Commissioning and Operation of a 50 mgd Ultrafiltration Advanced Reclamation Facility for Gwinnett County, Georgia

Transcription

1 Commissioning and Operation of a 50 mgd Ultrafiltration Advanced Reclamation Facility for Gwinnett County, Georgia ABSTRACT Robert A. Bergman*, Richard Porter**, Don Joffe***, Ed Minchew**** * CH2M HILL 3011 SW Williston Road Gainesville, FL **Gwinnett County, GA ***Jordan Jones and Goulding, Atlanta, GA ****CH2M HILL, Atlanta, GA Gwinnett County, located near Atlanta, Georgia, is one of the fastest growing counties in the nation. A new 20 mgd state-of-the-art water reclamation and processing facility, called the F. Wayne Hill Water Resources Center, became operational in December 2000 and is being expanded to triple its capacity to 60 mgd (maximum month flow basis). The expanded plant will produce high-quality product water discharging to Lake Lanier, which is the raw water supply source for the County s potable water treatment plant. The expansion includes a nominal 50 mgd ultrafiltration (UF) system treating chemical-clarified secondary effluent which will be fully operational in After an extensive 1-year testing period and an evaluation of performance, cost, and other factors, membrane filtration was selected as part of the advanced treatment processes to treat secondary effluent. The Zenon ZW-500c UF membrane process was the highest-ranked proposal following a selection process that evaluated eight different membrane filtration systems from five manufacturers. After successful proof testing with a pilot system operated at the asproposed design criteria and operating conditions, the UF system was incorporated into the final design of the full-scale facility as a package system to be supplied by Zenon as a subcontractor to the successful general contractor. The UF system consists of 16 parallel process trains with a high degree of redundancy for added reliability. Full production capacity 48 mgd permeate can be met with 14 trains in service, and all critical auxiliary equipment and piping are designed for redundancy with backups. Each train has a capacity of 3.6 mgd feed flow and operates at a minimum recovery of 96% at 20 C. The Phase 2 construction began on the AWT (advanced water treatment) facilities in the second quarter of The performance testing and commissioning of the membrane process trains is currently underway. At the time of this writing (June 2006), performance testing has been completed on the first eight trains and is in progress for the remaining eight trains. Before completion of testing and final acceptance by the County, 14 trains will be operated simultaneously. 2698

2 This paper summarizes the membrane process design criteria, start-up and commissioning activities, and performance test results to date. Previous publications have described pilot, demonstration, and proof testing and other facets of the project [1-6]. When the 50 mgd feed (48 mgd treated product) ultrafiltration system is fully-operational in 2006, it will be one of the largest membrane facilities in the world, the largest membrane facility in the eastern United States, and the largest capacity operating membrane filtration facility treating municipal wastewater in the country. KEY WORDS Ultrafiltration, Membranes, Reclamation, Reuse INTRODUCTION Gwinnett County is located near Atlanta, Georgia and is one of the fastest growing counties in the nation. A 20 mgd state-of-the-art water reclamation and processing facility costing approximately $200 million, named the F. Wayne Hill Water Resources Center, became operational in December 2000 and was designed for an expansion to triple its capacity to 60 mgd maximum month flow basis. The expanded plant will produce a very high-quality product water which discharges to Lake Lanier, the raw water supply source for the County s potable water treatment plant. Biological processes at the existing (Phase 1) plant were designed for complete nitrification, partial denitrification, and phosphorus reduction. Tertiary treatment processes currently include ferric chloride chemical coagulation/clarification followed by granular media filters, pre-ozone and granular activated carbon (biologically-enhanced activated carbon), and ozone disinfection. Originally, the tertiary treatment used a high-lime process followed by 2-stage recarbonation before media filtration, but the process was changed to metal coagulant addition because all treatment goals could be met with less chemical cost and less sludge to manage and dispose. The selected tertiary treatment process train for the Phase 2 plant expansion will treat secondary effluent and includes (See Figure 1): Metal salt coagulant addition/clarification for reduction of phosphorus, organics and solids Screening with a 500 micron drum-type in-channel automatically-cleaned prefilter (strainer) Particle filtration with vacuum-type ultrafiltration for turbidity and particle (pathogen) removal Blending with the existing granular media filter effluent followed by pre-ozone/gac for organics removal Final ozonation for disinfection. 2699

3 Figure 1 - Chemical Clarifier Ultrafiltration Tertiary Treatment Process Train Secondary Effluent Solids Contact Reactors (with FeCl 3 Addition) Coagulation Flocculation Clarification (with FeCl 3 Addition) 2-Stage Recarbonation/Clarification (No Longer Used for Treatment) Granular Media Filters Automatically-Cleaned Strainers Ultrafiltration Membrane System Pre-Ozone GAC Final Ozone Final Effluent After an extensive 1-year testing period and an evaluation of performance, cost, and other factors, membrane filtration using ultrafiltration was selected as part of the advanced treatment processes to treat secondary effluent. The Zenon ZeeWeed 500c membrane process was the highest-ranked proposal following a selection process that evaluated eight different membrane filtration systems from five manufacturers. In June 2002, a Membrane System Proof Test was started with the ZW-500c pilot-scale system to confirm operating and design criteria and projected operating costs. The proof testing also included an automatically-cleaning in-channel screen as membrane pretreatment. Simultaneously, preliminary design commenced incorporating the membrane system into the construction contract documents for the full-scale facility [membrane system sized for 50 mgd membrane feed flow]. The Phase 2 expansion, which included primary, secondary, and tertiary facilities has a construction cost of approximately $350 million. Figure 2 shows an aerial view of the site during final stages of the Phase 2 construction in May MEMBRANE SYSTEM DESIGN CRITERIA The full-scale membrane system design is based on treating chemically-clarified and screened secondary effluent. Up to 50 mgd of secondary effluent flows by gravity from the secondary clarifiers through the chemical clarifiers and the 500-micron screens to the membrane process tanks. Permeate pumps discharge membrane filtered water to a seal weir box, and then the permeate flows by gravity to downstream processes. During the course of construction, a decision was made to lower the elevation of the seal weir to reduce the discharge head of the 2700

4 Figure 2 - F. Wayne Hill Water Reclamation Facility Preliminary Treatment Chemical Clarifiers (Phase 1) Plate Settlers (Phase 2) Chemical Building (Phase 2) Biological Treatment Membrane Building (Phase 2) Granular Media Filters (Phase 1) Pre-Ozone Secondary Clarifiers GAC Final Ozonation permeate pumps. The permeate system is now connected to the vacuum system to assist in moving the permeate flow through the elevated permeate header pipes. Ferric chloride is added to the secondary effluent in rapid mix tanks. The flow then passes through flocculation tanks to inclined plate settlers. The effluent from the plate settlers can be injected with sodium hydroxide, if desired for ph control, before flowing by gravity to the drum screens (membrane pretreatment strainers). Sludge from the clarifier is blended with other process sludge and sent to thickeners and centrifuges before ultimate off-site disposal. Spent backwash water from the drum screens and the membrane system, as well as spent cleaning solutions, are blended with other plant waste flows and recycled to upstream processes for additional treatment. This blended flow can be recycled to three optional locations: the headworks, secondary clarifier effluent, or influent to the existing chemical solids contact clarifiers. The plant has 160 million gallons (MG) of on site storage (eight 20 MG ground storage tanks) which can receive and pump out flow from multiple sources (for example, raw influent, primary effluent, secondary effluent, and final effluent). This storage provides improved flexibility and reliability to the facility operation as well as flow equalization capability. Table 1 summarizes the design criteria for the full-scale membrane facilities. The design of the system is based on the proposal and proof testing results. 2701

5 Table 1 - Selected Full-Scale Membrane System Design Criteria Ultrafiltration membrane product Membrane pore size rating (micron) Total system feed capacity Total permeate flow rate Number of trains Criteria Value Zenon ZW 500c (immersed hollow fiber, vacuum-driven) 0.04 nominal; 0.1 absolute 50 mgd 48 mgd 16 (14 in service plus two additional) Number of membrane cassettes per train 13 (space for 14) Number of membrane elements (modules) per cassette Membrane surface area per element (module) 250 ft 2 Membrane train feed capacity Maximum instantaneous flux rate Minimum net flux rate mgd 34.0 gfd 30.8 gfd Minimum membrane system recovery 96 % Minimum net membrane system recovery 94 % Maximum TMP (vacuum) Minimum temperature-corrected 20C 12 psi 2.27 gfd/psi Minimum on-line service factor 94 % Backpulse (backwash) Cyclic aeration Maintenance cleans with hypochlorite or citric acid Minimum recovery clean interval Maximum individual train time off-line for hypochlorite or citric acid recovery clean seconds every 15 minutes 10-second on and 10-second off cycle 42-hour frequency 21 days (throughout a 10-year warranty period) 360 min each 2702

6 The design criteria and performance requirements shown above are based upon the mechanically screened membrane feed water shown in Table 2. Table 2 - Membrane Feed Water Criteria Criteria Value (Maximum values except as noted) Turbidity Total Suspended Solids (TSS) Chemical Oxygen Demand (COD) Total Phosphorus Total Iron Dissolved Iron 10 NTU 10 mg/l 50 mg/l 0.1 mg/l as P 2.5 mg/l 0.3 mg/l Temperature Between 20 and 30 C Table 3 summarizes the membrane system permeate water quality design criteria for each individual membrane train. Table 3 - Membrane Permeate Water Criteria Criteria Value (Maximum values except as noted) Turbidity (as measured over a 24-hour period) Less than or equal to 0.1 NTU 95% of the time Turbidity TSS Average Number of Particles Greater than 2 microns per 1 ml of membrane filtered water (as measured over a 24-hour period) 0.8 NTU 0.8 mg/l 10 particles 95% of the time 2703

7 MEMBRANE FACILITY DESIGN CRITERIA The design incorporates a very high level of redundancy for all major components to maximize process reliability. The membrane system itself is divided into 16 process trains, two of which provide redundancy. The design is based upon the philosophy that no piece of equipment, device, or piping header will cause major treatment shutdown. Four US Filter/Zimpro inclined plate settlers, each rated at 33 percent capacity (to provide spare capacity), have been constructed. The design loading rate is 0.35 gpm/ft 2. The plate settlers are covered to minimize algae growth and protect equipment. Figure 3 presents a cut-away isometric view the chemical clarifier building and Figure 4 shows a computer model of the interior and photographs taken during construction. The membrane process building contains two drum screens and 16 membrane process trains and other facilities (Figures 5 and 6). It is constructed with two levels, with a base level having an area of about 42,000 ft 2. Two Brackett Green USA, Inc. automatically backwashing drum screens (strainers), each rated at 100 percent capacity for redundancy, are installed in the membrane process building (Figure 7). The screens are constructed of stainless steel and have 500-micron openings. Plant finished water is used to supply the water jets that automatically clean the screens. The net water recovery of the screens is 99 percent minimum. Figure 3 - Chemical Clarifier Building Isometric Drawing Inclined Plate Settler (typ) Flocculation (typ) Rapid Mix (typ) Sludge Pump Area Inlet Distribution Box Facility has four clarification trains 2704

8 Figure 4 - Chemical Clarifier Building Interior Inclined Plate Settler Area Flocculators Inclined Plate Settler Figure 5 - Membrane Facility Layout Blower & Compressor Room Electrical Room Air Separator (typ) Train 8 Train 16 Control Room Building Mechanical Permeate Pump (typ) Membrane Process Tank (1 of 16) Dip Tank (typ) Train 1 Train 9 Bridge Crane (typ) Permeate Seal Weir Chemical Feed Pump Room (below) Drum Screens Backpulse & CIP System 2705

9 Figure 6 - Membrane Facility Isometric Drawing Membrane Process Trains (16 Total) Air Blowers Control Room Drum Screens Backpulse & CIP Systems Figure 7 - Membrane Pretreatment Drum-Screens Drum Screens (Strainers) 500 Micron Screen 2706

10 The major membrane-related facilities in the building include: 16 membrane process trains, each covered with checker plate with ventilation to outside the building Each train includes a process tank with 13 (space for 14) Zenon 36-module ZW-500c membrane cassettes, an air separator, and a permeate pump Three tanks for backpulse and membrane cleaning with two backpulse and two CIP pumps Two chemical feed systems for sodium hypochlorite and citric acid Six air blowers for membrane process tank aeration Seven vacuum pumps for air removal from the air separators and elevated permeate header piping Two air compressor systems (two compressors for membrane integrity testing air supply and two compressors for pneumatic valves) Two membrane DIP tanks for individual cassette maintenance as needed Membrane pilot (future) area Control room Electrical room Figures 8 and 9 show computer models of the interior of the membrane process building and photographs of the constructed facility. Figure 8 - Membrane Building Interior Drum Screens Area Permeate Pump Area 2707

11 Figure 9 - Membrane Process Trains Vacuum-type UF Process Trains Single UF Train Feed Tank (below cover plates) START-UP AND COMMISSIONING Following calibration of all instruments and completion of functional testing of each group of trains, the original plan was to segment the system into four groups of four trains for performance testing -- over a 60 day period for the first segment, 30 days for the second segment, and 24 days for the last two segments. However, during the testing phase, it was decided to modify the test plan to better meet the needs of the overall operating water reclamation facility. The modified plan included performance testing in three phases and totals about 80 days. One train is being run throughout the three testing periods to obtain longer term cleaning frequency and performance data. The first test phase included operation of four trains. The second phase tested nine trains simultaneously, and the final phase, currently in progress at the time of this writing, involved testing of all 16 trains, although some operate at reduced capacities depending on available membrane system feed flow rate. The Zenon ZW 500c membrane system is operated at the desired flux by adjusting the applied transmembrane pressure (TMP) using a variable speed permeate pump. The target membrane system recovery is attained by using a controlled reject stream flow rate from the process (feed) tank. During the performance testing the trains have instantaneous recovery setpoints of 96%. 2708

12 Cyclic aeration outside the membrane fibers is used to reduce accumulation of solids on the surfaces of the membrane fibers. Currently, one membrane integrity test (MIT) is performed on each operating train per week to confirm membrane and system integrity. MITs are performed by taking the train off-line, pressurizing the membrane permeate header with air, and monitoring the air pressure decay over a preset time period. The loss of more than 1 psi during the pressure-hold time (10 minutes) indicates an integrity failure (such as a membrane fiber break, valve leak, etc.) that requires corrective action. An automatic backpulse (backwash) is performed on a set time interval using permeate in a reverse-flow direction (inside-to-outside of the membrane fibers) to remove solids (see Table 1). Currently the trains are operating with backpulse frequency and duration of 15 minutes and 30 seconds, respectively. Additionally, the system receives automatic tank deconcentrations, automatic maintenance cleans, and semi-automatic (operator initiated) recovery cleans as necessary to maintain performance. Tank deconcentrations involve the complete draining of the membrane process tank for an individual train. Deconcentrations for each operating train are currently being performed after every 24 hrs of production. Maintenance cleans are currently being performed every 42-hour run time. Maintenance cleans involve the use of multiple permeate backpulse steps, with the addition of citric acid or sodium hypochlorite (200 mg/l). The chemical solution is backpulsed through the fibers in a reverseflow direction. Maintenance cleans typically result in the train being offline for approximately 30 minutes and each includes a tank deconcentration to remove solids from the system. Currently two citric acid and two sodium hypochlorite maintenance cleans are being performed per week on all operating trains. The cleaning schedule may be adjusted (in the remaining testing period) if needed, based on influent characteristics. If the influent total iron is greater than 1.5 mg/l or if the permeate total iron is greater than 0.1 mg/l, citric acid is used (instead of the sodium hypochlorite) to prevent potential post-precipitation of iron due to the high ph in the system during a sodium hypochlorite clean. Recovery cleans also involve pumping permeate with cleaning chemicals in a reverse-flow direction through the membrane fibers, however they typically take approximately 5 hours and include a membrane soak period with citric acid (ph 2.5) or sodium hypochlorite (250 mg/l). Recovery cleans also use a tank deconcentration to remove solids. Citric acid and sodium hypochlorite recovery cleans are performed as needed after a minimum of 21 days from the start of the performance test. After the minimal cleaning interval has been achieved, the membrane permeability (flux/psi) and transmembrane pressure (TMP) are used to determine the requirement for a recovery clean. When the membrane permeability is less than 4.5 gfd/psi or the transmembrane pressure is less than negative 8 psi for more then four hours during production, a recovery clean is performed. 2709

13 Typically, the interval between recovery cleans are extended beyond the minimum cleaning frequency based on membrane performance. However, membranes should not be operated at an elevated TMP for extended periods (approximately two to three months when operated at design capacity) as this may have a negative impact on long term membrane performance. If the influent water testing shows elevated total and dissolved iron (greater than 1.5 mg/l and greater than 0.1 mg/l, respectively), a citric acid recovery clean is selected. If iron values are not elevated, and organic fouling is assumed and a sodium hypochlorite recovery clean is performed. This recovery cleaning protocol may be adjusted once more full scale site experience is gained and previous fouling events are better understood. PERFORMANCE RESULTS TO DATE The performance test data presented below is based upon performance testing report submittals prepared by Zenon. The membrane system has performed well, even though it has received feed water quality exceeding the feed water design criteria limits at times. The 60-day performance test on Trains 1 through 4 was conducted in October-December The second phase of performance testing, which included 30 days of operation for Trains 1 and Trains 5 through 8, was conducted in April Water Quality Membrane Feed Water Quality Summaries of the membrane feed water quality during the Phase 1 and 2 performance testing are presented in Tables 4 and 5. The feed water quality during Phase 2 had higher concentrations of the listed parameters and constituents than during Phase 1. Table 4 - Membrane Feed Water Quality During Phase 1 Performance Testing Total Iron (mg/l) COD (mg/l) TP (mg/l) TSS (mg/l) Minimum Maximum Average # of samples Dissolved Iron (mg/l) Table 5 - Membrane Feed Water Quality During Phase 2 Performance Testing Total Iron (mg/l) COD (mg/l) TP (mg/l) TSS (mg/l) Minimum Maximum Average # of samples Dissolved Iron (mg/l) 2710

14 Membrane Permeate Quality The permeate turbidity requirements were met for all trains during Phase 1 and 2 test periods. Figure 10 shows typical permeate turbidity data for one train during Phase 1. Figure 10 Train 2 Permeate Turbidity During Phase 1 Performance Testing The overall membrane system produced average particle counts less the 5/mL for the nearly all of the testing period (see Figure 11). It is believed that some particle count spikes observed are being caused by biological regrowth within the sample lines and the degassing column and that some of the data are not representative of the actual permeate being produced. Sanitizing the particle counter degassing column several times a week indicated lower particle counts. Hydraulic Performance Transmembrane Pressure (TMP) The membrane system TMP was below the contractual maximum of (negative) 12 psi during the performance test period for all trains. As an example, Figure 12 presents the TMP data for Train 6 during Phase 2 performance testing. 2711

15 Figure 11 - Membrane System Permeate Particle Counts During Phase 1 Performance Testing Figure 12 Train 6 TMP During Phase 2 Performance Testing 2712

16 Temperature-Corrected Permeability (TCP) For the majority of the performance test period, with the exception of some fouling incidents that have occurred when feed water quality had excursions above the design criteria values (such as an incident which resulted in ferric chloride being significantly overdosed in the membrane pretreatment chemical clarification process), the 20-degree centigrade TCP was relatively stable and ranged between 7 and 9 gfd/psi. As an example, Figure 13 presents Train 6 TCP data before (lower data set) and after backpulsing (middle data set) during Phase 2 performance testing. The upper data set indicates TCP during backpulsing, which is used as a diagnostic indicator of foulants on the membranes. Figure 13 Train 6 Temperature-Corrected Permeability (TCP) During Phase 2 Performance Testing Currently, sodium hypochlorite recovery cleans are being performed to remove organic foulants from the membranes using a 5 hour soak in approximately 250 mg/l. Citric acid recovery cleans are conducted to remove inorganic foulants. Citric acid recovery cleans are typically performed at ph of approximately 2.5 and include several-hour soak times. Figure 14 shows a membrane cassette before and after a citric acid recovery clean during Phase 1 testing. The dark staining on the top of the membrane fiber indicated inorganic fouling. 2713

17 During the 30-day Phase 2 performance test period no recovery cleans were performed. This interval exceeded the contractual minimum requirement of 21 days. Figure 14 - Membrane Cassette Before and After Citric Acid Recovery Cleaning Throughout the performance test period weekly membrane integrity tests (MIT s) were performed. Trains are required to meet the criteria of pressure decay less then 1 psi/10 minutes during the MIT. The trains passed all MIT s performed during the test period. 2714

18 CONCLUSION The 50 mgd (feed flow) Zenon ultrafiltration system treating chemically-clarified secondary effluent has been constructed. At the time of this writing (June 2006), all 16 membrane trains have been started-up and the last phase of a three-phase performance acceptance test program is in progress. The membrane facility is expected to be fully operational and accepted by the owner, Gwinnett County, in ACKNOWLEDGEMENTS The authors would like to acknowledge Zenon Environmental Inc. for their involvement in the project as suppliers of the membrane system and the use of selected data and figures in this manuscript derived from two performance test reports they prepared dated January and May REFERENCES 1. Bergman, Robert A., Richard Porter and Scott Levesque. Evaluation of Dual-Membrane Treatment for Large-Scale Water Reclamation at Gwinnett County, Georgia. Proceedings of the American Water Works Association 2001 Membrane Technology Conference. San Antonio, Texas. March 4-7, Bergman, Robert A., Richard Porter, Tammy Watson, and Don Joffe. Advanced Wastewater Treatment Using State-of-the-Art Membrane Processes at Gwinnett County, Georgia. Proceedings of WEFTEC 2001, Water Environment Federation 74 th Annual Conference and Exposition. Atlanta, Georgia. October 13-17, Bergman, Robert A., Richard Porter, Brad Schulgen, and Joseph Elarde. Membrane Pilot and Demonstration-Scale Treatment for Water Reclamation at Gwinnett County, Georgia. Proceedings 2002 AWWA Water Sources Conference and Exhibition. Las Vegas, NV. January 27-30, Bergman, Robert A. and Richard Porter. Selection and Design of 40 mgd Membrane Filtration for Surface Water Augmentation. Proceedings 2002 AWWA Annual Conference. New Orleans, LA. June 16-20, Bergman, Robert A., Don Joffe, Nick Adams, and Richard Porter. Gwinnett County Water Reclamation with 50 mgd Ultrafiltration Proof Testing and Design. Proceedings of the American Water Works Association 2003 Membrane Technology Conference. Atlanta, Georgia. March 2-5, Bergman, Robert A., Don Joffe, Richard Porter, and Ed Minchew. Large Capacity 190,000 m3/day Ultrafiltration Facility for Reclamation in the Southeastern USA. Proceedings of the International Desalination Association 2006 World Congress on Desalination and Reuse. Singapore. September 11-16,