Conceptual demonstration of novel closed-loop pressure retarded osmosis process for sustainable osmotic energy generation

Applied Energy 132 (2014) 383–393

Contents lists available at ScienceDirect

Applied Energy journal homepage: www.elsevier.com/locate/apenergy

Conceptual demonstration of novel closed-loop pressure retarded osmosis process for sustainable osmotic energy generation Gang Han, Qingchun Ge, Tai-Shung Chung ⇑ Department of Chemical and Biomolecular Engineering, National University of Singapore, Singapore 117576, Singapore

h i g h l i g h t s

g r a p h i c a l a b s t r a c t

 One novel closed-loop PRO process

was demonstrated for osmotic energy capture for the first time.  Highly robust and permeable PRO hollow fiber membrane was developed.  Hydroacid complex draw solute with high water flux and negligible reverse solute flux was synthesized. 2  Power density of 16.2 W/m with an ultralow reverse solute flux (Js/ Jw < 0.062 g L1) was achieved at 12 bar.

a r t i c l e

i n f o

Article history: Received 4 February 2014 Received in revised form 17 June 2014 Accepted 10 July 2014

Keywords: Closed-loop PRO Renewable osmotic energy Hydroacid complex draw solute Hollow fiber membrane

a b s t r a c t For the first time, one novel closed-loop pressure retarded osmosis (PRO) process promoted by an effective hydroacid complex draw solution has been demonstrated for harvesting the renewable salinity-gradient energy. The complex draw solute was molecularly constructed to possess unique characteristics of high osmotic pressure, large molecular size and relative low viscosity, and easy regeneration. Compared to conventional PRO processes, the newly developed closed-loop PRO process exhibits promising advantages of sustainable high power output, negligible internal concentration polarization and low membrane fouling, as well as no problems of feed water pretreatment and brackish water discharge. Employing a highly permeable (A = 4.30 LMH/bar) and selective (B = 0.47 LMH) thin film composite PRO hollow fiber membrane, a power density of 16.2 W/m2 can be achieved with an ultralow reverse solute flux (Js/ Jw < 0.062 g L1) at 12 bar when using 1 M complex draw solution and deionized water as feeds. The diluted complex draw solution can be regenerated via a solvent precipitation process, and the outstanding PRO performance could be almost fully recovered. We believe the newly developed closed-loop PRO process shows great potential for salinity-gradient energy capture, although the specific benefits have to be fully defined through energy or cost analysis. Ó 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +65 65166645; fax: +65 67791936. E-mail address: [email protected] (T.-S. Chung). http://dx.doi.org/10.1016/j.apenergy.2014.07.029 0306-2619/Ó 2014 Elsevier Ltd. All rights reserved.

384

G. Han et al. / Applied Energy 132 (2014) 383–393

1. Introduction

2. Experimental

By using a semi-permeable membrane to control the mixing of solutions with different salinities, pressure retarded osmosis (PRO) has emerged as one of the most effective processes to harvest the renewable salinity-gradient energy (or osmotic power) in terms of electricity without causing adverse environmental impacts [1–7]. The global estimation of this osmotic power is 2.6 TW/year, and the value could be higher if the osmotic energy harvested from the high salinity retentate such as reverse osmosis (RO) desalination plants is taken into account [3,8]. However, utilization of PRO for practical power supply is still facing many difficulties such as the absence of the specialized PRO membrane, internal concentration polarization (ICP), membrane fouling, feed water pretreatment and energy consumption in the energy recovery device [3–8]. Recently, lots of effective flat-sheet and hollow fiber PRO membranes have been successfully developed [9–16]. Song et al. made a flat-sheet thin film composite (TFC) PRO membrane based on nanofiber substrate and obtained a power density of 15.2 W/m2 at 15.2 bar by employing 1.06 M NaCl as the draw solution and 0.9 mM NaCl as the feed [11]. Zhang et al. invented one TFC PRO hollow fiber membrane which can produce a power density of 24.0 W/m2 at 20.0 bar using 1 M NaCl as the concentrated brine [14]. Compared to flat-sheet membranes, the self-supported hollow fiber PRO membranes are of great interest due to the high surface to volume ratio and spacer-free module fabrication. Not only could this minimize the membrane deformation [17], but also eliminate the extra energy loss in the feed flow channel of flatsheet membrane modules [17–19]. However, some disadvantages have been observed in conventional PRO power generation processes. One of them is the requirement of effectively pretreating the feed waters in order to prevent membrane fouling, which would significantly increase the process costs [8]. Another disadvantage arises from the low osmotic pressure gradient between the natural feed water streams such as seawater and river water. As a result, there are no sufficiently high hydraulic pressures available for efficient power generation. In case the feed water contains a large amount of salts, the process efficiency would be diminished because of the dramatically amplified ICP and salt leakage [20]. The severe reverse salt flux across the PRO membrane is another shortcoming, which would induce other potential problems such as enhanced ICP, reduced driving force, exacerbated membrane fouling, and increased downstream processing. Furthermore, it would be difficult or impossible to operate the conventional PRO process in countries or areas with severe water shortages. These negative facts on the conventional PRO process have triggered the exploration of other novel PRO processes such as the closed-loop PRO [20–22]. Closed-loop PRO is a promising process for salinity gradient energy generation which still employs osmotic pressure as a medium but is no longer from natural streams. Therefore, the aforementioned shortcomings faced by the conventional PRO could be minimized or overcome by optimizing the draw solution and feed solution. The primary challenges to current closed-loop PRO process are the low power output and the low thermal efficiency since intensive energy input may be required for draw solution regeneration. In this study, a novel closed-loop PRO process is conceptually demonstrated to harvest the osmotic energy in terms of electricity. A molecularly designed hydroacid complex draw solution and freshwater resources with low salinities are used as the working fluids to maximize the membrane mass transfer. The complex draw solution possesses a high osmotic pressure, negligible reverse solute flux, and easy approaches in regeneration. The feasibility of the newly developed PRO process for power generation is studied in terms of water flux, power density and reverse solute flux by using a robust highly permeable TFC PRO hollow fiber membrane.

2.1. Materials and chemicals MatrimidÒ 5218 purchased from Vantico Inc. was used as the polymer material to fabricate the hollow fiber membrane substrate. N-methyl-2-pyrrolidone (NMP, >99.5%) and diethylene glycol (DEG, >99.0%) from Merck were utilized as the solvent and additive in the membrane fabrication, respectively. Polyethylene glycol with different molecule weights was ordered from Sigma– Aldrich to characterize the pore characteristics and molecular weight cut-off (MWCO) of the hollow fiber substrate. A 50/ 50 wt.% mixture of glycerol (Industrial grade, Aik Moh Pains & Chemicals Pte. Ltd., Singapore) and de-ionized water was prepared to post-treat the hollow fiber substrate before drying. Trimesoyl chloride (TMC, >98%) and m-phenylenediamine (MPD, >99%) from Sigma–Aldrich were employed as the monomers for the interfacial polymerization reaction. Sodium dodecyl sulphate (SDS, >97%, Fluka) and triethylamine (TEA, >99%, Sigma–Aldrich) were used as additives, while hexane (>99.9%, Fisher Chemicals) as the solvent for the TMC solution. Fe(NO3)39H2O (99%), citric acid (99%) and NaOH (99%) were purchased from Sigma–Aldrich for the synthesis of hydroacid complex draw solute. Ethanol (EtOH, 99%) was purchased from Acros Organics to regenerate the complex draw solute. The deionized (DI) water was produced by a Milli-Q unit (Millipore) with a resistivity of 15 MX cm. 2.2. Fabrication of hollow fiber membrane substrate The hollow fiber membrane support for TFC PRO membranes was fabricated via a wet–wet phase inversion spinning process. In order to obtain the desirable membrane structure and morphology during membrane formation, the dual-bath coagulation spinning process using a single-layer spinneret was employed to effectively control the phase inversion [13,23]. Table 1 summarizes the detailed spinning conditions of the hollow fiber substrate. The hollow fiber spinning was repeated for three times to ensure the membrane reproducibility. For module fabrication, the hollow fiber membranes were assembled into a module holder which consists of two Swagelok stainless steel male run tees connected by a perfluoroalkoxy tube 3/8 in. in diameter. Both ends were sealed with a slow cure epoxy resin (KS Bond EP231, Bondtec). 2.3. Fabrication of the TFC PRO hollow fiber membrane A polyamide selective skin was formed on the inner surface (lumen side) of the hollow fiber substrate via interfacial polymerization. As illustrated in our previous work [24–26], a 2 wt.% MPD aqueous solution containing 0.5 wt.% TEA and 0.1 wt.% SDS was Table 1 Spinning conditions of hollow fiber membrane substrate. Spinning parameter

Hollow fiber membrane substrate

Polymer dope solution (wt.%) Bore-fluid solution (wt.%) Polymer dope flow rate (ml/min) Bore-fluid flow rate (ml/min) Air–gap length (cm) Take-up speed (m/min) External coagulant

16.9/17/65.3/0.8 Marimid/DEG/NMP/H2O H2O/NMP 70:30 1.5 0.75 2.0 2.5 (free fall) IPA/Water (60/40 wt.%) then Tap water dual bathes of coagulation Single-layer spinneret (1.2–0.68–0.48) Ambient (23 °C ± 2) 50/50 wt.% glycerol/water for 2 days

Spinneret dimension (mm) Spinning temperature (°C) Post treatment

G. Han et al. / Applied Energy 132 (2014) 383–393

firstly fed into the lumen side of hollow fibers for 5 min at a flow rate of 4.25 ml min1. The excess MPD residual droplets were then removed by purging a sweeping air for 4 min using a compressed air gun. Secondly, the 0.15 wt.% TMC hexane solution was brought into contact with the MPD saturated membrane inner surface for 3 min at a flow rate of 2.50 ml min1, leading to the formation of the polyamide skin. The resultant TFC membranes were purged with air for 30 s to remove the residual hexane solution and then stored in deionized water before further characterizations. 2.4. Synthesis of the hydroacid complex draw solute The Na5[Fe(C6H4O7)2] (Na–Fe–CA) hydroacid complex draw solute was chemically synthesized through a simple but efficient onestep reaction [27,28]. In a typical reaction, 20 mmol of Fe(NO3)39H2O and 40 mmol of citric acid were dissolved in deionized water (700 mL) and stirred for 3 h at 60 °C. Then the pH of the solution was adjusted to 7.0 by stepwise adding NaOH. After stirring at 60 °C for another 24 h, cold ethanol (EtOH) was added into the solution to precipitate the product. It was then purified 3 times from H2O/EtOH. The resultant yellow solid Na–Fe–CA was dried under vacuum (yield > 93%) and ready for characterization and performance tests. Elemental analysis: calcd: C 24.7, H 2.1, O 43.9%; found: C 25.1, H 2.0, O 43.5%. 2.5. Closed-loop PRO process As shown in Fig. 1, one novel closed-loop pressure retarded osmosis (PRO) process was developed to harvest the renewable osmotic energy. In the closed-loop PRO process, the draw solution was pressurized to a hydraulic pressure lower than its osmotic pressure; thus water stream spontaneously flowed across the membrane into the pressurized draw solution driven by the osmotic pressure gradient between the two solutions. The incoming water flow expanded the volume and pressure of the draw solution, and work can be produced by depressurizing the high pressure draw solution through a hydro-turbine. The draw solution was recycled back to the membrane module until its osmotic pressure decreases to a certain value. The diluted draw solution would be regenerated to recover its osmotic pressure. A certain amount of feed water was inputted into the process to maintain its flow rate.

385

In the current study, the Na–Fe–CA solution with a certain concentration was prepared and used as the draw solution, and deionized water or synthetic river water (10 mM NaCl) was employed as the feed. Comparing to the conventional draw solutes such as NaCl, the Na–Fe–CA complex can provide a higher osmotic pressure at the same molar concentration due to the formation of abundant coordinated hydrophilic groups with a multi-charged anion and Na+ cations in its aqueous solutions. Importantly, the complex possesses an expanded octahedral structure which leads to a negligible reverse solute flux and provides relatively easy approaches for regeneration [27]. One TFC PRO hollow fiber membrane with high robustness and water permeation flux was fabricated and used to control the mixing. In this study, the Na–Fe–CA draw solution was regenerated at the end of the PRO performance tests. The regenerated draw solution was further tested in the closed-loop PRO process to evaluate its performance.

2.6. Regeneration of the diluted hydroacid complex draw solution Different from conventional PRO processes, the Na–Fe–CA draw solution used in the closed-loop PRO is recycled instead of directly discharging it after use. However, the recycled draw solution will be continually diluted and its osmotic pressure decreases. In order to recover the osmotic pressure, the diluted draw solution needs to be re-generated or re-concentrated after certain cycles in the closed-loop PRO. Nanofiltration and membrane distillation have been proposed to regenerate the Na–Fe–CA draw solute [27]. However, in order to reduce the energy consumption, a solvent induced precipitation process was used to regenerate the Na–Fe–CA draw solute in this study. Due to its unique characteristics, the Na–Fe– CA draw solute cannot dissolve in alcohol such as ethanol, while ethanol is water mixable. Therefore, the water in the diluted Na–Fe–CA draw solution can be extracted by ethanol and the draw solute will be precipitated out of the water. The small amount of water/ethanol in the mixture can be further removed by evaporation and the Na–Fe–CA draw solute can be fully regenerated. In a typical re-concentration process, a certain amount of ethanol was added into the diluted Na–Fe–CA draw solution with a small volume ratio depending on the degree of the dilution. The osmotic pressure and PRO performance of the re-generated Na–Fe–CA draw solution can be almost fully recovered. The loss of the Na–Fe–CA

Freshwater input

Freshwater

Pump Pressure exchanger Hydroacid complexes draw solution

Power

Regeneration

Water Closed-loop PRO power generation process

Hydroacid complex draw solution

Fig. 1. Schematic of the closed-loop PRO process for power generation and the Na–Fe–CA hydroacid complex draw solute. The feeds consist of fresh water as the working fluid and the complex as draw solution.

386

G. Han et al. / Applied Energy 132 (2014) 383–393

draw solute during this process is very low (yield > 99.0%). In addition, the small amount of salts permeated from the feed solution can be removed during this re-generation process. 2.7. Characterizations 2.7.1. Morphology, mechanical properties, porosity, MWCO, pore size, pore size distribution of the hollow fiber membranes Membrane morphology was observed via a field-emission scanning electron microscope (FESEM JEOL JSM-6700LV). Membrane samples for FESEM were firstly freeze dried, then fractured in liquid nitrogen and sputtered with platinum by a Jeol JFC-1100E Ion Sputtering device. Membrane mechanical properties in terms of the maximum tensile stress, Young’s modulus and maximum tensile strain were determined by an Instron tensiometer (Model 5542, Instron Corp.) at room temperature. A constant elongation rate of 10 mm min1 with a starting gauge length of 25 mm was applied. At least ten fiber samples were tested and the average value for each hollow fiber membrane. Membrane porosity of the hollow fiber substrate was measured using a protocol described in our previous studies [14,26]. Firstly, the excess water in the lumen side and on the outer surface of the wet membrane was carefully and quickly removed. After that, the wet membranes were immediately weighed (m1, g), freeze dried overnight, and immediately re-weighed (m2, g). Since the densities of both water (qw, 1.00 g cm3) and MatrimidÒ (qp, 1.17 g cm3) are known, the overall porosity e (%) of the membrane was obtained as:

e¼

ðm1  m2 Þ=qw ðm1  m2 Þ=qw þ m2 =qp

ð1Þ

For each sample, 10 random measurements were carried out and the average value was reported. Pure water permeability (PWP) (L m2 h1 bar1 or LMH bar1) of the hollow fiber support was measured using a lab-scale nanofiltration process [13–15,26]. Deionized water was pumped into the lumen side of the fibers at the flow rate of 0.1 ml/min and pressurized at 2 bar for 30 min before the permeate was collected from the shell side. The PWP was calculated as:

PWP ¼

Q Am DP

ð2Þ

where Q is the water permeation volumetric flow rate (L h1), Am is the effective filtration area (m2), and DP is the trans-membrane pressure drop (bar). The pore size, pore size distribution, and molecular weight cutoff (MWCO) of the hollow fiber supports were measured by solute rejection experiments using neutral solutes made of polyethylene glycol (PEG) with a concentration of 200 ppm [13–15]. For each test, the feed solution was circulated at 1 bar until the variation of rejection was less than 2% before the concentrations of both feed and permeate solutions were measured. Between runs of different solutes the membrane was rinsed thoroughly with deionized water. The concentrations of the neutral solutes in the feed and permeate were measured via a total organic carbon analyzer (TOC ASI-5000A, Shimadzu, Japan). The measured feed (Cf) and permeate (Cp) concentrations were used to calculate the effective solute rejection coefficient R (%) as follows:

  Cp  100% R¼ 1 Cf

ð3Þ

The relationship between solute Stokes diameter (ds, nm) and molecular weight (Mw, g mol1) of the neutral solutes can be expressed as [15]:

For PEG : ds ¼ 33:46  1012  M 0:557

ð4Þ

When the solute rejection R is plotted versus ds on a log-normal probability paper, a straight line is yielded and the mean effective pore size lp (diameter) is found at R = 50%, and the geometric standard deviation rp is found as the ratio of ds at R = 84.13% and R = 50%. The molecular cut-off (MWCO) is defined as the solute molecular weight at R = 90%. Then the pore size distribution of the membrane can be expressed as following [13,15]:

" # 2 ðln dp  ln lp Þ dRðdp Þ 1 pffiffiffiffiffiffiffi exp  ¼ 2 ddp dp ln rp 2p 2ðln rp Þ

ð5Þ

2.7.2. Determination of mass transport characteristics of the TFC membranes Intrinsic water permeability (A), the salt rejection rate (Rs), and the salt permeability (B) of the TFC PRO hollow fiber membranes were obtained via a lab-scale circulating RO filtration apparatus [13–15,26]. A was determined from the pure water permeation fluxes under a trans-membrane pressure of 1 bar. Rs was tested by one feed water solution containing 200 ppm NaCl and determined according to the conductivity measurements of permeate and feed solutions at 1 bar. B was determined according to the solution–diffusion theory as follows [14,15]:

1  Rs B ¼ AðDP  DpÞ Rs

ð6Þ

where DP and Dp are the pressure difference and osmotic pressure difference across the membrane, respectively. 2.7.3. Osmotic pressure and viscosity of the draw solution The relative viscosity (gr) of Na–Fe–CA and NaCl draw solutions compared to deionized water was calculated as [27]:

gr ¼

g tq ¼ g 0 t 0 q0

ð7Þ

where t (s) is the elution time of the solution measured by a AVS 360 inherent viscosity meter, q (g mL1) is the density of the draw solution measured by a DMA 35 potable density meter, and t0 (s) and q0 (g mL1) are the elution time and density of deionized water, respectively. The osmotic pressures of the draw solutions were measured using a model 3250 osmometer (Advanced Instruments, Inc.) [27]. 2.8. PRO performance tests The membrane water permeation fluxes (Jw) and reverse solute fluxes (Js) at different cross membrane hydraulic pressures (DP) were measured using a lab-scale PRO setup [13–15,26]. A variable-speed gear pump (Cole–Palmer, Vernon Hills, IL) was utilized to recirculate the feed water (deionized water) through the shell side of the hollow fiber module at 0.2 L/min, and a high-pressure hydra cell pump was employed to recirculate and pressurize the draw solution through the lumen side at 0.1 L/min. A pressure relief valve after the membrane module and a needle valve prior to the module were installed to adjust the hydrostatic pressure and the cross-flow rate. Digital pressure transmitters and flow meters were mounted to monitor the hydrostatic pressure and the flow rate of the pressurized draw solution, respectively. Before the tests, the PRO hollow fibre membrane was pressurized at 15 bar in the lumen for 20 min to stabilize its performance. Then the PRO performance of the stabilized membrane was evaluated by rapidly increasing the DP from 0 to 12 bar. At each pressure, the PRO test was continued for at least 20 min at room temperature (23 °C).

G. Han et al. / Applied Energy 132 (2014) 383–393

The water permeation flux (Jw) was determined by measuring the weight change of the feed solution with a digital mass balance connected to a computer data logging system. The Jw (L m2 h1, abbreviated as LMH) was calculated from the volume change of the feed or draw solution [13–15].

Jw ¼

DV A m Dt

ð8Þ

where DV (L) is the permeation water collected over a predetermined testing duration Dt (h) and Am (m2) is the effective membrane area. The salt concentration in the feed water was determined from the conductivity measurement using a calibration curve for the single solute solution. The salt leakage, salt reverse-diffusion from the draw solution to the feed, Js (g m2 h1, abbreviated as gMH), was thereafter determined from the increase of the feed conductivity [13,27]:

Js ¼

DðC t V t Þ Am Dt

ð9Þ

where Ct and Vt are the salt concentration and volume of the feed at the end of the tests, respectively. In terms of energy production, the power density (W) was obtained by the product of Jw and DP across the membrane [13– 15]:

W ¼ J W  DP ¼ AðDp  DPÞ  DP

387

within the boundary layer at the hollow fiber inner surface and the pressure build up in the fiber which considerably reduces the effective osmotic driving force [25]. Fig. 2(c) further compares the relative viscosity (gr), to deionized water, of the Na–Fe–CA and NaCl draw solutions at different concentrations. The Na–Fe–CA draw solution shows relative higher viscosity than NaCl at the same molar concentration because the former has a much larger molecular weight and a bulkier structure than the latter. Therefore, the pressure build-up in the lumen side of the hollow fiber membrane is more significant for Na–Fe–CA, as shown in Fig. 2(d). However, this viscosity is still quite low and acceptable as the draw solution when comparing to other macromolecular draw solutes such as polyacrylic acid sodium (PAA–Na) [29]. The adverse viscosity effects from this draw solution on PRO performance may be not so severe. With an increase in Na–Fe–CA concentration, the solution viscosity and the pressure build-up in the fiber rapidly increases. For example, the pressure build-up jumps to 1.2 bar when the concentration increases to 1.5 M. Therefore, a Na–Fe–CA solution of 1 M was used as the draw solution in the closed-loop PRO process in order to minimize the negative viscosity effects on membrane external concentration polarization and pressure drop in the hollow fiber membrane. The unique characteristics of the Na–Fe–CA complex, in terms of large molecular size, high osmotic pressure, and relative low viscosity, make it a feasible draw solution for the proposed closed-loop PRO power generation process.

ð10Þ 3.2. Characteristics of the TFC PRO hollow fiber membrane

3. Results and discussion 3.1. Characteristics of the Na–Fe–CA draw solution The hydroacid complex is a class of coordination materials normally consisting of a metal center and several ionic ligands. The ligands can be freely designed to have certain amount of charge groups [27,28]. In this work, a ferric complex with hydroxyl acids of citric acid (CA) as ligands was synthesized as the draw solute for the closed-loop PRO process. Fig. 1 shows the representative structure of Na–Fe–CA complex. The coordination of the CA ligands to the Fe3+ metal center has been verified by the Fourier transform infrared spectroscopy (FTIR) and thermal gravimetric analysis (TGA) characterizations in our previous work [27]. The Na–Fe–CA draw solute is supposed to have a large molecular size because of its high molecular weight and the expanded octahedral configuration [27]. Given the importance of the osmolality and viscosity of the draw solution in PRO processes, the osmotic pressure (P) and FO water permeation flux (Jw), and relative viscosity (gr) of the prepared Na–Fe–CA draw solutions were measured as a function of concentration and compared with NaCl solutions. As shown in Fig. 2(a), the synthesized Na–Fe–CA draw solution possesses a higher osmotic pressure than NaCl at the same molar concentration. For example, the Na–Fe–CA draw solution has an osmotic pressure of 58 bar at 1 M, while it is only 47 bar for NaCl at the same concentration. Furthermore, the osmotic pressure difference between Na–Fe–CA and NaCl increases with raising the solute concentration. The presence of the ionic CA ligands contributes the good water solubility under neutral conditions and the higher osmotic pressures. As a result, the Na–Fe–CA draw solutions show a higher water permeation flux (Jw) in FO tests (DP = 0) at the same concentration (Fig. 2(b)). In addition, the Jw increases with the raise of draw solution concentration. This is because of the increased osmotic driving force. However, the water flux seems to be leveled off at higher draw solution concentrations. This phenomenon is most likely due to the dilutive external concentration polarization

One thin film composite (TFC) hollow fiber membrane with an inner polyamide selective layer was fabricated and used as the PRO membrane in the closed-loop PRO process. Fig. 3 displays the representative morphology of the inner surface, outer surface, and cross-section micrographs of the hollow fiber substrate. In order to balance the trade-off between fiber strength and the resistance to the fluid flow inside the lumen, the fiber was designed to have a medium dimension with an outer diameter of 820 lm and an inner diameter of 520 lm, respectively (see Table 2). Macroscopically, the hollow fiber substrate shows good concentricity and has short finger-like straight macrovoids in its cross-section. These macrovoids were formed via an instantaneous demixing induced by the strong bore fluid (H2O/NMP = 70/30 (wt.%)) during fiber formation [26,30]. Microscopically, the hollow fiber substrate possesses a highly porous open-cell structure cross the fiber cross section, which is capable of reducing the transport resistance and minimizing the detrimental effects of internal concentration polarization (ICP). There is a less porous sponge-like layer of several tens of microns beneath the inner skin of the hollow fiber substrate and above the finger-like straight macrovoids. This micro-structure layer is essential to maintain the stability of the polyamide selective layer and effectively dissipate the stresses across the membrane under high pressures [12–14]. Since the bore fluid has a high amount of nonsolvent, the inner surface of the fiber exhibits relatively dense with small surface pores. This morphology is favorable for the formation of a less defective polyamide layer with a good water permeability during interfacial polymerization [25,31,32]. Benefiting from the dual-bath coagulation technology, a highly porous outer surface is observed with disconnected large pores. This outer morphology not only facilitates water and salt transport and thus reduces ICP in the support layer, but also provides robust mechanical strength. As listed in Table 2, water permeation and molecular weight cut-off (MWCO) tests show that the inner surface of the hollow fiber substrate has a relatively small mean pore size of 13.7 nm with a narrow distribution (rp = 1.47), which could function as

388

G. Han et al. / Applied Energy 132 (2014) 383–393 100

100

(a)

NaCl Na-Fe-CA

80

Water flux, Jw (LMH)

Osmotic Pressure (bar)

80

60

40

20

0

60

40

20

0 0.0

0.5

1.0

1.5

0.0

2.0

0.5

1.0

1.5

2.0

Concentration (M)

Concentration (M) 8

1.5

(c)

6

4

2

0

(d)

Na-Fe-CA

Pressure build-up (bar)

NaCl Na-Fe-CA

Relative viscosity

(b)

NaCl Na-Fe-CA

1.2

0.9

0.6

0.3

0.0 0.0

0.5

1.0

1.5

2.0

Concentration (M)

0.0

0.5

1.0

1.5

2.0

Concentration (M)

Fig. 2. Comparisons of osmotic pressure (a), water flux in FO (b), relative viscosity (c), and pressure build-up in the fiber (d) of the synthesized Na–Fe–CA and NaCl draw solutions. FO performance (DP) in (b) was obtained using deionized water as the feed. In (d), the feed velocity is 0.2 L/min and the draw solution velocity is 0.1 L/min.

an ultrafiltration (UF) skin with a MWCO of 146.4 kDa and a pure water permeability (PWP) of 284.4 L/(m2 bar h). These results are consistent with the observations in Fig. 3, as no large pore is detected on the membrane inner surface. In addition, the substrate possesses a relative low overall porosity of 70.3% due to the partial sponge-like structure. Since the robustness of the final TFC PRO membranes under high pressures is mainly determined by the robust stability of the substrate, the mechanical properties of the hollow fiber substrate were studied. As presented in Table 3, the hollow fiber substrate exhibits high tensile modulus of 326.1 MPa and tensile strength of 7.8 MPa due to the intrinsically tensile properties of the substrate polymer and the unique morphological structure of the hollow fiber membrane. Therefore, the substrate membrane possesses high toughness of 2.3  106 J/m3 and can withstand a hydraulic pressure larger than 15 bar. The polyamide skin was synthesized on the inner surface of the hollow fiber substrate by interfacial polymerization. Fig. 4 shows the cross-section and surface morphologies of the resultant TFC hollow fiber membrane. It can be clearly observed that a uniform thin layer (178.5 nm) with a ‘‘ridge-and-valley’’ morphology has been attached onto the inner surface of the hollow fiber substrate. Before tests, the as-prepared TFC hollow fiber membrane was pressurized to 15 bar for 20 min to stabilize the membrane permeability, selectivity, and membrane structure parameter [13,14,26]. After that, the membrane water permeability coefficient A, and salt permeability coefficient B were measured in a pressure driven mode; the membrane structure parameter S was estimated by using the water flux in FO tests. As listed in Table 4, the stabilized TFC PRO hollow fiber membrane displays a high A of 4.3 L/ (m2 bar h) accompanied with a low B of 0.47 L/(m2 h). In addition, a small S of 640 lm is obtained, indicating that the ICP effects

could be much reduced. The stretch in the radial direction in response to the applied high hydraulic pressure in the lumen of the TFC membrane during stabilization could help reduce S and improve the free volume of the polyamide layer with slightly compromised selectivity [13–16]. The basic osmotic performance of the stabilized TFC PRO hollow fiber membrane at DP = 0 was evaluated using deionized water as the feed solution and NaCl (1.0 M) or Na–Fe–CA (1.0 M) as the draw solution. As shown in Table 5, a volumetric water flux (Jw) of 49.2 L/(m2 h) was achieved by using NaCl as the draw solution in the PRO mode where the membrane selective layer faces the draw solution. When changing the draw solution to Na–Fe–CA, the TFC membrane shows an improved Jw of 62.1 L/(m2 h). Importantly, a negligible reverse solute permeation flux (Js) of 0.4 g/ (m2 h) was observed for Na–Fe–CA, which is much lower than that of NaCl (10.5 g/(m2 h)). Since the developed TFC hollow fiber membrane has characteristics of high Jw, ultralow Js (particularly to Na– Fe–CA), small S, and good membrane robustness (DP > 12 bar), it may be an effective membrane to achieve a high efficiency of power generating in the closed-loop PRO process. 3.3. Implications for osmotic power generation In order to effectively harvest the osmotic energy from the mixing process via the closed-loop PRO process, the choice of the feed water and draw solution is a critical issue. Different from the conventional PRO processes, one important advantage of the closedloop PRO is that deionized water could be used as the feed which is the most preferred one. However, an appropriate draw solution is very challenging, which determines the overall efficiency of the process [20,33,34]. In this study, the synthesized Na–Fe–CA hydroacid complex draw solution with a concentration of 1 M was used

389

G. Han et al. / Applied Energy 132 (2014) 383–393

Cross section

100 nm

Inner surface

Outer surface

10 µm

100 µm

Fig. 3. The representative morphology of the hollow fiber membrane substrate.

Table 2 Summary of the porous properties and dimension of the hollow fiber membrane substrate. Membrane (ID)

lp (nm)

rp

PWP (L/(m2 bar h))

MWCO (kDa)

Porosity (%)

Fiber OD/ID (lm)

Membrane substrate

13.7

1.47

284.4

146.4

70.3

820/520

Table 3 Summary of the mechanical properties of the hollow fiber membrane substrate.

a

Membrane ID

Tensile modulus (MPa)

Tensile strength (MPa)

Elongation at break (%)

Toughnessa (106 J/m3)

Burst Pressure (bar)

Membrane substrate

326.1 ± 24.5

7.8 ± 0.4

36.2 ± 3.1

2.3 ± 0.8

16

Toughness was calculated by taking the integral underneath the stress–strain curve.

as the draw solution to examine the applicability of the newly developed closed-loop PRO process. NaCl was used as a benchmark draw solute. 3.3.1. Water flux and power density Fig. 5 shows the experimentally measured water fluxes (Jw) and power densities (W) as a function of hydraulic pressure difference (DP) using deionized water as the feed, NaCl and Na–Fe–CA as the draw solutions. Since the TFC hollow fiber membrane possesses good mechanical strength, a wide range of operating pressure from 0 to 12 bar was operated in the closed-loop PRO process. It was observed that Jw decreased slightly with an increase in DP due to the combinative effects of reduced driving force and the internal concentration polarization (ICP). A high Jw of 62.1 LMH and 48.6 LMH could be achieved at 0 bar and 12 bar, respectively. When using NaCl as the draw solution, the Jw was 48.2 LMH at 0 bar and then Jw decreased to 41 LMH at 12 bar. Compared to NaCl,

the Na–Fe–CA complex draw solution shows a larger Jw at each DP. The improved Jw of Na–Fe–CA is attributed to its higher osmotic pressure at the same concentration and much lowered Js (Fig. 2 and Table 5). Owning to its relatively high viscosity and large molecular weight, diffusion to replenish its concentration at the membrane interface is slightly limited by the diffusivity of the Na–Fe–CA macromolecules. As a result, compared to NaCl, Na– Fe–CA shows a larger reduction in Jw with an increase in DP. Since the power density (W) is determined by the Jw and DP (Eq. (10)), W is almost proportional to Jw. A high W of 16.2 W/m2 could be achieved by the Na–Fe–CA draw solution at a DP of 12 bar, which is much higher than that of NaCl (13.7 W/m2). In addition, the newly developed closed-loop PRO process could achieve a W of 9.5 W/m2 even at a low DP of 6 bar. This power density is still much higher the proposed goal of 5 W/m2 by Statkraft [2,3,35,36], while the operating pressure is relative low which may reduce the overall cost. The ability of generating various high

390

G. Han et al. / Applied Energy 132 (2014) 383–393

(b)

(a)

10 µm

100 µm

(d)

(c) 178.5 nm

100 nm

100 nm

Fig. 4. The representative morphology of the TFC PRO hollow fiber membranes: (a) overall cross-section; (b) enlarged cross-section; (c) surface of the polyamide selective skin; (d) cross-section of the polyamide selective skin.

Table 4 Transport properties and rejection of the TFC PRO hollow fiber membrane. Membrane TFC PRO membrane a

a

PWP, A (L/(m2 bar h))

Salt permeability, B (L m2 h1)

Salt rejection, Rs (%) (200 ppm @ 1 bar)

S (lm)

Km (105 s m1)

4.30

0.47

88.29

640

4.32

The TFC PRO hollow fiber membrane was stabilized at 15 bar for 20 min.

Table 5 FO performance of the TFC hollow fiber membrane by using different draw solutions. Draw solution

NaCl Na–Fe–CA

FO performance (PRO mode) Water flux (LMH)

Reverse solute flux (gMH)

49.2 ± 3.2 62.1 ± 4.5

10.5 ± 3.2 0.4 ± 0.3

Draw solution concentration: 1 M; feed solution: deionized water; Flow rate: lumen side is 0.1 L/min; shell side is 0.2 L/min.

power densities implies the great process flexibility of the developed closed-loop PRO process. 3.3.2. Reverse solute permeation flux In addition to Jw and W, the reverse solute flux, Js, and the specific solute flux (Js/Jw) are other important parameters for the PRO process [12,37]. A large Js and/or a large Js/Jw is unacceptable because it will directly reduce the effective driving force and elevate ICP, thus greatly decreases Jw and W [37,38]. Particularly, a large Js in the closed-loop PRO process will significantly increase the cost of the PRO process and contaminate the feed solution due to the loss of the recycled draw solute. Fig. 6 compares Js and Js/Jw of the newly developed closed-loop PRO versus the trans-membrane pressure (DP) when using Na–Fe– CA and NaCl as draw solutions. NaCl shows a relative low Js of 10.5 gMH at 0 bar; however, Js increases rapidly to 32.7 gMH at 12 bar. The leaked NaCl solute from the draw solution accumulates in the porous support and reduces the effective osmotic driving force even though deionized water is used as the feed solution

[37–39]. Interestingly, the Na–Fe–CA draw solution has a negligible Js of 0.44 gMH at 0 bar due to its larger molecular size. The increment of Js with the increase of DP is also very small, which is still less than 3.0 gMH when DP goes up to 12 bar. The magnitude of this Js is more than one order of lower than that of the NaCl draw solution under the same conditions. The Js/Jw values of Na–Fe–CA exhibit similar results which are always lower than 0.062 g/L even at 12 bar, suggesting that the draw solute loss in the closed-loop PRO process is insignificant compared to NaCl (0.062 vs. 0.8 g/L). As a consequence, the replenishment cost to maintain a constant osmotic pressure may be minor. To the best of our knowledge, this PRO performance in terms of both W (Fig. 5) and Js (Fig. 6) outperforms all other PRO processes reported in the literatures [3–6,40]. Clearly, the Na–Fe–CA complex draw solution can significantly promote the closed-loop PRO process in terms of higher power density and negligible reverse solute flux. In addition, the ultralow Js of the Na–Fe–CA draw solution provides great flexibility to design PRO membranes which are often limited by the trade-off between permeability and selectivity. 3.3.3. Effects of feed salinity The feasibility of the newly developed closed-loop PRO process for power generation was further investigated using other fresh waters that contains low concentrations of salts such as synthetic river water (10 mM NaCl). As illustrated in Fig. 7, Jw shows a slightly decrease with an increase in feed water concentration. As a result, the power density drops from 16.5 W/m2 to 14.4 W/m2 at 12 bar. These reductions are mainly attributed to the combinative effects of the reduced osmotic pressure difference and the concentrative ICP in the porous substrate layer. It is worthy to note

391

G. Han et al. / Applied Energy 132 (2014) 383–393 80

(a)

60 50 40 30 20 10

(b)

NaCl Na-Fe-CA Power density, W (W/m2)

70

Water flux, Jw (LMH)

20

NaCl Na-Fe-CA

0

16

12

8

4

0 0

2

4

6

8

10

12

14

0

2

4

6

P (bar)

8

10

12

14

P (bar)

Fig. 5. Water flux, Jw (a) and power density, W (b) of the TFC hollow fiber membranes in the closed-loop PRO process by using NaCl and Na–Fe–CA as the draw solutions. The draw solution concentration is 1 M, and the feed solution is deionized water. 1.0

(a)

NaCl Na-Fe-CA

40

0.8

30

0.6

Js/Jw (g/L)

Reverse solute flux, Js (gMH)

50

20

(b)

NaCl Na-Fe-CA

0.4

0.2

10

0.0

0 0

2

4

6

8

10

12

0

14

2

4

6

8

10

12

14

P (bar)

P (bar)

Fig. 6. Comparison of the reverse solute flux, Js (a) and specific solute flux, Js/Jw (b) of the TFC hollow fiber membranes in the closed-loop PRO process by using NaCl and Na– Fe–CA as the draw solutions. The draw solution concentration is 1 M, and the feed solution is deionized water.

20

80

Water flux, Jw (LMH)

70

Power densdity, W (W/m2)

NaCl Na-Fe-CA

(a) 60 50 40 30 20 10

(b)

NaCl Na-Fe-CA 16

12

8

4

0

0 0

2

4

6

8

10

12

14

P (bar)

0

2

4

6

8

10

12

14

P (bar)

Fig. 7. The water flux, Jw (a) and power density, W (b) of the TFC hollow fiber membranes by using NaCl and Na–Fe–CA as the draw solutions. The draw solution concentration is 1 M, and the feed solution is the synthetic river water (10 mM NaCl).

that the Na–Fe–CA draw solution may be contaminated by the reverse diffused NaCl. Since the amount of NaCl in the river water is minor and the selectivity of the developed PRO membrane is good, the above negative effects are not significant. Ideally, the Na–Fe–CA draw solution and deionized feed water are preferred to be used in the closed-loop PRO process. 3.3.4. Regeneration of the diluted Na–Fe–CA draw solution For the closed-loop PRO process, an additional separation step is required to regenerate the diluted draw solution. The efficiency of the closed-loop PRO process is affected by this step since it may be

energy intensive when a non-effective draw solution is used. In the newly developed closed-loop PRO process, the diluted Na–Fe–CA draw solution was regenerated via a solvent induced precipitation process at room temperature. The regeneration process is believed to be relative economic since no additional energy is applied and the used solvent is cheap and the chemical feedstock consumption is low. Most importantly, the physicochemical properties of the regenerated Na–Fe–CA draw solution could be maintained. The PRO performance of the regenerated Na–Fe–CA draw solution was investigated under the same conditions using deionized water as the feed solution. As shown in Fig. 8, the Jw, W, Js and

392

G. Han et al. / Applied Energy 132 (2014) 383–393 80

Power density, W (W/m2)

70

Water flux, Jw (LMH)

20

Regenerated Na-Fe-CA As-synthesized Na-Fe-CA

(a)

60 50 40 30 20 10 0

Regenerated Na-Fe-CA As-synthesized Na-Fe-CA

16

12

8

4

0 0

2

4

6

8

10

12

14

0

2

4

6

P (bar) 50

8

10

12

14

P (bar) 1.0

Regenerated Na-Fe-CA As-synthesized Na-Fe-CA

(c) 40

0.8

30

0.6

Js/Jw (g/L)

Reverse solute flux, Js (gMH)

(b)

20

Regenerated Na-Fe-CA As-synthesized Na-Fe-CA

(d)

0.4

0.2

10

0.0

0 0

2

4

6

8

10

12

14

0

2

4

6

8

10

12

14

P (bar)

P (bar)

Fig. 8. The performance of the TFC PRO hollow fiber membrane in the closed loop PRO process by using the regenerated Na–Fe–CA draw solution: (a) water flux, Jw; (b) power density, W; (c) reverse solute flux, Js; (d) specific solute flux, Js/Jw. The draw solution concentration is 1 M, and the feed solution is deionized water.

Js/Jw values of the closed-loop PRO process are almost fully recovered. Particularly, Js and Js/Jw of the regenerated draw solution show the same levels compared with those of the as-synthesized draw solution. These results indicate that the Na–Fe–CA draw solution is successfully regenerated without disturbing its properties. In addition, the Na–Fe–CA draw solute was further purified during the regeneration process. Therefore, fresh water containing a small amount of NaCl can be also used in the closed-loop PRO process although the performance will be discounted (see Fig. 7). 4. Conclusions We have successfully developed and demonstrated one novel closed-loop pressure retarded osmosis (PRO) process for capturing the renewable salinity-gradient energy. The ferric hydroacid complex (Na5[Fe(C6H4O7)2]) has been synthesized and employed as the draw solute in the closed-loop PRO process. The characteristics of high solubility and easy dissociation in water, and structurally expanded configuration enable the complex solution to possess a high water flux but a very low reverse solute flux in PRO. Employing a highly permeable (A = 4.30 LMH/bar) and robust thin film composite PRO hollow fiber membrane, a power density of 16.2 W/m2 can be achieved with an ultralow reverse solute flux (Js/Jw < 0.062 g L1) at 12 bar when using 1 M complex draw solution and deionized water as feeds. The diluted complex draw solution could be easily regenerated through a solvent induced concentration process and its outstanding PRO performance could be almost fully recovered. PRO experiments indicate that the complex draw solute is superior to conventional salts as draw solutes in both PRO performance and subsequent regeneration processes. The newly developed closed-loop PRO process shows great potential to harvest the renewable osmotic energy, although its specific

benefits have to be further explored through extensive energy and cost analysis in the future.

Acknowledgments This research Grant is supported by the Singapore National Research Foundation under its Environmental & Water Technologies Strategic Research Programme and administered by the Environment & Water Industry Programme Office (EWI) of the PUB under the project entitled ‘‘Membrane development for osmotic power generation, Part 1. Materials development and membrane fabrication’’ (1102-IRIS-11-01) and NUS grant number of R-279000-381-279.

References [1] Loeb S, Norman RS. Osmotic power plant. Science 1975;189:654–5. [2] Thorsen T, Holt T. The potential for power production from salinity gradients by pressure retarded osmosis. J Membr Sci 2009;335:103–10. [3] Chung TS, Li X, Ong RC, Ge QC, Wang HL, Han G. Emerging forward osmosis (FO) technologies and challenges ahead for clean water and clean energy applications. Curr Opin Chem Eng 2012;1:246–57. [4] Logan BE, Elimelech M. Membrane-based processes for sustainable power generation using water. Nature 2012;488:313–9. [5] Zhao S, Zou L, Tang CY, Mulcahy D. Recent developments in forward osmosis: opportunities and challenges. J Membr Sci 2012;396:1–21. [6] Gerstandt K, Peinemann KV, Skilhagen SE, Thorsen T, Holt T. Membrane processes in energy supply for an osmotic power plant. Desalination 2008;224:64–70. [7] Loeb S, Hessen FV, Shahaf D. Production of energy from concentrated brines by pressure retarded osmosis. II. Experimental results and projected energy costs. J Membr Sci 1976;1:249–69. [8] Kim J, Lee J, Kim JH. Overview of pressure-retarded osmosis (PRO) process and hybrid application to sea water reverse osmosis process. Desalination Water Treat 2012;43:193–200.

G. Han et al. / Applied Energy 132 (2014) 383–393 [9] Zhang S, Fu FJ, Chung TS. Substrate modifications and alcohol treatment on thin film composite membranes for osmotic power. Chem Eng Sci 2013;87: 40–50. [10] Han G, Zhang S, Li X, Chung TS. High performance thin film composite pressure retarded osmosis (PRO) membranes for renewable salinity-gradient energy generation. J Membr Sci 2013;440:108–21. [11] Song X, Liu Z, Sun DD. Energy recovery from concentrated seawater brine by thin-film nanofiber composite pressure retarded osmosis membranes with high power density. Energy Environ Sci 2013;6:1199–210. [12] Bui NN, McCutcheon JR. Hydrophilic nanofibers as new supports for thin film composite membranes for engineered osmosis. Environ Sci Technol 2013;47: 1761–9. [13] Han G, Wang P, Chung TS. Highly robust thin-film composite pressure retarded osmosis (PRO) hollow fiber membranes with high power densities for renewable salinity-gradient energy generation. Environ Sci Technol 2013;47: 8070–7. [14] Zhang S, Sukitpaneenit P, Chung TS. Design of robust hollow fiber membranes with high power density for osmotic energy production. Chem Eng J 2014;241: 457–65. [15] Li X, Chung TS. Thin-film composite P84 co-polyimide hollow fiber membranes for osmotic power generation. Appl Energy 2014;114:600–10. [16] Chou S, Wang R, Fane AG. Robust and high performance hollow fiber membranes for energy harvesting from salinity gradients by pressure retarded osmosis. J Membr Sci 2013;448:44–54. [17] She Q, Hou D, Liu J, Tan KH, Tang CY. Effect of feed spacer induced membrane deformation on the performance of pressure retarded osmosis (PRO): implications for PRO process operation. J Membr Sci 2013;445:170–82. [18] Park M, Kim JH. Numerical analysis of spacer impacts on forward osmosis membrane process using concentration polarization index. J Membr Sci 2013;427:10–20. [19] Sivertsen E, Holt T, Thelin W, Brekke G. Pressure retarded osmosis efficiency for different hollow fibre membrane module flow configurations. Desalination 2013;312:107–23. [20] McGinnis RL, McCutcheon JR, Elimelech M. A novel ammonia–carbon dioxide osmotic heat engine for power generation. J Membr Sci 2007;305:13–9. [21] Loeb S. Method and apparatus for generating power utilizing pressureretarded-osmosis. US Pat 1975;3(906):250. [22] Reali M. Closed cycle osmotic power-plants for electric-power production. Energy 1980;5:325–9. [23] Widjojo N, Chung TS, Krantz WB. A morphological and structural study of Ultem/P84 copolyimide dual-layer hollow fiber membranes with delamination-free morphology. J Membr Sci 2007;294:132–46. [24] Cadotte JE. Reverse osmosis membrane. US Pat 1977;4039440.

393

[25] Han G, Zhang S, Li X, Widjojo N, Chung TS. Thin film composite forward osmosis membranes based on polydopamine modified polysulfone substrates with enhancements in both water flux and salt rejection. Chem Eng Sci 2012;80:219–31. [26] Han G, Chung TS. Robust and high performance pressure retarded osmosis (PRO) hollow fiber membranes for osmotic power generation. AIChE J 2014;30:1107–19. [27] Ge Q, Chung TS. Hydroacid complexes: a new class of draw solutes to promote forward osmosis (FO) processes. Chem Commun 2013;49:8471–3. [28] Matzapetakis M, Raptopoulou CP, Tsohos A, Papaefthymiou V, Moon N, Salifoglou A. Synthesis, spectroscopic and structural characterization of the first mononuclear, water soluble iron-citrate complex, (NH4)5Fe(C0H4O7)2 2H2O. J Am Chem Soc 1998;120:13266–7. [29] Ge Q, Su JC, Amy G, Chung TS. Exploration of polyelectrolytes as draw solutes in forward osmosis processes. Water Res 2012;46:1318–26. [30] Widjojo N, Chung TS. The thickness and air–gap dependence of macrovoid evolution in phase-inversion asymmetric hollow fiber membranes. Ind Eng Chem Res 2006;45:7618–26. [31] Alsvik IL, Hägg MB. Pressure retarded osmosis and forward osmosis membranes: materials and methods. Polymers 2013;5:303–327. [32] Li X, Wang KY, Helmer BJ, Chung TS. Thin-film composite membranes and formation mechanism of thin-film layer on hydrophilic cellulose acetate propionate substrates for forward osmosis processes. Ind Eng Chem Res 2012;51:10039–50. [33] Weingarten MH. Power generating means. US Pat 1971;3(587):227. [34] Loeb S. Method and apparatus for generating power utilizing pressure retarded-osmosis. US Pat 1980;4(193):267. [35] Skilhagen SE, Dugstad JE, Aaberg RJ. Osmotic power–power production based on the osmotic pressure difference between waters with varying salt gradients. Desalination 2008;220:476–82. [36] Skilhagen SE. Osmotic power—a new, renewable energy source. Desalination Water Treat 2010;15:271–8. [37] Zhang S, Chung TS. Minimizing the instant and accumulative effects of salt permeability to sustain ultrahigh osmotic power density. Environ Sci Technol 2013;47:10085–92. [38] She Q, Jin X, Tang CY. Osmotic power production from salinity gradient resource by pressure retarded osmosis: effects of operating conditions and reverse solute diffusion. J Membr Sci 2012;401–402:262–73. [39] Yip NY, Elimelech M. Performance limiting effects in power generation from salinity gradients by pressure retarded osmosis. Environ Sci Technol 2011;45:10273–82. [40] Achilli A, Childress AE. Pressure retarded osmosis: from the vision of Sidney Loeb to the first prototype installation—review. Desalination 2010;261: 205–11.