United States Environmental Protection Agency Risk Reduction EngineeringLaboratory Cincinnati OH 45268 Research and Development EPA/600/S2-89/022 Jan. 1990 4>EPA Project Summary Cost and Performance of Membranes for Organic Control in Small Systems: Flagler Beach and Punta Gorda, Florida J. S. Taylor, L. A. Mulford, W. M. Barrett, S. J. Duranceau, and D. K. Smith A membrane pilot study was conducted to determine if membrane processes could remove trihalo- methane (THM) precursors from highly organic surface and ground- waters and consistently maintain water production for 1 yr at each location. The scope of this 33-month project involved: site selection, mem- brane selection, 1 yr of membrane pilot plant operation at each site, and a cost comparison of the membrane process and the conventional water treatment process at each site. The highly organic Florida waters, a groundwater and a surface water, were selected. Several membranes were tested for pilot use by short- term testing at each site. Although four membranes controlled THMs to less than the THM Maximum Contaminant Level (MCL), the FilmTec N 70* nanofilter was selected at each site for long-term use because of both productivity and THM control. A three-stage membrane pilot plant operated at each site controlled THM formation potential (THMFP) to less than 0.10 mg/L. Organic and inorganic water quality, water production, and cost of treatment were reported for membrane utilization at both sites. •Mention of trade names or commercial products does not constitute endorsement or recommendation for use. This Project Summary was developed by EPA's Risk Reduction Engineering Laboratory, Cincinnati, OH, to announce key findings of the research project that is fully documented in a separate report of the same title (see Project Report ordering information at back). Introduction The purpose of this project was to determine if a membrane process could maintain consistent water production while keeping permeate THMFP below the 0.10 mg/L THM MCL. The scope of the project was to select both a groundwater and surface water site that had a highly organic raw water, select a membrane for extended operation on based on THM control and water production, operate a membrane pilot plant for 1 yr at each site, monitor THMFP as well as other water quality parameters and water production, and compare the cost of membrane and conventional processes at each site. The groundwater site selected for 1 yr of pilot plant operation was Flagler Beach. The 0.4 mgd Flagler Beach water treatment plant (FBWTP), which utilizes conventional lime softening, is located on the east coast of Florida near Daytona Beach and serves a population of 3,000. The THM concentration in the Flagler Beach distribution system averages more than 300 jig/L. Punta Gorda, located on the west coast of Florida about 80 miles ------- south of Tampa, was selected for the surface water site and used Shell Creek as a raw water source. At the start of the 1 yr of pilot plant operation at this site, the Punta Gorda water treatment plant (PGWTP) used an alum-coagulation process that was being operated in a fill- and-draw mode. While the pilot plant was in operation, a new PGWTP, using a continuous alum-coagulation process, was placed in operation. At the PGWTP, the average daily flow was 2.7 mgd for a service population of about 17,000. At present, the PGWTP uses chloramines for disinfection, and THMs in the distribution system average 80 jig/L. Before the switch from a free chlorine residual to chloramines, distribution system THMs exceeded 300 ng/L. Membrane Selection Eight spiral-wound, thin-film composite membranes were tested at Flagler Beach for THM precursors and permeate productivity. Single 4-in. diameter by 40- in. length membranes were installed in a small-scale test unit and tested at feed pressures of 105 to 140 psi at a recovery of 75%. Four of the membranes produced a permeate with a THMFP of less than the 0.10 mg/L THM MCL. Of these four, the FilmTec N 70 membrane was selected for the 1 yr pilot plant operation because its solvent (water) mass transfer coefficient (MTC) of 0.0137 day1 was the highest of all membranes tested at this site. Membrane testing at Flagler Beach showed that a membrane molecular weight cutoff (MWC) of 500 or less was needed to remove sufficient THM precursors for the permeate THMFP to be less than the 0.10 mg/L THM MCL. Similarly, an MWC of 300 or less was needed to reduce permeate total hardness to 150 mg/L as CaC03. Ten spiral-wound, thin-film composite membranes were tested at Punta Gorda for removal of THM precursors and permeate productivity. With the use of the small-scale test unit, the single membranes were tested under similar operating conditions using both raw and alum-coagulated feed waters. Four of the membranes were able to reduce the THMFP of the feed waters to less than the 0.10 mg/L THM MCL. Of these four membranes, the FilmTec N 70 had the highest solvent MTC (0.0106 day1 with a raw-water feed, 0.0048 day1 with an alum-coagulated feed) at this location and was again selected for the 1 yr pilot plant trial. Permeate solute concentrations were lower for all membranes when the feed had been alum coagulated first; however, all membranes whose permeate THMFP could meet the THM MCL using alum-coagulated feed, could also meet the THM MCL using raw-water feed. Since the FilmTec N 70 membrane showed greater permeate productivity (higher solvent MTC) on raw-water feed than on alum-coagulated feed, a raw- water feed was chosen to begin the Punta Gorda pilot plant operation. All membranes with an MWC of 500 or less were able to produce a permeate THMFP that could meet the THM MCL at the Punta Gorda site. Pretreatment The silt density index (SDI) was used to determine pretreatment requirements to prevent membrane fouling at Flagler Beach. The average of three SDI tests conducted on the Flagler Beach raw water was 2.2, which is less than the 3.0 maximum recommended for membrane plant feed waters. The solubility calculations made to define scaling problems indicated that calcium carbonate was the salt limiting recovery. Based on the manufacturer's recom- mendations, an initial dose of 6 mg/L of Pfizer FloCon 100 antiscalent was used. A flow diagram of the membrane pilot plant is shown in Figure 1. The mini-plugging factor index (MPFI) and the SDI were used at Punta Gorda to determine pretreatment requirements. Several different pretreatment schemes were investigated, including: -untreated raw water (R) -raw water plus sand filtration (RSF) -alum coagulated and settled water (ACS) -alum coagulated and settled water plus sand filtration (ACSSF) The lowest acceptable MPFI and SDI test values for membrane processes were for the RSF pretreatment option, which was selected for use at pilot plant start-up. Limiting salt calculations indicated that calcium carbonate also controlled recovery at Punta Gorda. A 5-mg/L dose of FloCon 100 was initially selected to control scaling. Operation at Flagler Beach The FBWTP personnel who operated the pilot plant at Flagler Beach kept daily records of pressure and flow. Pilot plant repairs, makeup of feed chemicals, and bi-weekly collection of samples were the responsibility of personnel from the University of Central Florida. The pilot plant only operated when the FBWTP operated, generally about 16 hr per day. From November 1986 to November 1 the pilot plant operated for 5,098 hr, 64 needed for maintenance. Additi operating time was lost because of re delays. Water production for the year of plant operation at Flagler Bead summarized in Table 1. The average 1 pressure was 141 psi, and the avei recovery was 79%. The average w flux from stage 1 (S1), stage 2 (S2), stage 3 (S3) was 12.5, 15.3, and g/sfd, respectively. Overall system averaged 12.8 g/sfd. As shown in Fie 2(a), system flux was very consis during the year of operation. Average values of the concentrati of selected water quality parameters the raw water, the FBWTP finished we and the membrane pilot plant perme are listed in Table 2. The perme THMFP averaged 20 ng/L and v consistently low throughout the year shown in Figure 2(b). Stage perme concentrations were typically in the or S2 < S1 < S3, which was the inverse the stage order water flux (S3 < S1 S2) as one would expect in a diffusi controlled process. The membranes were cleaned tw during the year at Flagler Beach. The f cleaning at 280 hours was necess, because iron fouled the FloCon 1 antiscalent and the antiscalent lost ability to prevent calcium carbon; scaling. The membranes were clean with sulfuric acid, and a sulfuric acid fe was substituted for the FloCon 100 prevent scaling. The membranes th operated until 3,891 hr when the seco cleaning was done to remove visit biological growths in the pressure vesse and transmission lines; it was not do because of any change in water quality production. If sulfuric acid had been us initially for the antiscalent feed, there no evidence that a cleaning to maint. water quality or production would ha been necessary during the year operation. Operation at Punta Gorda The membrane pilot plant operated total of 6,676 hr from November 1987 November 1988 at Punta Gorda. Pit plant maintenance required 198 hr downtime, and additional time was lo with PGWTP shutdowns and repa delays. The pilot plant was operated t PGWTP personnel, with maintenance ar other duties shared between PGWTP ar University personnel. Pilot plant operation was much difficult at the surface water site. Th ------- Supply Pump uptionai Holding lank C ) Optional Sand Fitten Alum or Raw Water Feed Line Concentrate Line Sample Point n Valve High Pressure Pump Permeate Discharge Pump Figure 1. Flow diagram For the F/lmTec N 70 Membrane System. pretreatment indices (SDI and MPFI) did not accurately predict the extent of membrane fouling that occurred. The membranes were cleaned on 20 occasions because of low flux caused by organic fouling. Several pretreatment changes were made during the year in an attempt to maintain the average system flux above 10 g/sfd. The most significant of these changes were: (1) the addition of a second sand pre-filter, (2) a change from an antiscalent feed to a dispersant/ antiscalent feed, (3) use of alum- coagulated and settled water from the PGWTP for the membrane feed, and (4) use of a sulfuric acid feed to dissolve any aluminum hydroxide colloids remaining in the alum-coagulated water after settling. The most successful pretreatment configuration was found to be the use of alum-coagulated and settled water from the PGWTP, following by two sand prefilters in series with sulfuric acid addition and 5-u final filtration. The average flux and average rate of flux loss for this pretreatment configuration was 8 g/sfd and 0 05 g/sfd2, respectively. Under these operation conditions, membrane cleaning was required about every 16 days to avoid a production loss of greater than 10%. Product recovery was also varied from 80% down to 40% in an attempt to maintain a higher velocity across the membrane surface to scour out foulants. Operation at 40% recovery produced the lowest average flux (7.1 g/sfd), but also least rate of flux loss. Water production for the year of pilot plant operation at Punta Gorda is also summarized in Table 1. As shown in Figure 2(c), system flux at Punta Gorda varied greatly because of numerous membrane cleanings and changes in the pretreatment configuration in an attempt to stabilize water production. Average values of the concentrations of selected water quality parameters for the raw water, the PGWTP finished water, Table 1 Average Water Flux, MTC and Pressure Drop by Stage for Flagler Beach and Punta Gorda FilmTec N 70 Membrane Plants. Flagler Beach Location Stage 1 Stage 2 Stage 3 System Flux glsfd 125 15.3 11 1 128 MTC" 1/day 0.015 0.009 0.008 0.005 Recovery % 38 38 43 79 Pressure Drop psi 57 101 100 141 "Mass Transfer Coefficient Punta Gorda Flux glsfd 8.0 8.0 9.0 8.2 MTC" 1/day 0.003 0.003 0.040 0.003 Recovery % 22 17 22 56 Pressure Drop psi 134 132 119 152 ------- Table 2. Average Values for Raw, Finished Plant and Membrane Water Quality at Flagler Beach and Punta Gorda FilmTec N 70 Membrane Pilot Plant Operations. Flagler Beach Parameter THMFP (itg/L) TOXFP (ng/L) Color (cpu) DOC (mg/L) TDS (mg/L) A/a (mg/L) TH (mgIL CaCOs) CaH (mg/L CaCOy) Cl (mg/L) Alk (mg/L CaCOJ PH Raw 367 1186 30 8.9 625 49 352 327 122 338 7.9 Finished 350 1105 5 7.6 530 49 111 91 122 98 8.3 Membrane 20 33 29 6.8 105 22 35 31 102 35 5.5" Raw 880 2737 46 20.6 411 44 227 165 85 121 7.4 Punta Gorda Finished 3821/80" 1414 3 6.4 473 57 222 161 105 116 8.1 Membrane 37 53 3 0.6 99 51 31 24 35 20 3.8" t THMs formed with free chlorine present over 96 hours " THMs formed in distribution system with chloramme disinfectant " Prior to stabilization and the membrane pilot plant permeate are listed in Table 2. The feed and permeate THMFP concentrations are shown in Figure 2(d) for the year of operation at Punta Gorda. The permeate THMFP exceeded the THM MCL on only one occasion when the new PGWTP was put on-line. The permeate THMFP averaged 37 ug/L, which was less than the average THM concentration of 80 vig/L in the Punta Gorda distribution system. An upward shift occurred in the concentrations of all organic and inorganic permeate parameters when the pretreatment was changed to alum- coagulated and settled feed water Although alum coagulation removed many foulants that impeded water flux, the presence of these foulants may have resulted in the formation of a tighter secondary layer on the membrane surface that would have increased solute rejection and decreased concentrations in the permeate. The order of stage flux before and after the change to alum-coagulated and settled-feed water was S1 > S2 > S3 and S3 > S1 = S2, respectively. The stage order for permeate concentration before and after the change to alum- coagulated and settled-feed water was typically S3 > S2 > S1 and 81 > 82 > S3, respectively. The stage pressure drops after the pretreatment change indicated that less fouling material was reaching S3, therefore S3 productivity was increased and S3 permeate concentrations were decreased. As at Flagler Beach, the stage and system solute concentrations were indicative of a diffusion-controlled process. Cost With the use of information obtained from the pilot plant operation, the estimated capital and O&M costs for installing a membrane water treatment plant of equivalent capacity (0.7 mgd) to the existing lime softening plant at Flagler Beach are shown in Table 3. A membrane plant (including concentrate disposal) installed at the Flagler Beach groundwater site would cost less to operate but would require more to build because the existing 0.70 MGD lime softening plant is debt free. Installation of a membrane plant (including concentrate disposal) at Flagler Beach with a 2.7 MGD capacity to meet the future demand for 20 years would essentially cost the same to build and operate as an equivalent capacity lime softening plant. The membrane plants would produce better water quality than the lime softening plants. Similar cost information for the surface water site at Punta Gorda is given (Table 3). Here, the use of a membrane process essentially requires the construction of a complete alum coagulation, sedimenta- tion, and filtration plant to be used as pretreatment in front of the membranes. Thus, the costs for the membrane operation are considerably higher. Operating a membrane plant on the highly organic surface waters found in Florida would require lower design flux and recovery, more frequent membrane cleanings, and very extensive pre- treatment than would a plant at groundwater site. The cost of membre treatment for this type of surface wa may be unreasonable unless futi regulations for THMs and disinfecti byproducts force further consideration. Summary A membrane plant, operati noncontinuously for a year on a grour water source at Flagler Beach, Floric produced a permeate of high chemk quality while maintaining consiste production. The permeate THMFP w always less than the current THM MCL 0.10 mg/L. On a unit cost basis, the cc of constructing and operating equivale capacity lime softening and membrai plants would be nearly equal, and tt membrane plant would produce superi quality drinking water. At the Punta Gorda, Florida, surfa< water source site, the membrane pil plant could not maintain consiste production because of severe organ fouling problems. The permeate THMF concentration consistently averaged lei than the current THM MCL, however, ar was also less than the distribution systei THMs produced by the existing alui coagulation plant using chloraminatioi Additional pretreatment involving alui coagulation, pH control, and san filtration was needed to improve produt tion and lessen the rate of flux los between membrane cleanings. The Punl Gorda data indicate that a membran process designed for a highly organi surface water source must be based onf lower design flux and recovery, mor ------- I 3 1 I 1 I 0. I 600 500 400 300 200 100 15 12 9 6 3 1500 1200 900 600 300 Operating Year Complete Operating Year Complete Permeate THMFP Flux Operating Year Complete Punta Gorda Feed THMFP Punta Gorda Operating Year Complete Permeate THMFP (A) (B) (C) (D) 1000 2000 3000 4000 Hours of Operation 5000 6000 7000 Figure 2. Water flux and permeate THMFP for the FilmTec N 70 Membrane System at Flagler Beach and Punta Gorda, Florida. frequent membrane cleanings, and exten- sive pretreatment. Although constructing and operating a membrane plant at Punta Gorda was shown to be more than twice as costly as the existing alum coagulation process, the plant would produce a better quality drinking water. The full report was submitted in ful- fillment of CR 813199 by the University of Central Florida under the sponsorship of the U.S. Environmental Protection Agency. Table 3. Capital and Operation and Maintenance (O&M) Costs for the Present (1988) and Future (2008) Flagler Beach and Punta Gorda Water Treatment Plants and Equivalent Capacity Membrane Plants. Present Location Capacity Flagler Beach Punta Gorda MGD 0.7 8.0 Conventional O&M $11000 gal 1.21 0.99 Capital $ 0 3,500,000 Membrane O&M $11000 gal 1 09 1.86 Capital $ 1,437,000 12,966,000 Future Capacity MGD 2.7 8.0 Conventional O&M $11000 gal 0.69 0.99 Capital $ 3,951 ,000 6,971.000 Membrane O&M $11000 gal 0.72 1.86 Capital $ 3,840,750 16,338,500 ------- J. S. Taylor, L A. Mulford, W. M. Barrett, S. J. Duranceau, and D. K. Smith are with the University of Central Florida, Orlando, FL 32816. J. Keith Carswell is the EPA Project Officer (see below). The complete report, entitled "Cost and Performance of Membranes for Organic Control in Small Systems: Flagler Beach and Punta Gorda, Florida," (Order No. PB 89-190 367/AS; Cost: $28.95, subject to change) will be available only from: National Technical Information Service 5285 Port Royal Road Springfield, VA22161 Telephone: 703-487-4650 The EPA Project Officer can be contacted at: Risk Reduction Engineering Laboratory U.S. Environmental Protection Agency Cincinnati, OH 45268 United States Environmental Protection Agency Center for Environmental Research Information Cincinnati OH 45268 US.OFFiCiALMAiL ,,•:,; 4' M i W/ATI- Official Business Penalty for Private Use $300 EPA/600/S2-89/022 -SI'/ATI- _ r -. ,'ji£S3CC| II ^ *1 - /» » j -• U «J »* - V , /- a METER v C.' \^s :;jro4'i41 00005833 S GE»C, ------- |