended solids such as sedimentation,
 59

FIGURE 1. Direct filtration with preflocculation and contact flocculation.

flocculation, hydrocyclones, etc. prior to filtration. In practice, the concept of direct filtration
has been developed along two different guidelines as illustrated in Figure l . 3 They are (1)
coagulation-flocculation as a preliminary step well-separated from filters; and (2 ) coagulation-
flocculation as an integral part of those mechanisms governing filtration.
These two concepts have led to different evaluations of similar conditions. As a matter
of fact, the target of coagulation-flocculation as a separate stage from filters is to obtain a
floe “ designed” to penetrate through the bed in depth and to resist the shear forces. The
different sizes of filter medium are necessary for “ sieving” floes of different size, whereas
the influence exerted by the intermixing of layers may be either positive or negative according
to whether or not it hinders the gradual action of mechanical filtration as highlighted by the
increase in pressure drop within the filtering bed. In the case of contact flocculation, achieved
by the injection of chemicals in line and with contact time and velocity gradient only just
sufficient to ensure the even distribution of chemicals in the water, the importance of filtering
layers is closely related to the forces governing the attachment and detachment mechanisms.
Therefore, it is important from a theoretical point of view to avoid suspended solids with
coarse sizes or scattered size distribution. These will enhance the detachment phenomenon
provoked by the avalanche effect, especially when the filtering cycle is in an advanced stage.
From this standpoint, the definition of direct filtration as previously established cannot satisfy
the principles that render it possible and convenient. In other words, a preliminary separation
of the coarser particles from water may be more appropriate without altering the principal
of direct filtration. If, on the contrary, the preliminary step of coagulation-flocculation led
to the formation of large and/or too weak floes, there would be an excess pressure drop and/
or a premature breakthrough of turbidity.
Still, direct filtration with a separate coagulation-flocculation step is based on interception-
occlusion principles which exploit the gravitational and inertial transport forces, whereas in
contact flocculation, Brownian motion plays an important role in addition to the above
forces. In contact flocculation, the filter is considered as a true reactor, and the chemical
60 Water, Wastewater, and Sludge Filtration

and physicochemical reactions which give rise to insoluble compounds occur within the
granular bed. When speaking of direct filtration, from now on it will mean filtration based
on the principles of contact flocculation, since from our experience this has proved to be
the most effective.

II. APPLICATION OF DIRECT FILTRATION

As previously explained, the concept of filtration is strictly connected to an idea of

clearness, which in turn is connected with a subjective sensory perception. The instruments
commonly employed to measure the turbidity of water are ineffective in providing objective
values of immediate application. This is the reason for the necessity to know the concentration
of suspended solids and the distribution of the particle sizes. For instance, the comparison
between the turbidity value in formazine turbidity units (FTU) and the concentration of
suspended matter can provide useful information which is known as “ fineness coefficient” ,
i.e., the colloidal degree of suspension. Rather than turbidity itself, this factor makes it
possible to anticipate the coagulant quantity to be employed for the treatment, thus rendering
direct filtration “ feasible ” .4 In fact, often it is the quantity of coagulant to be employed
that affects the treatment in terms of pressure drop and effluent turbidity, rather than the
removal of suspended solids itself.
In the literature one often reads that the optimal operation of filters is achieved when the
maximum permissible pressure drop matches the maximum permissible effluent turbidity.
If this statement is accepted, it follows that direct filtration is convenient when the above
target can be achieved only by means of filters capable of producing enough water in a
reasonably long stretch of time. From time to time, this concept, related to the periodical
operation of filters (service-regeneration or backwashing) has been differently defined in
terms of volumes/quantities, i.e.,

• Volume of product water per square or cubic meter of filtering media

• Total weight of retained particles per square or cubic meter of filtering media
• Percentage of water necessary to carry out regeneration as to product water
• Energy cost per cubic meter of treated water

These are a few examples of how the choice of treatment can be determined case by case,
according to a criteria of quality/quantity or economy. In actual fact, the economy parameter
comprises them all: the only variation is given by the “ weight” that each factor might have
according to local conditions. Many attempts at determining the limit of applying direct
filtration5 to the removal of turbidity (5 to 10 FTU up to 100 FTU and beyond) may be
found in the literature. However, these limits must be related to specific experimental
conditions by taking into account factors such as filter design, turbidity characteristics, etc.
In our experience, assuming that filters must operate continuously with a quantity of
backwash water not exceeding 1 0 % in relation to product water, the first limit to be taken
into account is the quantity of coagulant necessary to attain the required limpidity. For
instance, the alum quantity should not exceed 15 mg/€. In many cases, 5 mg/€ of alum
have provided excellent results, even with water having 1 0 0 mg/€ of suspended solids,
whereas in some other cases a turbidity content of 20 FTU in water has called for more
than 100 mg/€ of alum to attain an acceptable level of quality. In this regard, jar tests
provide useful indications. However, it must be considered that direct filtration with contact
flocculation, when compared to the jar test procedure, permits a decrease in the amount of
coagulant which ranges from a minimum of 50% to 90% and beyond. The reasonable average
concentration of suspended solids formed by silt and clay in surface water that can be treated
by direct filtration is up to 40 to 50 mg/ € . 6
 61

bQCl^ CHC13+OTHERS
/O H OH

/ COOH
HUMIC O X ID A T IO N ^ --------------­ ^ H O C ^ CHCl3.OTHERS
\ OH OH

CH3

CHC13+OTHERS
OH OH

FIGURE 2. Typical formation of chloroforms out of decaying products of humic acids.

Direct filtration can also be employed for the removal of dissolved substances. It is quite
common to employ filters to remove iron and manganese after oxidation. Filters also find
a widespread application in the removal of substances rendered insoluble by appropriate
chemical reactions (phosphates, tannin, humic and fulvic acids, etc.).
As far as iron and manganese are concerned, the concentration of just over 1 mg/€ gives
rise to drawbacks, mainly when the raw water is originally limpid, such as in the case of
well water. It may sound like a paradox, but naturally turbid water is less of a problem
since it can develop a particularly strong mechanism of attachment. In many cases, the level
of turbidity is of no significance. For instance, algae can easily clog the filters even when
water has little turbidity content. In the same way, the possibility of having direct filtration
to remove other pollutants, such as phosphorus, arsenic, color, etc. depends mainly on the
initial concentration, since coagulants are dosed as a function of it and give rise to high
pressure drops.

III. CHEMICAL CONDITIONING

A. Disinfecting Agents
To prevent both microbic growths in the filters and the formation of mud-balls, it is
necessary to keep the filtering bed disinfected. For this purpose, common disinfecting agents,
e.g., chlorine compounds, provide excellent results.
However, nowadays, chlorine derivatives are a source of great concern on account of
their capacity to give rise to organic halogenated compounds, chlorine-bromine in particular,
as methane derivatives (trihalomethanes, THM; Figure 2). This occurs when water contains
organic precursors, such as humic acids, products of algal metabolism, etc. Hence, the
importance of tackling this problem with great attention, for instance, by determining in
advance the THM formation potential (THMFP). This is the reason alternative disinfecting
chemicals, like chloramines, chlorine dioxide, ozone, have been taken into consideration.
However, none of them is faultless: chloramine has a very poor capacity to destroy bacteria,
viruses, spores, etc.; chlorine dioxide decays to chlorite which is harmful to human beings
(methemoglobinemia); besides being toxic, ozone may give rise either to mutagenic sub­
stances or to unknown substances whose degree of harmfulness cannot be determined easily.
Our experience leads us to favor those treatments that tend to remove, or at least to reduce,
the content of organic precursors, thus avoiding at the same time the need for a prechlorination
62 Water, Wastewater, and Sludge Filtration

or, generally speaking, a predisinfection. It is preferable to solve the disinfecting problems

related to filters by carrying out shock disinfections during backwash.

B. pH Adjustment
Almost all naturally soft water has a pH value ranging from 6 to 8.5. The pH values,
less than 7, are typical of surface water with little mineral content and telluric water with
a high concentration of carbon dioxide.
The importance of pH in water treatment is accounted for by at least three factors:

1. Together with other parameters, it determines the degree of aggressiveness of water

and, consequently, its corrosive action when in contact with filter components and the
construction material of distributing systems. On the contrary, high pH values lead to
scaling. It may seem incredible, but even today, important municipal water supplies
disregard this parameter, thus wasting a great amount of money in order to carry out
repairs and energy to pump water through water mains partially occluded with rough
walls owing to limestone deposits, not to mention the damaging effects to health due
to inorganic micropollutants dissolved in aggressive water, e.g., lead, zinc, cadmium,
copper, etc.
2. pH values affect the removal of humic acids and iron; optimal pH can be of vital
importance to the desired results.
3. Finally, pH determines the choice of coagulants. In fact, regardless of the degree of
efficiency of metal salts commonly employed, aluminum and iron sulfates or chlorides
in particular, the residual contents of cations after treatment cannot be ignored since
their concentration strictly depends on the operating pH. For instance, when employing
aluminum salts, pH should not be lower than 6 or higher than 7.5, thus avoiding an
aluminum concentration over the limit 0 . 1 to 0 . 2 mg/f.

For these reasons, pH adjustments are necessary both before and after filtration.

C. Metal Coagulants
Aluminum, iron sulfates, and chlorides are commonly employed. In theory, chlorides are
to be preferred owing to the high ratio between the number of anions to the cations. In
practice, the variations in the degree of efficiency of coagulants are not so high as to justify
the higher cost of one product in comparison with the other. Far more important are the
considerations relevant to pH, hydration, and polymerization degree of metal hydrolysis
products, and volume of polynuclear compounds giving rise to floes.
In fact, floes of metal hydroxide are almost incompressible and, therefore, take up a great
volume. As a result, in many cases, pressure drops are not caused by the substances removed
(hydrophobic colloids in particular), but by the metallic hydroxide precipitates which exert
an adsorption action on hydrophobic colloids. Therefore, there is a direct relationship between
volume of floes and dosage of coagulant, whereas there is an inverse relationship between
dosage of coagulant and length of filtration cycle .4
Consequently, in relation to this last consideration, it can be said that direct filtration may
not be an advantage when a dosage of metal ion above 2 to 2.5 mg/€ is required to obtain
an acceptable degree of coagulation. With regard to the above, direct filtration is more easily
applicable to hydrophobic colloids than to hydrophilic ones. The former require a small
quantity of coagulant, where its electrostatic action as counter-ion prevails, whereas in the
latter, adsorption is predominant if not a direct stoichiometric relationship of the coagulant
with hydrophilic colloids, e.g., humic and fulvic acids. This also accounts for the consid­
erable amount of coagulant employed in filtration, or, more generally, in clarification of
water with a moderate turbidity. As a matter of fact, in dilute suspension, colloids have
little possibility of colliding, and thus coagulants must act as flocculation support (sweep
 a COAGULANT b COAGULANT COAGULANT
c
DOSAGE DOSAGE DOSAGE

A B C

FIGURE 3. (A) Coagulation of concentrated hydrophobic colloids; (B) coagulation of dispersed hydrophobic
colloids with restabilization step; (C) coagulation of hydrophilic colloids.

FIGURE 4. Flocculation and bridging mechanism with polyelectrolytes.

flocculation) besides providing ions for the neutralization of colloidal electric charges
(Figure 3).

D. Filtration Aids
This is a blanket name for polyelectrolytes or those substances whose purposes are to
provide a bridging mechanism. Owing to electrostatic charges, many polymers can also act
as coagulant agents. In our experience, the use of polyelectrolytes as primary coagulants in
direct filtration leads to more drawbacks than advantages from an operational standpoint,
i.e., difficulty in determining dosages as a function of variations in water characteristics,
higher pressure drops, etc.
Commonly, the organic poly electrolytes of synthetic origin are employed, although ef­
ficient natural polyelectrolytes (alginates) or inorganic ones (activated silica) are available.
The choice of polymer flocculant is strongly affected by operating conditions and acquired
experience. The best possible dosage can be determined only by running trial tests. It is,
therefore, not necessary to go into further details. As a general rule, it can be said that the
cationic polymers will act better if the previous dosage of metallic coagulant is poor, whereas
nonionics and anionics provide a good operation only when the “ zeta potential” value is
almost neutral. It must be stressed that the proper use of polyelectrolytes is related to the
necessity of altering the composition of floes and of forming a chemical bridge between the
material deposited on the filtering bed and that still dispersed in water (Figure 4).

IV. FILTER DESIGN

Basically, two kinds of filters are employed in water treatment, i.e., single-layer sand
filter, and dual or multi-media filter.
The first kind is commonly employed in conventional treatment as a polishing unit after
the sedimentation. It is fed by gravity, and the filtering bed is made up of 1.5 to 2 m of
silica sand having a granule size ranging from 0.5 to 1 mm with a good degree of uniformity
64 Water, Wastewater, and Sludge Filtration

Table 1
TYPICAL FILTERING MATERIALS

Typical size Layer

Filter Density range depth
Layer medium (g/mf) (mm) (cm)

Coarse Plastic 0.4— 0.6 1.0— 3.0 40—60

Pumice 0.45 1.5— 2.5
Coal (anthracite, 0.7— 1.0 1.0—2.0
palm kernel)
Medium Sand 1.3— 1.5 0.4— 1.0 20— 30
Fine Garnet (or other 2.0— 2.4 0.2 —0.6 10— 20
heavy minerals)

(uniformity coefficient less than 1.4 to 1.5). Filtration velocities are fairly low, of the order
of 4 to 5 m 3/m 2-h, and even in the so-called rapid filters, velocity does not exceed 8 to 10
m 3/m 2-h. In many cases, the operation of this filter has been improved by converting it into
a dual-media filter with part of the sand being replaced by anthracite carbon.
Multimedia filters, commonly used in direct filtration, work under pressure, and the depth
of the filtering bed does not exceed 1 m. Filtering velocities are very high, up to 20 m 3/m2*h
and beyond. In the case of swimming pool water filtration, velocities can even reach 40 to
45 m 3/m 2,h.
Filtering material is selected so as to have a decreasing granule size along the direction
of water flow. Since water usually flows from the top to the bottom, filter media are chosen
according to their specific weight. Table 1 shows the standard characteristics of a multimedia
filter.
To continue with filter design, other filters which deserve mentioning are filters with
upward flow employing only one kind of filtering material of different granule size; bi-flow
filters with simultaneous downward and upward flow of water; filters with continuous wash
of filtering material.
The major differences in design are the distribution modes of the inlet water and, above
all, of the water employed for backwashing. The latter condition is essential for the good
operation of filters and considerably affects the choice of the filtering medium of the bed.
It is evident that regardless of the distribution system employed, this must ensure an even
distribution of water throughout the filtering bed. At the same time, backwashing must
guarantee the detachment and removal of the accumulated dirt by fully utilizing the shear
forces of water flow and possibly by adding chemicals enhancing the detachment or the
dissolution of strong links.
There are many contrasting opinions and solutions for this operational aspect of filters,
i.e., backwashing by water and air, backwashing by water only, control of bed expansion,
velocity and volume of water to be employed, etc. No final criteria can be established, as
it depends on the choice of filtering medium, operating modes, and experience of manu­
facturers. Another important aspect is a constant water flow as a function of operating
pressure. In filters operating at a constant rate, headloss increases gradually, more or less
automatically, as filters become clogged by dirt, whereas in filters operating on a declining
rate (constant pressure), flow rate decreases gradually as a function of the increase in pressure
drop. This latter condition is frequently experienced in filters working under pressure,
although the different characteristics of pumps may partially alter the trend of the declining
rate. This accounts for the application of declining rate to direct filtration systems. A further
reason is provided by the fact that as long as laminar flow conditions prevail in the filter,
which is normal when the filter is clean, the decrease in flow rate is directly proportional
to the pressure drop and is very often negligible. With the clogging of filter bed, a turbulent
 65

a) ____ PRESSURE DROP b) PRESSURE DROP

DECREASE IN FLOW RATE FLOW RATE

FIGURE 5. (a) Flow rate and pressure drop in clogged filters (turbulent flow); (b) flow rate and pressure
drop during filter run (laminar flow).

flow begins, and the decrease in flow rate will have a quadratic trend with the increase in
pressure drop. Therefore, by examining the pressure drop curve, it is possible to tell to what
extent the filtration cycle can be conveniently “ pushed” , as well as whether the filter has
become clogged owing to causes such as inadequate backwashing, mud-ball formation, etc.
(Figure 5).

V. LIMITS TO DIRECT FILTRATION AND ADVANCED SYSTEMS OF

FILTRATION

A. Consideration of Limits to Direct Filtration

The basic operational difference between sedimentation and filtration lies, in the first
instance, in the continuous removal of deposited sludge (this mechanism is governed by
simple rules which can be easily checked by average laboratory equipment), whereas in the
case of direct filtration, the build-up of separated particles continuously modifies the physical,
chemical, and chemicophysical characteristics of separation to such an extent as to render
impossible the application of any mathematical model. In other words, one has to go for
experimental verification of basic parameters relevant to filter characteristics, the suspended
substances to be filtered, and the interaction between the two.
Laboratory tests make it possible to establish the superiority of one filter design over
another, but it is impossible to apply the same conclusions to different conditions like different
raw waters, filter medium, flocculant, etc.
Qualitative descriptions are more appropriate than theoretical considerations in making
the results more meaningful and in indicating further steps to be taken. By way of example,
the classical continuity equation (Equation 1)

dc 1 —f da
( 1)
dl V dt

indicates that the decrease in suspension concentration (dc) with the bed depth (dl), is
inversely proportional to the filtration velocity and depends upon the subsequent rate of
build-up of deposited dirt (da). The removal efficiency is also a direct function of (1 —f)
where f is the porosity of the filter bed. However, neither “ f” nor “ v ” is constant. Further
66 Water, Wastewater, and Sludge Filtration

FIGURE 6. Optional exploitation of filter run (maximum pressure drop matches permissible effluent
quality).

differential equations are necessary to describe the variations of these two factors as a function
of time as well as the influence exerted by these variations on other parameters. The actual
practice of direct filtration is, therefore, based on experimental observations, assuming that
the optimal filtering cycle is obtained when the maximum permissible pressure drop matches
the maximum permissible effluent quality (Figure 6 ) . 7 11
Generally speaking, the best results offered by direct filtration are due to a thorough
exploitation of the filtering bed (deep bed filtration), whereas the efforts in conventional
filtration aim at achieving a layer of sludge on the surface of the bed as the most suitable
filtering medium. As is clearly shown by the continuity equation, it follows that in the case
of direct filtration, the most critical parameter is the effluent quality, whereas the pressure
drop gives rise to drawbacks. To be more accurate, even in the case of contact flocculation
both of the above conditions can be achieved by employing, for instance, poly electrolytes.
Even a condition described by Figure 7 may be possible. In other words, an apparent optimal
condition, maximum differential pressure matching maximum effluent quality, may be at­
tained even without having achieved an optimal filtration cycle. In fact, the nonlinear increase
in pressure drop indicates that the surface clogging has occurred together with premature
breakthrough related to the detachment mechanisms. This is precisely the reason why it is
difficult to employ filter aids in direct filtration. On the other hand, in the removal of natural
turbidity (without using polymers), occurrence of breakthrough due to effluent turbidity long
before the occurrence of a significant pressure drop is absolutely normal (Figure 8 ).
Hence, in this case, it becomes necessary to carry out backwashing of filters even when
the filters have not achieved their potential capacity of “ stocking” solids.
In the same way, the continuity equation shows the impossibility of obtaining an acceptable
level of effluent quality when the raw water turbidity is very high. These problems must
also be related to those arising from the operating procedures, like connection and discon­
nection of pumps, possible variations both in flow velocities and in inflow turbidity. Based
on past research, the following conclusions on the feasibility of applying direct filtration
can be drawn:
 67

<o
c/)
§<
LU
X

FIGURE 7. “ False” optimal exploitation of filter run.

(/)

§
CO

FIGURE 8. Typical direct filtration run: turbidity leakage anticipates significant pressure drop.

Water must have a moderate concentration of suspended solids

The possibility must exist of varying the dosage of conditioning chemicals, especially
polyelectrolytes, according to the variations in turbidity of raw water
Filtration velocities must be relatively low (10 to 15 m 3/m 2-h)
There should be cycle continuity by avoiding the above drawbacks as well as excess
variations in inflow turbidity

It must be understood that the negative effect of these drawbacks varies as a function of
the progression of the filtration cycle, i.e., with higher degree of specific deposit.
68 Water, Wastewater, and Sludge Filtration

RAW WATER

WATER OUTLET

VALVES OPERATING SEQUENCE

VALVE ^lo.
PHASE 1A 3A 4A 1 2 3 4 5
SERVICE O • # o o • • •
BACKWASH 1 • O 0 • • • • •
BACKWASH 2 O • • • • O O •
RINSE o • • 0 • • • O
LEGEND: #CLOSED OOPEN

FIGURE 9. OFSY system (Omnifiltration System) and operation of valves.

B. Example of Improvements in Direct Filtration: OFSY® In-Series Filtration

For the purpose of a theoretical model, besides the problems logically linked to direct
filtration, additional problems must be considered, such as those relevant to suspension (i.e.,
the nature of colloids and their physical and chemicophysical characteristics, particle size
and their size distribution, etc.) and chemical conditioning (i.e., Z-potential off-setting with
coagulants, use of filter aids and chemical bridging, etc.). By way of example, the use of
filter aids immediately after the addition of coagulants enhances the phenomena of prefloc­
culation which reduce the efficiency of chemical bridging in contact flocculation. In this
regard, in-series filtration with different dosages of chemicals has led to greater advantages
than those foreseeable in a mere double filtration. The former makes it possible to obtain a
better quality of filtrate, to treat highly turbid water, and is less sensitive to disturbing factors
and, finally, leads to a diminished necessity for check-up on any qualitative/quantitative
variation in water to be treated. In actual fact, further advantages are basically connected
to the specific and different physicochemical characteristics of outflow turbidity from the
first filter. These characteristics render turbidity particularly suitable for a dosage of polymer.
In fact, when polymers are used as flocculant aids in addition to the metal coagulants, they
exert their action in a much better way. The polymers enhance the adsorption and contribute
to an ionic change which favors chemical bridging.
As to the filter design, the possible filter medium sizes are different; for instance, the
first filtering unit can be packed with coarse size medium, whereas the second filtering unit
can be a traditional multimedia filter. 1213
The OFSY system (Figure 9) consists of a pair of absolutely identical multimedia filters . 14
In our experience, the advantages possessed by different filtering beds (in particular a
lower pressure drop in the first filtering unit when packed with a coarse medium) are counter­
balanced by disadvantages (viz. , rapid leakage of turbidity and greater pressure drop in the
second unit, different methods of backwash, greater control of chemical dosages, etc.).
 69

Table 2
TURBIDITY LIMITS (in NTU)

European Economic
World Health Organization Community U.S.
Max. Max. Max. Max. Max. Max.
Year suggested allowed Year suggested allowed Year suggested allowed

1972 5 25 1975 5 10 Before — 10

1962
1984 1 5 1980 0.4 4 1962 — 5
1977 1

Therefore, although in theory different filtering beds can be used, in practice this may be
a difficult option to use.
The different valves of an OFSY system with their functions are indicated in Figure 9.
The main operating steps are as follows:

1. During service, both filters operate at downflow mode.

2. The backwash of the first filter is independent of the second one.
3. The backwash of the second filter is done with water filtered by the first unit.
4. During rinsing, water takes the same path as the water obtained during service except
that the wash water is discharged to drain.

This sequence renders it possible to employ raw water for washing operation, thus saving
treated water. As to chemical injection, inorganic coagulant is employed immediately before
the first filter, and anionic or nonionic poly electrolytes are employed before the second
filter. It must be added that it is possible to have the system comply with widely varying
operating conditions by varying the chemical dose accordingly. Generally speaking, these
dosages must satisfy the following two purposes:

1. The purpose of the first filter is to collect as many substances to be removed as

possible, regardless of filtrate quality. This means that its pressure drop must increase
gradually and linearly, thus attaining an optimal saturation.
2. The second filter operates as a polisher and, therefore, the dosage of poly electrolytes
must primarily achieve this target of yielding required filtrate quality without causing
surface clogging.

The validity of these two purposes is demonstrated in actual practice of OFSY filtration
by the fact that the end of a cycle is always determined by the overall pressure drop which,
however, is predominant in the first filter. Anyhow, the optimum operating conditions are
always established on the thorough knowledge of the water to be treated. Precise estimation
of operating costs can be made by conducting pilot-scale studies.

VI. APPLICATION OF IN-SERIES DIRECT FILTRATION

A. Turbidity Removal
Water filtration has been greatly developed in view of obtaining potable water of acceptable
level of limpidity (Table 2). It must be noted that the turbidity of 1 nephelometric turbidity
unit (NTU) as per WHO Guidelines (1984) is explicitly referred to as the efficient degree
of disinfection and not to the agreeable aspect of water. For the same reason, the American
Water Works Association recommends a target of 0.1 NTU as a maximum value. There is
70 Water, Wastewater, and Sludge Filtration

FIGURE 10. OFSY performance with high turbid water.

a growing realization of the fact that the attainment of limpid water is the condition necessary
to achieve good disinfection. At the same time, the demand of filtration as a true means of
disinfection, such as for the removal of Giardia, bacteria, and also viruses, is on the
increase. 1516 Until recently, direct filtration was considered applicable only to water having
a turbidity content not higher than 10 NTU. Only after singling out the mechanisms governing
contact flocculation will direct filtration allow the treatment of water with a turbidity content
up to 50 to 60 NTU. However, generally speaking, a value beyond 20 NTU gives rise to
operational problems. Indeed, in-series filtration has been designed to overcome the problems
related both to water having a high turbidity content varying from 200 to 300 NTU and up
to 600 mg/€ of suspended solids, and filters operating at high velocity, i.e., 16 to 2 0
m 3/m 2*h. 14 Figure 10 shows the results of filtration cycles obtained with the OFSY system.
The diagram refers to two instances when raw water is of high turbidity content. The graphs
do not show the curves of turbidity at the outflow of the first filter. However, it must be
noted that upward curves develop with an initial value of 25 NTU up to even 80 NTU.

B. Algae Removal
Turbidity problems connected with the potabilizatioh of lake or artificial basin water are
quite rare because the maximum turbidity is of the order of 10 to 20 NTU. However, as a
consequence, there is significant algae growth in these water sources.
These microorganisms, represented by hundreds of different species, differ in growth
from place to place due to several factors, such as water temperature and relevant eutro-
phication level, inorganic or organic micronutrients, season, etc. The different kinds of algae
give rise to various drawbacks, particularly formation of bad tastes and odors in water, rapid
clogging of filtering systems, etc. The removal of algae is therefore important in water
treatment. Table 3 shows a comparison between results obtained with an OFSY system and
a conventional unit (i.e., coagulation-flocculation-sedimentation and rapid sand filtration),
treating the same water.
 71

Table 3
ALGAE COUNT (U/m€)

OFSY OFSYb
Treated Treated
Raw water water Conventional
Tests water8 (filter I) (filter II) treatment

No. 1 (after 2,000 18 3 460

1-hr run)
No. 2 (after 2,800 2.5 0.8 1.3
11-hr run)

Raw water turbidity: 12 NTU

Treated water turbidity: 0.4 NTU

Table 4
COLOR REMOVAL FROM SURFACE WATER

Raw water
Site 1 Site 2 Site 3 Site 4 OFSY effluent
Parameters Scotland Norway Norway Australia (Av)

Turbidity NTU 1.6—2.0 30— 36 0.8—0.2 18— 20 0.1—0.3

Color (true) Hazen 130 50— 80 135— 150 46—50 <2.5
CU
-£*■
00
L1/l

pH 6.5 6.8— 7.0 7.4— 7.5 According to optimum

o

adjustment: 6.0— 7.0

Conductivity p,S/cm 60 55 27 85— 90 Similar to raw water
Hardness mg/€ 16 19.7 5.0 21 Not modified
CaC03
Iron mg/€ Fe n.d. 1.0 1.1 1.5 0.01— 0.02
Suspended solids 6— 10 17— 21 5.0 20— 25 <1
mg/€
Organic substances 8— 10 24— 38 13— 15 20 Removal >60%
mg/€ KMn04 (as COD)
Aluminum mg/€ AL n.d. n.d. n.d. n.d. <0.05

Note: Operating conditions: filtration velocity, 15— 20 m3/m2*h; length of filtration cycle, 8— 12 h
max. pressure drop, 0.8— 1.0 bar; decrease in flow rate upon max. pressure drop, approx. 15%
dosage of chemicals, aluminum salts = 0.8— 2.0 mg/€ as Al3+ (according to color concentration)
and non-ion polyelectrolyte (PF 81) = 0.02 mg/€ or activated silica = 1 . 2 mg/€ as Si02.

C. Color Removal
Color in natural water is usually caused by organic acids of the humic and fulvic kind.
These acids are formed by the decay of vegetable matter, and it is a matter of discussion
whether they are to be considered hydrophilic colloids or dissolved matter.
Besides being a disagreeable factor from an organoleptic standpoint, color also strongly
hinders disinfection, as it reacts with chlorine, thus bringing forth THM formation . 17 20
In this case, the OFSY system has made it possible to solve the problem by exploiting
the reactions between these undissolved acids and aluminum salts . 2 1 24
The basic aspects of this process are the injection of the reagent (in a certain ratio to the
concentration of color) and pH control. The latter is especially important both to obtain the
ideal state of required reactions and to prevent aluminum from leaking into the treated water
beyond permissible limits (< 0 .1 mg/€). Table 4 shows the results obtained for color removal
in different countries.
72 Water, Wastewater, and Sludge Filtration

Table 5
CLARIFICATION OF BIOLOGICALLY
TREATED SEWAGE WATER

Treated
Parameters Raw water water

Suspended solids mg/€ 20— 80 <5

COD mg/€ 160— 250 30— 70
BOD5 mg/€ 7— 40 2— 12
Total organic carbon 10— 15 3— 8
Turbidity NTU 5— 50 <1

Note: Filtration velocity, 18 to 20 m3/m2-h; maximum

pressure drop, 0.6 to 1.0 bar. COD, chemical ox­
ygen demand; BOD5, biochemical oxygen demand.

D. Clarification of Effluents from Biological Treatment Units of Sewage Water

Clarification of effluents from biological units must meet the following requirements:

1. Removal of suspended solids exceeding the values set by waste water standards.
2. Removal of suspended solids to avoid any clogging in irrigational systems whenever
water is employed for this purpose.
3. Removal of suspended solids to ensure a stable and good disinfection of water whenever
employed in water meadows, gardens, or used to irrigate vegetables for raw con ­
sumption. In this case, too, the OFSY system has proved to be suitable, as is shown
in Table 5.

It must be noted that backwash water was recycled back to the biological unit. In many
instances, the capacity of the biological unit was improved by this operation, especially
when suspended solids in effluent were due to the carry-over of biological sludge (pin-point
floes) as a consequence of particular conditions such as low temperature, industrial waste,
or even sludge bulking.

E. Upgraded Removal of Phosphorus from Sewage Water

In many countries, eutrophication calls for an almost integral removal of phosphorus from
sewage water. In Switzerland, for example, the standard results obtained by conventional
units, based on the simultaneous precipitation of phosphorus within the biological unit (up
to 1 mg/€ of residual P), is clearly inadequate for restoring lakes to a safe oligotrophic
condition .26 In this case, too, the OFSY system has succeeded in further decreasing the
concentration of phosphorus to less than 0 . 1 mg/€.
Unlike simple clarification, phosphorus removal calls for injections of aluminum or iron
salts proportional to its concentration. Table 6 shows OFSY standard operations when applied
for this purpose.

F. Arsenic Removal
In many parts of the world there is water with a considerable content of arsenic, i.e.,
Central and South America, Taiwan, some areas of Europe, etc. According to the most
advanced technical knowledge, this polluting agent can be removed by ion exchange through
use of resins, reverse osmosis, adsorption by means of activated alumina, or carbon . 27 31
In other cases, conventional units based on chemical precipitation-sedimentation and
filtration have provided results not altogether stable, despite complex installations and tre ­
mendously high costs . 32' 34 Even in this case, the OFSY system has proved to be suitable in
removing arsenic while removing other pollutants (Table 7) . 35
 73

Table 6
PHOSPHATE REMOVAL

Average
Parameters Raw water Treated water removal

Total phosphorus 1— 2.5 0.1—0.2 90

(mg/f P)
BOD5(mg/f) 7— 13 2— 5 60—70
COD(mg/f) 45— 60 30— 40 30
Suspended solids 15— 20 2.5—4.0 80
(mg/€)
Enterococci 60,000—260,000 5,000—5,000 80— 95
(MPN/100 m€a)
Fecal coliforms 0.6 x 106— 3 x 106 1 x 104—0.5 x 70—99.5
(MPN/100 m€a) 106

Note: MPN, most probable number.

a Results obtained only by means of filtration; that is, without the use of chemical disin­
fectant.

Table 7
ARSENIC REMOVAL

Parameters Raw water Treated effluent

Arsenic (|Ag/f AS) 40— 120 <5

Iron (mg/€ Fe) 0.8 <0.02
Hydrogen sulfide (mg/€ H2S) 0.3—0.7 0
Ammonia (mg/f NH4+) <0.01
o

o
r

Note: Operating conditions: Filtration velocity, 20 m3/m2*h; length

of filtration cycle, less than 12 h; maximum pressure drop after
12 h service, 0.6 bar; dosages of chemicals, FeCl3 = 2 mg/f
(as Fe), polyelectrolyte = 0.01 mg/f.

VII. INFORMATION ON COSTS

Generally speaking, precise indications on costs cannot be made. However, the following
data compare various cost items of the OFSY system with conventional treatment (the item
cost of conventional system is taken as 1 0 0 ).

Investment cost 40— 60

Space taken up by installation 20— 30
(Figure 11)
Chemical treatment 10— 50

These considerable savings must be added to those connected with the simplicity of the
treatment, i.e., diminished stock of spare parts, greater flexibility of operation (OFSY can
tolerate variations in flow rates and being often arranged in parallel, it can also supply water
whenever a line undergoes maintenance), less labor for running the system, and a simpler
automation of the whole process.
74 Water, Wastewater, am/ Sludge Filtration

FIGURE 11. Comparison of a conventional water treatment system with an OFSY system of same capacity, (100
m3/h). A and B, Conventional system; C, OFSY system.
 75

REFERENCES

1. Frontinus, S. J., The water supply of the city of Rome (Herschel, C. transl.), New England Waterworks
Association, Boston, 1973.
2. Ben Aim, R., La filtration en question, Le Progres Technique, 1980, 19.
3. Janssens, J. G., Ceulemans, J., and Dirickx, J., Experiences with direct filtration: plant-scale evaluation
and pilot-scale investigations, Proc. 5th Asia Pacific Regional Water Supply Conf., Water Seoul ’85, Seoul,
1985, SS6-4, 1.
4. Bratby, J., Coagulation and Flocculation, Uplands Press, Croyden, England, 1980.
5. Culp, R. L., Direct filtration, J. AWWA, 69, 375, 1977.
6. Baumann, E. R., Granular media deep-bed filtration, in Water Treatment Plant Design, Sanks, L. R.,
Ed., Ann Arbor Science, Ann Arbor, Mich., 1978, chap. 12.
7. Adin, A. and Rebhun, M., High-rate contact flocculation-filtration with cationic poly electrolytes, J.
AWWA, 66, 109, 1974.
8. Adin, A., Baumann, E. R., and Cleasby, J. L., The application of filtration theory to pilot plant design,
J. AWWA, 71, 17, 1979.
9. Conley, W. R. and Hsiung, K. Y., Design and application of multi-media filters, J. AWWA, 61, 97,
1969.
10. Ives, K. J. and Gregory, J., Basic concepts of filtration, Proc. Soc. Water Treat. Exam., 16, 147, 1967.
11. Ives, K. J., Deep bed filtration: theory and practice, Filtr. Sept. 17, 157, 1980.
12. Baumann, E. R. and Oulman, C. S., Use of BDST Analysis Techniques for Design of Direct Filtration
Systems Using Coarse Media and Dual Media Filters In-Series, paper presented at NATO Advanced Study
Institute, Lagos-Portugal, January 4 to 15, 1982.
13. Kawamura, S., Two stage filtration, J. AWWA, 77, 42, 1985.
14. Coccagna, L., OFSY: A New Approach to In-Series Direct Filtration of Very Turbid Water, Filtech 1981
Conference, London, September 15 to 17, 1981.
15. Logsdon, G. S. and Lippy, E. C., The role of filtration in preventing waterborne disease, J. AWWA, 74,
649, 1982.
16. Logsdon, G. S., Symon, J. M., Hoye, R. L., Jr., and Arozanena, M. M., Alternative filtration methods
for removal of Giardia cysts and cyst models, J. AWWA, 73, 111, 1981.
17. Oliver, B. C. and Shindler, D. B., Trihalomethanes from the chlorination of aquatic algae, Environ. Sci.
Technol., 14, 1502, 1980.
18. Rook, J. J., Formation of haloforms during chlorination of natural waters, Water Treat. Exam., 23, 234,
1974.
19. Rook, J. J., Haloforms in drinking water, J. AWWA, 68, 168, 1976.
20. McCreary, J. J. and Sonoeyink, V. L., Reaction of free chlorine with humic substances before and after
adsorption on activated carbon, Environ. Sci. Technol., 15, 193, 1981.
21. Narkis, N. and Rebhun, M., Stoichiometric relationship between humic and fulvic acids and flocculants,
J. AWWA, 69, 325, 1977.
22. Mitchell, G. F. and Hill, D. O., Methods for Treatment of Color in Ground Water, Water Resources
Research Institute, Mississippi State University, Jackson, 1981.
23. Coccagna, L., Technique of Water Disinfection, 3rd Int. Congr. of Nature International Academy, Spoleto,
Italy, September 15 to 18, 1982.
24. Coccagna, L., Removal of Colour from Surface Water by Means of In-Series Direct Filtration at High
Velocity, 3rd Congreso Mediterraneo de Ingenieria Quimica, Barcelona, Spain, November 19 to 21,
1984.
25. Adin, A., personal communication, 1983.
26. Milani, M. B., Quatrieme phase d ’Epuration des Eaux, Proc. Conf. Association Romande pour la Protection
des Eaux et d l ’Air, Worben, Switzerland, March 11, 1983.
27. Gupta, S. K. and Chen, K. Y., Arsenic removal by adsorption, J. WPCF, 50, 493, 1978.
28. Clifford, D., Water Engineering & Management Reference Handbook, Scranton Gillette Communications,
Des Plaines, 111., 1982.
29. Huang, C. P. and Fu, P. L. K., Treatment of arsenic (v) containing water by the activated carbon process,
J. WPCF, 56, 233, 1984.
30. Calmon, C., Notes and comments, J. AWWA, 65, 568, 1973.
31. Clifford, D., private communication, May 1985.
32. Gulledge, J. W. and O ’Connor, J. T., Removal of arsenic (v) from water by adsorptiononaluminum
and ferric hydroxides, J. AWWA, 65, 548, 1973.
33. Shen, Y. S., Study of arsenic removal from drinking water, J. AWWA, 65, 543, 1973.
34. De Carlo, E. H. and Thomas, D. M., Removal of arsenic from geothermal fluids by adsorptivebubble
flotation with colloidal ferric hydroxide, Environ. Sci. Technol., 19, 538, 1985.
35. Coccagna, L. and Benak, J., Arsenic removal by direct filtration, Filtr. Sep., 23, 227, 1986.
 77

Chapter 4

APPLICATIONS OF DEEP BED FILTRATION IN WASTEWATER TREATMENT

Kenji Fujita

TABLE OF CONTENTS

I. Introduction....................................................................................................................... 78

II. Application Status in Nonindustrial Sources of Wastewater...................................... 78

A. Tertiary Treatment of Sewage........................................................................... 78
B. Water Recycling in Large Buildings................................................................. 79
C. Night Soil Treatment.......................................................................................... 79
D. Leachate from Solid Waste Dump S ites............................. 80

III. Industrial Wastewater Treatment....................................................................................81

A. Steel-Making Factory...........................................................................................81
B. Pulp and Paper Industry..................................................................................... 83
C. Beer F acto ry ........................................................................................................ 84
D. Soft Drink F actory .............................................................................................. 85
E. Plating Factory.....................................................................................................85
F. Research Laboratory............................................................................................ 87

References 100
78 Water, Wastewater, and Sludge Filtration

I. INTRODUCTION

Wastewater can be broadly classified into two groups: industrial and nonindustrial sources.
Nonindustrial sources of wastewater include sewage, night soil, leachate from a solid waste
dumping site, and recycling water in buildings. Industrial wastewaters differ in quality and
quantity from industry to industry and within industry itself, depending on the raw materials
used, required quality of product, production process, and quantity and quality of water
used in the production.
The treated effluent standard depends on local regulations and receiving water standards,
if it is discharged into receiving streams, and on the quality of water required in the production
process of the factory, if it is reused. Lately, a majority of industries take the initiative to
reuse treated wastewater, due to limited availability of water resources and severe waste
discharge regulations imposed on them.
Therefore, it is important to know both the raw wastewater quality and quantity required
to be reused in the factory when designing an industrial wastewater treatment plant.
Filtration processes have been used in industrial and municipal wastewater treatment, in
combination with other unit operations, such as activated sludge, chemical coagulation,
oxidation, reduction and ion exchange, etc.
The structure of the filter for industrial wastewater treatment is almost the same as that
used for potable water treatment. The filters used in industrial wastewater treatment are
generally made of steel instead of reinforced concrete, to facilitate the displacement of the
unit during factory expansion.
The size (D) and depth (L) of filter media depend on the raw water quality and the required
filter run. The ratio of L/D is usually in the range of 800 to 1000 in the case of potable
water treatment. 1 However, in wastewater treatment, a lower L/D ratio, between 500 to
700, can be used. In most cases, coarser media size and larger depths than those used in
potable water treatment are used. A dual-media filter consisting of anthracite and sand is
commonly used. Backwashing of the filter is combined with surface washing and/or air
scouring.

II. APPLIC A TIO N STA TU S IN N O N IN D U ST R IA L SO U RCES OF

WASTEWATER

A. Tertiary Treatment of Sewage

Rapid filters are used in tertiary treatment of secondary effluent from sewage plants.
Usually, such filters are used following coagulation and sedimentation or after coagulation
alone without the use of sedimentation. Figure 1* shows the layout of a tertiary treatment
plant in the Tokyo Metropolitan Water Works (Minamisenju Plant), which treats secondary
effluent from an activated sludge plant. This treated water is supplied for industrial use.
At the time of design, the influent water quality to tertiary treatment was 20 mg/€ as
COD, 30 mg/€ as color, and 30 mg/€ as turbidity .2 Since 1964, from the time the plant was
commissioned, it had been operated with prechlorination, coagulation, sedimentation, fil­
tration, and postchlorination.
Initially, the average alum dosage was about 100 mg/€. However, the influent water
quality has improved in recent times, as shown in Table 1. 3 Hence, the plant is now operated
without coagulation, except in special cases such as during malfunction of the sewage plant.
The plant is now supplying only about 40,000 m3/d of treated water, since many industries
have moved out of Tokyo. The total design capacity is 138,000 m 3/d.
The filters used in this plant are similar to the filters used for potable water treatment as
shown in Figure 2. However, corrosion problems are more severe in these filters than in

* All figures follow text, beginning on page 87.

 79

Table 1
WATER QUALITY OF MINAMISENJU PLANT3

Inflow Treated water

Max. Min. Av. Max. Min. Av.

Turbidity (mg/€) 28.2 11.1 20.8 4.6 0.2 1.1

Color (mg/€) 28 1.1 4.0 26 6 16
pH 7.6 6.5 7.0 7.5 6.5 7.0
Chlorides (mg/€) 94.1 21.7 62.0 92.4 34.4 64.3
Residual chlorine (mg/€) — — — 2.1 0.0 0.9
COD (Mn) (mg/€) 19 2.0 12 17 5.0 11
BOD5 (mg/€) 28 3.3 12 3.7 0.0 1.0
NH4+ - N (mg/€) 19 0.0 7.2 20 0.0 6.8
N 03 - N (mg/€) 8.7 1.7 4.8 10 2.0 5.2
Bacteria coli (No./m€) 26,000 500 5,600 0 0 0

filters used for potable water treatment, because the effluent from the sewage plant often
has a higher salt content. For instance, surface wash equipment corrodes rapidly at steel
parts which are in contact with brass.
The design data of the plant are shown in Table 2.
Recently, several filters have been taken out of operation and the averagefiltration rate
has been reduced to 75 m/d as water demand has dropped. The filter run timeis 4 days.
Raw water and treated water quality are as shown in Table 1.
Upflow type filters are also sometimes used in a tertiary treatment plant. A schematic
diagram of such an upflow filter is shown in Figure 3.

B. Water Recycling in Large Buildings

A significant number of large buildings in Japan have their own treatment facilities to
reuse wastewater as flush water of toilets, air-conditioning water, and floor-cleaning water.
Treatment facilities are usually designed to reduce 5-day biochemical oxygen demand (BOD5)
and suspended solids (SS) up to 10 mg/€ and 5 to 10 mg/€, respectively. Figure 4 shows
several possible treatment methods applicable for this purpose. Dissolved-air flotation is
used to remove grease obtained from kitchen waste.
Figure 5 shows a treatment process consisting of activated sludge followed by sand
filtration. The design data are as summarized in Table 3. The influent and effluent water
qualities of this treatment process are as follows:

Influent water quality

BOD5 = 200 mg/6
SS = 250 mg/6
NH4+ - N = 65 mg/6
Effluent water quality
BOD5 = 10 mg/6
SS = 5 mg/6
NH4+ - N = 5 mg/6

C. Night Soil Treatment

In some districts in Japan and Korea, night soil is collected by vehicles from individual
houses and treated by either aerobic or anaerobic digestion processes. In a few cases, the
night soil is treated by the Zimmermann or the Cover-Greenfield process.
Recently, many night soil treatment plants have incorporated tertiary treatment processes,
including biological denitrification, chemical coagulation, sedimentation, filtration, ozon ­
ation, and carbon adsorption.
80 Water, Wastewater, and Sludge Filtration

Table 2
DESIGN DATA OF UNITS IN
MINAMISENJU TERTIARY
WASTEWATER TREATMENT
PLANT

Solid contact clarifier

Type Accelator®
Capacity 34,500 m3/d x 4 units
Diameter 28,200 mm
Rise rate 55 mm/min
Detention time 112 min
Sand filter
Capacity 11,500 m3/d x 12 units
Filtration rate 150 m/d
Filter run 1d

Table 3
DESIGN CRITERIA OF DIFFERENT UNITS
USED

Unit Design criteria

Aeration tank Detention time, 11 h

BOD loading, 0.45 kg/m3/d
Primary settling tank Detention time, 5 h
Surface loading, 10 mm/min
Nitrification tank Detention time, 7 h
NHf- N loading, 0.25 kg/m3/d
Coagulation tank Detention time, 0.5 h
Secondary settling tank Detention time, 4 h
Surface loading, 15 mm/min
Filter Filtration rate, 150 m/d
Media size
Anthracite, 1.2 mm
Sand, 0.6 mm
Media depth
Anthracite, 200 mm
Sand, 400 mm
Backwash rate, 0.6 m/min
Air scour rate, 0.6 m/min
Chemical dosage Aluminum sulfate, 200 mg/€
Sodium hypochlorite, 20 mg/€
Caustic soda, 80 mg/€ (for coagulation)
Caustic soda, 80 mg/€ (for nitrification)

Figure 6 shows a flowchart of a night soil treatment plant. The typical quality of night
soil is shown in Table 4. The design data of each unit operation used in tertiary night soil
treatment are as presented in Table 5. The filter used in the above process is based on the
design criteria as shown in Table 6 .

D. Leachate from Solid Waste Dump Sites

The quality of leachate from solid waste dump sites varies, depending on the kind of
solid wastes dumped, dumping methods, and time during and after dumping. Table 7 shows
the quality of leachate from an anaerobic and an aerobic reclamation site.
Activated sludge or chemical coagulation-sedimentation is applied for treating the leachate.
 81

Table 4
TYPICAL
CHARACTERISTICS OF
NIGHT SOIL

pH 7— 9
BOD 13,500 mg/€
COD 9,000 mg/f
SS 21,000 mg/€
Total solids 30,000 mg/€
Total nitrogen 5,000 mg/€ as N
Chlorides 5,500 mg/€ as Cl
Phosphoric acids 1,000 mg/€ as P
Bacterial counts 104— 106 /m€
Bacterial coli 106— 107 /m t

Table 5
DESIGN CRITERIA AND PERFORMANCE OF
DIFFERENT UNITS USED IN TERTIARY NIGHT
SOIL TREATMENT

Unit Design criteria

Primary denitrification tank BOD loading, 1.1 kg/m3/d

F/M ratio, 0.16 kg BOD/kg MLSS/d
MLSS, 7,000 mg/€
BOD5 removal, 90%
Detention time, l i d
Nitrification tank BOD loading, 0.12 kg/m3/d
F/M ratio, 0.02 kg BOD/kg MLSS/d
Nitrogen loading, 0.45 kg/m3/d
F/M ratio, 0.07 kg BOD/kg MLSS/d
MLSS, 6,400 mg/€
Detention time, 10 d
Secondary denitrification tank MLSS, 6,400 mg/€
Detention time, 5 d
Re-aeration tank F/M ratio, 0.1 kg BOD/kg MLSS/d
MLSS, 6,400 mg/€
Detention time, 1.5 d
Sedimentation basin Surface loading, 9 m/d
Detention time, 6 h
Mixing tank Detention time, 5 min
Coagulation tank Detention time, 15 min
Disinfection tank Detention time, 15 min

Figure 7 shows a flowchart of a leachate treatment plant. The filters are used for refining
the discharged water and designed according to criteria given in Table 8 .

III. INDUSTRIAL WASTEWATER TREATMENT

A. Steel-Making Factory
The steel-making industry is one of the largest consumers of water. In a modem steel-
making factory,, water consumption is approximately 1 0 0 m3/ton of steel, and more than
90% of the above quantity is recycled after treatment.
The process of steel-making consists of conveying and storing raw materials, adjustment
of raw materials, the blast furnace, steel making, the rolling mill, and production of by-
82 Water, Wastewater, and Sludge Filtration

Table 6
DESIGN CRITERIA OF FILTER USED IN
NIGHT SOIL TREATMENT

Parameter Value

Filtration rate 200 m/d

Filter run 24 h
Filter media Anthracite E.S. = 1.2 mm, L = 200 mm
Sand E.S. = 0.6 mm, L = 400 mm
Gravel size 2— 40 mm
Backwash rate 0.6 m/min
Air scour rate 0.6 m/min

Note: E.S., effective size.

Table 7
TYPICAL QUALITY OF LEACHATE FROM
MUNICIPAL SOLID WASTE DUMP SITE

Anaerobic dumping5 Aerobic dumping6

During After During After
dumping dumping dumping dumping

BOD (mg/€) 40,000—50,000 200— 300 30,000 150

COD (mg/€) 40,000—50,000 1,000 20,000 1,000
NH4 - N (mg/€) 800— 1,000 500— 600 900 400
pH Approx. 6 7— 7.5
Transparency (cm) 0.9— 1.0 1.5 0.5 6— 7

Table 8
DESIGN CRITERIA OF FILTER USED
IN LEACHATE TREATMENT PLANT

Parameter Value

Filtration rate 140— 200 m/d

Filter run 24 h
Filter media Anthracite: E.S. = 1.2— 1.4 mm, L
= 200— 300 mm
Sand: E.S. = 0.6 mm, L = 400—600
mm
Backwash rate 0.6 m/min
Air scour rate 0.6 m/min

products. During the process, many kinds of wastewaters are discharged, including am ­
monium waste from the coke-making process, waste-containing dust collected from the coke
oven, waste gas washing from the blast furnace, waste-containing dust from the steel making
gas, waste containing iron from the continuous casting mill, blooming roll mill, and hot
strip mill, and oily, acidic, and alkaline plating wastes from the cold strip mill.
Filters are used for treating the wastewater from the continuous casting mill, hot strip
mill, blooming roll mill, and electric plating shop of the cold roll mill. Ineach case, the
treated water is reused in the factory.
The wastewater from the continuous casting mill, hot strip mill, and blooming mill are
usually treated by one of the processes shown in Figure 8 .
In the first treatment process stream shown in Figure 8 , the concentration of SS and oil
 83

Table 9
DESIGN CRITERIA OF FILTER USED IN
WASTEWATER RECYCLING PLANT IN A
CONTINUOUS CASTING MILL

Parameter Value

Diameter of filter 3,600 mm

Height of filter 8,000 mm
Filtration rate 40 m/h (960 m/d)
Filter media Anthracite: E.S. = 4.0 mm, L = 1,000 mm
Sand: E. S. = 2.1 mm
Filter run (designed) 8h
Backwash rate 1.5 m/min
Air scour rate 1.7 m/min

inflow into the filter are 10 to 50 mg/€ and 5 to 10 mg/€, respectively. SS and oil influent
to the filter in the second and third stream of Figure 8 are 30 to 50 mg/€ and 5 to 20
mg/€, respectively. Similarly, 3.5 to 5 mg/€ and around 20 mg/€ is in the fourth process
stream. The filtered water in each of the four processes is reduced to less than 5 mg/€ in
suspended solids and oil content.
Figure 9 shows a flowchart of a wastewater recycling plant in a continuous casting mill.
Filters are used after coagulation and sedimentation and have dimensions as presented in
Table 9. The performance of the filter is as follows:

Inflow concentration Suspended solids 20 to 30 mg/€

Oil 5 to 10 mg/€
Filtered water Suspended solids <5 mg/€
Oil <5 mg/€

Figure 10 shows a flowchart of the treatment of electric plating wastes from the cold roll
mill which includes alkaline, acidic, and chromium wastes. The alkaline waste contains 100
to 500 mg/€ of suspended solids, 70 to 150 mg/€ of oil, around 60 mg/€ of iron, and has
pH value of 10 to 13. The acidic waste contains 20 to 80 mg/£ of suspended solids, around
60 mg/€ of iron, and has pH value of 2 to 4. The chromium waste contains 10 to 20 mg/€
of suspended solids, 40 to 800 mg/€ of chromium, and has pH value of 2 to 7. After
treatment, the filtered water has less than 0.05 mg/€ chromate ion (C r04- 2), 1.0 mg/€ ferrous
ion, and 50 mg/€ suspended solids, and the pH value is between 5.8 to 8 . 6 .

B. Pulp and Paper Industry

This industry discharges wastewater from the pulping process and paper machines.
The pulp-making process, with wood as a raw material, uses one of three processes, i.e.,
mechanical, chemical, and chemico-mechanical process. Among them, the chemical process
used in the kraft pulp (K.P.) process with alkaline digestion is the most popular one.
The pulping wastewater contains chemicals dosed for digestion and organic materials
separated from wood, such as lignin, sugar, and hemicellulose. Due to the presence of these
contaminants, the wastewater has a high biochemical oxygen demand (BOD), chemical
oxygen demand (COD), and color. Table 10 shows the typical wastewater quantity and
quality discharged from different pulp- and paper-making processes. Pulp waste has been
treated by chemical coagulation and sedimentation in most cases. Few of them incorporate
a filtration process.
Wastewater from the paper-making plant using virgin pulp, which is relatively less con ­
taminated, is discharged after being treated with coagulation-sedimentation or dissolved-air
84 Water, Wastewater, and Sludge Filtration

Table 10
QUANTITY AND QUALITY OF PULP AND PAPER
MAKING WASTEWATER7

Quality
Quantity BOD COD SS
Process (m3/ton product) (mg/€) (mg/€) (mg/€)

DSP 400 1500 3000 188

DKP 400 370 350 192
SP 110 1640 2730 500
SCP (1) 100 2000 4000 440
SCP (2) 100 1400 1800 320
BKP 200 170 220 280
Bleached straw 403 1030 1170 435
Strawboard 333 710 950 1500
Newsprint paper 50 240 480 260
Cardboard 200 100 200 350

Note: DSP: Dissolved sulfite pulp; DKP: Dissolved kraft pulp; SP: sulfite
pulp; SCP (1): semichemical pulp for corrugating medium; SCP (2):
semichemical pulp for other than corrugated medium; BKP: Bleached
kraft pulp.

From Yukawa, N ., Wastewater treatment in pulp and paper making factories,

J. Ebara-Infilco, 66, 84, 1976. With permission.

Table 11
PERFORMANCE AND DESIGN CRITERIA OF
FILTER USED TO TREAT THE
SEDIMENTATION UNIT EFFLUENT

Wastewater White water

Pretreatment Chemical coagulation and sedimentation
Type Steel-made pressure filter
Inlet suspended solids 20—30 mg/€
Outlet suspended solids 2 mg/€
Diameter of filter 3600 mm
Height of filter 5250 mm
Filter media Anthracite; E.S. = 2.5 mm, L = 350
mm
Sand; E.S. = 1.37 mm, L = 700 mm
Filtration rate 540 m/d
Backwash rate 1.1 m/min
Air scour rate 1.1 m/min
Filter run 12 h

flotation. Lately, treated water has been reused after filtration. A filter applied for this
purpose has design conditions as shown in Table 11 and dimensions as shown in Figure 11.
Recently ultrafiltration has been applied to treat bleaching wastewater in a pulp factory.

C. Beer Factory
Wastewater from a beer factory is 30% from the brewery process and 70% from bottle-
washing process by volume.
The mixed wastewater has 500 mg/€ of 5-day biochemical oxygen demand (BOD5), around
550 mg/€ COD, 400 to 700 mg/€ SS and a pH between 3.5 to 11. The wastewater is treated
 85

Table 12
DESIGN CRITERIA OF DIFFERENT UNITS
USED TO TREAT BEER FACTORY
WASTEWATER

Unit Design criteria

Neutralizing basin Detention time, 8 min

Aeration tank Detention time, 22 h
BOD loading, 1.2 kg/m3/d
Sedimentation basin Detention time, 9 h
Surface loading, 10 m/d
Coagulated sedimentation basin Detention time, 2.5 h
Surface loading, 50 m/d
Activated carbon filter Linear velocity, 5— 20 m/h
Space velocity, 2.5 1/h
Filter Filtration rate, 200 m/d
E.S.
Anthracite, 1.2 mm
Sand, 0.6 mm
L
Anthracite, 200 mm
Sand, 500 mm
Backwash rate; 1.0 m/min
Air scour rate; 1.0 m/min
Filter run; 12 h

by activated sludge to produce an effluent of about 20 mg/€ BOD, 20 mg/€ COD, and 70
mg/€ SS.
Coagulation, sedimentation, and filtration added to the activated sludge process would
reduce COD, BOD5, and SS to about 5, 0, and 0 mg/€, respectively. Figure 12 shows a
flowchart of this process. The design criteria and the dimensions of filters used for this
purpose are summarized in Table 12. Part of the treated water is reused for bottle washing,
and the remainder is discharged into a receiving stream.

D. Soft Drink Factory

Soft drink factories discharge wastewater from bottle-washing machines and from the
production process. The wastewater from soft drink-bottling factories has approximately
1000 mg/€ of BOD5, and 70 mg/€ of SS.
An activated sludge process is applied after dosing urea and phosphoric acid to supply
the deficient nutrients to enhance biological treatment. Coagulation, sedimentation, and
filtration after the activated sludge process are used when a higher quality of treated water
is required. Figure 13 shows a flowchart of a wastewater treatment plant for a Coca-Cola
bottling factory, with design data as given in Table 13. The treated water is reused for the
bottle-washing process.

E. Plating Factory
Plating wastewater consists of cyanides, chromium, acid, alkali, and nickel. Cyanide
wastewater is usually treated by a two-step oxidation process with hypochlorite salts, while
chromium wastewater is treated by a reduction method. Acid and alkaline wastewaters are
treated by a chemical neutralization process. Wastewater containing nickel is treated either
by evaporation, ultrafiltration, or electrodialysis in order to recover nickel and to reuse the
treated water. Wastewater containing chromium and pyrophosphate copper-plating waste­
water are occasionally treated by ion exchange or ultrafiltration processes.
86 Water, Wastewater, and Sludge Filtration

Table 13
DESIGN CRITERIA OF DIFFERENT
UNITS USED IN THE TREATMENT
OF COCA-COLA WASTEWATER

Unit Design criteria

Aeration tank BOD loading, 0.85 kg/m3/d

Sedimentation basin Surface loading, 7.2 m/d
Aqua-pelletizing settler Surface loading, 288 m/d
Filter Filtration rate, 180 m/d
Backwash rate, 0.6 m/min
Air scour rate, 0.6 m/min
Filter run, 12 h
Filter media
Anthracite
E.S. = 1.2 mm
L = ­500 mm
Sand
E.S. = ­0.6 mm
L = 250 mm

Table 14
DESIGN CRITERIA OF FILTER USED
AS PRETREATMENT IN THE
TREATMENT OF PLATING
WASTEWATER

Parameter Values

Filter media Anthracite: E.S. = 1.2 mm, L = 200

mm
Sand: E.S. = 0.6 mm, L = 400 mm
Filtration rate 133 m/d for cyanamide waste
126 m/d for chromium-containing waste
139 m/d for acidic and alkaline waste
Filter run 12 h
Backwash rate 0.6 m/d
Air scour rate 0.6 m/d

Figure 14 shows the flowchart of a plating wastewater treatment plant, in which rapid
filter is used as a pretreatment unit prior to ion exchange and is designedto reducetheSS
content from 10 to 15 mg/€ to 5 to 10 mg/€. The design criteria of thefilter used aregiven
in Table 14.
Other unit operations are designed with the following detention times:

For wastewater containing cyanide

Electrolyzer, 22 h; 1st oxidation tank 1 h
Retention tank, 1 h; 2nd oxidation tank 1 h
pH Adjustment tank 30 min, coagulation tank 10 min
Sedimentation tank, 4 h
For wastewater containing chromium
Reduction tank, 30 min.; pH Adjustment tank 45 min
Coagulation tank, 10 min.; Sedimentation tank 5 h
For acidic and alkaline wastewater
Acidification tank, 5 min.; alkalification tank 30 min
Oxidation tank, 3 h; pH Adjustment tank 30 min
Coagulation tank, 10 min
 87

FIGURE 1. Layout of Minamisenju tertiary treatment plant.

F. Research Laboratory
There are many kinds of wastewater discharged from a research laboratory, including
mercury-, cyanide-, fluoride-, chromium -, organic solvent; bacteria; and radioisotope-con­
taminated wastes.
It is, therefore, very important to treat this wastewater. It is not advisable to mix these
wastes prior to pretreatment, because the mixture may form complex ions which are very
difficult to treat.
Figure 15 shows the flowchart of the wastewater treatment plant at the University of
Tokyo, in which the wastewater is classified under 8 categories, viz., mercury, cyanides,
fluorides, chromium, other heavy metals, alkaline materials, combustible solvents, and hard-
to-combust solvent wastes. Each of these is separately collected and treated. Filters are used
in the pretreatment step prior to the chelate resin adsorption process.
88 Water, Wastewater, and Sludge Filtration

FIGURE 2. Sand filters in Minamisenju plant.

AIR V E N T

TROUGHS

GRIDS

I I I 11111N 1~ ­m
SAND

| y FILTERED WATER

GRAVEL
Q Q Q Q Q

BLOW ER

FIGURE 3. Schematic diagram of an upflow filter in a tertiary treatment plant.

FIGURE 4. Treatment processes for wastewater reuse in buildings.

GRIT CHAMBER EQUALIZING TANK AERATION TANK 1ST SEDIMENTATION NITRIFICATION TANK
BASIN

FIGURE 5. Flowchart of a wastewater recycling plant for a bussiness building.

00
VO
 DILUTION WATER

RECEIVING TANK SCREEN STORAGE TANK N2 CH3 OH n 2 gas SEDIMENTATION BASIN

GAS
NIGHT SOIL
tl ns~i 1rs-i 1 □
1ST 2ND REAERA­
NITRIFICATION TION
DENITRIFICATION TANK DENITRIFICATION
TANK TANK TANK

— I GRINDER

SEPTIC TANK SLUDGE BLOWER

AI2 (804)3
NaOH SEDIMENTATION BASIN
POLYMER RAPID FILTERS DISINFECTION

NaOCI
TO FURNACE
Water, Wastewater, and Sludge Filtration

DISCHARGE
COAGULATION TANK

POLYMER

THICKENER ______ ^ CENTRIFUGE

X

X
SLUDGE CAKE

FIGURE 6. Flowchart of a night soil treatment plant.4

 1ST 2ND
EQ UALIZING DENITRIFICATION NITRIFICATION TANK DENITRIFICATION REA ERA TION SEDIM ENTATIO N
TANK TANK TANK BASIN

FIGURE 7. Flowchart of a leachate treatment plant of a solid waste reclamation site.4

 Water, Wastewater, and Sludge Filtration

FIGURE 8. Block diagram of wastewater treatment for continuous casting, hot strip, and blooming
mill.
 COOLING TOWER

HIGH RATE MAKE UP

VACUUM PUMP SUPERNATANT

PUMP

FIGURE 9. Flowchart of a wastewater recycling plant for a continuous casting mill.4

 Fc S 0 4 Co (OH)2 so
PICKLING WASTE W- POLYMER

Cr WASTE W. NoOH 0
0

f t

RESERVOIR REDUCTION NEUTRALIZATION FLOCCULATION

TANK TANK TANK

FUME WASHER WASTE W.

tu
SEDIMENTATION
TANK
FILTER

(P) f
PLATE W.W. FILTERED WATER
CLEANING W.W. RESERVOIR

REGENERATION W.W.

J-€ T h -O "
Water, Wastewater, and Sludge Filtration

RESERVOIR SLUDGE
RESERVOIR

FIGURE 10. Flowchart of a wastewater recycling plant for tin-plating wastewater in a steel-making factory.4
 95

A IR VENT

5 2 5 0 -----------------------------
WASTE
BACKWASH
DRAIN
INTERMEDIATE
700-1*

FIGURE 11. Pressure filter for a paper-making wastewater treatment plant.4

 NEUTRAL
Water, Wastewater, and Sludge Filtration

FIGURE 12. Flowchart of a brewery wastewater treatment plant.4

 RAW
WATER

BASIN ZING TANK

FIGURE 13. Flowchart of a wastewater treatment plant for a Coke-bottling factory.4

 THICKENER FILTER THICKENER FILTER THICKENER FILTER
HIGH / ___ ^ PRESS PRESS PRESS
CONC. CYANIDE 2 s©
oe
WASTE WATER

" H S l— | d

ELECTROLYZER
r
NaCIO NaCIO NaOH
ION EXCHANGE
LOW
CONC. CYANIDE
WASTE WATER

1ST OXIDATION 2ND OXI. pH ADJUSTING

TANK TANK TANK

CHROMIUM
WASTE WATER - 0 — j
Water, Wastewater, and Sludge Filtration

JlSwli
REDUCTION
TANK
pH ADJUST-
ING TANK
COAGULATION
TANK

h 2 so 4 NaOH
ACID
AND ALKALINE
WASTE WATER

Lie ACIDIFICA - ALKALIFI - OXIDATION pH ADJUST- COAGULA­ FILTERS

TION TANK CATION TANK ING TANK TION TANK
TANK

FIGURE 14. Flowchart of a wastewater treatment plant for a plating factory.

 99

FILTER OXIDATION/DECOMPOSITION CHELETE ADSORPTION

REDUCING TOWER
AGENT NoOH
WATER
A fA
MERCURY
CONTAINED
WATER TANK
'J o
— S~
H2SO4
WATER
NoOH NoClO

CYANIDE
CONTAINED
WATER TANK
- d L DECOMPOSITION
f k
TANK

FLUORIDE
CONTAINED
WATER TANK

A LK A LIN E
WASTE WATER
TANK

HEAVY METALS
CONTAINING
WATER TANK

CHROMIC
ACID WASTE
WATER TANK

H A R D -T O -
COMBUSTION
SOLVENT
TANK

COMBUSTIBLE
SOLVENT
TANK

INCINERATOR

FIGURE 15. Flowchart of a laboratory wastewater treatment plant at the University of Tokyo.4
100 Water, Wastewater, and Sludge Filtration

REFERENCES

1. Fujita, K., A study on media size and depth of rapid filter, J. Jpn. Water Works Assoc., 510, 2, 1977.
2. Takeda, T., Kohno, K., Hachiya, S., Usaka, H., and Fukaya, M., Water qualities of Kotoh industrial
water works, Water Purif. Liq. Wastes Treat., 15, 53, 1974.
3. Tokyo Metropolitan Water Works, Operation Log, Tokyo, 1985.
4. Fujita, K. and Kaneko, M., Water Treatment— Unit Operation and Industrial and Waste Water Treatment,
Gihodoh, Japan, 1982, 287.
5. Hanashima, M., Yoshida, C., and Matsufuji, K., Reclamation of solid wastes (3), J. Solid Liq. Wastes,
3, 22, 1973.
6. Hanashima, M., Yoshida, C., and Matsufuji, K., Reclamation of solid wastes (2), J. Solid Liq. Wastes,
3, 23, 1973.
7. Yukawa, N., Wastewater treatment in pulp and paper making factories, J. Ebara-Infilco, 66, 84, 1976.
 101

Chapter 5

MICROSTRAINING

E. P. Jackson

TABLE OF CONTENTS

Definition......................................................................... 102

Field of A pplication...................................................... 102

How Microstraining W orks.......................................... 102

Measurement of Water Q uality.................................... 102

Beaudrey P.C. Water Quality Measurement (B.P.C.) 104

Microstrainer F ab ric...................................................... 107

Special M eshes............................................................... 108

Results and Applications of M icrostrainers............... 108

A. Results Obtained with Microstrainer............... 108
B. Drinking Water Applications............................ 108
C. Tertiary Sewage Water Treatment................... 109
D. Industrial W ater.................................................. 109
E. Primary Sewage Treatment............................... 109

Combined Processes Using Microfiltration................. 110

Industrial Microstrainers............................................... 110

Installation....................................................................... 110

Sizing and Dimensions................................................... 111

Advantages of Microstrainers....................................... 111

Typical Ancillary E quipm ent....................................... 115

102 Water, Wastewater, and Sludge Filtration

I. DEFINITION

Microstraining, sometimes referred to also as microfiltration, microscreening, or micro-

sieving, is a water-screening process that uses very fine mesh fabrics to arrest all particles
larger than the mesh aperture and a large proportion of the particles smaller than the mesh
aperture.

II. FIELD OF APPLICATION

Microstraining is used in the following applications:

1. As a preliminary treatment for drinking water drawn from lakes, reservoirs, and rivers
2. As a complete treatment of water for industrial purposes
3. As a preliminary treatment for sewage
4. As a tertiary treatment for sewage
5. As a means of recovering valuable fine particles or plankton from effluent waters

III. HOW MICROSTRAINING WORKS

The water flows through a very fine mesh mounted onto the periphery of a rotating drum,
each section of fabric being washed once per revolution by passing under water jets (Figure
1). Examination of the many samples of water thus purified shows that fine-woven wire
cloth arrests particles of much smaller dimensions than those of the apertures of the filtering
medium, and the quantity thus arrested is increased with the increase in the period of
immersion. This is explained by the formation of a very fine filter of intercepted particles
of slight thickness on the surface of the wire cloth. This filtering mat, while allowing the
passage of water with a loss of head of only a few inches, holds back particles of a size
considerably smaller than the apertures of the fabric (Plate 1* shows plankton particles
arrested in relation to mesh size).
Too high a loss of head could destroy this filtering mat and force the arrested particles
throught the apertures of the wire cloth. In order to limit the loss of head to a permissible
value, adjustment of the speed of rotation of the drum (i.e., of the length of time during
which the straining fabric is immersed) is necessary when the quantity of water to be treated
and its quality varies. The efficiency and capacity of a straining drum depend upon several
factors, including the quality of the water to be purified, the characteristics of the fine
straining fabric, the speed of rotation of the drum, and the quantity of backwashing water.
The highest efficiency of a microstrainer, i.e ., when the quantity of arrested elements smaller
than the mesh aperture is the greatest, is obtained when operating under a constant head of
1 0 to 12 cm (4" to 5").

IV. MEASUREMENT OF WATER QUALITY

To lay down the basis of this technique for the purification of water (microstraining), it
is necessary to determine in a simple way the quality of the water. The various types of
turbidimeters, based upon interception, reflection, or refraction of a luminous beam, give
results which are often contradictory and not sufficiently precise, especially in the case of
water containing translucent matter such as plankton. Any system based on the settling of
particles is not adequate, since microstraining is especially effective in arresting particles
that do not settle.
In the 1940s when micro-mesh filtration was first studied, two pioneers, E. Beaudrey of

* Plate 1 follows page 104.

 103

France and Dr. Boucher of Great Britain, independently developed a method for measurement
of the quality of water which consisted of measuring the flow of water across a given area
of fine mesh of the type used in an industrial microstrainer. The Boucher method (filterability
index) consists of a constant flow system under a variable head. The Beaudrey method
(Beaudrey P. C.) consists of a constant head and decreasing flow system. The results given
by both systems are in near linear correlation. The Beaudrey system, which is easier to
handle, is described hereunder.
A Beaudrey Colmatometer (Figure 2) consists of a small silver-plated cup and a filter
element of one square centimeter, screwed to the bottom of the cup. When the flow of water
supplied exceeds that of the water filtered, the cup overflows. The water to be tested passes
through the filter element under a constant head of water.
When the water is perfectly clean, the rate of flow is maximum and constant. When the
water is polluted, the rate of flow continually diminishes. Similarly, in a rotary drum, the
rate of flow through each element of the fabric is maximum when it enters the water just
after it has been cleaned by the spray-water jets, and continually decreases as the fabric gets
progressively clogged during its immersion. Therefore, the quantity of water collected from
the Colmatometer over a given period of time, i.e., the rate of flow, is the same as that
passing through each 1 cm 2 of element of the same fabric fitted to the periphery of the drum,
immersed during the same length of time, and under the same head of water.
Measurements by means of the Colmatometer thus enable an assessment to be made of
the area of immersed fabric and the period of immersion required to treat a given rate of
flow of the sampled water under a known head of water, i.e., the size of the rotary drum
and its speed of rotation.
104 Water, Wastewater, and Sludge Filtration

water sample

FIGURE 2. Beaudrey Colmatometer.

V. BEAUDREY P.C. WATER QUALITY MEASUREMENT (B .P .C .)

As the rate of flow of polluted water through the filter element of the Colmatometer is
continually diminishing, the total volume filtered would reach a limit after a specific period
of time. For a given filter element and a given head, this limiting volume depends solely
on the quality of the water. The quality of the water is determined by the B .P .C ., defined
by the ratio 1 0 0 /Vm, Vm being the limiting volume expressed in cm 3 and determined upon
1 cm 2 of microstrainer fabric R1 under a head of 10 cm (see the next section for the
characteristics of R1 fabric).
When volume V has been filtered, the rate of flow decreases from Fo to F. Thus B.P.C.
can be calculated from the following formula:

(B.P.C.) x V x V l 4 = V Fo - V f

It takes only a few minutes to measure both Fo and F with the Colmatometer, and abacus
are available for direct determination of the value of the B.P.C.
The regularity of weave of the fabric, made of special stainless steel wire, allows the
production of filter elements which are identical.
PLATE 1. Comparison of size of plankton particles retained in relation to the mesh size.
 105

FIGURE 3. Laboratory-scale unit used for B.P.C. measurements.

Whenever possible, B.P.C. measurements should be carried out in the laboratory (Figure
3 shows a typical and simple set-up), but portable Colmatometers are also available for
B.P.C. field tests (Figure 4). The measurements of the B.P.C. have become a standard
method of determining the quality of the water and is used by many water authorities,
including the World Health Organization. The B.P.C. measure is a subject of AFNOR
Standard NF.T -90030 (February 1973), and French authorities use the B.P.C. criterion as
one of those to be met for drinking water (PC < 0.1).
A systematic determination of the B.P.C. allows records to be collected which indicate
the variations in the quality of the water to be treated.
106 Water, Wastewater, and Sludge Filtration

FIGURE 4. Portable Colmatometer.

A few typical B.P.C. measurements are

• Drinking water: less than 0.10

• Clear mountain lake: 0.10 to 0.70
• Lake or reservoir with visible growth: 1 to 6
• River water: 0.5 to 15
• Sewage water prior to tertiary treatment: 3 to 6
• River during flood periods: 100 or more
• Untreated incoming sewage water: 200 to 300
 FIGURE 5. Passage of water through the fabric.

VI. MICROSTRAINER FABRIC

Experiments have shown that plain-woven, very fine fabrics are fragile, and deformation
(particularly where synthetic fabrics are used) occurs when the microstrainer is subjected to
even small losses of head. It was also noted that the filtering mat did not adhere in the same
manner to those fabrics. An exhaustive research and numerous tests finally led to the choice
of a special weave known as reps or Dutch weave, that provides double straining obtained
across apertures in different planes. Water passes first through the apertures formed by two
adjacent weft wires on the horizontal plane (having the size of the diameter of the weft wire)
and then through the smaller innner apertures having a triangular shape (Figure 5). This
fabric is made up of special stainless steel alloy wire of a diameter small enough to constitute
a fine fabric, possessing high mechanical properties, and completely resistant to corrosion.
Double ply reps is unsuitable, because backwashing has proven to be impossible.
The aperture dimensions of these fabrics cannot be determined geometrically. Comparative
measurements with the Colmatometer of the filtrates obtained with plain-woven fabrics of
specific apertures and with microstrainer fabrics, have enabled the apparent size of the
apertures to be defined.
The above applies to all microstrainer applications, with the exception of primary sewage
treatment which requires plain weave to preserve backwashing efficiency. Particles arrested
in this case are sufficiently adherent to obviate the troubles otherwise met with plain weave.
Usual standard fabrics are

Grade Size of apertures (pun) Typical applications

R00
15 )
R0 22 > Tertiary sewage treatment, drinking
R1 35 ) water
R2 60 ) Industrial purposes, sand elimination
R3 90 ) Industrial plants supply
T2 50 (plain weave) \
T4 100 (plain weave) I
T5 160 (plain weave) \ Primary sewage treatment
T5 160 (plain weave) I
T6 280 (plain weave) J

Since such fine fabrics used in large surfaces suffer wear and tear due to metal fatigue,
the fabric is always bonded to a supporting grid with apertures ranging between 1/ 2 " ( 1 2
mm) to 2" (50 mm) depending on the fine mesh aperture (Figure 6 ).
108 Water, Wastewater, and Sludge Filtration

(h v I m m i

FIGURE 6. Fabric with supporting grid.

VII. SPECIAL MESHES

For some marginal applications, special meshes have been used. Some synthetic fibers
with apertures as small as 5 |xm have been used in the final treatment of paper mill liquors
and similar effluents. Each case needs a pilot plant study of over a year to make sure that
the process is stable and industrially viable. Generally, synthetic meshes require more area
per unit of flow as compared to their stainless steel counterparts.

VIII. RESULTS AND APPLICATIONS OF MICROSTRAINERS

A. Results Obtained with Microstrainer

The efficiency is measured by testing samples of water before and after microstraining
with a Colmatometer. The B.P.C. reductions depend on the mesh size; thus, 35-|xm mesh
yields 60 to 70% B.P.C. reduction, and 22-|xm mesh yields 80 to 90% B.P.C. reduction.

B. Drinking Water Applications

According to the B.P.C. of the water, there are two cases where a microstrainer can be
used to obtain good performance:

1. Clear waters with moderate plankton content (B.P.C.s of up to 0.6) and without any
silt (water from natural or artificial lakes of substantial size often come into this
category). In this case, microstraining with a mesh of 15 to 35 |xm can be used as the
sole physical treatment. Sterilization by chlorine or ozone completes the treatment.
2. For waters of all natures which contain suspended solids that do not easily settle,
microstraining can be used as a preliminary treatment (water from rivers and lakes
with plankton growth in the warm season are typical applications). The effect of
 109

pretreated effluent with microstrainer on the sand filters (downstream) is extremely

spectacular: with 35 fxm mesh, the time between two sand filter cleaning cycles is
increased 3- or 4-fold. With a 22-|xm mesh, the factor is anything between 5 and 6 .

C. Tertiary Sewage Water Treatment

Microstraining is often used as a polishing treatment for sewage water. Mesh of 22 (xm
is commonly used for this purpose. The performance depends very much on the characteristics
of the effluent produced by the secondary treatment which, in turn, depends among other
factors on the process used. Yet, when microstraining is chosen after careful assessment of
the case, both the BOD and the suspended solids are reduced by about two thirds. A standard
secondary effluent meeting the 30— 30 criterion is thus brought down to the 10— 10 standard,
which makes it safe to be returned to the natural cycle.

D. Industrial Water
Microstraining is used:

1. As a fine screening method for protecting coolers which are very susceptible to clog ­
ging, or when the process involves very fine spray nozzles. The choice of the mesh
aperture depends on the application.
2. As a method of arresting particles that are otherwise carried away by process water.
The microstrained water is then either recycled or discharged while the particles are
recovered for reuse or disposal. Here again the mesh size is specific to every case,
and the choice depends on prior exhaustive on-site testing. The wash water with a
high plankton concentration can be used for fish breeding purposes either as by-product
of a water treatment plant or as the sole purpose of a microstraining plant.
3. As a preliminary treatment for special water preparation plants (demineralization,
process waters, etc.). This case is not very different from drinking water pretreatment.

E. Primary Sewage Treatment

This aspect of the application of microstraining is one of the most delicate, yet one of
the most promising. The experimental work on this process is relatively new, since it
commenced after 1970. It replaces, in many instances, the primary settling basin and is
effective when one or more of the following constraints exists: very little space available
for treatment; high flow variations (holiday resorts or storm outspills); single-stage treatment
before returning the water to the sea.
Owing to the generally conservative approach of water authorities throughout the world,
this system has been applied only in very few cases. More work is required on the following
aspects in order to increase the applications of microstrainers:

• To optimize the mesh aperture for each type of water.

• To demonstrate its efficiency and thus gain acceptance from the ruling authorities.

The mesh size selection very much depends on the nature of the water, the effluent quality
required, and the available budget.
Typical reductions in suspended solids and BOD obtained by primary sewage micro­
straining are:
110 Water, Wastewater, and Sludge Filtration

Suspended
solids reduc­ bod5 COD B.P.C.
Mesh aperture tion reduction reduction reduction
(p,m) (%) (%) (%) (%)

50 41 38 32 61
100 16 20 13 19
160 20.5 20 20 23
280 12.5 8 13 9.2

Note: The figures quoted above are averages obtained over many months of experimentation of microstrainer on
sewage water in a large city. It is clearly apparent that for such water, 100 and 160 jxm give similar results.
Other tests conducted elsewhere sometimes show measurably better results with 100 p,m than with 160. In
each case, pilot plant studies are highly advisable.

IX. COMBINED PROCESSES USING MICROFILTRATION

Attempts have been made to have pretreatment prior to microstraining, to increase the
removal efficiency of microstrainer. Of these, the most successful pretreatment was preo-
zonization, which is supposed to act to some extent as a coagulant for the particles to be
removed. After microstraining, the water was ozonized again. To the author’s knowledge,
the process has not been used recently. While successful in some conditions, in most cases
it was costly, delicate to run, and gave mixed results.
Many attempts have been made to use flocculant as a means to capture small particles
and build them up into a sizable element that the fine mesh could arrest. Unfortunately, to
date, the floe has always broken up when it comes in contact with the mesh. Furthermore,
adding chemicals goes against the main advantage of the microstrainer, that of using no
consumables except a little electrical power. Yet the author believes that, in view of the
potential efficiency of such a process, the subject could be regularly investigated as new
flocculants are made available.
Attempts have also been made to backwash the microstrainer with compressed air to obtain
a nearly dry residue, but due to the energy required, no industrial application of the process
exists to our knowledge.

X. INDUSTRIAL MICROSTRAINERS

Microstrainers are self-contained, self-standing units that are normally installed in a con ­
crete tank. Figure 7 shows the structural layout and constructional details of a microstrainer.
One must look for sound structural and mechanical principles: one or two wheels according
to drum length, high endurance bearings, substantial construction, well proven long-lasting
end-seals, long-lasting drive spur ring and pinion. Most microstrainers have a diameter of
about 10' (3 m), smaller diameters only being used for very small or pilot plants. For high
capacity plants, a small number of very large machines up to 30' (9 m) in width were found
to be cost effective.

XI. INSTALLATION

All except the very small units that come in metal tanks are installed in concrete tanks.
As microstrainers require reasonably constant levels, and the pressure drop across the mesh
(4 to 6 " or 100 to 150 mm) must be able to vary freely, they are normally installed with an
inlet weir, an outlet weir (Figure 8 ), and an overflow weir. Layout depends on the size and
number of units required. A few typical layouts are shown in Figure 9. Longer units use a
both-end inlet layout. In all cases, the attainment of low velocities and even flows must be
the designer’s goal.
 I ll

IN S P E C T IO N GANGWAY DR IV E P IN IO N

FIGURE 7. Layout of a microstrainer.

XII. SIZING AN D DIMENSIONS

Microstrainer sizing must be done very carefully and must allow for some extra capacity
since water quality may vary beyond the expected B.P.C. values. Whenever possible, water
quality should be measured regularly with a Beaudrey Colmatometer over a long period
prior to laying out the plant.
For preliminary work, one can start from the graph (Figure 10), reading the required
microstrainer width (in meters) according to water quality, mesh aperture, and flow rate.
Using a supplier’s catalog, one can then decide upon the best combination of machines to
suit the plant requirement.
One should bear in mind that it is always cost effective to use fewer larger machines and
have a safety margin. An extra standby machine is often advisable, since it adds to the
safety margin at all times.

XIII. ADVANTAGES OF MICROSTRAINERS

• Very compact units, much smaller than any other type of water treatment units
• Very low head drop (67150 mm maximum in normal operation)
• High and constant efficiency, whatever the flow rate and water quality variations: 85%
B.P.C. reduction with RO mesh; 70% with R1
• Very low energy requirement, typically 15 kW power supply for a 100,000 m3/h plant
(25 million gallons per day)
• No chemicals of any sort required
• Very low spray water consumption (about 1%), most of which can be recycled
 MICROSCREEN
112

N OVERFLOW WEIR

EVENTUAL
INLET WEIR
OUTLET
WEIR

FIGURE 8. Hydraulic layout of a standard microstrainer.

Water, Wastewater, and Sludge Filtration
 113

inlet

V
outlet

inlet

o u t le t

B Y - P A S S VALVE
( O p tim al)
> M ICR OSCREEN
RAW WA TER MICR OS CRE E NE D

FIGURE 9. Typical plant layouts.

114 Water, Wastewater, and Sludge Filtration

FIGURE 10. Microstrainer-sizing chart.

 115

Removal of nonsettlable particles that no other system arrests efficiently

Very low maintenance cost (some plants have run 20 years without overhauls)
Completely automatic, no manpower requirements
No sludge treatment required
Four to five times less frequent washing of a sand filter needed when a microstrainer
is fitted upstream
Tertiary sewage needs only treatment by microstrainers to meet water quality criteria
in many cases

XIV. TYPICAL ANCILLARY EQUIPMENT

Head-loss monitoring equipment

Automatic speed adjustment to maintain the head loss constant around 4" (10 cm).
This is done by electrical or hydraulic speed controls and enables the microstrainers
to give peak performance
Inlet and outlet gates
Sterilization system to prevent proliferation on mesh panels (ultraviolet lamps, chlor­
inated spray, etc.)
Wash water pumps and pipework
A concentrator in systems used for primary treatment of sewage. The wash water goes
through a concentrator that concentrates the solids. The outflowing water is recycled.
 117

Chapter 6

PRECOAT FILTRATION

R. Illner

TABLE OF CONTENTS

I. Introduction..................................................................................................................... 118

II. Theory of Filtration........................................................................................................118

III. Precoat Filtration.............................................................................................................118

A. Pressure Filters................................................................................................... 118
1. Precoating..............................................................................................118
2. Body F eed ..............................................................................................120
a. Body Feed Practice...................................................................120
b. Design F acto rs..........................................................................122
3. W ashing................................................................................................. 122
B. Rotary Vacuum Precoat Filters........................................................................ 122

IV. Filter A id s ....................................................................................................................... 123

A. Diatomaceous E arth ...........................................................................................123
B. Perlite Filter A id s .............................................................................................. 124

V. A pplications....................................................................................................................124
A. Water F iltration..................................................................................................124
1. Economic Benefits of Precoat Filtration...........................................127
2. Economic Considerations.................................................................... 127
B. Swimming Pool Water Treatm ent................................................................... 127
C. Wastewater.......................................................................................................... 128
D. Water Quality and Source.................................................................................128
118 Water, Wastewater, and Sludge Filtration

I. INTRODUCTION

Two kinds of precoat aids are used commonly in precoat filtration, namely diatomaceous
earth and perlite. Other products like cellulose, wood flour, fly ash, and active carbon,
which are used occasionally, alone, or in mixture with diatomaceous earth and perlite, are
not discussed in this chapter, which provides general guidelines for the use of precoat filtration
in water and wastewater treatment.

II. THEORY OF FILTRATION

Precoat filtration is the process used in the removal of suspended solids from liquid by
filtering it through a cake made of filter aids. It depends upon the flow of liquid through
the cake and the factors influencing this rate of movement. Many theories have been de­
veloped on this subject, and most of them are based on the modified Darcy’s equation:

where Q = flow rate; A = area of the filter; L = cake thickness; AP = pressure drop
across the cake; |x = viscosity of the liquid; and K = permeability. If all parameters are
expressed in metric units, permeability will be in “ Darcy units.”
Most of the commercial literature gives the permeability value of different filter aids
proposed to the end-users.
Applying dimensional analysis (mass, time, length) to Darcy’s equation, one can see that
the Darcy unit has the dimensions of (length)2. A filter aid will have a permeability value
of 1 D when 1 m€ of liquid (with a viscosity of 1 cP) is filtered through a cake of 1-cm
thickness through a filter surface area of 1 cm2, in 1 s and using a differential pressure of
1 atm.
It is obvious that cake thickness is one of the most important parameters to be considered
in Darcy’s equation. Cake thickness is used to calculate the cake density: a product having
a cake density of 0.300 kg/dm 3 will build up a cake 1 mm thick, when using 0.300 kg of
filter aid per square meter of surface area. Cake thickness is an important property in precoat
filtration.

III. PRECOAT FILTRATION

A. Pressure Filters
Pressure filtration using filter aids is a three-step operation, namely: precoating, body
feed, and washing (Figure 1).

1. Precoating
A thin layer of filter aid is built up on the filter septum by recirculating a filter aid slurry.
This slurry is composed of water and filter aids, but may also contain a cellulose fiber to
give stability to the precoat cake (mainly when the septum is a wire mesh stainless steel
cloth). The mixture continues to circulate until the liquid from the filter septum becomes
clear. The most common filtration system using filter aids is the pressure system (Figure
2). This can be either a plate-and-frame, candle element, or a wire screen-type filter with
either a cloth, wire mesh, or ceramic filter septum. Most pressure filtration systems use a
precoat of filter aid to protect the septum and to facilitate cleaning when the filter cycle is
completed. Thickness of the precoat varies, but it is usually in the range of 2 to 3 mm
(1/16 to 1/8"), which normally requires 350 to 1050 g of dry powder per square meter of
filter area (7.5 to 25 lb/100 ft2, based upon the wet cake density of the filter aid).
 119

PRECOATING

FIGURE 1. Mock-up of precoat, body-feed, and cake washer. (Courtesy of Manville de France, Paris.)

Normally, a precoat cake of 2 mm thickness is enough to start the filtration process. An

amount of 350 to 700 g of filter aid per square meter of filter surface is used to prepare the
precoat slurry. The concentration of filter aids in the precoat tank used to make the precoat
slurry is about 6 to 15% (w/w).
The major advantages of using a precoat are the following: to protect the filter cloth, to
give an immediate clarity when filtering, and to improve cake removal and cake washing.
120 Water, Wastewater, and Sludge Filtration

Pressure
gauge

FIGURE 2. Pressure system using filter aids. (Courtesy of Manville de France, Paris.)

2. Body Feed
a. Body Feed Practice
When the precoat is made, the water to be filtered is allowed to flow throught the filter
by the use of a pump, and, thereafter, small amounts of filter aids are constantly fed into
the liquid. The body feed slurry should have a concentration averaging 10%. As filtration
progresses, the filter aid with the particles in the unfiltered water is deposited on the precoat.
Thus, a new filtering surface is continuously formed, the filter aid particles provide countless
microscopic channels which retain suspended impurities, but allow clear water to pass through
without clogging because the continuous body feed maintains an adequate porosity of the
cake.
Body feed addition is increased or decreased depending upon the turbidity of the unfiltered
water. The effects of varying the amount of body feed are illustrated in Figures 3 and 4.
 121

Maximum Differential Pressure

FIGURE 3. Cycle length (throughput at constant flow). (Courtesy of

Manville de France, Paris.)

12

10

3
xs
©
0J3
3
u©
/
-3
H
"3
©
H
t
1
o .1 .2 .3 .4 .5 .6 .7 .n .9 I
Percentage of Body Feed

FIGURE 4. Effect of body feed amount on throughput. (Courtesy of

Manville de France, Paris.)
122 Water, Wastewater, and Sludge Filtration

As filtration proceeds, the volume of the cake increases until the solids trapped in the bed
cause the flow to drop to an unacceptable level or the pressure to increase beyond the
acceptable limit. Hence, it is very important to determine the optimum quantity of body
feed to be used.
It is important to measure the differential pressure and flow rate during filtration.

b. Design Factors
The following factors should be considered during filtration:

(1) Filtration rate: this is generally expressed in the units of L/h/m2. The higher the flow
rate, the more permeable should be the filter aid in order to obtain the desired clarity.
This filtration rate is sometimes called surface loading rate. The higher the flow rate,
the smaller should be the surface area of the filter. But higher filtration rates result in
a rapid rise in pressure loss through the filter during the cycle. This results in shorter
cycles with more frequent precoating. Hence, the choice of filter aid, in order to
maintain desired clarity, is very important.
(2) Body feed rate: this is defined as the amount of filter aid that must be added to each
liter of raw water during filtration to ensure that the filter cake remains porous enough
to facilitate the water flow. Prior to filtration, it is recommended that the following
characteristics of the suspended solids be determined: percentage in weight, percentage
in volume, particle size, mean diameter, shape, and nature (rigid, deformable).
During the filtration operation, the limpidity of the water is, of course, a very
important parameter and should be known. Clarification efficiency can be controlled
by the following methods: conductometry, membrane filtration, and nephelometry,
etc.
As a rule of thumb, one can establish the following: when the particle is rigid, the
same amount of filter aid (in weight) as of suspended matter is added; whereas, when
the particle is deformable, 1 0 times the amount filter aid (by weight) is added.

3. Washing
Normally, the filtration cycle is terminated when the pressure drop exceeds its upper limit,
or when the flow rate drops to a very low value (which is not economical). When filtration
is terminated, the septum is cleaned, the cake is removed, and a fresh precoat is set up for
another cycle.
There are basically three different cake elimination processes, namely: sluicing, back-
washing, and elimination of the dry cake by vibration or centrifugation.
It is very important to check the filter septum carefully before preparing a new precoat.
A septum with incomplete washing will cause problems because filtration is handled not
only by the filter aid alone, but also by the septum. This will give rise to a very short cycle
run.

B. Rotary Vacuum Precoat Filters

A rotary vacuum precoat filter consists of a horizontal drum, 30 to 50% of which is
submerged in a filter bowl containing the water to be filtered (Figure 5). This drum is covered
with a clean septum in order to build up a precoat. A vacuum is created inside the drum by
means of internal piping. A filter of this type is commonly used in wastewater filtration.
From an economic point of view, inexpensive filter aids are used when the clarity of the
filtrate is not a critical factor. This process continues until the cake thickness reaches 60 to
100 mm. Then the drum is cleaned and precoated again.
 123

FIGURE 5. Rotary vacuum precoat filters. (Courtesy of Manville de France, Paris.)

Filtration efficiency depends upon several factors, such as: filter aid grade; drum sub­
mergence; drum speed; knife advance rate; vacuum; filter cloth (type, opening); knife sharp­
ness; and concentration of the filter aid slurry.
In any case, it is most important to select the appropriate filter aid which gives rise to a
high flow rate without allowing the solids to penetrate the precoat. A careful choice of the
above-mentioned factors will yield the most economical filter aid usage while maintaining
a high filtration rate.

IV. FILTER AIDS

As mentioned in the beginning of this chapter, only commonly used filter aids like
diatomaceous earth and perlite are considered.

A. Diatomaceous Earth
Diatomite is the skeletal remains of single-celled plants called diatoms. In life, these
microscopic algae have the unique capability of extracting silica from water to produce their
124 Water, Wastewater, and Sludge Filtration

Table 1
RELATIVE MERITS OF DIFFERENT GROUPS OF
DIATOMITE FILTER AIDS IN WATER FILTRATION

Clarification Flow Particle

Type of filter aid efficiency rate size

Natural Excellent Poor Very fine

Straight calcined Good Medium Medium size
Flux calcined Good to medium High to very high Medium to coarse

Courtesy of Manville de France, Paris.

skeletal structure or frustule. When diatoms die, they settle to form a diatomite deposit.
Although there are thousands of varieties of diatoms, not all of them are suitable for the
production of filter aid.
Manville, the leading producer of diatomite, manages five deposits around the world with
processing plants to produce the brand of filter aids called Celite. The most important
Manville diatomite deposit is in Lompoc, California. Others are in Iceland, France, Spain,
and Mexico.
Diatomite is processed by milling, calcining, and air classifying to give a finished, virtually
inert filter aid which is predominantly silica.
There are three groups of diatomite filter aids. Table 1 presents their merits in water
filtration. It is obvious that the final selection of the filter aids depends on the required
effluent quality. Table 2 presents the physical properties of the principal Celite diatomite
filter aids.

B. Perlite Filter Aids

Perlite is a naturally occurring form of siliceous volcanic glass that is left behind after a
period of volcanic activity of earth. Perlite ore can be found in deposits throughout the
world. However, few ores were used to produce filter aids which satisfy the U.S. Food
Chemical Codex . 1 Perlite ore is composed principally of aluminum silicate, with a combined
water content of approximately 3%.
To prepare filter aids, the perlite ore is expanded up to 20 times its original volume by
heating. As the perlite expands and “ pops” , an infinite variety of tiny glass-like bubbles
are produced, making the expanded perlite very light and possessing exceptional physical
properties. Grinding of the expanded perlite, followed by an appropriate air selection, will
give a complete range of perlite filter aids, as shown in Table 3.

V. APPLICATIONS

A. Water Filtration
The basic function of all water filters is to remove particulate matter from water. Precoat
filters accomplish this by physically straining solids out of the water. There is no chemical
reaction involved in this process. The grade of filter aid selected will offer the appropriate
performance with respect to clarity and flow characteristics. Particles as small as 0.2 |xm
can be removed by precoat filtration. This includes most surface water impurities. When a
soluble contaminant is present, it must be precipitated prior to filtration; and, where colloidal
matter or dispersed particles are present, precoat filtration alone may not be adequate to
reduce the turbidity to the desired level. Flocculation with flocculants, prior to the filtration,
is necessary in this case.
 Table 2
TYPICAL PHYSICAL PROPERTIES OF THE PRINCIPAL CELITE DIATOMITE FILTER AIDS

Filter Std. Hyflo

Grades Cel 505 577 SuperCel 512 SuperCel 501 503 535 545 560

Color Gray Pink Pink Pink Pink White White White White White White
Fibra-Cel/Diatomite 1 3 2 5 6 7 8 9 10 11 13
Blend Equivalent3
Approx. AP psi 1 3.2 2.0 1.2 0.7 0.4 0.1 0.07 0.06 0.03 0.02 0.005
gal/ft2/min with
0.15 lb/ft2 precoat
filter aid
b b b
Estimated gal/ft2/hr 10 20 50 65 130 240 370 1400
water 6 in. precoat
24 in Hg AP
Water, permeability 0.057 0.16 0.28 0.54 1.2 1.4 2.0 3.1 4.8
Darcies0
Density kg/m3 dry 110 130 130 130 130 145 150 150 190 190 210
Density kg/m3 wet 260 370 290 290 300 290 290 290 300 300 320
% Retained 150 2 — 2 4 7 6 8 9 10 12 40
mesh screen
% Moisture as 3.0 1.0 0.5 0.5 0.5 0.1 0.1 0.1 0.1 0.1 0.1
shipped
pH max 7.0 7.0 7.0 7.0 7.0 10.0 10.0 10.0 10.0 10.0 10.0
% Water solubles 0.15 0.15 0.10 0.10 0.10 0.15 0.15 0.15 0.15 0.15 0.15

Note: As expressed, these are typical or estimated physical properties, not specifications, and should be used accordingly.

a Fibra-Cel grades are designated by a number plus a letter (i.e., Fibra-Cel 7F). The number refers to the equivalent grade of Celite filter aid and the letter to the
percent cellulose in multiples of 2 1/2%; i.e., A = 2 1/2%, B = 5%, C = 7 1/2%, etc.
b Not applicable.
c A material having a permeability of 1-D unit passes 1 mf/sec/cm2 of a liquid of 1-CP viscosity through a cake of 1-cm thickness at a pressure differential of 1
atm.

Courtesy of Manville de France, Paris.

in
 Table 3
TYPICAL PHYSICAL PROPERTIES OF PERLITE FILTER AIDS

Grade

Property J-206 J-208 J-4 J -l J-2 J-10 J-100 J-150S J-200S J-250S

Color White White White White White White White White White White
Loose density dry (g/dm3) 65 60 80 65 75 60 60 75 85 85
Wet density (g/dm3) 240 160 200 140 175 150 150 150 150 150
% retained on 150 mesh 0.2 0.6 5.5 1 6.6 18 21 20 20 20
% H20 as shipped 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0 .3
PH 6.5— 7.5 6.5—7.5 6.5— 7.5 6.5—7.5 6.5— 7.5 6.5— 7.5 6.5— 7.5 6.5— 7.5 6.5— 7.5 6.5— 7.5

Floater % volume Tr 1 2.3 0.5 4 10 14 7.5 8 8.5

Permeability (D) 0.25 0.7 1.1 1.2 1.5 2.2 2.5 3.1 4.0 5.0

Courtesy of Manville de France, Paris,

Water, Wastewater, and Sludge Filtration
 127

1. Economic Benefits o f Precoat Filtration

Where the raw water source and other conditions are suitable, precoat filtration can offer
a number of economic benefits to the end user. They are as follows:

• Capital cost savings may be possible because of smaller land and plant building
requirements compared to sand filtration.
• Treatment cost is considerably lower than conventional coagulation/sedimentation/
granular-media filtration, when the concentration of filterable solids is low.
• The process is an entirely physical/mechanical operation and does not require operator
expertise in water chemistry relating to coagulation. The use of chemicals such as
aluminum sulfate, iron salts, and polymers, as in the case of granular-media filtration,
is not necessary.
• The volume of water used for cleaning the filter is less than that for granular-media
filtration.
• Diatomite filter media can easily be dewatered, and in some cases, the media may be
reclaimed for other uses, such as soil conditioning and land reclamation.
• Acceptable water quality is achieved as soon as the precoating is completed and
filtration commences. A filter-to-waste period is generally not necessary to bring
turbidity of the finished water within acceptable limits.

2. Economic Considerations
Once it has been found that precoat filtration and appropriate supplementary processes
can produce an acceptable finished water, several interrelated design parameters must be
considered in preparing design and specification documents. Since all these parameters
influence the total capital and operating costs of the system, it is important to evaluate them
individually and in relation to one another, so that the final design will provide water at the
most economic cost.

B. Swimming Pool Water Treatment

Proper filtration and chemical use play important roles in providing clean and clear
swimming pool water, a major concern to any pool owner. Effective filtration reduces the
amount of chemicals required for a sanitary pool. In turn, reduced use of chlorine, algicides,
and other chemicals, if necessary, eases the load on the filter.
The primary function of the filter is to remove dirt and other objects from the water.
Secondarily, effective filtration allows the chlorine to be utilized solely for disinfection
purpose, and not to form impurities that have to be removed by the filter. The most commonly
used filter is the diatomite precoat filter.
The other types of filter used for this purpose are either sand filters or sand and gravel
filters. In the case of sand filters, particles are trapped within the sand and the filter should
be cleaned when the contaminants have accumulated in the sand bed and restrict the flow
of water through the filter. Sand filters are cleaned by backwashing. During backwashing,
water is sent at a higher rate in the upward direction.
Even during heavy and continuous use, swimming pools equipped with a diatomite filter
will not undergo a change in water clarity, but pools with sand filters may. A possible
solution is to give the pool equipped with a sand filter a rest from swimmers, in order to
let the filter catch up with the accumulated contaminants. Use of alum will improve the
efficiency of a sand filter.
Heavy use of the swimming pool means the possibility of additional bacteria in the water.
This will lower the available chlorine residual to a great extent and a larger quantity of
disinfectant has to be used to combat bacteria. Chlorine residual levels should be maintained
at approximately 1 ppm, although pools equipped with diatomite filters may not require as
128 Water, Wastewater, and Sludge Filtration

much chlorine as pools equipped with other types of filters. This is due to the fact that
bacteria and other solid organic matters are removed by the diatomite filter cake.
If the swimming pool is subject to variable user loads, a 24 h/d filter use with a weekly
filter clean-up and a change of filter aid may be the most economical option. When pool
filters are seldom used, alga growth can become a problem, especially in outdoor pools. In
such cases, filtration and chlorination are supplemented by algicide addition.

C. Wastewater
Most of the time, the quality of the water to be filtered depends on its pretreatment.
Generally, a rotary vacuum precoat filter is preferred, since the build-up of a precoat is well
known in this case.

D. Water Quality and Source

Surface raw water sources should not only be evaluated on the basis of normal conditions
that may prevail for most of the year, but should also be examined for their effects due to
increased suspended material, on the filtration process during high-flow and run -off periods.
Many surface-water supplies, such as lakes or ponds, may have algae, color, or taste-
and-odor problems that would generally require additional treatment in conjunction with
precoat filtration. The use of a microstrainer prior to the precoat filter has been found to be
effective in removing microscopic material, including plankton organisms and amorphous
matter. This pretreatment also results in increased filter cycle length.
Ground water supplies may require filtration to remove suspended materials. In some
cases, mineral impurities such as soluble iron and manganese require pretreatment prior to
filtration in order to precipitate them from the raw water. The resulting precipitate can then
be removed by the precoat filter. Use of activated carbon can, in many instances, improve
the quality of the water by adsorbing undesirable contaminants.
Nowadays, water filtration is intended not only to remove algae, bacteria, soil particles,
virus, and others, but also the particulate contaminants, such as asbestos or giardia cysts.
Precoat filtration with diatomite is an effective way to obtain a high quality drinking water.

R EFEREN CE

1. U.S. Food Chemical Codex, FCC III Mononograph, 99; Food Chem. News Guide, 146.2
 129

Chapter 7

CARTRIDGE FILTRATION

Saravanamuthu Vigneswaran

TABLE OF CONTENTS

I. Introduction......................................................................................................................130

II. Theory.............................................................................................................................. 130

A. Filtration Mechanism.........................................................................................130
1. Surface Type F ilters..............................................................................130
2. Depth Type Filters.................................................................................130
B. Filter Param eters............................................................................................... 130
1. Differential Pressure vs.Flow R ate...................................................... 130
2. Efficiency of Dirt R em oval..................................................................132
3. Dirt-Holding C apacity.......................................................................... 133
C. Advantages and Disadvantages of Surface and Depth Filters.....................133

III. D esign........................ ; ................................................................................................... 134

A. Filtrate ................................................................................................................ 134
B. Filter Medium Selection....................................................................................134
C. Maximum Available Pressure D ro p ............................................................... 135
D. Scale - U p............................................................................................................. 135
E. Pretreatment........................................................................................................ 136

IV. Different Types of Filters............................................................................................... 136

A. Disposable C artridges....................................................................................... 136
B. Cleanable C artridges.........................................................................................137

V. Application ofCartridgeFilters.......................................................................................137
A. Industrial Ultrapure Water Production in the Pharmaceutical
Industry .............................................................................................................. 137
B. Decolorizing Liquids and G ases...................................................................... 137
C. Distilled Water Production...............................................................................138
D. Cartridge Filter as Pretreatm ent...................................................................... 138

References.................................................................................................................................... 138
130 Water, Wastewater, and Sludge Filtration

I. INTRODUCTION

Cartridge filters are used when particulate loads in fluids are low (0.01%, for example).
They are not suitable when the solids content is high.
Cartridge filters can clarify fluids to an optically clear stage. Another important use is in
the sterilization of liquids. The removed-particle size would range from submicron to 40
IJim. Cartridge filters have a cylindrical configuration made of a disposable or cleanable
filter medium. The supporting material, in most cases, is either plastic or metal. The assembly
should essentially be inert to the fluid in contact and should not contribute any material
from its component to the effluent liquid.
Cartridge filters are of two kinds: disposable cartridges and cleanable/reusable cartridges.
Commonly used materials for filter media construction in the case of disposable cartridges
are cotton, wool, rayon, cellulose, fiberglass, polypropylene, acrylics, nylon, asbestos,
cellulose esters, fluorinated hydrocarbon polymers, and ceramics.
For the cleanable or reusable cartridges, the filter media are typically made of such
materials as stainless steel, monel, ceramics, fluorinated hydrocarbon polymers, and other
metal alloys.

II. THEORY

A. Filtration Mechanism
Cartridge filters are classified on the basis of the mechanism of removal as “ surface”
and “ depth” types.

7. Surface Type Filters

The medium here acts as a sieve, trapping solids on its surface only. Examples of such
cartridges are membrane and wire mesh type cartridges.

2. Depth Type Filters

These tend to trap particles within the interstices of their internal structure. Examples of
such cartridges are felts, porous ceramics, asbestos pads, and the wound type cartridge.
Plastic membrane type cartridges have a wider pore distribution compared to microfilament
membrane types (Figure 1).
This distinction between plastic and fibrous type filters, based on pore size distribution,
becomes redundant when one considers the incident particle diameter. If the incident particle
is 0.5 |xm in diameter, both are surface filters; if it is 0.05 p,m diameter, both behave as
depth filters.
Nuclear type filters approach nearer to a true surface filter and have a wider pore size
range than either the plastic or microfilament filters.

B. Filter Parameters
The filtration properties of a cartridge filter depend upon its differential pressure vs. flow
rate characteristics, the efficiency of the particle removal, and the dirt-holding capacity. The
following parameters, too, are important in the selection of a filter: size, weight, cost,
compatibility with fluids used, strength, ease of servicing, cleanliness, economics of serv­
icing, lack of migration of filtering medium, workable temperature range of materials of
construction, and the fatigue life of the enclosure and the filtering medium.

7. Differential Pressure vs. Flow Rate

The flow rate through the filter is represented by the nondimensional Reynolds number
(Re), and the differential pressure by a loss coefficient K.
 131

a: 10
UJ

ft
° -<2
(\j
2 a
o
\ UJ
6
" o 10

o p-
a: cc
UJ UJ

10

=J 10

10
0 .1 0 0.12 0.14 0.16 0.18 0 .2 0 0 .2 2 0 .2 4

PORE DIA M ETER (IN JJLm )

FIGURE 1. Pore-size distribution of different membranes. (From Nickolaus,

N., in What the Filterman Needs to Know about Filtration, AlChE Symp.
Ser., 73 (171), 38, 1971 (Adapted by permission of the American Institute of
Chemical Engineers.)

Re = VD/v ( 1)

where D is an arbitary diameter, usually taken to be the diameter of the filter inlet port, V
is the velocity through the diameter D, and v is the kinematic viscosity of fluid.
This can also be written in terms of Q as,

V.D D 4 ^
Re =
v (11/4).D 2 v 11 vD
1.27
vD
( 2)

where Q = the volumetric flow rate.

Pressure loss, on the other hand, can be written as

V2
K ,- (3)

where y = the specific weight of the fluid, and g = the gravitational acceleration.
This can also be written as
132 Water, Wastewater, and Sludge Filtration

FIGURE 2. Typical pressure vs. air flow rate curve for different types of
cartridges. (From Nickolaus, N., in What the Filterman Needs to Know about
Filtration, AIChE Symp. Ser., 73 (171), 38, 1971. Adapted by permission of
the American Institute of Chemical Engineers.)

2PgA 2
K = (4)
Q7

where Q = volumetric flow rate, and A = cross-sectional area corresponding to diameter

D.
The differential pressure vs. flow rate curve, which depends on the characteristics of the
filter and the feed, can be established by a series of test runs. A typical pattern of pressure-
flow rate curve is given in Figure 2. Each point in this curve will give specific values for
Re and K, which can be calculated using Equations 2 and 4. Hence, every point in the
differential pressure vs. flow rate curve will give corresponding points to a curve that plots
log(K) vs. log(Re). The trend of K vs. Re curve in log-log scale is given in Figure 3.
The curve log(K) vs. log(Re) is specific for the filter (similar to a rating curve) and is
independent of temperature, feed viscosity, and specific weight, provided the feed liquid is
Newtonian and the internal geometry of the filter does not change throughout the working
range of flow rates and differential pressures.
This graph can thus be used to plot the differential pressure vs. flow-rates curve for any
feed (i.e., with different viscosity and specific weight), and the required differential pressure
can be obtained for any flow rate, by the use of this graph.

2. Efficiency o f Dirt Removal

The efficiency of particle removal, or degree of filtration, may be assessed by different
methods. It depends on the characteristics of the water to be clarified and the cartridge used,
in addition to the operating parameters.
The efficiency or transmission curve is a graph which indicates the percentage of particles
of each size retained or transmitted through the filter under a given set of test conditions.
 133

FIGURE 3. Typical trend of K vs. Re (diameter of inlet port = 1.27 cm; v

- 0.3 cm2/s; fluid specific gravity = 0.88 g/cm3). (From Wells, R. M., in
Process Engineering Technique Evolution — Filtration, Suttle, H. E., Ed.,
Morgan-Grampian, Kent, England, 1969, 56. With permission.)

A typical transmission curve is presented in Figure 4. The efficiency of particle removal

will vary with the operating conditions, and the transmission curve should be used as an
approximation only if the existing conditions are far removed from those prevailing during
the test.
Derivation of a transmission graph requires a test circuit into which test dusts are contin­
uously added. The samples of the test fluid are taken immediately upstream and downstream
of the filter for analysis. The test dust should contain particles of all sizes relevant to the
test filter, and it is necessary to estimate size distribution of particles in the feed and filtrate.
Frequently, the “ Coulter counter” is used for this purpose.

3. Dirt-Holding Capacity
The end of useful life of a filter element is determined by a marked reduction in efficiency
of particle removal, an increase in the differential pressure, or a reduction in flow rate.
In a depth filter, the pores of the medium become so clogged with dirt after a certain
time of operation that further contaminant will not be retained by the filter medium and will
result in a reduction of filtration efficiency. Flow surges, with resultant differential pressure
surges, will occur at this stage to cause detachment and reentrainment of previously collected
particles. The end of the service life of a surface filter also is governed by these three factors.

C. Advantages and Disadvantages of Surface and Depth Filters

• Both surface and depth filters are suitable for removal of noncompressible solids.
• Surface filters can readily be assigned an absolute rating, whereas this is not feasible
in depth filters due to its nonrigid structure.
134 Water, Wastewater, and Sludge Filtration

FIGURE 4. A typical transmission curve through a filter paper (test fluid =

mineral oil; v = 0.3 cm2/s; V = 7.62 cm/min; test dust = aloxite powder of
wide size range). (From Wells, R. M., in Process Engineering Technique Ev­
olution — Filtration, Suttle, H. E., Ed., Morgan-Grampian, Kent, England,
1969, 56. With permission)

• Depth filters lead to a long service life when handling compressible orgelatinous
contamination, whereas the surface filters become easily coated by compressible and
gelatinous solids, which would eventually stop the filtration process.
• Insurface filters, the rigid media would not unload accumulated solids until the end
of the filter run, whereas nonrigid media in depth filters unload accumulated solids
with variations in flow and pressure.

III. DESIGN

A. Filtrate
Of particular importance to the selection of the optimum cartridge filter system is the
quality of the filtrate required. Therefore, any bench-scale test must include techniques for
fully revealing these parameters.
Several established procedures, such as gravimetric count by using microscopes or au­
tomatic particle counters, can be used to quantify the amount and size distribution of con ­
taminant present in liquid samples. Specific analytical techniques are available for identifying
the particulate types which may be present in the feed and filtrate samples. Standard pro ­
cedures to evaluate the quantity and type of contaminants and their particle size distribution
are discussed in the literature.

B. Filter Medium Selection

The correct choice of filter medium will result in the production of a specified quality of
filtrate in an economic manner. Filter medium with finer size will be less economical, but
will produce filtrate of high quality. On the other hand, coarser filter medium will lead to
a longer filter run, but may not produce the required quality. However, in practice, consid ­
eration is often given to finer media which will give rise to increased reliability although
the cost may be high.
 135

AC FINE DUST ( GMS )

FIGURE 5. The characteristic pressure drop vs. capacity curve of filter cartridge. (From Howard, G. W.
and Nickolaus, N., in Solid/Liquid Separation Equipment Scale-Up, 2nded., Purchas, D. B. and Wakeman,
R. J., Eds., Uplands Press, London, England, 1987, chap. 9. With permission.)

The level of filtration depends on the specific application. For example, for silt control
of hydraulic systems only a 3-|xm absolute filter should be used, and for the removal of
yeast from wine a 0.65 -|xm filter is generally required .2
In addition to appropriate size, the filter medium must also be compatible with the process
liquid and withstand process temperature. One technique for confirming this compatibility
is to perform integrity tests, before and after the filter has been exposed to feed material.

C. Maximum Available Pressure Drop

Pressure drop is the driving force which moves the liquid through the filter assembly and
overcomes the flow resistance by the filter holder and filter medium. The resistance by the
filter medium increases with particle retention and once the ultimate storage capacity is
attained, the flow ceases.
Hence, it is necessary to provide sufficient pressure drop, not only to overcome the filter
holder resistance, but also to permit flow to continue at an acceptable rate as the medium
plugs, so as to utilize the effective dirt-holding capacity of the filter.
The characteristic pressure drop vs. capacity curve for filter cartridges is shown in Figure
5. The capacity of the filter is mostly consumed before the sharp increase in pressure drop.
Therefore, in practice, the maximum available system pressure should be equal to the knee
of the curve (Figure 5) plus the pressure drop through the filter holder.

D. Scale-Up
The selection of cartridge based on the laboratory-scale study may sometimes not be
reliable. This is due to the fact that the solid concentration and size distribution of particles
136 Water, Wastewater, and Sludge FUtration

FIGURE 6. The pressure drop vs. volume filtered for a properly matched prefilter and final filter. (From
Howard, G. W. and Nickolaus, N., in Solid/Liquid Separation Equipment Scale-Up, 2nd ed., Purchas,
D. B. and Wakeman, R. J., Eds., Uplands Press, London, England, 1987, chap. 9. With permission.)

in the suspension may vary with time in addition to the difficulty in controlling the process
variables in the laboratory-scale unit. Therefore, it is important to conduct an in-plant
sidestream test using a full-size cartridge unit prior to final filter selection.
However, in the absence of plant-scale study, one can use the following guideline in
translating the values obtained with laboratory-scale unit to plant-scale: for test samples with
moderate process conditions (temperature, pressure, viscosity), where the contaminant is
likely to be stable (unlike slimes, bacteria), where lab performance gives reproducible results,
and where the field filter area to test area ratio is about 3000 or less, then a safety factor
of 1.5 could be used. For other conditions, one should assign a higher value for safety
factor. Since cartridge filters and their holders are low in cost (compared to other filtration
techniques), oversizing does not result in an appreciable increase of cost. Oversizing may,
in fact, provide an inexpensive source of insurance against process upsets or unexpected
increase in flow.

E. Pretreatment
The applicability of a specific size cartridge filtration in achieving necessary throughputs
can be evaluated from laboratory-scale tests. In practice, if the liquid to be filtered contains
large size particles in excess quantities, then prefiltration should be incorporated for economic
reasons. The material used in prefiltration must be compatible with the feed.
The optimum prefilter will provide maximum protection to the final filter by removing a
major portion of those particles of approximately the same size as the pores of the filter.
Ideally, the service life of the prefilter should equal that of the final filter for optimum
performance (Figure 6 ).

IV. DIFFERENT TYPES OF FILTERS

Filter cartridges can be broadly classified into two types: disposable cartridges and clean-
able cartridges.

A. Disposable Cartridges
Commercially available disposable cartridges include:
 137

• Plastic membranes — Absolute removal ratings range from 0.2 to 8.0 jim. Has liquid
sterilizing applications.
• Inorganic microfilament membranes — Absolute removal ratings range from 0.2 to
25.0 |xm.
• Nuclear membranes — A recent development in membrane technology. Membranes
are available with pore sizes 0 . 2 |xm and larger.
• Thin-walled ceramics — Absolute rating range from 0.12 to 0.8 p,m.
• Felt materials — Fiber density is graded from upstream to downstream in order to
remove successively smaller particles as flow progresses.
• Wound materials — Specific removal rating can be achieved by controlling the char­
acteristics of the membrane.
• Cellulose, glass, plastics, asbestos, mixtures, and others — These are applied in coarse
filtration, with nominal ratings from 1 to 1 0 0 |xm.
• Filled adsorption materials — Mechanism of purification is adsorption rather than
filtration.

B. Cleanable Cartridges
Of the cleanable cartridges, the one commonly available is that with porous ceramics.
The cartridges are thick-walled cylinders with end-caps machined at the open ends and then
polished so as to be used with suitable gasketing . 3 A mixture of fine ceramic particles of
quartz, kaolin, and sand are sintered by heat and pressure to form an integral porous structure.
Such ceramic candles are capable of producing sterile effluents and have the economical
advantage of being cleanable and reusable. The cartridges are available with maximum pore
size diameters ranging from 0 . 2 to 1 0 0 fim.

V. APPLICATION OF CARTRIDGE FILTERS

There are numerous applications of cartridge filtration, including : 1 water filtration, aviation
fuel filtration, hydraulic oil filtration, natural gas filtration, automotive oil filtration, beer
filtration, compressed air filtration, and Decolorizing liquids and gases. However, this
chapter concentrates on the applications relevant to water and wastewater engineering.

A. Industrial Ultrapure Water Production in the Pharmaceutical Industry

The usual impurities found in water are dissolved inorganics and organics, particles, and
microorganisms. Various unit processes, such as distillation, reverse osmosis, chlorination,
flocculation and settling, sand bed filtration, activated carbon adsorption, ion exchange, and
ultrafine filtration are effective in the removal of these contaminants. 1
A typical pharmaceutical plant water system uses processes like ion exchange and distil­
lation. Although the influent water to ion exchange process may be relatively low in bacterial
contamination, the resin becomes coated with organics and encourages bacterial growth. In
this case, if cartridge filtration is used as a pretreatment, ion exchange run can be significantly
extended. Further, postfiltration with membrane cartridge filters will ensure a good quality
water with near zero bacterial and particulate counts entering the distribution system.

B. Decolorizing Liquids and Gases

Activated carbon can be used to remove the color and odor of a wide variety of liquids
and gases. Cartridge filters with activated carbon are used extensively for the removal of
excess chlorine and chloramines from water supplied to beverage-vending machines; dyes
and fatty acids from perchlorethylene used in dry cleaning machines; fumes and cigarette
smoke from aircraft cabins, etc.
138 Water, Wastewater, and Sludge Filtration

C. Distilled Water Production

Water distribution with a long piping system often necessitates the use of a storage tank.
The provision of a storage tank facility is still more important in distilled water systems
where production rate of water is low. During the storage period, bacteriological count may
increase. To overcome this problem, cartridge filters (with 0.45 |xm size) can be installed.
In the case of sterile water production, a cartridge filter of 0.2 |xm size can be used.

D. Cartridge Filter as Pretreatment

The reverse osmosis process is becoming popular in the production of ultrapure water for
certain industries. The reverse osmosis membranes can be easily fouled if no proper pre ­
treatment is given. Cartridge filters (with 3 to 30 fim size) can successfully be used for
pretreatment in this case.

REFERENCES

1. Nickolaus, N., The what, when and why of cartridges, in What the Filterman Needs to Know About
Filtration, AICHE Symp. Ser. 171, Vol. 73, Shoemaker, W., Ed., American Institute of Chemical En­
gineers, New York, 1971, 38.
2. Wells, R. M., Special units, in Process Engineering Technique Evaluation — Filtration, Suttle, H. K.,
Ed., Morgan-Grampian, Kent, England, 1969, 56.
3. Howard, G. W. and Nickolaus, N., Cartridge filters, in Solid/Liquid Separation Equipment Scale-Up,
2nd ed., Purchas, D. B. and Wakeman, R. J., Eds., Uplands Press, London, England, 1987, chap. 9.
 139

Chapter 8

R EV ER SE OSM OSIS

Saravanam uthu Vigneswaran

TABLE OF CONTENTS

I. Introduction..................................................................................................................... 140

II. Process Description........................................................................................................140

III. M em branes..................................................................................................................... 140

IV. Reverse Osmosis Devices..............................................................................................142

A. Tubular.................................................................................................................142
B. Spiral Wound.......................................................................................................144
C. Hollow F iber.......................................................................................................144

V. Fouling and Pretreatment for RO System s.................................................................144

VI. Theoretical Concepts...................................................................................................... 148

A. Membrane Structure and Rejection M echanism ........................................... 148
B. Transport Model: Flux Equation and Performance V ariables..................... 149
C. Concentration Polarization.................................................................................150
D. RO System D esign.............................................................................................152

VII. Applications....................................................................................................................152
A. Potable Water Treatm ent...................................................................................153
1. Desalination............................................................................................ 153
2. N 0 3 Removal in Potable Water Treatment........................................153
B. Production of Pure Water for Industrial Purposes......................................... 153
1. Ultrapure Water for the Electronics Industry.................................... 153
2. Sterilized Water for the Pharmaceutical Industry and for
Medical Purposes................................................................................. 155
C. Use of RO in Wastewater Treatm ent.............................................................. 155
1. Municipal Wastewater Treatm ent....................................................... 156
2. Use of RO in Industrial Wastewater Treatm ent............................... 156

VIII. A dvantages......................................................................................................................157

IX. Limitations....................................................................................................................... 157

References.................................................................................................................................... 158
140 Water, Wastewater, and Sludge Filtration

I. INTRODUCTION

Reverse osmosis is a separation technique involving the passage of water molecules through
a semipermeable membrane due to the application of pressure and involves no change either
in phase or temperature. Thus, the membrane has the quality of allowing the passage of
water while retaining up to 99% of the salts. 1

II. PROCESS DESCRIPTION

Reverse osmosis (RO) follows the well-known phenomenon of osmosis: i.e., during
osmosis, a concentrated solution of salt or sugar, separated by a semipermeable membrane
from a dilute solution, will result in transfer of water from the dilute solution to the more
concentrated until there is a balance, i.e., until the pressure difference from the volume
changes on the two sides is equal to the osmotic pressure (Figure la and lb). Osmotic
pressure is a property particular to the solution and is independent of the membrane. In
reverse osmosis, an external pressure is applied on the salt solution (a much higher pressure
than the osmotic pressure) to force the fresh water back through the membrane, i.e., reverse
its flow (Figure lc).
In its simplest form, an RO system consists of the membrane, a structure to support the
membrane, a vessel to contain the pressurized solution, and a pump to pressurize the solution
(water with dissolved solids). Pump pressure is the exclusive driving force for reverse osmosis
(Figure 2) . 1

III. MEMBRANES

The heart of the RO process is the membrane. Since Reid and Breton’s discovery in 19542
that cellulose acetate (CA) possesses good ion selectivity, CA has become the most uni­
versally studied membrane. 1 It was in the early 1960s that Loeb and Sourirajan discovered
how to make a cellulose acetate membrane with asymetric density having the specific
characteristics required for RO . 1
When cast as RO membranes, CA is an asymmetric film about 100 p,m in thickness. The
film has a dense layer of about 0.25 p,m above a thick, porous layer (Figure 3). Water
passes easily from the dense layer to the porous one, but with difficulty the other way
around. The dense layer on top of the membrane is made up of tightly packed and organized
chains of CA polymer that attract and hold water. Thus, water and solute are separated
because the water molecules can form hydrogen bonds with the acetyl groups on the polymer,
while other species may not. Thus, CA membranes have the following three essential
characteristics, which any RO membrane should have (1) high permeability to water; (2)
low permeability to salts; and (3) be easily molded into thin films.
The CA membranes, however, have two major limitations, namely: (1) they are susceptible
to degradation from biological attack, and (2 ) CA hydrolyses back to cellulose under acidic
and basic conditions. Therefore, the pH of the system must be controlled (usually 4.5 to
7.5) to minimize hydrolysis. The hydrolysis also limits the useful life of the CA membranes,
because most salt passes through as hydrolysis occurs. Sometimes diacetates and triacetates
of cellulose and mixed esters of cellulose (acetate butyrate or acetate-propionate or acetate
benzoate are also used to prepare the membranes. They, too, are very sensitive to hydrolysis.
To overcome these shortcomings of CA membranes, in 1970, du Pont commercialized RO
devices containing membranes made from an aromatic polyamide (aramid) polymer. 1 These
membranes are not susceptible to biological attack, resist hydrolysis, and can successfully
be used at a wide pH range of 4 to 11. However, because the membranes are degraded by
chlorine, water that contains Cl2 must be dechlorinated before contacting them. The aramid
 141

q) OSMOTIC FLO W b ) OSMOTIC EQUILIBRIUM c ) R E V E R S E O S M O S IS

FIGURE 1. Flow through semipermeable membrane.

PRESSURE VESSEL

CONCENTRATED

INFLUENT WAT ER PURE WATER EFFLUENT

HIGH PRESSURE
PUMP
-------------------------------- SEMIPERMEABLE
MEMBRANE

J VALVE

FIGURE 2. Pressure vessel of RO unit.

O .l - 1 .0 MICRONS m m n n m n x x x A A X A m m n x n A n n n A n x m * s k in m e m b ra n e
DO o o o o o o o o o o o oo oo
oo o o o o o o o o o o o o o o o
O O O O O O O O O O O O O O O O O
100 - 200
MICRONS O O O O O O O O O O O O O O O O O —------­ POROUS SUBSTRUCTURE
O O O O O O O O O O O O O O O O O
ooooooooooooooooo
OOOOOOOOOOOOOOOOO

FIGURE 3. RO membrane structure.

membrane has an asymmetric structure very similar to the asymmetric CA membranes.

Table 1 provides a comparison of CA membranes and hollow fiber polyamide membranes
from which it can be seen that the polyamide material offers a greater resistance to chemicals,
thus permitting a wider range of constituents to be used in any supplementary cleaning
processes, including both high and low pH cleansing solutions, to ensure optimum removal
of large organic molecules. Table 2 summarizes different types of RO membranes used in
desalination of sea water.
142 Water, Wastewater, and Sludge Filtration

Table 1
COMPARISON OF CA AND POLYAMIDE
MEMBRANES1

Characteristics Aromatic polyamide Cellulose acetate

Configuration Hollow Fibers

Physical resistance
Normal working pressure 28 bars 30 — 42 bars
Maximum back pressure 3.5 bars
of product water
Maximum service temper­ 35°C 30°C
ature
Maximum storage temper­ 40°C 30°C
ature
Chemical resistance
Permissible pH 4 — 11 4.5 — 6.5
Hydrolysis Insensitive Very sensitive
Bacterial attack Insensitive Very sensitive
Free chlorine (max. con­ pH < 8 for 0.1 ppm 0.5 — 1 ppm
tinuously admissible dose
pH > 8 for 0.25 ppm
Other oxidants Good resistance Moderate resist­
ance
Membrane life 3 — 5 years 3 years
Salt passage (on NaCl) 5 — 10% 5 — 10%

Table 2
TYPES OF RO MEMBRANES USED IN DESALINATION OF SEAWATER

Flux Rejection Pressure Parameter

Type of membrane Form (l/m2d*bar) rate (%) (bars) rate

Commercial
Cellulose acetate Plane, asymetric 448 99.2 69 6.5
Polyether/amide (PA 300) Plane, composite 733 98.7 56 13.3
Cellulose triacetale Hollow fiber 37 99.5 56 0.66
Arometric polyamide Hollow fiber 37 99.0 56 0.66
embranes under development
FT 30 Plane, composite 1200 99.6 69 17.6
PEC 1000 Plane, composite 300 99.8 56 5.4
Polyenzimidazole Hollow fiber 45 99.3 69 0.65
Cellulose triacetate Hollow fiber 50 99.7 55 0.91

From Aptel, P., Inf. Chim., 213, 189, 1981. With permission.

IV. REVERSE OSMOSIS DEVICES

The above-mentioned membranes have been incorporated in three different types of geo­
metries: tubular, spiral-wound, and hollow fiber.

A. Tubular
The first RO device, commercially marketed in the mid 1960s, was a tubular device using
a CA membrane. The membrane is either inserted into or coated onto the surface of a porous
tube designed to withstand the operating pressure. Feedwater under pressure is introduced
into the end of the tube, and the product water permeates the membrane and the tube, and
is collected on the outside (Figure 4). The concentrate exits from the far end of the tube.
 CONCENTRATE

I I ! I
i ! ! I CROSS SECTION OF THE TUBE
PERMEATE

FIGURE 4. Tubular RO device.

144 Water, Wastewater, and Sludge FUtration

The cost of this system, however, because of the small membrane area per unit volume of
container (approximately 2 0 m 2/m3), made it prohibitively expensive for treating large vol­
umes of solutions. The primary use of the tubular form is thus confined to low-volume and
high-value in-use applications.

B. Spiral Wound
Typically, spiral-would RO devices use membrane in the form of a flat film (Figure 5).
Two sheets of the flat membrane are adhered at their edges via their fabric support backing
with a tricot permeate channel cloth separating them to form a leaf . 1
The leaves are wound spirally about a plastic tube that receives the permeate from the
tricot and conducts it out of the device. The dimensions of some of the commercially available
membranes are given for information (Table 3). Development of the spiral-wound device
was a major advance in obtaining large surface area per unit volume of container.
The feed flow in a spiral device is in a straight axial path from the feed end of the cartridge
to the concentrate end and is parallel to the membrane surface. This feed flow pattern tends
to promote concentration polarization (a phenomenon where concentration of salts in a
boundary layer near the membrane exceeds that in the bulk solution). Therefore, a plastic
netting is placed in the feed channel to induce some turbulence and reduce concentration
polarization . 1 However, the conversion (fraction of feed converted to pure water) per spiral
cartridge is usually limited to 1 0 to 15%, to ensure an adequate feed flow rate and preclude
significant concentration polarization. To achieve reasonable conversions with a spiral-wound
system, a number of spiral cartridges are connected in series through their product tubes in
pressure vessels.

C. Hollow Fiber
The principal advantage of hollow fiber module is that this unit is very compact due to
its high membrane area per unit volume . 1 Another advantage is that these modules are
constructed without membrane support. For example, a hollow fiber developed by du Pont
in 1970, using aramid membrane, has an asymetric structure with an inside diameter of
about 42 p,m and an outside diameter of about 85 jxm (Figure 6 ). Up to 4.5 million of these
fibers are gathered into a bundle . 1 During forming, epoxy adhesive is applied to one end of
the bundle which, after curing, becomes a tube sheet. The other end of the fiber bundle is
sealed in epoxy to form a nub, which prevents short circuiting of the feed stream to the
brine outlet. The bundles are placed in pressure vessels up to 1.2 m long with diameters
between 10.2 cm and 25.4 cm.
Pressurized feed water enters the device through a porous distributor that extends length­
wise through the center of the unit. The feed water passes through the distributor wall and
flows radially through the fiber wall into the fiber bore. The desalted product water flows
through the bore of each fiber to the tube sheet, where it leaves the device. The salts and
other impurities remain in the concentrate, which flows to the outer perimeter of the fiber
bundle and exits through the concentrate port. 1
The characteristics of some of the commercially available hollow fiber membranes are
presented in Table 4. Table 5 compares different configuration of tube modules.

V. FOULING AND PRETREATMENT FOR RO SYSTEMS

The term “ fouling” means the retention of unwanted matter in the module or on the
surface of the membrane, but its causes and the required remedy vary according to the type
of matter retained in the module. Fouling can be attributed to many factors, which include1
precipitation of mineral salts, precipitation of metallic oxides, fouling by suspended solids,
fouling by colloids, and proliferation of bacteria.
 WATER FEED

SPIRAL WOUND
CARTRIDGE

FIGURE 5. Spiral wound RO device.

 146 Water, Wastewater, and Sludge Filtration

Table 3
CHARACTERISTICS OF SOME SPIRAL-WOUND MEMBRANES3

Dimensions
Commercial Diameter Length Producti
Manufacturer name Application (cm) (m) m3/d

Envirogenics ENRO 895-40 Brackish Water 20 1.02 20

Hydraunautics 4H25 10 1.02 3.8
UOP RoGA 30 6.1 40
Osmonics OsMO 554 10 0.99 3
UOP RoGA 1501 Seawater 15 1.02 7.6
Joray SC-5200 20 1.02 8

From Aptel, P., Inf. Chim., 213, 189, 1981. With permission.

FEED WATER

FIGURE 6. Hollow fiber RO device. (Adapted from Applegate, L. E., Membrane separation processes. Excerpted
by special permission from Chemical Engineering, 91, 64, 1984. Copyright © 1984 by McGraw-Hill, Inc., N.Y.
10020.)

Table 4
CHARACTERISTICS OF SOME HOLLOW FIBER RO MEMBRANES3

Dimensions
Commercial Diameter Length Production
Manufacturer name Application (cm) (m) m3/d

Dow Dowex RO 20K Brackish Water 27 1.24 75

Du Pont Permasep B9 0840 25 1.20 53
Toyobo Hollosep HR 8630 29.5 2.63 80
Dow Dowex Sea water 22 1.02 9.5
Du Pont Permasep B10 26.4 1.50 20
Toyobo Hollosep HR 8650 30.5 2.64 20

From Aptel, P., Inf. Chim., 213, 189, 1981. With permission.

Gradual fouling of the membrane surfaces by the numerous chemical and biological
contaminants transported in the feed water can significantly reduce energy efficiency and
cost effectiveness of the RO process. This accumulation is associated with an active decrease
in water flux and molecular rejection of the membranes.
For optimum performance of RO, the water which is to be treated by a RO module has
to be pretreated in order to eliminate the major portion of suspended particles. It is also
 147

Table 5
COMPARISON OF DIFFERENT CONFIGURATIONS OF RO
TUBE MODULES

Module type
Tubular Spiral wound Hollow fib<

Area (m2/m3)
Unit volume of membrane 300 1,000 15,000
Relative value 1 3 50
Feed flow rate at standard
conditions
L/m2/d 300—1000 300— 1000 30— 100
Relative value 1 1 0.1
Production/unit volume of mem­
branes
m3/m3 of membrane per day 100—300 300— 1000 450— 1500
Relative value 1 3 5
Pretreatment requirement Simple Average High
Extent of clogging Little Average High
Cleaning methods of membranes
Mechanical Possible Impossible Impossible
Chemical Possible Possible Possible

From Maurel, A., paper presented at Dept. Formation Continue I.C.P.I. Lyon, Lyon,
France, October 13 to 17, 1986. With permission.

ENERGY RECOVERY
(DIRECT OR INDIRECT)

LOW PRESSURE CONCENTRATION

CIRCULATION PUMP !
SAMPLE
1X3---­

o
HIGH PRESSURE
r - O HIGH PRESSURE SAMPLE SHUTDOWN SWITCH
PUMP

— O
INFLUENT PERM EATE
LOW PRESSURE CROSS - FLOW MICRO T R E VE R SE
PUMP FILTRATION OR ANY SAMPLE OSMOSIS
SUITABLE PRETREATMENT SAMPLE

FIGURE 7. RO unit with microfiltration pretreatment unit.

necessary to eliminate fouling of the membrane caused by precipitation and/or bacterial

growth.
In most cases, the following pretreatment methods are used:

• filtration
• adjustment of parameters that affect thesolubility in order to avoid precipitation (the
precipitation occurs due to an increase in concentration of sparingly soluble salts)
• coagulation of colloidal matter
• chemical treatment to avoid theformation of bacterialslime layer

Microfiltration as a prefilter is an attractive alternative to classical filters in removing

suspended solids, microorganisms, and colloidal matter. Additionally, energy recovered in
the RO process can be used in microfiltration. Figure 7 shows an illustration of the use of
microfilter as a pretreatment unit.
148 Water, Wastewater, and Sludge Filtration

Table 6
REQUIRED WATER QUALITY FOR RO
MODULE

Hollow fiber
Parameter module Spiral module

Turbidity < 0.5 < 1

Fouling index FI 15 < 3 < 4
Suspended solids concentration < 0.5 < 1

From Maurel, A., paper presented at Dept. Formation Continue I.C.P.I.

Lyon, Lyon, France, October 13 to 17, 1986. With permission.

The extent and type of the pretreatment depends on (1) the type of membrane, (2) the
physicochemical characteristics of water to be treated, (3) the operating conditions (viz.,
the conversion ratio, applied pressure, etc.).
As a guideline, the feed may be pretreated such that the frequency of washing the RO
module is not less than 1 month. Table 6 gives the required water quality that is suitable
for RO application.
Sometimes ultrafiltration is used as a pretreatment unit. It removes suspended matter
colloidal matter, high molecular weight organics, bacteria, etc. and produces excellent quality
of water (turbidity < 0.2 nephelometric turbidity units [NTU]. However, the cost incurred
in using UF as a pretreatment unit is higher than for the classical pretreatment methods.

VI. THEORETICAL CONCEPTS

A. Membrane Structure and Rejection Mechanism

The main aim of workers in this field of membrane separation has been to develop
technology to minimize flow resistance of the membrane by making it as thin as possible
while still maintaining structural integrity. For example, when cast as a RO membrane, the
membrane film has a thin dense layer of about 0.25 |xm above a thick, porous layer about
100 |xm thick. Pore size of the thin film is of approximately a nanometer size order. Water
passes easily from the dense layer to the porous one, but with difficulty the other way
around.
Filters and membranes differ in the mechanism by which the solute is retained. Filters,
such as paper, rely upon having the particles trapped within the fibrous network composing
them, which eventually results in a decreasing flux rate due to plugging. Membranes, on
the other hand, rely on the discreteness of the pore size opening in contact with the feed
solution and with the porous substructure under the thin skin. It is unlikely that particles
will become trapped within the membrane. The ability of a membrane to reject organic
compounds is based on this sieve mechanism . 6 The size and shape of the organic molecule
are important factors which determine whether the molecule will pass through the membrane
(Figure 8 ).
But this sieve-type mechanism is insufficient to explain the 97% (or more) salt rejection
obtained by CA membranes, since there is insufficient molecular size difference between
water and the common ions found in sea water. The high salt rejection occurs, in this case,
due to the repulsion of ions away from the membrane, and the preferential adsorption of
water molecules on the membrane surface (Figure 9) . 6 The pressure applied on the feed
forces the water through the membrane while retaining the salt in the bulk solution. Thus,
the ability of RO to separate ions from water depends not only on the pore size, but also
on the surface adsorption phenomena.
 149

M, LARGE M O L EC U LA R
PRESSURE WEIGHT ( = ­ 1 0 0 0 )

FEED F m 2 M ED IUM M O L E C U L A R
WEIGHT ( 3 0 0 - 5 0 0 )
m 3 SM ALLER MOLECULAR
WEIGHT ( - 2 0 0 )

SKIN

POROUS
SUBSTRUCTURE

FIGURE 8. Movement of molecules through membrane. (Adapted from Sundstrom, D. W. and Klei,
H. E., Wastewater Treatment, Prentice Hall, Englewood Cliffs, N.J., 1979, 274. With permission.)

PRESSURE
FEED FLOW

ADSORBED WATER LA YER

D IS C R E T E
PORE S IZE ( 2 D )
N a+ Cl

D SKIN

POROUS
S U B S TR U C TU R E

FIGURE 9. Preferential adsorption of water molecules on the membrane surface. (Adapted from
Sundstrom, D. W. and Klei, H. E., Wastewater Treatment, Prentice Hall, Englewood Cliffs, N.J.,
1979, 274. With permission.)

B. Transport Model: Flux Equation and Performance Variables

The solvent flow (Jw) can be represented by the following equation :6

Jw = A JA P - An) ( 1)
where AP = pressure on feed side-pressure on permeate side; AU = osmotic pressure of
feed-osmotic pressure of permeate; and Am = membrane permeability coefficient (or hy ­
draulic permeability).
The osmotic pressure of feed can be calculated from the following formula:

An = nRT (2)
where n is the molar concentration of solute, R is the gas constant, and T is the absolute
temperature.
However, the solute flux (Js), which does not change directly with the applied pressure,
can be calculated from the diffusion equation based on Fick’s law (Equation 3).
150 Water, Wastewater, and Sludge Filtration

0.9

10 50 A P - AZ^(BARS)

FIGURE 10. Plot of R vs. (AP - Air).

Js = - • (AC?) (3)
T

where D is the molecular diffusivity of the solute in the membrane, AC™ is the concentration
drop of solute across the membrane, and t is the membrane thickness.
From the mass conservation, the solute and solvent fluxes can be related by:

(4)

where C2 is permeate concentration of solute.

The rejection rate of solute (R) by membrane can be defined as follows:

where C 2 is the bulk solute concentration.

By combining the Equations 3, 4, and 5 with the assumptions that C™ = C2, we get

The plot R vs. (AP — Att), gives the following theoretical curve (Figure 10). But the curve
obtained from experimental values will be completely different from the theoretical curve
and can be accounted for by the phenomenon known as ‘’concentration polarization,” viz.,
the solute concentration of the solution adjacent to the membrane (C™) is considerably higher
than the concentration of bulk solution (C2) (i.e., R? > > C2).

C. Concentration Polarization
The enrichment of the salt concentration in a boundary layer near the membrane surface
with increased flow of water through the membrane and as salts are rejected by the membrane
is termed ‘‘concentration polarization.”
 151

dc
UmC - D ¥
Cp Vm

dz

FIGURE 11. Film theory model.

Assuming a film theory model such as shown in Figure 11, one can write the following
steady state equation describing the solute transport within the film.
Transport towards the membrane-back transport = net transport

- D ^ + VmC = ( 1 - Ra)VmQ ’ (7)

dz

Here, Vm is the velocity of the permeate through the membrane, C is the concentration of
solute at a distance Z, and Ra the intrinsic rejection.
The intrinsic rejection is given by:

R„ = 1 - (8)
Integration of Equation 7 gives:

C™ exp (Vm-8 /D)

(9)
C2 Ra + (1 - Ra) exp (Vm*8 /D)

For a tubular module, the above equation is modified as:

exp [25(Vm/u)-Re1/4-Sc2/3]
7 = (1 0 )
Ra + (1 - Ra) exp[25 (Vm/u)-Re‘'4-Sc2' !]
152 Water, Wastewater, and Sludge Filtration

CONCENTRATE

ST A G E 1 STAGE 2 STAGE 3 STAGE 4

FIGURE 12. Continuous operation, multi-stage (cascade).

where, Re = Reynolds number, u = velocity, Sc = Schmidt number ( = v/D), and 7 =

factor of polarization.
In reverse osmosis, the solution viscosity is relatively high due to recirculation of the
concentrate, and the coefficient of molecular diffusivity is high due to the smaller size of
particles to be removed (ions and low molecular weight molecules). This lowers the values
of both Sc and Re, which in turn reduces the value of 7 . This indicates that the concentration
polarization is lower in RO compared to ultrafiltration.
The concentration polarization thus results in the following adverse effects in RO: increase
of osmotic pressure (Att) (i.e., decrease in effective pressure AP — Att), decrease in solvent
flow, decrease in rejection rate, and precipitation of sparingly soluble salts.
The effect of concentration polarization can be minimized by having appropriate chemical
pretreatment (for example, control of pH or by pretreatment using UF or microfiltration to
reduce the colloidal solid load to RO).
All the transport equations derived above relate the flux rates to the concentration of solute
and its properties next to the membrane. It is often seen that solute concentration becomes
dependent upon the fluid dynamics and mass transfer properties of the boundary layer next
to the membrane.

D. RO System Design
A RO system essentially consists of a pump to pressurize the feedwater, a RO device,
and a throttling valve on the concentrate end to control conversion . 1
The basic design of a single-RO -device system is shown in Figure 7. As can be seen in
this schematic diagram, low pressure pumps for influent feed and circulation, a high pressure
pump to pressurize the feed, a RO module pretreatment facility, possibly an energy recovery
facility, and sampling facilities are the basic components.
For small-scale applications, single-stage systems can be used. It can be a batch process
with recirculation. But for large-scale applications, a multi-stage arrangement is used with
continuous operation mode (Figure 12).
While installing the RO system, one should consider the following: membrane selection,
permeate flow rate, membrane life, and the cleaning frequency of membranes. The parameter
of permeate flow rate is dependent on feed rate, trans-membrane pressure, temperature,
conversion rate, degree of pretreatment, and operation cycle time. It is necessary to consider
the above factors in order to arrive at an optimum design.

VII. APPLICATIONS

The reverse osmosis unit was first used for desalination of salt water. At present, there
are large installations with capacity of 2 x 1 0 6 m3/d to treat the sea water for domestic and
industrial uses . 5 In addition to this, there are small- and medium -sized RO installations
existing (for 0.4 m3/d to 95,000 m 3/d) to supply pure water for specific purposes like
petroleum platforms, agricultural purposes, sterilized water for hospitals, laboratories, etc .4
The applications of membranes are, therefore, classified into three groups and are discussed
 153

Table 7
ENERGY CONSUMPTION IN
SEAWATER DESALTING
PROCESSES

Energy consumption
Process (kWh/m3)

Multistage flash distillation 15.5

Vapor compression distilla­ 11
tion
Reverse osmosis 9
Reverse osmosis with en ­ 6.6
ergy recovery

From Ben Aim, R., Les Techniques a Membrane dans

le Traitement de 1’ Eau, unpublished technical report,
1986. With permission.

in the sections on potable water treatment, pure water for specific purposes other than
domestic purposes (industrial water supply), and wastewater treatment (both municipal and
industrial).

A. Potable Water Treatment

1. Desalination
The major application of reverse osmosis is the desalting of sea water into potable water.
Table 7 presents a comparison on energy consumption of different desalting processes.
The performance of RO installation with a capacity of 12,000 m3/d at Jeddah was reported
by Maurel in 1984.5 It consists of pretreatment, RO unit, and posttreatment as shown in
Figure 13. The pretreatment includes rapid sand filter (operating at 6 . 8 m 3/m 2h), acidification
by H2S 0 4 to maintain the pH at 6.4, injection of sodium hexametaphosphate (2 mg/€), and
cartridge filtration (25 p,m).
As shown in Figure 13, the pretreated water is further treated by two-stage RO units
(spiral-wound modules with PA 300 poly-ether-amide membranes). The treated water under­
goes posttreatment which includes pH adjustment by lime addition and disinfection by
chlorine addition. The energy consumption for the installation is about 12 kWh/m 3 of water
treated.

2. N 0 3 Removal in Potable Water Treatment

In some countries, the surface water, as well as water from aquifers, contains N 0 3 in
excess and, therefore, has to be treated. One of the methods to remove N 0 3 is by the use
of reverse osmosis.
The N 0 3 removal efficiency of RO depends on the applied pressure. For instance, in a
treatment plant, RO removed 85% of N 0 3 with an applied pressure of 30 bars, while at a
pressure of 60 bars, the efficiency increased to 95 % . 7 However, the cost factor limits the
application of RO in N 0 3 removal in potable water treatment.

B. Production of Pure Water for Industrial Purposes

7. Ultrapure Water fo r the Electronics Industry
One of the main industrial uses of the RO process is to prepare ultrapure water for the
electronics industry. With the task of preparing the smallest possible integrated circuits, one
should have an ultrapure water to clean these circuits. According to the present standards,
rinsing water should conform to the following criteria:8
 SAND FILTER
( 3 U N IT S ) C H EM IC A LS
Water, Wastewater, and Sludge Filtration

( PRESSURE = 3 5 BARS )

FIGURE 13. RO unit with post treatment. (From Maurel, A., paper presented at Dept. Formation Continue I.C.P.I., Lyon, France, October 13 to 17,
1986. With permission.)
 155

Table 8
QUALITY OF WATER REQUIRED FOR
CLEANING OF CIRCUITS OF
DIFFERENT CAPACITY

Capacity
Quality parameters 16K 64K 256K

Resistivity (Mft/cm at 25°C) 15 17 18

Si02 (ppb) 50 20 10
TOC (ppb) 1000 500 50
Particles
Size (p-m) 1 0.5 0.2
Number/cm3 10 10 100
Bacterial count/100 cm3 100 50 5
Dissolved oxygen (ppm) 8 8 0.3

From Dinard, H., L ’eau, Vindustrie, les nuisances, 94, 59,

1985. With permission.

• it should not have particles above 0 . 2 |xm

• the resistance of water should be 15 to 18 Mfl/cm
• it should not have any trace of contaminants, like dissolved organic matter, dissolved
minerals, bacteria, colloids, particles, and dissolved gases

Except for the dissolved gases, membranes can successfully remove the other contami­
nants. For the sake of comparison, the purity of water required for the cleaning of circuits
of two different capacities is presented in Table 8 . 9
The ion exchange process can produce water with resistivity of 18 M(l/cm, but cannot
remove the particles. The RO unit, on the other hand, can eliminate the particles as well,
but is easily clogged. Hence, it requires pretreatment like microfiltration or ultrafiltration.
The schematic diagram in Figure 14 gives an idea of the processes involved in the production
of ultrapure water.
Here the use of RO is to reduce the load to ion exchange process, by feeding bacteria-
free water to the ion exchanger. The clogging of the RO membrane is reduced by having
some pretreatment, such as cartridge filtration or microfiltration.
An application of a RO unit in an electronics industry in Pologue indicates that incor­
poration of reverse osmosis reduces the frequency of regeneration of ion exchange resin by
5 times, while improving the quality of water . 11

2. Sterilized Water fo r the Pharmaceutical Industry and for Medical Purposes

These industries require sterilized water (water absolutely free from bacteria) inaddition
to a very low SS level. Use of a RO unit will serve both purposes.
For example, a pharmaceutical industry in the south of England replaced its old process
(mixed bed ion exchange and submicron filtration) by a new process consisting of RO and
an ion exchanger. 12 To obtain 20 m3/d water in the older process, the ion-exchange resins
had to be regenerated every day with 70 € of acid and 60 € of base. This corresponds to a
daily chemical cost of 17.6 English pounds. In the new process where a RO unit is incor­
porated, the ion exchanger is regenerated once in 10 to 14 days, reducing the chemical cost
to 1 / 10 th of the original cost.

C. Use of RO in Wastewater Treatment

The use of membrane technology in the field of wastewater treatment was first studied
 156 Water, Wastewater, and Sludge Filtration

P REFILTER DEGASIFIER

FIGURE 14. Ultrapure water production. (From Judith, S., paper presented at Dept. Formation Continue I.C.P.I.,
Lyon, France, October 13 to 17, 1986. With permission.)

FIGURE 15. Tertiary treatment of municipal wastewater.

by EPA. The most common geometries of membranes used are either spiral or tubular
because of the presence of suspended particles .4

1. Municipal Wastewater Treatment

The main application of RO in municipal waste treatment is in tertiary treatment. RO is
used as a polishing unit of the effluent from an ultrafiltration unit (Figure 15). For example,
in Nakskoo in Denmark (17,000 in population), the above tertiary treatment is practiced . 13
The performance of the treatment plant is summarized in Table 9. In the use of reverse
osmosis in tertiary treatment, one should have an appropriate pretreatment to avoid the
clogging of the membrane by organics and inorganics. The common pretreatment for removal
of inorganics includes: softening, removal of Fe and Mn, and dechlorination (especially
when polyamide membranes are used). The microorganisms can be removed by either
chlorination or UV disinfection.

2. Use o f RO in Industrial Wastewater Treatment

In cheese production, 90% of the milk used is found in the form of lactoserum. If it is
disposed of without treatment, it causes a serious pollution problem in the receiving waters.
In most cases, it is either dried or separated into different components.
The reduction of significant volume of lactoserum is very essential before its ultimate
treatment and disposal. Two common processes used are evaporation and reverse osmosis.
As an example, in an installation producing 12,000 € of lactoserum per hour, it has been
 157

Table 9
EFFLUENT QUALITY OF NAKSKOV
PLANT AT VARIOUS STAGES413

Quality
parameters Secondary UF RO
(mg/€) effluent filtrate filtrate

Oxygen demand 390 150 15

bod5 190 84 12
Suspended solids 210 < 1 < 1
Total nitrogen 34 7 3
Total phosphorous 9.5 7 0.03
Chloride 790 760 3.9

found that reverse osmosis is much more economical compared to evaporation . 14 The mem ­
branes were replaced in the RO units every year.
The other uses of RO units in industrial wastewater treatment include: recovery of acids
and metal in the metal plating industry, and recovery of ZnCl2 in the paper industry.

VIII. ADVANTAGES

1. Energy requirements are low, because RO performs a separation without a phase

change. Brackish water plants typically consume 1.6 to 2.1 kWh/m 3 of product water.
Seawater plants consume about 9.2 to 10.6 kWh/m 3 of product water. Recent devel­
opments have led to a decrease in energy consumption for seawater plants by using
energy recovery equipment on the brine side.
2. In addition to desalting water, RO removes dissolved S i02, colloidal matter, and high-
molecular weight (>120) organics. This has led to RO being used, in conjunction
with ion exchange, by the power industry to produce high pressure boiler feed water
and by the electronics industry to produce ultrapure water for manufacturing electronic
components.
3. RO system is compact compared to other desalting systems (e.g., distillation or elec­
trodialysis).
4. Unskilled operators usually can be taught in a short period of time to operate the RO
plant since RO components are quite familiar — pumps, motors, valves, flow meters,
conductivity meters, and pressure gauges.
5. Automatic controls are designed to protect the system and prevent catastrophic failure;
and if shutdown occurs, less than half an hour is required for start-up.
6 . Modular design of a RO plant allows scheduled maintenance to be performed without
shutting down the entire plant. This modular design also makes expansion an easy
option.

IX. LIMITATIONS

1. From the fundamental transport equations derived earlier, it is seen that the applied
pressure of RO systems must exceed the osmotic pressure to obtain product flow and
separate the solute from the solvent. Therefore, RO is usually not applicable for
concentrated solutions.
2. There are limits to the maximum operating temperature for the different kinds of
membrane used. For CA, aramid, and their film composite membranes, it is 30 to
40°C, 35 to 45°C, and 45 to 50°C, respectively. Excessive loss of productivity occurs
158 Water, Wastewater, and Sludge Filtration

above these temperatures due to membrane compaction. Therefore, feed streams that
exceed these limits must be cooled before proceeding to RO.
3. Pretreatment is essential as RO systems are susceptible to fouling.
4. Oil and grease severely affect RO membranes and, therefore, must beremoved.
5. Chemical compatibility of RO devices for different kinds of wastewater may be un ­
known. Applications of RO to these involve pilot testing to ascertain both compatibility
and separation performance.

REFERENCES

1. Applegate, L. E., Membrane separation processes, Chem. Eng., 91, 64, 1984.
2. Reid, C. E. and Breton, E. J., Water and iron flow across cellulosic membranes, J. Appl. Polymer Sci.,
1, 133, 1979.
3. Aptel, P., Ultrafiltration et osmose-inverse preparation et caracteristiques des membranes et modules, Inf.
Chim., 213, 189, 1981.
4. Ben Aim, R., Les Techniques a Membrane dans le Traitement de l ’Eau, unpublished technical report,
1986.
5. Maurel, A., Application de l ’osmose Inverse au Dessalement de l ’eau de mer, Techniques Separatives sur
Membranes, presented at Dept. Formation Continue, Institute de Chimie et Physique Industrielles de Lyon,
Lyon, France, October 13 to 17, 1986.
6. Sundstrom, D. W. and Klei, H. E., Wastewater Treatment, Prentice Hall, Englewood Cliffs, N.J., 1979,
274.
7. Laprince, A. and Fiessinger, F., Le probleme des nitrates dans les eaux potables: Efficacite et cout des
solutions disponsibles, J. Fr. d ’Hydrologie, 15, 3, 279, 1984.
8. Short, J. L., Selection, application and optimization of hollow fibre UF membranes, Filtr. Sep., 5, 410,
1982.
9. Dinard, H., Le Traitement de l ’eau ultrapure utilisee dans Tindustrie de la microelectronique, L ’Eau,
VIndustrie et les Nuisances, 94, 59, 1985.
10. Judith, S., Millipore, Microfiltration dans lTndustrie Electronique, presented at Dept. Formation Continue,
I.C.P.I., Lyon, France, Oct. 13 to 17, 1986.
11. Daniel, L., L ’osmose Inverse et ses Separation, Techniques Separatives sur membranes, presented at Dept.
Formation Continue, I.C.P.I., Lyon, France, Oct. 13 to 17, 1986.
12. Cross, J., The Use of Reverse Osmosis Systems to Purify Water for Industrial and Medical Applications,
Filtech Conference, Filtration Society, London, 1983, 128.
13. Liquid Filtration Newsletter, The Mcllvaine Company, 44, 3, 1983.
14. Huchon, J. and Reynaud, J. F., Consommations energetiques coute d ’application des precedes de con­
centration, Le Lait, 64, 305, 1984.
 159

Chapter 9

U LTR A FILTR A TIO N

Saravanam uthu Vigneswaran

TA B LE O F CONTENTS

I. Introduction...................................................................................................................... 160

II. Process Description.........................................................................................................160

III. UF M em branes............................................................................................................... 160

IV. Types of UF D evices..................................................................................................... 160

V. Pretreatment for UF S ystem s........................................................................................162

VI. Theory of U F ...................................................................................................................163

VII. Applications................ 164

A. Domestic Wastewater Treatm ent.................................................................... 165
1. UF Instead of Secondary C larifier.............................................. 165
2. UF to Recycle the Biomass to Anaerobic Digester and
to Clarifythe Effluent from D igester.................................................165
3. UF in the Tertiary Wastewater Treatment..........................................165
B. Industrial Wastewater Treatm ent.................................................................... 166
1. Treatment of Lactoserum .....................................................................166
2. Use of UF in Fermentation of Lactose...............................................167
3. Other Applications in Industrial Wastewater Treatment..................167

VIII. Economy........................................................................................................................... 168

IX. Advantages...................................................................................................................... 168

X. Disadvantages................................................................................................................... 169

XI. Major Differences of UF from Other Membrane Processes.....................................170

References.................................................................................................................................... 170
160 Water, Wastewater, and Sludge Filtration

I. INTRODUCTION

The ultrafiltration (UF) membrane allows the passage of water and low molecular weight
solutes but retains macromolecules whose size is bigger than the pore size of the membrane
(Figure 1). UF pore sizes usually range from 0.001 to 0.02 p-m. 1 UF enables concentration,
purification, and fractionation of macromolecules in solution to be carried out at temperatures
close to ambient temperature and without phase change or addition of solvents. This protects
the biochemical structure and activity of the product, giving increased yield over conventional
technologies.

II. PROCESS DESCRIPTION

UF utilizes permeable membranes to separate macromolecules and suspended solids from

solution on the basis of size, separating compounds with molecular weights from 1 0 0 0 to
100,000 (1 to 100 nm in size). As a guideline, some of the chemicals used to calibrate the
pore sizes with their molecular weights and molecular diameters are presented in Table 1.
The application of high pressure to the feed side of the membranes enables the passage
of water through the membrane. This makes the higher molecular weight compounds con ­
centrate on the high pressure side, while the concentration of lower molecular weight com ­
pounds remains same on both sides of the membrane (Figure 1).
Two basic operation modes, namely, batch and continuous, are used. In the batch system,
the liquid is fed into the UF, permeate is withdrawn, and the concentrated liquid is returned
to the feed tank. There may be a single pass or multiple passes through the membrane until
the desired concentration or required purity is obtained. In a continuous system, the feed is
continually supplied, and the number of stages and feed flow rate are determined, based on
the required conversion ratio and required permeate quality. A continuous multistage system
is better than the single stage system because of its improvement in the average flux. UF
system size varies from 1.0 m 2 to 1000 m2. Average processing time of UF is 16 to 20 h.
Cleaning of membranes takes 2 to 4 h and may include an acid or caustic cycle, a sanitization
cycle, or combinations of these, with a water rinse interspaced between each cycle.

III. UF MEMBRANES

Typically, UF membranes have a thin (0.1 to 1 pm) surface “ skin” supported by a porous
substructure. Pore sizes (“ molecular weight cutoffs” ) can range from equivalent molecular
weights as low as 1 0 0 0 to as high as 1 0 0 ,0 0 0 .
A variety of polymers such as cellulose acetate, polyvinyl chloride, poly aery lonitrile,
polycarbonate, and polysulfone have been used to manufacture UF membranes (Table 2).
Membranes must be compatible with the feed solution and, since membranes are subjected
to surface fouling, they must be compatible with cleaning agents, also.

IV. TYPES OF UF DEVICES

There are three primary UF configurations, such as tubular, spiral-wound, and hollow
fiber.
A tubular UF device is similar in construction and operation to the tubular RO device.
As in RO tubular devices, this type is used in small-flow, high-value applications because
membrane area per unit volume is small. For applications anticipated with severe fouling,
this should be given serious consideration as they can be cleaned manually. The characteristics
of some commercially available UF tubular modules are summarized in Table 3 . 2,3
In a spiral-wound UF device, again, the general design is similar to that of the spiral-
 161

SOLUTION FEED • IONS

FIGURE 1. Movement of molecules through membrane.

Table 1
CALIBRATION OF
ULTRAFILTRATION MEMBRANES

Molecular
Molecular diameter
Sample weight (A)

Saccharose 342 10
Raffinose 594 12
Cytochrome 12,400 20
Myoglobin 17,800 22
Pepsine 35,000 25
Ovalbumin 45,000 27
Hemoglobin 64,000 36
Serum albumin 67,000 40
Dextran 250 250,000 57

Table 2
MEMBRANE MATERIALS USED IN INDUSTRIAL
APPLICATION4

Max temp
(°C) Chlorine Solvent
Material type pH Range at pH 7 resistance resistance

Cellulose acetate 4.5 — 9 55 Fair Poor

Polyamide 3— 12 30 Poor Good
Polysulfone 0— 14 30 Fair Fair
Polyacrylonitrite 2— 12 60 Good Fair
Polyfuran 2— 12 90 Poor Good

wound RO device. Since the membrane area is greater than the tubular type UF membrane,
it can economically treat large volumes. The disadvantage is that the spiral-wound UF device
is more vulnerable to fouling than the tubular device. Further, fouling can be controlled
only by chemical cleaning and flushing, since it cannot be cleaned mechanically.
The hollow -fiber UF device differs completely in design and mode of operation from the
RO hollow -fiber device. The membrane is generally made of acrylic copolymer, has a 0.1 -
162 Water, Wastewater, and Sludge FUtration

Table 3
CHARACTERISTICS OF SOME TUBULAR MODULES
OF UF

Tube
Internal Active
Commercial diameter Length surface
Manufacturer name (mm) (m) Number (m2)

Abcor UF-600-FE 25.4 3.0 12 2.4

Nitto P18A 11.4 2.6 18 1.6
Patterson-Candy B1 12.5 3.6 18 2.6
SFEC Carbosep 15.0 1.2 150 8.0
Wafilin WFS 17.2 6.0 18 4.9
Western Dynet 6x6 25.0 1.8 18 2.6

From Veyre, M., paper presented at Dept. Formation Continue, I.C.P.I., Lyon,
France, October 13 to 17, 1986. With permission.

Table 4
CHARACTERISTICS OF SOME HOLLOW FIBER
MODULE MEMBRANES OF UF3

Hollow fiber module

Internal Active
Commercial diameter Length Surface
Manufacturer name (mm) (m) Number (m2)

Romicon _ 1.1 0.63 660 1.4

Berghoff BST 1 0.6 __ —­ 0.5

Table 5
COMPARISON OF MAJOR MODULES CHARACTERISTICS4

Ease of Resistance
Module Hold-up Power mechanical to mechanical Field
configuration volume Prefiltration consumption cleaning damage replacement

Tubular High No High Good Good Simple

Spiral wound Low Yes Low Poor Poor Simple
Hollow fiber Low Yes Low Poor Good Simple

|xm skin on the interior, and is supported by an outer substructure. It is, therefore, very
necessary to have a high degree of pretreatment to avoid clogging of fibers. The characteristics
of some of the commercially available hollow fiber devices are given in Table 4 .2,3
Table 5 gives the comparison of different configurations of UF modules .4

V. PRETREATMENT FOR UF SYSTEMS

The fouling in UF can be due to concentration polarization and pore plugging or to a

slow consolidation of the gel layer (Figure 2). Past experiences indicate that fouling in UF
membranes can reduce productivity tenfold relative to that of water. It will also decrease
the molecular weight cutoff. This is unfavorable when macromolecular fractionation is being
performed and, thus, fouling must be considered in design. In order to obtain the desired
productivity, the effect of fouling must be reduced by increasing the membrane area and
 163

CONCENTRATE

. DIFFUSION .
OF •
SOLUTE-;;:
0
Ap .V;

c o n v e c t io n : i

PERMEATE
■ :
SOLUTION v
• TO BE i
TREATED ' f
MEMBRANE

FIGURE 2. Gel formation in UF.

feed pressure, and by limiting conversion. To reduce fouling, high feed flow rate and lowest
practicable hydraulic pressure difference across the membranes are used. Sodium hypo ­
chlorite, detergents, and sodium hydroxide are some reagents used for cleaning the mem ­
branes to remove foulants . 1 The cleaning agents must be compatible with the membrane.
Pilot tests must be performed to determine the fouling potential of feed streams, and an
appropriate pretreatment facility should be provided.

VI. THEORY OF UF

The classical theory for UF depends on mass balance for the solutes in the laminar boundary
layer formed at the membrane surface. 5 This can be written as:

JW.C - D ^ = Jw-Cj (1)

dZ

where Jw is the permeate flux, C is the concentration of solute, and Z is the distance from
membrane.
Integration of Equation 1 for constant J yields:
164 Water, Wastewater, and Sludge Filtration

C? - C?
Jw = kL ln c 2 - C l (2 )

where C2, C™, Q are the concentrations of solute in the bulk process stream, at the membrane
surface, and in the permeate, respectively; kL is the mass transfer coefficient (D/S)
Experimentally estimated Jw for various applied transmembrane pressure values indicated
that Jw is directly dependent on the applied pressure until a certain pressure is achieved.
Above this pressure value, the flux become plateaus, due to gel formation. The gel con ­
centration is generally of the order of 30 to 50% weight of solids.
The expression relating flux to transmembrane pressure (AP) can be written as

AP
Jw = ------------------- (3)
M(R»
- + Rc)

where Rm and Rc are the hydraulic resistance of the membrane and the polarized layer,
respectively . 5
Commercial UF units operate at 1 to 5 bars. A significant proportion of applied trans­
membrane pressure is necessary to overcome the hydraulic resistance of the gel layer.
Therefore, it is economical to operate until the plateau is reached, but not beyond “ gel
polarization” .
Rejection rate of salt increases with the applied pressure and can be defined as

c2 - a c?
R = — = 1 - pr (4)
V^2 ^2

From conservation of solute one can write the following equation

h = Jw C£ (5)

where Js is the solute flux.

In UF, the concentration polarization effect is more significant (because the viscosity is
high and diffusion coefficient is low, which makes the Schmidt number [Sc] and, thus,
factor of polarization [y] high). In UF, in addition to the primary concentration polarization
effect, there is formation of gel near the membrane surface as shown in Figure 2, which is
known as secondary polarization (Figure 3). This effect can be reduced by having optimum
operating conditions. For example, if one chooses a membrane corresponding to point 2,
flow rate may be initially high, but the concentration near the membrane also will be high,
which will reduce the flux rate rapidly. On the other hand, if one chooses a membrane
corresponding to point 1 , although flow rate may be a little lower, it will be stable with
time as the concentration near the membrane will not be that high.

VII. APPLICATIONS

The applications of UF, for the production of specifically treated water for industries, are
numerous. It is also widely used in industrial wastewater treatment where recycling of raw
materials, products, and by-products are of primary concern. It is also used in secondary
and tertiary treatment of municipal wastewater.
As discussed in the chapter on reverse osmosis, ultrapure water is necessary in the
electronics industry. UF units can be used in the pretreatment or in the treatment step for
production of ultrapure water.
 165

FIGURE 3. Plot of permeate flux vs. transmembrane pressure.

In the pharmaceutical industry and medical applications, UF is widely used to produce

sterilized water. Table 6 compares the performance and cost of UF with other processes
producing sterilized water.
In closed circuit aquaculture practices, a continuous treatment of water is necessary, and
unit operations, like sedimentation, denitrification, and sterilization are commonly used. For
sterilization, a UV sterilizer is conventionally used which is suitable only for low turbid
waters. Instead of a UV sterilizer, UF can be successfully used.

A. Domestic Wastewater Treatment

1. UF Instead o f Secondary Clarifier
UF can successfully be used instead of a secondary clarifier (Figure 4) in the secondary
treatment. As shown in the figure, the concentrated sludge is returned to the reactor, whereas
the permeate is discharged. The use of UF in this field becomes more useful in the following
cases:

• Wastewater from hospitals, where high risk is involved in discharging the treated
effluent, as it may contain chemicals
• When stringent standards are enforced for effluents to be discharged

The advantages of using UF here are that it produces suspended solids and bacteria-free
effluent, while withstanding high variations of pollutant load.

2. UF to Recycle the Biomass to Anaerobic Digester and to Clarify the Effluent from Digester
The UF can be used to recycle the biomass in the anaerobic digester, while the permeate
produced is free from suspended solids (Figure 5). The incorporation of UF will not only
give more stabilized functioning of the digester. It also will be helpful in reducing the volume
of the reactor because of the feasibility of increasing the loading in view of the resulting
high concentration of microorganisms. This will, in turn, reduce the capital cost of the
reactor significantly . 8

3. UF in the Tertiary Wastewater Treatment

The UF can be used as a pretreatment to RO or as a complete treatment in the tertiary
166 Water, Wastewater, and Sludge Filtration

Table 6
COMPARISON OF UF WITH OTHER PROCESSES PRODUCING
PURIFIED WATER

Hollow
fiber Reverse Filtration
Criteria UF Distillation osmosis (on charged media)

Removal efficiency > 105 > 105 < 105 102—103

Approximate invest­ X 8X 2.5X 0.7X
ment cost (membrane
surface area between
5 and 20 m2)

From Armand, A., Le Lait, 64, 298, 1984. With permission.

AIR SUPPLY

WATER TO BE_
TREATED

FIGURE 4. UF as secondary clarifier in domestic wastewater treatment. (From Lambert, S., paper pre­
sented at Dept. Formation Continue, I.C.P.I., Lyon, France, October 13 to 17, 1986. With permission.)

wastewater treatment. The performance of such a tertiary treatment plant in the town of
Nakskoo is discussed in the chapter on reverse osmosis (see Table 9, Chapter 8 ). The cost
data indicated that the cost (capital and operation) of UF for this purpose was about U.S.
$0.5/m 3 of secondary effluent treated . 9
In Japan, at least 50% of the wastewater produced in some buildings (i.e., water used
for washing, cooking, bathing, etc.) is collected, purified, disinfected, and used as toilet
flushing water. This water is then sent through the sewer to the municipal wastewater
treatment plant. 7
In most instances, the onsite treatment used is biological or physicochemical. The phys­
icochemical treatment could be either conventional or unconventional that includes the use
of ultrafiltration. One such treatment incorporating UF was developed by the Mitsui Company
(Figure 6 ). This type of treatment has been found to remove all suspended solids, 95% of
BOD, and 90% of COD. The flow rate of such a household UF unit is 100 l/m 2/h. The cost
of treatment is compared with classical biological treatment in Table 7 .4 This study indicates
that the proposed treatment reduces the cost by U.S. $0.7/m 3 of water treated .7

B. Industrial Wastewater Treatment

1. Treatment o f Lac toserum
During the treatment of lactoserum, the UF permits the retention and concentration of
proteins and greasy matter and allows the passage of lactose and mineral salts through the
 167

- Regulation of pH
• BIOGAS
- N utrient ( N ,P ) addition

PUMP R E C YC LE

PUMP — - P U R IF IE D WATER
HEAT ANAEROBIC
BIOMASS
WASTE REACTOR ULTRA FILTRATION
WATER -C Z H U N IT
EXCHANGER

EXCESS
BIOMASS
PUMP

FIGURE 5. UF in anaerobic digestion process.

UF WITH RECYCLE
(U F COMBINED WITH BIOLOGICAL OXIDATION)

FIGURE 6. Household UF used in tertiary wastewater treatment. (From Lambert, S., paper presented at Dept.
Formation Continue, I.C.P.I., Lyon, France, October 13 to 17, 1986. With permission.)

membrane as permeate. The protein concentrate can be used as animal and human food,
while the permeate is used for lactose preparation. As an indication, the UF of lactoserum
can be effected at 50°C under a pressure of 2.5 bars with a flow velocity of 2 m/s or greater. 10
Another example indicates that a hollow -fiber membrane module made of polysulfone with
a molecular weight cutoff size of 1 0 ,0 0 0 can be used to concentrate the protein in lacto­
serum . 11

2. Use o f UF in Fermentation o f Lactose

After the extraction of proteins, lactoserum contains essentially the fermentable sugar,
the lactose, at a concentration of 50 g/1. The fermentation of lactose in lactic acid seems
not only to be very economical (since lactic acid has high economic value) but also a suitable
way of treating lactoserum . 12 The use of ultrafiltration for anaerobic digestor effluent will
be helpful in enriching the bacteria which are recycled into the digester, while producing a
clear effluent which can be discharged without any problem . 8

3. Other Applications in Industrial Wastewater Treatment

Other uses of UF in industrial wastewater treatment applications include:

• Recovery of paints in the electrophoretic painting industry. The UF recovers the paint
in the waste stream which is considered to be about 35% of paint consumed. The
permeate can also be reused . 1013
• Use of ultrafiltration to treat and recycle the caustic extraction waste after the bleaching
process in the paper and pulp industry . 13 The average water consumption in the paper
168 Water, Wastewater, and Sludge Filtration

Table 7
COST COMPARISON OF
WASTEWATER TREATMENT IN
TOKYO

Treatment Approximate cost

U.S. $/m3

Potable water (water rate) 1.15

Municipal wastewater treatment 0.80
(sewer rate)
Total 1.95
Pure water after UF and biological 1.25
treatment (with UF treating 50%
of the recycled wastewater)

Note: Amortized cost as of 1982.

and pulp industry is about 50 m3/ton of pulp produced. The effluent produced by the
caustic bleaching process is highly colored due to dissolved organics. This may be
caused by lignins in the black liquor (which is not recovered during the washing
process) and degraded lignin products formed from the caustic bleaching process. The
classical biological treatment reduces the BOD and COD but not the color. If UF is
used in the process, not only can recovery be achieved, but also a high degree of
treatment efficiency (Figure 7). Past experience with UF shows that a UF membrane
with spiral configuration and a molecular weight cut off size of 5000 reduced the color
by 90% while reducing the wastewater production by 25 times.
• Removal of fiber in the textile industry. The synthetic fiber that escapes into the
wastewater of a textile factory is generally not biodegradable and creates problems in
wastewater treatment. If a UF unit is placed in the manufacturing process, not only
the fiber can be recycled, but the treatment efficiency, too, can be improved . 13 In ­
corporation of UF at various steps recovers the following materials: 1 continuous re ­
covery of textile lubricants with high BOD; concentration and recovery of valuable
dyes; recovery of caustic soda (NaOH).

VIII. ECONOM Y

The economics of UF is dependent on the specific application. For clean feed streams,
such as in the post treatment of ion exchange water, capital cost can be as low as about
U.S. $132/m3/d of installed capacity. Operating cost, which includes labor, cleaning
chemicals, power, replacement of UF devices every 4 years, and amortization, runs to about
U.S. $0.06/m 3 of perm eate . 1 For electrocoating paint applications, capital cost of
U.S. $1527/m3/d and operating cost of U.S. $1.75/m 3 of permeate have been reported . 1 The
capital and operating cost for concentration of enzymes by UF is comparable to that of
evaporation, but is lower than that for vacuum evaporation or vacuum evaporation followed
by salt or solvent precipitation. Also, the recovery is higher when UF is used (90 to 98%).1
Though the capital cost of UF is four times that of a precoat rotary vacuum filter, the
operating cost of UF is only 60% of it. The difference is made up in about 3 years.

IX. ADV ANTAG ES

1. There is no heat added and, hence, heat labile substances are not harmed.
2. Energy requirements are lower than for evaporation, and no change of state is required.
 169

WOOD IN CHIPS FORM

PULP FOR BLEACHING

PROCESS

FIGURE 7. UF application in paper and pulp wastewater treatment. (From Pichon, M., Muratore, and Morzie,
P., in Proc. 1st World Filtration Congr., Paris, May 14 to 17, 1974, F49-1. With permission.)

3. The products being recovered are not subjected to chemical desaturation as in the case
of salt precipitation or solvent extraction.
4. Concentration and purification may be achieved in one step.
5. Dangerous organisms and compounds may be totally separated from permeate.
6 . The permeate is free of macromolecules, such as pyrogens and suspended solids.
7. The UF may be used either directly as a continuous reactor by attaching cells or
enzymes to the membrane or by retaining cells and enzymes with the membrane.
8 . Wide range of molecular weight cutoffs are available.
9. Well suited for temperature-sensitive materials.
10. Compact and simple, requiring simple maintenance and easier installations.

X. DISADVANTAGES

1 . Some complexes, such as those of cyanide, may pass through quite easily.
2. Long residence time in batch systems may lead to microbial problems.
3. Membrane fouling resulting from concentration polarization and pore plugging or a
slow consolidation of the gel layer, reduces the molecular weight cutoff. This neces­
sitates frequent cleaning of membranes with such reagents as sodium hypochlorite,
detergents, and sodium hydroxide.
170 Water, Wastewater, and Sludge Filtration

Table 8
DIFFERENCES BETWEEN UF, RO, AND MICROFILTRATION4

Microfiltration UF RO

Membrane Porous (0.1— 10 Porous (10°A°—0.1 Nonporous (asymetric

p.m) isotropic jxm) asymetric or composite mem­
branes)
Transfer mechanism Sieving mechanism Diffusion law
The solutes migrate The solute migrates by
by convection diffusion mechanism
Law governing the Darcy’s Law Fick’s law
transfer
Type of solution Solution with solid Solution with colloids Ions, small molecules
Treated particles and/or macromolecules
Permeability range of 10— 100 m3 1 m3/m2 bar • d 0.01 m3/m2 bar • d
solvent (m2 • bar • day)
Pressure applied 1 bar 1— 5 bars 20— 80 bars

4. UF cannot be used to separate dissolved salts or low molecular weight (< 1000) species.
5. Reproducibility is poor.
6. Adsorption on pore walls and plugging is possible.

XI. MAJOR DIFFERENCES OF UF FROM OTHER MEMBRANE PROCESSES

Table 8 presents the differences between ultrafiltration, reverse osmosis (RO), and mi­
crofiltration (MF).

REFERENCES

1. Applegate, L. E., Membrane separation process, Chem. Eng., 91 (12), 64, 1984.
2. Aptel, P., Ultrafiltration et osmose inverse — preparation et caracteristiques des membranes et modules,
Inf. Chim., 213, 189, 1981.
3. Veyre, M., Les Milieux Filtrants Utilises en Ultrafiltration, Techniques Separatives sur Membranes, pre­
sented at Dept. Formation Continue, Institut de Chemie et Physique Industrielles de Lyon, Lyon, France,
October 13 to 17, 1986.
4. Ben Aim, R., Les Techniques a Membrane Dans le Traitement de l ’Eau, Institut de la Filtration et Techniques
Seperatives, Agen, France, 1986.
5. Fell, C. J. D., Wiley, D. E., and Fane, A. G., Optimisation of Module Design for Membrane-Ultrafil-
tration, Proc. 3rd World Chem. Eng. Congr., Tokyo, Japan, September 21 to 25, 1986, 266.
6. Armand, A., Elimination des bacteries et des pyrogenes par Vultrafiltration, Le Lait, 64, 298, 1984.
7. Lambert, S., Epuration et Sterilisation Simultanees d ’eaux residuaires de Petites Collectivites (Hopitaux,
Camping, Immubles), Techniques Separatives sur Membranes, presented at Dept. Formation Continue,
Institut de Chemie et Physique Industrielles de Lyon, Lyon, France, October 13 to 17, 1986.
8. Gailland, D., Reacteurs bioloqiques a membranes Dorr-Oliver: application a la digestion anaerobic du
lactoserum, Le Lait, 64, 252, 1984.
9. Liquid Liltration Newsletter, The Mcllvaine Company, 44, April 10, 1983.
10. Roulet, M., Les Application Industrielles de Tultrafiltration, Techniques Separatives sur Membranes,
presented at Dept. Formation Continue, Institut de Chemie et Physique Industrielles de Lyon, Lyon, France,
Oct. 13 to 17, 1986.
11. Short, J. L., Selection, application and optimization of hollow fibre UF membranes, Liltr. Sep., 5, 410,
1982.
12. Prigent, Y. and Franco, A., Etude d ’un fermenteur lactique a membranes: extraction par UF et ED du
lactate de sodium contenu dans le jus de fermentation de lactose supplemente, Le Lait, 64, 217, 1984.
 171

13. Hayward, M. F., Ultrafiltration and reverse osmosis: a survey of industrial applications, in Proc. 3rd
World Filtration Congr., Croydon, England, September 13 to 17, 1982, 572.
14. Pichon, M., Muratore, E., and Morzie, P., Elimination de matieres pollutantes par ultrafiltrationd ’ef­
fluents papetiers, Proc. 1st World Filtration Congr., Paris, May 14 to 17, 1974, F49-1.
 173

Chapter 10

MICROFILTRATION

Siegfried Ripperger

TABLE OF CONTENTS

I. Introduction...................................................................................................................... 174

II. Membranes for Microfiltration......................................................................................174

III. Crossflow Filtration Principles...................................................................................... 176

IV. Basic Plant Design.......................................................................................................... 180

V. Methods to Increase the Flux byCrossflow Microfiltration...................................... 184

VI. Economic Factors in Crossflow M icrofiltration......................................................... 187

List of Sym bols.......................................................................................................................189

References.................................................................................................................................... 189
174 Water, Wastewater, and Sludge Filtration

I. INTRODUCTION

Microfiltration (MF) is a pressure-driven membrane process for the separation of particles,

microorganisms, large molecules, and emulsion droplets. The filter medium is a microporous
membrane with a separation limit in the range of 0.02 to 10 ixm. 1 Therefore, MF is placed
between ultrafiltration and coarse filtration, which is not a membrane operation (Figure 1).
MF becomes an interesting separation technique since reliable separation of the above-
mentioned matter is difficult and not economical by the other separation methods (e.g.,
sedimentation, centrifugation, depth filtration). Hence, MF finds an important place in water
and wastewater treatment for removal of dissolved materials and colloidal particles. In many
instances, MF is used in combination with other separation processes.
In crossflow microfiltration (CMF), the fluid to be filtered flows parallel to the membrane
surface, i.e., crossflow with respect to the flow of the permeate. The crossflow reduces the
formation of a filter cake and keeps it at a low level, thus ensuring constant filter conditions
and, hence, permeate flow. This process is sometimes called tangential or dynamic filtration.
CMF is used for the production of pure liquids, the concentration of suspensions in order
to recover valuable products, and the regeneration of process liquids. In some applications,
CMF would lead to considerable cost savings, because recycling of water and/or other
valuable products would be made possible by this process.

II. MEMBRANES FOR MICROFILTRATION

A membrane is an intervening phase separating two phases and/or acting as an active or

passive barrier to the transport of matter between phases adjacent to it . 1 In MF, the membrane
has a microporous structure and separates particles according to the size of pores, from a
liquid or a gas phase. The separation is based on the sieve effect, i.e., the fact that the
separation effect is limited to the outer surface of the membrane. The microporous structure
of the membrane should have a narrow pore-size distribution in order to ensure the quantitive
retention of particles of a given size and type. Another important feature is the high porosity
of membranes which leads to high filtration rates at small differences in hydrostatic pressure.
Further requirements of membranes are high strength as well as chemical and thermal
stability.
The properties of microporous membranes are largely determined by the production process
and the materials used. The commonly used types are the polymer membranes. These
membranes, while fulfilling the requirements mentioned above, are also favorable in terms
of cost/benefit ratio. MF membranes based on inorganic materials have recently been put
on the market and are now being tested in specific applications. The porosity of these
membranes is much lower than that of polymer membranes, but their thermal stability enables
them to be used at high temperatures. Table 1 shows the major processes used in the
production of microporous membranes. Today, the majority of microporous membranes are
made by a phase inversion process where a homogeneous liquid phase consisting of a polymer
and a solvent is converted into a two-phase system, with the polymer being precipitated as
a solid phase and the liquid phase forming the pore system. The conversion of the homo­
geneous solution into a two-phase mixture may be brought about by different ways (Table
1). In the Accurel® process (Accurel is a trademark of AKZO), for instance, the polymer
is precipitated by lowering the temperature immediately after spinning the solution. Thus,
it is possible to produce self-supporting tubular or capillary membranes having a defined
small-pore diameter. The porosity of such membranes ranges between 60 to 80%.
Different methods are used to characterize the structure and separation limit of the mem ­
branes. Mostly, the separation limit is determined by a contamination test, using a specific
test organism. If at a typical contamination of the membrane (number of test organisms per
 0.01 0.1 1 10 100 pm

FIGURE 1. The separation range of microfiltration. (Courtesy of

AKZO, Wuppertal, West Germany.)

Table 1
MAJOR PRODUCTION PROCESSES FOR PRODUCTION OF
MICROPOROUS MEMBRANE STRUCTURES2

Process Material used

Phase inversion by Polymers

Change of temperature (e.g., polypropylene, polyamide,)
Solvent evaporation (e.g., polyamide, cellulose acetate)
Addition of precipitant (e.g., nitrocellulose, polysulfone)
Molding and sintering of fine-grained pow­ Polymers (e.g., polyethylene, polytetrafluo-
der roethylene); metals, metal oxides, ce ­
ramics
Irradiation and etching process (radioactive Polymers (e.g., polycarbonate, polyester)
radiation with subsequent etching of the
tracks formed)
Stretching of sheets of partially crystalline Polymers (e.g., polytetrafluoroethylene,
polymers polyethylene)
176 Water, Wastewater, and Sludge Filtration

Table 2
METHODS TO CHARACTERIZE MICROPOROUS MEMBRANE
STRUCTURES2

Method Characteristics derived

Electron microscopy Porosity, characterization of pore size, pore

size distribution
Liquid or gas permeation Hydrodynamic resistance, nominal pore di­
ameter
Filtration of test suspensions (e.g., with test Nominal pore diameter, sieve analysis curve
organisms, latex particles, test dust)
Bubble point method with gradually in ­ Maximum pore diameter, pore size distri­
creased pressure bution, nominal pore diameter
Porosimetry (e.g., mercury intrusion method) Pore size distribution, porosity
Volume determination and weighing Apparent membrane density, porosity
BET method Inner surface area

unit area) all organisms are held back, this will be the specific separation limit for the test
organism concerned. Membranes for sterile filtration have a nominal pore size of 0.2 |xm
which is determined with the bacterium Pseudomonas diminuta, using a typical contamination
of 108 to 1010 bacteria per cm2. Other definitions of separation limits and nominal pore sizes
are possible, although not generally used. The most common methods of characterizing
microporous membrane structures are listed in Table 2. The results obtained by the various
methods are not necessarily the same, although large variations are improbable. Pore size
distribution and mean pore diameters of porous structures may be determined by evaluation
of electron microscopical pictures. Another way to determine a hydraulic pore diameter is
by evaluation of flow rate measurements for liquids or gases. Polymer membranes are often
tested for maximum pore diameter by the bubble point method. This method is based on
the effect of surface forces preventing the penetration of gas within a given range in the
presence of a pressure difference.

III. THE CROSSFLOW FILTRATION PRINCIPLE

The equipment needed for filtration, the membrane area requirements, etc. are determined
from the filtration time and the other related parameters involved. Different requirements
have led to the development of different processes, equipment, and membrane units (modules,
filter cartridges).
Small quantities are filtered with disc filter or filter cartridges with a filter area from a
few cm 2 to 3 m2. The filtration method in this case is static or dead-end filtration. The
retained particles build up a layer on the membrane surface whose increasing thickness
rapidly reduces the permeate flow if the pressure difference remains constant. Depending
on the nature and quantity of the retained substances, the filtration process must be interrupted
at relatively short periods to clean or replace the membrane. Such a working mode has
disadvantages in industrial applications, where often considerable amounts of suspensions
with high solid contents must be treated. These industrial conditions gave rise to the de­
velopment of the CMF.
In crossflow-filtration, the fluid to be filtered flows parallel to the surface of the filter
medium. This parallel flow generates shearing forces and/or turbulences over the filter
medium and limits the thickness of the filter cake. In most cases, the formation of a thin
layer containing retained components (Figure 2) on the membrane surface cannot be com ­
pletely prevented. During the steady state case (i.e., the normal operation case), the thickness
of this layer becomes constant. For complete retention, the transport rate to the membrane
 177

CONCENTRATION OF
RETAINED SU BS TA N CES

FIGURE 2. Concentration profile at the membrane. (Courtesy of AKZO, Wuppertal, West

Germany.)

must then be equal to the one flowing back to the bulk stream. The mass balance will yield
the following equation:

CBJ + mback = 0 (1)

Here, cB J represents the transport of retained material to the membrane and mback, the back
transport away from the membrane into the bulk stream. This equation shows that the flux
can belimited by the back transport mechanism.
In the case of the separation of dissolvedsubstances as in ultrafiltration and reverse
osmosis, the back transport can be described with the laws of diffusion. A gradient in the
retained component concentration is established perpendicular to the membrane. This is the
phenomenon known as concentration polarization. The back transport can then be described
as:

. _ dc (2)
nw “ D dy

Combining Equation 2 with Equation 1 and integration, we obtain:

D (3)
J = — ln(cL/cB)
8c

This equation provides a relationship between the boundary layer thickness 8C and the
concentration of retained matter on the membrane surface cL. D/8Ccan be replaced by the
mass transfer coefficient K. Correlations for the value of K for laminar and turbulent flow
regimes and various flow channel geometries are available in the literature.3,4 The concen­
tration can be polarized to an extent that a gel slime is formed on the membrane surface
which considerably reduces the flux. The saturated concentration for gel formation depends
on the chemical and physical properties of the solution.
178 Water, Wastewater, and Sludge Filtration

F
y

MEMBRANE
J

FIGURE 3. Schematic representation of drag forces which act upon

deposited particles. (Courtesy of AKZO, Wuppertal, West Germany.)

In contrast to ultrafiltration and reverse osmosis, the retained matter in MF is mostly

colloids and particulates. It has been shown that the back transport from the membrane
surface into the bulk stream is driven, in this case, by some forces other than those for the
concentration gradient. 5,6
Porter5 discussed the back transport of particles using the “ tubular pinch effect.” This
effect manifests itself as a radial migration of particles away from the tube wall to an
equilibrium at an eccentric radial position. The radial migration arises from a slip-spin-force
similar to the Magnus force.
Green and Belfort7 made use of this phenomenon of radial migration of particles and
combined it with the theory of standard filtration. Experimental and theoretical studies arrive
at the following identical form for the migration velocity:

(4)

d
Where f(—) is a function of the radial position of the particles in the tubular channel. Different
dt
values for the constant C and the exponent a are reported in the literature depending on
assumptions, in the case of a theoretical calculation, or the kind of suspensions, in the case
of experimental results. The range of values of C and a are

C = 0.2 ­ 1 .8

a - 2.8 - 4.0

The experimental work was mostly limited to a laminar flow in the channel.
Ebner8 calculated critical conditions of particle depositions considering the drag forces
acting on a single particle (Figure 3). The drag forces are the result of the tangential flow
and permeate flow. For the description of the criteria for deposition he used a frictional
coefficient \x. In the case of globular particles and the consideration of coulomb friction
between particles and the membrane, the deposition does not take place if;

w > |X •J (5)

Here w is the mean tangential velocity acting on the particle. Figure 3 illustrates that this
velocity depends on the particle size. The requirements of this model have been confirmed
from the experimental results obtained . 6,8,9
 179

Schock6 was able to describe his experimental results under laminar and turbulent flow
conditions close to the particles, and established the following relationship from his exper­
imental results.

(6 )

Here, the viscosity and density are those of the suspension. He used the 50 percentile value
of the particle size distribution as a representative particle diameter.
Fischer and Raasch9 analyzed particle size distribution of the deposited solids at the
membrane. They found that the particle size distribution is shifted towards smaller particle
sizes with an increase of the tangential velocity. These experimental results are in accordance
with the conclusions drawn from the model (Figure 3). Fane 10 showed the influence of the
particle size on the flux. He also found that after a certain size, the flux will increase with
the particle size.
The flux is a result of the back transport and hydraulic resistance of the membrane RM
and resistance of the particle layer RL. The resulting flux is given by:

The hydraulic resistance RL of a filter cake layer consisting of particles can be deduced from
the Carmen-Kozeny equation:

If the layer is compressible, the porosity of the layer will decrease with pressure and the
hydraulic resistance of the particle layer will increase. Due to the retention of fine particles
on microporous membranes, the hydraulic resistance of the particle layer is often much
higher than the hydraulic resistance of the membrane itself. Equations 4 ,6 , and 8 show the
important role of the particle size.
The filtration models are necessary for the understanding of the basic mechanism of CMF,
but it is not possible to use them for the calculation of the flux of a crossflow unit because
the values of the coefficients appearing in these equations are not known in advance.
Moreover, the filtration models described above do not consider all the mechanisms re ­
sponsible for the retention of particles, for example, adsorption phenomena, chemical effects,
rejection characteristics of the particle layer, etc. Therefore, it is necessary, for the practical
application of CMF, to conduct tests with laboratory- or pilot-plants in order to find out the
steady state flux for different operating parameters.
One target of the pilot study is to find out the influence of the following operating
parameters: transmembrane pressure difference, tangential velocity, temperature, and con ­
centration factor.
The transmembrane pressure influences the flux together with the back transport conditions
and the thickness of the layer. This is one reason why the flux increase is not proportional
to the pressure difference (Figure 4). Flux will increase up to a certain pressure and then
remain constant. At this point, a close-packed layer is formed at the membrane. Any further
increase in pressure difference will result only in a temporary increase of convective transport
to the membrane which will not be balanced by the back transport mechanism. As a result,
there will be a net accumulation of colloids and particles on the membrane, resulting in a
180 Water, Wastewater, and Sludge Filtration

DIFFERENCE A p tM

FIGURE 4. Effect of transmembrane pressure and tangential velocity

on flux. (Courtesy of AKZO, Wuppertal, West Germany.)

thicker or more compact layer. The hydraulic resistance of the layer, therefore, increases
and the flux is commensurate with the fixed back transport of particles to the bulk stream.
The tangential velocity influences the flow conditions of the stream across the membrane.
A high velocity generates high forces on the particles in a parallel direction to the membrane.
In general, the higher the tangential velocity, the higher the flux. The consequence of a
higher flux is a higher pressure drop in the direction of the tangential velocity over the
membrane and a higher energy consumption. An optimal value, therefore, has to be found
if the CMF plant runs are to be reliable at minimum costs.
In most cases the flux will increase with the temperature due to a lower viscosity. The
influence of temperature can be high if components are precipitated or dissolve with tem ­
perature change. In many processes, the operating temperature is set by the product or the
process conditions; in others there is some flexibility in choosing an optimum temperature.
The concentration of retained particles in the bulk stream influences the mass transport to
and away from the membrane and the flow conditions. Therefore, the influence of the
concentration factor on the flux is rather complex. A certain value for the concentration
factor is assumed depending on the situation.

IV. BASIC PLANT DESIGN

CMF is the unit operation of a process which can include several posttreatment and
pretreatment procedures. In general, a CMF plant consists of several membrane modules
which are connected in a series and/or parallel. Different module configurations have been
developed. At this time, tubular and capillary configurations appear to be the most suitable
configurations for CMF. The feed solution flows through the inside of the membrane tubes
or capillaries. The permeate solution passes through the membrane and is collected in a
tubular casing. In this case, the flow conditions are defined and the biggest part of the energy
input serves to generate shearing forces on the membrane surface. Normally, a bundle of
capillary or tubular membranes is tightly scaled at both ends of the tubular casing. Tubular
membranes can either be self-supporting or are equipped with external supporting materials
for added mechanical strength. Generally they are produced with an inner diameter of 4 to
20 mm. The advantages of the tubular design are the ability to process high concentrations
of suspended solids and high viscous materials. Capillary membranes (hollow-fiber) with
inner diameters ranging from 0.5 to 2 mm allow a high packing density and are self-
supporting. Commercially available configurations with flat membranes include plate and
 181

STATIC PRESSURE p
CONCENTRATION OF RETAINED SUBSTANCES c
TANGENTIAL VELOCITY W

Pi

"1 ?2
-j c 2
Cl I—
Wi
W;
LENGTH
PERMEATE

/
CONCENTRATE
MEMBRANE MODULE

FIGURE 5. Profiles of static pressure p, concentration of retained sub­

stances c, and tangential velocity w across the length of a membrane
module. (Courtesy of AKZO, Wuppertal, West Germany.)

frame systems and spiral-wound systems. In the plate and frame system, flat sheet membranes
are placed between plates which form flow channels with heights of 0.5 to 1.5 mm across
the membrane. A large number of membranes and supports are sandwiched together and
form the module.
The spiral-wound configuration is obtained using sheet membranes wound around a central
permeate collection tube. The specific packing density depends on the size of the spacers
which form the flow channel. This configuration is not often used in CMF because the
possibility of blockage of part of the flow channel is high.
The static pressure within the module will decrease from a higher level px at the entrance
to a lower level p2 at the exit. The concentration of the retained substances will continually
increase (Figure 5). Since the single passage of the module arrangement will generally not
be sufficient for reaching the desired degree of concentration, the suspension is recirculated.
Since the filtrate is continuously removed, the tangential velocity will be lower at the exit
than at the entrance of the module. However, this difference is usually small in most of the
crossflow systems. The fraction of filtrate in a single pass is usually lower than 5%.
Figure 6 shows the basic flow chart of a CMF plant. The characteristic of the circulation
pump and the valve behind the module influences the flow conditions in the module and
the transmembrane pressure difference. The mean transmembrane pressure difference is
normally defined as:

(9)

The diameter of pipes in the recirculation loop is chosen in such a way that the flow velocity
in the pipe system is much lower than that of the tubular membranes. In such cases, the
transtube pressure drop in the module A ptx = p 2 — p 2 is the most important one in the
circulation loop. For modules with tubular and capillary membranes, this pressure drop can
be described with the following equation:
182 Water, Wastewater, and Sludge Filtration

PERMEATE

IL®
RECIRCULATION LOOP
—M--­ CONCENTRATE
r"

FEED
O
FEED RECIRCULATION M EM BRANE
PUMP PUMP MODULES

FIGURE 6. The basic flow chart of a CMF plant. (Courtesy of AKZO, Wuppertal, West Germany.)

PERMEATE

FIGURE 7. Flow chart: batch system. (Courtesy of AKZO, Wuppertal,

West Germany.)

AptT = ( k + ( 10)

Here k is a specific constant of module and reflects the pressure drop at the module entrance
and exit. Values for k are found to be in the range of 1 to 4, depending on the module
construction; X is the well-known friction factor for pipe systems. Correlations for \ are
available in the literature. It is also shown that the suction through the wall of membrane
tubes influences the value of A. . 11,12 The energy demand P for the circulation pump depends
on the circulated volumetric flow rate V ^ p , the pressure drop in the circulating loop ApLoop,
and the pump efficiency r]p.

P — A Ploojj * VL ( 11)
Vp

The plant design of a crossflow filtration system will vary according to the operation
mode. There are three basic operation modes: discontinuous operation (batch-processing
mode), one-stage continuous operation, and multi stage continuous operation.
The design of batch processing mode is simple (Figure 7). The initial feed volume is
 183

0 2 5 10 20 CONCENTRATION
| “ I I I I FACTOR
0 50 80 90 95 LIQUID REM OVAL
[% OF ORIGINAL
VO LU M E]

FIGURE 8. Flux-concentration relationship for some real suspensions and emulsions. (Courtesy of AKZO,
Wuppertal, West Germany.)

pumped continuously through the membrane modules to produce permeate. This operation
is characterized by the change of process conditions. These changes are mainly caused by
the gradual increase of the concentration of retained substances in the circulating liquid.
This happens if no concentrate is drawn off. Increasing concentration will change the Theo­
logical properties of the suspension; the pressure drop within the module will increase and
the circulating flow will be reduced if the centrifugal pump is not automatically adjusted.
The filtration is stopped when the predetermined degree of concentration has been reached
and the concentrated liquid is drawn off. The pressure drop in the circulation loop which
determines the energy demand P (Equation 11) will be the actual inlet pressure to the module
since the lowest pressure in the loop will be at atmospheric pressure. In most cases, the
flux decreases with the concentration of retained particulates. The flux vs. concentration
factor curves for different suspensions and emulsions are illustrated in Figure 8 . A disad ­
vantage of a batch system is that a storage tank for the whole volume is necessary. Therefore,
this mode is used for small volume applications.
The elementary feed and bleed system, single stage (Figure 6 ), serves for continuous
operation. This system differs from simple batch systems in that a working storage tank is
not necessary. The streams to and up from the units are adjusted in such a way that the
filtration takes place with a predetermined concentration factor. The process operates con ­
tinuously at the final concentration value, so that the flux is, in view of the flux-concentration
relationship, at its lowest value. In this case, increased membrane surfaces are required for
any given permeate rate, compared to the batch process. The pressure drop in the recirculation
loop is lower in this case, because the exit pressure is above atmospheric pressure.
If the influence of the concentration is significant, a multistage-design will give a better
result (Figure 9). In this system, only the last stage runs at final concentration.
The advantages of a design for continuous operations are that the hold-up volume is
minimal and that the required final concentration is achieved shortly after starting the op ­
eration. Continuous operation is desired for large-scale industrial CMF plants.
184 Water, Wastewater, and Sludge Filtration

p e rm e a te perm eate

I— 1. S T A G E ----------------------------------------------—
1 2. S T A G E ----------------------------------------------------- 1

FIGURE 9. Two-stage CMF plant. (Courtesy of AKZO, Wuppertal, West Germany.)

The service conditions will dictate the material of construction. Pressure rating, thermal
expansion and contraction, temperature stability, and mechanical strength are some of the
factors to be chosen to insure resistance against the chemicals connected with the application.
The devices of a CMF-plant in some applications, for example, water and wastewater
treatment, are sensitive to metallic corrosion. The most effective way to minimize corrosion
is to limit the use of metal. Engineering plastics and plastic composites should be used in
such cases. Polypropylene covers a wide range of requirements. In some cases PVC can be
used.
Some applications, for example, in the food, beverage, and pharmaceutical industry and
in biotechnology, include a hot sterilization or a hot cleaning step. For such applications,
stainless steel is used as the construction material.

V. M E T H O D S TO IN C R E A S E T H E F L U X B Y C R O S S F L O W
M IC R O FILTR A TIO N

The formation of a filter cake on the membrane surface is generally not required to keep
up the defined separation function. This is guaranteed by the membrane itself. Some of the
techniques described in the literature, to reduce the build-up of a layer on the membrane
are the periodic backflush with filtrate , 13 20 the use of abrasives, 20,21 pulsated flows , 20 and
electrophoresis .6
At present, only the periodical backflush technique is used in industrial application with
some regularity. The other techniques are still under development. The use of such methods
includes additional operating parameters which must be considered in the experiments.
Self-supported tubular or capillary membranes allow pressure differences from inside to
outside and vice-versa. Thus, periodic backflushing of the membrane in the opposite direction
of the permeate flow is possible. Under ideal circumstances, the backflush removes the layer
from the membrane in a short time. In other cases, it prevents compaction under the filtration
pressure and reduces the thickness of this layer. A backflush is a negative flux operation
lasting for a very short time, but it often serves for a high integrated filter flux. Usually,
the crossflow process is not stopped during backflushing. In this case, the backflush pressure
difference changes over the module length. In order to equalize the backflush effect, the
direction of the tangential velocity can be changed from time to time. The most important
parameters for incorporating this backflush technique are the time between two backflushes
 185

AtF [min]

FIGURE 10. Influence of a periodical backflush on the effective flux (specific backflush volume 0.5
L/m2). (Courtesy of AKZO, Wuppertal, West Germany.)

(backflush interval or filtration period AtF), the backflush pressure difference Apb, and the
duration of backflush Atb.
The combination of Apb and Atb with filtration conditions results in a certain specific
backflush volume. The practice shows that optimal values for specific backflush volume are
in the range of 0.2 L/m 2 to 1 L/m 2 for one backflush. The values are connected to backflush
times Atb from 1 to 4 sec. To find out the optimal parameters, it is necessary to run trials
to optimize the process.
The effective flux (Jeff) of the filtration with a periodic backflush is given by:

fr a if
Jo
= ° A,. + A,b
J(t)dt - Vb

<12)

The average flux J can be given by:

J(t)dt

Figure 10 presents the relationship between Jeff/J vs. filtration interval for different J
values. This graph is established for the situation where specific backflush volume is 0.5
L/m 2 (the value used in practical applications). Where the average flux is 100 L/m 2 and
backflush interval is 4 min, then the influence of the backflush is less than 10%.
The effect of periodical backflush with permeate depends on the particular application.
An optimization of the backflush parameters is necessary for each case.
Figure 11 shows the flow chart of a CMF plant with periodic backwashing. A small
storage tank must be provided from which filtrate can be backflushed by compressed air.
186 Water, Wastewater, and Sludge Filtration

Feed Circulation Membrane

pump pump modules

FIGURE 11. Flow chart: CMF plant with a system for periodical backflush. (Courtesy of AKZO, Wuppertal,
West Germany.)

FIGURE 12. Flow chart: Two-stage CMF plant with a backflush system. (Courtesy of AKZO, Wuppertal, West
Germany.)

Interval between two backflushes and the volume of the flushed liquid can be controlled
automatically by a timer and valves, respectively.
Large CMF plants contain several CMF units (blocks) parallel or in series. Each unit has
one pump and pipe system. The periodic backflush can be realized with a pump in com ­
bination with an air cushion. Due to the large number of modules, a quasi-continuous
operation of the pump is possible, if each module is separately backflushed (Figure 12).
Other methods to increase the flux include: a pretreatment of the suspension (e.g., the
 187

FIGURE 13. Flow chart: CMF plant with a cleaning-in-place (CIP) unit. (Courtesy of AKZO, Wuppertal,
West Germany.)

use of flocculants, presettling), and the use of filter aids .20 22 Filter aids can be applied either
as a precoat or as a mixture in the feed stream.
The crossflow technique, by itself or in combination with other techniques, cannot com ­
pletely prevent fouling. Most crossflow filter systems must be cleaned with some regularity.
The cleaning procedure and frequency depend on the product to be filtered and the resistance
of the membrane module. Most cleaning procedures are a combination of hydraulic cleaning
and chemical cleaning. The simplest hydraulic cleaning technique is a water rinse or a
backwash with filtered water. For chemical cleaning, different cleaning agents are used.
Their functions are mainly dispersion and/or solubilization. Cleaning agents, based on dis­
persion, separate deposits from the membrane and disperse them into smaller sized ones.
Detergents are the most common dispersing agents. In the food industry, enzyme detergents
with protease are used for hydrolyzing protein deposits.
In the case of solubilization, the deposits are dissolved by physical or chemical means.
The solubility of the deposits depends on the pH value and temperature. The chemical
inertness of the membranes used is a distinct advantage when it comes to cleaning. In the
case of polypropylene membranes, for example, nonoxidizing acids and caustic solutions
within the pH range of 1 to 14 can be used.
Cleaning products used for this purpose are generally a mixture of several purifying agents.
In some instances, it is economical to employ a hydraulic cleaning method with a shorter
detention time cycle and employ the chemical method after a longer time interval.
The cleaning procedure is mostly included in the operating process of the plant. Figure
13 shows a CMF plant with an integrated CIP unit (CIP = cleaning in place). In many
cases, a good cleaning result can be reached by flushing the cleaning solution from the
permeate side through the membrane as shown in Figure 13. This method minimizes the
demand on the cleaning solution.

VI. ECONOMIC FACTORS IN CROSSFLOW MICROFILTRATION

The principal factors to be considered in an economical analysis are

• The capital costs of the system depending on the flux

• Complexity of the system
• Required components other than membranes (pumps, measurement equipment etc.)
188 Water, Wastewater, and Sludge Filtration

TAN G EN TIAL VELOCITY [ m / s e c ]

FIGURE 14. Variation of cost items and total processing costs

with tangential velocity. (Courtesy of AKZO, Wuppertal, West
Germany.)

• The replacement cost of the membrane

• Membrane lifetime
• Energy requirements for pumps
• Labor costs (including time for membrane cleaning and membrane replacement)

Costs for the treatment or disposal of concentrate and permeate (if not reused) must also
be taken into consideration.
The required membrane surface depends on the flux. The membrane replacement costs
are proportional to the membrane surface. Due to the modular character of membrane plants,
the reduction of capital costs by reducing the number of modules is very small.* The flux
is highly influenced by the product and the tangential velocity. The product has to be filtered
as mentioned in Section II. In an optimization process, the optimal velocity can be found
out by comparing the increase of energy costs with the decrease of membrane and capital
costs.
Figure 14 shows typical variations of cost items with tangential velocity. The following
dependents are considered: flux increases with w in the order of w 1 26 (as in Equation 6 );
the pressure drop increases with w in the order of w2 (turbulent flow); these cause the energy
demand to increase with w in the order of w 3 (Equation 11). Capital costs can be reduced
by the use of large membrane modules . 16 Figure 14 shows that an optimal tangential velocity
exists. Other costs, for example, cost of operation and maintenance, labor, etc., do not vary
with the tangential velocity. The figure also shows that, in general, energy demand and
membrane replacement are the most important cost items. The amortized capital costs are
significant, but in general are not the controlling cost factor. The amortization schedule for
industrial applications is in the range of 5 to 10 years.
At the present time, the total costs for the industrial application of a CMF step are often
in the range of 0.3 to 10 DM/m 3 permeate. The cost assumes the life period of the membrane
to be in the range of 1 to 3 years. This wide range is a result of the different kind of products
and variation of the requirements encountered.

* Therefore the capital units are nearly proportional to the membrane surface.
 189

LIST OF SYMBOLS

A Area, m 2
c Concentration, kg m - 3
CB Concentration in bulk, kg m ~ 3
CL Concentration in particle layer, kg m ~ 3
d Inner diameter of membrane tubes, m
D Diffusity, m 2 s _1
dP Particle diameter, m
J Filtration flux, m s - 1
k Constant, dimensionless
K Mass transfer coefficient, m s - 1
L Module length, m
m Specific mass flow rate, kg m " 2 s _1
P Power, W
APtM Transmembrane pressure drop, Pa
AptX Transtube pressure drop, Pa
Pi Module entrance pressure, Pa
p2 Module exit pressure, Pa
Pf Permeate pressure, Pa
R m Hydraulic resistance of membrane, m - 1
R l Hydraulic resistance of particle layer, m r 1
t Time, s
AtF Filtration interval, s
Atb Backflush interval, s
vb Specific backflush volume, m 3
V Particle migration velocity, m s - 1
w Tangential velocity (mean value across the flow channel), m s
y Distance from membrane, m
8 Boundary layer thickness, m
sc Concentration boundary layer thickness, m
sL Particle layer thickness, m
e Void space, dimensionless
'O Dynamic viscosity, Pa s
V Kinematic viscosity, m 2 s - 1
s Density, kg m ~ 3
P' Coulomb friction factor, dimensionless
X Friction factor, dimensionless
Re Reynolds number, dimensionless

REFERENCES

1. Terminology for Pressure Driven Membrane Operations, Issued by European Society of Membrane Science
and Technology, Lund University, Desalination, 68, 77, 1988.
2. Ripperger, S. and Schulz, G., Microporous membranes in biotechnical application, Bioprocess Engi­
neering, 1, 43, 1986.
3. Blatt, W. F., David, A., Michaels, A. S., and Nelson, L., Solute polarization and cake formation in
membrane ultrafiltration, in Membrane Science and Technology, Flinn, J. E., Ed., Plenum Press, New
York, 1970.
190 Water, Wastewater, and Sludge FUtration

4. Rautenbach, R. and Albrecht, R., Membrantrennverfahren, Ultrafiltration and Umkehrosmose, Otto

Salle Verlag, Frankfurt, 1981.
5. Porter, M. C., Concentration polarization with membrane ultrafiltration, Ind. Eng. Chem. Prod. Res., 11,
234, 1972.
6. Schock, G., Mikrofiltration an iiberstroomten Membranen, dissertation, Technical High School— Aachen,
Aachen, West Germany, 1985.
7. Green, G. and Belfort, G., Fouling of ultrafiltration membranes: lateral migration and the particle trajectory
model, Desalination, 35, 129, 1980.
8. Ebner, H., Die kontinuierliche mikrofiltration von essig, Chem. Ing. Techn., 53, 25, 1981.
9. Fischer, E. and Raasch, J., Model tests of the particle deposition at the filter medium in crossflow
filtration, 4th World Filtration Congr., Ostende, 1986.
10. Fane, A. G., Ultrafiltration of suspensions, J. Membr. Sci., 20, 245, 1984.
11. Belfort, G. and Nagta, N., Fluid mechanics and crossflow filtration, Desalination, 53, 57, 1985.
12. Eickermann, U. and Werner, U., Laminare Stromung in permeablen Leitungen; turbulente Stromung in
Rohren mit permeablen Wanden, Chem. Ing. Techn., 58, S348, 1986.
13. Klein, W. and Holz, W., Crossflow microfiltration in the process industry, Chemical Eng., 10, 369, 1982.
14. Paul, H., Two phase ultrafiltration as an economical unit operation for recycling of detergents, Desalination,
47, 293, 1983.
15. Schmitz, F. J. and Dau, H., Die Crossflow-Mikrofiltration, Weinwirtschaft-Technik, 1984, 9.
16. van Gassel, T. J. and Ripperger, S., Crossflow microfiltration in the process industry, Desalination, 53,
373, 1985.
17. Ostermann, A. E., Advances in crossflow-microfiltration, Proc. 4th World Filtration Congr., Ostende,
1986.
18. Kroner, K. H. and Kula, M. R., Crossflow filtration in the downstream processing of enzymes, Process
Biochem., 19, 4, 1984.
19. Milisic, V. and Ben Aim, R., Developing a better understanding of crossflow microfiltration, Filtr. Sep.,
1, 28, 1986.
20. Milisic, V. and Bersillon, J. L., Anti-fouling techniques in crossflow microfiltration, Proc. 4th World
Filtration Congr., Ostende, 1986.
21. Bixler, H. J. and Rappe, G. C., U.S. Patent 3, 541, 006 November 1970.
22. Rahse, W. and Carduck, F. J., Mikrofiltration von Fermenterbriihen, Chem. Ing. Tech., 57, S747, 1985.
 191

Chapter 11

ELECTRODIALYSIS

Remy Audinos and Saravanamuthu Vigneswaran

TABLE OF CONTENTS

I. Introduction and Basic Principle..................................................................................192

II. Description of Electrodialysis Stack.............................................................................192

A. Assembling Cell Units ..................................................................................... 192
B. Spacer G askets...................................................................................................194
C. Solution Circulation in the Unit C e ll............................................................. 194
D. Flow System of the S tack................................................................................ 195
1. Cocurrent or Counter-Current Flow .................................................... 195
2. Internal or External Staging..................................................................196

III. Selectivity....................................................................................................................... 199

A. Description of Ion-Permeable Membranes (ipm)...........................................199
B. Models for Ion-Permeable M em branes.......................................................... 199
C. Transport M echanism s..................................................................................... 199
D. Transport Num ber............................................................................................. 200

IV. Perm eability....................................................................................................................202

A. Qualitative Considerations................................................................................202
1. Concentration Polarization...................................................................202
2. Critical C urrent......................................................................................202
3. Water Splitting....................................................................................... 202
4. Electrometathesis.................................................................................. 204
B. Ion Transport through Membrane U n it..........................................................204
1. Membrane U n it......................................................................................204
2. Basic Equations......................................................................................204
3. The Steady State....................................................................................206
4. The Critical C u rren t............................................................................. 207
C. Transport and Efficiency in a Unit C e ll........................................................ 207
1. Ionic Fluxes.............................................................................................207
2. Water Transfer....................................................................................... 208
3. Electrode Reactions............................................................................... 209
D. Current Efficiency of the Stack....................................................................... 210
1. The Stack R esistance............................................................................210
2. Concentration Potential........................................................................ 210
3. Electrodes and ManifoldLeakage.........................................................211
4. Overall Efficiency.................................................................................211

V. Process and Equipment D esig n ................................................................................... 211

A. Material B alance............................................................................................... 211
B. Demineralization F actor................................................................................... 213
C. Power Requirement............................................................................................214
192 Water, Wastewater, and Sludge Filtration

D. Instrumentation.................................................................................................. 215
1. S ensors................................................................................................... 215
2. Control................................................................................................... 215

VI. Operating C onditions...................................................................................................216

A. Pre- and Posttreatm ent.....................................................................................216
B. Current Reversal orPulsing..............................................................................216
C. Cleaning.............................................................................................................. 217

VII. Applications .................................................................................................................. 218

VIII. Conclusion..................................................................................................................... 218

Notations..................................................................................................................................... 218

References....................................................................................................................................222

I. INTRODUCTION AND BASIC PRINCIPLE

Electrodialysis (ED) is a physical method for extracting or concentrating ions in solutions

by migration, under the influence of an electric field, through anion-selective and cation-
selective membranes, without using the products of the electrode reactions . 1 As in every
membrane process, electrodialysis basically results from the interaction of three scientific
domains: membranes, energy, and medium (M E M); it is characterized by two main
parameters, permeability and selectivity.
ED is an electromembrane process in which ions are moved from a compartment to another
by migration through semipermeable membranes. 2 A direct-current electrical field inhibits
the ions from passing through appropriate membranes which allow neither diffusion nor
convection. Generally, membranes are in the form of flat sheets (Figure 1).
An alternative anion-permeable membrane (apm) and cation-permeable membrane (cpm)
are placed perpendicularly to a monodirectional electrical field. The anion-selective mem ­
brane permits only the anions to pass through, and the cation-selective membrane allows
only the migration of cations. The transport of anions and cations through the appropriate
selective membrane results in ion depletion in one cell and in ion concentration in the
adjacent cell.
This arrangement, consisting of an anion-permeable membrane, an ion-depleted cell, a
cation-permeable membrane, and an ion-concentrated cell, is called the “ unit cell” or cell
pair.
The solution exiting from the ion-depleted cell is the desalted or demineralized water or
dilute solution, and the solution leaving the ion-concentrated cell is the brine or concentrated
solution.
The consequence is that ED is both a desalting and a concentrating process.

II. DESCRIPTION OF ELECTRODIALYSIS STACK

A. Assembling Cell Units

In an ED stack, more than 100 but less than 1000 unit cells are placed between two
electrodes, resulting in hundreds of feed streams, dilute streams, and concentrate streams.
 193

mpc mpa mpc mpa

I I

0 ~
'^"O <^"0
43
i

FIGURE 1. Unit cell (b = concentrating compartment; d = diluting compartment; mpa = membrane permeable
to anions; mpc = membrane permeable to cations; © = anion; © = cation; □ = anode; E = cathode).

FIGURE 2. Electrodialysis stack without clamping device and parallel flow (ab = anode block; ar = anode
rinse solution; b = concentrate stream; cb = cathode block; cr = cathode rinse solution; d = dilute solution; ea
= anode; ec = cathode; mpa = membrane permeable to anions; mpc = membrane permeable to cations; sg =
spacer gasket).

The unit cells and electrodes can be arranged either in a horizontal or in a vertical direction.
This entire unit is called a stack or module and is commercially available (Figure 2).
Feed is introduced into alternate membrane compartments. Under the influence of an
applied DC voltage across the membrane assembly, the anions are attracted towards the
194 Water, Wastewater, and Sludge Filtration

anode, and the cations are attracted towards the cathode. But the anions are prevented from
reaching the anode by cation-permeable membranes, whereas cations are prevented from
reaching the cathode by anion-permeable membranes. This phenomenon causes the formation
of alternate concentrate compartments and dilute compartments.
The compartment into which the feed is introduced becomes the dilute compartment, and
the stream exiting from this compartment is referred to as the dilute stream. Similarly, the
stream exiting from concentrate compartments is the brine or concentrate solution.

B. Spacer Gaskets
Spacer gaskets separate the membrane sheets from one another. To minimize electrical
resistance, the distance between membranes, i.e., the cell thickness, has to be as small as
possible (around 200 p,m) . 3’4 In commercially available stacks, spacer gaskets are between
3 mm to 400 fim in thickness.
These spacer gaskets maintain the shape of the membranes, control the solution flow
distribution in the stack, and minimize the boundary layer thickness, and thus, limit the
effects of concentration polarization. It is very important to have uniform flow distribution
in the compartments and to prevent internal leakages, particularly from the ion-concentrating
cell to the ion-depleting cell.
Two basic spacer designs are in common use :2,5

• In the tortuous path design, solution travels between and along a pair of membranes,
from inlet to outlet, along well-defined channels (Figure 3). The width of the channel
is made up by plastic, thin strips which are kept in place by many small cross bars.
• Consequently, each channel covers many times the length between both the mem ­
branes. In these channels, the speed of the solution is about 30 to 50 cm/s. This kind
of thick spacer gasket is mainly used by an American manufacturer of ED stacks.
• In the sheet-flow spacer design, solution travels from the inlet to the outlet between
a pair of membranes in a single pass, but covers the whole surface of both opposite
membranes (Figure 4). This unique channel is lined with a thin plastic mesh fabric.
In this kind of thin spacer gasket, the speed of a solution is about 5 to 25 cm/s. These
spacer gaskets are primarily used by French, Japanese, and Russian manufacturers.
The drop in hydraulic pressure through a tortuous path stack is generally higher than
through a sheet-flow stack .6

C. Solution Circulation in the Unit Cell

Usually the supply ducts for the various solutions are formed by matching holes in the
spacer frames, membranes, gaskets, and frames. Each spacer frame is provided with solution
channels that connect the solution supply ducts with the solution compartment. Many kinds
of manifolding arrangements are possible for solutions undergoing dilution and concentration.
Electrode solutions have quite different specifications.

• In the case of a parallel circulation, the same compartments, diluting or concentrating,

are connected in parallel on the same pair of the four ducts: one for supplying and
one for issuing (Figure 5).
• In the case of water production from brackish water, there are often only three ducts,
with the compartments being supplied by the same duct (Figure 6 ).
• There are only three ducts in the case of brine production for the salt industry, mainly
in Japan. Two are devoted to the inlet and the outlet for the seawater, and the third
is the way out for the concentrated brine (Figure 7).
• With parallel circulation, the pressure drop is minimal and is the same in every
 195

FIGURE 3. Tortuous path spacer. (Courtesy of Ionics Inc., U.S.)

compartment. However, the depletion of salt is very limited; consequently, solutions

present quite a uniform concentration inside each compartment.
• In the case of circulation in series, the same solution, demineralizing or concentrating,
flows successively in all similar compartments. Therefore, the pressure drop is high,
and the demineralization is greater than in the earlier case. In consequence, the solution
at the outlet is very different from the feed concentration (Figure 8 ).

It may be noticed, from a general point of view, that since there are two fluid streams,
each one can circulate in parallel or in series, independently of the other one. Usually, fluids
are circulated by electric motor-driven pumps; but some gravity circulation is used for
production of drinking water.7

D. Flow System of the Stack

1. Cocurrent or Countercurrent Flow
To avoid large pressure differences across the membranes, it is customary to arrange the
dilute and the concentrate streams to flow through the stack in the same direction, which is
called cocurrent flow. When a single pass through the membrane stack is sufficient to achieve
196 Water, Wastewater, and Sludge Filtration

FIGURE 4. Screen mesh spacer and electrode block. (Courtesy of SRTI, France.)

ZZZSZZZZZZZZZZZZZZZZO
I> b

d Z T^y Z Z 77/77 77/77. 77/77

= o
FIGURE 5. Parallel circulation (b = brine concentrate stream; d = dilute stream).

the desired demineralization, the cocurrent flow system will also simplify the pipe and
manifold arrangement (Figure 5).
In some rare cases, dilute and concentrate streams flow in opposite directions through the
stack, which is known as countercurrent flow. In this case, the pressure difference at each
entrance may cause some undesirable water transfer from one stream to another.

2. Internal or External Staging

For an external staging, two or more groups of cells, each with its own set of electrodes,
 197

// 7 7 / / /LLLLL// / / / / Z Z Z { > d

IA
/
X> b
/
/
1 / 1
1 /
1 / •
,1 /
/
/
/
/
/ *
. / I
l
i /
/ 1
A
A

= Q
FIGURE 6. Parallel circulation (brackish water desalination, b = brine; d = fresh
water).

Ld Ld Ld
t> b

d 77C y - / / / / / 7 7 /7 7 7 7 /7 7 7 7 /7 7

FIGURE 7. Parallel circulation (brine production, b = concentrated brine; d = dilute

stream).

may be placed within the same clamping press. From the process point of view, these stages
are electrically independent, and several of them are interconnected to obtain a successful
degree of demineralization (Figure 9).
When there is an even number of groups of cells, the two extreme electrodes present the
same electrical polarity. With this arrangement, the electrical leakages in the ducts are
avoided.
In the case of internal staging, all the groups of cells are put between two electrodes.
But, in view of having some difference with the circulation in series previously described,
the group of cells must have circulation in series, and each cell in a group is subjected to
198 Water, Wastewater, and Sludge Filtration

17 7 7 7 / 7 / 1

\
/
11/
/
1/
/
, /
/
/
/ /
/
/
/
. /
1
• /

d ZZ0 / / / / / Z
/
/
*
///
\
zzz
I

2 Z O d

b = 0 = o b

FIGURE 8. Circulation in series (b = brine, d = dilute).

FIGURE 9. External staging of an electrodialysis stack.

n ~i ** n 1 r - r r
1 ~r~
I 11 'I I I I

FIGURE 10. Series parallel circulation.

 199

a parallel circulation — this is a classical series-parallel method of hydraulic circulation

(Figure 10). These stages are not electrically independent.

III. SELECTIV ITY

A. Description of Ion-Permeable Membranes (ipm)

Hellferich classified ion exchange membranes as homogeneous and heterogeneous, de­
pending on their microstructure . 8

• Heterogeneous membranes are those composed of more than one material. Typical
heterogeneous membranes are made by grinding ion exchange beads and mixing the
ion exchange material with a binder. This mixture is cast onto a support fabric that
imparts strength and dimensional stability to the membrane. Taking into account the
materials used, these membranes are named ion-exchange membranes.
• Homogeneous membranes are of uniform composition, even at the molecular scale,
and the physical properties of the membranes are essentially the same from point to
point.
• Microheterogeneous membranes are uniform from a macroscopic point of view, but,
at the molecular scale, contain some fragments, generally issued from the activation
process before grafting .9

Both homogeneous and microheterogeneous membranes are obtained by grafting func­

tional groups on a macromolecular chain.
Cation-selective membranes are usually made of cross-linked polystyrene that has been
sulfonated to produce sulfonate groups, —S 0 3H _ , which are attached to the polymer. Besides
these strong acidic groups, one can also find weak acidic groups such as - C O O H - .
Anion-selective membranes can be cross-linked polystyrene containing quaternary am ­
monium groups, N +. Currently, aliphatic anion-selective membranes are favored, because
they have lower electrical resistance.
Since an ion-permeable membrane is always electrically neutral, the functional groups
are electrically equilibrated by some mobile ions of opposite charge, the counterions. The
ions of the same polarity of the fixed charges are the co-ions.

B. Models for Ion-Permeable Membranes

One can assume, as did Meares , 10 that an ion-permeable membrane consists of an isotropic
three-dimensional network of polymer chains to which anionic or cationic groups are chem ­
ically bounded. This is swollen by water or by an electrolyte solution which constitutes an
internal aqueous phase (Figure 11). The fixed groups are completely ionized and are uni­
formly distributed throughout the polymer.
The internal aqueous phase is qualitatively similar to an ordinary aqueous phase. In this
internal aqueous phase, free ions and counterions balance the electrical charge of the fixed
groups. The polymer network is similar to a swollen rubberlike gel. The average distance
between junction points of the rubberlike polymer network is large compared to the dimen­
sions of the free ions of the internal solution. Therefore, the network exerts no mechanical
sievelike effect on the movement of the small free ions.

C. Transport Mechanisms
The most important principle involved in the ion exchange membrane process is the
Donnan equilibrium relationship . 11 For a dilute solution of univalent electrolyte, such as
NaCl, it can be shown that the relative concentration of a co-ion j in the membrane is
200 Water, Wastewater, and Sludge Filtration

FIGURE 11. Crosslinked ion permeable membrane (O

= fixed charge on macromolecule chain, # = co-ion,
O = counterion, A = water).

where Cjm and C^ are the concentration of the co -ion j, respectively, in the ion-permeable
membrane m and in the solution 1, and Cs is the charge concentration of the fixed groups
per unit volume of solvent in the membrane.
As the relative concentration of ions in a free solution is 1, its value is lowered in the
membrane for a co-ion, whereas the relative concentration of the counterion in the membrane
is increased (Figure 12).
If it is supposed that ions travel through the membrane by jumping from one fixed charge
to another, it appears that only counterions can pass through. Consequently, the main part
of the electrical current is transported by counterions and scarcely anything by co-ions.

D. Transport Number
The relative movement of different ions in a solution is denoted by transport numbers.
The transport number tj of a given ion in a solution is simply the ratio of the electric current
conveyed by that ion, ij? to the total current i . 12

Consequently, in an ion-permeable membrane, the transport number of the counterion is

near unity, and that of a co-ion is close to zero:
 1 0 -2 10
2
C jl

FIGURE 12. Relative concentrations according to Donnan

exclusion (Cj, = ion concentration in bulk solution; Cjm =
ion concentration in membrane phase; Cs = concentration of
fixed charges).

■cpm (6)
The sum of the transport numbers is always equal to unity.
However, in solution flowing on each side of both kinds of membranes, the transport
numbers of the anions and the cations are generally not very much different from each other.
The electric current is conveyed quite equally by each kind. For example, in the case of a
0.1 N KC1 aqueous solution, it is established that:

tcl - = 0.5102 tNa+ = 0.4898 (7)

and for 0.1 N NaCl solution that:

tcl - = 0.6146 tNa+ = 0.3854 (8)

Consequently, the selectivity of a membrane is well defined by its transport numbers.
Meanwhile, Winger et al . 13 put forward a definition for membrane permselectivity, i|/ipm,
based on the idea that the function of an ion-permeable membrane (ipm) is to alter transport
numbers of the individual ions, and that the maximum transport number attainable for a
hypothetical membrane with an infinite concentration of fixed charges is 1.0. This led to
the following equation:

(9)
202 Water, Wastewater, and Sludge Filtration

However, the determination of transport numbers by direct measurement is not easy, except
in the case of 1:1 electrolyte. Therefore, sometimes, it is preferred to characterize an ion-
permeable membrane by its membrane potential. 12

IV. PERM EA BILITY

A. Qualitative Considerations
1. Concentration Polarization
For ED, as for other membrane processes involving liquid phases, the concentration
polarization is an important feature. When a counterion is transported through an ion-
permeable membrane, from a dilute stream to a concentrated stream, it is depleted at the
membrane-solution interface in the dilute stream, and is not replaced at this point at the
same rate at which it is removed through the membrane. Momentarily, the co-ion is accu­
mulated at the membrane-solution interface, but, in order to meet the electroneutrality of
the liquid solution, the accumulated co-ion migrates away from the interface into the bulk
solution of the dilute stream. Thus, the total ionic concentration in the dilute compartment
is depleted at both interfaces between membranes and solution, as shown on Figure 13.
On the other hand, the total ionic concentration in the concentrate compartment is ac­
cumulated at both interfaces for analogous reasons.
It may be seen in Figure 13 that this effect is located on a thin zone, namely, the
concentration polarization layer (cpl) of 1 to 200 |xm. As such concentration polarization
occurs, the concentrate potential across the ion-permeable membrane is further enhanced,
and the diffusional leakage of ionic species through the membrane increases in an opposite
direction to that of the electric transference.
Therefore, the net ionic transport is severely retarded for a given electric potential. In the
presence of such concentration polarization, the ohmic potential drop in the dilute stream
also increases to a considerable magnitude. This occurs because the ohmic electric resistance
of an electrolyte solution is generally inversely proportional to its ionic concentration.
The occurrence of this concentration polarization is obvious for steady-state operation.
But it is equally true for batch type desalting or demineralizing operations in which the
current changes with time. As the movement of ions (towards this zone of transport number
change) is governed by potential gradient, concentration gradient, and physical mixing, there
is no method by which polarization can be prevented other than by stopping its cause, that
is, the electric current.

2. Critical Current
As a higher external electric potential is applied to obtain a higher electric current, the
dilute stream is more depleted of ionic species and is more polarized, often to such an extent
that the ionic concentration is virtually nil.
The corresponding electric current at which the interfacial concentration becomes zero at
some point in the apparatus is called the limiting current density or the critical current
density, because any further increase in current density would cause loss of the electric
efficiency of the separation.

3. Water Splitting
When the ionic concentration at the interface falls to zero, the electric current is still
transported by ions. However, in this case, the water solvent is electrolyzed and hydrogen
ions and hydroxide ions will be formed. These ions will compete with the original counterions
obtained from the salt, in the electric transport across the ion permeable membrane. This
phenomenon is called water splitting, and it retards the transport of the ionic species to be
separated for a given electric current.
 203

FIGURE 13. Concentration profiles in a unit cell given by interferometric holography.

As a matter of fact, the electric current is transported by all the ions in the solution, those
produced from salt and those produced as a result of water splitting. However, the mobilities
of these latter ions are 3 to 5 times greater than the mobilities of all other common ions.
As soon as the concentration of the ions produced from salt approaches V3 to V5 of the value
of 1 0 “ 7 mol/1 (i.e., the value of normal ionization of water), an appreciable amount of
current will be carried by hydrogen and hydroxyl ions.
204 Water, Wastewater, and Sludge Filtration

The water splitting also induces a change in the pH value at the interface. The anion-
permeable membranes are inherently unstable at high pH, and deterioration takes place.
Therefore, special care should be taken to reduce concentration polarization near the anion-
exchange membranes.
Furthermore, many inorganic hydroxides are insoluble. Consequently, a local increase of
the pH at the membrane-solution interfaces will induce the precipitation of insoluble hy ­
droxides, such as Ca(OH)2, Mg(OH)2, etc. on the surface of the membrane.

4. Electrometathesis
The concentration polarization, as in every membrane process in an electrodialysis unit,
also imposes a reduction in the production rate which is directly related to the current density.
But, as in the case of any membrane process, the existence of a concentration polarization
phenomenon may give rise to some favorable effect.
As water splits into proton and hydroxyl ions (for a particular value of the electrical
current), it is possible to make use of these ions in the production of corresponding acid
and base from the salt. This is done with bipolar membranes, where the dilute compartment
is so thin that the two flat sheets of anion- and cation-permeable membranes are combined
together. Nevertheless, the membrane stack has to be arranged in such a way that the
electrometathesis is effective .6

B. Ion Transport through Membrane Unit

7. Membrane Unit
An electromembrane module, such as an ED stack, is comprised of membranes and solution
phases. When a sufficiently high external electric field is applied to the membrane stack,
an electric current passes across the stack. In this case, the transport of ionic species from
one compartment to another is largely due to the electric current.
In the design of an ED module, it is essential to relate the external electric potential
applied at a given concentration distribution across the stack with the current density, the
ionic flux, and osmotic flow. This relationship should be discussed for the solution phase
and membrane phase in the stack. Special emphasis should be given to the membrane and
the two neighboring solutions, the so-called concentration polarization layers. This entire
assembly is called the membrane unit.

2. Basic Equations
The flux density Jj of a species j is classically related to its concentration Cj and its velocity
V
j*12

Jj = Cj • V, (10)

The velocity Jj is proportional to the driving force. When energy is supplied as electricity,
the driving force is related to the gradient of the electrochemical potential [Xj, the propor­
tionality coefficient being the mobility m^:

Vj = - m j • grad ^ (11)

For ions in aqueous solution, the driving force is always in equilibrium with the frictional
viscous forces; hence, velocity reaches its constant limiting value at a rapid rate. Further,
during ED process, it is taken that the flux of the different species are perpendicular to the
surface of the flat membranes.
In place of the mobility if one uses the diffusivity Dj as indicated by the Nemst-
Einstein relationship, then 14
 205

Dj = rrij • R • T (12)

where R and T are the perfect gas constant and the absolute temperature, respectively.
Substituting Equations 11 and 12 in Equation 10, then the flux density Jj of species j can
be expressed as:

Jj = - ^ (13)

The classical Gibbs expression for the electrochemical potential jlj is

[i° = juL
j + R • T • ln(aj) + v/P - P*) - s/T - T*) + ZjF<|> (14)

where jx° is a reference potential, aj? Vj, sj? and Zj are (respectively) the activity, the partial
molar volume, the partial molar entropy, and charge number of the ionic species j, respec­
tively, and P, T, and $ are, respectively, the hydrostatic pressure, the temperature, and
electric potential. The superscript* indicates the reference state.
Assuming that the process is running at a uniform temperature and constant pressure, the
flux density Jj takes the following form, which is known as the Nemst-Planck equation . 12

Jj = - D j • Cj g r S In aj - ^ • Cj • Dj ^ g r ^ 4> (15)

The conservation of the electric charge yields the value of the electric current density?:

T ^ v F - J j (16)
j

Combining flux Equation 15 and charge conservation Equation 16, the gradient of the
electric potential <\>, which is the same for all ionic species present in solution, can be
explicitly expressed.
Substituting the gradient of the electric potential obtained, in the flux Equation 15, an
expression for ionic flux can be written as follows:

J j = - D j • Cj [gracl ln(aj) - Zj 2 (Vzj) ln(aj)l + (Vaj) '* ( 17)

j

If solutions are dilute, the first two terms of the second part of the flux equation depend
only on the activity aj of the sole ion j. Thus the flux equation is simplified as the following:

1 } = ~ D }C}K m ^ a d ln(aj) + (t/Zj) • (?/F) (18)

In theabove equation, the coefficient Km is equal to unity in the case of a solution phase
and is very low within a membrane phase . 15
If the solution phase 1 is dilute, one can substitute the concentration Cj in place of the
activity aj? and thus flux density Jj! can be written as:

J* = - D jC jgralliK C j) + (19)
Zj ^
206 Water, Wastewater, and Sludge FUtration

m pa mpc m pa
cpI cpI cpI cpI cpI cpI

s \
E

FIGURE 14. Concentration profiles in bulk solution and in membrane unit (b =

concentrate solution; d = dilute solution; Cj = ion concentration; cpl = concentration
polarization layer; mpa = membrane permeable to anions; mpc = membrane permeable
to cations).

In the case of commercially available membranes, the following equation can be written
for the membrane phase m with current density being not too low and Km = 0.

T = (20)
z, F

Equation 19 is true for the two solution phases and for the four concentration polarization
layers, whereas Equation 20 is true for the two membrane phases of each unit cell.

3. The Steady State

At steady state, the migration and diffusion fluxes of counterion j through solutions b and
d and the concentration polarization layers (cpl) are the same as the migration flux through
the corresponding ion permeable membrane (ipm) and are equal to the total flux (Figure
14), i.e.:

*Jj j ,c p l,d , i p m ’J j.ip m j ,c p l,b , i p m (21)

Using the expressions of the flux densities through the membrane, Jj5ipm? and solution
phase (only in one concentration polarization layer of the solution phase), *Tj>cpUpm> (i.e.,
using the Equations 20 and 19) one obtains:

timi
Dj gracl q + (22)
zj F Zj F

In an electrodialysis stack, the flux is perpendicular to the surface of the flat membranes.
Furthermore, the integration of Equation 22 is quite easy and gives a simple relationship
(Equation 23) between concentration polarization layer thickness 8 , electrical current density
i, bulk concentration in the liquid solution, and concentration Cjn near the membrane for
a counterion.
 207

In the course of this integration, the values of the diffusity Dj and the transport numbers are
assumed constant and as the term i-8 /Zj is either negative or positive, the charge of ion j
accordingly has been taken as rS/|Zj|.

4. The Critical Current

When the concentration of the counterion near the membrane falls to zero, i.e.,

Cjn = 0 (24)

the critical or limiting current density, icrit, is expressed as:

,, F D,
ic r it - lZjl -------- C jl (2 5 )
jm ®

From the above equation, it is clear that the critical current increases with the increase in
bulk concentration C^ of the ion j in the liquid, and the diffusion coefficient Dj. However,
its value decreases as the difference in the two transport numbers, tjm— tj, and the concen­
tration polarization layer thickness 8 increase. In particular, the effect of the thickness is
very important. The concentration polarization layer in this case is different and a little
thinner than the hydrodynamic limit layer.
Classical hydrodynamics considerations show that the thickness is related to the Reynolds
number . 16 Therefore, a high limiting current density can be achieved by controlling the
thickness with a high stream flow rate in a thin channel.
Consequently, the term \critIC]X, called by Cowan and Brown 17 the polarization parameter,
is nearly independent of concentration, but is related to velocity 18 and, consequently, to the
Reynolds number. 19 Therefore, once this polarization parameter has been determined for a
system, it can be used to estimate the limiting current density for the same stack operating
with the same solution, but for different concentrations.

C. T ransport and Efficiency in a Unit Cell

The overall electrokinetic phenomenon in a membrane stack can be understood by com ­
bining all individual phenomena in existing membrane and solution phases. As previously
indicated, the discussion is confined to ED only, but this will be analogous to other electro ­
membrane processes as well.

1. Ionic Fluxes
For a unit cell, the flux of an ion j out of the dilute compartment will mainly result from
the transport through the corresponding ion permeable membrane, J j,jpm (in this case, j is
the counterion). However, because the transport numbers are not equal to zero or unity,
there is also a flux, Jjjim, for the counterion coming into the dilute stream through the other
membrane of the cell (in this case j is the coion). Here, the membrane jim is impervious to
the ion j .
Consequently, the net flux Jjuc of this ion j out of the unit cell (uc), is

J j, u c ~ J j. jp m “ J j. j i m (26)

Using Equation 20 for the flux density through a membrane, the net absolute flux of ion
j out of the unit cell can be written as:

J j ,uc ~ |z | p (tjJ p m I jj im ) (2 2 )
208 Water, Wastewater, and Sludge Filtration

Thus, one can define the current transport efficiency r\t as:

Introducing the value of the net flux, the current efficiency Tjt can be written as:

(29)

Hence, the electric current efficiency for ion transport is equal to the difference in transport
numbers of ion j through each kind of ion permeable membrane. From this point of view,
one can see that it is very important to use membranes with transport numbers as close as
possible to unity for the counterion and zero for the co - ion . 20

2. Water Transfer
In concentrated solution or in systems involving a high degree of desalination or de­
mineralization, water can be transported from the dilute compartment to the concentrate
one. In practice, since microporous membranes are used to obtain a high flux of certain
ionic species, the effect of the osmotic flow cannot be neglected.
Since a counterion is transported through the pore under the influence of the applied
electric potential, it experiences a frictional force due to the surrounding solvent and ex ­
changes its momentum (i.e., its mass velocity vector). In a neutral membrane, the momentum
absorbed by the solvent exactly balances with that transferred from the co-ion, which is
moving in the opposite direction. Therefore, the external electric potential has no effect on
solvent flow. However, if the membrane is electrically charged by means of fixed charges,
the counterion is more concentrated in the pores than the co-ion. Thus, the momentum
transferred from both ions to the solvent do not balance each other, and the solvent receives
a net force in the same direction of the movement of the counterion. If the convective flux,
in the presence of any pressure gradient across the membrane, is also taken into account,
one can write the osmotic flux for the solvent, i.e., water transfer, through the membrane
as 2

(30)

where tWm is the so-called electroosmotic transport number.

Physically, the transport number is the number of moles of solvent transferred by 1 mol
of electricity in the absence of hydraulic and osmotic pressure gradients.
Furthermore, most ionic species are transported through an ion permeable membrane in
their hydrated form. The corresponding transport of solvent water in the hydrated form is,
in general, of the same order of magnitude as that of pure electroosmosis, i.e.,

(31)

where hj is the hydration number of each ionic species j, and the summation is carried out
over all ionic species participating in transport across the membrane. 12
Thus, the total solvent water through the membrane is:
 209

Therefore, one can define an apparent electroosmotic transport number t*wm through the
membrane as:

Cm = + tjm hj (33)

This water transfer occurs in the same direction as the flux of the counterion; in other
words, it goes out of the diluting compartment into the concentrating compartment. Con­
sequently, it causes a dilution of the concentrate and a concentration of the dilute solution.
This action defies the very purpose of the ED method. So, for a constant salt transport, a
large water transfer will necessitate an additional current to meet the desalted or concentrated
product quality specifications. It also reduces the quantity of product obtainable from a given
amount of feed to the dilute stream.
The effect of water transport on the efficiency with which the current is used to separate
salt and water is quite ambiguous. If it is assumed that this water transport reduces only the
amount of electricity used for salt transport, the remaining quantity is proportional to (1 —
t ^ ) and the effect of water transport on current efficiency can be given by:

Tlw = (1 ” twm) (34)

3. Electrode Reactions
Throughout the diluting and concentrating cells of an electrodialysis stack and in the
membranes, electrical conduction is due to ionic transport. Thus, the relative movement of
electric charges carried by the ions is, in fact, the current itself. At the electrodes, the
mechanism of electrical conduction changes abruptly from ionic to electronic. Using noble
metal electrodes, such as platinum, this transition is accomplished by the addition of electrons
to the ions present in the solution at cathode and by removing the electrons at anode.
Thus, in many situations (not particularly in high sulfate low chloride solutions), it is
found that the anode reaction discharges oxygen from water with the simultaneous production
of hydrogen ions:

3 H20 - > 1/2 0 2 + 2 H 30 + + 2 e “ (35)

In consequence, the solution in the anodic compartment, the anolythe, becomes more acidic
with time. The formation of 0 2 or some oxidizing materials at the anode causes rapid
deterioration of the stack components, especially the electrodes.
The principal cations present in typical brine used at cathode are much less readily
discharged than the hydrogen ion, and the net result of this is the water electrolysis reaction
at the cathode.

2 H20 + 2 e~ —* H2 + 2 OH~ (36)

As a consequence, the solution in the cathodic compartment, the catholyte, becomes more
basic as the electrodialysis process proceeds. Alkaline solution, if allowed to enter the stack,
is likely to produce precipitation of insoluble hydroxides.
These electrode reactions are always associated with the industrial electrodialysis operation
and introduce two problems:

1. The products of electrode reactions may be harmful to the electrodialysis stack or may
interfere with the continued operation of the system. Because of this, it is usual to
provide hydraulic isolation of the two electrode stream compartments at the end of
the electrodialysis stack and to use some chemical resistant ion permeable membranes.
210 Water, Wastewater, and Sludge Filtration

The catholyte basicity can be controlled by the addition of inorganic acids, such as
sulfuric acid, to the cathode stream and reduced by mixing the anolyte.
2. Additional power must be supplied to provide the energy for electrode reactions.

D. Current Efficiency of the Stack

The total external voltage AU that is needed to operate a practical electrodialysis stack is
the sum of three principal potential drops. These three potential drops are associated with
the electrode AUe, the ohmic drops within solution and membrane phases AUr, and the
concentration potential across the membrane at the membrane-solution interface AUp:

AU = AUe + AUr + AUp (37)

1. The Stack Resistance

The resistance of the operational portion of an ED stack is itself the sum of several
resistances. The readily identifiable factors for a unit cell are the resistance of the bulk
diluted and concentrated solutions, rd and r5, the resistance of the four and concentration
polarization layers (the sum of these four is rcpl), and the resistance of the two membranes
used, rapm and rcpm,

ruc = rd + rb + rapm + rcpm + rcpl (38)

Among the eight resistances indicated, the main values are generally attributed to the
dilute stream and its concentration polarization layers and the membranes. If it is assumed
that the inner solutions of the membrane phases are similar to the dilute solution, it is obvious
that the resistance of a unit cell can be related to the concentration Cd of the dilute stream
as follows:

ruc = kd • C - g- (39)

where kd and gx are positive constants between two limited values of the concentration of
the dilute stream; kd is directly related to the equivalent conductance of the diluting solution.
In some cases it is possible to expand it into a series.
Thus, the voltage drop in the unit cell is

AUUC = AUr>uc + AUp,uc = ruc • i (40)

In this expression, for constant concentrations, when the density i of the electric current is
increased, AUr uc still remains proportional to i, but AUp uc increases exponentially. On the
other hand, as the applied voltage to the stack AU is increased, the current density i goes
towards an asymptotic value: i.e., towards the critical current icrit.

2. Concentration Potential
When a mole of electricity travels in a solution from point 1 to point 2, where concentration
of salts are not equal, it is partially carried by anion and partially by cation, as indicated
by transport numbers. In this case, the quantity of energy for the transport of an ion is equal
to the product of the change in its free energy (Gibbs energy) and the transport number. For
all the ions, the whole energy is the sum of these partial energies, and the corresponding
potential drop is given by the chemical junction potential AUj_2 such as : 12
 211

These potential drops exist in each layer of the unit cell where concentration varies. Con­
sequently, the total potential drop is the sum of these elementary cell voltages.
However, in the whole stack, all the cell voltages caused by the differences in the
concentrations of concentrated and diluted streams, as well as those coming from the con ­
centration polarization layers, cancel each other. Only the cell voltage caused by the mem ­
brane pairs remains, which, for a unit cell, reduces to:

(42)

This is expressed for the stack by a concentration polarization efficiency T]p.

3. Electrodes and Manifold Leakage

At useful current densities, the electrode potential AUe will generally be about the same
as that for the electrolysis of water, i.e., 2.3 volts. However, this voltage can be reduced
to a small fraction of the total voltage by increasing the number of unit cells used in series
between a given pair of electrodes. The additional power supplied in a stack to provide the
electrode reactions is expressed as an electric efficiency r\e.
The current leakage through the manifold system can, in principle, be completely elim ­
inated. To achieve high coulomb efficiencies, that is, high utilization of the quantity of
electricity added to the stack, it is important to minimize the current through the internal
dilute and concentrate ducts, since electric current through these two paths bypasses the
membranes of the unit cells and does not contribute to the separation by ED.
In the case of high concentration of salt, the electrical leakage can be minimized by feeding
or discharging solution to or from each compartment through separate distributors that are
connected to the main duct through high electric resistance hydraulic connections, generally
with a small diameter. Thus, the leakage current in a unit cell stack is usually less than 2%.
This is expressed for the stack by an electric efficiency %.

4. Overall Efficiency
The overall efficiency of an electrodialysis stack, t], is the product of the six principal,
individually identified efficiencies: the three voltage terms, T]e , T]r, and T|p , respectively,
associated with the electrode reactions; the total resistance, the polarization potentials, and
the three current terms r\{, r\n and associated with the manifold leakage, ions, and water
transport:

ti = 'ne-'nr'V'nr-'nr'n. (43)

The first and second terms of this overall efficiency are primarily functions of the stack
design; the third term depends on the operating conditions, salinity, and current density; the
fourth depends on design factors, such as cell spacers and membrane thickness, and on
operating conditions; and the fifth and sixth terms mainly functions of the membrane selected.
The major inefficiency in stack operation is associated with the product inp’T]r; the product
T|e*r|f is seldom less than 0.9, and the product will not be affected by any design in
which the electric resistance and concentration polarization losses have not been minimized
in a careful manner.

V. PROCESS AND EQUIPMENT DESIGN

A. Material Balance
Under the forced flow conditions imposed in conventional electrodialysis, relative mixing
212 Water, Wastewater, and Sludge Filtration

of the main flowing streams and concentration polarization layers gives an average effect
which is either demineralization or concentration, according to the compartment under
consideration.
The quantity of salt transferred from one stream to another is directly related to current
density i and effective area A of each unit cell. The total number of cells is N. As a mole
of electricity passes through a unit cell, it transports a given amount of salt. A mole of
electricity carries 96,500 C, i.e., 96,500 A*s or 26.4 A*h.
So, the number of moles of a z:z salt (i.e., a salt for which anion and cation have the
same absolute charge number z) theoretically transported in a unit time is

Anuc = ^ (44)
zF

In cell and stack design, serious considerations have to be given to minimize the current
that will flow through the manifold. Therefore, it is safe to assume that the total ionic flux
that flows through adjacent compartments is the same over the whole stack. According to
this, the quantity of electricity acts each time it passes through a unit cell, and thus it acts
N times. Taking into account the current efficiency tj, the total number of moles (An)
transferred per unit time in the stack is

AS - „ ^ (45)
zF

This quantityof mole is extracted from the dilution stream and brought into theconcentration
stream during a unit time with an absolute flux density J through themembrane area NA.
Therefore, An can also be written as follows:

An = NA • J (46)

If ED is operated in batch mode over a time t, the quantity of matter transferred from the
demineralizing volume Vd to the concentrating volume Vb induces a variation in concentra­
tions Cd and Cb, respectively. This can be written as:

An • t = —A(Vd • Cd) = A(Vb • Cb) (47)

where A refers to the difference with respect to the value at initial time.
If water transfer is not the main criterion, volumes do not vary much, but concentration
decreases or increases according to the stream, as given by the following equations:

An • t = —Vd • ACd = Vb • ACb (48)

If ED is a continuous process, assuming that streams are running through the ducts with
rates Qd and Qb, An can be written for this case as follows:

An = - A(Qd • Cd) = A(Qb • Cb) (49)

If water transfer is low and the flow rates are constant in each stream, the ionic transfer can
be written as:

An = - Qd • ACd = Qb • ACb (50)

 213

Equations 45 and 47 (or Equation 49) contain seven variables, and it is, therefore, necessary
to fix four of them. Frequently, volume or rate of the solution to be treated and its variation
in concentration are fixed by local considerations. In solving these equations, the following
two cases were considered.
In the first case, for the ED stack chosen, since the three variables N, A, i are always
together in the product form NAi, one has to calculate two of them by some other means.
The third one can then be calculated because the product of the three is known from the
above-mentioned set of equations. It has been previously shown that the current density i
is always smaller than that given by the polarization parameter. Knowing the value of i, the
product NA can be calculated. The fourth variable to be fixed is a choice between the
variation in concentration of brine Cd and its rate Qb or its volume Vb in order to solve these
equations. However, in this case, the ED run time t should also be specified.
In the second case, a suitable ED stack is used for which the product NA is fixed and
the current density i is obtained from the polarization parameter. The fourth parameter chosen
is either the variation in concentration or flow rate.

B. Demineralization Factor
For steady state conditions, the demineralization that can be achieved in the stack, or the
stack area required for a fixed degree of demineralization, can be derived by considering
the relationship between flux density and salt transfer. In this calculation, the ED stack is
considered to be a perfectly mixed reactor.
In the case of a differential membrane area of width X and length dz, a material balance
equation can be written for a constant flow rate Q.

Q*dC±J*N*Xdz = 0 (51)

Here, the flux density Jj>uc is taken as constant for the N unit cells with an absolute value
J.
Taking the value of J expressed with the current density i from Equations 45 and 46, and
introducing the voltage drop in the unit cell as indicated by Equation 40, the above material
balance equation can be rewritten as:

Q • dC — N • Xdz = 0 (52)
zF ruc

Since most applications of ED are concerned with demineralization of brackish water or

solutions, further discussions are confined to these cases.
For a diluting solution, taking into account the Equation 39 for the voltage drop in the
unit cell, the previous equation is modified to:

-n AU
(53)
Qd ' dCd + zF " k f Cd8,N ' Xdz = °

Supposing the coefficient g is equal to unity, which is always true in a restricted domain of
concentration, integration of Equation 53 from the inlet of the stack to the outlet (i.e.,
through the length of travel, Z), yields:

f AUuc NXZ -|
(54)
214 Water, Wastewater, and Sludge Filtration

where q is the demineralizing factor relating the concentration at entrance (Cd) and outlet
(Cj) for dilute and can also be written as follows:

Equation 55 may be used to estimate the degree of demineralization as a function of cell

dimension XZ, overall current efficiency r\, applied voltage drop per unit cell AUUC, average
equivalent conductance per unit length kd and flow rate Qd. It should be noted that the term
X/Qd is inversely proportional to linear velocity for any given thickness Y.
The ratio AUuc/Kd may also be expressed in terms of current as i/Cd in a form analogous
to the critical parameter. Substituting the above in Equation 54 yields:

(56)

It may be seen that for a stack with given geometry and for any flow rate Qd, this expression
is valid as long as the circulation is in a series or parallel mode. In the latter case, the flow
in each unit cell is Qd/N, the length of channel is NZ, and the ratio NZ/Qd is the same as
in the case of a circulation in series.
From the relationship established empirically between the critical parameter (i/Cd) and
linear velocity, useful ratios of i/Cd and X/Qd may now be chosen without any consideration
to resistance or voltage requirements.

C. Power Requirement
In conventional ED, the direct current electrical power P required for ion transport in the
stack, where solutions are circulated in parallel in N unit cells of area A is

P = N • (iA) • AUuc (57)

Substituting for AUUC from Equation 54, and for i from Equations 45 and 50, and in
combination with the following equation,

(58)

one obtains a value of the power requirement P on an absolute basis as:

(59)

To obtain the energy E for a unit product, P is divided by the flow rate Qd.

(60)

Equation 60 is particularly useful to estimate energy requirements for desalination or de­

mineralization, because the parameter kd is nearly constant for dilute solutions. The energy
consumption increases with concentration Cd and flow rate Qd of the solution to be de­
mineralized. It increases also as the concentration Cd in dilute is reduced; but i decreases
as the area (NA) of stack membrane is increased.
 215

D. Instrumentation
Electrodialysis is a process which involves the transport of ions from one solution to
another. So, the results of ED are directly related to the quantity and the kind of ions present
in the solution.

1. Sensors
When the objective of the electrodialysis process is the production of potable water or
demineralized solution of specified salinity, the most common method of quality and process
control is by means of electric sensing equipment. With proper temperature compensation,
the variable under control may be detected by a properly calibrated sensor.
The easiest method, and also the safest, is to use conductivity measurements. This gives
the overall ionic content of the solution within a precision of about 5%. As the sensing part
of a conductivity cell are the two metallic electrodes, there is no normal wear. But, some­
times, one can notice mechanical failures, or chemical or electrochemical corrosion, related
to the nature and the velocity of the solution. Hence, this kind of sensor may be used for
long periods without recalibration.
A more specific method is the use of potentiometry or amperometry. Theoretically, these
ionometric methods give the value or the amount of a known ion such as proton, pH, or
anions such as chloride, bromide, cyanide, fluoride, fluroborate, iodide, sulfide, or cations
such as ammonium, barium, cadmium, calcium, copper, lithium, potassium, sodium, etc.
when their concentrations are not too high. However, the given value may be wrong if there
are some interfering ions. As the sensing part of the electrochemical chain are the two redox
electrodes, they require daily maintenance. In some cases, there is fouling of the ion selective
membranes. Therefore, the maintenance should also include cleaning and periodical change
of the membrane. For a suitable use, daily calibration of the sensing equipment is necessary.

2. Control
For a continuous process, essential instrumentation would be no more than the use of a
light or an alarm system to indicate excess product salinity. Further control may entail
automatic plant shutdown and product diversion to some specific tank or waste, when the
product salinity is too high. Automatic voltage control from conductivity feedback signals
would be desirable, but this type of system would have to operate within polarization limited
voltages.
A batch process requires little more than the conductivity sensor and on -off control.
Recirculation is continued at any voltage current or salinity level until the required salinity
product is reached. At this time, the product flow is diverted to storage by actuating a three-
way valve. A liquid level control in the holding tank actuates a feed value when the tank
is nearly empty, and recirculation will be resumed when the conductivity sensing element
detects saline water or solution coming out of the stack.
Use of a conductivity sensor to initiate the batch recirculation and discharge function
guarantees a product solution of desired salinity, but does not control the time required for
demineralization. Therefore, variations in feed salinity or temperature will cause corre­
sponding variations in batch cycle time.
Similarly, feed-water variation in a continuous process would not alter the product flow
rate, but would yield a product with varying salinity. A conductivity sensor alone, in this
case, can do little more than sound an alarm or divert the product when its salinity is above
the specified value.
As ED is used in many cases for the production of drinking water, product with constant
salinity is more desirable than a constant production rate. The product water is usually stored
for subsequent use.
216 Water, Wastewater, and Sludge Filtration

VI. OPERATING CONDITIONS

A. Pre- and Posttreatment

To achieve optimum performance, ED requires pretreatment like other membrane sepa­
ration processes. In the case of drinking water production, the presence of suspended matter,
colloidal or organic materials, soluble salts near their saturation levels, or materials readily
oxidizable to an insoluble state in natural brackish water gives rise to problems in electro ­
dialysis systems. Thin compartments and narrow manifold channels become traps for sus­
pended particles flowing through them. Therefore, a fine prefiltration with a 10-prni cartridge
filter is essential.
Colloidal and organic matter tends to “ p°ison” membranes of an electrodialysis stack.
Most colloids are poly anions, such as the silicates, hydroxides of ferric ions, humic acids,
phenol derivatives, etc. which carry a net negative charge. In consequence, these substances
get collected on or in the anion permeable membranes. These foulants may form a film
which, in turn, may retard the transfer of the ion through the membrane or neutralize the
fixed charges of the membranes. Neutralization of the membrane charges and deposition
increase the electric resistance of the ED stack and reduce the ionic flux out of the diluting
stream. This frequently creates physical damage to the membrane structure. It has been
found that for raw water containing ferric ions (Fe3 +) in an amount greater than 0.3 mg/€,
or hydrogen sulfide (H2S) above 0.3 mg/€, or manganese ion (Mn2 +) above 0.1 mg/€,
pretreatment becomes essential. The pretreatment in this case involved oxidation of the
inorganic ions followed by filtration to remove the resultant insoluble species formed.
When process water contains colloids, detergents, organic degradation products, etc.,
clarification via coagulation, flocculation, and filtration may be used. Organic fouling could
be avoided by using an activated carbon cartridge when the turbidity exceeds 2.0 NTU.
Biological fouling can be prevented by the addition of chlorine to the raw water. However,
chlorine must be removed before the water enters the stack, because an excessive amount
of chlorine degrades membrane and electrode materials, since it forms oxidation products.
A good sterility may be obtained with some other products, such as sodium azide, etc.
Salts near their saturation level present another type of problem. Since ED is a process
which involves the transport of ions from one side of a membrane to another, the concentration
of salts in the concentration polarization layer on the concentrate side of each membrane
will be higher than in the feed or bulk concentrated stream. A solution containing significant
quantities of calcium carbonate and sodium bicarbonate is particularly likely to cause pre­
cipitation on the concentrate side of the anion membranes because concentration polarization,
at this point, results in the transport of hydroxyl ions into the concentrate stream.
The most common method for solving this problem is the acidification of either the
concentrate solution or the entire feed stream before it enters the membrane stack.
Saturation with respect to calcium sulfate presents a more severe problem because
acidification has little effect on its solubility. However, one can use acidification under
critical conditions, since the maximum solubility of calcium sulfate is at pH = 4.0, and
the supersaturation provides for precipitation after some limited delay. Water saturated with
calcium sulfate requires presoftening before treatment in an ED stack.
Posttreatment is required, in some cases, because of the change in pH brought about by
the process itself. The product stream will have a pH lower than the feed water because of
the polarization on the anion membrane. Therefore, it may be necessary to increase the pH
of the product water. Chlorination is also used as a posttreatment for the product water.

B. Current Reversal or Pulsing

The symmetry of the alternating anion and cation permeable membrane system made
obvious to the early investigators that reversing the polarity of the applied DC current
 217

automatically interchanges the functions of adjacent diluting and concentrating compart­

ments.
Whenever the membranes are suspected of producing a sieve effect, rather than acting as
a pure ion transfer phase, current reversal is practiced on a periodical basis, in an attempt
to discharge electromechanically entrapped colloids and/or macromolecules to restore original
membrane properties. Current reversal removes foulants from the membrane and brings them
into the bulk solution.
Although the effectiveness of this procedure depends on the type of membrane and on
the nature of contaminants that foul the membrane, the reversal of hydrogen and hydroxyl
ion transfer and the accompanying interchange of pH effect at membrane surfaces is the
significant defouling factor.
When fouling substances are weak acids or bases, or amphoteric, the reversal in pH results
in a change in the polarity of substances. When polarity reversal is used for an ED batch
process, care must be taken on the reversal of hydraulic streams in order to prevent mixing
of diluting and concentrating solutions. This is generally achieved by a set of conductimetric
cells on emerging streams which automatically acts on three-way valves when they notice
a conductivity lower or higher than the fixed value. The water content of both solutions is
directly related to the hold-up volume of the stack compartments. So, the time reversal is
generally about V2 h.
Current pulsing is not used, even if it alleviates scaling problems and improves efficiency,
because it needs specific electric power sources, particularly in the case of high flow rates.
Thus, polarity reversal, which is used more often, causes the dispersion of concentration
polarization layer and reduction of slime and colloid deposit. Moreover, in the case of water
production, electrodialysis with polarity reversal (EDR process) eliminates the need of
acidifying the brine stream. However, polarity reversal must be supplemented with regular
chemical cleaning.

C. Cleaning
In spite of good operating conditions and management, a decrease in efficiency of an ED
stack is noticed over a long period of time. This is readily seen by some decrease in the
electric current through the stack. In the case of given diluting, concentrating, and electrode
solutions, if the current differs from its nominal value when the same voltage is applied, it
implies the occurrence of some irreversible polarization. In this case, it is necessary to use
stronger cleaning methods than the previously mentioned methods.
Generally, chemical cleaning is a good method. First, the stack is alternately washed with
an acid solution and an alkaline solution. These solutions are carefully made with an acid
or alkali which acts on mineral deposits or some organics. If there is organic matter, use of
enzyme is beneficial. However, acids, alkalis, enzymes, etc. used have to be carefully
selected to restore the nominal value of transport fluxes and to avoid any irremediable failure
of the stack materials, primarily the membranes, but also the spacer gaskets, electrodes,
ducts, etc. Care must be taken with the temperature and duration of each step of the cleaning
process.
When good sterilization is not used, or generally after long rest periods, the growth of
some algae or other microorganisms may be noticed. As the membranes, spacer gaskets,
and electrodes are simply clamped together in an ED stack, it is very easy to dismantle them
in order to subject them to a thorough cleaning. Contrary to some other membrane processes,
it is possible to brush ED membranes without damage.
In the case of biological fouling, the main problem is the cleaning of the ducts, pipes,
and associated valves.
218 Water, Wastewater, and Sludge Filtration

Table 1
PRODUCTION OF POTABLE WATER FROM BRACKISH WATER

Plant P 1 Plant II29

Plant location Foss Reservoir, Okla., U.S. Nodroma, Algeria

Area of the plant 22 x 12 m2
Design of equipment
Number of units (parallel) 18 2
Number of stacks (series) 4 4
Number of unit cells (per stack) 320 840
Capacity of the plant 11,355 m3/d 2,530 m3/d
Overall conversion 70% 85%
Membrane area 1,680 m2
Membrane utilization 75%
Flow rate 20 cm/s
Spacer gasket Tortuous Sheet
Current density according to stage 1-2-3-4 23-10-5-2 mA-cm 2
Voltage per unit cell per stage 1.5-0.9-0.8-0.75 volts
Governing Parameters
Type of ED Continuous Reversal
Process Brine recycle Continuous
TDS of feed 1800 mg/€ 2600 mg/€
TDS of dilute solution 300 mg/€ 150 mg/€
Reduction in each stage 35%
pH 7.5—8.0
Temperature 25— 30°C
Economy
Energy consumption (total) 2.4 kWh/m3
Pretreatment By lime softening process, clari­ Antifouling 1.2 g/m3
fication, dual media filtration with
sand and anthracite followed by
10 |xm cartridge filter and ad­
justment of pH to 7.5— 8.0
Posttreatment Addition of sulfuric acid and hex-
amethaphosphate into brine to
minimize Ca scaling; product
water is chlorinated
Cleaning Required if TDS of the product 1 h/d with HC1 5.6 g/m3
water exceeds 500 mg/€; cleaning
with 3% vol. of 50% caustic soda
sol and 5% NaCl for about 5 h

VII. APPLICATIONS

ED finds wide application in potable water, industrial water, and wastewater treatment.
Some of the specific applications are highlighted in the following tables (Tables 1 to 7).

VIII. CONCLUSION

ED is an attractive electromembrane process for desalting water or diluting or concentrating

ionic species in solution. As electricity is used as the driving force for separation, the process
may be used for wide-ranging separation problems. Moreover, since there is no significant
mass or energy hold-up in the stack, it may be started or stopped according to the needs.
If operating parameters are chosen properly, it may be used without any trouble over a long
period of time.
 219

Table 2 Table 3
DEMINERALIZATION OF DESALTING OF LACTOSE
INDUSTRIAL WATER24 FROM PERMEATE OF
SULFURIC CASEIN29
Plant Location, U.S.S.R.
Plant Location, New Zealand
Design of equipment
Number of units — 1 to 3 Design of equipment
Number of stacks — 2 per unit Number of units — 02
Number of unit cells — 200 per stack Number of unit cells — 1,200
Membrane area — 0.48 x 0.75 m2 Membrane area = 600 m2
Membrane utilization — 60% Capacity of plant = 20,000 kg/h
Capacity of unit — 500 m3/d Governing parameters
Flow rate — 25 cm/s Type of ED — Reversal
Spacer gasket tortuous path — 900 cm Process — Batch
Current density — 33 mA/cm2 Feed
Voltage — Up to 400 V Insoluble — 5%
Governing parameters Ash content
Type of ED — Continuous 13% of insoluble
Process — Continuous 7.2 g/kg of feed
TDS of feed — 4000 mg/€ Demineralization factor — 50%
pH — 6.5— 8.5 Deacidification — 30%
Economy Temperature — 35°C
Current efficiency ( tj) — 40— 60% Costs
Investment — U.S. $550,000
Operational — U.S. $0.025/kg of insoluble

Table 4 Table 5
DESALTING OF CHEESE DEMINERALIZATION OF SKIM MILK
WHEY FOR BABY FOOD22 23 29
Design of equipment
Plant Location, France Number of stacks — 4
Number of unit cells = 150 per stack
Membrane area = 50 m2
Design of equipment
Flow rate = 30— 50 cm/s
Number of stacks = 1
Governing parameters
Number of unit cells = 700
Demineralization factor = 90%
Membrane area - 350 m2
Temperature = 18— 20°C
Capacity of plant = 200 kg/h
pH = 4.6 to 4.8
Governing parameters
Economy
Type of ED — Reversal
Energy consumption (transport) 0.1 to 0.3 kWh/m3
Process — Batch
Feed
Insoluble — 20%
Ash content
8.5% of insoluble
17 g/kg of feed
Demineralizing factor — 81 %
Deacidification — 62%
Temperature — 40°C
Costs
Investment — $300,000 U.S.
Operational — $0.042/kg of insoluble
220 Water, Wastewater, and Sludge Filtration

Table 6
WATER REUSE IN ELECTROPLATING INDUSTRY

Plant I25 26 Plant II27

Feed solution Electroplating rinse of Rinse water in cyanide

nickel galvanization copper plating
(NiS04)
Plant location Japan Limoges, France
Design of equipment
Number of units 1
Number of stacks 1, coupled with electro- 1 per unit
deposition bath
Number of unit cells 50
Membrane area 0.48 m2 per unit 0.5 x 0.5 m2 per unit
Membrane utilization 70 kg Ni/d 10 kg Cu/d
Flow rate 3 cm/s
Recovery rate 90% 94%
Spacer gasket 2 mm thick Sheet
Current density 10 mA/cm2
Governing parameters
Type of ED Continuous Reversal
Process Continuous Continuous 8 h/d
TDS of feed 5— 7 g/€ 1000 mg/€
TDS of dilute solution 222 mg/€
Temperature and pH Same as electroplating
bath
Demineralization fac­ >100
tor
Economy
Energy consumption 1 kWh/kg Cu
Pumps 1 kWh/kg Cu
Transport
Net profit $0.33/kg Ni $4/kg Cu

Table 7
RECOVERY OF SODA IN PAPER
MILL FROM BLACK LIQUOR28

Design of equipment
Number of stacks = 2 in series
Residence time = 80 min
Current density = 10— 15 mA/cm2
Governing parameters: TDS of feed = 7— 10%
Economy
Energy consumption transport = 21— 80 kWh/kg
Posttreatment: Electrogravitational precipitation of lig-
nin
 221

NOTA TIONS

Roman Letters

a Activity
A Area m2
C Concentration kg m ~ 3
D Diffusion coefficient m2 s - 1
E Energy per unit volume Jm 3
F Charge of one mole of electron Cb-eq g _1
hj Hydration number moles of water per mole
of ion
i Current density amp m 2
J Flux density kg m - 2 s - 1
k Coefficient
K Constant
m Mobility s 1
n Number of moles
N Number of unit cells
P Power W
P Pressure Pa
Q Flow rate m3 s - 1
r Membrane resistance ohm m 2
R Gas constant J kg"1 K 1
s Partial enthalpy J kg - 1 K " 1
t Time S
T Temperature K
U Potential V
v Partial volume m 3 kg - 1
V Volume m3
Velocity m s _1
x,X Width, overall width m
y,Y Thickness, overall thickness m
Z Overall length m
Zj Charge number of ion j eq g-kg- 1

Greek Letters

jjij Electrochemical potential J-kg- 1

i|ij Permselectivity
c|> Internal electric potential V
r| Efficiency
A Variation in
8 Thickness of the concentration polariza­
tion layer

Subscripts

b Of brine
d Of dilute
e Of electrode
j of ion j
222 Water, Wastewater, and Sludge Filtration

NOTATIONS (continued)

Subscripts

1 Of bulk solution
m Of membrane
n Near the membrane
p Of polarization
r Of resistance
s Of salt, of electrolyte
S Of fixed charge
t Of transport
W Of water
uc Of unit cell
apm Of anion permeable membrane
cpl Of concentration polarization layer
cpm Of cation permeable membrane
ipm Of ion permeable membrane
jim Of membrane impervious to ion j
jpm Of membrane permeable to ion j
EOO Of electroosmotic
HYD Of hydration
1,2 Points
Superscripts

* Of reference
' Apparent
e Of inlet
s Of outlet

REFERENCES

1. Audinos, R. and Isoard, P., [Glossary of technical terms in membrane processes], SFF/IDEXPO, Cachan,
Paris, 1986.
2. Shaffer, H. and Mintz, M., Electrodialysis, in Principles of Desalination, Spiegler, K., Ed., Academic
Press, New York, 1966, chap. 6.
3. Pierrard, P., Recent progress in electrodialysis, Ind. Aliment. Agric., 93, 569, 1976.
4. Hattenbach, K. and Kneifeld, K., The effect of cell thickness and flow velocity on water cost in desalination
by electrodialysis, Desalination, 58, 33, 1986.
5. Maurel, A., Water desalination by electrodialysis, J. Genie Chimique, 2850-1 to 2858-12, Techniques de
l ’lngenieur, Paris, 1978; Bonnin, A., Electro-dialysis, Genie Chimique, J 2840­1 to 2840-21, Techniques
de l ’lngenieur, Paris, 1988.
6. Lacey, R. and Loeb, S., Industrial Processing with Membranes, Wiley-Interscience, New York, 1972.
7. Narayanan, P., Harkare, W., Adhikary, S., Dave, N., Chauhan, D., and Govindan, K., Performance
of an electrodialysis desalination plant in rural area, Desalination, 54, 145, 1985.
8. Hellferich, F., Ion Exchange, McGraw Hill, New York, 1962.
9. Molau, G., Heterogeneous ion-exchange membranes, J. Membr. Sci., 8, 309, 1981.
10. Meares, P., Membrane Separation Processes, Elsevier, Amsterdam, 1976.
11. Donnan, F., The theory of membrane equilibrium in the presence of a non dialysable electrolyte, Z.
Elektrochem., 17, 572, 1912.
12. Lakshminarayanaiah, N., Transport Phenomena in Membranes, Academic Press, New York, 1969.
 223

13. Winger, A., Bodamer, G., and Kunin, K., Some electrochemical properties of new synthetic ion exchange
membranes, J. Electrochem. Soc., 100, 179, 1953.
14. De Groot, R. and Mazur, P., Non Equilibrium Thermodynamics, North-Holland, Amsterdam, 1963.
15. Sun Tak Hwang and Kammermeyer, K., Membranes in Separation, Wiley-Interscience, New York,
1975.
16. Vetter, K., Electrochemical Kinetics, Academic Press, New York, 1967.
17. Cowan, D. and Brown, J., Effect of turbulence on limiting current, Ind. Eng.Chem., 51, 1445, 1959.
18. Wilson, J., Demineralization by Electrodialysis, Butterworths, London, 1960.
19. Audinos, R., Conductimetric determination of the limiting current for low Reynoldsnumbers inelectro­
dialysis, Electrochim. Acta, 25, 405, 1980.
20. Passino, R., Biological and artificial membranes and desalination of water, Pontificae Academiae Scien-
tiarum Scripta Varia, Elsevier, Amsterdam, 1976.
21. Applegate, L., Membrane separation processes, Chem. Eng., 91, 65, 1984.
22. Okada, K., Tomita, M., and Tamura, Y., Electrodialysis in the treatment of dairy products. II. Devel­
opment of electrodialysis, Milchwissenschaft, 31, 1, 1977.
23. Leitz, F. and Eisenmann, J., Electrodialysis as a separation process, Am. Inst. Chem. Eng., Symposium
Series, 77, 204, 1981.
24. Shishliannikov, L. and Alzhanov, F., Operating experience of EDU-series electrodialysis plants used in
different industries in the USSR, Desalination, 58, 77, 1986.
25. Jorgensen, S., Industrial Waste Management, Elsevier, Amsterdam, 1979.
26. Itoi, S., Electrodialysis of effluent from treatment of metallic surfaces, Desalination, 28, 193, 1979.
27. Audinos, S., Improvement of metal recovery by electrodialysis, J. Membr. Sci., 27, 143, 1986.
28. Radhamohan, K. and Basu, S., Electrodialysis in the regeneration of paper mill spent liquor, Desalination,
33, 185, 1980.
29. Societe de Recherches Techniques et Industrielles (SRTI), private communication, 1986.
 225

Chapter 12

VACUUM FILTRATION

Saravanamuthu Vigneswaran

TABLE OF CONTENTS

I. Introduction..................................................................................................................... 226

II. Operation and Types of Vacuum F ilters.................................................................... 226

A. Principle of Operation....................................................................................... 226
B. Different Types................................................................................................... 226

III. Design of Vacuum Filters............................................................................................. 227

A. Design Based on Experience...........................................................................228
B. Design Based on Specific Resistance of C ake.............................................. 228
C. Design Based on Filter Leaf Test....................................................................229

IV. Operating Param eters.................................................................................................... 231

A. Vacuum Level.....................................................................................................231
B. Degree of Drum Submergence........................................................................ 232
C. Drum Speed.......................................................................................................232
D. Filter M edium ................................................................................................... 232

V. Perform ance....................................................................................................................232
A. Sludge Solids Concentration............................................................................. 232
B. Pretreatment........................................................................................................ 232
1. Chemical Conditioning......................................................................... 232
2. Application of Filter Aids.................................................................... 233

VI. A pplications....................................................................................................................233
A. Water Treatment Plant Sludges........................................................................ 233
B. Domestic Wastewater Sludge............................................................................235
C. Industrial Waste S ludge.................................................................................. 235
1. Metal-Finishing Waste Sludge............................................................235
2. Pulp and Paper Industry Sludge.......................................................... 235
3. Other Types of Sludges........................................................................235

VII. Advantages and Disadvantages ofVacuum Filtration.............................................. 235

References....................................................................................................................................236
226 Water, Wastewater, and Sludge Filtration

I. INTRODUCTION

Vacuum filters have been used in wastewater treatment for dewatering of sludges for more
than half a century. A vacuum filter is a cylindrical rotating drum covered with a filter
medium, a portion of the circumference being submerged in the sludge to be filtered, and
water is drawn through the filter medium by an applied internal vacuum. The main purpose
of vacuum filtration is to reduce the sludge volume significantly by dewatering in order to
facilitate the subsequent sludge treatment or disposal.
Although the vacuum filter achieves higher yields than other dewatering processes (es­
pecially when compared to pressure filtration), use of vacuum filters may be declining
nowadays because of high operating costs resulting from chemicals for sludge conditioning
and energy consumption. Higher initial purchase costs are also required for these machines.
The comparison of three commonly used dewatering equipments are summarized in
Table 1.

II. OPERATION AND TYPES OF VACUUM FILTERS

A. Principle of Operation
Figure 1 shows the main component of a drum or scraper-type rotary vacuum filter. The
unit consists mainly of a horizontal cylindrical drum that rotates and is partially submerged
(from 25 to 30%) in a vat of raw (or conditioned) sludge. The drum is divided into sectors
spanning the length of the drum, each of which may be placed under vacuum by means of
automatic valving . 3 As each sector revolves through the sludge vat, a vacuum is applied,
resulting in the formation of a layer of sludge on the filter medium. The vacuum remains
on this sector as it emerges from the vat, resulting in the continuous drainage of moisture
from the sludge layer. Drainage continues until the section reenters the sludge vat. At this
point the de watered sludge cake is automatically removed from the filter medium . 3
Figure 1 also illustrates the various operating zones encountered during a complete rev ­
olution of the drum. These zones are for cake forming, cake drying, and cake discharging.
The submerged surface of the drum is referred to as the cake-forming zone. Vacuum applied
to submerged drum sectors causes filtrate to pass through the medium and cake to be formed
on the surface of the medium. As the drum rotates, each section is successively carried
through the cake forming zone, where the cake is formed within the filter vat, and de watered
outside the vat. The cake drying zone represents from 40 to 60% of the drum surface and
terminates at the point where vacuum is shut off. At this point, the sludge cake enters the
cake discharge zone where the cake is removed from the medium .3

B. Different Types
The design problems are less severe in vacuum filters than in centrifuge or pressure filters
because of the low driving force used on vacuum filters. This facilitates the operation of
vacuum filtration on a continuous basis. By far the greatest number of applications for
vacuum filters is on continuous basis. However, batch filters have found their use in other
areas.
Following are the two types of batch vacuum filters used: vacuum leaf filter and vacuum
nutsche. Batch filters become suitable when process conditions change frequently, causing
the need for variable filtration or cake washing times .4
Continuous filters operate without interruption where the feed is fed continuously and the
filtrate and de watered sludge is continuously discharged. There are four types of continuous
vacuum filters: rotary drum, rotary disc, rotary horizontal, and horizontal belt or endless
cloth vacuum filters. This chapter will not review the different types in detail, as this topic
is discussed in detail in various literature. 3,4
 227

Table 1
DEWATERING EQUIPMENT COMPARISON

Vacuum Pressure
filter filter Centrifuge

Capital costs 3X 2X X
Operating costs (energy, chemicals, 3X 3X 2X
maintenance, labor)
Cake solid concentration 2X 3X X
Solid recovery 3X 3X X

From Spinosa, L. and Eikum, S., Characterization, Treatment and Use of

Sewage Sludge, L ’Hermite, P. and Oh, H., Eds., D. Reidel Publishing Com­
pany, Holland, 1981, 69. ©1981 ECSC, EEC, EAEC, Brussels and Luxem­
bourg. With permission.

FIGURE 1. Main components of a drum rotary vacuum filter. (From Svarovsky, L., Chem. Eng., July 2, 1979,
p. 62. With permission.)

III. DESIGN OF VACUUM FILTERS

The design of vacuum filter can be carried out in three ways as indicated below:

• Design based on experience

• Design based on filter leaf test
• Design based on specific resistance of cake
228 Water, Wastewater, and Sludge Filtration

Table 2
SPECIFIC RESISTANCE OF
SEWAGE SLUDGES

Sludge type r(m kg-1)

Raw sludge 10— 30 x 1013

Raw sludge after coagulation 3— 10 x 1011
Digested sludge 3— 30 x 1013
Digested sludge after coagu- 2— 20 x 1011
lation
Activated sludge A— 12 x 1013

From Coackley, P. and Wilson, F., Filtr. Sep.,

8, 61, 1971. With permission.

A. Design Based on Experience

The easiest way of designing a vacuum filter is making use of the data available from
previous experiences. Data available on similar sludge (with the same chemical conditioning)
can be used. As a guideline, typical filter yield values for different sludges are presented
in Table 2.

B. Design Based on Specific Resistance of Cake6,7

During the sludge dewatering by filtration, a cake of increasing thickness is built up on
the medium surface. In practical application of the vacuum filtration process, the hydraulic
resistance of the medium (Rf) is assumed to be negligible compared to that of the sludge
cake. The cake can be considered as a water-saturated porous medium through which the
liquid flows. Therefore, applying Darcy’s law,

dV _ AK AP
dt ~ |x L

dV AP
where — is the volumetric flow rate; — is the pressure gradient in the flow direction; A
dt L
is filter area; K is the Darcy coefficient of permeability; L is the sludge cake thickness; and
(x is the viscosity
It ismore convenient infiltration theory to correlate filtration rate and pressure gradient
by means of hydraulic resistance,in place of hydraulic permeability.

If one defines the resistance as R = ^, then Equation 1 becomes:

™ = A 4P
dt (J.LR

In a vacuum filter, although the major resistance is contributed by the filter cake, inclusion
of the resistance by the filter medium in Equation 2 gives,

dV _ APA
dt ~~ |x(LR + Rf) (3)

where Rf is the resistance of filter medium.

 229

The volume of the cake can be expressed as

LA = vV (4)

where v = volume of cake deposited per unit volume filtrate. Substituting for L in
Equation 3:

dV _ APA2
dt ~~ |jl(RvV + RfA)

It is more convenient to express the quantity of cake in dry weight (W) per volume instead
of volume of cake per volume of filtrate . 6 Similarly, the cake resistance can be represented
by r (resistance per unit weight) instead of R (resistance per unit volume). Thus:

dV _ APA2
dt ~~ |x(WrV + RfA)

where W = weight of dry cake solidsper unit volume of filtrate, and r = specific resistance.
Assuming constant pressure over a time t, and integratingEquation 6 with respect to
time t,

l = p.rW V jx R f
V 2APA2 APA

Equation 7 is a straight line of type y = a,x + a2, where

P - rW , P-Rf
a, = —----- ­ and a7 = ------­ (8 )
1 2APA2 2 AP• A

It should thus be possible to calculate the specific resistance of filter cake from an
experimental study with Buchner funnel apparatus (Figure 2) by measuring the filtrate volume
2 AP A 2aL
at various times, t. One can find the specific resistance (r = ------------­ ) from the linear
(xW
plot of t/V vs. V.
The experiment is conducted by pouring a constant volume of sludge into the funnel with
filter paper and imposing the vacuum at time zero. The amount of filtrate is then recorded
at various times. The pressure (AP) is measured with a vacuum gauge.
The cake deposited per volume of filtrate, W, can be calculated from Equation 9, which
is derived from the material balance equation.

C C
W = --------— -----­ (9)
100(CK - CG)

where Ck = cake solids concentration (%), and C0 = feed solids concentration (%).

C. Design Based on Filter Leaf Test6,8

Rotary vacuum filters are frequently designed and selected on the basis of leaf testing to
determine attainable filtering and washing rates.
A typical cycle of a vacuum filter consists of 30 sec of submerged operation, 60 sec of
drying under vacuum but not submerged, and 30 sec off the filter. In other words, 25% of
230 Water, Wastewater, and Sludge Filtration

VACUUM

FIGURE 2. Buchner funnel experimental set-up.

the drum circumference is submerged, and 25% is not covered by the fabric as the cake is
discharged. Sludge drying occurs on the remaining 50%.
Such a cycle can be simulated with a filter leaf, which is a small model of the prototype
filter (Figure 3). The filter leaf consists of a round disc of about 10 cm in diameter, over
which the filter medium is placed. This disc is connected to a vacuum source, through a
graduated cylinder which is used to collect the filtrate. The filter leaf is kept in a beaker
containing the sludge to be dewatered for a specified submerged time (usually in the order
of 30 sec). The filter leaf is then taken out of the beaker containing sludge and kept under
vacuum for a time period equal to the time that the filter would experience drying in prototype;
the vacuum is then turned off. The cake is scraped off the filter leaf and analyzed for solids
content, and the toal dry cake solids produced is calculated while the filtrate is analyzed for
suspended solids and thus the solids recovery.
The solids recovery in any dewatering device is calculated from a solids balance. If the
feed flow rate and solids concentration are Q0 and CG, the filter cake flow and solids
concentrations are Qk and Ck, and the filtrate flow and solids concentrations are Qf and Cf,
one can write the following equations.
From liquid flow balance,

Qo ~ Qf + Q k ( 10)

From solids flux balance,

QoCo = QfCf + q kc k (I D

Substituting Equation 10 in Equation 11,

Qo(C0 - Cf)
Qk - (12)
CK Cf
 231

FIGURE 3. Laboratory-scale leaf filter model. (From Cheape, D. W., Jr., Chem.
Eng., June 14, 1982, p. 141. With permission.)

Percent recovery of dry solids (% R) is

mass of dry solids as cake
x 100
mass of dry feed solids

- CKQk x 100
KVK
C0Q0

Ck(Cq - Cf)
x 100 (13)
Co(CK - Cf)

A great advantage of the filter leaf test is that the same medium used in the prototype is
also used in the filter leaf test. This leads to a realistic approximation of the prototype
operation . 6

IV. OPERATING PARAMETERS

The selection of vacuum level, degree of drum submergence, drum speed, and medium
type are very important in order to obtain optimum performance.

A. Vacuum Level
The vacuum applied affects the degree of de watering of sludge. Although the vacuum
imposed on the sludge can be adjusted, it usually does not exceed 50 cm of Hg (or 6 8 kPa),
232 Water, Wastewater, and Sludge Filtration

due to the power costs and compressible nature of wastewater sludges. Higher vacuums tend
to produce nonpervious cakes which decrease the sludge yield.

B. Degree of Drum Submergence

The higher the submergence of the drum, the greater the percentage of cycle time for the
pickup of solids on the filter surface. This will result in large cake thicknesses, but the
moisture content of sludge will be high. On the other hand, decrease of drum submergence
tends to decrease the time devoted for pickup of solids which results in production of thinner
but drier cakes.

C. Drum Speed
An increase in drum speed results in the reduction of sludge-drum contact time, which
will lead to the production of cake with higher moisture content. On the other hand, a slower
drum speed will result in a drier cake but with a lower filter yield.

D. Filter Medium
The filter medium is of a fabric type which is used to cover the drum. The fabric can be
categorized as an open or tight medium. Open media have large pores (or openings), while
tight media have small openings. The smaller the opening, the higher the removal of fine
particles. However, the opening cannot be so tight as to resist the flow through the medium.
Common media used are rayon, acrylic, polyesters, poly olefins, wire screens, and stainless
steel coils. Although stainless steel coils have the advantage of resisting wear and tear, they
are more expensive. The ideal medium should be chemically resistant to the materials to be
removed while providing minimum resistance to filtrate flow.

V. PERFO R M A N C E

Like the other types of mechanical dewatering equipments, optimum performance of

vacuum filters also depends on the type of sludge and its solids concentration, conditioning,
and filter operating conditions.

A. Sludge Solids Concentration3

The sludge solids concentration has a significant influence on the filter yield. An increase
in the feed sludge solids concentration usually results in a substantial increase in filter yields.
A practical upper limit of 8 to 10% is used because, at greater solids concentrations, the
sludge pumping becomes difficult and costly. The practical lower limit is kept at 3% because,
below this concentration, it is difficult to produce sludge filter cakes.
In practice, the sludge from various processes (both aerobic and anaerobic processes) have
low solids concentration, and pretreatment of the sludge is necessary for successful appli­
cation of vacuum filtration.

B. Pretreatment
7. Chemical Conditioning
Sludge is conditioned by biological, chemical, and/or physical treatment to enhance the
de watering characteristics of sludge. A variety of physical methods for altering sludge
characteristics are available to facilitate the de watering operation, like heating, freezing of
sludge, use of admixtures, ultrasonic vibrations, and solvent extraction. All these can be
used, although none of them are as yet in common use when compared with chemical
conditioning .9 Particle size is considered to be an important parameter affecting the de-
waterability of sludge. The primary objective of chemical conditioning is, therefore, to
 233

increase the particle size by adding chemicals which enable the particles to agglomerate into
fewer large particles, or floes. The formation of such floes aids the dewatering process of
the sludge. The common conditioning chemicals (or coagulants) for wastewater sludges are
FeCl3, Fe2 (S 04)3, alum, and lime. Before coagulants can combine with the solid fraction
of the sludge, it must satisfy the coagulant demand of the liquid fraction . 10 This is especially
true when the alkalinity of the sludge is excessive. As a precipitant of bicarbonate (alkalinity),
lime may be substituted for the portion of the coagulant that combines with the liquid
fraction. It should be noted that lime forms only a precipitate with the fraction and does not
form floes.
Coagulant or conditioner requirements should first satisfy the liquid-fraction requirement
approximated by the stoichiometry of the chemical reaction, i.e., for FeCl3. 10

2FeCl3 + 3Ca(HC0 3) 2 - > 2Fe(OH ) 3 ( j ) + 3CaCl2 + 6C 0 2 ( | )

1 mg/1 as C aC 0 3 alkalinity requires (2 X 162)/(3 X 106) =

1.08 mg/1 of FeCl3

The conditioner requirement should also satisfy the solid fraction requirement, which is
a matter of experience.
It should be noted that the coagulant demand of the liquid fraction (or alkalinity) can be
reduced either by lime addition (as a precipitant) or by washing out the alkalinity with water
of low alkalinity. This process is called elutriation . 10
Commonly, FeCl3 up to 2.5% of the weight of dry solids is used to condition raw or
digested municipal sludges, but up to 7% is used for activated sludge. In addition, approx­
imately 7 to 10% lime may be required. But, if the sludge is elutriated, the required FeCl3
may be reduced by as much as 80% and lime addition may not be necessary. Table 3 gives
the typical conditioner doses used for various sludges. While conditioning with FeCl3 and
lime is one of the most typical practices, the use of organic polymers or polyelectrolytes is
gaining in popularity, although the use of FeCl3 and lime will result in the production of
drier cakes. There are instances where the total quantity of water present in both sludge
solid cake and chemical solid cake produced from FeCl3 and lime conditioning is greater
than that produced from polymer conditioning.

2. Application o f Filter Aids

Many industries have adopted vacuum filtration process, using diatomite filter aids, in
order to reduce the cost of filtered products. Diatoms consisting predominantly of silica
which are insoluble in the process liquors form a strong filter cake/matrix, of great porosity
that allows liquids to pass through at a fast rate, but which traps suspended solids on the
outer surface, thus preventing cake penetration into the filtrates. 11
Hence, by the use of filter aids, very high flow rates or high de watering rate can be
obtained.

VI. APPLICATIONS

Vacuum filtration is commonly used in de watering of various kinds of sludges. Some of

them are highlighted below.

A. Water Treatment Plant Sludges

In water treatment plant, sludge is produced in different steps:
 Table 3
PERFORMANCE OF VACUUM FILTRATION FOR DIFFERENT TYPES OF SLUDGES12

Chemical
Total dosage Cake moisture Filtrate
solids Volatile Filter as % dry feed Pressure content rate
Type of sludge (%) (%) cloth sludge solids (mmHg) (%) (kg/m2-h)

Raw 4.0 54 Polypropylene Ca(OH)2 30 400—500 77 18.0

FeCl3 10
Digested (elutriated) 4.0 47 Polypropylene Ca(OH)2 45 400 —500 80 15.0
FeCl3 10
Activated 2.0 72 Polypropylene Ca(OH)2 60 400—500 83 8.0
FeCl3 18
Food industrial activated 2.3 — Polypropylene Ca(OH)2 20 400—500 86 17.6
Fly ash 20 — Polypropylene — 400— 500 32 21.6
Water, Wastewater, and Sludge Filtration
 235

• Alum sludge, from coagulation-sedimentation processes, is bulky and gelatinuous with

low solids content that is difficult to de water; solid content in this sludge is about 0 .1
to 0.5%
• Sludge from filter backwash, which has the solid content of 100 to 300 ppm

Water treatment plant sludge is inorganic in nature and does not exert any oxygen demand.
Clay or lime may be used with the alum as conditioning agents to improve thickening and
dewatering properties of this sludge.

B. Domestic Wastewater Sludge

The major application of vacuum filter is in this field since the mid-1920s, although the
era of vacuum filtration is declining. Table 3 gives the performance of vacuum filters for
different sludges. 12

C. Industrial Waste Sludge

1. Metal-Finishing Waste Sludge
The metal finishing process, for example, involves a wide variation of pH, in addition
to the presence of toxic heavy metals. Main waste results from rinsing and periodical
discarding of baths. The treatment methods which produce sludges include oil separation,
reduction of chromates, oxidative destruction of cyanides, and precipitation of metal hy ­
droxides from the treated waters. Average discharge solids from a vacuum filter is about
2 0 to 25% dry substance, and filtrate having high turbidity must be sent back to the clarifier
or polished. Feed concentration must be 4 to 8 % solids for successful operation of vacuum
filter. Therefore, a thickening tank is used in between clarifier and vacuum filter.

2. Pulp and Paper Industry Sludge

Sludges are obtained from coarse screening, SS removal, BOD removal, and color re ­
duction. Most of the sludges come from SS and BOD removals. Thickening precedes
de watering. Drum belt or coil types of vacuum filters are commonly used in dewatering.
Sludges from pulping and paper making produces cakes varying from 20 to 30% solids. A
filter rate of 30 kg/m2*h is normally used for dewatering of primary waste solids. Biological
wastes may be combined with primary wastes for vacuum filtration. Chemical addition aids
the vacuum filtration process. Chemical conditioners, such as ferric chloride, alum, or
polyelectrolytes, are observed to double the capacity of the vacuum filter for poorly filterable
sludges. Such chemical treatment is generally necessary when activated sludge is included
in the sludge to be de watered.

3. Other Types o f Sludges

In the inorganic chemical industry, for the complete removal of harmful solid wastes such
as metal (mercury, arsenic, etc.), sulfates are removed (recovered) by vacuum filtration.
Sludge from the titanium dioxide industry, cane sugar refining processes, the glass man­
ufacturing industry, etc., could also be de watered using vacuum filtration.

VII. ADVANTAGES AND DISADVANTAGES OF VACUUM FILTRATION

Advantages

• The proportion of solids in sludges is increased (almost doubled), which will result
in the reduction of sludge volume, thus the transport cost. Sludge handling, too,
becomes easier.
• Incineration costs are reduced by the increase in calorific value of sludge through the
reduction of moisture.
236 Water, Wastewater, and Sludge Filtration

• The reduction of moisture reduces the possibility of anaerobic digestion and, thus,
odor problems.
• The disposal through landfill is facilitated because of the reduced possibility of con ­
taminant leachate.
• It is compact in size and hence suitable for congested localities.
• It results in good performance in de watering digested sludges.
• It is a convenient process for the operator because of minimum contact with the sludge.
• Labor requirements for this process are minimal.
• It is a continuous process (stopped only for maintenance work).
• It is flexible in handling different types of sludges because the optimum operating
conditions may be determined through laboratory studies, such as filter leaf test, for
different sludges.
• It is well suited for hazardous wastes too.

Disadvantages

• It is not comparable to sand-bed treatment if land is available.

• It requires a high initial investment.
• Expert operating skills are needed.
• Chemical conditioning is needed in general and is a must for fresh sludges.
• Periodic maintenance of vacuum filtration is essential.
• It is not a complete process and further sludge treatment is needed before ultimate
disposal.
• It is not suitable for smaller towns and rural areas because of the lack of expertise and
high initial investment.
• Clogging of the filter is a common problem.
• Filter yield in vacuum filtration reduces with time because of a gradual increase in
clogging.
• Filter material may deteriorate due to certain sludge characteristics.
• Closer monitoring of operating parameters is essential for effective performance.

REFERENCES

1. Spinosa, L. and Eikum, S., Dewatering of municipal sludges, in Characterization, Treatment and Use
of Sewage Sludge, L ’Hermite, P. and Oh, H., Eds., D. Reidel Publishing, Holland, 1981, 69.
2. Svarovsky, L., Advances in solid-liquid separation I, Chem. Eng., July 2, 62, 1979.
3. Process Design Manual for Sludge Treatment and Disposal, EPA 625/1-79-011, Office of Technology
Transfer, U.S. Environmental Protection Agency, Washington, D.C., 1979.
4. Egee, L. P., Vacuum filtration, in Process Equipment Series, Vol. 1, Bhatia, M. V. and Cheremisinoff,
P. N., Eds., Technomic Publishing., Lancaster, PA, 1979.
5. Coackley, P. and Wilson, F., Flocculation with special reference to water and wastewater engineering,
Filtr. Sep., 8, 61, 1971.
6. Vesilind, A. P., Treatment and Disposal of Wastewater Sludges, Ann Arbor Science, Ann Arbor, Mich.,
1979.
7. Coackley, P. and Jones, B. R. S., Vacuum filtration I, Sewage Ind. Wastes, 28, 963, 1956.
8. Cheape, D. W., Jr., Leaf tests can establish optimum rotary-vacuum filter operation, Chem. Eng., June
14, 141, 1982.
9. Weber, W., Jr., Physico-chemical Process for Water Quality Control, Wiley Interscience, New York,
1972, 550.
10. Fair, G. Mi., Geyer, J. C., and Okun, D. A., Elements of Water Supply and Wastewater Disposal, John
Wiley & Sons, New York, 1971, 608.
11. Basso, A. J., Vacuum filtration using filter-aids, Chem. Eng., April 19, 1982, 159.
12. Sugaya, K., private communications, Ishigaki Mechanical Industry Co., Ltd., Tokyo, 1987.
 237

Chapter 13

PRESSU RE FILTRA TIO N

Saravanam uthu Vigneswaran

TA BLE OF CONTENTS

I. Introduction.................................................................................................................... 238

II. Principle and Theory...................................................................................................... 238

III. Different T ypes...............................................................................................................240

A. Plate and Frame Presses and Recessed-Plate P resses..................................240
B. Sheet Filters....................................................................................................... 240
C. Shell and Leaf Filters .......................................................................................241
D. Variable Volume Filters................................................................................... 241

IV. Design Parameters........................................................................................................ 243

A. Pressure D ro p .................................................................................................... 243
B. T emperature....................................................................................................... 244
C. Initial Mass F lu x ............................................................................................... 244
D. Filtration T im e .................................................................................................. 244
E. Downtim e...........................................................................................................244

V. Applications of Pressure Filters................................................................................... 245

References....................................................................................................................................247
238 Water, Wastewater, and Sludge Filtration

I. IN TRO D U C TIO N

Pressure filters are used for solid-liquid separation of sludge by the application of pressure
on the solid-liquid mixture in order to squeeze out the liquid through a filter medium. The
objectives of pressure filtration may be either:

1. To get a cake of higher solids content of 30 to 80% from slurries with a solids content
^ 5 to 40%, or
2. To obtain a very clear filtrate from a liquid containing very fine particles (=^0.05 pm ) 1

The filter medium, which retains the solids and passes the liquid through it, must withstand
the stress and strain during the application of high pressure. It is generally made of a cloth
of natural or synthetic fibers, coil springs, a wire mesh fabric, or sheets made of particulate
or fibrous materials such as diatomite, asbestos, or glass fiber bonded together with epoxy
resin for added strength.

II. PRIN C IPLE AND TH EO RY

When pressure is applied to a slurry, the liquid flows out from the pores of the particles
and through the filter medium, which acts as a support. Particles deposited on the filter
medium increase the thickness of the cake. Some of the particles may even penetrate the
medium, depending on the pore size of the medium. Retention of particles by the filter
medium and the growth of cake thickness increases the medium resistance to the flow of
liquid. If applied pressure is kept constant, then the feed flow rate will decrease. However,
constant feed flow rate could be maintained by either increasing the pressure or by limiting
the cake thickness on the filter medium.
The basic filtration equation, as derived from Darcy’s Law, is as follows:2

AP • A
Q = ( 1)

where Q is the volumetric flow rate of the feed suspension; V is the total volume of filtrate
passed through the filter in time t; A is the face area of the filter; Ap is the applied pressure
drop across the cake and filter medium, which may vary with time; |x is the filtrate viscosity;
Rf is the medium resistance; rav is the average specific resistance of sludge; and C is the
feed concentration.
If no loss of volume in the cake is assumed,

For higher feed concentrations, the volume of the feed slurry and the filtrate differ signif­
icantly. Therefore, a correction based on the effective concentration is necessary as follows:2

C,'C orrected (3)

1 _ (m - o
C ps p
 239

where m is the mass ratio of wet to dry filter cake; ps is the solids density; and p is the
liquid density.
The above correction is required in order to express the mass of solids in terms of the
filtrate volume.
Although the medium resistance (Rf) is theoretically constant, in practice, it varies with
time due to the penetration of solids onto the medium. The specific cake resistance is
dependent on the approach velocity, feed concentration, applied pressure drop, and the
degree of flow consolidation that the cake undergoes with time. It decreases with velocity
and feed concentration . 2
The effect of pressure on specific resistance (r) is conventionally expressed as:

r = r0(Apc)n (4)

where n is a constant of compressibility which takes a value of 0.5 to 1 . 1 ; Apc is the pressure
drop across the cake; and rGis the cake resistance per unit applied pressure drop.
Because each layer in a cake is subjected to a different pressure drop, an average value
for specific resistance (rav) has to be taken. This value is defined as follows:2

AP,
(5)
f Ap< d(Pc)
Jo r

Substituting Equation 4 in Equation 5,

rav = (1 - n)r0(APc)n (6 )

In Equation 1, substituting Equation 6 for rav and Equation 2 for Q,

dV AP • A
(7)
dt V
(1 - n)r0(APc)nc — + |xRf

At constant filter area (A) and pressure, integration of the above equation gives:

£ = (1 - n)r0(APc)"c|i. A + jiR f
V 2APA2 AAP

The values of rG, n, and Rf could be determined from pilot-scale filtration tests, bomb-filter
tests, or from a compression-permeability cell .2 Hence, Equation 1 may be solved for any
mode of operation, such as constant pressure, constant rate, or variable rate and variable
pressure, which are discussed in the latter sections.
One could use the following equation to calculate the pressing time (more exactly the
length of time to complete a press) which assumes that filtration continues until the chambers
of the press are full of cake . 3

= 0.321n)d2(Cf - C„)
APCo(100 - C J

where T is the length of time to complete a press (h), d is the distance between cloths in
cm; C0 and Cf are the sludge solids % during the initial and final stages; r is the specific
240 Water, Wastewater, and Sludge Filtration

resistance of the sludge (in cm/g); AP is the pressure in bars; and j] is the viscosity of the
filtrate in centipoise.

III. D IFFER EN T TYPES

Pressure filters are classified into two groups, namely, batch and continuous pressure
filters, based on their mode of operation. Even though the latter type of filters is desirable,
they are not common in practice because of their complex nature and high cost. The ad­
vantages of batch pressure filters are

• Rapid filtration of fine particles, which would otherwise be at a low rate

• The compact nature of the filters, with a high filtering area per unit plant space occupied
• The flexibility of operation provided by them at a relatively low initial cost1

The disadvantage with the batch pressure filter is its high operating cost, especially when
manually operated.
Batch pressure filters are operated in step-wise sequence for each cycle as mentioned
below:

• Forming the filter cake

• Removing unfiltered slurry
• Washing the filter cake
• Compressing the filter cake to remove excess liquid
• Blowing air or gas through the cake to discharge excess liquid
• Discharging the cake

Batch pressure filters can be grouped into three different types:

1. Filter presses (e.g., plate and frame, recessed plate, and sheet filters)
2. Leaf, plate, candle (tubular) filters (e.g., shell and leaf filters)
3. Variable-volume filters (e.g., membrane filters)

Cartridge filters and strainers, too, are pressure filters, but they are usually used online
to remove small amounts of solids from liquid streams.

A. Plate and Frame Presses and Recessed-Plate Presses

Filter presses consist of plates and frames in alternate sequence and can move along a
guide (Figure 1). Plates have corrugated surfaces over which a filter medium, normally a
woven cloth, is placed. Slurry feed enters through ports, and liquid flows out through the
filter medium. There are various physical combinations in feeding slurry, drain, and guide.
Another variation is the recessed-plate filter press which does not have any frame; the plate
itself is recessed to allow the formation of cake. The advantage of plate and frame filter
presses is that cake thickness could be varied by additional frames, which is important when
handling slurries having variable characteristics. Compatibility with mechanized discharge
and minimized leakage problems are the only advantages of recessed plate filter presses.
Feed pressures for filter presses range from 690 kPa to as high as 6900 kPa.

B. Sheet Filters
Thin sheets (2 mm to 6 mm ) 1 are used to trap fine particles in the pores, similar to that
in a deep bed filter, in order to obtain a clear filtrate. Trapped particles are flushed out and
sheets are used repeatedly until they produce a filtrate acceptable in quality.
 241

FIGURE 1. Plate and frame filter press. (From Moir, D. N., Chem. Eng., 89, 47, 1982. With
permission.)

C. Shell and Leaf Filters

In this category of filters, a number of hollow filter elements are used to separate solids.
A closed vessel houses the filter elements on which solids are deposited, while liquid flows
into the elements and out of the vessel. Major advantages of this type of filter are that it
could be used for inflammable and toxic substances, as the filtrate and cake are contained
in a closed vessel. The arrangement of a closed vessel also facilitates heating or cooling
during, or at the end of, the filtration cycle . 1 These filters are classified according to the
orientation of filter elements and their shape.

D. Variable Volume Filters

This class of filters has the additional feature of compressing the cake formed at the end
of pressure filtration, thus producing drier cake which is easier to handle. Compression is
achieved by means of inflating the membranes towards the filter medium once the feed is
stopped (Figure 2). Compression of cake in this type of filter is more uniform; thus, cake
washing is efficiently done. Membranes are used in rectangular and cylindrical shape filter
presses.
Another type of filter using compression to reduce the volume is the belt-filter press
(Figure 3). Two acutely angled belts compress the slurry within their narrow gap, their edges
being sealed by special belts. The cake is discharged at the end of a converging section.
The belts can be either vertical or horizontal. Cake solids concentration ranges from 35 to
60% and belts of width up to 2.5 m are available. 3 The operational cost of the belt-filter
press is much lower than that of other types of pressure filters, although it does not lead to
high solid recovery and cake solid concentration.
Filtering Compression Cake Discharge Washing of -fe.
Filter Cloths ts>

High p ressure
Water, Wastewater, and Sludge Filtration

FIGURE 2. Operating cycles of variable volume filter.7

 243

FEED

■FILTRATE

RIBBON OF
FILTER CAKE

FIGURE 3. Belt-filter press. (From Svarovsky, L., Chem. Eng.,

89, 62, 1979. With permission.)

IV. D ESIG N PARAM ETERS

In the design of pressure filters, as the characteristics and flow rate of slurry to be dewatered
are known, the pressure filter can be designed to produce the required quality of cake. The
following variables are to be chosen in order to obtain optimum performance of a pressure
filter.

• Pressure drop
• Slurry temperature
• Initial mass flux
• Filtration time
• Downtime

Apart from the above variables, the filtration constant for a given set of slurry, filter
medium, flow rate of slurry, and operating pressures has to be known, and can be calculated
from pilot-scale testing. To get reliable data, the slurry to be used in pilot-scale testing
should be similar to that to be dewatered in full-scale plant. The flow rate and operating
pressure to be tested should also cover the full-scale values to be used.
The batch pressure filters could be operated chiefly in the following modes: constant
pressure only, constant rate only, or constant rate followed by constant pressure. In all batch
filters, removal of cake, with or without washing, and precoating of the filter medium are
done during each cycle. This time is called the downtime.

A. Pressure Drop
The output per unit area of filter (mass flow rate) will increase with the pressure drop.
When selecting a pressure, it should be the maximum possible for a given material of
construction and for a given configuration of the filter press. Pilot-scale testing should be
conducted to verify whether the cake is effectively dewatered at the design pressure drop.
244 Water, Wastewater, and Sludge Filtration

INITIAL MASS FLUX

FIGURE 4. Typical average mass flux vs. initial mass flux curve.

B. Temperature
The work required to dewater the sludge includes the necessary work against viscous
forces during the expulsion of liquid. Increasing the temperature will decrease the viscosity
and the viscous forces, making the process of filtration easier. However, the design tem ­
perature should be chosen such that the slurry and the filter medium are not adversely
affected. The possibility of exploiting the waste heat by heat recovery or by cooling in
another process should be investigated, but heating purely for the purpose of increasing the
mass flow rate may not be economical.

C. Initial Mass Flux

Brown5 derived the following equation for average mass flow rate (W) (liquid) during
the total filter cycle,

_ A[2xK * AP0f - (KAP/G , ) 2] 1 2

w = (i u)
(ef + ed)x

where A is the area of the filter, m2; x is the mass fraction of the solids in the slurry
(expressed as a fraction of the liquid in the slurry); K is the filtration constant, kg m “ 3-s;
Ap = pressure drop, N/m2; 0f = filtration time, h; 0d = downtime, h; G! = initial mass
flux, kg/(m 2-h).
Figure 4 shows the effect of initial mass flux on the maximum average mass flux
(W/A)max for a given pressure drop (based on Equation 10). Once the pressure drop across
the filter has been decided, the initial mass flux could be chosen from Figure 4 such that it
is closer to the value of (W/A)max when the curve becomes asymptotic.

D. Filtration Time
Figure 5 shows the effect of filtration time on the average mass flux (based on Equation
10). This figure shows the existence of an optimum (W/A)max for a particular filtration time.
Therefore, filtration time can be chosen such that the average mass flux (for the given G,)
is maximum.

E. Downtime
As the downtime generally depends on the technique of cake removal, cake washing, and
precoating the filter medium, it is difficult to manipulate it once a filter has been constructed.
Careful choice of minimum downtime is a must to increase the output of the filter press.
 245

FIGURE 5. Typical average mass flux vs. filtration time curve.

Table 1
INDUSTRIAL APPLICATIONS OF DIFFERENT TYPES OF
PRESSURE FILTERS

Industry Application Type of filter

Chemical Separation of organic and inorganic Plate and frames; leaf,

salts from suspension; removal of un­ plate, and candle
dissolved particles; separation of un­
treated reagents; catalysts; liquid
clarification
Ceramics To prepare consistent raw materials for Recessed plate filter press
production
Coal wastes Recovery of coal and clay suspension
Dye stuffs Separation of precipitated materials Recessed and plate and
frame filter press
Wastewater treatment Dewatering sludges for disposal; pol­ Recessed plate, plate and
ishing of effluent frame filter presses, belt
filter press
Fats and food produc­ Squeezing, refining Recessed, plate and frame
tion
Kaolin, chalk Concentration before drying Recessed plate, tube
presses
Sugar industry Refining Circular plate, membrane
Paper industry Treatment of paper mill sludge Belt filter press

V. A PPLICA TIO N S O F PRESSU R E FILTERS

Many industries use filter presses either in their production or in their sewage treatment.
Different filter presses are used for similar industrial applications, due to the fact that
characteristics of slurries vary; also, various filter press designs have wide-ranging appli­
cations. Selection of a particular type of filter press depends mainly on the filtering char­
acteristics of slurry, physical and chemical properties, compressibility, and chemical nature
(organic, or inert), besides the objective of solid-liquid separation. Table 1 lists industries
and their application of pressure filters with the type most commonly used .6 The performance
of the pressure filter for different types of sludges is summarized in Table 2.
 Table 2
PERFORM ANCE OF PRESSURE FILTRATION FO R DIFFERENT TYPES OF SLUDGES 7

Chemical dosage Cake

as % dry feed
Total moisture Average
solids Filter sludge solids Pressure (kPa)
Volatile content mass flux
Type of sludge (%) (%) cloth Chemical Percent Filtration Compression (%) (kg/m2*h)

Raw 4.0 55 Polypropylene Ca(OH)2 25 294 1470 55 3.7

FeCl3 9
Digested (elutriated) 4.0 48 Polypropylene Ca(OH)2 40 294 1470 55 4.4
FeCl3 10
Activated 2.1 75 Polypropylene Ca(OH)2 55 294 1470 68 2.0
FeCl3 18
Water supply 5.1 _ Polypropylene 490 1470 52 2.5
Industrial waste 3.4 — Polypropylene Ca(OH)2 30 490 1470 68 3.3
Food industrial activated 2.6 — Polypropylene Ca(OH)2 20 490 1470 78 2.1
FeCl3 10
Chemical industrial activated 2.2 _ Polypropylene Ca(OH)2 16 490 1470 77 2.8
FeCl3 20
Water, Wastewater, and Sludge Filtration

Broken stone waste 39.0 _ Polypropylene 490 1470 23 20.0

Aluminum hydroxide 3.0 — Polypropylene — 490 1470 75 1.7
Chemical sedimentation 3.8 — Polypropylene — 490 1470 58 2.1
 247

REFEREN CES

1. Moir, D. N., Selecting batch filters, Chem. Eng., 89, 47, 1982.
2. Svarovsky, L., Advances in solid-liquid separation. I, Chem. Eng., 86, 62, 1979.
3. Jones, B. R. S., Vacuum sludge filtration. II. Prediction on filter performance, Sewage Ind. Wastes, 28,
1103, 1956.
4. Svarovsky, L., Advances in solid-liquid separation. Ill, Chem. Eng., 86, 72, 1979.
5. Brown, T. R., Designing batch pressure filters, Chem. Eng., 89, 58, 1982.
6. Hooton, J. A. and Thomas, C. M., Filter Presses, in Process Engineering Technique Evaluation —
Filtration, Suttle, H. K., Ed., Morgan-Grampian Publishers, Kent, England, 1969, 11.
7. Sugaya, K., private communication, Ishigaki Mechanical Industry Co., Tokyo, 1987.
 249

Chapter 14

C EN TRIFUGES FOR SLUDGE TR EA TM EN T

Christian Alt

TA BLE O F CONTENTS

I. Introduction.................................................................................................................... 250
A. Case H isto ry..................................................................................................... 250
B. Application of Centrifuges...............................................................................251

II. Sedimentation Centrifuges...........................................................................................252

A. Theory of Centrifugal Sedimentation............................................................ 253
B. The Unit Area Equivalent................................................................................254
C. Operation and Performancesin Wastewater Sludge Dewatering
and Thickening..................................................................................................255

III. Different T ypes.............................................................................................................. 256

A. The Scroll Discharge SolidBowl Centrifuge................................................. 256
1. Principles of O peration........................................................................256
2. Perform ances.........................................................................................260
3. Theoretical Prediction........................................................................... 263
4. Conclusions............................................................................................ 265
B. The Disc-Type Centrifuge................................................................................ 265
1. Principles of O peration........................................................................265
2. Application and Perform ance..............................................................270
3. Theoretical Prediction........................................................................... 271

IV. Conclusions..................................................................................................................... 272

References 272
250 Water, Wastewater, and Sludge Filtration

I. IN TRO D U C TIO N

A. Case History
Spin machines, used to remove a component from a multiphase mixture, may be among
the earliest devices used in human evolution. Even in the latter part of the 19th century,
there were centrifugal separation units which may be designated as centrifuges as defined
at present. First of all, centrifuges were applied in food processing and mineral separation
and later in the chemical industry. When environmental engineers realized the possible use
of the centrifuge in the field of sludge treatment, centrifuges had already reached an advanced
stage of development.
Though there were increasing local requirements for conventional drying beds for the
dewatering of sludges, it proved critical, especially in densely populated regions, due to
high land requirements in addition to unsightliness and odor problems. Attempts were made
to apply centrifuges in the beginning of the 20th century in Germany, using the centrifugal
filter (Hermann Schafer at Cologne-Riehl). This centrifugal filter was preferred because of
its enlarged filtration area compared to the conventional basket device. Moreover, this
machine was able to operate quasi-continuously, and gave a sufficient yield. The rotor
consisted of four tapered chambers crossed on a vertical shaft. A radial filter leaf on one
side of the chamber was able to retain the solids more or less completely, while the other
side of the chamber could be moved about automatically, discharging the filter cake.
In the year 1904, this centrifuge was improved and enlarged into a six-chamber rotor, in
collaboration with the manufacturer in Hannover-Linden. This type was used for many years
in Frankfurt-on-Main and other communities in Germany, before the substitution of the
filtering rotor by a solid bowl in which solids are removed by settling, and where the fine
particles of sludge are found in a lesser amount in the clarified liquid. At this time, it was
found to be impossible to remove the final 2 % of solids from the filtrate from the centrifugal
filter.
The new unit consisted of a base-bearing solid bowl with vertical axis of rotation, into
which the sludge was fed through a tube extending down to the bottom of the bowl. In the
course of the upward motion of the sludge, the solids migrated through the liquid radially
to the bowl wall by centrifugal forces. The liquid spilled over the solids and was discharged
by the machine via a skimmer arranged at the top and nearer to the shaft. The bowl could
rotate with an elevated speed up to 1320 rpm.
The rotation speed, as expressed, does not give adequate information about the increase
in the migration velocity of the solids; the acceleration number of the g-factor was introduced,
which describes the ratio of the acceleration in the centrifugal field to that in gravity. This
can be expressed as:

rco2
z = — = 5.59 x 10 - 4 Dn2 (1)
g

where z is the acceleration number, r the inner radius of the bowl, D the inner diameter of
the bowl, n the revolutions per minute, and co the angular velocity in radians per second.
The machine described above had an acceleration number (z) of 877, which in comparison
to present machines is remarkable; this unit had only a 900-mm bowl.
The solids were discharged automatically, in accordance with earlier patents of ter Meer
(leader in centrifuge development) by dropping the bowl wall hydraulically. Consequently,
this machine operated discontinuously with three distinct periods: loading, de watering, and
solids discharge. A modified device with a bipartite bowl was used in the U.S. (first in
Milwaukee and then in Baltimore) in the 1920s, mainly for fundamental studies . 1 It should
be noted that, from the present view, this machine was comparatively complicated. Among
 251

other types, there were drainage rings where retained solids formed a filter cake which acted
like a filter medium for the removal of finer particles. At that time, however, the pretreatment
of flocculation in order to remove fine particles was not so common as today.
Additional problems were observed with the inevitable variations in sludge composition
and characteristics of municipal wastewater sludges from plant to plant and within the plants
themselves. In those days, centrifuges were not very flexible in regulation parameters.
Furthermore, the parameters themselves were not studied in detail, and reactions, when the
parameters were adjusted, could not be evaluated. This was also the case in chemical
industries where centrifuges were applied widely. In conjunction with a lack of preventive
maintenance of centrifuges in conventional wastewater plants, use of centrifuges became
limited in the following decades. Moreover, vacuum filters and filter presses were introduced
and frequently preferred, particularly in wastewater sludge de watering, because these devices
appeared to be attractive in terms of cost.
From 1925 to 1950, numerous studies in centrifugal dewatering were conducted, and
since 1950, the centrifuges are being used again. This was mainly due to the improved scroll
discharge solid bowl centrifuge together with the application of flocculants.
Scroll discharge solid bowl centrifuges, frequently called decanters, operate continuously.
The separation efficiencies were remarkable, on the order of 60 to 70%, though this was
still not satisfactory in terms of economy. That may be due to the application of inorganic
flocculants, since recycle of the clarified liquid may give rise to some undesirable effects.
This was overcome by the use of poly electrolytes.

B. Application of Centrifuges
At present, centrifuges accomplish diverse tasks in wastewater treatment. Centrifuges are
essentially used in two stages of sludge separation: sludge dewatering and sludge thickening . 2
The following types of sludges can, in general, be separated by centrifuges of different
types: primary raw sludge to some extent with waste-activated or humus sludge, digested
primary sludge, mixed digested sludge, extended aeration sludge, septic tank sludge, and
excess activated sludge.
In general, centrifuges can treat a large quantity of sludge and requires relatively a smaller
space. The latter may be of interest where space is limited, and in the case of future extension.
Centrifuges are now extremely versatile in regard to adjustment of sludge characteristics (by
correct choice of operating parameters as well as by varied equipment and accessories). In
wastewater sludge separation, the closed housing of centrifuges is felt to be advantageous
to avoid odor problems.
Even though the success of centrifugal separation in wastewater treatment resulted mainly
from introduction of the polyelectrolytes, efforts are being made to use centrifuges without
flocculants, if possible, to save the comparatively high cost of the polyelectrolytes. For
example, in sludge de watering, 1.5 to 3 kg of flocculants per ton of solids are needed.
At present, wastewater treatment requires a comparatively high degree of separation.
Sludges, which are not easily separated, should have, for instance, 20 to 30% (or more)
solids mass fraction in the solids discharge, while the effluent should have less than 0.3%
in solids mass fraction. It should be noted that the solids density, which is an essential factor
in the settling, is close to that of water, and the original particle size distribution of the
nonflocculated solids may vary between 40 and 10 p,m. In the case of the activated sludge,
the particle size distribution is down to a few microns.
In activated sludge thickening, although the desired solids content in the thickened sludge
may be obtained without problems, the removal of the very fine particles from the liquid
requires comparatively high acceleration numbers of the centrifuge when flocculants are not
used (or are used in inadequate quantity).
Only a few categories of centrifuges are available for the dewatering and thickening of
252 Water, Wastewater, and Sludge Filtration

wastewater sludges: the scroll discharge solid bowl centrifuge, the disc-type centrifuge, and
the imperforate basket centrifuge. All of them belong to the class of sedimentation centri­
fuges.
The scroll discharge solid bowl centrifuge is, at present, preferentially used in municipal
wastewater sludge dewatering, compared to the imperforate basket centrifuge, which may
be found in Europe only in rare cases. The two could also be used with restrictions for
thickening of activated sludges, while the disc-type centrifuge is preferred in the thickening
process and especially in the case of handling bulking sludges.
The evaluation of centrifuges is based mainly on effectiveness of separation, apart from
reliability and economy. The effectiveness of separation could be estimated according to
the solids removal (in fact, the effectiveness of the separation of the two phases, the solid
and the liquid phase).
The mass recovery for the individual phase is given by:

For the solids:

= ~ - S || X 100 (2)
s0(s2 - s,)

where, @s is the solids mass recovery percent, s0 the solids mass fraction in the feed, s: the
solids mass fraction of the effluent, and s2 the solids mass fraction of the solids discharge.
For the liquid formed in concentrations:

0 L = - s ~­ ° l)(C^ ~ ^ x 100 (3)

(P s Co)(^ 2 Cl)

where 0 L is the liquid mass recovery percent, c0 the concentration of the feed (kg/m3), c t
the concentration of the effluent (kg/m3), c 2 the concentration of the solids discharge (kg/
m3), and ps the density of dry solids.
The separation efficiency E can thus be obtained as:

E = (@s + ©L - 100)

= (s2 - sQ)(s0 - st) x 100

x 100 (4 )
s0(s2 - S 0(100 - s0)

or for concentrations:

E = (Cj CoKCo O P s x 10Q (5 )

c0(c2 Ci)(Ps C0)

For example, in the case of dewatering of a raw primary sludge, when the feed has a
solids mass fraction (s0) of 1.5%, the effluent Sj of 0.3%, and the solids discharge 20%,
the solids mass recovery 0 S and the separation efficiency can be computed as 70% and
57.5%, respectively.

II. SED IM EN TA TIO N C EN TRIFUGES

Sedimentation centrifuges have proved to be sufficiently effective, reliable, and econom ­

ical in wastewater sludge de watering and thickening practice, and can be used in place of
 253

other separation devices like vacuum filters, filter presses, and press belt filters, etc. Though
the performances may frequently differ, economic considerations favor the centrifugal sep­
aration. Due to the very different compositions and characteristics of each and every waste­
water sludge, no general rule can be laid down to facilitate the selection and assessment of
the most appropriate separation device. Only in a few plants, especially the large capacity
ones, have experiments been carried out to study sludge characteristics in relation to operating
parameters.
Sedimentation centrifuges consist of a solid bowl or imperforate basket which has a vertical
or horizontal axis of rotation. The sludge fed into the bowl forms an annular pool caused
by the centrifugal forces. Solids are removed through migration of the particles across the
liquid towards the wall of the bowl. The centrifugal force makes the migration velocity very
much higher over the gravitation settling speed.
The liquid, as well as the solids, are then discharged via separate components of the
centrifuge.
The simplest type of centrifuge is the imperforate basket batch type (Figure 1 ) which
follows the principle of centrifugal sedimentation. It could be applied, with restrictions, to
the other types with an evaluation of the physical correlations.

A. Theory of Centrifugal Sedimentation

The settling of discrete particles under centrifugal forces can be given by Stokes’ law
(similar to gravitational settling).

(Ps - PL)dp
" “ i8„ ‘ <6)

where u is the migration velocity, ps the solids density, pLthe density of liquid, dp a significant
particle diameter, t] the viscosity of the liquid, and a the acceleration.
This equation is valid up to the laminar region, i.e., when the Reynolds number (Re =
u dp pL/nr|) is less than 1. Furthermore, in concentrated sludges, Equation 6 may not be
applicable, since this equation is developed for discrete settling where one assumes the
particle motion to be independent of the motion of the adjoining particles. The adjoining
particles retard the settling velocity of the particle considered. Thus, as a guideline, one
could say that Equation 6 is not applicable for a solids mass fraction above 2%. In this case,
one must apply empirical relationships. Although there are a number of such relations
available, none of them were developed for the case of centrifugal sedimentation.
Further, Equation 6 represents the motion of spherical particles. It has been found that
the effect of particle shape is greater in the case of concentrated sludges than in dilute ones.
In sludge dewatering with scroll discharge solid bowl centrifuges, therefore, the above
two effects may lead to a reduction of the settling velocity to between 20 and 30%.3
Introducing a reduction factor kw to consider these two effects, and combining Equations
1 and 6 , one gets:

, (Ps - PL)dp Z
u(c) = uStkw E- (7)
18t) g

where u(c) is the real migration velocity of the solids.

Substituting the Stokes equation in Equation 7 yields

u(c) = ustkwz (8)

254 Water, Wastewater, and Sludge Filtration

FIGURE 1. Schema of imperforate batch centrifuge.

where uSt is the settling velocity of discrete particle under gravity and in the Stokes’ law
region.

B. The Unit Area Equivalent

By analogy to the gravitational settling tank, the throughput in a sedimentation centrifuge
is given by:

Q = u(c)A = uSlkwzA (9)

where A is the unit area.

Taking into account that the unit area, as well as the acceleration number z, are functions
of the radius (affecting the particles within the annular liquid pool inside the bowl), Equation
6 can be written as:

2 ttw2L r? - r?
Q = uStk* ( 10)
3g

where L is the length of the bowl and ra and r, are the outer and inner radii of the annular
liquid pool.
Here, the geometry of the liquid ring is considered. Following the widely used 2 - concept ,4
if one substitutes 2 -value (Equation 11) in Equation 10,

2(h)2tt r;a Ai
2 = ( 11)
3 g ra

one will obtain Equation 12 for the capacity of the sedimentation centrifuge.

Q = uStkw2 ( 12)
 255

If the Reynolds number is greater than 1, the general transport equation is applied instead
of Stokes’ law. Unfortunately, the limitation of applicability of Equations 11 and 12 cannot
be found directly, since the Re number is itself dependent on the transport velocity u. By
dimensional analysis, the Stokes’ regime exists when approximately:

z d 3 x (Ps jO P L < ^
E ml

For example, this may occur in wastewater sludge de watering when the density difference
between the two phases is 2 0 0 kg/m3, the acceleration number is 2 0 0 0 , and the significant
particle size eliminated by the centrifuge in room temperature is over 20 |xm. This may not
be an abnormal case.
Using the universal relation for the motion of bodies in continua and gravity, one obtains:

= i r (Ps - p jd p g i 1
(14)
Ug 3 L fpL J

where f is the drag coefficient.

In this case the 2 value and the capacity (Q) can be written as:

^ (ra + fj)372
Xg = (2g)l/2 W7TL (15)

Q = ug2 g (16)

The 2 concept yields reliable results only when the flow of the fluid is undisturbed inside
the bowl (from the inlet to the effluent ports or weirs). Furthermore, it is applicable only
to the cylindrical one-chamber bowl. If the shape of the bowl is different from that of
cylindrical, then the 2 value must be adjusted accordingly. This has been discussed under
the section describing the two centrifuges, the scroll discharged solid bowl and the disc-
type centrifuge.

C. Operation and Performances in Wastewater Sludge De water ing and Thickening

In principle, there is no difference in sedimentation centrifuge procedure when it operates
either in the de watering or thickening mode. In fact, both methods are for separating solids
from water. Only the operating conditions are changed, for example, increasing or decreasing
the pond depth, or changing the shape of the rotor.
Sedimentation centrifuges consist of a solid bowl into which the sludge enters at one end,
in most instances through a feed funnel, or through appropriate equipment, while the liquid
overflows over the settled solids and flows out through effluent ports or weirs at the opposite
end of the bowl. In some cases, the liquid is removed by a skimmer or centripetal pump.
The types of the sedimentation centrifuges differ mainly in solids discharge and in some
structural details, as described in the following sections.
From the viewpoint of effective separation, it would be advantageous to apply a com ­
paratively high rotational speed, as can be seen from the theoretical relations. Thus, the
capacity depends on migration velocity u(c) (Equation 9), and acceleration number z (Equa­
tion 7). An increase in the acceleration number will lead to proportional increase of the
capacity. In the same way, increase of the migration velocity may produce a decrease of
the solids fraction in the effluent when the throughput is kept constant; this, in turn, results
in an increase of the solids mass recovery (Equation 2).
256 Water, Wastewater, and Sludge Filtration

The rotary speed of the bowl, however, is restricted by the mechanical stability of the
bowl itself. Considering that the admissible acceleration number is approximately given by
the balance of forces acting on the bowl wall, one gets:

z = 2 ' CTmn (17)

g • Pb • D

where a m is the mean tension stress in the bowl wall, pB the density of the bowl material,
and D the diameter of the bowl.
From the above equation, it follows that a full-scale centrifuge will always have a lower
acceleration number than that of a laboratory-scale unit with a smaller bowl diameter, a fact
which should not be neglected when tests are made in laboratory- or pilot-scale units.
On the other hand, high acceleration in the bowl will produce higher compression of the
settled solids and hinders the solids discharge.
In view of the discrepancies in the characteristics of different wastewater sludges, it is
not possible to give performances of dewatering or thickening of wastewater sludges without
exactly specifying their individual properties. As an indication, thickening of activated sludge
in a sedimentation centrifuge yields an increase of solids concentration from about 1 to 3%
up to 4 to 6 %, and at times up to 7 to 10%. The recovery, based on the solids content in
the liquid, may be between 80 and 97%, the higher value resulting by addition of flocculants
as high as 1.5 g/kg of solids. Table 1 presents performances of dewatering by centrifuges
in different plants.
Table 1 shows that digested sludge is more difficult to dewater than the raw primary
sludge, as observed with other techniques as well.

III. DIFFEREN T TYPES

A. The Scroll Discharge Solid Bowl Centrifuge

1. Principles o f Operation
The scroll discharge solid bowl centrifuge, frequently called decanter for short, again
became popular in the 1950s for sludge treatment. As mentioned previously, various designs
of this type of centrifuge were successfully used in chemical industries. In the design
development, much attention was paid to preventing floe shear, particularly during the feed
passage, and to the undisturbed motion of the separated phases. The most commonly used
type, at present, in wastewater sludge treatment, it consists of a solid bowl made of two
sections, one cylindrical and the other conical (Figure 2).
The cylindrical section of the centrifuge is responsible for the clarification of the water,
while the conical section effects drainage of the solids.
The sludge is fed through a concentric tube to an appropriate point, generally near the
connection of the two sections. The feed tube is adjusted to emphasize clarification or
de watering. The displacement of the feed tube in the direction of the cone increases the
residence time of the liquid in the bowl, and produces a more effective clarification, while
the residual moisture content of the solids is increased. From the feed pipe, the feed passes
through the conveyor hub and is accelerated. An exchangeable distribution impeller is
available here to control the acceleration in order to minimize the floe shear and any existing
erosion. The impeller may, in addition, be erosion protected when sandy particles are present
in the sludge. The sludge is then passed into the bowl where the centrifugal force causes
the liquid surface to become an annular ring (which is called the pond or pool), parallel to
the cylindrical bowl wall. Solids deposited against the wall are conveyed through the scroll
along the cylindrical section of the bowl towards the cone where they are taken out of the
pond. Drainage takes place as the liquid flows outwards and down the beach (conical section).
 257

Table 1
PERFORMANCE OF CENTRIFUGAL SLUDGE DEWATERING AND
THICKENING

Feed solids Cake solids Flocculants

concentration mass fraction (polyelectrolytes) Solids mass
Sludge type (kg/m3) (kg/kg) dosage g/kg solids recovery (%)

Raw primary 4— 5 20—30 1— 1.5 95— 97

Raw trickling filter 2— 3 9— 10 0 90—95
Raw primary plus 2— 3 12— 14 0.5— 1.5 93— 95
waste activated,
50:50
Aerobically di­ 1— 3 8— 11 0 80—95
gested 12— 14 0.5— 1.5 90—95
Aerobically di­ 1— 2 12— 14 0 75— 80
gested plus waste 10— 12 0.75— 1.5 85— 90
activated, 50:50
Waste activated 0.3— 1 6— 8 0 80—90

FIGURE 2. Schema of the scroll discharge solid bowl centrifuge. (Courtesy of Flottweg Werk Bird
Machine GmbH, Vilsbiburg, F.R.G.)

De watered solids are discharged through ports and ejected into the housing. In fact, the
slippage force may cause the conveyed solids to flow down the beach, too, in the opposite
direction to the solids transport, through the clearance of the conveyor when the sludge
consists of slimy solids. This is, unfortunately, found in municipal wastewater sludge. An
accurate design or a shallow beach may minimize the backflow. Therefore, conical sections
with an angle of 5, 8 , or 15° to the rotary axis are preferable. When a freshly thickened
activated sludge is required, a steeper cone with a 2 0 ° angle to the rotary axis is recommended,
particularly for small-to-medium sewage treatment facilities where a dual role (thickening
and dewatering) may be in operation.
Due to the drainage in the conical section of the centrifuge bowl, the decanter is not a
true sedimentation centrifuge, but rather a combination of two solid-liquid separation pro ­
cesses: sedimentation of fine particles due to high centrifugal velocity and filtration to increase
the dryness of the cake.
The clarified liquid spilling over the deposited solids into the pitches of the helical conveyor
flows in the form of a layer towards the end of the cylindrical section. (The clarified liquid
flow is in the direction of the opposite solid flow.) The liquid is discharged over a number
of adjustable ports (or weirs). The displacement of the ports nearer to the bowl shaft produces
an increase of the clarifying area inside the conical section while reducing the drainage area.
Thus, when the clarifying effect is improved, the discharged solids mass fraction decreases.
258 Water, Wastewater, and Sludge Filtration

1.0

• •
0.9 o o o
o

0.8
CD
N
CD
CL

0.7
CD

0 10 15 20
Differential Speed r.p.m.

FIGURE 3. Plot of separation efficiency over the differential speed in

the decanter centrifuge: • higher differential speed of conveyor, o lower
differential speed of conveyor.3,6

As the clarifying area is directly proportional to the pond depth, the clarification effect is
also improved with the increase in pond depth. Systematic investigations (with a wide range
of sludges) over some years in the U.K. verify that the increase in pool depth decreases the
polyelectrolyte requirement. 5 These indicate that saving on flocculants can be achieved by
increasing the pond depth.
This countercurrent transaction causes turbulences, mainly in the feed inlet region and
around the conveyor flights. This occurs, especially, in the case of relatively small density
differences of the two sludge phases, and in the presence of finer particles. Turbulences
reduce the settling space in the cylindrical section. Efforts are made to elongate the cylindrical
section but these efforts may be restricted due to the considerations of mechanical stress on
the cylindrical section.
In a concurrent design, the sludge enters the free cylindrical end of the bowl through
equipment similar to that of the countercurrent type. The liquid, as well as the solids, flows
towards the direction of the conical section. Here, the clarified liquid flows back to the inlet
end through return channels (arranged at the conveyor hub), and leaves the bowl. Hence,
the total length of the bowl is made available for sedimentation, while the effect of turbulence
is significantly suppressed. The latter is due to a smooth flow pattern, which minimizes the
floe shear.
To achieve solids transport through the bowl and then up the beach, the rotational speed
of the scroll and bowl should be different. In most practical cases, the rotational speed of
the scroll is slightly lower than that of the bowl, which leads to more effective removal
(than in the case of higher differential speed between the conveyor and the bowl). Systematic
studies have shown evident differences in separation efficiency, as shown in Figure 3. In
sludge de watering, the higher differential speed, on the other hand, results in a saving of
flocculant.
As can be seen in Figure 3, there is an optimum differential speed which results in highest
separation efficiency. The reason is as follows. At a lower differential speed region (which
is frequently preferred when a thicker layer of the discharging solids promoting the drainage
is desired), solids will accumulate. This occurs even if the solids mass rate in the feed is
not greater than the scrolling rate. On the other hand, higher differential speed produces
turbulences along the scroll flights and affects liquid clarity. The position of the optimal
differential (or scrolling) speed, however, depends on sludge characteristics. Studies made
 259

on dewatering and thickening of municipal wasterwater sludge by the aid of a 350-mm

concurrent decanter confirmed the existence of the optimal scrolling speed. Here, the op ­
timum recoveries are obtained when the differential speed is adjusted between 5 and 15
rpm, depending on the type of sludge.
Diverse drives for the bowl and the conveyor are available. Normally, the bowl and the
conveyor are joined through a reduction gear box in the form of planetary gears, special
curve-disk gears, or similar equipment. Two separate motors with exchangeable V-belt
pulleys are available to adjust the speed variation of the bowl on the conveyor. Different
designs are available to provide variation of the speeds while the centrifuge is in operation:
( 1 ) the tyristor-controlled direct current motor as well as the most advanced electric converter
back-drive system regulates the speed corresponding to a program; (2 ) the so-called hybrid-
dri ve, comprising hydromotor and hydraulic pump, controls the torque load of the scroll
through the oil rate. This is used in the case of varying solids feed fractions which would
cause clogging. The oil rate is automatically regulated to the optimum operation. Further,
the relative speed of the conveyor can be controlled in different designs by a belt transmission
between the sunwheel shaft of the planetary gear box and an intermediate shaft mounted on
the end of the frame.
The solids are mostly discharged by ejection at the open end of the conical section of the
bowl. Sticky or pasty solids need free discharge for which rotating gutters are provided.
There are instances in which the nozzle discharge yields more satisfactory operations and
increased cake dryness as a consequence of the solids compression. As a rule, discharging
the settled solids becomes more difficult as the centrifugal force increases. In the higher
centrifugal velocity range, increased pressure is set up between the solids and the bowl wall
as well as the scroll flights which makes the discharge of solids difficult. For this reason,
the low -speed decanter was introduced, i.e., the machine operating with the acceleration
number below 1000. The high-speed decanter operating above this acceleration number of
1000 up to 2500 produces, on the other hand, comparatively increased migration velocity
of the finer particles and may reduce the flocculant requirements. The latter is more effective.
Polyelectrolytes and special equipment to reduce floe shears are recommended. In general,
the rotation speed in wastewater sludge treatment is less compared to industrial applications,
although the solid content in the cake decreases.
Manufacturers of low -speed decanters prefer, obviously, the length to diameter ratio of
2 to 3, while in high speed decanters a ratio of 4 or more is used. Moreover, the geometry
of the bowl varies with the length of the cylindrical section (i.e., the section with constant
diameter) and the steepness of the conical section. The significance of the beach angle was
discussed earlier. The length of the cylindrical section has an influence on the clarity of the
effluent; for example, an increase of the length increases the retention time of the settling
solids in the bowl and thus facilitates the deposition of slowly settling particles. Therefore,
a saving on flocculants can be achieved, in practice, with the longer bowl. The longer bowl
will also produce, in general, a comparatively higher throughput when the separation effi­
ciency is held constant. These factors induced a recent supply of longer bowls with a length-
to-diameter ratio of more than 5.
There are variations in the scroll design. For example, in the case of flocculated wastewater
sludge, narrow scroll pitches in double leading helical conveyors minimize frictional forces
and turbulences between the flights although they have the risk of bridging. Thus, the optimal
design may be found from experience.
The machine components of standard models are made of normal carbon steel. In the
presence of corrosive materials in the sludge, all contact parts should be made of stainless
steel or diverse alloys. There is a special casting of the stainless steel bowl which may
produce special reliability. More frequently, there is a need for abrasive protection, especially
in the case of wastewater sludge separation. A number of different arrangements for protecting
machine components from abrasion are available. To name a few, hardfacing of the scroll
260 Water, Wastewater, and Sludge Fdtration

flights, sometimes resistant up to more than 300,000 ton of throughput, is applied as field-
replacable conveyor blades, ceramic tiles, and exchangable inserts in the feed pipe, in the
feed-accelerating device, and in solids discharge ports. The bowl can be protected by an
intentional solids heel upon the wall surface.
The bowl is mounted on fixed bearings. In order to minimize vibrations, a center-line
bearing system is now available for the longer bowl, similar to that of a disc-type centrifuge.
Attention is paid to the scroll torque. Any overload can lead to destruction. There are
different methods for torque control, some of which were mentioned before. The load of
the scroll is due to the solids friction at the flights which should be minimized. For this the
scroll can be polished, but the bowl should be roughened or grooved in order to support
scrolling. In the concurrent type, the longer solids conveyance may also increase friction,
for which special arrangements should be made to minimize it. In order to reduce energy
consumption, a slimmer bowl geometry may have an advantage over a shorter one of the
same capacity.

2. Performances
The number of adjustable parameters in this type, as discussed in the previous section,
yields a wide adaptability of this type for different sludge characteristics. In general, the
optimum conditions are only found from pilot-scale tests with the corresponding sludge. For
this purpose, movable equipment is available to make tests in the field. Frequently, it is
helpful to observe the immediate relations of the parameters to performance, as shown in
Table 2 (which is based on the results obtained from many field studies).
Above all, the capacity of a decanter can be theoretically evaluated because it depends
mainly on geometrical and operational factors. For the liquid throughput, the 2 - concept can
be applied (Equations 12 or 16). In the case of a bipartite bowl, the 2 - value becomes
approximately:

= (Dnf DV D/ H fl 4 H / H y '|
(IB)
1800 Lcy D + tana (.2 _ 3 D + YD/ j
In
D - H

where H is the pond depth, D the bowl diameter, a the beach angle, and Lcy the length of
the cylindrical section of a bipartite bowl.
The solids migration velocity (Equation 12) is related to the particle size of the sludge.
In flocculated sludge, the particle size may be very difficult to determine. Thus, the migration
velocity is estimated from laboratory tests using gravitational sedimentation. It should be
noted that this procedure is inaccurate, such that the 2 -concept is hardly brought in agreement
to the capacities obtained in full-scale operation. In addition, the theory using the 2 - concept
is based on spherical-shaped settling particles, on a plug flow within the settling region, on
the machine geometry which is frequently not identical with the actual, and on the existence
of an annular pond rotation without slippage in the bowl. Finally, effects of the conveyor
flights upon sedimentation as well as the instability of flocculated solids are ignored.
When information about the essential sludge characteristics is uncertain, insufficient, or
hard to obtain, the evaluation of the capacity, and, particularly, de watering effectiveness,
should be based on pilot studies. This offers an advantage of a more optimal assessment of
the full-scale centrifuge. The test runs made by the aid of a pilot-scale decanter of the same
type should correspond to the given centrifugal acceleration and to the specific throughput,
(i.e., the capacity related to the unit area). Frequently, the de watering effect depends on
the compressibility of the cake. In this case, the de watering process is adequate in the pilot
 261

Table 2
IM PLICATIONS OF OPERATING PARAMETERS IN DECANTER 8

Effect
Action Effluent clarity Solids dewatering Remarks

Increase of bowl speed Mostly better Better More wear; difficult transport
of the solids; more floccu-
lants
Increase of differential Up to the optimal differ­ Worse Release of the conveyor
speed of scroll ential speed (Figure 3): torque
better (less sludge block­
ing through the con­
veyor); above the
optimal differential
speed: worse (increase
of turbulences)
Increase of feed Worse Mostly worse, sometimes
better (in case of com­
pressibility of the cake)
Increase of depth of Better Worse
pond
Displacement of feed Better Worse In counter-current operation
inlet to cone only

9 O p e r a tio n p o i n t o f th e fu l.l-s c a le d e c a n te r

t
D
(' nJ 3 ') - 1 O E q u iv a le n t s o li d s c o m p re s s in g

A
D -2
( D ); -9- B e n e f i c i a l o p e r a t io n p o in t
P

~3
(-2 U n ifo r m a g g lo m e r a tio n
D
P

1 D 1/2 D
<r9)
P
<d^P
R o ta ry Speed R a tio

FIGURE 4. Operating conditions of pilot decanter centrifuge.8 Dg diameter of the full-scale centrifuge;
Dp diameter of the pilot centrifuge. (From Alt, C. and Gosele, W., Chem. Ing. Tech., 5, 425, 1982.
With permission.)

and full-scale centrifuge only when the rotational speed and the solids load are adjusted in
such a manner that a smooth pattern of flow as well as the consolidation of the solids is
achieved. This occurs when the rotational speed and the throughput ratio equals the diameter
ratio (see point “ C ” in Figure 4).
To provide a useful index to the true state of capacities in wastewater sludge dewatering
and thickening, typical capacities of available decanter models are listed in Table 3. This
 Table 3
262

TYPICAL CAPACITIES OF AVAILABLE DECANTER CENTRIFUGES

Capacity in waste water

sludge separation (m3/h)
Length to Rotation Connected
Bowl diameter diameter speed (rpm) motor Raw Indicated
(mm) ratio min—max (kW) primary Digested Activated (nominal)

220 2.7— 3.6 ­5500 11 2.5—3.5 (7— 10)

340— 355 (14") 2.1—4 1200— 4000 7— 30 4—7 3— 15 2— 3 6— 15 (3— 25)
425—450 2— 3 1000— 3300 15— 37 8 6— 26 4— 6 12— 18 (10—25)
500—530 3—4 1000— 2900 22— 55 10— 18 — 34 a 20— 30 (30)
600—635 2 .3 - 4 .2 1000—2900 37— 75 24—40 30—62 12— 30 40— 60 (23— 50)
735 3.1—4.2 1500— 2400 55— 110 a 52— 110 a (40—90)
900 2— 3 600— 1400 55— 75 48—64 36— 56 24—48
1020 (40") 3.5 1300— 1900 93— 150 a 120— 180 130— 180 (90— 220)

a No designation.
Water, Wastewater, and Sludge Filtration
 263

table can help the project engineer and planners of a municipal wastewater treatment system
to assess the centrifuge size. This table, however, is not complete. For instance, there are
models with bowl size above 1.02 m in operation. The centrifuges with larger bowl sizes
are more economical at larger installations.
The values calculated from the 2 - concept are always higher than that determined from
pilot-studies (or field-scale). Nevertheless, the 2 - concept should be considered as a basis
for quantitative evaluation, when operational as well as structural parameters are varied.

3. Theoretical Prediction
Attempts have been made to predict, theoretically, the expected performance. It was
initiated with the application of the fundamentals used in the 2 -concept. Additionally,
correction factors were introduced in order to adapt the theory to the actual condition of the
centrifuge. It was recognized that the complex process of the decanter can only be described
by mathematical models. In a recent work , 12 attention was paid to the region of the sludge
entrance where the feed is to be accelerated up to the bowl rotation speed and where solids
deposition may be an essential state. If the actual liquid flow in the inside of the bowl of
the decanter is considered, the (widely used) plug flow model for the annular pond will not
be satisfied with the prediction of the performance. Therefore, the influence of solids frictions
during the conveyance as well as the turbulence of the liquid should be included. These are
incorporated in a recent model. Equation 12 was rearranged in the following form to find
out the critical particle size that can be removed from the effluent.

T] • Ah • Q 0.5

d 100 = 1.69 (19)

kw • o) • WK • r(co • r • i • Ah • B + Q)Ap

where d 100 is the particle size of the solids removed completely from the effluent, r\ the
viscosity of the liquid, Ah the thickness of the above-mentioned liquid layer flowing towards
the effluent ports which can, in most cases, be assumed to be the half pond depth, Q the
feed volume rate, kw the same as in Equation 7, (o the angular velocity of the bowl, Wk the
number of scroll turns in the clarifying region, r the medium radius of the pond, B the pitch
of the scroll conveyor, i the number of scroll flights, and Ap the density difference of the
raw and dewatered sludge.
A comparison made between the experimental data 12 and theoretical results obtained from
various models9 11 is shown in Figure 5. Here the 50 percentile size is determined from the
following relationship:

To supplement the preceding relationship (Equation 2), the solids mass recovery results from
the summation of the several recoveries of the individual particle size ranges as given by:

(21)

where y is the separation grade (dp/d100) 2 (Equation 19), and qA the solids particle size
distribution density in the feed. If the mathematical function of ^ is not available, recoveries
of finite .particle size ranges may be summed up. Results obtained from a pilot-scale study
with various sludges are summarized in Figure 6 with the theoretically obtained values. The
difference in values obtained for different sludges are due to the disturbances in the settling
264 Water, Wastewater, and Sludge Filtration

FIGURE 5. Theoretical and experimental plots of the 50 percentile cut-off size vs. the feed volume rate
in the decanter centrifuge.12— •—Trawinski;9— ••— Nesterowich11;——— Frampton;10 o o Experimental.
(From Faust, T. and Gosele, W., Chem. Ing. Tech., 8, 698, 1985. With permission.)

FIGURE 6. A plot of solids mass recovery vs. throughput in a decanter centrifuge.12— theoretical
(Equations 19, 20, 21); V calcite untreated, undispersed; A calcite fed into the turbulence region; o calcite
untreated, dispersed; • calcite plus wetting agent, obstructed. (From Faust, T. and Gosele, W., Chem.
Ing. Tech., 8, 698, 1985. With permission.)
 265

region. For example, the solids mass recovery of an untreated calcite sludge decreases
steeply when the throughput is increased above 1.4 m 3/h. This may be due to the fact that
the sludge enters the turbulent region inside the conical section, when the throughput is
above 1.4 m 3/h.

4. Conclusions
There are numerous factors which should be considered in the selection of an appropriate
process for dewatering or thickening of sludge. The important operating parameters of
dewatering or thickening by centrifuge are described in the preceding sections. However,
the complexity of the parameters does not help the project engineer to make decisions. The
question of which type, a low- or high-speed decanter centrifuge, is the most suitable for
the plant under consideration, itself requires knowledge and experience. In addition, there
is a need for consultation because of structural diversities, and the capital, operational, and
maintenance costs involved with the high quality machine. Figure 7 presents the sectional
view of a commonly used solid bowl scroll discharge decanter centrifuge. Finally, external
influences, such as noise level, and installation with other wastewater treatment processes
should be taken into account, the emphasis of which varies from plant to plant.
Further, the operation of the centrifuge requires a series of accessories. The sludge pump
should feed the sludge at a constant rate, which requires automatic regulators. The equipment
for the solids discharge and effluent transportation should be made available in such a way
that safe removals are achieved from the machine. The conditioning device comprising
equipment for dissolving, mixing, and dosing the flocculants must have sufficient capacities
in order to ensure the constant operation of the entire system. Finally, these separation
devices may not be advantageous to remove coarser particles and fibers. The entry of these
materials will damage and block the machine. All equipment must be carefully checked so
that they adapt to each other. Very often, the reason for unsatisfactory operation of the
centrifuge is found in the deficient operation of one of the accessories.
Numerous publications are available on the application of the centrifuge in wastewater
treatment. The information, based on past experience and observations, varies widely from
one plant to another. The reason for this lies in the imperfect differentiation of sludge
characteristics. This could, moreover, lead to incorrect conclusions. The preceding arguments
may form a basis for evaluation and verification of the expected performance.

B. The Disc-Type Centrifuge

1. Principles o f Operation
Originally introduced for milk separation about 100 years ago through the basic patents
of de Laval (Stockholm) and von Bechtolsheim (Munich), it became an indispensable ma­
chine in the chemical and pharmaceutical industry. This sedimentation centrifuge was first
applied for wastewater sludge thickening in 1937.2
The disc-type centrifuge consists of a bowl with a vertical axis of rotation, usually
composed of two elements for opening and cleaning purposes. The special solids-ejecting
bowls are formed as a double cone to combine the removed and deposited solids (Figure
8 ). Inside the bowl, a number of conical superimposed discs are placed to distribute the
incoming sludge as extremely thin layers corresponding to the distance between adjacent
discs (which is generally of the order of 0.3 and 3 mm). This distance between the discs
depends on the characteristics of the sludge. The discs are significantly smaller in diameter
than the bowl shell to provide an adequate sediment-holding space. The settling time of the
solids is significantly reduced because of the centrifugally forced horizontal migration of
solids in the interspace of the discs.
The solids deposited on the underside of the discs slide down continuously in a cohesive
layer to the lower end of the discs, and from there in horizontal direction into the sediment-
266 Water, Wastewater, and Sludge Filtration

FIGURE 7. Sectional view of the solid bowl scroll discharge decanter centrifuge. (Courtesy of Sharpies Zentri-
fugentechnik GmbH, Dinslaken, F.R.G.)

holding space. There are nozzles, ports, or slits provided for solids discharge. The latter
are opened periodically through an automatically operating system. Because of the automatic
sludge disposal device, this type of centrifuge is also called the self-cleaning disc centrifuge.
The feed enters the bowl through a stationary feed pipe immersed into a funnel for sludge
distribution around the bottom of the bowl. Ribs are provided in order to accelerate the
sludge to the rotary speed of the bowl. The sludge then passes on to the disc stack where
liquid clarification takes place. The liquid is then discharged through the inner collecting
space of the bowl, over weirs. Alternatively, in closed systems, a paring disc or centripetal
pump skims off the effluent by utilizing the rotation energy.
The sediment-holding space (in this case a double cone) guides the settled solids directly
to the discharge openings which are located at the periphery of the bowl. The centrifugal
action consolidates and compresses the solids which increases the concentration of the solids
in the discharge. As mentioned above, there is different equipment available to discharge
the solids. For comparatively high solids concentrations (above 10 to 20% dry solids), the
nozzle discharge type is preferred (Figure 8 A), while for sludge below 10%, a periodically
operating self-cleaning system is used, where the deposited solids are accumulated to a
reasonable amount before ejection. If the solids slide easily, the nozzles are fitted in the
bottom portion of the bowl. 13 The number of nozzles and the diameter depend on the angle
of repose of the solids, and on the requested capacity, as well as on the sludge concentration.
If the solids content in the discharged sludge is not high enough or fluctuates, then the
recycling of the concentrate is employed. A part of the concentrate is returned to the bowl
through an additional pipe in the distributor directly into the sediment-holding space.
 267

r f

FIGURE 8. (A) Nozzle-type disc centrifuge (Courtesy of ALFA-LAVAL Industrietechnik GmbH, Hamburg,
F.R.G.); (B) self-cleaning disc type (Westfalia Separator, Self-thinker device): (a) sensing zone disc, (b) sensing
liquid clarifying discs, (c) and (e) sensing liquid pump, (d) flowmeter, (f) centripetal pump, (g) magnetic switch,
(h) timing unit. (Courtesy of Westfalia Separator AG, Oelde, F.R.G.)
268 Water, Wastewater, and Sludge Filtration

The nozzle discharge rate is derived from the hydrostatic head in the bowl as : 14

nrH^ ___________
Qn = kF • zN O) VG2 - rf) (22)

where QN is the discharge volume rate of the nozzle, kF the factor concerning the friction
and the constriction in the nozzle, (mostly 0.7 to 0 . 8 ), zN the number of nozzles, dN the
nozzle diameter, ra the radius of the bowl periphery, and r* the radius of the liquid surface
inside the bowl. Hence, the capacity of the nozzle -type disc centrifuge depends on the rotary
speed of the bowl, the number and the diameter of the nozzles, and on friction.
The sediment-holding space in the nozzle-type is furnished with interchangeable inserts
to prevent the retention of solids between the nozzles. There are also centrifugal valves or
annular pistons available to discharge the solids completely, when that is absolutely nec­
essary.
The self-cleaning type (Figure 8 B) provides for the closing of the openings, ports, or slits
during the time of accumulation and consolidation of the solids. Thus, the operation of the
self-cleaning system is intermittent. This may be an economical method to thicken dilute
sludges (below 10% dry solids) although different devices are available. Inside the bipartite
bowl forming the double cone, a movable bottom lining the lower bowl shell with a small
interval, is pressed against the upper shell through the hydrostatic head of the operating
water fed into the closing chamber of that interval. The centrifugal forces press this water
against the rim. When the sediment-holding space is filled, a hydraulically controlled disc
valve is opened, decreasing the hydrostatic pressure in the interval. The hydrostatic head of
the sludge inside the bowl then displaces the movable bottom, and the ports become free
for solids discharge. The sliding bowl bottom closes the ports automatically when the
hydrostatic head of the water in the closing chamber exceeds that of residual sludge in the
bowl. A different device provides for the motion of the sliding piston housed between the
two fixed lower bowl shells. The solids discharge control occurs in a similar way as described
above. In another design, axial discharge holes at the bowl periphery link solids pockets
inside the bowl to prevent the retention of solids; likewise, the segments of the nozzle type,
to valves which are closed by spring-loaded valve slide. For solids discharge, the valve slide
is lowered by compressed air, or gas, fed through the hollow bowl spindle. The sludge
discharge may be controlled so that the discharge occurs partially if increased consolidation
is required for an increased solids concentration in the discharge. Full discharge is used only
in the case of sludge containing sticky and fibrous solids. An automatically controlled solids
discharge device is also available for the nozzle-type disc centrifuge.
The bowl of the disc-type centrifuge, normally mounted on top of the vertical spindle
and supported on top and on the bottom end bearings (Figure 9), rotates at a faster speed
than the critical speed. This is because the center of gravity of the machine is not identical
to the structural axis of rotation. For that, sufficient clearance and an elastic bearing on top
must be provided. The bowl spindle is directly connected with the motor through a worm
gear. In a centrifuge with higher capacity (with the feed rate in the range of 70 to 150 m3/
h), the drive is arranged on top of the bowl spindle through V -belts, while the feed pipe
enters the bowl from the bottom. The bottom -drive system allows a short spindle configu­
ration for the simpler removal of the complete spindle assembly and to facilitate servicing
of the drive as shown for the 25,000 €/h machine in Figure 10.
The rotary speed of the disc-type centrifuge is comparatively higher than that of the
decanters. The rotational speed of the available species may range from 3,300 to 10,000
rpm corresponding to the acceleration number 3000 to 5000. Some machines are generally
equipped for infinitely variable speed through a tyristor regulator.
 269

FIGURE 9. Sectional view of the disc-type centrifuge, Type SB 80-06-076 (Courtesy of Westfalia Separator
AG, Oelde, F.R.G.)

Liquid discharge is provided with open, hermetic, and paring ring design. The feed and
discharge connecting pipes are attached to the rotating distributor in the closed systems by
means of airtight seals, since the feed enters under pressure through the feed pipe; an
alternative design supplies the feed, under pressure, through a hollow spindle.
270 Water, Wastewater, and Sludge Filtration

FIGURE 10. Schematic diagram of self-cleaning disc-type centrifuge with belt drive, Type SC 35, (Courtesy of
Westfalia Separator AG, Oelde, F.R.G.)

2. Application and Performance

Disc-type centrifuges are employed in industrial sewage treatment for concentration of
effluents and in municipal wastewater treatment mainly for thickening of sludges. Here, a
part of the activated sludge is returned to the aeration tank, while the excess activated sludge
undergoes further treatment before disposal. The sludge coming from the aeration tank is
normally thickened in the secondary settling tank up to 0.5 to 1% dry solids concentration.
The excess sludge from the secondary settling tank is generally concentrated up to 6 to 8 %
dry solids in order to reduce the capacity of the subsequent stages of the sludge treatment,
particularly, the digester. The thickening, up to the above-mentioned concentration, will
reduce the volume of the sludge by 90%. The thickening in the disc-type centrifuge can be
achieved without flocculants.
The thickening of different activated sludge, like the sludge coming from the upflow
settling tank, may occur with the use of the disc-type centrifuge in similar performance. In
all described cases the effluent returning normally to the aeration tank may be clarified down
to 0.03 to 0.1% dry solids concentration.
It is to be noted that the concentration of the thickened sludge, as well as the solids content
in the effluent, depends on the throughput of the machine. From Figure 11 it can be seen
that in the case of excess sludge thickening with an increase of throughput, the recovery
decreases while the dry solids content in the effluent increases slightly. In addition to the
throughput, the recovery also depends on the sludge volume index .2 The decrease in recovery
with the solids concentration in the thickened sludge is more significant when the sludge
volume index is high.
In general, the nozzle -type centrifuge with nozzle diameters around 1 mm requires a
preceding separation process to remove larger and fibrous particles from the sludge. That
can, for example, be achieved with a strainer. Moreover, in the nozzle-type centrifuge a
constant feed rate should be maintained in order to maintain a constant solids content in the
thickened sludge. To meet the fluctuations in municipal wastewater plants, the nozzle would
have to be changed. Since the self-cleaning machine may produce a more concentrated
 271

Cf)
TD
•H
*—I
o
C
O
M
O
a*
A
96
o'*

-QJ 91
>
O
O
(D
“ 86

Feed Rate m 3/h

FIGURE 11. Recovery and solids content of the effluent in the excess sludge thick­
ening with the nozzle-type disc centrifuge (Q = 40 — 50 m3/h, feed concentration
0.85% dry solids).14

sludge than the nozzle-type, it can also be used in combating the bulking sludge problem.
This problem arises when the sludge volume index is essentially over 150 m€/g (which
occurs as a result of the presence of filamentous bacteria).

3. Theoretical Prediction
Using the basic relations derived for the sedimentation centrifuge with the cylindrical
bowl, the following equation can be derived for the disc-type centrifuge:

r 2uol>2 n
Q = |^ustkw ctga(ii - r?)zDJ (23)

where Q is the throughput, ust the Stokes’ settling velocity, kw the correcting factor for
hindered settling and nonspherical particle shape, to the angular velocity of the bowl, a the
angle between the disc cone and the axis of rotation, ra the larger radius, r{ the smaller radius
of the disc cone, and zD the number of the discs.
The term within the brackets is equivalent to the 2 - value, for example, the unit area
equivalent, for the disc-type centrifuge. The direct prediction of the machine capacity using
Equation 23 may produce discrepancies with the throughput obtained in practice. For this
reason, manufacturers of the disc-type centrifuge recommend a modified empirical rela­
tionship. But this empirical relationship can be used only for the specified model, in order
to compare the capacities. Generally, the effective capacity may be assumed to be half as
much as the predicted capacity. The theoretical approach emphasizes the most important
parameters which exist and arise during the operation. Factors corresponding to material,
operation, and structure are considered by analogy to the decanter. Material factors are
concerned with:
272 Water, Wastewater, and Sludge Filtration

• the cut-off size (dp) removed in the centrifuge (for example, the minimum size of
particles removed is 0.5 p,m)
• the particle shape affecting the factor kw
• the density difference
• the viscosity (not significant in the case of wastewater)

The operating factor is

• the rotary speed, or the annular velocity co which is restricted by the bowl stability

The structural factors are concerned with:

• the geometry of the disc and consequently that of the bowl

• the cone angle, depending on the angle of the solids repose, and
• the number of discs, depending on the geometry of the bowl.

The material parameters are subject to change with the degree and type of pretreatment,
whereas the operational and the structural parameters are fixed for a particular centrifuge
considered.

IV. CONCLUSION S

The more essential models and devices of disc-type centrifuge used in wastewater sludge
treatment are described. Publications dealing with recent applications are limited, unlike the
case of the decanters.
In some instances, the theoretical prediction is restricted to comparisons and verifications.
In fact, the true prediction of performance may be in the hands of the experts using modified
theoretical methods which avoid the difficult evaluation of the different parameters in an
individual case; but frequently, it is the experimental result which yields the optimal operating
parameters.

REFEREN CES

1. Keefer, C. E. and Kratz, H., Jr., Experiments on the dewatering of sewage sludge with a centrifuge,
Sewage Works J., 1, 120, 1929.
2. Harrison, J. R. and Goodson, J. B., Process Design Manual for Sludge Treatment and Disposal, U.S.
Environmental Protection Agency, Washington, D.C., 1974.
3. Faust, T., Untersuchungen iiber die Stromungen und Absetzvorgange in der Klarzone von Dekantem,
Doctoral dissertation, University of Stuttgart, Stuttgart, 1983.
4. Miller, S. A., Liquid-solid systems, in Chemical Engineers’ Handbook, 5th ed., Perry, R. H. and Chilton,
C. H., Eds., McGraw-Hill Kogakusha, Tokyo, 1973, 19-93.
5. Ford, J., The sludge dewatering centrifuge, Inst. Chem. Eng. Ser., 41, Q l, 1975.
6. Alt, C., On the prediction of capacities and efficiencies of continuous decanter centrifuges, in Harnessing
Theory for Practical Application, 3rd World Filtration Congr., Downingtown, Pa., 1982, 364.
7. Reinmann, D., Schlammentwasserung mit Grosszentrifugen auf dem Klarwerk Wuppertal-Buchenhofen,
Kommunalwirtschaft, 9, 343, 1974.
8. Alt, C. and Gosele, W., Einsatzkriterien fur Dekanter, Chem. Ing. Tech., 5, 425, 1982.
9. Trawinski, H., Die aquivalente Klarflache von Zentrifugen, Chem. Ztg. Chem. Appar., 18, 606, 1959.
10. Frampton, G. A., Evaluating the performance of industrial centrifuges, Chem. Process Eng., August,
1963, 402.
 273

11. Nesterowich, A. A., Gelperin, N. J., and Shkoropad, D. E., Technologicheski raschet skorostnij os-
alitelnij shnekowij centrifug tipa nogsh, Kimicheskoye Neftyanoe Machinostroenie, 4, 15, 1967.
12. Faust, T. and Gosele, W., Untersuchungen zur Klarwirkung von Dekantierzentrifugen, Chem. Ing. Tech.,
8, 698, 1985.
13. Hemfort, H., Separators, Technical Scientific Documentation No. 1, Krochen H.-G., Ed., Westfalia
Separator AG, Oelde, F.R.G., 1984, 49.
14. Brunner, K., Experience in the use of centrifugal separators in the treatment of industrial effluent, separate
print, presented at VDI-Stuttgart, Westfalia Separator AG, Oelde, F.R.G., 1981.
 275

INDEX

A Cane sugar refining, 235

Capillary module, 180— 181, 184
Acceleration number, 250, 254— 256 Carbon, 137
Accurel®8 process, 174 Carman-Kozeny equation, 29, 47, 179
Acrylic copolymer, 161 Cartridge filtration, 129— 138
Adhesion, 18, 30, 33, 4 2 ^ 6 applications of, 137— 138
Algae, 70— 71 design for, 134— 136
Aluminum, 62, 71—72 filters for, 130— 137
Amperometry, 215 cleanable, 137
Anaerobic digestion, 166— 167 depth type, 130, 133— 134
Aquaculture, 165 disposable, 136— 137
Aramid, 140— 141 parameters of, 130— 133
Arsenic, 72— 73 surface type, 130, 133— 134
mechanism of, 130
B theory of, 130— 134
Casting plant, 83, 92—93
Back transport, 177— 179 Cationic polyelectrolytes, 38— 39, 45— 46, 49
Backwashing Cellulose acetate (CA), 140, 160— 161
in crossflow microfiltration, 184— 186 Centrifuges for sludge treatment, 249—272
in deep-bed filtration, 67— 69 application of, 251—252
methods for, 4— 7, 64, 72 case history of, 250—251
recycling water in, 72 performance of, 257
Batch system, 182— 183, 202, 212 sedimentation, 252— 256
Batch vacuum filters, 226 types of
Beaudrey Colmatometer, 103— 108, 111 base-bearing solid bowl, 250, 253
Beaudrey P.C. water quality measurement (B.P.C.), disc-type
104— 108, 111 application and performance of, 252, 270— 271
Beer factory, 84— 85, 96 principles of operation of, 265—270
Bessel function, 28 theoretical predictions of, 271—272
Biomass, 165 imperforate basket, 253— 254
Black liquor, 220 scroll discharge solid bowl
Body feed, 119— 122 early, 251— 252
Boltzmann equation, 34 performance of, 260—263
Boucher method, 103 principles of operation of, 256— 260
B.P.C., see Beaudrey P.C. water quality measure­ theoretical predictions of, 263—265
ment Cheese production, 156, 219
Brine, see Salt water Chloramines, 61
Brownian motion, 59 Chlorides, 62
Brownian particles, 36 Chlorine
Buchner funnel, 230 for disinfection, 61
effect of color on, 71
c excess, 137
membrane degradation by, 140
CA, see Cellulose acetate membrane resistance to, 161
Cake prevention of fouling with, 216
discharging of, 226 properties of, 58
drying, 226 sterilization by, 108
formation of, 226, 240 in swimming pool water, 127— 128
moisture content of, 234 Chlorine dioxide, 61
removal of, 244 Chloroforms, 61
resistance of, 228—229, 239—240 CIP, see Cleaning-in-place unit
thickness of, 118, 232, 238 Circulation in series, 197— 198, 214
washing of, 244 Clarifier, 165— 166
Cake filter, 18, 128, 184, 232 Clarifying area, 257— 258
Cake washer, 119, 122 Clean bed, 11— 12
Calcium carbonate, 216 Cleaning-in-place (CIP) unit, 187
Calcium sulfate, 216 CMF, see Crossflow microfiltration
276 Water, Wastewater, Sludge Filtration

Coagulants, 58— 59, 62— 63 87— 91

Coagulation, 78 principle of, 18
Coagulation-flocculation, 58— 59 schematic representation of, 19
Coal, 2, 64 theoretical approach to, 17— 52
Co-ions, 199— 202, 208 macroscopic filter behavior, 18— 30
Coke factory, 85— 86, 97 kinetic approach, 20— 28
Collector, 30— 33, 36, 37, 41, 42, 45—47 limitations of, 30
Colloids, 62—63 mass balance, 20
Color removal, 71 pressure drop, 28— 30
Concentrate streams, 192, 196, 206, 211 microscopic description of particle deposition,
Concentrating compartments, 194, 202, 209, 217 30—49
Concentration factor, 177— 180, 183 filter efficiency, 32— 49, see also Filter
Concentration polarization efficiency
in ED systems modeling flow through filter bed, 30— 32
in dilute stream, 210 Demineralization, 219
dispersion of, 217 Demineralization factor, 213— 214
effect of spacer gaskets on, 194 Deposition mechanisms, 24
efficiency of, 211 Desalination, 142, 152— 153, 157, 202, 219
equation for, 205—206 Detergents, 187
importance of, 202 Dewatering
reduction of, 204 by centrifuge, 252— 272
salt concentration in, 216 disc type, 256— 272
in RO systems, 144, 150— 152 application and performance, 270— 271
in UF systems, 162— 164 principles of operation, 265—270
Concentration potential, 210— 211 theoretical prediction, 271—272
Concentration profiles, 202—203 scroll discharge solid bowl, 252, 256— 265
Contact forces, 35 performances, 260— 263
Convective diffusion equation, 36 principles of operation, 256— 260
Coulomb efficiencies, 211 theoretical prediction, 263—265
Counterions, 199, 202, 206— 209 by filtration, 228, 251
Critical parameter, 214 solids recovery in, 230
Crossflow microfiltration (CMF) vacuum filters for, 226— 235, 251
basic plant design for, 180— 184 compared with other equipment, 227
economic factors in, 187— 188 for domestic waste sludge, 235
flux increase by, 184— 187 for industrial waste sludge, 235
principle of, 174, 176— 180 operating parameters for, 231—232
two-stage, 184, 186 performance in, 232—233
Current for plant sludges, 233—235
critical, 202, 207, 210 Diatomaceous earth, 118, 123
density of, 207, 213 Diatomite, 124— 127, 233
efficiency of, 208—212, 214 Differential pressure, 130— 132
reversal of, 216— 217 Diffusion coefficient, 207
transport of, 208 Dilute compartment, 194, 204, 207, 209, 217
Dilute streams, 192, 194, 196, 202, 210, 211
D Direct filtration, 57— 75
application of, 60— 61
Darcy’s equation, 118 chemical conditioning in, 61— 63
Darcy’s law, 228, 238 for conventional water treatment, 3— 4, 58— 60
Decanter, see Scroll discharge solid bowl centrifuge costs in, 73— 74
Decolorizing, 137 filter design in, 63—65
Deep bed filtration, 1— 52, 77—99 in-series application, 69—73
backwash methods in, 4— 7 limits to, 65— 69
design criteria for, 7— 9, 80— 86, see also Filter Dirt-holding capacity, 133
design Dirt removal, 132— 133
design optimization for, 7— 9 Disc-type centrifuge
improvements on, 2— 4 application and performance of, 252, 270— 271
for industrial wastewater treatment, 81— 87, 92— nozzle, 266— 268, 270—271
99, see also Wastewater principles of operation of, 265—270
liquid, 33— 36 self-cleaning, 266—268, 270
mathematical models of, 9— 14 theoretical predictions of, 271—272
for nonindustrial sources of wastewater, 78— 81, Disinfecting agents, 61—62
 277

Disinfection, 58, 70— 71 pressure within, 28

Donnan equilibrium, 199 zones of, 12
Donnan exclusion, 201 Filter behavior, 18— 30
Downtime, 243— 244 kinetic approach, 20— 28
Drag coefficient, 255 limitations of, 30
Drag forces, 18, 178 mass balance, 20
Drum speed, 232 pressure drop, 28— 30
Drum submergence, 232 Filter box, 5
Filter coefficient, 21, 25— 28
E Filter deposit, 22
Filter design
ED, see Electrodialysis for cartridge filtration, 134— 136
Effluent, 5— 7, 72, 84, 165 for crossflow filtration, 180— 184
Electrical double layer force, 34 for deep-bed filtration
Electric efficiency, 211 of beer factory wastewater, 85
Electrochemical potential, 204— 205 of casting mill wastewater, 83
Electrode block, 196 of coke plant wastewater, 86
Electrode reactions, 209—210 different units of, 80
Electrodes, 211 in leachate treatment, 82
Electrodialysis (ED), 191— 222 in Minamisenju plant, 80
applications for, 218—220 in night soil treatment, 81— 82
basic principle of, 192 optimization of, 7— 9
instrumentation for, 215 of plating factory wastewater, 86
operating conditions in, 216— 218 in sedimentation unit effluent treatment, 84
permeability in, 202—211 for direct filtration, 58, 63— 65
current efficiency, 210—211 for microfiltration, 180— 184
ion transport, 204— 207 for precoat filtration, 122
qualitative considerations, 202— 204 for pressure filtration, 243—245
transport and efficiency, 207—210 Filter efficiency, 21— 25, 32— 49
process and equipment for, 211— 215 for cationic polyelectrolyte, 49
selectivity in, 199— 202 in dynamic filtration phase, 46— 49
stack, see Electrodialysis stack empirical approaches for, 23— 25
Electrodialysis stack in initial phase, 32— 46
arrangement in, 204 for Brownian particles, 36
assembling cell units for, 192— 194 in liquid deep bed filtration, 33— 36
current efficiency of, 210— 211 for non-Brownian particles, 36— 46
design of, 212— 214 adhesion probability, 42— 46
electric current in, 217 transport efficiency, 36— 42
electrokinetic phenomenon in, 207, 209 for quartz particles, 48
flow system of, 195— 199 Filter layers, 31
flux in, 206 Filter leaf test, 229—231
overall efficiency of, 211 Filter media, 2, 64, 134— 135, 238
solution circulation in unit cell, 194— 195 Filter run, 66— 67
Electrometathesis, 204 Filters
Electronics industry, 153— 155, 164 backwash methods for, 4— 7
Electroosmotic transport number, 208—209 batch pressure, 240, 243
Electroplating industry, 220 belt, 241, 243, 245
Elutriation, 233 cake, 18, 128
cartridge, see Cartridge filtration
F centrifuge, 227
clogged, 64— 65
Feed and bleed system, 183 deep bed, see Deep bed filtration
Feed streams, 192, 195, 198 depth, 130, 133— 134
Fick’s law, 149 dual-media, 63
Film theory model, 151 effluent from, 5—7
Filter aids, 120— 124, 233 leaf, 240—241, 245
Filter bed materials for, 24, 43
clean, 30— 32 multi-media, 63— 64
clogged, 12— 14 plate, 240—241, 245
geometry of, 47 precoat, 18
homogeneous, 22 pressure, 118— 122, 227, 240
278 Water, Wastewater, and Sludge Filtration

rapid, see Deep bed filtration FTU, see Formazine turbidity units
sand, see Sand filters Fumic acid, 71
sheet, 240
shell, 240—241 G
surface, 130, 133— 134
upflow, 79, 88 Gel formation, 162— 163, 177
vacuum, 122— 123, 225— 231 Gibbs expression, 205
variable volume, 240— 242 Glass manufacturing, 235
Filter yield, 232 Gravity forces, 18, 23, 34
Filtrate, 9— 11, 18, 134 Gravity number, 25
Filtration
advanced systems of, 65—69 H
cartridge, see Cartridge filtration
dead-end, 176 Hamaker constant, 34, 36
declining-rate, 2— 3 Happel’s cell model, 29, 36, 38— 39
deep bed, see Deep bed filtration Headloss, 5, 7— 11, 64— 65, see also Pressure drop
direct, see Direct filtration Hollow fiber devices, 144, 146— 148, 161— 162
dual-media, 2 Hospitals, 165
dynamic, 174 Humic acid, 61—62, 71,216
high-rate, 3 Hydraulic cleaning, 187
in-series, 68— 73 Hydraulic resistance, 179
precoat, see Precoat filtration Hydrodynamic forces, 33
pressure, see Pressure filtration Hydrogen ions, 203, 209, 217
static, 176 Hydrophobic colloids, 62— 63
tangential, 174 Hydroxide ions, 202, 204
vacuum, see Vacuum filtration Hydroxides, 209
Filtration aids, 63 Hydroxyl ions, 203—204, 216— 217
Filtration coefficient, 9— 19
Filtration constant, 243 i
Filtration equation, 238
Filtration time, 244— 245 Imperforate basket centrifuge, 252—254
Fineness coefficient, 60 Inertia number, 25
Flocculant, 110 Interception number, 25
Flocculation Interfilter washing filtration, 5— 7
bridging mechanism and, 63 Iron, 61—62, 71— 72
coagulation-, 58— 59 Iron sulfates, 62
contact, 60, 66
control of, 3 K
pretreatment of, 251
prior to filtration, 124 Kozeny’s equation, 11— 12
Flocculation basin, 4
Flow rate, 2— 3, 130— 132 L
Flow systems, 195— 199
Flux Lactose, 219
of counterion, 209 Lactoserum, 156, 166— 167
density of, 204—208, 212— 213 Laplace transformation, 27
effect of back transport on, 177— 178 Leachate, 80— 82, 91
increase by CMF, 184— 187 Lignin, 168
ionic, 205— 208, 212 Limpidity, 122
mass, 244— 245 London-van der Waals' force, 34
permeate, 163
relation to membrane surface, 188 M
relation to transmembrane pressure, 164
Flux-concentration relationship, 183 Manganese, 61
Flux equation, 149— 150, 204— 206 Manifold, 211—212
Forces, 18, 23, 33— 36, 38 Mass balance, 18, 20, 177
Formazine turbidity units (FTU), 60 Mass flux, 244— 245
Foulants, 216 Mass recovery, 252, 255, 257, 264
Fouling, 144— 148, 162— 163, 217 Mass transfer, 18
Frame filter press, 240— 241 Medical applications, 155, 165
 279

Medium resistance, 239 special meshes for, 108

Membrane potential, 202 water quality measurement in, 102— 106
Membranes Milk, 219
anion-exchange, 204 Minamisenju plant, 78—80, 87— 88
anion-permeable, 192, 194, 204, 216 Molecular weights, 160— 162
anion-selective, 192, 199
bipolar, 204 N
for cartridge filter, 130
cation-permeable, 194, 204, 216 Nephelometric turbidity unit (NTU), 69— 72
cation-selective, 192, 199 Nemst-Einstein relationship, 204
heterogeneous, 199 Nemst-Planck equation, 205
homogeneous, 199 Night soil, 79—82, 90
hydraulic resistance of, 164 Non-Brownian particles, 36— 42
ion-permeable, 199— 202, 206— 209 N 03 removal, 153
ion transport through, 204— 207 Nozzle-type centrifuge, 266— 271
for microfiltration, 174— 176 NTU, see Nephelometric turbidity unit
microheterogeneous, 199
microporous, 174— 176 o
nuclear, 137
plastic, 137 OFSY®, see Omnifiltration System
pore size distribution in, 131 Omnifiltration System (OFSY®), 68—74
for reverse osmosis, 140— 152 Osmosis, reverse, see Reverse osmosis
cellulose acetate, 140— 142 Osmotic flow, 204, 208
fouling of, 144— 148 Ozone, 61, 108
in hollow fiber devices, 144, 146
polymer, 140— 142 P
in spiral wound devices, 144— 146
structure of, 141, 148— 150 Painting industry, 167
in tubular devices, 142— 144 Paper industry, see Pulp and paper industry
semipermeable, 192 Parallel circulation, 194, 196— 199
for ultrafiltration, 160— 161, 164 Particle deposition mechanisms, 30— 49
Membrane unit, 204, 206 in clean filter bed, 30— 32
Metal coagulants, 62— 63 filter efficiency of, 32— 46
Metal-finishing waste, 235 for Brownian particles, 36
MF, see Microfiltration in dynamic filtration phase, 46— 49
Michau diagram, 28 in liquid deep bed filtration, 33— 36
Microfiltration (MF), 173— 189, see also Micro- for non-Brownian particles, 36— 46
straining adhesion probability, 42—46
basic plant design for, 180— 184 transport efficiency, 36— 42
crossflow, see Crossflow microfiltration Particle detachment, 25— 28
differences from UF and RO, 170 Particle diffusion coefficient, 25
economics of, 187— 188 Particle size distribution, 179, 251
flux increase by, 184— 187 Particle transport, 18
membranes for, 174— 176 Perlite, 118, 124, 126
pretreatment with, 147 Permeability, 202— 211
Micro glass spheres, 44 current efficiency of stack, 210— 211
Microstraining, 101— 115, see also Microfiltration ion transport through membrane unit, 204— 207
application of, 102 quality considerations in, 202— 204
combined processes using, 110 transport and efficiency in unit cell, 207— 210
mechanics of, 102 pH adjustment, 62, 216
microstrainers for, 103— 115 Pharmaceutical industry, 137, 155, 165, 184
advantages of, 111, 115 Phosphorus, 72—73
ancillary equipment for, 115 Plate and frame press, 240— 241, 245
cross-sectional view of, 103 Plating factory, 85— 86, 98
fabric for, 107— 108 Polarization layers, 210
hydraulic layout of, 112 Polarization parameter, 207, 213
industrial, 110 Poly aery lonitrite, 160— 161
installation of, 110 Polyamide, 140— 142, 161, 175
plant layout of, 113 Polyanions, 216
results and applications of, 108— 110 Polycarbonate, 160
sizing and dimensions of, 111, 114 Polyelectrolytes, 3, 63, 69, 233, 251, 259
280 Water, Wastewater, and Sludge Filtration

Polyfuran, 161 Salinity, 215

Polymers, 68, 140, 174— 176, 233 Salt, 148, 211— 212, 216
Polypropylene, 184, 187, 234, 246 Salt water, 142, 152— 153, 192— 194, 197, 209, 213
Polystyrene, 199 Sand filters
Polysulfone, 160— 161 coarse, 4
Polyvinyl chloride, 160 in Minamisenju plant, 88
Pore size, 130— 131, 160, 176 quartz, 39-^40
Porosity, 9, 20 rapid, 18
Potentiometry, 215 single-layer, 63—64
Precoat filtration, 117— 128 slow, 2
applications of, 124— 128 for swimming pool water, 127
filter aids for, 118, 123 Sand filtration, 79
pressure filters for, 118— 122 Scroll discharge solid bowl centrifuge
rotary vacuum, 122— 123 capacities of, 262
theory of, 118 early, 251— 252
Preflocculation, 59, 68 high-speed, 259
Preozonization, 110 low-speed, 259
Presses, 240— 241, 243, 245, 251 operating parameters of, 261
Pressure drop, 18, 28— 30, 135— 136, 243, see also performance of, 260— 263
Headloss principles of operation of, 256— 260
Pressure filtration, 237— 247 sectional view of, 266
applications of, 245—246 theoretical predictions of, 263—265
design parameters for, 243— 245 Seawater, 142, 152— 153
filters for, 240— 243 Sedimentation, 78, 84
principle and theory of, 238— 240 Sedimentation basin, 4
Pretreatment, 136, 138, 144— 148, 162— 163, 186— Sedimentation centrifuges, 252—256
187, 216 Sensors, 215
Pulp and paper industry wastewater, 83— 84, 95, Separation efficiency, 258
167— 169, 220, 235, 245 Separation grade, 263
Sewage water, 72, 78— 79, 109— 110
Q Sieve-type mechanism, 148
o-concept, 254— 255, 263, 271
Quartz particles, 39—40, 44, 48 Skimmer, 250
Sludge, 79, 228, 232— 235, 260, see also Centri­
R fuges; Dewatering
Soda, 220
Rejection mechanism, 148— 150 Sodium bicarbonate, 216
Rejection rate, 164 Sodium hydroxide, 163, 169
Removal efficiency, 65—66 Sodium hypochlorite, 163, 169
Research laboratory waste, 87, 99 Soft drink factory, 85— 86, 97
Reverse osmosis (RO), 139— 158 Solid bowl centrifuge, 253
applications of, 152— 157 Solid waste dumpsites, 80— 82, 91
design for, 152 Solute flux, 149— 150
devices for, 142— 144 Solvent flow, 208
differences from UF and MF, 170 Spacer, 195— 196
fouling and pretreatment for, 144— 148 Spacer gaskets, 194, 217
limitations of, 157— 158 Specific deposit, 9, 21
membranes for, 140— 142 Specific resistance, 239— 240
process of, 140 Spin machines, 250
theoretical concepts in, 148— 152 Spiral-wound devices, 144— 148, 160— 162, 181
Ripening period, 22 Stack, see Electrodialysis stack
RO, see Reverse osmosis Steel-making factory, 81— 83, 94
Rotational speed Steric forces, 35
of bowl, 258, 263 Sterility, 216
comparatively high, 255 Sterilization, 217
of disc-type centrifuge, 269 Stochastic forces, 18, 23, 33
and migration velocity, 250 Stokes equation, 33, 253
restriction of, 256, 272 Stokes’ law, 254— 255
typical, 262 Stream function, 31— 32
Structural forces, 35
s Sulfonate groups, 199
Sulfuric casein, 219
 281

Surface protrusion, 39— 41 Void fraction, 31

Suspended solids, 4
Swimming pool water, 127— 128
w
T
Wastewater
Tangential velocity, 179— 181, 188 recycling, 79
Textile industry, 168 treatment of
THM, see Trihalomethanes by centrifuges, 251—252, 256— 257
Thomas equation, 27 for industrial sources
Titanium dioxide, 235 beer factory, 84— 85, 96
Transmembrane pressure, 164— 165, 179— 180 by microstraining, 109
Transmission curve, 133— 134 plating factory, 85— 86, 98
Transport efficiency, 33, 36— 42, 207— 210 pulp and paper industry, 83— 84, 95
Transport mechanisms, 30, 199— 200 research laboratory, 87, 99
Transport model, 149— 150 soft drink factory, 85, 97
Transport number, 200— 202, 207— 210 steel-making, 81— 83, 92—94
Trihalomethanes (THM), 61, 71 by vacuum filtration, 235
Tubular devices, 142— 144, 147, 160, 162, 180— for nonindustrial sources
181, 184 domestic, 165— 166, 235
Tubular pinch effect, 178 leachate, 80— 81, 91
Turbidity, 18, 61, 66— 70 municipal, 156, 252, 257— 259, 270
night soil treatment, 79— 90
u tertiary treatment of sewage, 78—79, 87— 88
water recycling in large buildings, 79, 88— 89
UF, see Ultrafiltration by precoat filtration, 128
Ultrafiltration (UF), 148, 159— 170 by reverse osmosis, 155— 157
advantages of, 168— 169 by ultrafiltration, 165— 168
applications for, 164— 168 Water
description of, 160 backwash, see Backwashing
devices for, 160— 162 brackish, 194, 197, 216, 218
differences from RO and MF, 170 demineralization of, 219
disadvantages of, 169— 170 distilled, 138
economics of, 168 drinking, 215—216
membranes for, 160 ground, 128
pretreatment for, 162— 163 industrial, 109, 219, see also Wastewater
pretreatment with, 148 limpid, 70
theory of, 163— 164 potable, 78—79, 153,215,218
Unit cells pure, 153— 155
assembling, 192— 194 reuse of, 220
concentration potential in, 210— 211 for RO module, 148
electricity through, 212 salt, see Salt water
flux density in, 213 sewage, see Sewage water
solution circulation in, 194— 195 sterilized, 155, 165
transport and efficiency in, 207— 210 surface, 71, 128
swimming pool, 127— 128
V ultrapure, 137, 153— 155, 164
Water splitting, 202— 204
Vacuum filtration, 225— 236 Water transfer, 208— 209, 212
advantages and disadvantages of, 235— 236 Water treatment
applications of, 233—235 conventional, 3— 4, 74
filter design in, 227— 231 different parameters in, 23
operating parameters in, 231—232 direct filtration in, 58—60
operation and types of, 226— 227 with microstrainer, see Microstraining
performance of, 232— 233 with precoat filtration, 124— 127
Van der Waals’ forces, 35—36, 38 sludges in, 233—235