WATER POLLUTION CONTROL RESEARCH SERIES DAST-4
Crazed Resin Filtration
of
Combined Sewer Overflows
U.S. DEPARTMENT OF THE INTERIOR FEDERAL WATER POLLUTION CONTROL ADMINISTRATION
-------
Crazed Resin Filtration
of
Combined Sewer Overflows
A Self-Adjusting and Self Cleaning
Filter Feasibility Study
FEDERAL WATER POLLUTION CONTROL ADMINISTRATION
DEPARTMENT OF THE INTERIOR
by
Hercules Incorporated
Allegany Ballistics Laboratory
Cumberland, Maryland
Contract No. lU-12-39
October 1968
-------
FWPCA Review Notice
This report has been reviewed by the Federal
Water Pollution Control Administration and
approved for publication. Approval does not
signify that the contents necessarily reflect
the views and policies of the Federal Water
Pollution Control Administration.
-------
ABSTRACT
This study was conducted with the support of the Federal Water
Pollution Control Administration under Contract 14-12-39 to evaluate
the feasibility of a self-adjusting and self-cleaning filter proposed
for use on combined sewer systems to remove a. large portion of suspended
solids from sewage influent during storm flow conditions. Increased
sewage flows during storm conditions in most combined sewer systems
dictate that most or all of the flow be diverted from treatment facilities
directly to the receiving body because the flow exceeds the treatment
capacity. If a large amount of the solids can be removed before discharge
to the receiving water, a major part of the pollutional load would be
removed.
The experimental filters tested in this program were cylindrical
structures about one foot in diameter by one foot in length. The raw
sewage influent is at the top, the filtrate comes through the cylindrical
walls, and the concentrate is discharged through a bottom discharge. The
materials of construction were fibers laid down in predetermined patterns
by a winding process and bonded in place by resins. The porosity of the
structure is imparted by a mechanical cracking or crazing of the resin
with care being taken not to rupture the fibers which hold the body
together. The self-adjusting feature is achieved by pressurizing the
structure slightly (as in a water head buildup during storm flow) to
slightly enlarge the pores. Water and a small amount of very fine
suspended solids pass through the structure, leaving behind a more
concentrated suspension of solids. The filtered water is then discharged
to the receiving body. Upon release of the pressure, the pores close to
shut off all flow through the walls and to divert the ertire flow through
the normal treatment operation .
In contrast to normal filters where a filter cake is allowed to
build up, the self-cleaning feature is achieved by forcing the sewage
flow to create a hydrodynamic scouring action against the smooth
cylindrical walls to keep the solids from building up. Thus, a large
part of the water goes through the walls, while most of the sewage
solids remain suspended in the concentrate stream which is then passed
through the normal treatment plant.
Design goals for the filter system were as follows:
(1) A filtration rate of 7 gpm/ft2 of filter surface. This
rate would be needed to design filters of such a size that
they would be compatible with the available space in typical
sewage treatment plants.
iii
-------
(2) A ratio of influent sewage flow to concentrate flow of ten to
one, the difference between influent and concentrate being
the filtrate flow (very low in suspended solids) which would
pass through the filter wall and be discharged into the
receiving body.
(3) A filtration system which could be continuously operated for
two to three hours under storm flow conditions without plugging.
(4) Sufficient design data to allow scale-up to an operational
filter.
The study consisted of an experimental fabrication and testing
program, which was supported by structural design analysis, and a
theoretical hydraulic analysis directed toward developing hydrodynamic
design parameters to aid the self-cleaning feature of the filter.
Filters were designed which had an initial filtration rate with
tap water of 25 gpm/ft*, considerably over the design rate with the
low pressures (4 psi) required for the proposed filter operation. The
fabrication technique needed for this filtration rate was achieved near
the end of the program, and it was not possible to evaluate such filters
extensively with raw sewage during the program. However, sufficient
data and fabrication techniques were developed to be able to reliably
design filters with these filtration rates.
Although the design ratio of ten to one of influent to concentrate
was not achieved, one experimental run showed a ratio of four to one of
influent to concentrate for a short time at the beginning of the test.
Solids content of the filtrate was 627» less than that of the influent,
indicating the degree of improvement in discharge quality to be expected
from this system. Although this was probably the best experimental
result obtained during the program, two factors should be pointed out.
The filter (04B) was one of the early ones, and as such had relatively
few pores and also relatively coarse pores which, when plugged, gave
low filtration rates. Filters developed toward the end of the program
should give both higher filtration rates and a greater reduction in
solids content in the filtrate due to the greater number of pores and
the smaller pore size. The second factor to be pointed out is that this
filtration rate, as in all of the runs, could not be maintained because
of plugging of the filter, and the filtration rate gradually decreased
during testing.
This plugging, mostly by fibers in the early filters and by slime
in the later finer porosity filters, is the major unsolved problem in
developing the filter concept. Data developed during experimental testing
iv
-------
showed that the hydrodynamic cleaning in conjunction with mechanical
flexing of the filter walls would in effect maintain or renew original
filtration rates. Several concepts were advanced to utilize these data
to devise alternate self-cleaning concepts in which the hydrodynamic
scouring action is augmented by mechanical flexing (automatically controlled)
of the filter walls. Such a concept is currently being evaluated in an
independent research and development investigation.
Although the self-cleaning capability was not demonstrated, it was
shown that the hydrodynamic self-cleaning approach would give partial
cleaning of the filter. If the hydrodynamic self-cleaning approach can
be successfully augmented by an automated mechanical flexing operation,
indications are that continuous operation of the filter should be
possible.
If the alternate self-cleaning approaches are successful in main-
taining the required filtration rates, then it is highly probable that
successful self-cleaning, self-adjusting filters can be developed. This
statement is based on the results that show that the salient features
of the filter except self-cleaning were achieved during the program.
The desired filtration rates and suspended solids reduction can be
achieved and would enable the filters to be used in combined sewer
operations, either at the treatment plant or at outfalls in the sewer
system.
An analysis of the storm flow influent to the Bowling Green Waste
Treatment Plant was made and showed a 20-fold increase in solids content
during the initial flush while the flow rate tripled.
-------
TABLE OF CONTENTS
Page
ABSTRACT iii
LIST OF TABLES viii
ILLUSTRATIONS ix
INTRODUCTION 1
SUMMARY 5
Problem and Approach 5
Accomplishments and Results 6
Conclusions and Recommended Future Work 12
DISCUSSION 15
Program 15
Materials Development 16
Filter Fabrication 20
Fluid Dynamics 32
Waste Water Characterization 37
Filter Feasibility Testing 42
Filtration Plant Costs 74
GLOSSARY 77
APPENDIX A Hydrodynamics of Sewage Filtration
APPENDIX B Sewer Filter Characterization Program
vii
-------
LIST OF TABLES
TABLE 1 Effect of Sewage on Candidate Filaments
TABLE 2 Effect of Sewage on Candidate Resins
TABLE 3 Filament-Resin Bondability Study
TABLE 4 Summary of Filter Fabrications
TABLE 5 Coulter Counter Data
TABLE 6 Sewer Filter Test #1443-6
TABLE 7 Sewer Filter Test #1443-7
TABLE 8 Sewer Filter Test #1443-9
TABLE 9 Sewer Filter Test #1443-9A
TABLE 10 Sewer Filter Test #1443-10
TABLE 11 Sewer Filter Test #1443-11
TABLE 12 Sewer Filter Test #1443-12
TABLE 13 Sewer Filter Test #1443-13
TABLE 14 Sewer Filter Test #1443-14
TABLE 15 Sewer Filter Test #1443-15
TABLE 16 Sewer Filter Test #1443-16
TABLE 17 Sewer Filter Test #1443-17
TABLE 18 Suspended Solids Removal Summary
TABLE 19 Summary of Filter Feasibility Testing
Page
17
18
19
31
43
52
53
54
55
56
57
58
59
60
62
64
65
68
72
viii
-------
ILLUSTRATIONS
Page
FIGURE 1 Self-Adjusting and Self-Cleaning Filter Concept 4
FIGURE 2 Filter Fabrication - Winding of First Helicals 22
FIGURE 3 Filter Fabrication - Winding of Doublers 22
FIGURE 4 Filter Fabrication - Stripping Filter from Mandrel 23
FIGURE 5 Filter Fabrication - Completed Filter 23
FIGURE 6 Filter 01A - Exterior View 24
FIGURE 7 Filter 01A - Interior View 24
FIGURE 8 Desired Crazing (10X) 27
FIGURE 9 Axial Crazing (Filter 04A) 27
FIGURE 10 Rapid Pressurization Crazing (Filter 03A) 28
FIGURE 11 Crazing and Gel Coat Separation (Filter 03B) 28
FIGURE 12 Filter 12B - Exterior View 30
FIGURE 13 Baffle Body - 10" x 6" Truncated Cone 34
FIGURE 14 Baffle Body - Ramp and Cylinder 34
FIGURE 15 Baffle Body - Deck Plate Cylinder 35
FIGURE 16 Dry Weather and Rainstorm Flow 38
FIGURE 17 Bowling Green Waste Treatment - Storm Flow Analysis 40
FIGURE 18 Particle Size Distribution - Tesc #1443-6 44
FIGURE 19 Particle Size Distribution - Test #1443-9 45
FIGURE 20 Particle Size Distribution - Test #1443-10 46
FIGURE 21 Particle Size Distribution - Test #1443-14 47
FIGURE 22 Overall Test Arrangement 49
FIGURE 23 Test Stand With Filter 49
ix
-------
Page
FIGURE 24 Detail of Subscale Filter Test Assembly 51
FIGURE 25 Filter 12B (After Testing) 71
FIGURE 26 Filter 12B - Pressurized 71
-------
INTRODUCTION
Pollution of natural waterways by sanitary sewage from communities has
become an increasingly serious problem as the population grows and moves to
urban areas. The sewerage systems receiving combined sanitary and storm
flows are a major source of this pollution. Waste treatment plants serving
these systems are designed to provide adequate treatment for normal con-
ditions; however, storm flows are frequently many times normal influent rates
and exceed the plant capacity for treatment. During these periods, the plants
are forced either to overload their treatment facilities, thereby reducing
efficiency, or to bypass a portion of the excess flow without treatment.
Either choice increases pollutional loading to the receiving waters.
The current solutions to the combined sewer problem are (1) to enlarge
treatment facilities to handle these intermittent excessive loads or (2) to
separate the sanitary and storm sewer systems. Both methods are extremely
costly and usually not feasible.
The Hercules approach to the storm flow pollution problem is the use
of a self-adjusting and self-cleaning filter that will concentrate the
solids loading of the influent to a combined waste treatment plant by
removing the surplus water that enters the system. The filtered water
would then be chlorinated and discharged into the plant outfall, thereby
reducing pollution of the receiving waters.
For some time it had been noted that filament-wound pressure vessels
would leak at pressures well below failure pressures, and this leaking rate
was found to increase directly with the amount of pressure applied. Micro-
scopic examination of the interior surface of the vessels showed that the resin
structure was cracked in a fine, random, hairline pattern, which has been
-------
termed "crazing." It was theorized that this inherent problem in composite
filament wound structures could be used to advantage in developing an
adjustable leak rate filter. This leak rate could be varied by controlling
the expansion of the vessel which, in turn, was determined by the winding
geometry and structural materials. The crazed resin would perform as the
filtering medium. It was felt that these crazings or pores could be kept
free from plugging materials by fluid flow induced shearing forces that
would prevent formation of a filter cake. These factors formed the basis of
the proposed self-adjusting and self-cleaning filter.
The original concept proposed use of low-modulus, high-elongation
filaments helically wound and bonded in place with a thermosetting resin
to yield a structure that would flex under low hydraulic pressures. The
inner surface of the structure was to be coated with a smooth resin layer,
or gel coat, that would craze and move with the supporting wound filaments,
producing a permeable surface to allow passage of liquids at a rate proportional
to the pressure applied. The shearing forces needed to keep the pores from blinding
were to be supplied by a specially designed baffle body that would cause the
solids containing concentrate liquid to constantly sweep the wall,allowing
continuous filtration.
To reduce this concept to practice, Hercules has conducted a six-month
study under PWPCA Contract No. 14-12-39 to demonstrate feasibility of the
system. This work was necessary to advance the technology from construction of
pressure vessels in which leaking is not desired to that of the filter
where controlled pore opening and controlled leaking are the main principles
involved. The goals set forth in this program were directed to producing a
feasibility model and fully characterizing this model.
-------
A schematic of the proposed method of filtration is shown in Figure 1.
The filtering element is placed in line with the plant influent to serve
as a conduit under normal flow conditions, passing the flow to treatment.
During storm flow conditions, the increased flow increases the hydrostatic
head in the filter. As the hydrostatic head increases, the filter struc-
ture expands, causing the wall to become permeable and allowing water to
pass to the outer chamber where it flows to a chlorination chamber for
treatment and subsequent discharge to the receiving water.
The rate of filtrate flow will depend upon the water level in the upper
chamber, which reflects the rate of storm water entering the plant. Because
of increased pressures, there will also be some increase in flow through
the bottom discharge of the unit; however, this stream will be controlled
within the limits of the treatment capacity of the plant. The solids re-
moved by the filter element will be carried within the element to the bottom
discharge and then to treatment.
-------
No Filtrate Flow
-Swirl Baffle
Filter
(Unexpended &
Impermeable)
NORMAL FLOW
CONDITIONS
To Treatment
Filtered Water
To Chlorination
INCREASED HEAD DUE
TO STORM FLOW
-Filter (Expanded)
STORM FLOW
CONDITIONS
To Treatment
FIGURE 1
SELF-ADJUSTING AND SELF-CLEANING FILTER CONCEPT
-------
SUMMARY
Problem and Approach
During storm conditions, the flow in a combined sewer system
greatly exceeds the capacity of the treatment plant. Consequently, a
part or all of the flow is diverted from the treatment plant directly
into the receiving body, resulting in increased pollution. A large
amount of the pollution caused in this manner is due to the suspended
solids contained in the sewage. Expecially during the initial flush,
the concentration of suspended solids can be 10 to 30 times higher than
in normal flow, even with the dilution caused by the storm water. Normal
filtration systems are not effective in removing suspended solids due to
plugging of the pores.
As an approach to solving this problem, a self-adjusting, self-
cleaning filtration system was conceived in which the incoming sewage
is separated into two streams. One stream, the filtrate, is much the
greater and passes through the filter, leaving the solids to build up
in the smaller stream, the concentrate, which is directed through the
normal treatment facilities. Filter cake build-up and resulting pore
plugging are prevented by a hydrodynamic scouring of the filter walls
by the sewage flow.
The areas of investigation included an examination of filter
materials, structural analysis of the composite filter element, hydro-
dynamic analysis of the flow field, and experimental evaluation of test
vehicles with domestic sewage.
-------
Accomplishments and Results
This study was conducted to evaluate the feasibility of a proposed
self-cleaning, self-adjusting filter concept for use on combined sewer
systems. Conventional filters, when used to filter or strain raw sewage,
plug very rapidly due to fiber and slime build-up in the pores.
In the present concept of a filter, a dynamic filtration system
was evaluated in which the pores of the filter structure expand (very
slightly) as pressure is applied as would happen during storm flow in a
combined sewer system. As these pores open up, large amounts of water
pass through the walls, including a small amount of very fine suspended
or colloidal sewage solids. When the pressure is reduced as when a storm
flow subsides, these pores contract and close, shutting off the flow
of excess water through the walls and directing the full flow again to
the treatment plant. This alternate opening and closing of the pores
under slight pressure variation constitutes the self-adjusting feature
of the filter.
In a conventional filter, all of the liquid passes through the walls
or filter media and the solids build up as a cake. When raw sewage is
passed through such a filter, complete plugging or blinding occurs very
rapidly. In the current study of sewage filtration, a self-cleaning
system is being developed to counteract this problem. In this concept,
the filter is designed to eliminate buildup of a filter cake and the
inflow is separated into two streams - the filtrate and the concentrate.
The filtrate stream (designed to be up to ten times the flow of the
-------
concentrate) passes through the filter walls and, being very low in
suspended solids, is treated by chlorination and passed directly into
the receiving stream while the concentrate stream (designed to be one-
tenth of the total flow) contains a high concentration of sewage solids
and is directed through the normal sewage treatment plant process. To
maintain the flow of water through the walls of the filter and to prevent
plugging of the self-adjusting pores, a hydrodynamic self-cleaning
technique is being evaluated. In this process, through the introduction
of high flow rates on the influent side of the filter body, a hydrodynamic
scouring action occurs to sweep the suspended solids away from the filter
pores. By utilizing thin walls, alternate slight fluctuations in pressure
effect variations in pore size and cause the particles which do become
trapped in the pores to be swept on through with the filtrate. This
constitutes the self-cleaning aspect of the system being studied in this
program. Design goals for the pressure fluctuations are 2 to 4 psi,
equivalent to a water head build-up of from about four to eight feet.
To combine these concepts into one filter, a filament-wound composite
structure was conceived as the test vehicle. The filter on which most of
the tests in this program were carried out, consisted of a cylindrical
structure one foot in diameter and one foot long made from fibers (poly-
propylene, polyester, or fiberglass) wound in predetermined patterns and
impregnated with liquid resins. Upon curing of the structure, the resins
harden to give a self-supporting, thin-walled (about 0.1 inch or less)
cylinder of fibers intertwined in a hardened resin matrix. End plates
are applied to the structure, resulting in a thin-walled pressure vessel.
-------
The desired porosity is imparted to the nonporous (at this point)
structure by flexing the structure (mechanically or with pressurization),
causing the resin portion of the structure to crack or craze into very
fine openings with the fibers preventing complete rupture. These cracks
then alternately open and close by the application or removal of slight
pressure heads. The self-cleaning aspects of the filter are imparted
to the structure by flow regulation, either by rate control into, through,
and out of the structure, or by center body baffles or by a combination
of these methods.
The study consisted of an experimental design, fabrication and
testing program, which was supported by structural design analysis, and
a theoretical hydraulic analysis directed toward developing hydrodynamic
design parameters to aid the self-cleaning feature of the filter.
A design filtration rate of 7 gpm/ft2 was selected as a goal
during this work. This goal was selected on the basis of practical
filtration units small enough for use in most combined sewer systems.
For example, a combined sewer system with a normal flow of 100,000
gal/day might experience a ten-fold increase or up to 1,000,000 gal/day
rate under storm flow conditions. It would be desirable to pass about
900,000 gallons of this flow through the filter. To do this at 7 gpm/ft
would require 90-100 sq. ft. of filtering area, equivalent to four 5 ft x
5 ft panels of filter surface or a cylinder a little more than 5 feet in
diameter and 5 feet high. Engineering design would optimize size, shape,
and placement of the filtering media to reduce the space required to a
minimum.
8
-------
Considerable effort was expended on developing fabrication techniques
for the filters. The initial filters produced had relatively coarse
crazing and relatively low filtration rates, while toward the end of the
program, filters were produced with the desired fine crazing and very
high filtration rates. All filters were first tested with tap water to
establish the ultimate capability of the filter. (It was found that the
tap water rate and the initial filtration rate with raw sewage were very
nearly the same.) A 6 in.-dia. filter made with polar fiber windings was
produced which had an initial filtration rate of 25 gpm/ft^ at 4 psi and
which passed no water at 0 psi. This filter was fabricated near the end
of the program and was not completely tested. It was only possible to
fabricate one 12 in.-dia. filter using the same technique as above.
Although it was not possible to completely reduce this filter concept
to practice during the period of performance, the salient features of
the concept, with the exception of self-cleaning, were demonstrated.
Filters were fabricated which were self-adjusting for varying water
flows at pressures varying between 0 and 10 psi.
Fabrication techniques were developed to produce filters capable of
?\
filtration rates of 25 gpm/ft at 4 psi, considerably above the design
goal of 7 gpm/ft . This was achieved with a 6 in.-dia. experimental
filter with polar windings.
One filter, 04B, was operated slightly under the design goal of
7 gpm/ft2 for several minutes before plugging occurred. During this test,
-------
the ratio of filtrate flow rate to concentrate flow rate was approximately
three to one. Suspended solids, as measured in the filtrate stream, were
only 38% of the incoming stream. This filter did not contain the ideal
porosity developed later but did clearly demonstrate the potential of
the system.
Filters were reactivated with no additional treatment after being
allowed to dry out with the pores plugged. Application of pressure to
the filters after drying restored the original rates of filtration. This
indicates the feasibility of on-off operation as would be experienced in
intermittent storm flows.
The experimental studies did not include evaluations of the theoretical
fluid flow analysis due to the timing in the program. Baffle designs and
fluid flow rates were made on a "best guess" basis. All of the filters
plugged after short periods of operation, showing the need for further
work in the self-cleaning area. However, the following promising results
were obtained:
1) The original filters had relatively coarse crazings and were
found to plug with fibrous materials. The improved filters
with fine crazing made toward the end of the program became
plugged but more with slime than with fibrous materials.
Solids passing through these filters also had much smaller
particle size as shown by Coulter Counter particle size data.
2) A hydrodynamic lift-drag theory was developed which suggests
that particles can be kept away from the filter wall, thus
10
-------
helping to prevent plugging. This theory was not tested
during the program.
3) Indications were that a purely hydrodynamic approach to self-
cleaning would not be sufficient to keep the filters from
plugging. Consequently, several approaches to a mechanical
augmentation of the hydrodynamic cleaning feature were conceived.
The.se concepts would also be capable of automatic operation
and are based on momentary intermittent backflushing of the
filters. Manual flexing of the filters tested during the
program restored the flow after plugging had occurred, showing
that this approach could be effective.
An analysis of the storm flow influent to the Bowling Green Waste
Treatment Plant was made and showed that during the initial flush, the
suspended solids content of the stream went from a normal content of
270 ppm to 4400 ppm within a two-minute period. For this storm, con-
sisting of 0.25 inch of rain over 30 minutes, the flow rate went from
the normal 250,000 gal/day to about 760,000 gal/day within 30 minutes
after the storm began. Flow rate and suspended solids content returned
approximately to normal about 2 hours after the end of the storm. This
shows that while in a strict sense, the Bowling Green system is not a
combined system (storm sewers do not purposely enter into the Sanitary
System) , there are unknown sources of storm water entering the sewer
system, being both old, unmapped storm sewers and illegally connected
downspouts. The effect is one of combined sewers except for the amount
of grit and street washings which would normally be expected in a combined
11
-------
system. During the program, it was also determined that there was
considerable ground water infiltration into the system during wet periods.
Additional experimental work is required before an operational
demonstration unit can be designed. However, preliminary design concepts
have been considered using assumed values where engineering data were
not available to develop a preliminary cost estimate. This estimate
suggests that a plant the size required by Bowling Green (0.1 MGD normal
flow) could operate with the filter system for a $30,000 capital invest-
ment with yearly operating costs of about $950. In larger plants the
operating cost appears to be about $0.01/1000 gallons of normal flow
through the plant.
Conclusions and Recommended Future Work
Conclusions drawn from results of this study are as follows:
1) Filters were developed with the desired fine pores which had
initial filtration rates greater than the design rate of
7 gpm/ft2.
2) Techniques were developed to design and build filters with
the desired fine porosity and high filtration rates.
3) In actual filtration runs, it was shown that suspended solids
could be reduced by as much as 62%. This test was made with
a filter having relatively coarse porosity, and indications
were that filters having the very fine porosity developed
toward the end of the program would have even better solids
removal effic iency.
12
-------
4) The original hydrodynamic self-cleaning concept was not
completely demonstrated. However, the studies suggest that
a lift-drag hydrodynamic cleaning with automatic mechanical
flexing of the filter could achieve the self-cleaning function.
Manual flexing of the filters during actual filtration runs
restored the filtration rate after plugging had occurred.
5) Sufficient data were generated during the study to show
promise that the filtration system could be used in combined
sewer systems when the self-cleaning feature is achieved.
6) When the self-cleaning feature is fully developed, the filter
system should be useful in combined sewer systems for concen-
trating solids in storm flows at the treatment plant and for
installation at remote outfalls in a combined sewer system.
V
Data were obtained to show that actual filtration rates almost
four times the design goal were achievable. If the self-
cleaning approach can be fully developed, such high filtration
rates would be very useful for the installation of the filter
in combined sewer treatment plants and at combined sewer outfalls.
7) Although the filter would not be applicable as a direct replace-
ment for primary or secondary sedimentation because the
concentrate would not be of sufficiently high solids content
to be equivalent to primary sedimentation sludges, it would
have potential in greatly reducing the size of sewage treatment
plants by acting as a concentrator of raw or aerated sewage.
13
-------
It is recommended that the following studies be conducted before
an operational filter can be demonstrated:
1) Test the lift-drag hydrodynamic theory for particle impingement
reduction on the filter walls as suggested by this study. (A
disadvantage of this theory is that flow regulating pumps and
other appurtenances would be required for the necessary flow
control.)
2) Evaluate alternate methods of self-cleaning to be used in
conjunction with the hydrodynamic cleaning of the filter.
The methods suggested by this study are pressure cycling and
momentary back-flushing utilizing a water-hammer effect.
3) Conduct a theoretical study of the filter design and fabrication
techniques needed to make this filter economically feasible.
A suggested approach which shows much promise is a filter
consisting of insertable flat panels rather than a cylinder.
The panels can be flexed and thus cleaned in either direction,
thus greatly easing the .self-cleaning problem. An additional
e>
feature would be the ease of replacement and repair.
4) A pre-pilot demonstration facility testing program to establish
the features suggested above and to establish the economic
requirements.
Hercules is currently carrying out feasibility testing of Items 1,
2, and 3 in an Independent Research and Development Program. When these
studies are completed in about three months, a follow-on proposal will
be submitted recommending the program to be followed to lead to a successful
operating filter.
-------
DISCUSSION
Program
Performance of the work on this program is reported in five sub-
sections to allow treatment of the closely related areas. Because of the
nature of the filter involved, each area is interdependent on the other
areas and a number of items will be discussed in more than one sub-section.
The overall program was geared to follow the plan presented in Hercules
Incorporated's Proposal No. 14-12-39, Revision No. 1. This program is
defined by the following goals:
1. Determine the configuration, winding geometry, types of
fabrication materials, and methods of fabrication to yield a
filter capable of opening at relatively low hydraulic pressures
to pass filtered water and a small percentage of solid particles.
2. Determine the fluid mechanic requirements to produce the
necessary forces along the interior walls of the filter element
to provide effective self-cleaning action.
3. Build scale-model units and install them at the Bowling Green
Waste Treatment Plant for filtering feasibility studies using
municipal raw sewage.
4. Evaluate methods of eliminating, minimizing, and/or controlling
slime growths and grease on the filter surfaces.
5. Perform preliminary analytical studies to measure velocity
profiles and turbulence levels.
6. Estimate the potential capital and operating costs for a full
scale filter operation.
7. Recommend an approach, based on the feasibility study, for full
scale filter demonstrations.
15
-------
8. Characterize the influent to a combined waste treatment plant during
storm flow conditions.
To achieve these goals, materials were selected and evaluated, test structures
were produced and tested, and a paper study was conducted to theoretically
develop design parameters for a workable self-cleaning unit.
Several of the program goals were not fully achieved during the span of
the program. In general, this was due to the problems of reduced filtration
rates during filter testing with raw sewage where self-cleaning was only partly
successful. Thi-s prevented any accurate cost estimate and necessitated a
series of assumptions for the order of magnitude estimate provided. Also,
a recommended program for full scale demonstration would be premature at
this time since feasibility has not yet been demonstrated on all functions of
this filter.
It should be noted that the design goal for filtration rate was 23
gallons per minute through the 3.14 square foot wall area of the scale
model test element. This is equivalent to a filtration rate of just
over 7 gpm/ft2.
Materials Development
The selection of the materials for fabrication of the filters was based
on desired mechanical properties and bonding characteristics. A study was
performed in parallel with the fabrications to determine the effects of the
operating environment on the candidate materials. This study was carried out
by placing samples of the candidate material in the wet well of the municipal
waste treatment plant. At intervals of 2, 8, and 16 weeks, the samples were
removed, examined, washed with water, and tested to determine their mechanical
properties. The examination and washing were steps in the evaluation of the
types of growths and deposits formed and the ease of their removal. Due to
16
-------
the fibrous nature of the filaments, it was difficult to determine whether
the sewage material was adhering or merely entrapped. The resin specimens
were easily washed clean except for a few minor white deposits that occurred
on the Epon 826 resin samples.
The initial candidate filaments, a polypropylene and a polyester, showed
some sign of degradation due to the exposure. The polypropylene decreased 117»
in tensile strength and 17% in elongation, while the polyester lost 28% in
tensile strength and 14% in elongation. The effects of the exposure appear to
occur during the early periods and then level off as though a resistant layer
had formed on the filaments. The levels of degradation exhibited would have
little adverse effect on any structure-fabricated from these materials.
Table 1 lists the results of the testing.
TABLE 1
Effect of Sewage on Candidate Filaments
Exposure Time Tensile Strength Tenacity Elongation
Filament (Weeks) fibs) (g/denier)
Polypropylene
Polypropylene
Polypropylene
Polypropylene
Polyester
Polyester
Polyester
Polyester
Control
2
8
16
Control
2
8
16
6.59
5.93
5.62
5.89
8.93
6.96
6.11
6.40
4.75
4.27
4.05
4.24
8.10
6.31
5.54
5.80
30.5
23.6
23.6
25.4
13.3
11.5
11.1
11.4
The candidate resins were cast into ASTM D638 Type 1 tensile specimens,
having long thin (1/4 in.) gage lengths to allow more surface exposure to the
sewage. The high elongation resins were more seriously degraded than the
lowest ones. The prime candidate, Epon 826 epoxy, suffered little effect
from the sewage regardless of the curing catalyst employed. The tests results
are shown in Table 2.
17
-------
TABLE 2
Effect of Sewage on Candidate Resins
Resin
System
Ingredients
Epon 953A & B
(100/16 pbw)
Epon 826/DETA,/
Cab-0-Sil
(100/11/3 pbw)
Epirez-Epicure
(10/7 pbw)
Epon 826/ZZL-0803
(100/37.5 pbw)
Epon 826/DETA
(100/11 pbw)
Epon 871/ZZL-0803
(100/82.5 pbw)
Exposure
Time
(Weeks)
Control
2
8
16
Control
2
8
16
Control
2
8
16*
Control
2
8
16
Control
2
8
16**
Control
2
8
16
Tensile
Strength
(psi)
1746
1647
1185
1321
7890
7643
7193
7908
1201
696
410
349
4639
9165
8778
8522
5013
9379
9547
9638
61
76
80
69
Elongation
(%)
38.9
51.9
40.0
46.6
2.7
3.0
3.0
,2.7
52.5
51.8
36.0
26.6
5.2
5.8
5.5
4.7
4.7
5.1
5.9
4.1
21.5
28.6
25.0
23.5
Modulus
(psi x 103)
21.2
12.3
6.8
11.8
284.6
247.8
216.7
338.4
15.7
4.7
2.5
17.0
137.4
192.3
190.0
330.9
114.6
246.9
186.9
353.4
0.385
0.422
0.378
0.330
*Results of one sample only.
**Results of two samples only.
18
-------
During the program a special high modulus, low elongation epoxy was
formulated for use as a gel coat. The stoichiometry of the epoxy was
unbalanced to yield a brittle smooth glassy coating that would craze in a
very fine pattern. This epoxy was Epon 826 with Catalyst D. No sewage
exposure tests were conducted due to short time remaining in the program.
The initial filters were fabricated using a two-resin system, one for
the inner liner or gel coat and a second resin for winding the supporting
structure. When bond failures occurred during crazing operations, a bonda-
bility study found that to prevent interface bond failures, similar resin
formulations should be used and the first resin applied should not be
completely cured prior to the application of the second. This was necessary
since most of the resin strength is developed by chemical rather than
mechanical bonding.
Colored polypropylene filaments were used in the filter fabrications
since previous testing found the undyed material quite difficult to
bond. The surface treatment required to color the filaments improves
the mechanical bond quality. Studies were made to determine the bond
strength developed between the candidate filaments and resins. The winding
resins, Epon 953 and Epon 826, formed good quality bonds to the polyester
and fiberglass and only fair bonding to the polypropylene (Table 3), with
the exception of the bulk polypropylene (crimped fibers) to Epon 826
high strength bond.
TABLE 3
Filament-Resin Bondability Study
Tensile Strength (psi)
Filament Epon 953A & B Epon 826/ZZL-0803
Polypropylene 310 263
(1050 denier)
Polypropylene 290 2050
(Bulk 3500 denier)
Polyester 970 1250
Fiberglass 2200 3000 19
-------
When difficulties in obtaining the desired crazing were encountered
(see following section, "Filter Fabrication"), several filters were
fabricated using fiberglass. The "S" or structural type fiberglass,
which has good chemical and moisture resistant properties, was immediately
available and was used for testing to prove the feasibility of high
modulus fibers for this application. No environmental testing was done
on the "S" fiberglass because insufficient time remained in the contract
performance period.
Tests of composite samples of candidate filaments and resins in a sewage
environment were not conducted since no acceptable quality test has been
developed to determine degradation of thin walled structures which have been
intentionally cracked in the manner required for this filtration study.
Filter Fabrication
The original concept of the self-cleaning self-adjusting filter was a
filament wound structure in which bands of filaments are precisely wound
over a rigid mandrel and bonded with a thermosetting plastic. These bands
are positioned in a helical pattern, and netting analysis calculations are used
to design the desired structural characteristics. Axial and hoop moduli of
die structure are varied by choice of helix angle, number of wound layers and
fiber-resin materials.
The model chosen to demonstrate the feasibility of this filter was a
12-inch diameter cylinder with a 12-inch long filtering area. The semi-rigid
structure was composed of a filament wound shell and a permeable resin liner
or gel coat. The gel coat was made permeable by cracking or crazing this
resin liner. Calculations were made for each fabrication using the mechanical
properties of the chosen materials and operating parameters to determine the
winding requirements.
20
-------
Figure 2 shows the mandrel set in the winding machine at the time of
winding the first helical bands. In this arrangement, the individual strands
(bundles of filaments) are pulled from storage spools by the rotation of the
mandrel. Each strand passes through the funnel-shaped resin cup where the
strands are wetted and positioned side by side into a band which is shown
being wound on the mandrel. When the desired amount of material has been
wound, hoop (or 90°) windings are placed over the structure at designated
intervals to act as reinforcements or "doublers." These become the filter
sealing and clamping surfaces (see Figure 3). It should be noted that the
length of the mandrel was selected to produce two separate filter elements
for each winding. After the assembly has been cured at an elevated tempera-
ture, a cut is made through the material at one end of the mandrel and the
filters are stripped from the mandrel as seen in Figure 4. Cutting the
filters to length completes the fabrication (Figure 5).
The first filter fabrication employed the low modulus polypropylene
fibers to wind the reinforcement doubler areas. The residual strain in these
fibers caused these areas to shrink after removal from the mandrel making
attachment to the test stand quite difficult. To remedy this, the remaining
fabrications were made using high modulus fiberglass roving for these doublers .
After fabrication of the first filter, it was found that the yarn did
not spread in as wide a band as predicted, producing an open weave structure
as shown in Figures 6 and 7. The inner resin liner or gel coat, if left
unsupported by the winding filaments, will break and fall out after crazing
leaving large holes through the filter wall. Yarn, as distinguished from
roving, contains a twist in the filaments making up each strand. This twist
prevents the individual filaments from lying side by side and forming a
wide band; instead, it produces a rope effect. Fabrication calculations
involve the number of filaments present in each unit area of the filter
21
-------
FIGURE 2
Filter Fabrication - Winding of First Helicals
FIGURE 3
Filter Fabrication - Winding of Doublers
22
-------
I
FIGURE 4
Filter Fabrication - Stripping Filter from Mandrel
^pw
_^w^w
FIGURE 5
Filter Fabrication - Completed Filter
23
-------
FIGURE 6
Pilfer 01A-Exterior View
24
FIGURE 7
Filter 01 A- Interior View
G-2141
-------
walls, so for a fixed amount of material to cover a given area without
leaving holes, the individual filaments must align themselves side by side
to form a wide thin band rather than series of ropes. Commercial polypropy-
lene-polyester fibers are intentionally given twist to make their handling
easier and to prevent fraying. Later filters (as noted in Table 4) were fabri-
cated with fibers which had the twist removed to yield a wider band.
The completed filter (Figure 5) shows an inherent feature of helical
wound structures - the crossover - about one-third the length from the left
doubler. This area is where the bands interweave. During crazing attempts,
when cracks were opening along the helix or winding angle, the crazings were
not produced at these crossovers, reducing the amount of filtration area.
A different technique of fabrication was tried - convolute or polar winding -
which forms no crossovers. In this type of winding, all the filaments in
each ply lie parallel along the helix angle, and successive plies are placed
at the negative angle to the previous layer.
The inner resin liner, or gel coat, was applied to the filter elements
by two methods: (1) the resin was brushed onto the mandrel and partially
cured before winding the filament structure; and (2) the resin was brushed
on the interior of the completed filter element. The former method takes
advantage of the pre-strain placed in the winding filaments by the fabrica-
tion technique, causing the gel coat to close tightly after the crazing or
cracking operation. This causes the filter to act as a fluid conduit at low
pressure. Gel coat thickness was varied to determine the amount required to
produce the best crazing. With the first fabrications, gel coat thickness had
no effect on crazing quality. All crazings were coarse.
25
-------
During early crazing tests it was found that the gel coating had a
tendency to break away from the wound structure. To overcome this, filters
were fabricated using a single resin for both applications. Resulting
filters were improved, but the problem was still present.
Producing the desired random hairline cracks, "crazing," proved to
be a major problem in the filter fabrications. The desired crazing,
exhibited by fiberglass pressure vessels, is shown in Figure 8. The first
method used was to apply internal pressure to the filters while mounted on
the test stand, rigidly supporting the ends of the filters. Pressurization
rates varied from 2 to 25 psi/minute. The few resulting cracks were oriented
axially (Figure 9). In an attempt to increase the amount of crazing, rapid
pressurization was produced by use of a pyrotechnic device (explosive) in
water filled filters. The cracks produced were still axially oriented but
more frequent (Figure 10). Figure 11 shows the separation of the gel coat
from the supporting structure that occurred during early crazing attempts.
Since the structures have a higher modulus in the axial direction than in
the hoop direction due to the filament helix angle, axial loading was applied
to the filters during internal pressurization to strain the gel coat more
uniformly. The resulting crazings were oriented along the fabrication helix
angle. Another method used was to mechanically break the gel coat with
rollers and hammering devices. The crazings produced were random but not as
numerous or fine as desired. During fabrication of filter 04A and B, the gel
coating was allowed to cure until hard and the surface scratched with coarse
sandpaper to cause stress lines; then the structure was wound over the
preconditioned gel coat. The stress lines had no effect on the crazing
pattern produced. The gel coat of filter 10A was reinforced with a polyester
cloth to support any pieces which might break free from the wound structure.
26
-------
FIGURE 8
Desired Crazing (10X)
FIGURE 9
Axial Crazing (Filter 04A)
G1976
27
-------
FIGURE 10
Rapid Pressurization Crazing (Filter 03A)
FIGURE U
Crazing and Gel Coat Separation (Filter 03B)
G1977
28
-------
This cured gel coat was cracked before overwinding by placing a wire cloth
over the coated mandrel and hammering on wire cloth. The star-like cracks
remained in the gel coating after fabrication was completed but the cloth
failed during internal pressurization, leaving larger than desired cracks in
the gel coat.
Micromechanical analysis of the crazing problem produced the following
results: (1) A high ratio of fiber modulus to resin modulus causes a strain
factor favorable to production of cracks parallel to each fiber and around
(or across) each resin-encapsulated fiber. (2) A higher number of cracks
will be produced by reducing the fiber diameter and increasing the number
of fibers used in fabrication. (3) The helical interweaving method previously
being used should be changed to a convolute or polar pattern, eliminating
crossovers within each layer. (4) Use of a high modulus fiber will result
in a structure whose expansion is less, opening any cracks in the structure
to a lesser degree.
The recommendations of this analysis were used in fabricating filter
12A and B and several sub-scale 6-inch-diameter, 6-inch-long filter elements.
The analysis also indicated that slow-rate pressurization (0.1 psi or less/
min) would help increase the number of cracks. Filter 12A was crazed in
this fashion but as each crack opened, the strain was relieved in the
adjacent areas, resulting in few cracks oriented along the helix angle.
Filter 12B (see Figure 12) and the sub-scale filters were crazed by
mechanically flexing the structures parallel to their longitudinal areas
and then slowly pressurizing. The crazing produced were quite fine,
numerous and oriented along the fibers. These were the fine cracks sought.
The summary of filter fabrications in Table 4 shows the progression of
changes made as tests were performed to produce crazing and to determine
operating characteristics in the sewage environment. Major changes were
-------
FIGURE 12
Filter 12B-Exterior View
G-2142
30
-------
TABLE It
Summary of Filter Fabrication
Filter
Number
01A
01B
02A
02B
03A
03B
04A
04B
05A
05B
06A
06B
07A
07B
08A
08B
09A
09B
10A
10B
HA
11B
12A
12B
Fiber Winding Resin
Polypropylene Epon 953 A&B
(1050 denier) " " "
Polypropylene Epon 953 A&B
(1050 denier)
Polypropylene Epon 953 A&B
(1050 denier) " "
Polypropylene Epon 953 A&B
(1050 denier) " "
Polypropylene Epon 826/ZZL-0803
(Bulk-3500 denier) " "
Fiberglass Epon 826/ZZL-0803
(S-994 Roving) " "
Polypropylene Epon 826/ZZL-0803
(1050 denier -low "
twist)
Fiberglass Cloth Epon 826/ZZL-0803
and Fiberglass " "
Polyester Epon 826/ZZL-0803
(1050 denier -low " "
twist)
Polypropylene Epon 826/ZZL-0803
(1050 denier -low " "
twist)
Polyester Epon 953
(1000 denier -low " "
twist)
Fiberglass Tape Vendor Proprietary
Fiberglass Tape "
and polyester
Gel Coat Resin
Epon 826/DETA
ii ii
Epon 826/DETA
Epon 826/DETA
n >i
Epon 826/DETA
n n
Epon 826/ZZL-0803
None
Epon 826/ZZL-0803
None
Epon 826/ZZL-0803
None
Epon 826/ZZL-0803
None
Epon 826/ZZL-0803
Epon 826/DETA
Epon 826/ZZL-0803
& polyester cloth
Epon 826/ZZL-0803
Epon 82 6 /Cat. D
None
None
None
Winding Angle
30°
If
26°
26°
u
15°
II
- 5 layer s
- 2 layers
n
- 2 layers
n
- 2 layers
15° - 1 layer
+ 90° overwind
ii n
15°
15°
It
15°
II
15°
II
15°
15°
- 2 layers
it
- 2 layer s
- 1 layer
- 2 layers
- 2 layers
it
- 2 layer s
ii
11° - 4 plies -
convolute or Polar
°° with 90° poly-
Gel Coat
Thickness
6-7 mils
n
16-17 mils
6-7 mils
M
6-7 mils
6-7 mils
None
6-7 mils
None
6-7 mils
None
6-7 mils
None
6-7 mils
If
6-7 rails
2-3 mils
6-7 mils
None
None
None
Crazing Method
Rapid
Rapid
Med & Rapid (Air)
Rapid (Hot Water)
Rapid (Cold Water)
Rapid
Rapid
Rapid
Rapid
Slow
Slow with axial compression
Slow cycling
Slow
Slow
Slow 0-60 psi cycles
Very slow, hoop reinforcement
Very slow
Slow
Slow and mechanical flexing
Slow and mechanical flexing
None
Ultra slow
Ultra slow
Results
Pyrotechnic sizing
Pyrotechnic sizing
Gel cast separation
Few axial cracks
Few axial cracks
Few axial cracks & gel coat separation
Medium axial cracks
Medium axial cracks
Coarse axial cracks
Coarse axial cracks
Helical cracks, structural
Helical cracks
Doubler unbending
Doubler unbonding
Helical failure
Coarse helical cracks
Coarse helical cracks
failure
Random cracks, cloth failures, reinforced
with 90° P STR overwinds
Fine random crazing
Coarse random cracks
Curing temperature caused
distortion
Few helical cracks
Many axial cracks (BEST PATTERN PRODUCED)
ester overwinds
-------
made in the filament helix angle, reducing the angle to increase flexing
and thus the filtration rate and finally changing to the fabrication technique
suggested by the analysis. Material changes also reflect the difficulties
experienced in obtaining optimum crazing. The two-resin system was replaced
by a single resin used for both the filament bonding and gel coat functions
to eliminate unbending of the gel coat from the winding resin. Untwisted
fibers to obtain a wider filament pattern and cloth reinforcement were used
to provide support for the gel coat after crazing.
The final fabrications were made incorporating the information developed
during the program. The crazing produced in Filter 12B possessed the fine
pores sought, although they were not randomly oriented as was previously
believed desirable. The cracks produced were numerous but the number was
below the expected level. The resulting filter required a higher pressure
to operate than was theoretically predicted for such a structure.
The knowledge gained from these fabrications provides sufficient
information for recommendations to be made for pilot demonstration models
containing the features of self-adjustment to hydraulic pressure and a
fine crazing pattern.
Fluid Dynamics
The test vehicle selected for this program was a cylinder. Selection
was based primarily on ease of fabrication of the filter elements and the
related test equipment. In this program self-cleaning was approached from
two directions. With the cylindrical test vehicle an experimental
approach was made based on the intuitive principle that cleaning will be
promoted by achieving the highest possible shear flow over the filter
surface. The second approach was the theoretical study which examined the
influence of a flow field on a particle in the field.
32
-------
To investigate self-cleaning experimentally, a smooth interior finish
on the filter element and high washing action were the two parameters on
which attention was focused. Normal fabrication techniques produced a smooth
initial surface which remained essentially smooth after being made porous.
High washing action was obtained by reducing the free area inside the filter
with a "swirl baffle." This terminology came from the original concept of
inducing a swirl flow for maximum fluid motion while moving longitudinally
through the filter area. Swirl flow was also induced by a tangential feed
at the top of the test apparatus.
Three smooth walled cone segments were fabricated to act as baffles
and were sized to provide different flow velocities across the filter
surface and also to allow acceleration or deceleration across the filter
based on the filter passing 7 gpm/ft^ (see Figure 13). Two other baffles
were also constructed. One was a spiral ramp configuration intended to provide
very high velocities along the wall (Figure 14) while the other was a rough
surfaced cylinder (constructed of deck plate) intended to cause high turbulence
(Figure 15) .
Most favorable results were obtained with the smooth walled cone which
produced the highest liquid velocity. Turbulence per se produced no noticeable
beneficial effect. All attempts at swirling the fluid over the filter by
tangential feed and/or use of the spiral baffle increased the tendency
for solids buildup on the filter of any tests run. A flow directed straight
along the wall with moderate velocity but little induced turbulence produces
the best cleaning. When this flow field was used with a filter which had
the finest porosity no fiber entrapment occurred and no solids buildup
developed on the inside surface of the filter; however, blinding was caused
by fine particle entrapment in the pores within the filter structure.
33
-------
FIGURE 13
Baffle Body-10" x 6" Truncated Cone
FIGURE 14
Baffle Body-Ramp and Cylinder
G-2143
-------
FIGURE 15
Baffle Body - Deck Plate Cylinder
G-2140
35
-------
The analytical studies which were conducted in the support of the
self-cleaning approach are discussed in Appendix A. Three different
concepts are examined in this analysis. One or more of their combined
effects can be directed to attain cleaning. These are briefly discussed
below. The discussion here is limited to physical understanding and no
attempt is made to consider the mathematics of the problem.
(1) Energy concept - If the distribution of the fluid energy is
made uniform on the entire cross-section of the filter conduit, all
suspended matter in the field including those adjacent to the filter
wall will be forced to move along in the direction of the bulk fluid
motion. Criteria for uniform distribution of energy are the uniformity
of the velocity distribution and the thickness of the boundary layer.
Thus, for the energy concept, the boundary layer development on the
filter wall is carefully examined and the layer thickness is decreased
to dimensions comparable to the particle dimensions to be filtered. As
a consequence, such particles are not "encouraged" to come to rest at
the wall for an eventual build-up of a filter cake.
(2) Particle lift concept - It can be shown that a particle trapped
in a shear flow is acted upon by two forces. One is the drag force
directed along the streamline and the other the lift force perpendicular
to it. When this shear layer is realized in the boundary layer region,
the lift force can be made to deflect the particle away from the boundary.
The particle size which penetrates the shear region and arrives at the
filter wall is a function of the fluid flow condition. Larger particles
are kept in the concentrate flow stream. Thus, for a given flow condition,
one can specify a filter which freely passes the smaller particles and
thereby maintains clean filter operating characteristics.
36
-------
(3) Particle migration concept - Again, the proper flow field is
established for concentrating the particles away from the wall forming a
ring in a pipe shear flow. This phenomenon has been observed repeatedly
in the past and was recently examined in detail. Its application to
filtration is examined here for the first time.
This analytical study (see Appendix A) has led to the following con-
clusions which are quoted: "The hydrodynamic approach to filtration and
the associated dynamic cleaning at the wall of a filter are shown to be
feasible concepts. It is shown also that optimum filtration schemes are
obtained when the Reynolds number of the flow is great enough to eatablish
a thin boundary layer over the surface of the filter. In particular, coni-
cal filter geometries and particle migration concept are found (analytical-
ly) to provide the best filtration schemes.
"Detailed methods for analysis and design of a filter are presented
with particular illustrative examples. A computer program which solves the
equation of particle motion in the boundary close to the filter surface and
for various geometries is the most direct approach to the design problem."
Waste Water Characterization
The site chosen for the feasibility testing of the filters with raw
sewage was the Bowling Green Waste Treatment Plant (BGWTP), Bowling Green, Maryland.
This community is located about 5 miles from Allegany Ballistics Laboratory. The
plant receives domestic sewage from approximately 2300 residents. No industrial
wastes are included in the treatment plant influent. During dry weather, the
average flow through the pump station is around 100,000 to 125,000 gallons
per day, but during storm flow the plant may receive in excess of 700,000
gallons during a 24-hour period, with peak flow rates above 1 MGD.
Figure 16 shows the average dry weather flow plotted with the flow created
37
-------
LEGKNP
AVERAGE DRV WEATHER FLOW
Storm of 10-25-67
Rainfall - 1.33 Inches {Influent Gate Closed
Back at 1 PM)
Storm of 10-18-67 & 10-19-67
Rainfall - 0.29 Inches and 0.03 Inches
Storm of 9-27-67 & 9-28-67
Rainfall - 0.25 Inches and 2.52 Inches
(Bypass Gate opened at 2:30 PM)
Storm of 9-20-67
Rainfall - 0.54 inches
Daily Rainfall Recorded at 7 A.M.
FIGURE 16
WASTE TREATMENT PLANT INFLUENT RATE
BOWLING GREEN, MARYLAND
DRY WEATHER AND RAINSTORM FLOW
HERCULES INCORPORATED
CONTRACT NO. 14-12-39
J. D. BANE 5/1/68
-------
by several rain storms. Since rainfall is recorded locally only as a daily
total, a 24-hour recording rain gage was transferred to this project for use
in determining hydrologic versus plant influent data.
When this sewerage system was installed, an attempt was made to eliminate
all connections except sanitary. Since the time of flow increase of the plant
influent is approximately 30 minutes after the start of a rain, it appears
that considerable runoff water is present. However, plant influent remains
above normal for an extended period after a storm and rises during a snow
melt leading to the conclusion that infiltration into the lines also is
significant.
During the contract negotiations, it was agreed that a characterization
of this plant's influent would be made during the rise from normal to storm
flow conditions. A sampling program was agreed upon and is presented in
Appendix B. There were no rain storms during the early month of the per-
formance period and then by the time the spring rains began, the influent
was being diluted by the infiltration from the saturated ground (caused by
the melting of a heavy snow) such that characterization at that time would
have been meaningless.
On June 16, 1968, a rain stormr producing 1/4 inch rainfall occurred
after a dry period of 8 days. The BGWTP influent was sampled according to
the plan. The hydrological and analytical data are presented in Figure 17.
The initial sample obtained was gray and appeared to be normal late night
flow, but after only two minutes the rate began to increase rapidly and
the sewage became black and gave off a very strong odor, indicating
septicity. This odor did not disappear until the flow again returned to
nearly normal. The results of Sample No. 2 indicate the flushing of grit
and putrescible material undergoing anaerobic digestion from the bottom
39
-------
0.8,
fTGURE 17
BOWLING GREEN WASTE TREATMENT PLANT
STORM FLOW ANALYSIS
DATE: 6/19/68
o
^0.2-
, i
Ho.i-
C
»r-t
a
r
30
Sample No.
Time (min)
Flow Rate (MGD)
Total Solids (ppm)
Total Solids-
Volatile (%)
Susp. Solids (Gdbch) (ppm)
(Glass) (ppm)
Susp. Solids-
Volatile (Gooch)(%)
(Glass)(%)
BOD (ppm)
COD (ppm)
pH
i
Normal
.25
655
41.9
274
380
86.5
86.0
204
500
7.2
60*
Time
2
0
.30
4866
55.8
4383
4678
57.4
36.2
1410
7400
6.5
(min.)
3
5
.76
1765
X
1652
1500
41.7
46.8
183
X
7.0
4
15
.66
1073
40.1
838
792
51.1
52.9
198
X
7.0
5
25
.66
1268
X
1179
1208
56.4
48.6
130
X
6.8
lie
6
45
.40
731
41.5
436
459
56.4
61.2
185
X
7.1
7
65
.32
534
42.9
190
163
64.2
79.7
102
X
7.3
155
8
120
.22
722
52
390
392
72.3
76.8
162
X
7.0
Immediate Oxygen Demand
(15 min) (ppn)
X - No analysis.
42
40
-------
of the sewers. The samples taken during the high flow conditions after
the initial flush indicate the presence of large quantities of grit.
A composite sample of the plant influent taken over a 4-hour period
during relatively normal flow was analyzed to be used in characterizing
the sewage to which the materials were exposed during their environmental
testing. The results of these analyses of raw sewage are tabulated as
follows:
Suspended Suspended Total Total
Source Solids Solids-Volatile Solids Solids-Volatile BOD COD pH
Raw Sewage 153 124 590 344 115 228 7.3
(4-hour
Composite)
All analyses reported were determined using the procedures outlined in
the Twelfth Edition of "Standard Methods for the Examination of Water and
Waste Water," with the exception of suspended solids determination. This
variation, approved by the FWPCA, involves the use of fiberglass filter
elements**, rather than the standard Gooch crucible technique.
Particle size distribution of the solids in the sewage was determined
by a Coulter Counter. This device is an electronic instrument which measures
particle size by circulating a conductive liquid containing the solid particles
through an orifice. Liquid conductivity is measured across the orifice. The
change in conductivity caused by a particle passing through the orifice is
then related to the particle size. As the liquid circulates, the counter
tabulates the size and number of particles passed. Raw data developed by
the counter is reduced by computer to give a normalized and standardized
size distribution readout which can be displayed by conventional methods.
** Whatman Glass Paper, grade GFC. "Rapid Solids Determination using glass
fiber filters" Bruce M. Wyckoff. "Water & Sewage Works" Nov. 30, 1964,
Reference Number, pg. R-349. &1
-------
Coulter Counter samples are separated into several size ranges to allow
accurate readout from an individual orifice. Normally the range of sizes for
which a single orifice is used is about one order of magnitude. Sewage samples
have utilized two ranges, 3 to 50 p, and 50 to 200 \i. The sample is filtered
to remove large size particles for the lower range. The small particles
(below the size desired for a run) do not interfere with counter operation,
and data reduction analytically ignores these sub-range size particles.
Table 5 illustrates the type of computer output obtained from a test
run. These dat;a are then plotted by a standard "Weight Percent Greater Than
vs. Size" method, as shown in Figures 18 through 21.
Filter Feasibility Testing
Two locations were prepared for use in the feasibility testing of the
filters. The first test site, located at Allegany Ballistics Laboratory
(ABL), is used for initial testing with water and simulated sewage (cellulose
acetate particles) to determine filtration rates, produce crazing, and measure
effects of swirl baffle body designs. The second test site is the municipal
waste treatment plant at Bowling Green, Maryland (BGWTP). This plant is
described in the Wastewater Characterization section of this report. A
schematic layout of the BGWTP testing facility is shown in Drawing 60225N10001.
The raw sewage for the feasibility testing was removed from the force main by
tapping the blind flange used to cap the connection for a future pump
installation. This tap is shown in the upper right hand side of Figure 22.
The sewage is piped to a screener housing a removable basket having 1/4-inch
openings. After passing through the test stand, the filtered and concentrated
flows are fed into a common flume and pumped back to the plant wet well.
The test equipment utilized 30° V-notch weir boxes for measurement of the
concentrate and filtrate flows (see Figure 23). The test site at the BGWTP
was designed and equipped to handle two filter test stands simultaneously.
42
-------
TABLE 5
COULTER COUNTER DATA
(Test #1443-10)
Se«|uence
1
2
3
4
5
6
7
1
2
3
4
5
6
7
8
9
10
Lower Threshold
Particle Diameter
(Microns)
146.36
116.17
92.20
73.18
58.08
46.10
36.59
29.07
23.08
18.32
14.54
11.54
9.16
7.27
5.77
4.58
3.63
Interval Weight
0.0
0.1992
0.1992
0.3984
0.9959
5.5771
19.5197
0.0978
0.7819
3.7142
6.3532
8.9922
9.3832
9.5787
8.4058
10.1651
15.6386
Total Particles Counted
Mean Diameter (Microns)
Specific Surface Area
Weight Percent
Greater Than
(*)
Midpoint
Particle Diameter
(Microns)
0.0
0.199
0.398
0.797
1.793
7.370
26.889
26.987
27.769
31.483
37.836
46.829
56.212
65.790
74.196
84.361
100.000
1
-
_
167.542
132.978
105.545
83.771
66.489
52.772
41.885
33.282
26.416
20.966
16.641
13.208
10.483
8.320
6.604
5.242
4.160
32,845,744
5.14
0.65355 m2/g
Interval Frequency
(7.)
0.0
0.0000
0.0000
0.0002
0.0010
0.0109
0.0764
0.0008
0.0122
0.1159
0.3964
1.1222
2.3419
4.7815
8.3920
20.2968
62.4518
Cumulative
Frequency
a)
0.0
0.0000
0.0001
0.0003
0.0012
0.0122
0.0885
0.0893
0.5015
0.2174
0.6138
7360
0779
8.8594
17.2513
37.5482
100.0000
1.
4.
-------
u»
-------
m ** so to
Particle Size Distribution
Filtrate Solids
Test #1443-9, Sample F-2
X >tTest #1443-9, Sample F-30
HERCULES INCORPORATED
CONTRACT #14-12-39
J. D. BANE
-------
Particle Size Distribution
Filtrate Solids
Test #1443-10, Sample F-30
HERCULES INCORPORATED
CONTRACT #14-12-39
J. D. BANE 3-8-68
-------
APPENDIX A
Hydrodynamics of Sewage Filtration
-------
SUMMARY
A new concept based on the hydrodynamic approach to filtration is intro-
duced and detailed design methods are outlined based on the analysis of the
flow in the boundary layer over the filter surface and the associated forces
which produce "dynamic cleaning." The feasibility of the concept is analyti-
cally demonstrated through the use of a simple but reasonable model.
CONCLUSIONS AND RECOMMENDATIONS
The hydrodynamic approach to filtration and the associated dynamic
cleaning at the wall of a filter are shown to be feasible concepts. It
is shown also that optimum filtration schemes are obtained when the
Reynolds number of the flow is great enough to establish a thin boundary
layer over the surface of the filter. In particular, conical filter
geometries and particle migration concept are found to provide the best
filtration schemes.
Detailed methods for analysis and design of a filter are presented
with particular illustrative examples. A computer program which solves
the equation of particle motion in the boundary close to the filter surface
and for various geometries is the most direct approach to the design problem
and, hence, is recommended. It is also recommended that the analytical
study be supported by laboratory experiments to test the concepts and
project these to the design of a full scale sewage plant.
A-l
-------
INTRODUCTION
Inasmuch as the separation or collection of suspended particles
appears to have been a process always known to man, filtration is often
thought of as an extremely simple operation. In reality, it is not; and
the practical difficulties in achieving proper filtration are numerous.
Generally, the source of these difficulties is the "plugging" or "binding"
of the filter medium in such a way that progressively greater pressures
are needed for continued filtration. The disadvantage in high pressure
driving forces lies not only in a more complex and expensive system but
also in that the suspended matter is often deformable and under such
pressures is compressed against the filter. The resulting closely packed
cake could eventually stop the flow due to excess resistance. In order
to get an idea of the magnitude of the driving force required to achieve
filtration across a porous medium, the following mathematical model is
often used. The formula was first obtained by Poiseuille for the laminar
flow of liquids under pressure through capillary tubes. While not
devised with filtration in mind, it nevertheless is widely used for flow
through sand and various porous media. It is one from which many filtration
equations have been derived.
A simplified form of Poiseuille's equation for smooth flow through
capillary tubes is
4
Q = -2LE2-
A-2
-------
where p is the pressure difference at tube ends, a is the internal capil-
lary radius, J- length of the opening, Q the quantity of filtrate, and ^
the dynamic viscosity of the fluid. Of particular interest here is the
strong dependence of the flow Q on the capillary radius. For example, by
reducing to one-eighth of its size, either the pressure must be increased
four thousand times to maintain the same flow or four thousand times re-
duction in the flow would result for the same pressure. From the point of
view of filtration, the large increases in the pressure are undesirable
and a substantial drop in the flow is considered a virtual "plugging".
A great variety of filtration methods are used in the chemical in-
dustry for the purpose of separating solid material from liquid or slurries.
The ordinary plate-and-frame filter press working on a batch system is
perhaps the most common type of filtration equipment. Rotary filters of
various specialized designs, however, are used in large scale continuous
processes. The essential mechanism in either case is to pump the liquid
through a porous filter cloth so that the solid matter adheres to the
cloth and builds up to form a "cake". The resistance through the cake is
then calculated through some modification of (1) and is dependent on com-
pressibility of the material forming the cake. If the cake is incompressible;
i.e., the porosity is uniform, simple algebraic relations are obtainable
for the "specific resistance" of the cake per unit thickness. Otherwise
an average specific resistance is obtained from the integration with re-
spect to "solid pressure" distribution in the cake layers.
A-3
-------
The interdependence of the characteristics of the filter medium
(i.e., the cake) and the slurry remains the key difficulty in the theoreti-
cal approach to the problem of filtration. This also explains why filtration
has been developed as a practical art rather than a science. The theories
have been seldom used in the actual design of a filter for a given opera-
tion. Instead they provided means for interpreting laboratory tests in
seeking the optimum conditions for filtration and in predicting effects of
changes in operating conditions. When examined from this point of view, it
is natural, therefore, to consider the advantages that filtration would
offer as it eliminates the dependence on the "cake" as a filter medium.
Nevertheless a filter medium with a fixed characteristic requires continuous
removal of the deposits from its surface. It is immediately clear that this
approach does not eliminate the conceptual or the practical difficulties
but shifts them to a new domain in technology. We are now concerned with
the hydrodynamics of the particle fluid motion and seek to direct the natural
forces in the flow field to promote "cleaning" at the porous wall. With this
approach, the attention is focused less on the flow in the porous medium it-
self and more on how to achieve through changes in geometry and flow pattern
the most optimum "cleaning" action. The flow conditions so determined and
the knowledge of the particle size, distribution, and consistency would in
turn provide information on the proper design of the filter medium itself.
It is important to note that the coupling action between the characteristics
of the filter medium and the flow field is thereby reduced to a conceptual
difficulty and not, as before, depending on the multitude of the unknown fac-
tors which make the "cake". In what follows, we shall develop these concepts
A-4
-------
and examine numerous possibilities which appear pertinent in the way
of achieving "self-cleaning" or even better, "dynamic cleaning."
ANALYTICAL CONSIDERATIONS
The development and formation of a hydrodynamic boundary layer over
a wall, be it porous or impermeable, influences in a significant way the
degree of "dynamic cleaning" that would otherwise be attained under the
influence of the full momentum of the free stream. For this reason it is
fruitful to examine the character of motion in the boundary layer and
to review some of its features in various geometries and under laminar
as well as turbulent free stream flows. The velocity Distribution over a
flat plate was first obtained by Blasius^ and is shown plotted in Figure 1
as a function of T\. From this, one finds that the boundary layer thickness
6 is related to the local Reynolds number at a distance x from the edge of
the plate by
* 5x
IF (2)
VKex
where R = (p u^ x)/(j,. For instance, at a distance x = 10 cm from the
leading edge, 6 is of the order of one half of a millimeter. For the flow over
a porous wall there is no similar general solution. Thus, under the assumption
that asymptolically du =0, it is obtained
ox
i 1 - el (3)
Uco
where v is the constant discharge velocity. The practical restriction on (3) is -the
o
requirement of small VQ, namely 0.0001 < v,,/''.. < 0.01. From (3) one can readily
obtain the expression for displacement or momentum thickness; i.e.,
A-5
-------
6* - o (4)
6 * &.
The velocity distribution is seen plotted in Figure 2, Curve I. Curve II,
drawn for the purpose of comparison, represents the boundary layer without
suction. It should be noted that the suction profile is fuller in this case.
Other solutions of the boundary layer are known for flow patterns which
can be associated with similar velocity profiles. Examples of such solutions
are obtained over a flat plate with a suction or blowing velocity v proportional
to 1/Vx and in a convergent channel with vo proportional to 1/x. The solutions
of the velocity profile over a flat plate are shown plotted in Figure 3a and
those in a convergent channel in Figure 3b. In these figures, the positive
values refer to suction and negative to blowing at the boundary. The decrease
in the boundary layer thickness with suction and increase with blowing should
be noted. The phenomenon implies a greater sensitivity of the boundary layer
thickness to blowing than to suction. With the application in mind of these
findings to filtration, one can expect "flashing" or backwashing of the filter
wall through a process of blowing. In Figure 3c the profiles in a convergent
channel and over a flat plate are shown compared for v = o and the former is
found to be much fuller.
It is known ' that the development of the boundary layer in the entrance
region of a channel or a pipe flow at high Reynolds number is identical to
that over a flat plate. Different problems, however, evolve when the Reynolds
number is not large.
A-6
-------
This problem has been treated^ as a perturbation to Poiseuille's flow and
solutions were obtained for very small discharge (suction). It is found that
the velocity profile in the major flow direction deviates from the Poiseuille
parabola by being flatter at the center of the channel and steeper in the
region close to the walls, the degree of deviation depending on the Reynolds
number for the flow through the channel walls (i.e., v h/v). Solutions exist
for a straight channel with impermeable walls for Reynolds numbers up to
500. A surprising result of this analysis is that the velocity profile for
small entrance distances is not convex, see Figure 4. It possesses a local
minimum on the axis y = 0 and symmetrically situated maxima on either side of
it.
The velocity distribution in the entrance region of a pipe may be
affected to a considerable degree if all or part of the wall is porous.
When all the wall is porous and there is a total absorption, a stream line
pattern such as shown in Figure 5 is obtained. A careful study of the
/Q\
latter case has been made recently. The streamlines for a total absorption
from this analysis are shown in Figure 6a. The absorbing wall is located
along the entrance length equal to 207o of the pipe radius. We note the bounding
streamline which meets at the downstream edge of the absorbing wall,runs along
the axis of the pipe to a stagnation point beyond this edge,and from there fans
out as a surface of revolution towards the walls. Beyond this surface are
circulations which produce a return flow along the axis.
More details about the velocity distribution are given in reference 8.
The results for a Reynolds number of 40 are shown plotted in Figure 6b. It
A-7
-------
It has been shown before that in the absence of absorption the initial
velocity profiles are not convex but have a local minimum on the axis and
symmetrically placed maxima on either side. However, we see here that the
effect of absorption is to draw out the velocity profile and so straighten
off the concave region.
Turbulent Boundary Layer
Transition into turbulent boundary layer may occur if the Reynolds number
R exceeds a critical value. The critical Reynolds number over a flat plate
ex
is found to be between 3 x 10 to 4 x 10 and is somewhat smaller if the free
stream is also turbulent. A similar trend is also expected with a rough sur-
face. The presence of a pressure gradient in the external flow has a strong
effect on the boundary layer transition. Accelerated flows (dp/dx < 0,
du/d > 0) favorable pressure gradient) are considerably more stable than
X
decelerated flows (d /d > 0 du/d < 0, adverse pressure gradient). When there
is a suction at the boundary, it is shown that critical Reynolds number
increases over one hundred times its value with no suction.
In a turbulent flow past a smooth boundary we may distinguish among three
regions^: (1) adjacent to the wall, an extremely thin region where the flow
is predominantly viscous (this is called the laminar sublayer and its thickness
is designated by 6/ ), (2) an intermediate buffer layer where the flow is tur-
bulent but directly influenced by viscosity, and (3) the rest of the flow region
where the direct viscous effects are negligibly small.
It is interesting to note that even with a perfectly smooth wall the
thickness of the viscous sublayer is of the order of .001 to .016 and that the
A-8
-------
mean velocity in this region increases linearly with the distance from the
wall. For the velocity distribution in the buffer zone, a simplified picture
of the flow is obtained in which the velocity distribution is often expressed
by a power law. Similar approximations are obtained for the velocity distribu-
tion in a pipe as seen in Figure 7. In the same figure is also seen plotted
the velocity profile in a fully developed laminar pipe flow. From the com-
parison of the two curves one obtains an idea of the extent of the momentum
penetration into the boundary layer and close to the wall of the pipe. In
the laminar flow, the entire flow field is a "boundary layer" whereas in a
turbulent flow the boundary layer is thin indeed. The two dotted curves
plotted in this figure are obtained from the boundary layer analysis at the
extreme region to the pipe, x/2a designating the proper location. The layer
thickness gradually increases along the pipe until it approaches the fully
developed profile at large distances downstream.
Experimental measurements of the boundary layer thickness over the walls
of a converging channel yield values much smaller than that over a flat plate.
The velocity profile obtained from such measurements is shown plotted in
Figure 8. In that figure u/u is plotted against the distance from the center-
line for the convergence angles shown and may be directly compared with that in
a turbulent pipe flow. The latter is designated by a zero convergence angle.
Again, it is seen that a considerable reduction in the boundary layer thickness
is obtained through a simple change in the flow geometry. This result has a
direct bearing on the dynamic cleaning in filtration.
A-9
-------
Completely new problems present themselves when the motion of the sus-
pended matter is also considered. This problem will be taken up next.
Dynamics of a Two-Phase Flow
Ideally, one should discuss at this point such related topics as the
size, distribution, concentration, and consistency of the pollutants; i.e.,
the "particles" in a sewage. However, the detailed consideration of such
matters will take us right outside the scope of the present work. It suffices
to note that the total solid content of the sewage is of the order of .1%,
sizes range from one to several hundred microns and the consistency of the
suspension is mostly organic matter. There are found sand particles, hair,
lint, oil and a multitude of other varieties of suspended matter in a sewage
effluent. Naturally an exact analytical treatment of a problem of this com-
plexity is not practical. An order of magnitude analysis of the problem
relevant to "dynamic cleaning" may be attempted, however, without too much
complication if we assume that the suspensions have regular geometric forms,
such as small rods or spheres. A sphere moving through a very viscous liquid
(10)
with velocity, w, relative to a uniform simple shear experiencesa lift force
L ~ 8U2M- wa k/v (6)
perpendicular to the flow direction. Here, a denotes the radius of the sphere,
and k the magnitude of the velocity gradient. Under restrictive conditions,
the above result was applied to the motion of a sphere in a Poiseuille flow
through a pipe simply by calculating the relevant velocity gradient and
substituting for k in (6). It was shown in reference (10) that the direction
A-10
-------
of the sidewise force is toward the center of the pipe when the particle
lags the fluid, and away from the center and towards the wall when the
particle leads the motion of the fluid. Stated in an equivalent way, the
down motion of a heavy particle causes it to drift toward the wall but the
same motion of a neutral or light particle causes it to migrate toward the
center. Since we are interested mainly in the region of the flow close to
the wall, the use of (6) requires some justification. It turns out that for a
very small particle, one can make the Reynolds number based on the velocity
gradient R = ka^/v, very small compared to unity. In a boundary layer
1/2
region this value is reduced even further if (a/x) is also small. In
other words, despite the presence of the wall, very small particles do not
feel its presence and (6) may still be applicable. With due regard to stated
technical difficulties we venture to employ (6) to examine the magnitude of
the side force developed at a point not too far away from the wall in a
Poiseuille flow and compare it with that obtained in a boundary layer over a
surface with absorption. Such comparison yields a lift force (25u) times
greater than that in a Poiseuille flow. It may be said, therefore, that the
effect of boundary layer is to promote cleaning in two ways - one due to
deeper penetration of the flow momentum close to the wall, and the other -
development of a useful side force on a particle.
The presence of the above sidewise force on a solid as well as deformable
particles have been experimentally confirmed by many investigators. The
apparatus used in most of these experiments was limited to capillary tubes in
the order of 1 cm and particles were close to neutral density having dimensions
A-ll
-------
in the order of several hundred microns. The first to show this phenomenon
was the experiments of Segre and Silberberg who demonstrated that a rigid
sphere transported along in Poiseuille's flow through a tube is subject to
radial forces which tend to carry it to a certain equilibrium position at
about 0.6 tube radii from the axis irrespective of the radial position at
which the sphere first entered the tube. A typical particle count distribution
and density histogram are shown in Figure 9. The reader is referred to
reference(ll)for the details of experimentation and data reduction. Experi-
(12)
ments on deformable liquid drops and tiny threads (which undergo shear
deformation) show that these also migrate toward the tube axis. Later
(13)
experiments showed that dense particles falling slowly through an upward
moving fluid migrate to the tube axis; buoyant particles in the same flow
migrate to the tube walls. More experimental data are presented in reference
(14) wherein the effects of initial particle concentration and entry length
on the migration of neutral density spheres from the wall are carefully
reported. It is noted in that reference that the region outside the con-
centration peak always consists of a particle-free layer of suspended fluid
indicating that the particles making up the concentration peak probably come
from this region. The height of the wall peak is shown to increase with the
distance from the entrance to the tube according to the plot seen in Figure 10.
The ordinate in Figure 11 is the plot of the relative peak (NR) to trough
(Nn) . particle counts. At the same tube location, the wall peak is shown
in Figure 11 to decrease rapidly with small concentration and then level off.
From these results it would appear that, while the initial creation of the peak
A-12
-------
takes place somewhere in the region of the mouth of the tube, it continues to
develop during the passage down the flow tube. This gradual increase in the
height of the peak is accompanied by a narrowing of the peak which is caused
by the passage of the particle into this region from faster moving streamlines.
(12)
In earlier experiments measurements at 350 pipe diameter downstream of the
inlet detected no particle migration, however. It is furthermore demonstrated
in reference (14) that the wall peak is a function of the initial concentration.
The possible reason for this is a particle-particle interation at high con-
centration.
The Criterion for Design of a Filter
In the foregoing discussion we have considered filtration from the hydro-
dynamic point of view in the sense that the characteristics of the filter
medium may be uncoupled from those of the suspension provided caking is pre-
vented at the filter wall. We also devoted particular attention to the
character of the flow in the vicinity of the wall under a variety of flow
conditions. We found that favorable flow conditions could be developed at the
wall in both turbulent and laminar flow. Under these conditions the momentum
of the fluid is made to penetrate deeper into the boundary layer and close to
the wall. This way the viscous flow is limited to an extremely thin region
next to the wall and the predominant portion of the flow field is thus imparted
with inertial forces transporting and imparting energy to the suspended matter
in the field. It was pointed out very clearly that the thin boundary layer
not only limits the portion of the slow moving fluid particles in the field to
a minimum but it also sets up a strong velocity gradient at the wall which acts
to deflect the approaching particles from the boundary. It is the purpose of
this present section to develop a better understanding of this mechanism and
A-13
-------
perhaps a confidence that it does indeed apply to filtration in a practical
sense. In the way of developing the ideas, we arrive at the criteria which
are so germain to the actual design of a filter. Naturally, our results will
be severely limited for two reasons: (l)to keep clarity of presentation we
examine only the most simple model, and (2) the full description of the flow
field, even if possible, is not available at the present time. Nevertheless,
we shall demonstrate for the selected model the feasibility of the concept as
a mathematical tool to arrive at a meaningful design of filters.
The schematic of a model to be considered is shown in Figure 12a.
Mathematically, the flow may be that over a flat plate and physically the
plate may be thought of as a part of the entrance to a channel. We assume
that the Reynolds number of the flow is great enough to allow development of
the boundary layer (Rex = 2 x 10 , say). We furthermore assume that there is
a uniform discharge through the filter wall with an effective constant
velocity v . To keep the analysis simple, we consider only the assymptotic
flow condition represented by the simplified equation (3). Unfortunately, the
consequence of this choice is a small v0, that is small in terms of filtration
requirements. We shall return to the discussion of this matter later. Mean-
while, let us emphasize that while different problems arise when there is a
total absorption at the wall, the effect of the thin layer and the associated
particle migration remain valid. Only the physical arrangement of the filtra-
tion system may undergo a suitable change with the requirements for dynamic
cleaning. Returning now to Figure 12a, we identify the following: the edge of
the boundary layer, the velocity vector tangent to the streamline at a region
A-14
-------
just outside the boundary layer and the longitudinal velocity profile inside
the layer. We note that the latter drops from its maximum value at the free
stream to zero at the wall, whereas the relative magnitude of VQ with respect
to u progressively increases. The predominance of the velocity component
towards the wall over that along it would direct the resultant velocity vector
more and more towards the wall until it eventually becomes perpendicular to it.
Meanwhile, there are two types of forces acting on a particle which is trapped
in this region - one is a drag force directed along the resultant velocity
vector whose magnitude is given by 6 rr (j, w a and the other is a lift force
perpendicular to the first and its magnitude is estimated by (6)
Now, consider again the point just outside the boundary layer in Figure 12a.
The corresponding force diagram acting on a particle at this point is sketched
in Figure 12b. Thus the underlying thought in the proper design is to arrive
at a resultant force on the particle acting away from or at most parallel to
the wall before the particle is too close to the wall. It turns out that for
a given flow condition progressively smaller particles approach the filter
wall as they penetrate the boundary layer. The larger particles are being
deflected away due to the strong lift producing forces and relatively small
drag. From the geometry of the streamlines, it is seen that whenever the drag
and lift forces are equal, the resultant force is parallel to the wall. The
maximum particle sizes for which this condition prevails at the wall determine
the criterion for the design of the filter. Clearly, the particles below this
size must be allowed to pass through the filter without appreciably changing the filter
resistance, otherwise a cake may form on the filter wall. Now, let us return to
A-15
-------
the mechanics of determining this particle size for the problem under con-
sideration.
The ratio of lift to drag forces is obtained from dividing equation (6)
by the drag 6 TT p, aw; i.e.,
(7)
The expression for the velocity gradient is available from a direct
differentiation of (3), namely,
u -y/6*
du
f
Substitutions of (8) into (7) and some algebra yields the basic design
equation -y/26*
L ~ 200 1- e (9\
D 6* **'
The*« is a particular equation applicable only to the problem under con-
sideration. In its derivation we assumed as stated earlier that Rex = 2 x 105 .
1/2
Also, we made use of the approximate relations Res = 5.5(R6x) = Re 6*
.34
which were obtained from (12).
Equation (9) states that the lift to drag ratio for a particle which is
(d/6*) times smaller than the displacement thickness varies exponentially with
the distance y from the wall, reaching its maximum value at the wall and vanishes
outside the boundary layer. A significant restriction on (9) is for (d/6*) to
be much smaller than unity. This is the condition for which (7) is most likely
A-16
-------
to apply. Consider now the following tabulated data obtained from evaluating
(9) for the three (d/6*) shown in Table II.
TABLE II
LIFT-TO-DRAG RATIO
y/6
0
.5
1.
1.5
2
3
4
The values on the top right hand side of this table are indicative of the
conditions wherein most likely dynamic cleaning would prevail. The column
for the particle size 1/200 shows that such particles reach the wall but
they probably slide over it. The particles of the size 1/100 are deflected
at a distance as close as 6* units from the wall. Once they are past this
distance they probably contribute to build up of a cake unless they are
absorbed by the filter.
The presentation in this section shows that under certain flow conditions
it would become possible to direct the forces in the fluid field to attain
dynamic cleaning. We purposely avoided complications in the attempt to
L/D
d/6*=l/200 1
1.00
.7788
.6065
.4724
.3679
.2231
.1353
d/6*=l/100 1 d/6*=l/20 ]
2.000 10.00
1.5576 7.788
1.2130
.9448
.7358
.4462
.2706
6.065
4.724
3.679
2.231
1.353
A-17
-------
demonstrate conceptual feasibility. Whether or not such conditions are reasonably
obtained under actual circumstances requires careful experimentation.
We have already examined the detrimental effect of the total
absorption in Figure 6. In this example only the upper part of the pipe
has a porous wall. The pattern of the streamline shows a large stagnation
region and some inverse flowing streamlines. The latter would bring to the
porous wall some suspensions from the central region of the pipe and accumulate
them at the wall. Thus for a good cleaning action the pattern of the stream
lines should be as straight as possible along the channel but undergo a
rapid curvature at a distance as close to the wall as possible. This pattern
of streamline cannot be obtained with total absorption. It is necessary that
a portion of the fluid leaves the pipe to avoid the build up of a stagnation
region. The presence of this region causes the reverse flow of the streamlines
which is so detrimental to the dynamic cleaning. The change in the geometry
of the channel would contribute to decreasing some of these effects in a total
absorption. A combined advantage of a suitable geometry and flow Reynolds
number would probably make a close to total absorption possible.
A quick review of Figures 1 through 8 brings us rapidly to the point we
wish to make with respect to the choice of flow geometry. Throughout this review
we must keep in mind: (1) energy is expended to move the fluid through conduit,
and (2) an efficient transport of the suspended matter requires a uniform
distribution of this energy across the conduit. This implies that viscous
dissipation of energy must be limited to as thin a region as possible at the
wall. Now, it is obvious that a parabolic distribution such as obtained in a
A-18
-------
fully developed pipe or channel flow (Figure 5) is the least effective scheme
from the above point of view. On the other hand, the fully developed tur-
bulent flow in a pipe (Figure 7) shows a substantial improvement over the
laminar flow. Note the penetration of the high velocities close to the
wall and into the boundary layer. The velocity profile in a convergent
channel or in a cone, be it for a laminar or turbulent flow, is more uni-
form than the corresponding flow in a straight channel or a circular pipe.
Naturally, the turbulent flow in a channel (Figure 8) has the most suitable
velocity profile. Note, however, that to obtain a turbulent flow, a penalty
has to be paid in terms of excess energy. For this reason, we look into the
boundary layer formation, first over a flat plate (Figure 1) and then in the
entrance region to a channel (Figure 4) and we note a boundary layer is almost
established at this region at considerably lower Reynolds numbers.
In a second proposed filtration scheme wherein energy is imparted to
the fluid but in a much lower level, one can make use of the particle migra-
tion phenomenon (Figure 9} . Since the particles are naturally concentrated
in a ring away from the wall, one can profitably consider a filtration system
suggested by the sketch in Figure 13. There are shown two concentric pipes.
The inner pipe is impermeable throughout its length. The outer pipe is made
impermeable only along a suitable length at its lower end. Only this part
envelops the inner tube. The major filtration process occurs in the annulus
and through the wall of the outer pipe. The flow is allowed to develop in
the outer pipe to a point where the greatest concentration distribution of
the suspended matter is expected. At a point along this pipe where the
A-19
-------
"necklace" effect is best developed, the segment of the porous wall begins.
Close to this point the inner tube should also begin. The diameter of the
inner tube should be so chosen that only the most dilute suspension is allowed
to pass into the annulus region. Provision should be made not to disturb the
distribution of the suspended matter. Presumably, then the most concentrated
matter follows into the inner pipe. This fluid may be collected and recir-
culated to obtain a further increase in the concentration using the same loop
or equivalent to it. The excess deposit in the annulus may be periodically
flushed out and recirculated. The necklace phenomenon is not reported in
very large pipes, however, In practice, there is a certain trade-off of
various parameters. A fully developed flow establishes itself after a long
distance (over 50 diameters) downstream of the entrance region. From this
point of view, there is an advantage in using smaller diameter pipes as
physical pipe length will be correspondingly shorter and filter units would
be easier to handle. For example, replaceable filter modules could be made
which are occasionally taken out and inserted in a fresh water cleaning cir-
cuit. In principle, the scheme suggested here operates on the concept of total
absorption which takes place in the annulus. The difficulties are reduced in
here because of the diluted concentration and allowance for some particle
deposition which is flushed out in a later cleaning process.
In an earlier discussion we have made a reference to the process of
blowing fluid into the boundary layer through the wall. It was shown
(Figure 3) the thickness of the layer increases appreciably with a relatively
small amount of blowing. The thickening of the boundary layer could be used
A-20
-------
with advantage in backwashing the walls of a filter. By means of an inter-
mittent blowing and rapid circulation of a clean fluid in the pipe, one can
induce a corresponding oscillation in the fluid near the wall and the
eventual removing of a build-up of deposits.
A-21
-------
LIST OF SYMBOLS
Q - quantity of flow
a - radius of a pipe or sphere
P - pressure
p, - dynamic viscosity
u - fluid velocity
p - dens ity
x - coordinate (usually longitudinal or the distance from the
leading edge of a flat plate and/or inlet of a pipe)
t - time
y - coordinate in the transverse direction
Ujjj - maximum velocity
u - initial velocity
UTO - free stream velocity
v - kinematic viscosity
6 - boundary layer thickness
6* - displacement thickness
6 - momentum thickness
6-j - laminar sublayer thickness
Re - Reynolds number depending on u, um, u or u0 or vo
v0 - suction velocity at the wall
"u - average velocity
v - transverse velocity
A-22
-------
X - y/h
h,b - channel half-width
w - relative velocity of a particle
D,F - drag force
L - lift force
k - velocity gradient
A-23
-------
a
1C
0.8
0.6
0*
02
um
J
/-Theory
" Btosius
':
10
X
3.0
V*x
~T~
. o ifcxlO5
7,-y
l/^
p«"
7.0
Figure 1. Velocity Profile in the Laminar Boundary
Layer Over a Flat Plate
_
(Jm
1.0-
0.8-
O.b
(/.*
0.2 >
0
<
f''
C
/
V
0
^
*"
,;
^ -
8 1.6 2* 12 *
Figure 2. Velocity Distribution in the Boundary Layer on a Flat Plate
at Zero Incidence
I. Uniform suction; "asymptotic suction profile"
II. No suction; "Blasius profile"
A-24
-------
KX5
20 2-5 3-0 3-5 AO
Figure 3a. Velocity Profiles in the Boundary Layer
on a Flat Plate with a Suction Velocity
Distribution vo~l//(x). (Reference 4)
KXD
0-75
^050
025
05 K) 1-5 ?0
Figure 3b .
Velocity Profiles in the Boundary Layer
of a Convergent Channel (Sink Flow) with
a Suction or Blowing Velocity vo ~ 1/x.
(Reference 4)
A-25
-------
Figure 3c. Comparison of Velocity Profile
for a Flat Plate (I) with That
in a Convergent Channel (II)
A-26
-------
Velocity profiles for R=20
x=0
Figure 40 Development of Velocity Profile in a Channel (Figure lOb)
-------
50
100
150 200
c/h
Figure 5. Streamline Pattern in a Total Absorption
250
A-28
-------
co
Figure 6a. Streamlines in the Case of Total
Absorption Rg = 0
x/a
Figure 6b. Velocity Profile at the Extreme Region
of a Pipe, Rg = 40, ^o =
A-29
-------
1.
.8
.6
6
.4
.2
entrance region
(boundary layer)
.2
.4
r/a
.6
.8
Figure 7. Comparative Study of the Fully Developed Velocity
Profile in a Pipe with Boundary Layer in the
Entrance Region
A-30
-------
1.
.8
.6
e
4
3
.2
Figure 8. Turbulent Profile in a Convergent Channel
A-31
-------
3. 14
.120
>
I
*>
s. is
2 3
r (mm)
Figure 9. Derived hit coincidence count 1% and hit density f for series S.14
and S.16; dependence on radial position r. (For data in case S.14,
see Table I. S16 has been performed under the same conditions but
with smaller Ax.) Standard errors in N^ are indicated. Full lines
reconstructed from f.
-------
i
80
H (cm)
Figure 10. Variation of Height of the Wall Peak Along
the Length of the Flow Tube at C = 2.90
particle/cm3, Re = 540
±
0 20 40
C0, particle/cm-^
Figure 11. Variation of Height of the Wall Peak
with Concentration
A-33
-------
external flow
filters _ .
\ I I I » » » I I
velogity- vector
streamline
velocity
profile
i T f I. I i I I I I
I I
Figure 12a. Boundary Layer Velocity and Streamline Diagram
filter
II i i iini i ii
streamline
.1 i i i i i i i i i i i i rrr
Figure 12b. Force Diagram
A-34
-------
centerline of the pipe
W
Ui
region of high particle
concentration
out-flow
^ -
region of low
nflT*t"f P!P -
concentration
wall of the outer pipe
1 t
"\
\
-^ "N 1
^\ I !
-
1 i 7 i
filtrate
\
> i
fs^i
l^xl Out" flow
Figure 13. Filtration Using Particle Migration Phenomenon
-------
REFERENCES
(1) Schlichting, H., Boundary Layer Theory, Fourth Edition, McGraw-Hill,
1960.
(2) Carslaw, H. S., and Jaeger, J. C., Conduction of Heat in Solids,
2nd Edition, Oxford Press 1959 (page 329).
(3) Schneider, P. J., Temperature Response Charts. John Wiley and Sons,
Inc., 1963 (page 33).
(4) Lachtnan, G. V., Editor, Boundary Layer and Flow Control, Volume II,
Pergamon Press, 1961.
(5) Berman, A. S,, "Laminar Flow in Channels with Porous Walls," Journal
of Applied Physics, Volume 24, #7, page 1232, 1953.
(6) Gillis, J. and Brandt, A., "Magnetohydrodynamic Flow in the Inlet
Region of a Straight Channel," The Physics of Fluids, Volume 9, #4,
page 690, 1966.
(7) Weissberg, H. L., "Laminar Flow in the Entrance Region of a Porous
Pipe," Physics of Fluids, Volume 2, #5, page 519, 1959.
(8) Friedman, M., and Gillis, J., "Viscous Flow in a Pipe With Absorbing
Walls," Journal of Applied Mechanics, Transaction of ASME, December
1967, page 819.
(9) Hinze, J. 0., Turbulence, McGraw-Hill, 1959.
(10) Saffman, P. G., "The Lift on a Small Sphere in a Slow Shear Flow,"
Journal of Fluid Mechanics, Volume 22, pages 385-400, 1965.
(11) Segre, G. and Silberberg, A., "Behavior of Macroscopic Rigid Spheres
in Poiseuille Flow," Part 1 and 2, Journal of Fluid Mechanics, Vol. 14,
pages 115-157, 1962.
(12) Goldsmith, H. L. and Mason, S. G., "Axial Migration of Particles in
Poiseuille Flow," Nature, June 17, 1961, No. 4781, page 1095.
(13) Jeffrey, R. C., and Pearson, J. R. A., "Particle Motion in Vertical
Laminar Tube Flow," Journal of Fluid Mechanics, Volume 22, Part 4,
pages 721-735, 1965.
(14) Maude, A. D. and Yearn, J. A., "Particle Migrations in Suspension Flows,"
Journal of Fluid Mechanics, Volume 30, Part 3, pages 601-621, 1967.
A-36
-------
APPENDIX B
SEWER FILTER CHARACTERIZATION PROGRAM
Waste Quality Testing
- Storm Flow Conditions -
Raw'2' Strainer Concentrated Strainer
Determination WaL>.e Effluent Stream Solids
1. pH d d d
2. Total Solids d-c d-c d c
3. Total Solids - Volatile d-c d-c d c
4. Suspended Solids(1) d d d c
5. Suspended Solids d d d
Volatile^1)
6. Biochemical Oxygen d d d
Demand
7. Chemical Oxygen Demand d-c d-c d
8. Particle Size Distri- d-c d-c d
bution
d = discrete samples
c = composite samples
Sampling Frequency - From the study of previous storm flow records at the
Bowling Green Waste Treatment Plant, sampling fre-
quency will be as follows: (1) discrete samples - 0,
5, 15, 25, 45, 75, 120 minutes and then hourly there-
after while flow remains significantly above normal
and (2) composite samples - 3 samples taken from first
30 minutes, 3 from next 90 minutes, and 3 from the
remaining duration.
The quality of the concentrated stream from the filter will be
determined by difference between raw waste influent and strainer (filter)
effluent with spot checks of the determinations indicated.
At the completion of each test run, the filter will be removed
and the deposited and entrapped material analyzed as shown.
(1) Suspended solids determinations will be made using fiberglass filter
elements rather than the standard Gooch crucible technique, with
occasional spot checks of both methods for comparison.
(2) Samples for quality analysis of the raw waste will be taken just
before the flow enters the filter test stand, which is at the plant
pump discharge after the waste passes through a screener sized to
remove materials larger than 1/4" diameter. This location was chosen
to provide a more accurate determination of the filter efficiency
while still yielding a good characterization of the plant influent.
-------
CO (0 O
Figure 21
Particle Size Distribution
Filtrate Solids
Test #1443-14 Composite Sample
HERCULES INCORPORATED
CONTRACT #14-12-39
J. D. BANE 3/15/68
J_
--
-.
-------
f I«W*1 M
N./ X 7'rt*»iit*
PLfc«TlC.»-
IU.
j^T /
~J~ 1 ' »T*wP
T- l0y *i*«*a<:-^
f^ffl1
H
j-r'v-sjTo -v»«S5»-
T-
, . k . J
1J...
DO HOT KALI OtUWIWO
««
-,.
«~_
-^r:
-.. rfr
r ,;
MIRCU1.EB INCOHFORATBD
F. 1021?
"S^i
-------
FIGURE 22
Overall Test Arrangement
FIGURE 23
Test Stand with Filter
49
-------
Each filter, after fabrication and crazing, was placed on the test stand
at ABL and its filtration rate determined at various hydraulic pressures. The
physical changes were also observed for use in checking the calculations
made for the filter's fabrication.
The test stand configuration is shown in Figure 24. The cylindrical
filter is clamped against the sealing 0-rings by the flanges. The tapered
doublers fit the clamping flanges so that a tensile load may be applied
to the filter if desired. The top plate assembly contains two-inch pipe
ports for feeding both tangent rally and axially. The bottom plate provides
the support for the upper assembly and baffle body, and ports for the filtered
and concentrated streams. Valves in the feed and concentrate lines provide
the means for adjusting the operating conditions of pressure and volume.
The testing program was directed toward obtaining engineering data on
filter operation for evaluation of the test vehicles. It became apparent
early in testing that although the data would not be complete, they would
indicate trends. These trends would allow a qualitative screening of the
effectiveness of various types of filters, baffles, etc.
Since a blank or control was almost impossible to incorporate in the
testing, the data from filter testing were reduced partly by direct comparison
and partly by experience. This was in part due to the change in sewage and
partly due to the differences in the filter fabrication methods. Also, since
all filters would partly plug during testing, no constant operating results
for an entire run could be obtained. This plugging also prevented using the
same filter twice so that a test could not be repeated with exactness. All
test data are recorded in Tables 6 through 17.
50
-------
Clear
Plastic
Shield
Bottom Plate
Assembly
Filtered
Water Drain
2" Standpipe
or Top Inlet
-2.
a
U3J
\A
Top Plate
/ Assembly
n
(f/^fs^/ '/ f S/ -^ ' \ljr/ ^i^^*
2" Tangential
Inlet
Y v-y
.
/ '/, '/////,'//// /, V,
Concentrated
Water Drain
Back-Up
Flanges
Tie Rods (6)
Concentrate Valve
FIGURE 24
Detail Of Sub-Scale Filter Test Assembly
-------
TABLE 6
Filter - 04A
Baffle - 11.6 x 10 cone
Feed - Top
Screen - 1/4"
I
TEST DATA
Time
(min)
5
10
15
20
21
25
28
30
31
36
37
38
43
44
0
0
0
5
Feed
Pressure
(psi)
9
10
10.4
10.5
12
12.6
14
0
13.2
13.7
0
13.7
13.7
0
13.2
7.7
12.5
13.5
Filtrate Flow
(gpm)
2.5
2.5
1.5
1
2.5
3.5
5
2
4
1.5
10
3
3.5
1.5
(«pm/ftz)
0.8
0.8
0.5
0.3
0.8
1.1
1.6
0.6
1.3
0.5
3.2
1.0
1.1
0.5
Concentrate Flow
(spin)
15
14
15
15
12
15
10
12
13
14
11
14
15
15
(gpm/ft2)
4.8
4.5
4.8
4.8
3.8
4.8
3.2
3.8
4.1
4.5
3.5
4.5
4.8
4.8
SEWER FILTER TEST
#1443-6
Date - 1/24/68
Bottom
Pressure
(psi)
7
7
7
9.5
10.5
9
10,5
10.5
10.5
10
4.5
9.2
10.2
Plant Influent - 0.2 MSD
Weather - Cold, Sunny
ANALYTICAL DATA
(mg/liter)
Remarks Source
Raw
Filtrate
Concent.
Raised Raw
Pressure Filtrate
Concent.
Raised
Raw
Filtrate
Concent.
Raised Pressure
Pressure Cycle #1
Pressure Cycle #2
Water Flush
Water Flush
Sewage Flow
Test End
SS
120
48
43
79
27
43
59
27
44
SSV
78
37
25
58
21
31
40
23
36
TS
538
499
434
126
476
482
484
323
487
TSV
145
74
23
54
55
73
100
18
89
Particle
BOD COD pH Dist.
41.3 7.4
31.2 7.5
7.2
7.6
7.7
7.5
7.5
7.3 Yes*1)
7.5
(1) 82% by weight particles smaller than 37(u.
-------
Filter - 04A
Baffle - 11.6 x 10 cone
Feed - Top
Screen - None
TABLE 7
SEWER FILTER TEST
#1443-7
Date - 1/31/68
Plant Influent - 0.26 M3D
(Controlled)
Weather - Warm, Sunny
Feed
Time Pressure
(tnin) (psi)
0
2
5
10
10.5
11
15
22
25
30
31
32
35
40
40.5
41
10.5
11.2
12
12.2
0
12.2
12.7
13.5
14
14.2
0
14
14.5
14.5
0
14
Filtrate Flow
(RPTO)
16
6
4.5
2
5
4.5
2
2
1.5
3
2
1.5
1.5
(gpra/ft*)
5.1
1.9
1.4
0.6
1.6
1.4
0.6
0.6
0.5
1.0
0.6
0.5
0.5
Concentrate Flow
(gpm)
10
15
15
15
15
15
15
15
15
15
5
5
5
(gpm/ft^T
3.2
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
4.8
1.6
1.6
1.6
Bottom
Pressure
(psi)
7.2
8
9
9
9
9.7
10.5
11
11.5
11
11.5
11.5
11
Remarks Source
Raw
Filtrate
Concent.
Raw
Filtrate
Concent.
Pressure Cycle #1
Dewatering Filtrate
Sample
SS
66
26
34
22
39
35
(mg/ liter)
Particle
SSV TS TSV BOD COD pH Dist.
33
17
24
13
30
14
Pressure Cycle #2
Concentrate flow reduced
Pressure Raw
Cycle #3 Filtrate
Concent .
67
49
64
44
32
47
44
14
1.5
0.5
2.2
11
Test End
-------
Filter- 04B
Baffle - 11.6 x 10 cone
Feed - Top
Screen - None
TABLE 8
SEWER FILTER TEST
#1443-9
Date - 1/31/68
Plant Influent - 0.26 MGD
(Controlled)
Weather - Warm, Sunny
TEST DATA
Time
2
5
8
10
15
20
25
30
40
43
45
50
51
55
60
Feed
Pressure
8
10.5
10.5
10.8
10.9
11.5
11.8
11.8
12.2
13.2
13.2
13.2
14.2
14.2
14.2
Filter Flos
16
12
7
5
3.5
3.5
3
2.5
1.5
2.7
2
1.5
2.5
2.5
2
(gpm/ft^)
5.1
3.8
2.2
1.6
1.1
1.1
1.0
0.8
0.5
0.9
0.6
0.5
0.8
0.8
0.6
Concentrate Flow
(gpni)
5
11
10
10
10
11
10
10
10
9
9
11
10
8
11
(Rpm/ft2)
1.6
3.5
3.2
3.2
3.2
3.5
3.2
3.2
3.2
2.9
2.9
3.5
3.2
2.5
3.5
Bottom
Pressure
(psi)
5
7.2
7.2
7.2
7.5
8
8.5
8.5
9
10
10.2
10
11
11
11
Remarks Source
Raw
Filtrate
Concent.
Raw
Filtrate
Concent.
Raw
Filtrate
Concent,
Pressure raised
Pressure raised
Raw
Filtrate
Concent.
ANALYTICAL DATA
ss ssv
92
35
38
36
24
34
52
34
39
42
33
43
72
22
27
23
16
23
38
23
2,9
32
23
32
(mg/ liter)
TS TSV BOD
402
383
590
414
385
388
397
387
403
424
391
420
96 22.6
79 19.3
181
105
95
97
95 22.7
90 22.7
103 22.6
109
93
123
COD
103
72
72
68
127
60
76
91
152
42
42
48
Particle
pH Dist.
7.4
7.2 Yes^1)
7.5
7.4
7.4
7.6
7'5 (2)
7.5 Yes^-1
7.5
-
7.5
7.6
7.5
61
Test End
(1) 88% by weight particles smaller than 37|u.
(2) 93% by weight particles smaller than 37^.
-------
Filter - 04B
Baffle - 11.6 x 10 cone
Feed - Top
Screen - None
TABLE 9
SEWER FILTER TEST
#1443-9A
Date - 2/15/68
Plant Influent - 0.3 MGD
Weather - Cold, Sunny
TEST DATA
Ul
Ln
Time
(min)
-
3
5
10
15
16
17
17.5
20
20.5
22
23
Feed
Pressure
(psi)
14.2
10
11
12
13
14
0
14
14
0
14
0
Filtrate
(gpm)
2
3.5
1.5
1.5
1.0
1.5
2
1
1
Flow
(gpm/ft*)
0.6
1.1
0.5
0.5
0.3
0.5
0.6
0.3
0.3
Concentrate
(gpm)
11
16
15
15
15
15
15
15
15
Flow
(ewn/ft^
3.5
5.1
4.8
4.8
4.8
4.8
4.8
4.8
4.8
Bottom
Pressure
(psi) Remarks
11 Filtering rate
at end of
previous test
7
8
9
10
10 Raised pressure
Pressure Cycle #1
10
10
Pressure Cycle #2
10
Test End
-------
Filter - 02A (Cloth Liner)
Baffle - Deck Plate
Feed - Top
Screen - None
TABLE 10
SEWER FILTER TEST
#1443-10
Date - 2/12/68
Plant Influent - 0.3 MGD
Weather - Cold, Sunny
TEST DATA
ANALYTICAL DATA
Time
(min)
2
5
10
15
16
17
25
29
30
34
35
39
40
41
Feed
Pressure
(psi)
1
5
9.5
11.2
14
0
14
10
0
13
0
14
0
14
0
Filtrate
(gpm)
18
7
4
1.5
1
3
1
3
1.5
1
Flow
(awn/ft2)
5.7
2.2
1.3
0.5
0.3
1.0
0.3
1.0
0.5
0.3
Concentrate Flow
(gpm) (awn/ft )
20 6.4
17 5.4
15 4.8
17 5.4
17 5.4
15 4.8
17 5.4
17 5.4
17 5.4
Bottom (mg/ liter) ...
Pressure o Particle
(psi) Remarks Source SS SSV TS TSV BOD COD pH Dist.
Water Flux
Rate Test
2
6 Raw 559 269 570 238 7.5
Filtrate 62 42 451 146 7.5 Yes
Concent. 182 132 531 224 7.3
8
11 Raised Pressure
Pressure Cycle #1
11
6 Reduced Pressure
Pressure Cycle #2
10.5
Pressure Cycle #3
10.5
Pressure Cycle #4
10.5
Test End
-------
Filter - 10A
Baffle - 11.6 x 10 cone
Feed - Top
Screen - None
TABLE 11
SEWER FILTER TEST
#1443-11
Date - 2/28/68
Plant Influent - 0.23 MOD
Weather - Cold, Clear
TEST DATA
ANALYTICAL DATA
Time
(min)
-
5
10
15
16
17
20
22
25
26
27
28
30
31
Feed
Pressure
(psi)
3
6
12
13
13.5
0
14
14
10
13
0
13.5
0
14
0
Filtrate Flow
(gpm) (gpm/ft^)
10 3.2
10 3.2
1 0.3
.75 0.2
.5 0.2
1 0.3
.5 0.2
5 0.2
1.25 0,4
1 0.3
Bottom (rag/liter)
Concentrate Flow Pressure Particle
(gpm) (gpm/f tz) (psi) Remarks Source SS SSV TS TSV BOD COD pH Dist_._
Water Flux
Rate Test
Water Flux Rate
Test with Hoop
Bands
15 4.8 9 Sewage Test
15 4.8 10
15 4.8 10.5
Pressure Cycle
15 4.8 10
Bands removed
11 3.5 7
11 3.5 10
Pressure Cycle
9 2.9 11
Pressure Cycle
7 2.2 11 Raw 159 136 485 229 51.2 7.25
'-' t iterate DJ *+/ Jo/ i**J
lest tna concent. 150 126 487 222
-------
TABLE 12
SEWER FILTER TEST
#1443-12
Date - 3/1/68
Plant Influent - 0.25 MGD
Weather - Cold, Clear
Filter - 10B with overwinds
Baffle - Ramp & Cylinder & 11.6 x 10 cone
Feed - Top
Screen - None
TEST DATA
Time
(rain)
-
5
6
0
5
9
10
12
-
0
5
9
10
Feed
Pressure
(psi)
10
14
0
8.5
14
0
14
0
15
13
12.5
13
0
14
Filtrate Flow
(Rpm) (gpm/ft2)
10 3.2
.75 0.2
.5 0.2
.75 0.2
.75 0.2
8 2.5
6 1.9
.5 0.2
.5 0.2
.5 0.2
Concentrate
(gpm) '
-
6
16
10
15
5
5
6
15
12
Flow
(gpm/ft^)
-
1.9
5.1
3.2
4.8
1.6
1.6
1.9
4.8
3.8
Bottom
Pressure
(psi)
5
5
11
11
12
10
7
10
11
Remarks
Water Flux
Rate Test
Ramp Baffle
Test End
Cone Baffle
Pressure Cycle
Water Flush
Sewage
Pressure Cycle
11
Test End
58
-------
TABLE 13
Sewer Filter Test
#1443-13
Date - 3/4/68
Filter - 11A Plant Influent - 0.3
Baffle - Cones (1) 10 x 11.6 (2) 7 x 11.6 Weather - Clear, Cold
Feed - Top
Screen - None
TEST DATA
Time
(tnin)
0
5
7
10
13.5
14
15
15.5
16
20
-
0
5
7
7.5
8
Feed
Pressure
(psi)
5.7
9.5
13.2
7
10.5
13
0
14
14
0
14
0
15
12
14
0
13.5
Filtrate Flow
(gpm) Cepm/ft2-L
5 1.6
10 3.2
15 4.8
.5 0.2
1.6 0.5
.75 0.2
1.5 0.5
.5 0.2
.5 0.2
15 4.8
.1 0.03
.5 0.2
5 0.2
Bottom
Concentrate Flow Pressure
(apm) (gpm/ft2) (psi) Remarks
Water flux
rate test
-
Cone (1)
16 5.1 4
16 5.1 6.5 Raised pressure
16 5.1 10 Raised pressure
Pressure cycle
15 4.8 11
15 4.8 11
Pressure cycle
14 4.5 11
Test end
Water flush
Cone (2)
15 4.8 8
15 4.8 10
Pressure cycle
15 4.8 10
9.5
10
15
0
14
0
.5
15
4.8
10
Pressure cycle
Test end
59
-------
TABLE 14
Sewer Filter Test
#1443-14
Date - 3/14/68
Filter - 12B
Baffle - Cone 7 x 11.6
Feed - Top (added booster pump)
Screen - 1/4"
TEST DATA
Time
(min)
.
-
0
5
10
10.5
15
_
_
0
7
14
15
25
26.5
27
34.5
35
36
37
44
45
55
60
61
70
71
80
82
83
90
Feed
Pressure
(psi)
15
20
19
20
0
22
20
22
20
21.5
23
22.5
0
23
0
23
0-23
21
0
22.5
22.5
0
23
24
0
23
0-23
25
23.5
Filtrate
(KPtn)
1.5
2.5
.5
.1
.1
1
1.5
.1
> .1
.1
> .1
.3
.3
.1
.2
> .1
.2
> .1
.1
.2
> .1
Flow
(gpm/ff)
0.5
0.8
0.2
0.03
0.03
0.3
0.5
0.03
> 0.03
0.03
> 0.03
0.1
0.1
0.03
0.1
> 0.03
0.1
> 0.03
> 0.03
0.1
> 0.03
Cor.centrate
(gpm)
-
-
10
10
6
_
-
7
7
4
3
3
3
3
3
2.5
3
5
5
5
5
Plant Influent - 0.45 MGD
Weather - Sunny, Cool
Flow
(gpm/ft2)
-
3.2
3.2
1.9
_
-
2.2
2.2
1.3
1.0
1.0
1.0
1.0
1.0
0.8
1.0
1.6
1.6
1.6
1.6
Bottom
Pressure
(psi)
16
17
19
18
19
20
20
20
20
18
20
20
20
21
20
23
21
Remarks
Water flux
rate test
Pressure cycle
Water flush
Pressure cycle
Pressure cycle
Pressure cycle
3 Pressure cycles
Pressure cycle
Pressure cycle
Pressure cycle
3 Pressure cycles
Test End
60
-------
TABLE 14 (CONTINUED)
Source
Raw
Filtrate
Concentrate
SS
75
20
46
ANALYTICAL DATA
(mg/liter)
SSV TS TSV BOD COD
Composite from Entire Test
49 630 162
9
33
650
657
112
150
Particle
Dist.
Yes
61
-------
TABLE 15
S3
Sewer Filter Test
#1443-15
Date - 3/15/58
Filter - 11A
Baffle - Deck
Feed - Top
Screen - 1/4"
Time
(min)
2
4
5
5.5
10
10.5
11
11.5
15
15.5
16
18
20
22
25
25.5
27
29
30.5
34
35
36
37
37.5
Feed
Pressure
(psi)
8
10
0
10.5
10
2
0
10
10
0
10
10
0
10
10
2
10
0
10
0
14.5
14.7
14.7
0
(Backf lushed)
Plate
TEST
Filter Flow
(gpm) (gptn/ft^)^
3.5 1.1
1.5 0.5
2 0.6
1 0.3
1 0.3
1 0.3
.7 0.2
.7 0.2
.5 0.2
.7 0.2
.5 0.2
.6 0.2
1.2 0.4
2 0.6
1.5 0.5
1 0.3
DATA
Plant Influent - 0.65 MGD
Weather - Warm, Sunny
Concentrate Flow
(gpm)
20
17
18
18
17
17
17
17
17
17
17
18
17
18
18
18
(tcpm/ff-)
6.4
5.4
5.7
5.7
5.4
5.4
5.4
5.4
5.4
5.4
5.4
5.7
5.4
5.7
5.7
5.7
Bottom
Pressure
(psi) Remarks
5.5
6
Pressure cycle
7
7
7 Open bottom valve
Pressure cycle
7
7
Pressure cycle
7
7
Pressure cycle
7
7
Open bottom valve
7
5 pressure cycles
7
Pressure cycle
11.5
11.5
11.5
Pressure cycle
-------
TABLE 15 (CONTINUED)
TEST DATA
Time
(min)
38
39
40
42
45
48
49
50
52
53.5
54
56
60
60.5
61
62
Feed
Pressure
(psi)
14.7
14.7
0
14.7
0
15
15
0
15
15
15
0
15
Filtrate Flow.,
(gpm)
1.5
1.0
1.7
1.0
0.5
2.5
1.5
1.5
0.7
0.5
1
(gpm/ft*)'
0.5
0.3
0.5
0.3
0.2
0.8
0.5
0.5
0.2
0.2
0.3
Concentrate Flow,
(gpm) (epm/ft )
18
18
17
--
15
15
15
15
15
15
5.7
5.7
5.4
--
--
4.8
4.8
4.8
4.8
4.8
4.8
Bottom
Pressure
(psi) Remarks
11.5
11.5
5 pressure cycles
11.5
5 pressure cycles
and drain.
12
12
Pressure cycle
12
12
12
Pressure cycle
12
Test End
-------
TABLE 16
Sewer Filter Test
#1443-16
Date - 3/15/68
Time
(niiO
0
5
5.5
10
14
14.5
15
17
18.5
19
22
24
26
28
30
32
33
34
35
Filter -
Baffle -
Feed
Screen -
Feed
Pressure
(psi)
7
9
0
10
10
0
10
10
0
10
10
0
10
0
10
10
15
10
10
6" Dia. x 6" long Plant Influent - 0.65 MGD
Cylinder
Bottom
1/4"
Weather - Warm, Sunny
TEST DATA
Bottom
Filtrate Flow Concentrate Flow Pressure
(gpm)
2
0.5
1.5
.3
1.0
0.3
1
0.3
0.5
1.2
0.2
2
0.5
_ ^gpm/ft2) (gpm) (gpm/ftz) (psi) Remarks
0.6
o.->
Pressure cycle
0.5
0.1
Pressure cycle
0.3
0.1
Pressure cycle
0.3
0.1 6 1.9
Pressure cycle
0.2
3 Pressure cycles
0.4
0.1
3 cycles overpressure
0.6
0.2
36
Test End
-------
TABLE 17
Sewer Filter Test
#1443-17
Dace - 3/22/68
Time
(min)
.Ji Hi.' "
_
_
_
_
-
1
3
6
9
10
11
13
13.5
15
17
17.5
21
22
23
25
26
30
.5
1
1.5
Filter -
Baffle -
Feed
Screen -
Feed
Pressure
(psi)
10
14
20
25
29
20
25
26
26
4
25.5
25.7
5
25
0
25
25.5
7
25
28
25.5
0
30
0
30
12B Plant Influent
Cone 10 x 11,6 Weather - Warm,
Top (added Booster pump)
None
TEST DATA
Bottom
Filtrate Flow Concentrate Flow Pressure
(Kpm) (gpm/£t*) (gpm) ^ftP/f *^). (psi)
-
- -
2.5 0.8
5 1.6 .
10 3.2
1.5 0.5 8 2.5 18
1.5 0.5 2 0.6 22.5
1.0 0.3 1 0.3 23
0.75 0.2 0.75 0.2 23
1 0.3 2 0.6 22.5
0.75 0.2 1,5 0.5 23
.75 0.2 3 1.0 22.5
0.75 0.2 2.5 0.8 22
0.3 0.1 2 0.6 22.5
0201 2 °-6 22
0.2 0.1 1-5 0.5 23
3 1.0
9 -2.9
- 0.5 MGD
Cloudy
Remarks
Water flux rate test
Slight weeping
Small streams
Streams
Adjust pressure
Weeping
Pressure cycle
Streams
Weeping
Pressure cycle
Few streams
Pressure cycle
Few streams
Few streams
Open bottom valve
Boost pressure with water
Test End
Water flush
Pressure cycle
. _ -
-------
Fifteen feasibility tests were performed at BGWTP using eleven different
filter elements and five swirl baffle bodies. The first tests were conducted
using tangential raw sewage feed to produce the swirling turbulent flow
described in the proposal. When it was found that the plugging solids appeared
to build up on the downstream side of the longitudinal cracks, the sewage
was fed to the test stand through the top opening to produce fluid flow parallel
to the longitudinal axis of the filter. Feeding parallel to the filter wall
noticeably decreased the rate at which the filter blinded. For this
reason, most of the tests were run with a nonswirling flow through the filter.
During the early testing it was noted that the screener basket was plugging
even though the screen was a %-inch mesh. The screen had been intended to remove
any large particles which might lodge in the narrow clearances between the filter
and the baffles in the sub-scale filter system. Since the strainer did plug and
cause a nonconstant feed composition for the filter, the screen was eliminated.
The first filters tested contained a limited number of cracks in the
gel coat. These few wide cracks allowed a filtration rate of 7 to 12 gpm/ft2
at 10 psi or less. During the tests with raw sewage, filtration rates
remained high several minutes and then began dropping, usually to 1 gpm/ft2
or less after ten minutes of operation. The filtration rate could be
restored somewhat by increasing the operating feed pressure, by flushing
with clear water, or by cycling the feed pressure to zero and back to
operating pressure. The mechanism of the latter method is attributed to
dewatering of the trapped solids which passed on through the wall at
repressurization.
66
-------
At the end of each test, the filters were removed and examined.
The interior walls showed no signs of deposit or build-ups other than
the material, mostly fibers, in the large cracks. Several filters which
contained fine pores were tested later. The filtration rate results
with these filters showed very little change from earlier filters in the
rate at which plugging occurred but, of far more importance, the character
of the plugging material changed. Few fibers were found trapped in these
fine pores.
Comparison of the results obtained with the swirl baffle bodies
yielded the following findings: (1) Fluid turbulence at the filter
surface did not reduce the plugging rate. (2) High fluid velocities
reduced the plugging rate. (3) Fluid flow parallel to the major axis
of the filter reduced the solids build-up.
One filter (04B) was allowed to dry out with the entrapped material
left in place. When testing of this filter was begun again, the filtration
rate (3.5 gpm at 10 psi) was higher than the rate before drying (2 gpm
at 14 psi), indicating that the dried material did not cause permanent
blinding of the pores.
Analytical data (Table 18) from the testing show a suspended solids
removal from the filtrate of up to 70%.
67
-------
TABLE 18
SUSPENDED SOLIDS REMOVAL SUMMARY
Raw Feed Filtrate
Filter
Number
04A
04A
04B
10A
12B
Susp. Solids
mg/1
120
79
59
66
92
36
52
42
159
75
Susp. Solids
Volat. mg/1
78
58
40
33
72
23
38
32
136
49
Susp. Solids
mg/1
48
27
27
26
35
24
34
33
55
20
Susp. Solids
Volat. mg/1
37
21
23
17
22
16
23
23
42
9
Eff.
O.)
60
66
54
61
62
33
34
22
65
73
This efficiency could be reached using both the coarse and finely
crazed filters. However, the results of the particle size distribution
analyses suggest a great difference in the character of the filtrate
solids. Figures 18 and 19 illustrate the particle size distribution
from filters with coarse openings, while Figures 20 and 21 are plots of
fine pore filtrate particle size distribution. It should be noted that
approximately 757o of the solids by weight passed by the fine pore structures
are 37 p. (0.0015 inch) or less in diameter.
The feasibility of filter 04B was tested during storm flow conditions
at the treatment plant. The influent flow rate was being controlled,
causing the feeder lines to act as a surge tank and allowing some sewage
to overflow and by pass treatment. The sewage test (#1443-9) lasted one
hour. Flow was maintained by raising the feed pressure when the rate
reached 0.5 gpm/ft2. After two minutes, the filtration rate at 8 psi was
68
-------
still above 5 gpm/ft2 and operation required only 1 gallon of concentrate
per 3 gallons of filtrate. A suspended solids removal of 60% was achieved.
Gradual plugging required increasing the concentrate flow to maintain the
pressure below 15 psi (the upper design limit for operation of this filter).
At the end of the test, the filtration rate was below 0.5 gpm/ft at
14.2 psi. The cracks in the gel coat were plugged with fibers that were
hooked in the openings, but the buildup was quite shallow and the remainder
of the filter interior was clean. This test indicated that the openings
must be finer to prevent fibrous blinding of the pores.
The filters which showed the greatest promise of fulfilling the
study goals were made in a polar winding pattern. A 6-inch-diameter,
6-inch-long filter element was polar wound using 6 plies of fiberglass
composed of 100 strands (20400 filaments) per inch of surface. This
filter was mechanically crazed by flexing parallel to the axis of the
cylinder. The structure would pass 25 gpm/ft* of tap water at 4 psi and
essentially 0 gpm/ft* at 1 psi. The sewage filtration test (#1443-16)
showed an early filtration rate of only 0.6 gpm/ft2 at 7 psi and rapid
plugging. However, an overpressurization cycle completely restored this
rate,indieating solids could be completely removed. The filtration
effectiveness was determined by comparing the filtrate and concentrate
visually. The concentrate was grey in color and loaded with fine particles
while the filtrate was clear with no visual particles. Filter 12B was
fabricated by hand laying two layers of 1-inch-wide preimpregnated fiber-
glass unidirectional tape on the mandrel at 0° with 90° overwinds of the
polyester fiber. Two layers were necessary to prevent gaps in the
69
-------
structure. Mechanical flexing produced fine cracks in the structure.
Due to the fabrication method, the resulting filter required a high
pressure to operate. Filtration rates of 0.8 gpm/ft2 at 20 psi and
3 gpm at 30 psi were reached with tap water. The initial rate with
sewage was 0.5 gpm/ft2 at 20 psi (Test #1443-17). Even though filtration
rates were low and plugging occurred rapidly, this filter is helieved to
offer great potential. By reducing the modulus of the structure or
increasing the diameter, the filter will be capable of operating at
lower pressure. A thinner wall will enable production of a higher number
of fine crazings and should allow the fine particles that enter the pores
to pass through more readily. Figure 25 shows the filter as it appeared
after feasibility testing. The fine cracks are partially plugged with
fine solids and very few fibers. Figure 26 shows the filter pressurized
to 30 psi. The fine sprays are indicative of the pore size.
A summary of the filter feasibility testing is shown in Table 19.
This evaluation brings out several points. Filter 04B produced
good flow but plugged with fibers which could never be made to pass the
filter. Both the 6 x 6 filter and the filter 12B plugged, but not with
fibers. All the filters showed good filtering efficiency. Plugging of
the fine pore filters could be relieved by overpressurization or by
passing the particles through the filter. While not demonstrated, the
results indicate that a polar wound structure can be made with the
following characteristics: (1) thin enough to craze in fine patterns with
a short,less-tortuous path for less trapping of particles, (2) capable
of high filtration rate, (3) strong enough to limit the amount of opening
70
-------
FIGURE 25
Filter 12B (After Testing)
FIGURE 26
Filter 12B (Pressurized)
71
-------
TABLE 19
SUMMARY OF FILTER FEASIBILITY TEST RESULTS
Test No. Filter
Baffle
1443-6 04A 11,6 x 10 Cone
(Normal sewage flow)
1443-7 04A 11.6 x 10 Cones
(Storm flow conditions)
1443-9 04B 11.6 x 10 Cone
(Storm flow conditions)
1443-9A 04B 11.6 x 10 Cone
(Slightly above normal flow)
1443-10 02A Deck Plate
(cloth
liner)
(Slightly above normal flow)
1443-11 10A 11.6 x 10 Cone
(Normal sewage flow)
1443-12 10B Ramp & Cylinder
and
11.6 x 10 Cone
(Normal sewage flow)
1443-13
11A
11.6 x 10 Cone
and
11.6 x 7 Cone
(Slightly above normal flow)
1443-14 12B 11.6 x 7 Cone
(Storm flow conditions)
Results
Low filtration rate and good sus-
pended solids removal. Pressure
cycling relieved blinding somewhat
but large gel coat cracks trapped
fibers.
High filtration rate and medium sus-
pended solids removal. Pressure
cycling relieved blinding somewhat
but cracks trapped fibers.
Very high filtration rate at start
and fair suspended solids removal.
One hour continuous test before
cracks trapped fibers.
Tested to determine filtration rate
after trapped solids dried in place.
Plugging was not complete but rate
could not be restored by cycling.
Rigid controlled pore cloth liner.
Low filtration rate which could not
be restored.
Low filtration rate, fiber plugging
could not be eliminated.
Low filtration rate, fiber plugging
could not be eliminated.
Low filtration rate, fiber plugging
relieved somewhat by water flushing.
Low filtration rate, blinding occurred
rapidly and high suspended solids
removal, but pores blinded with small
particles and relatively few fibers.
72
-------
TABLE 19(Cont)
SUMMARY OF FILTER FEASIBILITY TEST RESULTS
Test No.
1443-15
Filter
Baffle
Deck Plate
11A
(Back
flushed)
(Storm flow conditions)
1443-16 6" x 6" Cylinder
(Storm flow conditions)
1443-17 12B 11.6 x 10 Cone
(Storm flow conditions)
Results
Low filtration rate and blinding
occurred rapidly, but could be relieved
somewhat by cycling. Fibers trapped
in cracks.
High initial filtration rate but
blinding occurred rapidly. Plugging
relieved by pressure cycling and
overpressurization.
Low capacity pump added; low fil-
tration rate;could relieve blinding
somewhat but pores plugged with
fine particles and very few fibers.
73
-------
of the pores, (4) pores which will not trap fibers and (5) capable of
regeneration by overpressurization to remove solid particles. For these
reasons, Filter 12B is probably the best unit produced and offers the
route to development of a working unit. Accordingly, further experimental
evaluation of thin-wall filters with fine pores is recommended. Such
filters tested over a range of flow conditions (including both laminar
and turbulent) should establish whether or not the self-cleaning feature
can be realized.
Filtration Plant Costs
Data generated during the performance of this work were not sufficient to
allow design and costing of a complete filter plant utilizing the self-adjusting
self-cleaning filter. However, a rough order of magnitude cost can be projected
to allow preliminary evaluation of the system. These assumptions are based on
the best information we have at this time.
From what we now know, large filters would be costly to fabricate and would
be too space consuming in actual operation. A large filter would also require
a heavier wall which would mean more material per square foot of filter area
and a greater flow resistance. Thus, for this cost projection, an element size
of one to three feet in diameter will be considered. Again, due to manufacturing
cost considerations, length will be limited to about six feet.
The individual filter elements will be most easily handled if they are
grouped in a cluster. It is assumed that the ideal plant configuration would
be a two-section tank with half of the filter capacity on each section. In-flow
is assumed to be pump assisted to generate the necessary head during storm
flow conditions, thus allowing a gravity outflow. During dry flow only half
74
-------
the area would be used, allowing maintenance of the other half. With
the elements clustered in groups of four to seven on individual holders,
through-flow could be directed into separate drains located under each
element while filtered flow would be collected from the total area under
the filters. A bridge crane would be required to lift a cluster of
elements in and out for maintenance in a large plant.
The basis for cost calculations will be as follows:
(1) Sewer flow during storm conditions will be increased by a
factor of 10.
(2) Filter flux is estimated to be 7 gpm per ft^.
(3) Filter life is assumed to be one year.
(4) Elements will be clustered when a large number is used.
(5) Elements will be right cylinders 2 ft. diameter by 6 ft. long.
(6) Elements will support their own center bodies and will require
no other flow orientating arrangement.
A filter system for the Bowling Green Waste Treatment Plant would
be a typical small setup. For this plant, which has a dry weather flow
of 0.1 mgd and a storm flow of 1 mgd, three filter elements would be
needed. These would be installed in a metal filter holder and would be
operated at a positive pressure. A pump would be required to develop
the operating pressure. The pump house-filter area would require a
floor space of 10 x 12 feet and estimated cost would be in the neighborhood
of $30,000, most of which would be for the building and associated piping.
Assuming only 1 year filter life, operating cost would be about $950 per
year for filter replacement.
75
-------
No costs are given for all sizes of treatment plants because the
estimated costs are not accurate enough to allow a firm engineering
evaluation. For example, a lower filter flow rate would necessitate a
larger plant and a proportionally larger capital cost and operating cost.
A longer service life on the filter would allow a correspondingly lower
operating cost. Both of these factors would affect the costs in a nearly
linear fashion.
Naturally, as the size of the plant increases, the cost per 1000
gallon capacity would decrease. In a large plant, operating expenses
would be expected to be $0.01 to $0.02 per 1000 gallons of normal plant
flow,but firm costs must await results from a demonstration plant setup.
76
-------
GLOSSARY
BOD
COD
Catalyst D
Concentrate
Crazing
Crossover
Denier
DETA
Doubler
Epoxy Resin
Epon 826
Epon 953
End
Filament
Filtrate
Biochemical oxygen demand
Chemical oxygen demand
A curing catalyst for epoxy resin,
Shell Chemical Co.
Fluid retained within the filter-
ing element
Random hairline cracks in a resin
matrix
The point on a helically wound
structure at which the plus angle
filaments interweave with the minus
filaments
Grams weight of 9000 meters of roving
or yarn
Curing catalyst for epoxy resin
An area on a filament wound com-
posite structure reinforced for
special load bearing
Synthetic organic resin characterized
by two or more ring structures of
two carbon and one oxygen atoms
Epoxy resin, Shell Chemical Co.
Flexibilized epoxy resin, Shell
Chemical Co.
A group of filaments parallel to each
other, usually 50 to 204 in these
particular fabrications
A single fiber of material, usually
0.01 inch diameter or smaller
Fluid which passes through the filter
surface
77
-------
GLOSSARY (CONT)
GPM
Helical winding
Helix angle
M3D
Micromechanical analysis
Ply
Polar winding
Polyester
Polypropylene
Psi
Roving
Winding layer
Yarn
ZZL-0803
Gallons per minute
Filament winding technique that
places filaments in a helical
pattern on a surface of revolution
Angle of the filaments to the
central longitudinal axis of the
structure
Millions gallons per day
Prediction of composite structure
behavior from the knowledge of
constituent stress-strain properties
Filaments placed on a surface in a
single filament thickness
Filament winding technique in which
all filaments are parallel over a
surface without interweaving between
plies
Synthetic polymer produced from re-
action of ethylene glycol and terephthalic
acid or its derivatives
Synthetic long chain polymer produced
by polymerization of propylene (C3H6)
Pounds per square inch
One or more ends bundled together in
parallel without twist
A complete covering of the surface with
filaments placed at the plus (+) and
minus (-) helix angle
One or more twisted ends (or strands)
which are twisted together
A curing catalyst for epoxy resin,
Union Carbide Co.
78
-------
I BIBLIOGRAPHIC: Hercules Incorporated Allegany Ballistics "
Laboratory Cumberland, Maryland. Crazed Resin Filtration
j of Combined Sewer Overflows FWPCA Publication DAST-U 1968.
| ABSTRACT: The feasibility of developing a self-cleaning,
self-adjusting filtering device constructed of cylindrical
j structures of fibers laid down in predetermined patterns
by a winding process and bonded in place by resins was
| investigated. The permeability of the structure is im-
parted by a mechanical cracking or crazing of the resin.
| Filtration runs showed a 62$ reduction of suspended solids
in the filtrate. However, sustained runs could not be
I realized. The self-cleaning aspect was not demonstrated.
This report was submitted in fulfillment of Contract No.
I lU-12-39 between the Federal Water Pollution Control Admin
I istration and Hercules Incorporated.
J .
BIBLIOGRAPHIC: Hercules Incorporated Allegany Ballistics
| Laboratory Cumberland, Maryland. Crazed Resin Filtration
of Combined Sewer Overflows FWPCA Publication DAST-U 1968.
ABSTRACT: The feasibility of developing a self-cleaning
| self adjusting filtering device constructed of cylindrical
structures of fibers laid down in predetermined patterns
I by a winding process and bonded in place by resins was
investigated. The permeability of the structure is ira-
I parted by a mechanical cracking or crazing of the resin.
Filtration runs showed a 62$ reduction of suspended solids
I in the filtrate. However, sustained runs could not be
realized. The self-cleaning aspect was not demonstrated.
I This report was submitted in fulfillment of Contract No.
1^-12-39 between the Federal Water Pollution Control Admin
I istration and Hercules Incorporated.
BIBLIOGRAPHIC: Hercules Incorporated Allegany Ballistics
I Laboratory Cumberland, Maryland. Crazed Resin Filtration
of Combined Sewer Overflows FWPCA Publication DAST-U 1968.
ABSTRACT: The feasibility of developing a self-cleaning
I self-adjusting filtering device constructed of cylindrical
structures of fibers laid down in predetermined patterns
I by a winding process and bonded in place by resins was
investigated. The permeability of the structure is im-
I parted by a mechanical cracking of crazing of the resin.
Filtration runs showed a 62$ reduction of suspended solids
I in the filtrate. However, sustained runs could not be
realized. The self-cleaning aspect was not demonstrated.
I This report was submitted in fulfillment of Contract No.
lU-12-39 between the Federal Water Pollution Control Admin^
I istration and Hercules Incorporated.
ACCESSION No. I
KEY WORDS
Filter
Resin
Crazing
Volume Re-
duction I
Pollutant I;
Concentra-
tion j
"ACCESSION NO.
KEY WORDS
Filter
Resin
Crazing
Volume Re-
duction |
Pollutant |
Concentra-
tion |
ACCESSION No.
KEY WORDS
Filter
Resin
Crazing
Volume Re-
duction |
Pollutant |
Concentra-
tion j
GPO 882-977
------- |