Experimental Investigation of a Spiral-Wound ... - ACS Publications

Feb 11, 2013 - The SW PRO membrane module has two different flow paths, axial and ... in pressure-retarded osmosis: Module performance comparison with...
1 downloads 0 Views 4MB Size
Download PDF
Article pubs.acs.org/est

Experimental Investigation of a Spiral-Wound Pressure-Retarded Osmosis Membrane Module for Osmotic Power Generation Yu Chang Kim,* Young Kim, Dongwook Oh, and Kong Hoon Lee Department of Thermal Systems, Korea Institute of Machinery and Materials, Daejeon 305-343, Republic of Korea S Supporting Information *

ABSTRACT: Pressure-retarded osmosis (PRO) uses a semipermeable membrane to produce renewable energy from salinity-gradient energy. A spiral-wound (SW) design is one module configuration of the PRO membrane. The SW PRO membrane module has two different flow paths, axial and spiral, and two different spacers, net and tricot, for drawand feed-solution streams, respectively. This study used an experimental approach to investigate the relationship between two interacting flow streams in a prototype SW PRO membrane module, and the adverse impact of a tricot fabric spacer (as a feed spacer) on the PRO performance, including water flux and power density. The presence of the tricot spacer inside the membrane envelope caused a pressure drop due to flow resistance and reduced osmotic water permeation due to the shadow effect. The dilution of the draw solution by water permeation resulted in the reduction of the osmotic pressure difference along a pressure vessel. For a 0.6 M NaCl solution and tap water, the water flux and corresponding maximum power density were 3.7 L m−2h−1 and 1.0 W/m2 respectively at a hydraulic pressure difference of 9.8 bar. The thickness and porosity of the tricot spacer should be optimized to achieve high SW PRO module performance.

■

INTRODUCTION The emphasis on renewable and sustainable energy sources is continuously increasing because of the growing demand for energy throughout the world.1 The threat of global warming and the depletion of finite sources have directed more interest toward renewable energy sources such as solar, wind, and geothermal.2 Recently, osmotic power, salinity-gradient energy that can be converted into affordable electricity, has also been recognized as a potential renewable energy source because the salinity gradient that exists in seawater and river water estuaries is plentiful throughout the world.3−5 Pressure-retarded osmosis (PRO) is well-known as a salinitygradient power generation process based on the mixing of two solutions of different concentrations.6−10 PRO uses a semipermeable membrane as a means of mixing and requires both an osmotic pressure difference for water transport and a hydraulic pressure difference for energy conversion. In PRO, the draw solution (e.g., seawater) must be pressurized, but the applied hydraulic pressure difference must be lower than the osmotic pressure difference, thereby retarding the permeating water flow through the membrane. Like hydroelectric power generation, a hydroturbine is used to extract work from the pressurized diluted draw solution. A full-scale PRO osmotic power plant would require an enormous membrane area because the permeation rate through the PRO membrane is low due to osmosis retardation caused by the pressure required for energy conversion.11,12 The performance and price of the membrane will determine whether osmotic power generation is a viable process for real-world applications. Significant effort has thus been focused on designing the membrane structure to minimize the effect of © 2013 American Chemical Society

internal concentration polarization (ICP) that occurs within a porous support layer.13−16 ICP in osmotically driven PRO decreases the net driving force (i.e., effective osmotic pressure difference) and thus dramatically reduces the water flux and power density.13 However, PRO membranes for osmotic water permeation should also have the ability to withstand the high hydraulic pressure applied to the draw-solution side. Most recent studies on PRO have used a crossflow membrane test cell loaded with a flat-sheet membrane coupon.9,10,15,17 However, the operating conditions (flow rates and hydraulic pressures on both sides of the membrane) and the flow channel structure in a crossflow membrane cell is quite different from those in a spiral-wound (SW) PRO membrane module. Unlike in an SW forward osmosis (FO) module,18 the hydrodynamics in the feed channel of an SW PRO module are significantly influenced by the presence of the spacer because the draw-solution side is pressurized. Additionally, the dilution of the draw solution in the crossflow cell is negligible, whereas that in the membrane module could be significant because of the use of such a large membrane area. A Norwegian energy company, Statkraft, has recently conducted pilot tests with PRO membrane modules. However, no data about the module have ever been reported and no SW PRO membrane modules are commercially available. Therefore, a module study is necessary to design a real osmotic power plant and an optimal SW PRO membrane module structure. Received: Revised: Accepted: Published: 2966

October 6, 2012 January 15, 2013 February 11, 2013 February 11, 2013 dx.doi.org/10.1021/es304060d | Environ. Sci. Technol. 2013, 47, 2966−2973

Environmental Science & Technology

Article

Figure 1. (a) Flow rate effect on the inlet pressure in the SW PRO membrane module. Each inlet pressure was measured when there was a flow stream on only one side of the PRO membrane. (b) Effect of the hydraulic pressure difference (ΔP) on the flow rate of the draw and feed-solution sides in the presence of two streams on both sides of the membrane. Here, the draw inlet pressure was increased by adjusting only the back-pressure valve. (c) Effect of the draw inlet pressure on the feed inlet pressure. (d) Draw and feed inlet flow rates as functions of draw inlet pressure. In each case, tap water was used on both sides.

solutions (584, 2922, and 4675 mg/L NaCl) were used as the feed solutions in the PRO experiments. Because of the capacity of our PRO experimental unit using an SW element, we used tap water (approximately 65 mg/L in total dissolved solids) instead of deionized water to prepare the NaCl solutions. The initial volume of each tank was 160 L. During the PRO experiments, the draw and feed solutions were recycled into each solution tank. The membrane module was thoroughly flushed between experiments to remove any salt that may have accumulated in the porous support layer. The osmotic pressures of the NaCl solutions were calculated using a commercial software program (Stream Analyzer, OLI Systems, Inc., Morris Plains, NJ). PRO Experimental Setup. Figure S4 of the Supporting Information shows a schematic diagram of our PRO experimental setup in which a high-pressure pump (CRN 1− 25, Grundfos, Denmark) and a low-pressure pump (CRN 1−9, Grundfos, Denmark) were used to circulate the draw and feed solutions respectively in closed loops. The hydraulic pressure and flow rate of the draw-solution side were controlled by both a back-pressure valve installed at the outlet of the draw-solution side and a variable-frequency drive connected to the highpressure pump, whereas those of the feed-solution side were controlled only by a bypass valve. To measure the operating parameters including temperature, pressure, flow rate, and concentration, we installed several calibrated measuring devices, including thermocouples, pressure transmitters, turbine flow meters, and conductivity meters in the pipeline or tank. A chiller and an electric heater with a temperature controller were also installed in each solution tank to maintain the solution at the desired temperature. Data were collected every 5 s using the measuring devices connected to a computer and a dataacquisition switch unit (HP 34970A, Hewlett-Packard/Agilent Technologies, U.S.).

In this work, we investigated the relationship between two interacting flow streams (draw and feed solutions) using a prototype module to obtain experimental findings for an SW PRO membrane module. We also examined the effect of the tricot fabric spacer on the water permeation and PRO performance of the SW PRO membrane module. In addition, the water flux and power density of the SW PRO membrane module were analyzed under various salinity-gradient conditions using a pilot-scale PRO experimental setup.

■

MATERIALS AND METHODS

Spiral-Wound PRO Membrane Module. A prototype SW PRO membrane module (Woongjin Chemical Co., Korea), 0.2 m (8 in.) in diameter and 1 m (40 in.) in length, was used for our PRO study. The effective membrane area of this PRO module was approximately 29 m2, and 20 membrane envelopes were rolled into an SW configuration. According to the manufacturer, the membrane fabricated for PRO was a thin-film composite polyamide membrane (Figure S1 of the Supporting Information). The outer face of the membrane envelope was the active layer while the inner face was a porous support layer. Tricot fabric and biplanar extruded netting (Figure S2 of the Supporting Information) were used as feed and draw channel spacers in the inner and outer sides of the membrane envelope, respectively. Their measured thicknesses were approximately 0.4 and 0.8 mm, respectively. The Supporting Information includes a schematic diagram (Figure S3 of the Supporting Information) and details of the SW PRO membrane module. Two types of commercial SW reverse osmosis (RO) membrane modules (RE8040-SN and RE8040-BE; Woongjin Chemical Co., Korea) were also tested to confirm the adverse effect of the tricot fabric spacer used in the SW membrane modules. Draw and Feed Solutions. High-salinity sodium chloride (NaCl) solutions (35 000 and 70 000 mg/L NaCl) were used as the draw solution while tap water and low-salinity NaCl 2967

dx.doi.org/10.1021/es304060d | Environ. Sci. Technol. 2013, 47, 2966−2973

Environmental Science & Technology

Article

the flow rate and hydraulic pressure on the opposite side were affected (parts b and c of Figure 1). The increase in the draw inlet pressure resulted in both a decrease in the feed flow rate (red open squares, part b of Figure 1) and an increase in the feed inlet pressure (part c of Figure 1). As expected, manipulating the back-pressure valve changed the draw flow rate (blue open circles, part b of Figure 1). Even though the back-pressure valve was fully open, the feed inlet pressure increased up to about 3.5 bar because the membrane envelope was compressed by the draw-solution stream, unlike in the presence of a single feed-solution stream. In the presence of two solution streams, the compression effect resulted in a large pressure drop on the feed-solution side. Thus, the feed-solution stream of the inner side of the membrane envelope should overcome the flow resistance caused by the tricot fabric spacer. As the draw inlet pressure increased from 4 to 20 bar, the feed inlet pressure increased from 3.5 to 4.5 bar (part c of Figure 1), the draw inlet flow rate decreased from 41 to 22.5 LPM, and the feed inlet flow rate decreased from 12 to 6.5 LPM (part d of Figure 1). The pressure drop on the feed side was significantly high because of the spiral circuitous flow path, the use of a tricot spacer, and the membrane envelope compression.20−24 The pressure drop of the feed flow path (i.e., inner side of the membrane envelope) was related to the energy consumption.25 As the pressure drop of the feed-solution side increased, the hydraulic pressure of the draw-solution side should also increase to maintain the required hydraulic pressure difference. Accordingly, the energy consumed by the pump will increase. Transport Properties of the PRO Membrane in an SW Module. The water and salt permeability coefficients (A and B, respectively) of the membrane active layer determine the membrane performance.15,26 To determine the transport properties, RO experiments are usually conducted in an RO crossflow test cell loaded with a flat-sheet membrane coupon.15,26 However, in such experiments, the structural effect of the SW PRO module is not included in the calculation of the A and B values because the flow channel of the crossflow test cell is a simple rectangular structure. Conventionally, the flat-sheet membrane is supported by a highly porous sintered stainless steel plate inserted in the permeate side of the crossflow membrane cell, instead of a permeate channel spacer.17 Therefore, we investigated the membrane transport properties using an SW PRO membrane module. For this test, we performed RO experiments by running only the drawsolution (high-pressure) pump in our PRO experimental setup. Of course, the feed solution for RO experiments should flow through the draw-solution path (outer side of the envelope) of the PRO membrane module. Figure 2a shows the water flux (JW) and water permeability coefficient (A) of the SW PRO membrane module. Although the water flux increased steadily with increasing hydraulic pressure difference, the hydraulic pressure and water flux were not in direct proportion. That is, the water permeability coefficient (A) of the SW PRO membrane module was not constant because it was determined by JW/ΔP. However, the A value obtained using a flat sheet PRO membrane coupon in the crossflow membrane test cell was constant (Figure S5 of the Supporting Information). This difference occurred because of the presence of the dense tricot fabric spacer. The inevitable pressure drop caused by the tricot fabric spacer reduced the driving force available for membrane permeation. The net operating pressure required to obtain the desired permeate flow

Determination of PRO Membrane Transport Properties in an SW Module. For determining the water and salt permeability coefficients of the membrane (A and B, respectively), the SW membrane modules (one PRO and two RO modules) were used directly instead of a flat-sheet membrane coupon. The feed- and draw-solution flow paths of the SW PRO module were used as the permeate and feed flow paths respectively of the SW RO module. For comparison, flat-sheet PRO membrane coupons were also tested in a crossflow membrane test cell; details are in the Supporting Information. The water permeability coefficient (A) was determined in an RO experiment using tap water as the feed solution. Initially, the membrane module was compacted with tap water at an applied hydraulic pressure of 19.7 bar for 15 h. Next, the water permeate flow rate was measured at applied pressures decreasing from 19.7 to 3.9 bar in approximately 2-bar decrements. The water flux (JW) at each applied pressure was obtained by dividing the water permeate rate by the effective membrane area. The A value was then determined by dividing the water flux by the applied hydraulic pressure difference: A = JW/ΔP.9,17 The salt permeability coefficient (B) was determined in an RO experiment using 2922 mg/L (0.05 M) NaCl solution at hydraulic pressures of 7.8, 11.7, and 15.7 bar. The salt rejection (R) was obtained by measuring the conductivities of the bulk feed (CF) and permeate (CP) solutions. The B value was determined using B = A(ΔP − Δπ)(1 − R)/R.9,10

■

RESULTS AND DISCUSSION Influence of the Draw Stream on the Feed Stream in an SW PRO Module. The pressure drop caused by the tricot fabric spacer in an SW module has an adverse impact on the module performance.19,20 Therefore, we investigated the relationship between hydraulic pressure and flow rate for the prototype SW PRO module using tap water. Part a of Figure 1 illustrates the effect of the flow rate on the flow resistance. Naturally, the hydraulic pressure increased with increasing flow rate. The pressures on each side were measured when there was no flow stream on the opposite side of the PRO membrane. In our PRO setup, the maximum feed flow rate was 20 LPM resulting in an inlet pressure of 1.4 bar, and the maximum draw flow rate was 41 LPM resulting in an inlet pressure of 3.5 bar. The feed flow rate should be lower than the draw flow rate in the SW PRO module because, if the feed flow rate were higher, severe back-pressure would build up due to the large flow resistance caused by the tricot spacer inside the envelope. The pressure drop on the feed-solution side of the SW PRO module was 1.4 bar at a flow rate of 20 LPM (nearly equal to the inlet feed pressure), whereas that of the draw-solution side was approximately 0.16 bar at a maximum flow rate of 41 LPM. The feed inlet pressure was higher than the draw inlet pressure at an equivalent flow rate of approximately 20 LPM because the tricot fabric spacer generated greater flow resistance than the net spacer did, even though the presence of the back-pressure valve at the outlet of the draw side induced high flow resistance and there was no valve at the outlet of the feed side. We next investigated the relationship between the hydraulic pressure and flow rate in the presence of both solution streams in an SW PRO module. For this experiment, we used tap water on both sides of the membrane. As we increased the hydraulic pressure of the draw-solution side by adjusting the backpressure valve installed at the outlet of the draw-solution side, 2968

dx.doi.org/10.1021/es304060d | Environ. Sci. Technol. 2013, 47, 2966−2973

Environmental Science & Technology

Article

tube and lower in the membrane region furthest from the central tube. Furthermore, channeling and dead zones may reduce the available membrane area because of a somewhat complicated flow pattern by both the longitudinal drawsolution flow and the spiral circuitous feed-solution flow in the SW module. For these reasons, the water permeability coefficient (A) of an SW PRO membrane module is not constant, unlike that of a PRO membrane coupon. To confirm the adverse effect of the tricot spacer on water permeation, the RO test described above was performed for two SW RO membrane modules. The hydraulic pressure and corresponding water flux of the SW RO modules were not in direct proportion (Figure S6 of the Supporting Information), as the results above indicate for the SW PRO membrane module. The decrease in water permeation resulted from the flow resistance caused by the tricot spacer, but the drop was slight because the envelope of the SW RO module had no circuitous permeate path, unlike that of an SW PRO module.28 The water flux of the RO membrane was also much higher than that of the PRO membrane because the former was fabricated for a pressure-driven process and the latter for an osmotically driven process. The salt rejection (R) of the SW PRO membrane module was obtained by RO experiments using a 2922 mg/L (0.05 M) NaCl solution as the feed solution. The salt permeability coefficient (B) was determined by B = A(ΔP − Δπ)(1 − R)/R. Part b of Figure 2 indicates that the B value was not constant and increased slightly with increasing hydraulic pressure difference. This was because the salt rejection (R) increased but the water permeability coefficient (A) decreased with increasing hydraulic pressure difference. At hydraulic pressure differences (ΔP) of 7.8, 11.7, and 15.7 bar, the water permeability coefficients (A) were 0.81, 0.72, and 0.66 L

Figure 2. (a) Water flux (JW) and water permeability coefficient (A) of an SW PRO membrane module as a function of applied hydraulic pressure (P). RO experiments were conducted with tap water as feed solution. (b) Salt permeability coefficient (B) and salt rejection (R) as a function of P. RO experiments were conducted with a 0.05 M NaCl solution as the feed solution. All experiments were performed with solutions at a flow rate of 22.5 LPM at 25 °C.

rate should be increased. For a given average flux of the SW module, the local flux varies from region to region.27 That is, the flux is higher in the membrane region close to the central

Figure 3. Effect of the permeating flow on the flow rate and hydraulic pressure of the draw- and feed-solution sides in the SW PRO membrane module with the back-pressure valve at the outlet of the draw-solution side fully open. Tap water was used on both feed and draw sides for case A and 35 000 mg/L (0.6 M) and 70 000 mg/L (1.2 M) NaCl solutions were used as the draw solution for cases B and C, respectively. In all cases, the feed solution was tap water. 2969

dx.doi.org/10.1021/es304060d | Environ. Sci. Technol. 2013, 47, 2966−2973

Environmental Science & Technology

Article

m−2h−1bar−1 and the salt rejections (R) were 95%, 96%, and 96.7%, respectively. The corresponding salt permeability coefficients (B) were 0.24, 0.28, and 0.30 L m−2h−1. Influence of Water Permeate Flow on Hydraulic Pressure of the Draw-Solution Side. Water permeation across the flat-sheet membrane coupon in a crossflow membrane test cell is too small to affect the hydraulic pressure of the draw-solution side due to the small effective membrane area and short channel length, but this is not the case for the SW module with a large membrane area, as explained below. Figure 3 shows the effects of the permeating flow on the hydraulic pressure of the draw-solution side. In case A, tap water was used on both sides of the membrane module. Even when the back-pressure valve at the outlet of the draw-solution side was fully open, the measured inlet hydraulic pressures of the draw- and feed-solution sides were 3.92 and 3.53 bar respectively in the presence of two streams because of the flow resistance due to the back-pressure valve and the compression of the membrane envelope. Water permeated through the membrane from the draw- to the feed-solution side because there was a small pressure difference (ΔP = 0.39 bar) at the inlet, but the pressure drop of the feed side (Pdrop = 3.53 bar) was much higher than that of the draw side (Pdrop = 0.1 bar). Subsequently, NaCl solutions of 35 000 mg/L and 70 000 mg/L were used as the draw solutions for cases B and C, respectively. The hydraulic pressure difference of both sides of the membrane increased slightly with increasing salinity gradient (ΔP = 0.79 bar at 35 000 mg/L and ΔP = 1.26 bar at 70 000 mg/L NaCl) because water permeated through the membrane from the feed- to the draw-solution side. Notably, the feed inlet pressure decreased slightly with increasing salinity gradient because of the increase in osmotic water permeation into the draw-solution side. From these results, we concluded that the increase of hydraulic pressure of the draw-solution side was due to the combined effects of both draw-solution and permeate flow rate. This point should be considered in the design of a real osmotic power plant. Optimal Hydraulic Pressure Difference for the Peak Power Density. Because an optimal hydraulic pressure difference exists for the peak power density in the PRO process, we investigated the effect of hydraulic pressure difference on the water flux and power density. For this, we increased the hydraulic pressure difference (ΔP) from 0.9 to 15.4 bar in approximately 3.7 bar increments and then decreased ΔP again. Here, 35 000 mg/L (0.6 M) NaCl solution and tap water were used as the draw and feed solutions, respectively. The flow rates of both flow streams were not controlled because only the back-pressure valve was used. The flow rates of both streams varied with hydraulic pressure difference. Part a of Figure 4 shows the hydraulic pressure difference (ΔP) and the experimentally obtained water flux (JW) as a function of time. The predetermined ΔP was held at each level for approximately 150 s. As the hydraulic pressure difference increased, the water flux decreased, and vice versa. However, for any given value of ΔP, the left-hand water flux values were slightly higher than the right-hand values because the concentration of the draw solution became diluted and changed over time. On the basis of the ΔP and JW values in part a of Figure 4, we obtained the corresponding power density (W) (part b of Figure 4). A power density of approximately 0.9 W/m2 was obtained at a hydraulic pressure difference of approximately 7.7 bar.

Figure 4. (a) Effect of the hydraulic pressure difference (ΔP) on the water flux (JW) with time. (b) Corresponding projected power density (W) with time. The draw and feed solutions were 35 000 mg/L (0.6 M) NaCl and tap water, respectively. The draw inlet pressure was changed every 150 s, but the corresponding flow rate was not controlled in this experiment. The temperature of both solutions was held at 25 °C.

For experiments under identical conditions concerning concentration and flow rate, we replaced both solutions with new ones for each hydraulic pressure difference. That is, we adjusted the volume and concentration of the solutions to the initial conditions for each hydraulic pressure difference. We also controlled the flow rates of both streams using a variable frequency drive and a bypass valve for comparison under the same conditions. The controlled flow rates of the draw and feed solution were 22.5 and 7 LPM respectively for each hydraulic pressure. In the SW module, the water permeate flow across the membrane significantly diluted the concentration of the draw solution, as shown in Figure S7 of the Supporting Information. In particular, higher slopes and greater dilution occurred at a lower hydraulic pressure difference (ΔP) because the osmotic water permeation was higher. Theoretically, the power density should increase with increasing hydraulic pressure difference, but reach a maximum value (Wmax) when the hydraulic pressure difference (ΔP) is approximately half of the osmotic pressure difference (Δπ).9,15,17 Parts a and b of Figure 5 show JW and W as functions of ΔP. Higher ΔP corresponded to lower JW, whereas an optimal hydraulic pressure difference existed between 7.7 and 11.6 bar, and Wmax was approximately 1.0 W/m2. For this PRO membrane module, the optimal hydraulic pressure difference was approximately 9.8 bar, and the corresponding flux reversal pressure difference was approximately 20 bar. This reversal point was significantly lower than the ideal value (27.4 bar). This lower-than-expected flux reversal pressure is clearly the result of external and internal concentration polarization, and reverse salt diffusion.9,15,29 In addition to these effects, the shadow effect of the feed channel spacer was one of the performance-limiting phenomena.17 In PRO, as the hydraulic pressure of draw-solution side increases, the membrane area in 2970

dx.doi.org/10.1021/es304060d | Environ. Sci. Technol. 2013, 47, 2966−2973

Environmental Science & Technology

Article

Figure 5. Experimental water flux (JW) and the respective projected power density (W) as a function of applied hydraulic pressure difference (ΔP). The draw solution was 35 000 mg/L NaCl in (a) and (b), and 70 000 mg/L NaCl in (c) and (d). In all cases, tap water was used as the feed solution. The flow rates of the draw and feed solutions were fixed at 22.5 and 7 LPM, respectively. All experiments were performed with the solutions at a fixed temperature of 25 °C.

ICP in PRO is usually related to the salt concentration of the feed solution facing a porous support layer.13,14 Because the water flux decreases as the feed concentration increases due to the concentrative ICP effect even under the same osmotic pressure difference,10 we performed PRO experiments for various feed concentrations to examine the effect of the ICP phenomenon. Here, we used 584, 2922, and 4675 mg/L (0.01, 0.05, and 0.08 M, respectively) NaCl solutions as feed solutions to simulate river water, wastewater effluent, and brackish water, respectively.16,30−33 Because the 35 000 mg/L (0.6 M) NaCl draw solution was used in all the cases above, the osmotic pressure differences of the salinity-gradient pairs were not the same (Δπ = 27.3, 25.5, and 24.1 bar, respectively). Figure 6 shows the water flux and power density at a hydraulic pressure difference of 9.8 bar. As the concentration of the feed solution increased, the water flux decreased significantly. Even though the osmotic pressure difference between 35 000 mg/L NaCl (e.g., seawater) and 4675 mg/L NaCl (e.g., brackish water) solutions was as high as 24.1 bar, the water flux and corresponding power density were extremely low. To determine whether the flux decline was due to ICP or simply the decreased osmotic driving force, a PRO experiment using 30 400 mg/L (0.52 M) NaCl draw solution and tap water feed solution (Δπ = 24 bar) was also performed. Although the osmotic pressure differences of the above two cases were almost the same, a huge difference in PRO performance occurred because of the severe concentrative ICP effect. When the brackish water was used as the feed solution, a higherconcentration draw solution, such as brine from a desalination plant, should be apgplied.33−35 Implications. Because the feed-solution side was compressed in the PRO process, a tricot fabric spacer instead of a net spacer should be used to prevent membrane deformation. However, the use of the tricot spacer produced high flow resistance, resulting in a large pressure drop that reduced the efficiency of the SW module. The shadow effect of the dense

contact with the feed-channel spacer also increases, reducing the available area for osmotic water permeation from the feedto draw-solution sides. Therefore, the tricot fabric spacer for PRO should have sufficient thickness and porosity (or opening size) to form an open pathway for feed solution flow and water permeation. Of course, the use of a thicker spacer means the area of the membrane that can be packed into a module must be smaller. An opening size that is too large also results in membrane deformation.17 Influence of the Salinity Gradient on JW and W. A higher salinity gradient was expected to result in a higher water permeate flow.10,17 As the concentration difference between the draw and feed solutions increased, so did the hydraulic pressure difference necessary to generate the maximum power density. Thus, to investigate the effect of the salinity gradient, we used 70 000 mg/L (1.2 M) NaCl solution as the draw solution and tap water as the feed solution. The osmotic pressure difference of the two solutions was 56.9 bar. For the experiments using 70 000 mg/L (1.2 M) NaCl draw solution, we were unable to observe a peak power density over the range of 0.5 to 15.6 bar, and it was impossible to perform PRO experiments above 15.6 bar (ΔP) due to the limited capacity of our high-pressure pump. At a hydraulic pressure difference of 15.6 bar, the measured water flux was approximately 4.82 L m−2h−1 and the corresponding projected power density, obtained on the basis of the water flux and the hydraulic pressure difference, was 2.1 W/m2. The water flux did not increase linearly with the salinity gradient because the PRO process was an osmotically driven process with an ICP effect.13,14 The dilution effect was also larger for a higher salinity gradient than for a lower salinity gradient (data not shown). At a pressure difference of approximately 7.7 bar, the power density and water flux for each draw solution concentration (35 000 and 70 000 mg/L NaCl) were 0.96 W/m2 at 4.30 L m−2h−1 and 1.53 W/m2 at 6.72 L m−2h−1, respectively. 2971

dx.doi.org/10.1021/es304060d | Environ. Sci. Technol. 2013, 47, 2966−2973

Environmental Science & Technology

Article

Industrial Technology (KEIT) funded by the Ministry of Knowledge Economy (MKE), Korea. The authors also acknowledge Woongjin Chemical Co. for providing PRO membrane modules.

■

(1) Lior, N. Sustainable energy development: The present (2009) situation and possible paths to the future. Energy 2010, 35, 3976− 3994. (2) Banos, R.; Manzano-Agugliaro, F.; Montoya, F. G.; Gil, C.; Alcayde, A.; Gomez, J. Optimization methods applied to renewable and sustainable energy: A review. Renewable and Sustainable Energy Reviews 2011, 15, 1753−1766. (3) Achilli, A.; Childress, A. E. Pressure retarded osmosis: From the vision of Sidney Loeb to the first prototype installation−Review. Desalination 2010, 261, 205−211. (4) Ramon, G. Z.; Feinberg, B. J.; Hoek, E. M. V. Membrane−based production of salinity−gradient power. Energy Environ. Sci. 2011, 4, 4423−4434. (5) Logan, B. E.; Elimelech, M. Membrane−based processes for sustainable power generation using water. Nature 2012, 488, 313−319. (6) Loeb, S. Energy production at the Dead sea by pressure-retarded osmosis: challenge or chimera. Desalination 1998, 120, 247−262. (7) Loeb, S. One hundred and thirty benign and renewable megawatts from Great Salt Lake? The possibilities of hydroelectric power by pressure-retarded osmosis. Desalination 2001, 141, 85−91. (8) Skilhagen, S. E.; Dugstad, J. E.; Aaberg, R. J. Osmotic power− power production based on the osmotic pressure difference between waters with varying salt gradients. Desalination 2008, 220, 476−482. (9) Achilli, A.; Cath, T. Y.; Childress, A. E. Power generation with pressure retarded osmosis: An experimental and theoretical investigation. J. Membr. Sci. 2009, 343, 42−52. (10) She, Q.; Jin, X.; Tang, C. Y. Osmotic power production from salinity gradient resource by pressure retarded osmosis: Effects of operating conditions and reverse solute diffusion. J. Membr. Sci. 2012, 401−402, 262−273. (11) Loeb, S. Large-scale power production by pressure-retarded osmosis, using river water and sea water passing through spiral modules. Desalination 2002, 143, 115−122. (12) Thorsen, T.; Holt, T. The potential for power production from salinity gradients by pressure retarded osmosis. J. Membr. Sci. 2009, 335, 103−110. (13) Mehta, G. D.; Loeb, S. Internal polarization in the porous substructure of a semipermeable membrane under pressure-retarded osmosis. J. Membr. Sci. 1978, 4, 261−265. (14) Lee, K. L.; Baker, R. W.; Lonsdale, H. K. Membranes for power generation by pressure-retarded osmosis. J. Membr. Sci. 1981, 8, 141− 171. (15) Yip, N. Y.; Tiraferri, A.; Phillip, W. A.; Schiffman, J. D.; Hoover, L. A.; Kim, Y. C.; Elimelech, M. Thin-film composite pressure retarded osmosis membranes for sustainable power generation from salinity gradients. Environ. Sci. Technol. 2011, 45, 4360−4369. (16) Chou, S.; Wang, R.; Shi, L.; She, Q.; Tang, C.; Fane, A. G. Thinfilm composite hollow fiber membranes for pressure retarded osmosis (PRO) process with high power density. J. Membr. Sci. 2012, 389, 25− 33. (17) Kim, Y. C.; Elimelech, M. Adverse impact of feed channel spacers on the performance of pressure retarded osmosis. Environ. Sci. Technol. 2012, 46, 4673−4681. (18) Kim, Y. C.; Park, S.-J. Experimental study of a 4040 spiralwound forward-osmosis membrane module. Environ. Sci. Technol. 2011, 45, 7737−7745. (19) Kaschemekat, J.; Baker, R. W.; Wijmans, J. G. Membrane module. U.S. Patent 5,069,793, (1991). (20) Schwinge, J.; Neal, P. R.; Wiley, D. E.; Fletcher, D. F.; Fane, A. G. Spiral wound modules and spacers: Review and analysis. J. Membr. Sci. 2004, 242, 129−153.

Figure 6. Effect of the feed solution concentration on the water flux (JW) and the corresponding projected power density (W). Tap water and 0.01, 0.05, and 0.08 M NaCl solutions were used as the feed solutions, and 0.6 and 0.52 M NaCl solutions were used as the draw solutions. The results were obtained at an applied hydraulic pressure difference (ΔP) of 9.8 bar. All experiments were performed with solutions at a fixed temperature of 25 °C.

tricot spacer also reduced osmotic water permeation. Therefore, optimizing the thickness and porosity (or opening size) of the tricot spacer is essential for smooth flow circulation and good membrane support. To reduce the pressure drop of the feed-solution side, a new internal flow channel design of the membrane envelope should be considered. Furthermore, because the dilution of the draw solution was severe in the SW PRO membrane module due to water permeation, the osmotic pressure difference would be significantly reduced along a pressure vessel. Therefore, when several modules in a pressure vessel are connected in series, both solution streams should be counter-current flow.

■

ASSOCIATED CONTENT

S Supporting Information *

Details on the SW PRO membrane module; details on the determination of PRO membrane transport properties in a flatsheet membrane coupon; SEM micrographs of a TFC-PRO membrane with PET fabric layer; micrographs of feed and draw-solution channel spacers in an SW PRO membrane module; comparison of flow paths of the draw and feed solutions in an SW FO and an SW PRO module; schematic diagram of the PRO experimental unit; comparison of water permeability coefficient (A) and salt permeability coefficient (B) in an SW PRO membrane module and a flat-sheet PRO membrane coupon; comparison of water flux (JW) and water permeability coefficient (A) in three types of SW membrane modules; and the change of draw solution concentration by dilution resulting from osmotic water permeation with time. This material is available free of charge via the Internet at http://pubs.acs.org.

■

REFERENCES

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected], Phone: +82-42-868-7397. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors acknowledge the support of the Eco-Ener Plant Program (10034709) through the Korea Evaluation Institute of 2972

dx.doi.org/10.1021/es304060d | Environ. Sci. Technol. 2013, 47, 2966−2973

Environmental Science & Technology

Article

(21) Schock, G.; Miquel, A. Mass transfer and pressure loss in spiral wound modules. Desalination 1987, 64, 339−352. (22) Da Costa, A. R.; Fane, A. G.; Wiley, D. E. Spacer characterization and pressure drop modeling in spacer-filled channels for ultrafiltration. J. Membr. Sci. 1994, 87, 79−98. (23) Van Gauwbergen, D.; Baeyens, J. Macroscopic fluid flow conditions in spiral-wound membrane elements. Desalination 1997, 110, 287−299. (24) Li, Y.-L.; Tung, K.-L. The effect of curvature of a spacer-filled channel on fluid flow in spiral-wound membrane modules. J. Membr. Sci. 2008, 319, 286−297. (25) Guillen, G.; Hoek, E. M. V. Modeling the impacts of feed spacer geometry on reverse osmosis and nanofiltration processes. Chem. Eng. J. 2009, 149, 221−231. (26) Tiraferri, A.; Yip, N. Y.; Phillip, W. A.; Schiffman, J. D.; Elimelech, M. Relating performance of thin-film composite forward osmosis membranes to support layer formation and structure. J. Membr. Sci. 2011, 367, 340−352. (27) Johnson, J.; Busch, M. Engineering aspects of reverse osmosis module design. Desalination Water Treat. 2010, 15, 236−248. (28) Mehta, G. D. Further results on the performance of present-day osmotic membranes in various osmotic regions. J. Membr. Sci. 1982, 10, 3−19. (29) Yip, N. Y.; Elimelech, M. Performance limiting effects in power generation from salinity gradients by pressure retarded osmosis. Environ. Sci. Technol. 2011, 45, 10273−10282. (30) Yangali-Quintanilla, V.; Li, Z.; Valladares, R.; Li, Q.; Amy, G. Indirect desalination of Red Sea water with forward osmosis and low pressure reverse osmosis for water reuse. Desalination 2011, 280, 160− 166. (31) Fritzmann, C.; Lowenberg, J.; Wintgens, T.; Melin, T. State-ofthe-art of reverse osmosis desalination. Desalination 2007, 216, 1−76. (32) Greenlee, L. F.; Lawler, D. F.; Freeman, B. D.; Marrot, B. Moulin, P. Reverse osmosis desalination: Water sources, technology, and today’s challenges. Water Res. 2009, 43, 2317−2348. (33) Kim, Y. C.; Elimelech, M. Potential of osmotic power generation by pressure retarded osmosis using seawater as feed solution: Analysis and experiments. J. Membr. Sci. 2013, 429, 330−337. (34) Loeb, S. Production of energy from concentrated brines by pressure-retarded osmosis. I. Preliminary technical and economic correlations. J. Membr. Sci. 1976, 1, 49−63. (35) Loeb, S.; Hessen, F. V.; Shahaf, D. Production of energy from concentrated brines by pressure-retarded osmosis. II. Experimental results and projected energy costs. J. Membr. Sci. 1976, 1, 249−269.

2973

dx.doi.org/10.1021/es304060d | Environ. Sci. Technol. 2013, 47, 2966−2973