EPA-600/2-76-025
February 1976
ORNL/TM-5423
Environmental Protection Technology Series
CROSS-FLOW FILTRATION IN
PHYSICAL-CHEMICAL TREATMENT OF
MUNICIPAL SEWAGE EFFLUENTS
Prepared by
OAK RIDGE NATIONAL LABORATORY
rliOMHUm-
Prepared for
Municipal Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into five series. These five broad
categories were established to facilitate further development and application of
environmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The five series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to develop and
demonstrate instrumentation, equipment, and methodology to repair or prevent
environmental degradation from point and non-point sources of pollution. This
work provides the new or improved technology required for the control and
treatment of pollution sources to meet environmental quality standards.
Printed in the United States of America. Available from
National Technical Information Service
U.S. Department of Commerce
5285 Port Royal Road, Springfield, Virginia 22161
Price: Printed Copy $6.00 ; Microfiche $2.25
This report was prepared as an account of work sponsored by the United States
Government. Neither the United States nor the Energy Research and Development
Administration, nor any of their employees, nor any of their contractors,
subcontractors, or their employees, makes any warranty, express or implied, or
assumes any legal liability or responsibility for the accuracy, completeness or
usefulness of any information, apparatus, product or process disclosed, or represents
that its use would not infringe privately owned rights.
-------
EPA-600/2-76-025
ORNL/TM-5423
Contract No. W-7405-eng-26
Chemistry Division
CROSS-FLOW FILTRATION IN PHYSICAL-CHEMICAL TREATMENT
OF MUNICIPAL SEWAGE EFFLUENTS
by
H. A. Mahlman, W. G. Sisson, K. A. Kraus, and J. S. Johnson, Jr,
Oak Ridge National Laboratory
Oak Ridge, Tennessee 37830
Contract WQO 14-12-832
AEC Interagency Agreement 40-191-69
Project Officer
Warren A. Schwartz
Municipal Environmental Research Laboratory
Cincinnati, Ohio 45268
February 1976
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
MUNICIPAL ENVIRONMENTAL RESEARCH LABORATORY
CINCINNATI, OHIO 45268
Prepared by
OAK RIDGE NATIONAL LABORATORY
Oak Ridge, Tennessee 37830
operated by
UNION CARBIDE CORPORATION
for the
ENERGY RESEARCH AND DEVELOPMENT ADMINISTRATION
-------
DISCLAIMER
This report has been reviewed by the Municipal Environmental Research
Laboratory, U.S. Environmental Protection Agency, and approved for publi-
cation. Approval does not signify that the contents necessarily reflect
the views and policies of the U.S. Environmental Protection Agency, nor
does mention of trade names or commercial products constitute endorsement
or recommendation for use.
ii
-------
FOREWORD
Man and his environment must be protected from the adverse effects of
pesticides, radiation, noise, and other forms of pollution, and the unwise
management of solid waste. Efforts to protect the environment require a
focus that recognizes the interplay between the components of our physical
environment—air, water, and land. The Municipal Environmental Research
Laboratory contributes to this multidisciplinary focus through programs
engaged in
studies on the effects of environmental contaminants on the
biosphere, and
a search for ways to prevent contamination and to recycle valuable
resources.
This report deals with development of technology for municipal sewage
treatment.
Louis W. Lefke
Acting Director
Municipal Environmental Research
Laboratory
iii
-------
ABSTRACT
In separations of solids from liquids by filtration, the motion of
liquid is customarily normal to the filtering surface, and solids are left
on the filter. In a variation called by K. A. Kraus "cross-flow filtra-
tion," liquid is pumped parallel to the filtering surface. By this device,
thickening of flux-limiting filtercake is slowed, and the original stream
is separated into a large volume of filtrate and a concentrated slurry of
solids. Results reported here were obtained in cross-flow filtration used
as an element in physical-chemical treatment of municipal sewage, mostly
with the effluent from primary settling, but in some cases, with clari-
fied activated sludge effluent as feed.
Filtrate from passage through fabric tubes (1-inch fire-hose jackets)
of primary effluent, treated with iron or aluminum salts, with powdered
activated carbon (PAC), or with both hydrolyzable ions and PAC, was gener-
ally of quality superior in turbidity, organic carbon and other respects
to the effluent from biological secondary treatment. Effects on product
quality and flux of pressure, circulation velocity, additive concentration,
water recovery, pH and other variables were investigated. Based on pro-
duction rates obtained, estimates of treatment costs were made.
This report was submitted in fulfillment of WQO Order # 14-12-832
by the Water Research Program, Oak Ridge National Laboratory, under the
partial sponsorship of the Environmental Protection Agency.
iv
-------
CONTENTS
Foreword ill
Abstract iv
List of Figures vi
List of Tables ix
Acknowledgments x
Sections
Conclusions 1
Recommendations 3
Introduction 4
Equipment and Experimental Methods 8
Cross-Flow Filtration Equipment 8
Test Sections 10
Analyses 14
The Oak Ridge Municipal East Sewage Plant 16
The Oak Ridge National Laboratory Plant 20
Experimental Results 21
Operation Variables 22
Effect of Cross-Flow Velocity 22
Effect of Pressure 25
Effect of Water Recovery . 35
Effect of Temperature 41
Effect of Additives 41
Removal of P(V) by Hydrolyzable Ion Addition 41
Effects of Hydrolyzable Ions and pH on Flux 44
Use of Lime for Neutralization 60
Quality of Cross-Flow Filtrates from Effluents 66
Powdered Activated Carbon 75
Two-Stage Processing by Hydrolyzable Ions Followed by PAC ... 79
Porous Supports, Precoats, and Regeneration 83
Discussion 89
Costs 91
Cross-Flow Filtration 91
Cost of Chemicals 94
Other Costs 103
Comments 103
Future Work 106
References 109
Publications and Patents 114
Symbols and Abbreviations 115
v
-------
FIGURES
No. Paee
1 Schematic of Cross-Flow Filtration Test Unit 9
2 Cross-Flow Filtration Element 11
3 Cross-Flow Filtration Unit Installed in Laboratory Loop 12
4 Multielement External Flow Cross-Flow Filtration Installed
in Mobile Unit 13
5 Cross-Flow Filtration Test Module, Internal Flow Through
Fire Hose Jacket 15
6 Comparison of TOG Analyses of Primary Sewage by Different
Procedures 17
7 Variation of Constituents in Primary Sewage, Oak Ridge
Municipal Plant: 1970-71 18
8 Primary Sewage Analyses, Oak Ridge Municipal Plant: Daily
Variation 19
9 Fluxes in Filtration of Primary Sewage with and without
Cross-Flow 23
10 Cross-Flow Filtration of Primary Sewage: Effect of
Circulation Velocity 24
11 Cross-Flow Filtration of Primary Sewage 27
12 Cross-Flow Filtration of Primary Sewage with Powdered
Activated Carbon 28
13 Flux as a Function of Average Pressure in Cross-Flow
Filtration of ORNL Sewage 30
14 Fluxes for Different Combinations of Pressure and Circu-
lation Velocity 31
15 Approximate Pumping Energy Requirements in Cross-Flow
Filtration 32
16 Fluxes of Aged and Fresh Pairs of Modules 34
17 Low-Pressure, Low-Circulation-Velocity Cross-Flow
Filtration of Primary Sewage 36
18 Comparison of Fluxes During Initial Stages of Several
Runs: Pilot Unit at ORNL Sewage Plant 37
19 Fluxes, Precoated and Non-Precoated Modules ... 38
20 Flux Decline at Different Pressures and Recoveries 39
21 Fluxes at High Water Recovery 40
22 Effect of Temperature on Flux 42
vi
-------
No. Page
23 Dependence of Phosphate in Filtrate on Ratio of
Phosphate to Ferric in Feed . 43
24 Effect of Ratio of Al(III) to Polyacrylate on Flux
in Filtration . 45
25 Cross-Flow Filtration of Primary Sewage with and without
Fe(III) Addition 47
26 Fluxes at Two Hours, Cross-Flow Filtration of Sewage
Effluents 48
27 Correlation of Flux and Filtrate Quality with Fe(III) 49
28 Correlation of Flux and Filtrate Quality with Fe(III) and
Organic Carbon Level 51
29 Correlation of Average Flux with Organic Carbon in Feed 52
30 Correlation of Average Flux with Organic Carbon in Filtrate ... 53
31 Fluxes at Two Hours, Cross-Flow Filtration of Primary Sewage . . 54
32 Correlation of Flux and Filtrate Quality with Fe(III)
Addition: Set II 55
33 Correlation of Flux and Filtrate Quality with Fe(III) and
Organic Carbon in Primary Sewage: Set II 56
34 Effect of Acidity on Cross-Flow Filtration Flux 58
35 Comparison of Initial Fluxes in Operation at pH 6 and pH 7 ... 59
36 Flux at Two Hours as Function of Al(III) Addition 62
37 Flux and Filtrate Quality with Al(III) Addition to
Primary Sewage 63
38 Flux and Filtrate Quality as Function of ppm Al(III)
Addition per ppm TOC in Primary Sewage 64
39 Comparison of Cross-Flow Filtration of Primary Sewage
Neutralized with CaO after Addition of Iron Sulfate and
Iron Chloride 65
40 Phosphate and Metal Ion Leakage in Cross-Flow Filtration
of Primary Sewage as a Function of pH 69
41 Filtration of Fe(total) with Time, after Introduction
as Fe(II) 71
42 Iron Contamination of Filtrate as a Function of Acidity 72
43 Adsorption of Organic Carbon from Filtrate of Fe(III)-
Treated Primary Sewage after Cross-Flow Filtration 77
44 Cross-Flow Filtration of Primary Sewage Effluent after
Powdered Carbon Addition 78
45 Two-Stage Cross-Flow Filtration of Primary Sewage:
Low Initial PAC Addition 80
vii
-------
No. Page
46 Two-Stage Cross-Flow Filtration of Primary Sewage:
Medium Initial PAC Addition 81
47 Two-Stage Cross-Flow Filtration of Primary Sewage:
High Initial PAC Addition 82
48 Regeneration by Backwashing: Comparison of Precoats 85
49 Regeneration by Backwashing: Successive Runs 87
50 Regeneration of Firehose Jackets 88
51 Plan for a 1 Million Gallon per Day Cross-Flow Filtration
Plant 93
52 Elevation of a Cross-Flow Filtration Plant Module 98
53 Cross-Flow Filtration-Dependence of Operating Cost on
Plant Size 99
54 Cross-Flow Filtration-Dependence of Operating Cost on
Flux 100
viii
-------
TABLES
No. Page
1 Effect of Cross-Flow Velocity on Flux in Filtration
of Effluent from Primary Treatment of Sewage . 26
2 Average Fluxes for the First 24 Hours of Operation 33
3 Comparison of Fluxes after Neutralization by NaOH and
by CaO 61
4 Nitrogen Removal in Cross-Flow Filtration 73
5 Bacteria Contents of Cross-Flow Filtrates 75
6 Plant Characteristics 92
7 Module Characteristics 95
8 Cross-Flow Filtration - Cost Summary (Non-Automated Plant) . . 96
9 Cross-Flow Filtration - Cost Summary (Automated Plant) .... 97
10 Estimated Typical Chemical Costs in Cross-Flow Filtration
of Primary Sewage Effluent 104
11 Summary of Estimated Costs of Cross-Flow Filtration of
Primary Sewage Effluent . . 104
ix
-------
ACKNOWLEDGMENTS
We are indebted for the cooperation of Oak Ridge municipal officials,
in particular 0. K. Rickman, director of public works, Thomas C. Stephens,
supervisor of sewage plants during the early phases of this program, and
Jack Robinson, Jr., present supervisor. In operation of the pilot plant
at the Laboratory sewage plant, we have benefited greatly by the partici-
pation of J. D. Hutchins, ORNL Operations Division, and of J. W. Tester,
director, and R. W. Mayer, assistant director, of the Oak Ridge Station
of the MIT School of Chemical Engineering Practice during the period when
the work described in this report was carried out. Neva E. Harrison
carried out the bulk of the analyses, and contributions of others are
acknowledged at appropriate places in the report. Arthur J. Shor was
most helpful in providing consultation concerning equipment. Editing and
typing of this report was by Jane Beck, and figures were drafted by
B. S. Dunlap.
We are grateful for the advice and support of Warren A. Schwartz,
contract monitor, and J. M. Cohen of the U. S. Environmental Protection
Agency. The work was also substantially supported by the U. S. Energy
Research and Development Administration. We wish to thank in particular
F. L. Culler, deputy Laboratory director, for his continued interest.
-------
CONCLUSIONS
1. Cross-flow filtration of primary effluent treated with Fe(III) or
Al(III) produces filtrate typically containing 10 to 15 mg/£ of total
organic carbon (TOG), total phosphorus below 1 mg/£ (expressed as phos-
phate) , and turbidity below 1 Jackson Turbidity Unit (JTU). With settled
activated sludge effluent as feed, filtrate composition is similar,
except that TOC is lower, usually about 5 mg/&. Bacterial contents of
filtrates are usually low.
2. Filtrate from primary effluent treated with powdered activated
carbon (PAC) is of similar quality, except that little phosphate is
removed, and TOC is usually lower, about 5 mg/fc.
3. When Fe(III) or Al(III) additions are used, unless pH is adjusted
to the neutral range before filtration, filtrate is substantially contam-
inated with the additives.
4. An effluent low in phosphate and in TOC may be attained by a
second-stage cross-flow filtration of PAC-treated filtrate from primary
sewage, to which hydrolyzable ions had been added.
5. Fluxes increase with increasing cross-flow velocity. In opera-
tion at 4.6 meters/second (m/sec) or 15 feet/second (ft/sec), with Fe(III)
or Al(III) additives, average fluxes of 6 meters/day (150 gallons per
2
square foot per day (gpd/ft )) or higher seem attainable with 24-hour
intervals between backwashing or other regeneration procedures. Conflict-
ing results have been obtained with respect to the effect on flux of
Fe(III) additions above 50 mg/fc.
6. If neither hydrolyzable ions nor PAC are added to primary effluent,
fluxes in cross-flow filtration are much lower than if appropriate amounts
of these additives are present.
-------
7. High concentrations of PAC in the solutions being filtered are
necessary to attain favorable filtration rates. However, fresh primary
sewage may be brought into the feed without further PAC additions to
replace filtrate until the average amount of PAC used per volume treated
is lowered to the range used in other tests of this reagent. For example,
in a test in which 20 grams of PAC were added per liter of the initial
charge of primary sewage, the average mg of PAC per liter of primary
sewage processed in the first day was about 500, and the average flux
2
was vL6 ro/d (^400 gpd/ft ).
8. After initial transients, fluxes are not very dependent on pres-
2
sure in the range studied, between 35,000 and 690,000 Newtons/meter
2
(N/m ) or 5 and 100 pounds per square inch (psi).
9. Neither flux nor product quality are very sensitive to water
recovery (fraction or percent of initial feed volume in filtrate stream),
at least up to 90%.
10. Woven fabric tubes (2.5-cm (1-inch) diameter manufactured for
use as fire-hose jackets) are suitable as filtering surfaces. In more
limited tests, screens appeared to be acceptable alternatives. Fibrous
filteraid precoats are necessary with coarse screens (perhaps <100 mesh)
and seem helpful with fire-hose jackets in maintaining ability to regenerate
by backwashing in successive runs. There appears to be no significant
difference in performance of fire-hose jackets when flow of the solution
being filtered is outside (pressure support inside) or inside (no support).
r> f
11. For a 3800 m /d (10 gpd) plant and for average fluxes in the
lower range of those obtained in this study, preliminary estimates indicate
costs of cross-flow filtration of primary sewage with ferric chemical
3
treatment are about 12
-------
RECOMMENDATIONS
1. Further experiments, simulating practical operation and including
substantial water recoveries, are desirable to define more closely the
optima for certain variables. The most expeditious course is to carry out
simultaneous tests in parallel loops, with identical feed and conditions,
except for the variable under investigation. Proceeding in this way will
minimize uncertainties arising from variations with time of sewage composi-
tion. Variables which need further attention include comparison of additives
(Al(III) vs_ Fe(III); Fe(III) introduced as different salts; CaO vs_ NaOH in
neutralization); concentration of additive; pH of pretreatment and of
filtration, the latter particularly with PAC; and amount of filteraid
used in precoating.
2. More extensive exploration of filtration at low circulation veloc-
ities is needed.
3. A limited number of critical tests should be carried out on sewage
containing appreciable inputs from industries.
4. Modules suitable for practical systems should be designed, con-
structed, and tested.
5. Planning should be started for a demonstration. The most effec-
tive course would appear to be locating in our vicinity a small sewage
plant presently treating only by primary settling but planning in the near
future to add secondary. A cross-flow filtration system would be con-
structed to process the full primary output, so that the feasibility of
recycle of concentrated reject to the influent of the settler could be
determined. An alternative next step in development might be a package
plant designed for operation where space is limited (e.g., shipboard).
6. The usefulness of cross-flow filtration as an element in treatment
of industrial wastes should be evaluated. Developments of specific
schemes for individual feeds will be required.
-------
INTRODUCTION
Biological methods have dominated the treatment of municipal sewage
when it is carried beyond the usual primary settling stage. In spite of
the widespread further processing by aerobic activated sludge or trick-
ling filter methods, aimed at converting organic material into settleable
sludge, there are deficiencies in these usual secondary methods which
have kept interest alive in other approaches as supplements or alterna-
tives. Supplementary treatments (frequently referred to as tertiary)
would be aimed at improving the effluent from biological secondary
processes, e.g., removal of more oxygen-demanding constituents by treat-
ment with activated carbon, removal of phosphates by ferric or aluminum
addition, or removal of ammonia by addition of lime and sparging.
Some of these tertiary methods are also under test for application
directly to the primary effluent, or even raw sewage, as alternatives to
conventional secondary treatment—addition of activated carbon or hydro-
lyzable salts, for example. The objective is to improve effluent quality,
and perhaps more important, to gain reliability. Microorganisms on which
aerobic treatment depends can be seriously upset by substances intermit-
tently present in a sewage system or by flow of excessive volume through
the plant, and the restoration of their activity to give adequate treatment
can take undesirably long times.
After treatment with hydrolyzable salts or with activated carbon, if,
contrary to usual practice, it is introduced in powdered form, the bulk of
the water still has to be separated from additives and impurities. Fre-
1 2
quently settling is used, sometimes followed by filtration. * Settling
3
may not always be satisfactory. From tests at Mentor, Ohio, of additions
of ferrous-containing pickle liquor with lime neutralization, for example,
it was concluded that filtration might be required, since plant effluents
-------
were contaminated with about 10 milligram/liter (mg/£) iron. In similar
4 5
experiments at Milwaukee ' in which Fe(II) additions were in the range
8-15 mg/&, rather than the 30-50 mg/JJ, used in Mentor, effluent contained
typically one to two mg/£ Fe, only a small fraction of which was reported
as "soluble;" total phosphorus (reported by them as phosphorus P, rather
than phosphate used in this report) in the effluent was in the range
0.5-1 mg/X, (average 0.7). Since only about half of this was reported as
"soluble," more efficient separation of solids should have enabled meeting
the target of 0.5 mg/£ P. Total effluent phosphorus was reduced only by
half from what it was without iron addition. In Waukegan, Illinois, pick-
ling liquor additions to the secondary stage in amounts corresponding to a
1.9 mole iron per mole P in the primary effluent resulted in average P
in secondary effluent of 1.4 mg/£; polymer addition and/or filtration were
felt necessary for consistently higher removals.
Utilization of activated carbon in powdered form has many potential
advantages over granular—lower cost/lb, faster kinetics (smaller parti-
cles), for example—but its use also requires efficient solids-liquid
3 3
separation. In a 545 meter /day (m /d) (100 gallons/minute (gpm)), 2-year
7 8
study at Salt Lake City, ' coarse screened and comminuted raw sewage
was successively treated with Al(III) or Fe(III) salts and lime to remove
phosphate, then contacted with powdered activated carbon (PAC) in one or
two stages, settled, and finally filtered through a coal-sand bed. Regen-
eration of the carbon was also tested. The filtration generally reduced
suspended solids (effluents 1 to 3 mg/£) and turbidities (effluents 1 to 5
JTU) by a factor of 2 to 3 and usually effected some reduction of total
organic carbon (TOG), chemical oxygen demand (COD), and biological oxygen
demand (BOD). Removal increased with increased carbon doses. Typically,
COD of the effluent was about 10 mg/i when about 350 mg/£ of PAC was used
and feed COD was about 35 to 75 mg/£.
A somewhat similar 2-stage countercurrent scheme was tested at up
3 9
to 110 m /d (20 gpm) in Tucson, Arizona, the main difference being absence
of phosphate removal in a separate step, though alum was added to aid
solids-liquid separation. Effluents were typically well below 10 mg/£
in TOC, and the final filtration step did not seem to improve product
substantially over the second carbon step, except that turbidities of
-------
filtered effluent were substantially lower. Reduction of TOC was from
80% at 100 mg/£ PAC to 90% for 250.
3
Tests were carried out in Albany, N. Y., at ca. 380 m /d (100,000
gpd) level on municipal sanitary sewage, sometimes combined with storm
drainage. After treatment with powdered carbon in a pipe reactor, solids-
liquid separation was by a tube settler, followed in some cases by a
tri-media filter. The effluent contained on the average 17 mg/A BOD
with a turbidity of 0.6 Jackson turbity units (JTU).
Numerous other studies have indicated that filtration improves
removal of impurities. In tests of physical-chemical treatment, includ-
12
ing lime addition, at Washington, D. C., the average suspended solids
obtained over a year and a half operation was reduced from about 20 mg/&
in the clarified effluent to about 5 mg/£ by dual-media filtration; corres-
ponding reduction for phosphorus (as P) was from 0.3 to 0.2 mg/&. Even more
pronounced differences between clarified and filtered water were found
13
in alum-treated activated sludge effluent at the same station. Obser-
vations were similar with alum-treated, trickling-filter effluent in
14
Richardson, Texas. Filtration steps in treatment of waste-water in
Contra Costa County, California, also improved acceptability of the
effluent for industrial reuse.
The filtrations utilized for solids-liquids separations in the
references cited so far have operated with the flow of the solution being
filtered only in the direction perpendicular to the filtering surface.
The filters attain typically extremely high rates at low pressure drops
by arranging a coarse filter over a somewhat finer layer or layers, the
overall looseness of the system being indicated by the appreciable
turbidities and suspended solids in the filtrates. Material removed
accumulates in the filter and is removed periodically by backwashing.
An alternative mode, ' ' which has been called "cross-flow
18
filtration, is analogous to that used in hyperfiltration (reverse
osmosis). The solution is pumped parallel to the filtering surface
(or relative movement of solution and surface is attained in some other
way; for example, by mounting the filter on a rotor, in a variation
19 20
referred to as "axial filtration" ' ). The circulation will tend to
slow the thickening of flux-limiting filter cake, and one ends up with
-------
the bulk of the water purified by passage through the filter, and a
relatively small volume containing the concentrated removed substances.
While the approach does not obviate the necessity of periodic backwashes
or other regeneration procedures, it allows use of a relatively tight
filtering surface or filter cake, efficient in removal, and attainment
of higher filtering areas per unit volume than in many standard config-
urations, to balance against lower production rates per square foot and
energy required for pumping.
There appears to have been little done that would allow evaluation of
cross-flow in application to sewage treatment. Perhaps the closest work
involved cell-mounted commercial ultrafiltration membranes in bench-scale
21
operation for dewatering powdered carbon slurries. This report describes
exploratory investigations of the use of cross-flow filtration with tubular
filter elements to primary and secondary effluents treated with hydrolyz-
able salts and/or powdered activated carbon. The study started in 1970
and has continued intermittently up to Spring 1975.
-------
EQUIPMENT AND EXPERIMENTAL METHODS
CROSS-FLOW FILTRATION EQUIPMENT
The essential feature of the approach is a configuration allowing
filtration under conditions of relative movement of the fluid being
filtered and the filtering surface. In the work reported here, this was
accomplished by pumping the solution through the inside of a tubular
filter, or through the annulus between the outside of a tubular filter
and outer pressure jacket.
The bulk of the study was carried out in a unit located at the Oak
Ridge East Municipal Sewage Plant. The installation has been modified
and amplified from time to time. The main elements are shown schematic-
ally in Fig. 1.
The units for addition of hydrolyzable salts and acid and base
for the desired pH cycle are housed in a 2.3 x 3.7 m (7 1/2' x 12')
van. The effluent from primary or secondary treatment is pumped
into Tank I when called for by level controls, and Fe(II), Fe(III), or
Al(III) salts plus acid to the appropriate pH are metered in. In report-
ing results, mg/£ of Fe(lII) or Al(III) refer to the amount added to in-
coming primary effluent, rather than the amount in the circulating solu-
tion being filtered, in which the content will increase as filtrate is
removed. Powdered activated carbon can be added in addition or alter-
natively. Air sparging removes CO-, oxidizes Fe (II) to Fe(III),
and helps mixing. This solution moves to Tank II on call, and base is
added to bring it to the operating pH. Air is passed through to complete
oxidation to Fe(III).
Level controls in the feed tank III, located, along with the rest of
the system, in a 2.4 by 18.3 m (8' x 60') trailer, activate the transfer
of tank II solution into it. From the feed tank, the solution is pumped
8
-------
Q
iFEED
o
tE
I
400 i
TANK
PUMP
3501
TANK
VAN
FCV, Flow Control Valve
PVC, Pressure Control Valve
1
ACID
SE
FCV
rp
*
O Z
-I O
U. H
o
— ! (/)
^
^ ^
U UJ
1
^
F
5
^ o
O j_
^a
< ,_
2^
Ct LJ
LJ (—
1 —
UJ
-t — •• FILTRATE
•
r
FILTRATE
PCV
, I
FCV
COOLING
COIL
GOULDS PUMP
W/40 HP MOTOR (30kw)
~ 1100j!pm AT 8.6 bars '
~ 300gpm AJ_t25_psig |
TRAILER
Fig. 1. Schematic of cross-flow filtration test unit.
(pH and level controls, valves, instrumentation for flow measurement.
backwash equipment, and second-stage cross-flow loop not shown)
-------
through the filter test sections, pressure being maintained by passage
through a let-down valve before return to the feed tank. The filtrate is
only a small fraction of the solution passing through the test section and
consequently all parts of the test sections see essentially identical
solutions at a given time. Filtrate can be discarded, recycled to the
feed tank, or transferred to the feed tank of a second loop for tests of
second-stage operation. A D/P cell, connected with a Foxboro recorder,
monitors the pressure drop across each test section, and allows determin-
ation of the circulation velocity.
Two cross-flow loops are located in the trailer; one has provision
for two test sections in parallel, and is equipped with a 1135 liters/
minute (Jl/min) (300 gallons (U.S.)/minute (gpm)) at 862,000 Newtons/square
2
meter (N/m ) (125 pounds per square inch, measured by gauge (psi)) Goulds
centrifugal circulation pump; the other has provision for 5 parallel test
2
sections and is equipped with a 950 £/min (250 gpm) at 862,000 N/m (125
psi) Goulds pump. Originally the pumps were used with packed seals, but
rapid wear, particularly in runs with powdered carbon, led us to substi-
tute mechanical seals. Metal cooling coils in the feed tanks remove heat
generated in circulation of the solution.
Test Sections
Several different types of test sections have been used. Figure 2
is a picture of a typical unit not completely assembled. A 2.54 cm (1")
18
diameter polyester-nylon fire-hose jacket, (manufactured by Fabric Fire
Hose Co., Sandy Hook, Conn.) is slipped over a perforated stainless steel
tube. A transparent plastic jacket defines the outer wall of the annulus.
Figure 3 shows a similar unit installed in a laboratory cross-flow loop.
To the right is a test section in which feed is circulated on the inside.
Outside flow with supported hoses allows backwashing; with inside flow,
other regeneration methods must be used. Figure 4 shows some larger,
outside-flow configurations, operating in one of the loops in the mobile
unit. The test section in the center of the picture has four tubes in
2
parallel, which give a total of about 0.37 m (four square feet) of mem-
brane area.
10
-------
OAK RIDGE NATIONAL LABORATORY
23456789 K> 41 42
i I i I I i I i I i I i I i I i I i I i I i I i I t I i I i I
Fig. 2. Cross-flow filtration element.
-------
Fig. 3. Cross-flow filtration unit installed in laboratory loop.
(Left: external flow of pressurized solution. Right: internal flow.)
12
-------
Fig. 4. Multielement external flow module installed in mobile unit.
-------
An inside-flow unit of comparable filtering area is diagramed in
Fig. 5. A coil of about 5.5 m (181) of fire-hose jacket is wrapped around
a thin cylinder of sheet metal in a 114 & (30-gallon) stainless-steel
barrel. In contrast to the other units, there is a very substantial drop
o
in pressure through the unit—225,000 N/m (32 psi) at 4.6 m/sec (15 ft/sec)
circulation velocity, for example.
Filtering surfaces are not restricted to fire hose jackets. Screens
from 20 to 400 mesh, with a precoat of a fibrous asbestos or cellulosic
filteraid, were also used. Precoats also appeared advantageous with fire-
hose jackets.
The overall system is a test unit, rather than a pilot plant. Pro-
duction rates usually ranged between a hundred and a thousand gallons per
day. On the basis of results obtained with the mobile unit, a nominal
3 22 23
378 m /d (10,000 gpd) pilot plant ' was constructed at the Oak Ridge
National Laboratory sewage treatment plant. Although work done with it
is outside the scope of this program, tests of some operating conditions
were much simpler with the unit, and some results are reported here. The
filter surfaces were nominal 7. 6 cm diameter (3" I.D.) fire-hose jackets
supported on the inside by perforated stainless steel tubes, with feed in
the annulus between the jackets and hose of 10.2 cm (4") stainless
o o
steel pipe. Ten 2.44 m (8-ft) units (about 0.65 m (7 ft ) each) are con-
nected in series, with valving and bypassing allowing successive pairs
to be included or excluded from the circulating stream.
None of these configurations are projected for actual sewage process-
ing. Rather, they are experimental arrays, with defined flow patterns,
convenient for studying the effect of variables of performances.
ANALYSES
Evaluation of treatment was primarily based on total organic carbon
(TOG), phosphate, and iron in the filtrate. Turbidity (as measured by a
Hach 2100 Hach turbidimeter) was also monitored.
Total organic carbon (TOC) was measured with a Beckman 915 analyzer.
With this instrument, carbon as C0_ is measured by infrared absorption of
an oxygen stream through the reactors. Two samples are analyzed: one is
14
-------
CONTAINER
TOP
FILTRATE FROM
FEED
TANK
SIDE
Fig. 5. Cross-flow filtration test module,
internal flow through fire hose jacket.
15
-------
injected into a phosphoric acid reactor, which displaces carbon dioxide
from inorganic carbonate species; the other is a high-temperature catalytic
reactor, which oxidizes organic carbon components to C09, as well as
decomposing inorganic carbonates. The TOC is the difference in the carbon
determined in the two analyses. There have been questions raised about the
accuracy of this difference procedure, and some have recommended eliminat-
ing the inorganic step by acidifying and gas purging to remove inorganic
species, before injecting samples into the organic reactor. However, if
volatile organics are present, this procedure would lead to incorrect
estimates of the organic content. In Fig. 6 are plotted some results with
primary effluent from the Oak Ridge Municipal plant. These show a dis-
crepancy between TOC obtained by difference and TOC measured directly on
acidified, air-sparged samples. However, similar discrepancies are
observed with samples (TOC measured by difference) air sparged at approxi-
mately neutral pH, which still contain substantial inorganic carbon. It
appears that there is considerable volatile organic carbon present, and
we therefore report analyses by the difference procedure specified for the
instruments.
In analysis of total phosphate, samples were pretreated with persul-
fate, and color developed with potassium antimony tartrate and ascorbic
acid (Ref. 24, p. 236).
Iron analyses were by the orthophenanthroline method, after pretreat-
ment with hydroxylamine (Ref. 24, p. 156).
THE OAK RIDGE MUNICIPAL EAST SEWAGE PLANT
3
The facility is of about 5700 m /d (1.5 million gallons per day)
capacity. Treatment is by primary settling and activated sludge secondary
with chlorination of effluent and anaerobic digestion of sludge. The
plant is one of two handling the wastes of the city, essentially a bedroom
community of approximately 28,000 population. High flows occur during
heavy rains. Sewage during these periods is of low organic content.
A summary (Fig.7) of characteristics of the primary effluent, mea-
25
sured over a year, is reprinted from an earlier report. Some measure-
ments taken during the course of this work are presented in Figs. 6 and 8.
16
-------
Q.
Q.
o
o
a:
o
50
0
10
0
(
i
- «
I
— o
o
I
1 i ' i ' i
0
8\
« $<|
«*•
, ! -I ' 1
v £\ O
^A^\
,1 , 1 . 1
1200 0800 1200 1600
TIME
March 6, 1974 March
Fig. 6. Comparison of TOC analyses
o Difference between total and
1 1
O
t
1 REPEAT
ANALYSIS
18 MAR.
i 1
0
O
1
2000
7, 1974
of primary
1 1
o -
o
. 1
o
o
1
2400
sewage.
Analyzed
11-15 Mar.
inorganic carbon, as received
Difference between total and
inorganic carbon, air-sparged ,
neutral pH
Air-sparged after acidification
(all inorganic carbons <1 ppm )
15 Mar.
17
-------
—
DC
Co+UMg" C
(M] ' °0015
00010
60
TOTAL
PHOSPHATE ION
0
100
80
ORGANIC
CARBON 60
(ppm)
40
INORGANIC 60
CARBON
(ppm) 40
JTU
6O
:
5 15 25 5 15 25 5 15 25 5 15 25 5 15 25 5 15 25 5 15 25 5 15 25 5 15 25 5 15 25 5 15 25 5 15 25 5 15 25
JULY AUG SEP OCT NOV DEC JAN FEB MAR APR MAY JUNE JULY
1970 1971
Fig. 7. Variation of constituents in primary sewage effluent, Oak Ridge East Sewage Plant.
-------
E
a.
Q.
IbU
120
80
40
7.6
7 /i
f .*+
7.2
~r f\
7.Of
1
'- +
"
—
QO *•
• 8
1
A
A i
- A i
i
5 O C
0 C
CVJ ^
— C
'
Ooo(
I °GD .
a
t
A^A A
/
DOC
DOC
h CVJ ^
5 - cl
'
oo
I O°D
o •:
a
A A
A ^
/
\ ' L
5 § i
CVJ ^
J - CV
1
> o°o<
1 D
a
A
A
AA
, L
\ i i
I ° S
) O C
CVJ ^
j -- cv
> °00(
i °aa(
o
a
A
A A
\ *t
i
> 9 c
! s §
J — o
1
o
> oDa°<
r
a
A
i A ^^^1
' i
i o c
3 O C
r cvi o °-
o
Q
—
A -
i A ^i
i
> 0 C
> O C
CVJ rf
J - cv
TIME
DAY
WEATHER
WON.
CLEAR
TUES.
RAIN
WED.
RAIN
THUR.
CLOUDY
FRI.
CLEAR
SAT.
CLEAR
SUN.
CLOUDY
Fig. 8. Primary sewage analyses, Oak Ridge East Sewage Plant.
(March 4-10, 1974)
19
-------
Values of BOD in the figure were provided by the staff of the sewage
plant.
The composition of sewage is highly variable and consequently only
limited confidence can be placed in results of a single experiment. This
situation makes more difficult determination of the effects of variations
in experimental conditions and the optimization of processes. In the Oak
Ridge case, besides gross differences in feed composition after rainfall,
organic carbon contents were high when septic tank sludge or digester
supernate were discharged into the plant. We were not always aware of
the latter occurrences.
THE OAK RIDGE NATIONAL LABORATORY PLANT
The Oak Ridge National Laboratory plant, at the time of the pilot-
2
plant tests reported here, was of approximately 570 m /d (150,000 gpd)
capacity. Treatment was of sanitary wastes generated in the X-10 area
and was limited to primary settling. The effluent during working hours
was comparable in TOC to that from primary treatment of the municipal
plant, but TOC was much lower at night and on weekends.
20
-------
EXPERIMENTAL RESULTS
In this study, over two hundred and fifty runs, typically of about
24 hours duration, were made at the Oak Ridge municipal plant, plus some
sixty preliminary laboratory tests. It is not feasible to present a
detailed account of all experiments, even if operating difficulties and
errors in execution had not so flawed many as to compromise their value.
Consequently, the report will organize results to illustrate the effect
of important parameters. Among the variables of interest for which some
information is presented are:
• Feed—primary, secondary effluent, or other.
• Cross-flow velocity.
• Pressure.
• Performance as a function of water recovery (fraction of water
removed as filtrate).
• Temperature
• Additives—Fe(II) or Fe(III), or Al(III) salts; powdered activated
carbon (PAC); other chemicals.
• pH.
• Type of filtering surface.
• Precoating of filtering surface.
• Efficacy of backwashing or other regeneration procedures.
Because feedstock varied and because the system had many variables,
conclusions concerning optimum performance and conditions are tentative.
However, our experiences do indicate the promise of the approach, and the
situations in which it may be useful.
Before discussing in detail effects of additives, we shall present
results illustrating effects of operational variables.
21
-------
OPERATION VARIABLES
Effect of Cross-Flow Velocity
Pumping solutions parallel to the filtering surface has been shown to
•I Q
increase fluxes in filtration of various materials. Cross flow has a
similar effect with sewage. We have previously reported a laboratory
experiment in which a single unit was tested without circulation and,
after an intervening backwash, with circulation; flux decline was much less
sharp with cross flow than without (Fig. 12 of Ref. 26). Fig. 9 summarizes
a similar test carried out at the sewage plant, in which two external flow
units were exposed simultaneously, one with no cross flow and the other
with circulation at 4.6 m/sec (15 ft/sec). Both units were precoated
with a fibrous filteraid prior to exposure to ferric-treated primary sewage
effluent, and the fluxes with water at the end of pretreatment were essenti-
ally the same. Flux decline was much more rapid without cross flow. Fil-
trate quality was not very different—organic carbon was 10 to 15 mg/&
from both units during the first hours of operation and about 5 after
overnight; phosphate averaged 0.4 mg/£, and was in all cases less than
1 mg/£; and iron was about 0.1 mg/£.
The slope of -0.5 indicated in Fig. 9 requires comment. For constant
concentration of filterable material, if flow resistance of the filter is
negligible compared to that of the cake; if a constant fraction of filter-
able material contained in solution passing through the filter is deposited
in the cake; and if there are not changes in intrinsic permeability of the
19
cake, a slope of -0.5 is predicted for log flux vs log time (see also
Ref. 27). A slope steeper than -0.5 for constant water recovery, as in the
no-cross-flow case of Fig. 9, probably indicates a change in the cake with
time, perhaps from compaction or from change in the particle size from attri-
tion in pumping; such a change may account for the less dramatic effect of
cross flow than in Fig. 12, Ref. 26. In many subsequent figures, a slope
of -0.5 is indicated as a reference; it is not completely appropriate when
water recovery is not constant (i.e., the concentration of the solution be-
ing filtered increases with time). Fig. 10, a picture of test sections after
operation for 3 1/2 hours, illustrates qualitatively that fouling is less
at 4.6 m/sec (15 ft/sec) than at lower circulation velocities.
22
-------
100
50
10
5
0.5
i i i | i i 1 1|
o ^ '•
o .
o
SLOPE = -0.5
^
• 15 ft/sec (4.6 m/sec)
o No cross-flow
i i i 1 i i 1 1 1
0.10
1.0
10.0
HOURS
1000
500
OJ
100
50
10
100
Fig. 9. Fluxes in filtration of primary sewage effluent
with and without cross-flow.
(37 ppm Fe(III); pH 6.5; 37°C; fire-hose jackets,
40 psi [2.8 bar]; outside)
Q.
o>
23
-------
UNUSED
CIrc. Vel. = 15 ft/sec (4.6 m/sec)
Flux at 210 min = 345gpd/ft2
(14 m/d)
Circ. Vel. = 7.5 ft/sec (2.3m/sec)
Flux at 210 min = 255 gpd /ft2
(10.3 m/d)
Circ. Vel. = 3.75 ft/sec (1.1 m/sec)
Flux at 210min = 165 gpd /ft2
(6.7 m/d)
Fig. 10. Cross-flow filtration of primary sewage: effect
of circulation velocity.
Feed
T.O.C. 91 ppm
P04£ 40 ppm
Turbidity 55 JTU
Fe(ni) 98 ppm
(0.00177 M)
Filtrate
8 to 13 ppm
0.6 ppm (ave.)
0.88 JTU (ave.)
< 0.1 ppm (ave.)
-------
Figure 11 summarizes comparative laboratory tests with three
different cross-flow rates. Fluxes were higher at higher circulation
velocities.
Table 1 collects results for several runs in which two test sections
were operated simultaneously under conditions identical except for circu-
lation velocity. In every case, fluxes were higher at the higher circula-
tion velocity.
It is clear from these and from other results that energy spent in
pumping feed past the filtering surface effects higher fluxes.
Optimum circulation velocity is determined by many factors, includ-
ing energy cost, equipment cost, specific characteristics of feed, and
the premium placed on conserving space.
Effect of Pressure
For a fixed flow resistance, flux should increase linearly with
pressure difference across a filtering element. From many observations,
some of which are documented elsewhere in this report, it has appeared
that the fluxes attained in cross-flow filtration follow no such simple
pattern (see, for example, Fig. 3 of Reference 18). Indeed, independence
of pressure (so long as the filter cake has not previously been exposed
to higher pressure in previous operation) seems to be a better first
approximation. For example, when outside flow elements have been operated
simultaneously with the inside flow unit pictured in Fig. 5, pressures
2
vary typically at 4.6 m/sec (15 ft/sec) from 275,000 N/m (40 psi) to
255,000 N/m2 (37 psi) with the outside flow unit and from 275,000 N/m2
2
(40 psi) to 35,000 N/m (5 psi) over the 5.5 m (18 ft) length of hose in
the inside flow unit. Average fluxes for the two units usually are not
greatly different, however (see, for example, Fig. 12). In some cases,
flux through the inside unit has been higher, in spite of the lower
average pressure of the feed in it.
The ORNL pilot plant (see Equipment and Experimental Methods) is
admirably suited for testing pressure dependence; there is a large pressure
drop through the system and product streams from separate modules can be
individually monitored. Since only a fraction of the volume passing
through the system in a single pass is filtered, experimental conditions
25
-------
Table 1. EFFECT OF CROSS-FLOW VELOCITY ON FLUX IN FILTRATION
OF EFFLUENT FROM PRIMARY TREATMENT OF SEWAGE
(275,000 N/m2)
(40 psi)
Fe(III) addition
(mg/JL)
35
40
40
50
70
115
Q
ft/sec m/sec
7.5 2.29
10 3.05
15 4.57
Circulation
velocity (ft/sec)a
7.5
15
10
15
7.5
15
7.5
15
7.5
15
10
15
Flux
2 hr
225
370
125
190
355
710
240
390
185
290
270
455
(gpd/ft^)b at
24 hr
95
145
80
165
125
165
75
110
60
110
120
190
1 gallon (US) per day/square foot (gpd/ft2)
= 0.0407 meter/day (m/d)
= .00283 cm/min
26
-------
1 I I I
ro
1000
CM
-o
Q.
O>
X
_J
u_
100
i i
ITT I I
10
I I
I I I I
I .... I
4.6m/sec
15 ft/sec
2.3m/sec _
7.5 ft/sec -
1.14 m/sec
3.75 ft/sec _
. I ... i
50
40
30
20 ;o
E
10
100
1000
MINUTES
Fig. 11. Cross-flow filtration of primary sewage.
(Fe(III), ppm/TOC = 1.1; 30 psi [2.1 bar])
-------
00
10'
LU
X
Q.
O>
!5
or
e>
LJ
0
EXT. PRESS. UNIT
COIL
_L
10
40
20 30
TIME / (hours)
Fig. 12. Cross-flow filtration of sewage with activated carbon,
(Aqua Nuchar A, Concentration 20 g/£, pH4)
400
100
50
10
50
-------
in successive modules are essentially the same, except for average pres-
sure. Figure 13 presents results of a test carried out with six modules
23
on stream. It can be seen that fluxes have no simple dependence on
pressure, and tend to vary less with pressure with increasing time of
operation.
It may be that the filter cake becomes compressed when exposed to
higher pressure, though, if so, the tighter layer is not reflected in any
apparent variations in product quality from successive modules. An alter-
native explanation is that fluxes are much higher right after startup at
higher pressure and quickly build up a thicker flux-limiting cake than is
deposited on the surfaces exposed to lower pressures.
The behavior is similar to that sometimes observed in ultrafiltra-
tion of proteins, in which independence of flux with pressure is
approached at higher pressure ranges of operation. The limiting flux
increases with the stirring intensity. However, models currently used,
based on diffusive transport of particles away from the membrane surface,
28
do not seem adequate to explain observed fluxes.
Pumping solution past filtering surfaces consumes energy. Early
results have indicated that fluxes obtained in operation at low pressure
and low circulation velocity, though not as high as in the range of opera-
tion of most tests reported here, might be of interest. The sharp rise in
cost of energy in recent years has led us to reexamine briefly performance
at low circulation velocity.
In a test at the municipal plant, two test sections were operated
simultaneously under conditions identical except for pressure and circula-
tion velocity. One unit, in common with many experiments reported here,
2
was at 275,000 N/m (40 psi) and 4.6 m/sec (15 ft/sec) circulation velocity;
2
the other was at 69,000 N/m (10 psi) and 1.5 m/sec (5 ft/sec). Fluxes
for the higher pressure and circulation velocity were two to three times
as high (Fig. 14). An impression of the significance can be inferred from
Fig. 15. The computed values of pumping energy as a function of circula-
tion velocity and flux in the figure, taken from Ref. 29, apply strictly
only to the configuration assumed, but the relative values are nevertheless
informative. A flux obtained at 1.5 m/sec (5 ft/sec) would have to be
over twenty times as high at 4.6 m/sec (15 ft/sec) for the same kwhr
29
-------
bar
1000
S 500
x
TIME FROM START
1/2 hr
5 hr
MODULE NO. (10)
i I
RECOVERY
69%
0
20 40 60
AVERAGE OPERATING PRESSURE (psig)
Fig. 13. Flux us average pressure in cross-flow filtration
of ORNL primary sewage (Ref. 23).
(15 ft/sec [4.6 m/sec])
30
10
80
-------
-a
x.
E
x
ID
50
40
30
20
10
2
0.1
2.8 bar-4.6m/sec
40psi -15ft/sec
10 psi- 5 ft/sec •
0.7 bar- 1.5 m/sec
0.5
1.0
HOURS
Fig. 14. Fluxes for different combinations of pressure
and circulation velocity.
(Firehose jackets, ext. flow; pH 6.0; water recovery 0-62%)
1000
500
CVJ
CL
0>
100
75
50
J L
50
-------
m/d
50 75
(O
X
JC
10
-0.05
50 100 500 1000 2000
PRODUCTION RATE (gpd/ft2)
Fig. 15. Approximate pumping energy requirements in cross-flow filtration
(1 in x 20 ft [2.5 cm * 6.1 m] tubular filters)
32
-------
pumping energy consumption per kilogallon. Filtrate quality was essenti-
ally the same for the two test sections in Fig. 14, average (for the seven
samples analyzed) TOC for the higher velocity being 15 ppm, and for the
lower, 16 ppm. Average phosphate for both was 0.2 ppm.
Two pertinent tests were carried out at the ORNL plant, though circu-
lation velocity in the first was not outside the usual range of work
reported there. In it, operation with the first three pairs of modules
was at 4.6 m/sec (15 ft/sec). A fresh pair was valved in at twenty-four
hour intervals, and the system was adjusted so that the entrance pressure
2
for the fresh pair was 275,000 N/m (40 psi) (Fig. 16). Water recovery
was brought to 75% the first day, and was maintained there subsequently.
On the fourth day, a fourth fresh pair was introduced in the same manner,
except for a drop in circulation velocity to 3.6 m/sec (12 ft/sec) because
of limitations of pump capability. After a short time (1 hr.), circulation
velocity and pressure were lowered and the last pair of modules was valved
2
in. The pressure range across it was from 150,000 N/m (22 psi) to
2
55,000 N/m (8 psi), and the circulation velocity through the system was
3 m/sec (10 ft/sec). The average fluxes of the module pairs over the
first 24 hours onstream are listed in Table 2. It can be seen that in
operation at lower pressure and circulation velocity, flux was not greatly
less than the average of the first three modules, and was actually higher
than for the pair, operated at higher pressure and circulation velocity,
valved in a day earlier.
Table 2. AVERAGE FLUXES FOR THE FIRST 24 HOURS OF OPERATION
Modules
1 and 2
3 and 4
5 and 6
9 and 10
Water
rec.(%)
0-75
75
75
75
Circ.
m/sec
4.6
4.6
4.6
3.0
, vel.
(ft/sec)
. 15
15
15
10
Pressure
103N/m2 (psi)
275-180
275-180
275-180
150-55
40-26
40-26
40-26
22-8
Av.
m/d
8.3
7.5
5.3
6.1
flux
(gpd/ft2)
204
184
130
149
33
-------
LO
Pressure
1000
*U 800
x
Q.
60O
•b
X.
^
a. 400
200
0
Circ.Vcl.,ft/sec
H20 Recov.,%
II
^— psi bar
o 40-26 2.8-1.8
_
-
k
^^
^^^
- \^
^"*^-x>o
15 {4.6m/sec)
0 1
ii.i.
I '
psf bar
• 40-26 2.8-1.8
o 54-40 3.7-2.8
t
1
1
i
t
*v
X
^
" • o— t
~75~— *
1,1.1
1 ' 1
psi bar
A 40-26 2.8-1.8
• 54-40 3.7-2.8
o 68-54 4.7-3.7
t
1
!
1
*i- "
°12-X
1
i , i
I I
psi bar
* 22-8.0 1.5-0.6 "
A 37-22 2.6-1.5
A 51-37 3.5-2.6-
• 66-51 4.6-3.5
o 80-66 5.5-4.6-
-
i
i
i
I
n —
L
g^r=i=it™ |
b-J _.*....>
fr — ii^5 — -8
1 10 (3m /sec)
,i,i,
40
30
20
10
10
20
30
40
50
60
70
80
9O
10O
HOURS OF OPERATION
Fig. 16. Fluxes of aged and fresh pairs of modules.
(pH % 6)
-------
In the second test, all modules were operated from the beginning at
1.5 m/sec (5 ft/sec), and the pressure through the system varied from
2 2
275,000 N/ra (40 psi) to 28,000 N/m (4 psi) Fig. 17). Composite average
2
flux for the first 24 hours was over 2.4 m/d (60 gpd/ft ), and fluxes
for some of the modules at low pressure were higher than for those at
higher pressures.
In agreement with other observations of the effect of pressure on
product quality, no significant differences in TOG or phosphate in filtrate
from modules at the high and low pressure ends of the system were found.
In the run summarized in Fig. 16, most filtrate TOC values fell between
5 and 10 mg/A with a few being higher or lower; phosphates were 0.1 to 0.5
mg/£. In the run of Fig. 17, average TOC for 38 analyses was 7.6 mg/A.
These measurements cannot establish conditions for an economic
optimum, which will vary as relative energy and capital costs vary.
However, the tests cited strongly suggest that future work should pay more
attention to low-circulation-velocity, low-pressure operation than we have
accorded it so far.
Effect of Water Recovery
Fluxes might be expected to decline much more rapidly when the filters
are exposed to solutions from which a large fraction of the water has been
removed, since both impurities and added iron or aluminum are at higher
28
concentration. Experience, however, has been different. No great dif-
ferences, over scatter, in fluxes after the same period of operation have
been apparent in runs at least up to 90% water recovery.
The pattern, which was observed generally during the study, is illus-
trated by results of several runs at the ORNL pilot plant. Scatter in
fluxes during the early stages of different runs is substantial (Fig. 18)
but is not clearly related to the degree of recovery. In some runs,
fresh modules were substituted for adjacent modules (pressure range over
the modules would therefore be the same) after a certain water recovery
was attained. It can be seen in Figs. 19, 20, and 21 that fluxes were
not greatly lower for the modules exposed from the start to the concen-
trated solution than for those initially exposed to dilute feed.
35
-------
too
80
OJ
•o
a.
o»
60
x 40
ID
20
1 '
1 - 1
1111-
A
A
A
A
T
o
o o
%H20 RECOV.
_1_
' '
©
fcS)
-I 1 L.
2s
0.1
Fig. 17.
0.5 1.0 5 10
HOURS OF OPERATION
50
100
Low-pressure - low-circulation velocity cross-flow filtration
of primary sewage.
(65 ppm Fe; pH ^ 6; 5 ft/sec [1.5 m/sec])
Modules psL
1 a 2 40-
3 a 4 33-
5 a 6 26 -
7 a 8 18 -
9 a 10 11-
Composite
33
26
18
11
4
bar
2.8-2.3
2.3- 1.8
1.8- 1.2
1.2-0.8
0.8- 0.3
36
-------
40
30
T3 20
•X.
^x
E
X
^^^
^j j j /^
_J
L_
5
C
_ i i 1 i i —
Water Recovery
- 0 O.A.O 0-75%
- A^^^ G 0-50%
V a.
_D A
0 ^.
f o A^-.^
1111
) 1 23456
1000
500 ^
«*-
\
T3
Q.
100
k
HOURS
Fig. 18. Comparison of fluxes during initial stages of several runs
at ORNL pilot plant.
(15-80 psig [1,0-5.5 bar]; 15 ft/sec [4.6 ra/sec]; 50-70 ppm iron)
37
-------
u>
00
TD
E
^
x
•
__j
u_
40
30
20
10
5
2
_ i i i | i i i i i i i | i i i 1 1 i i i | i i —
X. ' I
7 ^-o o 1 :
Xfc. 9
0 ^ 5
^X A —
- A ^\ ee
- A A "^ v. ®
^^
n x»
— n "^ ^^^SLOPE =-05
D 50% recovery v.
- A no precoat; 50% recovery
O 0 + 50% recovery; 4- 6-hr interruption; ©continue at 50% recovery
i i i 1 i i i i
1000
500
CM
•t-
H-
Nv
-o
o.
o>
100
50
0.1 0.5 1 5 10 50
HOURS AFTER MODULES ON STREAM
Fig. 19. Fluxes, precoated and non-precoated modules.
(60-80 psi [A.1-5.5 bar]; 70 ppm iron; 15 ft/sec [4.6 m/sec])
-------
T3
\
E
40
30
20
10
5
2
- i "i — i rrn 1 1 — i 1 i 1 1 1 1 1 — i i i i — i
L ° -
- cT\ o .1
^ttv. °
* v^ •
-
On "^ -^
\ ASLOPE = -0.5
1 1 — i — 1 i i i 1 1 1 1 — i — 1 i i i i 1 i i i 1 i i i t
1000
500
0
100
50
0.1
0.5 1.0 5 10
HOURS MODULES ON STREAM
50
Fig. 20. Flux decline at different pressures and recoveries.
(15 ft/sec [4.6 m/sec]; 70 ppm iron)
Recovery
0-75%
75%
Pressure
60-80
4.1-5.1
15-30 psig
1.0-2.1 bar
-------
40
30
20
T3
E 10
2
0.1
i i i 1 1
*
O 0 +90% Recovery
A 90% Recovery
1
i i
SLOPE = -0.5
i i i 1 i i i t
1000
500
CVJ
TJ
O.
100
50
0.5 1 5 10 50
HOURS AFTER MODULE ON STREAM
Fig. 21. Fluxes at high water recovery.
(35-55 psig [2.4-3.8 bar]; 15 ft/sec [4.6 in/sec]; 40 ppm iron)
-------
The insensitivity to pressure and water recovery implies that in
projecting performance, variations in flux from pressure drop along the
system and from increased water recovery can be ignored to a first
approximat ion.
jSffect of Temperature
Temperature effects appear to be mainly on fluxes, which vary roughly
as the inverse of viscosity, after filtercakes are stabilized. Figure 22
summarizes results of two temperature scans made at different times in a
run. The earlier results were taken as fluxes were declining rapidly, but
the departure from dependence on fluidity involves more than flux decline,
since the 60°C point was the next to last measurement in the scan, just
before a repeat at 33°C. However, fluxes measured the next day paralleled
fluidity variations.
The only filtrate analyses available for this run were taken at 60°C
the first day. They seemed• fa'irly normal, though TOG was somewhat higher
than usual: TOG, 20 mg/fc (int),,19 (ext); phosphate, 0.5 mg/£ (int),
0.4 (ext); Fetotal, 0.06 (int), 0.01 (ext); turbidity, 1 JTU (int)»
<0.2 (ext). The run was carried out at pH about 7, and primary effluent
feed TOG ranged from 56 to 89 mgM and phosphate from 25 to 43 mg/£.
EFFECT OF ADDITIVES
Removal of P(V) by Hydrolyzable Ion Addition
An important purpose of the addition of Fe(III) and Al(III) salts
is removal of phosphorus. Since phosphorus in sewage can occur in many
chemical forms, prediction of necessary additive levels and of removal
30
efficiency is complicated. In addition, as one would expect, residual
phosphate is affected by pH, presence of other ions, and the completeness
of solid-liquid separations. Increased removal from acidic solution
caused by presence of sulfate and from basic solution by presence of
Ca(II) was attributed to agglomeration of finely dispersed ferric
31
phosphate precipitates.
Figure 23 summarizes results of tests of removal of P(V) at the
Oak Ridge municipal plant. Ferric chloride was added to the initial
41
-------
60
20
10
DEGREES C
50 40
30
Int.
* £ 4-6 hrs, 50-60% Recovery
•2 29-31 hrs, 86% Recovery —
2.9
3.0
3.1
1
TK
x 10
3.2
+3
3.3
Fig. 22. Effect of temperature on flux.
(15 ft/sec [4.6 m/sec]; 50 ppm Fe(III))
1000
500 3
u.
a:
LU
100
3.4
42
-------
30
u- 20
E*
Q.
CL
LJ
§10
CL
O
I
Q.
0.5 1.0 1.5 2.0
MOLARITY PI/ MOLARITY Fe(HI)
2.5
Fig. 23. Dependence of phosphate in filtrate on ratio of phosphate
to ferric in feed.
(40 psi [2.8] inlet; pH * 4)
Initial Fe(IH) , 18ppm
o a inside flow
I—I-outside flow
43
-------
charge of primary effluent at 18 mg/£ Fe(III) in one run and at 36 mg/£
in the other, and acidity was adjusted to pH 4 with H SO.. No further
ferric salt was added as primary effluent was brought in to replace
discarded filtrate, though the pH was kept about 4 by acid addition. By
periodic analysis of incoming sewage, with correction for P(V) in the
filtrate, the P(V)/Fe(III) mole ratio in the solution circulating past
the filters was estimated. It can be seen that phosphate in the filtrate
became substantial above mole ratios of total phosphorus to iron of
roughly 0.5; the difference between the two runs are within uncertainties
in concentration. Addition of more iron at the end of one run brought
phosphate in the filtrate again to a low level. There appears to be no
significant difference between removal with circulation outside and
inside the fire-hose jackets.
These results are reasonably consistent with trends of reported
•5Q •}-[
laboratory data. ' Mole ratios of Fe(III) to total P(V) of a little
higher than 2:1 appear adequate for good phosphorus removal. We have not
carried out similar tests at other pH values nor with Al(III), but our
experience at the Al(III) or Fe(III)/P(V) ratios used in this program,
usually higher than 2, indicates that filtrates were usually lower than
3_
1 mg/£ as PO, at the acidities of most runs (pH range 3 to 7).
Effects of Hydrolyzable Ions and pH on Flux
Besides removing phosphorus compounds as solids, Fe(III) and Al(III)
hydrous oxide precipitates also appear to react with other sewage constitu-
ents. Some sewage substances form on porous supports layers of low
32 33
permeability, filtering salt at high pressures. ' In the interest of
maintaining economically attractive fluxes, one would hope Fe(III) and
Al(III) additives would tie up components forming tight dynamic membranes.
Figure 24 summarizes an experiment with aqueous solutions (not sewage)
illustrating this possibility. Cross-flow filtration of a solution con-
taining polyacrylic acid (PAA) and an Al(III) salt was carried out, with
varying ratios of Al(III) to PAA. Change after 2.5 hours from a ratio
moles Al(III)/moles carboxylate groups of slightly under 1 to ca. 0.5
caused a precipitous increase in the rate of flux decline. Return to
the earlier ratio arrested decline, but did not restore flux. A further
44
-------
- 1000
- 500
- 100
o>
£
o
o
£
(ft
-------
increase in metal ion/polyelectrolyte ratio increased the decline rate.
Apparently there is an optimal ratio of the positively charged to the
negatively charged high molecular weight species.
Behavior of sewage is qualitatively similar. Figure 25 compares
fluxes obtained at the Oak Ridge municipal plant with three solutions:
primary effluent containing no additive; water and added ferric salt
(10 mg/£ Fe(III)); and primary effluent with added ferric salt (ca.
50 mg/£ Fe(III)). Fluxes with sewage feed are much higher when Fe(III)
has been added. More surprisingly, fluxes with primary effluent and
Fe(III) are somewhat higher than with water and Fe(III), even though the
concentration of iron was less in the water solution.
What additive level is optimal for flux is in doubt. The answer
from results discussed here appears to be that there is a rather broad
range above that necessary for phosphate removal in which flux seems
relatively insensitive to Fe(III) and Al(III) concentration. Filtrate
quality also appears not to vary much with additive level. These trends
disagree with preliminary results obtained in laboratory experiments,
reported earlier (Fig. 6.29, Ref. 34) (see Discussion).
We illustrate for Fe(III) with results from two sets of runs, each
run of approximately 24 hours' duration and carried out under similar
conditions, except as noted. The test sections were precoated with
asbestos fibers, and the sewage effluent was processed by the cycle out-
lined in Fig. 1, iron being introduced as FeCl~. On replacing the pre-
coating slurry with treated sewage, the initial filtrate is turbid, but
after a few minutes becomes clear. (Detailed histories of typical runs
can be seen in the first stage results of Figs. 45-47, to be discussed
in connection with two-stage treatment.) In both sets of runs discussed
here, filtrate was discarded. The water recovery attained varied with
flux and with area of filtering surface; filtrate volumes ranged between
64 and 96% of the volume of sewage effluent taken in.
The first set was obtained between February and May of 1973.
Figures 26 and 27 give the results as a function of Fe(III) concentration
observed at two hours and the average over 24 hours, respectively;
approximate experimental conditions are indicated in the figures. Although
there is considerable scatter, there does not appear to be any clear
46
-------
100
50
40
E 20
x
i 10
Q.
Q.
~1 1—I—I—I I I
1 1—I—i—r~TT
O
o
Fe(JII)~50ppm
PRIMARY SEWAGE
Fe(IH) ~10ppm
WATER
PRIMARY SEWAGE , NO
i i
2000
1000
500
400
300
200
100
50
15
10
5
n
— —
- • —
o
— —
i i j — 1 i i i 1 1 — i i i I i i . ,
0.1 0.5 1.0 5 10
TIME, hours
Fig. 25. Cross-flow filtration of primary sewage
with and without Fe(III) addition.
(Fire-hose jackets, external flow; 15 ft/sec [4.6 m/sec];
40 psi [2.8 bar])
47
-------
40
6 30
CO
or
ID
o
if, 20
10
0
0
0
<"
i 1
1
1000
<\J
500
•a
a.
100
50 100 150 200
AVERAGE Fe(HI), ppm
250
300
Fig. 26. Fluxes at two hours, cross-flow filtration of sewage effluent.
(Fire-hose jackets, internal and external flow; 40 psi [2.8 bar] at
inlet; 10 to 15 ft/sec [3 to 4.6 m/sec])
• pH 4.0 Primary Sewage
o pH 6.0 Primary Sewage (after air-sparge atpH3)
a> pH 4.0 Secondary Sewage
48
-------
.J"
X
T3
Q.
-
JI
CJ
o5
>
o
0>
^
X
ID
1
•^
u_
oT
o
4—
IT
•MB*
-|
O
h-
^«*.
yvjv
400
300
200
100
0
20
15
10
5
n
I i r " • —
o
-
o -
0
e 0
ID O
^f ^^
9
• A * •
0
9 m ^
*
-
i 1 i
~ • °
•
- *
O i i i
=120
-10
100 200 300
AVERAGE Fe(IH)f ppm
Fig. 27. Correlation of flux and filtrate quality with Fe(III) addition,
(Fire-hose jackets, internal and external flow; 40 psi [2.8 bar]
at inlet; 10 to 15 ft/sec [3 to 4.6 m/sec])
• pH~4.0 Primary Sewage
o pH^ 6.0 Primary Sewage
(Air-sparged at pH 3)
a> pH~4.0 Secondary Sewage
49
-------
influence of the Fe(III) concentration on flux or filtrate TOG under
these conditions. From these results, it appears that, with 24-hour
2
regeneration cycles, average fluxes of 8-12 m/d (200-300 gpd/ft ) might
be attained. There does not appear to be a significant difference
between fluxes obtained with primary effluent and fluxes obtained with
secondary effluent. Organic carbon in the filtrate, however, is
lower with secondary effluent as feed. In this set average fluxes obtained
in operation at pH 4 seem to be somewhat less than in operation at pH 6
(after air sparging at pH 3), though these results do not allow a defi-
nite conclusion because of the large scatter. The principal disadvantage
of the filtering at low pH is that the filtrate is contaminated with iron,
frequently several mg/Jl.
Conceivably, the amount of Fe(III) relative to organic contaminants
in the sewage is a controlling variable. Figure 28 presents the results
of Fig. 27, plotted against the ratio of Fe(III) to the average TOC con-
centration in the primary or secondary sewage effluent feed. Correlation
is no better than with Fe(III) concentration. There is also no apparent
correlation with the average TOC in the feed (Fig. 29), nor with the
average TOC in the filtrate (Fig. 30). (The set of experiments plotted
in Figs. 29 and 30 differs slightly from the set in Figs. 26 and 27.)
The second set of results was obtained September 1973-January 1974.
Figures 31 and 32 summarize these results (not all tests in Fig. 31 were
continued for 24 hours). The main difference between the sets is that
2
24-hour averages in Fig. 32 scatter about 6 m/d (150 gpd/ft ), while
fluxes in the earlier set were higher, those in Fig. 27 clustering around
2
10 m/d (250 gpd/ft ). Closer examination indicates fairly good agreement
between filtrations carried out near pH 4, the mean of the 24-hour
2
averages being a little above 8 m/d (200 gpd/ft ). However, the average
for those of the second set carried out near neutral pH was only 4.7 m/d
2 2
(115 gpd/ft ), in comparison with 14.3 m/d (350 gpd/ft ) in the first
set. The Fe(III)/TOC ratio appears also in the Fig. 32 set to be no more
significant than in Fig. 28 (Fig. 33).
Reasons for the differences are not clear. The pH at which the
neutral-range filtrations were carried out tended to be somewhat higher
in the second set (Fig. 32), but fluxes in more recent experiments in
50
-------
*4—
\
TJ
O.
D>
S2
-C
sj-
k_
0)
•>
O
0>
>
<
•*
X
_J
u_
k.
•»-
^-c
O
h-
0)
*J\J\J
400
300
200
100
20
15
L 10
i
n
O
"
—
o 0
00
00 ®
f
-•• * o
| 0
1 1 1 1 1 1
~ 0°
• £
I ®i i i° i i i
Fig. 28.
23456
AVE.Fe(m) /AVE. FEED TOG
20
10
8
Correlation of flux and filtrate quality with Fe(III)
and organic carbon level.
• pH - 4.0 Primary Sewage
o pH~6.0 Primary Sewage
(Air-sparged at pH 3 )
o pH- 4.0 Secondary Sewage
51
-------
J
H—
TJ
Q.
of
j=.
(NJ
k_
CD
O
O>
<
x~
_l
Lt_
DUU
400
300
200
100
0
I I I I I I I I v I
•
0 0
o *
• 00 0 PH 4.0
O o O • pH 6.0
_ u O
O
—
—
I I I I 1 1 1 I.I
20 30 40 50 60 70 80 90 100 167
- 10
AVERAGE TOC IN FEED
Fig. 29. Correlation of average flux with organic carbon in feed.
(Primary sewage)
-------
OUVJ
OJ
I 400
Q.
-^ 300
OJ
o 200
>
x 100
=>
o
III"
~
0 o
* o
O ° rL O °
0 0
~ O pH 4.0
• pH 6.0
I 1 I
=120
0
- 10
Fig. 30.
5 10 15
AVERAGE TOC IN FILTRATE
Correlation of average flux with organic carbon in filtrate.
20
-------
40
\
30
QL
O
x 20
CVJ
1-
X
3 10
u.
n
i i i i i i
—
*~ ^
• t>* •
0 0 • ° ""
0
— • pH 4.0
OpH 7.0
1 1 1 1 1 1
1000
CVJ
500 \
T3
Q.
O>
100
0 50 100 150 200 250 300
AVERAGE Fe(m), ppm
Fig. 31. Fluxes at two hours, cross-flow filtration of primary sewage,
(Fire-hose jackets, external flow; 40 psl [2.8 bar] at inlet;
15 ft/sec [4.6 m/sec])
54
-------
01
o 200
«VJ.
X
3 100
LU
20
LU
<
5
0
• pH 4.0
O pH 7.0
-a
B
\
1
10
5 E
200
50 tOO 150
AVERAGE Fe(fll), ppm
Fig. 32. Correlation of flux and filtrate quality with Fe(III) addition,
(Fire-hose jacket, external flow; 40 psi [2.8 bar] at inlet;
15 ft/sec [4.6 m/sec])
55
-------
(A
C\J
0^200
I*
-------
the pH range of the first set (including results at the ORNL plant in
Figs. 18-21} have been more consistent with those of Fig. 32 than with
Fig. 27. Average operating temperatures of the lower-flux set were
slightly higher than with the other, the wrong direction to explain the
discrepancy.
One difference in operation was that the liquid level maintained in
the feed tank was lower in the set of higher flux (Fig. 27) and there
consequently was considerable foaming; conceivably flux-limiting con-
stituents could have been extracted in the foam. However, in a later
test: carried out under similar conditions, with foam removed, the flux
2
was lower than in either set, about 10 m/d (250 gpd/ft ) at 2 hours. It
is perhaps worth noting that the foam did concentrate' constituents; foam
collected over the first eight hours of operation contained 530 mg/& TOC
and 355 rng/A Fe(III), and over the last two hours, 480 mg/£ TOC and
600 mg/& Fe(III); for comparison, the circulating concentrated solution
contained 180 mg/£ TOC and 170 mg/£ Fe(III) at termination of the experi-
ment.
In this connection, we carried out a pH scan by base and acid addi-
tions after overnight operation at pH ^6. The results (Fig. 34) indicate
no strong effect of acidity on flux, at least with this presumably sta-
bilized filtercake. In Fig. 35, the early stage of this run is compared
with the same stage of the subsequent run, for which the conditions were
similar, except that pH of operation was about 7. Fluxes were very simi-
lar, differences becoming apparent only after increase of water recovery
was started in the first run by discarding filtrate (the pH 6 results in
this figure were also included in Fig. 14.)
The relative independence of flux on pH is contrary to preliminary
observations reported previously, in which fluxes increased sharply with
increasing pH. We later found that when the results in Fig. 6.23 of
Ref. 35 were obtained feeds were contaminated with copper, sometimes tens
of ppm, coming from copper cooling coils in the feed tank during exposure
to acidic solutions in the initial addition of ferric salts. Since this
pattern of flux dependence on pH disappeared when the acid part of the
cycle was relocated to stainless steel tanks, we attribute it to copper
contamination. The copper cooling coils were replaced.
57
-------
Ui
CD
100
CM
80
T>
Q.
_J
u.
60
40
_ SEQUENCE NO.-
• 1,10
9 4
8
o INT. FLOW
• EXT. FLOW
1
7
8
PH
Fig. 34. Effect of acidity on cross-flow filtration flux.
(Primary sewage, 55 ppm Fe(III); inlet pressure 40 psi [2,8 bar];
""15 ft/sec |>4.6 m/sec]; 66% water recovery; measurements
between 21 and 26 hours from start)
-------
TJ
£
*»
x
_j
70
60
50
40
30
20
10
I I
0.1
£H_ Hater Recovery
O 7 0%
• 6 0-38% , t Start Concentrating
_L
I I I I
0.5
1.0
HOURS
J 1 L
5.0
1000
500
CVI
•a
a.
o>
10
100
Fig. 35. Comparison of initial fluxes in operation at pH 6 and pH 7.
(Primary sewage; fire-hose jackets, ext. flow; 50 ppm iron, introduced
as ferrous sulfate; 40 psl [2.8 bar]; 15 ft/sec [4.6 m/sec]; 30°C)
59
-------
With respect to differences in the sets reported in Figs. 27 and
32, we can advance only the unsatisfying suggestion that there was an
unknown difference in the composition of the sewage effluents during
the periods of the measurements.
We have carried out a similar, though less extensive, study with
Al(III). The results are summarized in Figs. 36-38. (Since the atomic
weight of aluminum is about half that of iron, the molar AI(III) concen-
tration corresponding to a given ppm of Al(III) will be about the same
as the molar concentration of Fe(III) at twice the ppm.) There appears
to be a trend to higher fluxes with increasing Al(III) concentration,
but the scatter is so large that a definite conclusion is not possible
on the basis of the few runs here. Fluxes are closer to those reported
in the second set with Fe(III), presented in Fig. 32, than to the other
set (Fig. 27).
Use of Lime for Neutralization
Per equivalent of base, CaO is much cheaper than NaOH, and if there
is no penalty in performance, it would presumably be the choice for
adjusting sewage treated with hydrolyzable ions to the pH selected for
filtration. Most of our experiments were carried out with NaOH, since
we had no equipment for automatic addition of solids, necessary for
unattended overnight operation, unless filtrates were recycled to the
feed tank (0% water recovery). However, several short-term runs allow
a comparison between fluxes obtained successively with NaOH and with CaO
neutralization (Table 3).
Scatter was great, but there appears to be no consistent flux advan-
tage from the use of one base or the other. There was an indication that
fluxes after lime neutralization were higher when the other reagents were
added as chlorides, rather than as sulfates. Simultaneous runs were made
with the two loops in the mobile unit to check this point with two batches
of primary effluent taken at the same time, one treated with chloride and
one with sulfate reagents (Fig. 39). There did not appear to be any
significant differences in flux, and it appears the variations in Table 3
resulted from random day-to-day changes in sewage. We conclude tentatively
that performance after neutralization with CaO is as satisfactory as after
neutralization with NaOH.
60
-------
Table 3. COMPARISON OF FLUXES AFTER NEUTRALIZATION
BY NaOH AND BY CaO
(40 psi [275,000 N/m2]; 15 ft/sec [4.6 m/sec])
Fe(III)
mg/£
100
100
100
100
100
100
100
56
9
75
50
59
59
50
48
54
37*
Base
added
NaOH
NaOH
CaO
CaO
CaO
NaOH
NaOH
NaOH
CaO
NaOH
CaO
NaOH
NaOH
NaOH
CaO
CaO
NaOH
Sulfate
added?
Yes
Yes
Yes
No
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
Yes
Approx.
pH of
filtration
Jan. - Feb. 1972
6
3.5
6
6
6
6
6
Aug. - Sep. 1973
4
7
4
6 to 7
4
Oct. 1974
7
7
7
7
5 to 6
% Water
recovery
0
0
0
0
0
0
0
50
75
75
75
75
65
70
76
75
79
Flux
m/d
at 3.
35
19
22
49
57
17
32
at 3
30
10
17
13
19
at 6
7
7
6.5
10
6.5
gpd/ft2
5 hrs
850
475
530
1200
1400
425
780
hrs
740
240
425
320
475
hrs
180
180
160
250
160
Al(III) additive
61
-------
X
20
p*
^0
.c
OJ
% 10
LL
LU
CC 15
Gig:10
o
0 5
0
• pH 4.0
O pH 7.0 (air sparged at 4)
O
O
-
~ O
II 1 1
o
1 1 1 1
) 10 20 30 40 51
-500
M
T3
O.
100
AKffl), ppm
Fig. 36. Flux at two hours as function of Al(III) addition
(IOC measured at 3 hrs)
62
-------
CSJ
i
"I.
i
o
^
CVJ
^
X
LL
uf
<
£
?|
•
UJ
§
200
150
100
50
15
•w
••
5
n
• pH 4.0 _
O pH 7.0 (air sparged at 4.0) °
O
•
—
A ^
—
0
-
•
-
1 1 1 1
I • 1
0
. • • o • °
1 1 1 1
5
TJ
£
o
\j
Fig. 37.
10 20 30 40
AVERAGE AI(HI) , ppm
Flux and filtrate quality with Al(III) addition
to primary sewage.
50
63
-------
41 ^uu
•o
a.
8?
o
.c
n
0
O '
• pH 4.0
O pH 7.0 (air sparged at 4)
—
O
~" * • •
-
£
-
1 1
0
- -
o
•
1 1
5
•o
E
0
0 0.5 1.0 1.5
AVE. Al(ni), ppm /TOC IN FEED, ppm
Fig. 38. Flux and filtrate quality as function of ppm Al(III)
addition per ppm TOC in primary sewage.
64
-------
ou
40
30
20
•o
E
x* 10
ID
U.
5
100
E
Q.
O.
O
0
L-»
10
5.0
0.
— —
1 8 :
- 8 8 -
D
° ° 8
r 0 CHLORIDE ADDITION
C SULFATE ADDITION
: ••FEED
..!.... , ,,!,,,,
—~ —
• • • • I
0 -
n 0 (fv D
O n n
D *^
.
r" o ~~T
o
i i i i i 1 1 1 i i i i i i 1 1
1000
500^,
CM
T3
Q.
O>
A l~\f\
1OU
0.5 1.0 5.0 10
OPERATING TIME (hrs)
Fig. 39. Comparison of cross-flow filtration of primary sewage
neutralized with CaO after addition of iron sulfate
and iron chloride.
(pH 7.5-8; 30°C; 40 psi [2.8 bar]; 15 ft/sec [4.6 m/sec];
fire-hose jackets; external flow; 0% water recovery)
65
-------
Quality of Cross-Flow Filtrates from Effluents Treated with
Hydrolyzable Ions
Filtrate and primary sewage were sampled for analysis at irregular
intervals of a few hours, mostly during the normal work day. Analyses
of filtrates taken during evening and nights were relatively rare; those
available from off-hours did not appear to depart greatly from the pattern
reported here.
TOG - Typical total organic carbon in filtrates of feed treated with
Fe(III) or Al(III) are recorded in Figs. 27, 32, and 37. From these and
other results, it appears that with primary effluent as feed, TOG can be
reduced most of the time to 10-15 mg/£ by this treatment. Over the range
studied, there does not seem to be much influence of additive concentra-
tion on product quality in this respect.
Removal of TOG by cross-flow filtration appears to be somewhat higher
than by the activated sludge secondary treatment of the Oak Ridge Munic-
ipal Plant. For twenty-five runs in the period January through April
1972, mostly with Fe(III)-treated primary effluent feed, a comparison
was made between TOG of filtrate from cross-flow filtration and effluent
from secondary settler of the municipal plant. The secondary effluent
was sampled about 7 hours after start of filtration, to allow roughly for
hold-up time in the activated sludge system. The average value for
cross-flow filtrate was 11.3 mg/Jl, with a range of 4 to 20, and for the
effluent from biological treatment, 14.4, with a range of 7 to 34.
Average primary effluent was 47.2 mg/Jl and removal of organic carbon
from primary sewage was 76% by cross-flow filtration, in comparison with
69% by the standard secondary treatment. This comparison should be
viewed with some caution. A wide range of variables was tested in the
filtration runs, and the average thus may not represent performance
under optimal conditions. On the other hand, most of these experiments
were short-term, with filtrate returned to the feed tank (0% water
recovery). However, the results are in the range of other values in
this study, and we believe the comparison is meaningful.
In this connection, we note that TOG content of the filtrate seems
generally not to vary significantly with water recovery, i.e., the
66
-------
percentage removal of TOG increases with TOG content of the concentrating
solution being filtered. This may imply that there is a certain fraction
of original TOG completely removed, and a fraction, presumably of low
molecular size and non-adsorbable on hydrous oxides, which passes through
completely with the water. Examples of filtrate TOG during increasing
recovery can be seen in Figs. 45-47, to be discussed later.
We pointed out in connection with Fig. 27 that cross-flow filtrates
from secondary effluent treated with Fe(III) were usually lower, in the
neighborhood of 5 ppra in TOG, than when primary effluent was the feed.
Turbidity - Turbidities of filtrates after the first few minutes of opera-
tion were usually well below 1 JTU, there being for example no values
above this and only three observations above 0.5 JTU in twenty-four runs
during the period the results of Fig. 26 to 32 were obtained. Filtrates
usually were visually similar to tap water. Primary effluent turbidities
varied between 10 and 100 JTU.
Phosphate - With sufficient Fe(III), removal of phosphate is usually good.
In the runs of Fig. 26, in only one analyzed sample was total phosphate
3_
(as PO, ) as high as 0.3 mg/&. In the runs of Fig. 32, removal for some
runs was poorer in several cases; filtrate samples in three runs contained
over 1 mg/£ phosphate. In one case of low Fe (21 ppm) and high average
molar ratio of feed phosphate to iron (1.25), there were 2.4 mg/fc in the
filtrate. In the six other runs, all samples were well below 1 rag/A
phosphate.
With Al(III) additions, phosphate removal was sensitive to pH. At
pH values below 5, several mg/£ of phosphate was found in filtrates, as
high as 40 mg/5- in one case. Above pH 6, phosphate removals were compar-
able to the results with Fe(III) treatment, the content usually being
well below 1 mg/Jl.
Contamination of filtrates by hydrolyzable ion additions - We mentioned
earlier that, at low pH, the filtrate contains iron; probably most pene-
trating the filter is Fe(II), remaining despite air-sparging in pretreat-
ment. Ferrous would be expected to penetrate more than Fe(III), since
it is less acidic, and therefore at a given pH tends less than ferric to
hydrolyze and form polymers and hydrous oxides. Iron removal will depend
67
-------
both on the equilibrium between Fe(II) and Fe(III) species and the kinet-
ics of reactions for reaching equilibrium and for hydrolysis of the
ferric. On account of hydrolysis, the equilibrium for Fe(II)-Fe(III)
shifts toward Fe(III) with increasing pH, and one expects iron removal
to increase with increasing pH.
The effect of acidity is illustrated in a run in which pH was
cycled from 3.5 to 10 and back to 4, with changes at one-hour intervals;
the points represent analyses of samples taken just before pH adjust-
ments (Fig. 40). Primary effluent was brought to 100 ppm.Fe(III) and
pH 3, digested overnight, and the pH was brought to 3.5 just before
starting the run. The analyses in the figure were from a fire-hose
jacket, operated with external flow of 6.4 m/sec (21 ft/sec), with
filtrate recycled to the feed tank (0% water recovery). Total iron in
the filtrate decreased with increasing pH, falling to 0.1 mg/£ or less
at pH of 6 and higher. However, (point 8) reversal of the metal
ion oxidation was illustrated by the fact that the filtrate again con-
tained iron when pH was decreased to 4. On return to low pH slightly
less than half the iron penetrating the filter was Fe(II); in all other
cases, Fe(II) accounted for all but perhaps 1 mg/£ of the iron in fil-
trates. Phosphate was also somewhat more completely filtered at higher
pH. Filtrate TOG scattered between 4 and 8 mg/Jl, with perhaps slightly
better removal at high pH.
In nineteen of the experiments of Figs. 27 and 32 utilizing primary
effluent, the following iron contents were obtained:
pH <5 >5
Average Range Average Range
pH 3.9-5 5.8-7.8
mg/£ Fe in filtrate 3.6 1.5-6.7 0.12 0.03-.22
In steady-state operation at pH >6 with added iron salts, iron con-
tamination of treated water will probably be acceptable for most situa-
tions. If the additive is introduced as Fe(II), however, which may be
desirable because of the solubility of ferrous sulfate in comparison
with ferric sulfate, and the corrosiveness of chlorides, the kinetics of
transformation to ferric becomes of concern. In experiments at the ORNL
68
-------
e
Q.
CL
UJ
<
"o
a:
a
X
Q_
cn
O
I
CL
100
50
10
1.0
0.1 (or less)
e
O Phosphate
D Fetotal
Numbers indicate hours from start
6 7
pH
8
10
Fig. 40. Phosphate and metal ion leakage in cross-flow filtration
of primary sewage as a function of pH.
(40 psi [2.8 bar]; fire-hose jackets)
69
-------
pilot plant, in which the additive was FeSO,, iron contamination of the
filtrate of several ppm has occurred, even in high pH operation. In
these cases, the base to neutralize the feed was added in the tanks from
which the pump circulates it past the filters. Since filtrate contamina-
tion was highest when production rate was high and fresh feed was dravn into
the feed tanks at frequent intervals, it seems clear that limited time
for the oxidation of Fe(II) is responsible.
The time necessary for oxidation of iron introduced as Fe(II) salts
to Fe(III) will depend of course on conditions affecting kinetics, such
as mixing and especially acidity. In one experiment, carried out at pH 6
in the mobile unit at the municipal sewage plant, iron was essentially
undetectable in the filtrate in the first analysis, on a sample taken after
about two hours operation. In another,(also discussed in connection with
Figs. 14 and 35),however (Fig. 41), the filtrate contained several mg/£
even after 6 hours operation, although levels were low the next day. At
higher pH, presumably because of faster oxidation and hydrolysis, good
removals occurred much earlier. In Figure 42, Fe in filtrate from one
of the test sections of Fig. 41 is compared with Fe in filtrate obtained
in another experiment, similar except that pH was 7.
Contamination of filtrate with iron even when added as ferrous can
therefore be avoided with short holdup time, if pH during filtration is
high enough. However, higher base additions are necessary. For example,
in the run at pH 7 of Fig. 42 about 1.3 millimoles of H-SO, per liter of
primary effluent were used,in addition to ferrous sulfate, to bring the
sewage initially to pH 4; 1.6 millimoles of NaOH per liter brought the
pH back to 6 and an additional 1.7 millimoles/liter to pH 7.
Al(III) was also high in the filtrate in operation at low pH. To
cite extremes, in one run at pH ^4, with average Al(III) addition of
32 mg/£, 25 mg/£ Al(III) was found in the filtrate; in another run at
pH 7.8, Al(III) was 0.01 mg/A. For the runs for Fig. 37, with one exception
(a value of 1.5 mg/A Al(III) at pH 7.5), Al(III) in the filtrate was
less than 1 mg/£ at pH > 5, and was several mg/fl. at pH < 5.
70
-------
CIRCULATING FEED
/40psi , 15 ft/sec \
FILTRATE V2.8 bar, 4.6m/sec /
Opsi , 5ft/sec \
bar , 1.5m/sec/
I
FILTRATE
10 15
TIME, hours
Fig. 41. Filtration of Fe(total) with time, after introduction as Fe(II).
(PH 6)
-------
e
Q.
Q.
50
10
5
LU
< 1.0
0.5
0.1
0.05
0.01
pH~ 6
HOURS
Fig. 42. Iron contamination of filtrate as a function of acidity,
(50 ppm Fe, introduced as Fe(II) sulfate; 15 ft/sec [4.6 m/sec];
40 psi [2.8 bar]; 30°C)
72
-------
Nitrogen compounds - We have accumulated little information on the
removal of nitrogen compounds by cross-flow filtration of sewage efflu-
ents .
In one experiment, with 100 mg/£ Fe(III) added to primary effluent
from the Oak Ridge Municipal plant, followed by digestion at pH V3 and
adjustment to pH 6 for filtration, analyses for organic nitrogen were
carried out by personnel of the Environmental Engineering Department of
the University of Tennessee, Knoxville (we are indebted to Prof. W. A.
Drewry for arranging these). Their results indicated 5.6 mg/£ as N in
primary sewage, and only a trace in the cross-flow filtrate. A sample of
secondary effluent from activated sludge treatment taken the same day con-
tained 0.6 mg/£. N. They also carried out chemical oxygen demand analyses
on the same samples and reported: primary, 136 mg/Jl; cross-flow filtrate,
8; and secondary sewage, 56. Our corresponding TOC analyses were primary,
53 mg/£; filtrate, 10; and secondary sewage, 26.
We analyzed samples for three cross-flow filtrations of effluents
from activated sludge treatment and one from primary effluent for total
nitrogen. Results are summarized in Table 4.
Table 4. NITROGEN REMOVAL IN CROSS-FLOW FILTRATION
(pH ^4)
Total nitrogen, mg/H*
Filtrate
Fe(III)
mg/X,
77
69
100
Time from
start ,hrs.
3
24
24
Water
rec.,%
80
96
90
Feed
0.5
2.0
13
Circulating outside
concentrate flow
Secondary Sewage Feed
1.0 0.35
11.6 0.3
13 10
Primary Sewage Feed
inside
flow
-
0.9
11
50
90
30
48
21
22
Kjeldahl, without prior ammonia removal.
73
-------
In another experiment with primary effluent (other aspects were
discussed in connection with Figures 14, 35, and 42) in which iron
(about 50 mg/£) was introduced as Fe(II) and filtration was at pH ^6, the
average of analyses for primary was 20 mg/£ total nitrogen (range 15 to
28) and 14 for filtrate (range 12 to 17), with little difference between
2
the test sections, one operated at 4.6 m/sec (15 ft/sec) at 276,000 N/m
(40 psi), and the other at 1.5 m/sec (5 ft/sec) at 69,000 N/m (10 psi).
Analyses of samples taken 22 hours after the start of the experiment
to be discussed in connection with Fig. 44 with powdered activated carbon
as additive, indicated essentially no removal—20 mg total N in both the
primary effluent and in the filtrate.
No firm conclusions can be inferred from such scattered results. Removal
of inorganic nitrogen compounds is, however, expected to be incomplete by
either hydrolyzable ions or PAC additions ; and if organic nitrogen is
degraded to inorganic by biological action, higher percentage removals
might be expected from primary effluents than from secondary. On the
other hand, one might expect activated carbon to remove substantial frac-
tions of organic nitrogen compounds, and the negligible reduction with
primary effluent is surprising. With either type of additive, nitrogen
removal appears erratic and incomplete.
Bacterial removal - The low turbidities of cross-flow filtrates suggest
that particles of bacterial dimensions are removed, and limited analyses
(carried out by the ORNL Industrial Hygiene Department) indicated that
there probably are substantial reductions of organisms. However, many of
the experiments were at low pH, 4 to 4.5, and the count in circulating
concentrate was usually lower than with primary feed. This may result
from carrying down of bacteria on hydrous oxide, but relative contribu-
tions of filtration and solution conditions unfavorable for organisms
are not clearly distinguished. On the other hand, there was not sterili-
zation of the filtrate side of the filter elements, and in some cases,
sample bottles were not sterilized. It is possible that multiplication
took place in the samples before assay. Table 5 summarizes some observa-
tions.
74
-------
Table 5. BACTERIA CONTENTS OF CROSS-FLOW FILTRATES
Total coliform/mJl
Filtrate
Fe(III)
mg/A
50
50
56
Time from
start, hrs.
70
6
2
Water
rec.,%
88
90
50
Primary
effluent
(pH 4 to
TNTC
1200
12000
17200*
Circulating
concentrate
A. 5)
TNTC
1
120
160*
outside
flow
0.4
0.2
40
<1*
inside
flow
-
<1
<1*
Filtrate
(pH ^6)
50 1
2
6
24
0
8
38
62
Neg
20
20
TNTC
(both outside)
Neg
Neg
Neg
Neg
Neg
Neg
Neg
Neg
TNTC indicates Too Numerous To Count
*
Total colonies/mfc
The measurements at pH 6 were all for a single run, for which the
primary effluent feed count for coliform and for total colonies ranged
from 2000 to TNTC per m£ of primary (measurements on 1 mH after 1 to
100 dilution). Prior to adjustment to pH 6, the original feed had been
maintained at pH -v4 with Fe(ll) addition for about 45 minutes. Perhaps
the exposure to acid conditions accounts for absence of detected bacteria
initially and the low count the first day in the circulating concentrate.
In any case, removal by filtration appears good (measurements on concen-
trate and filtrate on 1 mi after 10 to 1 or 2 to 1 dilution).
Powdered Activated Carbon
Potential advantages of powdered activated carbon (PAC) over granu-
lar were alluded to in the Introduction, and have been discussed in
36
greater length by Cohen. Our tests with primary sewage
although less extensive than with iron salts, show that PAC can be
75
-------
removed from sewage effluents essentially completely by cross-flow fil-
tration; that, with sufficient PAC, high fluxes are achieved, and that
the filtrate quality is good. Turbidity is low, and TOC is usually lower
(typically about half) than in filtrate after Fe(III) or Al(III) addi-
tions to primary sewage. Phosphate is not removed well by PAC. Many
other substances are adsorbed by powdered carbon, though their removal
was not evaluated in this study.
If satisfactory performance could be attained with powdered carbon
additions so low that PAC regeneration was unnecessary, the mode would be
particularly attractive. For reference, with PAC at 10c/lb and with
addition of 100 ppm and throw-away operation, the activated carbon cost
would be about 8$/kgal. However, our experience indicates that, except
in special situations, discard will likely be undesirably costly; fluxes
at low levels of carbon are low.
It appears that the distribution coefficient for the solutes com-
prising the TOC in sewage effluents is not high enough for removal to low
levels in the single-plate operation typical of powdered-carbon contact-
ing, and rather large amounts of PAC are required per volume of effluent.
Fig. 43 illustrates this with laboratory measurements of removal of TOC
from the filtrate from cross-flow filtration, carried out at pH 6, of
primary effluent treated with Fe(III). It appears that reduction of TOC
from vLO ppm to ^5 ppm required about 1000 ppm of PAC. In the tests to be
discussed, high ratios of PAC to feed or to the circulating concentrate
were therefore maintained.
Fig. 44 illustrates typical behavior in cross-flow filtration of
primary effluent with added powdered carbon. In this 14-day run, 20 g/£
of PAC was added to the initial charge of primary effluent, and no more
PAC was added as filtrate was discarded and fresh sewage brought in, until
the last part of the run. The pH was adjusted to about 4. Over the first
day, during which enough primary passed through to bring the mg PAC/liter of
sewage processed to about 500, average fluxes over 24 hours were "x-420
2 2
gpd/ft . Fluxes continued to decline, to below 25 gpd/ft after ten
days, and a backwash caused only a momentary increase. However, a back-
wash coupled with addition of enough fresh PAC to bring the average PAC/
liter total sewage processed (filtrate plus concentrate) to 500 mg, about
76
-------
10
8 —
CL
Q.
• 6 —
o
o
ID
Q
(/)
LJ
cr
4 —
2 —
0
O TOC Remaining after PAC Addition
• TOC Removed per mg PAC
10
— 0.04
— 0.03
— 0.02
o
Q
LJ
CD
o:
O
in
Q
o
o
100
POWDERED ACTIVATED CARBON , ppm
— o.oi I1
0
1000
Fig. 43. Adsorption of TOC on PAC from cross-flow filtrate
of Fe(III)-treated primary sewage.
77
-------
WATER
RECOVERY. %
• I
mgPAC/lfeed 20,000 5OO
99.0
250 500 400
0
50
100
150 200
TIME, hours
250
300
350
Fig. 44. Cross-flow filtration of primary sewage effluent
after powdered carbon addition.
(Fire-hose jackets; « external flow, 40-37 psl [2.8-2.6 bar];
O internal flow, 40-7 psi [2.8-0.5 bar]; 15 ft sec [4.6 m/sec])
78
-------
two days before termination, effected increase in flux to values in the
ranges initially observed at the same PAC level.
Total organic carbon of the filtrate scattered about 5 ppm after the
first day. Phosphates were 20 to 30 ppm, comparable to concentrations of
feed, and turbidities mostly <0.5 JTU.
The results, particularly the flux recovery on addition of fresh
carbon, suggest that maintenance of high fluxes requires a high enough
PAC/feed ratio to tie up flux-limiting organic substituents. It should
be noted that TOG in the filtrate does not completely define the level
of non-adsorbed TOG in the circulating concentrate, since a thin layer
of powdered carbon on the filter can provide in effect more than single-
37
plate separation.
Two-Stage Processing by Hydrolyzable Ions Followed by PAC
Addition of PAC to the cross-flow filtrate from primary effluent
treated with Fe(III) or Al(III), followed by a second-stage cross-flow
filtration, should give a low-phosphate product. In addition, it might
be hoped that fluxes high enough to make the second filtration cheap
might be attained even at low PAC levels. Fig. 45 indicates that higher
fluxes are obtained in a second stage than with primary sewage treated with
a similar low level of PAC. After twenty-four hours operation (pH >6
in this and the two-stage runs) at which time average PAC usage was about
65 mg/liter filtrate, flux of the second stage had fallen to about 8 m/d
2 2
(200 gpd/ft ) (24-hr average M.2 m/d (-^290 gpd/ft )), a reasonably high
value, but not enough to make trivial the cost of the second stage.
Phosphate in the first- and second-stage filtrates were 0.1 ppm or less.
Turbidities of first-stage effluent were 0.3 to 0.4 JTU and of second
stage, 0.15 JTU.
In two other two-stage runs, initial additions of PAC to the first
stage effluent were higher, and fluxes in the second stage were greater.
2
Fluxes at 22 hours were in one about 20 m/d (500 gpd/ft ) at which time
the average dosage was about 250 mg/liter (Fig. 46), and in the other
about 40 m/d (1000 gpd/ft2) for 750 mg PAC/liter filtrate (Fig. 47).
In the last mentioned run, filtrate was recycled to the feed tank over-
night, and final second-stage recovery was only 35%, in comparison to
79
-------
— 1000
-500
CVJ
Q.
Fig. 45. Two-stage cross-flow filtration of primary effluent.
(15 ft/sec [4,6 m/sec])
First Stage, ~85ppm Fe(m)
Second Stage
Primary Sewage
o
A Ext
•f
80
-------
100
50
\
E
x" 10
O
i—i 1—i—i—i—i—r
2 -
Q.
50-H
10
8 5.0
a:
2 1.0
0.5
o
0.1
8
A
8
A
1000
500
O.
O>
100
0 5 10 15 20
HOURS
Fig. 46. Two-stage cross-flow filtration of primary effluent,
(15 ft/sec [4.6 m/sec])
First Stage, ~75ppm Fe(HI)
Second Stage
Primary Sewage
Ext.
81
-------
2 -
o
o
50
10r
5 -
. 1-0
1 5
li-
ft)
f
(O
•a 1.0
(5
-< 0.5
*x
o
e
o»
0.1
0
Fig. 47.
6
12
HOURS
16
3000
1000
500
100
50
7 Start
2nd Stage
1
Recycle
2nd Stage Filtrate
1,1,1,
^Nr
Discards
2nd Stage:
Filtrate •
I
20
24
•o
O.
O>
Two-stage cross-flow filtration of primary effluent.
(15 ft/sec [4.6 m/sec])
First Stage, ~60ppm Fe(HI) °
Second Stage A
Primary Sewage +
Ext
82
-------
65-75% for the runs of Figs. 45 and 46. Second-stage filtrates were low
in phosphate (0.2 ppm or less) and were lower in TOG than first stage
filtrate.
If recovery and regeneration of powdered activated carbon from con-
centrates is feasible, it may be practical to obtain effluents lower in
TOG than from treatment with Fe(III) and Al(III) alone by contacting with
PAG in either single-stage or two-stage cross-flow filtration.
POROUS SUPPORTS, PRECOATS, AND REGENERATION
Most results reported here were obtained with fire-hose jackets.
Although these are attractive filter surfaces, they are by no means the
only option—porous tubes of various materials could be used. In the
earlier stages of the study, a considerable number of tests were made
with screens, mostly of stainless steel but also, in a few cases, of
nylon. They were mounted in external flow test sections, supported by
perforated stainless steel, in a manner similar to that illustrated in
Fig. 2. In most cases, they were precoated with fibrous asbestos or
cellulosic filteraids, though in some tests with fine screens (e.g.,
400 mesh), no filteraid coat was used.
In runs in which test sections of both screen and firehose jackets
were present, relative performances were mixed. Somewhat more care in
forming precoats appeared necessary with screens than with fire-hose
jackets, and, at least with screens coarser than 100 mesh, occurrence of
turbid product on occasion indicated instability of the filtercake.
Backwashing, surprisingly, was frequently more difficult with screens—
when part of the cake had been removed, loss of the pressure hindered
removal of the remainder. Attempts were made to assist backwash by
electrical pulses generating electrolytic gas. Although cake removal was
demonstrated, currents required seemed undesirably high.
With 20 and 40 mesh screens, somewhat higher fluxes were usually
obtained than through fire-hose jackets used in the same tests. For
three runs in which the circulation velocity was the same or higher for
the fire-hose jacket than for 20 and 40 mesh screens used in the same run,
fluxes measured at the same times (after 24 to 40 hours of operation)
83
-------
2
averaged 3 m/d (70 gpd/ft ) for the fire-hose jacket and 5 m/d (120
2
gpd/ft ) for 20 and 40 mesh screens. On the other hand, for ten short
runs (of a few hours) In which a 120-mesh stainless steel screen was
operated at either the same 4.6 m/sec (15 ft/sec) or higher circulation
velocity than a fire-hose jacket (both external flow of feed), measure-
2
ments near the end of the runs averaged V}3 m/d (^800 gpd/ft ) for the
2
fire-hose jacket and ^16 m/d (^400 gpd/ft ) for the screen.
Backwashing is the most obvious regeneration method, when internally
supported filter surfaces are used and pressurized solution is circulated
on the outside. Fig. 48 illustrates in successive short-term runs with
primary effluents (0% water recovery) the restoration of flux, and also
compares the effects of precoats of two different filteraids with no
filteraid. The precoating in this case was carried out at relatively
high circulation velocity, which results in a thin layer adherent when
circulation is stopped and pressure removed. In an alternative procedure
commonly used in this study, a much thicker precoat is obtained at low
circulation velocity, but the sewage effluent to be treated must be
introduced without shutting down, or the coat will tend to flake off;
since we wished to compare coats in this case, and a shutdown was required
for one, the high-circulation-velocity, adherent thin coat was used. It
has been customary to include a little powdered activated carbon (say,
10 mg/£) in the water slurry of filteraid used in precoating, to adsorb
substances left in the system which might be deleterious to flux.
For regeneration in the runs of Fig. 48, the inside of the test
2
section tube was first filled with water and 550,000 N/m (80 psi) com-
pressed air turned on for five seconds. This was repeated five times.
Tap water was then pumped past the test section at 4.6 m/sec (15 ft/sec)
to 6 m/sec (20 ft/sec) for ten minutes with the product valve closed.
The procedure is fairly typical of that used in the work described
throughout this report, though variations such as circulation of sulfuric
acid solutions were sometimes tried. Filter surfaces can also of course
be scrubbed, if desired.
The results in Fig. 48 .indicated success in restoring flux. The
lower rate of flux decline on the second and third days may result from
dilute feed, rainfall being heavy during this series.
84
-------
CO
50
T3
X
E
X
10
5
2 -
RUN
Slope =-0.5-*'
RUN 2
Slope = -0.5
RUN 3
1000
500 CM
13
CL
CT>
Slope=-0.5-
0.2 0.5 1.0
5 0.2 0.5 1.0
HOURS
5 0.2 0.5 1.0
Fig. 48. Regeneration by backwashing; comparison of precoats
(Fire-hose jackets; external flow; 40 psi [2.8 bar];
15 ft/sec [4.6 m/sec]; 50 ppm Fe; pH 7.5-8)
Precoot
° Medium Fiber Asbestos
a Cellulosic ( Solkafloc BW 200 )
A No Precoat
100
I
5
-------
In conformity with other experience during this study, asbestos and
cellulosic filteraid seemed advantageous in comparison with no precoat in
maintaining flux in successive runs. Fluxes with asbestos were a little
higher here than with cellulose, but the advantage seemed to diminish with
successive tests.
Figure 49 illustrates regeneration over a more extended series (over
seventy runs) with a single fire-hose-jacket test section. The fluxes com-
pared are those measured in each run after two hours operation. It appears,
when account is taken of the generally higher fluxes obtained with PAG than
with hydrolyzable ion additives, that regeneration is reasonably successful
by the procedures used. The apparent decline after sixty runs likely results
from the experiments being carried out, rather than hose deterioration,
since fluxes through fresh hoses run simultaneously were also lower.
Flow of solutions to be filtered inside tubes is attractive in compari-
son to external pressurization because of simplicity of equipment design
and the considerable savings realized from omission of pressure support—
2
fire-hose jackets have been operated at up to 3,450,000 N/m (500 psi) with
internal circulation without occurrence of sprays from stretching of fabric.
However, regeneration procedures involving backwashing are not feasible
without complicating design. Other methods to clean surfaces are conceivable,
such as chemical washes. Figure 50 indicates that this may well be possible.
Two lengths of hose jackets, A and B, were used in alternating runs with
inside flow in the test module of Fig. 5. Between runs, they were soaked
in 1 M HC1; in the exception noted in the figure the hose was allowed to
dry, and air was blown through before the next run. The results show that
the fluxes for the internal flow test section were comparable to an exter-
nally pressurized test section (regenerated by usual backwash procedures)
used simultaneously.
Although we have not yet developed a routine less cumbersome than
removing and soaking hoses, nor looked for optimum conditions for satis-
factory regeneration, we believe it likely that satisfactory procedures
can be worked out. Drying might be particularly attractive, if prelimin-
ary favorable results were confirmed in repetitive operations.
86
-------
CO
80 -
70
-= 60
CVJ
o 50
x
ID
40
30
20
10
0
o
o
00
o o
o
o o
1
1
1
T
ADDITIVE
2000
Carbon
Iron or Aluminum Salts
None
1
1
10 20 30 40 50 60
NUMBER OF RUN WITH SAME TEST SECTION
1500
CO
1000
•o
a.
500
o
100
70
Fig. 49. Regeneration by backwashing; comparison of fluxes
in successive runs.
(Fire-hose jacket; ext. pressure)
-------
, INTERNAL FLOW
, EXTERNAL FLOW
x x
=3 ID
-J _J
U- LL
I.O
1.6
1.4
1.2
1.0
0.8
0.6
n d.
I I I
• HOSE A
— O HOSE B —
* NO ACID
0 o
o • •
o o^
•"*
— —
I I I
0123
NUMBER OF REGENERATIONS
Fig. 50. Regeneration of fire-hose jackets.
(Outside flow - backwashing; inside flow - acid soak)
88
-------
DISCUSSION
The results reported here indicate that cross-flow filtration of the
effluent from primary treatment of municipal sewage, following addition of
salts containing (or producing on oxidation) Fe(III) or Al(III), produces
a filtrate superior by most of the usual criteria to the typical product
of biological secondary treatment—turbidities usually below 1 JTU; TOG,
10 to 15 mg/£; total phosphorus, below 1 mg/£, expressed as phosphate
(if expressed as P, the value would be about one third as large); and
low microorganism content. Filtrate after powdered activated carbon
treatment is comparable, except that it is usually lower in TOG, around
5 mg/£, and phosphate is not efficiently removed, the filtrate containing
about as much phosphate as the feed. Lower TOG contents can be obtained
with Fe(III) or Al(III) either by additional PAC treatment or by using
effluent from biological secondary processing as feed. Filtrates appear
superior to products of physical-chemical treatments discussed in the
introduction, when solids-liquid separations were effected by settling,
except perhaps when tube settlers were used.
Costs are of course a major factor in evaluation of the possible
usefulness. These in turn will depend heavily on the flux attainable.
Before discussing these aspects, certain reservations about conclusions
which can be drawn from the limited scope of work so far should be noted.
An important one is the question of disposition of the concentrate,
perhaps 5% of the total volume processed, but probably less. Hopefully,
it can be recirculated to the inflow of the primary settler, and the
solids removed in that stage, to be eventually disposed of with the rest
3
of the sludge by anaerobic digestion or otherwise. Other work has
indicated that presence of these additives does not have a deleterious
effect on digester operation, and in fact, it has been suggested that
89
-------
Fe(III) might improve operation, by reacting with sulfides and thus
lowering H~S in the CH, produced. Cross-flow filtration concentrates
appear to settle well, but a test on a scale of essentially the full
output of a primary unit will be necessary to establish that the primary
effluent is not affected to an undesirable extent by the addition of the
concentrate. Should this procedure prove unfavorable, other techniques,
including standard filtration approaches, likely will be feasible for the
relatively small volumes involved.
The tests reported here were for a waste waters of a suburban type
community, with relatively small inputs from industries. Conclusions
concerning performance with effluents generated by large urban complexes
require further work.
There are some differences between our experimental conditions and
those to which feeds in a practical system would be exposed. Perhaps
the most important is frequency of passing through the pumps, which sub-
ject the liquids to high shear. In a sewage plant, the filtration units
would presumably be long compared to those in these tests, and the circu-
lating solution would pass through pumps fewer times for a given water
recovery. Consequently, the particle size at the filter surface may be
larger in a practical situation. It might be hoped that fluxes would be
better than in these experiments, but this cannot be confidently predicted.
One further comment, on criteria, applies to sewage treatment
methods in general, not just cross-flow filtration. Evaluation in terms
of such inclusive measurements as TOC or BOD seems to us inadequate. A
more detailed inventory of the Components of the treated streams and their
effects appears needed. Since obtaining such information and determining
significance of various classes of impurities discharged into different
receiving waters is a formidable undertaking, use of the present non-
specific criteria is understandable. However, it is likely that the TOC
of water treated by different processes is made up of components having
quite different environmental effects, and final judgment on new processes
should be delayed until sounder bases of evaluation exist. The lower bac-
terial content of cross-flow filtrate might, for example, make more accept-
able discharge without disinfection should chlorination be restricted because
of concern about organochlorine compounds.
90
-------
COSTS
Discussion of costs can conveniently be divided into costs of cross-
flow filtration, costs of chemicals, and costs of operations included in
feed pretreatment.
Cross-Flow Filtration
Costs of cross-flow filtration were treated earlier in this program
by Wadia, Kraus, Shor and Dresner in Reference 29, and we shall only out-
line their approach and conclusions. Their analysis was of an operation
applicable to sewage treatment, but not restricted to it. A number of
parameters (Table 6) was investigated—plant size, average flux, and
circulation velocity, as well as plants with minimum instrumentation vs_
automated plants. The central set of conditions, from which the effect
o f.
of variations was assessed, was a 3800 m /d (10 gpd) plant, producing
2
filtrate at a rate of 6.1 m/day (150 gpd/ft ) when in operation at
4.6 m/sec (15 ft/sec) circulation velocity. Within the range of condi-
tions covered, costs of filtration for a plant of a given flux can be
estimated, and optimum circulation velocities can be selected, if flux
dependence on velocity has been established for the feed in question.
Effect of pressure and water recovery on flux were included in the
average flux; as the results presented in this report indicate, these do
not to a first approximation appear to affect flux greatly. A layout
for a typical plant is illustrated in Fig. 51, which helps to define
some of the terms in Table 6. "Tiers" are regions from which one or
more pumping circuits feed and into which reject is discharged. As fil-
trate is removed, and fresh feed brought into Tier 1, flow over a weir
into Tier 2 maintains recovery at a design steady-state figure. For
different sized plants, recovery varies from 50 to 66% in Tier 1, and
is brought to ca. 95% in second, and, in some cases, third tiers. Tiers
thus allow tapered plant design. The several recirculating "pumping
circuits" (p. c.) operating on each tier each incorporates one filtration
module. The capacity of the individual pumping circuits fixes the number
necessary and can be selected to match available equipment and to increase
reliability by use of smaller units, if desired. "Recycle ratio" is the
ratio of volume taken into a module to filtrate volume per unit time.
91
-------
Table 6. PLANT CHARACTERISTICS'
N5
Plant
Nominal
(gpd)
io5
106(A)
106(B)
106(C)
106(D)
106(E)
106(F)
io7
capacity Design parameters
Actual Velocity Flux
(gpd) (ft/sec) (gpd/ft2)
1.12
1.12
1.12
1.12
1.12
1.11
1.13
1.11
x IO5
x IO6
x IO6
x IO6
x IO6
x IO6
x IO6
x IO7
15
15
15
15
10
10
15
15
150
75
150
150
150
300
600
150
Number of mods. Circulation pump ratings
per tier Circ.
Tier Tier Tier Pres. head rate Motor hp
I II III (psi) (103N/m2) (gpm) consumption
1
2
4
2
4
2
2
6
1
1
2
1
2
1
1
4
50
105
1 60
70
1 30
30
50
110
345
724
414
483
207
207
345
759
630
3,500
1,370
2,800
980
1,180
1,100
5,000
25
250
60
135
25
25
45
365
No. of
pumps
circuits
2
3
7
3
7
3
3
10
Average
recycle
ratio
16.2
13.5
12.3
10.8
8.6
4.6
4.2
6.4
1106 gpd = 3785 m3/d ; 15 ft/sec - 4.57 m/sec; 150 gpd/ft2 = 6.11 m/d; 1 gpm - 3.785 A/mln.
hp signifies horsepower; 1 hp = 0.745 kilowatts.
-------
u>
PRODUCT
WATER
PRIMARY
SEWAGE
INLET
PRODUCT
WATER
27ft
82m
L.L .....
PRODUCT
WATER
TRENCH
/ 3ft DEEP\
\0.9m I
PUMP 5
CONCRETE PAD
'ADJUSTABLE GATES
TIER II
\
*.VS-V\\^\-^-.M ^ f.\^~-
.^^.v^^^v^
PUMP 6
u
Fig. 51.
14ft
4.3m
12ft
3.6m
Plan for a 3788 m3/day (106 gpd) cross-flow filtration plant.
30 ft
9.1 m
-------
Characteristics assumed for the modules are listed in Table 7, see
also Fig. 52. Circulating solution passes in parallel through groups of
tubes (for example, groups of 17 in modules of the 10 gpd plant, so that
there are four passes back and forth through the 68 tubes of the module).
The filters are one-inch diameter tubes with inside flow, fabricated to
allow backwashing. The system is designed so that feed circulation
reduces pressures at module exits to only a small fraction of pressure at
the inlet.
The costs in Reference 29 were estimated for July 1971, and we have
updated them to March 1975. For building costs, we used the ratio of
the construction index of Engineering News Record for the two dates,
1.335, and an average ratio of 1.30 for labor. Module costs were orig-
inally estimated by analogy with tube-in-shell heat exchangers—cost of
2 2
heat exchanger tubes were subtracted and $2/ft ($22/m ) of filter sur-
2 2
face was substituted. For this update, $2.80/ft ($30/m ) of surface was
used. Modules account for a quarter to a third of total capital cost.
A higher interest rate was also used, the amortization-interest assumed
now being 8.6%, equivalent to 6% interest over a twenty-year life.
Electricity costs were changed from 0.7
-------
Table 7. MODULE CHARACTERISTICS'
vo
Ui
Nominal
capacity
(gpd)
105
106(A)
106(B)
106(C)
106(D)
106(E)
106(F)
io7
Design parameters
Velocity Flux
(ft/sec) (gpd/ft2)
15
15
15
15
10
10
15
15
150
75
150
150
150
300
600
150
Total Production
module area per module
(ft2) (Kpd)
370
4,980
1,070
2,490
1,070
1,230
630
7,390
5.6 x
3.7 x
1.6 x
3.7 x
1.6 x
3.7 x
3.8 x
1.11 x
10A
io5
IO5
io5
io5
io5
io5
io6
Tube
length
(ft)
21.0
40.0
22.0
25.0
22.0
25.0
20.0
41.5
Approx.
shell dia.
(in)
12
32
20
27
20
20
16
36
Number of
tubes per
pass
17
95
37
76
37
47
30
136
Number of
tube passes
4
5
5
5
5
4
4
5
a!06 gpd = 3785 m3/d; 15 ft/sec = 4.57 in/sec; 150 gpd/ft2 = 6.11 m/d; 1 ft2 = 0.0929 m2; 1 in = 2.54 cm.
-------
Table 8 . CROSS-FLOW FILTRATION - COST SUMMARY
(Non-Automated Plant;* 95% Water Recovery)
Plant (gpd)
Flux (J, gpd/ft2)
Circ. vel. (u, ft/sec)
No. pump circuits
Capital Cost
$/daily gal
$/daily gal @ x=$2.8/ft2
c/kgal (AI=8.6%)t
Labor
(hrs/day)
c/kgal (@ $65/8 hrs)
Maintenance ( C/kgal)
(@ 3% of plant cost)
Subtotal (c/kgal)
C/ft2-day (CA)
Energy (kwh/kgal)
C/kgal (@ l.Oc/kwh)
Total Cost (c/kgal)
io5
150
15
2
.896 +
.009x
.921
21.7
3
24.4
6.2
52.3
7.8
8.4
8.4
60.7
106(A)
75
15
3
.494 +
.Ol8x
.544
12.8
5
4.1
3.7
20.6
1.5
12.7
12.7
33.3
106(B)
150
15
7
.427 +
.009x
.452
10.6
10
8.1
3.0
21.7
3.3
7.0
7.0
28.7
106(C)
150
15
3
106(D)
150
10
7
106(E)
300
10
3
.348 + .382 + .204 +
.009x
.373
8.8
5
4.1
2.5
15.4
2.3
6.9
6.9
22.3
.009x
.407
9.6
10
8.1
2.7
20.4
3.1
2.8
2.8
23.2
.005x
.218
5.1
5
4.1
1.5
10.7
3.2
1.3
1.3
12.0
106(F)
600
15
3
.183 +
.002x
.189
4.5
5
4.1
1.3
9.9
5.9
2.2
2.2
12.1
io7
150
15
10
.219 +
.009x
.244
5.8
20
1.6
1.6
9.0
1.4
6.2
6.2
15.2
Cases A, B, C, D, E, F - see Table 7.
AI signifies amortization-interest. 8.6% corresponds to 20 years amortization at 6% interest,
-------
Table 9 . CROSS-FLOW FILTRATION - COST SUMMARY
(Automated Plant;* 95% Water Recovery)
vD
Plant (gpd)
Flux (J, gpd/ft2)
Circ. vel. (u, ft/sec)
No. pump circuits
Capital Cost
$/daily gal
$/daily gal @ x=$2.8/ft
C/kgal (AI=8.6%)T
Labor
(hrs/day)
C/kgal (@ $65/8 hrs)
Maintenance (c/kgal)
(@ 3% of plant cost)
Subtotal (c/kgal)
C/ ft2 -day (CA)
Energy (kwh/kgal)
C/kgal (@ l.Oc/kwh)
Total Cost (c/kgal)
io5
150
15
2
1.
2 1.
29.
1.
8.
8.
45.
6.
8.
8.
53.
215 +
.009x
240
2
0
1
2
5
8
4
4
9
106(A)
75
15
3
.542 -
10
150
15
7
h
.018x
.592
13.9
1.5
1.2
4.0
19.1
1.4
12.7
12.7
31.8
13
2
2
3
19
2
7
7
26
'(B)
.539 H
10
150
15
3
h
.009x
.565
.3
.5
.0
.8
.1
.9
.0
.0
.1
9
1
1
2
13
2
6
6
6(C) 106(D) 106(E) 106(F) IO7
150 300 600 150
10 10 15 15
7 3 3 10
.396 +
.493 +
.254 +
.009x .009x .005x
.421
.9
.5
.2
.8
.9
.1
.9
.9
20.8
.518
12.2
2.5
2.0
3.5
17.7
2.7
2.8
2.8
20.5
.268
6.3
1.5
1.2
1.8
9.3
2.8
1.3
1.3
10.6
.231 +
.002x
.237
5.6
1.5
1.2
1.6
8.4
5.0
2.2
2.2
10.6
.235 +
.009x
.260
6.1
4.0
0.3
1.7
8.1
1.2
6.2
6.2
14.3
Cases A, B, C, D, E, F - see Table 7.
AI signifies amortization-interest. 8.6% corresponds to 20 years amortization at 6%.
-------
TO PRODUCT TRENCH
vo
00
PRECOAT
CONCENTRATE
CROSSFLOW FILTRATION MODULE
BACKWASH- PRECOAT
SUCTION
RECYCLE TO
SEWAGE TRENCH
PRODUCT DISCHARGE
PRODUCT TRENCH
Fig. 52. Elevation of a cross-flow filtration plant module.
-------
100
500 1000 2000
m3/d
5000 10,000
30,000
— 20
§>
O)
o
o
•B 10 —
ro
TOTAL COST
(SLOPE:-0.27)
ENERGY COST
AM.-INT. COST
{8.6 % AI)
LABOR + MAINT.
PLANT SIZE (gpd)
Fig. 53. Cross-flow filtration-dependence of operating cost
on plant size.
(15 ft/sec [4.6 m/sec]; 150 gpd/ft2 [6.1 m/d]; automated plant;
(p.c. = pumping circuits)
99
-------
m/d
100
•<> 10
O
O
20
40
— 20
— 10
TOTAL COST
(SLOPE:-0.52)
AM.-1 NT. COST
(8.6% AI)
LABOR -I- MAINT.
ENERGY COST
FLUX (gpd/ft2)
Fig. 54. Cross-flow filtration-dependence of operating cost on flux.
(106 gpd [3788 m3/d]; 3 pumping circuits; 15 ft/sec [4.6 m/d];
automated plant)
100
-------
Day-to-day variations in base and acid requirements emphasize the un-
certainty. We shall adopt for illustration a sequence of addition of
50 mg/Jl Fe(II) as ferrous sulfate (oxidized to Fe(III) by air) and enough
excess H^SO, to bring pH down to 4, then neutralization to pH 7 by lime
before filtration.
The appropriateness of the iron addition (^0.001 M) selected can be
questioned. It might appear from the molar ratio of iron to phosphate
upwards of 2 implied as necessary by Fig. 23 for removal of phosphate to
low levels that at higher phosphate concentrations encountered in feed
(Figs. 7 and 8) fifty mg/Jl of iron would be insufficient. However, with
few exceptions this level has reduced phosphate in filtrate to less than
1 mg/Jl and we believe it adequate. Possibly more phosphate is removed
when the Fe(III) hydrolyzes and precipitates in the feed, rather than when
the feed contacts iron added earlier, as was the case for most of
the results in Fig. 23. There also may be some iron or other substances
which remove phosphate in the primary sewage.
A more serious question is the optimum iron level for flux. We
mentioned earlier the apparent disagreement between, on the one hand,
the indication of Figs. 27 and 32 that ferric concentrations above those
necessary to remove phosphate had no clear effect on flux and on the other
hand the pronounced effect shown in Fig. 6.23 of Reference 35. A recent
experiment, in which primary sewage taken in at the same time was fil-
tered simultaneously in the two trailer loops at two different iron
levels (50 rag/A and 200 mg/Jl), also indicated a substantially higher
flux at higher iron concentration. It is of course possible that simi-
lar trends are lost in scatter in Figs. 27 and 32 arising from differences
in feed from day to day. Differences in test procedures are an alterna-
tive possibility. The experiments in which flux increased with additive
concentration were carried out at 0% water recovery, filtrate being
recycled to the concentrate tank. In those of Figs. 27 and 32, filtrate
was discarded, water recoveries being over 75% at the end of most 24-hour
runs. Since additive was added to feed at a constant amount per unit
volume, the circulating stream became continually enriched in iron; the
Fe(III) concentration at a given time was the number of mg/Jl added to in-
coming feed divided by (1 - fractional water recovery), less minor amounts
passed with filtrate.
101
-------
The most obvious way to settle the question would be simultaneous
experiments with increasing water recovery in the two trailer loops, under
conditions identical except for iron concentration. This would require
parallel feed pretreatment lines in the van (Fig. 1), which are not now
available. We have elected to base this discussion on results of the
pattern of Figs. 27 and 32, on the basis that it simulates more closely
practical operation than the other. Further, we shall use the conserva-
2
tive 24-hour average fluxes of about 6.1 m/d (150 gpd/ft ) suggested by
2
Fig. 32 (rather than the M.0.2 m/d (250 gpd/ft ) of Fig. 27). For situa-
tions where higher fluxes are realized, Fig. 54 allows estimation of the
benefits. The possibility, suggested by the results of Fig. 6.23 of
Ref. 35, of a tradeoff between cost of higher iron concentration and
savings from high flux should also be remembered; this would be particu-
larly attractive in locations where ferrous-containing waste streams are
available. '
Requirement of acid, in excess of that produced by Fe(III) hydrolysis,
and of base for neutralization vary with the sewage, though not in a way
readily predictable, at least from TOG. In four typical runs in which we
kept careful account of chemical requirements, H?SO, needed to bring the
pH to 4 (50 mg/fc iron) varied from 0.0011 to 0.0016 moles/A of primary; for
cost estimates we used the average, 0.0014 (Table 10). Sodium hydroxide
addition, after CO. had been removed by air sparging, needed to raise
pH 4 solutions to pH 7 ranged in three runs from 0.0018 to 0.0033 moles
NaOH/liter, the average being 0.0026. About half of this brought pH
from 4 to 6; other runs required about the same average amount to reach
pH 6 from 4. If NaOH at 16.5c/kg (6.5c/lb) were used, the neutralization
3
from pH 4 to pH 7 would cost nearly 1.8$/m (7
-------
3
The sum of these chemical costs (Table 10) is a little over 3.3c/m
(13c/kgal). Introduction of iron as other commercially available salts
did not alter cross-flow filtration performance significantly. There
would be little cost difference, according to current issues of Chemical
Marketing, if ferric chloride were used instead of ferrous sulfate and
ferric sulfate might be a little cheaper; there is some question concern-
ing the % Fe in the various industrial grades of all these chemicals.
Aluminum introduced as aluminum sulfate (at $68/metric ton ($62/ton),
17% AlJD.) would cost about a third more per mole than Fe(II).
Other Costs
Ref. 29 dealt only with the cross-flow filtration operation, and
did not include provision for the chemical pretreatment steps projected
here. We contemplate that these operations would be done by extending
the primary sewage effluent channel to include two more compartments,
also isolated by weirs, in the layout of Fig. 57. In the first, iron
salts and acid would be added with aeration, and in the second, pH would
be adjusted to the level selected for filtration, again with aeration to
complete oxidation of Fe(II) to Fe(III). It may be feasible to combine
the neutralizing compartment with the first tier, particularly if iron
is added as Fe(III) or if Al(III) is used. We have not estimated in
detail what the added cost would be, but by analogy with aeration pro-
38
cesses, an increment of about 4
-------
Table 10. ESTIMATED TYPICAL CHEMICAL COSTS IN CROSS-FLOW
FILTRATION OF PRIMARY SEWAGE EFFLUENT
Cost, flltratea
Additive Assumed price Level $/m3 c/kgal
Ferrous $57/tonnec $52/ton 50 mg/fc as Fe(II) 1.51 5.7
sulfateb
H2SOi+ $55/tonnec $50/ton 0.0014 raolesM primary 0.84 3.2
CaO $32/tonnec $28.75/ton 0.0026 equiv. OH~/£ .26 1.0
primary"
Total added chemicals 2.61 9.9
Filteraid 55c/kg 25c/lb 10 mg/cm2 filter surface .90 3.4e
(Solka-floc
BW-200)
Total, chemicals plus precoat 3.51 13.3
aAssumed 95% water recovery
^Assumed formula FeSOif»7H20, fraction Fe(II) = 0.2
cTonne is metric ton
dAssumed pH adjusted to 7 for filtration
eAssumed daily application, 6.1 m/d (150 gpd/ft2) average flux
Table 11. SUMMARY OF ESTIMATED COSTS OF CROSS-FLOW
FILTRATION OF PRIMARY SEWAGE EFFLUENT*
Cost
0/m3 C/kgal
Pretreatment , processing operation
Pretreatment , chemicals (Table 10)
Cross-flow filtration (Table 8, case B)
Total
1.1
3.5
7.6
12.2
4
13.3
28.7
46
asize: 3.8 x 10d md/d; 10b gpd
Flux: 6.1 m/d; 150 gpd/ft2
Circ. vel.: 4.6 m/sec; 15 ft/sec
Chemical treatment: Fe(II)
Nonautomated plant, 7 pumping circuits
104
-------
3 2
production rate had been 10.2 m /d (250 gpd/ft ) (suggested by Fig. 27),
3
costs would have been about 1.6/m (6c/kgal) lower. If backwashes with
coats of filteraid as thin as those of Fig. 48 laid at high circulation
velocity were able to maintain fluxes in successive regenerations as well
as those with the thicker coats typical of Fig. 49, the precoat cost of
Table 10 would be greatly reduced, perhaps to a negligible level. Use of
chemical or other regeneration procedures at intervals of a few runs may
make this feasible, even if thin coats give less protection.
Major potential for savings, particularly in view of the sharp recent
escalation in energy costs, may lie in operation at lower circulation
velocities. Fluxes will be lower, but capital costs, outside of expenses
for more filter surface, may not necessarily be substantially higher.
Lower circulation velocity entails lower pressure drop per unit length
which will tend to compensate increases in pumping volume necessary at
lower flux. If regeneration procedures illustrated in Fig. 50 were
proven adequate over extended periods, inside flow with hoses would make
shifts to lower circulation velocity particularly attractive, since
additions to filtering surface would be relatively cheap. Even in small
lots (305 m; 1000 ft), recent purchases of fire-hose jackets have been
2
only a little over $l/ft , and probably fabrics designed with optimum
characteristics for this application would not be much more expensive in
commercial quantities.
In the experiment of Fig. 44, PAC was added to primary sewage
at an initial high charge, with no further addition as sewage was processed,
If the run had been interrupted when the average usage of PAC was about
500 mg/Jl, in the range recommended by a group who ran tests with con-
tional separations procedures, average flux would have been in excess
2
of 16 m/d (400 gpd/ft ). At that rate, cost of the cross-flow filtration
•j
for the case comparable to Table 11 would be about 4.8
-------
filtration of PAC-treated sewage may well be attractive, though carbon
33 6
regeneration at the 3.8 x 10 m /d (10 gpd) level discussed here will
likely be somewhat costlier. We have filtered with cross-flow slurries at
concentrations in excess of 10 weight % PAC; from 10 cc of concentrate,
centrifugation separated a 5.5 cc bed of PAC. Filtration itself went
smoothly, and product was clear, though there was difficulty in pump
operation. The test took place before mechanical seals were installed,
and the slurry tended to degrade pump seals.
Comparisons with the cost of alternative sewage treatments are uncer-
tain at the present stage of development of this approach, but it appears
probable that costs, for example those summarized in Table 11, are consid-
erably higher than those of biological secondary for a plant of similar
size. However, additional treatments to the secondary effluent to bring
it to the quality of cross-flow filtrate with respect to phosphate content,
turbidity, and bacterial contamination would likely raise the total treat-
ment cost to a level comparable to or above that of filtration. The advan-
tages of the cross-flow filtration approach are likely initially to be
greater for a small installation in situations where reliability require-
ments are important and space is limited.
FUTURE WORK
The patterns of operation utilizing cross-flow filtration described
here, when used in place of standard biological secondary treatment, give
water of quality in many respects superior to the effluent from conven-
tional procedures. Operation should be more reliable, space requirements
less, and various options (Fe(III) and/or PAC, for example) allow flexi-
bility to meet variations in feed compositions and in effluent require-
ments. The cross-flow filtration units should be amenable to modular
construction, with the benefits which large-scale centralized production
might realize.
We enumerated in Recommendations the next steps we believe should be
taken in development of this concept. These can be classed as finding
answers to outstanding questions accessible to small-scale tests, compo-
nent development, and demonstration.
106
-------
In several significant areas, there are uncertainties which could be
cleared up by further experiments of the type discussed in this report.
Since many of these involve fixing more definitively the effect of vari-
ations in chemistry, the most expeditious route would be to add a second
pretreatment line for the other loop, so that simultaneous high-water-
recovery tests could be carried out under conditions identical except for
the variable under investigation. In this way effects of day-to-day
variations of sewage composition on results would be minimized. Some
important parameters whose effect should be better defined are effect of
Fe(III) concentration; differences in Fe(III) and Al(III), and of the
iron salt used; effect of pH, both in filtration and in the acid step
during Fe(III) or Al(III) pretreatment; effect of different patterns of
addition, e.g., heavy initial charge of Fe(III), and small additions to
incoming sewage vs constant additions to incoming sewage; effect of pH
on operation with PAC (there is preliminary evidence that fluxes after
PAC addition are higher at pH 4 than at higher pH); and variations in
filteraid precoats. We have expressed earlier the opinion that behavior
in operation at lower circulation velocities should be explored more com-
pletely, but this can be done with equipment at hand.
The unit should also be relocated for a limited time to assess per-
formance with sewage which includes substantial contributions from indus-
trial effluents. A relatively small number of critical tests should
indicate the extent experience with sewage from a bedroom-type community
typified by the Oak Ridge plant is applicable to wastes from more mixed
sources.
In component development, the important need is design, construction,
and testing of modules suitable for use in a practical plant. At a mini-
mum, configurations using both inside and outside flow should be tried.
The additional research proposed would not be absolutely necessary
for design of a plant, but we believe it advisable to define better
optimum characteristics. Concurrently with research and module develop-
ment, a small primary sewage plant, perhaps about 10 gpd, which is
faced with the necessity of adding secondary treatment in the near
future, should be found and steps taken to arrange for a demonstration
plant utilizing essentially the complete primary output. In addition
107
-------
to establishing the validity of design based on presently available
results and those obtained in continuing work, the demonstration would
settle whether or not the concentrate from cross-flow filtration can be
adequately handled by recycle through the primary settler. It would also
put cost estimates on a firmer basis.
In summary, the reported results indicate that cross-flow filtration
can be used in conjunction with physical-chemical additives to produce
reliably from primary sewage an effluent superior in quality to that from
conventional biological secondary treatment. Analogous capabilities for
19
other waste waters can be inferred, and other studies involving solids-
liquid separations, in removal of toxic substances and in treatment of
industrial wastes, suggest that the approach indeed may be useful. Dif-
ferent effluents are specific in character, however, and individual tests
and development of appropriate procedures are required before conclusions
concerning applicability can be drawn. Municipal sewage itself probably
varies enough in composition to require local tests before design of a
cross-flow filtration system for treatment.
108
-------
REFERENCES
1. Kreissl, J. F., and J. J. Westrick. Municipal Waste Treatment by
Physical-Chemical Methods. In: Proceedings of Conference on
Applications of New Concepts of Physical-Chemical Wastewater Treat-
ment) sponsored by The International Conference on Water Pollution
Research and The American Institute of Chemical Engineers, Vanderbilt
University, Nashville, Sept. 1972, Eckenfelder, W. W., and L. K.
Cecil (eds.). New York, Pergamon Press, 1972. pp. 1-11.
2. Weber, W. J., Jr. Physicochemical Systems for Direct Wastewater
Treatment. In: Proceedings of Conference on Applications of New
Concepts of Physical-Chemical Wastewater Treatment, sponsored by
The International Conference on Water Pollution Research and The
American Institute of Chemical Engineers, Vanderbilt University,
Nashville, Sept. 1972, Eckenfelder, W. W., and L. K. Cecil (eds.)
New York, Pergamon Press, 1972. pp. 13-25.
3. Alvord, E. T., D. M. Gaughan, C. M. Marr, T. Colpetzer, and D. B.
Rose. Phosphorus Removal by Ferrous Iron and Lime. Rand Develop-
ment Corporation and County of Lake, Ohio, performing organizations.
U. S. Government Printing Office Report WPCRS 11010 EGO 01/71. U. S.
Environmental Protection Agency, Jan. 1971. 71 pp.
4. Leary, R. D., L. A. Ernest, R. S. Powell, and R. M. Manthe. Phos-
phorus Removal with Pickle Liquor in an Activated Sludge Plant.
Sewerage Commission of the City of Milwaukee. U. S. Government
Printing Office Report WPCR 11010 FLQ 03/71. U. S. Environmental
Protection Agency. March 1971. 143 pp.
5. Leary, R. D., L. A. Ernest, R. S. Powell, and R. M. Manthe. 200 MGD
Activated Sludge Plant Removes Phosphorus by Pickle Liquor. Sewerage
Commission'of the City of Milwaukee. U. S. Government Printing
Office Report EPA-670/2-73-050. U. S. Environmental Protection
Agency. Sept. 1973. 127 pp.
6. Wilson, T. E. P. J. Yonikas, and G. Lukasik. Waste Pickling Liquor
for the Removal of Phosphorus in a Municipal Wastewater Treatment
Plant. In: Water - 1973, AIChE Symp. Series, Vol. 136, Bennett,
G. F. (ed.). New York. AIChE, 1974. pp. 350-357.
109
-------
7. Burns, D. E., and G. L. Shell. Physical-Chemical Treatment of a
Municipal Wastewater Using Powdered Carbon. Envirotech Corporation,
performing organization. U. S. Government Printing Office Report
EPA-R2-73-264. U. S. Environmental Protection Agency. August 1973.
pp. 657-669.
8. Shell, G. L., and D. E. Burns. Powdered Activated Carbon Application,
Regeneration and Reuse in Wastewater Treatment Systems. In:
Advances in Water Pollution Research, Proc. Sixth Int. Conf.,
Jerusalem, June 1972, Jenkins, S. H. (ed.). New York, Pergamon
Press, 1972. pp. 657-669.
9. Beebe, Richard L. Activated Carbon Treatment of Raw Sewage. Westing-
house Electric Corporation, performing organization. U. S. Govern-
ment Printing Office Report EPA-R2-73-183. U. S. Environmental
Protection Agency. March 1973.
10. Shuckrow, A. J., G. W. Dawson, and W. F. Bonner. Powdered Activated
Carbon Treatment of Combined and Municipal Sewage. Battelle Pacific
Northwest Laboratories, performing organization. U. S. Government
Printing Office Report EPA-R2-73-1A9. U. S. Environmental Protection
Agency, February 1973.
11. Gulp, R. L., and G. L. Gulp. Advanced Wastewater Treatment. New
York, Van Nostrand Reinhold Company, 1971. 310 pp.
12. Bishop, D. F., T. P. O'Farrell, A. F. Cassel, and A. P. Pinto.
Physical-Chemical Treatment of Raw Municipal Wastewater. U. S.
Government Printing Office Report EPA-670/2-73-070. U. S. Environ-
mental Protection Agency. September 1973.
13. Hais, A. B., J. B. Stamberg, and D. F. Bishop. Alum Addition to
Activated Sludge with Tertiary Solids Removal. U. S. Government
Printing Office Report EPA-670/2-73-037. August 1973.
14. Derrington, R. E., D. H. Stevens, and J. E. Laughlin. Enhancing
Trickling Filter Plant Performance by Chemical Precipitation. City
of Richardson, Texas, performing organization. U. S. Government
Printing Office Report EPA-670/2-73-060. U. S. Environmental
Protection Agency. August 1973.
110
-------
15. Horstkotte, G. A., Jr. Pilot-Demonstration Project for Industrial
Reuse of Renovated Municipal Wastewater. Contra Costa County Water
District, performing organization. U. S. Government Printing Office
Report EPA-670/2-73-064. August 1973.
16. Jahreis, C. A. Shriver Continuous Thicknener. U. S. Patent
2,364,366 (1944). T. Shriver and Co., Inc., Harrison, N. J.
17. Zhevnovatyi, A. I. The Thickening of Suspensions without Cake
Formation. Int. Chem. Eng. 4:124, 1964.
18. Dahlheimer, J. A., D. G. Thomas, and K. A. Kraus. Hyperfiltration
XVII. Application of Woven Fiber Hoses to Hyperfiltration of Salts
and Crossflow Filtration of Suspended Solids. I&EC Process Des. &
Develop. 9:565, 1970.
19. Kraus, K. A. Cross-flow Filtration and Axial Filtration. Paper
presented at 29th Annual Purdue Industrial Waste Conference,
Lafayette, Indiana, May 7-9, 1974. To be published in proceedings.
20. Nelson, F., H. 0. Phillips, and K. A. Kraus. In: Chemistry Division
Annual Report for the Period ending May 20, 1972. ORNL-4791.
pp. 88-90.
21. Desaulniers, C. W., and R. W. Hausslein. Ultrafiltrative Dewatering
of Spent Powdered Carbon. Amicon Corporation, performing organiza-
tion. U. S. Government Printing Office Report ORD 17020 DBA 03/70.
U. S. Federal Water Quality Administration. March 1970.
22. Perona, J. J., J.D. Hutchins, A. M. Rom, J. S. Johnson, Jr., and
W. G. Sisson. A Pilot Plant for Sewage Treatment by Cross-Flow
Filtration. ORNL TM-4659, 1974.
23. Murtha, S. A., R. Filiba, J. S. Newman, and R. D. Orlandi. Character-
ization of ORNL's Cross-Flow Filtration Plant for Treating Primary
Sewage Effluent. MIT School of Chemical Engineering Practice, Oak
Ridge Station. ORNL-MIT-189, 1974.
24. Standard Methods for the Examination of Water and Wastewater, American
Public Health Association, Inc., 12th Ed., Washington, D. C., 1965.
769 pp.
Ill
-------
25. Thomas, D. G., and R. B. Gallagher. Hydrodynamic Flux Control for
Waste Water Application of Hyperfiltration Systems. Oak Ridge
National Laboratory, performing organization. U. S. Government
Printing Office Report EPA-R7-73-228, U. S. Environmental Protection
Agency, May 1973.
26. Johnson, J. S., Jr. Polyelectrolytes in Aqueous Solutions—Filtration,
Hyperfiltration, and Dynamic Membranes. In: Reverse Osmosis Membrane
Research, H. K. Lonsdale and H. E. Podall (eds.). Plenum Publishing
Company, 1972. p. 379.
27. Coulson, J. M., and J. F. Richardson. Chemical Engineering, Vol. 2.
Pergamon Press, 2nd Ed., 1968. p. 50.
28. Porter, M. C. Concentration Polarization with Membrane Ultrafiltra-
tion. Ind. Eng. Chem. Prod. Res. Develop. 11:234, 1972.
29. Wadia, P. H., K. A. Kraus, A. J. Shor, and L. Dresner. Preliminary
Economic Analysis of Cross-Flow Filtration. Oak Ridge National
Laboratory Report, ORNL 4729, August 1973.
30. Hsu, Pa Ho. Complementary Role of Iron(III), Sulfate and Calcium in
Precipitation of Phosphates from Solution. Environmental Letters.
5(2):115-136, 1973.
31. Ghassemi, M., and H. L. Recht. Precipitation of Polyphosphates with
Aluminum and Ferric Salts. In: Proceedings of the 25th Industrial
Waste Conference, May 5-7, 1970, Purdue University, Part One, 1970.
pp. 356-364.
32. Savage, H. C., N. E. Bolton, H. 0. Phillips, K. A. Kraus, and J. S.
Johnson, Jr. Hyperfiltration of Plant Effluents. Water and Sewage
Works. 116:102, 1969.
33. Kraus, K. A., J. A. Dahlheimer, and W. R. Mixon. Application of
Hyperfiltration to Treatment of Municipal Sewage Effluents. Water
Research Program, Oak Ridge National Laboratory. U. S. Government
Printing Office Report WPCRS ORD 17030FOH01/70. January 1970. 71 pp.
34. Mahlman, H. A., K. A. Kraus, and W. G. Sisson. In: Chemistry
Division Annual Report for the period ending May 20, 1971. Oak Ridge
National Laboratory Report ORNL-4706, September 1971. pp. 148-149.
112
-------
35. Mahlman, H. A., W. G. Sisson, J. S. Johnson, K. A. Kraus, and
N. Harrison. In: Chemistry Division Annual Report for the period
ending May 20, 1972. Oak Ridge National Laboratory Report
ORNL-4791, August 1972. pp. 100-101.
36. Cohen, J. M. Reference 8, discussion.
37. Nelson, F., H. 0. Phillips, and K. A. Kraus. Adsorption of Inorganic
Materials on Activated Carbon. Paper presented at 29th Annual Purdue
Industrial Waste Conference, Lafayette, Indiana, May 7-9, 1974. To
be published in Proceedings.
38. Smith, R. Cost of Conventional and Advanced Treatment of Wastewaters.
Federal Water Pollution Control Administration, Cincinnati Water
Research Laboratory. PB 182158. July 1968.
113
-------
PUBLICATIONS AND PATENTS
Certain aspects of the work described in this report were discussed
in References 19 and 27. Preliminary results were also reviewed by
J. D. Henry, in a chapter on cross-flow filtration in "Recent Develop-
ments in Separation Science," Volume II, p. 205, edited by N. N. Li,
CRC Press, 1972.
Two patents were based on results obtained in this program:
Kraus, K. A., and H. A. Mahlman. Cross-Flow Filtration Process for
Removal of Total Organic Carbon and Phosphates for Aqueous Sewage Efflu-
ents. U. S. 3,733,265. May 15, 1973.
Mahlman, H. A., and W. G. Sisson. Cross-Flow Filtration Process.
U. S. 3,835,040. September 10, 1974.
A third was partially derived from this work:
Kraus, K. A. Process for Treatment of Aqueous Suspensions.
U. S. 3,778,366. December 11, 1973.
All three are assigned to the U. S. Government.
114
-------
SYMBOLS AND ABBREVIATIONS
bar
BOD
cm/min
COD
ft/sec (or fps)
gm
gpd/ft2 (or gfd)
gpm
J
JTU
kg
kgal
kwh (or kwhr)
a
m
m3
m/d
m3/d
m/sec
mg
Pressure (1 bar = 105 N/m2 = 0.987 atm = 14.50 psi)
Biological oxygen demand
Flux through filter in cubic centimeters per square
centimeter per minute (1 cm/min = 14.40 m/d =
354 gpd/ft2)
Chemical oxygen demand
Circulation or cross-flow velocity (flow of solution
being filtered parallel to filtering surface) in feet
per second
Gram
Gallons (US) per day (1 gallon (US) = 3.785 A)
Flux in gallons (US) per square foot of filtering sur-
face per day (1 gpd/ft2 = 0.00283 cm/min = 0.0407 m/d)
Gallons (US) per minute
Flux
Jackson Turbidity Unit (measured here with Hach 2100
turbidimeter)
Kilogram
1000 (US) gallons (1 kgal = 3.785 m3 or approximately
3.785 metric tons of water)
Kilowatt hour
Liter
Meter
Cubic meter; if water, approximately a metric ton
Flux in cubic meters per square meter per day (1 m/d =
24.6 gpd/ft2 = 0.0694 cm/min)
Cubic meters per day (1 m3/d = 264.2 gpd)
Circulation velocity in meters per second
Milligram
115
-------
N/m2 Pressure in Newtons per square meter (1 N/m2 =
1.450 x ICT1* psi - 10~5 bar)
PAA Polyacrylic acid
PAC Powdered activated carbon (Aqua Nuchar A used in this
work)
ppra Part per million; for dilute aqueous solutions here,
approximately mg/A
psi or psig Pressure in pounds per square inch, measured by gauge
(1 psi = 6895 N/m2)
TOG Total organic carbon, measured here by Beckman 915
analyzer
Tonne Metric ton (1000 kg, 2,204.6 pounds avoirdupois); if
water, approximately 1 m3
u Circulation velocity
116
-------
117
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-76-025
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
CROSS-FLOW FILTRATION IN PHYSICAL-CHEMICAL TREATMENT
OF MUNICIPAL SEWAGE EFFLUENTS
5. REPORT DATE
Completed November. 1975
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Mahlman, H. A., Sisson, W. G., Kraus, K. A., and
Johnson, J. S., Jr.
8. PERFORMING ORGANIZATION REPORT NO,
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Water Research Program, Oak Ridge National Laboratory''
P. 0. Box X
Oak Ridge, Tennessee 37830
10. PROGRAM ELEMENT NO.
1BC611
11. CONTRACT/GRANT NO.
WQO 14-12-832
AEC Interagency Agreement
40-191-69
12, SPONSORING AGENCY NAME AND ADDRESS
Municipal Environmental Research Laboratory
Environmental Protection Agency
Cincinnati, Ohio 45268
13, TYPE OF REPORT AND PERIOD COVERED
Final, 1970-1975
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
Operated by Union Carbide Corporation, for the U. S. Energy Research and Development
Administration
Cross-flow filtration was tested for separation of water from the effluent
of primary or secondary treatment of municipal sewage, after addition of physical-
chemical reagents. In this approach, the solution being filtered is pumped parallel
to the filter, to slow buildup of flux-limiting filtercake. In most cases, the filter-
ing surfaces were 1-inch fabric tubes, manufactured as fire-hose jackets. Results are
presented as functions of pressure, water recovery, circulation velocity, additive
concentration and other variables. With Fe(III), Al(III), or powdered activated carbon
(PAC) added to Oak Ridge primary sewage effluent, filtrate was superior in quality to
effluent from activated sludge secondary treatment, and compared favorably with report-
ed characteristics of sewage treated with similar reagents when solids-liquid separa-
tion was accomplished by settling. Turbidities were usually well below 1 JTU, and bac-
teria were substantially removed. With Fe(III), filtrate total organic carbon, (TOG)
typically was 10 to 15 mg/£, and phosphate below 1 mg/fc. With PAC, TOG was lower, usu-
ally about 5 mg/A, but phosphate removal was poor. With Fe(III) at about 0.001 M,
average fluxes of about 6 m/d (150 gpd/ft2) appear attainable at 4.6 m/sec (15 ft/sec)
circulation velocity with 24-hour backwash intervals. On this basis, we estimate fil-
trate cost for a 3,800 m3/d (106 gpd) plant using Fe(III) of 12c/m3 (46c/kgal). There
are reasonable possibilities of lower costs. Fluxes and product characteristics are
similar when feed is activated sludge secondary effluent, except that TOG is lower.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b. IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
*Filtration, *Activated carbon, *Chemical
precipitation, Tertiary treatment, Nutrient
removal, Organic matter, Phosphate, Iron,
Turbidity.
*Cross-flow filtration,
*Physical-chemical sewage
treatment, Iron salt
additives, Aluminum salt
additives, TOG removal,
Phosphate removal, Ultra-
filtration, Oak Ridge
13 B
13. DISTRIBUTION STATEMENT
Release unlimited
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
127
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
------- |