WATER POLLUTION CONTROL RESEARCH SERIES • 17O4O—O5/7O
RENOVATION OF
MUNICIPAL WASTEWATER
BY REVERSE OSMOSIS
ENVIRONMENTAL PROTECTION AGENCY • WATER QUALITY OFFICE
-------
WATER POLLUTION CONTROL RESEARCH SERIES
The Water Pollution Control Research Series describes the
results and progress in the control and abatement of pollu-
tion of our Nation's waters. They provide a central source
of information on the research, development, and demon-
stration activities of the Water Quality Office, Environ-
mental Protection Agency, through inhouse research and grants
and contracts with Federal, State, and local agencies, re-
search institutions, and industrial organizations.
j
Inquiries pertaining to the Water Pollution Control Research
Reports should be directed to the Head, Project Reports
System, Office of Research and Development, Water Quality
Office, Environmental Protection Agency, Washington, D.C. 20242,
-------
RENOVATION OF MUNICIPAL WASTEWATER BY REVERSE OSMOSIS
By
*
Mr. John M. Smith, Sanitary Engineer
Mr. Arthur N. Masse, Chief*
Mr. Robert P. Miele, Sanitary Engineer
MUNICIPAL TREATMENT RESEARCH PROGRAM
ADVANCED WASTE TREATMENT RESEARCH LABORATORY
Water Quality Office
ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO
**
LOS ANGELES COUNTY SANITATION DISTRICT
LOS ANGELES, CALIFORNIA
MAY 1970
For sale by the Superintendent ot Documents, U.S. Government Printing Olnce
Washington, D.C., 20402 - Price 65 cents
-------
ABSTRACT
The increasing demand for clean water makes clear the need for
improved treatment of municipal wastewater. Deliberate reuse of
wastewater requires efficient removal of "residual" organic and inor-
ganic solutes. A common and effective approach has been to supplement
biological processes with chemical and physical processes that, together,
are capable of achieving water quality consistent with reuse requirements,
A more direct approach is to use a single process such as reverse osmosis
which exhibits a potential to separate a broad spectrum of contaminants
from primary and secondary effluents.
The pilot plant studies reported herein were addressed toward
defining the problems associated with the reverse osmosis processing
of variously treated wastewaters. The effects of feed water quality
on membrane performance were evaluated, as were methods to control
membrane fouling. A comparison of three configurations; flat plate,
spiral-wound, and tubular was made to determine their relative merit
in treating wastewater. Finally, a cursory examination of costs
was made.
ii
-------
CONTENTS
Page
TITLE PAGE i
ABSTRACT ii
CONTENTS ill
INTRODUCTION iv
HISTORY OF REVERSE OSMOSIS DEVELOPMENT 1
PILOT PLANT INVESTIGATIONS 4
Pomona, California 4
Spiral-Wound Configuration 4
Tubular Configuration 16
Lebanon, Ohio 27
Flat-Plate Configuration 27
DISCUSSION AND INTERPRETATION OF PILOT PLANT RESULTS 51
CONTRACT AND RELATED INVESTIGATIONS 53
RELATIVE POSITION AND FUTURE ROLE OF REVERSE OSMOSIS
IN RENOVATING WASTEWATER 55
REFERENCES 59
iii
-------
INTRODUCTION
Since the fresh water resources of the world are limited, it is
axiomatic that our ever-increasing water needs will demand more
efficient use of our present water supplies.
One way of improving this efficiency is through multiple use of
available surface water. This,of course,requires treatment technology
capable of achieving water quality consistent with the planned future
uses of the water. In many cases,, present-day treatment techniques will
not be adequate, and must be supplemented or replaced with advanced
processes that can effect a high removal of undesirable contaminants.
Of primary concern is the removal of nutrients, dissolved mineral
salts, oxygen-demanding organics and undesirable bacteria and viruses
presently being discharged as a result of inadequate wastewater
treatment.
Advanced waste treatment of primary or secondary municipal effluents
is currently accomplished by a series of unit processes, each capable of
selectively removing a particular fraction of the residual waste load.
This approach has been successful but could be enhanced through the
development and utilization of a single process exhibiting removal
efficiencies of the combined unit processes.
Reverse osmosis is a process with the demonstrated capacity to
separate dissolved organics and inorganics as well as the larger
organic species from aqueous solution, and is therefore worthy of
evaluation as a potential advanced waste treatment process.
iv
-------
HISTORY OF REVERSE OSMOSIS DEVELOPMENT
The early work of Reid and co-workers at the University of Florida
first demonstrated the effectiveness of cellulose acetate as a desali-
nation membrane (1). Further investigation by Loeb and Sourirajan
(1959-1960) at the University of California, using aqueous magnesium
perchlorate as a swelling agent in the casting solution for cellulose
acetate membranes, resulted in a membrane that exhibited much higher
water permeation rates while still maintaining the solute rejection
characteristics of the earlier membranes. The original Loeb-Sourirajan
film formulation, identified as batch No. 25, is shown in Table 1
(2) (3) (4).
TABLE 1
CASTING SOLUTION BATCH NO. 25
COMPONENT (WT %)
Cellulose Acetate 22.2
(acetyl content 39.8%)
Acetone 66.7
Magnesium Perchlorate 1.1
Water 10.0
TOTAL 100.0
This formulation established the practicality of reverse osmosis
as a desalination process, and has been reported extensively in the
literature. Innumerable modifications of the early Loeb formulation
have been made since 1960 in an effort to improve the initial flux
and solute rejection characteristics of the membranes. These modifi-
cations have included basic changes in the casting formulation as well
as improvements in the casting and curing techniques. The basic cellu-
lose acetate designated by Eastman Kodak as E-398-3 is now used in
combination with additives and other cellulosic derivatives to produce
optimum membranes for the conversion of brackish and sea water. Much
of the membrane development work to date has been sponsored by the
U. S. Department of the Interior's Office of Saline Water.
- 1 -
-------
As the development of membranes progressed , the need for a practical
module configuration became apparent and the need was met by the design and
testing of three principal configurations; the plate and frame, the
spiral-wound, and the tubular modules. Extensive pilot plant evalu-
ation of these modules on brackish water illuminated the problems of
concentration polarization and scaling that could be expected with
these feedso The effect of concentration polarization was minimized
by maintaining turbulent flow conditions within the brine channels
while the precipitation of sparingly soluble salts was controlled by
depressing the feed stream pH to a value of 6.0 or less.
The application of reverse osmosis in treating municipal wastes
was first considered late in 1962 when preliminary investigation by
Aerojet-General Corporation at Azusa, California indicated that the
cellulose acetate membranes were an effective barrier for both organic
and inorganic solutes. Following preliminary testing, a contract was
awarded to Aerojet by the Public Health Service's Advanced Waste
Treatment Program (now FWQA) to conduct an in-depth bench-scale study
to evaluate the potential of reverse osmosis for renovation of muni-
cipal wastes.
This study was conducted on filtered municipal secondary effluent
using 3-inch diameter flat-plate test cells operated at pressures of
750 and 1500 PSIG. The feed stream was recirculated during the tests
and a 5-10 fold increase in feed concentration was produced. The pH
of the feed stream was maintained at 5.5 by the addition of sulfuric
acid. Three types of cellulose acetate membranes were used for these
tests with nominal flux rates that varied from 10 to 60 gallons per
square foot per day (GFD). Results of this study demonstrated that
high quality water could be produced by this process from secondary
effluent. Results of two tests are shown in Table 2 and indicate
greater than 95% removal of total dissolved solids (TDS) and chemical
oxygen demand (COD)(5).
It was also found that high flux (30-60 GFD) membranes were
unstable and, after a few hours of operation, exhibited flux rates
as low as the lower flux (10-20 GFD) membranes. Generally, the flux
rates were considered too low to be practical. Several possible
causes of the high flux rates were proposed, including the possibility
of organic fouling, poor hydrodynamic conditions and excessive concen-
tration of organic solutes. There was also some question regarding
the resistance of the membranes to biological attack.
This initial bench-scale evaluation using the flat-plate test
cells was followed in early 1965 by a series of experiments conducted
at the Pomona Pilot Plant located adjacent to the Pomona Water Renovation
Plant in Pomona, California. This study was a cooperative effort
between the County Sanitation Districts of Los Angeles County and the
- 2 -
-------
TABLE 2
WATER MINERAL ANALYSIS
INITIAL
BENCH SCALE STUDY
AEROJET-GENERAL . CORPORATION
TDS
ABS
COD
PH
Cations
Na+
K++
NH+
Ca^
M ++
Mg
Fa**-
Anions
Cl"
N0~
HC03
C°3
so;
Si°3
PO; (total)
Feed Water
(Typical)
ppm
550
4.5
95
51
85
40
25
125
50
N.D.
65
2
260
0
200
30
25
Product
(Test 18)
ppm
15
0.1
2
6.3
4.6
4.9
0
0
0.7
0
20.3
0
8.1
0
1.7
2.9
0
Product
(Test 28)
ppm
28
0.1
6.0
5.5
6.1
2.7
3.9
2.6
0
0
22.1
0
3.5
0
3.3
2.9
0
N.D. - not determined
1 - pH adjusted during test.
- 3 -
-------
General Atomic Division of the General Dynamics Corporation (b). The
experiments were conducted at a pressure of 200 PSIG at 5% recovery
using the spiral-wound modules developed by General Atomic. The feed
to the test loop was diatomaceous earth-filtered secondary effluent.
Results of these tests again demonstrated the high organic and
inorganic solute rejection seen in the earlier tests. Also noted
during these tests was a significant decrease in membrane flux with
time. This was attributed to fouling of the membranes by deposition
of organic material present in filtered secondary effluent. This
deposit could be partially removed by increasing local turbulence
at the membrane surface and by establishing a backward flow potential
through the membrane.
These preliminary results, while promising, clearly indicated
the need for continuing the investigation of reverse osmosis at the
pilot plant scale to further evaluate the fouling potential of various
treated waste streams, to determine optimum operating conditions, to
examine different module configurations, and to determine the required
pretreatment for sustained operation.
PILOT PLANT INVESTIGATIONS
(Pomona, California)
Spiral-Wound Configuration
The first pilot plant investigation of reverse osmosis was con-
ducted at the Pomona Water Renovation Plant under a joint FWQA-LACSD
contract in 1966. This study was conducted using a 5000 GPD unit
employing the spiral-wound module shown in Figure 1. The module
consists of 3/8" PVC pipe to which is attached a permeate flow channel
enclosed on each side by a cellulose acetate membrane. The membrane
is glued around the edge to form an envelope through which the per-
meate flows to the center collection pipe. The feed enters the
module axially and is distributed through the module by a polyethy-
lene vexar brine-side spacer. This spacer is rolled up between the
membranes to form a cylindrical module which is then placed into the
pressure vessel. A completed module is shown in Figure 2. The
arrangement of the individual modules within a pressure vessel is
shown in Figure 3, and Figure 4 is a flow diagram of the entire unit.
The first 11 months of operation of the spiral-wound module on
secondary effluent were plagued by problems of module construction
and mechanical failures. The major problems were excessive module
leaking, brine seal failures, and telescoping of modules. These
problems were compounded by a faulty pH control system and frequent
pump failures.
- 4 -
-------
ROLL TO
ASSEMBLE
FEED SIDE
SPACER
FEED FLOW
PERMEATE FLOW
(AFTER PASSAGE
THROUGH MEMBRANE)
PERMEATE OUT
PERMEATE SIDE BACKING
MATERIAL WITH MEMBRANE ON
EACH SIDE AND GLUED AROUND
EDGES AND TO CENTER TUBE
FIGURE 1. REVERSE OSMOSIS MODULE ASSEMBLY
-------
SEE DETAIL A
BACKING MATERIAL
MESH SPACER
MEMBRANE
PERMEATE TUBE
GLUE LINE
DETAIL A
FIGURE 2. MODULE FABRICATION
-------
FEED
VICTAULIC
COUPLING
ANTI-CORROSION
PLASTIC COATING
PRESSURE TUBE
MODULE
INTERCONNECTOR
PURGE VALVE
PURIFIED
T, WATER
TUBE
END CAP-
CONCENTRATE
OUTLET TUBE
FIGURE 3. ARRANGEMENT OF SPIRAL-WOUND MODULES IN A
MULTIMODULE PRESSURE VESSEL
-------
CARBON
TREATED
SECONDARY
EFFLUENT
! i
innir
PUMP
pH CONTROL
SYSTEM
I I MCl
I I SUPPLY
MIXING
CHAMBER
MOYNO
PUMP
SPIRAL-WOUND MODULES
\
PRESSURIZED
FEED
u
J.-I
I I
.L.l-1
PRESSURE REGULATING VALVE
f PRODUCT WATER
BRINE
FIGURE 4
FLOW DIAGRAM OF GULF GENERAL ATOMIC
REVERSE OSMOSIS TEST FACILITY, POMONA, CALIFORNIA
-------
Notwithstanding these early difficulties, it was concluded that
the spiral-wound modules could process secondary effluent with high
organic and inorganic rejection, but that some form of pretreatment
was necessary to prevent excessive clogging of the brine-side spacers
from suspended solids present in secondary effluent.
Bearing this in mind, and considering the possible fouling poten-
tial of soluble organic species, it was decided to pretreat the reverse
osmosis feed by passage through a granular carbon adsorption process.
Two runs were conducted on the carbon-treated secondary effluent.
Objectives of these two runs are shown below:
1. To determine the water-flux-time relationship for the
system when operating on an effluent from an activated
carbon plant.
2. To monitor the water quality produced by the membranes and
note any change in water quality with time.
3. To achieve a high rate of water recovery from the system
and to determine, if possible, any adverse effects on the
system as a result of high-water recovery.
4. To test the structural integrity of spiral-wound modules
being subjected to a long-term test under normal environ-
mental conditions.
5. To determine the materials responsible for any membrane
fouling which might occur.
6. To investigate procedures for minimizing or preventing
membrane fouling, along with methods for cleaning fouled
membranes.
A summary of the operating conditions for the two experimental
runs are shown in Table 3. It should be noted that for Run #1 the
diameter of modules within the seven series pipes are progressively
decreased to provide a relatively constant brine velocity, while Run
#2 was conducted at a higher initial, but gradually decreasing, brine
velocity since all of the modules are 4-inches in diameter.
- 9 -
-------
TABLE 3
SUMMARY OF CONDITIONS FOR EXPERIMENTAL
RUN IN GULF GENERAL ATOMIC REVERSE OSMOSIS UNITS
No. of Modules in Unit
Module Diameter
Total Membrane Area
Feed Water Source
Feed Pressure
Initial Feed Flow
Initial Brine Flow
Initial Water Recovery
pH Control
Chlorination
Run #1
85
45t4", 20-3", 20-2"
550 ft2
Carbon Effluent
400 psi
2.8 gpm (4040 gpd)
0.56 gpm (805 gpd)
807c
to 5.0 using HC1
5.0 mg/1 dosage
using NaOCl
4.8 gpm (6910 gpd)
1.0 gpm (1440 gpd)
5.0 mg/1 dosage
using NaOCl
A summary of the data from each experiment is shown in Figures 5
and 6. These figures show the time dependent variation of flux and
salt rejection throughout the run. Salt rejection was calculated for
these runs by use of the formula:
Salt rejection
Average feed conductivity - Product conductivity
Average feed conductivity
where,
Average feed conductivity
Feed conductivity + brine conductivity
2
All data was obtained from readings taken daily and then averaged
over 100-hour intervals.
The flux decline curve for Run #1 can be divided into 3 segments -
0 - 600 hours, 700 - 1600 hours, 1700 - 2500 hours. The abrupt
increase in flux rate between 600 and 700 hours is due principally
- 10 -
-------
8
o 7
Sl'
z
x 5
£ 4
3
2
1
n
^ % SALT REJECTION
"*""'"'x.-.x— x.x»x.^-N
•XB
\ • • —
-
FLUX (GFD)
r'x^,. ^^^
.-•— «x._^.-.
-
RUN -#1
FEED -CARBON TREATED
SECONDARY EFFLUENT
PRESSURE • 400 PSIG
i i i
1UU 5
0
95 S
ac
90 5
85 ^?
TIME IN HOURS
FIGURE 5
FLUX AND SALT REJECTION PERFORMANCE FOR GULF GENERAL
ATOMIC SPIRAL-WOUND REVERSE OSMOSIS MODULE, POMONA, CALIFORNIA
-------
10
=: 4
% SALT REJECTION 10°
FLUX |GFD)
"*~"*^»»«.»* m'* '***
"-•-•^.s ^,
RUN •# 2
FEED -CARBON TREATED
SECONDARY EFFLUENT
PRESSURE-400 PSIG
100055005500
TIME IN HOURS
FIGURE 6
FLUX AND SALT REJECTION PERFORMANCE FOR GULF GENERAL
ATOMIC SPIRAL-WOUND REVERSE OSMOSIS MODULE, POMONA, CALIFORNIA
95 2
90
-------
to a replacement of six modules in the unit and partially to slight
leakage of feed water through the membrane which developed during
that period. The increase in flux rate between 1600 and 1700 hours
resulted from an increased leakage of feed water through the membranes
which is also reflected in the salt rejection curve of Figure 5.
The only abnormality in the flux vs. time curve for Run #2
occurs between 2300 and 2600 hours. This is a result of two factors,
(1) the replacement of six modules in the unit; and (2) a one-week
depressurized flushing of the unit with tap water.
A summary of the feed and product water quality data for Runs #1
and #2 is shown in Table 4. The data for Run #1 is divided into two
time periods: 0 - 1600 hours, when all of the modules were functioning
properly, and 1600 -' 2500 hours, when several of the modules had
developed leaks. The differences in the relative level of N03-N and
NH3-N between the two time periods shown in Table 4 resulted from tha
method of operation of the Pomona secondary treatment plant. All of
the data shown in the two tables is based on average values of 24-hour
composite samples taken throughout the time period involved. Samples
were collected for analyses twice weekly.
As the product-water flow decreased throughout each run, the
water recovery also decreased. However, since the final brine flow
rather than the feed flow was being held constant, the decline in
water recovery did not accurately reflect the declining product water
flow. Thus, the decline in water recovery from the beginning to
the end of Run #1 was from 80% to 78%. Comparable values for Run ''2
were from 80% to 647o.
Attempts were made throughout both experiments to control the
flux decline in the membranes or to restore flux after decline had
occurred. Previous experiments at Pomona had indicated that a daily
15-minute air-tap water flush with the system in a depressurized
condition was successful in controlling excessive pressure drops
across the brine side of the modules. The excessive pressure drops
resulted from clogging of the brine passageways by suspended material
in the feed water. This procedure was followed throughout both
experiments, and while it was again successful in controlling head-
loss, it did not appear to have any significant effect on restoring
flux rates.
Along with the air-tap water flushing, two other methods of
cleaning the membranes were investigated. One procedure, which has
been successful in cleaning Gulf General Atomic modules fouled when
in use on a brackish water, was a citric acid recirculation through
the system. The other cleaning attempt was a procedure employing an
anionic detergent solution flush. Neither of these procedures were
successful in restoring membrane flux.
- 13 -
-------
TABLE 4
FEED AND PRODUCT WATER QUALITY
DATA FOR EXPERIMENTAL RUNS NO. 1 & 2
IN GULF GENERAL ATOMIC REVERSE OSMOSIS
Run No.l
TIME 0 - 1600 HRS Feed
COD (mg/1) 10.8
NH3-N (mg/1) 9.2
NO,-N (mg/1) 2.4
J
P04-P (mg/1) 10.1
TDS (mg/1) 623
Product
1.7
1.7
0.8
0.2
73
Brine
43.8
49.0
7.5
57.7
3402
UNIT
7, Rejection
93.8
94.2
84.0
99.4
96.4
TIME 1600 - 2500 HRS
COD (mg/1) 12.2 1.9 37.2 92.3
NH3-N (mg/1) 5.3 1.5 25.1 90.1
N03-N (mg/1) 12.2 6.0 34.9 74.6
P04-P (mg/1) 9.0 1.4 43.2 94.6
TDS (mg/1) 543 145 2738 91.2
Run No. 2
COD (mg/1)
NH--N (mg/1)
j
NO--N (mg/1)
j
PO. -P (mg/1)
b
TDS (mg/1)
11.4
17.1
2.1
9.7
552
0.7
2.9
0.8
0.07
51
37.4
59.9
8.9
39.8
2576
97.1
92.5
85.5
99.7
96.7
Notes: (1) All analyses run on 24-hour composite samples
(2) 7o Rejection calculated using average feed where,
Average feed = feed + brine
(3) Figures shown are average values.
- 14 -
-------
Each run was interrupted because of a mechanical problem with
the Moyno feed pump. In each instance, the unit was inoperative for
3 to 4 weeks while the pump was being repaired. During this time,
tap water was run through the unit at line pressure. When the pump
was repaired and the unit was returned to normal operation, most of
the modules had developed excessive leaks and were producing water of
unacceptable quality. The modules remained in operation for several
weeks after the run was resumed to see if the leaks would seal them-
selves; but, to the contrary, the product water quality continued to
deteriorate until the end of each run. Investigation of the membranes
indicated that the leakage was caused by a combination of membrane
deterioration, and failure of glue and seals in the modules. No
satisfactory explanation has yet been advanced for the cause of the
membrane deterioration.
The difference in initial flux rate between Run #1 and Run #2 is
due mainly to improvements in membrane casting and module fabrication
techniques made by Gulf General Atomic in the interim period between
the start of each run. However, some of the difference may be due to
the higher brine flows used during Run #2. While these higher brine
flows did not increase the brine velocity in the downstream modules
because of the constant diameter modules used during that run, they
did result in an increased brine velocity in the upstream modules.
This increased brine velocity may have decreased the boundary layer
enough to result in higher flux.
In only one instance during the two runs was the effect of
high water recovery on the system evident. At the start of Run #2,
the pH of the feed water was maintained at 5.5 rather than 5.0 as had
been the case in all previous runs. After 200 hours of operation,
the flux of the last set of modules on stream had shown a significant-
ly greater decline than the other modules in the system. A one-hour
acid flush of these modules returned the flux to a level consistent
with the remaining modules. Calculation of solubility levels, along
with consideration of boundary layer factors, indicated that a form
of calcium phosphate was possibly precipitating on the membrane surface,
After acid flushing, the pH was maintained at 5.0 and no further evi-
dence of excessive flux decline was observed in the last set of modules,
Results of these two experimental runs support the following
observations:
1. The extent of flux decline in the spiral-wound reverse
osmosis system operating on activated carbon effluent is
far greater than would be tolerable in a practical system.
2. The quality of the water produced by the reverse osmosis
process is excellent.
- 15 -
-------
3. An air-tap water-flushing procedure was successful in
controlling headless through the system.
4. All attempts to clean fouled membranes were unsuccessful.
5. The high-water recovery obtained in these experiments had
no adverse effect on the downstream modules at a pH of 5.0.
6. The cause of the apparent membrane deterioration during pro-
longed tap-water flushing has not been determined.
Tubular Configuration -
The second configuration to be evaluated at the Pomona Water
Renovation Plant was of tubular design, manufactured and furnished
by the Havens Company.* The module configuration for this unit is
shown in Figure 7. A module consists of seven, 8-foot long,
1/2-inch inside diameter fiberglass tubes coated on the inside with
a cellulose acetate membrane. The individual tubes are arranged in
series and are enclosed in a PVC shroud to collect the product water
that permeates from the feed stream side (inside) to the outside of
each tube. The complete unit included 48 modules, a high pressure
pump, acid addition equipment and the necessary piping, valves and
controls.
Six separate experiments were conducted during this phase of
the investigation utilizing three membrane types (designated as Type 3A,
4A, and 5A) on two waste streams, secondary effluent and granular
carbon-treated secondary effluent. Types 3A, 4A and 5A refer to the
relative "tightness" of the membranes. The operating conditions for
each of the six experiments are summarized in Table 5. The quality
of the feed and product water for both feed streams is tabulated and
shown in Table 6. The membrane flux for both clean and fouled modules
is compared to the "rated" value in Table 7. A normalized flux decline
pattern for both feed streams is shown in Figure 8.
These results demonstrated greater than 92% removal of both
organic and inorganic solutes with the exception of NH-j-N and N03~N,
which were poorly rejected at pH 5.0. Examination of Table 7 illus-
trates that the "clean module flux" values are much lower than the
"rated flux" for the two higher flux modules, and that the flux
decline was more severe for the higher flux module, i.e. 35% for the
3A module rated at 27.0 GFD and 10% for the 5A module rated at 10 GFD.
Comparison of the flux decline slopes shown in Figure 8 for each feed
indicates a two-week flux decline of 137<> for the carbon-treated second-
ary effluent and a 17% decline for the secondary effluent feed. While
this difference is not large, it may be attributed to higher COD and
turbidity of the secondary effluent.
* Havens Industries, San Diego, California. Mention of products by
name does not constitute endorsement by the Federal Water Quality
Administration.
- 16 -
-------
FIGURE 7
SCHEMATIC OF HAVENS
REVERSE OSMOSIS MODULE
OUT
1] FIBERGLASS TUBE
2) OSMOTIC MEMBRANE
3) END FITTING
4) PVC SHROUD
to collect product water
5) PRODUCT WATER
6) FEED SOLUTION
7) EFFLUENT
5
3
1
6
-------
TABLE 5
SUMMARY OF EXPERIMENTS ON HAVENS REVERSE OSMOSIS UNIT
Experi-
ment Feed
I Secondary
Effluent
II Secondary
Effluent
i
i— Ill Secondary
00 Effluent
i
IV Activated
Carbon
Effluent
V Activated
Carbon
Effluent
VI Activated
Carbon
Effluent
Feed
Pressure
(psi)
800
800
700
800
850
800
Feed
Flow
(gpm)
2.5
2.5
3.0
3.2
3.8
3.7
Type
Module
4A,5A
4A.5A
3A,4A,
5A
4A,5A
4A,5A
4 A, 5 A
Additive
To
Influent
50 mg/1
Sodium
Sulfite
5 mg/1
Calgon
CL-77
(See Note 1)
(See Note 1)
(See Note 1)
(See Note 1)
Duration
(hrs)
20
25
325
140
170
460
Initial
Recovery
(?.)
78
75
58
50
78
69
Final
Recovery
<%)
48
45
47
50
78
59
Reason
For
Termination
Rapid Flux
Decline
Rapid Flux
Decline
Pump
Failure
Pump
Failure
Excessive
Module
Breakage
Excessive
Module
Breakage
Notes: (1) 87» Sulfuric acid added to control product water pH at 5.0 *
(2) No acid addition during first 40 hours.
-------
TABLE 6
WATER QUALITY DATA FOR HAVENS REVERSE OSMOSIS UNIT
Secondary Number
Effluent of
as Feed Samples
COD (mg/1)
Turbidity (JTU)
TDS (mg/1)
P04 (mg/1)
NH--N (mg/1)
J
Activated Carbon
Effluent As Feed
COD (mg/1)
TDS (mg/1)
PO^ (mg/1)
NH--N (mg/1)
j
NOQ-N (mg/1)
J
Cl (mg/1)
8
6
4
6
4
7
5
6
3
4
2
Feed
45.3
12.8
607
30.5
11.4
8.9
536
26.3
10.9
3.9
121
Product
1.9
0.1
26
0.15
1.5
0.6
40
0.25
1.4
1.2
10
% Rejection
95.8
99.4
95.7
99.5
86.8
93.4
92.5
99.0
87.0
69.4
91.9
Note: All analyses run on grab samples.
- 19 -
-------
TABLE 7
COMPARISON OF MODULE FLUX
Module
Type
3A
4A
5A
Rated
Flux 2
(gpd/fO
27
18
10
"Clean" Modules
Flux (gpd/ft2)
Ave . Range
17.5
12.4
10.8
14.0-2C.8
11.3-14.0
8.6-12.0
"Fouled" Modules
Flux (gpd/ft2)
Ave . Range
11.4
9.7
9.5
9.1-14.0
8.0-12.4
8.2-11.5
% Decline
From Clean
State
35
22
10
Notes:
1. Rated flux obtained using 5,000 mg/1 sodium chloride solution at
600 psi, 25°C, flow rate of 1.5 gpm. Values as reported in Havens
Industries literature.
2. "Clean" and "Fouled" fluxes obtained at 800 psi, influent flow
rate of 2.5 gpm and influent IDS of 600 mg/1.
3. "Clean" modules flux represents values taken shortly after
experiment was begun.
4. "Fouled" modules flux represents values taken after recovery
had fallen to 45% - 50%.
5. "Clean" and "Fouled" modules flux represent readings from fourteen
Type 3A modules, seven Type 4A modules and ten Type 5A modules.
- 20 -
-------
12
4 Jfe
10
8
6
4
2
PRODUCT WATER CONDUCTIVITY
^<— • •
\ . A
-w\/\ ]
A
/-/ V\..--.. /s_
>. FLUX (GFD) N
v . ^/\
\''' "^X,.-^^ ^..^. xx- — ^
H^
EXPERIMENTS -#3 & #6
"" SECONDARY EFFLUENT FEED*AS SHOWN
— tARBUN IKtAltU
SECONDARY EFFLUENT
1 1 1 I ! 1 1 1 1 1 I 1
^ ° § s
ER C8HOUCTIVITY ( MHI
p"
40 5
Cd
ca
^
0 SO 100 150 200 250 300 350 400 450 500
TIME IN HOUtS
FIGURE 8
FLUX AND PRODUCT WATER CONDUCTIVITY FOR HAVENS
TUBULAR REVERSE OSMOSIS UNIT, POMONA , CALIFORNIA
-------
These early results, employing the tubular configuration, were
sufficiently encouraging to warrant continuing the investigation. A
Universal Water Corporation reverse osmosis unit was used for this
second-stage effort. This system utilized a 1/2-inch plastic perfor-
ated tube approximately 50 inches long, into which is fitted a preformed
tubular membrane sealed at each end with 0-rings. Eighteen of these
elements were connected in series and placed within a 4-inch diameter
PVC pipe to form a single module that contains 7 sq ft of membrane
surface. The complete experimental unit was comprised of thirty-two
of these modules arranged in banks of four modules each and plumbed
originally in a 5-3 array as shown in Figure 9. This arrangement
was modified after Run #1 to increase the initial system recovery
from 40% to 73%. Two experimental runs have been completed to date
with this unit using granular carbon-treated secondary effluent as
feed.
Run #1 was conducted at a feed pressure of 600 PSIG with the feed
acidified to pH 5.0 using sulfuric acid. A brine flow of 4.90 GPM
was maintained with a feed flow of 8.10 GPM and product water flow
of 3.20 GPM. The total headloss across the system increased from an
original value of 90 PSIG to a final value of 110 PSIG. The flux
decline and salt rejection for Run #1 are shown as a function of
time in Figure 10, and the feed and product water quality are pre-
sented in Table 8. Run #1 was terminated after 1000 hours in order
to replumb the unit in a 3-2-2-1 array in order to achieve higher
overall recovery. Following the shutdown, the modules were flushed
for one hour with a 0.75% bioenzyme solution* that had proven effec-
tive in restoring the flux decline in a Gulf General Atomic reverse
osmosis unit also being tested at Pomona. This technique increased
the flux from 18.0 GFD to 20.6 GFD, compared to an original flux of
23.9 GFD.
Run #2 was conducted under the same conditions as Run #1 except
for the recovery ratio, which was increased from 40% to 73%. This
caused the exit brine flow to decrease from 4.9 GPM, which for a
1/2-inch diameter tube is equivalent to a bulk velocity of 2.0 ft/sec,
to 1.0 GPM which corresponds to a bulk velocity of 0.4 ft/sec. The
effect of the decreased velocity and higher recovery can be seen by
comparing the flux decline for Run #2 shown in Figure 11, to that
exhibited by Run #1 shown in Figure 10. Flux decline for Run #1 was
11% during the first 300 hours compared to 41% during the first 300
hours of the high recovery run. The flux decline was observed to
be more pronounced for the downstream modules, confirming the detri-
mental effects of the poor hydrodynamic conditions experienced by
these modules. This condition was probably aggrevated by the in-
creased brine concentration in the later modules. Following the
first 300 hours of operation, the unit was depressurized to accom-
modate regeneration of the granular carbon columns. This caused a
significant, but temporary, increase in membrane flux as did an
enzyme detergent flush at 450 hours.
- 22 -
-------
PRESSURE
REGULATING
VALVE
BRINE
1 SECONDARY
TREATMENT
CARBON
ADSORPTION
COLUMNS
STORAGE
TANK
CHLORINE
PRODUCT WATER
PRESSURE
HIGH
IESSI
PUMP
i i i i
i i i
ACID
PUMP
pH CONTROL
SYSTEM
SULFURIC
ACID
FIGURE 9
FLOW DIAGRAM, UNIVERSAL WATER CORPORATION
TUBULAR REVERSE OSMOSIS UNIT, POMONA, CALIFORNIA
-------
TABLE 8
SUMMARY OF FEED AND PRODUCT WATER QUALITY FOR UNIVERSAL
WATER CORPORATION UNIT
EXPERIMENTAL RUN NO. 1
Chemical
Analysis
Total COD
Dissolved COD
N03-N
NH3-N
i
to 4
** TDS
Ca
Mg
K
Na
S°4
Cl
No. of
Tests
10
9
9
10
8
10
4
3
4
4
3
3
Feed
Ave . Range
9.4 7.2 - 12.7
8.3 6.3 - 12.0
3.6 0.1 - 12.0
11.7 2.5 - 18.1
10.1 8.8 - 11.7
6.9 5.4 - 711
39.6
24.5
12.1
111
227
85
Product
Ave . Range
0.3 0.0 - 1.0
-
0.7 0.0 - 2.2
0.7 0.2 - 1.4
0.12 0.0 - 0.4
39 20 - 72
>1
2.0
0.8
7.6
0
8
Brine Percent
Ave. Range Rejection
14.6
12.7
5.5
18.9
15.5
1018
55.4
39.3
17.9
156
383
153
11.1 - 19.0 96.8
9.7 - 18.6
0.1 - 19.0 80.6
3.5 - 36.5 94.0
12.9 - 18.9 98.8
633 - 1859 93.7
>97.5
91.8
93.4
92.8
100.0
90.6
Notes: (1) All analyses run on grab samples.
(2) Feed samples taken after acidification with H-SO,; hence values of SO, shown in table
include sulfate contributed by acid addition.
(3) All values shown are in mg/1.
-------
20
O 16
ik
O
z
X '
D
i 4
0
jioo 2
% SALT REJECTION
-------
o
5
z
X
u.
24
20
16
12
8
4
O
% SALT REJECTION
'•^—- ~>^i-^-.^. s \ .-••;
^•'~ " V
•••
mt»
FLUX (GFD)
\ A T^. /N-- s T*
X' \/x— >.. 'x«x\/ ^'' ^'
RUN - #2
FEED - CARBON TREATED,
SECONDARY EFFLUENT
PRESSURE - 600 PSIG
i i i i i i i i i i i i
100 o
u
95 %
Ul
90 £
to
*
100 200 300 400 500 600 700 800 900 1000
TIME IN HOURS
FIGURE 11
FLUX AND SALT REJECTION PERFORMANCE OF UNIVERSAL WATER
CORPORATION TUBULAR UNIT, POMONA , CALIFORNIA
-------
(Lebanon, Ohio)
Flat Plate Configuration
Experimental investigation of reverse osmosis at the Lebanon, Ohio
Pilot Plant has involved the performance evaluation of a 100 sq ft
(2000 GPD) flat-plate reverse osmosis unit on waste streams that
represent varying degrees of pretreatment. Principal objectives of
this study were to demonstrate the efficacy of treating wastewater
with a flat-plate configuration and to evaluate the performance
characteristics of the experimental unit as a function of feed water
quality.
The five waste streams evaluated during the course of this
investigation are shown in Figure 12. The modular arrangement of
the reverse osmosis cell and a flow diagram of the entire system are
shown in Figures 13 and 14, respectively.
The membrane plates are arranged in eight vertical modules within
the pressure vessel. The feed stream enters the uppermost module,
and passes in series through each succeeding module leaving the
eighth module as the reject stream. The product water flows from
each module through separate channels within a central collection
tube to allow individual product water quality monitoring from each
module. The plates are 14-inches in diameter and consist of two
molded phenolic half plates bonded together to form a completed plate
which encloses a product water channel. The enclosed channel conducts
the product water to the central collection tube. The support sub-
strates and cellulose acetate membranes are mounted on the outside of
a completed plate. Adjacent plates are then separated by a poly-
styrene spiral-flow-baffle that directs the feed stream across the
surface of the membrane. Construction of the half plates and the
spiral-flow baffle are shown in Figures 15 and 16, respectively. Each
successive module within the pressure vessel contains fewer plates
in order to maintain a relative constant brine velocity as the product
water is withdrawn from the feed stream.
Referring to Figure 14, each of the five waste streams evaluated
enters an equalization tank containing a thermostatically-controlled
immersion heater that acts as a heat exchanger to maintain the feed
temperature between 22 and 25°C. After leaving this tank, the pH of
the feed is automatically controlled at 5.0 "t 0.2 units by the auto-
matic addition of sulfuric acid. The five micron filters preceding
the high pressure pump were used as a precautionary measure during
Runs #1 and #2, but were removed for the remaining three runs. The
operating pressure was held within the range 450 to 485 PSIG through-
out the entire study.
- 27 -
-------
RAW
SEWAGE
PRIMARY
TREATMENT
SECONDARY LIME DUAL MEDIA
TREATMENT CLARIFICATION FILTRATION
RUN NUMBER
1
3
GRANULAR
CARBON
ADSORPTION
TREATED
EFH.UENT
5
FIGURE 12
SCHEMATIC DIAGRAM OF WASTE STREAMS
LEBANON, OHIO
-------
FEED
MODULE 1
MODULE 2
MODULE 3
MODULE 4
MODULE 5
MODULE 6
MODULE 7
MODULE 8
REJECT
PRODUCT FLOW
OF TOTAL PRODUCT
16.2C
13 5 /;
13 4°/=
1O 7
- 85
73
FIGURE 13
MODULAR ARRANGEMENT OF AEROJET GENERAL
FLAT PLATE CONFIGURATION
-------
FLUSHING SOLUTION
AND RECYCLE TANK
FEEDS -1-2-3-4-5
EQUALIZATION
AND
TEMPERATURE
CONTROL TANK
HIGH
PRESSURE
PUMP
LOW
PRESSURE
PUMP
ACHI
PUMP
i
I—
i
5 MICRON
FILTERS
pH CONTROL
SYSTEM
REVERSE
OSMOSIS
CELL
8 SERIES
MODULES
REJECT
PRODUCT WATER
FLOW MEASURING
AND RECORDING
pH
TEMPERATURE
CONDUCTIVITY
EIGHT SEPARATE
PRODUCT WATER
STREAMS -
FIGURE 14
FLOW DIAGRAM, AEROJET GENERAL
PLATE & FRAME REVERSE OSMOSIS UNIT, LEBANON .OHIO
-------
FIGURE 15
TWO PHENOLIC HALF PLATES
14-INCH DIAMETER AEROJET-GENERAL PLAT AND FRAME UNIT
-------
.
FIGURE 16
PHENOLIC HALF PLATE AHD SPIRAL FLOW BAFFEL
14-INCH DIAMETER AEROJET-GENERAL PLATE AND
FRAME UNIT
-------
The feed flow rate to the cell was controlled at 1.8 GPM by a
variable speed positive displacement pump. This flow rate resulted
in an initial volumetric recovery of 52% , which ultimately declined
during the course of the investigation to 35% due to the irreversible
membrane fouling that occurred.
The reverse osmosis unit was operated continuously during each
run unless interrupted by mechanical difficulties (which were infre-
quent after the first week of Run #1) . The desalination cell was
depressurized and stored in a 0.25% formaldehyde solution between runs,
or for periods of storage greater than two weeks.
The first experimental run was conducted on lime-clarified,
dual-media filtered, and carbon-treated secondary effluent. The flux
decline pattern and salt rejection performance for this run are
presented in Figure 17 and the rejection efficiency is tabulated
and shown in Table 9.
TABLE 9
FEED AND PRODUCT WATER QUALITY DATA
(Run #1)
Lime-Clarified, Filtered, Carbon-Treated
Secondary Effluent
PC>4 as P
NO -N
N03-N
NH3-N
Organic-N
TOC
Turbidity (JTU)
Feed
mg/1
0.60
0.90
3.0
< 0.1
< 0.2
1.0
< 0.5
Product
mg/1
< o.i
0.60
1.1
< 0.1
< 0.1
0.5
< 0.1
Results based on average of 10 determinations.
As can be seen from the above table, the feed concentration of
the parameters of interest are too low to allow meaningful interpre-
tation of the rejection performance of the membranes. The TDS rejec-
tion averaged 90.1% during the run as determined by feed and product
water conductivity measurements. The rejection performance of each
- 33 -
-------
16
14
Q
i^
0 10
z
X 8
3
ML 6
4
2
0
% SALT REJECTION
^^ •
"*"
<-B
FLUX (GFD)
x- .
_
RUN - #1
FEED - CARBON TREATED LIME
CLARIFIED, FILTERED
SECONDARY EFFLUENT
PRESSURE - 470 PSIG
-
i i i i i i i i i i i i i t i
z
0
^^M
on ^*
90 V
Ml
UJ
80 5
<
lie
02 4 6 8 TO 12 14 16 18 20 22 24 26 28 30
TIME IN DAYS
FIGURE 17
FLUX AND SALT REJECTION PERFORMANCE OF AEROJET
GENERAL FLAT PLATE CELL, LEBANON, OHIO
-------
module was examined independently during the first run to determine the
effect of the increasing brine and organic solute concentration on
rejection performance. The results indicated that both the salt and
water permeation rates were constant over the entire length of the
cell.
The pressure drop across the brine side of the cell was 17 PSIG
and remained constant for this run with a feed flow of 1.8 GPM.
The second run was made under the same conditions as Run #1
except that the granular carbon adsorption columns were omitted from
the pretreatment train. This run was interrupted on seven separate
occasions due to operating schedules, and was finally terminated after
127 days. The flux decline pattern and salt rejection performance
are shown in Figure 18, while the feed and product water quality data
are shown in Table 10.
TABLE 10
FEED AND PRODUCT WATER QUALITY
(Run #2)
Lime-Clarified, Filtered Secondary Effluent
Average Values
First 72 days
**
***
Average Values
72nd to 127th day
P04
NO--N
N02-N
NH3-N
ORG-N
TOC
Feed
1.0
1.5
1.9
9.6
1.8
10.6
Product
0.1
0.5
1.7
1.1
0.5
1.0
% Rejection
90
67
11
88.5
72
90.5
Feed
-
2.9
*
9.5
2.6
13.3
Product
-
1.2
*
1.0
0.4
1.7
% Re lection
-
58
*
89
84.5
95
Total Bacteria
Count x ID** Per
100 ml 7.3
Turbidity
JTU 1 . 3
0.8 99 -
< 0.1 - 2.2 < 0.1
Results in mg/1 unless noted.
* Feed and product NO -N less than 0.1 mg/1.
** Average values based on 28 determinations from samples withdrawn
at regular intervals during the run.
*** Average values based on 8 determinations from samples withdrawn at
regular intervals during the run.
- 35 -
-------
Comparison of the flux decline pattern for the first 8 days of
Run #2 to that exhibited by Run #1 demonstrates the effect of the
higher organic load as a principal contributor to the increased flux
decline. Several efforts to recover this flux decline were made,
including daily air and water flushing cycles combined with periodic
cell depressurizations as well as attempts to create a "back diffusion
potential" to reverse the permeate flow through the membrane. These
methods were only marginally successful as shown by Figure 18.
The most promising flux restoration was achieved by flushing
the cell for one hour using a 0.7% bioenzyme (BIZ) solution. This
procedure resulted in recovery of 63% of the total flux decline
experienced during the run. Several attempts to further improve
the cleaning efficiency during the flushing cycles were made following
this initial su'ccess. The most efficient procedure evaluated was to
fill the desalination cell with a 0.35% bioenzyme solution at pH 9.7
at a temperature of 25 to 30°C and allow the cell to "soak" for a
period of 45 minutes followed by a 15-to 30-minute high velocity flush
at an average cell pressure of 50 PSIG. The brine was recirculated
without filtration during the flushing cycle which was conducted at
twice the normal operating velocity. The TOC and organic nitrogen
removals accompanying the first five flushing cycles are given in
Table 11.
TABLE 11
REMOVAL OF TOC AND ORGANIC NITROGEN
FROM MEMBRANE SURFACE
USING A BIOENZYME FLUSHING SOLUTION
Flushing
Cycle
Initial Flushing
Solution
0.35% Bioenzyme
In Product Water
Reject Stream
After Completion of 45 min.
Soak and 15 min. Flushing
Cycle
1
2
3
4
5
TOC (Total) mg/1
62
44
67
69
93
ORG-N mg/l
1.1
0.4
1.1
2.1
4.0
TOC (Total) mjs/1
500
450
480
435
650
ORG-N mjj/1
80
71
25
44
94
The appearance of the flushing stream changes from a light blue
color before flushing to a light brown turbid solution following the
flushing cycle. Each flushing requires 20 gallons of water, ten of
which are contained in the cell, with the remainder used to fill the
- 36 -
-------
12
SALT REJECTION
FLUX (GFD)
h
CO
8
6
4
2
0
I
DP. • INDICATES CELL
WAS DEPRESSURIZED
32
48 M 80
TIME IN DAYS
ir> co
-1100 =
90
80
UJ
CO
96
112 128
FIGURE 18
FLUX AND SALT REJECTION PERFORMANCE OF AEROJET
GENERAL FLAT PLATE CELL, LEBANON, OHIO
-------
recirculating lines, pump and mixing tank. The TOC and organic nitro-
gen levels shown in Table 11 are representative of the increase that
occurs during each cleaning cycle and illustrate the organic nature
of the membrane deposit.
The TDS rejection capacity of the membranes remained relatively
constant at 92.1% during the run. The repeated exposures of the
membranes to a pH of 9.5 during the cleaning procedures have shown
no adverse effect on membrane performance thus far. Similarly, there
has been no evidence of biological degradation of the membranes.
The third run with the Aerojet-General reverse osmosis unit was
conducted on lime-clarified secondary effluent. System operating
pressure was maintained at 470 PSIG, with feed stream pH controlled
at 5.0 "t 0.2 units by the addition of sulfuric acid. The feed and
product water quality data for this run are shown in Table 12.
TABLE 12
FEED AND PRODUCT
WATER QUALITY
(RUN #3)
Lime-Clarified Secondary Effluent
PO, as P
4
NCyN
N03-N
NH3-N
ORG-N
TOC
Turbidity
(JTU)
Feed
2.0
< 0.2
0.2
mo
3.0
20.8
7.5
High
0.1
0.4
0.4
1.7
1.6
9.0
Produc t
Average
0.1
< 0.2
0.1
1.1
0.7
2.8
< 0.1
Low
0.1
< 0.1
< 0.1
0.6
0.1
0.8
Ave.% Rejection
95
*
*
90
86
87
99
Results in mg/1 unless noted.
* Feed concentrations of NO«-N and NO--N are too low for meaningful
rejection data.
Membrane performance was satisfactory throughout this run with
TDS and TOC rejection averaging 92% and 87%, respectively. The TDS
rejection and flux decline pattern are shown in Figure 19.
- 38 -
-------
•
10
9
8
7
6
S
4
3
2
1
0
SALT REJECTION
FLUX (GFD)
RAW ENZYME AND
ENZYME DETERGENT
COMBINATIONS
7 8 9 10
1 1 I I 1 I
_l J_ 1 1 i
2,2,3,32
1 1 1 1
RUN -#3
FEED-LIME CLARIFIED
SECONDARY EFFLUENT
PRESSURE-470 PSIG
1 1 1 1 1
1 1 1
0123
141511 1818202122232425
353837383840414243444548
TIME IN DAYS
FIGURE 19
FLUX AND SALT REJECTION OF AEROJET
GENERAL FLAT PLATE CELL, LEBANON, OHIO
100 *
Cd
LU
90 =
CO
80 >,o
-------
Several standardized flushing cycles were employed during this
run using BIZ, as well as a number of raw enzyme detergent combinations.
The raw enzymes used were the protease-amylase enzymes manufactured
specifically for the detergent industry. A comparison of the relative
effectiveness of these enzymes is presented in Table 13. Although
results of these flushing procedures are not quantitative, some gene-
ralities concerning the effectiveness of the enzymes can be seen.
The raw enzymes appear to be more effective when used with a
detergent. Enzyme concentrations less than 35 mg/1 and detergent
concentrations less than 3500 mg/1 tend to limit the cleaning effec-
tiveness for a service interval of 24 hours. The three enzymes
evaluated seem to be equally effective in removing the organic fouling
layer. An enzyme/detergent ratio of 1 to 100 appears to be the most
effective combination evaluated. This ratio approximates the reported
concentration of enzyme contained in the BIZ formulation. The effect
of pH has not been well defined by the evaluation but information from
enzyme suppliers indicates that the optimum pH for the detergent
enzymes is between 7.5 and 9.5.
The percent flux increase associated with each flushing cycle is
highly variable, depending on the amount and nature of the organic material
deposited during the previous service cycle. A portion of this fouling
layer seems relatively easy to remove as evidenced by the flux increase
(6.3 to 7.4 GFD) resulting from the flush with a detergent solution on
March 3, 1969. Following completion of this flushing cycle, a BIZ
solution resulted in an additional flux recovery (from 7.4 to 8.7 GFD).
The average flux during this run was 7.6 GFD compared to 8.7 GFD
for Run #2. This decrease was apparently due to both the higher
organic and suspended solids load of the feed and the less efficient
flushing cycles employed while comparing the raw enzymes.
The pressure drop across the brine side of the membrane increased
from an average value of 20 PSIG for Run #2 to 28 PSIG for the present
run. This increased pressure drop is a result of the higher suspended
solids concentration of the feed stream which tends to accumulate in
the feed channels. Continued accumulation of these solids is easily
controlled, however, by the higher velocities associated with the
flushing cycles.
The fourth run with the Aerojet-General flat-plate configuration
was conducted on secondary effluent. This run, lasting 22 days, was
interrupted after the llth day for a period of 60 days and was then
resumed for an additional eleven days. The system pressure averaged
485 PSIG with the feed stream pH controlled at 5.0 T 0.2 units during
the run. The desalination cell was stored in a 0.25% formaldehyde
solution during the 60-day shutdown to prevent possible biological
hydrolysis of the membranes.
- 40 -
-------
TABLE 13
EVALUATION OF ENZYME FLUSHING SOLUTIONS
Date
2/11/69
2/13/69
2/14/69
2/18/69
2/20/69
2/25/69
2/26/69
2/27/69
2/28/69
3/1/69
3/2/69
3/3/69
3/3/69
3/4/69
3/5/69
Flushing
Solution
BIZ
BIZ
BIZ
HT-Proteolytic
HT-Proteolytic
HT-Proteolytic
HT-Proteolytic
with TIDE
HT-Proteolytic
with TIDE
Milezyme
with TIDE
Milezyme
with TIDE
Milezyme
TIDE
BIZ
BIZ
WT-2
with TIDE
Service
Preceding
Flush in HRS.
24
48
24
24
48
24
24
24
24
24
24
24
0
24
24
24
Concentration
of Flushing
Solution in
PPM
3500
3500
3500
3.5
10
10
35
10
35
3500
35
35
500
10
500
3500
3500
3500
35
3500
pH of
Flushing
Solution
9.5
9.5
9.4
5.2
5.2
9.4
7.0
_
8.8
-
8.8
8.6
8.6
8.8
9.4
9.5
8.8
Flux in
GFD
Before-After
7.6 -
6.5 -
7.8 -
7.4 -
6.8 -
6.8 -
6.8 -
6.0 -
6.8 -
6.8 -
6.5 -
6.3 -
7.4 -
6.6 -
6.8 -
9.6
9.1
9.1
8.4
8.3
8.2
8.0
8.4
8.4
8.4
7.6
7.4
8.7
8.9
8.3
% Flux
Increase
26.3
42.0
16.3
13.5
22.0
20.6
17.5
40.0
23.6
23.6
17.0
17.5
17.5
35.0
22.0
Membrane
Rejection
Performance
Good
Good
Good
Good
Good
Good
Good
Good
Good
Good
Good
Good
Good
Good
-------
The rejection performance of the cell is shown in Table 14 for
both the period preceding and following the shutdown. In general,
rejection of the divalent ions was greater than 94% while the rejec-
tion of the monovalent ions was varied from 80 to 90%.
The effect of the 60-day storage on membrane performance is
somewhat anomalous. The TDS rejection declined from 91.5% before
storage to 86% after storage, a drop of 5.5% as measured directly
(by evaporation of composited feed and product water samples).
Conductivity measurements show this decline to be 9.3%, from 90.3%
before storage, to 81% after storage. This apparent discrepancy can
be resolved by considering the increasing percent rejection that
occurred following the storage period as shown by Figure 20, and by
noting that the TDS samples were composited during the 20th and 23rd
day. Conductivity measurements on the product water from the indi-
vidual modules indicated the decrease in membrane performance was a
result of a generalized condition, exhibited by all eight modules
rather than a localized condition. The increase in rejection per-
formance with time may result from the "healing" of imperfections
present in the membrane. This rationale presumes that previously
deposited organic materials were leached out during the storage period,
thereby causing the increase in salt permeation.
Organic rejection, as measured by TOC, increased slightly from
89% to 94% during the storage period. No particular significance can
be attached to this increase, however, in view of the relatively low
TOC levels involved.
The flux decline characteristics for this run are shown in
Figure 20, and indicate that a moderate flux decline can be maintained
by employing a one-hour daily enzyme flush.
It should be noted that the quality of the secondary effluent
was exceptionally good during this run with TOC averaging 14.9 mg/1
for the first 11 days and 8.3 mg/1 for the remaining 11 days. Feed
turbidity similarly decreased from an average of 6.1 JTU during the
first half of the run to 2.4 JTU during the last half of the run.
The fifth run was made on primary effluent. This run was con-
ducted at an average operating pressure of 480 PSIG with a feed
stream pH of 5.0 1" 0.2 units. The flux decline characteristics and
rejection performance for this run are shown in Figure 21. The flux
declined from a daily average of 5.4 GFD on the third day of operation to
3.8 GFD on the 12th day. This represents a 30% decline in flux that
occurred in nine days with daily bioenzyme flushing. The 5.4 GFD
is significantly lower than the daily average of 7.8 GFD that was main-
tained during the run on secondary effluent. The feed and product
water quality data for this run are presented in Table 15. The TDS
rejection decreased from a high of 90% to a low of 84% during this
run. This decline in rejection is not surprising in view of the
lower flux and poor distribution of the feed caused by plugging of
the feed channels.
- 42 -
-------
.p-
oo
TABLE 14
FEED AND PRODUCT WATER QUALITY
(Run #4)
Secondary Effluent
Before 60-day Shutdown After 60-day Shutdown
Constituent
Calcium (Ca)
Magnesium (Mg)
Sodium (Na)
Potassium (K)
Iron (Fe)
Manganese (Mn)
Phosphate (PO.asP)
Sulfate(SO.)
Chloride (Cl)
Silica (SiO.)
Nitrite (NOp
Nitrate (NCp
NH_-N
Organic-N
Total Alk. (CaCO_)
TDS (Conductivity
TDS (Evaporation)
TOC (Unfiltered)
Turbidity (JTU)
Cone.
Feed
104
30.5
95.5
8.5
<0.3
<0.1
6.6
113
133
12.4
<0.2
<0.2
8.7
2.1
332
-
726
14.9
6.1
in mg/1*
Product
5.2
0.4
15.0
1.5
<0.3
<0.1
0.4
14
26
4.1
<0.2
<0.2
1.7
0.3
39
-
66
1.7
<0.1
7o Rejection
95
99
84
82
-
-
94 (1)
88C1)
81
67
-
-
80
86
88
90.3
91.5
89
-
Cone.
Feed
110
28.6
91.0
8.2
0.3
-
6.8
107
131
13.8
-
8.6
1.0
0.9
326
-
644
8.3
2.4
in mg/1*
Product
2
.5
14.0
1.3
0.1
-
0.6
7
26
3.4
-
4.7
0.3
0.1
40
-
88
0.5
<0.1
% Rejection
98
98
85
84
-
-
91
93
80
75
-
45
-
89
88
81
86
94
"•
% Change
+3
-1
+1
+2
-
-
-3
+5
-1
+8
-
-
-
+3
0
-9.3
-5.5
+5
~
* Average of two 24-hour composite samples .
(1) Feed water sample taken before addition of H-SO, .
-------
. •-"—-*_ SALT REJECTION
10
O 8
ik
2*
X 4
D
£ 2
0
FLUX (GFD)
KjANNSTv
RUN - #4
FEED - SECONDARY
EFFLUENT
PRESSURE - 485
PSIG
BEFORE 60 DAY
SHUTDOWN
AFTER 60 DAY
SHUTDOWN
90
80
70
V)
fc?
0 2
6 8 10 12 14 16 18 20 22 24 26 28 30 32
TIME IN DAYS
FIGURE 20
FLUX AND SALT REJECTION PERFORMANCE
OF AEROJET GENEUl FLAT PLATE CELL. LEBANON, OHIO
-------
9
8
7
0 6
u.
® 5
Z
X
2 3
Ik
2
1h
\
SALT REJECTION
_.—-^
FLUX IN (GFD)
RUN -#5
FEED - PRIMARY
EFFLUENT
PRESSURE - 480 PSIG
12
14
16
Z
90 2
18
'024 6 8 10
TIME IN DAYS
FIGURE 21
FLUX AND SALT REJECTION PERFORMANCE OF
AEROJET GENERAL FLAT PLATE CELL, LEBANON, OHIO
-------
TABLE 15
FEED & PRODUCT WATER QUALITY
(Run #5)
Primary Effluent
Constituent Concentration mg/1
Calcium (Ca)
Magnesium (Mg)
Sodium (Na)
Potassium (K)
Iron (Fe)
Manganese (MN)
Phosphate (PO^as P)
Sulfate (S04)
Chloride (Cl)
Silica (Si02)
Nitrate (NO )
Nitrite (N03)
NH3-N
Organic-N
Total Alkalinity (as CaCCO
TDS (Conductivity)
TDS (Evaporation)
TOG (Unfiltered)
TOC (Filtered)
Turbidity (JTU)
Feed
mg/1
138
30.7
141
11.0
1.5
*
9.7
173
168
26.7
*
*
12.5
6.4
395
-
917
97
35.0
52.0
Product
mg/1
2.0
0.4
24.7
2.6
0.1
*
0.4
5
47
6.0
*
*
2.1
0.6
23
-
128
13.4
12.0
< 0.1
% Rejection
99
99
82
76
93
-
96
97(D
72
80
-
-
83
91
94
86
86
89
66
-
* Concentration < 0.1 mg/1.
(1) Feed water sample taken before addition of H.SO,.
(2) Feed filtered through glass-fiber filter paper, Type 1106 BH,
Mine Safety Appliances Company, Pittsburgh, Pennsylvania.
- 46 -
-------
The most notable change in operating conditions during this run
was the more pronounced pressure drop that was experienced across the
desalination cell. This drop ranged from a low of 40 PSIG after the
daily flushing to over 80 PSIG near the end of the daily service cycle.
In many cases it was necessary to decrease the velocity of the feed
stream to maintain the pressure drop across the cell to values less
than 100 PSIG. This lower brine velocity may have significantly
affected the flux decline rate resulting in a distorted comparison
to the performance characteristics exhibited by the previous feed
streams.
A comparison of the rejection of principal constituents for the
five runs conducted at Lebanon is shown in Table 16. These results
indicate that the rejection performance was similar for the waste
streams evaluated.
TABLE 16
SUMMARY OF REJECTION PERFORMANCE
FOR
LEBANON WASTE STREAMS
Constituent Run #1 Run #2 Run #3 Run #4 Run #5
P04 as P
NCyN
NH3-N
Organic-N
TOC
TDS
Turbidity(JTU)
*
*
*
*
*
90
*
91
68
88
89
85
92
99
90
63
89
78
89
92
99
93
*
80
87
94
91**
99
96
*
83
91
89
90
99
* Feed concentration too low for meaningful rejection data.
** TDS rejection before 60-day storage period.
Experience has shown that the time-dependent flux decline
pattern of a membrane system operated on a feed containing both
organic and inorganic solutes at constant pressures can be linearized
by plotting log flux vs log time; the slope of the resulting straight
line of best fit is commonly referred to as the "flux decline slope"
and is used to quantitatively describe the change in flux with time.
- 47 -
-------
The flux decline slopes for Runs #1 and #2 are compared in
Figure 22. The differences in the flux decline slopes for these
two runs indicate the effect of carbon adsorbable material on flux
decline. It should be remembered that the feed for both Runs #1
and #2 was passed through a 5 micron cartridge filter prior to en-
tering the desalination cell. This, combined with the increased
feed TOC exhibited by Run #2, supports a positive role for soluble
and colloidal organic carbon in causing flux decline.
A similar comparison of flux decline slopes for Runs #3, #4,
and #5 is presented in Figure 23. These runs were made after removal
of the 5 micron filters and reflect the effect of the particulate
material in the respective waste streams. It is encouraging to note
that, by employing a one-hour bi-daily cleaning schedule for Run #3 -
lime-clarified effluent, and a daily cleaning schedule for Run #4 -
secondary effluent, respective flux decline slopes of -0.03 and
-0.04 could be maintained. These slopes are less than the slope of
-0.05 experienced when operating on carbon column effluent, a much
higher quality water. These data support the concept of controlling
membrane fouling by employing periodic cleaning procedures as an
alternate to pretreating the feed to prevent fouling.
The importance of maintaining a reasonable flux decline slope
and membrane lifetime can be seen from an examination of Table 17.
TABLE 17
FLUX DECLINE SLOPE ACCUMULATIVE 7. LOSS OF
LOG FLUX VS LOG TIME ORIGINAL FLUX AFTER YEARS
-0.01
-0.05
-0.10
(1)
5
25
44
(2)
5.5
28
47
(3)
6
30
51
The flux decline characteristics shown above indicate that the
first-year flux decline represents well over 80% of the decline that
may be expected for any reasonable membrane lifetime and that the
one-year flux value would provide a rational design figure.
- 48 -
-------
2.6
2-
0
o
z
X
10-
9
8
7
6 -
5
4-
3-
2.5-
2-
1.5-
RUN • #1 LIME CLARIFIED, FILTERED, CARBON TREATED
SECONDARY EFFLUENT, FLUX DECLINE SLOPE = -0.05
TURBIDITY
JTU
TOC mg/l
MEMBRANE
CLEANING
RUN - #2 LIME CLARIFIED, FILTERED SECONDARY
EFFLUENT FLUX DECLINE SLOPED -0.14
FEED CONDITIONS
RUN #1
0.5
1.0
NONE
RUN #2
1.3
10.6
NONE
1 10 22.63 4 S678910 10 2263 4 66789 1O
TIME IN DAYS
FIGURE 22
COMPARISON OF FLUX DECLINE SLOPES FOR EXPERIMENTAL
RUNS #1 AND #2, USING AEROJET GENERAL FLAT PLATE CELL, LEBANON, OHIO
-------
2.5.
2 •
1.5-
SUMMARY OF OPERATING CONDITIONS
TURBIDITY TOG FLUX DECLINE
JTU
SLOPE
FLUSHING
INTERVAL
RUN #3 7.4
RUN #4 6.1
RUN #5 52
20.8
14.9
97
-0.3
-0.04
-0.19
48 HRS.
24 HRS.
24 HRS.
RUN #5
RUN #3
RUN #4
i tiii
J I
1J5
2.5 3
507891O 15 2
TIME IN DAYS
2.5 3
6 7 8 9 10
FIGURE 23
COMPARISON OF FLUX DECLINE SLOPES FOR EXPERIMENTAL RUNS 3, 4 & 5
USING AEROJET GENERAL FLAT PLATE CELL, LEBANON, OHIO
-------
DISCUSSION AND INTERPRETATION OF PILOT PLANT RESULTS
The foregoing experiences in operating small reverse osmosis
modules on variously treated waste streams have demonstrated, in
all cases, that excellent separation of the contaminants of major
concern can be achieved.
In general, the membranes were found capable of rejecting 93 to
95% of the total dissolved solids, 95 to 99% of the phosphate, 80 to
90% of the ammonia nitrogen, 60 to 70% of the nitrate nitrogen, 99
to 100% of the particulate matter, 90 to 95% of the total organic
carbon and greater than 907<> of the COD. The rejection performances
exhibited by the three types of modules evaluated were similar, with
the small observed differences being attributed to intrinsic differ-
ences in the membranes evaluated.
In many cases, the studies were conducted with prototype units
which resulted in an initially high incidence of mechanical failures.
Many of the earlier malfunctions have been eliminated by design
changes resulting in much improved equipment.
All of the experiments were conducted with modules equipped with
membranes designed for the conversion of "brackish water," which is
characteristically low in organic content and much higher in IDS
than the normal municipal wastewater streams of interest.
Work conducted early in the program showed that pH control of the feed
was necessary to control the precipitation of calcium carbonate, even
at moderate recovery ratios. Possible precipitation of a complex
calcium phosphate at 80% recovery was noted once during the investi-
gation, and was controlled by reducing the pH of the brine to 5.0
from an original value of 5.5. No further instance of inorganic
fouling was observed during the investigation.
Organic fouling of the membranes was experienced in all of the
experimental runs and can be cited as the principal cause of the flux
declines that occurred. There is considerable interest and some
controversy at the present time regarding the principal cause of
organic fouling. The major argument centers around the correct
identification of the class of material responsible for flux decline
due to fouling. Most investigators agree that the material is organic
in nature but dispute the particle size and size distribution respon-
sible for the observed declines. Experimental evidence, including
data reported herein, can be cited that supports equally well both
sides of the issue in question. The argument is partially academic
when one realizes the diverse nature of raw and treated wastewaters
which are candidate feeds for treatment by reverse osmosis._,These „
streams encompass a particle size range that varies from 10 to 10
microns with molecular weights that range from less than one hundred
- 51 -
-------
to several million. Attempts to characterize these waste streams as
the first step in understanding the interfacial phenomena and
fouling mechanism that occur at the membrane surface have not been
successful, largely because of the complex nature of the waste. It
is becoming increasingly clear, however, that soluble, colloidal,
and suspended species are all fundamentally involved in the fouling
that takes place at the membrane surface. Figure 22 shows that even a
water free of gross particulate matter (turbidity of 1.3 JTU) exhibits
a fouling potential sufficiently high to prohibit economic treatment
by a membrane separation process unless some provision is made to
control deposition of the fouling material, which can range from mole-
cularly disperse systems through the colloidal range to the larger
suspended particles.
The most successful approach in controlling the organic fouling
of reverse osmosis membranes employed to date is to routinely depres-
surize the unit and to flush the brine sides of the membranes using
the enzyme detergent flushing solution previously described. Success
of the method may be due to protein hydrolysis of a slime layer that
is physically deposited during operation coupled with the dispersing
and cleaning action of the accompanying detergent. This concept is
supported by investigations conducted at the Pomona Pilot Plant by
Gulf General Atomic under FWQA Contract No. 14-12-181.
Although very little investigation of the hydrodynamic factors
affecting fouling was conducted under this program, limited experi-
ences with superficial velocity increases during the cleaning opera-
tions on the flat plate unit at Lebanon indicate some promise for
this approach. Experimental investigations conducted at the Oak
Ridge National Laboratory on filtered and unfiltared lake water indi-
cate that substantially lower flux decline rates can be achieved by
increasing the axial velocity in a 1/2" diameter tube. Additional
studies employing turbulence promoters have shown promise of improving
the efficiency of water transport through a cellulose acetate membrane (7)
A fundamental question in utilizing either increased velocity or
turbulence promoters is whether these approaches will offer an econo-
mic advantage when the greater pumping cost is balanced against the
increased productivity. Perhaps the greatest advantage will be
realized in periodically increasing the velocity during a flushing
operation. The question of membrane attrition must be dealt with
and may prevent the continued use of axial velocities greater than
3-5 ft/sec (6). Preliminary investigation of fouled modules taken
from the Universal Water Corporation unit operating on carbon-treated
secondary effluent at Pomona, California showed some evidence of
physical attrition of the membrane near the end of the module. The
severity of this type of problem has not been fully assessed at this
time and will require further investigation (8).
- 52 -
-------
CONTRACT AND RELATED INVESTIGATIONS
In addition to the above-mentioned pilot-plant investigations,
the FWQA has supported, in cooperation with other government agencies,
a great deal of contract research in reverse osmosis.
This research has been directed toward three goals; the develop-
ment of improved high-flux membranes tailored specifically for waste-
water renovation, investigation of new methods to control flux decline
due to organic fouling of the membranes, and improvements in modular
design and contacting systems.
Improved cellulose acetate membranes for wastewater renovation
have been developed in the laboratory by two investigators. North
Star Research and Development Institute, under Contract No. 14-12-587,
investigated the application of ultrathin membranes (200-2000 A) for
the renovation of municipal effluents. These ultrathin membranes
were prepared on polysulfone supports and demonstrated 600 PSI fluxes
of 58 to 91 OFD with salt rejections of 76 to 86% when tested in the
laboratory against a 0.1% sodium chloride solution. These membranes
were tested with filtered secondary effluent and demonstrated average
fluxes of 17 to 38 GFD, depending on the pore size of the filter used
to pretreat the feed. Evidence indicated that particles smaller than
1 micron were responsible for nearly 80% of the observed flux decline.
It was also demonstrated that about one quarter of this decline could
be eliminated by chemical clarification of the feed followed by
passage through a 1 micron filter.
Another contractor, Aerojet-General Corporation, working with
assymetric cellulose acetate films under Contract No. 14-12-553,
developed three classes of membranes suitable for wastewater renova-
tion. One membrane, a crosslinkable cellulose acetate methacrylate
(CAM) membrane, synthesized by methacrylation of Eastman Type E-360-60
cellulose acetate having a degree of substitution (d.s.) of 2.09 to
2.12 to a total d.s. of 2.34 to 2.39, produced 700 PSI fluxes of 70
to 90 GFD with a TDS rejection of 80 to 90% when tested in the labora-
tory in flat sheet form against a synthetic feed solution.
Another membrane formulation using Eastman E-383-40 cellulose
acetate as the base polymer with a d.s. of 2.28 produced 600 PSI fluxes
of 64 to 105 GFD with a sodium chloride rejection of 61 to 67% when
tested in flat sheet form against a 0.1% sodium chloride solution.
The third class of membranes, a blend of diacetate and triacetate,
produced 600 PSI fluxes of 60 to 70 GFD with a sodium chloride rejec-
tion of 70 to 80%.
- 53 -
-------
These three classes of membranes were all tested on secondary
effluent and demonstrated satisfactory rejection of both organic
and inorganic species of interest. The flux of these membranes was
sharply reduced by organic fouling when tested without cleaning.
It was demonstrated, however, that average fluxes as high as 50 GFD
could be sustained when processing secondary effluent by periodically
cleaning the membrane with an enzyme solution.
The above results are encouraging and indicate that, by
sacrificing a small amount of salt rejection, the intrinsic flux of
reverse osmosis membranes can be substantially increased. It was
also shown that as the membrane flux increases the fouling problem
becomes more severe.
The problem of membrane fouling has received attention from
several investigators. Aerojet-General, under Contract No. 14-12-184,
conducted an extensive investigation of pretreatment procedures as
a means of controlling flux decline due to fouling. This study
involved over ninety controlled experiments to evaluate the effects
of feed quality and pretreatment on flux decline. Several classes
of feed additives such as chelating agents, dispersants, and floccu-
lating agents were evaluated on a number of waste streams including
primary effluent, secondary effluent, and carbon-treated secondary
effluent. In general, the flocculating agents were found to be the
most successful cleaning agents, especially when followed by rapid
sand filtration as a pretreatment for reverse osmosis.
Another approach taken by Gulf General Atomic, under Contract
No. 14-12-181, was the examination of flushing procedures that can
be periodically employed during operation to limit flux decline to
acceptable levels. Gulf General Atomic first used the enzyme presoak
formulations as membrane cleaning agents to aid in the removal of
organic slimes. This has proven to be the most successful method
presently available to maintain reasonable fluxes when treating
wastewater.
Other methods of controlling membrane fouling under considera-
tion are the periodic or sustained use of higher brine velocities,
the use of integral and detached turbulence promoters, gas bubble
generation at the membrane surface, generation of electric current
at the membrane surface, charged membranes, and membranes containing
attached enzymes. Of these methods, turbulence promoters, charged
membranes, and periodic velocity increases appear most promising.
In addition to the development of high-flux membranes and
fouling control techniques, there is some interest in developing
more efficient modules and module configurations in an effort to
lower the capital cost of reverse osmosis systems.
- 54 -
-------
Developments in this area include studies of in-situ membrane
replacement, cast-in-place membranes, membrane regeneration techniques,
and replaceable membranes for the popular configurations. Also under
study is the development of new module configurations that employ
hollow fibers spun from cellulose acetate. These promise a substan-
tial increase in module productivity, but are just entering the
deve1opment stage.
RELATIVE POSITION AND FUTURE ROLE OF REVERSE OSMOSIS IN RENOVATING
WASTEWATER
At the present time, the role of reverse osmosis in the
reclamation of wastewater is not clear. Factors that will affect
the utility of this process are three-fold; first - the need to
demineralize wastewater for reuse applications; second - the economic
attractiveness of reverse osmosis when compared to competing proc-
esses such as ion exchange, electrodialysis, distillation for the
demineralization of wastewater on an equitable basis; and finally
the potential of using reverse osmosis as a pollution control device
with discharge to a receiving stream.
Since each domestic use of a water supply adds 300 to 400 mg/1
of IDS it is clear that continued reuse of a single supply will
require demineralization. The amount of demineralization required
to maintain drinking water quality standards of 500 mg/1 TDS will
depend upon the TDS concentration of the source and the consumptive
use ratio of the particular community. A recent study sponsored by
the Office of Saline Water (9) indicates that, of the approximately
20,000 communities in the United States, about one-half are served
by public water supply systems. The study further estimates that
the water distributed through over 98% of these community water
systems has a total dissolved solids content below the USPHS standard
of 500 mg/1. Municipal water use figures in the United States
indicate that the consumptive use averages about 13% (10). Wide
variation in this figure occurs with as high as 50% comsumption
recorded in some arid or semi-arid regions. At the latter locations,
demineralization of wastewater for reuse at 907,, volumetric recovery
will require about 25% make-up water and a demineralization effi-
ciency of 60% except for nitrogen.
There have been several cost estimates made for the competing
processes under consideration for the demineralization of waste-
water. These include ion exchange, electrodialysis, distillation
and reverse osmosis. Of these, ion exchange and reverse osmosis
appear to offer an economic advantage over electrodialysis and
distillation. A recent cost study made by the County Sanitation
Districts of Los Angeles County compared the cost, exclusive of brine
disposal, of electrodialysis, ion exchange and reverse osmosis for the
- 55 -
-------
demineralization of 1000 mg/1 IDS water to 667 mg/1 IDS (11). In
this study, the feed to each demineralization process was carbon-
treated secondary effluent. The total flow from the carbon adsorption
columns was treated by the electrodialysis unit for a total cost of
26.9C/1000 gallons including the cost of the carbon pretreatment. The
cost for ion exchange and reverse osmosis was based on demineralizing
377o of the carbon column effluent and blending the product water with
carbon column effluent to achieve a blended product quality of 667 mg/1
IDS. The total cost of the blended product for these two processes
including carbon pretreatment was 19.5c/1000 gallons for ion exchange
and 24.8c for reverse osmosis. This study indicates an advantage for
ion exchange for the demineralization of carbon-treated secondary
effluent. The cost for reverse osmosis used in this study was based
on an average flux of 10 GFD for a membrane lifetime of two years.
Using the same assumptions, but increasing the flux to 15 GFD (a
demonstrated flux rate for carbon-treated secondary effluent) decreases
the cost for reverse osmosis to 20.Sc/1000 gallons. In general, the
cost of reverse osmosis is quite sensitive to flux rate, membrane life-
time and membrane cost. Another consideration is that, depending on
the intended reuse of the demineralized water, the carbon adsorption
pretreatment could be replaced with mixed-media filtration as pre-
treatment for reverse osmosis for an additional savings of 5c/1000
gallons.
Consideration of the successive elimination of required pretreat-
ments poses the question, can reverse osmosis be considered a potentially
useful AWT process if demineralization for reuse is not the principal
objective? It has generally been accepted that treatment of municipal
wastes by reverse osmosis with discharge of product to a receiving
stream is not practical. A little reflection, however, can demonstrate
a potential in this area. The total costs for a conventional biologi-
cal treatment system followed by tertiary processes for nutrient and
residual organic removal are given below for a 10 MGD plant (12).
Treatment Total Cost
Step c/1000 gal.
Primary 6.2
Activated Sludge 4.8
Lime Clarification 7.8
with Recalcination
Ammonia Stripping 3.6
Filtration Through Sand or
Graded Media 4 gpm/sq ft 2.8
Activated Carbon Adsorption 10.2
TOTAL 35.4
- 56 -
-------
Examination of the treatment sequence and associated costs suggest
several possible treatment schemes incorporating reverse osmosis.
Several possible combinations and the corresponding competitive costs
for reverse osmosis are presented in the following table.
Pretreatment and Cost for Reverse Osmosis Total
Associated Cost to be Competitive c/1000 gal Cost
Primary 6.2 29.2 35.4
Primary and
Sand Filtration 9.0 26.4 35.4
Primary and
Activated Sludge 11.0 24.4 35.4
Primary and
Activated Sludge
and Sand
Filtration 13.8 21.6 35.4
This comparison clearly indicates that for reverse osmosis to
become competitive, it should be employed near the primary end of a
treatment train. For the situation outlined above, the application of
reverse osmosis to primary or filtered primary effluent at a cost of
29.2 and 26.4 cents per thousand gallons of treated water appears
competitive with present-day technology for achieving similar water
quality.
The cost of brine disposal must be included in the cost of
reverse osmosis treatment, and, in many cases, may prove to be a
limiting factor in inland areas. A recent investigation of the dis-
posal of brines produced in the renovation of municipal wastewater
was conducted by Burns and Roe, Inc. under FWQA Contract No. 14-12-492.
This study indicated that brine disposal cost varies from $0.052/1000
gallons to SO.76/1000 gallons for solar evaporation of a Colorado River
water source near Denver. Brine disposal at Tucson, Arizona is esti-
mated at $0.13/1000 gallons by deep-well injection. Another study of
deep well disposal of desalination brine waste was conducted by Dow
Chemical Company under OSW Contract No. 14-01-0001-1691 and reported
in Research and Development Program Report No. 456. This study
estimates the cost of deep-well injection to vary from 2.5C/1000
gallons at Arkansas City, Kansas to 35C/1000 gallons for Morgan,
Colorado. The cost of inland brine disposal, whether by deep-well
injection or by solar evaporation, is highly variable depending on
local geologic and climatic conditions as well as on current land uses
and value. In particular, disposal of reverse osmosis brines will
undoubtedly require some form of pretreatment to remove the organics,
especially if applied to wastewater streams such as primary or filtered
primary effluents.
- 57 -
-------
The reverse osmosis process, barely ten years old, developed
primarily as a demineralization process is now competitive for that
purpose. Present day costs of this process are prohibitively high
for water renovation, unless a high premium is placed on the need
to demineralize wastewater for reuse. Recent emphasis by FWQA in
supporting development of membranes and membrane systems specifically
designed for wastewater treatment, the progress being made in control-
ling organic fouling, and technological improvements being made by
the industry itself should substantially improve the posture of
reverse osmosis in the foreseeable future. Widespread use of this
process will ultimately depend, of course, on the economic advantage
that will be demonstrated over competing processes in any particular
treatment scheme.
- 58 -
-------
REFERENCES
1. Reid, C.E. and Breton, E.J., Applied Polymer Science, 1, 133,
(1959).
2. Loeb, S.J. and Sourirajon, S., "Sea Water Demineralization by
Means of a Semipermeable Membrane," Department of Engineering,
University of California, Los Angeles, Report No. 60-60 (1961).
3. Loeb, S. and Sourirajon, S., Advanced Chemistry Series. 38,
117 (1963).
4. Loeb, S., Sourirajon, S., and Weaver, D.E., U. S. Patent No.
3,133,137 (May 12, 1964).
5. "Reverse Osmosis as a Treatment for Wastewater," - A final
report to U. S. Public Health Service, Contract No. 86-63-227,
Report No. 2962 (January 1965).
6. Bray, D.T. and Merten, U., "Reverse Osmosis for Water Reclamation,"
General Atomic Division, General Dynamics Corporation, Augustas,
Max, County Sanitation Districts of Los Angeles County.
7. Sheppard, J.D. and Thomas, D.G., "A Study of a Hydrodynamic Aspect
of Reverse Osmosis," (Hyperfiltration) Monthly Reports, July 15,
1968 to March 15, 1970, OSW Contract No, 14-01-0001-534.
8. Personal Communication from James Gratteau, Project Engineer,
Research and Development, Los Angeles County Sanitation District,
Los Angeles, California.
9. Office of Saline Water Research and Development Report 462,
"Communities of Over 1000 Population with Water Containing in
Excess of 1000 PPM of Total Dissolved Solids."
10. Senate Select Committee on National Water Resources, Water Resource
Activities in the U. S. Water Supply and Demand, Committee Print
No. 32, (1960).
11. Carry, C.W., "Demineralization of Wastewater by Electrodialysis,
Ion Exchange, and Reverse Osmosis," Presented at the 42nd Annual
Conference of the Water Pollution Control Federation, Dallas,
Texas, October, 1969, Session No. 9.
12. Middleton, P.M., "Advanced Treatment of Municipal Wastewaters in
the United States of America," Presented at the Biennial Conference
of the South African Branch of the Institute of Water Pollution
Control, Capetown, South Africa (March 19, 1970).
- 59 -
-------
1
Accession Number
w
5
2
Organization Environmental
Federal Water
Subject Field &
05D
Protection
Oualitv Adi
Group
SELECTED WATER RESOURCES ABSTRACTS
INPUT TRANSACTION FORM
Agency
ninistration
Municipal Treatment Research Program
Advanced Waste Treatment Research Laboratory, Cincinnati, Ohio
Title
RENOVATION OF MUNICIPAL WASTEWATER BY REVERSE OSMOSIS
10
Authors)
Smith, John M.
Masse, Arthur N.
Miele, Robert P.
16
Project Designation
ORD-17040---05/70
21
Note
22
Citation
23
Descriptors (Starred First)
*Reverse osmosis, *Desalination processes, *Membranes, *Demineralization, *Semi-
permeable membranes, osmosis, separation techniques, pressure, membrane processes,
water reuse, water renovation, wastewater treatment, fouling, organic fouling
25
Identifiers (Starred First)
*Ultrafiltration, *Cellulose acetate membranes, *reverse osmosis modules, Hyper-
filtration, Water conversion, Reverse osmosis modules tubular, Reverse osmosis
modules spiral wound, Flux, Flux decline
IT
Abstract
The increasing demand for clean water makes clear the need for improved treat-
ment of municipal wastewater. Deliberate reuse of wastewater requires efficient
removal of "residual" organic and inorganic solutes. A common and effective approach
has been to supplement biological processes with chemical and physical processes
that, together, are capable of achieving water quality consistent with reuse require-
ments. A more direct approach is to use a single process such as reverse osmosis which
exhibits a potential to separate a broad spectrum of contaminants from primary and
secondary effluents.
The pilot plant studies reported herein were addressed toward defining the
problems associated with the reverse osmosis processing of variously treated waste-
waters. The effects of feed water quality on membrane performance were evaluated,
as were methods to control membrane fouling. A comparison of three configurations;
flat plate, spiral-wound, and tubular was made to determine their relative merit in
treating wastewater. Finally, a cursory examination of costs was made.
(Smith-FWQA)
AbstTotin M. Smith
Institution
Federal Water Quality Administration
WR:'02 (REV. JULY 1969)
WRSIC
SEND. WITH COPY OF DOCUMENT. TO: WATER RESOURCES SCIENTIFIC INFORMATION CENTER
U.S. DEPARTMENT OF THE INTERIOR
WASHINGTON. D. C. 20240
* GPO: 1970-389-930
------- |