EPA-670/2-75-050a
June 1975
Environmental Protection Technology Series
National Environmental Research Center
Office of Research and Development
jnmental Protection Agency
Cincinnati, Onto 45288
-------
EPA-670/2-75-050a
June 1975
DIRECT FILTRATION OF LAKE SUPERIOR
WATER FOR ASBESTIFORM FIBER REMOVAL
By
Black § Veatch, Consulting Engineers
Kansas City, Missouri 64114
Program Element No. 1CB047
Contract No. DACW 37-74-C-0079
Interagency Agreement EPA-IAG-DA-0388
USEPA, Region V and Corps of Engineers, St. Paul
Project Officer
Gary S. Logsdon
Water Supply Research Laboratory
National Environmental Research Center
Cincinnati, Ohio 45268
NATIONAL ENVIRONMENTAL RESEARCH CENTER
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
REVIEW NOTICE
The National Environmental Research Center, Cincinnati, has reviewed this report
and approved its publication. Approval does riot 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.
11
-------
FOREWORD
Man and his environment must be protected from the adverse
effects of pesticides, radiation, noise and other forms of pollution,
and the unwise management of solid waste. Efforts to protect the
environment require a focus that recognizes the interplay between
the components of our physical environment -- air, water, and land.
The National Environmental Research Centers provide this multidis-
ciplinary focus through programs engaged in
studies on the effects of environmental
contaminants on man and the biosphere, and
a search for ways to prevent contamination
and to recycle valuable resources.
This report presents the results obtained and conclusions
drawn from pilot plant filtration research for the removal of
asbestiform fibers from drinking water. The appendices, which are
available separately, present detailed information on water quality,
pilot plant equipment and operation, individual filter run data,
asbestiform fiber and amphibole mass concentrations in raw and
filtered water, and diatomite filter optimization.
Arrangements for performance of this research were made
through an interagency agreement between EPA Region V and the Corps
of Engineers, St. Paul, Minnesota.
A. W. Breidenbach, Ph.D.
Director
National Environmental
Research Center, Cincinnati
in
-------
ABSTRACT
Pilot plant research conducted in 1974 at Duluth, Minnesota, demonstrated that
asbestiform fiber counts in Lake Superior water could be effectively reduced by
municipal filtration plants. During the study, engineering data were also obtained
for making cost estimates for construction and operation of both granular and
diatomaceous earth (DE) media filtration plants ranging in size from 0.03 to
30 mgd.
Both dual and mixed-media granular filters using alum and nonionic polymer,
employing flash mix and flocculation without settling, and DE filters with alum
coated DE as precoat and/or body feed or with Catfloc B added to raw water,
produced effluents with amphibole fiber counts below electron microscope
detection limits. Turbidity was not a direct measure of fiber count, but amphibole
counts were generally lowest at effluent turbidities sO.l TU. Chrysotile removal
was more difficult, but mixed media granular filtration with alum and nonionic
polymer, and DE filtration with anionic polymer conditioned DE frequently
reduced chrysotile fiber counts markedly.
Systems for economic reasons recommended for consideration during design
studies are:
1. Mixed media direct filtration, 5 gpm/ft2, multiple-stage flash mix.
2. Dual media filtration, 4 gpm/ft2, single stage flash mix.
3. Pressure DE filtration 1 gpm/ft2, alum conditioning of precoats and body
feed, or alum conditioning of precoat only, and cationic polymer fed to raw
water.
IV
-------
CONTENTS
Abstract ii
List of Figures iv
List of Tables vi
Acknowledgments viii
Sections
I Conclusions 1
II Recommendations 4
III Introduction 6
IV Equipment Design, Installation and Operation 9
V Sampling and Analysis 27
VI Results 33
VII Discussion 70
VIII References 102
IX Glossary 103
X Index of Appendices 105
-------
FIGURES
No. Page
1 Flow Diagram for MM-1 13
2 Flow Diagram for MM-2 14
3 Alternative Mixing Modes for MM-2 15
4 Flow Diagram for BIF (Alt. 1) 20
5 Flow Diagram for ERD 22
6 Raw Water and Effluent Turbidity Curves 39
Unit MM-1 Run 31
7 Raw Water and Effluent Turbidity Curves 40
Unit MM-1 Run 32
8 Head Loss and Effluent Turbidity Curves 45
Unit MM-2 Run 107
9 Head Loss and Effluent Turbidity Curves 46
Unit MM-2 Run 109
10 Raw Water and Effluent Turbidity Curves 53
Unit ERD-2 Run 47
11 Raw Water and Effluent Turbidity Curves 54
Unit BIF Run 25
12 Head Loss and Effluent Turbidity Curves 55
Unit BIF Run 110
13 Head Loss and Effluent Turbidity Curves 56
Unit ERD-2 Run 79
14 Head Loss and Effluent Turbidity Curves 57
Unit ERD-2 Run 21
15 Head Loss and Effluent Turbidity Curves 58
Unit ERD-2 Run 34
16 Typical MM- 2 Filtered Water Turbidity Readings 66
Obtained with HACH 2100A and HACH 1720 Turbidimeters
17 Typical MM-2 Raw Water Turbidity Readings 67
Obtained with HACH 2100A and Monitek Turbidimeters
18 Typical ERD Filtered Water Turbidity Readings 68
Obtained with HACH 2100A and Monitek Turbidimeters
VI
-------
FIGURES (CONTINUED)
No.
age
19 Typical MM-1 Filtered Water Turbidity Readings 69
Obtained with HACH 2100A and Monitek Turbidimeters
20 Relationship Between Raw Water Turbidity at Duluth 71
Lakewood Intake and ORF Amphibole Fiber Counts
21 Relationship Between Raw Water Turbidity at Duluth 72
Lakewood Intake and ORF Chrysotile Fiber Counts
22 Relationship Between Raw Water Turbidity at Duluth 73
Lakewood Intake and NWQL Amphibole Mass Concentration
23 Relationship Between Raw Water Turbidity at Duluth 74
Lakewood Intake and NWQL Suspended Solids Concentration
24 Effluent Turbidity vs. Amphibole Fiber Count 79
Granular Media
25 Effluent Turbidity vs. Chrysotile Fiber Count 81
Granular Media
26 Effluent Turbidity vs. Amphibole Fiber Count 85
Vacuum DE Filtration
27 Effluent Turbidity vs. Chrysotile Fiber Count 86
Vacuum DE Filtration
28 Effluent Turbidity vs. Amphibole Fiber Count 87
Pressure DE Filtration
29 Effluent Turbidity vs. Chrysotile Fiber Count 88
Pressure DE Filtration
VII
-------
TABLES
No. Page
1 Dimensions of Unit Process of Granular Filters 11
2 Sieve Analysis of Filter Media 17
3 Polymers Utilized in Filter Runs 18
4 Dimensions of DE Units 19
5 Equivalent DE Filter Aids 24
6 Comparative Physical Properties of Amphibole and 30
Chrysotile Fibers in Lake Superior Water
7 Granular Media Filtration (MM-1) Summary of Asbestiform 35
Fiber Removal by Category
8 Granular Media Filtration (MM-2) Summary of Asbestiform 36
Fiber Removal by Category
9 Number of Runs Made at or Near Indicated Filter Rate 41
10 Results Obtained Using Dual and Tri-Media Filter Beds 43
11 Relationship of Backwash Solids to Sludge Volume 47
Unit MM-2
12 Vacuum Diatomaceous Earth Filtration Summary of 50
Asbestiform Fiber Removal by Category
13 Pressure Diatomaceous Earth Filtration Summary of 51
Asbestiform Fiber Removal by Category
14 Relationship of Backwash Solids to Sludge Volume 64
DE Units
15 Granular Media Filtration (MM-2) Amphibole and 75
Chrysotile Removal (ORF Data)
16 Comments on Chrysotile Removal 76
17 Granular Media Filtration (MM-2) Filter Rates 78
5 gpm/ft2 and Greater (ORF Data)
18 Vacuum (BIF) and Pressure (ERD) Diatomaceous Earth 82
Filtration Selected Filter Runs (ORF Data)
19 Water Treatment Plant Economic Analysis Granular Media 90
Filtration 30 MGD Plant Design Lake Superior Intake
at Lakewood
Vlll
-------
TABLES (CONTINUED)
No. Page
20 Water Treatment Plant Economic Analysis - Pressure 92
Diatomaceous Earth Filtration 30 MGD Plant Design
Lake Superior Intake at Lakewood
21 Ranges of Initial Capital Costs of Various Plant 95
Capacities Utilizing Granular Media and Pressure
DE Filtration for Asbestiform Fiber Removal
22 Summary of NWQL Results 97
23 Comparison of Amphibole Mass and EM Results in 98
Filtered Water Samples
24 Mean Suspended Solids in Filtrate 100
IX
-------
ACKNOWLEDGMENTS
The cooperation and the assistance of the City of Duluth, Minnesota, throughout
the pilot filtration study, is gratefully acknowledged. A special thanks is due to
Mr. Dave Peterson, Director of Water and Gas, for his help in providing working
space, electrical power, and other assistance during, the study. Invaluable assistance
was also provided by Mr. Art Biele, Chief Chemist, and Mr. Ted Olson, Chief
Operator, at the Lakewood Pumping Station. Other City personnel assisting in the
study were Mr. Tom Parr, Mr. Bill Holmes, Mr. Arvid Benson, and Mr. Terry
Schumann, the pilot plant operator provided by the City.
The assistance of several local Federal personnel from the U. S. Environmental
Protection Agency-National Water Quality Laboratory at Duluth and the United
States Army Corps of Engineers at Duluth also is acknowledged. The laboratory
and administrative services they provided were essential to a successful study.
They were:
Dr. Donald Mount National Water Quality Laboratory
Duluth, Minnesota
Mr. Mike Lubratovich National Water Quality Laboratory
Duluth, Minnesota
Dr. Phil Cook National Water Quality Laboratory
Duluth, Minnesota
Dr. Gary Glass National Water Quality Laboratory
Duluth, Minnesota
Mr. Court Mueller U. S. Army Corps of Engineers
Duluth, Minnesota
Mr. Leo Zygmanski U. S. Army Corps of Engineers
Duluth, Minnesota
The cooperation and assistance of the following manufacturers and their
representatives who provided both materials and assistance is gratefully acknow-
ledged :
Mr. E. E. Halverson Northwestern Power Equipment Co., Inc.
St. Paul, Minnesota
Mr. Clifford Powell Johns-Manville Products Corp.
Chicago, Illinois
-------
Mr. William Graves
Mr. Roger Winters
Mr. Bryce Johnson
Mr. Bob Miller
Mr. T. W. Hubner
Mr. Paul Coffman
Mr. Dean Newby
Mr. John Ebeling
Mr. Patrick Jayne
Mr. Don Anspach
Mr. Henry Mueller
Mr. Walter Conley
Mr. Frederick Siegele
Mr. J. W. HcHugh
Eagle-Picher Industries, Inc.
Minneapolis, Minnesota
Jeno's
Duluth, Minnesota
Johns-Manville Sales Corp.
N. Kansas City, Missouri
Dicalite Division-Grefco, Inc.
St. Louis, Missouri
Calgon Corp.
Tulsa, Oklahoma
Coffman Industries, Inc.
Builders Iron Foundry
Kansas City, Kansas
Eagle Picher Industries, Inc.
Cincinnati, Ohio
Eagle Picher Industries, Inc.
St. Louis, Missouri
Dow Chemical Co.
Shawnee Mission, Kansas
R. P. Adams Co.
Buffalo, New York
Neptune MicroFLOC, Inc.
Corvallis, Oregon
American Cyanamid Co.
Chicago, Illinois
Builders Iron Foundry
West Warwick, Rhode Island
The assistance of Professor Walter Johnson of the University of Minnesota
involving the loan of a pilot water filter apparatus is sincerely appreciated.
XI
-------
SECTION I
CONCLUSIONS
GENERAL
1. The raw water turbidity experienced at the Duluth Lakewood Intake during
the pilot studies ranged from 0.4 to 6.3 TU, but was less than 1.0 TU 90 per cent
of the time.
2. Heavy rainfall and high winds resulted in increases of suspended solids,
turbidity, and bacteriological counts as well as changes in temperature in the
Duluth raw water; however, only high winds, generally from the east and north-
east, resulted in increases in amphibole fiber concentrations.
3. A general link was evident between the turbidity and the suspended solids
concentration of the raw water from the Duluth Lakewood Intake.
4. The data indicate a close connection in turbidity readings utilizing a Hach
2100A laboratory turbidimeter, a Hach 1720 in-line turbidimeter, and a Monitek
in-line turbidimeter.
5. No discernible tie was evident between the Duluth raw water turbidities and
the asbestiform fiber levels.
6. At finished water turbidities of less than 0.1 TU, the amphibole fiber count
and mass determinations were usually below the detection limits of the analytical
method used.
7. A general association was indicated between the NWQL amphibole mass
concentration and the ORF amphibole fiber counts in the Duluth raw water.
8. No relationship was observed between the counts of the amphibole and the
chrysotile fibers in the Duluth raw water.
9. Proposed treatment facilities are capable of meeting the new Federal Safe
Drinking Water Act standards which establish a maximum turbidity in drinking
water of 1.0 TU.
-------
TREATMENT FOR ASBESTIFORM FIBER REMOVAL
Granular Media Filtration
Based on achieving BDL or near it, 32 of 34 MM-2 (granular) runs and 21 of 23
MM-1 runs were successful for amphibole removal. Only 8 of 34 MM-2 runs and 2
of 23 MM-1 runs were successful for chrysotile removal.
Diatomaceous Earth Filtration
For the ERD (pressure) runs, 19 of 27 were successful for amphibole removal but
only 4 of 27 were successful for chrysotile removal. Vacuum DE filtration (BIF)
was not found suitable for treating the raw water being tested.
TREATMENT FOR TURBIDITY REMOVAL
Granular Media Filtration
1. Within the range of raw water turbidities experienced during the pilot study,
the effluent turbidity, length of filter run, sludge production, and asbestiform
fiber removal were more affected by the choice and amount of chemicals than by
the raw water turbidity.
2. Alum was a more effective coagulant than ferric chloride in the Duluth raw
water, producing lower effluent turbidities and longer filter runs.
3. Of the polymers tested in the pilot study, a nonionic polymer, 985N, was the
most effective in preventing turbidity breakthrough.
4. Resistance to turbidity breakthrough was greater with two-stage flash mixers
than with in-line mixers. Both systems were more effective in terms of filter run
length and turbidity removal than single-stage mixing.
5. Sedimentation prior to filtration increased filter run length in proportion to
the amount of solids removed in the sedimentation process, but did not increase
fiber removal.
6. A sludge production of approximately 0.03 pound per 1000 gallons and a
backwash volume of approximately 1.4 percent of the treated water resulted
from an average granular media filter run.
7. Amphibole fiber removal accomplished by the tri-media filter exceeded that
accomplished by the dual media filters and the DE filters.
-------
8. Tri-media filtration with two stage flash mixing resulted in the lowest
chemical costs.
9. For raw water conditions similar to those experienced during the pilot
studies, granular filters operated under positive head conditions would minimize
air binding of the filter.
10. Granular filtration rates approaching 352 m3/m2 day (6 gpm/ft2) were
effective in the removal of turbidity and amphibole fibers.
11. The amphibole content of settled backwash wastewater was found to be the
same order of magnitude as that of the raw water.
Diatomaceous Earth Filtration
1. The vacuum DE unit was affected more adversely by the release of dissolved
gases from the raw water than the pressure DE unit, resulting in turbidity break-
through and passage of DE through the filter septum.
2. Total precoat weights of 0.731 to 0.974 kilogram per square meter (0.15 to
0.2 pound per square foot) of filter area were required to eliminate high effluent
turbidity and to maintain stability in the precoat.
3. A change in grade of DE in the body feed application from fine to coarse
resulted in an increase in the filter run length and higher effluent turbidities.
4. The rate of head loss buildup and length of filter run in the DE units were
affected more by the type of precoat and the type and amount of DE and
chemicals in the body feed process than by the turbidity of the raw water.
5. A medium grade precoat and a fine grade body feed were most effective in
turbidity and asbestiform fiber removal.
6. For the Duluth raw water, two treatment conditions, alum and soda ash
added to the precoat with a cationic polymer introduced to the raw water, and an
anionic polymer added to the precoat, were most effective in turbidity and
asbestiform fiber removal.
7. Although an optimum flow rate of 94 m3/m2 day (1.6 gpm/ft2) is theoreti-
cally obtainable, the rate of 60 m3/m2 day (1.0 gpm/ft2) as employed in the
pilot study will result in substantial turbidity and asbestiform fiber removal.
-------
SECTION II
RECOMMENDATIONS
GENERAL
1. Quality control of the filter operation should be based on maintaining a
finished water having a turbidity not greater than 0.1 TU.
2. In-line turbidimeters should be used for continuous monitoring of the raw
and finished water turbidity.
3. Electron microscope analysis for asbestiform fiber counts should be per-
formed periodically on raw and finished water samples to check on the effective-
ness of the treatment processes for fiber removal.
4. Analysis for amphibole mass by X-ray diffraction in raw and finished water
should be performed more frequently than electron microscope analysis due to
the lower costs and more rapidly obtainable results involved.
5. Costly and time-consuming electron microscope analysis for asbestiform fiber
counts should be replaced by improved analytical techniques as soon as they
become available.
6. For large capacity plants, granular filtration of Lake Superior water is re-
commended over diatomaceous earth filtration. For small plants, diatomaceous
earth filtration should also be considered.
GRANULAR MEDIA FILTRATION
1. Two stage flash mixing followed by flocculation is recommended for condi-
tioning the raw water prior to filtration.
2. A positive head tri-media filter designed to operate at a rate of 290 m3/m2
^
day (5 gpm/ftz) at design flow is recommended.
3. For treatment of Lake Superior water at the Duluth Lakewood Intake, alum
at 12-20mg/l and 985N or an equivalent nonionic polymer at 0.05 mg/1 is re-
commended when the raw water turbidity is near 1.0 TU.
4. Although sedimentation is not required for fiber removal, presedimentation
could be used to reduce the solids loading to the filters and to extend filter run
lengths.
-------
5. Provision should be made for the introduction of polymer to the flash
mixers, the flocculation chambers, the filter beds, or to any combination of the
three locations. In addition, provision for the addition of a second polymer
should similarly be made as a potential aid in chrysotile removal.
6. Backwash wastewater should be discharged to a sludge lagoon, with return of
the supernatant to the treatment plant.
D1ATOMACEOUS EARTH FILTRATION
1. For the conditions experienced during the pilot studies, pressure DE filtration
is recommended with provisions for addition of chemical aids to the precoat, the
body feed, and the raw water.
O n
2. The DE filters should be designed for a flow rate of approximately 60 m /m
day (1.0 gpm/sq ft2).
3. A medium grade DE at approximately 0.974 kilogram per square meter (0.2
pounds per square foot) of filter area coated with alum at O.Olg/g of DE and
soda ash at 0.005 g/g of DE and a cationic polymer at 0.5 mg/1 added to the raw
water or an equal grade and amount of DE coated with an anionic polymer at
0.00005 g/g of DE should be used for precoating for treatment of Lake Superior
water at the Duluth Lakewood Intake.
4. A fine grade of DE at a rate of 20 to 30 mg/1 should be used for body feed
for treatment of Lake Superior water at the Duluth Lakewood Intake.
-------
SECTION III
INTRODUCTION
PURPOSE OF STUDY
The pilot filtration study was conducted to determine which of the filtration
systems studies was most effective in the removal of asbestiform fibers, and to
obtain for each system the necessary data for design of a full-scale water treat-
ment plant. Data to be obtained included raw and filtered turbidities, asbestiform
fiber removal efficiencies, length of filter run, rate of head loss buildup, and
sludge produced per 1,000 gallons (gal) of water treated.
BACKGROUND
Prior to initiation of the pilot filtration study, little was known about the removal
of asbestiform fibers from water. However, studies conducted by the Corps of
Engineers did indicate a substantial portion of the asbestiform fibers could be
removed from the water using pressure diatomaceous earth (DE) filters. In
addition, studies at the U. S. Environmental Protection Agency (EPA) laboratory
in Cincinnati indicated that granular filters could also effectively remove much of
the asbestiform fibers from the water. The selection of granular and DE filters for
testing during the pilot filtration study was made on the basis of the work
conducted by the U. S. Army Corps of Engineers and the EPA.
SCOPE OF STUDY
During the period from April through September, 1974, four pilot filter units
were operated in Duluth, Minnesota, on Lake Superior water. The primary goal of
the pilot filtration study was the investigation of the ability of various filtration
systems to remove the asbestiform fibers and turbidity found in the water
supplies of the Duluth-Superior-Cloquet area, with emphasis on removal of fibers
and turbidity from the raw water obtained at the Lakewood Intake in Duluth,
Minnesota. Filtration processes investigated included granular filtration, pressure
DE filtration, and vacuum DE filtration. The study was originally scheduled to be
completed after a period of 4 months, but results obtained during this period
indicated further study would be beneficial. The study was therefore extended
for 5 weeks, with additional variables in operation investigated, especially to
increase the removal of chrysotile fibers. A total of 227 granular filter runs and
228 DE filter runs were conducted. Limitations of time and money prevented
study of other methods of water treatment which might have been considered.
-------
Variables studied during the pilot plant operation are summarized below:
ELEMENTS OF STUDY
Granular Media Filtration
1. Filtration with and without prior sedimentation.
2. Effectiveness of combinations of iron salt and polymers and aluminum salt
and polymers.
3. Dual media versus tri-media filter bed.
4. Effect of filtration rate.
5. Rate of head loss buildup.
6. Sludge production.
7. Effect of seasonal conditions.
8. Effect of raw water turbidity on filter performance.
9. One-stage versus two-stage flash mixing.
10.* Flash mixing chambers versus in-line mixers.
11.* Techniques for increased removal of chrysotile as well as amphibole fibers.
12.* The effect of higher raw water algal counts on filter performance (no algal
blooms occurred).
Diatomaceous Earth Filtration
1. One-step versus two-step precoating.
2. Effect of body feed concentration.
3. Addition of chemical aids to precoat and/or body feed.
4. Rate of head loss buildup.
5. Sludge production.
6. Effect of seasonal conditions.
7. Effect of raw water turbidity on filter performance.
* Variables studied during the 5 week extension.
-------
8.* Techniques for increased removal of chrysotile as well as amphibole fibers.
9.* The effect of higher raw water algal counts on filter performance.
10.* Generation of adequate data for input to the POPO computer program.
* Variables studied during the 5 week extension.
-------
SECTION IV
EQUIPMENT DESIGN, INSTALLATION, AND OPERATION
GENERAL
The pilot filtration study was conducted with two granular filters and four DE
filters installed at the Lakewood Pumping Station in Duluth, Minnesota. The
granular filters were a skid-mounted and a trailer mounted "WATER-BOY" unit
as manufactured by Neptune MicroFLOC, Incorporated, Corvallis, Oregon. The
DE filters were two pressure DE units supplied by the Corps of Engineers and
two vacuum DE units as manufactured by the BIF Corporation, Providence,
Rhode Island.
RAW WATER SOURCES
With the exception of three runs on the MM-2 units, all runs on the pilot filtra-
tion units were made using the Lakewood wetwell as the raw water source. The
intake of the Lakewood Pumping Station extended approximately 1,500ft out
into Lake Superior and was about 70 ft below the surface of the lake. Suction
hoses were placed directly in the wetwell and raw water was pumped to each of
the filter units so that the raw water supplied to the units was the same as that
being supplied to the City of Duluth.
Raw water from the Cloquet Pipeline Intake was used for three runs on the MM-2
units. The Cloquet raw water was transported from the Cloquet raw water
pumping station on Minnesota Point to the Lakewood Pumping Station in a 4,800
gal tank truck. After the truck arrived at Lakewood, the suction of one of the
pilot plant raw water pumps was connected to the discharge from the tank truck
and the water pumped to the MM-2 unit.
PERSONNEL
The operation of the pilot filtration systems was conducted by three operators
and a field engineer. The pilot filters were operated in 8-hour shifts, 24 hours a
day, 5 days a week for a period of 5 months extending from April 19, 1974 to
September 20, 1974. The study was conducted in three phases:
1. A preliminary test phase to train the operators, develop testing and data
recording procedures, and gain plant operating experience.
-------
2. A treatment efficiency phase to determine which treatment methods gave
promise of effectively removing asbestiform fibers from the raw water.
3. An "optimum-design" phase. The operation of the DE filters during this
phase was conducted to determine filter cake resistance measurements necessary
for optimum design of DE filters to provide least cost of filtration of Lake
Superior water. The granular filters were operated during this phase with several
variations in the physical set-up of the filter systems to determine the best process
scheme for asbestiform fiber removal.
AMBIENT CONDITIONS
To determine if any relationship existed between ambient conditions and the
quality of the raw water at the Lakewood Pumping Station, daily records were
maintained on temperature, wind velocity and direction, precipitation, and lake
level. The daily temperatures and wind velocities and directions were obtained
from records kept by the operators at the Pumping Station. The Pumping Station
records are not official, but because of the wide discrepancy between weather
conditions at the Pumping Station and the National Weather Service Station, it was
decided that the unofficial records would better represent the conditions at the
Pumping Station Intake. No record of precipitation was kept at the Pumping
Station. Therefore, official National Weather Service data were used for precipi-
tation values. Lake level data were obtained from the Corps of Engineers station in
Duluth, and are based on International Great Lakes datum. Records of tempera-
ture, wind velocity and direction, precipitation, and lake level are presented in
Appendix A-l.
A record was also maintained of the Lakewood raw water pumping schedule. This
record is presented in Appendix A-2. All data in the tables refer to Pump No. 2 as
this was the only raw water pump which the City operated during the course of
the pilot filtration study.
GRANULAR UNITS
Because the turbidity of Lake Superior water is generally low and in such
circumstances direct filtration has been found to be successful for turbidity
removal, this process was tried for fiber removal. In direct filtration, water is
conditioned with an inorganic coagulant and polymer and filtered (usually
10
-------
through dual or mixed media) without sedimentation. Selection of the particular
granular media used was based on the wide and successful experience with the
media chosen.
Size and Type
The two granular units used in the filtration study were designated as MM-1 and
MM-2. The only difference in the basic set-up of the two filters was in the size of
the settling basins. MM-1 had a settling basin with a volume of 1,708 liters (451
gal) and MM-2 a settling basin with a volume of 693 liters (183 gal) allowing for
approximately 30 centimeters of freeboard at the design rate utilized. Dimensions
of each of the granular units are presented in Table 1. Further details appear in
Appendices B-l and B-2.
TABLE 1. DIMENSIONS OF UNIT PROCESSES OF GRANULAR FILTERS
Pilot Unit
MM-1
Mixing chamber
Flocculation chamber
Sedimentation chamber
Filter chamber
Clearwell
MM-2
Mixing chamber
Two-stage mixing chamber0
Kenics mixers
Flocculation chamber
Flocculation chamber0
Sedimentation basinc
Filter chamber
Clearwell
Length
cm
62.2
65.5
154.2
62.7
-
62.0
52.7
62.0
152.4
152.4
62.2
152.4
Width
cm
29.5
62.2
91.2
62.2
137.93
29.2
5.08a
61.5
30.0
30.0
62.0
91.4
Depth
cm
182.9
182.9
Varied 163 to
182.9
152.4
184.2
184.2
Varied 157 to
Varied 157 to
184.2
177.8
183
178
178
a Diameter rather than width.
b 208 liter (55 gal) drum.
c Used for flocculation on all runs except Run 140.
Mixing and Settling Variations
A flow diagram of the MM-1 (granular) unit is presented on Figure 1. Raw water
entered the flash-mixing chamber of the unit where the coagulant was added.
Detention time in the mixing chamber was 5.1 minutes (min) at a nominal flow
rate of 87 cubic meters per day (m^/day), or 16 gallons per minute (gpm).
11
-------
Calculation of all detention times was based on flow rates and unit volumes.
From the mixing chamber, the water entered the flocculation chamber where
the polymer was added. With sedimentation, the polymer was added at
the entrance to the sedimentation chamber (Figure 1). Detention time in the
flocculation chamber was 12.3 min. After flocculation, the water was discharged
either into the sedimentation basin or directly onto the filter, depending upon
which mode of operation was being used. A dual media filter bed with an
effective surface area of 0.37 square meter (m2), or 4 square feet (ft2) was used in
all runs conducted with the MM-1 unit. Flow through the filter bed was main-
tained at a constant rate by a float valve located on the effluent discharge line. A
water meter was installed in the effluent line to determine the rate of flow
through the unit and the total volume of water treated during each run.
Sedimentation ahead of the filter was provided by removal of a bypass line and
installation of tube settlers in the sedimentation chamber. The tube settlers were
hexagonal-shaped, installed on a slope of 5 degrees, with a diameter of 2.54 cm
(1 in) and a length of 1.2 meters (4 ft). A detention time of approximately
28 min was provided in the sedimentation chamber at a nominal flow rate of 87
m3/day (16gpm).
Flow diagrams of the MM-2 (granular unit) are presented on Figures 2 and 3. Raw
water was introduced into one of several points described below for chemical
mixing:
1. A flash mixing chamber.
2. A sequence of two 208 liter (55 gal) external flash mixing drums.
3. A sequence of Kenics in-line static mixers external to the unit.
The flash mix chamber provided a detention time of 5.0 min at a filter rate of 87
m^/day (16 gpm). Flow from the flash mix chamber was discharged into a floc-
culation chamber, with a detention time of 10.3 min, where the polymer was
added.
For the sequential backmix system, flash mixing occurred in two steps in two 208
liter (55 gal) drums. The primary coagulant was added at Point C, (Figure 3)and
the polymer was added at Point D (Figure 3). At a filter rate of 87 m^/day
(16 gpm), the theoretical detention time was 2.8 min in the first drum and 2.2
min in the second drum. When three treatment chemicals were used in sequence,
the third was added at Point D (Figure 2).
12
-------
-FROM LAKEWOOD
WETWELL
FLASH MIXER
FLOCCULATOR DRIVE
FLOAT VALVE TO
CONTROL FLOW
THRU FILTER
-THROTTLING
VALVE
BACKWASH
TO WASTE
FILTER
MEDIA-
HOSE BIBB TO
MONITEK
TURBIDIMETER
O WATER METER
THROTTLING
VALVE
.
TUBE
SETTLERS
BACKWASH
VALVE
''OVERFLOW
TO WASTE
EFFLUENT PUMP
BACKWASH PUMP
MIXING FLOCCULATION SEDIMENTATION
CHAMBER CHAMBER BASIN
FILTER
BED
CLEARWELL
FIGURE I . FLOW DIAGRAM FOR MM-I.
-------
FLASH
MIXER
FLOCCULATOR
DRIVE
WATER METER
COMPRESSED
AIR-
FLOAT VALVE TO
CONTROL FLOW
THRU FILTER
FILTER
MEDIA"
BACKWASH
TO WASTE
BACKWASH
VALVE
/HOSE BIBB TO
( HACH IN-LINE
1 TURBIDIMETER
?
RAW WATER
Cm/i
OVERFLOW
TO WASTE
EFFLUENT PUMP
BACKWASH PUMP
MIXING
CHAMBER
(SEE FIGURE 3
FOR ADDITIONAL
MIXING MODES)
FLOCCULATION FLOCCULATION FILTER
CHAMBER CHAMBER BED
CLEARWELL
FIGURE 2 . FLOW DIAGRAM FOR MM-2.
-------
1725 RPM MIXER-) TO FLOC
CHAMBER
TO FLOC CHAMBER
rl fZJ) NPM MIXEK1
X] D
"t
^a
9
1
^a
9
.TO FLOC
/CHAMBER
V
55 GALLON DRUMS'
RAW WATER
KINECS CORP.
STATIC MIXER
RAW WATER
KINECS CORP.
STATIC MIXER
RAW WATER
XXHX
FIGURE 3 .ALTERNATIVE MIXING MODES FOR MM-2 .
-------
With the sequential in-line mixing, shown on Figure 3, a combination of three or
six 5.08 cm (2 in) diameter Kenics static mixers were used. Provision was made
for adding up to three different treatment chemicals at Points E, F, and G.
Mixing of water and chemical occurred in either one or two mixers before the
next chemical was added. At a filter flow rate of 87 m^/day (16gpm), the
detention time in each static mixer module was 1.1 seconds.
Flow from the mixing portion of the process was discharged into a flocculation
chamber, with a detention time of 10.3 min at a filter rate of 87 m^/day,
(16 gpm). An additional 11.4 min of flocculation was provided by removing the
tube settlers from the sedimentation chamber of the MM-2 unit and bubbling air
up through the flow as it passed through the chamber.
From the sedimentation chamber, the flow was discharged onto the filter. A dual
media filter bed was used in 76 runs and a tri-media filter bed in 64 runs
conducted with the MM-2 unit. The filter bed had an effective surface area of
9 9
0.37 mz (4 ftz). Flow through the filter was maintained at a constant rate by a
float valve located on the effluent discharge line. A water meter was installed in
the effluent line to establish filter flow rates and to determine the total volume of
water treated during each run.
Filtration Rates
All runs on the MM-1 unit were conducted at a flow rate of approximately 232
o 9
cubic meters per square meter day (nr/mz day), or 4 gallons per minute per
square foot (gpm/ft2).
Most of the runs on the MM-2 unit were conducted at a flow rate of approxi-
-3 ^ s-\
mately 232 m /m day (4 gpm/ft ). However, to determine the effects of flow
rate on the performance of the filter, 14 runs were conducted at a flow rate of
approximately 290 m3/m2 day (5 gpm/ft2), 3 runs at about 116 m3/m2 day
(2 gpm/ft2), and 22 runs at approximately 350 m3/m2 day (6 gpm/ft2).
Media Description
The results of a sieve analysis on the various medias and the composition of the
dual and tri-media beds used during the pilot study are given in Table 2.
16
-------
TABLE 2. SIEVE ANALYSIS OF FILTER MEDIA3
Media
Anthracite
Brady sand
Fine illmenite
Coarse illmenite
Very coarse illmenite
Effective Size
mm
1.10
0.42
0.22
1.02
1.75
Uniformity
Coefficient
1.60
1.52
1.66
1.55
2.06
Media
Thickness
Dual Media Tri Media
cm cm
53.3
22.9
--
41.9
22.9
11.4
3.8
3.8
a Analysis performed by Neptune MicroFLOC, Incorporated.
The dual media bed was used for all runs on MM-1 and 76 runs on MM-2 and was
placed directly on the carborundum filter bed supports provided with each of the
granular units. The tri-media bed was used for the last 64 runs en the MM-2 unit.
To insure that none of the fine illmenite would pass through the carborundum
support, two coarser sizes of illmenite were placed on the carborundum support
prior to installation of the bed.
The dual media beds were installed under the supervision and direction of a
representative of Neptune MicroFLOC, Inc. The tri-media bed was installed by
the field engineer with assistance from the pilot plant operators. The placement of
the beds was accomplished by adding an excess of the bottom layer of media,
backwashing the filter, skimming about one-half the excess media off, back-
washing again and skimming off the remainder of the excess media. This procedure
was repeated for each media.
Chemical Additions
Alum and ferric chloride (FeCl-^) were used as the primary coagulants in the
testing of the granular filters. The coagulant solutions were mixed daily at the
pilot plant site to a 1.19 percent solution strength using dry chemicals and raw
water from the Lakewood Intake.
The polymers used in the pilot filtration study were selected partially on the basis
of availability and partially on the basis of jar tests conducted at the Lakewood
laboratory. The polymers utilized in the filtration study are presented in Table 3.
17
-------
TABLE 3. POLYMERS UTILIZED IN FILTER RUNS
Polymera
A-23
847A
Separan NP-10
Catfloc B
C-31
573C
985N
N17
Type
Anionic
Anionic
Weak Anionic
Cationic
Cationic
Cationic
Nonionic
Nonionic
Manufacturer
Dow
Cyanamid
Dow
Calgon
Dow
Cyanamid
Cyanamid
Dow
Home Office
Midland, Mich.
Chicago, 111.
Midland, Mich.
Pittsburg, Pa.
Midland, Mich.
Chicago, 111.
Chicago, 111.
Midland, Mich.
a Each of these polymers is approved by the EPA for potable water use.
The liquid polymers were generally mixed for at least 30 min before use and the
powdered polymers were mixed in excess of 1 hour before use.
Backwashing
Backwashing of the granular units was initiated for one of two reasons, either
excessive head loss through the filter or excessive effluent turbidity. If
effluent quality was satisfactory, the filter was backwashed whenever the head
loss through the filter exceeded 8 ft of water. Effluent quality controlled
initiation of the backwash cycle whenever the turbidity exceeded 1.0 TU or the
turbidity of the raw water, whichever was lower, or whenever it became evident
that turbidity breakthrough had occurred. Turbidity breakthrough was evidenced
by a sudden and continuous rise in effluent turbidity readings.
A water only backwash with no surface wash was used for cleaning the granular
unit filter bed. To determine the amount of backwash water used and the rate of
backwashing, the length of backwash, and the level of water in the backwash tank
before and after each backwash cycle were recorded. Backwashing rates and length
"30 0
of backwash varied, but were usually about 755 m^/m day (13gpm/ftz) and 7
min respectively.
DIATOMACEOUS EARTH UNITS
Four DE filter units were installed at the pilot plant site, two vacuum filters and
two pressure filters. Both vacuum and pressure units were installed to determine if
18
-------
release of dissolved gases under vacuum conditions would be detrimental to the
filtration process. Although two vacuum filters were installed, only one was used
for the pilot study, with the second filter provided as a standby unit. The
vacuum DE filter was designated as the BIF unit. The pressure DE filters were
designated ERD-1 and ERD-2,but the two pressure filters were identical and all
data and discussion will be presented under the heading ERD. The dimensions of
the units are given in Table 4.
TABLE 4. DIMENSIONS OF DE UNITS
Pilot Unit Length
cm
Width
cm
Depth
cm
BIF
Filter section 78.7 20.32 Varied 72.4 to 76.2
Raw water tank - 68.6a 92.7
DE slurry or polymer tank - 35.Oa 50.8
ERD
Water treatment section - Varied 67.3 to 221.03 147
Filter section - Varied 7.6 to 40.6a 137
DE slurry or polymer tank - 61.0 91
a Diameter rather than width
Vacuum Unit
A flow diagram of the BIF unit is presented on Figure 4. Raw water entered the
mixing chamber where chemical or body feed additions, if any, were made and
mixing was accomplished. The water level in the mixing chamber was controlled
by a float valve installed on the influent raw water line. Dry body feed was added
by a helix feeder. If alum-coated body feed was being used, a DE slurry was
mixed in a slurry tank and fed to the mixing chamber. From the mixing chamber,
the water flowed into the filter section of the unit where the water passed
through the filter elements due to suction from the effluent pump, and was dis-
charged into the waste sump. The filter had an effective surface area of 0.93 m^
T
(10ftz). A water meter was installed in the effluent line to set flow rates and
determine the total amount of water treated during each run.
19
-------
Isi
O
FINISHED
WATER
WATER
METER
HEADLOSS
^-
f
TEN SQ.FT.
DE FILTER
-0*3-
BODY FEEDER
475 RPM
MIXER
HOSE BIBB FOR
BACKWASH-
RAW
WATER
DRAIN TO WASTE
VRAW WATER PLUS
BODY FEED
^RECIRCULATION LINE
EFFLUENT AND
RECIRCULATION PUMP
^VALVE CONT-
ROLLED BY
WATER LEVEL
IN TANK
^DRAIN TO WASTE
FIGURE 4 .FLOW DIAGRAM FOR BIF(ALT.I)
-------
Pressure Unit
A flow diagram of the ERD unit is presented on Figure 5. The unit consisted of a
water treatment section, a filter section, and a drain tank. The water treatment
section was designed to remove organic and suspended matter in the raw water to
a level suitable for further treatment in the filter section. However, the quality of
the raw water at the Lakewood Pumping Station was such that treatment of the
water ahead of the filter section was not considered necessary. Therefore, no
coagulants were added in the water treatment section. Because of the piping
configuration of the unit, water did flow through this section.
Each DE filter was a two-part cylinder containing six cylindrical elements, with
9 9
each element having an effective surface area of 0.15 rcr (1.67 ftz), providing a
total filtration area of 0.9 m^ (lOft^) per filter. Each filter was equipped with
gages for indicating influent and effluent pressure, a precoat funnel, an air release
valve, and a rate of flow controller in the effluent discharge line.
Filtration Rates
All runs on the BIF and ERD units were made at a flow rate of approximately
^ 0 "~)
58 m /m day (1 gpm/ft ), using raw water from the Lakewood Intake, and were
made without sedimentation prior to filtration.
Precoat Methods
Although the ERD unit is a pressure type filter and the BIF unit a vacuum type
filter, the principle of operation for the units is essentially the same. DE filtration
as conducted during the pilot filtration study consists of three basic steps:
1. Precoating
2. Filtration
3. Backwash
The purpose of the precoat is to give immediate clarity and to prevent the filter
septa from becoming clogged by impurities contained in the raw water. Precoating
is accomplished by circulating a slurry of DE through the filter septa. Since
most of the DE particles are larger than the openings in the filter septum, they
form a precoat by bridging these openings. After the precoat is formed on the
filter septa, the filtration process is begun by positioning appropriate valves on the
unit from the recirculation to the filter position, and initiating body feed to the
flow of water. This normal precoat procedure was altered somewhat when it was
21
-------
RECIRCULATION LINE
BODY FEEDER-*
r
UPFLOW CLARIFIER
TO MONITEK
RAW WATER-
-M-
EDUCTOR
AIR RELEASE
VALVE
FILTER PUMPH
TO BIF
FILTER WASH
VALVE
PRESSURE
kGAGE
FILTER
RECYCLE
VALVE
-TO WASTE
SUMP
FINISHED WATER
RATE OF
FLOW
CONTROLLER
FLOW METER
FIGURE 5 . FLOW DIAGRAM FOR ERD (ALT. I)
-------
discovered early in the research through analysis that a portion of the DE was
passing through the filter into the treated water. A two-step precoating procedure
was initiated in an effort to eliminate the DE bleed-through. The procedure when
precoating in two steps was similar to that used when precoating in one-step
except that two DE slurries were mixed, one using a coarse DE and one using a
fine DE. After mixing the two slurries and filling the filter section with raw
water, the recirculation pump was started and the coarse DE slurry poured into
the filter section. The unit was placed in recirculation until the water cleared and
then the fine DE slurry was poured into the filter section. The unit continued to
recirculate until the water once again cleared, at which time the recirculation
valve was closed and the effluent valve opened to the position giving the desired
flow through the filter unit. Bleed-through of DE through the filter septa was
discovered during the first week of testing. The two-step precoat method utilized
was successful in eliminating the bleed-through problem.
Body Feed
The purpose of the body feed is to maintain the permeability of the filter cake.
The amount of body feed added is dependent upon the concentration of sus-
pended solids contained in the raw water and must be carefully controlled to
achieve the maximum filter cake permeability. Too much DE relative to the
suspended solids concentration will add excessive thickness to the filter cake and
reduce the cake permeability. Too little DE will not sufficiently offset the effect
of the raw water suspended solids, and will either allow breakthrough of solids or
result in a reduction in the filter cake permeability.
DE Filter Aids
DE filter aids are produced by several manufacturers in several grades. Table 5 lists
the filter aids utilized in the pilot filtration study. The table lists the trade name
of filter aid grades from several manufacturers that are "equivalent". Filter aids
are considered "equivalent" when they produce approximately the same flow rate
and clarity under the same operating conditions.
23
-------
TABLE 5. EQUIVALENT DE FILTER AIDS
Brand Name
Celatom FW-6a
Celatom FW-12a
Celatom FW-20a
Celatom FW-50a
Celatom FW-60a
Mean Particle
Size
urn
6
8
12
15
19
Structure
Flux
Calcined
Flux
Calcined
Flux
Calcined
Flux
Calcined
Flux
Calcined
Equivalent Filter Aid
Celite 5126
Special Speedflowc
Hyflo Super Celb
Speedplusc
Celite 503b
Speedexc
Celite 53 5b
4200C
Celite 545b
4500C
a Manufactured by Eagle-Pitcher Industries, Inc.
Manufactured by Johns-Manville Products Corporation.
cManufactured by Dicalite Division Grefco, Inc.
DE Conditioning
Modifications made in the operation of the DE units during the course of the
pilot filtration study to enhance asbestiform fiber removal included:
1. Addition of chemical aids to the precoat
2. Two-step precoating
3. Addition of chemical aids to the body feed
4. The addition of polymer along with body feed into the raw water
Runs made with a single precoat and no chemical aids in the precoat or body
feed did not have satisfactory finished water turbidities. In an attempt to lower
the effluent turbidities, an anionic or cationic polymer was added to the precoat.
The precoat procedure when polymer was used in the precoat was unchanged
from the above except that polymer was added to the DE slurry prior to pre-
coating. The anionic polymer was added to the DE slurry as a 0.01 per cent
solution and from 0.00001 to 0.00005 gram of polymer per gram of DE (g/g DE)
24
-------
was added. The cationic polymer was added as a 0.1 per cent solution and from
0.0003 to 0.0006 g/g DE was added.
Several two-step precoats were made using a chemical aid in both the coarse and
fine steps. Polymers were added to the precoats in the same manner as when a
single precoat was used. Alum was added to the precoats as a 5 per cent solution
and from 0.01 to 0.11 g/g DE was added. In addition, whenever alum was added
to the precoats, sufficient soda ash, partly purified sodium carbonate, was added
to prevent depression of the pH to a value lower than 7.0. To allow time for alum
coating of the DE, the alum, soda ash, and DE were added to 3.78 liters (1 gal)
of water, the mixture was stirred, restirred after 5 min of quiescent settling, again
restirred after another 5 min of quiescent settling, and then poured into the filter
section. All other steps in the precoat procedure were the same as previously
described in this section.
The use of chemical aids in both precoats sometimes resulted in plugging of the
filter septa. To prevent plugging, the chemical aid was added only to the
second precoat, with the first precoat being used as a protective coat for the filter
septa.
Backwashing
As the permeability of the filter cake decreased, the head loss through the filter
increased until it reached a point where the unit must be backwashed. To control
the operation of the filters, the turbidity of the raw and finished water and the
head loss through the filters were recorded hourly for each of the DE units. The
total head loss through the filter which could be tolerated varied due to several
factors, with one of the more important being type of filter system used, i.e., a
pressure or a vacuum system. A pressure system could be operated at a much
higher pressure differential than a vacuum system. During the backwashing of the
unit, the DE, including the precoat, was completely flushed from the filter septa.
The unit was then ready for a subsequent precoat-filtration-backwash cycle.
Backwashing of the units was initiated for one of three reasons:
1. Excessive head loss through the filter
2. Excessive effluent turbidity
3. Time limitations
25
-------
If effluent quality was satisfactory and the length of filter run had not exceeded
24 hours, the BIF unit was backwashed when the head loss through the filter
reached 17 ft of water (7.5 psi).
If effluent quality was satisfactory and the length of filter run had not exceeded
24 hours, the ERD unit was backwashed when head loss through the filter
reached 103 ft of water (45 psi).
At the beginning of the pilot filter study, if the turbidity of the finished water
exceeded 1.0 TU or the turbidity of the raw water, whichever was lower, the
units were backwashed regardless of head loss or length of filter run. However, as
the study progressed, it was decided that a lower limit on effluent turbidity was
necessary and during the last 6 weeks of the study, the units were backwashed
whenever the effluent turbidity exceeded 0.20 TU.
If the effluent turbidity was satisfactory and the head loss through the units had
not reached terminal limits within 24 hours, the units were backwashed due to
time limitations. It was believed that any useful data which could be obtained
from a particular run would be obtained within this period and that any
extension of run length would not be beneficial.
26
-------
SECTION V
SAMPLING AND ANALYSIS
SAMPLING
A 1 liter grab sample of raw water and treated water from each unit was collected
during each 8-hour shift for analysis in the laboratory at the Lakewood Pumping
Station. Color, pH, alkalinity, hardness, odor, and temperature were run on each
sample and certain samples were also analyzed for aluminum and iron to
determine if there was any carryover of coagulant into the treated water.
Aluminum content was analyzed to determine the amount of alum which passed
through the filters as a measure of treatment effectiveness. Grab samples of treat-
ed water were taken during periods when coliform counts in the raw water were
high to determine the efficiency of the units in removing bacteria. When problems
occurred with air binding of the filter units, samples were collected to determine
the dissolved oxygen content of the water.
Backwash water samples were collected from Run 43 and all subsequent runs on
MM-2, and Run 14 and all subsequent runs on MM-1. Depending upon which
mode of operation was being used, i.e., sedimentation or no sedimentation, the
backwash sample was collected either from above the filter bed and also the drain
from the sedimentation basin, or only from above the filter bed. Most backwash
samples were collected by grabbing a portion of backwash water at the beginning,
the middle, and near the end of the backwash cycle. However, subsequent to Run
120, a continuous sampling technique was used to collect backwash samples from
MM-2. The continuous sampling technique consisted of placing a siphon hose in
the backwash water discharge and collecting a sample throughout the duration of
the backwash cycle.
Grab samples for asbestiform fiber analysis were usually collected twice per week
from the Lakewood Intake wetwell and also from the effluent of each of the
pilot filters. The 1 liter samples occupied approximately 2/3 of the volume of the
sample bottles. Most samples were collected in triplicate, with a portion going to
Ontario Research Foundation (ORF), to the School of Medicine at the University
of Minnesota at Duluth (UMD), and to NWQL at Duluth. Samples for ORF and
UMD were collected in 1 liter plastic containers and samples for NWQL were
collected in 18.9 liter (5 gal) plastic containers. The containers used to collect
samples for asbestiform fiber analysis were either new or had been specially
27
-------
cleaned prior to use. In addition, each container was thoroughly rinsed three
times with sample before sample collection.
CONVENTIONAL WATER QUALITY ANALYSES
Except for temperature, which was taken at the time of sample collection,
analyses of the 1 liter samples collected during each shift were made daily. Color
tests were conducted using Nessler tube color standards prepared according to the
13th Edition of Standard Methods.1
The pH of each sample was determined using a Beckman Model H-2 pH meter.
Odor tests were subjective and results were recorded only as odor or no odor.
Alkalinity, hardness, aluminum, and iron were determined using premeasured
reagents obtained from Hach Chemical Corporation, Ames, Iowa. Bacteriological
tests were made according to the 13th Edition of Standard Methods1, using the
membrane filter technique, and included total coliform, fecal coliform, and fecal
streptococci tests. The suspended solids analyses made on the backwash samples
and the dissolved oxygen tests were also conducted as set forth in the 13th Edi-
tion of Standard Methods.1 Results of these tests are provided in Appendices A-3,
A-4, and A-5.
TURBIDITY ANALYSES
Turbidity measurements were made with three different lurbidimeters, a Hach
Model 2100A Laboratory Turbidimeter, a Hach Model 1720 Low Range In-Line
Turbidimeter, and a Monitek Model 215/130 In-Line Turbidimeter. The Hach
2100A unit was used during the entire study period, but the two in-line units
were used only during the last 6 l/2 weeks of the study. The two types of turbidi-
meters were studied in order to learn whether 15° forward scatter or 90° side
scatter turbidimeters would be more effective in detecting fiber or turbidity
breakthrough. No significant difference was observed.
The Hach 2100 A utilizes a nephelometric principle of operation. Light is passed
up through the sample and light striking any suspended matter in the sample is
scattered at right angles to the light path. The scattered light is received by a
photomultiplier tube which converts the light energy into an electrical signal for
readout on a meter. Standardization of the instrument is accomplished with a set
of permanent turbidity standards which simulated Formazin solutions.
The Hach low range turbidimeter also utilizes a nephelometric principle of
operation but is a continuous flow unit requiring a sample flow rate of from % to
28
-------
Vi gpm. Light passes through the sample and is scattered at right angles and
received by two photoconductive cells which convert the light energy into an elec-
trical signal for readout on a meter. Standardization can be performed with a
standard reflectance rod equivalent to 5.0 turbidity units (TU), but during the
pilot filtration study, the unit was standardized using the Hach 2100A unit.
The Monitek Model 215/130 turbidimeter utilizes a combination of forward light
scattering and dual-beam ratio computation. The unit consists of two sub-systems,
the Model 215 transmitter and the Model 130 converter. The transmitter
measures the turbidity of the liquid flowing through it by projecting a thin ribbon
of light through the flowing stream. A direct beam detector measures the
intensity of the source light on the far side of the sight glass and the scattered
light from particles in the flow is measured by a scattered beam detector. The
scattered beam detector measures only light scattered in the 15° forward di-
rection. Signals from the transmitter are received by the Model 130 converter for
amplification and indication. Standardization of the unit was accomplished using
a Formazin solution.
ASBESTIFORM FIBER ANALYSES
Asbestos is a non-mineralogical term for asbestiform mineral material which can
be used commercially. Most of the world's asbestiform minerals are either chryso-
tile, a fibrous serpentine, or members of the amphibole group of silicate minerals.
Chrysotile is a hydrous magnesium silicate.2 Electron micrographs of chrysotile
indicate that the fibrils are in the form of a hollow tube and chrysotile fibers,
therefore, have an extremely large surface area. The chemical composition of
chrysotile is hydrous magnesium silicate with trace quantities of iron and calcium.
One of the most important properties of chrysotile is its positive surface electrical
charge, compared to the negative surface electrical charge of fibrils in the
amphibole group and almost all of the commonly used filter media such as sand,
diatomite, and cellulose. This may account, in part, for the difficulty in the
removal of chrysotile fibers as compared to amphibole fibers.
The amphibole group consists of crocidolite, anthophyllite, and tremolite.
Amphibole fibers contain primarily iron, magnesium, calcium, and sodium and are
solid and larger in diameter than chrysotile, but with a much smaller surface area
on a per mass basis. A physical comparison of amphibole and chrysotile fibers is
presented in Table 6. Several analytical techniques have been developed for
analysis of asbestiform fiber content in water. In order to provide the most
definitive quantitative and qualitative data on fiber removal in the pilot water
29
-------
treatment units, all available techniques were investigated prior to operation as to
their effectiveness. An attempt was made to balance the accuracy of each method
against the cost and time involved prior to making the final selection.
TABLE 6. COMPARATIVE PHYSICAL PROPERTIES OF AMPHIBOLE
CHRYSOTILE FIBERS IN LAKE SUPERIOR WATER
Diameter
Fiber ^m in x id15"
Amphibole 0.084-3.377 3.36-135.08
Chrysotile 0.054-0.168 1.92- 5.96
Transmission Electron Microscopy
The transmission electron microscope combined with electron diffraction has
proven quite effective in the analysis of asbestiform fibers in water samples.
Electron diffraction patterns from fibers of chrysotile have been measured with
diameters in the range of 0.02 Mm to 0.025 Mm and lengths in the range of 0.10
Mm to 0.15 Mm. With electron diffraction, a specific identification of the minerals
present can be made using wavelength dispersive X-ray spectrometers. Two general
methods of specimen preparation technique are utilized but in both cases the
fibers are specifically identified by electron diffraction patterns as they are being
counted. Results from this technique are reported in number of fibers per volume
of sample analyzed for both fiber groups.
The Ontario Research Foundation (ORF) in Sheridan Park, Mississaugu, Ontario,
Canada utilized the transmission electron microscopy process and reported results
as the number of f/1 for both the serpentine (chrysotile) and amphibole groups.
Results given as fiber counts and separated into amphibole and chrysotile counts
will refer to data from ORF. This laboratory has been employed by the EPA on a
contract basis for related western Lake Superior studies.
The ORF procedure involved collecting solid material from the water sample on a
0.1 Mm pore size membrane filter by filtering approximately 200 ml of sample.
The membrane filter was then ashed in a clean glass vial at a temperature of 70 °C
using a plasma microincineration technique. While no decomposition of the
mineral fibers occurs, organic materials in the water and the filter itself are
oxidized to carbon dioxide. The resultant residue was then redispersed ultra-
sonically in filtered distilled water and centrifuged onto a 1 cm diameter glass
30
-------
cover disc. The disc was dried and a carbon coating applied by evaporation that
was scored and floated off onto water, carrying the fibers with it. Pieces of this
material were then picked up on 200 mesh copper grids. A maximum of 10 grid
squares were searched for asbestos fibers at a magnification of about 25,000 using
a JEOL Model No. JEM 100U transmission electron microscope. The fibers were
identified by electron diffraction and measured for both length and width. The
fiber counts were processed by a computer program which calculated and plotted
the fiber number and mass concentrations. The process was sensitive to a
concentraton of 2 x 104 f/1 in water, below which the data were reported as below
detectable limits (BDL).3
Results of all ORF analyses are presented in Appendix E.
X-Ray Diffraction
The X-ray diffraction analysis process is also specific to the mineral type present
and will provide identification by fiber group. It will not distinguish, however,
between massive forms of the minerals and their fibrous counterparts. The
detection limit of X-ray diffraction is generally not as slow as that of transmission
electron microscopy. It does provide, however, a more rapid and inexpensive
determination of amphibole mass concentration in water which compares well
with average electron microscope fiber counts for the same samples.
The EPA National Water Quality Laboratory (NWQL) in Duluth utilized the X-ray
diffraction technique and reported the concentration of suspended solids and
amphibole mass in the water samples. Results from this technique were reported in
mg/1 of suspended solids and amphibole mass concentration. Results which refer
to amphibole mass will refer to data from NWQL. The NWQL has had consider-
able experience in the analysis of water for asbestiform fibers through sampling
provided for related studies involving many raw water and potable water intake
sources in the western Lake Superior area. The results they reported could be
readily compared with data from other sources they evaluated as well as provide a
determination of the effectiveness of the various removal processes tested in the
pilot plant operation. In addition, unavoidable sample storage time due to ship-
ment delays prior to analysis could be held to a minimum to preclude any
potential problems associated with long-term sample containment.
The NWQL procedure involved pressure filtering, at approximately 50 psi, of 25
to 40 liter samples through a preweighed Millipore 0.45 pm membrane filter. The
filter was dried in an oven at 70 ° C and the total suspended solids in mg/1 deter-
31
-------
mined by the difference. A weighing correction factor was applied to compensate
for a small filter weight loss due to leaching. The dry membrane filter with the
sample was then fastened to a glass slide using a thin film of lacquer and
examined with a Philips Model No. APD-3500 vertical X-ray diffractometer
system. Amphibole mass analysis was generally sensitive to 0.005 mg/1, with the
exact detection limit dependent upon the individual sample. When results were
reported as "less than", the lower detection limit of the technique had been
reached. Work by Cook'* indicates a standard deviation of ±3 per cent for
determining amphibole concentrations in typical samples with 0.1-0.3 mg/1
amphibole. For samples having lower amphibole concentrations (< 0.1 mg/1) and
suspended solids ( >1.0 mg/1), this precision is reduced to ±25 per cent. ^
Results of all NWQL analyses are presented in Appendix F.
Laboratory Selection
The two laboratories utilized for asbestiform fiber analyses, ORF and NWQL,
were selected by EPA for use in the pilot plant study as representative of a
number of qualified laboratories. One additional facility, the University of Min-
nesota at Duluth, was utilized for fiber analysis. Data from this facility are pre-
sented in Appendix G.
32
-------
SECTION VI
RESULTS
SEASONAL CONDITIONS
Raw water quality parameters such as pH, alkalinity, hardness, aluminum, and
iron are affected very little by ambient conditions. Parameters which are affected
by ambient and seasonal conditions include turbidity, water temperature,
suspended solids, amphibole concentration, and bacteriological counts. Changes in
climatologically related raw water quality parameters generally follow the climato-
logical events by 1 to 2 days. This relationship is evidenced by the relatively high
raw water turbidities of 5.6 TU and 6.3 TU which occurred on June 7, 1974 and
June 10, 1974, respectively. These high turbidities were each preceded by a day
on which rainfall in excess of 1 in, and wind velocities of 10 miles per hour
(mph) or higher were recorded. Other related data are presented in Appen-
dices A-3, A-4 and A-5.
Work by Cook^'5 shows that the suspended solids concentration of the raw water
may be increased following heavy rainfall or extended periods of high winds from
the east or northeast. If the suspended solids increase is caused by heavy rainfall,
generally no increase in amphibole concentration occurs. Rainfall does not cause
an increase in amphibole concentrations because the suspended solids increase is
the result of river runoff which contains very little amphibole mass. However, a
suspended solids concentration increase caused by high wind usually results in a
corresponding increase in amphibole concentration. Most of the amphibole mass
contained in the raw water comes from a taconite tailings discharge at Silver Bay,
Minnesota. The prevailing water circulation in western Lake Superior is counter-
clockwise, which results in the amphibole fibers contained in the tailings from
Silver Bay being transported to the Lakewood Intake. Winds from the northeast
and east promote this counterclockwise circulation, and during periods of
sustained east-northeasterly winds, water with increased amphibole concentration
is transported to the Lakewood Intake. High winds may also cause the resus-
pension of amphibole-rich sediment in the vicinity of the intake. Ice cover, which
begins in January and remains until late February or March, prevents wind-
generated resuspension of lake sediment.
According to Cook,^ amphibole concentrations in raw water from the Lake-
wood Intake decrease during the summer until fall overturn occurs. Peak
amphibole concentrations occur in spring and in late fall.
33
-------
This trend is also evident in the amphibole data obtained by the X-ray diffraction
technique during the pilot filtration study. Maximum raw water amphibole
concentrations occurred from April 19 to July 19, 1974 and ranged from 0.07 to
0.26 mg/1, with an average concentration of 0.15 mg/1. From July 19, 1974 to the
conclusion of the pilot filter study, the raw water amphibole concentration ranged
from 0.02 mg/1 to 0.10 mg/1, with an average concentration of 0.067 mg/1.
During the summer months, changes in raw water temperature are often wind-
related. Offshore winds cause upwelling which brings colder water to the intake
and easterly or northeasterly winds can push warm surface water into the
Lakewood Intake area, causing higher water temperatures. This phenomenon is
evidenced by the dramatic increase in raw water temperature which occurred on
July 9, 1974. The water temperature increased from 40 °F to55°F during a period
of sustained high winds from the east.
GRANULAR MEDIA FILTRATION
Asbestiform Fiber Removal
For purposes of comparison and evaluation, experimental filter runs conducted on
the pilot units have been placed in categories that employed similar treatment
configurations. The design and operational differences of the several pilot units
have been described previously as have the physical modifications of the units,
made to vary the treatment parameters. A summary of the results from each
category of operation for the granular media filters is presented in Tables 7 and
8. The tables present results for asbestiform fiber removal summarized for each
category in terms of a selected fiber removal goal. The successful goal chosen for
either amphibole or chrysotile fiber removal is below detectable limits (BDL), or
40,000 fibers per liter, and for amphibole mass removal less than 0.005 mg/1.
Details of each individual granular media run are provided in Appendices C-l and
C-2.
Turbidity Removal
It is difficult to made definitive statements concerning the effect of raw water
turbidity on the effluent turbidity from the pilot filter units, because of the
narrow range of raw water turbidities which occurred during the pilot filtration
study. In the range of raw water turbidities which occurred, the effluent turbidity
was affected more by type and amount of chemicals added than by the turbidity
of the raw water, particularly with the granular media filters.
34
-------
TABLE 7.
GRANULAR MEDIA FILTRATION (MM-1) SUMMARY OF ASBESTIFORM FIBER REMOVAL BY CATEGORY
Categories
Alum & Anionic Polymer
Filtration w/o
MM-1-
Dual
Media
Sedimentation
4 gpm/sq.ft.
(Runs 1-40)
- (Separan NP-10 or 847A)
(Runs 1,2,20-24)
Alum & Nonionic Polymer
- (N-17 or 98SN)
(Runs 4-19, 25-40)
Alum & Nonionic
Sedimentation
Tube Settlers
4gpm/sq.ft.
(Runs 41 -87)
i Polymer (98SN)
(Runs 41-44, 64-86)
Alum & Nonionic Polymer
(985N) and Coagulant
Aid (Bentonite)
(Run 87)
FeCl3 & Nonionic
1 Polymer (985N)
(Runs 45-63)
NWQL Amphibole
Mass Results
Samples
Analyzed
with Detection Samples with
Limit <0.005 mg/1 filtrate £0.005 mg/1
0
3
12
1
2
0
2
12
1
1
ORF Amphibole
Fiber Count Results
Samples with
Samples <0.04x!06
Analyzed f/1
0
7
13
1
2
0
6
13
0
2
ORF Chrysotile
Fiber Count Results
Samples with
Samples
-------
TABLE 8. GRANULAR MEDIA FILTRATION (MM-2) SUMMARY OF ASBESTIFORM FIBER REMOVAL BY CATEGORY
. Categories
Alum & Nonionic
Polymer (985N)
Cloquet
Dual
»*- ji_
(Runs 1-76)
(Runs 75,76)
2 gpm/sq.ft.
(Run 76)
Alum & Nonionic
Polymer(985N)
4 gpm/sq.ft.
(Run 75)
Alum & Nonionic
Polymer (985N)
Lake wood
(Runs 1-74)
2 gpm/sq.ft.
(Runs 54-56)
FeCls & Cationic
Polymer (C-31)
4 gpm/sq.ft.
(Runs 1-13,16)
FeC13 & Nonionic
_ Polymer (N-17)
4 gpm/sq.ft.
(Runs 19-23)
Alum & Nonionic
Polymer
(N-17 & 985N)
4 gpm/sq.ft.
(Runs 24-53)
Alum & Nonionic
Polymer (985N)
6-8 gpm/sq.ft.
(Runs 57-74)
NWQL Amphibole
Mass Results
Samples
Analyzed
with Detection Samples with
Limit £0.005 me/1 filtrate £0.00 5 mg/1
1
0
2
0
0
2
0
1
0
2
0
0
2
0
ORF Amphibole
Fiber Count Results
Samples with
Samples * 0.04x106
Analyzed f/1
1
0
2
2
0
3
3
1
0
1
2
0
3
2
ORF Chrysolite
Fiber Count Results
Samples with
Samples *0.04xlo6
Analyzed f/1
1
0
2
2
0
3
3
o
0
1
0
0
0
-------
TABLE 8.
(CONTINUED)
Categories
Chem. Add. Alum & Nonionic Polymer
/~*ln<-...A* +~ 4...~. 171 t, Jt «._« /~n f* SnocKT\
Tri.
Media
(Runs
77140
uri
^4
*^»w*fi*w» i\j iwu i loan -T gfjui/ 314.1 1. \_7oji^y
(Run 106) Mixers (Run 106)
Chem. Add. Alum & Nonionic Polymer
Lakewood
(Runs 77-105,
107-137)
-to Mixing 4 gpm/sq.ft. (985N)
Chamber (Runs 77-81, 84-92)
Alum & Nonionic Polymer
4 gpm/sq.ft. (985N)
Chem. Add.
-to two Flash
Mixers
(Runs 93-105,
107-113,115,140a)
Alum & Nonionic Polymer
2 gpm/sq.ft. (98 5N)
(Run 114)
Alum&Anionic&Cationic Poly.
(A-23&Catfloc B or C-31)
4 gpm/sq.ft. (Run 116,118,139)
Alum & Cationic Polymer
4 gpm/sq.ft. (Catfloc B)
(Run 117)
Alum & Anionie & Nonionic
Polymers (A-23 & 985N)
4 gpm/sq.ft.
(Runs 119,120,138)
Alum & Nonionic Polymer
Mixers
1 4 gpm/sq.ft. (98 5N)
(Runs 121,122,123)
Alum & Nonionic & Anionie
Polymers (985N & A-23)
4 gpm/sq.ft. (Runs 124,127,128)
Alum & Anionie & Cationic
Pr»l vtn»»re ( A 0 "\ XT C* "31^
r oiyrners \z\ i j QE> v^ o i )
4 gpm/sq.ft. (Runs 125,126)
Alum & Nonionic Polymer
6 gpm/sq.ft. (985N)
(Runs 129-133)
Alum & Cationic Polymer
1 4 gpm/sq.ft. (C-31)
(Runs 134-137)
NWQL Amphibole
Mass Results
Samples
Analyzed
with Detection Samples with
Limit SO. 00 5 mg/1 filtrate iO.OO 5 mg/1
1
1
5
1
2
0
2
1
2
1
2
1
1
1
5
1
2
0
2
1
2
1
2
1
ORF Amphibole
Fiber Count Results
Samples with
Samples <0.04xl06
Analyzed f/1
1
1
5
1
2
0
2
1
2
1
2
1
1
1
5
1
2
0
2
1
2
1
2
1
ORF Chrysotile
Fiber Count Results
Samples with
Samples S0.04xl06
Analyzed f/1
1
1
5
1
2
0
2
1
2
1
2
1
1
0
2
0
0
0
0
1
0
1
0
0
Sedimentation Tube Settlers (Run 140)
-------
A plot of raw water turbidity and effluent turbidity versus time for Run 31 on
MM-1 is presented on Figure 6. Although the raw water turbidity increased from
0.8 TU to 4.6 TU during the run, the effluent turbidity changed only slightly,
increasing from about 0.07 TU to 0.16 TU. No changes in chemical feeds or flow
rate were made at the time of the turbidity increase. A similar plot of the
subsequent run on the MM-1 unit on Figure 7 indicates that although the raw
water turbidity varied from 1.4 to 5.6 TU, the effluent turbidity only ranged
from 0.07 to 0.27 TU and the low and high effluent turbidities did not follow the
low and high raw water turbidities. The chemical dosages on this run were slightly
lower than those of Run 31 and turbidity breakthrough was beginning to occur at
the conclusion of Run 32.
The effect of raw water turbidity on filter run lengths of the granular filters was
twofold. First was the direct effect: the higher solids loading associated with a
higher turbidity resulted in a more rapid rate of head loss buildup. Second was
the indirect effect: higher turbidities required higher chemical dosages to maintain
satisfactory effluent turbidity and these higher chemical dosages resulted in a
more rapid rate of head loss buildup. For the range of raw water turbidities
experienced during the pilot filtration study, the effect of the increased chemical
dosages was much greater than the effect of the increased solids loading associated
with a higher turbidity.
Operational Variations
Where possible, analysis of the effects of the various treatment configurations
included all of one or more of the major categories listed in Tables 7 and 8. How-
ever, in many instances, a further breakdown of a major category into subsets or
series of runs within that category was necessary. In this manner, a base line of
treatment could be established in which all treatment variables would be held
constant except for those being compared.
Runs conducted on the MM-2 unit during the period of high turbidity were made
at flow rates of 116 m3/m2 day (2 gpm/ft2) and 350 m3/m2 day (6 gpm/ft2) and
results were similar to those obtained on the MM-1 unit. Seventy-three runs with
sufficient chemical dosages had satisfactory effluent turbidities regardless of raw
water turbidity (within the range of raw water turbidities experienced during the
pilot filtration study), and those without sufficient chemical addition had high
effluent turbidities even when the raw water turbidity was less than 1.0 TU.
Table 9 shows the number of runs made on granular filters at filter rates of
4 gpm/ft2 or higher. All of the MM-1 runs were made at 4 gpm/ft2 and this unit
38
-------
6.0
5.0
4.0
3.0
2.0
p: os
0.8
t °7
5 0.6
CD
(T 0.5
0.4
0.3
0.2
O.I
.09
.08
.07
.06
I
rv
*
RAW WATER
EFFLUENT
1 1III 1 1II 1
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34
HOURS
FIGURE 6 . RAW WATER AND EFFLUENT TURBIDITY
CURVES. UNIT MM-I RUN 31 .
39
-------
6.0
5.0
4.0
3.0
2.0
=> i.o
I- 0.9
>_" 0.8
t 0.7
5 0.6
CD
* 0.5
0.4
0.3
0.2
O.I
.09
.08
.07
.06
I 1 I
WATER
- -* EFFLUENT
I
V
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34
HOURS
FIGURE 7 . RAW WATER AND EFFLUENT TURBIDITY
CURVES. UNIT MM-I RUN 32
40
-------
~\
was operated exclusively with dual media. Thirty-six runs at 5 gpm/ftz and higher
rates were made on the MM-2 filter unit. Of these, 23 were made with dual media
and 13 with tri-media.
TABLE 9. NUMBER OF RUNS MADE AT OR NEAR INDICATED
FILTER RATES
Pilot Unit
MM-1
MM-2
4 gpm/ft2
73
85
5 gpm/ft2
0
14
6 gpm/ft2
0
8
7 gpm/ft2
0
14
Chemical Additions
Either alum or FeClg was used as a coagulant in all of the granular media unit
filter runs. A comparison of their relative effectiveness in the MM-1 unit can be
made by examining the data from two major categories in which a nonionic
polymer, 985N, was utilized. The data in App.G4show that, overall, the average
raw and filtered turbidities appear to be quite similar for Runs 41-44 and 64-86
in which alum was used and Runs 45-63 in which FeCl^ was used. The length of
those filter runs in which alum was used was longer, however; averaging 23 hours
as compared to 13 hours in those runs using
A more direct comparison of the two coagulants may be made between Runs
41-44, utilizing alum, and Runs 45-49, utilizing FeC^. In these runs, the polymer
was introduced to the floe chamber of the MM-1 unit in both cases,- essentially
equating the treatment processes except for the difference in coagulant. The
data, from Appendix C-l, show that a slight but significant increase in turbidity
removal was achieved with alum as a coagulant when the average raw and effluent
turbidities of the alum runs, 0.97 TUand 0.09 TU, are compared to the average
raw and effluent turbidities of the FeCl3 runs, 0.68 TU and 0.13TU. In these
runs, the alum dosage averaged 16.5mg/l and the FeCl-^ dosage averaged
12.1 mg/1. Filter run lengths again were longer with the alum runs, averaging 22.3
hours compared to 16.8 hours with the FeQ3. Sludge solids produced and back-
wash volume as a per cent of the treated water were both higher in the FeCl-^
runs, averaging 0.051 lb/1000 gal and 2.42 per cent, as compared to 0.018
lb/1000 gal and 1.88 per cent in the alum runs.
A similar comparison can be made between alum and FeClo in the MM-2 unit by
examining Runs 24-53 in which alum was used and Runs 19-23 in which
41
-------
was used. In both categories, the coagulants were used with a nonionic polymer,
N-17 or 985N. As shown in App.O2 the overall results in the two categories
demonstrated a slightly better turbidity removal with alum as the coagulant by
comparing average effluent turbidities of 0.20 TU with alum to 0.36 TU with
FeCl^. Longer filter runs were also achieved with alum as the coagulant in these
categories. The data from Appendix C-2 show an average length of 14.6 hours as
compared to 10.6 hours with
Comparing directly Runs 20, 22, and 23, in which FeCl3 was used, with Runs
24-27 and 34-41, in which alum was used, the data from Appendix C-2 again
show that slightly better turbidity removals were achieved in the alum runs. In
both sets of runs, the coagulant was introduced to the mixing chamber of the
MM-2 unit and the polymer, N-17 in both series, was introduced to the first
flocculation chamber of the unit. The average raw and effluent turbidities of the
FeC^ runs were 0.68 TU and 0.35 TU while the average raw and effluent
turbidities of the alum runs were 1.09 TU and 0.20 TU. The lengths of filter runs
also were better with those runs using alum, averaging 15.1 hours as compared to
1 2. 1 hours in the FeCl^ runs. In these runs, the FeCl^ dosage averaged 1 1 .4 mg/1
and the alum averaged 18.6 rng/1 while the dosage of N-17 averaged approximately
0.25 mg/1 in both sets of runs.
A total of four different coagulant aids or polymers were utilized in the MM-1
unit to improve the results of the use of the coagulants. The polymers were
intended primarily to extend filter runs and prevent breakthrough in the filter by
strengthening the chemical floe. Two anionic polymers, Separan NP-10 and 847 A,
and two nonionic polymers, N-17 and 985N, were used.
Mixing accomplished by the external flash mixers and the in-line mixers was more
effective than the mixing provided by the pilot unit mixing system as evidenced
by better turbidity removals and longer filter runs. That the external flash mixers
and in-line mixers consisted of more than one stage is the probable reason for
increased effectiveness. The staged systems greatly reduced the possibility of short
circuiting during the mixing process and, therefore, improved the overall pilot
plant operation.
As indicated in Table 10 certain filter runs with the in-line mixers produced a
lower effluent turbidity than the two-stage external mixers, but at a higher cost
and with a much greater possibility of turbidity breakthrough. In addition, the
effluent turbidity of the runs conducted with the two-stage mixers could probably
be lowered by a slight increase in alum dosage.
42
-------
Dual and Tri-Media Filter Beds
The filter bed in the MM-2 unit was utilized in both a dual and a tri-media con-
figuration to investigate the differences that the two filter beds would have on the
raw water at the Duluth Lakewood Pumping Station. The results obtained using
dual and tri-media filter beds are presented in Table 10. All runs selected for use
in calculating the values presented in the table met the following conditions:
1. Filtration without sedimentation
2. Filter rate of 214-277 m3/m2 day (3.4-4.4 gpm/ft2)
3. Filtered water temperature of 37° F to 44° F
4. Raw water turbidity of 0.46 to 0.95 TU
5. Run length is equal to time required to reach 8 ft of water head loss or 0.20
TU in effluent
6. Alum and 985N used for chemical treatment
TABLE 10. RESULTS OBTAINED USING DUAL AND TRI-MEDIA
FILTER BEDS
Media
Dual
Tri
Tri
Tri
Mixing No. of Avg. length
mode runs of runs,
hr
Alum @ rapid mix 5 19.4
Poly @ flocculator
Alum @ rapid mix 3 11.1
Poly @ flocculator
Two-stage 7 1 8.9
flash mix
In-line mixers 3 20.3
Avg. effl. Avg. chem.
turb., costs
TU S/1000 gal
0.12 0.0051
0.10 0.0044
0.10 0.0039
0.06 0.0042
Based on data presented in Table 10, the tri-media filter bed produced lower
effluent turbidities at lower chemical costs than did the dual-media filter bed. The
lowest chemical cost was obtained using two-stage flash mixing prior to the tri-
media filter bed and the lowest effluent turbidity was obtained using the Kenics
in-line mixers prior to filtration. However, problems with turbidity breakthrough
were encountered when using the in-line mixers which did not occur when using
two-stage flash mixing. An increase in alum dosage probably would have lowered
the effluent turbidity during those runs with two-stage flash mixing.
43
-------
Filter runs conducted with Duluth raw water in the dual media bed configuration
in MM-2 included three categories of filter rates. Runs 54-56 were conducted at
116m3/m2 day (2 gpm/ft2), Runs 42-53 at 232 m3/m2 day (4 gpm/ft2) and
Runs 57-74 at 348-464 m3/m2 day (6-8 gpm/ft2). For comparison purposes,
those runs utilizing alum and a nonionic polymer, 985N, introduced at the same
point in the treatment process were analyzed. The data in Table 8, Appendix C-l,
and Appendix C-2 again show that little difference resulted in the finished water
quality, with average effluent qualities of 0.14 TU, 0.18 TU, and 0.18 TU,
respectively. The rate of head loss buildup did vary proportionately, however,
with average filter run lengths of 41.8 hours, 18.1 hours and 10.3 hours,
respectively.
Two comparisions of filtration rates can be made in the tri-media filter bed
configuration in the MM-2 unit involving alum and a nonionic polymer, 985N.
The first, utilizing chemical addition to the two external flash mixers as shown in
Table 8 shows a slightly better turbidity removal at 116 m3/m2 day (2 gpm/ft2),
^ 9 "*)
as compared to 232 m^/m^ day (4 gpm/ftz). However, since only one run was
^
involved at the 2 gpm/ftz rate, no definitive conclusions can be made.
The second comparison involves chemical addition through in-line mixers at fil-
tration rates of 232 m3/m2 day (4 gpm/ft2) and 348 m3/m2 day (6 gpm/ft2). As
shown in Table 8, turbidity removals were approximately the same, with average
effluent turbidities of 0.06 TU and 0.10 TU, respectively. As the filter runs at
O O f\
232 nrVmz day (4 gpm/ftz) were not carried to terminal conditions, no definitive
comparisons of filter run lengths can be made in these two categories.
Head Loss, Filter Run Length, and Terminal Turbidity
The rate of head loss, the length of the filter run, and the terminal turbidity
were factors used to evaluate the performance of the various filter configurations
utilized in this study. Ideally, terminal head loss at about 8 ft of water and the
beginning of turbidity breakthrough would occur at the same time with a
resultant optimum filter run length.
Figure 8 shows the head loss and effluent turbidity curves of Run 107 of the
MM-2 unit. These curves show that turbidity breakthrough began to occur at
about the 28th hour of the filter run, a point where the head loss was about
4.5 ft of water, considerably below a terminal head loss condition.
Figure 9, the head loss and effluent turbidity curves of Run 109 of the MM-2
unit, shows terminal head loss and turbidity breakthrough occurring at about the
44
-------
HEAD LOSS IN FEET OF WATER
O
c
:o
m
09
3) CD I
c m
z 5 >
H O
0 2
c w
3 0)
<
m >
o> z
o
m
ro
o>
ro
c
x
i
p
01
ro
b
ro
bi
OJ
6
t
I
I
ro
O
ro o»
TURBIDITY,TU
ro
ro
ro
V
i
i *
5 o
pi
b
a
N
P
g
-------
10.0
9.0
8.0
CC
£7.0
u_
O 6.0
I-
UJ
UJ
^ 5.0
V)
V) 4.0
O
< 3.0
UJ
X
2.0
1.0
Average Raw Turbidity = 0.62 TU
HEAD LOSS
TURBIDITY
-cr
.36
.32
.28
.24
.20 «
>-
H
O
cr
h-
.12
.08
.04
I
I
I
10 14 18 22
HOURS
26 30 34 38
FIGURE 9 .HEAD LOSS AND EFFLUENT TURBIDITY
CURVES. UNIT MM-2 RUN 109.
46
-------
same time. This particular run was conducted for 4 hours before the run was
terminated, with an average raw turbidity of 0.62 TU and an average effluent
turbidity of 0.10 TU.
The terminal head loss, in feet of water, terminal turbidity, in TU, and the length
of filter run, in hours, are presented in Appendices C-l and C-2 for each filter run
of the granular media units. These data reflect the progressively increasing filter
run lengths that occurred with no sacrifice in effluent quality. This was due to
the optimum selection of coagulants and polymers and their dosage rates that
occurred as the study progressed and operating experience was gained. In
addition, utilization of certain unit processes that provided better chemical mixing
and flocculation, as well as better filtration, also produced a marked increase in
filter run lengths as well as increased turbidity and asbestiform fiber removal. For
the most part, the type and method of coagulant addition or the rate of filtration
had the most effect on the rate of head loss buildup and filter run lengths.
Individual head loss and turbidity curves for each granular media run are provided
in Appendices D-l and D-2.
Backwash Solids
The relationship of backwash solids to sludge volume is presented in Table 11.
The values presented are based on SS tests and settleable solids tests performed
on backwash water samples collected during the pilot filtration study. Sludge
production given in the table is in terms of sludge produced per 1000 gallons of
water treated.
TABLE 11. RELATIONSHIP OF BACKWASH SOLIDS TO SLUDGE
VOLUME - UNIT MM-2
Run
132
133
134
137
138
139
Backwash
solids,
lb/1000 gala
0.060
0.061
0.046
0.055
0.038
0.043
SS
mg/1
166
L06
227
1-24
208
132
Cone, after
60 min,
%
0.72
0.62
0.38
0.44
1.09
1.47
Gal of
sludge
per 1000 gal
1.00
1.17
1.46
1.49
0.42
0.35
'Based on theoretical solids production
47
-------
Backwash Water
In order to determine the asbestiform fiber concentration of the backwash water,
a sample of the backwash water from the MM-2 unit, Run 138, was allowed to
settle for approximately 10 days. Supernatant from the sample was withdrawn
and submitted for analysis. The results indicate a low concentration of asbesti-
form fibers in the supernatant with 0.0152 x 10° f/1 of amphibole and 0.57 x
10" f/l amphibole and 1.3 x 10" f/1 chrysotile as reported by this same
laboratory. This would indicate that supernatant from settled backwash water
could be recycled for treatment after a sufficient period allowed for settling.
Filtration With and Without Sedimentation
The MM-1 unit was operated in two configurations, filtration without sedimenta-
tion in Runs 1-40 and filtration preceded by sedimentation in tube settlers in
Runs 41-87. The purpose of this dual operation was to investigate the effects of-
sedimentation on effluent turbidity, filter breakthrough, and filter run lengths.
A comparison of the relative effects of the two configurations may be made with
the results of two series of filter runs: Runs 2540 which were conducted without
sedimentation, and Runs 77-86 which were conducted with sedimentation prior to
filtration. Runs 77-86 were selected for comparison as a measurement of the
solids retained in the sedimentation basin which was made independently of the
solids backwashed from the filter bed. In both series of runs, alum was used as
the coagulant at approximately 25 mg/1 and 985N as the polymer at approx-
imately 0.05 mg/1. Turbidity removal was slightly better with sedimentation than
without sedimentation. Average raw and effluent turbidities for Runs 25-40 were
1.65 TU and 0.13TU while the average raw and effluent turbidities for Runs
77-86 were 0.79 TU and 0.09 TU.
In addition, the data from Appendix C-l indicate that the average run lengths
without sedimentation were considerably shorter than those with sedimentation,
17.8 hours as compared to 25.1 hours. An examination of the backwash solids for
those runs with sedimentation revealed that the solids loading on the filter was
reduced by approximately 27 per cent by the sedimentation step. It would be
expected that the filter run length would be increased 37 per cent by the re-
duction in solids loading. In actuality, the run length increased 41 per cent.
48
-------
This observation is reinforced by examining the data in Appendix C-2 for Run
115 and Run 140 on the MM-2 unit. Both of these runs were conducted with
alum as the coagulant at approximately 14.5 mg/1 and 985N as the polymer at
approximately 0.045 mg/1. Run 115 was conducted without sedimentation and
Run 140 was conducted with sedimentation prior to filtration. Again, the average
raw and effluent turbidities were about the same, 0.77 TU and 0.09 TU for Run
115, and 0.47 TU and 0.06 TU for Run 140. The run length, however, was longer
with Run 140 which included sedimentation, 33.6 hours as compared to 24.8
hours with Run 115. Analysis of backwash solids samples from Run 140 showed
that the majority of the total solids removed were removed from the filter bed,
70 per cent, as compared to 30 per cent from the sedimentation basin.
DIATOMACEOUS EARTH FILTRATION
Asbestiform Fiber Removal
Two major types of DE media units were used in the pilot plant study. The
design and operational differences of the two units have been described pre-
viously. An examination of the results of the various processes utilized is
presented in this section and is summarized below.
The DE unit designated as BIF was used exclusively to provide data on vacuum
filtration. One other vacuum DE unit was available, but was held in reserve. Data
from vacuum DE filtration runs will be referred to as BIF data. Two DE units,
designated as ERD-1 and ERD-2, were used exclusively to provide data on
pressure filtration. Since the physical configuration of these two units was identi-
cal, data gathered from filter runs on these two pressure units have been com-
bined and will be referred to as ERD data.
The data listed in Tables 12 and 13 summarize the results of the BIF and ERD
runs and are used to compare sets of runs with similar treatment configurations.
Details of each individual DE run are provided in Appendices C-3 and C-4.
Filter runs with the BIF unit and especially with the ERD units were quite long
when conducted to their respective maximum head loss differentials. Under good
quality effluent production conditions (effluent turbidity 0.2 TU or less), filter
run lengths of 30 to 40 hours could be achieved easily with the BIF unit. Under
similar conditions, the ERD unit could easily produce filter runs of over 60 to 70
hours duration. Once a maximum filter run length for a particular test condition
had been established, filter runs were limited to a maximum of 24 hours to
49
-------
en
O
TABLE 12. VACUUM DIATOMACEOUS EARTH FILTRATION SUMMARY OF ASBESTIFORM FIBER REMOVAL BY CATEGORY
Categories
One
Step
Precoat
BIF-
Igpm/sqft
(Runs
1T-14T
-Precoat & Body Feed only (Runs 1T-5T,6,7)
-Anionic Polymer (A-23) to Precoat (Runs 8T-14T.8)
-Nonionic Polymer (985N) to Precoat (Runs 1-5)
-Cationic Polymer (C-31 or S73C) to Precoat (Runs 9-17)
1-17)
Medium
Coarse
Two
Step
L Precoat
Igpm/sqft
(Runs
18-122)
Step of
Precoat
p Cationic Polymer (S73C) to
Total Precoat (Runs 18-24,26,29-35)
Anionic Polymer (A 23) to
Total Precoat (Runs 36-41)
-Alum to 2nd Step of
Precoat (Runs 42,46,47,49-51)
Alum & Soda Ash to 2nd
Step of Precoat (Runs 60,61)
- Alum & Soda Ash to 2nd Step
of Precoat and to Body Feed
(Runs 62)
Anionic Polymer (A-23) to 2nd
Step of Precoat, Alum & Soda
Ash to Body Feed (Runs 63-66,69-74)
Very
Coarse
LDE in 1st
Step of
Precoat
p Cationic Polymer (573C) to
Total Precoat (Runs 25,28)
Alum to 2nd Step of
Precoat (Runs 52-57)
- Alum & Soda Ash to 2nd Step
of Precoat (Runs 58,59,89-103)
-Anionic Polymer (A 23) to 2nd Step
of Precoat, Alum & Soda Ash to
Body Feed (Runs 75-78,111, 112)
Anionic Polymer (A 23) to 2nd
Step of Precoat (Runs 79, 80)
- Cationic Polymer (Catfloc B) to
Raw Water (Runs 82-84, 86-88)
- Alum & Soda Ash to 2nd Step of
Precoat and to Body Feed
(Runs 104-109,120-122)
Alum & Soda Ash to 2nd Step of
Precoat, Cationic Polymer (Catfloc B)
to Raw Water (Runs 110,113-119)
Mass Results
Samples
Analyzed Samples with
with Detection SO. 00 5 mg/1
Limit sn.nnfi mg/l Amnh. mass
0
2
0
0
3
0
1
0
0
3
0
0
2
2
1
1
2
4
0
1
0
0
2
0
1
0
0
2
0
o
2
2
0
1
2
4
ORF Amphibole Fiber
Count Results
Samples with
Samples <0.04xl06
Analyzed f/1
0
3
0
' 1
4
1
2
0
0
3
1
o
2
2
1
3
3
4
0
2
0
0
1
1
o
0
o
3
0
o
o
2
1
3
2
4
ORF Chrysotile Fiber
Count Results
Samples with
Samples <0.04xlfl6
Analyzed f/1
0
3
0
1
4
0
2
0
o
3
1
0
2
2
1
3
3
4
0
1
o
0
o
1
0
o
1
0
0
0
0
1
o
o
o
-------
TABLE 13. PRESSURE DIATOMACEOUS EARTH FILTRATION SUMMARY OF ASBESTIFORM FIBER REMOVAL BY CATEGORY
Categories
One
ERD 1&2
Step
Precoat
Igpm/sq.ft.
(Runs 1A-6A,
1-27)
Two
Step
Precoat
Igpm/sq.ft.
(Runs 2 8-86)
Precoat & Body Feed only
(Runs 2,7-11)
Cationic Polymer to Precoat
(C 31 or 573C)
(Runs 1A-6A, 12-15,20)
Alum to Precoat
(Runs 21-27)
Anionic Polymer to Precoat
1 (A-23) (Run 3)
Anionic I'olymer (A-23) to 2nd Step
of Precoat, Alum & Soda Ash to
Body Feed (Run 72)
Alum to 2nd Step of Precoat (Runs 28-37)
Alum & Soda Ash 10 2nd Step of Precoat
(Runs 38-44, 54-60, 62-66)
Cationic Polymer to Raw Water
(Catfloc B) (Runs 45-51;
Alum & Soda Ash to 2nd Step of Precoat
and to Body Feed
(Runs 67-71,82,85,86)
Alum & Soda Ash to 2nd Step of Precoat,
Cationic Polymer (Catfloc B) to Raw Water
(Runs 73-76, 78-81)
NWQL Amphibole Mass
Results
Samples
Analyzed Samples with
with Detection < 0.005 mg/1
Limit SO. 005 mg/1 Amphibole mass
1
5
0
0
1
1
6
5
3
3
1
5
0
0
1
1
6
5
3
3
ORF Amphibole Fiber
Count Results
Samples with
Samples = 0.04x106
Analyzed f/1
1
6
1
0
1
2
6
5
3
3
0
3
1
0
1
0
3
5
3
3
ORF Chrysotile Fiber
Count Results
Samples with
Samples S0.04xlfl6
Analyzed f/1
1
6
1
0
1
2
6
5
3
3
0
1
0
0
0
0
1
2
0
0
-------
permit variations of treatment within the test condition. Because of this,
comparisons of filter run lengths cannot always be made and quality of effluent
and asbestiform fiber removal must be relied upon for judgment criteria.
Turbidity Removal
High raw water turbidities occurred from June 6 to June 15. No other extended
periods of raw water turbidities over 1.5 TU were experienced during the pilot
filtration study and fluctuations in raw water turbidities in the range of 0.8 to
1.5 TU did not appear to affect effluent turbidities. A plot of raw water and
effluent turbidities versus time for Run 47 on the ERD-2 unit is given on
Figure 10. Raw water turbidities during this run varied from 0.84 to 1.3 TU and
effluent turbidities varied from 0.08 to 0.18TU. However, fluctuations in effluent
turbidity did not follow the increases and decreases of the raw water turbidity.
As with the pressure DE unit, all runs conducted on the BIF unit during the
period of high turbidities, which began on June 6, 1974, had very high effluent
turbidities. A plot of raw water and effluent turbidities for Run 25 on the BIF
unit is given on Figure 11.
A review of the data for BIF in Appendix C-3 shows that, on runs where the
average raw water turbidity was greater than 1.5TU, the lowest average effluent
turbidity attained was 0.3 TU. This inability to attain low effluent turbidities
would seem to be the result of high raw water turbidity.
Two head loss and effluent turbidity curves are presented on Figures 12 and 13.
Figure 12 representing Run 110 of the BIF unit, demonstrates the steady, almost
straightline head loss buildup and rather constant effluent turbidity characterized
by vacuum DE units. Figure 13, representing Run 79 of the ERD-2 unit, demon-
strates a typical head loss curve with a gradually diminishing rate of head loss
buildup. This type of head loss curve is due to the cylindrical shape of the filter
septa in which the filter surface increases as the filter cake increases, thus negating
somewhat the head loss buildup in the filter cake.
Figures 14 and 15 are included to demonstrate the effect of the two-step precoat
process on the initial effluent turbidity. On Figure 14, a single-step precoat pro-
cedure was used with a subsequently high initial effluent turbidity. Approximately
4 hours were required to attain the necessary filter permeability to produce a
steady effluent turbidity. On Figure 15, however, a two step precoat application
produced an almost immediate steady state level in the effluent turbidity. With
52
-------
6.0
5.0
4.0
3.0
2.0
1.0
0.9
0.8
0.7
0.6
g05
CD
tr. 0.4
0.3
0.2
O.I
.09
.08
.0 7
.06
RAW WATER
- EFFLUENT
v\
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34
HOURS
FIGURE 10 . RAW WATER AND EFFLUENT TURBIDITY
CURVES. UNIT ERD-2 RUN 47.
53
-------
6.0
5.0
4.0
3.0
2.0
(- 9.0
>T8.0
±7.0
m
5.0
4:0
3.0
2.0
0. I
.09
.08
.07
.06
RAW WATER
EFFLUENT
i i i i
i i i
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34
HOURS
FIGURE II . RAW WATER AND EFFLUENT TURBIDITY
CURVES. UNIT BIF RUN 25.
54
-------
HEAD LOSS IN FEET OF WATER
IV)
Ol
0>
C
3J
m
3J H I
c c rn
z 3) >
CD °
Ol
O |-
o >
c z
m m
C/) T|
-n
||
CD
(D
x
O
CO
o
no
r
i
I 2
5 o
8
"O
I
M
O
O
0)
O
O>
fO
j
ro
TURBIDITY TU
-------
30
25
UI2O
i
a
u.
Av»rog* Raw Turbidity = 0.42 TU
O HEAD LOSS
^ TURBIDITY
V)
V)
O
10
UJ
X
-AY7
.08
.06
- .04 Q
- .02
m
IE
I
I
I
10 14 IB 22
HOURS
26
30
34
FIGURE 13 . HEAD LOSS AND EFFLUENT
TURBIDITY CURVES. UNIT
ERD-2 RUN 79.
56
-------
60
Average Raw Turbidity - 1.16 TU
55
50
45
40
tr
LJ
h-
UJ
LU
30
25
CO
(O 20
O
Q
< 15
LU
X
10
HEAD LOSS
TURBIDITY
-» .36
.32
.28
.24
.20
Q
CD
.16
.12
.08
.04
10
14
18
HOURS
22
26
30
34
FIGURE 14 . HEAD LOSS AND EFFLUENT TUR-
BIDITY CURVES. UNIT ERD-2
RUN 21.
57
-------
64
60
cr
LJ
< 56
Average Raw Turbidity = 0.66 TU
HEAD LOSS
' TURBIDITY
62
Ld
LU
48
CO
CO
3
Q
X
44
I
10
14 18
HOURS
22
26
30
.16
.12 13
.08 H
Q
CD
.04 ?5
34
FIGURE 15 . HEAD LOSS AND EFFLUENT TUR-
BIDITY CURVES. UNIT ERD-2
RUN 34.
58
-------
properly designed filter septa, two-step precoating might not be necessary for
production of low initial effluent turbidities.
Operational Variations
The data listed in Tables 12 and 13 summarize the results of the BIF and ERD
runs and are used to compare sets of runs with similar treatment configurations.
Filter runs with the DE units were quite long when conducted to their respective
maximum head loss differentials. Under good quality effluent production
conditions (effluent turbidity 0.1 TU or less), filter run lengths of 30 to 40 hours
with the BIF unit and 60 to 70 hours with the ERD unit could be achieved
easily. Therefore, filter runs for the DE units were limited to a maximum of 24
hours to permit variations of treatment configurations.
Individual head loss and turbidity curves for each DE run are provided in
Appendices D-3 and D-4.
Precoat Application
Several series of runs were conducted in both the BIF and ERD units utilizing a
single level of DE for the precoat. Substantial operational and effluent turbidity
problems were encountered with this procedure.
The data from Tables 12 and 13 show the relatively high effluent turbidity levels
that resulted with the single precoat application. In Runs 1T-14T and 1-17 in the
BIF unit, average effluent turbidity results when using a single precoat, averaged
0.47, 0.18, 0.26, and 0.35 TU. In addition, Appendix C-3 shows that several
additional attempts at filter runs in the BIF unit were aborted shortly after
initiation, due to initially high effluent turbidity, in all cases greater than 1.0 TU.
Similar results were found when a single grade DE was used as a precoat in the
ERD units. Runs 1A-6A and 1-27 produced average effluent turbidities of 0.30,
0.22, 0.10, and 0.82 TU. As with the BIF unit, several runs in the ERD single
precoat series were aborted for reasons of initially high effluent turbidity, again
greater than 1.0 TU, shown in Appendix C-4.
Results of the amphibole mass analysis from the NWQL listed in Tables 12 and 13
show that a significant amount of DE was passing through the filter septa in both
the BIF and ERD units upon application of a single precoat of DE. This ac-
counted,^ part, for the relatively high effluent turbidity results. These conditions
were particularly noticeable when the single layer of DE involved was a relatively
fine grade of DE. It is probable that the openings in the filter septa used
59
-------
during the pilot study were not designed to retain the finer grades of DE which
were required to attain the low effluent turbidities desired. For instance, Run 2T
in the BIF unit and Run 2 in the ERD units utilized Hyflo Super Cel as a precoat
media, a relatively fine grade of DE, with resultant average effluent turbidity
levels of 0.65 and 0.56 TU, respectively.
Initially, in an attempt to prevent the DE passage through the filter septa and
improve the effluent turbidity levels, two grades of DE, fine and coarse, were
mixed and then coated with alum or polymer. Results from Table 12 for
the BIF unit indicate that some success was achieved. Average effluent turbidity
levels dropped from 0.47 TU to 0.18 TU in the case of Runs 8T-14T and Run 8
in which an anionic polymer, A-23, was added to a precoat composed of a fine
and a coarse DE. Similarly, results from Table 13 for the ERD units indicate
average effluent turbidity reductions from 0.30 to 0.10TU in Runs 21-27 in
which alum was added to a precoat composed of a fine and a coarse DE.
The runs conducted implementing the above two steps produced a serious opera-
tional problem involving backwashing of the filter septa. Single step precoat
applications to which coagulants were added so coated the filter septa with the
precoat that ordinary backwashing procedures were ineffective in removing the
solution in preparation for a subsequent run.
The two-step precoat procedure was used in a series of runs with the BIF unit. In
some runs, a medium coarse DE such as FW-20 or Celite 503 was used as the first
precoat step while a very coarse DE such as FW-50 or Celite 545 was used in the
remaining runs as the first precoat step. Polymers were added to both the first
and second precoat steps in some runs. Although the average effluent turbi-
dities were somewhat improved by the addition of polymers, the operational
problem of the precoat adhering to the filter septa during backwashing still per-
sisted. Runs in which either coagulants or polymers were added to only the
second step of the precoat eliminated this operational problem. In addition,
average effluent turbidity results were improved to levels as low as 0.10 and
0.08 TU in some chemical-precoat combinations.
The series of runs in the ERD units utilizing the two-step precoat procedure
produced even better results. The problem of the precoat adhering to the filter
media was eliminated and average effluent turbidities were reduced to levels
consistently below 0.10TU in almost all chemical-precoat combinations. In
addition to the elimination of the aforementioned operational problem and the
passage of DE through the filter septa, asbestiform fiber removal results became
60
-------
more consistent between the three reporting laboratories, especially in those runs
in which optimum concentrations of chemicals were utilized.
One other variable that was considered in the precoat application process was the
amount of weight of precoat per square foot of filter surface area. The initial runs
conducted with the BIF unit (Runs 1T-11T) used 0.48 kg/m2 (0.1 lb/ft2). The
precoat was increased to 0.73 kg/m2 (0.15 lb/ft2) in Runs 12T-14T and 1-24 in
an effort to build a more cohesive precoat and prevent DE passage through the
filter media. This same amount, 0.73 kg/m2 (0.15 lb/ft2), was also used in the
ERD units in Runs 1-27. It was found, however, that during the initial phases of
the runs, the effluent turbidity was quite high and required as long as 1 hour to
be reduced to a consistent and acceptable level. The amount of DE in the precoat
was then increased to 0.968 kg/m2 (0.20 lb/ft2) in a successful effort to reduce
the initially high effluent turbidity. The results of this practice are reflected in the
data from Runs 25-100 in the BIF unit and Runs 28-86 in the ERD units. It was
also found that the stability and integrity of the precoat were enhanced in that
hydraulic surges during the initial stages of the filter run would not cause the
precoat to rupture. The amount of precoat used with the BIF unit was increased
to 0.25 lb/ft2 in Runs 101-122.
Body Feed
Body feed application in the BIF unit was generally divided into two major
groupings. Runs 1-61 utilized three basic grades of DE in varying concentrations
without the addition of any chemicals. Runs 62-122 utilized these same grades in
varying concentrations, with the addition of chemicals such as alum and soda ash
and a cationic polymer, Catfloc B. The three basic grades of DE utilized in Runs
1-61 included fine grades such as FW-6, Celite 512 or Speedflow, medium grades
such as FW-12 or Hyflo Super Cel, or medium coarse grades such as Celite 503.
The initial selection of a fine grade DE for Runs 1T-14T, 1-16 and 46-61 was
made to produce a finished water with a low effluent turbidity level. However,
with the discovery of the DE bleed-through problem discussed previously, and in
an effort to lengthen the filter runs by reducing the rate of head loss buildup, a
change was made to a coarser or medium grade DE in Runs 17-42. This change,
in combination with the two step precoat process, was successful in reducing the
passage of DE through the filter septa and preventing the accumulation of DE on
the septa during the backwash procedure. Data from Table 12 have already been
cited as showing the subsequent improved effluent turbidity levels.
61
-------
The rates of body feed application in Runs 1-61 varied from approximately
40 mg/1 to 340 mg/1 in the case of fine media, 60 mg/1 to 300 mg/1 for medium
grade media, and 370 mg/1 for the medium coarse media. The general effect of
increasing the dosage rate in all cases was to decrease the filter run length due to
a faster rate of head loss buildup and increase the level of effluent turbidity re-
moval. The increase in rate of head loss buildup with an increase in body feed
rate would indicate that the body feed rates were above the optimum feed rate.
Varying the grade and dosage of body feed DE was accomplished in BIF Runs
62-122 in which chemicals were added to the body feed. A fine grade of DE was
used with alum and soda ash in Runs 62-78, 107-109, and 111-112. Dosage rates
varied in this series of runs from approximately 15 mg/1 to 110 mg/1. The data
show the general decrease in filter run lengths with an increase in dosage rate of
body feed DE. For example, in Runs 107-109, body feed dosages of Celite 512
varied from 30 mg/1 to 100 mg/1, respectively, while the filter run lengths
decreased from 20.5 hours to 8.5 hours. This same observation was made in
Runs 104-106 in which Hyflo Super Cel, a slightly coarser grade of DE, was
used as the body feed. Filter run lengths varied from 22 hours with a dosage rate
of 12 mg/1 to 9 hours with a dosage rate of 84 mg/1.
The concentration of body feed in the ERD units varied considerably, ranging
from 30 mg/1 to 200 mg/1 for the finer grades and 25 mg/1 to 100 mg/1 for the
coarser grades such as Celite 545 and FW-50. It was found that the coagulants
and polymers used in the precoat and body feed processes produced a greater
effect on the effluent turbidity and the rate of head loss than did the grade and
amount of body feed. This was probably due to the greater pressure differential
produced across the filter media in the ERD units that tended to dampen the
effects on the head loss produced by the various grades of media. In addition,
contrary to the operation of the BIF unit, effective filter area of the ERD units
tends to increase with increasing amounts of DE applied as body feed due to the
cylindrical shapes of the septa. Consequently, the rate of head loss buildup would
not be as rapid for these particular units.
Length of filter run of the DE units appeared to be affected more by changes in
raw water turbidity than was the filter run length on the granular units. However,
filter run length of the DE units was also affected by the amount of body feed
used in proportion to the raw water suspended solids. If the body feed rate is not
increased sufficiently whenever an increase in raw water suspended solids occurs, a
higher rate of head loss buildup will occur, with a resultant reduction in filter run
62
-------
length. If the body feed is increased too much in proportion to the suspended
solids, excessive thickness will be added to the filter cake, again resulting in rapid
head loss buildup and a shorter filter run. Optimum body feed rates, based on the
suspended solids concentration of the raw water, were not used during the pilot
filtration study.
Chemical Additions
Two series of runs were conducted utilizing an anionic polymer, A-23, introduced
to the total precoat in Runs 36-41 and to the second step of the precoat in Runs
79 and 80. Comparatively better effluent turbidity levels were achieved, with
averages of 0.15 and 0.22 TU. In addition, it was found that introduction of the
polymers to the second step of the precoat eliminated the backwashing problems
previously noted. Subsequent runs were then conducted utilizing this procedure.
Alum was used in two series of runs in which it was introduced in the second
precoat step. These were Runs 42, 46, 47, and 49-51 and Runs 52-57 which
produced average effluent turbidity levels of 0.27 and 0.29 TU. An additional two
series of runs were conducted in which soda ash was introduced with the alum.
This procedure was designed to maintain the pH of the precoat solution above
7.0 so that aluminum hydroxide would form on the DE precoat. The data
from these two series, Runs 58-61 and 89-103, show average effluent turbidity
levels of 0.35 and 0.23 TU.
The addition of alum and soda ash to the second precoat step and to the body
feed was investigated in Runs 62, 104-109, and 120-122. It was felt that the
initial positive charge resulting from the addition of alum and soda ash only
in the second step of the precoat was counteracted by the addition of the DE in
the body feed. Constant addition of the body feed would eventually produce a
net negative charge in the filter cake. With the addition of alum and soda ash in
the body feed, a filter cake with a net positive charge could be maintained.
Following this procedure, average effluent turbidities of 0.15 TU were produced.
In Runs 75-78, 111, and 112, an anionic polymer, A-23, was introduced in the
second precoat step and alum and soda ash were added to the body feed. In the
series in which a very coarse DE was used in the first step of the precoat, ex-
tremely good turbidity results were achieved, with an average effluent turbidity of
0.10TU.
A cationic polymer, Catfloc B, was introduced to the raw water prior to filtration
in Runs 82-84 and 86-88. Effluent turbidity levels averaged 0.23 TU. In Runs 110
63
-------
and 113-119, this procedure was carried a step further with the addition of alum
and soda ash to the second step of the precoat. Extremely good effluent turbidity
results, averaging 0.08 TU, were recorded.
A program of chemical addition similar to the one followed with the BIF unit
was conducted utilizing the ERD units. A single run was conducted, with an
anionic polymer, A-23, added to the second precoat step,and alum and soda ash
added to body feed. Excellent results were achieved, with an average effluent
turbidity of 0.05 TU.
Three series of runs were conducted using alum without polymers. In the first,
Runs 28-37, alum was added only to the second step of the precoat. The data
show that effluent turbidities of 0.12 TU were achieved. As in the case with the
BIF unit, soda ash was added to the alum in the next series, Runs 38-44, 54-60,
and 62-66, to promote the alum coating process. In this series of runs, the
effluent turbidities averaged 0.10TU. In the final series of runs in this grouping,
alum and soda ash were added both to the second precoat step and to the body
feed. Effluent turbidities were extremely low, averaging 0.06 TU.
Runs 45-51, 73-76, and 78-81 utilized a cationic polymer, Catfloc B, added to the
raw water prior to filtration. In the first series, Runs 45-51, an effluent with an
average turbidity of 0.12TU was produced. In Runs 73-76 and 78-81, with the
addition of alum and soda ash to the second precoat step, the average effluent
turbidity was further reduced to 0.06 TU.
Backwash Solids
The relationship of backwash solids to sludge volume is presented in Table 14.
The data given is based upon results of settleable solids tests performed on
samples of backwash water from the BIF and ERD units. Sludge production shown
in the table is in terms of sludge produced per 1000 gallons of water treated.
TABLE 14. RELATIONSHIP OF BACKWASH SOLIDS TO SLUDGE VOLUME
DE UNITS
Pilot
Unit
BIF
BIF
ERD-2
Run
11-4
121
85
B.W. SS,
lb/1000 gala
2.05
0.90
0.93
B.W. SS,
mg/1
89,156
20,112
14,504
Cone. @
60 min,
%
36.4
15.5
90.6
Sludge
volume,
gal/ 1000 gal
0.68
0.70
0.122
a Based on theoretical solids production
64
-------
COMPARISON OF TURBIDIMETERS
Parallel turbidity readings on the effluent from the MM-2 unit were made using
the Hach 2100A and the Hach 1720 turbidimeters. Typical data from the 2100A
and 1720 units are plotted on Figure 16 and all the data collected with the units
are tabulated in Appendix H. Readings obtained with the 1720 unit correlate
reasonably well with those obtained with the 2100A turbidimeter.
Parallel readings on the raw water were taken with both Monitek turbidimeters
and the Hach Model 2100A turbidimeter. Readings on the Monitek units were
very erratic and did not correlate to readings obtained with the Hach 2100A unit
until after one of the Monitek units was disassembled, the sight glass cleaned, and
the unit reassembled with an air purge connected. The air purge had been tried
previously with no effect on the performance of the units, apparently because the
sight glass was so dirty that the source light was being deflected as it passed
through the sight glass.
After the. sight glass was cleaned and raw water readings on the Hach 2100A and
the Monitek appeared to correlate, parallel readings were also taken on the
finished waters from the ERD unit and the MM-1 unit. Typical data from the
Hach 2100A and the Monitek unit are presented on Figures 17, 18, and 19 and
the data are tabulated in Appendix H.
All turbidity measurements discussed in comparisons of the various pilot filter
units tested will refer to data collected with the Hach 2100A unit.
65
-------
99
QZL\ HOVH QNV
HUM aswviao soNi
U31VM Q383inid
v ooiz HOVH
Aiiaigyni
nVOIdAl
o
o
p p
ro
o
OJ
o
m
-_ ro
fi
I?
Oj
CD
TURBIDITY,TU
p p o p
O ro c>J
a>
o
o
O
o
o
o
p p
b -
o
cr>
o
o
O
O
O
00
o
o
\l\3
Ag
i\S
Ol
o
I X
> >
o o
I I
ro
ro O
00
-------
Z.9
Sd3i3wiaiaani xaiiNow QNV v ooiz
HOVH HUM Q3NIV180 S9NIQV3d
HVOIdAl
L\
TURBIDITY , tu
p
bi
o
p
61
m
3
o
Q.
O
-------
o
o ^
CM UJ
X
II
o
O
Uj
to
6
CVJ
o
I I I I
o
o o
O
CD
Q C
Q- **v^
I I I I
O
o
* «»
CVJ k.
3
O
.c
UJ
o o 6 o
o
6
FIGURE 18 .TYPICAL ERD FILTERED WATER
TURBIDITY READINGS OBTAINED WITH
HACH 2100 A AND MONITEK TURBIDI-
METERS.
68
-------
I
8
a
o
UJ
o
o
q
d
FIGURE 19 . TYPICAL MM-I FILTERED WATER
TURBIDITY READINGS OBTAINED
WITH HACH 2IOOA AND MONITEK
TURBIDIMETERS.
69
-------
SECTION VII
DISCUSSION
RELATIONSHIP OF TURBIDITY, ASBESTIFORM FIBER COUNT AND
AMPHIBOLE MASS IN RAW WATER
Figures 20, 21, 22, and 23 present the relationships of raw water turbidity, ORF
amphibole and chrysotile fiber counts, and NWQL amphibole and suspended
solids concentrations, respectively. It was determined from the raw water asbesti-
form data that no discernible link was evident between the raw water turbidity
and the asbestiform levels. Appendices E, F and G give asbestiform analytical data
reported by ORF, NWQL and UMD, respectively.
GRANULAR MEDIA FILTRATION
Asbestiform Fiber Removal
In terms of fiber removal, either amphibole or chrysotile, a test run has been
considered successful if BDL or a value near BDL ( < 40,000 f/l)was reached in
the filtered water.
Table 15 summarizes the ORF data on amphibole and chrysotile removal for the
MM-2 runs and shows that 32 of 34 MM-2 runs for amphibole removal were
successful. The two unsuccessful runs (54 and 57) were made early in the study
and used dual media and single-stage mixing, which were not as effective as
tri-media and two-stage mixing. Table 7 presented data on MM-1 and indicated 21
of 23 runs achieved BDL or near it for amphibole removal, but only 2 of 23 were
successful for chrysotile removal. There appears to be evidence that contamination
of samples may have occurred either in the process of collection or perhaps in
sample preparation in the laboratory. The effects of long storage (2-7 weeks) on
the detection results are not definitely known but this, too, might have had a
bearing on some of the fiber counts. Hopefully, future studies will remove any
uncertainties and provide more reliable chrysotile data. In spite of these points,
certain evaluations on treatment effectiveness for chrysotile removal can be made
and are presented in Table 16.
To achieve BDL, or close to BDL for chrysotile, it was found that the minimum
alum dosage should be approximately 15 mg/1. Although MM-2 Runs 96, 105 and
122 prove to be exceptions to this by reaching BDL or near it with alum dosages
less than 15 mg/1, Runs 53, 54, 55, 57, 61, 71, 99, 101, 111 and 131 (which used
70
-------
4.0
3.0
Q Q
8
CD
£ cr
LL) ID
o:
LJ
I-
2.0
Q <
cr I.O
1.0 2.0 3.0
AMPHIBOLE FIBERS /LITER X I06
ONTARIO RESEARCH FOUNDATION
4.0
FIGURE 20. RELATIONSHIP BETWEEN RAW WATER TURBIDITY AT DULUTH
LAKEWOOD INTAKE AND ORF AMPHIBOLE FIBER COUNTS.
71
-------
4.0
3.0
z >
o m
s
Q <
o:
2.0
1
I
1.0
2.0
3.0
4.0
CHRYSOTILE FIBERS/LITER X |Q6
ONTARIO RESEARCH FOUNDATION
FIGURE 21 . RELATIONSHIP BETWEEN RAW WATER TURBIDITY AT DULUTH
LAKEWOOD INTAKE AND ORF CHRYSOTILE FIBER COUNTS.
72
-------
4.0
3.0
Q Q
O m
_l
UJ
IS
§1
cr
2.0
I
1
O.I 0.2 0.3 0.4
AMPHIBOLE FIBER CONCENTRATION (mg/l)
NATIONAL WATER QUALITY LABORATORY
FIGURE 22 . RELATIONSHIP BETWEEN RAW WATER TURBIDITY AT DULUTH
LAKEWOOD INTAKE AND NWQL AMPHIBOLE MASS CONCEN-
TRATION .
73
-------
4.01
Uj ^
5 30
-------
TABLE 15. GRANULAR MEDIA FILTRATION (HH-2) AHPHIBOLE AND CHRYSOTILE REMOVAL (ORF DATA)
Run Date
1 H/19
7 4/25
37 5/16
50 5/30
53 6/4
- 54 6/6
55 6/11
57 4/13
61 6/17
71 6/24
76 6/28
78 7/3
96 7/19
99 7/23
101 7/25
105 7/30
106 7/31
107 8/1
109 6/6
III 8/8
113 8/13
114 8/15
118 8/20
119 8/23
122 8/28
124 8/30
126 9/4
128 9/6
131 9/10
133 9/11
137 9/13
133 9/16
139 9/17
140 9/19
Filter
Rate
gpm/ft2
4.96
4.26
3.86
4.01
4.15
1.97
1.86
6.31
6.77
6.86
1.79
3.15
3.75
4.41
4.55
4.41
4.17
3.20
3.38
4.47
3.88
1.97
4.01
4.24
4.33
3.92
4.00
4.12
6. 18
5.98
3.98
4.29
4.35
3.85
6 Alum dosage probably
Cheni ca)
Turbidity
Amphi bole
Chemicals (mg/l)
FeCl3(8.22). C-3l(0.il8)
FeCI3(8.6l), C-3IJ0.626)
Alun(l8.62), N-I7J0.24)
Alumjie.O), 985-NJ0.054)
Alunjll. 19), 985-N(0.053)
Alun(l2.8), 985-N(0.052)
Alun(l3.72), 985-N(0.050)
A1um(M.I7), 985-N(0.05l)
AlunjlO.25), 985-K(0.037)
Alun (7.10), 985-0(0.079)
Alum(26.89), 98S-NJO.I59)
Alum(|iL66), 985-KJ0.079)
Alum(6.98), 985-N(0.060)
Alun(8.77), 985-NJ0.029)
Alun(9.99), 98E-N(0.0!3)
Alun(l0.97), 985-H(0.029)
Alira(20.97), 985-NJO.I22)
AIun(l6.l9), 985-HJ0.04)
Alum(l5. 10), 985-N(0.043)
Aluffl(ll.73), 985-NJ0.034)
Alum(l5.30), 98£-N(0.04l)
Aluro(l8.93-), 985-HJ0.042)
Alum(l5.43), C-B(0.56), A-23(0.65)
Alun(l4.24), 985-N(0.0>}2), A-23(0.28)
Alurajl 1.58), 985-11(0.049)
A-23JO.I60), Alun(l4.93), 965-11(0.064)
A-23(O.I58), Alun (16.95), C-3IJO.S67)
A-23(O.I24), Alum(l3.38), 985-NJ0.060)
Alum(l3.36), 985-NJ0.053)
Alum(24.56), 985-NJ0.077)
Alum(l9.66), C-3I(0.790)
Alun(l4.62). A-23(O.I26), 985-K(0.065)
Alum(l4.70), A-23JO.I23), C-3I(0.622)
Alum(l4.02), 985 11(0.066)
too low for effective chrysotile removal
Media
Dua
Dua
Dua
Dua
Dua
Dua
Dua
Dua
Dua
Dua
Dua
Tri
Trl
Tri
Tri
Tri
Tri
Tri
Tri
Tri
Tri
Tri
Tri
Tri
Tri
Tri
Tri
Tri
Trl
Tri
Tri
Tri
Tri
Tri
Stage
Mixing
2
2
2
2
2
2
2
2
2
2
2
2
2
in-lin
in-1 i n
in-1 in
in-lin
in-1 in
in-lin
in-1 in
2
2
2
Fibers x I06/l
Raw
0.304
0.348
2.61
1.43
1.74
1.5
1.0
0.48
0.61
1.0
0.91
0.56
0.52
0.54
O.I!
0.11
0.33
0.22
0.6
0.06
0.13
0.09
0.30
0.72
0.39
0.78
1.61
0.72
0.6
0.3
0.02
0.9
0.2
0.6
Filt.
BDL
BDL
BDL
BDL
BDL
0.09
BDL
0.15
0.04
0.02
BDL
0.02
0.02
BDL
BDL
BDL
BDL
BDL
BOL
BDL
BDL
0.02
0.02
BDL
BDL
0.02
BDL
BDL
BDL
0.02
0.02
0.02
BDL
BDL
Chry«otil«
Fibers x I0«/l
Raw
0.217
0.174
1.35
1.83
2.91
2.3
1.6
0.41
2.15
3.9
3.35
3.57
0.35
0.09
1.43
O.I 1
0.22
0.15
0.3
0.09
0.2
0.17
0.72
0.44
0.48
0.33
0.3
0.39
0.5
0.3
0.06
O.I
0. 1
0.09
Filt.
0.04
0.348
0.174
0.913
1.87
0.09
2.3
2.02
0.37
0.98
0.52
0.5
0.04
0.20
0.15
0.04
BDL
BDL
8DL
0.09
1.4
0.22
2.1
0.28
0.04
0.30
BDL
0.37
0.6
0.3
O.I
0.3
O.I
0.3
Filter Sun
Hrs.
14.6
15.0
12.2
23.7
24.0
67.3
33.1
15.6
11.6
12.8
9.4
23.7
7.8
17.9
14.7
25. 8
4.0
30.8
34.3
25.6
22.6
£8.6
20.2
21.8
19.9
19.6
17.8
21.0
10.0
7.4
9.6
21.8
21.4
31.9
Tin* to
Sample
Hrs.
6.5
4.5
9.8
7.5
4.7
25.6
26.9
13.5
4.0
4.3
4.9
3.3
2.1
7.4
13.1
4.3
2.4
17.9
23.5
9.9
22.3
45.7
1.9
18.5
3.7
3.7
9.6
8.7
2.9
4.3
5.3
1.9
I.S
i.e
Avg Filtered
Turbidity
T.U. Fenarkt
0.4
0.3
0.2
O.I
O.I
O.I
O.IS
0.25
O.IS
O.IS
0.07
O.I
O.IS
O.I ,d,f
0.15
0.10
0.06
0.08
0.07
0.09 ,d
0.06 ,d,f
0.06 ,f
0.10 ,f
0.08
0.06
0.07
0.06
0.07
0.07 a,c
g
0.07 c,d
0.06 d.e
0.07 d,«
0.06 b.d
breakthrough in less than 9 hours
d
8 A-23 interference wi
.
removal sat
sf actor i ly
Contamination of sanple or laboratory problems suspected
Great variation in final turbidity - poor run
-------
TABLE 16. GRANULAR MEDIA FILTRATION - COMMENTS ON
CHRYSOTILE REMOVAL (MM-2)
Run
Chrysotile
106
107
109
126
Chrysotile
96
105
122
Runs with
37a
soa
76a
78b
113C
114°
137C
139C
140°
118C
119C
124C
138°
Filt. Rate
gpm/ft2
Runs Which Achi
4.17
3.20
3.38
4.00
Chemicals Media
eved BDL
Alum (20.97)
985-N (0.122) Tri
Alum (16.19)
985-N (0.04) Tri
Alum (15.10)
985-N (0.43) Tri
Alum (16.95)
A-23 (0.158)
C-31 (0.567) Tri
Mixing
2 Stage
2 Stage
2 Stage
In-Line
Filtered
Chrysotile
(Fibers x 106/1)
BDL
BDL
BDL
BDL
Runs Which Achieved 0.04 x IO6 fibers/1
3.75
4.41
4.33
Similar Treatment
3.86
4.01
1.79
3.15
3.88
1.97
3.98
4.35
3.85
4.01
4.24
3.92
4.29
Alum (6.98)
985-N (0.060) Tri
Alum (10.97)
985-N (0.029) Tri
Alum (11.58)
985-N (0.049) Tri
Which Did Not Achieve Desired Goal
Alum (18.62)
N-17 (0.24) Dual
Alum (16.0)
985-N (0.054) Dual
Alum (26.89)
985-N (0.1 59) Dual
Alum (14.66)
985-N (0.079) Tri
Alum (15.3)
985-N (0.041) Tri
Alum (18.9)
985-N (0.042) Tri
Alum (19.66)
C-31 (0.790) Tri
Alum (14.7)
C-31 (0.622)
A-23 (0.125) Tri
Alum (14.02)
985-N (0.066) Tri
A-23 (0.65)
Alum (15.43)
Catfloc B (0.56) Tri
A-23 (0.28)
Alum (14.24)
985-N (0.042) Tri
A-23 (0.160)
Alum (14.93)
985-N (0.064) Tri
A-23 (0.126)
Alum (14.62)
985-N (0.065) Tri
2 Stage
2 Stage
In-Line
Single
Single
Single
2 Stage
2 Stage
2 Stage
In-Line
2 Stage
2 Stage
2 Stage
2 Stage
In-Line
2 Staee
0.04
0.04
0.04
0.174
0.913
0.52
0.50
1.4
0.22
0.1
0.1
0.3
2.1
0.28
0.30
0.3
a Utilized dual media and single stage flash mixing.
bNo apparent reason for not achieving desired goal.
Alum dosage too low.
76
-------
low alum dosages and 985-N) support this statement in that BDL was not
reached. Runs 37, 50, and 76, although utilizing alum in excess of 15 mg/1, were
not successful probably due to the use of single stage flash mix and perhaps dual
media. Run 78 did not achieve BDL, and there are no apparent reasons why it
did not. Runs 113,114, 137, 139, and 140 failed to achieve BDL principally as a
result of the effect of low chrysotile in the influent. When influent chrysotile
values are low, differentation between influent and effluent values is apparently
lost. Finally, Runs 118, 119, 124, 138, and 139 were not successful due to the
suspected interference of A-23 and alum. This apparently was overcome in Run
126 by the presence of C-31 where an effluent chrysotile of BDL was achieved.
In Run 118, Catfloc B was unable to offset the substantial dosage of A-23. Run
126 employed three in-line mixers arranged in series, with three separate chemicals
fed, one upstream from each mixer.
Turbidity Removal
Table 15 shows filtered water average turbidity achieved in all MM-2 runs where
asbestiform analyses were made. Raw water turbidity usually was quite low during
the pilot plant study, generally less than 1.0 TU and only rarely increasing to
about 4 TU. Filtered water turbidity averages in Table 15 reached a maximum
of 0.4 TU for Run 1 but decreased steadily as experience led to better chemical
treatment, mixing, and tri-media filtration. As shown in the table, the average final
turbidities quickly dropped to less than 0.2 TU and during the last 16 runs,
nearly all values were less than 0.1 TU. It is felt that granular filtration employing
tri-media, together with proper chemical treatment and mixing, is capable of
producing consistently a filtered water turbidity less than 0.1 TU.
Although turbidity removals were approximately the same in runs where alum was
used as the coagulant, compared to runs in which FeCl^ was used, filter run
lengths were consistently longer in those runs where alum was used.
Design Filter Rate
Table 17 shows data on 25 additional MM-2 runs where no fiber removal analyses
were made plus 6 runs with fiber analyses data. All runs were made at about
^\
5 gpm/ft or higher. This information has significance when interpreted on the
basis of Figure 24 which shows the relationship between filtered water turbidity
and amphibole fiber count. This figure clearly shows that when a turbidity of 0.2
TU or less was achieved, the corresponding amphibole count was 0.02 x 10° f/1 in
the vast majority of the samples. This fiber level was reported as BDL by ORF.
77
-------
TABLE 17. GRANULAR MEDIA FILTRATION (MM-2) FILTER RATES
5 gpm/ft2 AND GREATER
Run
Group
30
51
58
59
60
62
63
64
65
66
67
68
69
70
72
73
74
85
86
87
88
89
129
130
132
Group
1
57
61
71
131
133
Date
Filter Avg. Turb.
Rate Treated Water
gpm/ft^ TU
Filtered Water
Amphibole Chrysotile
Fibers x 10°/1
1 (No Fiber Analysis Made)
5/10
5/31
6/13
6/14
6/14
6/18
6/18
6/19
6/19
6/20
6/20
6/20
6/21
6/21
6/25
6/25
6/25
7/10
7/11
7/11
7/12
7/12
9/9
9/10
9/11
2 (Fiber Analysis
4/19
6/13
6/17
6/24
9/10
9/11
5.28
5.09
5.33
5.65
6.82
6.84
6.91
6.86
6.87
7.82
6.82
6.89
7.47
6.78
6.17
6.64
6.91
4.85
4.91
4.94
4.86
4.79
6.16
6.09
5.97
Data Available)
4.96
6.31
6.77
6.86
6.18
5.98
0.09
0.43
0.10
0.13
0.13
0.19
0.17
0.18
0.14
0.12
0.12
0.15
0.22
0.30
0.12
0.15
0.35
0.12
0.08
0.14
0.13
0.08
0.12
0.07
0.16
0.40
0.25
0.15
0.15
0.07
_
.
-
BDL 0.04
0.15 2.02
0.04 0.37
0.02 0.98
BDL 0.50
0.02 0.30
78
-------
1.0
0.14
0.12
0.10-
₯. 0.08
CO
-------
For purposes of this study, final fiber counts of 40,000 f/1 or less, or BDL were
considered to be successful. As shown on Figure 25, the same relationship was not
apparent between effluent turbidity and chrysotile fiber count.
Table 17 shows under Group 1 that 22 of 25 runs at filter rates of 5 gpm/ft^ or
greater achieved an average turbidity of 0.2 TU or less and quite likely would
have reached BDL or close to it, had asbestiform analyses been made. This data
suggests that, as an initial operational guide,filter runs be terminated when turbi-
dities reach 0.2 TU. Because some of the runs should have been terminated
earlier, this would have avoided some high turbidity data and could have
produced additional successful runs.
Runs 1, 61, 71, 131, and 133 shown in Table 17 under Group 2 were operated at
^
filter rates of 5 gpm/ftz or greater and were successful for amphibole removal.
^
The only other run at above 5 gpm/ftz (Run 5 7)was unsuccessful for amphibole
removal and this is attributed to dual media and single-stage mixing. On this basis,
5 of 6 runs at high filter rates where fiber removal analyses were made, were
successful for amphibole removal. One of the 6 runs under Group 2 was suc-
cessful for chrysotile removal. Poor chrysotile removals may have been due to low
alum dosage, dual media, and single stage mixing in Runs 57, 61 and 71. Low alum
dosage could have been the reason in Run 131. Possible sample contamination,
long storage, or laboratory problems might explain why Run 133 was unsuccessful
for chrysotile removal.
Based on the above data, it is reasonable to conclude that a filter rate of
^
5 gpm/ftz with proper chemical treatment and good mixing should 'achieve
excellent results (BDL or near it) for amphibole fiber removals.
DIATOMACEOUS EARTH FILTRATION
Asbestiform Fiber Removal
Table 18 shows fiber removal data for selected BIF (vacuum) DE runs. Three runs
(8, 10T, and 12T) employing A-23 polymer in one precoat step treatment
achieved success for amphibole removal in two of the runs but in only one for
chrysotile removal. Four runs (37, 70, 72, and 79) used A-23 in the second pre-
coat step and all runs were successful for amphibole removal. Two of the four
runs reached BDL or near it for chrysotile removal. Runs 77 and 111 used A-23
in the second precoat step but in addition added alum and soda ash to the body
feed. Both runs were successful for amphibole removal, but neither was suc-
cessful for chrysotile removal. Runs 82, 84, and 88 used Catfloc B
80
-------
23x10°
187 xlO6
Z.OxlO6
2.1 » IO
1.6
1.4
1.2
en
a:
a
UJ
O 0.8
CO
>-
o:
x
o
Z 0.6
UJ
u.
UJ
0.4
0.2
0.2
0.4
0.6
0.8
1.0
BDL'S
EFFLUENT TURBIDITY, TU
FIGURE 25 .EFFLUENT TURBIDITY VS. CHRYSOTILE
FIBER COUNT GRANULAR MEDIA.
-------
TABLE 16. VACUUM (BIF) AND PRESSURE (ERD) DIATOHACEOUS EARTH FILTRATION SELECTED FILTER RUNS (ORF DATA)
Run
Date
Vacuum
8 5/16
IOT
121
37
79
70
72
no "
OO
to '"
82
84
88
113
116
117
h8
Pre!
11
73
78
79
45
46
48
49
51
5/7
5/9
6/24
8/1
7/19
7/23
7/30
9/6
8/6
8/8
8/13
9/9
9/11
9/13
9/16
syre
9/6
9/9
9/13
9/16
7/30
8/1
8/6
8/8
8/13
Filter
Rate
jpm/ft*
0.948
1.035
0.979
0.866
0.877
0.925
0.920
0.958
0.946
0.818
0.826
0.914
1.002
0.862
0.330
0.826
.108
.146
.093
.092
.01
.087
.261
.266
.059
Feed To
Raw Mater
lbs/1000 gal
- - - .
....
.
....
Cat Floe 8(0. Oil)
Cat Floe 8(0.0016)
Cat Floe 6(0.0032)
Cat Floe 6(0.0024)
Cat Floe 8(0.0046)
Cat Floe 6(0.0036)
Cat Floe 6(0.0049)
...
Cat Floe 8(0.0047)
Cat Floe 8(0.0050)
Cat Floe 6(0.0028)
Cat Floe 8(0.0040)
Cat Floe 8(0.0037)
Cat Floe 8(0.0044)
Cat Floe 8(0.0044)
Cat Floe 8(0.0057)
Fir.t Precoat
Ibj/IOOO gal
FW-50(0.079), FW-6(0.039). A-23(0. 000001 . 18)
512(0.070), A-23(0. 000002.1)
Speed flow (0.138), A-23(0. 000004. 1 )
503(0.174)
546(0.066)
FW-20(0.099)
503(0.063)
546(0.063)
FW-50(0. 166)
645(0.051)
546(0.114)
545(0.036)
FW-50(O.I09)
FW-50(0. 131)
FM-5P(O.I30)
F»-50(0. 141)
FW-50( 0.067)
FW-50J0.068)
FW-50(O.III)
FW-50( 0.077)
545(0.036)
546(0.054)
545(0.050)
545(0.030)
FW-50(0.066)
FW-6(O.I74)
545(0.056),
512(0.099),
512(0.063),
545(0.053),
512(0.111),
512(0.051)
Second Precoat
lbi/1000 gal
. - - -
....
_ _ _ _
A-23(0.000007.0)
A-23(0. 000002. 82)
A-23JO. 000001. 98)
A-23JO. 000001. 89)
A-23(0.000002.67)
A-23(0.000003.32)
Super Cel (0.114)
612(0.036)
512(0.072),
612(0.088),
.512(0.086),
'512(0.094),
512(0.067),
512(0.068),
512(0.111),
512(0.077),
512(0.036)
512(0.054)
512(0.050)
Super Cel(0
Super Cel(0
Alum(0.0060), Soda A»h(0.0040)
Alum(0.0096), Soda Ath(0.0048)
Alum(0.0095), Soda A»h(0.0048)
Alum(O.OIO), Soda Asn(0.0052)
A-23( 0.000002. 02)
Alum(0.0074), Soda Ash(0.0037)
Alum(O.OI2), Soda Ash(0.006l)
Alum(0.0085), Soda Ash(0.0042)
030)
066)
Body Feed
lbi/1000 gal
FV-6(0.550)
512(0.483)
512(0.51!)
FW-6(0.677)
512(0.577)
512(0.243)
512(0.174)
512(0.213), Alun(O.OI2), Soda Ath(O.OOSI)
512(0.229). Alum(0.025), Soda A«h(O.OI3)
512(0.306)
512(0.469 )<
FW-50(0.877)
Super Cel (0.952)
6I2(0.4CO)
Aqua Cel(0.58B)
Aqua Cel(l.29l)
512(0.233). Alun(0.026), Joda A.h(O.OI3)
Super Cel (0.171)
512(0.407)
Aqua Cel (0.228)
545(0.486)
545(0.882)
545(0.351)
FW-SO(0.48t)
FW-SOJ0.676)
Aiwnibola
Fiben > IC*/|
Raw
2.61
0.522
0.870
1.0
0.72
0.52
0.54
0.11
0.72
0.6
0.06
0.13
1.0
0.3
0.02
0.9
0.72
1.0
0.02
0.9
0.11
0.22
0.6
0.06
0.13
Filter
0.565
BDL
BOL
BDL
BDL
BDL
BDL
BDL
0.02
BDL
BDL
BDL
O.OK
BDL
0.02
BDL
BDL
BDL
BDL
0.02
BDL
BDL
0.02
BDL
0.04
ChrytotiU
Fiberi > 106/1
RaH
1.35
0.130
1.78
3.9
0.15
0.35
0.09
0.11
0.39
0.28
0.09
0.2
oh
0.3
0.06
0.15
0.39
0.5
0.06
O.I
0.11
O.IS
0.3
0.09
0.20
Filter
0.652
1.43
B'DL
0.2
0.04
BDL
0.76
0.07
0.59
0.06
0.4
O.I
1.0
0.13
0.4
1.3
1.67
1.0
0.06
0.06
0.07
0.13
IDL
0.04
0.5
-------
added to the raw water and all three were successful for amphibole removal but
none for chrysotile. Runs 113, 115, 117, and 118 had Catfloc B added to the raw
water and alum and soda ash to the second precoat step. All runs were successful
for amphibole removal but none for chrysotile.
It is clear that several treatment processes in vacuum DE filtration will remove
amphibole fibers successfully but the picture for chrysotile removal is far less
clear. Part of the problem may be due to sample contamination, long storage, or
laboratory problems. Based on the data, however, the most promising treatment
for chrysotile removal of those attempted appears to require the use of A-23 in
the second precoat step. This would appear to be a good starting point in any
future research.
Table 18 shows fiber removal data for selected ERD (pressure) DE runs. Run 72
employed A-23 in the second precoat step and alum and soda ash in the body
feed and achieved BDL for amphibole but was unsuccessful for chrysotile
removal. Run 73 employed Catfloc B added to the raw water and alum and soda
ash added to the second precoat step. This run resulted in BDL for amphibole but
was unsuccessful for chrysotile removal, Runs 78 and 79 were performed with
Catfloc B added to the raw water and alum and soda ash added to first precoat
step. Amphibole removal was successfully accomplished in both runs, but neither
run achieved satisfactory chrysotile removals. In Runs 45, 46, 48, 49 and 51,
Catfloc B was added to the raw water. Amphibole removal was successful in all
5 runs but of chrysotile removal was successful in only 2 of the 5 runs.
It is evident that a number of treatment processes in pressure DE filtration will
remove amphibole fibers but successful treatment for chrysotile removal is less
apparent. Reasons for the difficulties in chrysotile removal may not all be in treat-
ment as mentioned previously, but also in collection, transporting, and processing
samples in the laboratory. Using the available data, however, the most promising
treatment for chrysotile removal of those attempted appears to require addition
of Catfloc B to the raw water. Addition of alum and soda ash to the second
precoat step may also be worth further testing. Together with testing addition of
Catfloc B to the raw water, these two processes appear to be good starting points
for future studies.
Turbidity Removal
The effectiveness of DE media filters for removal of suspended material was
clearly shown by the results of the pilot study. Because vacuum DE filtration
83
-------
caused problems due to dissolved gases, this treatment was not considered suitable
for treatment of Lake Superior water. Figures 26, 27, 28, and 29 show the rela-
tionship of amphibole and chrysotile fibers vs. effluent turbidity for both vacuum
and pressure DE filtration.
Appendices C-3 and C-4 provide data on pressure DE filtration. Data in these
Appendices show that initial runs (1-30) produced final turbidities which generally
were below 0.5 TU but sometimes approached 0.8 TU. Runs 31-60 produced
lower final turbidities,generally about 0.1 TU, due to improved chemical treatment
resulting from the experiences gained in the earlier runs. Beginning at Run 62
and continuing to the final Run 86, 21 runs produced final turbidities which were
consistently below 0.1 TU. Of the 21 runs, 8 employed addition of Catfloc B to
the raw water, and 8 others utilized alum and soda ash in the body feed. All 21
used alum and soda ash in the second precoat step. It is apparent that for the raw
water turbidity range encountered during the pilot plant study it is possible to
produce consistently a final water turbidity of 0.1 TU, or less, with proper
chemical treatment and mixing.
SUMMARY OF COST ANALYSIS
Analysis of the results presented in Section VII indicates several categories of treat-
ment that were essentially similar in their effectiveness for removal of asbestiform
fibers and other suspended solids. These categories were examined initially in
Section VII considering such parameters as success in amphibole fiber and mass
removal, turbidity removal, length of filter run, rate of head loss buildup, and
amount of suspended solids in backwash water. Due to the similarity of results in
the various categories and in order to compare the two basic types of treatment
processes, granular media filtration and diatomaceous earth filtration, an economic
analysis was conducted. In this manner the cost of treating water on a per 1,000
gallon basis could be determined for each category selected and an economic com-
parison made among all the categories.
In the economic analysis the construction or capital costs were based on recent
costs of construction for facilities with similar design and operating parameters
and on costs supplied by equipment manufacturers. A treatment plant with a
nominal capacity of 30 mgd was selected for a detailed analysis utilizing the same
water supply as tested in the various pilot units. All cost estimates were based on
December 1974 price levels. A 25 percent contingency allowance for omissions,
engineering, legal and administration costs was included in the capital costs. No
factor for inflation was assumed.
84
-------
1.6
*>
QL
UJ
CD 1 0
Ul
o
X
Q.
2
Ld
0.6
UJ
0.4
0.2
1.
0.4
0.6
0.8
BDL'S
0.2
EFFLUENT TURBIDITY, TU
FIGURE 26. EFFLUENT TURBIDITY VS. AMPHIBOLE
FIBER COUNT VACUUM DE FILTRATION.
-------
I.IXIO*
1.6 r
1.4
1.2
I
2] 1.0
OD
UJ
O
(O
>-
cc
o
o
do
0.6
UJ
0.2
1.83 X (O*
BDL'S
f^ri
0.4
0.6
0.8
1.0
EFFLUENT TURBIDITY, TU
FIGURE 27. EFFLUENT TURBIDITY VS. CHRYSOTILE
FIBER COUNT VACUUM DE FILTRATION.
86
-------
1.6
O
tr
iil
UJ
O
50.8
i
CL
2
<
£0.6
UJ
u.
u.
UJ
0.4
0.2
fttit
XX XXXX XX
X X X X
BDL'S
0.2 04 06 08 1.0
EFFLUENT TURBIDITY,TU
FIGURE 28. EFFLUENT TURBIDITY VS . AMPHIBOLE
FIBER COUNT PRESSURE DE FILTRATION.
87
-------
. 53XI06 , 9.3 XIO6
t 6 r
1.4
1.2
CO
(r
ui
00 1.0
LU
08
CE
X
o
0.6
UJ
u.
0.4
0.2
BDL
0.2 0.4 0.6 0.8
EFFLUENT TURBIDITY, TU
i.o
FIGURE 29 .EFFLUENT TURBIDITY VS. CHRYSOTILE
FIBER COUNT PRESSURE DE FILTRATION.
88
-------
Operation and maintenance cost estimates were based on those chemical costs
actually incurred in the pilot plant operation and by existing water treatment
plants with similar treatment processes as those considered. Solids produced in the
backwash process were based on calculated amounts rather than the lower
amounts found in samples analyzed during the runs, to correct for any potential
sampling error. An average flow rate of 20 mgd was assumed as the basis for
determining operation and maintenance costs.
A 50 year economic analysis was performed on each treatment process selected
utilizing a discount rate of 5.625 per cent as required by the Federal Water
Resources Council for making economic comparisons of alternative projects. Total
first costs, replacement costs, average annual costs, and cost per 1,000 gallons of
water treated were determined.
Granular Media Filtration
Three categories of treatment were selected from the runs utilizing granular media
filtration. The selection of these categories had, as a common base, amphibole
fiber count and mass removal to a level near or below the detectable limits (BDL)
of the analytical equipment and an average effluent turbidity of 0.1 TU or less.
Economically, the three categories were quite attractive, with average chemical
costs of 0.35, 0.4, and 0.5 £/1000 gallons, respectively, as compared to the range
of average chemical costs for all granular media runs of 0.35 to 1.26 9/1000
gallons.
The first category, or treatment process, selected utilized dual media filtration with
single-stage mixing as represented by MM-1 Runs 28, 29, 30, 39, and 40 and
MM-2 Runs 50 and 53. Alum and a nonionic polymer, 985N, were introduced at
average rates of 15 and 0.05 mg/1, respectively, with a filter flow rate of
^9 9
235 mj/m day (4.0 gpm/ftz), producing an average effluent turbidity of
0.1 TU. Amphibole fiber count and mass removals were excellent,with 4 out of 4
amphibole fiber counts and 4 out of 4 amphibole mass removals achieving the
above stated removal goal. The economic analysis of this treatment process, based
on a 30 mgd treatment plant design, is presented in Table 19 and shows that
treated water could be produced for approximately 7.2 (j;/1000 gallons.
The second granular media filtration category selected utilized a tri-media filter
o 9
bed with two-stage flash mixing at a filter flow rate of 293 irr/mz day (5.0
s\
gpm/ftz). This category is represented by MM-2 Runs 93, 94, 103, 104, and
105. Alum and a nonionic polymer, 985N, were introduced at average rates of 11
89
-------
TABLE 19.
WATER TREATMENT PLANT ECONOMIC ANALYSIS - GRANULAR MEDIA FILTRATION
30 MOD PLANT DESIGN - LAKE SUPERIOR INTAKE AT LAKEWOOD
Process Category
Dual Media Bed
Q = 235 m3/m2 day
Single Stage Mixing
Alum & Nonionic Polymer
<£>
O
Capital Costs
First
Replacement-present worth
of replacement equipment
Total-present worth of all
capital costs, 50 year
analysis at 5-5/8%
Actual Costs
Tri-Media Bed
Q = 293 m3/m2 day
Two-Stage Flash Mixing
Alum & Nonionic Polymer
5.50
0.66
6.16
Actual Costs
$x!Q6
5.25
0.63
5.88
Tri-Media Bed
Q = 293 m3/m2 day
In-Line Mixing
Alum & Nonionic Polymer
Actual Costs
SxlQ6
5.25
0.63
5.88
Average Annual Costs
First
Replacement
O&M - calculated at 20 mgd
Total
Costs per 1000 gallons
First
Replacement
O&M - calculated at 20 mgd
Total
SxlO6
0.33
0.04
0.13
0.50
6/1000 gal
4.70
0.60
1.90
7.20
SxlO6
0.32
0.04
0.12
0.48
6/lOOOgal
4.49
0.57
1.73
6.79
SxlO6
0 32
0 04
0 12
0.48
$/l OOP gal
449
0 57
1 79
6.85
-------
and 0.03 mg/1, respectively, producing an average effluent of 0.1 TU. Amphibole
fiber count and mass removals again were excellent, with the desired removal goals
achieved for both the amphibole fiber count and mass removal. The economic
analysis of this category of treatment, based on a 30 mgd treatment plant design,
is presented in Table 19 and shows that treated water could be produced for
approximately 6.79 6/1000 gallons.
The third granular media filtration category selected also utilized tri-media
filtration at a filter flow rate of 293 m3/m2 day (5.0 gpm/ft2) and is represented
by MM-2 Runs 121, 122, and 123. Alum and a nonionic polymer, 985N, were
introduced through a series of in-line mixers at rates of 12 and 0.05 mg/1,
producing an average effluent turbidity of 0.06 TU. Amphibole fiber count and
mass removals were excellent, with amphibole fiber count and amphibole mass
removal achieving the desired goal for the single test performed. With a treatment
plant designed for 30 mgd, treated water could be produced for approximately
6.85 6/1000 gallons as shown by the economic analysis presented in Table 19.
Diatomaceous Earth Filtration
Three categories of treatment also were selected for economic analysis from the
runs utilizing pressure diatomaceous earth filtration. The selection of these cate-
gories had, as a common base, amphibole fiber count and mass removal to a level
near or below the detectable limits of the analytical equipment and an average
effluent turbidity of 0.1 TU or less. The three categories were the most eco-
nomically attractive of the various categories screened, with average chemical and
DE costs of 2.7, 2.3, and 2.9 6/1000 gallons, respectively, as compared to the
range of average chemical costs for all pressure DE runs of 2.2 to 4.5 £/1000
gallons.
The first category selected utilized a medium grade DE precoat applied at 0.73
kg/m2 (0.15 lb/ft2) of filter surface area coated with alum and soda ash at rates
of 0.11 g/g of DE and 0.055 g/g of DE, respectively. This category is represented
by ERD-2 Runs 67, 68, and 69. A fine grade body feed was applied at a rate of
27.5 mg/1, with alum at 3 mg/1 and soda ash at 1.5 mg/1 added to the body feed.
A filter rate of 61 m3/m2 day (1.04 gpm/ft2) was utilized to produce an effluent
with an average turbidity of 0.06 TU. Amphibole fiber count and mass removal
achieved the removal goal in 2 out of 2 tests. The economic analysis of a treat-
ment plant designed for 30 mgd and based on this treatment is presented in
Table 20. The analyses indicate that a treated water could be produced for 11.97
6/1000 gallons.
91
-------
TABLE 20.
WATER TREATMENT PLANT ECONOMIC ANALYSIS - PRESSURE DIATOMACEOUS EARTH
FILTRATION 30 MOD PLANT DESIGN - LAKE SUPERIOR INTAKE AT LAKEWOOD
Process Category
*}
Q = 61 rrr/mz day
Alum & Soda Ash to
Precoat and Body Feed
Fine Body Feed
Q = 67 m3/m2 day
Alum & Soda Ash to
Precoat-Cationic
Polymer to Raw Water
Fine Body Feed
Q = 64 m3/m2 day
Alum & Soda Ash to
Precoat-Cationic
Polymer to Raw Water
Medium Fine Body Feed
Capital Costs
First
Replacement-present worth
of replacement equipment
Total-present worth of all
capital costs, 50 year
analysis at 5-5/8%
Actual Costs
$xlQ6
6.50
0.50
7.00
Actual Costs
$xlQ6
6.10
0.46
6.56
Actual Costs
$xlQ6
6.35
0.49
6.84
Average Annual Costs
First
Replacement
O&M - calculated at 20 mgd
Total
Cost per 1000 gallons
rirst
Replacement
O&M - calculated at 20 mgd
Total
$x!06
0.39
0.01
0.43
0.83
£/1000gal
5.43
0.14
6.40
11.97
SxlO6
0.37
0.01
0.39
0.77
6/1 000 gal
5.09
0.14
5.80
11.03
SxlO6
0.38
0.01
0.44
0.83
£/1000gal
5.30
0.14
6.56
12.00
-------
^
The second category utilized a medium grade DE precoat applied at 0.73 kg/m
(0.15 lb/ft2) of filter surface area coated with alum and soda ash at rates of
0.11 g/g of DE and 0.055 g/g of DE, respectively. This category is represented by
ERD-2 Runs 73, 74, and 75. A fine grade body feed was applied at a rate of
20.5 mg/1. A cationic polymer, Catfloc B, was added to the raw water at a rate of
0.56mg/l. A finished water with an average effluent turbidity of 0.06 TU was
produced. Amphibole fiber count and mass removals were excellent, achieving the
removal goal in 2 out of 2 tests. The economic analysis presented in Table 20
shows that for this category a treated water could be produced for 11.03 6/1000
gallons.
<}
The third category utilized a medium grade DE precoat applied at 0.73 kg/m
(0.15 lb/ft2) coated with alum and soda ash at rates of 0.11 g/g and 0.055 g/g
of DE, respectively. This category is represented by ERD-2 Runs 79, 80, and 81.
A fine grade of body feed was applied at a rate of 27.3 mg/1. A cationic polymer,
Catfloc B, was added to the raw water at a rate of 0.33 mg/1 to produce a
finished water with an average effluent turbidity of 0.06 TU. Amphibole fiber
count and mass removals were excellent, achieving the removal goal in 2 out of 2
tests. The economic analysis.presented in Table 20 shows that for this category
treated water could be produced for approximately 12.0 £/1000 gallons.
Two other pressure DE categories with equally attractive chemical costs that were
examined utilized only alum and soda ash applied to the precoat. These categories
were represented by ERD Runs 59, 60, and 62 and ERD Runs 55, 57, and 58,
Turbidity removal was not as successful, however, with average effluent turbidities
in both categories greater than 0.1 TU.
A theoretical optimization was performed for the three selected pressure DE
filtration categories to simulate the least cost performance for each process and
the optimum design characteristics of the treatment plants in each case. Pilot
study data from the three runs involved in each of the three categories were
utilized to predict the buildup of filter aid density for that category. From this
value, values for the filtration rate, terminal head loss, body feed rate, filter area,
optimum length of filter run, and the optimum filter operating costs were
calculated. For each category, a filter flow rate of 94 m3/m2 day (1.6 gpm/ft2)
was predicted as compared to 60 m3/m2 (1.0 gpm/ft2) utilized in the pilot
studies. Body feed rates calculated were substantially lower than those actually
run in the pilot studies, averaging between 5 and 10 mg/1 as compared to 20 to
30 mg/1. Details on the DE optimization process are provided in Appendix I.
93
-------
The optimization resulted in lower capital and average annual operation costs for
all three of the selected categories. The costs in £/1000 gallons of treated water
were optimized at 10.55, 8.50, and 8.14, respectively for the three categories, as
compared to costs in
-------
pressure DE filtration. The O&M costs, composed of chemical, power, and labor
costs, constitute approximately 25 per cent of the total costs for the granular me-
dia filtration as compared to approximately 53 per cent for the pressure DE filtra-
tion. This owuld mean that increases in O&M costs due to inflation would affect
future costs in C/1000 gallons for pressure DE filtration more than granular
media filtration.
The choice between pressure or vacuum DE filtration and granular media filtration
would be dependent upon the specific application. Several factors,such as quality
of raw water and quantity of water to be treated, would have to be examined
before a definite selection could be made. In the case of a treatment plant at
Duluth designed for 30 mgd and utilizing Lake Superior water at the Duluth
Lakewood Intake, the choice can be made based on the economic evaluation
presented. However, the choice of treatment plants for very small capacity instal-
lations utilizing water with different quality parameters would require a detailed
study of the specific conditions at each potential site.
TABLE 21. RANGES OF INITIAL CAPITAL COSTS OF VARIOUS PLANT
CAPACITIES UTILIZING GRANULAR MEDIA AND PRESSURE
DE FILTRATION FOR ASBESTIFORM FIBER REMOVALa
Plant Capacity (mgd)
Process
Granular Media
Pressure DE
0.03
$x!06
0.04-0.05
0.05-0.06
1.0
$x!06
0.19-0.24
0.22-0.25
10.0
$x!06
2.9-3.4
3.1-3.5
30.0
$x!06
5.2-5.5
6.1-6.5
a Costs presented do not consider replacement costs or costs associated with pretreatment.
Items such as existing equipment or facilities, existing distribution and storage, raw water
quality, and site location would modify the general cost ranges presented.
MONITORING PLANT OPERATION
Purpose of Monitoring
It is customary to monitor the quality of drinking water after it is treated by a
water utility. This is done for a number of reasons, including assuring both the
consumers and regulatory authorities that the water meet appropriate standards,
and assuring utility operators and managers that the treatment plant is being
95
-------
operated properly and that the goals of the treatment processes are being
achieved.
Monitoring techniques vary. The efficacy of clarification can be monitored by
measuring turbidity. Waters are disinfected to kill pathogenic microorganisms, but
judgments on the achievement of this goal are made on the basis of the presence
or absence of indicator microorganisms, along with measurements of turbidity and
chlorine residual, rather than on specific tests for pathogens.
Because at the present time there is no rapid analytical method for detection of
asbestiform fibers in water, treatment plant operators shall have to rely on indi-
cators and the use of proper operational procedures to assure the production of
water low in asbestiform fiber content, just as operators now rely on indicators
for pathogens.
Effluent Turbidity
One treatment goal that should aid in the production of water with amphibole
fibers present in low quantities is the attainment of an effluent turbidity at or
below 0.1 TU. Turbidity is not a direct measure of amphibole fiber count in
water, but Figure 24 shows that high amphibole fiber counts were more likely to
occur when effluent turbidities exceeded 0.2 TU. Filters should be backwashed
when the effluent turbidity rises to 0.2 TU. For chrysotile, however, there did not
appear to be any relationship between effluent turbidity and fiber count. High
fiber counts occurred with both high and low effluent turbidities, as shown on
Figures 24 to 29. This relationship was demonstrated for certain types of filters
and processes treating Lake Superior water. It should not be assumed to hold true
in other parts of the country until demonstrated so by experimentation.
X-Ray Analysis for Amphibole Mass
Even though an effluent turbidity goal can be stated, it is highly desirable that
some monitoring capability exists, so that an operator can verify that following
established procedures actually does cause the reduction of asbestiform fiber
count. Ultimate proof of this now is obtained by electron microscope analysis of
water samples, a costly and time-consuming technique. To assure that amphibole
fibers are being removed, a type of indicator test, X-ray diffraction analysis can be
performed. This method gives results in terms of milligrams of amphibole mass
per liter of water, and it is quite useful.
96
-------
The amphibole mass tests done at NWQL for this research were used as a guide to
planning further filtration research, because of the time delay involved in getting
back electron microscope results. In addition, the amphibole mass data provided
independent verification of the ability of filters to remove amphibole fibers.
The performance of the four filtration processes as measured by amphibole mass
analysis is presented in Table 22.
TABLE 22. SUMMARY OF NWQL RESULTS
Process
Total No. Of
Samples Submitted
No. of samples with measurable
Amphibole cone, and/or
detection limit <0.005 mg/1
No. of samples
with amphibole
cone. C0.005 mg/1
Dual Media
Mixed Media
Vacuum DE
Pressure DE
32
24
29
27
24
24
21
25
22
23
17
23
Data in this table are presented in three categories because the detection limit of
the X-ray method varied, depending in large measure on the volume of water
filtered. This in turn was a function of the suspended solids content of the water,
with higher quantities of suspended solids plugging the membrane filter sooner,
resulting in higher detection limits. Thus we are concerned not only with how
many samples were submitted for X-ray analysis, but also with the number of
samples to which a definite amphibole mass concentration could be assigned, or
which had an amphibole mass concentration < 0.005 mg/1. Finally, the number of
filtered samples having ^0.005 mg/1 amphibole mass are tabulated.
In general, each process performed reasonably well for amphibole removal.
However, the vacuum filtration process failed to achieve 10.005 mg/1 in the
filtrate in four instances vs. one or two for the other processes.
Amphibole Mass vs. Fiber Count - Filtered Water
The amphibole mass X-ray analysis was the principal analytical method relied
upon for making week-to-week operating decisions during the filtration research.
In addition, it has provided independent verification of the capability of filtration
processes to remove amphibole fibers. Finally, X-ray analysis offers the operator.
97
-------
of a municipal filtration plant the opportunity to monitor for amphibole removal
at a cost much lower than electron microscope analysis.
It can be shown statistically that X-ray analysis is useful for monitoring in place
of electron microscope testing. In analysis of data from this research, a desired
goal for amphibole fibers in filtered water was <0.04 x 106 f/1. A desired goal for
amphibole mass was <0.005 mg/1. One can ask to what extent the desired goals
were simultaneously either attained or not attained. If <0.04 x 10^ f/1 and
£0.005 mg/1 generally tend to be observed together in filtered water, then X-ray
analysis would be a valuable tool for monitoring filter effluent.
In order to do such a statistical analysis, it is necessary first to pair data from
NWQL and ORF. The NWQL X-ray data had a varying detection limit, because of
the suspended solids in water filtered for analysis. It is necessary to know if
amphibole mass exceeds 0.005 mg/1, so for samples with amphibole mass
expressed as an indefinite value below a detection limit, the data could be
considered only if the detection limit was 0.005 mg/1 or less. For example, a value
expressed as 0.01 mg/1 could be greater than or less than 0.005 mg/1, so such data
would have to be omitted in this analysis. Data are grouped according to filtration
process, not unit, arid the only runs not included in the statistical analysis are
those rejected because of the high amphibole mass detection limit.
The results are presented in Table 23.
TABLE 23. COMPARISON OF AMPHIBOLE MASS AND EM RESULTS IN
FILTERED WATER SAMPLES
Pairs of data Pairs of data with simultaneous
amphibole mass known occurrence of 0.04 x 106 f/1
to be above or and 0.005 mg/1 or
Process below 0.005 mg/1
Granular Dual Media
Granular Mixed Media
Pressure DE
Vacuum DE
23 -
23
24
21
<0.04 x 106 f/l and < 0.005 mg/1
20
23
18
14
Since the simultaneous attainment of the selected values occurred in all but three
data pairs for dual media filtration and for every data pair for mixed media
filtration, analysis was made on the basis of probability rather than by using a
98
-------
chi-square table. For each process, the hypothesis was made that there is no
relationship between EM and X-ray data. Thus the probability for simultaneous
occurrence of high or low values becomes 0.5, and the significance of the ex-
perimental results can be tested. At the 99 per cent confidence level, simultaneous
occurrences of high or low EM and X-ray values was significantly greater than one
would expect for random events for dual media, mixed media, and pressure DE
filtration. The difference was not statistically greater for vacuum DE data.
On the basis of the analysis of the probability of the observed events, it can be
stated that for dual and mixed media granular filtration and for pressure DE fil-
tration, the X-ray amphibole mass analysis can be used to indicate if the
amphibole fiber count of the filtered water would be equal to or less than a
count of 0.04 x 106 f/1 by ORF. Although amphibole mass should not be used as
an indicator for chrysotile, the X-ray analysis is faster and cheaper than EM
analysis and it is a useful indicator method for monitoring water for amphibole
fibers.
Suspended Solids Data - NWQL
The difference in performance between the vacuum filtration process and the
other processes can be related in part to the inability of the vacuum filter unit to
prevent the passage of DE through the septum. In the suspended solids data
reported by NWQL, there is often a footnote beside the BIF (vacuum DE) data
stating that the solids were mostly DE. This judgment was based upon visual and
optical microscope observations.
It can be demonstrated statistically, by comparing mean filtered suspended solids,
that the vacuum filtration process was not as effective for removing suspended
solids. For purposes of this comparison, data are grouped by process, so that dual
media filtration includes data from MM-1 and MM-2. In order to assess the
capabilities of the processes, early runs involving less successful techniques were
excluded. Thus in dual media filtration data, runs involving alum and anionic
polymer and all ferric chloride runs are deleted. In the mixed media data, a run
deliberately sampled after turbidity breakthrough was deleted. All single precoat
step runs and all runs involving conditioning with 575 C polymer were deleted
from ERD and BIF data. The results are presented in Table 24.
99
-------
TABLE 24. SUSPENDED SOLIDS IN FILTRATE
Process
Granular Dual Media
Granular Mixed Media
Pressure DE
Vacuum DE
No. of
Samples
28
23
20
20
Mean
Suspended
Solids Standard
mg/1 Deviation
0.084 0.066
0.067 0.040
0.079 0.146
0.408 0.190
Significantly different
from BIF @ 99%
Confidence Level
Yes
Yes
Yes
It is evident that the mean suspended solids from the granular filtration and the
pressure DE processes were considerably lower than the solids from the vacuum
DE filter, and these differences were statistically significant. This is additional
evidence that the vacuum filtration process experienced operating problems
throughout the testing period, at least in part because of bubble formation in the
DE, a problem brought about by low water temperatures and the saturation of
the raw water with dissolved gases.
Efforts to Develop Rapid Detection Methods
A limited effort to learn about rapid detection of asbestiform fibers was made in
this research, but the principal objective was to learn about fiber removal by fil-
tration. Other efforts to develop a rapid fiber detection method are underway,
sponsored by EPA and other Federal agencies. One method being investigated
involves placing a water sample in a laser light beam measuring scattered light
from incident angles of about 10° and 135° and relating variations of light
intensity and incidence angle to the types of particles present in water. One of
the goals of these efforts is to provide a technique that is practical for monitoring
both amphibole and chrysotile asbestiform fibers in water at filtration plants.
Such a technique should be rapid enough to permit a plant operator to make
meaningful changes in the treatment process in order to hold fiber content of the
filtered water to a minimum. Until a more rapid method is available, water filtra-
tion plants on Lake Superior should use the X-ray diffraction method, along with
occasional EM analyses.
100
-------
FUTURE RESEARCH
Information developed in this pilot plant research permits a number of questions
to be formulated for future study. Among the ideas that could be investigated are
the following:
1. Ways to improve chrysotile removal by anionic polymer conditioning of DE
or by the use of three-stage mixing and combinations of three conditioning
chemicals in granular filtration.
2. Effect of high algal counts on filter performance.
3. Fiber removal during times of highest amphibole mass and fiber count.
4. Verification of POPO optimization.
5. Additional filtration experiments at 5 to 6 gpm/ft^ with granular filters.
6. Effect of mixing intensity on filtration, and a comparison of back-mixing vs.
in-line mixing.
7. Further laboratory development, followed by pilot plant tests, of an
operator's method for monitoring the presence of asbestiform fibers in water.
A number of the suggestions for future work represent an extension of past work
into promising study areas. Additional research is needed to increase the
knowledge of the water treatment profession on the topic of asbestiform fiber
removal by filtration.
101
-------
SECTION VIII
REFERENCES
1. American Public Health Association. Standard Methods for the Examination
of Water and Wastewater, Thirteenth Edition. New York, American Public
Health Association, 1971. 874 p.
2. Gusmer, J. H. Asbestos-Containing Filter Materials. Filter Materials, Inc.
(Presented at the Joint Filtration Symposium of the 78th National Meeting
American Institute of Chemical Engineers. Salt Lake City. August 19, 1974.)
12 p.
3. Chatfield, E. J. Quantitative Analysis of Asbestos Minerals in Air and Water.
Ontario Research Foundation. (Presented at the 32nd Annual Process
Electron Microscopy Society of America. St. Louis 1974.) 2 p.
4. Cook, P.M. Semi-Quantitative Determination of Asbestiform Amphibole
Mineral Concentrations in Western Lake Superior Water Samples. Environ-
mental Protection Agency. (Presented at the 23rd Annual Denver Conference
on Application of X-ray Analysis. Denver. August 10, 1974.) 11 p.
5. Cook. P.M., G. E. Glass, and J. H. Tucker. Asbestiform Amphibole Minerals
Detection and Measurement of High Concentrations in Municipal Watei
Supplies. Science. 185: pp 853-855, September, 1974.
6. Fairless, B. Laboratories That Have Demonstrated An Ability To Analyzi
Water Samples For Asbestiform Fibers. Environmental Protection Agency
Chicago, 111. Memorandum. February 24, 1975.
7. Baumann, E. R. Diatomite Filtration of Potable Water In: Water Quality an
Treatment, The American Water Works Association, Inc. New York,
Hill Book Company, 1971. p. 280-294.
102
-------
SECTION IX
GLOSSARY
alum aluminum sulfate
B.W. backwash
BDL below detectable limits
BIF vacuum diatomaceous earth filtration unit
cm centimeter
m^ cubic meter
3 }
rn^/m^ day cubic meters per square meter day
°C - degree Centigrade
°F degree Fahrenheit
DE diatomaceous earth
DO dissolved oxygen
S/1000 gal - dollars per 1000 gallons of treated water
EPA - Environmental Protection Agency, U. S.
ERD pressure diatomaceous earth filtration units
ERD-1 - standby ERD unit
ERD-2 - main ERD unit
FeCl^ ferric chloride
f/1 fibers per liter
ft - foot
gal gallon
gpm gallons per minute
^
gpm/ftz gallons per minute per square foot
g - gram
g/g DE grams per gram of diatomaceous earth
HL - head loss
hp horsepower
in inch
kg - kilogram
103
-------
kg/hr kilograms per hour
^
kg/mz - kilograms per square meter
m meter
Mm micrometer
mph miles per hour
mg milligram
mg/1 milligrams per liter
ml milliliter
MGD million gallons per day
MM-1 - granular media filtration unit designated No. 1
MM-2 - granular media filtration unit designated No. 2
min minute
NWQL - National Water Quality Laboratory
ORF - Ontario Research Foundation
Ib pound
Ib/hr pounds per hour
^
lb/ftz pounds per square foot
psi pounds per square inch
lb/1000 gal - pounds per 1000 gallons
rpm revolutions per minute
soda ash partly purified sodium carbonate
9
cmz/g square centimeters per gram
ft square foot
9
m square meter
SS suspended solids
TU turbidity unit
UMD University of Minnesota at Duluth, School of Medicine
w/o without
yr - year
104
-------
SECTION X
INDEX OF APPENDICES
A-l Weather and Lake Level Data, Duluth Lakewood Pumping Station,
1974.
A-2 Raw Water Pumping Schedule, Duluth Lakewood Pumping Station,
1974.
A-3 Raw and Effluent Water Sample Parameters, Duluth Lakewood Pumping
Station, 1974.
A-4 Dissolved Oxygen Data, Duluth Lakewood Pumping Station Raw and
Finished Water, 1974.
A-5 Bacteriological Data, Duluth Lakewood Pumping Station Raw and
Finished Water, 1974.
B-l Equipment Design and Installation
B-2 Operation of Pilot Plant Filters
C-l Granular Media Filtration (MM-1) Summary of Individual Runs
C-2 Granular Media Filtration (MM-2) Summary of Individual Runs
C-3 Vacuum Diatomaceous Earth Filtration (BIF) Summary of Individual
Runs
C-4 Pressure Diatomaceous Earth Filtration (ERD) Summary of Individual
Runs
D-l Granular Media Filtration (MM-1) Head Loss and Turbidity Curves
D-2 Granular Media Filtration (MM-2) Head Loss and Turbidity Curves
D-3 Vacuum Diatomaceous Earth Filtration (BIF) Head Loss and Turbidity
Curves
D-4 Pressure Diatomaceous Earth Filtration (ERD) Head Loss and Turbidity
Curves
E Ontario Research Foundation Electron Microscope Analysis Results .of
Pilot Water Treatment Units, Raw Water From Duluth Lakewood Intake
F EPA National Water Quality Laboratory X-Ray Diffraction Analysis
Results of Pilot Water Treatment Units, Raw Water From Duluth
Lakewood Intake
G University of Minnesota at Duluth Electron Microscope Analysis Results
of Pilot Water Treatment Units, Raw Water From Duluth Lakewood
Intake
105
-------
H Comparison of Turbidimeters, Duluth Lakewood Pumping Station,
1974.
Diatomaceous Earth Filtration Optimization
106
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-670/2-75-050a
2.
3. RECIPIENT'S \CCESSIOWNO.
4. TITLE AND SUBTITLE
DIRECT FILTRATION OF LAKE SUPERIOR WATER FOR
ASBESTIFORM FIBER REMOVAL
5. REPORT DATE
June 1975; Issuing Date
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT NO.
Black § Veatch, Consulting Engineers
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Black £ Veatch, Consulting Engineers
1500 Meadow Lake Parkway
Kansas City, Missouri 64114
10. PROGRAM ELEMENT NO.
1CB047; ROAP 21AQB; Task 024
11. CONTRACT/OBMflXT NO.
DACW 37-74-C-0079
IAG #EPA-IAG-D4-0388
12. SPONSORING AGENCY NAME AND ADDRESS
National Environmental Research Center
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
This work conducted through interagency agreement between EPA Region V and the Corps
of Engineers, St. Paul District. See also EPA-670/2-75-OSOb, c, d, e, f, and g.
16. ABSTRACT
Pilot plant research conducted in 1974 at Duluth, Minnesota, demonstrated that
asbestiform fiber counts in Lake Superior water could be effectively reduced by
municipal filtration plants. During the study, engineering data were also obtained
for making cost estimates for construction and operation and both granular and
diatomaceous earth (DE) media filtration plants ranging in size from 0.03 to 30 mgd.
Both dual and mixed-media granular filters using alum and nonionic polymer, employing
flash mix and flocculation without settling and DE filters with alum coated DE as
precoat and/or body feed or with Catfloc B added to raw water, produced effluents
with amphibole fiber counts below electron microscope detection limits. Turbidity
was not a direct measure of fiber count, but amphibole counts were generally lowest
at effluent turbidities £0.1 TU. Chrysotile removal was more difficult, but mixed
media granular filtration with alum and nonionic polymer, and DE filtration with
anionic polymer conditioned DE frequently reduced chrysotile fiber counts markedly.
Systems for economic reasons recommended for consideration during design studies are:
(1) mixed media direct filtration, 5 gpm/ft2, multiple-stage flash mix; (2) dual media
filtration, 4 gpm/ft2, single stage flash mix; and (3) pressure DE filtration, 1 gpm/
ft2, alum conditioning of precoats and body feed, or alum conditioning of precoat
only, and cationic polymer fed to raw water.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COS AT I Field/Group
Asbestos
Amphiboles
Serpentine
Water supply
Filtration
Water treatment
Pilot plants
Mixed media filtration
Diatomaceous earth fil-
tration
Asbestiform
Chrysotile
Fiber removal
Duluth (Minnesota)
Lake Superior
13B
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This Report)
UNCLASSIFIED
21. NO. OF PAGES
119
20. SECURITY CLASS (Thispage)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
107
U. S. GOVERNMENT PRINTING OFFICE: 1975-657-59V539't Region No. 5-1 I
------- |