EPA-670/2-74-032
April 1974
Environmental Protection Technology Series
THE ROLE OF POLYELECTROLYTES
IN FILTRATION PROCESSES
PR
National Environmental Research Center
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
-------
EPA-670/2-74-032
April 1974
THE ROLE OF POLYELECTROLYTES IN FILTRATION PROCESSES
by
Charles R. O'Melia
Department of Environmental Sciences and Engineering
University of North Carolina
Chapel Hill, N.C. 27514
Program Element No. 1BB043
Grant No. R-800351
Project Officer
Sidney A. Hannah.
Advanced Waste Treatment Research Laboratory
National Environmental Research Center
Cincinnati, Ohio 45268
NATIONAL ENVIRONMENTAL RESEARCH CENTER
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
FOREWORD
Man and his environment must be protected from the adverse effects
of pesticides, radiation, noise and other forms of pollution, and the
unwise management of solid waste. Efforts to protect the environment
require a focus that recognizes the interplay between the components of
our physical environment—air, water, and land. The National Environ-
mental Research Centers provide this multidisciplinary focus through
programs engaged in
• studies on the effects of environmental contaminants on
man and the biosphere, and
• a search for ways to prevent contamination and to recycle
valuable resources.
As part of these activities, the study described here presents
results of an investigation to determine applications of destabilizing
chemicals in filtration for the treatment of wastewaters.
A. W. Breidenbach, Ph.D.
Director
National Environmental
Research Center, Cincinnati
il
-------
REVIEW NOTICE
The National Environmental Research Center—
Cincinnati has reviewed this report and approved its
publication. Approval does not signify that the
contents necessarily reflect the views and policies
of the U.S. Environmental Protection Agency, nor
does mention of trade names or commercial products
constitute endorsement or recommendation for use.
iii
-------
ABSTRACT
This research ha.s been conducted (i) to determine how destabilizing
chemicals function in enhancing the effectiveness of filtration processes,
and (ii) to consider selected applications of destabilizing chemicals
in filtration for the treatment of wastewaters.
The investigations have included (i) laboratory experiments using
polymers and latex suspensions, (ii) laboratory and pilot plant experi-
ments using alum, polymers, and trickling filter effluent, and (iii) lab-
oratory experiments using polymers and calcium phosphate suspensions.
The report includes conclusions regarding the mechanisms of polymer
action in filtration, the results that can be obtained using polymers
as filter-aids, and the application of this knowledge to the design of
filtration processes for wastewater treatment.
iv
-------
CONTENTS
Section Title Page
I CONCLUSIONS 1
II RECOMMENDATIONS 5
III INTRODUCTION 7
Objectives and Scope 7
The Filtration Model 7
The Transport Step 8
The Attachment Step 10
IV EXPERIMENTAL APPARATUS AND PROCEDURES 13
Filtration Experiments Using Latex Sus-
pensions 13
Destabilization Experiments 16
Filtration of Trickling Filter Effluent. ... 19
Laboratory Experiment 19
Pilot Plant Experiments 19
Filtration of Calcium Phosphate Suspensions. . 25
Phosphate Precipitation Tests 25
Destabilization Tests 26
Filtration Experiments 26
V RESULTS AND DISCUSSION 27
Filtration and Destabilization Experiments
Using Latex Suspensions 27
Filters with Uncoated Media 27
Filters with Coated Media 29
Jar Test Anomalies , . 32
The Mechanism of Polymer Action in Fil-
tration 34
Stoi.chlometry in Filtration 41
Molecular Weight of the Polymer and
Filter Performance 41
-------
Section Title
Suspensions of Low Particle Surface Concentration. . 43
Suspended Particle Size and Filter Performance. ... 46
Filtration of Trickling Filter Effluent 49
Laboratory Studies 49
Pilot Plant Studies 52
Destabilization Studies 52
Effects of Pretreatment on Filtration 55
Effects of Filter Characteristics 61
Procedures for Filter Design 63
Filtration of Calcium Phosphate Suspensions. 67
Coagulation Studies 67
Filtration Tests 69
Development of Filtration—A Theoretical Perspective. 71
VI REFERENCES 75
VII PUBLICATIONS 79
VIII LIST OF SYMBOLS 80
vi
-------
FIGURES
Number Title Page
1 Schematic Diagram of Filter Apparatus,
Laboratory Experiments 14
2 Schematic Diagram of Filter Apparatus,
Pilot Plant Experiments 20
3 Comparison of Jar Test Results (A) with
Filter Performance, Uncoated Media (B,C) . • . . 28
4 Comparison of Jar Test Results (A) with
Filter Performance, Precoated Media (B,C) • • • 31
5 Comparison of Jar Test (A) and Pilot Filter
(B) for Determining the Optimum Polymer Dosage
for Filtration 33
6 Effects of PEI-18 on Residual Turbidity (A) and
Electrophoretic Mobility (B) of 0.109 y Latex
Suspension 38
7 Effects of PEI-18 on Residual Turbidity (A) and
Electrophoretic Mobility (B) of 1.099 V Latex
Suspension 39
8 Stoichiometry in Filtration: Optimum Polymer
Dosage (PEI-18) as a Function of Colloid Con-
centration 42
9 Effects of Molecular Weight of Polymer on
Filter Performance at the Optimum Polymer
Dosage 44
10 Effects of Polymer on Coagulation (A) and Fil-
tration (B,C) of a Low Turbidity Suspension • • 45
11 Effects of Suspended Particle Size on Filter
Performance at the Optimum Polymer Dosage • • • 48
12 Effects of Polymer on Coagulation (A) and Fil-
tration (B,C,D) of a Secondary Effluent from
Mason Farm Wastewater Treatment Plant 50
v±±
-------
Number Title Page
13 Effects of Alum on Coagulation of Secondary
Effluent. Residual Turbidity vs. Alum Dose- . • 53
14 Effects of Alum on Coagulation of Secondary
Effluent. Residual BOD^, TOC, SS, and TP vs.
Alum Dose 54
15 Effects of Purifloc A-23 on Coagulation of an
Alum-Treated Secondary Effluent 56
16 Effects of Purifloc A-23 on Filter Performance. 57
17 Effects of Purifloc A-23 on Filter Performance. 58
18 Effects of Purifloc A-23 on Filter Performance. 59
19 Effects of Purifloc A-23 on Filter Performance. 60
20 Effects of Media Size on Filter Performance.
Head Loss vs. Filtration Time 62
21 Effects of Media Size on Filter Performance.
Residual Turbidity vs. Filtration Time 64
22 Determination of Optimum Media Size 66
23 Effects of Cat-floe on the Coagulation of
Calcium Phosphate ....... 68
24 Effects of Cat-floe on the Filtration of
Calcium Phosphate 70
viii
-------
TABLES
Number Title Page
1 Wastewater Characteristics 21
2 Characteristics of Filter Media 24
3 Latex Charge and the Optimum Polymer Dose- • 36
4 Charge of Polyethylenimine and Optimum
Dose at pH = 8.2 37
5 Development of Filtration Processes 72
ix
-------
ACKNOWLEDGEMENTS
This project was directed by Professor Charles R. O'Melia, Professor
of Environmental Sciences and Engineering, University of North Carolina
at Chapel Hill. Dr. Mohammad T. Habibian conducted the studies dealing
with the filtration of latex suspensions, the mechanisms by which
destabilization is accomplished by polymers in filtration, and also ini-
tiated studies of the filtration of trickling filter effluent. Mr. Cheng
H. Chiang conducted the investigations of the filtration of calcium phos-
phate suspensions. Mr. Ali K. T. Basaran conducted the main portion of
the investigations using trickling filter effluent, with the assistance
of Mr. William B. Snodgrass.
Appreciation is extended to Professor James C. Brown, Director of
the UNC Wastewater Research Center, for his support of the project and
for making the facilities of the Center available for this study.
The financial support of this project by the Environmental Protection
Agency and the assistance provided by Dr. Sidney A. Hannah, the Grant
Project Officer, are greatly appreciated.
-------
SECTION I
CONCLUSIONS
1. Particles are removed from suspension in wastewater filtration by
being transported to the vicinity of the external surface of the
filter grains, and then attaching to the solid material at that
interface. Particle transport is determined by the design and
operation of the filters, and by the size and density of the parti-
cles to be removed. The attachment within the bed generally re-
quires chemical destabilization in a pretreatment step prior to
filtration.
2. Particle transport within the pores of a filter bed is a very ef-
fective process in conventional water and wastewater filtration.
Given proper chemical pretreatment, conventional filter beds are
capable of removing virtually all particles applied to them, re-
gardless of suspended particle size. Viruses (0.01 microns) and
biological floes (100 microns) should be filterable. When effective
filtration is not obtained, failure should be attributed to poor
chemical pretreatment.
3. Regarding mechanisms of particle transport within filter beds, the
following statements can be made:
a. There exists a size of particles for which the transport (and
hence removal) efficiency is a minimum. For conventional fil-
tration practice, this critical suspended particle size is
about 1 micron. Nevertheless, sufficient particle transport
Is, provided in these systems to permit effective filtration.
b. For suspended particles larger than 1 micron, transport within
the filter pores is accomplished by sedimentation and (or)
simple fluid motion (interception). The removal which results
increases with increasing size of the suspended particles.
c. For particles smaller than 1 micron, transport is accomplished
by diffusion (Brownian motion). Removal increases with decreas-
ing size of the suspended particles.
-------
d. Mathematical models for particle transport within filter beds
have been developed which are in qualitative agreement with
the results obtained experimentally in laboratory systems.
The chemical destabilization required for effective filtration is
similar to that required for effective coagulation. This chemical
pretreatment was observed to be required for effective filtration
of all systems considered in this research, including suspensions
of latex particles, synthetic calcium phosphate precipitates, and
trickling filter effluent.
When chemicals are used to provide for the destabllization of sus-
pended particles, jar tests may be used to determine the proper
type and dosage of chemical. The optimum dosage in a jar test
(a batch coagulation test) is the chemical dosage that will pro-
vide the best effluent quality in a filtration process.
For many conventional filters, this best effluent quality is pro-
duced at the expense of short filter runs due to rapid build-up of
headloss. Effective attachment within filter beds requires the
use of bi- or multi-media filters, upflow filters, moving bed fil-
ters, etc.
When polymers are used as destabilizing chemicals, overdosing can
occur due to restabilization of the suspended particles. A stol-
chiometry exists between the optimum polymer dosage and the concen-
tration of material to be filtered from the water.
For the cationic polymer series investigated (polyethylenimines),
the optimum polymer dosage was independent of the molecular weight
of the polymer, but the removal efficiency of filters operated at
this optimum dosage increased with increasing molecular weight.
For direct filtration, precoating of the filter media is essential
for effective removal of suspended particles. Precoating is also
useful even when prior coagulation and settling are employed, since
it greatly reduces the initial "ripening" period during a filter
run. It is probable that precoating can be accomplished during the
backwashing step of a conventional filtration cycle.
-------
10. For a properly conditioned influent, filter efficiency is, to a
first approximation, independent of the concentration of suspended
materials. Direct filtration of low-turbidity systems is feasible
and can lead to lower chemical requirements than when a conventional
rapid mix-flocculation-settling-filtration sequence is used. Di-
rect filtration will require instrumentation for adjusting the
chemical dosage, and the services of competent operators.
11. The head loss developed within a filter bed varies inversely with
the size of the particles removed from suspension. Stated another
way, sub-micron particles removed in conventional filters produce
much, greater head leases than an equal weight of particles of
larger size.
12. Trickling filter effluent apparently contains large quantities of
sub-micron colloidal material. This is evidenced by C&) the need
for large polymer dosages in coagulation and filtration, (b) the
shape of the jar test curves in the underdosed region, with some
polymer doaages. producing increased turbidities, and (c) the ex-
cessive head, losses produced during filtration.
13. Studiea of the direct filtration of trickling filter effluent led
to the following conclusions:
a. Direct filtration, without chemical pretreatment and using con-
ventional filters was ineffectual in removing suspended solids,
BOD, and TOC.
b. Direct filtration using conventional filters and a cationic
polymer as the sole destabilizing chemical is possible, but
polymer dosages are high and head losses are excessive.
c. Direct filtration using alum for chemical destabilization pro-
duces floes, which, readily pass through conventional filter beds.
d. It was not possible to find an. anionic polymer capable of use
as the sole destabilizing chemical for the filtration of trick-
ling filter effluent.
e. Alum and an anionic polymer (Purifloc A-23) used in series pro-
vided effective destabilization for the direct filtration of
trickling filter effluent. Efficient removal of suspended
-------
solids, turbidity, BOD, TOG, and total phosphorus could be
achieved.
f. The optimum doses of alum and Purifloc A-23 for producing the
best effluent quality in direct filtration could be determined
using jar tests.
g. Short filter runs due to head loss build-up occurred using alum
and Purifloc A-23 in series prior to conventional filter beds.
The use of larger media provided longer filter runs. For the
conditions of this research a media size of about 5 mm permitted
breakthrough at about the same time as the available head loss
was consumed.
14. For the direct filtration of calcium phosphate suspensions prepared
at pH 8 to 9.5 in the laboratory, the following conclusions can be
made:
a. Direct filtration without chemical destabilization is ineffec-
tive in achieving removal.
b. With proper chemical destabilization, 80 to 90% of the phos-
phorus in the suspensions was removed by direct filtration.
c. It was possible to determine the optimum type and dosage of
chemical (a polymer was finally used) for this filtration using
jar tests.
15. In any deep-bed filtration, process, effluent quality is primarily
determined by the chemical pretreatment used. The rate of head
loss increase, however, is also affected by the size of the filter
media. A suitable design procedure could be to select a pretreat-
ment process which is effective by using jar tests, and then to de-
sign filters to treat this destabilized suspension using pilot fil-
tration experiments.
-------
SECTION II
RECOMMENDATIONS
1. The design of filter beds cannot be separated from the design of
pretreatment systems without risking serious difficulties when the
full-scale filter system is placed into service. No filter should
be designed for any application without consideration of the pre-
treatment that may be required to permit effective filtration. The
technology is. available to permit a wide range of selection among
pretreatment processes (^operating costs) and filters (capital costs)
for achieving stated objectives for water quantity and quality.
2. Consideration should be given to changing current procedures for
designing filters. The engineer could first design the pretreat-
ment system, and then design a filter system to remove the suspen-
ded materials which have been destabilized by this pretreatment.
If appropriate, several pretreatment-f liter bed combinations could
be developed and compared on a coat basis. Batch coagulation (jar)
teats can be used to provide the basis for the design of the pre-
treatment system; pilot filtration experiments are then needed to
design the filter system.
3. Direct filtration should be considered in the treatment of low tur-
bidity waters, thereby permitting a reduction in chemical require-
ments. This, process requires excellent operation. Present develop-
ment of instrumentation to provide feedback control of the required
chemicala should be continued.
4. The feasibility of reducing excessive head losses in some direct fil-
tration processes by providing flocculation without sedimentation
should be investigated.
5. Precoating of filter beds by the addition of appropriate chemicals
during backwashing should be considered in the design and operation
of filter systems.
-------
6. The methods of backwashing filter beds containing large media (5 mm
or so) should be investigated.
7. Pilot plant experiments of the direct filtration of calcium phos-
phate suspensions should be conducted. These investigations should
use polymers as filter-aids, and use the pH range 8 to 9.5.
-------
SECTION III
INTRODUCTION
Objectives and Scope
This research has been conducted (i) to determine how destabilizing
chemicals function in enhancing the effectiveness of filtration processes,
and (ii) to consider selected applications of destabilizing chemicals in
filtration for the treatment of wastewaters.
The investigations have included (i) laboratory experiments dealing
with the effects of polymers on the removal of colloidal and suspended par-
ticles from aqueous suspensions by packed-bed filters, (ii) laboratory and
pilot-plant experiments dealing with the direct filtration of trickling-
filter effluent from the wastewater treatment plant in Chapel Hill, N.C.,
and (iii) laboratory experiments dealing with the effects of polymers on
the filtration of suspensions of calcium phosphate. Associated with all
these filtration experiments were investigations of the effectiveness of
polymers and other chemicals as destabilizing agents for the suspended ma-
terials to be removed by filtration. Some limited experiments to delineate
the mechanisms of this destabilization were also conducted.
The approach used in this research is based upon a model for water and
wastewater filtration which has been developed during the past few years
(1-6). A brief summary of this model is included in this report to outline
the rationale used in developing the experimental program.
The Filtration Model
The filtration of suspended and colloidal particles from a water or
wastewater can be considered as involving two separate and distinct steps:
(i) the transport of the suspended particles to the immediate vicinity of
the filter grains, and (ii) the attachment of the suspended particles to
the filter grains or to another particle which has previously been deposited
-------
in the bed (1, 7). When viewed in this perspective, filtration and coagu-
lation processes are quite similar. Previous investigators have evaluated
the overall rate of a coagulation process by determining the rate at which
collisions occur between particles by fluid motion (orthokinetic floccula-
tion) and by Brownian diffusion (perikinetic flocculation), multiplied by
a "collision efficiency factor" which reflects the ability of chemical co-
agulants to destabilize colloidal particles and thereby permit attachment
when contacts occur. A similar approach is used herein to describe filtra-
tion. Efficient particle removal requires (i) that the filter bed be de-
signed and operated to provide sufficient contact opportunities and (ii)
that the particles to be filtered be rendered unstable or "sticky", usually
by the addition of an appropriate type and dosage of chemical either di-
rectly to the filter influent or in a prior coagulation process.
The Transport Step
Based on theories developed and tested in the field of air filtration
(8), a model for the transport step in water and wastewater filtration has
been proposed (2) and developed (3-5). Three physical mechanisms were
found to be significant in such processes, viz., interception, sedimentation,
and Brownian diffusion. The relative importance of these mechanisms de-
pends upon several physical parameters, the most significant of which can
often be the size of the particles to be filtered from suspension. Based
on this model and some preliminary experiments, the following conclusions
have been reached:
1. Both theory and experiment show that there exists a critical size
at which the suspended particles have a minimum contact opportu-
nity and hence a minimum removal efficiency. This critical sus-
pended-particle size is in the order of one micron.
2. For suspended particles which are smaller than this critical size,
filter efficiency Increases with decreasing particle size. Effec-
tive particle transport is provided by Brownian diffusion. This
permits the conclusion that a conventional rapid sand filter pro-
vides ample contact opportunities for the removal of viruses (ca.
0.01 to 0.1 microns in size). Some authors have mistakenly con-
cluded that such transport is not possible C9» 10).
8
-------
3. For particles larger than the critical size region, sedimentation
and interception combine to produce sufficient contact opportuni-
ties so that efficient removal is achieved in conventional filter
beds when attachment is effective. Filtration efficiency increases
with increasing suspended-particle size in this region.
The important work of FitzPatrick and Spielman (11, 12) should also
be noted. These investigators have developed a combined physical-chemical
model which includes particle transport and suspended particle-filter media
interactions. This model is conceptually more complete than the work re-
ported here.
Some useful similarities exist between particle transport in coagula-
tion processes and in filtration processes. Particle transport in both
processes can be considered separately in two distinct regions (e.g., or-
thokinetic and perikinetic flocculation). In both processes diffusion con-
trols the transport rate for particles smaller than about 1 micron (e.g.,
perikinetic flocculation). In both cases mathematical models are avail-
able for making quantitative estimates of the effects of pertinent para-
meters on the transport rate in either region. Smoluchowski's equations
as extended and applied by Camp (13) provide a quantitative basis for the
design of flocculators; Yao's model has the potential of providing a simi-
lar base for the design of filters (3).
An important difference arises in comparing coagulation and filtration
processes. The detention time which must be provided to achieve a given
degree of aggregation by coagulation increases as the concentration of par-
ticles to be aggregated decreases. In contrast, the theoretical model for
water and wastewater filtration (3) indicates that removal efficiency in
filtration, to a first approximation, is independent of particle concentra-
tion. Theory and laboratory experiments (5) indicate that a conventional
sand bed can effectively remove particles of any size if the attachment
step is effective. As a result, filtration without prior coagulation and
sedimentation but with direct polymer addition may offer economies for the
removal of colloidal particles from waters containing low but objectionable
concentrations of colloidal materials.
-------
Attachment Step
The attachment step is affected by both physical and chemical factors.
The significance of some chemical parameters has been demonstrated by se-
veral investigators. Conley and Pitman observed that the addition of a
polyacrylamide to the filter influent markedly improved the removal of alum
floe particles (14). More recently Tchobanoglous demonstrated that cati-
onic polyelectrolytes can be used as filter aids to enhance the removal of
suspended particles from secondary effluent (15). These improvements in
particle attachment are due to the significant changes brought about by
polyelectrolytes in the colloid-chemical characteristics of filter influ-
ents* Despite suck remarkable effects, little is known about selecting
the type and dosage of filter aids. The significance of the molecular
weight of the polymers in enhancing the filter performance and the mecha-
nism (s) of filter-aid action are not known. The process is not well under-
stood and no comprehensive conceptual models are available.
Experiments reported in this report were designed to test the proposal
that a better understanding of the role of chemical parameters in particle
attachment in filtration could be obtained by considering the chemical as-
pects, of filtration and coagulation as analagous. At present levels of
knowledge of colloid chemistry, the similarity between the attachment step
in filtration and coagulation cannot be shown in mathematical terms. The
concept, however, becomes more evident by considering that in both processes
the particles should be made "sticky" (destabilized) so that they may attach
to each, other or to the filter grains upon contact.
At the interface of a liquid phase with solid particles such as the
colloids found in natural waters, a diffuse layer of charges exists. When.
two colloidal particles are brought into close proximity, the repulsive in-
teraction of their diffuse layers can produce an energy barrier which op-
posea interparticle attachment. To make the colloidal particles "sticky"
or destabilized the potential energy barrier between the contacting parti-
cles must be reduced or nullified.
At the beginning of a filtration cycle an energy barrier between the
particles and the filter grains can control the attachment of particles to
the filter grains. However, as filtration proceeds, the suspended particles
10
-------
coat the filter grains and further removal occurs primarily by the attach-
ment of particles in suspension to particles already deposited on filter
grains. Therefore, shortly after the commencement of a filtration cycle,
the magnitude of the energy barrier between particles in suspension and
particles deposited on filter grains controls the particle attachment in
the filtration process. The same interaction also governs interparticle
attachment in the coagulation process. Thus, similarities between the
attachment steps in coagulation and filtration should be expected.
The destabilization of colloids in a coagulation process frequently
involves the adsorption of a destabilizing chemical (coagulant) on the sur-
face of the colloid. Charge neutralization and/or bridging which can result
from this adsorption then permit aggregation when contacts occur. The fol-
lowing statements summarize pertinent aspects of this process:
1. An optimum dosage of polymer exists for the coagulation of a given
suspension.
2. Overdosing (restabilization) can occur due to charge reversal or
saturation of bridging sites.
3. The optimum dosage is directly related to the concentration of
colloid.
4. The effectiveness, of many polymers is related to their charge and
molecular weight.
5. The type and optimum dosage of chemical for a coagulation process
can be estimated using jar (batch) tests.
Based on the hypothesis that the effects of destabilizing chemicals are
similar in both coagulation and filtration', experiments were performed to
teat for the existence of, these phenomena in filtration. The possible ap-
plications of these concepts to the filtration of trickling filter effluent
and of calcium phosphate suspensions were then investigated.
11
-------
SECTION IV
EXPERIMENTAL APPARATUS AND PROCEDURES
The experiments may be divided into four groups: (a) filtration ex-
periments which were designed to evaluate the mechanisms by which polyelec-
trolytes function as filter aids through the use of latex suspensions, (b)
destabilization experiments which were conducted to determine the optimum
polymer dosages necessary for coagulating a given suspension and also to
study the mechanisms by which polymers bring about the destabilization of
colloids, (c) filtration experiments to test the application of these con-
cepts to the filtration of trickling filter effluent, and (d) filtration
experiments to test the application of these concepts to the filtration of
calcium phosphate suspensions.
Filtration Experiments Using Latex Suspensions
Filtration experiments were conducted in which the filter influents
received dosages of polymer equal to, less than and greater than the opti-
mum polymer dosage determined from a jar test. To provide for such experi-
ments a filter apparatus with six identical laboratory filters was construc-
ted (Fig. 1). Characteristics of the filter bed (size and shape of the
media grains, bed depth and porosity, surface characteristics of the filter
media), characteristics of the operation of the filter (filtration rate,
direction of flow, backwashing procedure), and the pH and ionic strength of
the filter influent were kept constant. The size and concentration of the
suspended particles and the dosages and molecular weight of the polymers
added to the filter influent were varied in a controlled manner. The six
filters were operated in parallel in the conventional downward direction
at a constant rate of 2 gpm/sq. ft. During each run one of the filters
did not receive any polymer and served as a blank, two of the filters were
underdosed, two were overdosed, and the sixth filter received a continuous
polymer dose equal to the optimum dose determined from a jar test. The
similarities between filtration and coagulation were then determined by
13
-------
OVERHEAD CLEAN WATER TANK OVERHEAD SUSPENSION TANK
OVERHEAD POLYMER TANK
BACKWASH TO WASTE
6 filter '<
columns, "
piping shewn!
for I filter
FROM POLYMER PUMP
iu TO OVERHEAD
POLYMER TANK
FROM BACKWASH LINE
SIX FLOW CONTROLLERS
frrfl
— OVERFLOW
IT RETURN TO
f-* POLYMER JARS
POLYMER
PUMP
POLYELECTROLYTE
SOLUTIONS
2 - DISTILLED WATER
3-GLASS WOOL
4-CONG. H2S04
SUSPENSION
TANK
Figure I. Schematic Diagram of Filter Apparatus,
Laboratory Experiments.
14
-------
comparing the performance of these filters with each other and with the
results of jar tests conducted on the same suspension.
Laboratory filters were made of plexiglass tubes having an internal
diameter of 2.63 cm. Glass beads (class B, Minnesota Mining and Manufac-
turing Co.) with a mean diameter of .38 mm and a standard deviation of .04
mm were used for filter media in most of the filter runs. The bed depth
and porosity were 14 cm and 36 percent respectively. Initial experiments
used uncoated media. Subsequently the filter beds were precoated with a
solution of the filter-aid prior to use. The precoating was found neces-
sary for the reduction of a long ripening period which was observed when
uncoated filter beds were employed.
Latex beads (Daw Chemical Go.) with mean diameters of 0.109, 1.099
and 7.6 microns were, utilized for the influent suspensions. These sizes
were selected to provide for evaluating the effectiveness of the continu-
ous, addition of polyelectrolytes in both, "orthokinetic" and "perikinetic"
filtration as well as at the critical particle size. Demineralized water
was used for diluting the concentrated latex suspensions provided by the
manufacturer. Sufficient sodium chloride and sodium bicarbonate were
added to the demineralized water to provide a concentration of 10~3 mole/1
of each in the final latex suspension. The suspension was brought in
equilibrium with, the laboratory atmosphere by air agitation. The tempera-
ture in. the laboratory was almost constant» 27°C 4^1, during the entire
period of the investigation. The pH of the suspension was approximately
8.2.
Eive homologs of polyethylenimine (PEI seriea, Dow Chemical Co.) with
molecular weights of 600, 1200, 1800, 40-60., QOQ and 50-100,000 were used
as filter-aids. In a few filter runs Cat-floe (Calgon Corporation) and
Primafloc C-7 (Rohm and Haas Company) were utilized. All of these poly-
mers are cationic. To reduce the adsorption of these polymers on glass-
ware used in polymer preparation, the glassware was coated frequently with
Sillclad (Clay Adams Co.).
A portion of the filter effluent was collected at appropriate time
intervals in a test tube for analysis. Absorbance measured with a Beckman
Model DB Spectrophotometer was used to determine the latex concentration in
15
-------
the effluent. Measurements were made at wavelengths of 270, 490, and 700 mp
for suspensions of .109, 1.099 and 7.6 y latex particles respectively. The
absorbance of the suspensions was observed to be a linear function of par-
ticle concentration, and hence the ratio of the optical density of the
filter effluent and influent was used as an index of the fraction of the
particles remaining in the filter effluent. The absorbance in such sys-
tems is affected by particle size, BO that the existence in the filter
effluent of particles which might have grown in size during the course of
filtration can reduce the accuracy of this evaluation. However the possi-
bility of significant particle growth is believed to be very small (6).
Filters were equipped with piezometer tubes above and below the beds
of media. The head losses across the filters were measured at the begin-
ning of each experiment, and at frequent time intervals during the filter
runs.
Destabilization Experiments
Destabilization experiments were conducted to determine the signifi-
cance of charge neutralization and/or bridging in bringing about the desta-
bilization of colloidal particles in filtration. It was intended to com-
pare the magnitude of the negative charge existing on suspended particles
with the positive charge existing on that quantity of polyelectrolyte ne-
cessary for optimum destabilization of these suspended particles. Such a
comparison was expected to provide some evidence for determining the dominao
mechanism(s) of polyelectrolyte action. To accomplish this, the optimum
polymer dosage required for the destabilization of a given suspension, the
charges of the suspended particles, and the charges of the polymer used for
particle destabilization were determined.
The optimum polymer dosage was determined by conducting jar tests using,
a six-position magnetic stirrer. The polyelectrolyte was added to the sus-
pension under vigorous agitation. This rapid stirring continued for 2 mi-
nutes followed by an appropriate period of slow stirring. The residual
light absorption (absorption by the suspension after coagulation and set-
tling) was used as the index of the extent of particle aggregation.
16
-------
Absorbance measurements were made as described previously.
The surface charge of the latex particles is due to an anionic etnulsi-
fier physically adsorbed on the particle surface and to the initiator anion
chemically bonded to the starting point of the polymer chains (16). Al-
though the exact chemical natures of the emulsifier and initiator anions
have not been reported by the latex manufacturer, the literature suggested
that oleic acid and the persulfate ion respectively might have been em-
ployed (16). Organic acids such as oleic acid have pK values of about
4.8 (17) and the sulfate radical would have a pK of about 1.7. The pH of
a dilute latex suspension was found to be about 6.0. Therefore, it seemed
plausible to measure the charge of the latex particles due to the emulsi-
fier anion by titration of a latex suspension with an acid. Conductotnetric
titration was employed using a Serfass Model RCM 15 Bl conductance bridge
with a platinum electrode having a cell constant of .098. The conductivity
measurements were conducted in accordance with the procedures described in
Standard Methods (18). Hydrochloric acid was used as the titrant.
The charge of the latex particles due to the initiator could not be
determined by this method. This charge is so small that the end point of
the conductometric titration is heavily shadowed by the interference of
other ions existing in the bulk solution. Procedures developed by
Vanderhoff et al. (16) were employed for this purpose and followed precise-
ly. The technique utilized exchange resins to remove all interfering ions
and to exchange the Na+ and K+ cations associated with the sulfate end
groups with H+ ions. The number of these H+ ions and thus the charge due
to the initiator anions can then be determined by a conductometric titra-
tion of the exchanged solutions. Dowex SOW cation exchange and the Dowex
1X8 anion exchange resins were used.
Acid-base titrations (17) were used to determine the number of charges
on the polyethylenimine molecules. The technique is based on the principle
of electroneutrality in the polymer solution. If pn+ is the total number
of positive charges on a given amount of polymer at a given pH, Na+ and Cl*~
the amount of charge due to the addition of base and acid during the titra-
tion, V the total volume of the solution, and if AQ~ and Co+ are the total
equivalent amounts of anions and cations initially present in the polymer
17
-------
solution, one may write the following:
C_+ + pn+ + V-H+ + Na+ = Cl" + VOH" + AQ (I)
If A and C + are known, then pn+ at any given pH can be determined from
the above equation.
Sodium and potassium were considered as the only cations that might
exist in the polymer s:ocks. Correspondingly, chloride, sulfate and
nitrate were considered as the probable anions in these polymer stocks.
The concentrations of sodium and potassium in the polymer solution were
measured using an atomic absorption spectrophotometer (Perkin-Elmer 303).
The chloride concentration was measured by potentiometric titration of a
polymer solution with silver nitrate. A Corning pH meter model 12 equipped
with glass and silver-silver chloride electrodes was used. The concentra-
tion of sulfate anion was measured by a turbidimetric method as described
in Standard Methods (18). In this method, sulfate ion is precipitated in
a hydrochloric acid medium with barium chloride in such a manner as to form
barium sulfate crystals of uniform size. The Beckman Model DB spectropho-
tometer was employed for measuring the absorbance of the resulting barium
sulfate suspension. Nitrate concentrations were determined by the brucine
method following the procedures listed in Standard Methods (18). The re-
action between nitrate and brucine yields a sulfur yellow color which can
be analyzed by colorimetric methods. The Beckman Model DB spectrophotome-
ter was used.
Titration of polymer solution was begun by addition of enough standard
hydrochloric acid to a given polymer solution to lower its pH to about 2.
The resulting solution was then titrated with standard sodium hydroxide
under a nitrogen atmosphere. The pH value was measured after each addition'
of a suitable increment of the titrant. The polymer charge at each pH was
then calculated by substituting the data obtained into Equation 1.
18
-------
Filtration of Trickling Filter Effluent
Laboratory Experiment
Initial experiments of wastewater filtration used the laboratory
filter apparatus described previously (Figure 1). Effluent from the
Mason Farm Wastewater Treatment Plant, Chapel Hill, N.C., was collected
and transported to the School of Public Health for use in the filtra-
tion experiments. Characteristics of the plant and its waste are de-
scribed in a subsequent section of this report.
Several polymers were tested in jar tests for their potential as
filter aids. Primafloc C-7 was selected for further study. Filter beds
were precoated with polymer prior to each run. Filter beds were com-
prised of coarse sand with a size of 1.3 mm (geometric mean of adjacent
sieve sizes) and had a depth of 5.5 inches. Filter runs used polymer
applied continuously at the optimum dose (based on jar tests), at an
"underdosed" concentration, and at an "overdosed'1 concentration. One
filter received no polymer after the precoating, and served as a blank.
Pilot Plant Experiments
A filtration apparatus comprised of four filters and associated
appurtenances was constructed (Figure 2). Secondary effluent was
pumped continuously from a secondary clarifier at the Mason Farm Treat-
ment Plant (Chapel Hill, H.C.) to a fifty-five gallon drum. A submer-
sible pump located inside a screen housing at mid-depth in the drum
was used to -pump the secondary effluent to otie compartment o£ a small
constant head tank located above the pilot filters. Secondary effluent
was then withdrawn to be treated by the pilot filter apparatus after
the appropriate chemical destabilization. The overflow from the constant
head tank was discharged to waste as was the overflow from the drum.
Pertinent characteristics of the secondary effluent are presented in
Table 1, together with a description of the raw sewage received by the
Mason Farm treatment plant. This treatment facility, a major part of
the TJNC-CH Waste Water Research Center, is a conventional high-rate
trickling filter plant.
19
-------
overflows
2
«
«
o
M \r
-txl-
filter Influents
four filter columns
used, only one is
shown here.
a.
i
1
1
1
1
1
— jn-^nir rvioa»a
i • ~*f
\
M
6
c
c
3
8
b.
**
s-
T
: backwash
to drain
it
T
- r
f
T.,
— T—
1
) — »^l
/ --
float
manometers
backwash main
7/rv7
flow
control
assembly
adjustable
drain
Figure 2, Schematic Diagram of the Apparatus, Pilot Plant Experiments
20
-------
TABLE 1
WASTEWATER CHARACTERISTICS
(after Hanson et al. (20))
TABLE 1
Flow (MGD)
BOD5 (mg/1)
Suspended Solids (mg/1)
Total Phosphorus (mg/1)
Total Dissolved Phosphorus
(mg/1 )
Total Organic Carbon (TOC)
Influent
2.77
172
245
11.8
178
I
Effluent
43.5
60.
9.2
5.85
61.1
21
-------
Based on a series of jar tests, aluminum sulfate (reagent grade
A12(SO^)3 '18H 0, hereafter termed alum) and Purifloc A-23 (Dow Chemi-
cal Co.) were selected as destabilizing chemicals. Purifloc A-23 is a
negatively-charged polyelectrolyte with a molecular weight in the order
of ten million (manufacturer's description). Alum was used as the first
coagulant or filter-aid; Purifloc A-23 was added to destabilize further
the floes which were formed by alum addition.
Alum stock solutions (31.5 gm/1) were prepared in five-gallon
carboys by dilution with. Chapel Hill tap water. The alum solution was
then pumped to a second and separate compartment in the small influent
constant-head tank (Figure 2), An appropriate quantity was withdrawn
to mix with the filter influent; the overflow from the constant-head
tank was returned to the carboy- containing the alum stock solution.
Purifloc A-23 stock solutions CO.1455 gms/1) were also stored in five-
gallon carboys. To facilitate dissolving the polymer, sufficient pow-
dered Purifloc A-23 to prepare 19 liters of stock solution was added to
about 75 ml of methanol in a one-liter bottle. While providing con-
stant agitation, 500 ml of tap water were added to the polymer-methan-
ol suspension. This mixture waa stirred vigorously until the polymer
dissolved, after which, it was poured into the five-gallon carboy and
diluted to 19 liters with tap water. During a filtration run, stock
polymer solution was pumped to a third and separate compartment In
the small constant-head tank (Figure 2). An appropriate quantity was
withdrawn to add to the secondary effluent-alum, mixture. Overflow
from the constant-head tank was returned to the carboy containing the
polymer stock solution. The secondary effluent, alum, and polymer
flows were metered individually using flowmeters. It was later found
that the methanol used to dissolve the polymer contributed soluble BOD
and TOG to the filter Influents.
Provision was made to provide a period of slow mixing prior to
filtration. The alum—secondary effluent mixture was allowed to flow
to a second and larger constant-head tank (Figure 2). This tank was
divided into three compartments and equipped with a paddle stirrer.
22
-------
Flow was in series through the three compartments. Depending upon the
number of filters in operation, the filtration rate, and the number
of compartments used, the flocculation period could be varied from 10
to 120 minutes. Provision was also made to permit by-passing of the
flocculation tank; the latter procedure was used in most experiments.
Polymer was added to the effluent from the flocculation tank (or to the
by-passed secondary effluent-alum mixture) just prior to filtration
(Figure 2).
Four pilot filters were constructed using plexiglas tubes having
an internal diameter of 5.5 inches. Filter media were prepared from
crushed granite which was screened using sieves made from aluminum
afreets and a nail punch set. Siave openings were uniform (1/32, 1/16,
3/32, 1/8, 5/32, 3/16, 1/4, 9/32, and 5/16 in., or 0.8, 1.6, 2.4, 3.2,
4.8, 6.4, 7.2, and 8.Q mm).. Media were sized by using the two adjacent
sieve sizes and also by Hazen's count and weigh, method C21)• A summary
of the. filter "beds and media sizes used in this research is presented
in Table 2. For media larger than 3 mm, wall effects were probably not
negligible., so that the actual removal by full-scale filters could be
even better than that observed in the pilot experiments.
Provision was made to operate the filter in both downflow and up-
flow directions. Piezometer tubes for measuring head loss were located
above and below: the beds, and at several points within the beds. Pro-
vision, for obtaining samples was made above and below the filters.
Filtration, was. at a constant rate, usually 2 gpm/sq. ft. The
flow rate through each filter was maintained constant using a float-
operated needle valve, controlled by a syphon C6> 22).
Samples were obtained every half-hour from the secondary effluent,
the filter influents., aad the filter effluents. Each sample was im-
mediately analysed for pH. and turbidity. A Hack model 2100 turbidi-
meter was used for turbidity measurement. Samples were then refrige-
rated and later analysed for total organic carbon (TOG), five-day bio-
chemical oxygen demand CBOD,.), suspended solids CSS), and total phos-
phorus CTP). The head loss across each filter was observed at half-
hour intervals. Floe penetration was observed visually at hourly internals.
TOG measurements vere performed with- a Beckman Model 915 two
23
-------
TABLE 2
CHARACTERISTICS OF FILTER MEDIA
ro
Media
No,
1
2
3
4
5
6
7
Sieve Openings
(in.)
Passing
2/32
3/32
4/32
6/32
8/32
9/32
10/32
Retained
1/32
2/32
3/32
4/32
4/32
5/32
5/32 !
Media Size,
Hazen (21)
(mm)
1.65
2,36
3,08
4. 1)
5.9
6.3
7.5
Specific
Gravity
(gm/cu.cm)
2.64
2.64
2.64
2,64
2.64
2.64
2.64
Media
Depth
(in.)
36
36
36
34
33
32
30
Bed
Porosity
0.4
0.4
0.4
0.4
0.4
0.4
0,4
-------
channel carbon analyser using procedures prescribed by the manufacturer.
Measurements of BOD and SS were made in accordance with Standard
Methods (18). Total phosphorus was determined by the stannous chloride
method using the Technicon Auto Analyser (23).
Filtration of Calcium Phosphate Suspensions
These experiments can be further subdivided into three parts:
(i) phosphate precipitation tests to evaluate the effectiveness of lime
under various conditions, (ii) coagulation tests to evaluate types and
dosages of polymers which could provide effective destabilization of
calcium phosphate suspensions, and (iii) filtration tests to evaluate
the effectiveness of these polymers for the removal of calcium phos-
phates from suspension by filtration.
Phosphate Precipitation Tests
Precipitation tests were conducted to describe the influence of pH,
inorganic carbon, calcium, and fluoride on the removal of phosphate by
lime. All experiments were conducted using deionized water and reagent
grade chemicals. Initially, sufficient aliquots of stock solutions of
NaHC03, Ca(N03)2«4H20, KH^O^-aH,^ and NaF were added to deionized
water to give a phosphorus concentration of 10 rag P/l and specified con-
centrations of inorganic carbon, calcium, and fluoride. The solution
was then transferred to 600 or 1000 ml beakers, after which the pH was
adjusted to the desired level with HC1 or NaOH and maintained constant
within + 0.1 pH units. The resulting suspensions were rapidly stirred
for 2 minutes, and then slowly stirred for an additional 20 minutes un-
less the effects of time on precipitation were under consideration.
Finally, 30 minutes of quiesent settling were provided. Two samples were
then withdrawn from each supernatant. One sample was immediately fil-
tered through a membrane filter (0.45 y), after which the phosphorus re-
maining in both samples was determined using the vanadomolybdophosphorlc
acid method (18).
25
-------
Destabilization Tests
Conventional jar tests were used. Polymers of several types and
ranging from 1 pg/1 to 100 mg/1 were added to calcium phosphate suspen-
sions. Polymers were added either with the other reagents (NaHCO ,
KH2PO^'2H20, etc.) or two minutes after these reagents had been mixed
and rapidly stirred. Two minutes of rapid stirring were provided
after polymer addition, followed by 20 minutes of slow stirring (floc-
culation) and 30 minutes of settling were provided. Residual phosphate
with and without membrane filtration was determined as described above.
In some cases electrophoretic mobility measurements were made on
the calcium phosphate suspensions, with and without added polymer.
A glass Briggs cell was used and the procedures recommended by Black
and Smith (31) were followed.
Filtration Experiments
The filtration apparatus used for filtering calcium phosphate sus-
pensions was adapted from that used for filtering latex suspensions
(Figure 1). Calcium phosphate suspensions were prepared and stored in
a 50 gallon polyethylene tank prior to use. Procedures used were
essentially those described previously for the precipitation tests,
save that continuous air agitation was used to provide flocculation and
to maintain the precipitates in suspension. Filter media were precoated
with polymer. The dosages of polymer to be used were selected on the
basis of jar tests. One filter received no polymer after the start of
a run, one received a continuous dosage of the optimum concentration
determined in the jar tests, two filters were underdosed, and two were
overdosed. Filter media were uniform sand with a size of 0.9 mm; bed
depth was 5.5 inches, bed porosity was 43 percent, and the filtration
rate was 2 gpm/sq. ft.
26
-------
SECTION V
RESULTS AND DISCUSSION
Filtration and Deatabilization Experiments Using
Latex Suspensions
Filters with Uncoated Media
Typical data obtained during experiments using uncoated filter
media are presented In Figures 3-A, B and C. Residual turbidity in jar
tests is plotted as a function of polymer dosage (Fig. 3-A); effluent
quality and head loss are plotted as functions of filtration time (Figs.
3-B and 3-C, respectively). In these experiments a suspension of .1 y
latex particles was prepared and the optimum dosage of PEI-18 for co-
agulating thi's suspension was determined to be .06 mg/1 from jar tests
(Fig. 3-A). The latex suspension was then introduced into six identi-
cal filter beds containing uncoated glass beads. The rate of flow con-
trollers for two filters ceased to function in the early stages of this
experiment, so that the results produced by these filters are not con-
sidered here.
Filter 1 received no polymer throughout the duration of the run;
filter 4 received the optimum polymer dose as determined by the jar
test (0.06 mg/1); filter 3 was underdosed (.018 mg/1), and filter 5 was
overdosed (0.093 mg/1) for the main portion (0 < t < 328 min.) of the
experiment. Near the end of the filter run (after 328 minutes of fil-
tration) , the polymer dosages were increased by 61 per cent, so that
filters 2, 4 and 5 then received 0.029, 0.096, and 0.145 mg/1 of poly-
mer respectively.
As may be noted from Figs. 3-B and 3-C, filter 1 which received
no polymer also did not remove any latex particles from the suspension.
Soon after the beginning of the filtration cycle the latex concentra-
tion in the filter effluent approached the concentration of latex in
the influent. No increase in head loss across the filter bed was
27
-------
QI20
CD
100
60
J340
ac.
I
I0~* 10"' I
DOSAGE OF POLYMER (mg/l)
dp= .1 microns
Co*9>7 mq/l
L = 5.5 in.
Polymer = PEI-18
POLYMER DOSE
FILTER NO. I* NONE
ii it 2=0.018 mg/l
•i u 4=0,060 •• - O
5»0.093 11 - n
O-OO-O-O-
120 180 240 300 360
FILTRATION TIME (min.)
* POLYMER DOSE WAS INCREASED BY 61% AT M 328 min.
Figure 3. (A,B,C)-Comparison of Jar Test Results
(A) with Filter Performance, Uncoated
Media (BondC).
28
-------
observed during this seven hour run.
Prior to changing the polymer dosage (t < 328 min.), the removal
efficiencies are filter 5 > A > 2 (Fig. 3-B). Correspondingly, the
rate of head loss increase is filter 5 > 4 > 2 (Fig. 3-C). The occur-
rence of a long ripening period during which the removal efficiencies
of filters 2, 4 and 5 continuously improved is also evident (Fig. 3-B).
This long ripening period would have obvious disadvantages in practice
and must be reduced or eliminated.
Increasing the polymer dosage during the run produced rapid chan-
ges in the performance of these filters. The rate of head loss in-
crease after changing the polymer dosage (t > 328 min.) is seen to be
filter 4 > 5 - 2 (Fig. 3-C). The removal efficiency of filter 5 de-
teriorates near the end of the run (Fig, 3-B).
These results indicate that the role of particle attachment can be
dominant in the filtration process. Unless provision is made for effi-
i
cient particle attachment to the filter grains or to the particles pre-
viously deposited there, a filter bed will be very inefficient in re-
moving impurities from the filter influent. It may be concluded that
the removal efficiency of a filter first increases and then decreases
as the applied polymer dose is increased from zero. This indicates
.that an optimum dosage of polymer for particle removal by filtration
does exist and that polymer underdosing and overdosing also occur. In
this experiment using uncoated media it appears that the optimum poly-
mer dose in filtration (ca. 0.096 mg/1) is greater than the optimum
polymer dose in coagulation (0.06 mg/1). The shape of the experimental
curves in Figure 3-B indicates that if the experiment had been conti-
nued for a longer period of time, a "ripening effect" might have con-
tinued to the ;point that filter 4 would have produced the best efflu-
ent. In other words, the optimum polymer dose for particle removal in
filtration might approach the optimum dosage observed for coagulation,
as filtration time .proceeds. Similar phenomena have been reported by
Tchobanoglous (15).
Filters with Coated Media
In order to reduce the long ripening period, a series of experiments
29
-------
was conducted in which the filter media were coated with polymer pri-
or to the beginning of the filtration cycle. Data obtained in a typi-
cal experiment are presented in Figures 4-A, B and C. All experimental
conditions were similar to those reported in Figure 3 except that a
different polymer (Cat-floe) was used and the filter media were pre-
coated.
When no polymer was added to the influent of a precoated filter
bed (filter 1) the effluent concentration reached a plateau soon after
the commencement of the filtration cycle (Fig. 4-B). The filtration
efficiency remained constant for a short period thereafter and then
rapidly deteriorated. No noticeable head loss increase was observed.
Similar results have been reported by Yao (3, 5). Filters which re-
ceived underdosed influent (filters 2 and 3) showed the same pattern
of performance.
The. latex particles possess a negative charge in wat/er. A resi-
dual negative charge persists in an underdosed suspension (6). The
attachment of these negatively charged particles to the positively-
charged precoated filter grains is manifested by a rather efficient
filtration for a very short period at the beginning of the filtration
cycle.. Soon the precoat capacity for a "monolayer" of suspended parti-
cles is exhausted and the removal efficiency deteriorates.
At the optimum polymer dose as determined by the jar tests
C.07 tug/I, Fig. 4-A) the effluent concentration Cfilter 4, Figure 4-B)
reaches a maximum soon after the.' initiation of the filtration cycle
and then quickly drops to a low value. Filtration is seen to be so
effective that the 7.5 feet of available head are utilized in less
than three hours (Fig. 4-C). It is significant to note that this oc-
curs even with, particles with, a size in the order of . ly» approximately
1/20QO of the pore openings in the filter bed.
The. effluent concentrations of filters 5 and 6 rapidly reached
very high values and their removal efficiency remained poor (Fig. 4-B).
The increase in head loss across filters 5 and 6 was also small. These
data clearly indicate that both underdosing and overdosing occur, that
30
-------
^ ^
fi?
9.7mg/l
dp » O.I microns
POLYMER "CATFLOC
I 10
DOSAGE OF POLYMER (mg/l)
FILTER SYMBOL POLYMER
NO. DOSE (mg/l)
2 A & 0.0014
3 • • 0.007
4 O O 0.07
5 O D 0.7
4 6 A A 7 B
1 1
1,2 and 3
POLYMER » CATFLOC
d » 0.397 mm
L* 5.5 in.
v*2gpm/tq.ft.
f »0.36
T «25°
C
50
100 ISO 200
FILTRATION TIME (min.)
250
300
Figure 4. (A,B,C) Comporison of Jar Test Results (A)
with Filter Performance (BandC), Precooted
Media.
31
-------
a polymer dosage which provides maximum particle removal by filtration
exists, that this polymer dosage when precoated filter beds are used
is equal to the optimum dosage determined in jar tests, and that Cat-
floc, an efficient coagulant for latex suspensions, is an efficient
filter-aid for the same suspension. Many other experiments conducted
using various concentrations and sizes of latex particles confirmed
the validity of these conclusions. The results of these experiments
can be found elsewhere (6).
Jar Test Anomalies
While conducting a trial filtration run of a suspension having a
low turbidity, it was observed that a filter which had received an op-
timum polymer dosage determined by the jar test did not produce clear
effluent. The sample volume used in each beaker in the jar test for
this experiment was 40 ml. The jars were 100 ml beakers which had
been coated with Siliclad, the suspension had a concentration of 7.55
mg/1 of .ly latex, and the optimum polymer dosage for coagulation was
found to be about .07 mg/1 of PEI-18. This was somewhat higher than
the expected value. Another jar test was conducted with the same sus-
pension, but using a sample volume of 900 ml in 1000 ml beakers. The
optimum polymer dose was found to be 0.025 mg/1 of PEI-18, considerably
lower than with the smaller beakers. The results of these jar tests
are presented in Fig. 5-A. Similar results were obtained when suspen-
i
sions containing small concentrations of 1.099 y latex particles were
used (6).
The optimum polymer dosage necessary for the maximum particle re-
moval when filtering the above suspension of .1 y particles was deter-
mined by a pilot filter technique. This technique uses the response
in filter performance to the changes in polymer dosages. This is very
rapid and had a time lag of only about 5 minutes for the apparatus used
in these experiments. The results of these experiments are shown on
Fig. 5-B. It can be seen that the optimum polymer dosage necessary for
filtration is equal to the optimum dosage found from the jar tests
(Fig. 5-A) when the larger sample volume was used. Thus, the equality
32
-------
180
160
2! 140
>-
5 120
CO
§ 100
< 80
(O
ID
OC
40
20
I I
Suspensions7.55mg/l, .lu Latex A
Polymer »PEI-18 r
SYMBOLS (Figure A only)
O Volume of Suspension in Each Jar * 40 ml
A Volume of Suspension in Each Jar • 900 ml
id"3 id"2 10"'
DOSAGE OF POLYMER (mg/l)
o
z
L* 5.5 in.
Polymer Dose mg/l
B
.025 .012 .025 .068
60 120 180
FILTRATION TIME (Min.)
240
Figure 5. (A,B) Comparison of Jar Tests (A) and
Pilot Filter (B) for Determining the
Optimum Polymer Dose for Filtration.
33
-------
In optimum dosages for coagulation and filtration can be observed
if the jar tests are performed "correctly11.
Some other investigators have also failed to observe an equality
in the optimum polymer dosage for coagulation and filtration. For ex-
ample, Mints (24) has written, "It should be pointed out that the mini-
mum dose which ensures the required degree of clarification in filtra-
tion and the minimum dose for the formation of large quick falling floes
are not identical in most cases. The former is usually smaller than
the latter and for the treatment of slightly turbid water may indeed
be several times smaller". The discrepancy between Mints1 observation
and the results of this study can be due to the adsorption of polymer
onto the surfaces of glassware utilized in conducting jar tests. The
surfaces of glass (silica) beakers used in this research for the jar
tests even when coated with chemicals such as Slliclad can compete with
the colloidal particles in adsorbing the polymer added to the jar for
particle destabilization. Only a fraction of the total amount of poly-
mer added to a jar will be available to react with colloidal particles.
The magnitude of this "available" fraction of polymer depends on the
ratio of the surface area of the. beaker in contact with the suspension
to the surface area of the suspended particles in each beaker. The
higher this ratio, the lower the fraction of polymer available for re-
action with particles. This implies that the optimum polymer dosage
determined by jar tests can be higher than the actual dosage necessary
for optimum coagulation. This effect would be more pronounced for
suspensions of low turbidity, where the total area of colloidal parti-
cles in the sample used for the jar test may have the same order of
magnitude as the beaker area in contact with the suspension. The ef-
fect would also be pronounced if the polymer is adsorbed to a greater
extent on the beaker surfaces than on the colloidal particles,
The MechanisgL^of Polymer Action in Filtration
To determine whether charge neutralization or (and) bridging Is
the dominant mechanism by which polyelectrolytes affect filter per-
formance, the charges of the-latex particles and the various polymers
34
-------
used in this research were measured. The charges due to the emulsi-
fier anion, the initiator, and the total charge (the sum of the emul-
sifier and initiator charges) for both .109 and 1.099 y particles are
presented in Table 3. The charges of various polymers at pH = 8.2 as
determined by acid-base titrations are presented in Table 4. In the
same table the optimum polymer dosages necessary for the coagulation
and filtration of a 48 mg/1 suspension of 1.099 y latex particles are
also shown. The optimum dosages of all polymers except PEI-6 are about
equal. The charge per unit weight of polymers except PEI-6 are also
almost equal. The optimum dosage for coagulation using a series of
polyethylenimines has also been reported by Dixon et al. (25) to be
independent of molecular weight above a certain minimum molecular
weight. This analysis indicates that charge interactions play a sig-
nificant role in the coagulation of latex particles by polyethyleni-
mine.
The magnitude of the positive charge of the optimum dosages of
PEI-18 and the negative charge of the latex particles for some of the
experiments conducted in this research are compared in Table 3. Even
if it is assumed that all of the polymer added to the system is ad-
sorbed on the latex particles, the data indicate that optimum coagu-
lation occurs and that particle restabilization commences before the
negative charges associated with the latex particles can be completely
neutralized by adsorbed cationic polymers.
To provide additional evidence, the electrophoretic mobilities of
the latex particles were determined at several polymer dosages. The
results are presented in Figs. 6 and 7. Here the residual turbidity
of the supernatent in the jars and the mobility of the latex particles
are plotted as a function of the polymer dosage added to the latex
suspension. The data indicate that optimum particle destabilization
occurs at negative mobility, confirming that complete charge neutrali-
zation is not required for effective coagulation. Furthermore, the
data indicate that particle restabilization occurs before the charge
on the latex particles has been reversed. This phenomenon has also been
35
-------
TABLE 3
LATEX CHARGE AND THE OPTIMUM POLYMER DOSE
Particle
Diameter
(Microns)
0.109
1.099
Emulsifier
Charge
ry
ycoul/cm
7.96
46.4
coul/gm
4.16
2.40
Residual
Charge
9
ycoul/cm
1.53
3.24
coul/gm2
0.802
0.168
Total
Charge
o
ycoul/cm
9.49
49.64
coul/gm
4.96
2.57
Ratio of
Polymer Charge
to Latex Charge
at Optimum Dose
0.5
0.2
CO
ON
-------
OJ
TABLE 4
CHARGE OF POLYETHYLENIMINE AND OPTIMUM DOSE AT pH = 8.2
Polymer
PEI-6
PEI-12
PEI-18
PEI-600
PEI-1000
Molecular
Weight
600
1200
1800
40-60,000
50-100,000
Charge at pH = 8.2
(Coulomb/gm)
9.3 x 102
7.9 x 102
7.6 x 102
7.4 x 102
Optimum Dose (mg/5,)
to Coagulate a Latex
Suspension
(l.lu, 48 n«/A)
0.050
0.029
0.032
0.024
0.025
-------
*> -0-5
-1.0
II
i~
-1.3
-2.0
-2.5
-3.0
0.20
DOSAGE OF POLYMER (mg/l)
0.30
Figure 6. Effects of PEI-18 on Residuol
Turbidity (A) and Electrophoretic
Mobility (B) of 0.109 p. Latex
Suspension.
38
-------
120
100
80
a
60
40
o
13 20
0.05
-8.0
0.00 0.05
DOSAGE OF POLYMER (mg /1}
Figure 7. Effects of PEI-18 on Residue!
Turbidity (A) and Electrophoretic
Mobility (B) of 1.099/1 Latex
Suspension.
39
-------
observed by many others in several coagulant-colloid systems (e.g., 26)•
These findings indicate that charge neutralization is not the sole
mechanism by which polyethylenimine destabilizes latex particles, since
optimum coagulation did not occur at zero mobility. However, the close
correlation between the optimum polymer dosages and the charge per unit
weight of polymers (Table 4) suggests that the charge interactions are
significant. Theae results are consistent with the following postulate.
Segments of polymer molecules are adsorbed onto the surfaces of latex
particles by electrostatic and chemical interactions between the posi-
tive sites of the polymer and the negative sites of the latex, while
the remainder of the polymer molecules extend into the solution. These
extended positively charged segments will be adsorbed onto the free ne-
gative sites of other particles when the particles are brought in close
proximity. The polymer molecules provide bridges which enhance the
attachment of the colloidal particles to each other.
It is generally accepted that the polymer molecule should have
a rather large size (greater than, the thickness of the diffuse layer)
to act as a bridge in bringing about particle destabilizatlon. Howev-
er, it should be noted that the adsorption of the cationic polymers on
negative colloids reduces the effective, charge of the colloidal parti-
cles and therefore also reduces the thickness of the double layer sur-
rounding such, particles. It may also alter the van der Waals attractive
forces. It is plausible that bridges can be formed even by low mole-
cular weight polymers in such systems. Alternatively, steric effects
can produce restabilization.
These conclusions are in accord with the findings of Dixon et al.
(25) who studied the coagulation of silica suspensions by polyethyl-
enimines., and Black et al. C27), who investigated the coagulation of
clay suspensions with. Cat-floe.
The high dosage of PEI-6 (Table 3) may be explained by considering
that the low molecular weight polymer may not be completely adsorbed
on. the latex particles. Thus, the amount of PEI-6 which, must be added
to achieve a certain coverage of the latex surface can be greater than
the corresponding value for larger polymers which, are more completely
40
-------
adsorbed. This is also in accordance with the findings of Dixon et al.
(25).
Stoichiometry in Filtration
The results of experiments conducted to test for Stoichiometry are
presented in Fig. 8. The optimum dosage of the polymer (mg/1 PEI-18)
is plotted as a function of the concentration of suspended particles
(mg/1). A linear relationship for both 0.1 and 1.1 y particles is ob-
served. Again, a similarity between filtration and coagulation is ob-
served .
The data presented in Fig. 8 indicate that the optimum polymer do-
sages required for coagulating suspensions of different particle size
but with equal concentration of particle surface are not equal. For
the same weight concentration, a suspension of 0.1 P latex spheres will
have a surface concentration of (1.1/0.1) or 11 times that of a suspen-
sion of 1.1 y latex spheres. However, a suspension of .1 y latex par-
ticles would need only about one-half of the polymer which is necessary
for optimum coagulation of a suspension of 1.1 V latex particles at
an identical particle surface concentration. This result is in disagree-
ment with the findings of Black and Vilaret (28) who studied the coagu-
lation of latex suspensions with Cat-floe. These investigators obser-
ved a linear relationship between the optimum polymer dosage and latex
particle surface concentration, regardless of particle size. This lat-
ter finding seems questionable in view of the fact that the total charge
per unit area of the latex particles increases markedly as particle size
increases (Table 3). The same trend in the residual charge density of
the latex particles has been demonstrated by Vanderhoff et al. (16).
The discrepancy may be due to inaccuracies in determining the optimum
polymer dosages in jar tests. Alternatively, Cat-floe may interact
with and adsorb on the surface of the latex particles in a manner which
is not primarily dependent upon electrostatic effects.
Molecular Weight of the Polymer and Filter Performance
The results of an experiment designed to evaluate the influence
of molecular weight on the effectiveness of polymers as filter-aids are
41
-------
.20
ro
0»
E
tr
5
o
GL
.15
.10
LJ
O
-05
Q.
O
a dp=0.lu
o dp= l.lu
10 20 30 4O
CONCENTRATION OF LATEX SUSPENSION (mg/l)
Figure 8. Stoichiometry in Filtration: Optimum Polymer
Dosage (PEI-18) as a Function of Colloid
Concentration.
50
-------
Presented in Figure 9. A suspension of 1.1 y latex particles at a
concentration of 48 mg/1 was used as the filter influent. The opti-
mum dosages of five homologs of polyethylenimine were determined by
the pilot filtration technique (Table 4). Five shallow precoated fil-
ter beds (1.4 inch deep) were used, each of which received the latex
suspension and one of the polyethylenimine homologs at the correspond-
ing optimum dosage.
The removal efficiencies slightly increase as the molecu-
lar weight of the applied polymers increase, except for filter 3 (Fig.
9). A second experiment using only filter 3 produced results which
are consistent with this observed slight increase in removal efficiency
with increasing molecular weight. The results presented for filter 3
in Figure 9 were probably due to poor experimental technique, such as
improper preparation of the stock polymer solution. The slight im-
provement in removal efficiency as the molecular weights of polymers
increases is consistent with, the head loss data and with visual obser-
vation of the filter beds. The rates of head loss development across
the filters increase as the molecular weights of the applied polymers
increase except for filter 3 (Fig. 9). The depth, of significant par-
ticle penetration into the filter bed was observed to decrease with in-
creasing polymer molecular weight. Particles were removed within a
very thin top layer to the filter bed when PEI-1000 (molecular weight -
50-100,000.) was used. At the other extreme, when PEI-6 (molecular
weight « 600) was used, the particles penetrated into deeper layers.
particle penetrations, into filter beds when using PEI-18 and PEI-6QO
have intermediate molecular weights were observed to be between
these two extremes. Based on these results it is concluded that when
polymers, of this type, are applied at the optimum dosage in filtration,
the efficiency of particle attachment increases as the molecular weight
o£ the applied homolog of polymer is increased.
.Suspensions of Low Particle Surface Concentration
The coagulation of suspensions containing low concentrations of
suspended particles generally have been found to be quite difficult.
Experimental results presented in Figure 10-A illustrate this problem.
43
-------
40 r
dp* Up
c0= 48mg/l
L= 1.4 In.
0
80
60
3 40
LJ
X
20
-Poly-
mer
Do**'
(rog/l)
Filter No.2 =.050 PEI-6
,. .. 3« ,029 PEI-12
» ii 4* .032 PEI-18
n n Ss.025 PEI-600
n « 6s.025 PEI-IOOO
6
60 120
FILTRATION TIME (Min.)
Figure 9. Effects of Molecular Weight
of Polymer on Filter Performance
at the Optimum Polymer Dosage.
44
-------
120
IOO
m
oc 80
(O
ui
60
40
_L
SUSPENSION* 14.4
POLYMER: PEI-18
dp =7,6/1
I I
45min
STIRRING
6hr. STIRRING
_L
J_
I0~2 10"' 1 10
DOSAGE OF POLYMER (mg/l)
3
dp*7.6/i,
C0sl4.7mg/t
L« 5.5 in.
o
2
X
POLYMER DOSE (PEI-18) SYMBOL
h FILTER WO. I'NoM 9
- « 2 = 0.0014 mg/l A
N » 3*0.0068 H •
» « 4*O068 «
» « 5= 0.545 -
.. 6= 5.450 »
B
60 120 180 240 300 360
FILTRATION TIME (min.)
420 480 540
Figure 10. (A,B,C) - Effects of Polymer on Coogulotion (A) and
Filtration (B,C) of o Low Turbidity Suspension.
-------
Here a suspension of 7.6 y latex particles at a concentration of 14.7
mg/1 was treated with PEI-18. The coagulation of this water was quite
slow. The suspension treated with the proper amount of the polymer re-
quired six hours of stirring before any significant coagulation could
be observed. This problem arises from the low rate of interparticle
contacts in this system; in other words, the process is limited by
mass transport.
Data presented in Figures 10-B and 10-C, however, indicate clear-
ly that such a suspension can be easily and effectively filtered. The
data are consistent with the prediction of Yao's transport model for
water filtration (3, 5) in which the transport efficiency is considered
independent of the particle concentration. The efficiency of coagula-
tion, however, depends markedly on particle concentration. Therefore,
it is concluded that the direct filtration of waters containing low con-
centrations of particles, in contrast to their coagulation, is quite
effective. This concept already has been put to practical use in the
field by others (29).
It is useful to note that the filter which did not receive any
polymer dosage except the precoat (filter 1) is as efficient as filter
4 which received the optimum polymer dosage, while the head loss across
filter 1 was much smaller than that of filter 4 (Fig. 10-B, C). This
indicates that when suspensions of low turbidity are filtered through a
precoated filter bed, it may be profitable not to add any polymer to the
filter influent until the capacity of the precoat is almost exhausted.
This procedure would reduce the rate of head loss development and thus
provide a longer filter run.
Suspended Particle Size and Filter Performance
Yao's theoretical and experimental studies showed that the size
of the suspended particles affects the filter removal efficiency mark-
edly (3, 4). To evaluate the validity of this prediction when polymeric
filter-aids are continually added to the filter influent an experiment
was conducted in which three suspensions of approximately equal concen-
tration of .109, 1.099 and 7.6 y latex particles were filtered through
three shallow filter beds (depth =0.8 inch). The results of these
46
-------
filter runs are presented in Fig. 11. It is seen that the trend in re-
moval efficiency as a function of particle size is in qualitative agree-
ment with the prediction of Yao's transport study. The removal effici-
ency for filtering suspensions of .109 }J latex particles is rather high
and remains high until the end of the experiment when all of the 85
inches of available head are exhausted. Filtration of the suspension
of 1.099 p latex particles is the least efficient and the breakthrough
is observed at a head loss of 5 inches. Furthermore, filtration of 7.6 y
latex suspensions produces the best removal efficiency. Similar trends
of filtration efficiency as a function of particle size have been ob-
served in aerosol filtration (8, 30).
Tht head loss data (Fig. 11) show a remarkable dependence upon the
size of the suspended particles. Analysis of the results of other ex-
periments conducted in this research indicates a similar dependence of
head loss on suspended particle size (6). Such analysis indicates that
for any given amount of cumulative particle removal, the head losses
across filters receiving suspensions of .1 y latex particles at the op-
timum polymer dosages are much greater than the corresponding head loss-
es when suspensions of 1.099 and 7.6 u latex particles are filtered.
A plausible explanation for the dependence of head loss across a
filter bed upon the size of the suspended particles is that suspended
particles treated with the optimum polymer dosage have a tendency to be
removed within the very top layer of the filter bed. This can cause the
formation of a surface mat which is responsible for rapid head loss de-
velopment. Under such circumstances the particle deposit (or the sur-
face mat) acts as the filtering medium except for a very short period
at the beginning of the filter run. A filter mat which is formed from
suspended particles would show a resistance to flow which is an inverse
function of the size of the suspended particles. The formation of a
surface mat and the consequent rapid rate of head loss buildup is one of
the main drawbacks of polymer-aided filtration. The prevention of sur-
face mat formation has practical significance.
The formation of mats when filtering suspensions of .1 y latex
particles could not be prevented even at flow rates as high as 10 gpm/sq.
47
-------
O dp=.l/x, C0=50mg/l
L=.8in.
POLYMER =.176 mg/l PEI-18 -
, C0=48mg/l
L= .8 in.
POLYMER = .032 mg/l PEI-18
dp= 7.6/x, C0 =52mg/l
L=.8in
POLYMER = .024 mg/l PEI-18
60 120
FILTRATION TIME (Minutes)
Figure II. Effects of Suspended Particle
Size on Filter Performance ot
the Optimum Polymer Dosage.
180
48
-------
ft. The rate of head loss development was observed to be rapid and ex-
ponential while the removal efficiency remained quite high. In contrast,
the use of larger filter grains was found to be effective in reducing
°r even eliminating the mat formation. Although this retardation
°f head loss build-up was associated with deterioration in removal ef-
ficiency, it seems plausible that efficient particle removal without
excessive head-loss development can be obtained by use of deep filter
beds of large grains even when suspensions of sub-micron particles are
to be filtered. It is also plausible that a short period of floccula-
tion to increase the size of sub-micron particles before reaching the
filter can reduce the rate of head loss development.
Filtration of Trickling Filter Effluent
laboratory Studies
Conventional wastewater treatment plants generally remove from 55
to 95% of the BOD and suspended matter from the raw wastewater. A major
fraction of the remaining impurities is in colloidal form. Therefore,
the knowledge obtained in this study regarding the filtration of colloidal
suspensions should be, at least in principle, applicable to the further
treatment of conventionally treated wastewater by filtration using poly-
meric filter aids.
Preliminary filter runs using trickling filter effluent without
chemical pretreatment produced very low removals of turbidity using 0.8
""a glass beads in 5.5 inch deep filter beds. Results of the filtration
°f the effluent from the Mason Farm Wastewater Treatment Plant, Chapel
Sill, N.C. using Prima-floc C-7 as a filter aid are presented in Fig.
12. In this experiment four filter beds (1, 3, 4 and 5) containing sand
with a geometric mean diameter of 1.3 mm were precoated with a concentra-
ted polymer solution; the depth of each filter bed was 5.5 inches. Filter
1 did not receive any additional polymer and served as a blank. Filter
4 received waste treated with the optimum polymer dosage determined from
tests (14.2 mg/1). Filters 3 and 5 were underdosed (2.87 mg/1) and
-------
140
II"
£w
80
UJ
cc
40
o 160
z *""*
z£ 140
35 60
u5
2^ 40
Is 20
o
0 10 20 30 40 50
FILTRATION TIME (min-1
10 100 1,000
DOSAGE OF POLYMER (mg/l)
80
760
340
O
SUSPENSION :
SECONDARY EFFLUENT
POLYMER:
PRIMAFLOC C-7
0 10 20 30 40 50
FILTRATION TIME (min.)
20 40 60
FILTRATION TIME (min.)
Figure 12. (A,B,C,D)- Effects of Polymer on Coogulotion (A)
ond Filtrotion (B,C, D) of o Secondory Effluent
from Mason Farm Wastewater Treatment Plant.
50
-------
overdosed (71.2 mg/1), respectively. The similarity between the results
of jar tests and filtration can be seen again from these results. Filter
1 which did not receive any polymer except the precoat showed a very
low efficiency (about 20%) in terms of the removal of turbidity and to-
tal organic carbon. No significant head loss development across
this filter was observed. The turbidity reduction achieved by filter
3 (the underdosed filter) and the corresponding head loss were very close
to those of filter 1. The effluent of filter 3 does contain more organic
carbon than the untreated wastewater influent. This may be due to or-
ganics in the polymer solution. A greater increase in TOC in the efflu-
ent of the overdosed filter (no. 5) was observed. The turbidity reduc-
tion achieved by this filter was also poor.
Filter 4 which received the optimum polymer dosage as determined
from jar test data showed the best removal efficiency (about 60%) both
in terms of turbidity and total organic carbon. It also had the highest
rate of head loss increase.
Economic considerations can be a drawback for wastewater filtration.
As can be inferred from Fig. 12, the optimum polymer dosage required
for coagulation or filtration of the conventionally treated wastewater
is rather high (14.2 mg/1 in this case). At the present level of poly-
mer cost (about $1 per pound) the application of Primafloc C-7 at a do-
sage of 15 mg/1 would result in chemical costs of about 12 cents per
thousand gallons of waste to be treated and so is not feasible at pre-
sent. However, the number of polymers commercially available is numerous
and it may be possible to find or develop a polymer which would be more
acceptable.
The rapid increase in head loss is another drawback for polymer-
aided filtration of conventionally-treated wastewater. It might seem
that the removal of significant amounts of large particles could be re-
sponsible for this rapid head loss development by filling and blocking
the pores in the bed. A comparison of results obtained by conducting
jar tests on wastewater treatment plant effluent (Fig. 12-A) and on a
suspension of 0.1 y latex particles (Fig. 3-A) provided another plausible
51
-------
explanation for, and perhaps a key to, the solution of this head loss
problem.
As can be seen from Figs. 12-A and 3-A, in both cases an increase
in residual absorbance is observed at polymer dosages surrounding the op£
mum polymer dosage. This phenomenon is due to the growth in the size of
sub-micron particles (6). This comparison suggests that the majority of
the colloidal impurities in trickling filter effluents are sub-micron
in size. In this connection it is useful to note that the head loss
development when suspensions of 0.1 micron latex particles were filtered
was very rapid CFigure 3-C, 4-C). The similar rapid head loss build-up
which was observed when the conventionally-treated wastewater was fil-
tered can thus be attributed to the existence of sub-micron colloidal
particles in wastewater treatment plant effluent.
gilot Plant Studies
These experiments can be subdivided into three areas for discussion:
(i) destabilization studies, (ii) filtration studies dealing with pre-
treatment, and (iii) filtration studies dealing with filter characteris-
tics. Experimental results are presented and discussed in this perspec-
tive.
Destabilizatiem. Studies
Several anionic polymers were tested as primary coagulants using
jar tests. The effects, of calcium ions, pH, and flocculation period on
the effectiveness of these polymers were also evaluated. Regardless of
the combination of anionic polymer type and dosage, [Ca^+], pH, and
flocculation period, effective coagulation was not observed.
Alum was investigated as a primary coagulant for trickling filter
effluent. Reaulta of a typical jar test are presented in Figure 13.
Residual turbidity is plotted as a function of alum dose. Data obtained
in another jar test for the removal of 5-day BOD, TOG, SS, and TP are
presented in Figure 14. An optimum alum dose is observed (Figs. 13 and
14). In these and most other tests this optimum dose was found to be
about 150 to 160 mg/1 tAl2(SOA)3-18H20). Floe quality Cslze and settling
characteristics) was especially poor in the underdosed and overdosed
52
-------
50
100
150 200
ALUM DOSE (ppm)
250
300
350
Figure 13. Effects of Alum on Coagulation of Secondary Effluent. Residual
Turbidity vs. Alum Dose.
-------
100 150 200
ALUM DOSE (ppm)
250
300
Figure 14.
Effects of Alum on Coagulation of Secondary
Effluent. Residual BOD5, TOC, SS and TP
vs. Alum Dose.
54
-------
regions. Even at the optimum alum dose the floe did not settle well.
Attempts to lower the optimum alum dose and to improve floe character-
istics by changing the pH were unsuccessful.
Preliminary filter runs were made using the optimum alum dose,
The floes which were produced penetrated the filters very easily, and
filter runs were terminated in an hour or less because of complete
breakthrough. To improve the ability of the pilot filters to remove the
alum-treated secondary effluent, it was decided to investigate the use
of polymers as secondary coagulants. Again, jar tests were made to de-
termine the types and dosages of polymers which had potential for fil-
tration.
Results of jar tests using Purifloc A-23 as a secondary coagulant
are presented in Figure 15. Residual turbidity is plotted as a function
of polymer dose (mg/1, log scale). In these tests all jars received the
optimum alum dose (in this case 150 mg/1). After two minutes of rapid
stirring, the desired polymer dosage was added, after which the conven-
tional jar test procedure was followed. Some slight improvements in
residual turbidity were obtained. At a polymer dosage of 1 mg/1, tur-
bidity was 1 jtu, SS were 1 mg/1, and TP was 0.4 mg/1. Most dramatic
changes occurred in the appearance of the floes. For polymer dosages
between 0.5 and 20 mg/1, large and rapidly settling floes were formed.
Typical of polymer coagulation, regions of underdosing and overdosing
were observed (Fig. 15). Based on these results, Purifloc A-23 was se-
lected for further study in filtration tests.
Effects of Pretreatment on Filtration
Results of experiments using alum plus polymer are presented in Fi-
gures 16 to 19. Each figure contains results using a different media
size. Residual turbidity (log scale to delineate differences) and head
loss are plotted as functions of filtration time. Residual turbidity is
expressed as the ratio of the filter effluent turbidity to the turbidity
of the secondary effluent (per cent). Alum dose is constant (150 mg/1);
polymer dosage was varied from filter to filter within a given run. Media
size was varied from run to run; four sizes ranging from 1.65 mm to 4.08
-------
8
3
56
b
Q
CD
DC
Alum = ISOppm
/ -I
I I I
CJ
O
in
O
« (O f^ 00 0>O
6 d o 6 d_;
CM
10
o
(M
8 § 2
DOSAGE OF PURIFLOC A-23 (ppm)
Figure 15. Effects of Purifloc A-23 on Coogulotion of on Alum-
treoted Secondory Effluent. Residual Turbidity vs.
Polymer Dose.
-------
Ul
A -
0.2 ppm
d= 1.65 mm
A-23do»*
Alum = 150 ppm
0.5 ppm
1234
FILTRATION TIME (Hr».)
12345
FILTRATION TIME (Hrs)
Figure 16. Effects of Purifloc A-23 on Filter Performonce.
-------
00
A -
Incrtas* polymer
dos« by 100%
1.0 ppm
ppm - A-23 do««
Alum «ISO ppm
2 3
FILTRATION TIME (Hrt.)
100
80
60
111
x 40
20
1 1
B
-f I.Opprn
(234
FILTRATION TIME (Mrs.)
Figure 17. Effects of Purl floe A-23 on Filter Performance.
-------
100
Ui
vo
IU
9
_ 8
^ 7
5* i
t 5
Q
54:
QC
?3
_J
<
D 0
0 ^
-------
100
d = 4.08 mm
ppm = A-23 dose
Alum = ISO ppm
2 3
FILTRATION TIME (Hrs.)
01234
FILTRATION TIME (Mrs.)
Figure 19. Effects of Purifloc A-23 on Filter Performance
-------
nan were used.
Polymer applications ranged from underdosing up to the lower end
of the optimum region for coagulation (Figure 15). Removal is observed
to be dependent upon polymer dose while effects of media size on removal
are slight (Figs. 16 to 19). In all cases a ripening period exists;
filter efficiency improves at the start of a run, reaches a constant
level and then, if sufficient time is provided, it deteriorates. The
initial ripening period is probably due to less than complete attach-
ment of the suspended particles to the media grains. After the initial
deposit is accumulated, the attachment of newly applied particles to
similar ones previously retained In the bed is more effective.
A simple experiment emphasizing the importance of the chemistry
of the system is shown In Figure 17. One filter receiving a continuous
polymer dosage of 0.5 mg/1 broke through after 2.5 hours of operation.
At that time 55 per cent of the turbidity of the secondary effluent
was passing the filters. The applied polymer dose was then in-
creased to 1.0 mg/1, in the optimum region of the jar tests. Within an
hour the effluent turbidity fell to 14 per cent, and in another half
hour was reduced to 4 per cent. At this time the available head loss
was exhausted.
For these filter beds, the use of effective destabilizatlon caused
rapid and excessive head loss. The rate of head loss increase depended
upon the polymer dosage. Unlike removal efficiency, however, the rate
of head loss is also very dependent upon media size. Larger media pro-
duce, lower rates of head loss increase while having only minor effects on
effluent quality. To investigate further the trade-offs between pre-
treatment and filter characteristics, a series of experiments using se-
veral media sizes was conducted.
Effects of Filter Characteristics
The results of experiments In which, media size was varied are sum-
marized in Figure 2.0. Head loss is plotted as a function of filtration
time for seven sizes of filter media. The optimum alum dose for runs
using 1.65, 2.36, 3.08, and 4,08 mm media was 150 mg/1. The optimum
alum dose for a second run with 4.08 mm media and also for filters using
61
-------
to
2.36mm
I
A-23 s 1.0 ppm
Alum = 150 ppm (
Alum - 75-125 ppm (
mm - Medio size
4.0 5.0 6.0 7.0 SO
FILTRATION TIME (Mrs.)
Figure 20. Effects of Media Size on Filter Performance. Head Loss vs. Filtration Time.
-------
5.9, 6.3, and 7.5 mm media varied from 75 to 125 mg/1 during the course
of the run. For all experiments a polymer dosage of 1.0 mg/1 Purifloc
A-23 was used. As expected, the head loss increase during filtration
decreases with increasing media size. Differences between the two runs
using 4.08 nan are ascribed to the different alum dosages required. These
head loss data will be discussed in a different perspective subsequently.
Here one can note that filters using small media (1.65 mm is this case)
will have their runs terminated due to utilization of available head,
while filters using larger media (7.5 mm in this case) may have their
runs terminated because of poor effluent quality.
The residual turbidity (jtu, log scale) observed in these experi-
ments is plotted as a function of filtration time in Figure 21. In this
case removal is only moderately dependent upon media size. In all cases
a ripening period occurs. The upset after 5 hours for the 7.5 mm bed is
probably due to an increase in the alum dose from 75 to 125 mg/1 at that
time. An increase in the polymer dose after 10 hours provided immediate
improvement, probably because the 1.0 mg/1 dose of Purifloc A-23 is at
the low end of the optimum region (Fig. 15). These data are considered
again in a different perspective in the following discussion.
Procedures for Filter Design
Filter design involves consideration of the quantity and quality
of water produced, and of the costs of this production. Conventionally,
rather standardized filters are specified (e.g., filtration rate -
2 gpm/sq ft, bed depth = 24 inches, media size = 0.6 mm, available head
loss » 8 feet). The operator is then required to produce a desired
quantity and quality of finished water by adjusting the pretreatment of
the water prior to filtration. Effective destabilization in pretreat-
ment can. reduce water production by shortening filter runs through rapid
head loss buildup, while poor destabilization can impair water quality
and hence reduce water production by the breakthrough of solids in the
filter effluent. In most cases an optimal (least sort) combination of
pretreatment and filter costs to produce a desired effluent quality is
not obtained, or even sought.
-------
60
50
40
30
T I 1
I I
I
I
r
I
i
i
\
I
T
1 I I I
Polymer incrtose
by 30%
E E I E E
o
o
d = 1.65 mm
d = 2.36 mm
d = 3.08 mm
d s 4.08 mm
d = 4.08mm
• d = 5.9mm
A d = 6.3 mm
A d = 75 mm
E= Secondery Effluent
I = Secondory Effluent + Alum.
_L
t
L
I
45678
FILTRATION TIME (Mrs.)
10
II
]
Figure 21.
Effects of Media Size on Filter Performance.
Residual Turbidity vs. Filtration Time.
64
-------
Now destabilization in pretreatment is difficult to vary gradually.
Coagulants tend to either work or not work. In contrast, particle trans-
port in filter beds can be varied continuously over a wide range by
changes in filtration rate and media size. Conventional filter design
specifies what is easy to vary (particle transport), and leaves unre-
solved what is difficult to control (particle destabilization). It is
useful here to consider a reversal of this procedure—one could specify
an effective destabilization procedure and then design a filter to pro-
duce a desired quantity and quality of water using this pretreatment. By
considering several pretreatment procedures (together with their asso-
ciated filter designs), the least cost combination of filter and pre-
treatment costs could be selected. To implement this approach, experi-
ments are needed. The data obtained in this research will be used to
illustrate this approach.
Length of filter run is plotted as a function of media size in
Figure 22. Pretreatment consisted of 150 mg/1 of alum followed by 1.0
mg/1 of Purifloc A-23. The filtration rate was 2 gpm/sq. ft. Filter
runs were terminated because the available head loss (80 inches) was ex-
ceeded, or because the effluent quality exceeded 10 jtu. An optimum
media size of about 5 mm exists for this pretreatment condition (Fig. 22).
For smaller media, the runs were terminated because of head loss buildup;
for larger media the runs were terminated because of poor effluent qua-
lity. The maximum length of run was 11 hours, corresponding to 1320
gals/sq. ft. of filter area.
Based on these results the following experimental procedure is sug-
gested for consideration in filter design:
1. Select a suitable pretreatment procedure using jar tests. Deter-
mine a desirable type and concentration of destabilizing chemical,
or a combination of chemicals*
2. Using this pretreatment procedure, evaluate the effects of media
size using pilot filters at a constant filtration rate. If de-
sired, consider dual and multimedia beds, biflow filtration, etc.
3. Repeat (2) using other filtration rates.
65
-------
12
ti
10
Run Term i no fed by
Utilization of Available
Head (80 in.)
H
H
Run Terminated by
Breakthrough ( Filter
Effluent Turbidity >
lOjtu)
Optimum Media
Size = 5 mm
Alum = 150 ppm
A-23 * 1.0 ppm
v = 2 gpm/ sq.ft.
4 6
MEDIA SIZE (mm)
8
10
Figure 22. Determination of Optimum Media Size
66
-------
4. Considering the results of (2) and (3), select the combination
of media size and filtration rate that provides maximum filtered
water production given constraints of available head and desired
effluent quality.
5. If desired, use the results of (2) and (3) to determine the ef-
fects of design head and effluent criteria on the costs of the
process.
6. Design suitable backwash facilities. For larger media this can
require air wash.
7. Repeat steps (1) through (6) for other suitable pretreatment
procedures.
8. Compare the costs of each combination of pretreatment, filter,
and backwashing system. Select the final design, considering
also such factors as variability in raw water quality.
This general approach requires experimentation for design, and is
more time-consuming and expensive for the design engineer. It adds ex-
perimentation at the pilot plant level prior to design, in order to re-
duce experimentation at the full-scale plant level after design and con-
struction. It shifts some of the responsibility for pretreatment from
the operator to the designer. It may also provide for more effective
operation by selecting a pretreatment process that will be effective in
destabilizatlon (recall that destabilization is hard to control) and pro-
viding a filter that is tailored to this pretreatment (recall that parti-
cle transport is more readily controlled).
Filtration of Calcium Phosphate Suspensions
Coagulation Studies
Jar tests were conducted to determine the feasibility of using poly-
mere as filter aids in the direct filtration of suspensions of calcium
phosphate precipitates. Results of experiments using a cationic polymer
(Cat-floe) are presented in Figure 23. Residual phosphorus (per cent) is
plotted as a function of polymer dosage (mg/1, log scale). Phosphorus
concentrations present in the supernatent after settling (3.11 mg P/l)
or in the filtrate after settling and membrane filtration (2.71 mg P/l)
6.7
-------
120
oo
INITIAL CONC
{mg/0
Inorganic C
POLYMER = CAT-FLOC
Settled, P-3.llmg/l at
Polymer Dosage = 0
Filtered, P=2.7I mg/t at Polymer Dosage-O
I I
1.0"' I 10
POLYMER DOSAGE (mg/l)
100
1000
23. Effecfs of Cat-Floe on the Coagulation of Calcium Phosphate
-------
when no polymer was added have been used to calculate the percentages
of residual phosphorus, rather than the initial phosphorus in the jars
at the start of the test (10.75 mg P/l).
The optimum polymer dosage, defined here as the dosage which pro-
duces the minimum residual phosphate, is about 1 mg/1. The minimum
concentrations of phosphorus are 1.59 mg P/l (after settling), and 1.19
mg P/l (after membrane filtration). Both underdosing and overdosing
occur. A rather broad optimum region is observed, suggesting that the
process could be easily controlled in a continuous-flow system.
Prima-floc C-7 (a cationic polymer) and Purifloc A-23 (an anionic
polymer) were tested and produced effective destabilization in jar tests,
with results similar to those presented in Figure 23 (32). Cat-floe was
selected for further evaluation in direct filtration tests using labora-
tory filters (Figure 1).
Filtration Tests
The results of a typical filtration experiment are presented in
Figure 24A and B. Residual phosphorus (per cent) is plotted as a function
of filtration time (hours) in Figure 24A. Filter 1 received no polymer
and hence was "underdosed"; filter 4 received a continuous dosage of 1 mg/1
of Cat-floe, the optimum dosage. Filters 5 and 6 received 10 and 100
mg/1 of Cat-floe and hence were overdosed (see Figure 23). All filter
beds were precoated with a Cat-floe solution. Removal efficiency is fil-
ter 4 > filter 5 > filter 6 > filter 1. Head loss (inches) is plotted
versus filtration time (hours) in Figure 24B. The rate of head loss in-
crease is filter 4 > filter 5 > filter 6 > filter 1.
Filter 4 received the optimum polymer dosage, produced the best ef-
fluent, and exhibited the greatest head loss. Over 90 per cent removal
of phosphorus was achieved after 3 hours without prior sedimentation.
Underdosed and overdosed filters were much less effective. It can be
concluded that an optimum polymer dosage for the filtration of calcium
phosphate suspensions does exist, and that polymer underdosing and over-
dosing also occur. While not evaluated experimentally, stoichiometry
is presumed to occur. The role of particle destabilization in the
69
-------
60
-40
to
iu 20
x
0
80
60
UJ
40
Q.
(/>
°20
T
T
POLYMER
SYMBOL FILTER DOSE (mg/l)
X I 0
A 4 I
• 5 10
• 6 100
INITIAL P = 10 mg/l
pH = 9
POLYMER = CAT - FLOC
B
/ 1
1 1 1
01234
FILTRATION TIME (Mrs.)
Figure 24. Effects of Cat-Floe on the Filtration
of Calcium Phosphate.
70
-------
filtration of these suspensions is demonstrated. The coagulation and
filtration of these materials are analogous. Jar tests may be used to
evaluate the effectiveness of polymers as filter aids.
Development of Filtration—A Theoretical Perspective
Consider first the efficiencies of particle aggregation in coagula-
tion and particle removal in filtration. Coagulation efficiency may be
related to the dimensionless product acGt, where ac is a collision
efficiency factor that reflects the chemistry of the system, is the vo-
lume of solid material (colloidal and suspended particles) per unit
volume of suspension, G is the mean velocity gradient which reflects the
rate at which contacts occur between particles by mass transport, and t
is the detention time in the flocculation tank (33). In a similar way
the efficiency of particle removal in filtration may be described by a
dimensionless product aF(l-f)n(L/d), where ctF is a collision efficiency
factor that reflects the chemistry of the system, (1-f) represents the
volume of filter media per unit volume of filter bed (f « porosity),
n is a "single collector efficiency" that reflects the rate at which par-
ticle contacts occur between suspended particles and the filter bed by
mass transport, and L/d, the ratio of bed depth to media diameter, repre-
sents the number of single collectors in the system and is somewhat
analogous to detention time in coagulation processes. The single collector
efficiency is an inverse function of media size, filtration rate, and
water temperature, and also depends upon the size and density of the
particles to be filtered (5).
The development of packed-bed filtration processes in water and waste-
water treatment is presented in Table 5 in terms of the dimensionless
product aFn(L/d). For the purposes of this development the differences
in bed porosity among the processes are small and have insignificant ef-
fects. Let us begin by considering Infiltration galleries, where surface
waters are filtered through porous strata prior to withdrawal by wells.
Similar phenomena occur in ground water recharge with treated wastewater.
No destabilization is attempted in these processes, and the suspended
71
-------
TABLE 5
S3
DEVELOPMENT OF FILTRATION PROCESSES
Filtration
Process
Infiltration
Gallery,
Ground Water
Recharge
Slow Sand
Filtration
Rapid Sand
Filtration
Ultra-high
Rate
Filtration
Destabili-
zation
None
Biopolymers
in Schmutz
decke
Alum
Alum
plus
Polymers
Stability L
Factor * d
(
-------
particles do not adhere effectively to the filter media. It is reason-
able to consider that most contacts do not result in attachment; aF is
in the order of 10~-^> meaning that only 1 particle out of every hundred
thousand that strike the filter media is actually removed. Effective
filtration is accomplished by providing many collectors (L/d = 10 ) and
by achieving efficient mass transport (n-1). The overall value of the
dimensionless product is about 1.
Higher yields and (or) smaller filters are provided by slow sand
filtration. Here some partial destabilization is accomplished by the
production of destabilizing chemicals on and within the filters. These
destabilizing chemicals are biopolymers which are produced by microor-
ganisms in the surface layers of the bed and which can flocculate
colloids (34). Attachment is improved; ay increases by two orders of
magnitude to 10 . This chemical process permits a reduction in trans-
port effectiveness; L/d and n are each reduced by an order of magnitude
to 10 and 10 respectively. The dimensionless product is again about
1. Effective filtration is accomplished by trading off filter size for
chemicals.
Still higher filtration rates and smaller filters are provided by
conventional rapid sand filters. Here chemical pretreatment with alum
is required. Attachment is again improved by chemical means, thereby
permitting a reduction in transport effectiveness. The dimensionless
filter product is again estimated to be about 1. The resultant trade-off
of filter size for chemicals still permits effective filtration.
Alum does not provide complete destabilization. In practice this
fact has led to the use of polymeric "coagulant-aids" or "filter-aids,"
ranging from activated silica to present synthetic organic polymers.
These polymers function by improving the efficiency of particle attach-
ment in flocculation tanks and in filter beds. When used, attachment
becomes so effective that particle removal is accomplished within the
uppermost portion of conventional rapid sand filter beds, resulting in
excessive head losses. Polymeric filter aids therefore not only permit
but actually require that transport efficiency be reduced by using larger
filter media, multimedia beds, upflow filters, etc. This effective
73
-------
attachment also permits the use of higher filtration rates (ultra high-
rate filtration, Table 5). The product <* n(L/d) is again about 1. Poly-
mer is used as a trade-off for filter size. However, a limit to filter
development can arise here. Particle attachment cannot be more than
100 per cent effective (a-sl). It is probable that the use of alum and
r
polymers produces virtually complete destabilization. Further trade-offs
between chemical and filter costs are not possible. Complete destabili-
zation can permit the use of large media (d=5mm, L/d^lO^), and high fil-
"? —3
tration rates (v=4-10gpm/ft , n-10 ), but additional reduction in par-
ticle contact opportunities cannot be accomplished by increasing the
effectiveness of the chemical component of the process. The continuing
development of the filtration process from infiltration galleries where
no chemicals are used (ctT,=10-5) and transport is very effective (n=l) to
r
ultra-high rate filters in which chemicals are very effective (aF-l) and
transport is reduced (n-l&~3) is completed.
74
-------
SECTION VI
REFERENCES
1. O'Melia, C.R., Discussion of "Theory of Water Filtration," by T.R.
Camp, Proceedings of the American Society of Civil Engineers.
Journal of the Sanitary Engineering Division, Vol. 91, No. SA2,
92-98, (1965).
2. O'Melia, C. R, and Stumm, W., "Theory of Water Filtration," Journal.
American Water Works Association, Vol. 59, 1393-1412, (1967).
3. Yao, K.M., "Influence of Suspended Particle Size on the Transport
Aspect of Water Filtration", unpublished Ph.D. dissertation, Univer-
sity of North Carolina, Chapel Hill, N.C., 1968.
4. Yao, K.M., and O'Melia, C.R., "Particle Transport in Aqueous Flow
Through Porous Media," paper presented at 16th Annual Conference of
the Hydraulics Division of ASCE, MIT, Cambridge, Mass., available as
ESE Publication No. 210, Department of Environmental Sciences and
Engineering, University of North Carolina, Chapel Hill, N.C., 1968.
5. Yao, K.M,, Habibian, M.T., and O'Melia, C.R», "Water and Wastewater
Filtration: Concepts and Applications," Environmental Science and
Technology. Vol. 5, 1105-1112, (1971).
6. Habibian, M.T., "The Role of Polyelectrolytes in Water Filtration",
unpublished Ph.D. Dissertation, University of North Carolina, Chapel
Hill, N.C., 1971.
7. Ives, K.J., and Gregory, J., "Basic Concepts of Filtration",
Proceedings of the Society for Water Treatment and Examination.
Vol. 16, 147-169, (1967).
8. Friedlander, S.K., "Theory of Aerosol Filtration," Industrial and
Engineering Chemistry, Vol. 50, 1161-1164, (1958).
9. Bishop, D. F., Marshall, L.S., O'Farrell, T.P., Dean, R.B., O'Connor,
B., Dobbs, R.A., Griggs, S.H., and Villiers, R.V., "Studies on
Activated Carbon Treatment", Journal, Water Pollution Control
Federation. Vol. 57, 1547-1560, (1965).
75
-------
10. O'Melia, C.R., and Crapps, D.K. , "Some Chemical Aspects of Rapid
Sand Filtration", Journal, American Water Works Association, Vol.
56, 1326-1344, (1964).
11. FitzPatrick, J.A. ,"Mechanisms of Particle Capture in Water Filtration,"
unpublished Ph.D.dissertation, Harvard University, Cambridge, Mass., 1972,
12. Spielman, L.A., and FitzPatrick, J.A., "Theory of Particle Collec-
tion under London and Gravity Forces for Application to Water Fil-
tration," Journal of Colloid and Interface Science, in press.
13. Camp, T.R., "Flocculation and Flocculation Basins," Transactions
of the American Society of Civil Engineers, Vol. 120, 1-16 (1955).
14. Conley, W.R. and Pitman, R.W., "Test Program for Filter Evaluation
at Hanford," Journal, American Water Works Association, Vol. 52,
205-218, (I960).
15. Tchobanoglous, G., "Filtration Techniques in Tertiary Treatment,"
Journal, Water Pollution Control Federation, Vol. 42, 604-623,
(1970).
16. Vanderhoff, J.W., Van Den Hul, H.J., Tausk, R.J.M., and Overbeek,
J.Th.G., "The Preparation of Monodisperse Latexes with Well-Charac-
terized Surfaces" in "Clean Surfaces", ed. by Goldfinger, G., Marcel
Decker, Inc., New York, 1970.
17. Sawyer, C.N. and McCarty, P.L., "Chemistry for Sanitary Engineers."
McGraw-Hill, New York, p. 110, 1967.
18. Standard Methods for the Examination of Water and Wastewater, 12th
Edition, American Public Health Association, Inc., New York, 1965.
19. Tanford, C., Physical Chemistry of Macromolecules, John Wiley and
Sons, New York, 1962.
20. Hanson, R.L., Walker, W.C., and Brown, J.C., "Variations in
Characteristics of Wastewater Influent at the Mason Farm Wastewater
Treatment Plant, Chapel Hill, North Carolina," UNC Wastewater
Research. Center Report No. 13, (1970).
21. Hazen, Allen, Annual Report of the Massachusetts Board of Health
0-892).
22. Basaran, A.K.T., "Direct Filtration of Alum-Treated Secondary
76
-------
Effluent," unpublished master's report, Department of Environmental
Sciences and Engineering, University of North Carolina, Chapel Hill,
N.C., (1972),
23. Environmental Protection Agency, Methods for Chemical Analysis of
Water and Wastes, 259-269, Water Quality Office, Analytical
Quality Control Laboratory, Cincinnati, Ohio.
24. Mints, D.M., "Aids to Coagulation," International Water Supply
Association, Sixth Congress, Stockholm, 1964.
25. Dixon, J.K., La Her, V.K., Li, Cassian, Messinger, S., and Linford,
H.B., "Effect of the Structure of Cationic Polymers on the Floccula-
tion and the Electrophoretic Mobility of Crystalline Silica,"
Journal of Colloid and Interface Science, Vol. 23, 465-473, (1967),
26. Singley, J.E., Maulding, J.S. and Harris, R.H., "Ferric Sulfate as
a Coagulant," Water Works and Wastes Engineering, Vol, 2,No. 3, 52,
(1965).
27. Black, A.P., Birkner, F.B. and Morgan, J.J., "Destabilization of
Dilute Clay Suspensions with Labeled Polymers," Journal. American
Water Works Association, Vol. 57, 1547-1560; (1965).
28. Black, A.P., and Vilaret, M.R., "Effect of Particle Size on Turbidity
Removal," Journal, American Water Works Association, Vol. 61, 209-214,
(1969).
29. Rob.eck, G.G,, Dostal, K.A., and Woodward, R.L., "Studies of Modifi-
cations in Water Filtration," Journal, American Water Works Associa-
tion, Vol. 56, 198-213, (1964).
3Q. Spurny, K.R., Lodge, T.P., Jr., Frank, E.R., and Sheesley, D.C.,
"Aerosol Filtration by Means of Nucleopore Filters-Structural and
Filtration Properties," Environmental Science and Technology. Vol.
3, 453-468, (1968).
31. Black, A.P., and Smith, A.L., "Determination of the Mobility of
Colloidal Particles by Microelectrophoresis," Journal, American
Water Works Association, Vol. 54, 926-934, (1962).
32. Chiang, C.H., "Filtration of Calcium Phosphate Suspensions Using
Polymeric Filter—Aids," unpublished master's report, Department of
77
-------
Environmental Sciences and Engineering, University of North
Carolina, Chapel Hill, N.C., (1973).
33. O'Melia, C.R., "Coagulation and Flocculation," Chapter 2 in
Physiochemical Processes for Water Quality Control, edited by W.J.
Weber, Jr., John Wiley and Sons, New York, 61-109, (1972).
34. Busch, P.L., and Stumm, W,, "Chemical Interactions in the Aggrega-
tion of Bacteria: Bioflocculation in Waste Treatment," Environmental
Science and Technology. Vol. 2, 49-53 (1968).
78
-------
SECTION VII
PUBLICATIONS
1. Yao, K.M., Habibian, M.T., and O'Melia, C.R., "Water and Wastewater
Filtration: Concepts and Applications," Environmental Science
and Technology, Vol. 5, 1105-1112, (1971).
2. Habibian, M.T., "The Role of Polyelectrolytes in Water Filtration,"
unpublished Ph.D. dissertation, University of North Carolina, Chapel
Hill, N.C., (1971).
3. Basaran, A.K.T., "Direct Filtration of Alum-Treated Secondary Efflu-
ent," unpublished master's report, Department of Environmental
Sciences and Engineering, University of North Carolina, Chapel Hill,
N.C., (1972).
4. Habibian, M.T., and O'Melia, C.R., "Particles, Polymers, and Perform-
ance in Filtration," accepted for publication in Proceedings of the
American Society of Civil Engineers, Journal of the Environmental
Engineering Division.
5. Basaran, A.K.T., O'Melia, C.R., and Snodgrass, W.B., "Filtration of
Wastewater," paper presented at the 45th Annual Conference, Water
Pollution Control Federation. The manuscript is in preparation
and will be submitted to Journal. Water Pollution Control Federation.
6. O'Melia, C.R., "Physical and Chemical Considerations in Filtration
Practice," in Filtration of Water and Wastewater;, University of
Michigan, Ann Arbor, Michigan (in press).
7. Chiang, C.H., "Filtration pf Calcium Phosphate Suspensions Using
Polymeric Filter-Aids," unpublished master's report, Department of
Environmental Sciences and Engineering, University of North Carolina,
Chapel Hill, N.C., (1973).
79
-------
SECTION VIII
LIST OF SYMBOLS
AQ" = Total equivalent amount of anions in a polymer solution.
BODS = Five-day biochemical oxygen demand.
C = Total Equivalent amount of cations in a polymer solution.
o
C = Concentration of particles in a filter influent.
d = Size of the filter media.
d = Size of the particles in the filter influent.
P
f = Porosity of the filter bed.
G = Mean velocity gradient.
L = Bed depth of a filter.
p = Total number of positive charges on the polymer molecules in a
polymer solution.
SS = Concentration of suspended solids.
t = Time, also the detention time in a flocculation tank.
TOG = Concentration of total organic carbon.
TP = Concentration of total phosphorus.
v » Filtration rate.
80
-------
V = Total volume of a polymer solution.
a = Sticking factor in coagulation.
a^, = Sticking factor in filtration.
F
H = Single collector efficiency in filtration.
cfi = Floe volume fraction, the volume of suspended material per unit
volume of suspension.
81
-------
TECHNICAL REPORT DATA
{Please read Ifts&uctions on the reverse before completing}
1. REPORT NO.
EPA-670/2-74-032
2.
3. RECIPIENT'S ACCESSIO(*NO.
4.TITLE ANDSUBTITLE
THE ROLE OF POLYELECTROLYTES IN FILTRATION PROCESSES
5, REPORT DATE
April 1974j Issuing Date
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Charles R. O'Melia
8. PERFORMING ORGANIZATION REPORT NO
9. PERFORMING ORG MMtZATION NAME AND ADDRESS
Department of Environmental Sciences & Engineering
University of North Carolina
Chapel Hill, N.C. 27514
10. PROGRAM ELEMENT NO,
1BB043; ROAP 21-ASQ; Task 11
11. CONTRACT/GRANT NO.
R-800351
12. SPONSORING AGENCY NAME AND ADDRESS
National Environmental Research Center
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
13. TYPE OF REPORT AND PERIOD COVERED
Research 971/71 to 2/28/73
14. SPONSORING AGENCY CODE
5. SUPPLEMENTARY NOTES
16, ABSTRACT
This research has been conducted (1) to determine how destabilizing chemicals
function in enhancing the effectiveness of filtration processes, and (2) to con-
sider selected applications of destabilizing chemicals in filtration for the treat-
ment of waste-waters. The investigations have included (1) laboratory experiments
using polymers and latex suspensions, (2) laboratory and pilot plant experiments
using alum, polymers, and trickling filter effluent, and (3) laboratory experiments
using polymers and calcium phosphate suspensions.
The report includes conclusions regarding the mechanisms of polymer action
in filtration, the results that can be obtained using polymers as filter-aids,
and the application of this knowledge of the design of filtration processes for
wastewater treatment.
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. cos AT I Field/Group
* *
Filtration, Polyelectrolytes, Filters,
Polymers, Phosphorus
Trickling filter efflu-
ent, Destabilization
mechanisms, Nutrient
removal, Tertiary treat-
ment, Sewage effluents
13B
14B
3. DISTRIBUTION STATEMENT
Release to public
19. SECURITY CLASS (ThisReport}
UNCLASSIFIED
21. NO. OF PAGES
92
20. SECURITY CLASS (Thispage)
.UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
82
U.S. GOVERNMENT FEINTING OFFICE! 1974 - 757-582/51!"
------- |