MEMBRANES FOR REMOVING ORGANICS
FROM DRINKING WATER
by
C. A. Fronk, B. W. Lykins, I K. Carswell
Proceedings of 1990 American Filtration Society Annual Meeting
Washington, D.C, March 18-22,1990.
-------
ABSTRACTS
ATION
American Filtration Society
Annual Meeting
Filtration and Separation
Providing Solutions
to the Technical Problems
of the 1990fs
March 18-22,1990
Stouffer Concourse Hotel
Arlington, Virginia
-------
MEMBRANES FOR REMOVING ORGANICS FROM DRINKING WATER
by
C.A. Fronk, B.W. Lykins Jr., and J.K Carswell
ABSTRACT
Membranes have historically been used to remove salts and other inorganic compounds
from water but recently both bench-scale and field studies have shown their effectiveness for
removing organic compounds from drinking water. Two different membrane types have been
evaluated by the U.S. Environmental Protection Agency: high pressure membranes and low
pressure membranes. High pressure membranes are those using pressures between 150 to 400
psig. These membranes are commonly called reverse osmosis membranes. During bench-scale
studies, reverse osmosis membranes tested included cellulose acetate, polyamide, and thin-film
composites. These membranes were used to treat multisolute, aqueous solutions in the
concentration range of 6 to 153 ug/L. Removal efficiencies for alkanes, alkenes, aromatics, and
pesticides showed that thin-film composite membranes were more effective than the polyamide
or cellulose acetate membranes.
At a research site in Suffolk County, New York, removal of agricultural contaminants by
reverse osmosis was evaluated on the bench and in a pilot plant. Percent removals for long term
pilot plant evaluation for aldicarb sulfone, aldicarb sulfoxide, 1,2-dichloropropane, and
carbofuran ranged from 53% to more than 95%.
Low pressure membranes are usually operated at or below 150 psig. These membranes,
normally called ultrafiltration membranes, were evaluated at various sites in Florida to
investigate their efficiency for removing disinfection byproduct precursors. After membrane
selection trials were completed, a mobile trailer was used to evaluate the performance of the
selected membrane. With a system recovery (permeate flow/raw water flow) of 75 percent at
1
-------
one groundwater site, the average reduction of trihalomethane formation potential and total
organic halide was 95 percent and 96 percent from raw water averages of 456 ug/L and 977
ug/L, respectively.
INTRODUCTION
Increasingly, many volatile organic compounds and non-volatile organic compounds such
as pesticides are being found in our nation's ground and surface waters. The Safe Drinking
Water Act (SDWA), passed in 1974, required the U.S. Environmental Protection Agency
(USEPA) to establish recommended maximum contaminant levels (RMCLs) for each
contaminant that may have an adverse effect on the health of persons. The SDWA also requires
that National Primary Drinking Water Regulations (NPDWRs) establish maximum contaminant
levels (MCLs) or treatment techniques, as well as secondary drinking water regulations(l).
Under the 1986 Amendments to the SDWA, the EPA is to set maximum contaminant
level goals (MCLGs) (formerly called RMCLs) and NPDWRs for 83 specific contaminants(2).
Of these 83 contaminants, MCLGs were proposed for 26 synthetic organic chemicals (SOCs)(3).
Other synthetic organics, both volatile and non-volatile, are to be regulated under the SDWA
Amendments of 1986(4).
According to the SDWA Amendments, granular activated carbon was specified as being
feasible for control of synthetic organic chemicals, but alternative technologies, such as
membrane technologies, could be used to meet MCL requirements. The Drinking Water
Research Division (DWRD) of the USEPA, in Cincinnati, is responsible for evaluations of
various technologies that may be feasible for meeting the MCLs. A type of technology that has
shown considerable promise is reverse osmosis (RQ). Until recently, reverse osmosis was
mainly evaluated for removal of salts and other inorganic compounds but both bench-scale and
-------
field studies have shown its effectiveness for removing SOCs from drinking water. When
alternative sources of water are unavailable, treatment of existing sources becomes imperative.
In addition, if regulations become more .strict regarding trihalomethanes and other disinfection
byproducts, alternative treatment processes will be needed to meet the new standards. This paper
presents data collected during research activities by EPA's DWRD on the performance of
membranes for .removing SOC's, pesticides, and precursors of disinfection byproducts. Different
types of membranes and variety of waters were used in these tests.
If membranes are selected by a community, utility, or individual homeowner as the best
treatment for their situation, performance will be only one selection criteria. Other major
concerns will include concentrate disposal and the cost of the membrane treatment when
compared to other available treatment options.
MEMBRANES
Reverse osmosis does not remove chemicals from water by filtration or adsorption;
instead, it rejects compounds based on molecular properties as well as membrane characteristics.
RO is commonly defined as diffusion through a semipermeable membrane with applied pressure
and is commonly used to desalinate seawater. Pure or "cleansed water" passes through the
membrane and is called the permeate stream. The water that does not pass through the
membrane becomes increasingly more concentrated with impurities and is called the reject. This
technology has been successful in rejecting inorganic compounds at various operating pressures,
but little-work has been performed on organic compounds. Thin-film-composite membranes,
introduced in the 1970's, showed promise for removing certain low molecular weight organics
(MW < 200) including alkanes, alkenes, aromatics, and pesticides(S). 'Molecular characteristics,
such as water solubility, acidity, hydrogen bonding, and branching affect the molecule's ability
-------
to pass through the membrane. Organic molecule characteristics can be ascertained for most
compounds, but the difficulty in predicting their rejection lies in the limited knowledge of the
membrane's characteristics and the subsequent interaction (rejection) of the compound with the
membrane.
Reverse Osmosis
Reverse osmosis membranes vary in modular configuration and polymeric chemical
structure. Two configurations that are commonly used are the hollow fiber and spiral wound.
Semipermeable, hollow fiber membranes are usually produced using aromatic
polyamides. The membrane material is spun into hairlike hollow fibers having an outer diameter
of 85 to 200 um. These fibers are bundled together in either a U-shaped configuration for brine
flow on the outside or in a straight configuration for brine flow that flows inside of the fiber.
The fibers are wrapped around a support frame and the open ends of the looped fibers are
epoxied into a tube sheet(6). Figure 1 shows a typical hollow fiber module.
For most operations, raw water is pumped under pressure (200-400 psig) through a
distributor tube and flows outward through the fiber bundle. A portion of the pressurized feed
water permeates through the wall of each hollow fiber and into the bore, leaving most of the
dissolved solids, organics, and bacteria in the concentrated reject water. The permeate forced
into the bores is withdrawn at the epdxy tube sheet end of the membrane shell. A flow screen
inserted between the bundle and shell permits the concentrate to exit the shell through a reject
port. Product recovery of permeate is usually in the range of 50 to 60 percent of the feed flow
rate(6).
Figure 2 shows a typical spiral wound module. The spiral wound module contains two
layers of semipermeable membranes separated by a woven fabric. The semipermcable
-------
membranes can be constructed of various materials including cellulose acetate, cellulose
triacetate, or thin-film composites. The woven fabric can consist of nylon or dacron, or in the
case of the thin film composites, a polyester web with a polysulfone coating. A flexible
envelope is formed by sealing the edges of the membrane on three sides with the fourth open
side attached to a perforated tube. A sheet of plastic netting placed adjacent to the membrane
envelope separates the membrane layers and promotes turbulence in the feed stream during
operation. The envelope and netting are wrapped around the central tube in a spiral
configuration. Pressurized feedwater permeates through the membrane into the fabric where it is
directed to the perforated central tube for collection and removal as product water. Percent
recovery of permeate element usually ranges from 5 to 15 percent of the feed flow rate, but
with elements arranged in series the percent recovery can be as much as 90 percent.(6)
Ultrafiltration
Ultrafiltration has been used in industry to exclude large organic molecules either for
purification of the permeate or concentration of a marketable retenate. Until recently it had
received little attention as a drinking water treatment process. Ultrafiltration membranes
typically have pore sizes ranging from 40 to 1,000 Angstroms (10"3 to 10"1 microns) and may be
employed for the removal of submicron colloidal particles, microorganisms, silt and
large-molecular-weight organic compounds. They may be considered to be intermediate in pore
size between reverse osmosis (RO) membranes [(1 to 20 Angstroms)(10"5 to 2x10"4 microns)]
and microfiltration membranes [(103 to 104 Angstroms)(10"1 to 10 microns)]. Indeed, some of the
first "ultrafiltration" membranes used in a drinking water treatment application were RO
membranes that were simply operated at a lower ( < 200 psig) (14.06 kg/cm2) feed pressures.
-------
REVERSE OSMOSIS
BENCH-SCALE RO EVALUATION- SOCs
During bench-scale studies, several membranes were evaluated by the DWRD to
determine their potential for removing organic compounds.(7) Table 1 shows the varying
configurations and polymeric chemical structure of the membranes that were tested. These
membranes included cellulose acetate, polyamide, and three types of thin-film composite
membranes. As shown in Table 2, multisolute, aqueous solutions in the concentration range of 6
to 153 ug/L were tested.
Using a single pass system, removal efficiencies were determined for over twenty
volatile organic compounds including alkanes, alkenes, aromatics, and pesticides on spiked
distilled water, spiked groundwater, and organically contaminated groundwaters. Using pressures
of 150-250 psig (10.5-17.6 kg/cm2) and 2 to 4 inch (5.1 - 10.1 cm) diameter membranes,
percent removals were determined for various classes of organic compounds, as shown in Table
2. For most compounds (where data is available) membrane C (one of the thin-film composites)
was superior in removing the organics. This is shown graphically in Figure 3, where average
percent removals for four compounds - Chloroform, 1,1,1-Trichloroethene,
cis-l,2-Dichloroethylene, and Trichloroethylene - vary from 4 percent for cellulose acetate
membranes to 72 percent for one thin-film composite membrane.
FIELD EVALUATIONS OF RO MEMBRANES AT SUFFOLK COUNTY, NY - SOCs
Since 1978, Suffolk County has examined groundwater for agricultural and organic
contaminants as well as for their decay products. During this testing, 101 agricultural or organic
-------
compounds were evaluated; 41 were found in the groundwater. Many of these contaminants
were present in trace quantities (1-10 ug/L), but four agricultural compounds were found to be
present at levels that were at times > 100 ug/L: aldicarb, carbofuran, 1,2-dichloropropane
(1,2-DCP), and 1,2,3-trichloropropane (1,2,3-TCP). Nitrates from fertilizer applications were
also present in quantities exceeding the primary drinking water standard (up to 15 mg/L).
4
Concern about this contamination led to a study to determine a cost-effective system for
removing agricultural chemicals from drinking water.
A cooperative agreement was initiated by the DWRD to examine the cost effectiveness
and removal efficiency of certain water treatment systems for removing agricultural chemicals
from Suffolk County groundwater.(S) Two parallel treatment systems were evaluated for
pesticide and organics removal: (a) granular activated carbon (GAC) plus ion exchange and (b)
reverse osmosis. A knowledge base was developed at pilot-plant flows for application to a
full-scale municipal system.
Because data for removal of aldicarb sulfone, aldicarb sulfoxide, carbofuran,
1,2-dichloropropane, 1,2,3-trichloropropane, and nitrate by membranes was not available, reverse
osmosis membrane manufacturers were asked to supply commercially available hollow fiber and
thin-film configuration membranes for evaluation. One cellulose acetate and six polyamide
membranes were received.
Each membrane was evaluated using bench-scale reverse osmosis units ranging from 2
to 4 inches (5.1 - 10.1 cm) in diameter. All units were operated continuously at a pressure of
160-200 psi (11.3-14.1 kg/cm2), and each membrane operated for periods ranging from 5 to 24
weeks. The systems were operated in a one pass mode and were fed from a common raw,
ground water source. Results from this evaluation are shown in Table 3. The cellulose acetate
membrane effectively removed carbamates (aldicarb sulfone, aldicarb sulfoxide, and carbofuran)
but was not efficient in removing the other organics evaluated. Polyamide membranes removed
-------
carbamates, the other organics, and nitrates. Removals ranged from 95 to > 97 percent for
aldicarb sulfoxide, 94 to 98 percent for aldicarb sulfone, 4 to 88 percent for
1,2-dichloropropane, 0 to > 85 percent 1,2,3,-trichloropropane, 86 to > 93 percent for
carbofuran, and 74 to 96 percent for nitrates.
Additional data reported by Eisenberg and Middlebrooks support the conclusions
reported above for membrane performance(6). Cellulose acetate membranes showed the least
overall rejection of organics while composite membranes of polyamide with cross linked surface
structure were more effective in limiting organic penetration. The percent rejection for
compounds with molecular weights less than 100 was uncertain. As the molecular weight
increased, rejection by reverse osmosis membranes increased. For chlorinated hydrocarbons and
organophosphorous pesticides such as DDT, aldrin, parathion, endrin, chlordane, PCBs,
methoxychlor, and malathion, high removals were attained.
During a follow-up 12 month pilot plant study at Suffolk County, using a hollow fiber
polyamide membrane, 3.9 million gallons (14.76. million L) of water were treated producing 2.6
million gallons (9.84 million L) of potable water. This resulted in approximately 67 percent
recovery for an influent of 8 gpm (30.3 L/min), using 400 psi (28.1 kg/cm2) feed pressure at a
water temperature of 55°F (41.4°C). The unit consisted of three membrane cells piped to give
parallel flow to Cells 1 and 2, with the concentrate from each passing through Cell 3 before
disposal.
This reverse osmosis unit removed both volatile and non-volatile organics from Suffolk
County's groundwater. Table 4 shows the arithmetic averages for these compounds. For aldicarb
sulfoxide and aldicarb sulfone, the removals were always .greater than 91 percent; for
carbofuran, removals were > 76 percent, and for'1,2-dichloropropane, removals varied from 53
to 71 percent. Preliminary indications are that certain RO membranes are very effective for
removing a wide range of organic chemicals.
8
-------
ULTRAFILTRATION
BENCH-SCALE EVALUATION OF ULTRAFILTRATION MEMBRANES - SOCs
To investigate the performance of an ultrafiltration (UF) membrane to reject selected
SOCs on a short term basis, a small pilot plant was constructed on the University of Central
Florida campus(9). The pilot plant housed a single FilmTec N 70 membrane element and
included prefiltration and antiscalent feed. SOC spiked water was used for these tests.
Thereafter, SOC-contaminated streams from the pilot plant were discharged to the campus
wastewater treatment plant. Approval for this discharge was obtained from the Florida
Department of Environmental Regulation.
The SOCs selected for this study were alachlor, chlordane, heptachlor, methoxychlor,
dibromochloropropane (DBCP) and ethylene dibromide (EDB). Alachlor was obtained as a
water-soluble, formulated compound and was directly dissolved to prepare a stock feed solution
using permeate water. The other SOCs were first dissolved in either acetone or methanol and
then diluted with permeate water. The SOCs were fed, individually, into the pretreated well
water to achieve feed concentrations generally less than 100 ug/L.
The pilot plant was operated for a one-month period of continuous operation for each
compound. During the month, operating conditions were varied according to a pre-determined
pattern of recovery, with and without recycle. Physical and chemical parameters were monitored
during the operation and additional samples were collected at the end of each run to determine
if any adsorbed SOCs could be flushed from the membrane.
The results from this study showed that EDB was not rejected by the membrane, DBCP
was partially rejected ( 19 to 52%) and all of the remaining SOCs were completely rejected.
Mass balances conducted on the pilot plant system showed that the three SOCs chlordane,
-------
heptachlor and methoxychlor - were adsorbed onto the membrane, but did not desorb during the
one-month operation. Desorption, resulting in contamination of the permeate, may occur with
longer periods of operation however.
FIELD EVALUATION OF UF MEMBRANES - PRECURSOR CONTROL
Removal of THM Precursors - Short-Term Studies
"The DWRD funded, in 1983, bench-, pilot-and plant-scale studies at the University of
Central Florida to evaluate the costs and performance of several new treatment technologies for
reducing the concentration of trihalomethanes in drinking water (10). Four drinking water
treatment plant sites in the State of Florida were selected for this study. Each site used a highly
organic ground- or surface water as a raw water source and served a population of less than
30,000 persons. Several processes were investigated at these sites for THM precursors removal:
low-pressure membrane treatment (UF), polyvalent aluminum chloride (PAC1) coagulation,
dissolved air flotation (DAF), lime softening succeeded by alum coagulation as well as
optimization of conventional lime softening and alum coagulation.
The UF studies were conducted at two sites in Palm Beach County, Florida that used
local, rather shallow aquifer sources. At each site, membrane selection trials, using single,
spiral-wound, thin-film composite membrane modules [4 in. (10.16 cm) diameter by 40 in.
(101.6 cm) length], were conducted using a small (1000 gpd)(3785 Lpd) test unit. Candidate
membranes obtained from several manufacturers, had molecular weight cut-offs (MWCs) that
varied from 100 to 40,000 daltons. Analysis of the water quality data from these trials showed
that a membrane with a MWC of 400 daltons or less would be required to produce a permeate
that, when chlorinated, would meet the THM MCL. On this basis, and considering other water
10
-------
quality parameters and the required operation pressure, the FilmTec N-50 membrane, with a
molecular weight cut-off of 400 daltons, was selected for use in the pilot plant studies.
While the membrane selection trials were being conducted, a mobile UF pilot plant with
a maximum permeate capacity of 18,000 gpd (68,130 Lpd) was installed in a 30-ft. (9.14 m)
trailer. The original configuration of this pilot plant was four pressure vessels connected two
each in series to form two pressure stages. The trailer was equipped for prefiltration, antiscalent
feed, degasification, chlorination, stabilization and permeate storage. First stage feed pressure
could be varied from 80 to 120 psig (5.62 - 8.44 kg/cm2) and system recovery varied from 50 to
90 percent.
The mobile pilot plant was moved to the first test site in January 1985 and was operated
for 365 non-continuous hours over a two-month period. The raw groundwater at this site had a
trihalomethane formation potential (THMFP) that ranged from 400 to 700 ug/L. The THMFP of
the system permeate ranged from 8 to 28 ug/L. Typical distribution system THMs at this
location, which used conventional lime softening treatment, were in excess of 600 ug/L. The
results of the tests showed that the UF system could produce a permeate that easily met the
existing THM MCL.
The mobile pilot plant was moved to the second test site in March 1985 and was
operated for 1020 non-continuous hours over a three-month period. The raw groundwater at this
second site had a THMFP that ranged from 300 to 500 ug/L. As before, the system permeate
(THMFP: 18 to 43 ug/L) easily met the THM MCL of 0.10 mg/L. Conventional lime softening
was used in this location and typical distribution system THMs were in excess of 275 ug/L.
Removal of THM Precursors - Long term Studies
Encouraged by the substantial rejection of THM precursors-demonstrated by a UF
11
-------
membrane during short-term pilot-scale studies, the DWRD decided to fund a second project to
further examine UF treatment of drinking water(ll). This project, also conducted by the
University of Central Florida, had as its primary objective the documentation of the performance
of a UF membrane to reject THM precursors on a long-term (one year) basis. A requirement of
this project was that it be conducted at project sites where the drinking water utility served a
small ( < 10,000) population. Two sites were required, one with a groundwater source, and a
second with a surface water source. Each site was to be used for one year of pilot-scale UF
operation. In addition, estimation of the capital and operation and maintenance costs of UF
treatment from pilot plant data, and consideration of the cost of disposal of the UF system
concentrate was required.
For this project, the mobile UF pilot plant used in the first project was reconfigured
using the same number (4) of pressure vessels. Three pressure stages were used, with the
concentrate from the previous stage used as the feed for the succeeding stage. A higher pressure
feed pump was installed and auxiliary equipment was either replaced or repaired. The modified
pilot plant had a rated permeate capacity of 12,500 gpd (47,313 Lpd) at 75 percent recovery and
a first stage feed pressure of 150 psig (10.6 kg/cm2).
The groundwater test site selected for the UF pilot plant operation was a small water
utility serving a population of about 3,000 persons in Flager County, Florida. The raw water
source was the Floridan aquifer. Typically, the THMFP of the raw water was about 450 ug/L
although this value varied when different combinations of the seven supply wells were used.
The water treatment plant used conventional lime softening, with THMs in the distribution
system averaging more than 300 ug/L. Generally, this water treatment plant was only operated
for about 16 hours per day, limiting the operation of the pilot plant to those hours.
Membrane selection trials were conducted at this site using eight spiral wound, thin-film
composite UF membranes, and the FilmTec N 70 membrane was selected for use in the
12
-------
one-year pilot plant study. The pilot plant was placed in operation in November 1986 and
operated for 5,098 non-continuous hours until November 1987. Operating conditions for this
period were fixed at first stage feed pressure of 150 psig (10.6 kg/cm^and a system recovery of
75 percent. The membranes were cleaned twice during this period; once because of a problem
with the antiscalent feed, and once to remove visible biological growths in the pressure vessels
and interconnecting tubing.
Physical and chemical parameters were monitored during the one year period of
operation. The average reduction in THMFP for the period was 95 percent (average raw
THMFP = 456 ug/L average permeate THMFP = 20 ug/L). The permeate THMFP concentration
was always less than the MCL of 0.10 mg/L. Also, the average reduction in total organic halide
(TOX) was 96 percent (average raw TOX = 977 ug/L average permeate TOX = 34 ug/L). In
general, the permeate was of high chemical quality and the production of permeate was
consistent throughout the operating period.
In November 1987, the mobile pilot plant was moved to the surface water source site in
Charlotte County, Florida, where raw water was obtained from a tributary of the Peace River.
The water utility at this site used alum coagulation treatment to serve a population of about
17,000 persons. Chloramines are used for disinfection, and the THM concentration in the
distribution system averaged 80 ug/L, although the potential to produce THMs with free
chlorine was much higher.
Membrane selection trials, using ten spiral-wound, thin-film composite UF membranes,
were conducted, and the FilmTec N 70 membrane was selected for the one year pilot plant
operation. Pilot plant operation, utilizing the pretreatment scheme of antiscalent feed and
prefiltration, was initiated in November 1987, using a 150 psig (10.6 kg/cm2) first stage feed
pressure, a system recovery of 75 percent and 24-hour per day operation. System flux losses
indicated fouling and the membranes had to be chemically cleaned every two weeks.
13
-------
Although consistent, trouble-free operation of the UF system was not attained at the
surface water site, significant rejection of selected organics did occur. The average percent
rejection of THM precursors, as measured by THMFP reduction, was 94 percent. This
percentage includes the removal of THM precursors by the extensive pretreatment required to
retard membrane fouling. Similarly, the average TOX reduction was 97 percent.
Removal of DBF (other than THMs) - Long term Studies
Methods of analysis for disinfection byproducts (DBFs) other than THMs have been
developed and are now being used to understand the formation and control of these drinking
water contaminants. In addition to the trihalomethanes, these methods will measure
haloacetonitriles, halogenated solvents, chlorinated ketones, haloacetic acids, chlorinated phenols
and several miscellaneous compounds. The availability of these methods led the DWRD to fund
a third project, again with the University of Central Florida, to document the performance of a
UF membrane to reject the precursors of a broad spectrum of DBFs on a long-term (one year)
basis. This current project, which began in August 1988, is structured in a manner similar to
that of the preceding THM precursor removal project.
Two Florida sites have been selected for this project; a groundwater source (Floridan
aquifer) in Volusia County and a surface water source (St. Johns River) in Brevard County. At
the groundwater source site, membrane selection was conducted using a 1,000 gpd (3785 Lpd)
test unit with eleven spiral-wound, thin-film composite UF or low pressure RO membranes.
Selection criteria included rejection of DBF precursors as evidenced by DBF formation potential
comparisons of the raw water and permeate, productivity, required feed pressure, other chemical
parameters and availability of the membrane for pilot plant use. The DuPont 201117 membrane,
which was tentatively, selected based on these criteria, was then subjected to a month-long flux
14
-------
reduction test to determine the efficacy of the antiscalent and five micron prefiltration
pretreatment selected for this site. Flux loss was minimal during this test and the DuPont
membrane was approved for use in the year-long pilot plant operation.
The mobile pilot plant used in the previous two projects was extensively modified into a
three pressure stage configuration containing seven pressure vessels in a 4-2-1 array. All piping
and pressure vessels were replaced with stainless steel to permit higher operating pressures, if
required. Pressure gages, flow meters and feed pumps were replaced and the electrical control
system was completely revised.
Pilot plant operation at the groundwater source site began in May 1989. To date, there
have been no major operating problems with the system. To gain additional data, the system is
being operated with changes in first stage feed pressure and system recovery at monthly
intervals according to a fixed schedule. These variables will range over the one-year test period
from a high of 170 psig (11.95 kg/cm2) feed pressure and 90 percent system recovery to a low
of 110 psig (7.73 kg/cm2) feed pressure and a system recovery of 70 percent. The quality of the
permeate is typical of an RO plant, very soft, low in total dissolved solids and poorly buffered.
Preliminary analysis of the available DBF data indicates that the membrane system is reducing
the formation of disinfection (chlorination) byproducts (converted to their chloride equivalent
concentrations) in the system permeate by greater than 95 percent over those formed through the
chlorination of the raw water.
COST SUMMARY
Reverse osmosis could be considered in three water supply areas: (1) direct water supply
treatment, (2) dual water supply systems, and (3) point-of-use and point-of-entry systems. It
could be considered as a water supply treatment option, primarily because of the minimum labor
15
-------
and technology needed to maintain the systems. If reasonable power costs ($0.03-0.05 kW*h)
and suitable concentrate disposal are available, this unit process can be effective for nitrate,
pesticide, and specific organics removal.
For point-of-entry systems, RO is an expensive first-cost technology ($1000 to
6000+/home).(12) Maintenance, however, would be minimal. Reverse osmosis for point-of-use
systems can be used under-the-sink to provide up to 5-10 gpd (19-38 L/d) of potable water, with
a first cost of $100-1000 and a minimum amount of monitoring and maintenance.
EFFECT OF CONCENTRATE DISPOSAL
Alternatives for concentrate disposal vary widely depending on the location of the
membrane plant and include discharge to the sewer, construction of separate pipelines to a
saline body of water such as the ocean or lagoon, trucking to a landfill, spray irrigation,
deep-well injection, etc. If the concentrate disposal cost is a large percent of the total system
cost, the net cost for both treatment and disposal may increase dramatically.
PROJECTED COSTS OF MEMBRANE TREATMENT AT THE FLORIDA
SITES
With the use of information obtained from pilot plant operations on long-term removal
of THM precursors in the previously mentioned Florida studies, capital and O&M costs for the
installation of a membrane water treatment plant (including concentrate disposal) to meet a 20
year future demand (2.7 MGD) at a groundwater source site (Flager County, Fla.) were
estimated to be essentially equal to the capital and O&M costs to build and operate a
conventional lime softening plant of equivalent capacity. At this location, the membrane plant
would produce drinking water of better quality.
16
-------
-------
At the surface water site, Charlote County, Florida the use of the membrane process
would require the construction of a complete alum coagulation, sedimentation and filtration
plant to be used as pretreatment. Operating a membrane plant on the highly organic surface
water found in the State of Florida would require lower design flux and system recovery, more
frequent membrane cleaning and extensive pretreatment. The cost of membrane treatment for
this type of surface water may be unreasonable unless future regulations for control of DBFs in
drinking water force further consideration of this technology.
CONCLUSIONS
Membranes for removing organic compounds and precursors of organic compounds from
drinking water show considerable promise. Bench and pilot plant studies on actual waters have
shown that several organics proposed for regulation can be removed by membranes. As
membrane technology improves, rejection of more difficult to remove compounds is expected to
improve. Also, smaller volumes of concentrate are expected to be produced which can be
handled more cost effectively.
One major concern with the use of membranes is concentrate disposal which may
increase the overall cost of treatment and disposal. The cost of membranes is very sensitive to
such factors as recovery, economies of scale, systems configuration, membrane type, and electric
power cost. In certain situations, membranes are a viable treatment option that is not cost
prohibitive.
17
-------
REFERENCES
1. "Safe Drinking Water Act", Public Law 93-523, Dec. 16, 1974.
2. "The Safe Drinking Water Act Amendments of 1986", Public Law 99-339, June 19,1986.
3. "National Primary Drinking Water Regulations; Synthetic Organic Chemicals, Inorganic
Chemicals, and Microorganisms: Proposed Rule", Federal Register, 40CFR Part 141,
46936-47022, November 13, 1985.
4. "Drinking Water; Substitution of Contaminants and Drinking Water Priority List of
Additional Substances Which May Require Regulation Under the Safe Drinking Water
Act", Federal Register, Vol. 53, No. 14, January 22, 1988.
5. Probstein, R.F., Calmon, C., and Hicks, R.E., "Separation of Organic Substances in
Industrial Wastewater by Membrane Processes", from Control of Organic Substances
in Water and Wastewater, EPA-600/8-83-011, April 1983.
6. Eisenberg, T.N. and Middlebrooks, E.J., "Reverse Osmosis Treatment of Drinking Water",
Ann Arbor Science Publishers, Inc., 1986.
7. Fronk, C.A., "Removal of Low Molecular Weight Organic Contaminants from Drinking
Water Using Reverse Osmosis Membranes", 1987 Annual AWWA Conference
Proceedings, Kansas City, MO, June 14-18, 1987.
8. Baier, J.H., Lykins, B.W., Fronk, C.A., And Kramer, S.J., "Using Reverse Osmosis to
Remove Agricultural Chemicals from Groundwater", JAWWA, August 1989.
9. Taylor, J.S., et al. "SOC Rejection by Nanofiltration", USEP A/600/2-89/023; February
1989.
10. Taylor, J.S., et al., "Cost and Performance Evaluation of In-Plant Trihalomethane Control
Techniques", USEP A/600/2-85/138, January 1986.
11. Taylor, J.S., et al., "Cost and Performance of Membranes for Organic Control in Small
Systems", USEP A/600/2-89/022, May 1989.
12. Lykins, B.W., Clark, R.M., Fronk, C.A. "Reverse Osmosis for Removing Synthetic
Organics from Drinking Water: A Cost and Performance Evaluation". 1988 Annual
AWWA Conference Proceedings, Orlando, Florida, June 19-23, 1988.
18
-------
o
I
.DJ
O
O
0
2
J2
0
LLJ
<
S
^
LLJ
X
o
"z.
o
*Ł.
CL
O
LL.
O
O
LLJ
Q_
LLJ
Z
DC
m
Ťt^
LLJ
5
0
Ť-j
al
00
o
CO
Jv-
1
^t
°o
^^
se Acetate (Diacetate
0
^
"0
0
TJ
C
O
^
"2
"a.
CO
"0
o
0
CO
^
^3
"0
O
ic Diamide
ic Acid Chlorides
to to
E E
2 2
< <
1
u_
.0
"5
X
0
_H
CO
^%
"o
Q.
0
$
^%
g
_Q
i_
o
C/J
.Ł
oj
"c
o
O
0
c
CO
1
O
CO
0
"52
Q_ O
0
"55
O
Q.
E
o
O
-Q
C
0
"co
'a.
CO
0
c
2
.a
E
0
<
0
4-
'co
o
Q.
E
o
0
E
LL
c
Ł
E
a
0
-* ,
C CO
' c
CO O
CO
Ow
CO
0 O
.E O
Qj
'c co
0 C
">. fo-tf
Ł C^O
5,8 |
c?^^
0 *- ?
S CV o
^02=
<-> C U
2 CD CO
+- 3 _>ť
co -5 o
0)1-0.
5 E §
8.C 2
P -s °
Q- $ OL
jD
"co
O
Q.
. E
o
0
-a
c
o
"2
"a.
CO
0
c
CO
o
E
0
CD
S
1
Q.
E
0
O
Ł
u_
c
*-
CO
c
0
0
CO
co
CD
0
c
,f2
-
en ^~
^^ ^
CO "O
0 c
g-5
0 1
ll
. *y
o ca
S CD
s
'co
o
a.
E
o
O
~a
c
u
o
2
"o.
CO
0
2
E
0
0
0
+-*
"co
o
Q.
E
o
O
_E
LL
c
H
-------
TABLE 2. REMOVAL OF LOW MOLECULAR WEIGHT ORGANIC CONTAMINANTS
BY VARIOUS REVERSES OSMOSIS MEMBRANES *
MEMBRANETYPE
COMPOUNDS
ALKANES
1 ,2-DICHLOROETHANE
1 ,2-DICHLOROPROPANE
CHLOROFORM
1 , 1 ,1 -TRICHLOROETHANE
CARBON TETRACHLORIDE
BROMODICHLOROMETHANE
DIBROMOCHLOROMETHANE
BROMOFORM
ALKENES
CIS-1 ,2-DICHLOROETHYLENE
TRANS 1,2-DICHLOROETHYLENE
TRICHLOROETHYLENE
TETRACHLOROETHYLENE
AROMATICS
BENZENE
TOLUENE
ETHYLBENZENE
0-XYLENE
P-XYLENE
CHLOROBENZENE
O-DICHLOROBENZENE
P-DICHLOROBENZENE
M-DICHLOROBENZENE
BROMOBENZENE
1 ,2,4-TRICHLOROBENZENE
PESTICIDES
ETHYLENE DIBROMIDE
ALACHLOR
METOLACHLOR
MOLECULAR
WEIGHT
98
113
119
133
154
164
208
252
97
97
132
166
78
92
1.06
1 HR
1 UO
106
112
147
147
147
157
182
188
270
284
CELLULOSE
ACETATE
PERCENT
REMOVAL
10
0
15**
7
0
16
0**
20
0**
Ť
2**
10
34
22
~
10
11
17
-
POLY
AMIDE
PERCENT
REMOVAL
61
33
88**
44
32
38
19**
0
31**
~~
18**
"
0
"
-
A
PERCENT
REMOVAL
15**
90
47
100**
--
79
78
81
14**
30
37**
~~
16**
"~
50
65
0
64
--
35**
100**
100**
THIN FILM
COMPOSITE
B
PERCENT
REMOVAL
38
55
97
12
41
71
-
""
54
"
..
-
C
PERCENT
REMOVAL
71
82
97
96
32
75
92
-
-
--
-
87
-
.
-
* RUN LENGTHS RANGED FROM 13-286 HRS.
"AVERAGE OF DISTILLED AND GROUND WATER
TESTS
- TESTS NOT CONDUCTED
ALL TESTS VERIFIED BY MASS BALANCES
ALL REMOVALS REFLECT STEADY STATE CONDITIONS,
ONE PASS TESTS
-------
M
CD
O
O
§
CO
.0
nJ
15
LJJ
0
I
jQ
E
0
CO
o
E
CO
O
0
Ł2
0
0
CC
od
_0
S
i
o
-S
co
O
.2 CD
u co
*§
^= co
O Q.
TJ 2
-g CD
| §
is ^
< CO
e-i
S'g
TJ fc
CO
s
**o bŁ
o** Cť
CD
DC
TJ ^j
CD O)
U. 3
"to
5*
<*i
CD
DC
"§=!
CD O5
U. =
1
vO Zl ,
0s* ^
CD
CC
"S =^
CD O5
U. 3
to
o 0
0s* C
0
DC
TJ _|
CD D)
U_ 3
"co
* §
DC
U_ =>
"co
. §
5s* c
CD
DC
TJ _]
CD =1
CD D5
U_ 3
CD
JS o
DC
mbrane
CD
2
s
CD
Q.
H
^
CO
CO
fs.
in
CO
o
CO
CM
*"
CM
CO
CO
05
u
^
O)
A
cn
co
o
CO
%
1
o
<
o
CO
_O
"CD
O
in
CO
co
CO
O)
O)
CO
CO
in
CO
o
m
CM
0)
f^
O)
J^
rt
CO
A
CO
co
o
o
TJ
C
o
1
Q.
CO
"c
3
Q
CO
O)
co
CO
O)
O)
CO
CO
CO
12
CM
h-
m
05
r-.
^t
CO
A
CO
co
o
o
in
CD
iZ
1
0
<4^
3
Q
co
in
CO
CO
CO
CO
m
CO
co
co
in
co
CM
N.
in
05
^
^
S3
A
cn
co
o
CO
TJ
C
33
O
1
'o.
CO
CO
co
co
TJ
X
co
m
CO
CO
CO
O)
CO
co
CO
in
co
CM
°i
O)
A
N.
^
o
A
0)
CO
O
o
T
TJ
1
2
Q.
CO
u
0)
E
u_
CO
CO
O5
CO
O5
O)
CO
co
co
CM
O5
CM
O5
O5
A
f-
^
CO
A
O5
CO
O
in
TJ
o
2
Q.
CO
o
0>
E
u.
CO
O5
CO
CO
cn
m
CO
co
co
in
co
CM
CO
8
^
0)
^t
CO
CO
co
0
CO
TJ
C
o
1
Q.
CO
CO
to*
TJ
U.
TJ
0)
5
=
8
Q.
CO
CO
o
TJ
0)
D5
2
o>
.2
co
to
Q
S
o
-------
o
O
Q
Q.
|
Q.
CO
"8
0)
o
0)
o
.3
I
,
OJ
Q) -U
<2 CB
0) O
5 1
cc <
5
CO
CD
I
3
CO
u
T3
K
UJ
0)
CO
rocooococooocococot^-r^N-
AAAAAAAAAAAA
oooooooooooo
vvvvvvvvvvvv
OOOi-00 o r^ co"
CMCMCMT--1-CMCMT-CMT--I-
pcMcqcqcjcnincoco^T-
AAAAAAAAAAA
p p O O p O O O O O p
'v^v'v'v'v'v^v'v^^v'v
p CM <* CM r*; p * p p co_ CM
ort*hs-co^4"^*cococoT~T~
CMl--I-l--I-1-^T-T-1-T-
coc\joo-
CO
-------
LU
Q
O
LU
<
DC
S
CE
I
O
LU
DC
LL
-------
LJJ
Q
O
Q
O
E
CO
CM
LLJ
DC
O
-------
CD
ca
o
_O CD
II
5 Ť
-c of
O c
o o
E o
ir-g
|Q
CD V
I
------- |