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— >t—Test #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  i—i—n—i  i i—i
                                                         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

-------