Nanofiltration/Reverse Osmosis for Treatment of Coproduced Waters Subrata Mondal, Ching-lun Hsiao, and S. Ranil Wickramasinghe Department of Chemical and Biological Engineering, Colorado State University, Fort Collins, CO 80523-1370; wickram@engr.colostate.edu (for correspondence) Published online 15 April 2008 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ep.10271 Current high oil and gas prices have lead to renewed interest in exploration of nonconventional energy sources such as coal bed methane, tar sand, and oil shale. However oil and gas production from these nonconventional sources has lead to the coproduction of large quantities of produced water. While produced water is a waste product from oil and gas exploration it is a very valuable natural resource in the arid Western United States. Thus treated produced water could be a valuable new source of water. Commercially available nanofiltration and low pressure reverse osmosis membranes have been used to treat three produced waters. The results obtained here indicate that the permeate could be put to beneficial uses such as crop and livestock watering. However minimizing membrane fouling will be essential for the development of a practical process. Field Emission Scanning Electron Microscopy imaging may be used to observe membrane fouling. Ó 2008 American Institute of Chemical Engineers Environ Prog, 27: 173–179, 2008 Keywords: coal bed methane, membrane filtration, oily-waters, produced water INTRODUCTION Produced water (PW) refer to water that is coproduced during oil and gas exploration. Traditional oil and gas reservoirs have a natural water layer (formation water) that lies under the less dense hydrocarbons [1]. Often, water is injected into the reservoir to force the oil to the surface. As the oil field is depleted, the volume of PW increases as the reservoir fills with water. Ó 2008 American Institute of Chemical Engineers Environmental Progress (Vol.27, No.2) DOI 10.1002/ep In this work we focus on PW from nontraditional oil and gas sources in the Western United States. Unlike conventional oil and gas wells, production of oil and gas from nontraditional sources is usually accompanied by coproduction of large quantities of water. For example, the Wellington oil field in Wellington CO produces about 98% water and 2% oil [2]. In Wyoming, from 1987 to 2004, 2.9 billion barrels of water were produced while recovering 1.5 trillion cubic feet of coal bed methane (CBM). In 2003 alone, 577 million barrels of water were produced. The total volume of PW in Wyoming, if all known reserves of recoverable CBM were extracted, is estimated to be 55.5 billion barrels [3]. Management of PWs has become a major factor in the feasibility of gas field development [4]. Today more than 60% of the PWs are reinjected into wells that are geologically isolated from underground sources of drinking water. Reinjection costs vary from $0.40 to $1.75 per barrel, while installation costs vary from $400,000 to $3,000,000 per well [5]. Surface discharge of large volumes of PWs has already had many adverse environmental affects such as stream bank erosion, change in natural vegetation, salt deposition, etc [6]. Development of economical treatment processes for PWs is vital for two reasons:  Regions where CBM and other nonconventional oil and gas exploration are occurring in the western USA lack drinking and irrigation water. This situation will become much more severe as the population in the western states increases.  Economical and environmentally friendly methods of disposal of PWs are vital in order to prevent July 2008 173 Table 1. Common components in PW. Organic compounds Aliphatic, aromatic, polar compounds, e.g. fatty acids, oil, grease, benzene, phenol Inorganic components Na1, K1, Ca21, Mg21, Cl2, SO422, CO322, silicates (H4SiO2), borates (H3BO3) Production chemicals Emulsion breakers to improve separability of oil and water, corrosion inhibitors Table 2. Properties of membranes tested at 30 L m22 h-1, 2000-ppm solute, 258C, pH 7–8 and 10% recovery [13]. Membrane NF270 NF90 BW30 Feed pressure (psi) 50 70 150 Rejection (%) NaCl 80 CaCl2 50 MgSO4 99.3 NaCl 90–96 NaCl 98.6 CaCl2 98.8 MgSO4 99.7 serious environmental damage and allow development of new energy resources. Currently most PW is treated as waste; yet, as the demand for fresh water (especially in the western states) surpasses available supplies, treatment of PW could provide a source of new water for beneficial use. The composition of PWs varies widely since they originate from different geological formations. Table 1 gives some typical species that may be present. Conventional treatment of PWs has included gravity separation and skimming, dissolved air flotation, deemulsification, coagulation and flocculation [7–10]. However, there are numerous disadvantages associated with these unit operations. For example, gravity separation may not produce effluents that meet discharge limits; use of chemical emulsion breakers requires customization for each site to determine the types and quantities of chemicals needed; large volumes of sludge are often produced; and operation costs can be high. The use of membrane filtration processes such as nanofiltration and reverse osmosis offer many advantages.  The technology is more widely applicable across a range of industries.  The membrane is a positive barrier to rejected components, thus variation in feed water quality will have a minimal impact on permeate quality.  No addition of chemicals is required.  Membranes can be used in process to allow recycling of selected waste streams. Both polymeric and ceramic membranes have been used for PW treatment [11]. In general, polymeric membranes are cheaper, but they must be operated at lower temperatures and are less tolerant of harsh cleaning conditions. Here we have studied 174 July 2008 Comments MWCO between 200-400: NF270 has higher MWCO and NF90 is the tighter pore size Brackish water treatment membrane operates at lower pressure than salt water treatment membranes the use of commercially available polymeric nanofiltration and low-pressure reverse osmosis membranes. The membranes have different surface roughness and rejection behavior for salts. These membranes are used for brackish water treatment and may be well suited for treatment of PWs [4, 12]. The change in permeate flux as a function of time at constant pressure driving force has been determined for three different PWs. A reduction in permeate flux may be related to SEM images that indicate deposition of foulants on the membrane surface. EXPERIMENTAL PW was obtained from three sources in Colorado. The PWs used in these experiments were not kept in an oxygen free environment. The two PWs, labeled CBM1 and CBM2, were from CBM manufacturing facilities in Walsenburg, southern Colorado. The third PW, labeled ODG, was obtained from Wellington in northern Colorado and is associated with oil production. The PWs were characterized using Inductively Coupled Plasma Emission Spectroscopy (ICP AES). In addition, total organic carbon (TOC) and total dissolved solids (TDS) were also determined. Three membranes, two nanofiltration and one low pressure reverse osmosis, were donated by FilmTec Corporation, Dow Chemical, Midland, MI. These Filmtec membranes consist of three layers: a polyester support web, a microporous polysulfone interlayer and an ultra thin polyamide barrier layer on the top surface (See Table 2). NF270 is a piperazine-based semi-aromatic polyamide thin film composite membrane while NF90 and BW30 are fully aromatic polyamide thin film composite membranes. Tang et al. [14] have measured the following root mean square roughness for Environmental Progress (Vol.27, No.2) DOI 10.1002/ep 0.01 0.01 1.30 All values are in mg/mL. pH 8.52 8.41 8.70 675 547 1940 TDS TOC 68.8 47.7 136.4 <0.01 <0.01 <0.01 V Si 7.4 10.1 14.4 <0.005 <0.005 0.008 Pb Ba <0.01 0.01 10.1 0.26 0.21 2.90 B Sr 0.06 0.08 1.00 <0.01 <0.01 0.01 Cr Cd <0.005 <0.005 <0.005 <0.01 <0.01 <0.01 Mo Mn <0.01 0.01 0.02 0.01 0.05 0.07 Fe Al P K Mg CBM 1 CBM 2 ODG Na Environmental Progress (Vol.27, No.2) DOI 10.1002/ep Sample Ca Table 3 gives the water quality results. The ODG PW has a much higher TDS, TOC and salinity than the CBM PWs from southern Colorado. Since the pH of the PWs tested is around 7–8, the all three membranes are negatively charged. Figures 1–3 give the variation of permeate flux with pressure. Figure 1 gives results for CBM 1, Figure 2 for CBM2 and Figure 3 for ODG PW. As Table 3. PW analysis. RESULTS <0.01 0.75 2.20 where, Cf1 is the original conductivity of the feed water and Cf2 is the conductivity of the permeate sample. Clean and fouled membranes were inspected using Field Emission Scanning Electron Microscopy (FESEM). To prevent collapse of the pores during sample preparation, all membranes were immersed in ethanol for 4 h. The ethanol was then replaced by dry ice for about 1 h after which the system was heated to critical point of dry ice. Then, all membranes were coated with 10 nm of gold. FESEM images were taken using JSM6500F FESEM equipped with an in-lens thermal field emission electron gun. All membranes were imaged at a voltage of 15 kV and a magnification of 10,000 times. 0.01 0.08 <0.01 (1) 1.2 1.3 10.5 Cf2 Þ=Cf1 3 100% 314 250 782 Apparent salt rejection ¼ ðCf1 1.7 2.4 11.0 the NF270, NF90, and BW30 membranes 9.0 6 4.2, 129.5 6 23.4, and 68.3 6 12.5 nm, respectively. In addition they measured the zeta potential of the virgin membranes. At pH 7.0 the zeta potential of the NF270, NF90, and BW30 membranes are 232.6, 226.5, and 25.2 mV, respectively. The negative charge on the membrane is most likely due to the deprotonation of the carboxylic and amine groups at higher pH [15–17]. Prior to use, the membranes were not precompacted. However the membranes were washed with a 1:10:9 (volume basis) mixture of sulfuric acid/ethanol/DI water [18]. Dead end filtration experiments were conducted using a YT30 142 HW, Millipore Corp, Bedford, MA filtration cell. The membrane diameter was 140 mm. The feed was pressurized using a nitrogen cylinder attached to the feed reservoir. The feed volume was 500 mL. All experiments were conducted at room temperature. In the first series of experiments, the feed pressures varied from 20–100 psi. Once the system reached the required pressure, the pressure was held for 2 min to ensure equilibrium was reached. Next the permeate was collected over a 5-min interval. Then the pressure was set to the next value and the procedure repeated. In the second series of experiments the variation of the apparent salt rejection (as measured by conductivity of the permeate) with time at a constant pressure of 80 psi was determined. The permeate was collected for 20-min intervals and the TDS and conductivity determined using a handheld conductivity/TDS/temperature Meter, Oakton Instruments, Vernon Hills, IL. Apparent salt rejection was calculated by using the following equation. July 2008 175 Figure 1. Variation of permeate flux with pressure for Figure 3. Variation of permeate flux with pressure for CBM 1 PW. The looser the membrane pore structure (larger pores) the higher the permeate flux. Diamonds, squares, and triangles represent results for NF270, NF90, and BW30 membranes. ODG water. The looser the membrane pore structure (larger pore) the higher the permeate flux. The lowest permeate fluxes were obtained with the ODG water. Diamonds, squares, and triangles represent results for NF270, NF90, and BW30 membranes. Figure 4. Variation of salt rejection with time. Solid Figure 2. Variation of permeate flux with pressure for CBM 2 PW. The looser the membrane pore structure (larger pore) the higher the permeate flux. Diamonds, squares, and triangles represent results for NF270, NF90, and BW30 membranes. expected the permeate flux increases with increasing pressure. As can be seen the looser (larger pore size) NF270 membrane gives the highest flux for all three PWs. The permeate flux at the same pressure is lowest for the ODG water for all three membranes. Though similar the permeate flux for all three membranes is higher for CBM 2 compared with CBM 1 PW. Figure 4 gives the variation of apparent salt rejection, as measured by the conductivity of the permeate, with time. As expected, as the contents of the feed reservoir are concentrated, the apparent salt rejection decreases since Eq. 1 uses the initial conductivity of the feed. In general, the salt rejection at 176 July 2008 symbols are for NF90, open symbols are for BW30 membrane respectively. Diamonds, squares, and triangles represent CBM1, CBM2, and ODG PWs. any time is greater the smaller the pore size of the membrane. The highest salt rejection at any time is obtained for the ODG PW using the BW30 membrane followed by the NF90 membrane. Since rejection by the NF270 membrane will be less than the NF90 membrane, salt rejection data for the NF270 membrane are not included. FESEM images are given in Figures 5–10. Figures 5 and 6 give FESEM images for a clean and fouled NF270 membrane. Figures 7 and 8 give analogous results for the NF90 membrane while Figures 9 and 10 give results for the BW30 membrane. Comparing Figures 5, 7, and 9 with 6, 8, and 10 deposition of solute species present in CBM1 PW on the membrane surface is clearly visible. Further the NF270 membrane has the smoothest surface. Surface roughness Environmental Progress (Vol.27, No.2) DOI 10.1002/ep Figure 5. FESEM image of clean NF270 membrane. Figure 8. FESEM image of fouled NF90 membrane. The membrane was used to treat CBM1 PW. Figure 6. FESEM image of fouled NF270 membrane. The membrane was used to treat CBM1 PW. Figure 9. FESEM image of clean BW30 membrane. Figure 7. FESEM image of clean NF90 membrane. Figure 10. FESEM image of fouled BW30 membrane. The membrane was used to treat CBM1 PW. Environmental Progress (Vol.27, No.2) DOI 10.1002/ep July 2008 177 Table 4. Salt in bulk permeate estimated using the average apparent rejection coefficient. Water ODG ODG CBM1, CBM2, CBM1, CBM2 Membrane BW30 NF90 BW30 NF90 Permeate flux (L m22 h21) 5 5 10 10 Permeate flow rate (mL min21) 1.3 1.3 2.6 2.6 can have a large effect on the degree of membrane fouling [19]. DISCUSSION As the PWs used in these experiments were not kept in an oxygen free environment the pH of the original PW may be different. In these experiments it was decided not to store the PW under a controlled environment (e.g. under high purity argon) as actual treatment of PW will occur well after it is first exposed to the atmosphere. Table 3 indicates that there is great variability in the quality of PW. TDS as high as 170,000 mg L21 have been reported [20]. The TDS of a PW depends strongly on the geological formation and the origin of the water. Thus treatment requirements will vary. Further the degree of treatment also depends on the beneficial use for the PW. The recommended TDS for potable water is 50 mg L21 and 1000–2000 mg L21 for other beneficial uses such as stock ponds or irrigation. Comparing Table 1 and Figures 1–3 it appears that the higher the TDS the lower the permeate flux. Further the permeate flux is much higher for the NF270 membrane. This is expected given the NF270 membrane has a much more open pore structure than the NF90 membrane [13]. In fact the permeate fluxes for the NF90 and BW30 membrane are similar for all three PWs. In Figure 4 apparent salt rejection is determined by using Eq. 1. Since Eq. 1 uses the conductivity of the feed and permeate it gives the overall apparent rejection of ionic species not just NaCl. However from Table 3 it can be seen that the majority of the cations present are Na1. Consequently Eq. 1 is used to approximate the apparent salt rejection. Comparing Table 2 and Figure 4 it appears that the actual level of salt rejection obtained is less than specified by the manufacturer. However Figure 4 gives the apparent rejection coefficient based on the initial feed concentration while the manufacturer gives the actual rejection coefficient based on the actual feed concentration. At small run times the apparent and actual rejection coefficients approach each other. Figure 4 indicates that the apparent rejection coefficient approaches the manufacturer’s stated value at small run times. Further the test conditions used by the manufacturer are quite different to the real PWs tested here. In particular the presence of other charged species will affect the observed rejection of 178 July 2008 Rejection after 20 min (%) 70 70 70 60 Rejection after 140 min (%) 50 30 20 20 Salt in bulk permeate (mg/L) 315 390 175, 140 190, 150 salt [21]. It can also be seen form Figure 4 that even though ODG PW has a higher Na1 concentration the apparent salt rejection is higher than for CBM 1 and CBM 2 PWs. This is probably due to the much higher TDS of the ODG PW. In an earlier study Wickramsinghe et al. [22] noted that a high TDS improved the rejection of arsenic bound to ferric salts during microfiltration. Using the average value of the apparent rejection coefficient at 20 and 140 min, the salt concentration in the bulk permeate was estimated and is given in Table 4. The results suggest that the permeate could be used for beneficial uses such as livestock watering. Figures 5–10 indicate that FESEM images may be used to observe membrane fouling. Our results indicate that the PWs tested here may be treated with the BW30 or NF90 resulting in a permeate that that may be used for beneficial uses such as livestock and crop watering, the feasibility of using membranes depends on the degree of membrane fouling and consequently the frequency of cleaning. FESEM imaging of the membranes could provide a quick way to observe the degree of membrane fouling. In this work we have used dead end filtration to evaluate the performance of three commercially membranes. While dead end filtration is a quick way to assess membrane performance, tangential flow filtration data is needed prior to building a commercial facility. CONCLUSIONS Three commercially available nanofiltration and low pressure reverse osmosis membranes have been used to treat three PWs. Two of the membranes, BW30 and NF90 produce a permeate that could be used for beneficial uses such as livestock and crop watering. However the feasibility of using membranes to treat PWs depends on the degree of membrane fouling. FESEM imaging of the clean and used membranes may be a rapid method to observe the degree of membrane fouling. ACKNOWLEDGMENTS Funding for this work was provided by the National Science Foundation IIP 0637664 and CBET 0651646. Dr William Mickols, Dow-FilmTec, Edina, MN provided the membranes. Environmental Progress (Vol.27, No.2) DOI 10.1002/ep LITERATURE CITED 1. Wills, J. (2000). Muddied waters, environmental effects of drilling waste discharges. Available at: http://www.offshore-environment.com/producedwaters.html (accessed November 2006). 2. Pomeroy, B. (2006). Personal communication, Denver, CO: Wellington Operating Company. 3. The Ruckelshaus Institute of Environment and Natural Resources. (2005). Water production for coalbed methane development in Wyoming: A summary of quantity, quality and management options, prepared for The Office of the Governor, State of Wyoming. Available at: http://www. uwyo.edu/enr/ienr/CBMWaterFinalReportDec2005.pdf (accessed November 2006). 4. Xu, P., & Drewes, J.E. (2006). Viability of nanofiltration and ultra-low pressure reverse osmosis membranes for multi-beneficial use of methane produced water, Separation and Purification Technology, 52, 67–76. 5. Hayes, T., & Arthur, D. (2004). Overview of emerging treatment technologies. In Proceedings of the 11th Annual International Petroleum Environmental Conference, Albuquerque, NM, October 12–15. 6. Keith, K., Bauder, J., & Wheaton, H. (2003). Frequently asked questions, coal bed methane (CBM), Montana State University, Bozeman: The department of land resources and environmental sciences. Available at: http://waterquality. montana.edu/docs/methane/cbmfaq.shtml (accessed November 2006). 7. Cheryan, M., & Rajagopalan, N. (1998). Membrane processing of oily streams. Wastewater treatment and waste reduction, Journal of Membrane Science, 151, 13–28. 8. Beisinger, M.G., Vining, T.S., & Shell, G.L. (1974). Industrial experience with dissolved air flotation. In Proceedings of the 29th Purdue Industrial Waste conference, (p. 290–301), West Lafayette, IN. 9. Gardner, N.A. (1972). Flotation techniques applied to the treatment of effluents, Effluent and Waste Treatment Journal, 12, 82–85. 10. Lash, L.D., & Kominek, E.G. (1975). Primary waste treatment methods, Chemical Engineering, 82, 49–61. 11. Zaidi, A., Simms, K., & Kok, S. (1992). The use of micro/ultrafiltration for the removal of oil and suspended solids from oilfiled brines, Water Science and Technology, 24, 163–176. Environmental Progress (Vol.27, No.2) DOI 10.1002/ep 12. Hilal, N., Al-Zoubi, H., & Darwish, N.A., Mohammad, A.W. (2005). Nanofiltration of magnesium chloride, sodium carbonate, and calcium sulphate in salt solutions, Separation Science and Technology, 40, 3299–3321. 13. See Dow Water Solutions. http://www.dow. com/liquidseps/ (accessed July 2007) 14. Tang, C.Y., Kwon, Y.-N., & Leckie, J.O. (2007). Fouling of reverse osmosis and nanofiltration membranes by humic acid—Effects of solution composition and hydrodynamic conditions, Journal of Membrane Science, 290, 86–94. 15. Tang, C.Y., Kwon, Y.-N., & Leckie, J.O. (2007). Characterization of humic acid fouled reverse osmosis and nanofiltration membranes by transmission electron microscopy and streaming potential measurements, Environmental Science and Technology, 41, 942–949. 16. Tang, C.Y., Kwon, Y.-N., & Leckie, J.O. (2007). Probing the nano- and micro-scales of reverse osmosis membranes—A comprehensive characterization of physiochemical properties of uncoated and coated membranes, XCPS, TEM, ATR-FTIR, and streaming potential measurements, Journal of Membrane Science, 287, 146–156. 17. Childress, A.E., & Elimelech, M. (1996). Effect of solution chemistry on the surface charge of polymeric reverse osmosis and nanofiltration membranes, Journal of Membrane Science, 119, 253– 268. 18. Kulkarni, A., Mukherjee, D., & Gill, W.N. (1996). Flux enhancement by hydrophilization of thin film composite reverse osmosis membranes. Journal of Membrane Science, 114, 39–50. 19. Nicolaisen, B. (2002). Developments in membrane technology for water treatment, Desalination, 153, 355–360. 20. Rice, C.A., & Nuccio, V. (2000). Water produced with coal bed methane, USGS Fact Sheet FS 15600, USGS, Denver, CO. 21. Mukherjee, P., Jones, L.K., & Abitoye, J.O. (2005). Surface modification of nanofiltration membranes by ion implantation, Journal of Membrane Science, 254, 303–310. 22. Wickramasinghe, S.R., Han, B., Zimbron, J., Shen, Z., & Karim, N.M. (2004). Arsenic removal by coagulation and filtration: Comparison of groundwaters from the United States and Bangladesh, Desalination, 169, 231–244. July 2008 179