FLUORIDE REDUCTION
in
COMMUNITY WATER
SUPPLIES
.Je:
prepared for the •.
.
STATE OF SOUTH CAROLINA
DEPARTMENT OF HEALTH & ENVIRONMENTAL CONTROL
WATER SUPPLY DIVISION
» JOINT VENTURE
J.E.SIRRINE COMPANY & AWARE, INC.
Summervilli, South Carolina Brentwood, TIMISSII
volume ONE
-------
FLUORIDE REDUCTION
IN
COMMUNITY WATER SUPPLIES
Prepared For
STATE OF SOUTH CAROLINA
DEPARTMENT OF HEALTH AND ENVIRONMENTAL CONTROL
WATER SUPPLY DIVISION
A Joint Venture Of
J. E. SIRRIME COMPANY and AWARE, INC.
Summerville, S. C. Brentwood, TN
VOLUME ONE
-------
ABSTRACT
Forty-three community water supplies located along the east coast of South
Carolina contain fluoride concentrations which exceed the limit estab-
lished by law. During the course of this study, each high fluoride sup-
ply was evaluated in an effort to identify one or more viable alternatives
that might be implemented to achieve compliance.
The primary option proposed for the majority of systems was blending.
By constructing shallow wells that tap a low-fluoride aquifer and mix-
ing their yield with that of the existing deep wells, fluoride concen-
trations can be attenuated to acceptable levels in the combined supplies.
For blending to be successful, twenty—five coninunities would be required
o drill 204 new wells. The total estimated capital cost of blending is
$15,305,000.
The remaining communities can achieve compliance by implementing one of
four other alternatives. Those alternatives and their attendant capital
costs are as follows: seven would replace existing wells with new ones
at a cost of $385,000, eight would abandon existing supplies and purchase
water from an adjacent utility at a cost of $200,000, two would employ
activated alumina treatment to reduce fluoride at a cost of $10,472,000,
nd the remaining one would construct a separate distribution system to
.erve exempt users exclusively at a cost of $230,000.
iiplementation of the primary alternative by all forty-three communities
)uld require a state-wide capital expenditure of $26,592,000.
1
-------
ACKNOWLEDGEMENTS
This report represents the combined efforts of a great many individuals.
It would be impractical to express appreciation to every contributor;
however, the authors wish to acknowledge the assistance of Bob Heater of
Heater Well Company, Larry West of the S. C. Water Resources Commission,
and Al Zack of U.S. Geological Survey in providing groundwater data.
The authors also wish to express their gratitude to the various equip-
ment vendors, especially Hydronautics Water Systems and UOP Fluid Sys-
tems Division, for the invaluable information that they provided on
water treatment processes . A special thanks is due the personnel that
manage and/or operate the water systems included in this report for the
information and advice that they offered. Two of those managers, Ron
Bycroft of Mt. Pleasant and Robert Winfield of Conway, made significant
contributions to this study by providing accommodations and facilities
for the bench scale testing that was conducted. Another noteworthy con-
tributor to the testing effort was Culligan Water Conditioning, Inc. who
provided the equipment. Finally the authors thank Fred Soland of the
South Carolina Department of Health and Environmental Control for his
valuable guidance and suggestions.
11
-------
TABLE OF CONTENTS
CHAPTER
1 Fluoride Reduction in South Carolina
Introduction
PAGE
1—1
1— 1
• . 2—1
• . 2—1
• . 2- 1
• . 2-4
• • 2-4
• . 2-6
• • 2-6
• . 2-7
• . 2-7
• . 2-7
• . 2-10
• . 2-13
2-1 3
Proposed Fluoride Reduction Methods
1-2
Blending
1-2
Abandonment
1-3
Treatment
1-5
Unique Solutions
1—6
Regional System
1-6
Fluoride Reduction
Cost
1-9
Summary
1-10
a.
2 Treatment for Fluoride Reduction
Incidental Fluoride Removal
Lime Softening
Activated Carbon Adsorption .
Ion Exchange Using Synthetic Resins
Ferric Salts
Zeolite
Fluoride Removal Using Bone-Related Medi
Bone
Bone Char
Synthetic Bone
Sorption Using Aluminum Compounds . .
Aluminum Sulfate
Ill
-------
TABLE OF CONTENTS (cont.)
CHAPTER PAGE
2 Activated Alumina 2-18
Demineralization by Reverse Osmosis . . 2-28
Configurations of Membranes 2-28
Prevention of Fouling and Pretreatment Requirements 2-31
Bench-Scale Testing 2-34
Installations Employing RO . . . 2-35
Summary 2-38
3 Activated Alumina Pilot Studies . . . 3-1
Experimental Procedures . . . 3-1
Experimental Results 3-3
Discussion of Results • • 3—8
4 Basis of Design and Treatment Costs for Activated
Alumina 4-1
Basis of Design 4-1
Activated Alumina - Cost Estimates . 4-9
Process Design - Regional Water Treatment Facility . 4-12
iv
-------
FIGURES
FIGURE PAGE
2-1 Effect of Influent-Water Composition on Fluoride
Exchange Capacity Utilizing Amberlite XE—75 2-5
2-2 Relation of Exchange Capacity and Fluoride Ion
Concentration in Influent Utilizing Activated Alumina . . 2-21
2—3 Relation of Exchange Capacity and Alkalinity of
Influent Utilizing Activated Alumina 2-22
2-4 Relation of Exchange Capacity and pH of Influent
Utilizing Activated Alumina 2-23
2-5 RO Spiral Wound Module Configuration 2—30
2-6 RD Hollow Fine Fiber Module Configuration 2—32
3—1 Mt. Pleasant Pilot Plant Run 3-6
3—2 Conway Pilot Run 3-7
4-1 Activated Alumina System - Capital Cost 4-10
4—2 Activated Alumina System - Annual 0 & M Cost 4—11
4-3 Process Flow Schematic - Pee-Dee River Regional Facility 4-13
V
-------
TABLES
TABLE
1-1 Regional Water System Data
1-2 Summary of Fluoride Reduction Alternatives
2—1 Fluoride Reduction at Three Lime-Softening Plants
2-2 Bench-Scale Study of Aluminum Sulfate
2-3 Removal of Fluorides from Salts in Water
p1 1 Values by ALUM FLOC
2—4 EPA Comparative Costs
3—1 Mt. Pleasant Fluoride Removal Pilot Test
Summary
3—2 Mt. Pleasant Fluoride Removal Pilot Test
Summary
3-3 Conway Fluoride Removal Pilot Test Run I
3-4 Conway Fluoride Removal Pilot Test Run 2
4—1 Process Design Summary (100 GPM System).
4-2 Process Design Summary (250 6PM System).
4-3 Process Design Summary (500 6PM System).
Process Design Summary (1,000 GPM System)
Basic Column Information
Process Design Summary
At Various
Run I Data
Run 2 Data
Data Summary
Data Summary
PAGE
1-8
1—1 1
2—3
2-15
2—17
2-39
3-4
3-4
3-5
3-5
4-2
4-3
4-4
4-5
4-8
4-14
4-4
4-5
4-6
V . ’
-------
CHAPTER 1
FLUORIDE REDUCTION
IN SOUTH CAROLINA
INTRODUCTION
As of the 24th of June, 1977, water supplies throughout the United States
were required to comply with the Environmental Protection Agency (EPA)
National Interim Primary Drinking Water Standards. Those standards es-
tablished a maximum contaminant level (MCL) for fluoride and nine other
inorganic chemicals. Enforcement responsibility for the standards was
requested by, and subsequently granted to, the South Carolina Department
of Health and Environmental Control (SC DHEC). As an initial step toward
winging South Carolina water supplies into compliance with the law, the
State authorized this study of fluoride reduction alternatives. During
the course of the study, which was funded by EPA and conducted under the
auspices of SC DHEC, forty-three community water supplies were evaluated
and conceptual solutions for reducing fluoride concentrations were form-
ulated for each of the affected systems. For systems where treatment of
an existing or alternate water source was indicated, desk-top system de-
signs were developed; where fluoride attenuation by blending the exist-
ing supply with the yield from proposed shallow wells was indicated,
designs were based upon estimates of water quality and quantity obtained
from well drillers, the South Carolina Department of Water Resources,
arid United States Geological Survey personnel. Planning-level cost esti-
mates were prepared in 1980 dollars for all alternatives and served as a
‘sis for ranking the various solutions in their order of viability.
1— I
-------
Individual reports documenting the results of the study, as applicable
to a specific system, were prepared and transmitted to the respective
purveyor responsible for each supply determined to be in non-compliance.
Copies of those reports are contained in Volume Two of this document.
PROPOSED FLUORIDE REDUCTION METHODS
All high fluoride water utilized as a couniunity source of supply in South
Carolina is drawn from wells. Reduction of the fluoride concentration in
the yield from those wells can be achieved in one of three ways. The
water can be treated to remove a portion of the fluoride, it can be
blended with a low fluoride source, or the source can be abandoned in
favor of a more acceptable supply. Depending upon a host of site-spe-
cific conditions, variations of all three basic fluoride reduction methods
can be implemented by the various purveyors to bring their respective
systems into compliance.
Bi ending
Blending low fluoride water with existing supplies to achieve an accept-
able mix was determined to be a viable option for twenty-seven systems.
Of those, blending ranks • s the least expensive solution for twenty-five.
Implementation of the primary alternative in all systems would require the
construction of 204 new shallow blending wells having a combined capacity
of 11,573 gallon per minute (GPM).
The cost of designing and constructing the blending facilities was esti-
mated at $15,305,000. The blended supply would be distributed to 29,465
consumers (approximately 100,00 people).
1—2
-------
Water Quantity . The scope of this study did not include the construc-
tion of test wells; consequently, availability and capacity were estima-
ted from existing information obtained from private contractors and
governmental agencies. Some capacity variation will occur in all wells
constructed in response to the information provided in this report.
However, the projected yield should be attainable in most instances.
Water Quality . Blending shallow well water with existing supplies will
create iron-related aesthetic problems in most communities. The sever-
ity of those problems will vary with the actual iron concentrations that
are encountered and with the blending ratios required to reduce the
fluoride content of the combined supply to acceptable levels.
Lacking accurate quality data, the assumption was made that sequester-
ing would provide a relatively inexpensive means of controlling red water
in most systems. Iron removal was included in the blending scenario for
three communities where the success of sequestering is doubtful. Those
communities are the City of Conway, Bulls Bay Rural Community t later
District, and North Myrtle Beach.
Shallow wells are generally more susceptible to bacterial contamination
than are deep wells. Consequently, gas chlorination equipment and con-
tact tanks were included in the estimated cost of several specific blend-
ing alternatives.
Abandonment
Abandonment of existing wells in favor of another source of supply was
determined to be a viable option for twenty-five communities. Purchas-
1-3
-------
ing water from another utility and constructing new wells were the two
variations of abandonment found to be feasible.
Purchase . Buying water from an adjacent utility was presented as a pri-
mary option for eight systems. Of those, five would be able to obtain
a low-fluoride supply, the remaining three would tie to a larger system
that is in noncompliance. The rationale behind connecting to a larger
non-complying utility lies in the economy-of-scale that would be realized
by the large purveyors in reducing fluoride as opposed to the small sys-
tem operator. For example, Crystal Lakes Mobile Home Park would incur
an annual per consumer cost increase of $540.00 by installing an acti-
vated alumina treatment system on its existing supply. However, by pur-
chasing water from the City of 4yrtle Beach, which would also employ
activated alumina treatment, Crystal Lakes could reduce its annual per
consumer cost increase to $133.11. Comparison of the two treatment
alternatives in terms of cost per unit of installed capacity verifies
the economy-of—scale to which the significant decrease in consumer cost is
attributed. That comparison is as follows:
Crystal Lakes Mobile Home Park
• Installed A.A. system capacity 125 GPM
• Annual cost increase $108,000
• Annual cost increase per GPM of installed capacity $864.00
City of Myrtle Beach
• Installed A.A. treatment capacity 9460 GPM
• Annual cost increase $2,222,800
• Annual cost increase per GPM of installed capacity $234.00
1-4
-------
Implementation of the primary alternative in all eight systems would re-
sult in the abandonment of 1127 GPM of water supply capacity. The
service to 1378 consumers, approximately 4,800 people, would shift to a
larger municipal supply.
Construction of New Wells was indicated as the primary option for seven
communities. As was the case with blending wells, quality and quantity
were estimated from existing information. Therefore, actual conditions
encountered will vary somewhat from results predicted in this study
utilizing assumed data.
The new wells, if constructed, will replace 775 GPM of existing capacity
which will be abandoned. The replacement facilities will serve 305
consumers, approximately 1,100 people.
Treatment
Initially, two methods of water treatment were identified as possible
fluoride reduction alternatives. One was reverse-osmosis, the other was
an ion-exchange process utilizing activated alumina as the exchange med-
ium. Further efforts, which included a complete review of available
fluoride treatment literature and two limited bench-scale tests that
were conducted as part of this study, resulted in the selection of acti-
vated alumina as the most viable process for fluoride reduction treat-
ment.
It was noted that both treatment processes have demonstrated technical
feasibility in many instances. However, full scale fluoride reduction
is a relatively new area of endeavor for municipal water purveyors.
1-5
-------
Therefore, the reader is cautioned that any design of full scale treat-
ment facilities should be preceded with adequate bench scale and pilot
plant testing of each source of supply.
Activated alumina treatment was determined to be a possible solution
for twenty communities. Of those, only two, the City of Myrtle Beach
and the Town of Edista Beach, exhibit existing situations that favor the
selection of treatment as the primary fluoride reduction alternative.
Unique Solutions
Of the forty-three communities evaluated, only the Town of Kingstree pre-
sented a situation requiring a unique solution. The primary alternative
for the community proposes the construction of a separate distribution
system which would serve exempt users exclusively. Five major consumers
two schools, two industries, and a hospital are located in close
proximity to the only municipal well that contains excessive concentra-
tions of fluoride. By constructing a mini-system to serve the exempt
users from the high fluoride well, Kingstree can achieve compliance with-
out constructing treatment facilities or shallow blending wells.
Regjonal System
In addition to the individual alternatives that were presented for each
community, a regional water system concept was developed for Horry and
Georgetowr Counties. The source of supply for the system would be the
Great Pee Dee River. A 35 MGO conventional surface water treatment facil-
ity would be located on the northeast side of the river along U.S. High-
way 701.
1-6
-------
Transmission mains varying in diameter from 14 to 60 inches would convey
the water in a westerly direction as far as Conway, in a southerly direc-
tiori as far as Pawley’s Island, and in a northerly direction as far as
North Myrtle Beach.
Management and operation of the proposed system would be effected under
a joint agreement of all political subdivisions involved.
Each community system served would purchase water on a bulk basis for re-
sale to its consumers.
The capital cost of the regional system has been estimated at $85,000,000.
Assuming that the system operated during calendar 1980, annualized costs
would total $11,695,000. The bulk water purchase rate for that period was
calculated to be $2.95 per 1000 gallons. That computation was made as
follows:
• Debt Service at 12% for 30 Years $10,551,900
• Estimated Operating Cost 1,143,100
Annualized Cost $11,695,000
• Current Annual Demand (in thousands)
(10,860)(365) = 3,963,900
• Cost Per 1000 Gallons
S11,695,000 = $2.95
3,963,900
Design, securing permits and approvals, solicitation of proposals, con-
tract negotiation and award, and construction of the proposed regional
system can be accomplished within 60 months of completion of required
referendums, rate structure studies, funding procurement, etc.
1—7
-------
TABLE 1-1
REGIONAL WATER SYSTEM DATA
COMMUNITY
North Myrtle Beach
Briarci iffe
Myrtle Beach
Myrtle Beach Air
Force Base
Crystal Lakes MHP
GSW&SA Socastee
Area
GSW&SA Garden City
Area
Forest Acres Trailer
Park
Platt Water Company
Surfside Beach
Garden City
Pawley’s Island!
Murrell’s Inlet
Wagon Wheel Farms
Inlet Oaks Trailer
Park
Conway & Conway Rural
Mike Williamson MHP
Oakey Swamp Trailer
Park
Bucksport Water Corn-
pa fly
CURRENT DEMAND (MGD)
1.571
0. 120
5.700
0.173
0.032
0.338
0.135
0.003
0.085
0.336
0.402
0.500
0.008
0.020
1.330
0.008
0.008
0.061
10. 860
1990 DEMAND (MGD)
5. 100
0.309
13. 330
0.311
0.058
3. 300
0.385
0.007
0.323
1 . 400
2.160
3.820
0.018
0.036
4.000
0.014
0.012
0.220
34.803
1-8
-------
A summary of the community data utilized in developing the regional water
system concept is tabulated in Table 1-1.
FLUORIDE REDUCTION COST
The financial information presented in this report is based upon 1979
consumer levels and 1980 planning-level cost data. The annual costs pre-
sented do not represent the total cost of any alternative . They do repre-
sent the increase that would be incurred if the proposed facilities were
constructed and became operational during calendar 1980. For example,
blending alternatives include the estimated annualized capital cost and the
estimated operating cost for all proposed facilities. They do not include
similar costs for existing facilities which will remain in operation. Con-
sequently, the annual cost data presented in this document represents the
increase that will result from fluoride reduction, not the total operating
cost that will be borne by the community.
Another noteworthy item is that all capital expenses were assumed to be
financed completely with long term loans repayable over a 30-year period
at a 12% rate of interest. The assumption was made primarily to place the
alternatives on an equal basis for comparison. Secondarily, evaluation of
the fiscal position of each purveyor included in the study was not possible
within the scope of this project. In actuality, the funding program selected
by each community will have a significant impact on the per consumer cost
increases listed throughout this writing. In most cases - grants, utiliza-
tion of existing cash reserves, and lower interest rates will act singularly
or in concert to reduce the stated incremental cost increases. For example,
at an interest rate of 12% for 30 years, annual debt service on $1,000,000
1-9
-------
would be $124,140. By reducing the interest rate to 8% for 30 years,
the annualized expense would be reduced to $80,590.
Implementation of the primary alternative by all communities included in
this study would result in a state-wide capital expenditure of $26,592,000.
That translates to approximately $177 for each of the 150,000 people who
would have the fluoride concentration of their water supply reduced as a
result of the proposed changes.
SUMMARY
The results of this study are summarized in the following table.
1-10
-------
TABLE 1-2
SUMMARY OF FLUORIDE REDUCTION ALTERNATIVES
FOR COMMUNITY WATER SYSTEMS IN SOUTH CAROLINA
_____________ — ANNUAL COST
INCREASE
COMMUNITY
L L
CI)
end I . . .
ALTER NA TI VES
b
Lj
‘ - ‘._)‘__
American Heritage MHP
1
65
44
50
0.1
6.3
97.05
Construct shallow well and blend
Aynor
Construct shallow wells and blend
Treat existing deep well supply with
tivated alumina
1
2
325
150
150
90
340
0.8
27.0
11.2
42.2
36.92
212.95
Belmont S/D
Purchase water from Summerville
1
32
N/A
35
1.6
4.4
187.50
Briarcliffe S/D
Construct shallow wells and blend
Purchase water from Myrtle Beach
Treat existing deep well supply with
activated alumina
1
2
3
417
300
N/A
300
110
35
397
1.2
70.3
38.5
13.7
4.3
49.3
35.63
178.91
210.51
Bucksport Water Company
Construct shallow wells and blend
Purchase water from Conway
Tie to regional system
Treat existing deep well supply with
activated alumina
Construct surface water treatment
facility on Pee Dee River
1
2
3
4
5
303
355
N/A
150
175
175
180
100
453
70
500
0.5
40.7
6.0
46.7
77.9
22.3
12.4
56.2
58.3
62.1
75.39
175.43
205.00
346.53
462.05
1—11
-------
TABLE 1-2
SUMMARY OF FLUORIDE REDUCTION ALTERNATIVES
FOR COMMUNITY WATER SYSTEMS IN SOUTH CAROLINA
COMMUN/ T Y
and
4LTERNA TI YES
Bull
Construct
s Bay Water Di
shallow wells
strict
with iron
treatment
facilities
and
blend
Central MHP
Construct new well and abandon
isting supply
Conway
Construct shallow wells with iron
treatment facilities and blend
Treat existing deep well supply with
activated alumina
Tie to regional system
Crystal Lakes MHP
Purchase water from Myrtle Beach
Treat existing deep well supply with
activated alumina
1
2
1
2
200
N/A
125
30
355
22.9
63.9
3.7
44.1
133.11
540.00
Edisto Beach
Treat existing deep well supply with
activated alumina
Treat abandoned brackish well
yield with reverse-osmosis
750
360
360
1,077
1,100
71.3
185.4
133.7
136.5
273.33
429.33
Forest Acres MHP
Purchase water from GSW&SA
struct new wells and abandon
existing supply
1
2
14
8
N/A
20
45
1.6
0.1
2.5
5.6
290.07
402.64
ANNUAL COST
INCREASE
3,000
1 3,930
2 3,000
22.8
290.5
2,340
2,270 207.0
79.72
281.8
124.38
3 2,777 10,427 138.7 1,294.5 364.67
1—12
-------
TABLE 1-2
SUMMARY OF FLUORIDE REDUCTION ALTERNATIVES
FOR COMMUNITY WATER SYSTEMS IN SOUTH CAROLINA
ANNU4L COST
C ) iNCREASE
COMMUNITY —
L
and
‘ I)
ALTERNA Ti VES
L
Lu
3Q
Garden City Beach
Construct shallow wells and blend
Tie to regional system
Treat existing deep well supply with
activated alumina
1
2
3
—
1,319
1,580
1,500
1,580
1,000
3,134
2,890
4.0
41.7
216.2
124.2
389.0
357.5
97.16
326.54
434.97
GCbI&SA Pawley’ s/Murrel l’s
i11 shallow wells and blend
Tie to regional system
Treat existing deep well supply with
activated alumina
1
2
3
2,555
1,810
2,653
1,810
900
2,139
2,255
2.7
28.5
208.0
111.7
265.5
279.9
44.80
115.05
190.97
GSW&SA Garden City Area
Construct shallow wells and blend
Tie to regional system
Treat existing deep well supply with
activated_alumina
—
1
2
3
800
400
267
400
310
1,058
840
1.5
14.1
62.0
38.5
131.3
104.3
49.90
181.70
207.85
GSW&SA Socastee
Construct shallow wells and blend
Treat existing deep well supply with
activated alumina
Tie to regional system
1
2
3
1,571
1,500
1,500
2,292
900
1,590
2,650
3.4
156.0
35.2
111.8
197.4
328.9
73.28
224.94
231.80
Henu i ngway
Construct shallow wells and blend 1 532 785 205 0.3 25.5 48.30
-iat existing deep well supply with 2 500 435 52.0 54.0 199.25
ivated alumina
1-13
-------
TABLE 1-2
SUMMARY OF FLUORIDE REDUCTION ALTERNATIVES
FOR COMMUNITY WATER SYSTEMS IN SOUTH CAROLINA
44’NUAL COST
INCREASE
COMMUNITY
L
-
L
and
ALTER /VA TI VES
c ) . j Q;:
ZCr
INLET OAKS MHP
Construct shallow wells and blend
Purchase water from GCW&SA
1
2
95
38
N/A
40
25
0.2
21.8
5.0
3.1
54.38
262.15
Isle of Palms Water Company
Construct shallow wells and blend
eat existing deep well supply with
activated alumina
1
2
1,439
1,000
1,000
340
734
0
72.5
42.2
82.6
29.47
107.71
James town
Install pump in existing shallow well
and blend
1
90
180
28
0.1
3.5
39.22
Ki ngs tree
Isolate well and construct mini-
system to serve exempt consumers
Construct shallow wells and blend
1
2
1,800
500
785
230
250
0
0.5
28.6
31.0
15.86
17.52
Lane
onstruct shallow well and blend
1
135
145
60
0.1
7.5
55 .93
Little River
3nstruct shallow wells and blend
irchase water from GSW&SA
neat existing deep well supply with
:tivated alumina
1
2
3
290
259
N/A
325
110
0
331
1.0
21.4
26.0
13.7
0
41.1
50.5
73.71
231.3
Loris
)nnect selected existing wells and
lend
existing low fluoride wells to
!move iron
eat existing high fluoride we
ith activated alumina
1
2
870
440
450
400
100
400
405
3.3
19.5
To
12.4
49.7
18.0f
- 7 9 .4ç
T 4
-------
TABLE 1-2
SUMMARY OF FLU RIDE REDUCTION ALTERNATIVES
FOR COMMUNITY WATER
COMMUN/ TV
and
AL TERNA TI VES
Mike Williamson MHP
o Construct new well
• Purchase water from Conway
construct shallow wells and blend
• Treat existing deep well supply with
activated alumina
SYSTEMS IN SOUTH CAROLINA
Myrtle Beach
• Treat existing deep well supply with
activated alumina
• Tie to regional system
1
2
10,000
9,460
J,257
9,395
44,665
,056.5
593.9
1,166.3
5,543.5
222.2
613.75
Myrtle Beach Air Force Base
• Construct shallow wells and blend
• Purchase water from Myrtle Beach
1
2
865
325
N/A
380
45
0.2
102.0
47.2
5.6
54.7
124.3
North Myrtle Beach
• Construct shallow wells and blend
• Treat existing deep well supply with
activated alumina
• Tie to regional system
1
2
3
4,70
4,400
4,400
3,542
5,100
4,915
12,535
32.0
461.0
166.7
633.2
610.2
1,556.1
141.5
227.9
366.5
North Tranquil Acres S/D
• Purchase water from Summerville
—
1
175
N/A
13
7.1
1.6
49.71
Oakey Swamp MI-IP
urchase water from Conway
1
20
N/A
4
2.8
0.4
162.6(
ANNUAL COST
/NC/?EA SE
1
2
40
1
2
5,200
2,460
,250
1,470 0 182.5 35.
1,650 167.2 204.8 7l.5
1—15
-------
TABLE 1-2
SUMMARY OF FLUORJDE REDUCTION ALTERNATIVES
FOR COMMUNITY WATER SYSTEMS IN SOUTH CAROLINA
COMMUN/ r Y
and
AL TERNA TI VES
ANNUAL COST
INCREA SE
p.’
H
L
c.)
L
—4
L
c. ‘-
- —‘
‘4 i
L
(r
22
Pine Ridge MHP
urchase water from regional water
ystem. Note: regional system is
n planning stage. Cost and iniple-
entation schedule not available at
Ms time.
shallow well and blend
Plantersville Water System
instruct shallow wells and blend
rchase water from Brown’s Ferry
stern
1
2
N/A
33
29
0.1
1
2
240
164.77
3.6
150
N/A
55
0
0.3
8.0
6.8
0
29.70
33.22
Platt Water Company
truct shallow wells and blend
at existing deep well supply with
:ivated alumina
to regional system
1
2
3
450
270
200
224
160
350
666
0.7
29.6
8.9
20.0
43.5
82.7
45.56
162.22
203.39
Red Hill Water System
1
2
40
28
N/A
35
45
0.1
1.3
4.3
5.6
111.13
172.93
struct new well
chase water from Brown’s Ferry
tern
Rock Bluff S/D
:ruct new well
1
21
8
45
0.1
5.5
268.24
Rose Hill Water System
;truct new well
hase water from Brown’s Ferry
I
2
32
N/A
42
35
42
0. 1
1.5
4. 3
5.2
106.07
158.95
1-16
-------
TABLE 1-2
SUMMARY OF FLUORIDE REDUCTION ALTERNATIVES
FOR COMMUNITY WATER SYSTEMS IN SOUTH CAROLINA
COMMUNITY
0/70’
AL TERNA TI VES
ANNUAL COST
iNCREASE
cr
H
L
(, J
L
(- ‘ -.
J —
LI
(r
Sangaree Sf0
Purchase water from CCPW
1
2
N/A 96 27.5
11.5 47.85
South Tranquil Acres S/D
1
100
N/A
2.0
2.9
0.3
31.00
Purchase water from Surnmerville
Town of Stuckey
1
86
200
-135
0
16.7
194.31
Lonstruct new well
Sull-ivans Island
Constructidditiona] shallow wells
and blend
Construct new deep well and treat
with activated alumina
1
2
790
500
500
18
865
0
50.7
2.2
107.4
2.83
200.00
Surfside Beach
Construct shallow wells and blend
Treat existing deep well supply with
activated alumina
Tie to regional system
1
2
3
1,485
1,167
1,170
972
700
1,115
2,632
3.4
102.8
35.0
86.9
138.4
326.8
60.78
162.40
243.63
Sycamore Acres S/D
Transfer water from adjacent system
nd blend
1
91
175
50
0
6.2
68.22
Wagon Jhee1 Farms
onStruct new well
‘urchase water from GCW&SA
40
1
2
19
45
0
5.6
140.23
N/A
4
5.4
0.5
147.93
1—17
-------
CHAPTER 2
TREATMENT
FOR FLUORIDE REDUCTION
Previous experiences with the removal of fluoride were reviewed in order to
evaluate methods which have been used to remove fluoride from drinking
water. The purpose of this review was to define the process capabilities
and limitations of demonstrated systems, and to establish the most appro-
priate technologies for the systems being evaluated on this project.
INCIDENTAL FLUORIDE REMOVAL
There are a number of processes employed in water treatment systems which
may incidentally remove fluoride. The following technologies fall into
this category and are briefly discussed with respect to their defluor-
idation capabilities:
• Removal of fluoride utilizing the lime softening process.
• Removal of fluoride using carbon adsorption.
• Removal of fluoride utilizing ion exchange resins.
• Precipitation of fluoride using ferric salts.
• Fluoride removal utilizing zeolites.
Lime Softening
Lime is best considered a viable defluoridation alternative for harder
waters having moderately low fluoride levels, between 3.0 and 4.0 mg/i. 1
2-1
-------
Laboratory Experiments . In 1934 Boruff 2 discovered that the addition of
lime to fluoride spiked waters resulted in partial fluoride removal,
reducing 5.0 mg F71 to 3.0 mg F/l. Co-precipitation of the fluoride was
found to occur when the carbonate and magnesium hardness precipitated arid
the phenolphthalein alkalinity was greater than or equal to the total
alkalinity. A detailed analysis of the mechanism of fluoride reduction
during lime softening was made by Scott. 3 Scott observed that fluoride was
adsorbed on the gelatinous magnesium hydroxide precipitate formed in the
lime/magnesium reaction. Scott also found a mathematical relation between
magnesium and fluoride removal which holds that the fluoride reduction is
approximately equal to 7 percent of the initial fluoride multiplied by the
square root of the magnesium removed. Consequently, only a high degree of
magnesium removal will result in significant reductions of fluoride.
Strict control of pH and caustic alkalinity must be maintained to secure
proper magnesium precipitation.
Installations Employing Lime Softening for Defluoridation . Early reports
of fluoride reduction by three small lime softening plants 3 showed modest
removal capabilities, as shown in Table 2-1. Culp and Stoltenberg 4 ran
experiments on a local water to determine the increase in fluoride reduc-
tion when adding sufficient magnesium in a lime softening process. Their
magnesium dosages, according to Scott’s formula, resulted in good fluoride
removals (from 3.2 mg F/l influent to less than 1 mg F/l effluent).
However, scale-up calculations indicated that the additional magnesium
requirement was excessive. Several plants in states with hard water wells,
such as Ohio, Indiana and Illinois were able to achieve modest fluoride
2- 2
-------
TABLE 2-1
FLOURIDE REDUCTION AT THREE LIME-SOFTENING PLANTS (Scott, et al )
Plant 1
Plant 2
Plant 3
Raw
Eff
Raw
Eff
Raw
Eff
Fluoride
1.5
0.7
2.1
1.6
1.7
0.95
Magnesium
46
14
23
9
42.7
16
Total Hardness
523
121
234
65
449
99
Alkalinity
337
42
270
101
373
88
pH
7.8
10.3
7.7
10.0
7.1
10.0
2-3
-------
levels economically 5 because these waters contained adequate amounts of
magnesium. The operating costs of these softening plants were much lower
than any of the more specialized processes, but the initial capital
required to build such a plant was high.
Activated Carbon Adsorption
Few studies have been done on the potential of carbon adsorption because of
the high costs associated with lowering (and subsequently raising) the pH
of the water. 6 In 1934, McKee and Johnston experimented with four differ-
ent types of carbon. One carbon type was successful in removing fluoride,
an acid-treated carbon discarded from a soda pulp industry. The carbon
removal efficiency was excellent on those waters having a pH of 3.0 or
less. During this study fluoride levels were reduced from 8.0 mg F71 to
less than 1 mg F71. Tests on water at pH values above 3 showed no fluoride
reduction.
Ion Exchange Using Synthetic Resins
The Rohm & Haas Amberlite IR-4B was first thought to be promising for the
removal of fluorides. However, later studies showed that its capacity was
entirely due to the precipitated aluminum oxide which had formed in the bed
during alum regeneration. The floc caused severe flow restriction through
the column and the process had to be abandoned. 8 Thompson and McGarvey 9
discovered that the strong basic Amberlite XE-75 (also made by Rohm & Haas)
was quite effective for removing fluorides. The efficiency of the column
was observed to be a function of the ratio of fluorides to the amount of
total anions in the raw water as seen in Figure 2-1. Maier 8 calculated that
2- 4
-------
400
350
0
U
300
‘I
i
p250
>.
I-
C.)
4
A.
4
C.)
U I
C,
z
4
150
x
U. ’
UI
0
1
U.
50
0
0.16 0.20
RATIO OF FLUORIDE TO TOTAL ANIONS
FIG. 2 -1 . EFFECT OF 1NFLUENT-WATER COMPOSITION ON
FLUORIDE EXCHANGE CAPACITY UTILIZING
AMBERLITE XE-75
0 0.04 0.08 0.12
2-5
-------
if the water being treated at Bartlett, Texas (with a fluoride-total anion
ratio of 0.0147) were treated with this resin, the salt regeneration alone
would cost about $200/mil gal (1953 dollars). He also pointed out that
most high-fluoride waters have very low fluoride—total anion ratios,
exemplified by the 0.0083 ratio for Britton, South Dakota water which
contained an average of 6.7 mg F/l. Due to practically obtainable capa-
cities ranging from 100-400 grains F1cu ft, 9 resinous ion exchange is not
economically competitive with the more fluoride-specific removal
processes. 1
Ferric Salts
Experiments by Boruff 2 were performed to determine removal of fluoride by
complexing using ferric salts. The floc produced is a ferric fluoride
complex which is slightly ionized. Applications of 34 to 85 mg/l of ferric
salts on waters containing from 1.8 to 5.0 mg F7l at pH 7.2 achieved
minimal fluoride removal. When this treatment was accompanied by 170 to
340 mg/l lime, waters containing 1.8 to 5.0 mg F7l were reduced to 1.6 and
4.7 mg F/l, respectively. Ferric salts were found to be largely ineffec-
tive for practical fluoride removal.
Zeolite
Boruff 2 determined that natural zeolite (the particular type was not spec-
ified) has a small removal capacity for fluoride, based on laboratory
experiments with a 5.0 mg Ff1 water. Using a miniature contact bed,
Boruff found that fluoride residuals of 0.6 mg F71 were obtainable after
passage of 6 liters of the stock water. After the first regeneration with
2- 6
-------
a saline solution, the minimum effluent fluoride concentration was 0.9 mg
F7l in the 4- to 6-liter fraction and 4.0 mg F71 in the 44- to 47-liter
fraction. Subsequent saline regenerations showed rapidly deteriorating
capacity. Such a limited removal capability precludes zeolite from con-
sideration as a viable treatment alternative.
FLUORIDE REMOVAL USING BONE-RELATED MEDIA
Bone and its related materials have demonstrated fluoride removal capabil-
ity. Fluoride removal efficiencies are found to be much greater than those
found with the more general treatment methods already discussed.
Bone
The use of bone as a substrate for defluoridating water was the earliest
reported methodology shown to be effective, but it imparted an unwelcome
flavor to the water. Bone is essentially a calcium phosphate, Ca 3 (P0 4 ) 8
CaCO 3 , in which the carbonates are replaced by fluoride when fluoride-
bearing waters are contacted with the media. The resulting fluorapatite
compound is later flushed with a caustic solution to remove the attached
fluoride, leaving behind a hydroxy apatite compound available f or further
fluoride sorption. Matters of economy led to the obsolescence of bone as a
defluoridation media, as the use of charred bone and synthetic tricalcium
phosphate/hydroxyapatite mixtures became widespread.’
Bone Char
Bone char is more economical and asthetically pleasing than bone. Conse-
quently, it replaced bone as a media in contact filters. The concept of
2— 7
-------
using bone char was borrowed from the sugar industry, where it had been
employed in the decolorization of syrups. Charred bone, which is an
apatite and carbon mixture,’ has proved its viability in many years of use
as the media in the Britton, South Dakota plant and five California plants.
Laboratory Experiments . Burwell, et al ’° ran field tests on 30-40 mesh
virgin bone black and obtained excellent fluoride removals on waters con-
taining 5.0 mg F/l. At an average flow rate of 20 gal/hr. 700 gal of the
water could be passed through a 10 lb bed before effluent fluoride levels
reached 1 mg F71. The authors found in a separate test that fluoride
removal efficiencies quadrupled when 900 mg/i bicarbonate was reduced (by
the addition of hydrochloric acid) to 100 mg/i bicarbonate. Trisodium
phosphate regenerant was used instead of the cormion, but difficult to
handle, caustic soda solution for easier handling in a small plant applica-
tion. This method was compared to the caustic/hydrochloric acid method.
It was found that although trisodium phosphate and sodium acid phosphate
were more expensive, their superior performance and ease of handling jus-
tified these costs. However, other reports 8 reported that this pair of
regenerant compounds reduced bed capacities by 12 percent after the ini-
tial regeneration. Regardless of the choice of regenerant, other studies
have shown that bone char is not very durable and periodic media replace-
ment is required.”
Installations Employing Bone Char . Bone char has been found to be a
successful defluoridation media in several small installations in
California. 12 The Camp Irwin plant serves a 200-family housing unit and a
50-space trailer court, treating a natural water with fluoride levels
2-8
-------
between 9 and 12 mg F7l. The plant has two separate contact units. Each
unit is capable of treating 15,000 gal/cycle at an average flow of 10 gpm.
Automatic controls enable the second unit to cut on when the first unit
becomes exhausted. Tests taken during the first 30 days’ operation showed
that the fluoride content of the treated water was reduced to 0.6 to
0.8 mg F/l. Sodium hydroxide and phosphoric acid are employed as regen—
erant chemicals. 13 Regeneration costs were high ($1.20/1,000 gal in
1955). Media loss of 2 to 4 in. annually has been a problem, along with
decreasing cycle times with each regeneration. The Camp Irwin plant
changed to an activated alumina media in July of 1959, but returned to bone
char in December of 1960.12
The oldest continously operated defluoridation plant in the United States
ran from 1948 to 1971 at Britton, South Dakota; this plant began operation
with bone char in 1953.14 The supply consisted of three wells containing
6.7 mg/l F which were pumped to an enclosed steel tank at rates varying
from 38 to 206 gpm. The contact tank contained 300 cu ft of 28-48 mesh
bone char, resting on 12 in. of gravel. Regeneration was found to be
necessary about every 450,000 gal. The media would first be backwashed
with treated water (370 gpm for approximately 18 mm) to remove the accumu-
lated sand followed by 5,000 gal of a 1 percent caustic solution. The
caustic was then removed by rinsing with raw water (7,000 gal at 150 gpm)
and applying a weak carbon dioxide solution to the bed. When the p 1 - i of the
effluent approached that of the raw water, the end of the cycle was
reached. 5 After it was demonstrated through laboratory investigations that
lowering the p 11 of the raw water could optimize the use of bone char, the
Britton plant included such a modification in its process. An existing
2-9
-------
sulfuric acid dilution system was altered to feed acid continuously to the
raw water. However, due to the solubility of bone char in acids, the pH
was maintained at levels of 8.4 to 7.1 which are above the range in which
optimal removal is accomplished. 14
Synthetic Bone
Tricalcium phosphate and hydroxyapatite can be formed by reacting phos-
phoric acid and lime. The use of this material as a fluoride removal media
stermied from the need for a cheaper alternative to natural bone or bone
char. In powder form, this synthetic bone can be added to raw fluoride-
bearing water to precipitate the insoluble apatites. The material can also
be made in the form of coarse granules for use as a contact medium in a
f liter. 1
Laboratory Experiments . Investigations conducted by Adler et al ’ 5 on
tricalcium phosphate showed promise for its defluoridation capabilities.
In column tests on a 30 mg F71 water, the synthetic bone lowered the
fluoride concentration to 0.3 mg F71 in the first 3-liter fraction.
Adler et al found that neither influent fluoride levels nor raw water pH
have any significant effect on the fluoride removal capacity of tricalcium
phosphate. Influent water hardness was found to have a slight effect
(lower amounts being optimal). Comparative tests indicated that, pound
for pound, tricalcium phosphate had roughly twice the capacity of activa-
ted alumina. Testing of several regenerants indicated that sodium hydrox-
ide followed by hydrochloric acid was most cost effective. However, this
method resulted in a loss of 2.5 to 3 percent of the tricalcium phosphate
per regeneration.
2-10
-------
Citing this loss of media, Behrman and Gustafson 16 conducted tests to study
regenerant effects on the tricalcium phosphate. A fluoride-bearing water
(5 or 10 tug F/l) was passed downward through a 10-in, depth of 20 to
40 mesh tricalcium phosphate granules, at the rate of 1 gal/sq ft/mm.
Regeneration was accomplished by passing 1 liter of sodium hydroxide solu-
tion successively through the unit at 0.61 gal/sq ft/mm. Behrman and
Gustafson found that the fluoride removal efficiency of the bed became
unacceptable in 25—35 cycles. Loss of efficiency occurred even though the
2 percent media loss per regeneration was being replaced at regular inter-
vals. The actual overall loss per cycle was far greater than the 2 percent
visible shrinkage observed by Adler at al . Careful examination of the
tricalcium phosphate showed that the bulk density had increased from the
original value of 35 lb/cu ft to about 68 lb/cu ft, and that the original
porosity of 72 percent had decreased to about 45 percent. Furthermore, the
average particle size of the spent material was appreciably greater than
that of the original. Upon examination of the spent regenerating solu-
tions, Behrman and Gustafson hypothesized that the hydrochloric acid had
dissolved certain constituents from the upper part of the bed. These
materials were being partly redeposited in the lower portion of the bed,
with a consequent increase in density and particle size and a corresponding
decrease in porosity. A modified method of tricalciurn phosphate regenera-
tion was proposed, which neutralized with carbon dioxide instead of with
hydrochloric acid. More tests were done with the same bed, increasing the
water wash to 2 liters, and substituting the hydrochloric acid with 3
liters of an aqueous solution of carbon dioxide containing 1.5 g C0 2 /l.
The rate of flow during the application of the carbon dioxide solution was
2—11
-------
increased to 1.8 gal/sq ft/m m. At the end of one hundred thirty cycles,
the following conditions were observed:
• The original volume of tricalcium phosphate had not decreased.
$ The fluoride-removal capacity had remained the same.
• There was no apparent change in the density, porosity, or parti-
cle size of the granules.
• The fluoride concentration was reduced from 10 mg F7l to
0.1 mg/l after 2,377 LV.
The city water in which these tests were conducted had a pH of 7.8 to 8.1
and an alkalinity of 110 to 115 mg/i.
Studies were also made by Goodwin and Litton, 1 7 who ran pilot plant experi-
ments on natural waters containing 5.2 mg F7l fluoride. The pilot plant
consisted of a 1.3 c i i ft bed of a tricalcium phosphate/hydroxyapatite
mixture, through which the raw water was passed at the rate of 1.5 gpm.
Goodwin and Litton achieved an effluent fluoride level of 0.42 mg F71,
averaged over 50 runs. It was found that approximately 1 lb of caustic was
required for each cubic foot of phosphate, and the authors estimated that a
phosphate replacement of 5 percent might be necessary for every three
hundred cycles of operation.
Installations Employing Synthetic Bone . Synthetic bone had its first
full-scale application in 1937 at the Climax, Colorado water plant.
Scooba, Mississippi followed in 1940. Both plants were abandoned in 1949.1
The Climax plant, which treated a 14 rug F71 water, consisted of four
tanks, each charged with 45 in. of tricalcium phosphate. The combined
capacity of the four units was a net of 300,000 gal of treated water in 24
2-12
-------
hours. One or two of the units were regenerated each day, using sodium
hydroxide and carbonic acid (carbon dioxide in solution). Loss of media
from the filter was reported to be negligible. 18
The previously described plant at Britton, South Dakota, began its opera-
tions in 1948 using Fluorex, a synthetic hydroxyapatite, as its media. The
existing water supply system at Britton consisted of three 1,000-ft deep
wells, drawing water with an exceptionally high mineral content, and a
fluoride concentration of 6.7 mg F/l) 4 Regeneration was performed every
third day with a caustic solution followed by neutralization with carbonic
acid. 19 The Fluorex media, although efficient for fluoride removal, was
subject to high losses through regeneration. These losses through attri-
tion reached 52 percent per year. Subsequently, the plant was forced to
change to a bone char media in 1953.14
SORPTION USING ALUMINUM COMPOUNDS
A variety of compounds containing aluminum have been proposed for removing
fluorides from water, including bauxite, sodium aluminate, aluminum sul-
fate, and activated alumina. Early research indicated bauxite and sodium
aluminate had minimal removal capacity. 2 Aluminum Sulfate and activated
alumina will be discussed with respect to their potential as viable full-
scale defluoridation media.
Aluminum Sulfate
Aluminum sulfate has a high fluoride absorption capacity, but would be
considered an effective treatment methodology only for certain waters.
Coagulant aids such as clays and activated silicas may increase aluminum
2-13
-------
sulfate capacity.’ Certain cations also reduce capacity. In many cases,
the reduction in capacity renders this method economically impractical.
Laboratory Studies . Regardless of potential cation interference, several
laboratory experiments have demonstrated the effectiveness of aluminum
sulfate in defluoridation. Kempf et al 20 added anhydrous alum, Al 2 (SO 4 ) 3 ,
to a city water containing 7.5 mg 171. The pH of the raw water had been
adjusted from 8.4 to 7.15. The vessel was stirred for 30 mm and allowed
to stand, samples being removed at intervals. Fluoride levels were found
to be 0.85 mg 171 after two hours, and 0.40 mg F7l one day later. Tests
were made on a separate city water with a fluoride concentration of
8.5 mg 171 to determine the effect of pH on defluoridation. The results
of varying alum doses at two pH levels are shown in Table 2-2. Fluoride
levels were measured 2.5 hr after mixing.
Boruff 2 found that a dose of 170 mg/i aluminum sulfate was necessary to
reduce fluoride concentrations from 5.0 mg F/l to 1.0 mg F/l in a stock
water. The alum was added to a vessel, mixed for 30 mm and allowed to
stand. The optimum pH range for the renioval of fluorides with aluminum
sulfate was found to lie between 6.25 and 7.0. Boruff observed that the
chemical composition of the water was not of great importance in fluoride
removal efficiencies. Boruff found that even chloride and sulfate levels
as high as 1,000 mg/l had no effect on defluoridation. Treatment of waters
containing 256 mg/l sodium silicate was found to decrease the removal capa-
bilities of aluminum sulfate.
Kempf et al 2 ’ published a report two years after their original study
describing successful fluoride removal in a pilot plant using aluminum
2-14
-------
TABLE 2-2
BENCH-SCALE STUDY OF ALUMINUM SULFATE (Kempf, et al )
Aluminum Sulfate
(mg/i)
pH =_788
(mg F /1)
pH
(mg
= 6.95
F /1)
110
6.5
5.85
220
3.75
3.25
440
1.13
0.45
2-15
-------
sulfate. The plant, attached to the mains in a school, consisted of three
mixing tanks, a settling basin, and sand filters. Fluoride levels, which
originally ranged from 7 to 10 mg F7l, were reduced to between 1.5 and
2 mg F71 with alum doses of approximately 340 mg/l.
Scott et al 3 treated stock waters of various fluoride concentrations with
aluminum sulfate, thoroughly mixing each flask and allowing it to settle.
The results of this work indicated that required alum doses exceeded those
found adequate by Boruff. 2 Scott et al lowered initial fluoride levels of
1.7, 3.0 and 6.0 mg F/l to 1.0 mg F/l by the addition of 200, 340, and
890 mg/l alum, respectively.
Later work by Boruff et a1 22 investigated a discrepancy noted between
fluoride removal in natural waters and the removal earlier demonstrated
with his laboratory—made sodium fluoride solutions, the natural waters
showing a decrease in removal efficiencies. Tap water samples were pre-
pared with 4 mg/i calcium fluoride, magnesium, fluoride, aluminum fluo-
ride, sodium fluoride, and sodium fluorosilicate, respectively. All sam-
ples were treated with 9.72 grains of Al(S0 4 ) 3 18 1 120 per gal (167 mg/l),
mechanically stirred, and allowed to settle. This study indicated that the
cation associated with the fluoride ion in water greatly affected the
degree of its removal by alum floc. The magnesium and aluminum fluoride
salts exhibited a modest removal rate; sodium fluoride salt demonstrated a
higher degree of defluoridation. The study results are summarized in
Table 2-3.
2-16
-------
TABLE 2-3
REMOVAL OF FLUORIDES FROM SALTS IN WATER AT
VARIOUS pH VALUES BY ALUM FLOC
Salt
— pH 7 • 4 a
54 a
6 a
pH
I
F
I
F
I
F
I
F
NaF
4.0
1.7
4.0
1.8
4.0
1.6
4.0
1.0
CaF 2
3.7
1.7
-—
--
4.0
3.0
4.0
2.3
MgF 2
3.9
1.5
3.9
2.2
4.0
3.5
4.0
3.3
A1F 3
3.6
1.6
3.6
2.8
4.0
3.4
4.0
2.6
Na 2 SiF 6
4.0
1.9
4.0
2.2
4.0
3.1
4.0
1.2
acoagulant added over
a period of 2-3 mm.
bCoaguiant added over
a period of 30
mm.
C 1 initial fluoride
concentration in mg/i
concentration
in mg/i; F =
final
fluoride
2—17
-------
Activated Alumina
Known chemically as garuna aluminum oxide (y.A1203), activated alumina is
manufactured both in gram quantities as an analytical reagent and in large
quantities as a desiccant. It becomes “activated” when neutral alumina is
treated with acid, cormionly sulfuric or hydrochloric. The crystal struc-
ture of activated alumina is known to contain gaps in its cation lattice
which result in local sites of negative charge. However, the availability
of positively-charged sites is also indicated by the electroneutrality of
the substance. Said sites acconinodate anion “adsorption”. Ionic adsorp-
tion of cations and anions may take place simultaneously but one path of
exchange usually dominates. Experimental data presented in 1947 indicated
that activated alumina was preferential to particular anions over others.
In this hierarchy fluoride ranks third, after hydroxide and phosphite.
Fluoride was followed by sulfite, ferrocyanate, chromate, sulfate, ferro-
cyanite, dichromate, nitrite, bromide, chloride, nitrate, et al , respec-
tively. in treatment processes, the preferred ions can be used to displace
(or regenerate) lesser-preferred ions. Activated aluminas’ high prefer-
ence for fluoride is a direct reversal of fluorides’ low selectivity with
most synthetic anion exchange resins. Since many ions are usually present
to compete for adsorption sites, activated alumina is more efficient for
fluoride uptake than any other anion-exchange media. 23
In all applications of activated alumina this preferred—ion substitution
principal holds true. If, for example, the media was manufactured with
hydrochloric acid activation, the initial exchange upon water application
would be fluoride taking the place of chloride, as long as the most-
2-18
-------
preferred hydroxides are absent. Addition of hydroxide, which is more
preferred than fluoride, is necessary to regenerate the fluoride-contain-
ing adsorbent. The hydroxide added at this point must be in dilute solu-
tion because strong caustic solution has been found to dissolve the
alumina. Finally, applying dilute acid restores the fluoride removal
capacity and readys the acidic alumina for another adsorption cycle. 23
This procedure is suriinarized as follows:
• Alumina-HC1 + NaF Alumina — HF + NaC1
• Alumina-HF + NaOH Alumina - NaOH + NaF + H 2 0
• Alumina-NaOH + 2 HC1 Alumina - HC1 + NaC1 + 1120 (23)
Maler’ states that activated alumina can be reused indefinitely, but mini-
mal replacement, 3 percent annually, of the media has been found to be
required.
Bench Scale Testing . Various laboratory experiments substantiated chem-
ists’ early claims of activated alumina’s affinity for fluoride. In 1934,
Boruff 2 reduced the fluoride concentration of 30 liters of water from 5.0
to 1.6 mg F/l by passing the liquid through 300 g of activated alumina in
a contact filter. Results of the 2 1/2-hour run indicated a bed capacity
of 119 gr F/cu ft. A total of nine such runs were made, during which
regenerants and application rates were varied. Boruff also conducted
tests on the effectiveness of adding powdered activated alumina to a
stirred reactor containing water with 5 mg F/l. The addition of as much
as 50 mg/i powdered alumina failed to reduce the fluoride concentration
below 2 mg F/l.
2-19
-------
Two years later, Fink and Lindsay 24 achieved a fluoride removal
capacity of 334 gr/cu ft in a slightly larger contact apparatus. In their
experiment, 250 gal of 5 mg F7l water was passed through the unit before
the effluent reached 1 mg F7l. Fink and Lindsay tried different regener-
ant solutions, but found that only 8 percent NaOH accomplished complete
renewal of the activated alumina.
Swope and Hess 25 reduced a 6.4 mg F71 water to 1.1 mg F/l using the same
apparatus as that described by Fink and Lindsay. Interestingly, they
achieved their highest bed capacities at their most rapid application rate
(12 gal/hr), after the bed had been regenerated with 5 percent caustic. In
addition, Swope and Hess attributed a significant reduction in water hard-
ness to contact with the activated alumina.
In their laboratory studies on activated alumina, Savinelli and Black 26
used aluminum sulfate as a regenerant. Figures 2—2, 2—3, and 2—4 show
relationships found for exchange capacity and influent fluoride concentra-
tion, influent alkalinity, and influent pH. Increasing regenerant dosages
up to 12 lb alum/cu ft resulted in rapidly increasing fluoride removal
efficiencies, approaching 2,000 gr/cu ft. Likewise, increasing regenera-
tion time from 1 hr to 5 hr increased bed capacity from 600 to 1,750 gr/cu
ft with the same regenerant dose. Concentration of the alu m regenerant was
found to have no effect on fluoride exchange capacity. No relationship was
found for influent sulfate or chloride ion concentration and fluoride
removal efficiency. Savinelli and Black determined that the most critical
parameter affecting bed capacity was the influent alkalinity. Their data
indicates that a fluoride removal efficiency of 3,400 gr/cu ft is achieva-
ble when the p H of the influent water is lowered to 5.6.
2-20
-------
2,000
I I I
0
U 4 :
1,500 —
--S
<2
(.)0
w
-
)(..1
LuL l-
500 —
3 9 12
FLUORIDE ION IN INFLUENT — mg/I
FIG. 2-2 RELATION OF EXCHANGE CAPACITY AND FLUORIDE ION
CONCENTRATION IN INFLUENT UTILIZING ACTIVATED
ALUMINA
0
I I
I
6
-------
2,000
0
1,500
1,000
x-j
wu-
500
0
ALKALINITY OF INFLUENT WATER —mg I CaCO 3
FIG. 2-3 . RELATION OF EXCHANGE CAPACITY AND
ALKALINITY OF INFLUENT UTILIZING
ACTIVATED ALUMINA
50 100 150
-------
4,000
It
r 3,200
0
L u
a
g 2,400
1,600
4
(3
‘ U
C,
2
4
x
CI ,
‘C
‘ U
0
10
pH OF INFLUENT
FIG. 24 .RE IATION OF EXCHANGE CAPACITY AND pH OF INFLLJENT
UTILIZING ACTIVATED ALUMINA
5 6 7 6 9
-------
The small bench column used by Bishop and Sansoucy 27 achieved a capacity of
1,424 gr/cu ft when reducing an influerit 10 mg F71 to 0.5 mg F71. The
capacity was 600 gr/cu ft when treating a 5 mg F71 influent to the same
level. This finding illustrated the increasing difficulty in removing
smaller quantities of fluoride. Their studies also confirm the feasibil-
ity of a fluidized activated alumina reactor. The increased surface area
of the alumina granules provides greater exchange capacity than column
applications while allowing high flow rates. Wu 28 , also experimenting
with a fluidized system, added activated alumina to stirred 2-liter reac-
tors containing fluoride solutions. Wu found that the ratio of fluoride
concentration to activated alumina dose was an important factor in removal
efficiency. He also determined that removals increase with lower fluoride
to activated alumina ratios. Fluoride uptake was also found to be most
efficient at a pH of 5 in this type of system.
Installations Employing Activated Alumina . Few full-scale fluoride
removal plants exist. Those which have incorporated activated alumina
technology have shown great improvement over the years in optimizing bed
capacities. Three decades ago, bed capacities of 4001 and 50029 gr/cu ft
were considered acceptable. Recently, capacities of over
3,000 gr/cu ft 28 ’ 30 have been reported.
Activated alumina filtration was employed at the Bartlett, Texas plant
from the time the plant opened in 1952 until its shutdown in 1977.11 The
plant was designed to reduce the fluoride concentration from 8.0 mg F71 to
an average of 1.0 mg F71 and was capable of treating 400 gpm well water. 8
The 500 cu ft of activated alumina would require regeneration after approx-
imately 450,000 gal of the fluoride-bearing water had been treated. A
2-24
-------
1 percent sodium hydroxide solution was applied countercurrently to the
bed at the rate of 235 gpm, and discarded to the sewer. After a rinse of
28,000 gal raw water, dilute sulfuric acid was passed through the media
until the effluent alkalinity was the same as the influent. The total
amount of water used per regeneration amounted to approximately
68,500 gal. In 1960, a new well was drilled at Bartlett that contained
3.0 mg F/l fluoride; this extended the average cycle length to
1.5 mu gal. However, the overall bed capacity was reduced from 700 gr F
/cu ft to 400 yr F7cu ft. 5
The Army plant at Camp Irwin, California (normally employing bone char)
temporarily used activated alumina as its filter media, but with little
success. 12 It reportedly achieved an initial bed capacity of
500 gr E/cu ft, but this dropped sharply after several regenerations.
Large amounts of rinse water were mandatory to maintain acceptable efflu-
ent fluoride levels. The effective capacity was reduced to approximately
100 gr/cu ft after 30 to 35 cycles. Lee and Haras state that the plant
study was similar to laboratory tests projecting a capacity loss rate of
about 30 gr F7cu ft per cycle for activated alumina.”
Although reports of capacity loss are unconunon, consideration should be
given to frequent citations of bed cementing and blinding. Several
authors 11 ’ 12, 27 have noted a tendency of activated alumina beds to become
clogged, thus creating shorter cycle times and highly inefficient removal.
Harmon and Kalichman 12 cite the Elsinore, California plant as suffering
from this “cementing ’ from an early point in its operation. Plant engi-
2-25
-------
neers found it necessary to remove, dry, and crush all the media in the
unit, as filter fines had aggregated to form a wet, grey plastic material.
The problem at Elisnore was remedied by increasing caustic and acid doses.
Rubel and Woosley 3 ° dismiss these failures of activated alumina to ignor-
ance and/or shortcutting by plant operators. They recently studied three
fluoride removal plants with widely different raw water characteristics
and outlined a program by which activated alumina’s routine removal capa-
bility could be rnaiitained above 2,000 gr/cu ft. Rubel and Woosley’s
recommendations for maximizing removal efficiency are reported to involve
installation and operating costs compatible with limited public budgets
and funding programs. Their optimization plan focuses mainly on six
aspects of operation:
• Treatment Mode . Rubel arid Woosley have determined that a pH of 5.5 is
the optimum pH for activated alumina to attract fluoride ions, and
advise that raw water pH be carefully held to this value. This
recommendation is paramount to process optimization. If raw water pH
goes above 6.0 or below 5.0, experiments have shown activated alumina
bed capacity to decrease to 500 gr/cu ft. Earlier breakthrough also
occurs when pH is not held at the optimum, serving as further incen-
tive for careful pH control.
• Backwash Mode . The authors recommend backwashing with raw water
prior to regeneration for two reasons:
1. to remove those suspended solids which tend to blind the filter
2. expand the bed to break up any channeling or wall effects. They
cite an adequate backwash rate of 5 to 6 nini/s, which will expand
the bed approximately 50 percent.
• Regeneration Mode . The most complete regeneration should proceed as
follows:
1. Upflow the regenerant while the bed is still expanded from back—
washi ng.
2. Follow regeneration with an upflow rinse, draining to the top of
the media bed
3. Downf low the regenerant and immediately follow with neutraliza-
t i on
2-26
-------
Regeneration, say Rubel and Woosley, employs 1 percent sodium hydrox-
ide solution flowing at 1.7 m Is, and assuming a standard bed depth of
1.5 m, each step takes 35 m m. Chemical requirements in each regener-
ation step are 27 liters of 50 percent NaOH per cubic meter of media
(0.2 gal/cu ft). The intermediate upflow rinse flows at a rate of 3.4
mm/s for 30 mm. The regenerate solution is usually an in-line
di lutipn of 50 percent NaOFI with raw water.
• Neutralization Mode . As soon as the downf low regeneration is com-
pleted, raw water with ph adjusted to 2.5 is fed downflow at the
normal treatment flow rate. As the effluent pH neutralizes down to
9.0 to 9.5, the raw water p 1 - I is adjusted up to 4.0. When the treated
water p h reaches 7.5, the final adjustment of the influent is made to
5.5 and carefully maintained.
• Initial Start-up Mode . Proper start-up procedures are critical when
the activated alumina is first put on line. Rubel and Woosley
strongly recommend that the contact vessel be half-filled with water
before any bed material is placed into it. The presence of the water
serves several important purposes. First, it can prevent cementing
of the bed because it helps to dissipate the heat generated when the
alumina becomes wet. Fines present in the bed material re separated
out by the water from the more granular component, initiating strati-
fication. Also, the water will protect the underdrain assembly from
impact by the filling of the bed with media. An initial backwashing
is also recommended to flush out all of the alumina fines. The
backwashing should be carried out for an extended period of time.
• Blending . A large portion of the run time shows an effluent fluoride
concentration far below the required level; likewise, there is a
gradual increase to beyond maximum limits of concentration as break-
through occurs. As a very cost effective method of increasing cycle
time and cutting operating expenses, blending fluoride—bearing waters
to ot*a n 1 n average acceptable concentration is not a new con-
cept. ‘ ‘ For the three plants they studied, Rubel and Woosley
found that the treated water can still be allowed to flow to storage
or distribution until its fluoride concentration reaches one and one—
half to two times the maximum allowable level. In systems without a
large reservoir in which the major volume of a treatment run can be
stored, blending can still be accomplished by staggered regenerations
in plants that have two or more treatment units.
Other recommendations for the optimization of activated alumina technology
include the hiring of fully qualified operators, proper selection of con-
struction materials, and the disposal of wastewaters in accordance with
local discharge standards. 3 °
9_9 1
-------
DEMINERALIZATION BY REVERSE OSMOSIS
The reverse osmosis process was reported to be employed in more than five
hundred plants having capacities greater than 25,000 gpd in 1977.31 The
principle underlying desalination by reverse osmosis CR0) is that fresh
water will diffuse out of a brine solution across a membrane when the
pressure applied to the brine side is greater than the osmotic pres-
sure. 32 ’ The applied pressure, the ambient water temperature,
and the permeability of the membrane are the major factors determining the
quality and quantity of the product water. As an indication of fresh water
flux across the membrane, approximately 1,000 mg/l of total dissolved
solids (lOS) is equivalent to 10 psig osmotic pressure; the volume of
desalinated effluent will be directly proportional to the difference
between the externally applied pressure and this osmotic head. 31
Configurations of Membranes
In terms of its effect on overall process performance, the membrane is the
most critical cost-determining item in an RU installation. 33 ’ Several
different membrane configurations are in common use at reverse osmosis
facilities: spiral wound, tubular, hollow fiber, and plate and frame.
Whatever the physical arrangement, the membrane must be extremely well
supported in order to withstand the high pressure drop across it. 32 There
are also a large number of available membrane materials, although not all
are readily convertible to any desired configuration of the menibrane. 33
Comercially available membranes are commonly categorized as either cellu-
losic or non-cellulosic, with the most popular cellulose acetate being
responsible for the recent strides in making RO an economically competi-
tive technology. 36 ’ 37 Reinforcing material or backing is compulsory for
2-28
-------
adequate mechanical support of the film as the semipermeable cellulose
0
acetate is often made in sheets only 1,000 A thick. This backing material
0
must consequently have pores less than 1,000 A in diameter. Certain
noncellulosic materials like DuPonUs polyamidic nylon can be formed into
hollow fibers that do not require any external support to withstand
1,000 psi or more. 33 Applicability of the different membrane materials
also depends on certain influent water quality parameters, with the cellu-
losic membrane systems being much more constrained by their limited pH
tolerance (approximately 3.5 to 7.5) than the non—cellulose forms (tolera-
ting 3 to 11). However, these latter forms which include mostly polya—
mides, are unable to withstand free chlorine. 36
Often employing cellulosic materials, the spiral wound RO element is com-
posed of two reinforced membrane sheets sandwiching a product water col-
lection tube, rolled up as illustrated in Figure 2—5. These spiral RO
elements are assembled in series to produce the required flow. This
particular configuration is known to have fairly good resistance to foul-
ing, fair im munity to plugging, good production per unit volume, and high
flux and rejection ranges. A broad span of operating pressures is attain-
able with spiral wound elements, and their costs/gal are relatively low. 36
The tubular configuration utilizes its tube wall as a pressure vessel and
the tube surface as its membrane support structure, thus combining two
functions. The membrane is usually placed on the inside of the tube, and
product water passes through to the outside. 34 Tubular systems have the
widest range of operating pressures for reverse osmosis, from 400 to
1,000 psig. They are also the easiest of all configurations to clean and
2-29
-------
RoH to assemble
Product-water-side backing
material with membrane on
each side, glued around
edges and to center tube
(after passage throug
\ ‘ - membrane)
Product- water
side back ng
M
Brine- side
FIG. 2-5. RO SPIRAL VVOUND MODULE CONFIGURATION
Br In —side
Product wate-
, -Brine flow
Product-water f
\
2-30
-------
have a high resistance to plugging and fouling. These qualities make the
tubular system the best choice for waters which contain very high levels of
suspended solids. Compared to the spiral wound, however, production per
unit volume is poor and they have a high capital cost/gal of water pro—
duced. 36
When membranes are cast in the form of hollow fibers, product water is
collected in the inside, where it flows out axially to a collection mani-
fold. This structural form may range from the size of a human hair (the
polyamide hollow fiber) to four tinies larger (cellulose acetate hollow
fibers). 34 ’ 36 Sometimes referred to as “permeators ”, the hollow fiber
configuration manifests high salt rejection, fair resistance to fouling,
and very good tolerance toward chemical cleaning agents. Their low flux is
compensated for by the extremely large surface area, which makes the gross
production per unit very high. Costs/gal of water produced are relatively
low for hollow fiber membranes. 36 The hollow fiber configuration is pre-
sented in Figure 2-6.
In plate and frame construction, membranes are mounted on both sides of
solid plates into which product water channels have been cut. These plates
alternate with brine feeding frames with the entire array housed in a
pressurized vessel. 34 This configuration, which resembles a filter press,
is not as widely employed as the other forms.
Prevention of Fouling and Pretreatment Requirements
There are two major causes of membrane fouling:
2-31
-------
Feed wafer
, 0
Product Brine
water
Brackish feed water
Product wafer collec ted in hollow fibers
Product water removed
Brine removed
Typical - OD:ID - 44 . 6 i: 24 .9p
FIG. 2-6. RO HOLLOVV FINE FIRER MODULE CONFIGURATION
0
Hollow Fibers
ED
ED
ED
ED
2—32
-------
• The presence of colloidal material and/or certain dissolved
salts in the untreated water. 38 ’ 39 (However, these materials
can be dealt with through pretreatment, which will be discussed
later.)
• The lack of maintenance of a minimum brine flow rate, a param-
eter which is coninonly reduced with the intention of increasing
product water f low. 3 1 ’ 40 (This practice, however, robs the brine
membrane surface of the turbulence necessary to prevent precipi-
tation (scaling) and solids buildup, thereby causing what is
known as concentration polarization.)
Polarization encourages dissolved ion breakthrough and deteriorating pro-
duct water quality. However, as pointed out previously, certain membrane
configurations are more inherently prone to this problem than other
forms. 4 °
Precipitates that are most apt to foul RO membranes include calcium carbon-
ate, calcium sulfate, and various iron and manganese oxides. 4 ° Silica,
humic acids, bacteria, and fine clay particles are also potential foul-
ants. 38 ’ 3 The expediency of pretreatment for RO is dependent upon the
case and effectiveness of periodic membrane flushing or cleaning, the
frequency of such cleaning, and the membrane’s resistance to cleaning
agents. The need for pretreatment is indicated when the frequency of
cleaning is unacceptable to the customer, or if the raw water contaminants
cause irreversible fouling. Typical pretreatment consists of micron fil-
tration for suspended solids removal, pH adjustment or softening, and some
times activated carbon for organic removal. 38 Controlling pH to 5.0 to 6.0
2-33
-------
will minimize the calcium carbonate scale. Calcium sulfate solubility can
be increased by sequestering the calcium with the addition of sodium hexa-
rnetaphosphate at temperatures below 96°F. Manganese levels to 1.5 mg/i and
iron levels of 4.0 mg/i can be economically controlled with very small
quantities of coninercially available chelants, dispersants and sequester-
ing agents. 4 °
Bench—Scale Testing
Pilot plant data has yielded much encouraging information concerning the
efficacy and economy of reverse osmosis plants. Cruver and Sleigh 41 con-
ducted studies at a 1,000 gpd sea water pilot plant at Sea World in’ San
Diego, California to compare single-stage and two—stage reverse osmosis
schemes. The authors admit that a single-stage system initially appeared
more cost effective, since 20 percent less equipment was required. Their
data, •however, showed that the capital cost savings would be completely
offset by the higher membrane costs. They found that their single-stage
system required a membrane with 10 times the salt rejection capability’of a
two-stage system to achieve the same product water quality, one that would
reject 99 percent of the salt. Such a membrane is certainly available, but
the price is prohibitively high. The decision by Cruver and Sleigh’ to
choose a double-pass system was also made on the basis of the increased
reliability margin with such a system.
Schmitt and Hurley 42 report that the U.S. Army is testing portable RO units
on surfacewaters to develop a transportable packaged system suitable for
emergency use. Being employed in the studies are cellulose acetate tubular
modules, rated at 10,000 gpd with 90 percent salt rejection. In tests on
2-34
-------
Potomac River water, excellent results were found with these units, except
for three instances of stoppage due to damaged parts. Separate experiments
on spiral wound units showed a greater tendency toward fouling by suspended
material. The authors admit that further study is needed before a recom-
mendation can be made on a unit suitable for field army use.
Installations Employing RO
Reverse osmosis systems have been installed to provide suitable water
supply for both domestic and industrial use. It is currently employed at
the Manassas, Virginia IBM semiconductor plant. The modules are used for
demineralization of raw water prior to being fed to a mixed bed demineral-
izer. The influent water is drawn from a municipal surface water reser-
voir, having between 95 and 180 mg/i lOS. Pretreatment for the polyamide
RO units includes acidification, cationic polymer coagulation, sand filtra
tion, and carbon adsorption. The reverse osmosis system has been operating
at the semiconductor facility since 1972 without any major problems. 43
Crabbe 44 described the success of a double-stage RO system for the produc-
tion of water for pharmaceutical use. He cites the major reason for
selecting the double—pass system as added protection from biological con-
tamination. The 25 gpm installation includes softening, activated carbon,
and a 10 micron filter as pretreatment. The influent water characteristics
include 207 mg/i bicarbonates and 330 mg/i TDS, which are reduced to 4.9
and 7.3 mg/i, respectively.
A reverse osmosis system at a utility company is used as an effective
pretreatment for the cation-anion deniineralizer train f or condensate water
makeup. Wadiington 45 reports that the installation of the three-stage RO
2—35
-------
unit (and its own pretreatment units) has resulted in substantial savings
for the utility in regenerant chemicals. The number of pressure vessels in
the three stages is staggered 7-4-3 (70 membrane modules) in the sumner and
altered 8-6-3 (185 membrane modules) in the winter to reduce increased
system pressure. Membranes were replaced at this facility after two years
of operation when water quality began to deteriorate. A similar experience
is noted by Hollier 46 in his observations on an RO demineralizer for boiler
feedwater at a Louisiana power station. Product water flux and quality had
been seen to deteriorate, with the original output of 250 gpm dropping in
2½ years to 200 gpm, and conductivities from both stages increasing approx-
imately 75 percent over that time period. The problem was found to be
caused by iron oxides which were causing the membranes to foul. This was
remedied later by acid pretreatment of the weliwater and installation of a
different media in the prefilter.
Wastewater is being reclaimed with the aid of reverse osmosis in the
Fountain Valley, California “Water Factory 21” plant. The final effluent
from this 15 mgd facility is injected into wells, to be withdrawn later for
irrigation, domestic, and industrial use. One-third of the flow passes
through the RO unit, which utilizes large spiral wound cellulose acetate
modules in a three-stage process. Six parallel pressure vessel assemblies
each contain 35 vessels arranged in a 20-10-5 array, enabling the plant to
attain 85 percent water recovery. Sodium, sulfate, chloride, electrical
conductivity and COD are all reduced approximately 95 percent. 47
Responsible for desalinating Red Sea water, the 3.2 mgd RO plant at Jeddab,
Saudia Arabia is reported by Al-Gholaikah et a1 48 to be performing well.
2-36
-------
Spiral wound polyamide membrane modules are employed in a two-stage proc .-
ess, the first-stage units running at a 30 percent recovery and the second
stage at 85 percent recovery. Pretreatment equipment for the reverse
osmosis system includes acid storage and injection pumps, hexametaphos-
phate injection pumps, and cartridge filters. Waste brine from the first
stage is discharged back into the Red Sea; reject from the second stage,
being cleaner than the raw influent, is recycled to the front of the plant.
The Yuma, Arizona desalting plant is designed to operate on a 3,200 mg/i
TDS feedwater. It will use reverse osmosis to produce a low lOS product
water that can be blended with raw water to achieve the allowed effluent
solids limits. 1 to 5 mgd will be treated by membrane desalination, with
product water salinity approximately 386 mg/i lOS. The reject brine flow
will be passed through an energy-recovery turbine and then routed to the
Santa Clara Slough in Mexico. 49
2-37
-------
SUMMARY
The review of demonstrated treatment technologies for the removal of
fluoride indicate that there are three processes which have the capability
to provide required water quality:
• Activated alumina.
• Bone char.
• Reverse osmosis.
The relative costs for the three processes have been prepared by the
EPA 5 ° and The State of Arizona 51 . The EPA comparative costs for each of
the three processes utilizing a 500 gpm system are presented in Table 2-4.
The Arizona Report indicates a total cost of 35 /1,OOO gal utilizing acti-
vated alumina, 95 /1,0OO gal utilizing bone char, and 90 /1,OOO gal
utilizing reverse osmosis for a 500 gpm system.
Based on the outlined costs, and operational considerations, activated
alumina has been chosen as the most viable technology for fluoride removal.
2-38
-------
TABLE 2-4
EPA COMPARATIVE COSTS
Process
Construction
Costs ($)
O&M
Costs ($lyr)
Activated Alumina
100,000
19,000
Bone Char
130,000
20,000
Reverse Osmosis
450,000
100,000
2—39
-------
CHAPTER 3
ACTIVATED ALUMINA PILOT STUDIES
The findings of the literature survey, conversations with operators and
engineers of existing fluoride removal facilities, and general cost sur-
veys indicated that the activated alumina process is the most viable proc-
ess for fluoride removal. Significant quantities of information are
available on the use of activated alumina for fluoride removal. In order
to confirm the application of activated alumina for fluoride removal,
pilot studies were conducted at two sites in South Carolina.
The purpose of these pilot studies was to establish the fluoride removal
capacities for a high and low fluoride water. The two pilot studies were
conducted at the communities of Mt. Pleasant and Conway. Both sites were
selected based on the fluoride content of the raw water. Mt. Pleasant’s
water supply contains a relatively high fluoride level, 4.3 mg F/l.
Conway’s wells yield a relatively low fluoride level, 2.2 mg F/l.
The methods utilized, experimental results, and conditions are presented
below.
EXPERIMENTAL PROCEDURES
Identical columns were employed for both tests. Each column had a diameter
of 4.5 in. and was charged with sufficient activated alumina to provide a
bed depth of 2 ft. The total bed volume for each column was 0.8836 Cu ft.
3- 1
-------
Each column was prepared by partially filling the vessel with water and
adding a predetermined quantity of activated alumina. The virgin column
contents were neutralized by downf lowing water with a pH of 4-5 units in
the Culligan laboratory. The pFI was controlled by the addition of muriatic
acid. When the pH of the column effluent reached the level of 8.0, the
columns were transported to the test site. The time required to reach an
effluent pH of 8.0 was 45-90 mm equivalent to 87-175 Bed Volumes (BV) at a
flow of 1.72 gpm.
The column was operated on-site at a pH of 5.5 for a period of 30 mm at the
test flow. At this time, the initial column operating conditions were
established. Each column, which had a detention time of 5 mm, was opera-
ted until fluoride breakthrough occurred. Throughout the run flow, influ-
ent fluoride, effluent fluoride and pH data was collected. Analytical
procedures utilized through the course of study conformed to Standard
Methods for the Examination of Water and Wastewater , 14th Edition, 1975.
After column exhaustion, the unit was transported to the Culligan labora-
tory for regeneration and neutralization. The column was first backwashed
in the upf low mode for a period of 30 mm at a flow of 10 gpm, equivalent to
approximately 23 gpm/sq ft. The backwash rate forced the media out of the
column into a device which allowed collection and continued backwash of the
media.
The column was recharged with the backwashed media and regenerated with
approximately 1 percent caustic. Caustic regeneration was conducted for a
period of 30 mm at a flow of 1.72 gpm. This flow is equivalent to a
loading rate of 3.9 gpm/sq ft.
3- 2
-------
The column was again backwashed (rinsed) for a period of 30 mm after
regeneration. The backwash flow was 5 gpm, equivalent to a loading rate of
11.3 gpm/sq ft. Following backwash the column was neutralized in the
downf low mode utilizing raw water with a pH of 4-5 units using the same
procedure as outlined previously.
EXPERIMENTAL RESULTS
Experimental operating data is sun iiarized in Tables 3—1 through 3—4 for
each of the four experimental pilot runs. These data have been plotted to
determine capacities for the two waters (see Figures 3—1 and 3—2). The
test results indicate that approximately 10,000 gal of Mt. Pleasant raw
water was treated without exceeding the MCI for fluoride utilizing 1 cu ft
of media. The fluoride exchange capacity at Mt. Pleasant was approximately
0.424 lb F/cu ft of activated alumina.
The Conway data indicates that 8,500 gal of raw water was treated success-
fully with 1 cu ft of media. The fluoride exchange capacity at Conway was
0.18 lb F7 cu ft of activated alumina.
3- 3
-------
TABLE 3-1
MT. PLEASANT FLUORIDE REMOVAL PILOT TEST
RUN 1 DATA SUMMARY
Cumulative
Flow
(gal) (BV)
Fluoride Concentration
inital final
mg/i mg/i
initial
pH
final
0
4.4 0.82
5.5
9.1
1,740 263
4.2 0.15
7.0
7.5
2,250 340
4.3 0.12
6.5
7.0
3,620 548
4.3 0.43
4.5
4.8
4,220 638
4.2 3.8
5.0
4.9
TABLE 3-2
MT.
PLEASANT FLUORIDE REMOVAL PILOT
RUN 2 DATA SUMMARY
TEST
Cumulative
Fluoride Concentration
Flow
inital final
pH
(gal) (By)
mg/i mg/i
initial
final
0 0
4.3 0.95
5.5
9.5
570 86
4.4 0.24
5.7
5.4
1,920 290
4.3 0.23
5.5
5.3
2,340 354
4.3 0.23
5.7
5.3
3,840 581
4.3 0.25
5.9
5.8
4,490 679
4.2 0.28
5.5
5.2
5,700 862
4.2 0.28
5.8
5.6
6,230 943
4.3 0.35
5.9
5.8
7,660 1,159
4.2 0.32
6.5
6.4
8,070 1,221
4.2 0.48
5.7
5.8
9,070 1,372
4.3 2.4
5.8
5.7
3- 4
-------
TABLE 3-3
CONWAY FLUORIDE REMOVAL PILOT TEST
RUN 1 DATA SUMMARY
Cumulative
Flow
(gal) (BV)
Fluoride
inital
mg/i
Concentration
final
mg/i
pH
initial final
0 0
2.6
0.85
5.5
9.3
1,540 233
2.2
0.15
6.0
7.9
1,670 253
2.3
0.10
4.9
6.5
3,680 557
2.0
0.11
5.5
6.0
5,380 814
2.2
0.10
6.0
5.8
5,540 838
2.1
0.20
5.5
5.7
7,260 1,098
2.3
0.91
5.5
5.6
7,420 1,123
2.3
1.40
5.5
5.7
TABLE 3-4
CONWAY FLUORIDE REMOVAL PILOT TEST
RUN 2 DATA SUMMARY
Cumulative
Flow
(gal)
(BV)
Fluoride
inital
mg/i
Concentration
final
mg/I
pH
initial final
0
0
2.3
0.43
5.5
8.3
1,390
210
2.2
0.24
5.5
5.2
1,630
247
2.0
0.25
5.5
5.1
3,450
522
2.1
0.17
5.4
5.8
3,620
548
2.2
0.24
5.6
5.7
5,290
800
2.2
0.91
7.3
7.1
5,650
855
2.2
0.60
5.6
5.8
6,920
1,047
2.2
1.80
5.7
5.7
3— 5
-------
7
0 151 302 453 604 755 906 1057 1208 1359
CUMULATIVE FLOW. BV 1000 gal
3.0
2.0
U.
1.0
0
I
a.
I-
z
w
U
1
U.
2
5
1510
FIG. 3-1 . MT. PLEASANT PILOT PLANT RUN
-------
3.0
7
1.0
0
0 151 302 453 604 755 906 1057 1208 1359
CUMULATIVE FLOW, BV 1000 gal
E
2.0
U-
0.
I-
2
-J
U.
2
5
1510
FIG. 3-2 .CONWAY PILOT RUN
-------
DISCUSSION OF RESULTS
The data indicate that influent pH control is of primary importance in the
operation of the alumina column. AWARE had difficulty controlling pH
during the initial runs at both Mt. Pleasant and Conway. The systems were
equipped with a pulse type acid feed pump. During the initial runs pH was
monitored utilizing grab samples. The pH was adjusted based on the grab
sample analysis. Due to the pulsing of acid, this mode of operation was
discontinued during the second run. During run 2 pH was monitored on a
small composite sample. As a result of this modification, the run time was
significantly greater at the Mt. Pleasant site during the second run. The
Conway experimental data is somewhat erroneous. During the second run,
acid feed was discontinued for a period of less than 24 hr. It is felt
that this loss of acid feed was responsible for not obtaining the antici-
pated throughput of 10,000 gal/cu ft.
3- 8
-------
CHAPTER 4
BASIS OF DESIGN AND TREATMENT COSTS FOR ACTIVATED ALUMINA
Activated alumina fluoride removal systems were designed for systems in
which this technology might be feasible. The process design summaries for
a 100, 250, 500, and 1,000 gpm system are presented in Tables 4-1, 4—2,
4—3, and 4-4, respectively. The basis for design and associated capital
and O&M costs for each system are described below.
BASIS OF DESIGN
Activated alumina column beds were designed based on maintenance of a 5 mm
bed detention time, 12 BV/hr. The detention time selected was based on the
experience of operating activated alumina fluoride removal facilities and
pilot studies conducted as part of this project. The column surface area
was based on a hydraulic loading rate of 9 gpm/sq ft utilizing the design
flowrate. The hydraulic loading rate of 9 gpm/sq ft was selected based on
the flow required to expand the media bed during backwash. The column
length was designed based on a 60 percent bed expansion volume.
Acid feed equipment would be required to adjust the column feed p11 and
adjust the alkaline byproducts generated during the regeneration and neu-
tralization operating modes. No information on acid requirements was
available during the investigation period. Experience of other facilities
indicated that sulfuric acid usage was equivalent to 1 tank truck load
(approximately 3,400 gal) per year for a 100 gpm plant operated 16 hr/day.
4-1
-------
TABLE 4-2
PROCESS DESIGN SUMMARY
250 gpm System
Bed Depth in Single Column, ft 6.0
Length of Single Column, ft 9.6
Column Surface Area, sq ft 28.27
Initial Activated Alumina Inventory
cu ft 169.62
lb 9,329.10
Time Between Regenerations, days 7.07
Operating Mode Downf low
Flow Rate, By/mm 0.197
gpm 250
gpm/sq ft 8.84
Backwash Mode Upflow
Backwash Rate, BV/min 0.197
gpm 250
gpm/sq ft 8.84
Backwash Duration, mm 10
Backwash Volume, gal 2,500
Regeneration Mode Upf low
Regeneration Rate, BV/min 0.06675
gpm 84.7
gpm/sq ft 3.0
Regeneration Duration, mm 35
Regeneration Volume, gal 2,541
Rinse Mode Upf low
Rinse Rate, BV/min 0.1335
gpm 169.4
gpm/sq ft 6.0
Rinse Duration, mm 5.0
Rinse Volume, gal 847
Neutralization Mode Downf low
Neutralization Rate, BV/min 0.197
gpm 250
gpni/sq ft 8.84
Neutralization Duration 240
Neutralization Volume 60,000
4-3
-------
TABLE 4-3
PROCESS DESIGN SUMMARY
500 gpm System
Bed Depth in Single Column, ft 6.0
Length of Single Column, ft 9.6
Column Surface Area, sq ft 56.75
Initial Activated Alumina Inventory
cu ft 340.5
lb 18,727.5
Time Between Regenerations, days 7.09
Operating Mode Downflow
Flow Rate, BV/min 0.196
gpm 500
gpmlsq ft 8.81
Backwash Mode Upflow
Backwash Rate, BV/min 0.196
gpm 500
gpm/sq ft 8.81
Backwash Duration, mm 10
Backwash Volume, gal 5,000
Regeneration Mode Upf low
Regeneration Rate, By/mm 0.06675
gpni 170
gpm/sq ft 3.0
Regeneration Duration, mm 35
Regeneration Volume, gal 5,950
Rinse Mode Upf low
Rinse Rate, BV/min 0.1335
gpm 340
gpm/sq ft 6.0
Rinse Duration, mm 5.0
Rinse Volume, gal 1,700
Neutralization Mode Downflow
Neutralization Rate, BV/min 0.196
gpm 500
gpm/sq ft 8.81
NeutraFizatjon Duration 240
Neutralization Volume 120,000
4-4
-------
TABLE 4-4
PROCESS DESIGN SUMMARY
1,000gpm System
Bed Depth in Single Column, ft 6.0
Length of Single Column, ft 9.6
Column Surface Area, sq ft 113.5
Initial Activated Alumina Inventory
,cuft 681
lb 37,455
Time Between Regenerations, days 7.09
Operating Mode Downf low
Flow Rate, BV/min 0.196
gpm 1,000
gpm/sq ft 8.81
Backwash Mode Upf low
Backwash Rate, BV/min 0.196
gpm 1,000
gpm/sq ft 8.81
Backwash Duration, mi i i 10
Backwash Volume, gal 10,000
Regeneration Mode Upf low
Regeneration Rate, BV/min 0.06675
gpm 340
gpm/sq ft 3.0
Regeneration Duration, mm 35
Regeneration Volume, gal 11,900
Rinse Mode Upflow
Rinse Rate, BV/mun 0.1335
gpm 680
gpm/sq ft 6.0
Rinse Duration, mm 5.0
Rinse Volume, gal 3,400
Neutralization Mode Downf low
Neutralization Rate, BV/min 0.196
gpm 1,000
gpm/sq ft 8.81
Neutralization Duration 240
Neutralization Volume 240,000
4—5
-------
The reviewed facilities did not require acid neutralization of spent
regenerant. The acid holding tanks were sized based on the volume occupied
by 40,000 lb of 92 to 96 percent H 2 S0 4 (approximately 3,500 gal).
Caustic storage was based on the use of 50 percent NaOH shipped in drums
for the 100 gpm and 250 gpm systems. The vessels were sized based on the
feed dilution of 1 percent caustic with sufficient volume for one regener-
ation cycle. The caustic storage tanks for the 500 gpm and 1,000 gpm
systems were based on bulk delivery of 50 percent caustic. Caustic would
be added in-line for dilution to 1 percent. The size of the caustic
storage tank was based on the volume of 1 tank truck of 50 percent caustic,
i.e., 4,000 gal. The volume is sufficient for 26 wk of 500 gpm system
operation and 13 wk of 1,000 gpm system operation.
Raw water storage will be required for those systems which do not have
sufficient storage to maintain pressure during bed regeneration. A survey
of the systems which do not have sufficient storage indicated that the
maximum system flow was 300 gpm. A raw water storage tank was sized and
costs were estimated based on a 250 gpm system. The storage tank volume is
70,000 gal. The developed costs are included in the total treatment costs
for those systems without storage.
A brine storage tank may be required at each site. The storage tank has
three basic functions:
• Batch neutralization of the alkaline regenerant.
• Equalization for sewered systems.
• Holding vessel for contract disposal.
Costs were developed with and without brine storage. Brine storage may not
be required in some of the larger sewered coimiunities.
4- 6
-------
The brine storage tank volume was based on the anticipated quantity of
regenerant, 52 By.
Pump capacities were sized based on the required flows to backwash, regen-
erate, rinse and neutralize the bed. Those facilities with adequate water
storage will utilize the well water pump to operate, backwash, regenerate,
rinse, and neutralize the bed. This mode will require the use of a control
valve and recycle loop to control flow.
Those systems which do not have adequate raw water storage will be supplied
with a groundwater storage tank equipped with a variable speed pump to
backwash, regenerate, rinse and neutralize the bed.
Operation of the facility requires strict pH control. Acid addition for
maintenance of a feedwater pH of 5.5 will require pH controlled acid
addition. Acid addition will be required to neutralize the bed. The
neutralization acid feed system will be isolated to allow the continued raw
water treatment in the second column. Acid addition will also be required
for regenerant pH adjustment prior to disposal. The regenerant neutrali-
zation system will utilize the bed neutralization acid feed system on a
batch basis.
The 100 and 250 gpm systems will not require in-line dilution for caustic
regeneration. Caustic feed pumps have been based on a flow of 0.67 BV/min.
The 500 gpm system and 1,000 gpm system will utilize 50 percent caustic
diluted in—line to 1 percent followed by a static mixer. The caustic feed
pump sizes are based on flows required for dilution of 50 percent caustic
to 1 percent. A summary of the basic equipment sizes and flowrates are
presented in Table 4-5.
4- 7
-------
TABLE 4-5
BASIC COLUMN INFORMATION
System
100
250
500
1,000
Media Volume, cu ft
Column Surface Area, sq
ft
71.76
11.04
169.62
28.27
340.5
56.75
681
113.5
Media Depth, ft
6.5
6.0
6.0
6.0
Column Height, ft
Backwash Flow, gpm
Regeneration Flow, gpm
Rinse Flaw, gpm
Neutralization Flow, gpm
Brine Volume, Gal
10.4
100
40
72
100
28,000
9.6
250
85
170
250
66,000
9.6
500
170
340
500
133,000
9.6
1,000
340
680
1,000
265,000
Backwash Feed Tank, Gal
NA
70,000
NA
NA
Acid Storage Tank, Gal
Caustic Storage Tank, Gal
5,000
1,500
5,000
3,000
5,000
4,000
5,000
4,000
-------
ACTIVATED ALUMINA - COST ESTIMATES
Cost estimates were developed for 100, 250, 500, and 1,000 gpm activated
alumina fluoride removal system models. Three systems were designed for
each system capacity. The first system is a basic column design. The
second system includes a brine storage or raw water storage vessel. The
third system includes both raw water and brine storage facilities. The
cost curves developed for the model systems are presented in Figure 4-1.
Capital costs were developed for each individual system based on the
required system to achieve a 1.6 mg F71 blended water supply. Annual capi-
tal costs are based on a 30 year amortization at an interest rate of
12 percent.
Operating costs are based on labor and chemical costs. Additional energy
costs were assumed to be minimal. The anticipated labor requirements were
based on the following:
• 100 gpm System 3 M-H per day
• 250 gpm System 5 M—H per day
• 500 gpm System 6 M-H per day
• 600 gpm System 8 M-H per day
Labor costs were assumed to be $10/M-H. Chemical costs were based on the
anticipated caustic and acid requirements FOB, Columbia, South Carolina.
Annual maintenance costs were based on 3 percent of the total capital cost.
The developed annual 0&M cost curve is presented in Figure 4-2.
4- 9
-------
9
8
7
6
5
C
C
o4
0
0
‘ft
1-
‘3
-J
I .-
C.,
2
1
FLOW (gpm)
• ACTIVATED ALUMINA SYSTEM
• SYSTEM WITH BRINE OR RAW WATER STORAGE
£ SYSTEM WITH BRINE AND RAW WATER STORAGE
10 50 100 200 300 400 500
1000
FIG. 4-1 ACTIVATED ALUMINA SYSTEM — CAPITAL COST
-------
0
0
0
0
0,
10
9
8
7
6
4
3
2
1
100 200 300 400 500
FLOW (gpm)
FIG. 4-2. ACTIVATED ALUMINA SYSTEM — ANNUAL 0 & M COST
1000
-------
PROCESS DESIGN - REGIONAL WATER TREATMENT FACILITY
A conventional surface water treatment facility was designed in order to
develop capital and O&M costs. A process flow schematic is presented in
Figure 4—3. The facility consisted of flash mixing, flocculation, sedi-
mentation and filtration. Ancillary facilities consisted of gravity
thickening and filter presses for sludge dewatering. A settling pond is
also provided for sand filter backwash. A process design sumary is
presented in Table 4—6.
The estimated total capital cost for the treatment plant is $26,250,000.
The estimated plant operating cost is $509,000 per year. These costs
include energy, chemical, and labor costs. The estimated maintenance cost
for the facility is $675,000 per year.
4-12
-------
ALUM
GE
FLASH MIX
(Two Chambers)
FLOCCULATOR
SEDIMENTATI ON
BASINS (13)
THICKENERS 12)
TANKS (3)
DISTR1 BUTION
SYSTEM
FiLTER CAKE
FILTER
PRESSES
(2)
FIG. 4-3 . PROCESS FLOW SCHEMATIC — PEE—DEE RiVER REGIONAL FACILITY
-------
TABLE 4-6
PROCESS DESIGN SUMMARY
Flash Mix Basins
Detention Time, sec 20
Number of Basins 3 2
Basin Volume, gal x 10
Basin 1 54
Basin 2 54
Basin Dimensions, ea.
Length, ft 8.5
Width, ft 8.5
Depth, ft 10
SWD, ft 12
Power Requirement, ea
Basin 1, HP on line 40
Basin 2, HP on line 40
Flocculation Basins
Number of Basins 4
Detention Time Total, in 40
Basin Volume, gal x 10
Compartment 1 224
Compartment 2 224
Compartment 3 224
Basin Dimensions, ea
Length ioo
Width 30
Depth io
SWO 12
Power Requirement, ea
Compartment 1, HP on line 11
Compartment 2, HP on line 7
Compartment 3, HP on line 1
Paddle Area, sq ft 500
4-14
-------
TABLE 4-6 (Cont.)
PROCESS DESIGN SUMMARY
Paddle Velocity
Compartment 1, fps 2.54
Compartment 2, fps 1.72
Compartment 3, fps 0.93
Revolutions per minute
Paddle Compartment 1 6
Paddle Compartment 2 4
Paddle Compartment 3 2.25
Alum Dosage Supply
Alum, ppm 3 40-100
Alum Dose, lb/day x 10 , maximum 37.5
Alum Supply, days 3 30
30-day supply, gal x 10 208.5
Alum Storage Tanks 3 2
Volume, ea, gal x 10 108
Tank Dimensions
height, ft 15
diameter, ft 35
Sedimentation Basins
Surface Overflow Rate, gpm/sq ft 0.25
Detention Time, hr 3 5
Basin Volume Total, gal x 10 9.724
Number of Basins 13
Basin Dimensions
Length, ft 200
Width, ft 50
Depth, ft 10
SWD, ft 12
Flow per basin, gpm 2,400
Flow Velocity, fpm 0.6
Weir Overflow Rate, gpm/ft 8
Weir Length Total, ft 3,900
Weir Length, per Basin, ft 300
Thickener
Mass Dry Solids, lbs/mg 490
Volume Underf low Sludge, g 1/mg 2,930
Thickener Surface Area, ft 4,400
Number of Thickeners - 2
Detention Time, days 3.9
4-15
-------
TABLE 4-6 (Cont.)
PROCESS DESIGN SUMMARY
Thickener Dimensions
Height, ft 15
Diameter, ft 3
Sludge Removal Rate, gal x 10 75
Sludge Mass Removed, lb/day x 10 21.9
Filter Beds
Filter Bed Components
Coarse material Anthracite
Fine material 2 3 Sand
Filter Bed Area Total, ft x 10 12.6
Number of Filter Beds 2 6
Surface Area, ft 2,100
Filter Run Time
Accepted Contemporary Maximum, hrs 60
Anticipated, Firs 48
Solids Accumulation Rate, lb/hr 6.6
Backwash Flow Rate, g rn 31,500
Backwash Flow, gpm/ft 1 5
Backwash Time,_min - 5—10
Backwash Holding Tank ,
Volume, ea, gal x 10” 423
Volume, Total, gal x 10 1,269
Number of Tanks 3
Tank Dimensions
height, ft 20
diameter, ft 60
Settling Pond
Number of Ponds 3 1
Volume for Settling, gal x 10 1,683
Pond Dimensions
Length, ft 300
Width, ft 75
Depth, ft 10
Detention Time, days 1.8
4-16
-------
TABLE 4-6 (Cont.)
PROCESS DESIGN SUMMARY
Filter Press
Condition Solids, lb Lime/lb sludge solids 0.15
Coagulant Condition, lb/ton dry SS 5
Cake Solids, % 3 25
Cake Density, lb/ft 70
Cake Thickness, in. 1.2
Filter Chamber Volume, ft 1.88
Cycle Time, hr 2
Filter Pressure, psig 225
Dry Weight Solids, 3 1b x 10 /day 25.2
Cake Solids, ib/lO /da 100.75
Filter Cake Volume, ft 1,450
Number of Chambers/day 775
Cycles per Day 8
No. of Presses 2
No. of Chambers, ea press 50
Chlorination
Chlorine Dose, ppm 3
Contact Basin 3
Volume, gal x 10 945
Detention Time, mm 30
4-17
-------
REFERENCES
1. Maier, F. J. Fluoridation . CRC Press, Cleveland, Ohio (1972).
2. Boruff, C. S., “Removal of Fluorides from Drinking Waters.” md. Eng.
Chem. , 26, 1, 69 (1934).
3. Scott, R. 0., et al., “Fluoride in Ohio Water Supplies.” Journal
AWWA , 29, 17 1937).
4. Cuip, R. L., and Stoltenberg, H. A., “Fluoride Reduction at La Crosse,
Kansas.” Journal AWWA , 50, 3, 427 (1958).
5. Maier, F. 3., “Partial Defluoridation of Water.” Public Works , 91,
11, 90 (1960).
6. Sigworth, E. A., and Smith, S. B., “Adsorption of Inorganic Compounds
by Activated Carbon.” Journal AWWA , 64, 6, 386 (1972).
7. McKee, R. H., and Johnston, W. S., “Removal of Fluorides from Drinking
Water.” md. Eng. Chern. , 26, 8, 849 (1934).
8. Maler, F.J., “Defluoridation of Municipal Water Supplies,” Jour-
nal AWWA , 45, 8, 879 (1953).
9. Thompson, 3., and McGarvey, F. X., “Ion—Exchange Treatment of Water
Supplies.” Journal AWWA , 45, 2, 145 (1953).
Burwell, A. L., et al., “Fluoride Removal from Drinking Water.” Okia.
Geo. SurveyTW No. 25 (1945).
11. Lee, L. P., and Haras, E. “Fluoride Removal.” Paper presented at
Permutit Seminar, July 19, 1979.
12. Harmon, J. A., and Kalichnian, S. G , “Defluoridation of Drinking
Water in Southern California.” Journal AWWA , 57, 2, 245 (1965).
13. “Treating a High Fluoride Water.” Public Works , 86, 2, 67 (1955).
14. Maier, F. J., “Water Defluoridation at Britton: End of an Era.”
Public Works , 102, 6, 70 (1971).
15. Adler, etal., “Removal of Fluorides from Potable Water by Tricalcium
Phosphate.” md. Eng. Chem. , 30, 2, 163 (1938).
-------
16. Behrman, A. S., and Gustafson, H., “Removal of Fluoride from Water.”
md. Eng. Chem. , 20, 9, 1011 (1938).
17. Goodwin, R. C., and Litton, J. B., “Removal of Fluorides from Public
Water Supplies.” md. Eng. Chem. , 33, 8, 1046 (1941).
18. Warnsley , R.., and Jones, W. E., “Fluoride Removal.” Wtr & Sew. Works ,
94, 8, 272 (1947).
19. “How to Take Fluoride Out of Water.” Engr. News—Record 144, 3, 40
(January 19, 1950).
20. Kempf, C. A., et al., “Removal of Fluorine from Water.” Iowa Acad .
Sd., 41,TSfl1934).
21. Kempf, C. A., et al., “Studies on the Removal of Fluorine from Drink-
ing Wate rT the State of Iowa.” Iowa Acad. Sd. , 43, 191
(1936).
22. Boruff, C. S., et al., “Adsorption of Fluorides from Salts by Alum
Floc.” InJ Ei . Chem. , 29, 10, 1154 (1937).
23. Clifford, D., et a]., “Activated Alumina: Rediscovered ‘Adsorbent’
for Fluoride, Humic Acids and Silica.” md. Water Eng. , 15, 7, 6
(1978).
24. Fink, G. J., and Lindsay, F. K., “Activated Alumina for Removing
Fluoride from Drinking Water.” md. Eng. Chem. , 28, 8, 947
(1936).
25. Swope, H. G., and Hess, R. H., “Removal of Fluorides from Natural
Waters by Defluorite.” md. Eng. Chem. , 29, 4, 424 (1937).
26. Savinelli, E. A., and Black, A. P., “Defluoridation of Water with
Acti vated Alumina.” Journal AWWA , 50., 1, 33 (1958).
27. Bishop, P. L., and Sansoucy, G., “Fluoride Removal from Drinking
Water by Fluidfzed Activated Alumina.” Journal AWWA , 70, 10,
554 (1978).
28. Wu, Y. C., “Activated Alumina Removes Fluoride Ions from Water.” Wtr.
& Sew. Works , 125, 6, 76 (1979).
29. Choi, W. W,, and Chen, K. V., “The Removal of Fluoride from Waters by
Adsorption.” Journal AWWA , 71, 10, 562 (1979).
-------
30. Rubel, F., and Woosley, R. 0., “The Removal of Excess Fluoride from
Drinking Water by Activated Alumina.” Journal AWWA , 71, 1, 45
(1979).
31. Kosarek, L. J., “Purifying Water by Reverse Osmosis, Part I.” Plant
Engrg . 33, 16, 103 (August 9, 1979).
32. Homer, W. A., and Scharpf, C. A., “Principal Factors Relating to the
Selection of Single-Purpose and Dual-Purpose Desalting Plants.”
Chem. Eng. Prog. Symp. Series 67, 107, 184.
33. Podall, 1. E., “Recent Advances in Reverse Osmosis Membranes for
Desalination.” Chem. Eng. Prog. Symp. Series 67, 107, 260.
34. Channabasappa, K. C., “use of Reverse Osmosis for Valuable By-
Products Recovery.” Chem. Eng. Prog. Symp. Series 67, 107, 250.
35. Lim, H. S., and Johnston, H. K., “Reverse Osmosis as an Advanced
Treat ment Process.” Journal WPCF , 48, 7, 1804 (1976).
36. Quinn, R. M., and 1-lendershaw, W. K., “A Comparison of Current Membrane
Systems Used in Ultrafiltration and Reverse Osmosis.” md.
Water Eng. , 13, 3, 12 (1976).
37. Beasley, J. K., “The Evaluation and Selection of Polymeric Materials
for Reverse Osmosis Membranes.” md. WAter (rig. , 15, 3, 6
(1978).
38. Beach, W. A., and Epstein, A. C., “Summary of Pretreatment Technology
for Membrane Processes.” md. Water Eng. , 12, 4, 13 (1975).
39. Crits, S. J., “Some Characteristics of Major Types of Reverse Osmosis
Modules.” md. Water Eng. , 13, 6, 20 (1977).
40. Koserek, L. J., “Purifying Water by Reverse Osmosis, Part I!.” Plant
Engrg. , 33, 19, 183 (Sept. ?O, 1979).
41. Cruver, J. F., and Sleigh, J. H., Reverse Osmosis-—The Emerging
Answer to Seawater Desalination.” md. Water Eng. , 13, 3, 8
(1976).
42. Schmitt, R. P., and Harley, S. M., “Development in Use of Desalting by
the United States Military.” Chem. Eng. Prog. Symp. Series 67,
43. Caprio, C., et al., “Reverse Osmosis Provides Reusable Water from
ElectroniEsVaste.” md. Water Eng. , 14, 6, 24 (1977).
-------
44. Crabbe, D. C. M., “A Double Pass Reverse Osmosis System for the
Produc tion of USP Water.” md. Water Eng. , 13, 6, 12 (1977).
45. Wadlington, M., “Chemical Regenerant Savings Can Pay for a Reverse
Osmosis Unit.” md. Water Eng . , 13, 3, 17 (1976).
46. Hollier, M., “Experiences with Reverse Osmosis Demineralizing for
Boiler Feed Water.” md. Water Eng. , 15, 3, 20 (1978).
47. “Water Factory 21.” Technology Transfer Water Research Capsule
Report.
48. Al-Gholaikah, A., et al., “The World’s First Large Seawater Reverse
Osmosis DesalT tion Plant at Jeddah, Kingdom of Saudi Arabia.”
NWSIA Journal , January, 1979.
49. Taylor, I. G., and Haugseth, L. A., “Yuma Desalting Plant Design.”
md. Water Eng. , 14, 2, 17 (1977).
50. “Estimating Water Treatment Costs, Vol. 3.” U.S. EPA Report
60/2—79-162C, August, 1979.
51. “Defluoridation Study for The Arizona Department of Health Services”
December, 1978 (Unpublished).
------- |