WATER POLLUTION CONTROL RESEARCH SERIES • 17040EUNO2/71
DEMINERALIZATION OF WASTEWATER
            BY THE
  TRANSPORT-DEPLETION PROCESS
  U. S. ENVIRONMENTAL PROTECTION AGENCY

-------
             WATER POLLUTION CONTROL RESEARCH SERIES
The Water Pollution Control Research Series describes the
results and progress in the control and abatement of pollution
in our Nation's waters.  They provide a central source of
information on the research, development, and demonstration
activities in the Environmental Protection Agency, through
inhouse research and grants and contracts with Federal,
State, and local agencies, research institutions, and
industrial organizations.

Inquiries pertaining to Water Pollution Control Research
Reports should be directed to the Head, Publications Branch,
Research Information Division, R&M, Environmental Protection
Agency, Washington, D. C. 20^60.

-------
             DEMINERALIZATION OF  WASTEWATER  BY THE
                   TRANSPORT-DEPLETION PROCESS
                              by
                  Southern Research Institute
                  Birmingham, Alabama  35205
                            for the
                ENVIRONMENTAL PROTECTION AGENCY
                         Project  #17040 EUN
                         Contract #14-12-812
                           February 1971
For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402 - Price 65 cents

-------
                EPA Review Notice
This report has been reviewed by the Environmental
Protection Agency, and approved for publication.
Approval does not signify that the contents neces-
sarily reflect the views and policies of the
Environmental Protection Agency, nor does mention
of trade names or commercial products constitute
endorsement or recommendation for use.
                     ii

-------
                         ABSTRACT
The transport-depletion process was investigated for
demineralizing clarified or filtered municipal secondary
effluent.  Wastewater containing 850 mg/1. total dissolved
solids and having a turbidity of 1 to 10 JTU, was deminer-
alized continuously for up to 500 hours in a pilot plant
operating at a rate of 3,800 gpd.

Although regenerated-cellulose membranes, generally
regarded as neutral membranes, were found to be satis-
factory in regard to physical durability, the coulomb
efficiencies attained with these membranes were only
0.14 to 0.28.  The major problems encountered in deminer-
alization of wastewaters by conventional electrodialysis,
fouling and scaling, were, however, largely overcome by
the use of a special anion-selective membrane and periodic
flushing of the stack with sodium chloride solution.  With
the special anion-selective membrane, current densities
up to three times the conventional limiting current density
could be used without precipitation of pH-sensitive salts
and coulomb efficiencies of 0.70 were obtained.

Membrane fouling, which caused the electrical resistance
of the demineralizer stack to increase, was largely over-
come by flushing the stack with a sodium chloride solution.
This flushing technique should be useful for cleaning
conventional electrodialysis stacks.

Cost estimates for a 10 mgd plant indicated a demineral-
ization cost of 25.7
-------
                         CONTENTS

                                                    Page
CONCLUSIONS                                           1
RECOMMENDATIONS                                       3

INTRODUCTION                                          5

PILOT-PLANT EQUIPMENT AND OPERATION                   9

   Equipment and Operation                            9
      Pretreatment of secondary effluent              9.

   Demineralizer Stack                               16
   Power Supply                                      17
   Instrumentation                                   17
   Typical Operating Procedure                       17

MEMBRANES                                            19
   Procedures for Evaluating Membranes               19

      Transference number                            20
      Electrical resistance                          20
      Thickness                                      20
      Electrolyte diffusion rate                     20
      Water permeability                             21

   Membrane Properties                               21
   Stability of Regenerated-Cellulose Membranes      23
RESULTS AND DISCUSSION                               27

   Chronology of Demineralization Runs               27
   Membrane Fouling                                  31

      Effect of fouling on cell-pair resistance      32
      Effect of membrane fouling on coulomb
        efficiency                                   34
   Treatments to Remove Fouling                      34
   Coulomb Efficiency                                42

      Neutral membranes                              42
      lonac IM-12 anion-selective membranes          43

   Alum Pretreatment of Wastewater                   45
   Effect of Removing Spacer Screens from
     Depleting Compartments                          46
   Effect of Solution Velocity and Product:Waste
     Ratio                                           47
   Ion-Removal Selectivity                           48

ECONOMIC CONSIDERATIONS                              51

ACKNOWLEDGMENTS                                      57

REFERENCES                                           59
                            -v-

-------
                              FIGURES


^'                                                          PAGE

1      DIAGRAM OF A TRANSPORT-DEPLETION STACK                  7

2      TRANSPORT-DEPLETION PILOT PLANT (IN TRUCK TRAILER)     10
       AT THE SHADES VALLEY SEWAGE TREATMENT PLANT OF
       JEFFERSON COUNTY

3      TRANSPORT-DEPLETION STACK AND CONTROL PANEL            11

4      FLOW DIAGRAM OF TRANSPORT-DEPLETION PILOT PLANT        12

5      LOCATION OF MAJOR ITEMS IN THE TRANSPORT-DEPLETION     13
       TRAILER-MOUNTED PILOT PLANT

6      CHANGES IN CELL-PAIR RESISTANCE DURING TYPICAL         33
       OPERATION

7      CELL-PAIR RESISTANCE AND FEED TURBIDITY  IN RUNS        39
       97-105

8      CELL-PAIR RESISTANCE AND FEED TURBIDITY  IN RUNS        40
       106-114

9      CELL-PAIR RESISTANCE AND FEED TURBIDITY  IN RUNS        41
       "115-122
                                -vi-

-------
                             TABLES
No.

1      Analysis of Wastewater from Shades Valley             14
       Sewage Treatment Plant

2      Composition of Typical Feed Solution Used             15
       in Pilot Plant

3      Properties of Membranes                               22

4      Decrease in Burst Strengths of Regenerated-           25
       Cellulose Membranes Exposed to Various Types
       of Water

5      Conditions and Results for Demineralization           38
       Runs 97-122

6      Conditions for and Results of Demineralization        47
       Runs 94-96

7      Ion-Removal Selectivity                               49

8      Capital Costs for 10 mgd Demineralization             53
       Plant

9      Operating Costs for Demineralizing 10 mgd of          55
       Secondary Wastewater
                                -Vll-

-------
                        CONCLUSIONS
This report includes work performed from June 28, 1968
to December 22, 1970, under Contracts 14-12-443 and
14-12-812.

The main conclusions are:

    - A process which uses cation-selective membranes
      combined with special anion-selective membranes
      (IM-12 membranes from lonac Chemical Company)*
      appears to be technically feasible for long-term
      demineralization of secondary sewage effluents
      if the stack is cleaned periodically by 'flushing
      with an 18% sodium chloride solution.

    - The estimated total cost for demineralizing
      secondary effluents by the above process is
      slightly less than the cost of demineralization
      by conventional electrodialysis combined with
      activated-carbon pretreatment  (estimated on the
      same basis).

Other conclusions are:

    - Neutral membranes made of regenerated cellulose
      retain essentially all of their burst strength
      after 3 months service in the stack, however, the
      coulomb efficiencies with these membranes were
      only 0.14 to 0.28, which makes their use econom-
      ically unattractive for demineralization.

    - The special anion-selective membrane, lonac IM-12,
      was useable with current densities up to three
      times the conventional limiting current density
      without encountering the usual precipitation of
      pH-sensitive salts.  Coulomb efficiencies of
*Mention of proprietary equipment or products is for
 information only and does not constitute endorsement
 by the Environmental Protection Agency.
                            •™ 1 —

-------
about 0.70 were obtained with this membrane when
demineralizing wastewater in a long-term operation
of more than 500 hours.

Removing the spacer screens from the depleting
compartments of the stack did not improve the
effectiveness of removing the materials that
caused fouling.

Variations in solution velocity from 5 to 14 cm/sec
and in product:waste ratio from 5:1 to 15:1 had no
significant effects on cell-pair resistance or
coulomb efficiency.

Bicarbonate and calcium ions were removed to a
greater degree than were other ions.
                      -2-

-------
                      RECOMMENDATIONS
It is recommended that the IM-12 membranes and the
sodium chloride cleaning technique be considered for
use in various electromembrane processes for demineral-
izing secondary sewage effluents.
                            -3-

-------
                       INTRODUCTION
Use of the conventional electrodialysis process for
demineralization of secondary sewage effluents, while
promising, presents a number of operating problems.
Major problems are temporary fouling of membranes by
colloidal organic materials, permanent fouling of anion-
selective membranes by both colloidal and dissolved
organic materials in the feed, and scaling at the anion-
selective membranes because of precipitation of pH-
sensitive compounds.  Brunner1 found that the first two
problems could be minimized by removing essentially all
of the organic material from the wastewater by clarifying
it and treating it with activated carbon.  More recent
work at the Pomona, California Advanced Waste Treatment
Facility showed that granular carbon treatment of
effluent from a well-operated secondary plant was
sufficient for operation of electrodialysis.  The cost
of the activated-carbon treatment, however, is significant.
The transport-depletion process is a variant of electro-
dialysis that offered promise for solving some of the
problems encountered with electrodialysis without the
necessity of activated carbon pretreatment.

The transport-depletion process is similar to conventional
electrodialysis in that it removes soluble salts from
water by passage of electric current through an array of
ion-permeable membranes and thin solution compartments.
The term "transport depletion" could properly be applied
to all electromembrane processes/ because their common
basis for demineralization is the depletion  (or enrich-
ment) that occurs because of the transport of ions across
ion-exchange membranes.  However, the name "transport
depletion" generally has been used to designate the
electromembrane process in which near-neutral and cation-
selective membranes are used, in contrast to conventional
electrodialysis in which highly anion-selective and cation-
selective membranes are used.

Even though we used a highly specialized type of anion-
selective membrane in this program as well as near-neutral
membranes, we have referred to both processes as transport
depletion in this report.
                            -5-

-------
A diagram of a transport-depletion stack is shown .in
Figure 1.  The alternating cation-selective and neutral
membranes cause the solution in every other compartment
to become less concentrated in ions and that in the other
compartments to become more concentrated in ions when
direct electric current is passed through the stack.
When direct electric current is passed through a transport-
depletion assembly, depleted and enriched boundary layers
form at the opposite sides of the cation-selective mem-
branes, but there is no such effect at the nonselective
neutral membranes.  The neutral membranes serve only to
separate the depleted and enriched streams from each
other and, in doing so, they create alternating deplet-
ing and enriching compartments.

Previous studies at Southern Research Institute had shown
that transport depletion offered several advantages for
demineralizing waters that normally cause operating
problems when conventional electrodialysis is used.  The
scaling of anion-selective membranes that is often
encountered when electrodialysis is used for deminerali-
zation is avoided since no conventional anion-selective
membranes are used.  Transport depletion permits the use
of much higher current densities and greater throughput
rates than can be used in conventional electrodialysis.
Neutral membranes, such as the regenerated-cellulose
films -that are ordinarily used in the transport-depletion
process are lower in cost than the anion-selective mem-
branes that they replace—less than $0.04 per square foot
compared to $1 to $3 per square foot.  The major dis-
advantage of the transport-depletion process is that it
requires substantially more electrical energy for the
removal of a given amount of salt than does electro-
dialysis.

Southern Research Institute carried out an initial
feasibility study of the demineralization of wastewater
 (secondary sewage effluent) by the transport-depletion
process under Contract No. 14-12-443.  The primary
objectives of the work were to determine on a pilot-
plant scale the technical feasibility of the transport-
depletion process for reducing the dissolved solids of
secondary effluent and to determine whether the transport-
depletion process could overcome the problems encountered
when conventional electrodialysis is used to treat waste-
water.  The results obtained under Contract No. 14-12-443
                            -6-

-------
jm
          CONCENTRATED
             STREAM
DEMINERALIZED
   STREAM



1
C


NA+-





i






i
\
\
s
\
\
\
\
\
\
-^*m

t
N


^NA+-



CL-^

i

^
—
—
—
~~"
—








C


+ NA+-






-CL"
-ONE CE
i

\
\
^
\
\
\
\
:L

i
N


-NA+-


>
CL"^
L PAIR-
j

—
—
—
1 "
—


-r—
->•
| B

\
C


+ NA+-



-CL"

i

k
N
\
"\
\
\
\
\
\
\


+~





A

ANODE FEED (NACL TYPICAL) CATHODE
RINSE RINSE
     C   CATION-SELECTIVE MEMBRANE
     N   NEUTRAL MEMBRANE

      Figure  1.   Diagram of a Transport-Depletion Stack
                            —1 —

-------
indicated that the transport-depletion process had
potential for demineralizing wastewater at a lower, cost
than electrodialysis if certain problems could be solved,
A summary report covering the work under Contract No.
14-12-443 was prepared but not published because the
work was continued under Contract No. 14-12-812.

The primary objectives of the work under Contract No.
14-12-812 were to investigate in a transport-depletion
pilot plant promising techniques for overcoming the
effects of fouling of membranes, to evaluate neutral
membranes that had not been studied, and to demonstrate
the technical capability of operation for a period of
at least 500 hours with the more promising operating
conditions and membranes, as determined in initial
studies.  Other objectives included obtaining engineer-
design data and providing preliminary cost estimates
to indicate the economic promise of transport.depletion
for demineralizing wastewater.
                            -8-

-------
            PILOT-PLANT EQUIPMENT AND OPERATION
A mobile pilot plant was constructed to allow us to carry
out on-site studies of the treatment of secondary effluent
from a sewage plant.  The site chosen was the Shades Valley
Sewage Plant in the southwest part of Jefferson County,
Alabama.  This sewage-treatment plant, built in 1960, uses
the activated-sludge process for secondary treatment of
sewage and has a design capacity of 10 mgd.  Under normal
conditions, the flow rate varies from 5 to 8 mgd.  In
periods of heavy rainfall, infiltration into the sewerage
system increases the maximum flow rate to 15 mgd.  The
Shades Valley plant treats domestic sewage and a small
amount of industrial sewage from the southern part of the
Birmingham metropolitan area.  The industrial sewage some-
times includes some plating wastes that contain cyanide
ions and metal ions, such as iron, zinc, cadmium, copper,
and chromium.

Equipment and Operation

The transport-depletion pilot plant was installed in a
truck van trailer, 32 feet long with standard rear doors
and a 42-in. side door.  Figures 2 and 3 are photographs
of the trailer and the transport-depletion stack and
control panel in the trailer, respectively.  A flow
diagram of the pilot plant is shown in Figure 4.  The
location of major items of equipment in the trailer is
shown in Figure 5.  The pilot plant was designed to
operate continuously with minimal attention.

Pretreatment of secondary effluent

Analyses of samples of the wastewater effluent from the
Shades Valley plant taken at 2-hour intervals over a
24-hour period gave the results shown in Table I.  During
this period, the plant flow rate ranged from 5.5 to 9.6 mgd.

The total' concentration of ions in the Shades Valley waste-
water was lower than in typical wastewater, which normally
contains 700 to 850 mg/1.  Therefore, concentrated solutions
of calcium chloride, sodium sulfate, and sodium bicarbonate
were metered into the wastewater to augment the dissolved
solids content so that the feed water used in our studies
contained about the same concentrations of ions as an
average secondary effluent.
                            -9-

-------

Figure 2.  Transport-Depletion Pilot Plant (in truck trailer)  at the
      Shades Valley Sewage Treatment Plant of Jefferson County

-------
Figure 3.  Transport-Depletion Stack and Control Panel
                          -11-

-------
ALUM
FEEDER
SECONDARY


1
DC


I




* *

WASTEWATER
                    FEEDER
           300-GAL.
         FLOCCULATION
            TANK
SAND FILTERS
                   275-6AL.
                  SURGE TANK
                                                                                                          DEM1NERAUZED
                                                                                                          WATER
                                                                                                           ENRICHED WATER
                                                                                                           TO WASTE
                              CHEMICAL FEEDERS
                               TO AUGMENT TDS
              AQUA-CHEM
                WD6-2
                STACK
5-MICRON
CARTRIDGE
 FILTER
ENRICHED SOLUTION
SURGE AND RECYCLE
    SYSTEM
                 Figure  4.   Flow Diagram  of Transport-Depletion Pilot  Plant

-------
                         1.  FLOCCULATION TANK                 11.
                         2.  ALUM METERING PUMP                12.
                         3.  ALUM FEED TANK                   13.
                         
-------
                          Table I
               Analysis of Wastewater from
           Shades Valley Sewage Treatment Plant
                                     Concentration
     	Component	        range, mg/1.

     Calcium                        16-30
     Magnesium                      4-7
     Total hardness  (as CaC03)      78-90
     Bicarbonate                    127-152
     Sulfate                        32-42
     Chloride                       15-20
     Iron                           0.03-0.34
     Manganese                      0.4-0.6
     Chromate                       0.05-0.08

     Total ions                     194-252

     Total solids                   216  (170-380)a
     Suspended solids               16  (16-56)a
     Dissolved solids               200  (150-320)a
     Biochemical oxygen demand      28  (17-36)a
     a.  Shades Valley plant records.
The secondary effluent was supplied to the pilot plant at
a pressure of about 65 psig.  When alum pretreatment was
used, an alum solution was added continuously to the
incoming wastewater stream with a metering pump as the
water flowed into a 300-gal., galvanized-steel, floccu-
lation tank.  The tank provided a retention time of
35-90 minutes to permit flocculation by the alum.  When
alum pretreatment was not used, the flocculation tank
was by-passed and the wastewater flowed directly to the
sand filters.  In either case, sodium hypochlorite
solution was added continuously with a metering pump to
add about 5 mg/1. of chlorine to the wastewater before
it entered the sand filter.
                           -14-

-------
Two sand filters were used in parallel.  These were
Aqua-Guard MSF-2048 filters with epoxy-lined steel tanks
20 in. in diameter.  Each contained 31 in. of 0.7-nuti
anthracite supported by a shallow layer of 0.3-0.6-mm
calcite, graded-size gravel, and a distributor plate.
The filters could be used singly or.in parallel.  The
flow rate through these filters did not exceed 2 gpm/ft2
at the maximum feed rate used.  The chlorinated and
filtered wastewater flowed into a 275-gal. polyethylene
surge tank at a rate that maintained a small overflow
from the tank.

The wastewater was pumped from the surge tank and the
flow was divided into two equal streams.  A concentrated
calcium chloride solution was added continuously with a
metering pump to one stream and a solution of sodium
sulfate and sodium bicarbonate was added to the other.
Each stream flowed through a chamber with internal
baffles to cause mixing.  The two wastewater streams
with the added salts were combined, and fed to the
demineralizer stack.  The composition of a typical feed
solution is shown in Table 2.
                         Table 2.
           Composition of Typical Feed Solution
                    Used in Pilot Plant


                                 Concentration,
     	Component	   	mg/1.	

     Calcium                           121
     Magnesium                           7
     Total hardness  (as CaCO3)         331
     Sodium                            112
     Potassium                          11
     Ammonium                            7

     Bicarbonate                       263
     Chloride                          173
     Sulfate                           111
     Phosphate                          11

     Total ions                        816

     Total dissolved solids            850
                           -15-

-------
As shown in Figure 4, part of the enriched stream was
recirculated to the stack after the addition of makeup
feed solution.  This was done to make the solution
velocities equal in the enriching and depleting compart-
ments of the stack even though the amount of depleted
solution from the stack was several times greater than
the amount of enriched solution going to waste.  This
method of operation resulted in essentially equal
pressures on both sides of each membrane and prevented
membrane distortion and poor distribution of solutions
within the stack.  The flow rates of the recycled
enriched stream and of the makeup feed solution were
adjusted to give the desired ratio of deminefalized
product to enriched or waste stream.

Demineralizer Stack

The demineralizer stack was an Aqua-Chem WD6-2 with
8 x 23.6 in.  (20 x 60 cm) membranes.  The effective
area of each membrane was 116 in.2   (750 cm2).  The
stack was normally assembled with 10 cell pairs.
lonac Chemical Company MC-3470 cation-selective mem-
branes were used throughout the entire pilot-plant
program.  During most of the study Union Carbide
Corporation Zephyr Z neutral membranes (made of re-
generated cellulose) and lonac Chemical Company IM-12
special anion-selective membranes were evaluated in the
pilot plant.  The stack was assembled with alternating
cation-selective and neutral or anion-selective membranes/
with a cation-selective membrane adjacent to both electrode
compartments.  The cell separators were of the sheet-flow
type, 0.040 in.  (0.10 cm) thick, and served as perimeter
gaskets and membrane spacers.  The inner part of the cell
separators consisted of 11-mesh plastic screen.  The
hydraulic pressure drop through the stack was about 3 psi
when the superficial solution velocity was 10 cm/sec.  A
superficial solution velocity of 10 cm/sec resulted in a
flow of 0.277 gpm (1050 ml/min) through each compartment.

Electrode compartments, 0.10 in. (0.25 cm) thick,
separated the platinized-titanium anode and the 316
stainless steel cathode from the cell pairs in the
stack.
                           -16-

-------
A solution of sodium carbonate was recirculated from a
50-gal. polyethylene tank through the anode compartment
to neutralize the hydrogen ions formed at the anode and
to provide a highly conductive electrical path between
the anode and the cell pairs.  Similarly, a solution of
acetic acid was recirculated through the cathode compart-
ment to neutralize the hydroxyl ions formed at the cathode
and to provide an electrical path between the cathode and
the cell pairs.  Acetic acid was selected for use in"the
cathode rinse solution because of its buffering capability.

Power Supply

The direct current for the stack was obtained by recti-
fication of three-phase 230-volt alternating"current
from isolation transformers.  The a-c input voltage to
the three-phase full-wave rectifier bridge was controlled
by a continuously adjustable autotransformer.   The d-c
power to the stack could be varied continuously from 0
to 380 volts, with a maximum current of 50 A.   The iso-
lation transformers were found to be necessary to prevent
electrical interference with measurements of conductivity
and pH.

Instrumentation

A Leeds and Northrup 12-point recorder was used to record
primary data during pilot-plant operation.  Data that were
recorded continuously were:  temperature of the feed and
depleted  (product) streams, conductivities of the feed
and depleted streams, stack current, and d-c voltage
applied to the stack.

Typical Operating Procedure

In typical operation of the pilot plant, the filtered,
chlorinated wastewater flowed continuously into the surge
tank at a rate sufficient to maintain a small overflow.
The feed-control valve was adjusted to give the desired
flow rate of feed solution to the depleting compartments
of the stack.  The flow rates of the recycled enriched
solution and of the makeup feed solution were adjusted
to provide the desired product-to-waste ratio and to
give a flow rate into the enriching compartments that
was the same as that into the depleting compartments.
The metering pumps for the supplemental salt solutions
were adjusted to give the desired concentrations of
                           -17-

-------
salts in the feed solution to the stack.  Conductivity
readings were used as indications of the salt concen-
tration in the feed solution.  When lonac IM-12 anion-
selective membranes were used, the enriched stream was
maintained at a pH of 2.5-3.5 by adding 1 N hydrochloric
acid to the enriched-stream recycle tank with a metering
pump.

The input voltage to the rectifier bridge was adjusted
to provide the desired d-c current or voltage 'to the
stack.  Except for slow changes caused by fouling of the
membranes, steady-state conditions were attained within
15 minutes after adjustments of voltage or flow rates.

Samples of the feed and depleted streams were taken
periodically for determinations of TDS and concen-
trations of the principal ions.  Turbidity measurements
on the feed solution were made with a Hach Laboratory
Turbidimeter, Model 2100.  The sand filters were back-
washed and the solutions for augmentation of TDS of
the feed were replenished as needed without interrupting
the continuous operation of the demineralizer stack.
                           -18-

-------
                         MEMBRANES
The properties of various types of films were determined
in an effort to find better neutral membranes.  A truly
neutral membrane is one in which the transference numbers
of ions are the same as their transference numbers in
solution, e.g_. , a cation transference number  (t+) o.f
0.39 in sodium chloride solutions.  Other desirable
characteristics of neutral membranes for the proposed
use are low electrical resistance, good mechanical
strength and stability, and low cost.

In prior work on the transport-depletion process,2'3 more
than 80 materials were evaluated at Southern Research
Institute for possible use as neutral membranes.  Films
made of regenerated cellulose were found to have the
most promising combinations of properties for use as
neutral membranes.  However, the transference numbers
(t ) of most of these materials were significantly
higher than 0.39 in sodium chloride solutions, especially
at the salt concentrations of about 0.015 N_ which are
typical of wastewater.                    ~~

The search for better neutral membranes was continued
in work under the present contract.  Eight potential
neutral membrane materials that were not available for
our earlier studies were evaluated.  None of the near-
neutral membranes investigated were found to have a
completely satisfactory combination of properties.  As
a result we investigated the lonac Chemical Company IM-12
anion-selective membrane material.  lonac IM-12 is a
special type of anion-selective material that is
relatively porous and permits hydrogen ions to pass
through it at a greater rate than through conventional
anion-exchange membranes.  These characteristics reduce
the tendency of the membranes to become fouled or scaled
compared to conventional anion-exchange membranes.

Procedures for Evaluating Membranes

Where possible, the methods given in the Office of Saline
Water Test Manual1* were used for the evaluation of mem-
branes.  Other test methods were developed when required.
                           -19-

-------
      Transference number

The procedure used to determine the transference number
of a membrane was similar to Method 602.1 in the Office
of Saline Water Test Manual.  A membrane specimen was
mounted in a concentration cell so that it separated
two sodium chloride solutions of different concentrations.
Each solution flowed through the cell at a rate of 1 l./min
to maintain constant salt concentrations and to minimize
polarization effects.  The potential between silver-silver
chloride electrodes in the two sodium chloride solutions
was measured with a potentiometer.  The transference
number, t+, was calculated by dividing the measured
potential by the theoretical potential for the two
sodium chloride concentrations.

      Electrical resistance

The a-c electrical resistance of membranes was deter-
mined by a procedure similar to Method 601.1 in the
Office of Saline Water Test Manual.  A membrane specimen
was clamped between two identical half-cells, each contain-
ing a platinized platinum electrode.  Sodium chloride
solutions of the same concentration flowed through each
half-cell.  The resistance between the electrodes was
measured with the membrane specimen in place and again
after it had been removed.  The membrane resistance was
calculated from the difference in the resistance of the
cell with and without the membrane and the area of mem-
brane exposed to the solution.

      Thickness

The thickness of a wet membrane was measured at a minimum
of five locations with a Randall and Stickney dial micro-
meter having a 6 DW weight and a pressure foot 0.75 in.
in diameter.

      Electrolyte diffusion rate

The procedure for determining the electrolyte diffusion
rate consisted of clamping a membrane specimen between
two compartments of a Plexiglas cell and adding 1.5 N
sodium chloride solution to one compartment and a   ~
measured amount of deionized water to the other.  The
contents of both compartments were stirred.  The concen-
tration of sodium chloride in the compartment that
                           -20-

-------
initially contained deionized water was determined by
conductivity measurements at 10-min intervals for 1 hr.
From these data, the electrolyte diffusion rate was calcu-
lated as the equivalents of sodium chloride diffused per
square centimeter of membrane per hour and the membrane
diffusion coefficient was calculated from the electrolyte
diffusion rate and the membrane thickness.

      Water permeability

The water permeability of a membrane was determined by
placing a membrane specimen in a pressure-filter apparatus
with .water on top of the specimen and applying pressure
to the water.  The rate of water transfer through the
membrane was measured and the water permeability, in
ml/min-cm2-psi, was calculated.

Membrane Properties

The properties of the various membranes that were evaluated
in this program for use in the transport-depletion process
are given in Table 3.

All of the neutral membranes shown in Table 3 had at "least
one property which made them less than ideal for use as
membranes in the transport-depletion process.  The trans-
ference numbers, t+, of most of the membranes were
significantly higher than the desired value of 0.39 at
the salt concentration typical of wastewater (about
0.015 N).

The properties of the small sample of Yumicron Y-101
membrane obtained originally indicated that this material
should be evaluated as a neutral membrane in the pilot
plant.  However, when Y-101 membranes from a new lot were
installed in the stack, leakage between the enriching and
depleting compartments was excessive, and the water
permeability of the new membranes was found to be 0.26
ml/min-cm2-psi, compared to a value of 0.03 ml/min-cm2-psi
for the original sample.  The supplier replaced the Y-101
membranes with Y-601 membranes, but these were not suit-
able for use because of the high t+ value (0.71).

The Yumicron Y-121, Y-203, and Y-621 membranes were dropped
from consideration because of their high permeabilities to
water.  The Yumicron Y-201 membrane was dropped from con-
sideration because of its extremely high electrical resist-
ance.
                           -21-

-------
        Table 3
Properties of Membranes
Transference number, t
Membrane 0.1 vs 0.2 N 0.01 vs 0.02 N 0.005 vs 0.01 N
designation NaCl NaCl NaCl
Neutral membranes
Yumicronc Y-101d 0.41 0.52 0.57
Yumicronc Y-101e -
Yumicron0 Y-121 0.42 0.56 0.60
Yumicronc Y-202 0.36 - 0.58k
Yumicronc Y-203 0.40 0.48 0.51
Yumicronc Y-601 - 0.71
Yumicronc Y-621 0.40 0.50 0.55
Eastmanf HT-00 0.37 - 0.37
Eastmanf HF-35 0.38 - 0.40
UCC^ Zephyr Z 0.41 0.58 0.70
AHTh 4465-A2-4.9 0.38 0.47
Ion-exchange membranes
lonac IM-121 . 0.04 0.02 0.04
lonac MC-34703 0.98 0.96 0.96
Electrical
resistance, Electrolyte Membrane
a-c, in diffusion diffusion
0.05 N NaCl, rate," Thickness, coefficient,
ohm cm2 eq/cm -hr wet, cm cm2/sec

6 161xlO~E 0.010 287xlO~8
12
39,000 -
4 231xlO~5 0.006 274xlO~e
26 83xlO~s 0.009 134xlO~8
10
4 171xlO~E 0.009 284xlO~6
6 132xlO~5 0.011 268xlO~8
15 70xlO~E 0.012 149xlO~8
23 32xlO~; 0.021

12 14xlO~5 0.014 38xlO-e
35
Water
permeability
ml/mir.-
cm^-psi

0.03
0.26
5.1
0
0.14
0.001
5.4
0.001
0.001
-

-
a. Determined from measurements of concentration potentials between the concentrations shown.
b. From 1.5 N NaCl solution to deionized water.
c. Yuasa Battery (America), Inc.
d. Original sample.
e. Lot installed in stack.
f. Eastman Chemical Products, Inc.
g. Union Carbide Corporation, regenerated cellulose casing.
h. Arthur H. Thomas Co. dialyzer tubing, 4465-A2-4.9.
i. lonac Chemical Company, anion-selective.
j. lonac Chemical Company, cation-selective.
k. Value is questionable because of extremely high resistance









of membrane.











-------
As shown in Table 3, all of the values for the Eastman
HT-00 and HF-35 membranes were very good.  However,
satisfactory seals could not be obtained with either
type of Eastman membrane in the pilot-plant stack.
Although the stack clamping bolts were tightened to
twice the normal torque, there was leakage of solution
between the enriching and depleting compartments of the
stack.  Both of these Eastman membranes were thin and
easily torn.  Neither of these membranes is suitable
for use in the Aqua-Chem WD6-2 stack.

The properties of the AHT 4465-A2-4.9 regenerated-
cellulose membrane material that had been used in our
prior work2'3 on the transport-depletion process were
slightly better than those of the Zephyr Z material.
However the AHT material is not available in sufficient
width for use in the Aqua-Chem WD6-2 stack.

Consideration of the properties of the neutral membranes
listed indicated that the Union Carbide Corporation
Zephyr Z regenerated-cellulose membrane was the best
available neutral membrane for use in the pilot-plant
studies.  The Zephyr Z membrane was, therefore, used
in the pilot-plant studies with neutral membranes.

The lonac IM-12 anion-selective membrane was evaluated
in the pilot plant because of its unique properties.
The IM-12 membrane is relatively porous and permits
hydrogen ions to pass through it at a greater rate
than through conventional anion-exchange membranes.
The manufacturer states that this porosity causes the
IM-12 membrane to have less tendency to be fouled by
organic and inorganic materials than conventional anion-
exchange membranes.

The lonac MC-3470 cation-exchange membrane was selected
for use in the pilot-plant studies because of its very
good physical and electrochemical properties.   (Table 3)

Stability of Regenerated-Cellulose Membranes

Prior to this study, a question was raised regarding the
ability of regenerated cellulose to withstand biological
attack by the secondary effluent unless the wastewater
were chlorinated before it entered the demineralizer
stack.  Also, metal ions, such as those from plating
wastes, might promote oxidative degradation of the
neutral membranes.
                           -23-

-------
Specimens of the AHT 4465-A2-4.9 and UCC Zephyr Z mem-
branes were placed in the final clarifiers of the Shades
Valley and Patton's Creek sewage treatment plants, both
of which are operated by Jefferson County.  As mentioned
earlier, the Shades Valley plant sometimes treats sewage
containing metal plating wastes, but the Patton's Creek
plant treats only domestic sewage.  The Mullen burst
strengths of the specimens were measured after 15 days
of exposure to the wastewater.  The results of these
measurements, Experiments 1 and 2 in Table 4, showed
that the burst strengths of the specimens of both mem-
brane materials decreased by at least half in 15 days
of exposure.  There was no significant difference in the
rates of degradation at the two plants, which indicates
that the small amounts of metal ions from the plating
wastes that are treated by the Shades Valley plant did
not affect the physical stabilities of the membranes.

Studies of the effects of metal ions and chlorination on
the degradation of the neutral membranes were made in our
laboratory.  Specimens of membranes were exposed to
chlorinated and continuously aerated Birmingham tap water
to which salts had been added to provide concentrations
of 1 mg/1. of chromium, 1 mg/1. of copper, 1 mg/1. of
iron, and 3 mg/1. of manganese.  Chlorinated and aerated
Birmingham tap water was used as a control.  The results
of Experiments 3 and 4 in Table 4 indicate that this
water, with or without the addition of the metal ions,
caused no significant loss in the burst strengths of the
membranes.

Membrane specimens exposed to continuously aerated Shades
Valley wastewater in the laboratory  (Experiment 5) were
degraded to about the same degree as those exposed at
the plant.  The purpose of Experiment 6 was to determine
whether chlorination of the wastewater would reduce the
rate of degradation of the neutral membranes.  Since
chlorine was added only once or twice each day, the
residual chlorine concentration varied from 0 to 0.7
mg/1., and usually reached a value of zero overnight.
The results indicate that this method of addition of
chlorine was not effective in reducing the degradation
of the regenerated-cellulose membranes.  The experiment
with chlorinated wastewater was repeated  (Experiment 7)
with continuous addition of chlorine to provide a residual
chlorine level of 0.3 to 1 mg/1.  After 8 weeks of exposure
to this wastewater, the AHT 4465-A2-4.9 membrane retained
93% of its original burst strength and the UCC Zephyr Z
membrane retained 86%.

                           -24-

-------
                            Table 4

 Decrease in Burst Strengths of Regenerated-Cellulose Membranes
               Exposed to Various Types of Water
Exper-
iment
Type of water
Plant clarifier exposures

  1    Shades Valley wastewater

  2    Patton's Creek wastewater

Laboratory experiments

  3    B'ham tap water, aerated,
         chlorinated

  4    B'ham tap water, aerated,
         chlorinated, with metal
         ions added0
  5    Shades Valley wastewater,
         aerated,

  6    Shades Valley wastewater,
         aerated, chlorinated
         intermittently

  7    Shades Valley wastewater,
         aerated, chlorinated
         continuously

Used in demineralization stack

       Shades Valley wastewater,
         chlorinated
                         ..   ,  Decrease in burst
                     Length of     strength, %
                     exposure,	—-*-—-—	
days
                         15

                         15
                         14
                         14
                         56
                         90
AHTa



 59

 50
         57
Zephyr Z



   63

   76
14
14
1
47
4
76
           36
                   14
a.  Arthur H. Thomas Co. dialyzer tubing, 4465-A2-4.9.

b.  Union Carbide Corporation Zephyr casing, Z.

c.  Water contained:  1 mg/1. of Cr, 1 mg/1. of Cu, 1 mg/1. of Fe,
    3 mg/1. of Mn.

                             -25-
                                            AWBERC LIBRARY  U.S. EPA

-------
These results indicated that the neutral membranes could
be expected to remain serviceable in transport-depletion
stacks for periods of 2 months or more, if the wastewater
feed to the demineralizer were chlorinated.

The UCG Zephyr Z membranes that were used in the demineral-
izer stack in the pilot plant exhibited excellent resist-
ance to degradation.  The membranes that were removed
from the stack for burst tests had been in the stack for
3 months.  For most of this period, the stack was filled
with primary or secondary effluent containing 0.1 to 3.0
mg/1. of residual chlorine.  The average Mullen burst
strength of these membranes was 23 psi, which was the same
as that found for control  (unexposed) specimens taken from
the same roll of regenerated-cellulose film.  A second set
of Zephyr Z membranes that was subjected to similar con-
ditions in the stack for 2.5 months showed no evidence of
degradation.

The excellent physical condition of the Zephyr Z membranes
after 3 month's use in the demineralizer stack strongly
suggests that these membranes would be useable for a much
longer time.
                           -26-

-------
                  RESULTS AND DISCUSSION
All results given in this report are from deminerali-
zation runs made with the stack containing ten cell
pairs.  In all runs, sodium hypochlorite was added to
provide a residual chlorine content of 1-3 mg/1. in the
feed solution to the stack.  The runs are numbered in
chronological order.  After Run 25 a run number was
assigned to each calendar day of pilot-plant operation.
Letters after a hyphen in some run numbers indicate
periods of operation with different conditions.  Union
Carbide Corporation Zephyr Z neutral membranes  (made of
regenerated cellulose) were used in Runs 1-39.  lonac
Chemical Company IM-12 special anion-selective membranes
were used in all subsequent pilot-plant runs.

Chronology of Demineralization Runs

Runs 1-10 were short runs  (about 1 hour each) with
secondary effluent to determine the effects of super-
ficial solution velocities of 5, 10, and 15 cm/sec and
current densities of 3.3 to 43.3 mA/cm2 on demineral-
ization performance.  Membrane fouling was encountered
in this series of runs, which limited the amount of
reliable data that could be obtained on the effects of
these two operating parameters.

Runs 11 and 12 were not typical of normal operation
because the wastewater from the sewage treatment plant
was essentially primary effluent that had been subjected
to one additional stage of clarification.  Because of
heavy rainfall, infiltration of ground water into the
sewerage system greatly diluted the raw sewage entering
the plant, and secondary treatment was stopped to con-
serve fuel gas used for compressing air.

Runs 13-15 were made after the sewage treatment plant
resumed normal secondary treatment.  These runs were
made without disassembling and cleaning the stack
following Run 12.  The secondary wastewater feed to the
demineralizer was treated with alum to reduce the tur-
bidity and the fouling of the membranes.  However, the
fouling that remained from Runs 11 and 12 severely
hampered the operation of the pilot plant.
                           -27-

-------
Runs 16-18 were made with alum-treated secondary waste-
water after the stack had been disassembled and cleaned.
The alum dosage was 35 to 50 mg/1., which laboratory jar
tests showed was optimum for this water.  Runs 16-18,
lasting 7 to 11 hours each, were made on consecutive
days with overnight shutdowns.  Although membrane foul-
ing caused the electrical resistance of the stack to
increase, no operating difficulties were encountered
in Runs 16-18.

Runs 19-25 were made over a period of 139 hours of almost
continuous operation.  The duration of the individual
runs ranged from 5.5 to 32.7 hours.  Before each run,
one of several treatments to reduce the fouling on the
membranes was tried.

When pilot-plant operations were resumed under Contract
No. 14-12-812, Runs 26 and 27 were made to check the
operability of the pilot plant with a sodium chloride
feed solution.  The stack and other pilot-plant equipment
were found to be in good operating condition.

Runs 28-39 were made to determine the effectiveness of
flushing the stack with short bursts of nitrogen for
controlling the fouling of membranes.  Nitrogen flush-
ing was tried with the stack current off and with the
polarity of the stack current reversed.  The technique
of nitrogen flushing with the polarity of the stack
current reversed was more beneficial in reducing mem-
brane fouling.  However, it was not adequate to  prevent
substantial increases in stack resistance and decreases
in coulomb efficiency over long periods of operation.

lonac IM-12 anion-selective membranes were installed
in the stack after Run 39 and were used in all subsequent
pilot-plant runs.  The results of Runs 40-42 with a sodium
chloride feed solution showed that coulomb efficiencies
0.90 or higher were obtained with the IM-12 membranes.

Runs 43-55 were made to determine the effectiveness of
cleaning the stack, and thus reducing the effects of
membrane fouling, by circulating a solution of an enzyme-
active laundry material through the stack at 12-hr
intervals. However, the high pH of the calcium chloride
solution used to increase the calcium concentration of
the wastewater feed solution caused precipitation in
the feed lines and in the depleting compartments of the
                           -28-

-------
stack and caused the data from these runs to be unreliable,
The high pH was caused by the use of a new lot of calcium
chloride that was found to contain more than the usual
amount of calcium hydroxide.

The purpose of Runs 56-65 was the same as for Runs 43-55.
The results of Runs 56-65 indicated that circulation of
a solution of an enzyme-active laundry material through
the stack for 1 hr at 12-hr intervals was effective for
cleaning the stack and reducing the cell-pair resistance.

Runs 66-70 were made to determine the effectiveness of
a concentrated solution of sodium chloride in place of
the solution of enzyme-active laundry material for clean-
ing the stack at 12-hr intervals.  The results of Runs
66-70 indicated that a concentrated sodium chloride
solution was more effective than the enzyme-active
laundry material in cleaning the stack and reducing
cell-pair resistance.

The purpose of Runs 71-74 was to determine the effect
of variations in solution velocity and product:waste
ratio on coulomb efficiency and cell-pair resistance.
The erratic results obtained in Runs 71-74 indicated
that the runs should be repeated before drawing con-
clusions about the effects of solution velocity and
product:waste ratio on demineralization performance.

Runs 75-85 were made to determine whether alum treat-
ment of the wastewater would significantly reduce the
problems with fouling in the stack.  During these runs
the stack was flushed at 24-hr intervals with a sodium
chloride solution.  The results with alum-treated waste-
water indicated that, although the use of alum reduced
the turbidity of the feed solution to the stack, the
cell-pair resistances before and after flushing with
the sodium chloride cleaning solution were not sig-
nificantly different from those obtained in runs with
wastewater that had not been treated with alum.

Runs 86-89 were made to determine whether removing the
spacer screens from the depleting compartments would
facilitate the removal of the membrane-fouling material
from the compartments by flushing.  The results of the
runs indicated that removing the spacer screens from
the depleting compartments was not beneficial.
                           -29-

-------
Runs 90-93 were made to determine the effect of solution
velocity and productrwaste ratio on demineralization
performance.  However, it was found that the lonac IM-12
membranes in the stack were irreversibly fouled.  After
Run 93 the membranes in the stack were replaced with new
lonac MC-3470 cation-selective and lonac IM-12 anion-
selective membranes.

In Runs 94-96 it was determined that variations in
solution velocity and product:waste ratio over the ranges
investigated had no significant effects on coulomb effi-
ciency or cell-pair resistance."

Runs 97-122 were made over a period of approximately
600 hours of continuous operation to determine the
long-term operating characteristics of the process and
to obtain operating data for use in estimating the costs
of demineralization.  During Runs 97-118 the stack was
flushed with a sodium chloride solution at 24-hr inter-
vals.  During these runs, cleaning the stack with the
sodium chloride solution was effective in restoring the
coulomb efficiency and cell-pair resistance to near the
original values".  The data from Runs 97-118 indicated
that the average cell-pair resistance* increased from
about 1050 ohm cm2 to about 1550 ohm cm2 and remained
at about this level although there were some changes
with the turbidity of the feed.  The average coulomb
efficiency values were essentially constant at about
0.70 during the last 300 hr of this series.

During, Runs 119-121 the long-term operation was con~
tinued with the same conditions except that the stack
was flushed with tap water instead of the sodium chloride
solution.  The tap water was less effective than the
sodium chloride solution in cleaning the stack and there
was a significant decrease in coulomb efficiency during
these runs.
*The time-average cell-pair resistance was calculated
 for each run.
for each run.


                          -30-

-------
In Run 122, flushing the stack with a sodium chloride
solution was less effective in cleaning the stack than
it had been in the runs before water was tried for
flushing.

Membrane Fouling

Fouling of. membranes or the effects of fouling were
observed in most runs with wastewater, regardless of
the operating conditions or the type of pretreatment
used.  Whenever the stack was opened for inspection a
brown, flocculent, slightly slimy material was found
to be fairly evenly distributed throughout each deplet-
ing compartment.  When the stack was inspected after
being flushed with solutions of an enzyme-active laundry
material or of sodium chloride the amount of fouling
material was small, but some was always present.  The
fouling material was found both in the mesh of the
spacer screens and on the sides of-both of the membranes
facing the depleting compartments.  Both the UCC Zephyr Z
neutral membranes and the lonac IM-12 anion-selective
membranes became more heavily coated with the fouling
material than the lonac MC-3470 cation-selective mem-
branes.  Most of the flocculent fouling material could
be removed easily by rinsing with water; a thin layer
of finely divided material remained on the neutral and
anion-exchange membranes, but this was easily removed
by light rubbing.  The enriching compartments contained
only traces of the fouling material.

When the fouling material was heated in a flame, most
of the material was consumed and only traces of ash re-
mained, indicating that it was mostly organic material.
Polishing'filters with 5-micron cartridges collected
only a small quantity of material and had no significant
effect on the turbidity of the feed solution which
indicates that the particle size of the material that
caused fouling was less than 5 microns.  Presumably,
the material consisted mostly of negatively-charged
particles of organic compounds or colloids that were
too large to pass through the neutral or anion-selective
membranes.
                           -31-

-------
      Effect of fouling on cell-pair resistance

The most important effect of the membrane fouling was an
increase in the cell-pair resistance and an associated
decrease in stack current for demineralization  (when
the stack voltage was held constant).  The RCp values
in this report are normalized for a feed-solution temper-
ature of 25°C and a log mean concentration of 675 mg/1.
(0.01 N) in the depleting compartments.  The RCp values
were normalized to 25°C by applying the temperature
coefficient for the resistance of the solution being
demineralized.   (Data reported by Casolo and Leitz5
showed that the temperature coefficient of resistance
for an electrodialysis stack was essentially the same
as that for the solution in the, stack.)  The normalities
of the solutions were calculated from the concentrations
in mg/1. by use of the average equivalent 'weight of the
salts in solution (67.5 g/eg), as determined from analyses
of samples of feed and product (demineralized) water.
Prior work3 with the transport-depletion process showed
that the product of cell-pair resistance and the log
mean normality of the depleting solution, RCpN, is
constant over the range of concentrations used in this
program.  Normalizing the cell-pair resistance values
to a given mean concentration permits comparisons of
cell-pair resistances with a given stack regardless of
the feed concentration and the degree of demineralization.

The cell-pair resistance generally increased rapidly
during the first 1 to 2 hr of operation after the mem-
brane fouling had been removed and then continued to
increase at a fairly constant, but much lower rate, for
the duration of the run.

Figure 6 shows the change in cell-pair resistance during
typical operation in the continuing series of runs.

There was no significant difference in the rates of
fouling-with UCC Zephyr Z neutral membranes or lonac
IM-12 anion-selective membranes.

Although no quantitative correlation was found between
the turbidity of the feed solution and the rate of
increase in cell-pair resistance (or rate of fouling),
the cell-pair resistance did increase when the turbidity
of the feed solution increased.
                           -32-

-------
I
co
to
I
    X
    o
    LU
    o
    to
to
LU
    Q.
     I
    UJ
    O
      1600
      1400
      1200
      1000
       800
       600
       400
       200
          8
                    RUN  98
                                           RUN 99
           12
16    20
24

 0
12
16   20
                                4     8

                              TIME,  HR

Figure 6.  Changes in Cell-Pair Resistance During Typical Operation

-------
      Effect of membrane fouling on coulomb efficiency

 The overall results of the demineralization runs  indicate
 that membrane  fouling generally caused  some reduction in
 coulomb  efficiency.  However, correlation between coulomb
 efficiency and the degree or rate of membrane  fouling was
 not good.  For example, in some runs the coulomb  efficiency
 increased or remained constant although the increases in
 cell-pair resistance indicated that severe fouling was
 taking place.  In contrast, during other runs,  the coulomb
 efficiency decreased although' the increases in cell-pair
 resistance indicated the rate of fouling was much lower.

 There is reason to believe that the membrane fouling
 should cause lower coulomb efficiencies.  If large
 negatively-charged particles, such as colloids or large
 organic  anions, were present in the feed water, they
 would be electrically driven toward the anode  but would
 be stopped by  the neutral or anion-selective membranes
 and would deposit on these membranes.   This deposit of
 negatively-charged particles would cause the neutral or
 anion-selective membranes to become more cation selective
 and to have a  higher effective cation-transference number.
 A higher cation-transference number for the neutral or
 anion-selective membranes would result  in lower coulomb
 efficiency because the theoretical coulomb efficiency is
 the difference in the cation-transference numbers of the
 cation-selective and the neutral or anion-selective mem-
 branes.  The fouling material that was  observed on the
 cation-selective membranes and the cell separators
 probably adhered to them because of the slightly  sticky
 nature of the  fouling material.  The material  on  the
 cation-selective membranes and cell separators would
 increase the electrical resistance of the stack,  but
 probably would have little effect on the coulomb
 efficiency.

 Treatments to Remove Fouling

Two types of  remedial treatments were found to be helpful
 in removing  fouling material from the stack without dis-
assembly.  They were flushing with short bursts of
nitrogen and  flushing with a solution of an enzyme-active
laundry  detergent or sodium chloride.
                           -34-

-------
The procedure for nitrogen flushing was to admit three to
six 1-sec bursts of nitrogen (at a pressure about 4 psi
greater than .the pressure of the feed solution) into the
depleting compartments in a 1-min period while the normal
solution-flow rates were maintained.  It was hoped that
the bursts of gas would cause vigorous agitation in the
depleting compartment and loosen the fouling material so
that it would be carried out of the compartment, by the
feed solution.  Nitrogen was used for flushing because
it was readily available under pressure, but compressed
air probably would have been equally satisfactory.
Nitrogen flushing had almost no effect when the stack
power was on, which indicates that the fouling material
was electrically charged.  Nitrogen flushing with the
stack power off resulted in an average reduction of about
14% in the cell-pair resistance in Runs 13-25, but did
not reduce the cell-pair resistance to its original value.
In these runs, the nitrogen flushing was carried out after
the stack had been in operation for periods of 5 to 33 hr.

In Runs 29-39, nitrogen flushing was carried out at 20-min
intervals in an attempt to prevent the membranes from
becoming heavily fouled.  During Runs 29-31, in which the
nitrogen flushing was done with the stack power off, the
average cell-pair resistance increased from cycle to cycle
(a cycle was one 20-min period of operation followed by
nitrogen flushing).  The increase in the average cell-
pair resistance over a number of cycles during a run was
about 1.3%/hr.  Six 1-sec nitrogen bursts were no more
effective than three bursts during a flushing period.
Typically, a considerable quantity of finely divided
material was removed by the first burst of nitrogen/ a
little by the second, but practically none by the
additional bursts.

In Runs 32-39, the polarity of the stack current was
reversed while nitrogen flushing was being carried out.
Nitrogen flushing with current reversal removed consider-
able quantities of material from the stack even though
the stack had been flushed with nitrogen with the stack
power off immediately prior to current reversal.  During
Runs 32-39, the average cell-pair resistance increased
about 0.4%/hr.  This rate of increase in cell-pair
resistance was considered to be intolerable as it would
result in doubling the resistance in about 10 days.  It
was concluded that nitrogen flushing did not provide
adequate removal of membrane fouling to permit long-
term operation.
                           -35-

-------
Runs 56-65 were made to determine the effectiveness of
removing the membrane fouling, and thus reducing the
cell-pair resistance, by circulating a solution of an
enzyme-active laundry material through the stack at
12-hr intervals.

The procedure for cleaning the stack was to turn off
the electrical power and solution flows to the stack
and then to circulate 3 gal. of a solution of an enzyme-
active laundry detergent (trade name "BIZ") through the
depleting compartments of the stack for 1 hr, at a solution
velocity of about 10 cm/sec.  The concentration of BIZ in
the cleaning solution was 15 g/gal., which was that
recommended for soaking laundry.  When cleaning was com-
pleted, the stack was returned to service with the con-
ditions being used for the run.

During Runs 56-65, the average cell-pair resistance after
the stack was cleaned with the BIZ solution was about
1030 ohm cm2 compared to about 1520 ohm cm2 before cleaning,
or a reduction of 32%.

In Runs 66-70, the same procedure was used for cleaning
the stack except that the cleaning solution was 3 gal.
of an 18%  (3.5 N) sodium chloride solution instead of
the enzyme-active detergent solution.  It was thought
that solutions of high ionic strength, such as a concen-
trated sodium chloride solution, might break the colloidal
structure of the fouling material and make it more soluble
or more easily removed from the membranes.  During these
runs, the average cell-pair resistance after the stack
was cleaned with the sodium chloride solution was about
750 ohm cm2 compared to about 1400 ohm cm2 before cleaning—
a reduction of 46%,  The cell-pair resistance of 750 ohm
cm2 was about the same as the values obtained with clean
membranes and comparable operating conditions.

The coulomb efficiencies in Runs 56-70 were not signif-
icantly affected by cleaning the stack with either the
sodium chloride solution or the solution of enzyme-
active laundry material.

The results of the experiments discussed above indicated
that flushing the stack with a 3.5 N sodium chloride
solution was the most effective method for cleaning the
                           -36-

-------
stack to reduce membrane fouling without disassembly
of the stack.  Therefore, sodium chloride flushing was
used for cleaning the stack in all subsequent demineral-
ization runs.

The effectiveness of the sodium chloride flushing in
cleaning the stack was demonstrated in over 450 hr of
continuous operation during Runs 97-116.  The conditions
and results for Runs 97-122 are given in Table 5.

The variations in cell-pair resistance and feed turbidity
during Runs 97-122 are shown in Figures 7, 8, and 9.  The
curves in these figures are smoothed curves -obtained from
the data during each run.  The dotted lines connect points
just before and after the stack was flushed.  In Runs 97-
122, the interval between cleanings was increased from
12 to 24 hr in order to increase the fraction of elapsed
time that the pilot plant was on stream.  During Runs-
97-116, the average cell-pair resistance before the
stack was cleaned was 1830 ohm cm^ and after cleaning
it was 620 ohm cm2—a decrease of 67%.  These data
indicate that cleaning the stack with the sodium chloride
solution was effective for removing most of the fouling
material from the membranes.

In Runs 117 and 118, which were a continuation of the
long-term operation, the cell-pair resistance after
cleaning was about 800 ohm-cm  compared to values of
about 620 in previous runs.

Tap water  (with nothing added) was used to flush the
stack in Runs 119-121, but it was less effective than
the sodium chloride solution in reducing the cell-pair
resistance.  Sodium chloride solution was used for
flushing in Run 122, but it only reduced the cell-pair
resistance to 1250 ohm cm , compared with an average
of 620 ohm cm2 in Runs 97-116.

As discussed above, the sodium chloride cleaning procedure
was effective in reducing the cell-pair resistance to
about 620 ohm cm  after cleaning during most of the
long-term operation.  As shown in Figures 7, 8, and 9
the maximum cell-pair resistance reached in a run appeared
to depend on the turbidity of the feed solution.  In Runs
107 and 108, the turbidity of the feed increased to about
                            •37-

-------
                                        Table 5.
                Conditions and Results for Demineralization Runs 97-1223


Run
97
98
99
100
101
102
103
104
105
106
107
108
109
, 110
w 111

-------
   2800
Q
I—1
00
UJ
UJ
0.
      8

      6
97     98    99    100    101     102
                      RUN NUMBER
103    104
105
    Figure  7.   Cell-Pair Resistance and Feed Turbidity in Runs 97-105

-------
o
I
       2800



       2400
    o
    o  2000
    UJ
    o
    to
       1600
    1C  1200
    CL
     I
    UJ
    O
    =)
    f-
        800
400
         10

          8

          6
    CO
    oe.
    UJ
    UJ
    u_
     106
112
113
114
             107    108    109     110    111
                             RUN  NUMBER

Figure 8.  Cell-Pair Resistance  and Feed-Turbidity in Runs '106-114

-------
   2800
   2400
o
o  2000
UJ
o
01
a.
I
   1600
   1200
    800
g   400
H-
t—i

Q
i—i
CO
UJ
UJ
     10

      8

      6
         115
116
                                                        122
                        117    118    119   120-   121

 "~                            RUN NUMBER

Figure 9.  Cell-Pair  Resistance and Feed Turbidity in Runs  115-122

-------
8 JTU and the maximum cell-pair resistance increased
to about 2400 ohm cm2, which was higher than in any pre-
ceding runs in the series.  When the feed turbidity
decreased to about 2 JTU in Runs 109-114, the maximum
cell-pair resistance during each run reached a fairly
constant value of about 2000 ohm cm2.  The feed turbidity
was high and was increasing during Runs 117-121 (Figure 9),
and the maximum cell-pair resistances increased to 2400
to 2600 ohm cm2.  We believe the maximum cell-pair resist-
ances were higher in Runs 117-121 because of the high
turbidity of the feed and because the stack was flushed
with water instead of with the sodium chloride solution.
The increase in the cell-pair resistance after cleaning
in Runs 119-121 probably was caused by the failure of
the water flushing to clean the stack effectively.

It is concluded that cleaning the stack with the 3.5 N
sodium chloride solution at 24-hr intervals reduces the
effects of membrane fouling so that demineralization can
be carried out at reasonable rates for extended periods.
This technique of cleaning the stack by flushing with
a concentrated sodium chloride solution may be equally
effective for removing organic fouling from conventional
electrodialysis stacks and reverse osmosis demineralizers.

Coulomb Efficiency

      Neutral membranes

The coulomb efficiencies that were obtained with UCC
Zephyr Z membranes and sodium chloride feed solutions
ranged from 0.28 to 0.36, which were in good agreement
with the value of 0.33 that was predicted from the trans-
ference numbers of the membranes, solution concentrations,
and productrwaste ratio.  The cation-transference number,
t+, of the cation-selective membrane (0.98) minus the t+
of the neutral membrane  (0.62) is equal to'the theoretical
or. maximum value of coulomb efficiency (0.36) for the
transport-depletion process with the lonac MC-3470
cation-selective and UCC Zephyr Z neutral membranes and
the solution concentrations typical of wastewater.  Our
prior work indicated that a product:waste ratio of 10:1
will cause the coulomb efficiency to be about 10% lower,
or about 0.33.
                           -42-

-------
The coulomb efficiencies in Runs 28-39 with Zephyr Z
membranes and wastewater decreased from 0.26 to 0.14.
This decrease in coulomb efficiency indicated that
membrane fouling was severely affecting the membranes.
The low coulomb efficiencies obtained with the Zephyr Z
membranes indicated that better membranes must be found.
As discussed in the section on membranes, the properties
of other membrane materials were determined in an effort
to find better neutral membranes.  However, none of the
materials investigated appeared to be better than the
Zephyr Z membranes in overall properties.

      lonac IM-12 anion-selective membranes

lonac IM-12 anion-selective membranes were used in place
of the Zephyr Z membranes in Runs 40-122.  Coulomb
efficiencies of 0.90 to 0.99 were obtained when 0.012 N
(700. mg/1.) sodium chloride solutions were demineralized
with the IM-12 membranes.  A decrease in the pH of the
depleted stream indicated that the limiting current
density with the IM-12 membranes was about 3.5 mA/cm2
with-a solution velocity of 10 cm/sec and a product:waste
ratio of 10:1.  However, when the enriched stream was
acidified to about pH 3 as recommended by lonac Chemical
Company, much higher current densities  (6 to 10 mA/cm2)
could be used without the precipitation of pH-sensitive
salts that is usually encountered with anion-selective
membranes at the limiting current density.

The coulomb efficiencies obtained in the first runs with
IM-12 membranes and wastewater. feed  (Runs 43-47) ranged
from about ,0.70 to 0.90.  The coulomb efficiencies de-
creased in succeeding runs until the range of coulomb
efficiencies was only 0.50 to 0.62 in Runs 81-:85.  The
coulomb efficiencies obtained with these membranes when
demineralizing sodium chloride solutions were 0.46 to
Q;.62, compared to the values of 0.90 to 0.99 when the
membranes were new.

The stack was disassembled and the cation-transference
numbers, t+, of membranes from the stack were determined.
The t+ values for the lonac MC-3470 cation-selective mem-
branes ranged from 0.97 to 0.99.  The average t+ value
of 0.98 for the MC-3470 membranes from the stack was
the same as that for new MC-3470 membranes.  The t
values for the lonac IM-12 anion-selective membranes
from the stack ranged from 0.26 to 0.35, with an average
                           -43-

-------
value of 0.32, compared to a value of 0.05 for new IM-12
membranes.  This change in the t+ values of the IM-12
membranes would be expected to cause the maximum attain-
able coulomb efficiency to decrease from about 0.93
(0.98-0.05) to about 0.66 (0.98-0.32).

In an attempt to restore the original t+ values of the
IM-12 membranes,-two membranes from the stack were
cleaned by a procedure similar to that which is used to
clean electrodialysis stacks used for demineralization
of whey.  The membranes were treated with a solution of
hydrochloric acid (pH 2) for 20 min, rinsed with water,
and treated with a solution of sodium hydroxide (pH 13)
for 20 min.  This cleaning procedure did not significantly
change the t+ values of the membranes.

These results indicated that the IM-12 membranes had been
so badly fouled by exposure to wastewater that the fouling
could not be easily removed to restore the electrochemical
properties of the membranes.

New IM-12 anion-selective and MC-3470 cation-selective
membranes from lonac Chemical Company were installed in
the stack before the long-term operation  (Runs 97-122)
was begun.  As shown by the data in Table 5, the average
coulomb efficiency decreased from 0.65 in Run 97 to 0.46
in Run 103, so that the average coulomb efficiency for
Runs 97-103 was 0.57.  The sodium chloride solution used
to flush the stack in Run 104 was acidified to pH 2.5
with hydrochloric acid.  The coulomb efficiency increased
to 0.70 after this flushing and remained in the range of
0.65 to 0.73, for an average value of 0.70, during Runs
105-118, although unacidified sodium chloride solutions
were used for flushing the stack in all of these runs.
(The addition of the acid to the sodium chloride cleaning
solution did not appear to enhance its efficacy in re-
ducing cell-pair resistance even though the coulomb
efficiency was increased.)

The coulomb efficiencies decreased slightly to an average
of 0.68 in Runs 119-121, in which the stack was flushed
with water containing no sodium chloride.  Flushing the
stack with sodium chloride solution in Run 122 did not
restore the coulomb efficiency to its previous level.
                           -44-

-------
We conclude that average coulomb efficiencies of about
0.70 can be maintained during long periods of operation
when demineralizing wastewater with lonac IM-12 and
MC-3470 membranes if the stack is flushed at 24-hr
intervals with an 18% (3.5 N) sodium chloride solution.

Alum Pretreatment of Wastewater

Runs 75-85 were made to determine whether alum treatment
of the wastewater would significantly reduce the problems
of fouling in the depleting compartments of the stack.
An alum solution containing 75 g/1. of commercial "17%
alum" was added continuously to the incoming wastewater
stream with a metering pump as the water flowed into
the 300-gal. tank.  The tank provided a retention time
of about 90 min to permit flocculation by the alum.  The
alum-treated wastewater flowed from the flocculation tank
to the sand filters and feed-solution system.  The total
dissolved solids concentration of the wastewater was in-
creased to about 800 mg/1. by the addition of calcium
chloride, sodium sulfate, and sodium bicarbonate.

The alum solution was added at a rate to give a concen-
tration of 30 mg/1. of alum  (as aluminum sulfate) in
the incoming wastewater.  This concentration of alum
reduced the turbidity of the wastewater to about 0.6
JTU in laboratory jar tests.  The stack was cleaned at
24-hr intervals by flushing with a sodium chloride
solution in the usual manner.

The data for Runs 78-85  (with alum-treated wastewater)
show that the cell-pair resistances after the stack was
cleaned ranged from about 680 to 970 ohm cm2.  These
values are about the same as the range of 600 to 900
ohm cm2  (after cleaning) which was found in Runs 66-69,
which were made without alum treatment of the wastewater.
Also, there was little difference between the coulomb-
efficiency range of 0.43 to 0.65  (after cleaning) in
Runs 78-85 and the range of 0.45 to 0.60  (after cleaning)
in Runs 66-69.  The results from these runs indicated
that, although pretreatment with alum reduced the
turbidity of the feed solution entering the stack, it
did not significantly reduce the problem of fouling in
the stack.
                           -45-

-------
Effect of Removing Spacer Screens; from Depleting Compart-
ments

Runs 86-89 were made to determine whether removing the
spacer screens from the depleting compartments would
facilitate removal of the fouling material from the
compartments by flushing.  The screens were removed
from the depleting-compartment spacer frames/ leaving
only the perimeter gaskets and the distribution disks.
During assembly of the stack, proper location of the
distribution disks was maintained by the use of guide
tubes which were removed after the stack assembly had
been clamped together.  The other conditions for Runs
86-89 were the same as those described above for Runs 78-
85, including alum treatment of the wastewater.  During
Runs 86 and 87, the hydraulic pressure in the depleting
compartments was maintained 3 psi higher than the pressure
in the enriching compartments to keep the membranes from
touching and reducing the solution flow rate through the
•depleting compartments.  In Runs 88 and 89, the pressure
differential was increased to 6 psi.

The results from Runs 86-89 showed that operation without
spacer screens in the depleting compartments resulted in
generally slightly higher cell-pair resistances, before
cleaning, than the corresponding values in Runs 78-85
 (with spacer screens).

However, the cell-pair resistances after cleaning were
significantly higher in Runs 86-89, without spacer screens,
than in Runs 78-85 with spacer screens.  The coulomb
efficiency values were slightly lower in the runs without
spacer screens than in the runs with spacer screens.

These results indicate that removing the spacer screens
from the depleting compartments made the flushing less,
rather than more, effective for removing the fouling
material from the compartments.  It is probable that the
higher cell-pair resistances that were obtained when the
screens were removed were a result of the presence of
high-resistance depleted boundary layers in the deplet-
ing compartments.  The spacer screens promote turbulence
in the solutions, which tends to decrease the thickness
of the boundary layers and reduce the cell-pair resistance.
                           -46-

-------
Effect of Solution Velocity and Product;Waste Ratio

Runs 94-96 were made to determine the effect of changes
in solution velocity and product:waste ratio on cell-pair
resistance and coulomb efficiency.  The feed solution for
these runs contained about 810 mg/1. of sodium chloride
in Birmingham tap water.  The wastewater feed solution
was not used for these experiments because membrane foul-
ing caused rapid changes in cell-pair resistance and other
parameters, which would have masked the variations caused
by the changes in solution velocity and product:waste
ratio.  The conditions for and results of Runs 94-96 are
given in Table 6.
                          Table 6
               Conditions for and Results of
                Demineralization Runs 94-96
Solution Product:
velocity, waste
Run cm/sec ratio
94
95-A
-B
-C
96-A
-B
5
14
10
10
10
10
10:1
10:1
10:1
5:1
15:1
10:1
Current Cell-pair
density, resistance,3
mA/cm ohm cm2
5.4
11.9
9.6
9.0
9.3
9.2
261
327
319
333
330
343
Coulomb
efficiency
0.86
0.88
0.84
0.74
0.83
0.82
a.  Normalized to a feed solution temperature of 25°C
    and a mean concentration of 585 mg/1.  (0.01 M) of
    sodium chloride in the depleting compartments.
The data in Table 6 show that coulomb efficiencies of
about 0.83 and cell-pair resistances of about 320 ohm cm2
were generally obtained.  However, in Run 94, with a
solution velocity of 5 cm/sec, the cell-pair resistance
was only 261 ohm cm2; and in Run 95; with a product:waste
'ratio of 5:1, the coulomb efficiency was only 0.74.  The
reason for these two anomalous results is not known.
Because of the generally small variations in coulomb
                           -47-

-------
efficiency and cell-pair resistance observed with the
various conditions, it was concluded that variations in
solution velocity from 5 to 14 cm/sec and in producttwaste
ratios from 5:1 to 15:1 had little effect on cell-pair
resistances or coulomb efficiencies.

Ion-Removal Selectivity

In the transport-depletion process, all ions are not
necessarily removed from the feed solution to the same
degree.  The fraction of each ion that is removed de-
pends on the solution composition and the types of mem-
branes used.

Chemical analyses were made on seven sets of feed and
product samples taken during approximately 600 hours of
continuous pilot-plant operation with lonac IM-12 anion-
selective membranes (Runs 97-122).  Ion-selectivity data
from these seven sets of samples are given in Table 7.
The fractions of all ions removed were based on the total
normalities of the corresponding feed and product samples
assuming no ions other than those shown were present in
significant quantities.  The combination of lonac MC-3470
cation-selective and lonac IM-12 anion-selective membranes
had significantly higher selectivity for the removal of
bicarbonate and calcium ions than for other ions.  The
lowest selectivities were for the removal of sodium and
phosphate ions.  Although there was considerable scatter
in the data, the results indicate that there was no
significant change in the ion-removal selectivity during
the approximately 600 hours of continuous operation with
these membranes.  This indicates that the membrane fouling
that caused an increase in stack resistance had little
effect on the selectivity characteristics of the mem-
branes.

Only a few chemical analyses were made of the demineralized
(or product) wastewater from runs with the Zephyr Z mem-
branes.  The results of the analyses that were made indi-
cated that there was no significant selective removal of
any ion.
                           -48-

-------
                        Table 7
               Ion-Removal Selectivity3
    Ion
Bicarbonate
Calcium
Ammonium
Potassium
Magnesium
Sulfate
Chloride
Sodium
Phosphate
   Fraction of specific ion removed
divided by fraction.of all ions removed
       Average           Range
        1.42
        1.21
        1.00
        0.92
        0.62
        0.80
        0.67
        0.56
        0.36
1.33-1.49
  97-1.41
  60-1.20
  56-1.70
  16-1.09
  52-0.94
  57-0.83
0.38-0.74
0.04-1.36
a.  Based on analyses of seven sets of samples.
                         -49-

-------
                  ECONOMIC CONSIDERATIONS
The results obtained in our pilot-plant operations with
IM-12 membranes indicate  that cleaning the stack with a
sodium chloride solution makes continuous operation
practical with feed water having a turbidity in the range
of 1 to 10 JTU.  Occasional shutdowns to permit more
thorough cleaning of the stacks might be necessary.

Estimates of the costs of demineralization using the
fouling-resistant IM-12 membranes were made for comparison
with those published by Brunner1 for demineralization of
wastewater using electrodialysis membranes designed for
treatment of brackish water.  The same conditions were
assumed insofar as was possible when making the estimates.
Our cost estimates were made with the same assumptions
about the costs of piping, instrumentation, contingencies,
and similar items so that they could be compared directly
with the costs given by Brunner.  The prices of equipment
and materials were obtained from published information or
quotations from suppliers.  It was assumed for both cases
that some degree of coagulation and clarification of the
secondary wastewater would be required and that activated-
carbon pretreatment would be required for brackish water
membranes but not for the IM-12 membranes.  No credit
was given for the improved water quality obtained when
carbon treatment is used.

The operating conditions used for the cost estimates
were based on the data obtained in our pilot plant
during approximately 600 hours of operation during
Runs 97-122.  The coulomb efficiency assumed for the
cost estimates was 0.70, which was the average value
obtained in Runs 105-118.  A conservative value of
500 was assumed for the polarization parameter, i/N,
(current density in mA/cm2, divided by the log mean
normality in the depleting compartments) although the
average value obtained in Runs 97-118 was 620.  The
assumed cell-pair voltage of 7.1 volts was the average
value in Runs 97-118.
                           -51-

-------
The other conditions assumed for the cost estimates
were:

      solution velocity - 10 cm/sec
      product:waste ratio - 10:1
      production rate - 10 mgd for 365 days/yr
      cation-selective membranes - lonac MC-3470
      anion-selective membranes - lonac IM-12
      membrane cost - $3/ft2
      total dissolved solids concentration in feed
        water - 850 mg/1.
      total dissolved solids concentration in product
        water - 500 mg/1.
      average equivalent weight of dissolved salts - 67.5
      feed-water temperature - 25°C
      feed-water turbidity - 1 to 10 JTU

The feed water was assumed to have a composition similar
to that used in our pilot-plant studies.  It was assumed
that the stacks would be flushed with an 18% sodium
chloride solution for 1 hr at 24-hr intervals.

The details of the capital costs are given in Table 8.
For the IM-12 membrane system, 198 stacks, each with 150
cell pairs, would be required.  For the cost estimates
this number was increased by 15%, to 228 stacks, to provide
spare stacks for use during periodic maintenance of others.
The quoted price of the stacks without membranes was $5265
each.  Membranes  (at $3/ft2) cost $3883 per stack, increas-
ing the total cost to $9148 per stack.  For the brackish
water membrane system, stack costs are those of a different
manufacturer and date back to 1966.  These were not up-
dated and are likely to be slightly low.

The rectifier capacity needed for the IM-12 membrane
system would be 2900 kw, compared to 370 kw for the
brackish water membrane system because of the lower
cou-lomb efficiency and higher cell-pair resistance
values of the former.

The size of the building to house the IM-12 membrane
system was based on an area of 30 ft2 per stack, which
a floor plan of the projected plant indicated was more
than adequate for the type of stacks used.
                           -52-

-------
                                   Table 8.
                Capital Costs for  10 mgd Demineralization Plant


                                                         Cost, $
                                                               Brackish water
                    Item                     IM-12 membranes      membranes3
   Electrodialysis stacks with membranes       2,085,700         1,557,200
   Installation - 15% of component cost          312,900           233,600
   Rectifiers  - installed                        203,000            43,900
   Pumps and motors - installed                   73,000            98,500
   Piping  - 30% of installed cost of
     stacks and pumps                            741,500           566,800
   Instrumentation and controls -
,     10% of stacks and pumps                     247,200           188,900
<"  Acid storage tank                              11,000            11,000
i   Building -  at $13/ft2                         177,800b          114,300C
   Contingencies - 10% of investment             385,200           281,400
   Engineering - 5% of above                     211,900           154,800
   Interest on investment during
     construction - 4% of above                  178,000           130,000

                      Total capital cost       4,627,200         3,380,400
   a.  See Reference 5.

   b.  Assumed 30 ft2/stack.

   c.  Assumed 100 ft2/stack.

-------
The details of the operating costs are given in Table 9.
The costs of disposal of the concentrated waste stream
and of flocculation and filtration are not included
because these costs were considered to be common to
both processes.  The estimated cost of electrical
energy with the IM-12 membrane system is about 4.3£/kgal
higher than with brackish water membranes because of the
lower coulomb efficiency and higher cell-pair resistance.
The estimated cost of sodium chloride used to flush the
stack was based on the concentration and quantity used
in our pilot-plant studies.  In the pilot plant 1.31 Ib
of sodium chloride was used per kilogallon of product
water, and the cost of the sodium chloride, at 1.3£/lb
was 1.7C/kgal of product water.  We did not attempt to
determine the minimum amount or concentration of sodium
chloride solution that was effective in cleaning the
stack.  In a larger installation, the cost of sodium
chloride could almost certainly be reduced several-fold
by the use of a smaller volume of the sodium chloride
solution relative to the volume of product water or by
reusing the solution.  We believe the cost of 1.7£/kgal
for sodium chloride given in Table 9 in very conservative.

The amortization costs given in Table 9 were based on
a 25-yr period at 6% interest to make them more realistic.
The amortization cost based on 25 years and 4% interest
given by Brunner was updated to a basis of 25 years and
6% interest so that the two estimates would be more nearly
comparable.

The most recent data that was available on the costs
of activated-carbon treatment of wastewater was that
given by Smith and McMichael.6  They estimated the total
treatment costs for a 10 mgd activated-carbon treatment
plant to be lOolC/kgal and the capital cost of the plant
to be $1,600,000.  An adjustment of the amortization costs
from their basis of 4.5% for 25 yr to the basis of 6% for
25 yr increased the amortization cost from 3.0*/kgal to
3.4
-------
                                                Table  9.
                  Operating  Costs  for  Demineralizing 10 ragd  of Secondary  Wastewater
                                                                 Cost,  C/kgal  of  product water
                                                                                 Brackish water
                                Item                            IM-12 membranes      membranes3
       Electric  energy
          Rectifier  energy  at  90%  rectifier  efficiency
          and  power  at 0.7C/kwh                                        5.18              0.69
          Pumping energy  at 0.7
-------
The slightly higher cost for demineralization by con-
ventional electrodialysis is primarily due to the high
cost of the activated-carbon pretreatment.  If the
technique of flushing with sodium chloride solution is
as effective in cleaning brackish water membrane stacks
as it is in cleaning the IM-12 membrane stacks, activated-
carbon pretreatemnt probably could be eliminated and the
total operating cost of the brackish water membrane system
might be lower than for the IM-12 membrane system.
                           -56-

-------
                      ACKNOWLEDGMENTS
This research was directed by Mr. Robert E. Lacey and Mr.
Edward W. Lang, and was carried out by Mr. Everett L.
Huffman with the aid of Mr. Samuel L. Edward,

The cooperation and assistance of Mr. H. A. Snow, Jefferson
County Engineer, and Mr. W. M. Mclnnis, Superintendent of
the Shades Valley Sewage Treatment Plant, are gratefully
acknowledged.

The support of this study by Water Quality Office Contract
No. 14-12-812 and the advice and assistance of Dr. Carl A.
Brunner, Project Officer, are also acknowledged.
                           -57-

-------
                        REFERENCES
1.  C. A. Brunner, J. Water Pollution Control Federation,
    39_, 10, Part 2, Rl (Oct. , 1967) .

2.  Southern Research Institute, "Demineralization by
    Transport Depletion", Office of Saline Water Research
    and Development Report 80 (1963).

3.  Southern Research Institute, "Development of Trans-
    port-Depletion Processes", Office of Saline Water
    Research and Development Report 439  (1969).

4.  Bureau of Reclamation, "Test Manual  (Tentative) for
    Permselective Membranes", Office of  Saline Water
    Research and Development Report 77  (1964).

5.  A. J. Casolo and F. B. Leitz, 133rd  National Meeting
    of the Electrochemical Society, Boston, Massachusetts-,
    May, 1968.

6.  Robert Smith and W. F. McMichael, "Cost and Performance
    Estimates for Tertiary Wastewater Treating Processes",
    Report No. TWRC-9, Robert A. Taft Water Research Center,
    FWPCA, Cincinnati, Ohio,  (June, 1969).
                           -59-

-------
1
Access/on Number



5
2
Subject Field & Group


03A

SELECTED WATER RESOURCES

ABSTRACTS
INPUT TRANSACTION FORM


Organization
       Southern Research Institute
       Birmingham, Alabama
      Title
       DEMINERALIZATION OF WASTEWATER  BY THE  TRANSPORT-DEPLETION PROCESS
 10
      Authors)
       Lacey, Robert E., and
       Huffman, Everett L.
                                       16
Project Designation

  FWQA Contract Nos. 14-12-443 and  14-12-812
                                             21
                                                 Note
 22
Citation

 Final report to Water Quality Office under Contract Nos. 14-12-443 and 14-12-812, February, 1971.
 86  p, 9  tab, 9 fig,  5 ref.
 23
Descriptors (Starred First)

      *Desalination  processes,  *Transport depletion, *Wastewater treatment
      *Electrodialysis,  Demineralization, Membrane processes. Sewage effluents
 25
      Identifiers (Starred First)
 27
     Abstract
    The transport-depletion process was investigated for demineralizing municipal  secondary  effluent.   Waste-
    water containing 850 mg/1. total dissolved solids and having a turbidity of  1  to  10 JTU, was demineralized
    continuously for 500 hours in a pilot plant at a rate of 3800 gpd.

    Although regenerated-cellulose membranes were found to be satisfactory  in regard  to physical durability, the
    coulomb efficiencies attained with these membranes were only 0.14 to  0.28.   The major  problems  encountered  in
    demineralization of wastewaters by conventional electrodialysis, fouling and scaling,  were, however,  largely
    overcome by the use of a special anion-selective membrane and periodic  flushing of the stack with  sodium
    chloride solution.  With the special anion-selective membrane, current  densities  up to three times the
    conventional limiting current density could be used without precipitation of pH-sensitive  salts and coulomb
    efficiencies of 0.70 were obtained during the 500 hours of operation.

    Membrane fouling, which caused the electrical resistance of the demineralizer  stack to increase, was  largely
    overcome by flushing the stack with a sodium chloride solution.  This flushing technique should be useful for
    cleaning conventional electrodialysis stacks.

    Cost estimates for a 10 mgd plant indicated a demineralization cost of  25,7<:/kgal for  transport depletion
    compared to 27.8C/kgal for electrodialysis and required activated-carbon pretreatment.  The estimated cost
    for transport depletion is slightly lower because it does not require activated-carbon pretreatment.

    This report was submitted in fulfillment of Contract Nos. 14-12-443 and 14-12-812 under  the sponsorship of
    the Water Quality Office of the Environmental Protective Agency.
Abstractor
    Everett L. Huffman
                                Institution
                                 Southern  Research Institute.  Birmingham. Alabama
  Wd.102
  WRSIC
        (REV. JULY IB89I
                                                    SEND TO:  WATER RESOURCES SCIENTIFIC INFORMATION CENTER
                                                             U.S. DEPARTMENT OF THE INTERIOR
                                                             WASHINGTON. O. C. ZOI4O
                                                                                                 * CPOI l»e»-3S»O39

-------