Minimizing the Instant and Accumulative Effects of ... - ACS Publications

Jul 30, 2013 - ABSTRACT: We have investigated the instant and accumu- lative effects of salt permeability on the sustainability of high power density ...
1 downloads 0 Views 4MB Size
Download PDF
Article pubs.acs.org/est

Minimizing the Instant and Accumulative Effects of Salt Permeability to Sustain Ultrahigh Osmotic Power Density Sui Zhang and Tai-Shung Chung* Department of Chemical and Biomolecular Engineering, National University of Singapore, Singapore 117576 S Supporting Information *

ABSTRACT: We have investigated the instant and accumulative effects of salt permeability on the sustainability of high power density in the pressure-retarded osmosis (PRO) process experimentally and theoretically. Thin-film composite (TFC) hollow-fiber membranes were prepared. A critical wall thickness was observed to ensure sufficient mechanical stability and hence a low salt permeability, B. The experimental results revealed that a lower B was essential to enhance the maximum power density from 15.3 W/m2 to as high as 24.3 W/m2 when 1 M NaCl and deionized water were feeds. Modeling work showed that a large B not only causes an instant drop in the initial water flux but also accelerates the flux decline at high hydraulic pressures, leading to reduced optimal operating pressure and maximal power density. However, the optimal operating pressure to harvest energy can be greater than one-half of the osmotic pressure gradient across the membrane if one can carefully design a PRO membrane with a large water permeability, small B value, and reasonably small structural parameter. It was also found that a high B accumulates salts in the feed, leads to the oversalinization of the feed, and largely lowers both the water flux and power density along the membrane module. Therefore, a low salt permeability is highly desirable to sustain high power density not only locally but also throughout the whole module.

■

INTRODUCTION

From 2005, the attention on forward-osmosis (FO) membranes has been exponentially increased, and breakthroughs have been made on the membrane structure to minimize the ICP effects and ultimately enhance the FO water flux by more than 10 times compared to traditional RO membranes.12−14 In 2009, Statkraft built the first PRO prototype plant in Norway and demonstrated power densities of less than 1.5 W/m2 by employing cellulose acetate membranes.11,15 Later, lab-scale experiments using Hydration Technology Innovations (HTI) FO membranes produced a maximum power density of 5.06 W/m2 at around 9.7 bar when concentrated brine (1.03 M NaCl) was used as the draw solution.16 The thin-film composite (TFC) hollow-fiber membranes fabricated by Chou et al. can achieve a power density of 10.6 W/m2 using 1 M NaCl seawater brine and 40 mM NaCl wastewater as feeds, but the fibers were broken at 10 bar.17 Recently, fundamental studies and solid improvements on membrane stability and permeability were reported by Zhang et al. and Li et al. by modifying the physicochemical properties of the support layers and post-treating the TFC layer.18,19 The subsequent TFC flat-sheet and hollow-fiber membranes

To mitigate the dependence on crude oil and reduce greenhouse gas emissions, research on alternative renewable energy sources has received worldwide attention for sustainable development. Pressure-retarded osmosis (PRO) is an environmentally friendly membrane-based process to harvest the salinity gradient energy from seawater or concentrated brine1−3 in which a semipermeable membrane is placed between two solutions of different salinity. Because of the osmotic pressure difference across the membrane, water permeates naturally from the dilute solution to the concentrated solution (referred to as the draw solution hereafter) and increases the hydraulic pressure in the draw solution compartment. The pressurized permeate then drives the hydroturbine and converts the energy to electricity. The global osmotic energy is estimated to be 2.6 TW per year, with even more expected if seawater brine is taken into account.3,4 Early studies in the 1970s investigated the technical and economic feasibility of PRO for power generation,1,5,6 but very low energy was obtained when reverse-osmosis (RO) membranes were employed resulting from the internal concentration polarization (ICP).7−9 It has been calculated that for the process to be commercially viable, the minimum power density per unit membrane area should reach 3 or 5 W/ m2 depending on the use of hollow-fiber or flat-sheet membranes, respectively.10,11 © 2013 American Chemical Society

Received: Revised: Accepted: Published: 10085

June 18, 2013 July 29, 2013 July 30, 2013 July 30, 2013 dx.doi.org/10.1021/es402690v | Environ. Sci. Technol. 2013, 47, 10085−10092

Environmental Science & Technology

Article

demonstrated a peak power density of up to 12.0 W/m2 at 15 bar20 and 14 W/m2 at 16 bar, respectively, using 1 M NaCl and deionized water as feeds.21 Meanwhile, Song et al. made a flatsheet TFC membrane based on nanofiber substrates and obtained a power density of 15.2 W/m2 at 15.2 bar by employing 1.06 M NaCl and 0.9 mM NaCl as feeds.22 Thus far, the power density of almost all reported membranes falls below 16 W/m2. Although it is well known that a high salt permeability decreases the water flux in the initial state of PRO processes (i.e., at no hydraulic pressure),23 no special emphasis has been given on the effects of salt permeability on the power density at high pressures and the overall PRO productivity. The current publications on PRO membranes either did not provide the salt-reverse fluxes in high pressure PRO processes16,17,20,21 or simply presented quite high salt-reverse fluxes of up to several hundred gm2 h−1 at high pressures.22 Yip et al.23 developed a method to simultaneously increase the water permeability, A, and salt permeability, B, of the TFC membranes and expected 9.21 W/m2 using seawater and river water as feeds by balancing the A and B values. However, no in-depth analysis of the effects of salt permeability on high-pressure PRO processes was reported. Most recently, Zhang et al.24 have fabricated hollow-fiber membranes with an ultrahigh power density and a low salt-reverse flux to water flux ratio of less than 1 g/L at 20 bar, but the underlying reasons for this high PRO performance have not been fully explored. In this study, both the experimental and theoretical analyses of the instant and accumulative effects of salt permeability on PRO membranes within the full range of operating hydraulic pressures are presented. TFC hollow-fiber membranes with different salt permeabilities are fabricated, and the influences of salt permeability on the initial and high-pressure power density in the PRO process are carefully examined. A modeling work is carried out to understand the different aspects of the instant impact of salt permeability. Furthermore, a simulation of a 4 in. module reveals the importance of salt accumulation on the average power density of the whole system. This study may provide insightful perspectives on the sustainability of high osmotic power density in the PRO process.

permeability B (LMH) were measured under the RO mode following ref 28. Osmotic Power Generation by PRO Processes. The PRO tests were conducted using a lab-scale cross-flow PRO set up.18 A variable-speed gear pump (Cole-Palmer, Vernon Hills, IL) was utilized to recirculate the feed solution (DI water) through the shell side of the hollow fibers at 0.2 L/min, and a high-pressure hydra cell pump was employed to recirculate the draw solution (1 M NaCl) through the lumen side at 0.15 L/ min. First, the burst pressure of the TFC membranes at which the direction of the water flux reversed was determined by gradually increasing the hydraulic pressure in the draw solution during PRO tests. Second, the as-prepared TFC hollow-fiber modules were tested in the PRO process at no hydraulic pressure and then stabilized for 20 min at their highest stable pressures (e.g., 17.3, 20.0, and 23.0 bar for fibers A, B, and C, respectively). Third, their PRO performance was conducted at various hydraulic pressures from 0 bar to their corresponding highest stable pressures. Subsequently, at least one hysteretic PRO test was carried out for each membrane, namely, the hydraulic pressure was increased and then decreased to ensure good reproducibility and reversibility of the PRO data. The water flux (Jw) was determined by monitoring the weight changes of the feed solution, and the salt reverse flux (Js) was calculated on the basis of the conductivity measurements. The power density (W) can be obtained from the product of the water flux and applied hydraulic pressure in the draw solution: W = Jw Δp

(1)

■

THEORETICAL BASIS Mass Transport Across the Membrane in the PRO Process. The mass transport across the membrane in the PRO process has been discussed in detail elsewhere.14,15,18 In the ideal case, the water flux, Jw, is related to the water permeability, A, osmotic pressure difference, Δπ, and hydraulic pressure, Δp, in the following way: Jw = A(Δπ − Δp)

■

(2)

Combining eqs 1 and 2 followed by differentiation indicates that the maximum power density is achieved at Δpm = 1/2 Δπ. Here, the optimal operating pressure Δpm is defined as the hydraulic pressure where the maximum power density is reached. However, in the real case, Jw is highly affected by the saltreverse flux and concentration polarization as described in eq 3

EXPERIMENTAL SECTION Fabrication of PES TFC Hollow-Fiber Membranes. A wet−wet spinning process employing coextrusion technology through a dual-layer spinneret was conducted to spin the poly(ether sulfone) PES hollow-fiber membranes as the support.25,26 The spinning conditions for the PES hollowfiber supports are listed in Table S1. The detailed procedures for the dope preparation, spinning process, and post treatment can be found in the Supporting Information. After post-treatment by a 50 wt % glycerol aqueous solution for 2 days followed by air drying, the hollow fibers were made into modules before interfacial polymerization between mphenylenediamine (MPD) and trimesoyl chloride (TMC) monomers. The interfacial polymerization was conducted at the inner surface of the hollow-fiber supports. The details are provided in the Supporting Information. Characterizations of PES Hollow-Fiber Supports and TFC Membranes. The burst pressure, pore size, pore-size distribution, porosity, morphology, and mechanical strength of the hollow-fiber supports were characterized on the basis of the procedures in the Supporting Information.27,28 The water permeability A (LMH/bar), salt rejection R (%), and salt

Jw = A

1+

B Jw

Jw S

( ) − AΔp (exp( ) − 1)

πD,b − πF,b exp

D

Jw S D

(3)

where D is the solute diffusion coefficient, S is the structural parameter, and πD,b and πF,b are the bulk salt osmotic pressures in the draw and feed solutions, respectively. The structural parameter S is determined by the porosity, ε, tortuosity, τ, and thickness, χ, of the membrane support: S=

τχ ε

(4)

Salt-reverse flux Js can be expressed as a function of Jw using van’t Hoff factor i18: 10086

dx.doi.org/10.1021/es402690v | Environ. Sci. Technol. 2013, 47, 10085−10092

Environmental Science & Technology ⎞ B ⎛ Jw ⎜ + Δp⎟ ⎠ iRT ⎝ A

Js =

Article

(5)

Notice that from eq 3 the water flux is inversely related to the salt permeability B if B is large. The salt-reverse flux causes an immediate drop in the effective osmotic driving force across the membrane and leads to a decreased water flux. Mass Variation along the Flow within the Membrane Module. As illustrated in Figure S1B, the salt concentration in the feed and draw solutions are changing along with the solution flow in the membrane module because of the saltreverse flux and water permeation. In the feed channel, the volumetric flow rate at the outlet (VF,o) is lower than its inlet counterpart (VF,i) because water is permeated continuously from the feed to the draw solution. Meanwhile, salt is permeating into the feed. As a result, the salt concentration at the feed outlet, CF,o, is higher than CF,i at the inlet. Similarly, in the draw channel, VD,o > VD,i and CD,o < CD,i. Consequently, the water flux along the module is not a constant but changes as a function of the location. At location x of the module, the salt concentration and volume of the feed and draw solution could be determined using the following equations d(C F, xVF, x) dx d(C D, xVD, x) dx dVF, x dx dVD, x dx

= Js l

(6)

= Js l

(7)

= −Jw l

(8)

= −Jw l

(9)

Figure 1. The cross section and surface morphologies of the PES hollow-fiber support (representative images of fiber C).

the magnified image. The inner surface is relatively dense with small pores, and the outer surface is quite porous. Table S3 shows that the inner surfaces of all fibers have similar pore sizes and pore-size distributions, but their overall porosity decreases slightly from fiber A to C, possibly because of the retarded outflow of solvents from the dope during phase inversion and the reduced ratio of n-methylpyrrolidone (NMP) over the polymer dope at the outer channel when the dope flow rate is increased. As a result, the maximum tensile strength is increased. Because burst pressure is related to the maximum tensile strength, outer diameter, and wall thickness on the basis of Barlow’s formula,24,29 the burst pressure of the hollow-fiber supports can be estimated as presented in Table S3, which follows the same trend as the experimental values. In general, the burst pressure is enhanced from 18.5 to 22.5 bar for fibers A, B, and C. After interfacial polymerization and stabilization, the resultant TFC membranes exhibit the same trend as that shown in Table 1. Their burst pressures increase with increasing fiber-wall thickness. It is noteworthy that the burst pressure of fiber C is 23.8 bar, which outperforms all of the PRO membranes reported so far. As illustrated in Figure S3, the initial water flux at Δp = 0 bar in the PRO process decreases slightly from fibers A to C, and their initial saltreverse fluxes are almost the same. After stabilization at their highest stable pressures, the water flux and salt-reverse flux of all fibers increase, which is most likely due to the cracking of defective spots and the rearrangement of polymer chains at high pressures, especially at the interface of the polyamide thin film and the support layers. The most dramatic increase in saltreverse flux is observed for fiber A, which possesses the weakest mechanical strength among the three. The salt-reverse fluxes for fibers B and C are almost the same, suggesting that above the critical thickness of around 200 μm, the salt rejection of the fibers is stabilized. Table 1 presents the A, B, and S values of the stabilized fibers. Impressively, small B values are found for fibers B and C. However, fiber C possesses the largest structural parameter among the three because of its relatively low porosity and large wall thickness, which explains its lower stabilized water flux in the PRO process. The PRO performance at different hydraulic pressures using 1 M NaCl as the draw solution and DI water as the feed is plotted in Figure 2. Interestingly, despite fibers A and B show a similar water flux of around 65 LMH at Δp = 0 bar, their changes against hydraulic pressure are remarkably different. Fiber B has a mild water flux change as a function of pressure, whereas fiber A has a sharp flux decrease with an increase in pressure. As a result, fiber B produces a maximum power

where l is the overall perimeter of the cross section of all fibers within the module. Js and Jw can be obtained by eqs 3 and 5, where πD,b = iRTCD,x and πF,b = iRTCF,x. Notice that Js and Jw are functions of location x as well. The overall length of the module is L0. When the volumetric flow rate and concentration of both the draw and feed solutions at the inlet are known, the mass distribution along the module with given parameters can be resolved. The stage cut of the feed solution, ϕ, which represents the percentage of the feedwater permeating across the membrane for power generation, can be defined as follows: ⎛ VF,o ⎞ ⎟⎟ × 100% ϕ = ⎜⎜1 − VF,i ⎠ ⎝

(10)

The feedwater permeating across the membrane is approximately the amount of pressurized water to drive the turbine in steady-state cases.

■

RESULTS AND DISCUSSION Characteristics and Performance of the PES TFC Hollow-Fiber Membranes. As shown in Figure S2 and Table S2, the morphology of hollow-fiber supports A, B, and C are similar to each other, but their wall thickness increases with the increment in dope flow rate during the spinning process. Figure 1 displays representative images of fiber C. Macroscopically, the cross section is half covered by fingerlike macrovoids and half covered by a spongelike structure. Microscopically, there is a thin spongelike dense layer of several hundred nanometers beneath the inner skin that can be observed from 10087

dx.doi.org/10.1021/es402690v | Environ. Sci. Technol. 2013, 47, 10085−10092

Environmental Science & Technology

Article

Table 1. Burst Pressure, Water Permeability, Salt Permeability, Structural Parameter, And Maximum Power Density of the Stabilized PES TFC Hollow-Fiber Membranesa fiber

burst pressure (bar)

A (LMH/bar)

NaCl rejection (%)

B (LMH)

S (μm; structural parameter)

Wmax (W/m2 feed = 10 mM NaCl)

Wmax (W/m2 feed = 40 mM NaCl)

A B C

18.0 21.0 23.8

4.0 ± 0.2 3.3 ± 0.2 3.8 ± 0.4

92.8 ± 0.1 97.6 ± 0.1 97.4 ± 0.3

1.65 0.31 0.39

350 450 490

15.0 24.0 23.2

12.3 19.2 17.7

The maximum power density was tested at 17.3, 20.0, and 23.0 bar for the TFC hollow fibers A, B, and C, respectively, using 1 M NaCl as the draw solution and synthetic river or brackish water as the feed. a

Figure 2. Water flux (a), power density (b), and salt/water flux ratio (c) of the PES TFC hollow-fiber membranes in the PRO processes for osmotic power generation. The draw solution is 1 M NaCl, and the feed solution is DI water.

Figure 3. Modeled relative water flux (Jw,Δp/Jw,Δp = 0 bar, panels a, b, and c) and power density (panels d, e, and f) in PRO processes as a function of water permeability, salt permeability, and structural parameter, respectively. A = 3.3 LMH/bar, B = 0.3 LMH, and S = 450 μm were used for the simulation when they are not variables. The draw solution is 1 M NaCl, and the feed solution is DI water.

density of 24.3 W/m2 at 20.0 bar, which is much higher than the 15.3 W/m2 of fiber A. In the meantime, fiber C shows a similar decreasing trend in water flux as that of fiber A as well as a slightly smaller maximum power density resulting from its lower water flux. The most probable reason for the different decreasing trends in water flux arises from the salt permeability. Instant Effects of Salt Permeability. It has been well understood from eq 3 that a high water flux in PRO is achieved

by large A, small B, and small S values. Previous reports have emphasized the requirements of a small structural parameter and balanced A and B values to ensure a high water flux.19−24,30 However, much less attention has been paid to the influences of A, B, and S on the decrease of water flux against rising hydraulic pressure. Figure 3 shows the modeled results of the normalized water flux (i.e., Jw,Δp/Jw,Δp= 0) and power density as a function of the hydraulic pressure at different A, B, or S values. (A 10088

dx.doi.org/10.1021/es402690v | Environ. Sci. Technol. 2013, 47, 10085−10092

Environmental Science & Technology

Article

Figure 4. Experimental and modeled water flux, power density, and normalized salt-reverse flux of the PES TFC hollow fiber A (panels a, b, and c) and fiber B (panels d, e, and f) in the PRO processes for osmotic power generation. The draw solution is 1 M NaCl, and the feed solution is DI water.

structural parameter reduces the water flux significantly through the ICP effect, and such an effect is significantly mitigated at high pressures when Jw is smaller. On the contrary, a large B value not only enhances the ICP effect on Jw but also imposes an adverse effect on Δpm. A comparison of the ideal, modeled, and experimental PRO results in Figure 4 clearly demonstrates the instant effects of salt permeability, B. Notice that despite its larger A and smaller S, the initial water flux at Δp = 0 of fiber A is only around the same as that of fiber B resulting from the instant effect of its high salt permeability. Furthermore, the Δpm of fiber A is apparently lower than fiber B. Ideally, the water flux would be reduced to 0 at around 46.5 bar. However, the modeled water flux for fiber A has a steep decline and approaches 0 at around 41 bar. A much milder flux drop is observed for fiber B because of its small B value. Interestingly, the modeled water flux for fiber B at 20.0 bar remains at 71.4% of its initial value, which is much higher than the 57.0% calculated for the ideal situation. Consequently, fiber B not only has a calculated Δpm of around 28 bar (i.e., 5 bar more than the ideal case) but also has a predicted peak power density of 29 W/m2. Clearly, a high salt permeability exerts two negative impacts on PRO performance: reducing the initial water flux at Δp = 0 and accelerating the water-flux decline against hydraulic pressure. In addition, the modeling results imply that the optimal operation pressure to harvest energy in PRO processes can be greater than one-half of the osmotic pressure gradient across the membrane if one can carefully design a PRO membrane with large A, small B, and reasonably small S values. Figure 4 also compares the modeled water flux and power density with the experimental data. The former is slightly higher than the latter. However, there is a big difference in the normalized salt-reverse flux (i.e., Js,Δp/Js,Δp= 0) between the modeled and experimental values, as shown in Figure 4c,f. Obviously, the experimental salt-reverse flux increases much faster than eq 5, indicating that the salt permeability is increasing rather than remaining as a constant at high pressures,

display of the actual water flux is presented in Figure S4.) For the simulation, A = 3.3 LMH/bar, B = 0.3 LMH, and S = 450 μm are used when they are not variables. One molar NaCl is employed as the draw solution, and DI water is the feed. The contour lines representing the relative water flux values of 0.5 are marked in Figure 3a−c. The horizontal dashed lines are drawn to guide the readers to the relative water flux against the hydraulic pressure at the same A, B, or S values, and the dashed vertical lines are used to indicate the relative water fluxes at 23 bar, which is around one-half of the osmotic pressure of the 1 M NaCl solution. Clearly, to keep a relative water-flux ratio of 0.5 for power generation when A or S is increased, one must significantly increase the hydraulic pressure in PRO to lower the dilution effect of the draw solutions or the ICP effect of the sublayer. However, the hydraulic pressure in PRO must be reduced to keep a water flux ratio of 0.5 when B is enhanced so that the effect of reverse-salt flux on the effective osmotic driving force is lessened, especially in the ranges where 1 < A < 4 LMH/bar, 1 < B < 3 LMH, or 150 < S < 400 μm, as guided by the white dashed lines drawn in the figure. Consequently, as shown in Figure 3d−f, the hydraulic pressure at which the maximum power density is reached (Δpm) is changing rapidly to either below 20 bar or almost 30 bar, depending on the A, B, and S values, rather than remaining constant at around 23 bar, as in the ideal case. The delay of flux decline by a large A can be understood from eq 3. A high initial water flux at Δp = 0 is achieved with a large water permeability and gradually drops as the hydraulic pressure increases. Because of the ICP, the flux-reduction term, exp(JwS/D), decreases exponentially with a smaller Jw. As a result, ICP initially reduces the water flux, Jw, significantly at low pressures but gradually lessens because Jw becomes smaller at high pressures. Because of the nonlinear relationship between Jw and Δp, the optimal operating pressure, Δpm, where the maximum power density occurs, deviates from Δp = 1/2 Δπ. Hence, Δpm can increase to around 30 bar. Similarly, a large 10089

dx.doi.org/10.1021/es402690v | Environ. Sci. Technol. 2013, 47, 10085−10092

Environmental Science & Technology

Article

Figure 5. Comparison of the modeled module length, feed concentration at the outlet, average water flux, and normalized average water flux (panels a, b, c, and d, respectively) at different stage cuts of the feed solution in the PRO process for hollow-fiber membranes with increasing salt permeability B.

which is probably due to the reversible polymeric-chain relaxation. As a result, the experimental water flux becomes lower than expectation. Notice also that the deviation between the modeling and experimental results of fiber A is much more severe. The experimental salt-reverse flux at 17.3 bar is twice the value of the modeled one. As a result, fiber A has more serious deviations in terms of power density between the modeled and experimental values than fiber B. Hence, a small B value is highly desirable to ensure a more stable salt permeability and a high water flux with an increase in hydraulic pressure. Accumulative Effects of Salt Permeability. It has been shown that a large salt permeability significantly reduces the power density at high hydraulic pressures because of its instant effect on the effective osmotic driving force. Moreover, in the long-term PRO run or in a more practicable large-membrane module, the permeated salts will accumulate in the feed solution, increase its salinity, and ultimately cause a reduction in the power density of the module. Figure S5 shows the decrease of water flux and power density as a function of salt concentration in the feed. The water flux declines very fast with an increase in feed concentration because of the reduction of the salinity gradient and ICP effects. When the draw solution is 1 M NaCl and the feed concentration reaches 264 mM, the water flux drops to less than one-fifth of the value obtained from DI water and so does the power density. Therefore, to maintain a high power density during PRO operations, feed solutions must not only contain minimal salts but also minimal salt accumulation from the draw solution to the feed. To calculate the salt accumulation in the feed solution and investigate its effect on PRO performance, the transport phenomena and salt distribution along a 4.0 in. (in diameter) module were simulated using eq 3 and eqs 5−9. The operating pressure is 20 bar, and more details on the module parameters can be found in the Supporting Information.

Figures S6−S8 depict the simulation results as a function of module length. It is quite interesting to notice that the water flux Jw varies greatly along the module, especially when L0 is large. At x = 0, the feed solution is DI water, whereas at x = L0, the feed solution becomes a salty solution because of the accumulation of the salt-reverse flux. Therefore, a lower water flux at x = L0 in comparison to that at x = 0 is mainly attributed to the enhanced salt concentration in the feed. Figure S6 shows that with a large L0 the water flux is slightly lower at the feed outlet. Such an effect is magnified in Figures S7 and S8 where salt permeabilities become higher and higher. A very significant drop in water flux is found in Figure S8, which shows no more water flux across the membrane near the feed outlet. In this case, the feed concentration increases to as high as 450 mM at the feed outlet. Figure 5 compares the modeled module length, feed concentration at the outlet, and average water flux Jw and normalized Jw as a function of the feed stage cut defined as in eq 10. The average water flux is calculated on the basis of eq 11:

Jw =

VF,i − VF,o lL0

(11)

Clearly, a much longer module is required to achieve the necessary stage cut of the feed if the salt permeability B is large. If B is too large as 3.6 LMH in this study, then the stage cut can never reach 90%, even though the module is 120 cm long, because the water flux has already dropped to 0 below that. Meanwhile, the feed concentration at the outlet increases with increasing the stage cut. The increment is not significant when B is 0.3 LMH, but it becomes more serious as B grows bigger. Figure 5c shows the average water flux as a function of the stage cut. A smaller Jw is consistently observed when B is bigger because of the combined instant and accumulative effects of salt permeability. Figure 5d depicts the normalized water flux at different stage cuts. The lower normalized average water flux at 10090

dx.doi.org/10.1021/es402690v | Environ. Sci. Technol. 2013, 47, 10085−10092

Environmental Science & Technology

Article

(7) Mehta, G. D.; Loeb, S. Performance of Permasep B-9 and B-10 membranes in various osmotic regions and at high osmotic pressures. J. Membr. Sci. 1979, 4, 335−349. (8) Jellinek, H. H.; Masuda, H. Osmo-power. Theory and performance of an osmo-power pilot plant. Ocean Eng. 1981, 8, 103−128. (9) McCutcheon, J. R.; Elimelech, M. Influence of membrane support layer hydrophobicity on water flux in osmotically driven membrane processes. J. Membr. Sci. 2008, 318, 458−466. (10) Skilhagen, S. E. Osmotic powera new, renewable energy source. Desalin. Water Treat. 2010, 15, 271−278. (11) 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. (12) McCutcheon, J. R.; Elimelech, M. Modeling water flux in forward osmosis: implications for improved membrane design. AIChE J. 2007, 53, 1736. (13) Wang, K. Y.; Chung, T. S.; Qin, J. J. Polybenzimidazole (PBI) nanofiltration hollow fiber membranes applied in forward osmosis process. J. Membr. Sci. 2007, 300, 6−12. (14) Wang, R.; Shi, L.; Tang, C. Y.; Chou, S.; Qiu, C.; Fane, A. G. Characterization of novel forward osmosis hollow fiber membranes. J. Membr. Sci. 2010, 355, 158−167. (15) Gerstandt, K.; Peinemann, K.-V.; Skilhagen, S. E.; Thorsen, T.; Holt, T. Membrane processes in energy supply for an osmotic power plant. Desalination 2008, 224, 64−70. (16) 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. (17) Chou, S.; Wang, R.; Shi, L.; She, Q.; Tang, C. Y.; Fane, A. G. Thin-film composite hollow fiber membranes for pressure retarded (PRO) process with high power density. J. Membr. Sci. 2012, 389, 25− 33. (18) Zhang, S.; Fu, F. J.; Chung, T. S. Substrate modifications and alcohol treatment on thin film composite membranes for osmotic power. Chem. Eng. Sci. 2013, 87, 40−50. (19) Li, X.; Zhang, S.; Fu, F. J.; Chung, T. S. Deformation and reinforcement of thin-film composite (TFC) polyamide-imide (PAI) membranes for osmotic power generation. J. Membr. Sci. 2013, 434, 204−217. (20) Han, G.; S. Zhang, Li, X.; Chung, T. S. High performance thin film composite membranes: Break the bottleneck of power output to harvest renewable salinity-gradient energy via pressure retarded osmosis. J. Membr. Sci. 2013, 440, 108−121. (21) Han, G.; Wang, P.; Chung, T. S. Highly robust thin-film composite pressure retarded osmosis (PRO) hollow fiber membranes with outstanding power density for renewable salinity-gradient energy generation. Environ. Sci. Technol. 2013, 47, 8070−8077. (22) Song, X.; Liu, Z.; Sun, D. D. Energy recovery from concentrated seawater brine by thin-film nanofiber composite pressure retarded osmosis membranes with high power density. Energy. Environ. Sci. 2013, 6, 1199−1210. (23) 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. (24) Zhang, S.; Sukitpaneenit, P.; Chung, T. S. Design of robust hollow fiber membranes with high power density for osmotic energy production. Chem. Eng. J., submitted for publication. (25) Liu, Y.; Koops, G. H.; Strathmann, H. Characterization of morphology controlled polyethersulfone hollow fiber membranes by the addition of polyethylene glycol to the dope and bore liquid solution. J. Membr. Sci. 2003, 223, 187−199. (26) Sukitpaneenit, P.; Chung, T. S. High performance thin-film composite forward osmosis hollow fiber membranes with macrovoidfree and highly porous structure for sustainable water production. Environ. Sci. Technol. 2012, 46, 7358−7365.

a higher stage cut can be attributed to the increased feed concentration alone when the recovery rate or the degree of the dilution is the same. In other words, the normalized average water flux in the case of B = 0.3 LMH drops to 0.9 at the recovery of 90%, whereas it drops severely to less than 0.5 at close to 90% recovery when B = 3.6 LMH because of the more serious accumulation of salts in the feed solution. Because power density is proportional to the water flux, the average power density of the module at a high stage cut is reduced by more than half if a high salt permeability is present. To sustain a high power density during the PRO process and to achieve a high stage cut, a small salt permeability is highly needed. Taking into consideration that the experimental salt-reverse flux is usually several times higher than predicted, as discussed from Figures 4c,f, the requirement to minimize salt permeability becomes even more stringent.

■

ASSOCIATED CONTENT

S Supporting Information *

Materials and experimental methods; dimension, morphology, and transport characteristics of hollow-fiber supports and TFC membranes; and additional modeling results for the instant and accumulative effects of salt permeability. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

* Tel: +65-65166645; Fax: +65-67791936; E-mail: chencts@ nus.edu.sg. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This research was funded by the Singapore National Research Foundation under its Competitive Research Program for the project titled “Advanced FO Membranes and Membrane Systems for Wastewater Treatment, Water Reuse and Seawater Desalination” (grant no. R-279-000-336-281) and was also supported by the Singapore National Research Foundation under its Environmental & Water Technologies Strategic Research Programme, administered by the Environment & Water Industry Programme Office (EWI) of the PUB under the project titled “Membrane Development for Osmotic Power Generation, Part 1. Materials Development and Membrane Fabrication” (1102-IRIS-11-01) and NUS grant no. R-279-000381-279.

■

REFERENCES

(1) Norman, R. S. Water desalination: A source of energy. Science 1974, 186, 350−352. (2) Thorsen, T.; Holt, T. The potential for power production from salinity gradients by pressure retarded osmosis. J. Membr. Sci. 2009, 335, 103−110. (3) Wick, G. L.; Schmitt, W. R. Prospects for renewable energy from the sea. Mar. Technol. Soc. J. 1997, 11, 16−21. (4) Jones, A. T.; Finley, W. In Recent developments in salinity gradient power, Oceans Proceedings, San Diego, CA, Sept. 22−26, 2003; The Marine Tehcnology Society: Piscataway, NJ, 2003; Vol. 1, pp 2284− 2287. (5) 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. (6) Loeb, S.; Norman, R. S. Osmotic power plant. Science 1975, 189, 654−655. 10091

dx.doi.org/10.1021/es402690v | Environ. Sci. Technol. 2013, 47, 10085−10092

Environmental Science & Technology

Article

(27) Zhang, S.; Wang, K. Y.; Chung, T. S.; Chen, H. M.; Jean, Y. C. Molecular design of the cellulose ester-based forward osmosis membranes for desalination. Chem. Eng. Sci. 2011, 66, 2008−2018. (28) Zhang, S.; Wang, K. Y.; Chung, T. S.; Chen, H. M.; Jean, Y. C.; Amy, G. Well-constructed cellulose acetate membranes for forward osmosis: Minimized internal concentration polarization with an ultrathin selective layer. J. Membr. Sci. 2010, 360, 522−532. (29) Machinery’s Handbook 25; Industrial Press: New York, 1996; pp 268. (30) Sivertsen, E.; Holt, T.; Thelin, W.; Brekke, G. Pressure retarded osmosis efficiency for different hollow fiber membrane module flow configurations. Desalination 2013, 312, 107−123. (31) Lee, K. L.; Baker, R. W.; Lonsdale, H. K. Membrane for power generation by pressure retarded osmosis. J. Membr. Sci. 1981, 8, 141− 171.

10092

dx.doi.org/10.1021/es402690v | Environ. Sci. Technol. 2013, 47, 10085−10092