Membrane Applications 2010

Application of Nanofiltration and Reverse Osmosis Membranes for Produced Water Treatment
Arun Subramani, Ron Schlicher, Jim Long, Jack Yu, Geno Lehman, Joseph Jacangelo MWH Americas Inc. 618 Michillinda Avenue, Suite 200 Arcadia, California 91007 ABSTRACT Produced water is a term used to describe water that is obtained along with oil and gas production. Produced water constitutes the single largest waste stream from oil and gas exploration and production activities and contains high levels of oil and grease, total dissolved solids (usually sodium chloride), hydrocarbons, and refractory organics. If treated appropriately, produced water can be employed as a true water resource to augment existing surface water streams and creeks. Due to stringent surface discharge limits being imposed in the United States, produced water needs to be managed and treated before being discharged to surface water streams and creeks. Certain discharge limits require a chloride concentration of less than 230 mg/L in the treated water. Treatment of such wastewater streams to meet low chloride, selenium, and boron discharge limits requires a technology, such as nanofiltration (NF) and reverse osmosis (RO), which can serve as an absolute barrier for various contaminants. In this study, different types and configurations of NF and RO processes were pilot tested to determine their applicability in treating produced water obtained from natural gas wells at a location in the western United States. In order to reduce the fouling potential on NF and RO membranes, dissolved air floatation (DAF), ceramic ultrafiltration (UF), MYCELX cartridges, and organoclay filters were tested as pretreatment alternatives. It was determined that the fouling potential of NF and RO membrane was not substantially different for the various pretreatment processes utilized. In order to handle high silica concentrations in the feed water and increase the overall feed water recovery, a two pass NF-RO system was tested. The first pass NF system was used to remove hardness and alkalinity from the feed water. The pH of permeate from the first pass NF system was increased to 10.0 to increase silica solubility and used as feed to a second pass seawater RO system. A combination of spiral wound and disc tube configuration was effective in achieving more than 90 percent recovery for the first pass NF membranes while an overall feed water recovery of more than 70 percent was achieved for the entire NF-RO membrane system and also resulted in meeting the discharge limits.

KEYWORDS Oily wastewater, silica polymerization, organic fouling, recovery optimization.

Copyright 2010 Water Environment Federation. All Rights Reserved.

491

Membrane Applications 2010

INTRODUCTION Produced water, water that is co-produced during oil and gas extraction, represents the largest source of oily wastewaters (Mueller et al., 1997). The volume of produced water can be as much as ten times the volume of oil extracted (Mondal and Wickramasinghe, 2008). Produced water consists of a combination of organic and inorganic compounds and production chemicals. Typical organic compounds present are aliphatic, aromatic, and polar compounds. Inorganic components include sodium, potassium, calcium, magnesium, chloride, sulfate, carbonate, silicates, and borates. Production chemicals can include emulsion breakers to improve separation of oil and water and corrosion inhibitors (Mondal and Wickramasinghe, 2008). The concentration of these contaminants can vary significantly due to natural variation in the geological formation and the type of oil-based product being produced (Franks et al., 2009). The typical method of dealing with produced water is deep well injection. A portion of the produced water is reinjected into oil producing zones to improve oil recovery through water or steam flooding (Visvanathan et al., 2000). The other portion of produced water is disposed off through deep well injection. Deep well injection is limited by the capacity of the injection wells and discharge limits set by local governing agencies. Due to this limitation in discharging the produced water, oil production companies are looking into options for treating the produced water for surface discharge. Example of discharge limits based on the class and purpose is listed in Table 1. Certain discharge limits (Class 2AB) require a chloride concentration of less than 230 mg/L in the treated water. Treatment of such wastewater streams to meet low chloride and boron discharge limits requires a technology, such as NF and RO, which can serve as an absolute barrier for various contaminants. The use of membrane technology also offers other advantages such as a smaller footprint, high level of automation, and applicable for both onshore and offshore oil exploration (Mondal and Wickramasinghe, 2008). Table 1. Example of surface water discharge limits for produced water.
Class 3D 3B Purpose Beneficial use for livestock watering Aquatic life, ephemeral streams Discharge limit TDS < 5000 mg/L, Oil and grease < 10 mg/L, 6.5 < pH < 9, Turbidity < 10 NTU, Chloride < 2000 mg/L, Boron < 5 mg/L TDS < 5000 mg/L, Oil and Grease < 10 mg/L, 6.5 < pH < 9, Turbidity < 10 NTU, BOD5 < 30 mg/L, TSS < 30 mg/L, Cadmium < 0.54 ug/L, Selenium < 5 ug/L, Sulfur < 2 ug/L, Chloride < 2000 mg/L, Boron < 5 mg/L TDS < 5000 mg/L, Oil and Grease < 10 mg/L, 6.5 < pH < 9, Turbidity < 10 NTU, BOD5 < 30 mg/L, TSS < 30 mg/L, Cadmium < 0.54 ug/L, Selenium < 5 ug/L, Sulfur < 2 ug/L, Chloride < 230 mg/L, Boron < 5 mg/L, Benzene < 2.2 ug/L, Toluene < 1 mg/L, Ethylbenzene <0.53 mg/L

2AB

Potable water, preserving fish life

OBJECTIVES Although high-pressure membrane processes such as NF and RO have been used in the past for produced water treatment (Bartels and Dyke, 1990; Tao et al., 1993; Mohammadi, et al., 2003; Franks et al., 2009), limited studies are available which have looked into the application of advanced pretreatment technologies for controlling fouling and recovery optimization for NF and

Copyright 2010 Water Environment Federation. All Rights Reserved.

492

Membrane Applications 2010

RO processes. In this study, pilot-scale evaluation of pretreatment using ceramic ultrafiltration, MYCELX filtration, DAF, and organoclay filtration were performed and recovery optimization of NF and RO process was conducted for produced water obtained from natural gas wells at a location in the Western United States. Given the range of discharge standards, the pilot system was designed to provide recommendations on meeting the most stringent case (which includes a chloride limit of 230 mg/L). Specific objectives of the study were as follows: Assess the applicability and performance of DAF, organoclay filtration, MYCELX filtration, and ceramic ultrafiltration as pretreatment ahead of NF and RO membrane system to minimize fouling. Evaluate the performance of NF and RO membranes with different operating configurations (i.e., double pass systems and multistage systems) to maximize water recovery. MATERIALS AND METHODS Three different treatment schemes were tested at the pilot-scale. A schematic of the treatment scheme is shown in Figure 1. The three different treatment schemes were tested one at a time. The first pretreatment scheme was DAF Ceramic UF. The second pretreatment scheme was DAF Organoclay. The third pretreatment scheme was DAF MYCELX filtration.
Permeate

UF 1 Spiral NF 2 DAF Organoclay 3 DT NF

Concentrate

Treated Water Caustic

Spiral RO

MYCELX Figure 1. Treatment schemes used at the pilot-scale.

Reject Water

Copyright 2010 Water Environment Federation. All Rights Reserved.

493

Membrane Applications 2010

Dissolved Air Floatation Produced water from natural gas wells were first treated in gun barrel tanks for oil and water separation. After the gun barrel tanks, the produced water was fed to a DAF system (VanAire Inc.). The DAF system was used for further reduction of oil and grease and turbidity of the produced water. A hydraulic retention time of 60 to 90 minutes was used in the DAF. A 50% recycle of feed and an air flow of 1 standard cubic feet per hour were used during testing. To enhance flocculation and settling, 1 to 3 mg/L of polymer and 50 to 100 mg/L of aluminum chlorohydrate (Baker Petrolite Chemicals) was tested in the DAF system. Ceramic Ultrafiltration The first prettreatment scheme used was ceramic ultrafiltration. A ceramic UF system (Membralox Unit X15) was leased from PALL Water Processing (New York, U.S.A.). A schematic of the ceramic UF membrane element and structure is shown in Figure 2. The UF system consisted of seven ceramic elements with a total membrane area of 2.5 m2. A 50 nm alumina-zirconia type of ceramic filtration membrane was used for the study. Ceramic membranes have several advantages over polymeric membranes such as: higher flux (up to 5 times higher), higher operating temperature (up to 95 0C) and pressure (up to 150 psi), longer membrane life (up to 5 years), resistance to harsh chemical cleaning, and full pH compatibility. A range of crossflow velocities (3.0 to 4.5 m/sec), transmembrane fluxes (100 to 150 gfd), feed water recoveries (75 to 90 percent), and backpulse intervals (3 to 5 minutes) were evaluated.

Figure 2. Schematic of ceramic membrane module and filtration layer (Source: PALL Corporation).

Copyright 2010 Water Environment Federation. All Rights Reserved.

494

Membrane Applications 2010

Organoclay Filtration The second pretreatment scheme utilized was organoclay filtration. Organoclay was obtained from Aquatech Inc., Wyoming. Organoclay are chemically altered volcanic ash consisting of montmorillonite. They have a surface area of approximately 750 m2/gm and used for the removal of oil and grease, turbidity, metals, and solvents. An organoclay filtration bed with an empty bed contact time (EBCT) of 15 minutes was used to the study. Backwashing was performed when the pressure drop across the bed exceeded 30 psi. MYCELX Filtration The third pretreatment scheme utilized was MYCELX filters obtained from Mycelx Technologies Inc., Georgia. MYCELX technology was used as an alternative to organoclay filtration for the removal of oil and grease. MYCELX filters are considered to be effective in the removal of hydrocarbons, oil sheen, synthetic oil, and natural oil. In this study, 25 micron filters arranged in series were used. Effluent from the DAF was used as the feed to the MYCELX filters. The filters were coated with polymeric surfactant technology to enable the removal oil droplets. NF and RO Membranes High-pressure membranes were used for reduction of TDS and target constituents such as chloride, boron, and selenium. Two types of NF and one type of RO membrane was used. The NF membranes used were NF270 (DowFilmtec) and NF90 (DowFilmtec). The RO membrane used was TM810L (Toray). The NF270 membrane is considered to be a loose nanofiltration membrane with 40-60% rejection of CaCl2 and less than 35% rejection of NaCl. In contrast, the NF90 membrane is considered a tight nanofiltration membrane and has more than 85% rejection of NaCl. The TM810L membrane is used for seawater desalination and is a cross linker fully aromatic polyamide composite membrane and has more than 99.75% rejection of NaCl. Disc Tube Technology A DT system was leased from PALL Water Processing (New York, U.S.A.). An image of the DT system module is shown in Figure 3. The DT system consists of commercial flat sheet membranes installed in a plate and frame configuration. The length of the module is 1 m (3.3 ft) and the membrane area in the module is 7.65 m2 (82.3 ft2). The module consists of a fiber glass housing and can withstand pressures up to 1000 psi. The module consists of unique crossflow construction with stacked membrane discs. The disc membrane stack was housed in an 8-inch (diameter) pressure vessel and assembled on a center tension rod using stainless steel end flanges. The extremely short feed water path across the membrane surface, followed by a 180o flow reversal greatly reduces concentration polarization on the membrane surface, reducing fouling and scaling potential. The DT module is capable of operating at high particulate loading (2500 mg/L). The system was operated at various recoveries (50 - 85%). Flux was maintained constant (18-20 gfd) for all the tests. To minimize the scaling potential due to sparingly soluble salts, 8 mg/L of MDC150 (# 3617, Eastern Technologies Inc.) antiscalant was added to the system.

Copyright 2010 Water Environment Federation. All Rights Reserved.

495

Membrane Applications 2010

Figure 3. Illustration of DT module (Source: PALL Corporation). Spiral Wound Technology A spiral wound (SW) RO skid was leased from PALL Water Processing (New York, U.S.A.) and used for pilot testing. In a SW system the membrane is wound on a central tube in a spiral configuration. Membrane leaves are separated by a feed spacer used to promote turbulence and reduce concentration polarization. A permeate spacer transports the permeate water to the center tube. The concentrate stream leaves the element to be further treated by another SW membrane element or discharged. The SW system consisted of 8 pressure vessels which could be configured as 8 SW elements connected in series. Each pressure vessel accommodated a SW membrane element with nominal dimensions of 4 x 40. The system was also configured as 4 SW elements connected in series with a fraction of the concentrate stream recycled back to the feed of the first SW element. For the first pass testing, NF270 membranes were installed in the SW system. For the second pass testing, TM810L membranes were used. The system was operated at various recoveries (50 - 75% for first pass and 50 80 % for second pass). Flux was maintained constant (less than 10 gfd) for all the tests. To minimize the scaling potential due to sparingly soluble salts, 8 mg/L of MDC150 (# 3617, Eastern Technologies Inc.) antiscalant was added to the system for the first pass. Before the second pass tests, pH of the water was raised to 10.0 by addition of sodium hydroxide to increase the solubility of silica (Sheikholeslami and Tan, 1999). To further minimize silica scaling on the second pass membranes, an antiscalant specific to silica (Formula 3680, Eastern Technologies Inc.) with a dosage of 8 mg/L was used. Chemical cleaning of the membrane systems were performed with cleaning chemicals provided by PALL Water Processing (New York, U.S.A.). Low pH cleaning solution used was RO Cleaner C which was composed of 10 30% citric acid. The high pH cleaning solution used was RO Cleaner A which was composed of 0 -5% sodium hydroxide and 0 5% ethylene diamine tetra acetic acid. Cleaning was performed with a low pH solution (pH~2.0) followed by use of high pH solution (pH ~11.0). Each cleaning cycle was performed for a period of 2 hours at with clean water flush in between cleaning cycles. For high pH cleaning of NF membranes, special cleaning solution (Cleaner NFE) was used. High pH cleaning for NF membranes was performed

Copyright 2010 Water Environment Federation. All Rights Reserved.

496

Membrane Applications 2010

at pH ~9.0. All chemical cleanings in the DT system was performed at room temperature. Chemical cleaning in the SW system was performed at 40 deg C. RESULTS Feed Water Quality Average feed water quality to the pilot plant is listed in Table 2. The contaminants included suspended solids, oil and grease, dissolved organics, volatile organic compounds, metals, soluble salt. The pH of the produced water was approximately 7.5. The concentration of gasoline range organics (GRO) was substantially higher than diesel range organics (DRO). The 5-day biological oxygen demand (BOD5) and chemical oxygen demand (COD) were also high in the feed water. The temperature of the produced water was as high as 80 0C. Since the operation of NF and RO membranes is limited to less than 40 0C, a heat exchanger was installed to cool the influent feed water to the pilot plant. The concentration of silica varied between 60 and 100 mg/L in the feed water. Operating the RO process at 50 % recovery with a feed water silica concentration of 100 mg/L would result in 200 mg/L of silica in the concentrate stream (concentration factor of 2) exceeding the solubility limit of silica. Table 2. Average feed water quality to the pilot plant.
Parameter Concentration (mg/L) Oil and Grease 125 DRO 4 GRO 113 BOD-5 771 COD 1470 TDS 6280 pH 7.5 Alkalinity, as CaCO 3 2690 Sodium 3132 Chloride 1776 Calcium 31 Magnesium 3 Sulfate 8 Barium 5.8 Iron 0.8 Selenium 0.1 Silica 60 - 100 Methanol 280 Benzene 20 Toluene 30 Ethylbenzene 1 Xylene 1

The presence of suspended solids, oil, and grease in the feed water can severely hinder the performance of NF and RO process due to fouling. Thus, several pretreatment techniques were

Copyright 2010 Water Environment Federation. All Rights Reserved.

497

Membrane Applications 2010

evaluated for the removal of suspended solids and oil and grease. The performance of different pretreatment processes evaluated is described below. Performance of DAF For enhanced flocculation and settling of suspended particles, hydrocarbons, and oil and grease, chemicals were added to the feed water of the DAF. Optimum chemical dosage was determined through jar testing. During pilot testing, turbidity removal between 60 and 90% was achieved consistently in the DAF system. The percentage removal of oil and grease varied between 25 and 90%. This is attributed to significant variation in the oil and grease content of the feed water. An increase in the chemical dosage to the DAF was necessary to achieve higher turbidity removal. Although automated chemical injection system with feed-back control was not tested at the pilot scale, flow-paced chemical injection is recommended to automatically vary the dosage of chemicals in order to deal with the feed water quality changes. Overall, the DAF system was efficient as the first stage of pretreatment in reducing the concentration of oil and grease and suspended solids. Thus, for all further testing, the DAF system was used as the default pretreatment step followed by either organoclay (or) MYCELX filters (or) ceramic UF. Performance of Organoclay Filtration Organoclay filters have been used commercially for many years to remove oil from wastewaters. Column bench-scale testing results conducted using the clay showed good removal of oil and grease as long as adequate contact time was allowed. However, the bench-scale tests were not able to adequately determine the capacity of the organoclay for the oil and grease constituents during pilot-scale testing. Therefore, a unit was pilot tested. For the pilot test, DAF effluent was fed to the organoclay filter which operated as a down flow pressure filter. As solids built up in the organoclay media, the differential pressure increased and the unit was backwashed. When the system was operated at a high hydraulic loading rate, backwashing was necessary almost every day. Later in the pilot study, the unit was operated at a much lower hydraulic loading rate (1 gpm/sq ft) which alleviated some of the operational problems. But, the system achieved only 30 to 45% of oil and grease removal based on the field measurements. During the initial period when the filter was hydraulically overloaded, the EBCT, which is a measure of the time that contaminants have to adsorb to the media, was only 3 to 5 minutes. Even after increasing the EBCT to 15 minutes, the oil and grease removal did not improve. Overall, the relatively poor performance of the organoclay filter may be due to the fact that the constituents being measured as oil and grease were not really oil and were dissolved or liquefied at the elevated temperature of the water. Performance of MYCELX Filtration MYCELX filters were evaluated as an alternative to the organoclay filter for oil and grease removal. Bench-scale results indicated that the filters can effectively remove oil and grease at the ambient temperature of the laboratory. The test showed an oil and grease reduction from 74 mg/L down to 9 mg/L after passing through two filters in series and down to 5 mg/L after passing through a third filter. Based on these results, MYCELX cartridge filters and snippets were shipped to the site for pilot testing. The arrangement for testing these filters at the pilot site

Copyright 2010 Water Environment Federation. All Rights Reserved.

498

Membrane Applications 2010

involved routing the DAF effluent through a bag filter filled with snippets and then through two 25 micron MYCELX cartridge filters in series. Even with the DAF achieving good turbidity removal, the first MYCELX cartridge filter plugged within one day of operation. In terms of performance for oil and grease removal, the MYCELX filter did not perform as well as expected based from the bench-scale testing. Only 18% removal of oil and grease from field measurements was achieved. It is also possible that the filters might perform better after the water was cooled down and the oil and grease material precipitates out of solution. The bench testing was conducted at 65F whereas the temperature of the water during the pilot operation was around 83F. Based on the limited testing, the use of MYCELX filters was recommended only as a polishing step. Performance of Ceramic UF As with the organoclay and MYCELX filters, oil and grease removal through the ceramic UF was much lower than expected, averaging only 25% removal based on field measured oil and grease measurements. The low oil and grease reduction by the ceramic UF was likely associated with the dissolved or liquefied organics in the water because solids and oil particles should not pass through the small 50 nm pores of the ceramic UF. Performance of the ceramic UF is shown in Figure 4. During startup a water recovery of 80% was achievable without observing any decline in the water flux. For the first 100 hours of operation, no substantial decrease in the specific flux was observed. After the initial operation period, the specific flux decreased steeply requiring chemical cleaning of the membrane to restore the specific flux to initial values. Recoveries greater than 80% lead to steep decline in the specific flux. A flux decline of about 14% was observed during 550 hours of operation. A three step cleaning of the ceramic UF involved bleach and caustic soda in the first step, an alkaline cleaner in the second step, and nitric acid in the third step. Cleaning at elevated temperature (140F) was found to be more effective than cleaning at ambient temperature.

Copyright 2010 Water Environment Federation. All Rights Reserved.

499

Membrane Applications 2010

14.0

Ceramic UF
Specific flux @ 25 deg C Recovery Flux @ 25 deg C Chemical Cleaning

240 220 200 180

12.0

Specific Flux @ 25 deg C gfd/psi

8.0 160 6.0 140 120 4.0 100 2.0 80 0.0 0 100 200 300 400 500 600 700 60

Time of Operation, hr

Figure 4. Performance of ceramic UF. Limiting Recovery of RO Process To determine the maximum recovery possible in a single pass, tests were conducted using a tight NF membrane (NF90). The NF90 membrane is capable of achieving more than 85% rejection of salts such as NaCl. Hence, the NF90 membrane installed in the DT system was tested to determine if the discharge limits can be met in a single pass and also determine the maximum achievable recovery and fouling potential. Since the fouling potential of pretreated water was no known, the DT system was utilized for estimating the limiting recovery. As mentioned earlier, the DT system is capable of handling fouling prone feed water. The temperature corrected flux, recovery, and temperature corrected specific flux for DT system with NF90 membranes is shown in Figure 5. Pretreatment consisted of DAF and ceramic UF membranes. The recovery of the system was increased from an initial value of 45% up to a final recovery of about 65%. The operational flux was maintained constant at 20 gfd. At a recovery of less than 50%, no substantial decrease in specific flux was observed. But, as the recovery was increased further (more than 55%) a gradual decrease in the specific flux was observed. At a recovery of 65%, the specific flux was found to be 0.060 gfd/psi compared to the initial specific flux of 0.085 gfd/psi. Hence, within 120 hours of operation a decrease of more than 25% in the specific flux was observed suggesting that fouling/scaling of the membrane occurred leading to decrease in performance. During the entire duration of operation no substantial increase in the differential pressure across the module was observed. At the end of operation, chemical cleaning was performed to determine if the specific flux could be recovered to initial values. Even after

Copyright 2010 Water Environment Federation. All Rights Reserved.

500

Recovery, % Flux @ 25 deg C, gfd

10.0

Membrane Applications 2010

cleaning with both low pH and high pH solutions, the specific flux was not recovered to initial values suggesting irreversible fouling/scaling has occurred on the membrane surface.
DT-NF90
100.0 90.0 80.0 1

Flux @ 25 deg C, gfd Recovery, %

70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 0.00

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 20.00 40.00 60.00 80.00 100.00 120.00 140.00

Time, hr

Figure 5. Performance of NF90 membrane. Silica rejection was more than 90% by the NF90 membrane. During the testing of NF90, silica concentrations in the feed were between 90-100 mg/L. Hence, with more than 90% rejection of silica by the membrane, the concentration of silica in the concentrate stream of the NF90 membrane would be over 245 mg/L (concentration factor of 2.72 at a recovery of 65%) exceeding the silica solubility limits. Although a silica specific antiscalant at a dosage of 8 mg/L was used as pretreatment for the NF90 membrane the irrecoverable specific flux after chemical cleaning suggested that a hard silica scale had formed on the membrane surface. Also, antiscalants are not efficient when silica concentrations exceed more than 200 mg/L in the concentrate stream of membrane systems. Hence, due to high silica concentration in the feed water the overall feed water recovery of the NF90 membrane system was restricted to less than 65%. Since silica scaling of the membrane was restricting the overall feed water recovery of the membrane process, further membrane tests were conducted to increase the silica solubility limit in the feed water by increasing the pH of the solution. In order to increase the pH of the solution hardness related to calcium needed to be reduced to prevent the precipitation of calcium carbonate (CaCO3). Hence, a two pass membrane system was implemented. In the two pass system, the permeate from the first pass is used as feed to the second pass membrane. In the first pass a loose NF membrane (NF270) was employed to remove hardness and alkalinity so that pH can be raised to 10.0 in the second pass utilizing a RO membrane.

Copyright 2010 Water Environment Federation. All Rights Reserved.

501

Specific Flux @ 25 deg C, gfd/psi

Recovery Chemical Cleaning Flux @ 25 deg C Specific Flux @ 25 deg C

0.9 0.8

Membrane Applications 2010

Performance of First Pass NF Membranes For the first pass, NF270 membranes installed in SW configuration were evaluated. The temperature corrected flux, recovery, and temperature corrected specific flux for spiral wound system with NF270 membranes used for the first pass is shown in Figure 6. The performance parameters for the tests are listed in Table 3. The system was operated with different pretreatment schemes. From 0 90 hours (region 1 in the plot), the SW-NF270 membrane system was operated with pretreated water obtained from the DAF Ceramic UF system. From 90 135 hours (region 2), the system was operated with pretreated water obtained from the DAF Organoclay filter. From 135 157 hours (region 3) the system was operated with pretreated water obtained from the DAF MYCELX cartridges. From 157 204 hours (region 4), the system was operated with pretreated water from DAF Organoclay filter. For the first 140 hours of operation the feed water recovery was maintained at 65% and hence the fouling potential of different pretreatments (ceramic UF, MYCELX, and Organoclay) on the SW-NF270 membrane could be compared. Although the specific flux seemed to fluctuate within the first 140 hours of operation there was no substantial difference in the rate of decrease in the specific flux for the membrane when operated with three different pretreated waters. Hence, it is not possible to conclusively determine the best pretreatment process for the spiral NF270 membrane with respect to fouling. After the initial tests, feed water after DAF Organoclay filter was used to obtain performance data at higher recoveries. Hence, the recovery of the system was increased to 70% and 75% from 140 200 hours of operation. When the recovery was increased, a gradual decrease in the specific flux was observed. Within 60 hours of operation at recovery greater than 70% the specific flux decreased from 0.15 gfd/psi to 0.13 gfd/psi signifying a 13% decrease. Hence, membrane fouling occured when the recovery of the system was increased higher than 70%. During the entire duration of operation the differential pressure across the module increased from 20 psi to 27 psi. No chemical cleaning cycles were performed for the SW-NF270 membrane system.

Copyright 2010 Water Environment Federation. All Rights Reserved.

502

Membrane Applications 2010

SW-NF270
100.0 90.0 80.0 70.0 0.3

Flux @ 25 deg C Recovery Specific Flux @ 25 deg C

60.0 1 50.0 40.0 30.0 0.15 20.0 10.0 0.0 0.0 50.0 100.0 150.0 200.0 0.1 250.0 2 3 4 0.2

Time, hr Figure 6. Performance of SW-NF270 membrane.

Note: 1: Pretreatment is DAF Ceramic UF; 2: Pretreatment is DAF OrganoClay; 3: Pretreatment is DAF MYCELX; 4: Pretreatment is DAF OrganoClay

Table 3. Performance parameters for SW-NF270 tests.

Te st 1 2 3 4 Spe cific Flux (Start) Specific Flux (End) Net Operating Pressure (Start) Ne t Ope rating Pressure (End) gfd/psi @ 25 de g C gfd/psi @ 25 deg C psi psi 0.24 0.26 26 29 0.22 0.24 32 28 0.38 0.39 32 32 0.35 0.3 22 32

To determine the rejection capability of the first pass SW-NF270 membrane and its applicability as first pass membrane system, water quality parameters were monitored during operation. The rejection of BOD-5, COD, TDS, and alkalinity was low. The rejection of BOD-5 and COD was approximately 35% and the rejection of TDS and alkalinity was less than 15%. Among the individual ion rejections, the rejection of calcium was approximately 10% and rejection of magnesium was more than 40%. The rejection of chloride was negligible. The rejection of sodium was also low (less than 15%). The rejection of sulfate was greater than 85% and the Ammonia-N was rejected less than 5%. The rejection of barium was about 30% where as

Copyright 2010 Water Environment Federation. All Rights Reserved.

503

Specific Flux @ 25 deg C, gfd/psi

0.25

Flux @ 25 deg C, gfd Recovery, %

Membrane Applications 2010

selenium rejection was approximately 5%. Rejection of boron and selenium were low (less than 10%). In order to increase the feed water recovery of the first pass, a DT-NF270 system was used to treat the concentrate obtained from the SW-NF270. The system was operated for a total duration of approximately 100 hours after. The recovery of the system was maintained at about 71%. The flux was maintained constant at about 10 gfd. The initial specific flux of the system was determined to be ~ 0.52 gfd/psi. The specific flux decreased to 0.26 gfd/psi, a 50% decrease in 24 hours. Although the specific flux decreased substantially within the first 24 hours of operation, it was recovered to initial value by flushing only with RO permeate. The recovery of specific flux suggested that the foulant layer deposited on the membrane surface was not irreversibly adhered and was loosely deposited on the membrane surface. During the entire duration of operation the differential pressure across the module was constant and did not increase. No chemical cleaning cycles were performed for the DT-NF270 membrane system while treating the SW-NF270 concentrate stream, only flush with RO permeate was performed. Performance of Second Pass RO Membranes The second pass membrane system consisted of a seawater RO membrane, TM810L. The second pass membrane system consisted of the spiral wound configuration operated at a recovery of 70 80%. The operating flux was maintained constant at less than 10 gfd. The pH of the feed water was increased to 10.0 for the second pass membrane operation to increase the solubility of silica. Temperature corrected flux, recovery, and temperature corrected specific flux for SW system with TM810L membranes used for the second pass is shown in Figure 7. The system was operated in 4 different batches since the permeate water from the first pass (SW-NF270) membrane needed to be stored and pH adjusted. The four batches of operation presented in Figure 7 are from 0-29 hours, 29-40 hours, 40-80 hours, and 80-114 hours. For all the tests, on average the specific flux decreased by more than 65%. The feed pressure increased from about 490 psi to more than 700 psi for a majority of the tests. The net operating pressure increased significantly for tests 1 and 4. The differential pressure increased from 16 psi to 27 psi for the entire duration of the tests. A summary of the performance parameters for the tests is listed in Table 4. The significant increase in feed pressure requirement, increase in net operating pressure, decrease in specific flux, and, increase in differential pressure suggested significant fouling and scaling of the membranes. The membranes were cleaned after 80 hours of operation using only high pH (12.0) cleaning solution. The specific flux recovered to initial value after chemical cleaning suggesting that cleaning was efficient.

Copyright 2010 Water Environment Federation. All Rights Reserved.

504

Membrane Applications 2010

SW-TM810L
100.0 90.0 80.0 70.0

Flux @ 25 deg C Recovery Specific Flux @ 25 deg C

0.30

Flux @ 25 deg C, gfd Recovery, %

0.20

60.0 50.0 40.0 30.0 20.0 10.0 0.0 0.0 20.0 40.0 60.0 80.0 100.0 0.00 120.0 0.10

2 3 4

Time, hr Figure 7. Performance of SW-TM810L membrane. Regions 1,2,3, and 4 represent different tests performed with varying recovery.

Table 4. Performance parameters for SW-TM810L tests.

Te st 1 2 3 4 Spe cific Flux (Start) Specific Flux (End) Net Operating Pressure (Start) Ne t Ope rating Pressure (End) gfd/psi @ 25 de g C gfd/psi @ 25 deg C psi psi 0.08 0.05 115 225 0.14 0.05 412 413 0.12 0.02 78 65 0.09 0.02 104 382

Note: Test 2 was performed with 8 elements without recycle.

The rejection of TDS and alkalinity was more than 95%. The rejection of BOD-5 was 72% and the rejection of COD was 63%. Since the TM810L membranes are RO membranes with high rejection capability, TDS and alkalinity rejection was high and easily met discharge limits. The rejection of B0D-5 was low as it was associated with the passage of methanol through the RO membrane. The rejection of methanol is low through a RO membrane due to its low molecular weight (32.04 gm/mol). All ions were rejected greater than 95 % except for ammonia-N and boron. The rejection of Ammonia-N was less than 25% and boron rejection was approximately 75%. Ammonia rejection by RO membranes is a function of pH. The feed pH was 10.0 (0.1) for the tests with SW-TM810L membranes. Lower pH leads to the formation of ammonium (NH4+) ions (higher rejection). Beyond a pH of 9.5, the dominant form of nitrogen compounds is ammonia, which is an uncharged molecule and difficult to reject by RO membranes hence the rejection of Ammonia-N is low (Yoon and Lueptow, 2005). Among the metals, selenium rejection was approximately 73%. The rejection of boron (75%) was better when compared to

Copyright 2010 Water Environment Federation. All Rights Reserved.

505

Specific Flux @ 25 deg C, gfd/psi

Membrane Applications 2010

the SW-NF270 membrane since boron speciates into the borate anion form at high pH and results in better rejection by the SW-TM810L membranes. Determination of the constituents in the produced water limiting the performance of the membrane process was critical to optimize the treatment process and achieve higher recoveries. Autopsy of membrane elements removed from the first pass SW-NF270 and second pass SWTM810L were performed to determine the nature of foulant and scalant deposited on the membrane surface. Determination of Organic and Inorganic Content of Foulant Layer A loss on ignition (LOI) test was performed to determine the organic content of the foulant material deposited on the membrane surface. A LOI value in excess of 35% represents the presence of significant organic content of the foulant layer. A LOI value of 85% was found for the SW-NF270 membrane suggesting the presence of significant organic content deposited on the membrane surface. The LOI value for the SW-TM810L membrane was 22%, suggesting low concentrations of organic matter deposited on the membrane. Since the SW-NF270 membrane was used for the first pass, organic matter not removed from the pretreatment processes reached the membrane surface and eventually deposited. Since the organics are removed by the SWNF270 membrane and the permeate was used as feed to the second pass SW-TM810L membrane, deposition of organic matter was restricted on the second pass membrane. A Fourier Transform-Infrared (FT-IR) analysis performed on the two membrane elements indicated the presence of carbohydrates, polysaccharide-like, and protein-like material found on both the membrane surfaces. Also, gram-positive and gram-negative bacteria were identified on both SWNF270 and SW-TM810L membranes. The presence of bacteria on the membrane surface can lead the biofouling issues eventually and cause a decrease in the performance of the membrane and auxiliary equipments. To determine the inorganic constituents of the deposit layer, Energy Dispersive X-ray Analysis (EDX) was performed in combination with the SEM. Since the SW-NF270 membrane was fouled by predominantly organic matter (from LOI results) the weight percent of inorganic elements such as silicon was low (less than 4%) suggesting the presence of only a small amount of clay on the membrane. For the second pass membrane, a high silicon content (24%) was found on the SW-TM810L membrane which suggested the presence of significant amounts of silica scale and some inorganic clay matter in the form of and aluminum silicates. The presence of silica as aluminum silicates suggested the co-polymerization of silica with aluminum. DISCUSSION The first pass membrane system consisting of NF270 membranes (SW and DT configuration) were fouled significantly with organic matter. The use of antiscalant in the first pass to prevent BaSO4 precipitation was efficient as no BaSO4 scaling was found on the membrane surface. The presence of significant organic matter on the NF270 membrane suggested that a large fraction of the organic matter passed through the pretreatment system and reached the NF membrane. Hence, optimization of pretreatment or utilization of better pretreatment processes could decrease the fouling potential of the first pass membranes and thereby decrease the cleaning

Copyright 2010 Water Environment Federation. All Rights Reserved.

506

Membrane Applications 2010

interval of the membranes. Although organic fouling of the first pass NF270 membrane occurred, chemical cleaning was found to be efficient in recovering the specific flux to initial values. Hence, operating the SW-NF270 system at 70% recovery and the DT-NF270 system (treating concentrate of SW-NF270) at 71% recovery was feasible and more than 90% overall recovery in the first pass was achievable. The second pass membrane system consisting of TM810L was fouled significantly with silica. Although the pH of the feed water was increased to 10.0 and a silica specific antiscalant was used, silica precipitation was predominant on the membrane surface. The precipitation of silica restricted the recovery of the second pass system to less than 75%. Hence the overall recovery of the entire system (both first and second pass) was restricted to a feed water recovery of less than 68%. To increase the recovery of the second pass system silica levels in the feed water needs to be reduced using a pretreatment process or a combination of pretreatment process to enable higher recovery of the membrane system. The second pass membrane systems using a seawater RO membrane (TM810L) was effective in removing target contaminants and meeting discharge limits in the treated water. Certain constituents such as benzene, methanol (constituting a large fraction of BOD), and selenium discharge criteria were not met by the RO membranes. But, the seawater TM810L membranes are also capable of meeting the discharge limits for boron. Hence, further post-treatment steps were necessary to meet discharge limits for contaminants such as methanol. Restriction of silica scaling on the membrane system is essential to achieve high (more than 80 %) feed water recoveries. Efficient pretreatment process for removal of silica might be necessary to achieve higher feed water recoveries. Efficient removal of silica in the feed water to the membrane surface can facilitate the use of a single pass brackish water reverse osmosis membrane or a tight nanofiltration (NF) membrane and also achieve overall feed water recoveries of more than 80%. CONLUSIONS High-pressure membrane processes such as nanofiltration and reverse osmosis were used to achieve reduction in TDS and chloride content of produced water to meet discharge limits satisfying a chloride concentration of less than 250 mg/L in the treated water. A two pass membrane system was used to optimize the recovery of membrane treatment and meet discharge limits. The first pass consisted of NF270 membranes and the second pass consisted of TM810L membranes. Fouling potential of NF and RO membranes was not substantially different when ceramic UF was used as pretreatment as opposed to the use of organoclay or MYCELX filtration. However, the pressure drop across organoclay filter bed and MYCELX cartridges increased substantially within a few hours of operation. Initial analysis based on feed water quality determined that the recovery of the NF and RO membrane system would be restricted due to the precipitation of silica, barium sulfate, and calcium carbonate on the membrane surface. The first pass membranes were used to remove a large fraction of scaling precursors such as calcium, sulfate, and barium from the feed water so that scaling issues due to calcium carbonate and barium sulfate was restricted in the second pass.

Copyright 2010 Water Environment Federation. All Rights Reserved.

507

Membrane Applications 2010

The first pass membrane removed 10-40% of calcium, more than 70% of sulfate, and more than 30% of barium. Due to removal of large fraction of calcium by the nanofiltration membrane the pH of the first pass permeate could be increased to 10.0 to increase silica solubility and thereby increase the feed water recovery of the second pass. More than 95% reduction of TDS in the feed water was achieved by the use of second pass membrane. The concentration of chloride in the permeate of the second pass RO membrane was consistently lower than the discharge limit of 230 mg/L. The second pass membrane also rejected more than 60% of COD and more than70 % of BOD. Alkalinity removal of more than 95% was also achieved by the RO membrane. Boron and selenium removals were more than 70% by the second pass RO membrane. Although a variety of discharge limits were met by using the two pass membrane system, the overall feed water recovery was restricted to less than 70 % due to organic fouling on the first pass membranes and silica scaling on the second pass membranes.

REFERENCES Bartels, C.; Dyke, C. (1990) Removal of organics from offshore produced water using nanofiltration membrane technology, Environmental Progress, 9, 183. Franks, R.; Bartels, C.; Nagghappan, LNSP (2009) Performance of a reverse osmosis system when reclaiming high pH high temperature wastewater, Proceedings of the AWWA Membrane Technology Conference, Memphis, Tennessee. Mohammadi, A.V.R.; Kazemimoghadam, M. (2003) Modeling of membrane fouling and flux decline in reverse osmosis during separation of oil in water emulsion, Desalination, 157, 369. Mondal, S.; Wickramasinghe, S.R. (2008) Produced water treatment by nanofiltration and reverse osmosis membranes, Journal of Membrane Science, 322, 162. Mueller, J.; Cen, Y.W.; Davis, R.H. (1997) Crossflow microfiltration of oily water, Journal of Membrane Science, 129, 221. Sheikholeslami, R.; Tan, S. (1999) Effects of water quality on silica fouling of desalination plants, Desalination, 126, 267. Tao, F.T.; Curtice, S.; Hobbs, R.; Sides, J.; Wieser, J.; Dyke, C.; Touhey, D.; Pilger, P. (1993) Conversion of oilfield produced water into an irrigation/drinking quality water, Society of Petroleum Engineers Exploration and Production Environmental Conference, San Antonio, Texas. Visvanathan, C.; Svenstrup, P.; Ariyamethee, P. (2000) Volume reduction of produced water generated from natural gas production process using membrane technology, Water Science Technology, 41, 117 123. Yoon, Y.; Lueptow, R.M. (2005) Removal of organic contaminants by RO and NF membranes, Journal of Membrane Science, 261, 76.

Copyright 2010 Water Environment Federation. All Rights Reserved.

508