EPA-600/2-75-047
October 1975
DEMORALIZATION OF WASTEWATER
BY
ELECTRODIALYSIS
by
Harold H, Takenaka
U.S. Environmental Protection Agency
San Francisco, California 94111
Ching-Lin Chen
Robert P. Miele
County Sanitation Districts of Los Angeles County
Whittier, California 90607
Contract No. 14-12-150
Project Officer
John M. Smith
Wastewater Research Division
Municipal Environmental Research Laboratory
Cincinnati, Ohio 45268
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
MUNICIPAL ENVIRONMENTAL RESEARCH LABORATORY
CINCINNATI, OHIO 45268
-------
DISCLAIMER
This report has been reviewed by the Municipal
Environmental Research Laboratory, U.S. Environmental
Protection Agency, and approved for publication. Approval
does not signify that the contents necessarily reflect the
views and policies of the U.S. Environmental Protection Agency,
nor does mention of trade names or commercial products
constitute endorsement or recommendation for use.
-------
TECHNICAL REPORT DATA
(Please read litunjcrians on the reverse before completing)
1, REPORT NO,
EPA-600/2-75-047
3.
4. TITLE ANDSUBTITLE
DEMORALIZATION OF WASTEWATER BY ELECTRODIALYSIS
5. RETPORTDATE
October 1975 (Issuing Date)
6. PERFORMING ORGANIZATION CODE
7. AUTHOH(S)
Harold H. Takenaka, Ching-Lin Chen, and
Robert P. Miele
8. PERFORMING ORGANIZATION REPORT NO.
9, PERFORMING ORG "VNIZATION NAME AND ADDRESS
County Sanitation Districts
1955 Workman Mill Road
Whittier, California 90607
of Los Angeles County
10. PROGRAM ELEMENT NO.
1BB043/21-AST
it.
NO,
14-12-150
12. SPONSORING AGENCY NAME AND ADDRESS
Municipal Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
13. TYPE OF REPORT AND PERIOD COVERED
Final, 4-70 to 12-70
14. SPONSORING AGENCY CODE
EPA-ORD
15, SUPPLEMENTARY NOTES
15. ABSTRACT
Demineralization of carbon-treated secondary effluent by a 45.5 1 pm (12 gpm) Ionics
electrodialysis pilot plant was investigated at the Pomona Advanced Wastewater
Treatment Research Facility, Pomona, California.
Slime formation (organic fouling) on the membranes and spacers of the electrodialysis
stack during continuous operation significantly decreased the effectiveness of the
electrodialysis process to demineralize municipal wastewater. If the total COD of
the feedwater was maintained at or below 10 mg/1, a weekly enzyme-detergent flush
maintained the total dissolved solids (TDS) removal in the design range of 30-35-
percent.
A cost estimate for a 10 MGD single stage electrodialysis plant based on the
operating results obtained at Pomona was made. The estimated cost of 19.4<£ per
1,000 gallons was based on the use of carbon-treated secondary effluent with an
average TDS concentration of 540 mg/1 and a total COD of 10 mg/1 or less to
produce a product water with 30-35 percent reduction in TDS.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFlEHS/OPEN ENDED TERMS
c. COSATI Field/Group
Sewage
Effluents
Membranes
Cost analysis
Microorganism control (sewage)
Electrodialysis
Demineralizers
Waste water
13B
3. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19, SECURITY CLASS (This Report)
UNCLASSIFIED
21. NO. OF PAGES
20. SECURITY CLASS (Thispage)
UNCLASSIFIED
EPA Form 22ZO-1 (9-73)
1976 — 657-695/5338 Region 5-11
-------
Page Intentionally Blank
-------
FOREWORD
Man and his environment must be protected from the adverse
effects of pesticides, radiation, noise, and other forms of pollution,
and the unwise management of solid waste. Efforts to protect the
environment require a focus that recognizes the interplay between
the components of our physical environment—air, water, and land.
The Municipal Environmental Research Laboratory contributes to this
multidisciplinary focus through programs engaged in
• studies on the effects of environmental contaminants
on the biosphere, and
• a search for ways to prevent contamination and to
recycle valuable resources.
Electrodialysis has long been used to desalt brackish water.
This report describes a pilot scale evaluation of this process for
the partial demineralization of wastewater.
-------
ABSTRACT
Demineralization of carbon-treated secondary effluent by a 45.5 1pm
(12 gpm) Ionics electrodialysis pilot plant was investigated at the
Pomona Advanced Wastewater Treatment Research Facility, Pomona,
California.
Slime formation (organic fouling) on the membranes and spacers of the
electrodialysis stack during continuous operation significantly de-
creased the effectiveness of the electrodialysis process to demineral-
ize municipal wastewater. Other forms of stack fouling such as mem-
brane scale were not as serious a problem as organic fouling. An
enzyme-detergent (Biz, manufactured by Proctor and Gamble) cleaning
technique was developed to minimize organic fouling and thus maintain
steady state demineralization conditions. If the total COD of the
feedwater was maintained at or below 10 mg/1, a weekly enzyme-deter-
gent flush maintained the total dissolved solids (TDS) removal in the
design range of 30-35 percent.
A cost estimate for a 10 MGD single stage electrodialysis plant based
on the operating results obtained at Pomona was made. The estimated
cost of 19.4
-------
CONTENTS
Page
Abstract iv
List of Figures V1-
List of Tables V1--j
Acknowledgments viii
Sections
I Conclusions 1
II Recommendations 3
III Introduction 4
IV Pilot Plant Description 7
V Membrane Cleaning Operations 8
VI Performance Data 26
VII Membrane Life 30
VIII Process Cost Estimate 34
-------
FIGURES
No. Page
1 The Electrodialysis System 5
2 The Effect of DC Current on 10
Membrane Fouling
3 Daily Enzyme-Detergent Cycle 13
4 Every-Other-Day Enzyme-Detergent 16
Cycle
5 Weekly Enzyme-Detergent Cycle 20
6 Every-Other-Week Enzyme-Detergent 22
Cycle
-------
TABLES
No. Page
1 Summary of Pilot Plant Operations with Different Modes 15
of Daily Enzyme-Detergent Flushing Procedures
2 Summary of Conductivity Reduction 25
3 Major Ion Removal and Selectivity Data 27
(April 1970 through September 1970)
4 Typical Heavy Metal Concentration Data 28
(May 6, 1970)
5 Removal of COD, TOC, TDS and Turbidity 29
(April 1970 through September 1970)
6 20 Cell-Pair Stack Membrane Quality Control Tests 31
7 65 Cell-Pair Stack Cation Membrane Quality Control Tests 32
8 65 Cell-Pair Stack Anion Membrane Quality Control Tests 33
9 Electrodialysis Process Cost Estimate 35
Vll
-------
ACKNOWLEDGMENTS
The authors, Harold H. Takenaka, Ching-lin Chen, and Robert I? Miele,
wish to express their appreciation to the operating and laboratory
staff of the Pomona Advanced Wastewater Treatment Research Facility.
Mr. John N. English and Mr. John R. Burgeson, former EPA Project
Engineers at Pomona Research Facility, were instrumental in initia-
ting the pilot plant study.
The advice and suggestions given by Dr. John Arnold of Ionics,
Incs. during the course of study were important contributions to
the success of the study.
Helpful review and discussion of the project progress was provided
throughout the duration of the investigation by Mr. Charles W.
Carry, Assistant Department Head, Technical Services Department,
County Sanitation Districts of Los Angeles County.
vm
-------
SECTION 1
CONCLUSIONS
1. Stack fouling due to slime was evidenced whenever an increase in stack
pressure was accompanied by a decrease in stack current (amperage)
or demineralization.
2. An enzyme-detergent (Biz) was successful in removing most of the
slime on the membranes, however, heavy slime formation in the
stack, viz, stringy slime found on spacer webbings, was not
removed.
3. Scale formation was effectively removed by an acid solution.
4. The degree of organic fouling dictated the frequency and length
of the enzyme-detergent portion of the flush cycle {e.g. 30 or
60 minutes). However, see (2) above.
5, An acid solution rinse was required after the enzyme-detergent
flush portion of the cycle to remove the detergent and restore
the stack amerage.
6. No strong evidence was obtained to demonstrate that a 60 minute
rather than a 30 minute acid rinse was more effective.
7. No strong evidence was obtained to demonstrate air-water back-
flushing of the dilute compartments was effective for other
than removing loose scale, sand, and carbon particles if present
in the stack.
8. The quality control tests did not indicate any significant
changes in exchange capacities for the cation membranes after
16,136 hours of operation. A 30 percent reduction of the
exchange capacity of the anion membranes was noted during the
same period of operation. There was no evidence of accelerated
membrane deterioration as a result of enzyme-detergent cleaning.
9. To minimize cleaning cost, realize a 30-35 percent conductivity
removal, and experience continuous steady-state operation, a
weekly enzyme-detergent flush, and stack hand-cleaning and in-
spection every two-months is necessary with an influent total
COD level of about 10 mg/1.
-------
10. A bench-scale current reversal ED unit showed that polarity
reversal had no significant effect in retarding slime formation.
11. The estimated cost to produce a product water with 30 to 35
percent IDS reduction in a 10 MGD single stage electrodialysis
plant is about 19.4<£ per 1,000 gallons of product water, ex-
clusive of the costs of carbon adsorption pretreatment and
brine disposal.
-------
SECTION II
RECOMMENDATIONS
The results of this study indicated that the major problem associated
with the electrodialysis process for wastewater demineralization was the
formation of scale and slime on the membranes. Although these membrane
fouling substances can be effectively removed by regular enzyme-
detergent and acid flushing as demonstrated in this study, this membrane
cleaning technique may not be the most cost-effective way of operating
the electrodialysis plant for wastewater demineralization. Therefore,
some other means of preventing the formation of scale and slime on the
process membranes require further investigation to optimize the process
operation and improve the process cost effectiveness. This may require
the removal of calcium to prevent scale formation, and the use of
bactericidel techniques to prevent slime growth in the membrane stack.
-------
SECTION III
INTRODUCTION
Electrodialysis is a process used to demineralize, desalt, or de-
ionize water by transporting ions in a direct current (D.C.) electric
field through alternating cation and anion membranes. The process
was initially used to desalt brackish water.
Technically, electrodialysis should be effective in partial demin-
eralization of municipal wastewater to remove the approximately 300
mg/1 total dissolved solids (TDS) added to potable water as a result
of domestic usage. Because municipal wastewater (Pomona TDS approxi-
mately 540 mg/1) has a lower mineral content than brackish water
(TDS approximately 1000 - 10,000 mg/1), the operational cost should
be less; however, wastewater contains organic contaminants which
adversely affect the operation and performance of the electrodialysis
process.
The electrodialysis process (Figure 1) utilizes cation and anion
permeable membranes, arranged in an alternating pattern, and placed
in the path of an electric current. The "stack" consists of a series of
of cell-pairs with each cell-pair consisting of an anion membrane,
a spacer, and a cation membrane. The membranes are thin sheets of ion
exchange material (0.102 to 1.016 mm thick) containing active ionic
groups and a substantial amount of water by weight when in use to
enable the active ionic groups to function. The membranes are com-
posed of an inert cross-linked, plastic polymer to which ionizable
radicals such as sulfonic acids or amines are attached by a chemi-
cal bond. The entire structure contains 15 to 40 percent by weight
of water distributed in millions of very fine capillary pores which
honeycomb the structure. Membrane pore sizes are of the order of
10 to 100 Angstroms. The fineness of its passages makes the ion
transfer membrane virtually impermeable to water at normal pressures.
The anion membranes, which contain ionic amine groups with a fixed
positive charge, are capable of repelling other positive ions (cations).
The cation membranes, which contain ionic sulfonic acid groups with a
fixed negative charge, are capable of repelling other negative ions
(anions). The oppositely charged ions are allowed to enter into the
fine capillary pores of the membrane and travel through the membrane
under the influence of a D.C. electric field, while the identically
charged ions are repelled and cannot pass through the membrane. By
-------
FEED
ANODE
C_n
WASTE BRINE
PRODUCT
A-ANION MEMBRANE
C-CATION MEMBRANE
S - SPACER
cc- CONCENTRATE COMPARTMENT
dc - DILUTE COMPARTMENT
FIGURE I : THE ELECTRO DIALYSIS SYSTEM
-------
alternating anion and cation membrane a series of diluted and concen-
trated compartments are established. The spacers are made of a plas-
tic material and provide a path for the water to travel across the
face of each membrane in both dilute and concentrating compartments.
The stack may be regarded as one electrolytic cell, and the standard
electrochemical principles apply. The direct current traverses every
compartment perpendicular to the membranes between the anode and the
cathode. Design and operation of the electrodialysis system depends
upon utilization of appropriate membrane properties combined with
application of Ohm's and Faraday's Laws. The movement of ions in
solutions or through membranes is governed by Faraday's Law and the
voltage requirement for an electric membrane system is governed by
Ohm's Law.
The specific objectives of this electrodialysis pilot plant study
were (1) to investigate the feasibility of wastewater demineraliza-
tion by the electrodialysis process; (2) to investigate the membrane
fouling problems and their control methods; (3) to develop the
optimum operating conditions; (4) to demonstrate the reliability of
the process performance> and (5) to produce a realistic process cost
estimate based on the pilot plant operation experience.
The pilot plant study was conducted with the carbon-treated secon-
dary effluent of the Pomona activated sludge plant from July 1968 to
December 1970.
-------
SECTION IV
PILOT PLANT DESCRIPTION
The electrodialysis pilot plant was designed and constructed by
Ionics Inc. and employed a single-stage with 65 cell-pairs that
was capable of removing about 40 percent of the influent total
dissolved solids. The membranes were of the standard brackish water
demineralization type and were 45.7 cm x 50.8 cm (18" x 20") with
an effective area of 1460 sq. cm. Spacers for the stack were the
tortuous path type, 0.102 cm (0.040") thick, which also served as
perimeter gaskets. The unit received 45.4 1pm (12 gpm) of carbon-
treated secondary effluent and produced 39.7 1pm (10.4 gpm) of
product water and 5.7 1pm (1.5 gpm) waste brine (88 percent water
recovery). The flow rate through a cell-pair was 10 ml/sec with a
residence time of about 15 seconds, while the flow rate through each
of the electrode (anode and cathode) compartments was maintained at
18 ml/sec. The brine stream leaving the concentrate compartments
was partially recirculated with feed water back into the stack.
This method of operation kept the flow rate through the concentrate
compartments approximately the same as through the dilute compart-
ments. This resulted in nearly equal pressure on both sides of each
membrane and prevented membrane and spacer distortion.
Sulfuric acid was injected into the brine recirculation and cathode
streams with diaphragm pumps to prevent scaling problems, A pH
control system was used to control brine stream pH between 3.5 and
4.5. Cathode stream pH was held at about 1.5. Totalizing flow
meters were provided for the feed and product streams.
Direct current for the stack was provided by silicon cell rectifica-
tion of a 3 phase, 440 volt source. The electrode system consisted
of a stainless steel cathode and a platinized columbium anode.
A 20 cell-pair stack and a current reversal unit were also used for
a short period of time to determine the effects of enzyeme-detergent
flushing and polarity reversal in retarding slime formation on the
membranes.
-------
SECTION V
MEMBRANE CLEANING OPERATIONS
From the beginning of the electrodialysis pilot plant operation
on the carbon-treated secondary effluent, membrane fouling pro-
blems constantly interfered with the system performance. The
fouling was caused by scale and slime formation on the membranes.
This in turn caused the operating stack current to decrease from
its initial setting of 4 amps to the range of 2 to 3 amps, and
the stack pressure to increase from 2.1 kg/sq cm (30.5 psig) to
as high as 3.4kg/sqcm (49.5 psig) within a few days of on-stream
operation. Therefore, a great deal of research effort was devoted
to the development of membrane cleaning techniques.
A 20 cell-pair experimental unit furnished by Ionics, Inc. was oper-
ated at the Pomona Research Facility to simulate the operation conditions
of the 65 cell-pair pilot plant, namely flow rate of 10 ml/sec per cell-
pair and current density of approximately 3 ma/sq cm. The 20 cell-pair
stack was allowed to foul and attempts were made to clean the membranes
in-place using the Biz enzyme-detergent, which had shown promising
effectiveness in cleaning the reverse osmosis membranes used in the
concurrent reverse osmosis pilot plant studies. The use of Biz enzyme-
detergent for cleaning the 20 cell-pair stack at Pomona and also some
other compatability tests performed by Ionics laboratory did not show
any detrimental effects on the membranes. Therefore, the enzyme-
detergent cleaning technique was applied to the 65 cell-pair pilot plant.
The enzyme-detergent cleaning solution was prepared by dissolving 1.5
kg of Biz enzyme-detergent in 60 liters of warm tapwater (about 40°C).
The flushing procedure was as follows:
(1) Recirculating the Biz enzyme-detergent solution 30 to
60 minutes through the membrane stack;
(2) Flushing the enzyme-detergent solution out of the stack
with tapwater for 30 to 60 minutes;
(3) Rinsing the stack with sulfuric acid for 30 to 60 minutes
at pH of 2.5 to 3.0; and
(4) Returning the unit to normal operations.
-------
Throughout the flushing cycle, equal flow through the dilute and con-
centrate compartments was maintained to prevent excessive pressure on
one side of the membrane which could cause membrane distortion. The
stack current was off during all flushing periods.
To determine if membrane fouling was enhanced by the DC current elec-
trical attraction of the charged organic materials, the 20 cell-pair
experimental unit was operated with the rectifier off for a two week
period. The rectifier was turned on for 30 minutes once a day to
allow an amperage reading. The results of this special study are shown
in Figure 2. As indicated in this figure, there was very little dif-
ference in the amperage decline rate with or without the operation of
the rectifier. Therefore, the DC current is not a significant factor
in causing membrane fouling in the electrodialysis process for waste-
water demoralization.
Ionics, Inc. also furnished a two-stage electrical and a three-stage
hydraulic bench-scale current reversal unit for the purpose of evalu-
ating the effect of current reversal on slime fouling trends. The
unit was operated under the similar flow rate and current density as
the 65 cell-pair pilot plant. The results of operating the current
reversal unit showed that polarity reversal had no effect in retarding
slime formation. Membrane sliming occurred even though the electrode
polarity was reversed every 15 minutes. This finding supports the
finding of no change between rectifier on and off operation as dis-
cussed above. However, the reversal of electrode polarity did provide
a means for controlling scale that did not require the use of acid .in
the concentrating compartments.
Organic fouling of membranes constituted the primary and most significant
disadvantage to the electrodialysis process in demineralizing wastewater.
Chemical characterization of the slime material scraped from the
electrodialysis membranes at Pomona indicated a formula of Cj^pOs^j
which is slightly different from the reported Cj-HgO^N for the cnemical
composition of biological slimes in anaerobic treatment systems.
Microscopic examinations were also performed on the slimes scraped
from the membranes. These examinations always showed bacteria as well
as fungi, protozoa, worms, and other debris. In addition, microbio-
logical studies of these slimes showed heavy growth within 24 hours
on an anaerobic culture media and medium to heavy growths on aerobic,
fungi, and mold culture media.
Removal of these biological slimes required manual cleaning of the
membranes and normally entailed the following:
(1) Stack disassembly;
(2) Cleaning of each individual membrane and spacer with a
scrub brush or fine steel wool, and a dilute acid
solution; and
-------
o
7
6
co
Q.
< 5
w
h-~
z
UJ
cc 4
cc ^
:D
o
o 3
CO
2
I
0
i
i 1 1
B-BIZ FLUSH (1-hr.)
-
8^^> o^
o^vr°x>
u U *^
-o V,
°^l
0 t
f ^
R °)
B z>
_l
- u.
N
CO
I—
5)-°>xx
N
-
—
o
—
)
w
o3o B B B B
H~ Ijj |O _
O I O
^ C^LL
1
0 5
1 1 1
10 15 20
i |
25 30
3S
DAYS ON STREAM
FIGURE 2- THE EFFECT OF DC CURRENT ON MEMBRANE FOULING
-------
(3) Stack reassembly.
Approximately 6 to 8 man-hours were required to accomplish this task
on a Mark II 65 cell-pair stack.
To minimize the "out of service" time required to clean the membranes,
a program was initiated to evaluate the use of enzyme-detergent
flushing procedures. The following five procedures were evaluated for
the length of time shown:
(1) Enzyme-detergent flush the unit on an as-required basis
when the demineralization level dropped below 25 percent
- 273 days;
(2) Enzyme-detergent flush the unit daily - 51 days;
(3) Enzyme-detergent flush the unit every other day - 68 days;
(4) Enzyme-detergent flush the unit weekly - 92 days; and
(5) Enzyme-detergent flush the unit every-other week - 72 days.
The unit operation and performance of each of these five cleaning
sequences will be discussed individually.
As - Required Enzyme-Detergent Cycle
During this 273 day period the unit was allowed to operate continuously
until the demineralization level dropped below 25 percent. At this
point, the stack current had decreased from 4 amps to approximately 2
amps and the stack pressure had increased from 2.1 kg/sq cm (30.5 psig) to
approximately 3.3 kg/sq cm (48 psig). To remove the organic foulants the
stack was flushed with an enzyme-detergent solution according to the
procedures described earlier. A total of six enzyme-detergent flushings
were applied to the unit during the 273 day cycle.
Poor stack performance, characterized by reduced TDS removal, low stack
amperage and high stack pressure, necessitated stack disassembly and in-
spection 47 days after start-up. The inspection showed:
(1) Slime formation throughout the stack membranes;
(2) One anion membrane was moderately scaled and had to be
replaced; and
(3) Slight scaling on all anion membranes at spacer webbings.
Approximately 77 days after start-up, high stack resistance was noted
and resulted in reduced current and demineralization performance and
dictated stack disassembly, inspection, and hand cleaning. This occurred
11
-------
despite two enzyme detergent flushings that were conducted a week apart.
The reduction in current was not accompanied by high stack pressure, so
scaling rather than slime fouling of the membranes was assumed. Stack
disassembly showed scaling in the anion membranes in the concentrate
compartments. The scale was removed by using a dilute sulfuric acid
solution and by lightly rubbing the membranes with steel wool.
In addition to the presence of scale, particles of undissolved light
bluish colored enzyme-detergent were detected under spacer webbings.
To more effectively remove the foulants and the cleaning agent, the
piping was modified to allow by-passing the brine recirculation pump
and to provide identical flows through both the dilute and concentrate
compartments during the flushing operation.
Eighty-two days after start-up, abrupt stack pressure increased and
stack amperage declined with resultant decline in demineralization.
This coincided with periods of methanol leakage from the carbon column
that supplied water to the electrodialysis unit. Methanol was used
in the carbon column during a dem'trification study. COD values as
high as 50 mg/1 were recorded in the carbon effluent feed to the unit.
Stack inspection showed a large build-up of slime particularly in the
dilute compartments.
The enzyme-detergent was not effective in removing the slime found
in the dilute compartments caused by the high feed COD, nor was it
effective in removing slime that accumulated during long periods of
operation between cleanings. On several occasions shortly after
enzyme-detergent flushing, the stack was inspected and the stringy
slime formation on the spacer webbings remained. Unit operation
under these conditions resulted in a current of 3.2 to 3.5 amps with
about a 34 percent conductivity reduction. These conditions ordi-
narily reflect clean membranes. However, due to the unremoved slime
the stack pressure was increased from 2.1 kg/sq cm (30.5 psig) to
3.3kg/sq cm (48 psig) and the flow decreased from 38 1 pm (10 gpm)
to 36 1 pm {9.6 gpm). To restore normal flow, hand-cleaning of the
spacers with a scrub brush was necessary.
Occasional air-water backflushing of the dilute compartments was
effective in removing loose scale, sand and carbon particles if pres-
ent in the stack and thereby helped lengthen the operating time
between enzyme-detergent flushings throughout the 273-day study.
The air-water backflushing procedure normally required 10 to 20
minutes to complete.
Daily Enzyme-Detergent Cycle
Unit performance for this 51-day study is shown in Figure 3. The
influent total and dissolved COD averages were 11.6 mg/1 and 7.8 mg/1,
respectively. Stack pressures above 2.3 kg/sq cm (34 psig) shown in
12
-------
. 6
<
h-~
UJ
I 4
o
I
'.40
UJ
§
o
z
o
o
20
10
STACK CURRENT
STACK PRESSURE
CONDUCTIVITY REMOVAL
I
20 30
DAYS ON STREAM
40
50
u
cr
2.8^
O)
Of.
Z)
2 1
•* • '
oe.
Q.
u
1.4
60
FIGURE 3: DAILY ENZYME- DETERGENT CYCLE
-------
Figure 3 were a result of influent COD concentrations of 15 mg/1 to
20 mg/1. The unit operated for 1,087 hours at approximately 90 per-
cent water recovery. The average feed and product conductivities
were 869 p.mhos/cm and 583 umhos/cm, respectively, a reduction of 33
percent.
Several flushing methods were investigated during the daily enzyme-
detergent flush cycle study. These were:
(1) 30 minutes enzyme-detergent, 30 minutes water and 30
minutes air-water flush;
(2) 30 minutes enzyme-detergent, 30 minutes water and 30
minutes acid solution flush;
(3) 30 minutes enzyme-detergent, 30 minutes water, 30
minutes acid solution and 10 minutes air-water flush;
(4) 60 minutes enzyme-detergent, 60 minutes water, 60
minutes acid solution and 10 air-water flush; and
(5) 30 minutes enzyme-detergent, 30 minutes water and 60
minutes acid solution flush.
The results of employing these various methods of flushing the mem-
branes are identified and shown in Table 1.
It was concluded that a 90 minute flushing period was adequate and
that an acid solution rinse was necessary in the enzyme-detergent
cycle to insure stack amperage recovery. These results confirmed
earlier studies by Ionics that showed a high resistance film, remov-
able by hydrogen ion .adhered to the membrane after soaking in an
enzyme-detergent cleaning solution. However, it should be noted
that the stack amperage loss without the acid flush was temporary;
the amperage readings would usually return to normal the day after
the membrane flushing. The four methods that included an acid rinse
were all effective in recovering the stack current. There was no
strong evidence that: (1) 60 minutes of acid flush was more effective
than 30 minutes of acid flush; and (2) air-water flush was effective
other than to remove loose scale, sand, and carbon particles if
present in the stack. The time (30 or 60 minutes) required in the
enzyme-detergent portion of the cycle would be dependent upon the
extent of slime formation on the membranes and spacers in the dilute
compartments.
Every-Other Day Enzyme-Detergent Cycle
Unit performance for the 68 day study is shown in Figure 4, The
total and dissolved feed COO averages were 5.9 mg/1 and 5.1 mg/1,
14
-------
TABLE 1
SUMMARY OF PILOT PLANT OPERATIONS WITH DIFFERENT MODES OF
DAILY ENZYME-DETERGENT FLUSHING PROCEDURES
of
Flushing
A
B
C
D
E
Testing
Period
( day )
10
10
11
10
10
Average Stack
( psig
Before
31.6
34.8
31.2
31.4
30.5
Pressure
) (2)
After
29.8
32.6
29.7
29.9
29.9
Average Stack Current
( amps )
Before After
3.2 2.5
2.9 3.1
3.2 3.4
3.2 3.5
3.7 3.9
(1) Mode of Flushing:
A - 30 min. Enzyme-detergent, 30 rnin. water and 30 min. air-water flush.
B - 30 min. Enzyme-detergent, 30 min. water and 30 min. acid flush.
C - 30 min. Enzyme-detergent, 30 min. water, 30 min. acid and 10 min. air-
water flush.
D - 60 min. Enzyme-detergent, 60 min. water, 60 min. acid and 10 min. air-
water flush.
E - 30 min. Enzyme-detergent, 30 min. water and 60 min. acid flush.
(2) 1 psig = 0.0703 kg/sq cm
-------
E
u
o .
O)
LU
13
LU
DC
u
£2.1
I
< 4
LU
IT
a:
o
CO
J 40
LU
a:
30
o
o
z
o
u
20
STACK PRESSURE
STACK CURRENT
60 WIN.
ACID FLUSH
CONDUCTIVITY REMOVAL
10 20 30
DAYS ON STREAM
40
FIGURE 4: EVERY-OTHER-DAY ENZYME - DETERGENT
CYCLE
16
-------
E
u
l/» '
O>
UJ
IT
EC
ac
o
g
STACK PRESSURE
STACK CURRENT
•60 MIN. ACID
FLUSH
CONDUCTIVITY REMOVAL
40
50 60
DAYS ON STREAM
70
80
FIGURE 4; (CONTINUED)
17
-------
respectively. The average feed and product conductivities for the
study were 982 and 658 ^mhos/cm, respectively, a reduction of 33
percent. A routine stack inspection was conducted at 30 days into
the study to see the effect of the flushing procedure upon the mem-
branes. The inspection did not reveal any excessive slime or scale
build-up. The stack was exceptionally clean with only a few isolated
carbon and sand particles. This could be partially attributed to the
additional 10 minute air-water flush employed between enzyme-detergent
flushings. However, this mode of membrane cleaning was not able to
restore the stack current to the expected level. On 38th day of the
study, an hour of acid flush was applied in addition to the routine
enzyme-detergent flushing. This additional acid flush was effective
in restoring the stack current from 3.3 amps to 4.7 amps. The con-
ductivity removal through the unit was also improved from 28 percent
to 34.5 percent as a result of the acid flush.
Significant changes in unit performance occurred on the 48th and 49th
days, as the conductivity removal declined to less than 25 percent.
Enzyme-detergent flushing did not restore the performance to the expected
level. An hour of acid flush was able to restore the stack current
but not the stack pressure. The stack was finally inspected on the
51st day to determine the cause of the high stack pressure. The
inspection showed that the membranes and spacers were exceptionally
clean with very little slime and scale formation. However, there was
a heavy accumulation of calcium carbonate on the cathode plate and in
the cathode compartment. The calcium carbonate deposits were removed
by a dilute acid solution wash and rubbing with fine steel wool. The
membranes and spacers were rinsed in tapwater before the stack was
reassembled and put back on stream with normal stack pressure and
current readings of 2.3 kg/sq cm (33.5 psig) and 4.2 amps, respectively.
The excessive deposits of calcium carbonate occurred on two separate
occasions during a 7 day period because no acid was being added into
the cathode compartment to prevent scaling or calcium carbonate build-
up. Each occasion lasted for 12 hours and was caused by air-locking
of the acid pump.
The stack was inspected at the end of the study to determine the cause
of the sudden stack pressure increase as shown in Figure 4. Just
prior to disassembling the stack it was probed. Stack probing involved
indirectly determining the resistance of the stack using volt meter.
The probing was started at the top of the stack and the voltage for
every inch of the entire stack was taken. A high voltage reading,
corresponding to a high internal resistance, as compared to the other
readings of any one inch segment of the stack would indicate the
location of a possible problem area. The results of the stack probing
did not locate any trouble spot which could be responsible for the
stack pressure increase. Consequently, the stack was disassemDleci
to inspect the condition of the membranes. Very little slime and
18
-------
scale formation on the membranes were noticed during the inspection.
The cathode compartment was found free of calcium carbonate deposits
that were noted previously. The reason for the pressure increase
toward the end of the study was not readily apparent.
Weekly ...Enzyme-Detergent Cyc 1 e
Performance for this 92 day study is shown in Figure 5. The unit
operated for 2,019 hours at approximately 91 percent water recovery.
The average feed and product conductivities were 804 and 555 jumhos/cm,
respectively, a reduction of 31 percent. The influent total and
dissolved COD averages were 9.1 and 6.9 mg/1 , respectively.
At the start of the study, the stack was disassembled and the membranes
and spacers were rinsed in tapwater to remove small amounts of sand,
carbon particles, and slime in the dilute and concentrate compartments.
Ten days later, because of high stack pressure and poor performance,
the stack was inspected and hand-cleaned to remove the slime and small
amounts of scale under spacer webbings. The stack was not disassembled
or hand-cleaned again until the end of the study, at which time a sig-
nificant amount of slime was present on the membranes and spacers in
the dilute compartments.
It was found that the weekly enzyme-detergent cleaning schedule was
not able to maintain a steady stack current between the cleaning cycles.
The conductivity removal was reduced as a result of the decline of the
stack current. Therefore, a 30 minute acid rinse with or without a
10 minute air water flushing, depending on the extent of the build-up
of the stack pressure, was employed between the enzyme-detergent
cleaning cycles to maintain a rather steady unit performance. As in-
dicated in Figure 5, the conductivity removal showed a definite de-
cline trend during the last 25 days of the study. This could be
attributed to the inadequate membrane cleaning procedures. The stack
inspection at the completion of the study did show a significant
amount of slime on the membranes in the dilute compartments.
E very Other WeekEnzyme-Deterjent Cycl e
The unit was operated with this cleaning schedule for a total of 72
days. The various performance data are shown in Figure 6. The over-
all average of the water recovery during this study period was approx-
imately 91 percent. The average feed and product conductivities were
806 and 583 jumhos/cm, respectively, a reduction of 27.6 percent. The
influent total and dissolved COD averaged 7.1 and 5.4 mg/1, respectively,
The stack was hand-cleaned prior to initiating the 72 day study.
to declining stack performance, the stack was inspected after th
Due
the first
19
-------
E
u
IT
•A
3.5"
0£
3
CO
0£,
a.
U
2 8
* •**
2.1
I
<
UJ
tr
or
o
o
5s 40
o
s
UJ
IT
30
o
o
o 20
O
STACK PRESSURE
UJ
o o o o o o o o * 9
UJ < < UJ < UJ < <<< UJ« UJ<
STACK CURRENT
Ac-ACID
Aw-AIR WATER
R-RINSE
Hc-HANDCLEAN
E-ENZYME DETERGENT
•STACK OFF
9 HOURS
CONDUCTIVITY REMOVAL
10
20 30
DAYS ON STREAM
40
50
FIGURE 5: WEEKLY ENZYME-DETERGE NT CYCLE
20
-------
E
u
3.5
STACK PRESSURE
<
H-
CE
-------
E
v
UJ
tt
V)
v>
UJ
0£
O-
o Q
2.8
to
Q.
2
UJ
I
o
X.
o
t
CO
40
UJ
QC
30
o
8 20
o
<
UJ
UJ <
STACK CURRENT
Ac - ACID
E-ENZYME DET
He- HANDCLEAN
CONDUCTIVITY REMOVAL
10 20
DAYS ON STREAM
FIGURE 6: EVERY-OTHER-WEEK ENZYME-DETERGENT
CYCLE
-------
E
u
U)
32.8
to
to
a.
O
2.1
UJ
£E
CC
3
40
UJ
CC
30
-STACK PRESSURE
o u o o
I < < iu iu < UJ
STACK CURRENT
CONDUCTIVITY REMOVAL
50 60
DAYS ON STREAM
70
80
FIGURE 6: (CONTINUED)
23
-------
15 days of operation and showed the presence of large quantities of
brownish colored slime (previous slime color was gray) on the membranes
and spacers in the dilute compartments. Scaling was also evident on
some of the membranes. Slime was also found on the glass of the feed
and brine make-up rotometers so they were dismantled and cleaned.
The declining performance during the first 15 days of operation could
be due to the deteriorated influent quality. The average COD of the
feed for a seven-day period prior to the stack inspection on the 15th
day was about 13.1 mg/1. The stack was hand-cleaned and it was re-
turned to service receiving influent water with a COD of less than
10 mg/1.
On the 44th day, the stack was again inspected due to an overall de-
cline in unit performance. The enzyme-detergent and acid flush did
not maintain stack performance. Stack inspection showed very little
slime and scale build-up on the membranes and spacers; however, one
of the spacers was found to be placed incorrectly in the stack. The
mistake had occurred during the hand-cleaning on the 15th day. The
am"on membrane associated with the misplaced spacer was completely
covered with scale and showed evidence of deterioration. Corrective
action was taken, and the unit was back on stream four days later.
In spite of cleaning the stack and replacing the anion membrane,
further operation with this mode of membrane cleaning schedule con-
tinued to show a definite decline in the unit performance.
The overall unit performance, as reflected by the conductivity re-
moval, under the above five different modes of membrane cleaning
procedures is summarized in Table 2 . As expected, the enzyme-
detergent cleaning of the stack at more frequent intervals resulted
in cleaner membranes, hence higher conductivity removals. However,
when the expensive cost for high frequency membrane cleaning is
considered together with the slight gain in the conductivity re-
moval as indicated in Table 2 , it is better to use a less frequent
cleaning schedule for the unit operation. A weekly enzyme-detergent
cleaning cycle will be employed as the basis for process cost estimate
in this report.
Z4
-------
TABLE 2
SUMMARY OF CONDUCTIVITY REDUCTION
Frequency of Period of Average
Enzyme-Detergent Study Conductivity Reduction
Cleaning ( days ) ( % )
As-Required 273 29
Daily 51 33
Every-other-day 68 33
Weekly 92 31
Every-other-week 72 28
25
-------
SECTION VI
PERFORMANCE DATA
The average concentrations of the major ions found in the feed and
product waters for a six month period, April 1970 through September
1970, are shown in Table 3. The ion removal and ion selectivity,
which is defined as the ratio of the percentage removal of a given
ion to the percentage removal of all ions, are also shown in Table 3.
As indicated in Table 3, the bivalent cations seemed to be more
efficiently removed than the monovalent cations. However, the mono-
valent anions demonstrated higher removal than the bivalent anions.
The ion selectivity as defined above can be affected by many para-
meters. These may include pH, influent constituents, membrane
structure, polarization, membrane fouling, etc. The selectivity for
sulfate ion was found to decrease significantly with an increase in
membrane fouling. The phosphate ion selectivity was also found to
behave similarly as sulfate ion but not as significantly. Generally,
as expected, the ion removal by the electrodialysis process was not
specific.
A typical set of the heavy metal concentrations in various process
streams is shown in Table 4. The heavy metal concentrations in the
process feed water were too low to provide any conclusive evaluation
of the process capability in heavy metal removal.
As shown in Table 5, the electrodialysis process was able to remove
some of the refractory organic substances remaining in the carbon-
treated secondary effluent, which was used as the feed water for the
electrodialysis pilot plant operation. The organics removal could be
attributed to the biological activity in the form of slime formation
in the membrane stack.
26
-------
TABLE 3
MAJOR ION REMOVAL AND SELECTIVITY DATA
(April 1970 through September 1970)
Ions
Na+
K+
NH4+
Ca++
Mg++
Cl"
NO/
so4=
YV
HP04=
HC03"
Feed
mg/1
118.5
11.8
12.1
62.6
12.7
109.9
12.4
71.1
14.6
19.1
268
Product
mg/1
92.9
8.6
8.5
35.5
7.2
61.6
7.5
53.3
11.7
14.0
200
% Removal
22
27
30
43
43
44
39
25
18
27
25
Selectivity
0.691
0.867
0.953
1.384
1.385
1.406
1.256
0.812
0.591
0.854
0.814
-------
TABLE 4
TYPICAL HEAVY METAL CONCENTRATION DATA
( May 6, 1970 }
Heavy Metal
Aluminum, mg/1 Al
Arsenic, mg/1 As
Boron, mg/1 B
Cadmium, mg/1 Cd
Chromium, mg/1 Cr
Copper, mg/1 Cu
Iron, mg/1 Fe
Lead, mg/1 Pb
Lithium, mg/1 Li
Manganese, mg/1 Mn
Nickel , mg/1 Ni
Zinc, mg/1 Zn
Feed Water
0.00
0.00
0.82
0.006
0.01
0.02
0.00
0.001
0.03
0.00
0.26
0.06
Product Water
0.00
0.00
0.73
0.006
0.01
0.02
0.00
0.001
0.02
0.00
0.16
0.02
Waste Brine
0.07
0.00
1.10
0.008
0.01
0.02
0.00
0.001
0.03
0.00
0.62
0.20
28
-------
TABLE 5
REMOVAL OF COD , TOC^! IDS , AND TURBIDITY
{ April 1970 through September 1970 )
Parameter
Total COD, mg/1
Dissolved COD, mg/1
Total TOC, mg/1
Dissolved TOC, mg/1
TDS, mg/1
Turbidity, JTU
Feed Water
7.9
6.2
2,2
1.9
536
1.0
Product Water
6.3
5.2
1.6
1.4
361
0.6
% Removal
20
16
27
26
33
40
(1) COD = Chemical oxygen demand.
(2) TOC = Total organic carbon.
(3) TDS = Total dissolved solids.
29
-------
SECTION VII
MEMBRANE LIFE
As described earlier, membrane fouling was the major problem en-
countered with the operation of the electrodialysis process for
wastewater demineralization. However, the fouling was not permanent,
since cleaning the membranes and spacers resulted in restoration of
the design performance conditions. Occasionally, the membranes with
extensive scaling had to be replaced because the scale would pene-
trate and could permanently damage the membranes. Also, the anionic
detergents present in the wastewater are known to damage anion mem-
branes permanently; however, the carbon adsorption pretreatment had
reduced the concentration of the methylene blue active substances
(MBAS) to an insignificant level of 0.05 mg/1 or less.
Membrane life data as measured by Ionics quality control tests are
shown in Tables 6, 7 and 8. The tests were performed on membranes that
had been in operation for 1,320, 4,100, 9,930, and 16,136 hours in
the 65 cell-pair stack and on membranes that had been utilized for
600 hours of operation in the 20 cell-pair stack. The quality control
tests did not indicate any significant changes in exchange capaci-
ties for anion membranes were reduced approximately 30 percent on the
tested membranes after one year of on-stream operation. The resist-
ances of the anion membranes were also found to increase substantially
in the membrane control tests. However, deterioration of the anion
membrane was not well demonstrated in the pilot plant operation since
the average conductivity removal during the 2.5 years of pilot plant
study always ranged between 28 to 33 percent, depending on the type
of membrane cleaning procedure being employed.
Based on the membrane performance during this study and adequate
maintenance and competent operating personnel, anticipated membrane
life will be 5 years, i.e., 20 percent replacement every year. This
figure was used to prepare the process cost estimate.
30
-------
TABLE 6
20 CELL—PAIR STACK MEMBRANE QUALITY CONTROL TESTS
Membrane
Batch No.
Time (hrs)
Leak Test
Mullen Burst (psi)
Resistance
(ohm-cm^)
Thickness (cm)
Water Content (%}
Strong Capacity
(meq/dgr)
Weak Capacity
(meq/dgr)
cation
773C
0
-
95
11.4
0.053
42.7
3.20
-
600 (1)
2 pin
holes
111
12.2
0.052
42.8
2.80
-
cation
811A
•o
126
12.5
0.056
48.8
1.42
j
600 (1)
tight
122
17.6
0.049
48.4
1.40
0.75
(1) Including 2 enzyme-detergent cleanings.
31
-------
TABLE 7
65 CELL-PAIR STACK CATION MEMBRANE QUALITY CONTROL TESTS
Membrane
Batch No.
Time (hrs)
Leak Test
Mullen Burst (psi)
Resistance
(ohm-cm2)
Thickness (cm)
Water Content (%}
Strong Capacity
(meq/dgr)
Weak Capacity
(meq/dgr)
cation
673A
0
-
133
12.3
0.054
46.4
2.61
-
1,320
tight
127
14.6
0.056
47.6
2.32
-
cation
673A
0
-
133
12.3
0.054
46.4
2.61
-
4,100
tight
136
13.4
0.054
47.8
2.60
-
cation
673A
0
-
133
12.3
0.054
46.4
2.56
-
g.gsoC1)
-
135
15.8
0.062
48.7
2.46
-
cation
671B
0
126
15.6
0.058
48.4
2.66
-
16,136(2)
(3)
135
12.5
0.054
48.6
2.73
-
(1) Including 30 enzyme-detergent cleanings.
(2) Including 115 enzyme-detergent cleanings.
(3) Leaking entire spacer outline.
-------
TABLE 8
65 CELL-PAIR STACK ANION MEMBRANE QUALITY CONTROL TESTS
Membrane
Batch No.
Time (hrs)
Leak Test
Mullen Burst
(psi)
Resistance
(ohm-cm?)
Thickness (cm)
Water Content (%}
Strong Capacity
(meq/dgr)
Weak Capacity
(meq/dgr)
anion
658A
0
-
119
12.9
0.058
44.2
1.67
-
1,320
tight
130
18.3
0.058
45.7
1.49
0.73
anion
659C
0
-
119
13.7
0.059
44.9
1.60
-
4,100
tight
137
16.8
0.054
46.6
1.68
0.69
anion
(1
0
-
-
-
-
-
-
-
)
9,930(2)
-
130
25.2
0.058
47.3
1.16
0.44
anion
(1
0
-
-
-
-
-
-
-
)
16,136(3
tight
146
20.6
0.058
48.9
1.02
-
(1) No visible batch no. identification, assume original analysis similar to batch no. 658A or 659C for
comparison purposes.
(2) Including 30 enzyme-detergent cleanings.
(3) Including 115 enzyme-detergent cleanings.
-------
SECTION VIII
PROCESS COST ESTIMATE
3
A cost estimate for a 37,850 m /day (10 MGD), single stage electro-
dialysis plant based on the operating results obtained at the Pomona
Research Facility is shown in Table 9. More specifically, the cost
estimate is based on the following parameters that were developed
from the Pomona pilot plant study:
(1) The influent of the electrodialysis plant is a carbon-
treated secondary effluent with a total COD level of
10 mg/1 or less;
(2) The influent TDS level is approximately 550 mg/1, and a
30 to 35 percent TDS removal is achieved by the single
stage electrodialysis plant;
(3) A weekly, enzyme-detergent and acid flush are required
for membrane cleaning;
(4) The membrane stack is disassembled once every two months
for membrane inspection and maintenance, and
(5) A five-year membrane life is assumed for an electro-
dialysis process for demineralizing wastewater.
The other factors that were used to determine the costs are included
in Table 9. The capital cost of $1,764,000 includes site prepara-
tion and 20 percent for contingencies. The cost of the carbon
adsorption pretreatment to reduce the secondary effluent total COD
from 40 mg/1 to 10 mg/1 or less is not included in the total process
cost estimate. The carbon adsorption pretreatment cost is estimated
to be about 9.6 tf/1000 gallons. The cost of brine disposal is also
not included in the process cost estimate. The brine disposal cost
may vary substantially from one plant to another depending on the
type of brine disposal and the geographical location of plant.
34
-------
TABLE 9
ELECTRODIALYSIS PROCESS COST ESTIMATE
37,850 m3/day ( 10 MGD ) PLANT
Amortl zati on of Capital
$1,764,000; 20 years at 6%
Operati on and Maintenance
Sulfuric Acid ($40/ton)
Enzyme-Detergent ($1.2/Kg)
Membrane Replacement (5 year life)
Maintenance Materials
Power ($0.01/kWh)
Labor (5 man-days/day)
1/1000 gallons
4.7
4.0
1.2
4.2
0.9
2.8
1.6
14.7
TOTAL 19,4
35
------- |