EPA-60Q/2-77-170
August 1977
Environmental Protection Technology Series
REVERSE OSMOSIS FIELD TEST: TREATMENT OF
COPPER CYANIDE RINSE WATERS
Industrial Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL PROTECTION TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-77-170
August 1977
REVERSE OSMOSIS FIELD TEST:
TREATMENT OF COPPER CYANIDE RINSE WATERS
by
Kenneth J. McNulty
Robert L. Goldsmith
Arye Gollan
Sohrab Hossain
Donald Grant
Walden Division of Abcor, Inc.
Wilmington, Massachusetts 01887
for
The American Electroplater's Society, Inc.
Winter Park, Florida 32789
Grant No. 800945
Project Officer
John Ciancia
Industrial Pollution Control Division
Industrial Environmental Research Laboratory - Cincinnati
Edison, New Jersey 08817
INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
DISCLAIMER
This report has been reviewed by the Industrial Environmental Re-
search Laboratory - Cincinnati, U.S. Environmental Protection Agency, and
approved for publication. Approval does not signify that the contents
necessarily reflect the views and policies of the U.S. Environmental
Protection Agency, nor does mention of trade names or commercial products
constitute endorsement or recommendation for use.
-------
FOREWORD
When energy and material resources are extracted, processed, con-
verted, and used, the related pollutional impacts on our environment and
even on our health often require that new and increasingly more efficient
pollution control methods be used. The Industrial Environmental Research
Laboratory - Cincinnati (IERL-CI) assists in developing and demonstrating
new and improved methodologies that will meet these needs both efficiently
and economically.
This report is a product of the above efforts. It was undertaken to
demonstrate the effectiveness and economic feasibility of using reverse
osmosis for closed-loop control of metal finishing rinse wastes under
actual plant conditions. The reverse osmosis system concentrates the
chemicals for return to the processing bath while purifying the wastewater
for reuse in the rinsing operation. The results of the report are of value
to R&D programs concerned with the treatment of wastewaters from various
metal finishing, non-ferrous metal, steel, inorganic and other industries.
Further information concerning the subject can be obtained by contacting
the Metals and Inorganic Chemicals Branch of the Industrial Pollution
Control Division.
David G. Stephan
Director
Industrial Environmental Research Laboratory
Cincinnati
-------
ABSTRACT
Field tests of reverse osmosis (RO) were conducted on copper cyanide
rinse waters at two different sites: Whyco Chromium Co. and New England
Plating Co. At both sites, closed-loop treatment was used with plating
chemicals recycled to the bath and purified water recycled to the rinsing op-
eration. The objective of the tests was to establish, under actual plating
conditions, the feasibility of RO treatment for copper cyanide plating
wastes.
At the first field-test site (Whyco Chromium Co.), both the flux and
rejection of the membrane modules (duPont B-9 hollow fiber permeators) de-
clined within a period sufficiently short to make RO unattractive on the
basis of membrane replacement costs. The decline in performance is believed
to be the result of chemical degradation of Reemay wrap material (used as
a flow distributor within the permeator) as well as chemical degradation
of the membrane itself. Supporting laboratory tests indicate that degrad-
ation of the Reemay component was related to exposure of the module to the
brightener in the bath. Furthermore, in these laboratory tests the membrane
appeared highly resistant to all major bath constituents, including the
brightener; thus the constituent responsible for membrane attack during the
field tests at Whyco Chromium Co. was not identified.
At the second field-test site (New England Plating Co.), the flux and
rejection of the membranes were much more stable. As determined by NaCl
solution performance tests, the flux did not decline significantly during
100 days of operating time. However, a moderate decline in NaCl rejection,
from 90% to 85%, was observed over the same test period.
IV
-------
It is concluded that RO can be used to close the loop in copper
cyanide plating. However, care must be taken to insure that adequate
membrane life can be achieved. Where membrane life approaches that in
traditional RO applications, the capital and operating costs for RO, com-
pared to those for alternative treatment processes, are attractive. The
cost attractiveness of RO will depend on several factors specific for
each installation. Bases for assessing capital costs, operating costs,
and process credits are presented.
-------
CONTENTS
Foreword jij
Abstract .jv
Figures viii
Tables x
Acknowledgement X1-
I Conclusions 1
II Recommendations 3
III Introduction 4
IV Phase I: Field Test at Whyco and Related
Investigations 6
V Phase II: Field Tests at New England Plating 45
VI -Discussion 75
VII References as
vi i
-------
FIGURES
Number Page
1 Process and Instrument Diagram for Whyco Field
Test System 8
2 Flow Schematic for Laboratory Life Test System 11
3 Corrected Productivity vs. Operating Time (Whyco) 17
4 Conductivity Rejection of Plating Salts vs. Operating
Time {Whyco) 20
5 Membrane Life Tests without Brightener, with Brightener,
and with Actual Plating Solution 25
6 Schematic Diagram of Mini-permeator 27
7 Mini-permeator Conductivity Rejection vs. Operating
Time without Brightener 29
8 Mini-permeator Conductivity Rejection vs. Operating
Time with Organic-based Brightener 30
9 Mini-permeator Conductivity Rejection vs. Operating
Time with Selenium-based Brightener 31
10 Comparison of Mini-permeator Rejections for Various
Feed Solutions 32
11 Conductivity Rejection vs. Operating Time for Life Test
without Brightener 35
12 Productivity vs. Operating Time for Life Test without
Brightener 36
13 Conductivity Rejection vs. Exposure Time for Life Test
with Brightener 40
14 Productivity vs. Exposure Time for Life Test with
Brightener 42
15 Sodium Chloride Rejection vs. Exposure Time for Life
Test with Brightener 43
16 Flow Schematic for New England Plating Field Test System 47
17 Corrected Productivity vs. Operating Time (NEP Co) 56
18 Corrected Productivity for Standard NaCl Solution vs.
Operating Time (NEP Co) 59
viii
-------
19 Rejections for Roche!le Copper Cyanide Rinse Waters 61
20 Corrected Conductivity Rejection of Plating Salts vs.
Operating Time (NEP Co) 64
21 Corrected Copper Rejection of Plating Salts vs. Operating
Time (NEP Co) 66
22 Corrected TDS Rejection vs. Operating Time (NEP Co) 69
23 Conductivity Rejection vs. Feed Conductivity (NEP Co) 72
24 Corrected Rejection for Standard NaCl Solution vs. 75
Operating Time (NEP Co)
25 Schematic of Closed-Loop RO Recovery System for 80
Copper Cyanide Bath at New England Plating Co.
26 Concentration in Second Rinse vs. Rejection for 81
Half-Size B-9 Module Operated at 75% Conversion
27 Typical Capital Costs for RO Systems 86
28 Typical Operating Costs for RO Systems as a 87
Function of Capacity and Membrane Life
ix
-------
TABLES
Number Page
1 Composition of Field Test Copper Cyanide Baths 7
2 Flux and Pressure as a Function of Operating Time 16
3 Effect of Operating Time on Conductivity Rejections 19
4 Results of Laboratory Tests with Cyanide Bath Chemicals
with and without Brightener, and with Actual Bath Solution 24
5 Life-Test Data for Feed Solution without Brightener 34
6 Life-Test Data for Feed Solution Containing Brightener 38-39
7 Bath Concentrations as a Function of Operating Time 51
8 Rinse Concentrations as a Function of Operating Time 53
9 Module Productivity as a Function of Operating Time and
Operating Conditions 55
10 Sodium Chloride Flux and Rejection 58
11 Conductivity Rejection as a Function of Operating Time
and Operating Conditions 63
12 Copper Rejection as a Function of Operating Time and
Operating Conditions 65
13 Total Solids Rejection vs. Operating Time and Operating
Conditions 68
14 OH~ Rejection vs. Operating Time and Operating Conditions 70
15 Free CN~ Rejection vs. Operating Time and Operating
Conditions 71
16 Conductivity Rejection at Various Feed Concentrations 73
17 Break-Down of Operating Costs for New England Plating 82
18 Credits Realized for RO Operation at New England Plating 83
-------
ACKNOWLEDGMENT
The authors gratefully acknowledge the help and cooperation of Mr. Jack
Hyner, President, Whyco Chromium Co., Thomaston, Conn., and Mr. Bruce Warner,
President, New England Plating Co., Worcester, Mass., who provided the field test
sites for this program, including support facilities and maintenance personnel.
Their fine spirit of cooperation was invaluable to the successful completion of
this program.
Mr. Roger Lisk and Mr. J.erry Wheelock were responsible for the day to day
operation of the field test systems. This included obtaining all necessary data
and samples, and performing maintenance and modifications required to keep the
systems running properly.
Direction was received throughout the program from members of the American
Electroplaters1 Society Project Committee: Charles Levy (District Supervisor),
Lawrence E. Greenberg (Committee Chairman), Arthur A. Brunei!, Joseph Conoby, Dr.
Robert Mattair, and Robert Michaelson. The EPA project officer, John Ciancia, also
contributed substanially to the program direction.
XI
-------
SECTION I
CONCLUSIONS
1. During field tests at Whyco Chromium Co., both the flux and rejection of
the membrane modules declined within a period sufficiently short to make
RO unattractive on the basis of membrane replacement costs.
a. Tests on one of the field modules revealed that the decline in
performance was the combined result of chemical degradation of the
Reemay wrap-material/flow-distributor and chemical degradation of
the membrane skin.
b. Degradation of the Reemay component was simulated in the laboratory
by exposure of a module to massive doses of brightener. For copper
cyanide applications, the manufacturer should replace the Reemay
component of the module with a more chemically inert material.
c. During laboratory tests, the membrane itself was highly resistant
to all major constituents of the bath including brightener; there-
fore, the constituent responsible for chemical degradation of the
membrane skin during the Whyco Chromium Co. field test remains
unidentified.
2. During field tests at New England Plating Co., the flux and rejection of
the membrane module were much more stable. As determined by the standard
NaCl solution performance tests, there was no substantial decrease in flux
during the test period of 100 days. The NaCl rejection decreased from 90%
to 85% during the same test period.
-------
3. The economics of RO recovery of copper cyanide are closely tied
to the membrane life which, at present, can be determined for
each application only by field tests. For the specific field
test at New England Plating Co., the net savings per day for RO
recovery were insufficient to make the capital investment attrac-
tive on a purely economic basis (i.e., no positive return on in-
vestment). However, RO may still be the most attractive waste-
water treatment alternative available, especially if zero discharge
is required.
4. The dragout rate for most copper cyanide plating lines will
greatly exceed the dragout rate observed during the field test at
New England Plating Co. As the dragout rate increases, the credits
resulting from closed-loop recovery increase. Provided these
credits are not off-set by a shorter membrane life, the economic
attractiveness of closed-loop recovery will increase with the drag-
out rate. The economics can become very attractive if, at higher
dragout rates, the membrane life remains comparable to that observed
at New England Plating Co.
-------
SECTION II
RECOMMENDATIONS
1. In the light of the differences between the two field tests even though
the bath compositions were similar, a plater s'nould obtain some advance
assurance from the membrane equipment supplier that the membrane module
life will be adequate for his particular plating bath.
2. The design of a rinse system using RO recovery should be optimized for
each installation. In the overall design, water conservation and efficient
rinsing (e.g., countercurrent, spray, agitated, etc.) should be used to
reduce the required capacity of the RO unit. The purity of the final
rinse must be specified, based either on the allowable drag-in to a
subsequent processing step or the appearance of the dried part. Means
should be considered to control the rate of bath evaporation to give an
optimum evaporation to drag-out ratio. The optimum ratio will be set by
a balance between energy costs for bath evaporation and RO treatment costs.
3. It would be desirable to identify the cause of membrane deterioration in
the Whyco Chromium Co. field tests in order to better define the limitations
of RO for the treatment of copper cyanide plating wastes.
4. Field demonstrations of reverse osmosis should be extended to other baths.
Mew membranes should be evaluated as they become available on a commercial
or semi-commercial basis.
-------
SECTION III
INTRODUCTION
Most platers recognize the need to reduce the amount of toxic substances
discharged by the metal finishing industry. The basis for the extent of re-
duction which must be achieved by metal finishers is the Federal Effluent
Guidelines for 1977 - , as well as receiving water standards. Moreover, the
Federal Water Pollution Control Act Amendments of 1972 declare that it is the
national goal to eliminate the discharge of pollutants into the navigable
waters of the United States by 1985.
In the plating industry a major source of polluting effluent results
from the discharge of spent rinse waters. The conservation of rinse water by
countercurrent rinsing is always good practice, but in many cases countercur-
rent rinsing alone cannot eliminate rinse water effluent. In looking toward
the national goal for 1985, increasing attention is being focused on clo^ed-
loop processes operating on the rinse water from a specific plating bath.
These processes recover purified water that can be reused in rinsing and con-
centrated plating chemicals that can be recycled to the bath.
Reverse Osmosis (RO) can be used to recover plating chemicals and puri-
fied water from rinse water in a closed-loop system. The advantages and limi-
(2 3)
tations of RO have been discussed previously--. As part of this program
(2 3}
in-house pilot plant tests were conducted-- to determine the feasibility
of treating a variety of plating baths with the commercially available mem-
branes. The results indicate that RO shows promise for the treatment of a
number of plating bath rinse waters.
Before recommending that plating facilities purchase RO equipment, it is
essential to demonstrate the capabilities of RO under realistic conditions.
This can best be done by operating a full-size RO demonstration system in an
actual plating shop. As part of an on-going program to investigate the appli-
cability of RO to metal finishing waste treatment problems, field tests were
conducted on copper cyanide plating baths at two different locations:
— Whyco Chromium Co., Thomaston, Conn.; and
~ New England Plating Co., Worcester, Mass.
4
-------
The objectives of the field tests were:
--to obtain information on the potential limitations of closed-
loop RO treatment;
— to determine the performance (flux and rejection) of the RO
modules, and the deterioration of performance with time; and
--to assess the economics of the closed-loop RO recovery process.
Copper cyanide was selected for the RO field tests for the following reasons:
--Cyanide wastes make a significant contribution to the plating waste
problem.
--Many plating shops contain copper cyanide baths.
—The RO treatment of cyanide wastes has not been previously de-
monstrated.
--Copper cyanide baths operate at elevated temperatures (M50°F)
with significant bath evaporation. Auxiliary evaporation of the
RO concentrate is not required before returning the concentrate
to the bath.
(ft
The RO modules used in the field tests were duPont B-9 Permasep w Permeators
which contain the polyamide membrane in hollow fine fiber configuration. The
polyamide membrane is the only commercially available membrane material which
can withstand the high pH of the cyanide solutions. The other commercially
available membrane (cellulose acetate) is limited to a pH of 2.5 to 8.
Membrane performance was evaluated by measuring the flux and rejection as
a function of operating time. The flux is defined as the rate at which permeate
passes through a unit area of membrane surface under specified conditons. For the
duPont modules, productivity (permeate flow rate per module) is reported rather
than flux per se. The rejection is a measure of the degree to which dissolved
substances are prevented from passing through the membrane. Rejection is defined
by the equation: C - C
Rejection = -^—- 100%
where:
I
Cp= Concentration in the permeate
Cr= Concentration in the feed
-------
SECTION IV
PHASE I: FIELD TESTS AT WHYCO AND RELATED INVESTIGATIONS
GENERAL
Field tests of RO for treatment of copper cyanide rinse waters were
initiated at Whyco Chromium Co. The copper cyanide bath was part of a copper-
nickel-chrome line used to plate a variety of die-cast parts. The plating bath
was about 4,000 gal in size. It was preceded by a copper strike and followed
by two rinses, an acid dip, and a final rinse. The RO system operated in con-
junction with the two rinse tanks between the plating bath and the acid dip.
The composition of the plating bath is shown in Table 1.
The plating line was an automatic rack line that was operated two
shifts per day when the work load was heavy. However, during the period of the
field tests, the copper line was operated on the average of one shift per day.
After the RO system had been in operation for about 400 hours, the line was shut
down for extensive modifications. Some additional data were obtained after
the shut-down by operating the RO unit under "simulated" plating conditons.
Plating was simulated by using a metering pump to transfer 40 gallons per day
(estimated daily dragout for three operating shifts) from the bath into the
first rinse and from the first rinse into the second rinse. The plating bath
was maintain!
evaporation.
was maintained at its normal operating temperature (155°F) to simulate bath
Because of the rapid deterioration in membrane performance observed
in the Whyco field tests, a number of laboratory investigations were undertaken.
The results of these investigations are reported in this section along with the
field test results.
EXPERIMENTAL
Field Test System
A detailed process and instrument diagram of the unit is shown
in Figure 1. The centrifugal booster pump (PI: Flotec Model C6P8) was used to
withdraw about 5 gpm of feed from the first rinse and pass it through two one-
6
-------
TABLE 1. COMPOSITION OF FIELD TEST COPPER CYANIDE BATHS
Whyco
New England Plating Co.
Na2Cu (CN)3
(Cu as metal
NaCN
Rochelle salts
Brightener**
PH
Temp.
Purification
18 oz/gal
6 oz/gal)
2.5 oz/gal
4.0 oz/gal
2000 ppm vol.
11-12
155°F
Continuous active
carbon filtration
K2Cu (CN)3 21 oz/gal
(Cu as metal 6 oz/gal)
K CN 3.5 oz/gal
Cuprolite 20* 6% vol.
Brightener** 2000 ppm vol.
pH 13-13.5
Temp.
140°F
Purification Continuous active
carbon filtration
* A Udylite Rochelle substitute
** MacDermid CI Bright Copper
-------
Evaporation
oo
Line 3 gpm SOV-1 | Rinse 12
Hater Flow
Control
CONCENTRATIONS
Cu (PP») CM (ppn)
roiirt
A
*
C
0
E
f
C
H
I
J
K
L
M
M
0
»
q
*
s
T
u
v
•
X
T
Strike Dili-In
Strike Evaporation
latk Drag-in
Rath Evaporation
Bath Dra'f-out
First Rinse Drag-out
Second Rinse Overflow
Second Rinse Prag-out
PI MaVeup Hater
Line Kater for Rinsing
Feed to BO Unit
Feed to Stage II
Sta|e fl Concentrate
Stage «J Concentrate
Stages «1«3 (^circulation
Concentrate Flow
Stage '1 Peraeate
Stage *2 fertttfate
Nigh-pressure Safety Overflow
Feed to Stage 'S
Stage 13 Recirculation
Stag« «5 Concentrate
Low-pressure Safety Recycle
Stage '] Peracate
First Rinse Overflow
Figure 1. Process and Instrument Diagram
for Whyco Field Test System.
Flux assiaed equal to 2.S gpm for ««ch aodul*
Cu r*i*ction«: 0.995 for «<**«• t«2; 0.96 far «tttg« 5
O4 r»)«ction»: O.DdU for it»f»» 1(2; 0.92 for it«j« 3
-------
micron cartridge filters (Filterite Model U1AW20U) in parallel. The pressure
drop across the filters was measured (PI) to determine when cartridge replacement
was necessary. A spare set of filters allowed cartridges to be changed without
shutting the system down. The temperature was measured (TI) at the discharge
from the filters.
A high pressure piston pump (P2: Cat Pump Corp. Model 01001)
pressurized the feed to 300-400 psi. The discharge pressure was controlled by
a needle valve (NV-3) in the pump by-pass. An accumulator on the pump dis-
charge was used to dampen pressure pulsations. A high pressure switch and alarm
(PA-HI) and a pressure relief valve (PRV) protected the RO modules from over-
pressurization, and a low pressure switch and alarm (PA-LO) prevented the pump
from running dry in case of fluid loss. The feed was passed through two RO
modules (duPont B-9 Permasep Permeators Model 0440) in series. These modules
are designated as "stage 1" and "stage 2". The total pressure drop across the
modules was measured (PI). A portion of the concentrate was returned to the
strike and/or bath, and the flow rate was maintained constant by the flow control
valve (FC). The remainder of the concentrate was recycled to the suction of the
high pressure pump. This recycle was required in order to maintain the recom-
mended flow rate through the duPont modules.
The permeate from stages 1 and 2 was combined and repressurized by
a second piston pump (P3: Cat Pump Corp. Model 00501) which was fitted with the
same type of accumulator and high and low pressure safety devices as the pump
for stages 1 and 2. The feed was passed through a third RO module, stage 3,
which was identical to stages 1-and 2. The concentrate from stage 3 was returned
to the first rinse and the permeate to the second rinse. A high-pressure over-
flow line (line "S") and a low-pressure recycle line (line "W") were included
to keep the pump suction for stage 3 between 0 and 30 psi.
In addition to measuring the pressures (PI) and flows (FI) at various
points in the RO unit, conductivity probes and a recorder (CR) were used to
continuously monitor the rinse water quality. A conductivity alarm in the second
rinse would shut down the RO system if the preselected conductivity set-point
were exceeded. If the RO system were shut down, normal countercurrent rinsing
-------
would resume automatically using a line-water flow of 3 gpm.
During closed-loop operation of the RO system the only non-plating
loss of chemicals was by dragout from the second rinse; the only loss of water
was by evaporation from the bath (and strike). Deionized water was used for
make-up since the use of line water would introduce salts into the system which
would be recovered and recycled to the bath along with the plating chemicals.
These could eventually build up to such an extent that plating quality would be
adversely affected.
The calculated flow rates and concentrations of copper and cyanide
are shown in Figure 1 for various points throughout the RO system and plating line.
The assumptions upon which the calculations are based are given at the bottom of the
tabular insert. The calculated copper concentration in the second rinse is about
O.b :-
-------
Permeate
To
Rectifier
Drain
Concentrate
Cooling Water
IPS
HPS
High
Pressure
Pump
BPR V_n
RO Module
Figure 2. Flow Schematic for Laboratory Life Test System.
-------
Samples were taken periodically for conductivity analyses.
Feed samples were withdrawn from the feed tank; concentrate and permeate samples
were withdrawn from the end of their respective return lines by removing them
from the mixing tee. Flow rates were measured by the bucket-and-stopwatch tech-
nique.
In general, the feed solution was prepared at 20% of bath
strength and the pH was adjusted to 11. The full-strength bath composition is:
CuCN 8.6 oz/gal
NaCN 11.9 oz/gal
Roche!le Salts 4.0 oz/gal
Brighteners 2000 ppm
Assays
The only assay performed on a regular basis during the field
test was conductivity. Since the deterioration in membrane performance was so
rapid, the expense of more complete analytical work did not appear justified.
Several weekly samples were taken and analyzed for total dissolved solids
(gravimetric technique), and the TDS results showed a close correspondence to
the conductivity results.
For the laboratory life tests, conductivity was again the only
analysis performed regularly. The conductivity was measured with a battery-
operated conductivity meter that was calibrated with a NaCl standard.
RESULTS OF FIELD TEST
General Operation
The field demonstration unit was operated intermittently over
a three-month period. During this time certain problems became evident. The
most serious problem was a gradual deterioration in performance of the membrane
modules which finally necessitated a temporary halt in the field test program.
A few minor mechanical problems, associated with the staged permeate mode of
12
-------
operation, were resolved during field testing and, mechanically, the unit operated
satisfactorily.
Since the rinse water was treated in a closed loop, a build-
up of temperature was anticipated. There were two sources of heat input:
1) heat was transmitted to the first rinse via parts and drag-in from the warm
Plating bath; and 2) heat was introduced in the RO unit where pumping energy was
converted to frictional heat. There were also two sources of cooling: 1) heat
transfer to the surroundings; and 2) addition of deionized water at a rate sufficient
to compensate for bath evaporation. Although rinsing is more efficient at higher
temperatures,the duPont modules are not recommended for use above 95°F. On very
Warm summer days the temperature of the feed to the RO system climbed above this
limit. In applications where the ambient temperature can exceed the 95°F limit
a heat exchanger should be installed on the feed to the RO unit with the cooling
Water thermostatically controlled. A separate cooling-water drain system is
Preferred so that the volume of waste going to the chemical treatment system
is not needlessly increased with spent cooling water.
Deionized water was added to the final rinse at a rate of about
One-third gallon per minute. Rapid exhaustion (15 days) of the exchange resins
Proved to be an annoying and costly maintenance problem. The regeneration fre-
quency could be greatly reduced by pretreating the line water with RO before
deionization, or perhaps by using RO alone.
productivity (or Flux)
The productivity of a given module is the rate at which permeate
is produced under specified conditions. The productivity is dependent on
temperature, pressure, feed concentration, and conversion. The measured productivity
Was corrected to 400 ps1 and 77°F (normal operating conditions for the duPont
Module) using the duPont Technical Information Manual. However the data were
Hot corrected for variations in feed concentration and conversion.
The conversion is the ratio of the permeate flow to the feed
'flow. For a module operated at near-zero conversion, the concentrations of the
"feed and concentrate streams are nearly the same. Thus the average concentration
13
-------
on the feed/concentrate side of the membrane is very nearly the same as the feed
concentration. On the other hand, a module operated at high conversion will
produce a concentrate stream that is much higher in concentration than the feed
stream. In this case, the average concentration on the feed/concentrate side of
the membrane will be substantially greater than the feed concentration. For two
modules operated at the same pressure, temperature, and feed concentration, the flux
will be lower for the module operated at the higher conversion since flux decreases
with increasing average concentration on the feed/concentrate side of the mem-
brane.
The concentrate withdrawal from stage 2 was fixed at 1.50
gpm. The conversion for stage 2 is given by:
P
2 ~P2 + 1.50
where V^ is the productivity (permeate flow rate) of stage 2. The conversion
for stage 1 is given by:
p,
+1.50
The feed to stage 3 was maintained approximately constant at 5 gpm so that the
conversion for stage 3 is given by:
P
From these equations the conversion for stage 1 decreased from 41% to 33%, for
stage 2 from 65% to 53%, and for stage 3 from 64% to 40% as the operating time •
increased (i.e., as flux decreased ).
The feed concentrations to the RO modules were not fixed
and could vary greatly depending on the amount of work passing through the rinse.
The uncontrolled feed concentration is a much greater source of error than the
rather small variations in conversion.
14
-------
The productivities of the three modules are given as a function
of operating time in Table 2, and corrected productivities are plotted in Figure 3.
The "operating time" gives the cumulative hours during which the RO unit
was running, as opposed to "exposure time" which is the cumulative time during
which the modules were in contact with the feed solution.
Over the first 400 hours (normal operation) the flux declined
rather rapidly with time but the decline seemed to taper off and approach a
plateau value. The decline was the greatest for stage 2 which was exposed to the
most concentrated feed solution and was the least for stage 3 which was exposed
to the most dilute solution. The flux for stage 2 declined to 60% of the initial
flux after only 300 hours (12.5 days) of operation. At 300 hours a new module
was installed in stage 2, and the flux began to drop as before.
During simulated operation (metering pumps used to simulate
dragout from the bath and first rinse) the drop in flux was much more rapid
than during normal operation. It is quite likely that the simulated dragout rate
was higher than the average normal dragout rate. If the curves for simulated
operation are extrapolated, the predicted fluxes at 750 hours ( 1 month) will be
about one third their initial values.
The drop in flux is much too large to be explained in terms of
compaction of the fibers. The other possibilities include plugging by particulates
in the feed, plugging by precipitation of a sparingly soluble salt during concen-
tration, and chemical attack by some constituent in the feed. Of these possibilities
the latter is the most likely. Destructive tests on the stage 2 module showed
no sign of plugging or fouling.
Rejection
The rejection measures the degree to which plating salts are
prevented from passing through the membrane. The rejection based on feed
concentration (r= 100 [Cp - Cp]/Cp) depends on the operating pressure, the con-
version, and the feed concentration. The operating pressures and conversions
were reasonably constant for the three modules so that only minor corrections
15
-------
TABLE 2. FLUX AND PRESSURE AS A FUNCTION OF OPERATING TIME
o>
Operating
Time (hrs
2
4
32
85
88
244
300
340
400
416
440
480
540
560
Temp.
70
74
88
84
88
94
77
80
90
62
63
81
70
70
Pressure ipsija
Stage 1
400
395
325
375
360
355
380
378
375
340
340
375
350
350
Stage 2
385
380
300
357
345
340
365
355
355
315
315
345
338
325
Stage 3
315
300
275
288
282
265
325
300
325
350
350
300
375
350
Stage 1
2.58
2.75
2.55
2.55
2.57
2.50
1.76
2.25
2.15
1.50
1.40
1.90
1.40
1.30
Flux (gpm)
Stage 2
2.25
2.40
2.40
2.10
2.10
1.95
1 .60
2.65
2.10
1.50
1 .50
1.90
1.40
1.40
Stage 3
2.52
2.60
2.50
2.48
2.50
2.70
2.15
2.20
2.40
1.75
1.75
2.00
1.65
1.65
Corrected Flux (gpm)
Stage 1
2.94
3.06
2.66
2.45
2.43
2.18
1 .85
2.30
2.19
2.27
2.12
1.91
1 .80
1.67
Stage 2
2.79
2.70
2.74
2.13
2.06
1.76
1.75
2.88*
2.26*
2.45*
2.45*
2.01*
1.87*
1.94*
Stage 3
3.74
3.76
3.16
3.16
3.06
3.20
2.65
2.80
2.80
2.60
2.60
2.51
1.98
2.10
a) Pressure measured at feed. For Stage 2 feed pressure was determined by dividing the AP equally
between Stages 1 and 2.
b) Flux corrected to 400 psi and 77°F using duPont correction factors from Technical Information
Manual.
* A new module was installed in Stage 2 at 300 hours operating time.
-------
c
Q.
#i
o
[Z
0)
rtJ
OJ
Q-
3.0,
£ • O i
2.6
2.4
2.2
2.0
1.8
1.6
1.4
1.2
1.0
.8
.6
.4
^~I — ~^~~-y^~^
\ ^- — Stage 3
n\P
-\ °\
7
V>-^^D *—
>--._.
^— Stage 2
—
O Stage 1
D Stage 2
~ A Stage 3
\ \
100
Stage 1
-Q--
New Module
Installed
in Stage 2
200
j L
300
Operating Time, Mrs
1
400
Simulated Operation
Started
500
Figure 3. Corrected Productivity vs. Operating Time (Uhyco).
-1
600
-------
for variation in pressure and conversion would be anticipated. However, the feed
concentration to a given module varied significantly depending on the amount of
work passing through the rinse. The variability of the feed concentration
represents the largest source of error in the data. It is not possible to conve-
fiiently correct the data to a fixed feed concentration as will be discussed in
the following section. Therefore no correction factors were applied to the rejection
data.
The conductivity rejection of plating salts is given as a function
of operating time in Table 3 which also gives the conductivity and (in some cases)
pH of the feed to each stage. The conductivity rejections are plotted as a function
of operating time in Figure 4. .
There are several interesting trends in this rejection data.
The obvious trend is the rapid decrease in rejection for stages 1 and 2. During
a total operating time of about 23 days, three good modules were consumed. At
a replacement cost of $1206.00 this represents an unacceptable operating ex-
pense. The overall rate at which the rejection decreases is greater for stage 2
suggesting that the greater decrease is related to the higher feed concentration
for stage 2.
During simulated operation (after 416 hours) the rejection for
stages 1 and 2 decreased at a significantly greater rate. At 560 hours the
rejection of stage 1 had dropped to 28%, and stage 2, to 21%. At this point
the conductivity alarm in the second rinse was triggered and the entire RO unit
was automatically shut down. The second rinse alarm was set at about500 y mhos/cm
which is three times the conductivity of line water used for normal counter-
current rinsing. At this point it was decided to terminate the field tests
until the cause for the decrease 1n rejection could be determined.
One of the most significant features of Figure 4 is that the
rejection for Stage 3 remains substantially constant over the entire test period.
This cannot be entirely attributed to the lower concentration in stage 3 since,
as shown 1n Table 3, the feeds to stage 3 and stage 1 are not really that much
different in terms of conductivity. The high constant rejection for stage 3
Indicates that the constituent responsible for the decrease in rejection does not
pass through the membrane of stages 1 and 2 even though their rejections are lowv
-------
TABLE 3. EFFECT OF OPERATING TIME ON CONDUCTIVITY REJECTIONS
Feed Conditions by Staqe
Operating
Time (hrs
5
85
185
245
270
300
340
400
416**
440**
480**
540**
560**
I Staqe 1
Stage 2
Stage 3
>) pH Conductivity pH Conductivity pH conductivity
(umho-cm) (ymho-cm) (ymho-cm)
3,900
9.97 7,500
10.65 3,500
10.28 6,600
10.28 8,800
12,000
2,650
10.5 8,400
2,500
10.1 2,500
2,500
1,800
10,000
7,000
10.09 11,300
10.63 7,200
10.29 10,200
10.35 14,000
21 ,000
3,600
10.51 11,300
3,100
3,000
3,100
2,100
12,000
410
9.75 2,800
10.53 2,000
10.21 2,100
10.10 3,700
3,400
600
10.39 4,600
1,100
1,650
3,250
1,300
8,000
% Conductivity Rejections
by Stage
Stage 1
97
92
90
87
84
72
75
81
80
60
45
33
28
Stage 2
90.5
60
77
67
63
97*
85*
78*
79*
74*
70*
57*
21*
Stage 3
90
92
90
90
93
94
94
95
94
94
94
92
94
* A new module was installed in Stage 2 at 300 hrs operating time.
**During this period the rack plating operation was stopped. However, the plating solution dragin
and dragout were simulated by pumping solutions with metering pumps.
-------
ro
o
100
90
80
£ 70
o
cu
"S? 60
oc
I" 50
? 40
-o
c
o
30
20
10
0
D
O Stage 1
D Stage 2
A Stage 3
100
200
New Module
Installed in
Stage 2
400
500
300
Operating Time, Mrs
Figure 4. Conductivity Rejection of Plating Salts vs. Operating Time (Whyco).
600
-------
INVESTIGATIVE TESTS ON FIELD MODULE
When the original stage 2 module was removed after 300 hours of
operation its conductivity rejection for plating chemicals had dropped to 60%.
Several cleaning procedures were performed for the removal of common foul ants,
but no improvement in rejection was obtained. During the cleaning procedures
the NaCl rejection of the module was measured under standard!zed conditions (400psi,
77°F, 75% conversion, 1500 ppm feed). The measured rejections were 58% and 50%
as compared to 90% for a new module. This confirms that significant damage did
occur in the field.
The module was returned to duPont for more extensive tests. The
duPont tests involved opening the fiber bundle, visually inspecting the various
internal components, and making physical tests on the fibers. These tests
resulted in the following findings:
1. There was no evidence of mechanical defects.
2. There was no evidence of any scaling, particulates, or
foulants in the fiber bundle.
3. There was no significant deterioration in tensile pro-
perties or collapse resistance of the fibers.
4. The Reemay spacer (which functions as a flow distributor
in the fiber bundle) showed significant deterioration in
physical properties.
5. Permeation tests, conducted by making a mini-permeator
from about 150 of the fibers in the module, showed
high salt passage caused by severe skin damage.
These results indicate that the decrease in rejection was caused by
a chemical attack of the membrane fiber rather than by particulate plugging,
precipitation, or deposition of foulants on the membrane surface. The decrease
cannot be attributed to direct hydrolysis at high pH since the upper pH limit of
11 was not exceeded. In addition, the rejection for stage 3 did not decrease
even though the pH o* the feed to stage 3 was nearly the same as for stages 1
and 2 (see Table 3).
21
-------
LIFE TESTS ON FULL-SIZE PERMEATOR
If direct attack by OH~ is ruled out, there must be some
other consitutent of the bath which was responsible for the rejection decline.
The constituents of the Whyco bath are: copper cyanide, sodium cyanide, sodium
hydroxide, Rochelle salts, and a selenium-based brightener (MacDermid CI Bright
Copper^. None of the major constituents of the bath appeared to be likely
candidates for attacking the membrane. Previous life tests with other cyanide
solutions showed that the membrane has good cyanide resistance. Copper is also
an unlikely candidate and in any case could not be removed from the bath.
Sodium hydroxide can be easily handled provided a pH of 11 is not exceeded.
Aside from the possibility that some impurity in the bath was responsible for the
decline in rejection, the selenium brightener, which contains an inorganic
oxidizing agent, appeared to be the most reasonable choice as the attacking
constituent. Since chlorine, also an inorganic oxidizing agent, is known to chem-
ically attack the membrane, the oxidizing agent in the brightener could con-
ceivably exhibit similar behavior. It is anticipated that all selenium-based
brighteners would behave in essentially the same way, so the problem would not
be specific to the particular brand of selenium brightener used.
Laboratory life tests were conducted to determine what effect the
brightener had on membrane performance. The laboratory life test system (total
recycle) was described previously. One of the limitations encountered in using
a total-recycle system is that a faulant in the feed tank can interact with the
membrane in several passes through the module and thus be removed from the system.
The only observed effect is a very slight (usually undetectable) drop in membrane
performance. In an actual system the membrane is continually exposed to the foulant
resulting in a gradual decline in performance. Thus in a total-recycle system
it is necessary to ensure that the membrane receives the same total exposure
to the foulant as would be received in an actual system.
Life tests were conducted with three different feed solutions: a
synthetic solution of plating bath chemicals at 20% of bath concentration but
without the brightener; the same solution with a considerable excess of brightener
added at various times to simulate a continuous input of brightener,; and the
actual plating bath solution (with the recommended concentration of brightener
but with no excess added) diluted to 17% of bath strength.
-------
The flux and rejection data for the three life tests are given in
Table 4 and are plotted as a function of exposure time in Figure 5. Except
for the periods shown in Figure 5, the test system was operated 24 hours per day
so that exposure time and operating time are not greatly different in this plot.
Over the first 150 hours using the synthetic feed solution without brightener, no
decrease in rejection was observed. Over the next 150 hour period brightener was
added to the feed solution at several points as shown in Table 4. A gradual
but definite decrease in rejection was observed. Over the 150 hour period the
rejection decreased from 96.5% to 92%. Tests with the actual plating bath solution
at 17% of bath concentration showed an initial decrease in rejection, but at
longer exposure times the rejection appeared to decrease much more slowly. (This
behavior would be expected for a total recycle system as explained above.)
No substantial difference can be observed in the rate of flux
decline for the three life tests. The discontinuity in the flux curve can be
attributed to a higher feed concentration for the actual bath.
The rejection results of Figure 4 tend to confirm the suspicion that
the brightener is the constituent responsible for the rejection decline. A
definite decrease in rejection was observed when the brightener was added to the
feed solution, but it remains to be answered as to whether the magnitude of the
decrease is comparable to the decrease observed in field tests. If the brightener
reacts rapidly with the membrane then the degree of degradation is a direct
function of the amount of brightener fed to a given module. A total of 0.134
gal (5,380 ppm in 25 gallons) was added to the feed solution during the labora-
tory tests. For the recommended bath concentration of 2000 ppm vol and a drag-
out of 40 gpd, the RO demonstration plant would be fed 0.134 gal of brightener
in approximately 40 hours. The decrease in rejection for the laboratory tests
(96.5% to 92%) is at least reasonably consistent with the decrease observed in
field tests (Figure 4) for an equivalent brightener exposure.
23
-------
TABLE 4. RESULTS OF LABORATORY TESTS WITH CYANIDE BATH CHEMICALS
WITH AND WITHOUT BRIGHTENER. AND WITH ACTUAL BATH SOLUTION
Exposure
Tine , Mrs
Temp .
(JC)
Pressure
(psi]
Flux Corrected
(gpm) Flux (aprc
* % Con- Conductivity, nmho/cm
) version Feed Permeate « Rejection
Life Test with Bath Chejuicals Without Brightener in the Feed (20% Bath Concentration)
1
7
23
20
System pump was
74
91
15
32.5
49.5
Li fe
400
Tes
ppm
20
16
19
22
23
t with
355
355
stooped running
.5
.5
360
335
350
355
355
Bri gntener
of MacDermid
Added
Brightener
1
1
for
1
1
1
1
1
.55
.51
about 60
.54
.38
.45
.51
.51
to the
Above
Added to 25
1 .
1.
hrs
1.
2.
1.
1 .
1.
86
97
5
5
during weeke
99
00
90
86
81
5
3
5
5
5
Feed Solution
qal
Feed
Solution
152.5
155.5
173
Added 1660
196
Added 1660
245
Added 1660
251
268
276
292
20
20
20
350
350
355
ppm of MacDermid
20 355
ppm of 'lacDermi-d
20 355
ppm of 'lacDermld
20 355
19 355
19 355
19 355
1.43
1.43
1.43
Brightener
1.47
Brightener
1.42
Brightener
1.37
1.37
1.35
1.35
1 .90
1.90
1.87
1 .90
1.87
1.79
1.85
1.82
1 .82
56.5
56.5
56.5
57
58
58
58
60
58
Life Test with Actual Bath Solution (20!li Bath Concentration)
294 Started test
295
299.5
Added 5
316.5
324
339.5
364
Pump was
481 .5
488
504
534
535
600
20
21.5
365
350
gal of water to the
19
17
15
18.5
stopped
19
19
19
18.5
19
20
350
355
350
350
runnina for
350
350
350
350
370
350
1.18 1.50
.98 1.21
feed solution
1.03 1.41
.97
.97
1.08
about 115 hrs dur
.95
.98
.98
.90
.40
.50
.50
ng weekend
.30
.34
.34
.30
1.0 1.30
1.0 1.33
40
42
50
48.5
48
50
47
48
46
52
55.5
51
End of Tests
16.000
16.000
15.000
15,000
15,400
16,000
16,500
16,000
16,000
16,000
15,100
15,500
15,100
15,100
15,000
15,000
24,900
25,000
21,000
20,500
20,000
21.000
20.800
20.500
20,500
20,000
19.800
22.SOO
680
545
675
440
530
560
580
670
620
655
640
950
950
1 .070
1,180
1 ,150
1 .750
2,450
2,100
1,880
1 .710
2,000
2,400
2.250
2,100
2,400
2,500
2.720
95.75
97.68
95.50
97.06
96.56
96.50
96.48
95.81
96.12
95.90
95.76
93.87
93.70
92.91
92. 13
92.13
93.00
90.20
90.00
90.80
91.4
90.5
88.5
89
89.8
88
87.4
87.9
Flux corrected to 400 psi and 77°F using duPont correction factors from Technical Information Manual.
24
-------
§
•H
+->
O
O
en
X
U
T3
§
u
c\°
100
98
96
94
92
90
88
IV3
x
^H
P-
2.0
1.8
1.6
1.4
1.2
1.0
I I I
I I I
i i r
i 1 T
i i r
i 1
O
o
o
20% Bath Chemicals
without Brightener
as Feed
20% Bath Chemicals
with Brightener as
Feed
17% Actual Bath Solution as Feed
Not
_ Operating
-•—i
Not Operating
(Q)
J L
100
200 300
Exposure Time, Hrs
400
500
O
600
Figure 5. Membrane Life Tests Without Brightener, With Brightener,
and With Actual Plating Bath Solution.
-------
LIFE TESTS ON MINI-PERMEATORS
Tests conducted on the full-size module were somewhat inconclusive
because of the rather small drop in rejection and the small total exposure to
the brightener. In order to investigate the stability of the polyamide membrane
in more detail, tests were conducted with mini-permeators obtained from the
Permasep Products Division of duPont.
The details of a mini-permeator are shown in Figure 6. The active
portion of the permeator consists of one strand (150 filaments) of polyamide
hollow fibers. The strand is looped as shown, and both open ends are sealed in
an epoxy pot. Permeate is withdrawn at one end of the fiber strand after slicing
the permeate tube to expose the open fiber ends. The normal four-inch permeator
con^ins about 900,000 filaments so that, in terms of surface area, the nvini-
-4
permeator is less than 2 x 10 times the size of a full-scale module.
The mini-permeators were operated at 400 psi, approximately 77°F,
and essentially zero percent conversion. The initial flux was on the order of
2 cc/min or less while the feed and concentrate flow rates were on the order of
1000 cc/min.
The feed solution was prepared from laboratory grade chemicals
and distilled water. The bath conposition, shown below, was diluted
Component Concentration
CuCN 8.6 oz/gal
NaCN 11.9 oz/gal
NaOH 2.5 oz/gal
Rochelle Salts 4.0 oz/gal
to 20$ of its original concentration and the pH was adjusted to 11.0 (maximum
for the polyamide membrane) with hydrochloric acid.
The brighteners were added in considerable excess over the recommended
bath concentration. A total of one pint of brightener solution was added to
five gallons of feed solution giving a concentration of 25,000 ppm. This is
26
-------
ro
Permeate
Epoxy Pot
Stainless
Steel Fitting
Shell
1 Strand of
150 Filaments
Concen-
trate
Stainless
Steel
Fitting
Feed
Figure 6. Schematic Diagram of Mini-permeator.
-------
about an order of magnitude above the recommended bath concentration and about
two orders of magnitude above the maximum concentration that the RO system sees.
The exposure of a mini-permeator to one pint (.125 gallons) of
brightener is equivalent to the exposure of a full size permeator to 625 gallons
(.125/2 x 10 ) of brightener. At 40 gallons per day dragout of 2000 ppm brightener,
this is equivalent to an operating time at Whyco of 7,800 days or 21.5 years!
Results will be presented for three mini-permeators, each operated
on a different feed solution. All feed solutions contained plating chemicals at
20% of bath concentration adjusted to pH 11.0. The first feed solution tested
contained no brightener, the second contained an organic-based brightener
(Allied-Kelite Isobrite 625), and the third contained a selenium-based brightener
(Mac Dermid CI Bright Copper); (The use of these particular brighteners is not
intended as an endorsement.)
Rejection data for the three feed solutions are shown in Figures
7, 8, and 9. The curves from these three figures are compared in Figure 10.
The best overall performance was obtained with the feed solution containing the
selenium-based brightener, and the poorest performance was obtained with no
brightener in the feed. The difference in performance between the feed solutions
containing the organic-based and the selenium-based brighteners is not significant.
The rejection results of Figure 10 indicate that the polyamide
membrane is quite resistant to both organic-based and selenium-based brighteners.
Both the concentration and exposure of the membrane to the brightener were far in
excess of the concentration and total exposure a typical membrane would receive
in an actual system. This conclusion would appear to contradict the results
of Figure 5 which indicate a definite decrease in rejection performance of the
module when the brightener is added. The difference in results may be due
to deterioration of some portion of the module (in particular, the Reemay spacer)
other than the polyamide fiber. (Note that the mini-permeator does not contain a
Reemay spacer.) This will be discussed in more detail below.
28
-------
ro
100
80
o
o
ce.
70 —
•(J
O
| 60
o
o
50
40
o
o
~T
O
_l
100
200
Operating Time, Hrs
_____ L
300
400
Figure 7. Mini-permeator Conductivity Rejection vs. Operating Time Without Brightener.
-------
co
o
o
OJ
CD
o:
o
3
T3
C
O
100
90
80
70 L
60
50
I
40
0
-O
O
o
100
7
O O
O
o
200
Operating Time, Mrs
300
400
Figure 8. Mini-permeator Conductivity Rejection vs. Operating Time With Organic-based Brightener.
-------
100 i—
() O
90
80
O
O)
O)
70
60
O
o—o-
50
40
50 100
00
O Ql O
1 "T
o—o-
J L_
150
200 250
Operating Time, Hrs
.Of-).
1
300
O O
350 400
Figure 9. Mini-permeator Conductivity Rejection vs. Operating Time With Selenium-based Brightener.
-------
CO
ro
100
90
80
0
o
CD
O)
70
u
1 60
o
o
50
40 L
No Brightener
Organic-based Brightener
Selenium-based Brightener
50
100
150 200 250
Operating Time, Mrs
300
..L
350
400
Figure 10. Comparison of Mini-permeator Rejections for Various Feed Solutions.
-------
LIFE TESTS ON HALF-SIZE PERMEATOR
Because of the apparent contradiction in results between the full-
size permeator tests and the mini-permeator tests, further life studies were
initiated using half-size B-9 permeators. The configuration of the half-size
module is identical to the full-size module except that the half-size module is
only about half as long. Laboratory tests were conducted with two separate
modules: one operated on feed solutions containing no brighteners, the other
on feed solutions containing a selenium-based brightener (Mac Dermid CI Bright
Copper). The same laboratory life test system described previously was used.
Feed Solution Without Brightener
A half-size B-9 module was operated for 1360 hours on a feed
solution containing plating chemicals at 20% of bath strength (pH 11) but without
the brightener. The data are given in Table 5. No corrections for feed concen-
tration, conversion, pressure, or temperature were applied to the data. The
conductivity rejections are plotted as a function of exposure time in Figure 11.
For these tests the operating time was very nearly the same as the exposure time
since the system operated continuously. The rejection declined gradually from
about 96.5% initially to an extrapolated value of 92.5% over three months
(2200 hours) of exposure time.
Flux data are shown in Figure 12. The productivity dropped
to about 1/2 of its initial value over the first 450 hours of exposure. The module
was cleaned at that point using the procedure noted in Table 5, and the flux
gradually recovered and remained close to its initial value for the remainder of
the test. It is possible that the cleaning procedure removed some iron hydroxide
deposits from the membrane that could have resulted from corrosion within the
test system.
The data of Figures 11 and 12 indicate that the flux and rejec-
tion are quite stable tothemajor constituents of the bath.
33
-------
TABLE 5. LIFE-TEST DATA FOR FEED SOLUTION WITHOUT BRIGHTENER
Cumulative
Exposure Feed
Time Pressure
(hrs) (psi)
76
77
91
99
115
123
140
147
211
219
235
260
268
291
309
316
386
432
452
464
483
486
562
625
634
650
656
675
697
706
721
730
804
919
949
964
988
1012
1021
1084
1092
1108
1164
1251
1284
1310
1336
1360
1422
200
280
310
320
300
290
300
290
300
250
310
310
310
280
290
290
290
290
280
System
300
270
300
340
310
310
340
330
340
340
340
330
300
300
300
300
300
300
300
300
300
300
300
300
300
300
200
225
200
Flux
(1/min)
1.83
1.85
2.28
1.64
1.64
1.68
1.64
1.76
1.96
1.44
1.68
1.61
1.80
1.36
1.40
1.40
1.36
1.10
.96
Cleaned*
1.32
1.28
1.68
2.32
2.00
1.92
2.00
1.76
1.68
1.76
1.84
1.68
1.72
1.70
1.72
1.72
1.82
1.80
1.78
1.80
1.76
1.84
1.96
1.88
1.84
1.76
2.40
2.10
Conversion
(%)
45
45
56
40
40
41
40
43
48
35
41
39
44
33
34
34
33
27
23
32
31
41
56
49
47
49
43
41
43
45
41
42
41
42
42
44
44
43
44
43
45
48
46
45
43
58
51
Conductivity
(as ppm Nad )
Feed xlO Permeate xlO"
3.2
3.4
3.7
3.2
3.1
3.0
3.4
3.2
3.7
3.4
3.3
3.2
3.6
3.3
3.3
3.1
3.2
3.1
3.3'
3.4
2.1
1.4
1.4
1.4
1.4
1.4
1,4
1.4
1.4
1.4
.4
.4
.6
.6
.4
.4
1.4
1.4
1.4
1.4
2.1
2,1
1.9
1.9
1.9
2.1
1.9
1.9
0.9
1.5
1.2
1.0
1.2
1.0
1.2
1.2
0.9
1.2
1.0
1.2
1.4
1.6
1.4
1.4
1.2
1.0
0.9
0.8
0.7
0.7
0.9
0.8
0.8
0.6
0.7
0.7
0.5
0.6
0.7
0.9
0.9
0.9
0.7
0.7
0.7
0.8
0.8
0.9
1.3
1.2
.1
.2
.3
.5
.1
.5
, Rejection
6 m
97
95
97
97
96
96
96
96
97
96
96
96
96
95
•96
95
96
96
97
98
97
95
94
95
95
97
95
95
98
97
95
93
94
94
95
95
95
95
95
94
94
94
94
94
93
93
94
92
* Module was flushed with a 2% solution of citric add adjusted to pH 4.0 with
ammonium hydroxide.
34
-------
O
o
O)
OJ
cc:
100
90
80
70
oo
tn
o
1 60
o
50
o o
•o —o oo o-
00
_J. I L
_L
200 400 600 800
L ......... L_l ___ L ..... _L_ J_~~L ...... I
1200 1400 1600 1800
1000
Operating Time, Mrs
-_ L ___ L ...... i .....
2000 2200 2400
Figure 11. Conductivity Rejection vs. Operating Time for Life Test Without Brightener.
-------
% a
<* -o
3.0
2.8
2.6
2.4
2.2
2.0
1.8
1.6
1.4
1.2
1.0
.8 -
.6 -
.4 _
.2 -
0
o
o
o
Module Cleaned
J L
J L
_L
O
°cP
(JO
I I l L
-L
0 200 400 600 800 1000 1200 1400
Operating Time, Hrs
T 1 r
1600
L—.JL L__
1800 2000
2200
— t
2400
Figure 12. Productivity vs. Operating Time for Life Test Without Brightener.
-------
Feed Solution With Brightensr
A half-size B-9 module was operated on an identical
feed solution to which brightener was added at various points. The data are
summarized in Table 6. The uncorrected conductivity rejections are plotted as
a function of exposure time in Figure 13. The notes indicate when and to what
extent brightener was added.
Over the first 1222 hours enough brightener was added to the
feed tank to give a concentration, after the final addition, of 17,500 ppm. This
is 8.75 times the recommended concentration (2000 ppm) for the bath, and about 87.5
times the maximum concentration the RO system should see (assuming the concentrate
stream is about 10% of bath strength). Therefore, in terms of brightener con-
centration, the life tests were severe. The total exposure after point D is
1.05, of brightener (17,500 ppm in 58£). This is an exposure equivalent to 10
days operating time at Whyco (40 gpd dragout of 2000 ppm brightener for eight hours
per day). A significant decline in rejection was observed in the field tests
over 10 days (240 hours) of operating time. The results of Figure 13 indicate
that the decline in rejection observed at Whyco cannot be directly attributed
to the action of the brightener. The results are summarized below for equivalent
exposures.
Conductivity
Total Exposure Max. Cone. Rejection Decline
Whyco (over first 240 hours) 1 & 200 ppm 90% to 65%
Laboratory Simulation 1 i 17,500 ppm 97% to 92%
In order to determine whether the brightener had any detrimental effect at all on
module performance, the module was exposed to a massive dose of brightener
(1.5 gal brightener in 10 gal water) at point E of Figure 13. At point F the
brightener solution was replaced by a solution containing plating chemicals at
20% of bath strength (without brightener), and the rejection compares favorably
with the rejection at point D, before the massive brightener dose.
37
-------
TABLE 6. LIFE-TEST DATA FOR FEED SOLUTION CONTAINING BRIGHTENER
Cumulative
Exposure
Time
(hrs)
* 0
18
44
52
67
70
148
210
219
236
241
260
291
306
A-314
389
504
535
B — 549
573
581
597
606
669
677
693
C--749
848
869
894
909
933
1037
1060
1079
1107
0^1177
1205
E-1222
1223
1224
1241
1263
1264
1265
1329
1338
1379
Feed
Pressure
(psi)
300
300
300
300
300
300
300
310
310
275
285
290
290
290
290
300
300
300
300
300
300
300
300
300
300
300
300
300
300
300
300
300
310
310
310
310
320
320
310
320
320
320
320
330
330
330
330
330
Flux
(1/min)
1.12
1.20
1.12
1.13
1.16
1.04
1.28
1.20
.80
0.88
0.96
-._
0.96
0.96
1.16
0.88
0.90
0.96
0.90
0.92
0.88
0.88
0.84
0.84
0.80
0.80
0.80
0.80
0.80
0.72
0.72
0.64
0.64
0.68
0.68
0.56
0.59
0.52
3.5
3.5
3.6
3.8
3.56
3.6
3.4
3.5
3.48
Conversion
(X)
28
30
28
28
29
26
32
30
20
22
24
--
24
24
29
22
22
24
22
23
22
22
21
21
20
20
20
20
20
18
18
16
16
17
17
14
15
13
88
88
90
95
89
90
85
88
87
it
Feed xlO
3.4
3.4
3.4
2.1
2.5
2.5
2.4
2.4
2.4
2.4
2.4
2.4
2.4
2.2
2.3
2.3
2.3
2.3
2.3
2.3
2.5
2.5
2.4
3.6
3.6
3.5
3.5
3.5
3.5
3.5
3.5
3.5
2.8
2.8
2.7
2.7
2.7
2.6
0.12
0.11
0.11
0.07
0.042
0.036
0.038
0.080
0.060
Conductivity
s ppm Nad ) ,
Permeate- xlO
2.2
2.0
2.0
1.8
2.0
1.4
1.4
1.4
1.8
1.5
1.6
1.5
1.5
1.5
2.0
1.9
1.9
2.2
2.0
2.0
2.1
2.3
2.3
3.5
3.8
3.4
3.0
2.6
2.8
2.8
2.8
2.5
2.4
2.5
2.5
2.7
2.6
2.6
0.31
0.32
0.34
0.32
0.13
0.13
0.09
0.12
0.27
Rejection
(3)
94
94
94
92
92
94
94
94
92
94
93
93
93
89
91
92
92
90
91
91
92
91
90
90
89
90
91
92
92
92
92
93
91
91
91
90
90
90
74
71
69
54
69
64
76
85
55
38
-------
TABLE 6 (continued)
Cumulative
Exposure
Time
(hrs)
1507
1509
1529
1554
1625
1652
F-1652
1673
1676
1693
1700
1788
G*-1788
1814
1838
1952
1981
1999
H<-2071
2096
2150
2268
2408
Feed
Pressure
(psi)
320
360
350
350
350
350
350
320
320
350
340
350
350
350
350
350
350
350
350
350
350
350
350
Flux
(1/min)
3.20
3.6
3.52
3.60
3.24
3.36
0.96
0.96
0.80
1.56
1.52
1.52
1.12
1.12
1.12
1.12
1.12
1.12
1.12
1.32
1.30
1.40
1.24
Conversion
(%)
80
90
88
90
81
84
24
24
20
39
38
38
28
28
28
28
28
28
28
33
32
35
31
Feed x 10"4
0.039
0.036
0.036
0.036
0.036
0.036
2.6
2.4
2.4
1.6
1.6
1.2
2.4
2.4
2.4
2.7
2.4
2.4
2.4
1.8
1.7
1.7
1.7
Conductivity
(as ppm NaCl) ,
Permeate x 10
0.09
0.11
0.09
O.-l
0.09
0.11
2.1
2.8
2.7
1.3
1.4
1.3
2.7
3.0
3.1
3.6
3.6
3.6
3.7
2.7
3.0
2.8
2.7
Rejection
/0/\
(a)
77
69
75
69
75
69
92
88
89
92
91
89
89
88
87
87
85
85
84
85
82
83
84
* Initial Brightener Concentration: 145 ml in 58J!, solution = 2,500 ppm
A 290 ml Brightener Added (7,500 ppm Total)
B 290 ml Brightener Added (12,500 ppm Total)
C 145 ml Brightener Added (15,000 ppm Total)
D 145 ml Brightener Added (17,500 ppm Total)
E 1.5 qal Brightener plus 10 gal Water
F 20% Plating Chemicals, No Briqhtener, with Plating
G U Brightener Added (10,000 ppm), with Plating
H 1.9£ Brightener Added (30,000 ppm), with Plating
39
-------
100
c
o
0)
az
0
A
J_
90 -
80 -
70
60 -
(D
c
°^
50
J L
J L
D E
•V-!-
i
x-
i—r—r~"~r r
09,
r
c£r>fx
200
400
600
800
1000
L_
1200
1400
1 ____ L ____ L-.
1600 1800
J._. ..I ....
2000
2200
2400
Exposure Time, hrs
(Initial Brightener Concentration:
A 290 ml Brightener Added ( 7,500 ppm Total)
B 290 ml Brightener Added (12,500 ppm Total)
C 145 ml Brightener Added (15,000 ppm Total)
D 145 ml Brightener Added (17,500 ppm Total)
145 ml in 582, solution = 2,500 ppm)
E 1.5 gal Brightener plus 10 gal Water as Feed
F 20% Plating Chemicals, No Brightener, with Plating
G 1 £ Brightener Added (10,000 ppm), with Plating
H 1.9 H Brightener Added (30,000 ppm), with Plating
Figure 13. Conductivity Rejection vs. Exposure Time for Life Test With Brightener.
-------
It was hypothesized that the difference in results between the
Whyco field test and the laboratory test was associated with the presence of an
electric field in the Whyco plating tank. This field could affect the valence
state of selenium thus making it more reactive toward the membrane. To simulate
the plating process in the laboratory tests,a copperplate anode and a stainless-
steel-pipe cathode were mounted in the feed tank. A DC power supply operated at
1.5 volts and 1/2 amp was used to plate copper onto the cathode. The decline in
conductivity rejection with plating is shown in Figure 13 beyond point F. Between
points F and G the feed solution contained no brightener. Brightener was added
at points G and H. The rate of decline in rejection is about 4.7 times greater
with plating than without.
The productivity of the module during the life tests with brightener
is shown in Figure 14. There is a definite decline in flux over the first 1222
hours to approximately one-half of the initial value. However, after the massive
dose of brightener the flux recovered and remained stable although the level varied
somewhat at the points where the life test was interrupted for NaCl tests. The vari-
ation in flux level could be the result of flushing during the NaCl tests or of
variations in operating conditions before and after the life test interruption.
Standard sodium chloride rejection tests were conducted (400 psi,
77°F, 75% conversion, 1500 ppm feed) to follow the membrane performance after the
massive brightener dose. The NaCl rejection is shown as a faction of operating time
in Figure 15. Over the first 1222 hours the NaCl rejection declined only slightly
from the factory test value of 95%. When the module was exposed to the massive
brightener dose (point E) the rejection declined substantially (from 93% to 81%).
This indicates that the brightener can indeed attack the module if the total ex-
posure and/or concentration are sufficiently great.
The decline in rejection when plating was on-going in the feed
tank (both with and without brightener) is very interesting. This suggests that the
plating process does have some effect on the membrane performance. Comparing the
magnitude of the rejection decline for NaCl (Figure 15) and plating chemicals
(Figure 13), the greater decline for NaCl is to be expected. When rejection declines,
the species that have the lowest initial rejection (small univalent ions such as
sodium and chloride) have the highest rate of decline. Large multivalent ions
41
-------
Q.
+J
o
3
•o
o
D.
2.0
1.8
1.6
1.4
1.2
1.0
.8
.6
.4
.2
0
0
D E
JLL
8
I I I I I I
200 400
600
800
F G
rLr V-
QDJ
o
r-~r
i
1
1000 1200 1400
Operating Time, hrs
._...!__. __1. I i I i._ ._!.. 1
1600 1800 2000 2200 2400
(Initial Brightener Concentration: 145 ml in 58£ solution = 2,500 ppm)
A 290 ml Brightener Added ( 7,500 ppm Total)
B 290 ml Brightener Added (12,500 ppm Total)
C 145 ml Brightener Added (15,000 ppm Total)
D 145 ml Brightener Added (17,500 ppm Total)
E 1.5 gal Brightener plus 10 gal Water as Feed
F 20% Plating Chemicals, No Brightener, with Plating
G 1 i Brightener Added (10,000 ppm), with Plating
H 1.9 I Brightener Added (30,000 ppm), with Plating
Figure 14. Productivity vs. Exposure Time for Life Test With Brightener.
-------
50
200
400
600
800
1000 1200 1400
Exposure Time, Mrs
1600 1800
2000
2200
2400
(Initial Brightener Concentration 145 ml in 58£ solution = 2,500 ppm)
A 290 ml Brightener Added ( 7,500 ppm Total)
B 290 ml Brightener Added (12,500 ppm Total)
C 145 ml Brightener Added (15,000 ppm Total)
D 145 ml Brightener Added (17,500 ppm Total)
E 1.5 gal Brightener plus 10 gal Water as Feed
F 20% Plating Chemicals, No Brightener, with Plating
G 1 SL Brightener Added (10,000 ppm), with Plating
H 1.9 £ Brightener Added (30,000 ppm), with Plating
Figure 15. Sodium Chloride Rejection vs. Exposure Time for Life Test With Brightener.
-------
(plating solution) are initially rejected very well and show a slow rate of decline.
The results of Figure 13 are sufficiently interesting to warrant a more detailed
investigation of the interaction between brighteners (with and without plating)
and the B-9 membrane. It may be that the massive brightener dose initiated
the rejection decline, and the same rate of decline may have occurred without any
plating in the feed tank.
The module used in the brightener tests (Figures 13 - 15)
was returned to duPont for destructive analysis. Unfortunately, the fiber
length in a half-size module is too short to fabricate nrini-permeators. There-
fore permeation tests on the fibers themselves could not be made. The
findings were:
1. The NaCl rejection of the module,upon receit by duPont ,
was only 35% as measured under standard conditions (400 psi, 77 F, 75% conversion,
1500 ppm feed concentration).
2. There was no evidence of any mechanical defects.
3. There was no evidence of any scaling, particulates,
or foulants in the fiber bundle.
4. There was no significant deterioration in the tensile
properties or collapse resistance of the fibers.
5. The Reemay spacer showed significant deterioration in
physical properties and was visibly, damaged to such an extent that very poor
flow distribution of feed throughout the fiber bundle would occur during normal
operation.
6. Dye tests performed on the polyamide fibers to check
for skin damage showed a slight positive indication of skin damage. The subjective
evaluation of these tests indicated that a rejection as low as perhaps 80% could
be explained by the amount of skin damage observed, but there was insufficient
skin damage to account for a rejection of 35%.
These results indicate that the major cause of deterioration
in module performance may be attack of the Reemay spacer/flow-distributor
by the brightener rather than attack of the membrane itself. This will be
discussed in more detail in Section VI.
44
-------
SECTION V
PHASE II: FIELD TESTS AT NEW ENGLAND PLATING
GENERAL
Because of problems encountered in the plating operation at Whyco
Chromium Co., the RO field tests at that site had to be discontinued. These
problems were not directly related to the presence of the RO unit. After approx-
imately six-month's delay, a new field test site was located: New England Plating
Co., Inc., Worcester, Massachusetts and tests were resumed on a small copper
cyanide bath at this new location.
As shown in Table 1, the copper cyanide bath at New England Plating
is similar in composition to the Whyco bath. The major differences are that New
England Plating uses a potassium bath rather than sodium, and the pH is higher.
The same selenium-based brightener (MacDermid CI Bright Copper) is used at both
locations.
The plating line was a manual rack line which operated one shift per
day. The plating tank was about 450 gallons in size. It was preceeded by a two-
stage rinse (after an acid dip) and was followed by a two-stage rinse, an acid
dip, and a final rinse. Only one type of work was plated in this operation:
smooth round discs . Since the dragout for these parts was uncharacteristically low,
a drip-tank of plating solution was installed over the rinse during the final
week of operation to simulate a higher continuous dragout from the bath.
Because of the great difference in size of the two plating lines, the
demonstration unit used at Whyco was greatly oversized for New England Plating.
Therefore a smaller system was designed and installed at New England Plating.
This system was operated over a four month period during which membrane performance
Was monitored. During the last two months of operation the system ran largely
unattended. Sodium chloride tests were performed once every two weeks during this
period to monitor membrane performance.
45
-------
EXPERIMENTAL
Field Test System
A simplified flow schematic of the plating line and RO field
demonstration unit is shown in Figure 16. Feed was pumped from the rinse tank
by a booster pump (FlotecC6P8 centrifugal) and passed through a one-micron
cartridge filter. The pressure of the filtered feed was increased to the desired
operating pressure by a high-pressure, positive-displacement pump (Yarway
Cyclophram Model 072). Pressure pulsations were dampened by accumulators on the
pump suction and discharge. The feed was separated into a concentrate stream
and a permeate stream by a half-size duPont B-9 Permasep permeator (model 0420-021),
The permeate stream from the RO module was returned directly to the rinse tank.
The concentrate stream passed through a back-pressure regulator (BPR) which
controlled the operating pressure in the module. Most of the concentrate stream
was recycled to the suction of the high-pressure pump to maintain a sufficiently
high flow through the module. A float valve operating off the bath level returned
concentrate to the bath as needed to compensate for evaporation.
Pressures were measured before and after the filter to deter-
mine when the cartridge should be replaced. Pressures were also measured before
and after the RO module to determine the operating pressure and the pressure
drop. The system was protected against overpressurization by a pressure relief
valve and high pressure switch, and the pump was protected against running dry
by a low pressure switch.
The flow rates of the permeate and concentrate-to-bath
were measured. In addition the output of the high pressure pump was measured on
several occasions by the "bucket-and-stopwatch" technique and was found to be
constant at 1.03 (± .02) gpm. The concentrate recycle flow was determined by
difference.
Samples of the feed, permeate, and concentrate were obtained
through the sample valves shown in Figure 16. Because of the variations in the
rinse concentration, samples were generally taken during the afternoon after
the concentration in the rinse tank had reached a steady value for the day.
46
-------
Deionized
Make-up
Water
(C)
Evaporation
(A) (B)
Bath
(Concentration=B)
I
-X(J)
en
4PS
LPS
CG)
High Pressure
Pump (K)
RO Module
SV
SV
Figure 16. Flow Schematic for New England Plating Field Test System,
Point
A
B
C
D
E
F
G
H
I
J
K
Concentration
(Fraction of Bath)
0
0
B
.0023B
0
.0023B
.00858B
.056B
.056B
.056B
.00086B
-------
Permeate samples were taken first since the loss of permeate would not affect the
concentration of either the feed or concentrate. Feed samples were taken second
since the feed concentration is used in calculating rejection. Concentrate samples
were withdrawn third and could have been low in concentration if a large feed
sample was withdrawn just previously. The conductivity in the rinse tank was
continuously monitored by a conductivity probe and recorder.
Calculated flows and concentrations are shown at various points
in Figure 16. The evaporation from the bath was estimated by measuring the drop
in bath level with time when no make-up water was added. The drag-out rate was
estimated by measuring the increase in copper concentration with time in a still
rinse following the bath. The calculated flows (gallons per minute) and concen-
trations (fraction of bath concentration, B) are based on an assumed rejection of
90%, a conversion of 75%, and a maximum high-pressure pump output of 1.0 gpm.
The calculated rinse concentration is 0.3% of the bath concentration which
meets the requirement of a two-order-of-magnitude drop in concentration for each
rinse. This requirement was agreed to by New England Plating (and also by Whyco
in the first field test). The permeate from the RO module was returned to the
first rinse since its concentration was too high to be returned to a second
or third-stage rinse. The rinse shown in Figure 16 was inserted into the line
for the purposes of the RO demonstration. It was followed by a two-stage counter-
current rinse in order to assure well-rinsed parts regardless of the performance
of the RO system.
In addition to operating the unit in the normal mode shown in
Figure 16, the module was periodically tested with a standard 1500 ppm NaCl solution
at fixed conditions. For these tests the NaCl solution was mixed in an auxiliary
tank. Feed to the RO system was withdrawn from the tank, and the permeate and
concentrate were returned to the tank. (This mode of operation is identical to
that for the life tests described in Section IV). When steady state was reached
feed and permeate samples were analyzed for conductivity.
48
-------
Operating Conditions
The duPont Technical Information Manual for Permasep® products
recommends that the B-9 module be operated at 400 psi and 25 to 90% conversion.
The conversion is defined as the ratio of permeate flow to feed flow. Conversion
is limited to 90% in order to maintain a good flow distribution of feed through
the fiber bundle. If it is assumed that this limit is based on the rated produc-
tivity of the module (1.25 gpm of permeate for the half-size module), the rate of
concentrate withdrawal should be 0.14 gpm to maintain sufficient flow through the
fiber bundle.
Since the pump output was only one gallon per minute the module
had to be operated considerably below the 400 osi optimum in order to decrease the
Permeate flow rate to some reasonable fraction of the feed flow rate. Using the
criterion of a 0.14 gpm minimum concentrate withdrawal rate, the pressure should
be decreased to the point where the permeate flow rate is 0.86 gpm (86% conversion).
For the most part conversions ranged from 73 to 90% with an
average of 84%. The pressure varied from 135 to 205 psi with an average of 180 psi.
For the data reported, the feed temperature varied from 72 to 80°F.
For measurements on a standard NaCl feed solution, -the average
operating conditions were: conversion 74%, pressure 185 psi, and temperature
79°F.
Assays
Assays were performed for conductivity, pH, total solids (TS),
copper, and free cyanide. Most of the conductivities were measured with a
battery-operated hand conductivity meter, although in the field, conductivity was
often measured with the probes to the conductivity recorder. Good agreement
Was obtained with the hand meter. All other assays were performed by the
Walden Research Division of Abcor, Inc. A pH meter was used for pH, a gravi-
metric technique for total solids, atomic absorption for copper, and an ion selective
electrode for free cyanide.
49
-------
FIELD TEST RESULTS
Mechanical Operation
Aside from a few minor problems, the mechanical operation of
the system was satisfactory. As usual, problems that were encountered were associated
with the high-pressure pump. The original pump (which had been used for about
1000 hours previously) had to be replaced after about 250 hours of operating time.
Pressure pulsations associated with the high-pressure pump were, at times,
excessive but could be controlled by careful bleeding of all air from the lines
and keeping the accumulators charged to the proper pressure.
The temperature build-up in the rinse and feed to the RO system
(mainly because of pump energy imput in a closed loop) was less than observed
at Whyco. This can be accounted for by the difference in number and type of pumps
and by the lower ambient temperatures at New England Plating. The maximum
observed temperature of the feed to the RO system, during an operating period from
mid-August to late October, was 89°F. This is comfortable below the maximum
recommended operating temperature (95°F) of the B-9 module.
The level of suspended solids in the plating bath and rinse tank
was very low. The cartridge filter did not require replacement, and no significant
increase in pressure drop across the filter was noted during the entire field
test.
Bath and Rinse Concentrations
The plating bath was analyzed for copper metal, free cyanide and
caustic twice weekly by New England Plating and additions based on these analyses
were made twice weekly if necessary. Bath samples were obtained periodically
throughout the field test and analyzed by the Walden analytical laboratory in
order to verify that the bath composition remained constant. The results of these
analyses are given in Table 7. For the most part the bath composition remained
quite constant. The most notable exception is the free cyanide, and this may be
the result of the analyses rather than an actual change in bath concentration.
50
-------
TABLE 7. BATH CONCENTRATIONS AS A FUNCTION OF OPERATING TIME.
Cumulative
Operating
Time
(hrs)
72
46
105
138
180
225
326
418
487
575
644
736
809
922
967
1061
1130
Total
Solids
(mg/1 )
222,000
236,000
329,000
348,000
339,000
335,000
326,000
331 ,000
239,000
236,000
245,000
239,000
247,000
244,000
257,000
255,000
233,000
Copper
(mg/1)
19,000
24,000
44,000
48,000
43,000
48,000
42,000
42,000
43,000
46,000
49,000
49,000
49,000
47,000
44,000
46,000
40,000
Free
Cyanide
(mg/1)
3,600
3,600
10,000
8,800
8,800
10,000
10,000
22,000
31,000
19,000
19,000
Conductivity
(ymhos/cm)
290,000
290,000
250,000
280,000
270,000
270,000
260,000
250,000
280,000
270,000
290,000
260,000
270,000
240,000
280,000
285,000
250,000
PH
13.3
13.2
13.1
13.2
13.2
13.2
13.2
13.2
13.4
13.4
13.5
13.4
13.3
13.4'
13.3
13.4
13.3
51
-------
Rinse concentrations are given in Table 8 and depend primarily
on the amount of dragout from the bath prior to sampling. Starting at 967 hours a
drip tank was installed to continuously add bath to the rinse. During this period
rinse concentrations were significantly higher than before the drip tank was installed.
Flux
Data Correction - Flux is defined as the rate at which permeate passes through
a unit area of membrane surface when operated under specified conditions. The flux
is given by the equation:
J] = K] (AP - An) (1)
where:
J, = Flux (usually reported in gallons per sq. ft. per day)
K. = Constant (dependent on membrane properties and temperature)
AP = Difference in applied pressure across the membrane
An = Difference in osmotic pressure across the membrane.
In general the pressure and osmotic pressure on the permeate side of the membrane
are negligible relative to their respective values on the feed side. Equation
(1) then simplifies to:
J, = KI (p-n) (2)
where P and n are, respectively, the applied and osmotic pressure on the feed side
of the membrane. When significant conversion occurs the average feed-side osmotic
pressure must be used in equation (2).
The constant, K, is directly proportional to the diffusivity
of water through the membrane. As the temperature increases, the diffusivity
increases, and, by equation (2), the flux increases. All fluxes were corrected to
a temperature of 77°F using empirically determined data for the B-9 module from the
Technical Information Manual.
52
-------
TABLE 8. RINSE CONCENTRATIONS AS A FUNCTION OF OPERATING TIME
Cumulative
Operating
Time
(hrs)
22
46
105
138
180
225
326
418
487
575
644
736
809
922
967
1061
1130
Total
Solids
(mq/1)
319
283
236
102
182
201
140
68
310
195
291
10
465
303
895
1208
1689
Copper
(mq/1)
50
29
44
10
40
28
13
9
40
27
47
3.7
73
49
148
204
208
Free
Cyanide
(mq/1 )
29
13
29
2.7
39
55
88
110
110
Conductivity
(umhos/cm)
450
270
470
240
380
500
260
120
600
270
430
42
900
600
1500
2000
3300
PH
10
9.8
10.1
10.0
10.5
10.4
10.4
10.0
10.6
10.1
10.7
9.6
10.6
10.1
10.7
11.0
11.6
53
-------
Flux decreases with increasing feed concentration and con-
version (increasing It) as shown by equation (2). For dilute solutions as en-
countered in the field tests, the osmotic pressure is almost negligible compared
to the applied pressure. All flux data were corrected to a feed concentration of
1500 ppm using correction factors from the Technical Information Manual. These
correction factors were based on sodium chloride solutions so are not strictly
applicable to plating solutions. However, the correction was only minor: in only
two cases did it exceed 4%.
The flux is quite strongly dependent on operating pressure,
and since the module was operated significantly below its maximum (optimum)
pressure, a rather substantial correction factor (on the order of 100%) was applied
to correct the flux to 400 psi. This correction factor was taken from the duPont
Technical Information Manual arid closely approximated a direct proportionality to
P as given by equation (2) (with n neglig ble).
Normal Operation - Flux data for the field demonstration are given in Table
9 and Figure 17. The flux is presented in terms of the module productivity
(gpm of permeate). The operating time (pressurized operation on plating waste)
during which flux and rejection were measured for plating rinse water was 1130
hours (47 days). The corresponding exposure time (pressurized and non-pressurized
exposure to plating waste) was 1500 hours (62 days). The unit was operated for
a total of 100 days, but data were not obtained on the flux and rejection of plating
salts during the latter stages of operation. The flux (corrected to 400 os1,
75% conversion, 1500 ppm feed concentration, and 77°F) decreased only slightly,
from 2.3 gpm initially to 2.0 gpm after 1130 hours.
A decrease in flux is usually attributed either to membrane
compaction or fouling. The observed decrease (15% in two months) is greater than
expected for compaction (5% in one year at these conditions). It is therefore
possible (based on these data) that some fouling occurred. In many cases foulants
can be removed by a simple cleaning procedure. However, no attempt was made
to clean the module following the field test.
54
-------
TABLE 9. MODULE PRODUCTIVITY AS A FUNCTION OF
OPERATING TIME AND OPERATING CONDITIONS
Cumulative
Operating
Time
(hrs)
22
45
105
138
180
225
326
346
418
485
487
574
575
642
644
736
806
809
922
967
1056
1061
1128
1130
Feed
Pressure
(psi)
170
180
136
165
165
160
170
170
170
180
205
170
200
180
195
175
185
200
180
200
195
200
200
195
Conversion
(X)
87
87
72
90
83
90
88
90
89
83
88
88
86
89
86
87
88
84
76
77
78
77
58
73
Feed
Conductivity
(umhos/cm)
1400
800
825
550
380
800
400
320
120
470
600
520
350
420
950
88
600
1100
1250
3800
3100
4100
17500
7750
Temp.
(°C)
26
23
27
56
27
27
27
27
28
24
23
28
26
26
23
27
25
23
22
21
22
23
27
24
Measured
Productivity
(qpm)
.90
.90
.74
.93
.86
.93
.91
.93
.92
.86
.91
.91
.89
.92
.89
.90
.91
.87
.78
.79
.80
.79
.60
.75
Corrected
Productivity
(qom)
2.25
2.34
2.34
2.42
2.15
2.48
2.24
2.29
2.18
2.16
2.08
2.16
1.86
2.16
2.10
2.08
2.17
2.01
2.07
1.94
2.00
1.92
1.93
2.04
55
-------
tn
g:
O_
t!
£
» *
-o
o
O-
4.0
3.5
3.0
2.5
2.0
1.5
1.0
.5
0
1 I 1 I i 1 I r r
—
—
I-^x_n 0 0
~ O
—
1 1 II 1 L 1 .. ,,l JL
100
~T—
J I
200 300 400 500 600 700 800 900 1000 1100
Operating Time, Mrs
Figure 17. Corrected Productivity vs. Operating Time (NEP Co.).
-------
NaCI Tests - At various times during the four month operation, the
RO system was shut down, flushed, and operated on a standard 1500 ppm NaCI so-
lution with total recycle. The results of these periodic tests are given in
Table 10. The flux is plotted as a function of operating time in Figure 18.
The flux decline with time is very slight. In fact, a line of zero slope
would fit the data points quite well. The loss of flux calculated by an
extrapolation of the curve in Figure 18 is about 9% per year. Thus, from
the NaCI tests, the loss of flux can be attributed primarily to compaction of
the polyamide fiber. In any case, it can be concluded that the flux remains
quite stable with operating time.
Rejection
Data Correction - The rejection of a membrane is defined by
r = CF " CP - 1 - 5E. (3)
CF CF
where:
r = Rejection usually expressed in %
Cp = Concentration of species in feed
Cp = Concentration of species in permeate.
The dependence of rejection on operating pressure and osmotic pressure can be
derived by noting that the passage of salt through the membrane is given by:
J2 = K2(AC) = K2(CF - C ) (4)
where:
J2 = Flux of solute
1C = Constant characteristic of membrane
AC = Solute concentration difference across membrane,
57
-------
TABLE 10. SODIUM CHLORIDE FLUX AND REJECTION
umulative
perating Feed
i »\
Corrected^
Time Pressure Convers1onv"' Temperature Conductivity ( mhos)
(hrs) (psi) (X)
0 195 81
178 190 75
347 175 79
485 220
643 165 75
806 210 79
967 165 62
967 195 75
1128 180 75
1462 245 76
1918 245 78
2400... 240 75
2400(f 210 78
2400^9) 400 68
(°F) Feed
3125
2700
85 2750
80 3150
88 3100
77 2700
75 3000
76 3000.
87 2200
64 1400
68 1400
Permeate
470
700
650
800
700
800
650
1000
\ 800 i \
I) 60° e
f 450 !
65 1600 l( 410)^
74 1500ie; 37Q(e)
1950
265
Rejection
(%)
85
74
76
75
77
70
78
67
64
57
68
74
75
86
Flux
(gpm)
1.84(c)
r.7o(c)
1 .71, .x
1.37(d)
1.66
1.65
1.69
1.71
1.50
1.68
1.60
1.66
l!70(c)
Corrected^*
Rejection
(*)
95
88
92, .,
86(d)
91
88
86
90
83
74
83
84
87
83
Conversion based on 1.0 gpm
Corrected to 77°F, 400 psi,
Temperature assumed 77°F.
) Conversion assumed 75%.
ppm Nad.
) Measured at Wai den pilot lab
) Measured at Abcor pilot lab
of feed to module. Measured flux in gpm =
con version/ 100.
1500 ppm Nad , and 75% conversion.
after return of module.
after return of module.
-------
E
o>
£
£
(J
T3
O
S-
Q.
..„,,,,, | i i i r — r--i —
1.80-
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
n
— O-JD r>^ 0 (-} P
^J ^J * ^ «. rg
0
O
— —
—
— —
— —
— —
— —
1 1 1 1 1 1 1 1 1 1 1 1
0 400 800 1200 1600 2000 2400
Operating Time, Mrs
Figure 18. Corrected Productivity for Standard NaCl
Solution vs. Operating Time (NEP Co.).
59
-------
The concentration of solute in the permeate (C ) is equal to the flux of solute
(Jg) divided by the total flux of material (J-, + J«)» or» since J2 « J-|> equa-
tions (1), (3), and (4) combine to give:
1
r- - - (5)
^ (AP - An)
Since both K, and Kp have the same temperature dependence, the rejection is
essentially independent of temperature.
Equation (5) indicates that as the feed concentration and
conversion increase (All increases), the rejection decreases.
The correction is not difficult to apply if the average feed
concentration can be estimated and the average osmotic pressure calculated.
However, for plating solutions the rejection does not follow the dependence
on feed concentration given by equation (5). Figure 19 shows the rejection
as a function of feed concentration for the Rochelle copper cyanide bath
(2 3}
tested in the in-house pilot phase of the program v- -: The rejection
increases with increasing feed concentration up to a total dissolved solids
concentration of about 5%. The ionic equilibria are more complex for plating
solutions than for simple salts. In solution, cuprous ions and cyanide ions
associate to form the following complexes:
Cu
Cu (CN)
Cu (CN)|
The degree to which the above complexes are formed depends on the molar con-
centration of the solution. As the solution becomes increasingly dilute the
complexes tend to dissociate. Since rejection increases with ionic size and
ionic charge, rejection can be expected to increase with increasing concen-
tration until a point is reached where the formation of larger complexes no
longer outweighs the effects on rejection of the increase in osmotic pressure
(equation [5]).
60
-------
0
30
60
90
93
96
o
LJUI
LU
99
99.3
99.6
99.9
O Total Dissolved Solids
Q Cu+
A CN"
O Conductivity
10 15 20
TOTAL DISSOLVED SOLIDS
25
Figure 19. Rejections for Rochelle Copper Cyanide Rinse Waters,
61
-------
Because of this unorthodox dependence of rejection on concentration, the data
were not corrected for feed concentration and conversion.
The correction curve given in the duPont Technical
Information Manual was used to correct for the effect of operating pressure on
rejection. This curve is consistent with equation (5) over the pressure range
of interest for the great majority of the data. For very concentrated solutions,
the duPont correction curve is inapplicable and equation (5) was used directly.
Dependence on Time - Conductivity rejections are given in Table 11 and Figure 20.
The rejection decreased over the first 300 hours of operating time to a value of
about 70% and then appeared to increase again although there is considerable scatter
in the data. The scatter may be due, in part, to the dependence of rejection
on feed concentration. The feed concentration could not be controlled at a set
value, and the data were not corrected for variations in feed concentration. A
drip tank was installed at 967 hours of operating time to continuously add bath to
the rinse tank, simulating a continuous dragout. During this period the feed
concentration was greater than the maximum feed concentration observed without the
drip tank. The rejections between 967 and 1130 hours were better, on the average,
than the rejections prior to the installation of the drip tank. This suggests that
the rejection increases with increasing feed concentration which is contrary to the
theory and typical behavior for simple salts.
Copper rejections are given in Table 12 and Figure 21.
The variation in copper rejection with operating time is similar to the variation
of conductivity rejection shown in Figure 20. The copper rejection goes through a
minimum of about 80% at 500 hours. During the drip tank operation the copper
rejections were very good.
62
-------
TABLE 11. CONDUCTIVITY REJECTION AS A FUNCTION OF OPERATING TIME AND OPERATING CONDITIONS.
CJ
Cumulative
Operating
Time
(hrs)
22
45
105
105
138
180
225
326
326
346
348
418
418
485
487
487
547
575
575
642
644
644
7-36
736
806
809
809
922
967
1056
1061
1061
1128
1130
1130
Feed
Pressure
(psl)
170
180
135
135
165
165
160
170
170
170
175
170
170
180
205
205
170
200
200
180
195
195
175
175
185
200
200
180
200
195
200
20H
200
195
195
Conversion Conductivity (pmhos/cm)
(5K) Feed Permeate Concentrate
87
87
72
72
90
83
90
88
88
90
89
89
89
83
88
88
88
86
86
89
86
86
87
87
88
84
84
76
77
78
77
77
58
73
73
1400
800
825
1100
550
380
800
400
3330
320
260
120
140
470
600
800
520
350
430
420
880
950
88
TOO
600
1100
1200
1250
3800
3100
4100
5000
17500
7750
6000
200
150
200
240
220
290
340
240
210
225
180
91
no
250
370
435
340
185
210
310
400
350
38
52
345
470
620
360
8000
800
1100
1200
4100
1000
1800
8000
3800
4175
5000
3000
5000
3500
1700
1350
1700
1800
350
600
1500
2500
3200
2500
1100
1500
1400
3250
2800
100
135
1550
4250
5000
4000
10000
9500
13600
13000
32000
9750
18000
Rejection
(%)
86
81
76
78
60
32
58
40
36
30
31
24
21
47
38
46
35
47
51
26
54
63
57
48
42
57
48
71
79
74
73
76
76
87
70
Corrected
Rejection
(%)
94
91
92
93
84
73
84
75
73
71
70
68
67
76
67
72
73
73
75
66
77
82
81
77
73
78
73
87
89
87
86
88
88
94
85
-------
(J
OJ
CD
o:
o
3
•o
O
100
90
80
70
60
50
40
30
20
10
0
Drip Tank
Installed
1
1
100 200 300 400 500 600 700
Operating Time, Mrs
800
900
1000 1100
Figure 20. Corrected Conductivity Rejection of Plating Salts vs. Operating Time (NEP Co).
-------
TABLE 12. COPPER REJECTIONS AS A FUNCTION OF OPERATING TIME AND OPERATING CONDITIONS
01
Cumulative
Operating
Time
(hrs)
22
46
105
138
180
225
326
418
487
575
644
736
809
922
967
1061
1130
Feed
Pressure
(psi)
170
180
135
165
165
160
170
170
205
200
195
175
200
180
200
200
195
Conversion Copper Concentration (mg/1 )
W Feed Permeate Concentrate
87
87
72
90
83
90
88
89
88
86
86
87
84
76
77
77
73
190
120
140
50
90
80
40
12
70
48
78
83
136
134
425
510
660
10
7
11
11
15
23
15
7
27
20
27
51
35
23
60
81
58
1250
710
740
380
50
530
230
65
400
190
560
113
615
580
2000
2180
2440
Rejection
(*)
95
94
92
78
83
71
62
42
61
58
65
38
74
83
86
84
91
Corrected
Rejection
(%)
98
97
97
91
93
89
84
76
79
78
82
73
87
92
93
92
96
-------
O
o
O)
s-
ai
o.
CL
o
o
100
90
80
70
60
50
40
30
20
10 L_
0
I
0 TOO 200 300 400 500 600 700
Operating Time, Mrs
800
900
*
A Drip Tank ~
JLi __________ L.._.
1000 1100
Figure 21. Corrected Copper Rejection vs. Operating Time (NEP Co).
-------
Similar behavior is also noted for the total solids
rejections given in Table 13 and Figure 22.
The pH's of the feed, permeate, and concentrate were also
measured and used to calculate the hydroxide ion rejections given in Table 14.
In many cases negative rejections were obtained indicating that the rate of transport
of OH~ through the membrane was faster than the rate of transport of water. The
average rejection of OH" calculated from the values of Table 14 is very nearly zero
percent.
Free cyanide concentrations were measured only during the
latter part of the field test. The concentrations and rejections (uncorrected)
are given in Table 15.
Dependence on Concentration - The rejection of copper cyanide plating salts
appeared to improve during the period that the drip tank was operated, suggesting
a positive correlation between rejection and feed concentration. The corrected
conductivity rejection is plotted against the conductivity of the feed in Figure
23. Although there is considerable scatter in the data, a positive correlation
between rejection and feed concentration is obtained.
The relation between feed concentration and rejection
was investigated directly. Feed solutions of various concentrations were prepared
in an auxiliary feed tank by diluting a portion of the plating bath. The RO
unit was operated in a total recycle mode on each feed solution and samples were
analyzed for conductivity. Results are given in Table 16. A modified procedure
was used to correct the data because of the high osmotic pressure of some samples.
The correction procedure is given in Table 16. Both the uncorrected and corrected
rejections show the same trend: an increase in rejection with increasing feed
concentration. This follows the theory outlined previously, i.e., dissociation
at low concentration to species which are poorly rejected.
67
-------
TABLE 13. TOTAL SOLIDS REJECTION VS. OPERATING TIME AND OPERATING CONDITIONS
Cumulative
Operating
Time
(hrs)
22
46
105
138
180
225
S 326
418
487
575
644
736
809
922
967
1061
1130
Feed
Pressure
(psi)
170
180
135
165
165
160
170
170
205
200
195
175
200
180
200
200
195
Conversion
87
87
72
90
83
90
88
89
88
86
86
87
84
76
77
77
73
Feed
1106
600
65
982
753
418
215
56
433
293
603
83
755
846
2495
3348
4647
Permeate
154
169
136
112
134
154
73
16
228
161
239
51
332
223
410
599
857
Concentrate
5926
3966
3341
1968
3170
2488
1162
361
2105
936
2820
113
3872
3000
12100
12370
16480
Rejection
86
72
—
88
82
63
66
71
47
45
60
38
56
74
84
82
82
Corrected
Rejection
(X)
94
87
--
95
93
86
86
88
72
72
80
73
77
88
92
91
91
-------
10
C
o
o
-------
TABLE 14. OH" REJECTIONS VS. OPERATING TIME AND OPERATING CONDITIONS
Cumulative
Operating
Time
(hrs)
22
46
105
138
180
225
326
^ 418
0 487
575
644
736
809
922
967
1061
1130
Feed
Pressure Conversion
(psi)
170
180
135
165
165
160
170
170
205
200
195
175
200
180
200
200
195
(X)
87
87
72
90
83
90
88
89
88
86
86
87
84
76
77
77
73
Feed
10.3
10.0
9.7
9.6
10.0
10.6
10.1
9.9
10.7
9.8
10.6
10.0
10.5
10.3
10.9
11.2
12.0
pH
Permeate
9.9
9.6
10.2
10.0
10.5
10.6
10.5
10.0
10.7
10.1
10.6
7.0
10.6
9.7
10.6
11.0
11.5
OH"
Concentrate
10.8
10.4
9.4
10.0
10.0
10.1
10.1
9.5
10.9
9.7
10.7
9.2
10.0
10.0
11.4
11.8
12.3
Rejection*
(*)
60
60
-68
-60
-68
0
-60
-20
0
-50
0
**
-21
75
50
37
68
* Rejection
rejection
based on concentration
is defined by r = (CF
(moles
- V".
per liter)
j. Thus the
of hydroxide ipn.
minimum rejection
For negative rejections
is -100%.
the
** Analyses questionable.
-------
Cumulative
Operating
Time
(hrs)
487
575
644
736
809
922
967
1061
1130
Feed
Pressure
(psi)
205
200
195
175
200
180
200
200
195
Conversion
(X)
88
86
86
87
84
76
77
77
73
Free
Feed
34
20
36
6
55
190
700
190
940
Cyanide Concentration
Permeate
18
10
23
1.2
3.4
24
39
42
39
(mg/1 )
Concentrate
120
52
135
24
260
220
880
780
780
J.UIXO
Rejection
(X)
47
50
36
80
94
87
94
78
96
-------
•—I
ro
•r-
>
100
90 —
80 —
70
60
50
40
O
3
-o
§ 30
20
10
0
TOO
1
200 300 500 1000 2000
Feed Conductivity (ymhos)
5000
"^0-
10,000 20,000
Figure 23. Conductivity Rejection vs. Feed Conductivity (NEP Co).
-------
co
TABLE 16. CONDUCTIVITY REJECTION AT VARIOUS FEED CONCENTRATIONS
(TOTAL RECYCLE MODE OF OPERATION)
Run
1
2
3
4
Pressure
(psD
215
205
225
305
Conversion
(X)
75
75
75
76
Temperature Flux
(°F) (gpm)
60
64
68
73
0.75
0.75
0.75
0.75
Conductivity (ppm as NaCI)
Feed Permeate Concentrate
170
500
2300
9000
60
150
450
1800
500
1800
9000
29000
Rejection
(%)
65
70
80
80
Corrected*
Rejected X
75
82
89
86
The following procedure was used to correct the data. The fluxes were all corrected to a temperature of 77°F. Equation (1) was
applied to Run No. 1 with ATI assumed equal to zero, and ^ was calculated. Using this valur of Kj along with the temperature-
corrected flux and the measured pressure P = AP, values of An were calculated for runs 2, 3, and 4 using equation (1). Salt
passages were then corrected using these values of ATI in equation (5).
-------
NaCI Rejections - While the rejection of various plating salts is important
in determining the extent to which plating chemicals can be recovered, it is
difficult to determine the rejection stability from these data because of the
dependence of rejection on feed concentration. The true measure of rejection
stability of the membrane with operating time is obtained from the standardized
NaCI tests. These rejections are given in Table 10 and are plotted in Figure 24
as a function of operating time. Only a moderate decline in rejection occurred
over the 100 days of operating time: from 90% to 85%. This decrease is
acceptable for certain plating applications as will be shown in Section VI.
It should be emphasized that the rejection decline for
NaCI will be greater than for the plating chemicals. When the membrane re-
jection declines, the rejection declines most rapidly for species which are
poorly rejected (small monovalent ions such as sodium and chloride). The
d< .line in rejection is slower for species which are highly rejected (large
multivalent ions such as copper cyanide complexes). Thus, the rejection
decline for NaCI should give a conservative estimate of the rejection decline
for copper cyanide plating salts.
74
-------
o
OJ
O)
o:
o
=3
-o
O
o
o
to
100
90
80
70
60
50
40
30
20
10
0
1 1 1 1 1 1
•O1
0
"O"
o
1 1 1 1
1
1 1
1 1
o
I 1
400 800 1200 1600 2000
Operating Time, Mrs
2400
Figure 24. Corrected Rejections for Standard NaCl Solution
vs. Operating Time (NEP Co).
75
-------
SECTION VI
DISCUSSION
WHYCO FIELD TEST
During field tests at Whyco Chromium Co., both the flux and rejection of
the membrane declined within a period sufficiently short to make RO unattractive
on the basis of membrane replacement costs. Based on the results of tests
conducted on one of the Whyco modules, the reason for the decline in performance
appears to be two-fold:
1. Chemical and physical degradation of the Reemay wrap-
material/flow-distributor, and
2. Chemical degradation of the skin of the hollow fiber
membranes.
Laboratory tests conducted to identify the bad-actor constituent were
successful in simulating the degradation of Reemay when massive doses of
brightener were used. In the operation of a module, feed is distributed radially
outward from a porous tube running down the axis of the module. In passing from
the distributor tube to the outer radius of the module, the feed passes through
concentric layers of hollow fine fibers, each layer separated by a thin paper-
like material called Reemay. In a four-inch module, there are about 16 of these
concentric fiber layers. The Reemay wrap material acts as a flow distributor
by holding the fibers in position. Without this material, the feed solution
would create channels through the fiber bundle.
The Reemay removed from the brightener-life-test module showed a regular
pattern of destruction particularly in outer layers. The most severly damaged
portions lined up to form a channel of low flow resistance from the central
distributor tube to the outer radius.
The poor rejection performance observed during the latter part of the
life test with brightener (Figure 15) was probably the result of poor flow
distribution in the module. In pockets where the flow 1s very low, the con-
76
-------
centration builds up because of the permeation of water through the fibers.
In addition, when the flow is low the boundary layer is thicker, and the
concentration at the membrane surface builds up relative to the bulk concentration
(concentration polarization). In regions of very low flow the osmotic pressure
at the membrane surface can approach the operating pressure of the module.
Since the salt flux through the membrane is directly proportional to the con-
centration of salt at the membrane surface, poor flow distribution leads to
high salt passage or low rejection. Thus the degradation of the Reemay results
in low overall rejections even though the hollow fiber membranes themselves
remain intact.
While deterioration of Reemay was simulated during the laboratory tests,
no significant deterioration of the hollow fiber membranes was observed during
these tests even in the presence of massive doses of brightener. This is
particularly evident from the high and stable rejections observed during mini-
permeator tests. Since mini-perinea tors do not contain Reemay and are not limited
by poor flow distribution, they give a direct indication of membrane performance
as opposed to module performance. In addition, dye tests on some of the fibers
from one of the laboratory test modules indicated that membrane attack was not
the major reason for the decline in module performance when exposed to massive
doses of brightener. It is concluded that the polyamide membrane is highly
resistant to the brightener and to the other major constituents of the bath.
Since chemical degradation of the membrane fiber could not be simulated
in the laboratory tests, it is concluded that the constituent responsible for
chemically attacking the membranes at Whyco is not a major bath constituent.
At present, its identity has not been determined.
It is evident that a wrap-material/flow-distributor is essential to the
proper operation of a duPont hollow fiber permeator. However, it appears that
Reemay is not sufficiently resistant for copper cyanide applications. Contacts
with the manufacturer have Indicated that it would be possible to substitute a
more chemically resistant material for Reemay on a special-order basis. It
would appear that application of hollow-fiber permeators to cyanide baths
77
-------
should be based on such modules.
The field test at Whyco illustrates the danger of extrapolating
laboratory results to actual applications. In terms of module performance the field
test was unsuccessful even though laboratory life tests were very promising.
It is recommended that field tests be conducted on the specific waste stream to be
treated prior to the purchase of RO equipment. Meaningful field tests can be
conducted with a relatively low level of effort. The system can consist of
little more than a cartridge filter, a pump, and a half-size module operated on the
overflow from the first rinse tank. It is not necessary to return either the
concentrate or permeate to the plating operation. A sodium chloride flux and
rejection test before and after three months of field operation should give a
good indication of membrane stability.
Alternatively, it is recommended that a performance guarantee be obtained
from the supplier of membrane equipment. This guarantee will likely require the
supplier to conduct field tests on the bath unless previous experience indicates
the application is a highly successful one.
NEW ENGLAND PLATING FIELD TEST
The results of the field test at New England Plating appear much more
favorable. The flux stability, as determined by the standard NaCl performance
tests, was quite good. Within the scatter of the data, fouling was not signi-
ficant. The rejection stability, as determined by the standard NaCl performance
tests, was much better than at Whyco, but a moderate decline was observed. This
decline may again be related to degradation of the Reemay wrap material.
The economics for closed-loop RO treatment of this particular plating bath
can be estimated from the data obtained during the field test. A more generalized
presentation of economics is given in the following section.
The present rinsing system at New England Plating consists of two tanks
operated countercurrently; no chemicals are recovered by this system. The
maximum allowable concentration in the final rinse is 100 mg/1 of total solids
78
-------
-4
or about 4x10 times the bath concentration. The RO system wotild be designed
to operate as shown in Figure 25. A single half-size B-9 module would be op-
erated at 75% conversion to given 2.0 gpm of permeate. The capital cost for
such a system would be about $8,500.
Figure 26 shows the second rinse concentration as a function of rejection.
The rejection can decrease to about 65% before the concentration limit is ex-
ceeded. For design purposes it is assumed that the decline in rejection follows
that measured for NaCl. Thus, by extrapolation of Figure 24, the life of the
module is 500 days.
The breakdown of the operating cost is shown in Table 17. The total
operating cost for the RO system is $2.94 per day.
The operating cost of the RO system can be offset by credits resulting
from closed-loop recovery. The major credits result from recovery of plating
chemicals and from the savings in destruction chemicals previously used to
oxidize cyanide and precipitate copper. The credit resulting from the reduction
in water usage is minor in comparison. Table 18 gives the breakdown of credits
for New England Plating. Based on one operating shift per day, the total credits
amount to $2.65 per day.
The operating cost is almost entirely offset by the credits resulting from
closed loop operation; however, for this particular plating line the credits are
insufficient (in relation to the operating cost) to make the capital investment
attractive on a purely economic basis. Therefore, an RO system would be recom-
mended for New England Plating only if complete closed-loop treatment using the
present two-stage rinse were required. Since this particular plating line is a
manual rack operation, closed-loop treatment could be achieved by adding more
countercurrent rinse stages. Theoretically, a three-stage rinse would give a final
rinse purity very close to the specified concentration.
79
-------
DI Make-up
water 0.023 gpm
Evaporation
0.023 gpm
00
o
Dragout
0.0013 gpm
Recycle
0.647 gpm
Feed
2.67 gpm
Half-Size B-9 Module
75% Conversion
Permeate
2.0 gpm
Bath
Concentrate Return To
Bath 0.023 gpm
Figure 25. Schematic of Closed-Loop RO Recovery System for Copper Cyanide Bath at New England Plating Co.
-------
QJ
U
o
o
ra
CO
o
0
10
in
OJ
o
o
a;
o
tD
•P
C
(U
U
C
o
o
0 10 20 30 40 50 60 70 80
Rejection, %
90 100
Figure 26. Concentration in Second Rinse vs. Rejection for Half-Size
B-9 Module Operated at 75% Conversion.
81
-------
TABLE 17. BREAK-DOWN OF OPERATING COSTS FOR NEW ENGLAND PLATING
1. Power (at $0.036/kwh)
Major power requirement is for high pressure pump
(flow rate = 2.67 gpm; AP = 400 psi ; motor/pump efficiency = 50%)
Power Consumed = sQ = 1-25 hp
Daily Cost = (1 .25)(.745)(8)(0.036) = $0.27 per day
2. Module Replacement ($720 each); 500 day life
Daily Cost = - = $1.44 per day
3. Maintenance (5% of capital investment per year)
Daily Cost = (-05)(^50°)= $1.16 per day
365
4. Deionized Water
(Based on cost of $2.00 per 1000 gal from central deionizer
which uses RO for water pretreatment)
Daily Cost = (.023)(1440)($.002) = $0.07 per day
TOTAL OPERATING COST = $2.94 ($1.02 per 1000 gal permeate)
82
-------
TABLE 18. CREDITS REALIZED FOR RO OPERATION AT NEW ENGLAND PLATING
1. Chemical Credits
Bath composition and unit cost of chemicals:
Constituent Concentration Unit Cost
CuCN 8.5 oz/gal $1.87/lb
(Cu as metal) (6.0 oz/gal)
KCN 16.0 oz/gal $0.61/lb
Rochelle Soln 6% Vol $3.55/gal
Brightener 2000 ppm Vol $5.20/gal
Value of plating solution = $1.83/gal
-4-
Minimum recovery of RO system = 1 " ^x 10 — = 99.96%
Daily Savings per shift = (.9996)(.0013)(1440)($1.83) = $] J4
J day
2. Water and Sewer Credits
Assume water and sewer costs at^ $0.50/1 000 gal
Present water requirements for two-stage countercurrent rinse and
final rinse concentration of 4 x 10"^ times bath concentration =
62 gpd (one shift per day)
Daily Savings = (62) ($0.0005) = $0.03 per day (one shift per day)
3. Chemical Treatment Credits
Total cyanide concentration in bath = 8.9 oz/gal
Daily dragout = (•0013'8-9) = 0.35 Ib/day
Requirements for chemical destruction: Caustic = 1.0 Ib/lb CN
Chlorine = 8.0 Ib/lb CN
Cost for chemicals as used: Caustic • $0.22/lb NaOH from 50% soln
Chlorine = $0.50/lb C19 from 15% NaOCl
* soln
Treatment cost = $4. 22/1 b CN
Daily Savings = (0.35)($4.22) = $1.48 per day
TOTAL CREDITS = $2.65 per day
83
-------
GENERAL ECONOMIC PROJECTIONS
Care must be exercised in comparing the results from these two field tests
and in extrapolating the results observed at New England Plating to other
plating operations. The dragout at New England Plating was very small, and,
compared to Whyco, a much longer operating time was required to give the membranes
an equivalent exposure to plating chemicals. In addition, the ratio of evapor-
ation to dragout at New England Plating was about 18 compared to about 10 at
Whyco. Therefore, the concentrate returned to bath at Whyco was more concentrated.
If the deterioration in module performance is related to the concentration of
plating chemicals, a more rapid decline would be expected as the evaporation-
to-dragout ratio decreases. Since it is impossible on the basis of present
information to accurately extrapolate the life test data from one plating bath
to another, it is recommended that a life test be conducted on the particular
bath to be treated.
The capital cost for a closed-loop RO recovery system depends primarily
on the size of the system in terms of the gallons of permeate per day that it
can produce. Beyond this rather broad generalization, there are many factors
which can significantly affect the capital cost but are often related in a
complex way to the specific requirements for a particular installation. For
example, the flux has a direct influence on the amount of membrane surface
area required to achieve a given system output (in gallons of permeate per
day). As the flux declines, the required number of membrane modules increases,
and the capital cost increases.
The flux is determined, in part, by the intrinsic permeability of the
membrane to water, the extent of compaction and fouling, the conversion at
which the module is operated, and the degree to which the rinse waters must
be concentrated. The degree of concentration depends on the ratio of bath
evaporation to dragout which can vary widely from application to application.
For baths with a low ratio of evaporation to dragout, the concentrate returned
to bath must be highly concentrated resulting in a low flux. For these baths
it may be more efficient to partially concentrate the rinse water with RO
('to a concentration at which the flux becomes uneconomically low) and then
84
-------
use an auxiliary evaporator to reduce the volume of the RO concentrate
to be returned to the bath.
The capital cost of an RO system can also be affected by membrane
rejection. If the rejection is too low to meet the platers' specification for
the final rinse purity, additional purification will be required. The
permeate from the RO system could be treated with a second RO system or with
ion exchange. This would add significantly to both the capital and operating
costs.
Nevertheless, approximate capital costs are shown as a function of
system capacity in Figure 27. These capital costs are based on the rated
productivity of B-9 modules as determined with a 1500 ppm NaCl solution at
400 psi, 77°F, and 75% conversion. Also shown in Figure 27 is the capital cost
for membranes alone. This curve can be used to estimate the additional cap-
ital cost for applications where the average productivity is lower than the
rated productivity. The cost for membranes, based on the rated capacity,
varies from about 10% to 25% of the total capital cost for the range of capacities
covered by Figure 27.
Typical operating costs as a function of system capacity are shown in
Figure 28. These costs are based on the same assumptions as given in Table 17,
but they do not include the cost for deionized make-up water which must be
based on the cost and usage for each particular application. In addition,
these operating costs are based on the rated productivities of the modules.
The operating cost given in Table 17 tor New England Plating is somewhat lower
than the cost indicated 1n Figure 28. This is due to the high productivity for
this application which allows a half-size rather than a full-size module to be
used. For applications where the average productivity 1s below the rated
productivity, the operating cost will be higher than shown 1n Figure 28 since
more membrane modules must be replaced.
It should be emphasized that these costs are only approximate. For more
accurate costs, quotes should be obtained for the specific plating bath to be
treated from manufacturers of membrane equipment.
85
-------
100,000
10,000
00
O
o
a.
«t
CJ
1,000
100
i i i i
i i
1,000
Capital Cost for
RO System
r
Capital Cost for
Membranes Only
10,000
100,000
RO SYSTEM CAPACITY, gal of permeate per day
Figure 27. Typical Capital Costs for RO Systems.
86
-------
100
GO
O
o
10
CL.
O
I I I \s\ \ \ 1_
I I I I I I I I
1000 10,000 100,000
RO SYSTEM CAPACITY, gal of permeate per day
Figure 28. Typical operating costs for RO systems as a
function of capacity and membrane life.
(Does not include cost of DI make-up water.)
87
-------
VII. REFERENCES
1. Federal Register. March 28, 1974 pp. 11510-14
2. Donnelly, R.G., Goldsmith, R.L., McNulty, K.J., and Tan, M., "Reverse Osmosis
Treatment of Electroplating Wastes", Plating. 61. (5) 432 (1974)
3. Donnelly, R.G., Goldsmith, R.L., McNulty, K.J., Grant, D.C., and Tan, M.,
"Treatment of Electroplating Waste by Reverse Osmosis", Draft Final Report,
EPA Contract No. R-800945-01
4. Steward, F.A., "EPA Discharge Regulation", Metal Finishing. 72. (9), 47 (1974)
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-77-170
2.
3. RECIPIENT'S ACCESSIOf*NO.
4. TITLE AND SUBTITLE
REVERSE OSMOSIS FIELD TEST:
COPPER CYANIDE RINSE WATERS
TREATMENT OF
5. REPORT DATE
August 1977 issuing date
6. PERFORMING ORGANIZATION CODE
Kenneth J. McNulty, Robert L. Goldsmith, Arye
I. Gollan, Sohrab Hossain, Donald Grant, Wai den
Research Div. of Abcor. Inc.. Wilmington. MA 01887
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
10. PROGRAM ELEMENT NO.
The American Electroplater's Society, Inc.
Winter Park, Florida 32789
1BB610
11. CONTRACT/GRANT NO.
R-800945
12. SPONSORING AGENCY NAME AND ADDRESS
13. TYPE OF REPORT AND PERIOD COVERED
Industrial Environmental Research Lab.
Office of Research & Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
- Cin., OH
Final
14. SPONSORING AGENCY CODE
EPA/600/12
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Field tests of reverse osmosis (RO) were conducted on copper cyanide rinse
waters at two different sites: Whyco Chromium Co. and New England Plating Co.
At both sites, closed-loop treatment was used with plating chemicals recycled
to the bath and purified water recycled to the rinsing operation. The objective
of the tests was to establish, under actual plating conditions, the feasibility
of RO treatment for copper cyanide plating wastes.
It was concluded that RO can be used to close the loop in copper cyanide
plating. However, care must be taken to insure that adequate membrane life can
be achieved. Where membrane life approaches that in traditional RO applications,
the capital and operating costs for RO, compared to those for alternative treat-
ment processes, are attractive. The cost attractiveness of RO depends on several
factors specific for each installation. Bases for assessing capital costs,
operating costs, and process credits are presented.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
COS AT I Field/Group
Electroplating
Waste Treatment
Membranes
Copper Cyanide
Rinse Water*
Closed-loop Treatment*
Reverse Osmosis
Treatment Costs
13 B
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (ThisReport)
Unclassified
21. NO. OF PAGES
101
20. SECURITY CLASS (Thispage)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
89
ftl).S.«OVBMI»ITnmmM OFFICE: Mn-757-05W6Ul
------- |