Desalination 350 (2014) 86–94

Contents lists available at ScienceDirect

Desalination

j ourna l homepage: www.e lsev ie r .com/ locate /desa l

Energy and thermodynamic analysis of power generation using a naturalsalinity gradient based pressure retarded osmosis process

Wei He, Yang Wang, Mohammad Hasan Shaheed ⁎School of Engineering and Materials Science, Queen Mary University of London, E1 4NS, London, UK

H I G H L I G H T S

• Power density and energy generation by PRO processes are mathematically interpreted.• Discharge behaviour of PRO processes based on natural salinity is obtained and analysed.• Membrane consumption of a full scale PRO discharge is studied.• Configurations of two-stage PRO process are proposed and discussed.• Energy generation is compared for different configurations.

⁎ Corresponding author. Tel.: +44 20 7882 3774.E-mail addresses: w.he@qmul.ac.uk (W. He), yang.wa

m.h.shaheed@qmul.ac.uk (M.H. Shaheed).

http://dx.doi.org/10.1016/j.desal.2014.07.0150011-9164/Crown Copyright © 2014 Published by Elsevie

a b s t r a c t

a r t i c l e i n f o

Article history:Received 5 March 2014Received in revised form 9 July 2014Accepted 12 July 2014Available online xxxx

Keywords:Available energyDesalinationDischarge behaviourPressure retarded osmosisTwo-stage PRO process

This study presents a thermodynamic and energy analysis of the discharge behaviour of a single-stage pressureretarded osmosis (PRO) process which is then expanded into a proposed two-stage process to enhance total en-ergy extraction in a practical application. A thermodynamic model describing the operational conditions for theoptimal power density and the extraction of energy from a single-stage PRO process is introduced. The dischargebehaviour of the power generated from the process is analysed and the profiles of water flux, power density, andextracted energy are obtained. The membrane consumption is also studied with respect to different hydraulicpressures on the draw solution, and the flows of both the draw and feed solutions. The inherent inconsistenciesin the operational conditions with regard to achieving maximal power density and available energy is discussedand interpreted based on the discharge behaviour. A two-stage PRO process with two alternative feed arrange-ments (continuous feed and divided feed) is then proposed and its operations are simulated and analysed. Theresults indicate favourable energetic performance of the two-stage versus the one-stage PRO process in termsof the reduced frictional loss and unused energy involved in the process.

Crown Copyright © 2014 Published by Elsevier B.V. All rights reserved.

1. Introduction

Rapidly increasing global population and climate change are two ofthe greatest challenges of our time [1]. Continued dependence on fossilfuels is unsustainable due to the finite fuel supply and climatic effectsarising from the emission of carbon dioxide [2]. Renewable energy tech-nology represents an apparent solution in this context [3]. Technologiesinvolving wind, solar, geothermal and biomass have attracted signifi-cant attention but have achieved only limited capacity, so far, for localand global electricity generation without or with lower emission ofgreenhouse gases [4,5]. Research into alternative technologies con-tinues with the goal of providing practical and reliable renewable ener-gy sources (RES) [6–8].

ng@qmul.ac.uk (Y. Wang),

r B.V. All rights reserved.

Osmotic energy fromnatural salinity gradients has been identified asa candidate RES since the 1950s [9], due to its substantial potential en-ergy capacity, estimated to be 2 TW, or about 13% of the current worldenergy consumption [10]. Research groups worldwide have, investigat-ed the feasibility of capturing energy from the mixture of freshwaterand seawater by means of PRO [11]. This technology, is an osmoticallydriven membrane process that takes the advantage of hydraulic pres-sure developed in the draw solution to convert osmotic energy intoelectricity by hydro-turbine [12]. Following rapid developments in thefield over the last decade, this technology is now in operational use. In2009, the world's first PRO plant was launched in Norway with a4 kW capacity [13].

Prior investigations to improve the performance of a PRO processhave focused on developing high performance membranes and settingup appropriate operational conditions tomaximize energy yields. Detri-mental effects during transportation across the membrane have beenidentified as a major problem [14], limiting the performance of PRO

87W. He et al. / Desalination 350 (2014) 86–94

technology [15,16]. Primary performance-limiting phenomena com-prise internal concentration polarization (ICP) in the support layer[17], external concentration polarization (ECP) on the draw solutionside [18], and reverse solute flux (RSF) [19]. Different membrane typesand orientations, aswell as varying operational conditions of a PRO pro-cess have been investigatedwith the aimofminimizing these detrimen-tal phenomena [20–22].

The current study also draws on prior investigations of the PRO pro-cess not only as an independent power generator but also as a pre- orpost-treatment mechanism to recover osmotic energy from high con-centration brine discharge, in a hybrid process integrated with reverseosmosis (RO) and forward osmosis (FO) processes [23,24]. Previousstudies havemostly focused on improvingwaterflux andpower densityof theprocess [19] and only a fewalso focused on energy generation [25,26]. Data from these studies indicate that the potential energy from themixed seawater and freshwater can be calculated using the concept ofGibbs free energy [27]. However, the potential chemical energy repre-sents themaximal energy involved in themixing process, which cannotbe fully used due to intrinsic thermodynamic inefficiency. Taking thisunavoidable frictional loss and unutilized energy into account [28], theconcept of extractable energy has been incorporated into the energyanalysis.

Like power density, the maximum energy generated is also signifi-cantly influenced by the applied hydraulic pressure of the draw solu-tion. In the current study, a thermodynamic analysis of the dischargebehaviour of a PRO process is carried out to investigate the relationshipbetween thepower density and the extracted energy. A thermodynamicmodel of the PRO process is introduced and used as a basis for furtherinvestigations. The study focuses on the influences of different parame-ters and operational conditions on the discharge behaviour of the pro-cess. The inherent inconsistency of optimal conditions for maximalpower density andmaximal extractable energy generated by a constantpressure PRO process is analysed and discussed.

Furthermore, similar to the advantages of energy efficiency in two-stage RO configurations [29], two-stage PRO process is also potentiallyefficient in salinity energy harvest. The configuration is capable to de-crease the frictional loss by altering the hydraulic pressure applied onthe draw solution, and to increase the energy generation due to thetwo-stage generation by increasing the water permeation and reducingthe unutilized energy loss. However, a review of the literature revealsno investigations to date of two-stage PRO processes, or analysis oftheir potential for increased total energy extraction. Therefore, a studyof two-stage PRO process is also carried out. Relevant targets for studyinclude the analysis and optimization of the possible configurationsand operations of two-stage processes. To address this omission, thecurrent investigation defines and analyses the performance of a two-stage PRO with two different feed water operations: continuous feedtwo-stage PRO and divided feed two-stage PRO. The results show thecharacteristic of preferable energy generation capacity of the two-

Fig. 1. A schematic illustration of a PRO process. The water permeates through the membrane ftion), and then is expanded in hydro-turbines.

stage PRO process and the influences on the performance due to varia-tion in operational conditions and the available volumes of feed anddraw water for use in the PRO process.

2. Osmotic energy generated by a PRO

PRO uses the natural phenomenon of osmosis to permeate wateracross a semi-permeable membrane from a side with low solute con-centration and low hydraulic pressure to a side with high concentrationand high pressure. The permeated water is then used to generate elec-tricity in a hydro-turbine (Fig. 1). In the PRO process, the draw solutionis pressurized by the pump and the energy recovery device (ERD). Aswater is transported across the membrane, the draw solution becomesprogressively diluted and the concentration of the feed solution rises.For simplicity, the efficiency of pumps and turbines is assumed to be100% in this study.

Two of the key parameters determining the performance of PROpro-cesses are the conditions of available salinity streams and the PRO oper-ations. The two salinity streams determine the total energy capacity thatcan be harvested. The Gibbs free energy is released when the twostreams mix, and is converted into electricity by the PRO process. How-ever, this energy cannot be fully harvested. In addition to the inherentmembrane performance, the efficiency of the PRO process also signifi-cantly depends on the hydraulic pressure applied on the draw solution,because the applied pressure determines not only the flow rate but alsothe pressure head of the pressurized permeated water.

Variables in the available water conditions comprise the concentra-tion and volume of the draw and the feed water. Power generation bya PRO process from natural salinity gradients could utilize high concen-tration salinewater such as seawater and brackishwater as drawwater.The total dissolved solid (TDS) of brackishwater is in the range of 1000–5000mg/L, and the TDS of seawater is larger than 35,000mg/L. Normal-ly, water with TDS smaller than 1000 mg/L is classified as fresh water.This includes water from rivers, sewage, private effluents and industrialwastewater, but water from such sources would require pre-treatmentbefore use as feed water in a PRO system to prevent membrane fouling.

The available volume of draw and feed water is also very importantbecause it significantly influences the variation of net driving force and,as a consequence, determines the change of osmotic pressure differ-ence, water flux and power density along the membrane. The van'tHoff equation for osmotic pressure (π) applies to dilute, ideal solutionsand is given by,

π ¼ νRTc ð1Þ

where ν is the number of ionic species each salt molecule dissociates,Ris the gas constant, T is the temperature, c is the concentration of thesolution. In this study, for simplicity, the seawater is regarded as a hypo-thetical solution with 35,000 mg/L TDS. Accordingly, the van't Hoff law

rom the low concentration side (feed solution) to the high concentration side (draw solu-

88 W. He et al. / Desalination 350 (2014) 86–94

is used to approximate the osmotic pressure difference between thedraw and feed solutions. Thus, it can be expressed as,

π ¼ νRTc ¼ νRTV

0D

V0D þ V

c0D−

V0F

V0F−V

c0F

!ð2Þ

in which,V0D andV

0F are the initial volume rate of draw and feed, c0D and

c0F are the initial concentration of draw and feed, andV is the permeated

water volume rate. Furthermore, if a dimensionless feed volumetricratio, ϕ, is defined as ϕ ¼ V

0F= V

0F þ V

0D

� �, the osmotic pressure differ-

ence can be expressed as,

π ¼ νRT1−ϕ

1−ϕ 1− V

V0F

! c0D−

1

1− V

V0F

c0F

0BBBB@

1CCCCA ð3Þ

Based on Eq. (3), the influence from the feed volumetric ratio on theosmotic pressure difference across the PROmembrane can be obtained.The osmotic pressure difference between seawater (35,000 mg/L) andfreshwater (100 mg/L) as a function of dimensionless permeated vol-ume and volumetric ratio based on the van't Hoff law is illustrated inFig. 2, at a temperature of 293 K. The results indicate that the wateravailability has a significant influence on the change of osmotic pressuredifference. The concentration difference of draw and feed water deter-mines the initial osmotic pressure difference and the volume affectshow the difference disappears. It is obvious that all the curves originateat the same point due to the same initial concentrations of the draw andfeedwater. However, different volumetric ratio leads to different trajec-tories of osmotic pressure difference during energy generation by thePRO process.

Fig. 2. The osmotic pressure difference between seawater (35,000 mg/L) and fresh water(100mg/L), as a function of dimensionless permeated volume and volumetric ratio of thefeed solution to the sum of both the draw and feed solution volume is presented. In therepresentative plot, temperature is 293 K, Δπ represents osmotic pressure difference,and ΔV and V0

F are volume of permeate water and feed water, respectively.

2.1. Power density

Membrane power density is the power that can be generated perunit membrane area [18], and is a key factor to signify the performanceof a PRO process. With a larger power density, a smaller area of mem-brane is needed to achieve certain level of power generation [30]. Thepower density (W) is the product of water flux and applied hydraulicpressure,

W ¼ JwΔP ð4Þ

in which the water flux JWð Þ can be further expressed as,

JW ¼ LP π−Pð Þ ð5Þ

whereLP is themembrane permeability,P is the applied hydraulic pres-sure on the draw solution, and π represents the osmotic pressure differ-ence. The pressure drop through the flow channel is negligible in boththe feed and the draw side [31,32]. Hence, the applied pressure is con-stant over the length of both channels.

The equation above describes the ideal power dynamic characteris-tic of a PRO process. However, in practice, the osmotic pressure differ-ence between the two sides of the membrane, π−P, is lower than theosmotic pressure difference between the bulk draw and feed solutionsdue to the polarization effects of ICP within the porous support layer,ECP near the membrane surface in the draw solution side and RSFwhich is common phenomenon inmembrane processes [18]. The polar-ization will reduce the flow rate of the permeation and result in lowerpower density of the membrane and less energy generation from theharvesting process. Several researchers have carried out a series of stud-ies to improve the PRO membrane in order to identify the dynamics ofthe polarization [33,34] and reduce the possibility of these detrimentaleffects [35,36]. In this investigation, it was decided to carry out a ther-modynamic analysis on the PRO process, focusing on the connectionsbetween the water flux, power density and energy capacity of a PROprocess. Therefore, for the purpose of simplicity, concentration polariza-tion effects are neglected and the salt concentration near themembranesurface is assumed equal to the bulk concentration of the flow stream.

2.2. Osmotic energy generated by a PRO process

Asmentioned earlier, themaximum energy released from amixtureof two different composition solutions is the difference in the Gibbs freeenergy between the final mixture and the two initial solutions. In fact,theoretically, it is the maximum energy generation of a reversible PROprocess in which infinitesimal water flux is maintained through the os-mosis process by applying a hydraulic pressure negligibly smaller thanthe osmotic pressure difference. It is defined as the reversible PRO (R-PRO) process energy [28]. The R-PRO energy (ER−PRO) can be represent-ed as the integration of osmotic pressure difference through the perme-ating process as,

ER−PRO ¼Z

VSUM

0πd Vð Þ ð6Þ

in whichV represents the volume of the permeating water, and VSUM isthe accumulated permeate volumewhenwaterflux is terminated. In re-ality, due to natural thermodynamic inefficiency, the extracted energy isless than the R-PROenergy. In an actual PROprocess, a constant hydrau-lic pressure is applied on the drawsolution. Thewater flux is terminatedwhen the net driving force equals zero. Therefore, the final permeatedvolume of water is less than the total volume permeated in a reversibleprocess. The work done by a constant PRO (C-PRO) process can beexpressed as,

EC−PRO ¼ P � VSUM ð7Þ

89W. He et al. / Desalination 350 (2014) 86–94

in which, EC−PRO is represented as the energy generated by the C-PROprocess. The C-PRO energy is the energy that can be extracted in practice[28]. The difference between the R-PRO and C-PRO energy representsthe irreversible energy loss due to the frictional resistance in the trans-portation of water inside the membrane. This frictional force betweenthe water molecules and membrane gives rise to a hydraulic resistance[37], and thus a partial of osmotic driving force is consumed to compen-sate for the resistance. This part of energy loss is due to the frictionalloss. Furthermore, in a C-PRO process, the actual permeate volume issmaller than the volume of water that would permeate into the drawsolution without applied hydraulic pressure. Thus, the energy embed-ded in the “non-permeated water” cannot be extracted by a C-PRO pro-cess. This part of energy loss is unused energy. With no pressure lossassumed in both draw and feed flow channels, the extractable energyand energy losses of the PRO process from seawater and river waterare illustrated in Fig. 3. In the figure, the rectangle represents the actualenergy extracted by the C-PRO process. The frictional loss and unusedenergy are denoted by the blue shaded areas alongside the y- andx-axis.

Because the final permeated volume of water in a C-PRO process,VSUM , is a function of the applied pressure, it can be obtained bysubstitutingΔP= Δπ into Eq. (1). EC − PRO is only dependent on the hy-draulic pressure applied. Different applied hydraulic pressure on drawsolution results in different capacity of extracted energy and the corre-sponding energy losses. The optimal applied hydraulic pressure forextractingmaximal energy by a C-PRO process can be obtained by solv-ingdEC−PRO=d Pð Þ ¼ 0, based on Eq. (7) inwhich the dimensionlessfinalpermeated volume VSUM=V

0F is obtained by substituting ΔP = Δπ into

Eq. (3). Accordingly, the optimal applied pressure can be expressed as,

PMAXC−PRO ¼ νRT 1−ϕð Þc0D−ϕc0F þ 2ϕ−1ð Þ

ffiffiffiffiffiffiffiffiffiffiffic0Dc

0F

q� �ð8Þ

wherePMAXC−PRO represents the optimal applied hydraulic pressure to ob-

tain the maximum extracted energy. Conversely, the maximum powerdensity of a certainmembrane is achieved at the applied hydraulic pres-sure equalling half of the initial osmotic pressure difference according to

Fig. 3. The different types of energy involved in the PRO process are illustrated. It consistsof extractable energy (C-PRO energy), frictional loss and unused energy. In this represen-tative plot, the draw solution is seawater and freshwater is feedwater. The volume ratio offeed water to total volume is 0.5 and temperature is 293 K.

dW=d Pð Þ ¼ 0 based on Eq. (4) [38]. The applied pressure to achieve themaximum power density of a PRO membrane is expressed as,

PMAXW ¼ 1

2νRT c

0D−c

0F

� �ð9Þ

where PMAXW is the optimal applied pressure to obtain the maximum

power density.Therefore, based on Eqs. (8) and (9), the applied pressure to achieve

the maximum extracted energy and power density of a PRO processwith respect to the different volumetric ratios of feed water can be ob-tained (Fig. 4). Usually, there is incompatibility in operations to achievethe optimal power density and themaximumenergy extracted simulta-neously excluding the volumetric ratio of feed equalling 0.5. In the rangeof volumetric ratio from zero to 0.5 or from 0.5 to 1, the optimal appliedpressure to achieve maximum extracted energy is different from theone to achieve maximum power density.

3. Discharge behaviour of the PRO process

Based on the preceding analysis, it can be observed that the requiredoperation for maximizing power density is not usually synchronizedwith the one for achievingmaximal extracted energy. Actually, themis-match on the optimal applied hydraulic pressure between the powerdensity and extracted energy is determined by their different physicalrepresentations. Power density represents the utilization efficiency ofthemembrane. Conversely, extracted energy is the efficiency of salinityenergy harvesting by a C-PRO process with given water conditions ofboth draw and feed solutions. Power density is related to the waterflux of the membrane. In contrast, the extracted energy is determinedby the accumulated permeate water volume from the feed side to thedraw side along the membrane. The water flux of a steady PRO processis, in fact, the velocity of permeated water flow along the membrane.From this viewpoint, the water flux can be expressed as,

d Vð Þ ¼ Jwd AMð Þ ð10Þ

where AM is the area of the membrane. Furthermore, if the expressionfor water flux with respect to dimensionless permeated water volume,V� ¼ V=V

0F, is considered, combining Eq. (10) with Eqs. (2) and (4), an

ordinary differential equation (ODE) for the permeated water volumet-ric rate V along the membrane can be obtained as,

dV�

dAm

¼ A νRT1−ϕð Þc0D

1−ϕ 1−V�ð Þ−c0F

1−V�

!−P

!: ð11Þ

Fig. 4. The optimal pressures to achieve the maximums of extracted energy and powerdensity with different volumetric ratio. In this figure, curves are obtained from the PROprocess with the draw solution as seawater (35,000 mg/L) and feed solution as freshwater (100 mg/L). The temperature is 293 K.

Table 1Power density from selected publications with membrane having similar waterpermeability.

Reference Water permeability(L m−2 h−1 bar−1)

Draw Power density(W m−2)

[42] 1.88 Seawater 6.1[43] 1.4 Seawater brine

(1.0 M)8.9

1.7 9.21.9 11

[40] 1.74 Seawater 6.091.42 5.244.12 5.71

[41] 0.72 Seawater 2.42.23 5.5

90 W. He et al. / Desalination 350 (2014) 86–94

The solution to this differential equation represents the accumulatedwater permeated volume along the membrane during the discharge ofthe PRO process. A study to find the changing discharge processeswith different operations was conducted. In the investigation, draw so-lution was seawater and feed solution was fresh water with a volumet-ric ratio of 0.5. Themembrane area was assumed large enough and thusthe full-scale PRO discharge and the corresponding required minimumarea can be studied. The influence of different applied pressure onwater flux is depicted in Fig. 5. At each operation, both the water fluxand power density, due to the largest net driving force between thetwo sides of the membrane are reached at the entry of the membranemodule and decrease gradually along the membrane. The water fluxat the entrance decreases with increased applied pressure through dif-ferent trajectories, resulting to different required area ofmembrane ter-minating the water permeation. The power density at the entrance isnot proportional to the applied pressure and themaximum power den-sity at the entrance is achieved with the applied pressure equalling tohalf of the initial osmotic pressure difference.

However, in practice, membrane properties are the major con-straints to affect the water flux and power density [39]. It significantlyaffects the membrane performance and determines the membranearea required in energy generation. In this study, thewater permeabilityof the PRO membrane was selected as 1 L m−2 h−1 bar−1. The consis-tency was obtained by comparing the obtained results of the waterflux and power density with the previous studies using similar waterpermeability membrane [40,41]. Power densities of different PRO pro-cesses from some selected publications are illustrated in Table 1. The se-lected data indicate that the higher power density is not guaranteedwith the higher water permeability as the high performance of a mem-brane is a result of the trade-off between permeability and selectivity[36]. Furthermore, experimental results also indicate that the detrimen-tal effects would reduce the theoretical water flux and power densitysignificantly [32].

The harvest of salinity energy from seawater and freshwater duringthe PRO discharge is illustrated in Fig. 6, including the extracted energy,frictional loss and unused energy in kWh per cubic meter of feed waterby a C-PRO process. According to the results, the distribution of the ex-tracted energy and energy losses is different with respect to differenthydraulic pressure applied. In Fig. 6(a), due to low applied pressure,the net driving force of water permeation is high. Therefore, less areais required to harvest the salinity energy by a C-PRO process. Althoughmost of the salinity energy is used according to the low unused energy,only part of the energy is extracted by the C-PRO process as a result ofhigh frictional loss. Based on the results in Fig. 6(b) and (c), with the in-creased hydraulic pressure, frictional loss decreases and unused energyincreases with the increase of hydraulic pressure.

Fig. 5. Calculated water flux and power density curves in time domain with respect to differenwater permeability is assumed to be 1 L m−2 h−1 bar−1. Volumetric ratio of feed water is 0.5.

With the discharge behaviour of the PRO process in Figs. 5 and 6, it iseasy to understand the mismatch in hydraulic pressure to obtain themaximal power density and maximal energy extraction as describedin Fig. 4, because they are different inherently although closely interre-lated. The general definition of power density (Eq. (4)) is dependent onthe water flux. Power density is a transient characteristic of the PROmembrane. In contrast, extracted energy is an accumulated PROproper-ty along the membrane at its steady state and shows the performanceof a full-scale PRO discharge. In Fig. 4, the power density is a valuerepresenting the efficiency of unit membrane at the entrance. In con-trast, extracted energy is the integration of the power generation ofthe entire system.

The membrane consumption of a full scale PRO process varies withdifferent hydraulic pressure applied on the draw solution as illustratedin the Figs. 5 and 6. The relation between the water permeation V�ð Þand the membrane consumption AMð Þ is non-linear as described inEq. (11). In fact, the membrane cost of a PRO process depends onsome influencing factors, including the hydraulic pressure applied onthe draw solution, the dimensionless flow rate and membrane perme-ability. Consequently, the problem in this study can be representedmathematically as follows,

AFULLM ¼ AMjJw¼0

s:t: 0 ≤ P ≤ PMAX

ϕ ¼ 0:2; 0:5 and 0:8LP ¼ 1:74 and 1:9 L �m‐2 � h‐1 � bar‐1

ð12Þ

whereAFULLM is themembrane area consumed by a PRO processwith full

scale water permeation in which the water flux reaches the zero flux.PMAX is the maximum of the applicable hydraulic pressure on the

draw solution that it is the osmotic pressure difference at the entrance,which can be expressed as P

MAX ¼ νRT c0D−c

0F

� �. In the simulation,

t applied hydraulic pressure are represented in (a) and (b), respectively. The membrane

Fig. 6. The discharge behaviour of the PRO process with different applied hydraulic pressures are represented in the figure, in which applied pressure equalling to 7 bar in (a), 14.5 bar in(b) and 25 bar in (c). In these representative plots, the draw solution is seawater (35,000mg/L) and feed is freshwater (100mg/L), the volumetric ratio of feed is 0.5. Themembranewaterpermeability is assumed to be 1 L m−2 h−1 bar−1.

91W. He et al. / Desalination 350 (2014) 86–94

three dimensionless flow rates (0.2, 0.5 and 0.8) were selected and rep-resented for the possible flow rates of the available salinity gradients indifferent regions. In addition, two membrane permeability parameterslisted in Table 1 (1.74 L m−2 h−1 bar−1 and 1.9 L m−2 h−1 bar−1)[43] are studied. The results are presented in Fig. 7.

The results indicate that the membrane consumption with differentapplied pressure has a maximum which varies depending on the di-mensionless flow rate. For a certain dimensionless flow rate, when theapplied pressure increases from zero, the membrane consumed in afull scale PRO discharge first increases and reaches its maximum, thendecreases to zero when the maximum hydraulic pressure is applied. Inaddition, higher water permeability of the membrane reduces themembrane consumption according to increased water flux across themembrane at each applied hydraulic pressure. Furthermore, comparingthe profiles of membrane consumption in different dimensionless flowrates, it is found that the maximum of the membrane consumptionchanges. This is due to the extraction of the salinity energy during thefull scale PRO discharge in which the optimal pressure is differentwith respect to the dimensionless flow rates (Fig. 4).

4. Analysis of two-stage PRO processes

In this study, a two-stage PRO process and its several operations areproposed in order to further increase the efficiency in terms of the ener-gy harnessed. The two-stage PRO process is illustrated in Fig. 8. In thefigure, two PRO processes are series connected in draw water flow.The dilute draw solution from the first stage PRO process is the first par-tially used in pressurizing the raw draw solution, then combined andfully used to be the draw solution in the second stage PRO process. De-pending on the different feed water flows two types of two-stage PRO

Fig. 7. The schematic illustration of membrane consumption of a PRO process with different hyability. The result of the membrane cost with dimensionless flow rate of 0.2 is illustrated in (a)

processes are defined. In one case, the concentrated feed water fromthe first stage is continuously used as the feed water for the secondstage. This flow scheme is defined as a continuous feed two-stage PROprocess which is illustrated by the solid line in Fig. 7. The other is a dif-ferent feed flow scheme, because the feed water is divided into twoflows at the beginning and is used as feed water to two PRO processesseparately. This flow scheme is defined as divided feed two-stage PROprocess and is illustrated in Fig. 8 with dashed line to represent feedwater flows. In this section, the performance of the two-stage PRO pro-cess and single-stage PRO process with the same water availabilityamong different feed water flow schemes and operations are analysedand compared.

4.1. Continuous feed two-stage PRO process

In the case of the continuous feed two-stage PRO process, the feedwater is used to generate power twice through the two PRO processes.With the maximal C-PRO energy extraction in both two PRO processes,the total energy generation is illustrated in Fig. 9(a). The energy gener-ated by the first stage PRO process is represented by the area of the leftdashed rectangle, and the energy generated by the second stage is de-noted by the area of the smaller dashed rectangle. It is noted that,with the same water availability at optimal operation, the operationand performance of the first stage PRO process in the continuous feedtwo-stage configuration is exactly the same as that of the single-stagePRO process. In other words, the energy generated by the secondstage PRO process is the extra energy from the continuous feed two-stage PRO process as compared to the single PRO process, which is thefurther extraction from unused energy of the first stage PRO process.

draulic pressure applied on the draw solution, dimensionless flow rates and water perme-, whilst 0.5 and 0.8 are shown in (b) and (c), respectively.

Fig. 8. The schematic illustration of two-stage PRO process. Depending on the feed watercondition, continuous feed two-stage PRO process (a) and divided feed two-stage PROprocess (b) are illustrated.

Fig. 10. Energy generation difference between continuous feed two-stage PRO and singlePRO process with respect to different operational pressure on the draw water in the firststage PRO process.

92 W. He et al. / Desalination 350 (2014) 86–94

From the viewpoint of overall energy generated from a continuousfeed two-stage PRO process, being operated at its optimal conditionseparately, the overall C-PRO energy extraction is not guaranteed tomeet the global maximum. If the operation of the first stage varies,the total C-PRO energy and performance of the second stage PRO willalso change. As in Fig. 9(b),when the applied pressure is increased, it re-sults in two varied performance in the case of both two stages. There-fore, an investigation is carried out to find the optimal operationwhich achieves the maximal C-PRO generation by a continuous feedtwo-stage PRO process. In order to emphasize the potential energy ca-pacity generated by the continuous feed two-stage PRO process, thecomparison between continuous feed two-stage PRO and single-stagePRO process is analysed on the basis of the same water condition. Theextra energy generated by a continuous feed two-stage PRO processcompared to the maximum extracted energy by the single-stage PRO(0.3335 kWh/m3) is depicted in Fig. 10. The draw solution is seawater(32,000 mg/L), the feed is freshwater (100 mg/L) and the volumetricratio is 0.5. From the figure, the energy extracted by the continuousfeed two-stage PROprocess ismore than themaximumenergy harvest-ed by the single-stage PRO formost of its operations. Furthermore, it canbe seen that the optimal conditions separately do not result in theoverall maximal energy generation of the two-stage PRO. If the appliedhydraulic pressure were to continue increasing, the total energy gener-ation would increase as well. The overall maximal energy generated bythe continuous feed two-stage PRO process is achieved at the appliedpressure equalling about 18 bar (respectively, the optimal applied pres-sure in the second stage is about 12.6 bar).

Fig. 9. Illustration of energy generated by the continuous feed two-stage PRO process. These re(32,000 mg/L), and a volumetric ratio of feed water of 0.5. In (a), the two processes are operatcondition to obtain the total optimum C-PRO energy.

4.2. Divided feed two-stage PRO process

In the case of the divided feed two-stage PRO process, the feedwateris used to generate power separately and the drawwater is used contin-uously by the two PRO processes. The energy generated by the dividedfeed two-stage PRO process with equally distributed feed water is illus-trated in Fig. 11. Each PRO process is operated at its optimal C-PRO ener-gy condition. Similarly, the energy generated by the first and the secondstage PRO process is represented by the area of two dashed rectangles.As indicated in the figure the divided feed two-stage PRO processtakes advantage of the two operations to rearrange the distribution ofthe energy. The extra energy is generated by reducing frictional lossand unused energy compared to that of the single-stage PRO process.

In divided feed water condition, the influence from the first stage tothe second stage is less significant than the one in continuous feed two-stage PRO process because only draw solution is utilized continually.Therefore, an investigation is carried out on thewater distribution of di-vided feed two-stage PRO process when the two PRO processes are op-erated at their optimal C-PRO energy conditions. The water distribution

sults are achieved, with the feed water as fresh water (100 mg/L), drawwater as seawatered at their optimal C-PRO conditions. In (b), the two-stage PRO process is operated at the

Fig. 11. Illustration of energy generated by the divided feed two-stage PRO process. Thewater distribution factor is 0.5 whichmeans water is distributed evenly for two PRO pro-cesses. The results are achievedwith the feedwater as freshwater (100mg/L), drawwateras seawater (32,000 mg/L), and a volumetric ratio of feed water of 0.5. The two processesare operated at their optimal C-PRO conditions.

93W. He et al. / Desalination 350 (2014) 86–94

factor here is defined as the fraction of water utilized in the first stagePRO process. The divided feed two-stage PRO process with differentwater distribution factors is simulated and the results are presented inFig. 12.

The performance of divided feed two-stage PRO process in terms ofenergy generation with different water distribution is presented andcompared to that of the single-stage PRO process operated at its optimalC-PRO condition. From Fig. 12, it is evident that the energy generated bythe divided feed two-stage PRO process is larger than themaximum ex-tracted energy of the single-stage PRO process for all the operations. Themaximal C-PRO energy in this operation is achieved when about 42% ofthe feed water is allocated to the first stage PRO.

Therefore, based on the results, both continuous feed two-stage anddivided feed two-stage PRO processes have shown their ability to har-ness more energy from a certain level of water availability. It is neces-sary to note that more energy could be recovered with additionalstages but this would incur an additional economic cost.

Fig. 12. Energy generation difference between divided feed two-stage PRO and single PROprocess with respect to different water distribution factors on the draw water in the firststage PRO process. V1

F and V2F are the volume distributed to the first and the second

feed stage, respectively.

5. Conclusions

In this study, the discharge behaviour of water flux, power densityand energies involved in PRO process are explored to develop a compre-hensive understanding of the relationship between the power densityof the membrane and energy extraction from natural salinity gradientsby a C-PRO process. In addition, a further investigation is carried out onthe performance of two configurations of a two-stage PRO process, acontinuous feed two-stage PRO and a divided feed two-stage PRO pro-cess. Based on the results obtained, several conclusions can be drawn:1) water availability is a significant factor affecting the performance ofa PRO process, including the concentration and the volume of the feedand draw water available. 2) With a certain level of water availability,applied hydraulic pressure needs to be varied to achieve better perfor-mance with different operational conditions. And for different optimalobjectives, different applied pressure can be selected. 3) The operationsof a two-stage PRO processes is advantageous in harnessingmore ener-gy from the same level of water availability compared with a single-stage PRO process.

NomenclatureAm membrane area [m2]c concentration of solution [kg m−3]E extractable energy [kW]LP membrane water permeability [m3 m−2 h−1 Pa−1]JW water permeation flow rate [m3 m−2 h−1]P pressure [Pa]R gas constant [J mol−1 K−1]T temperature [K]V volumetric rate of solution [m3 h−1]W power density [Wm−2]ν number of species in the solutionπ osmotic pressure [Pa]

AbbreviationsPRO pressure retarded osmosisRO reverse osmosisRES renewable energy sourcesERD energy recovery deviceICP internal concentration polarizationECP external concentration polarizationRSF reverse osmosis flowTDS total dissolved solidR-PRO reversible PRO processC-PRO constant PRO process

References

[1] S. Fuss, J. Szolgayová, N. Khabarov, M. Obersteiner, Renewables and climate changemitigation: irreversible energy investment under uncertainty and portfolio effects,Energy Policy 40 (2012) 59–68.

[2] P. Grassini, K.G. Cassman, High-yield maize with large net energy yield and smallglobal warming intensity, Proc. Natl. Acad. Sci. 109 (2012) 1074–1079.

[3] P. Bajpai, V. Dash, Hybrid renewable energy systems for power generation in stand-alone applications: a review, Renew. Sust. Energ. Rev. 16 (2012) 2926–2939.

[4] B. Gadde, C. Menke, R. Wassmann, Rice straw as a renewable energy source in India,Thailand, and the Philippines: overall potential and limitations for energy contribu-tion and greenhouse gas mitigation, Biomass Bioenergy 33 (2009) 1532–1546.

[5] H.C. Kim, V. Fthenakis, J.-K. Choi, D.E. Turney, Life cycle greenhouse gas emissions ofthin-film photovoltaic electricity generation, J. Ind. Ecol. 16 (2012) S110–S121.

[6] H. Bevrani, A. Ghosh,G. Ledwich, Renewable energy sources and frequency regulation:survey and new perspectives, Renew. Power Gener. IET 4 (2010) 438–457.

[7] G.L. Park, A.I. Schäfer, B.S. Richards, Renewable energy-powered membranetechnology: supercapacitors for buffering resource fluctuations in a wind-poweredmembrane system for brackish water desalination, Renew. Energy 50 (2013)126–135.

[8] A. Schäfer, A. Broeckmann, B.S. Richards, Renewable energypoweredmembrane tech-nology. 1. Development and characterization of a photovoltaic hybrid membrane sys-tem, Environ. Sci. Technol. 41 (2006) 998–1003.

94 W. He et al. / Desalination 350 (2014) 86–94

[9] R.E. Pattle, Production of electric power bymixing fresh and salt water in the hydro-electric pile, Nature 174 (1954) 660-660.

[10] F. La Mantia, M. Pasta, H.D. Deshazer, B.E. Logan, Y. Cui, Batteries for efficient energyextraction from a water salinity difference, Nano Lett. 11 (2011) 1810–1813.

[11] S.E. Skilhagen, J.E. Dugstad, R.J. Aaberg, Osmotic power—power production based onthe osmotic pressure difference between waters with varying salt gradients, Desali-nation 220 (2008) 476–482.

[12] G. Han, S. Zhang, X. Li, T.-S. Chung, High performance thin film composite pressureretarded osmosis (PRO) membranes for renewable salinity-gradient energy genera-tion, J. Membr. Sci. 440 (2013) 108–121.

[13] S.E. Skilhagen, Osmotic power—a new, renewable energy source, Desalin. WaterTreat. 15 (2010) 271–278.

[14] T.Y. Cath, M. Elimelech, J.R. McCutcheon, R.L. McGinnis, A. Achilli, D. Anastasio, A.R.Brady, A.E. Childress, I.V. Farr, N.T. Hancock, J. Lampi, L.D. Nghiem, M. Xie, N.Y. Yip,Standardmethodology for evaluating membrane performance in osmotically drivenmembrane processes, Desalination 312 (2013) 31–38.

[15] J.T. Arena, B. McCloskey, B.D. Freeman, J.R. McCutcheon, Surface modification of thinfilm composite membrane support layers with polydopamine: enabling use of re-verse osmosis membranes in pressure retarded osmosis, J. Membr. Sci. 375 (2011)55–62.

[16] Y.C. Kim, M. Elimelech, adverse impact of feed channel spacers on the performanceof pressure retarded osmosis, Environ. Sci. Technol. 46 (2012) 4673–4681.

[17] Y. Gao, Y.-N. Wang, W. Li, C.Y. Tang, Characterization of internal and external con-centration polarizations during forward osmosis processes, Desalination 338(2014) 65–73.

[18] A. Achilli, T.Y. Cath, A.E. Childress, Power generation with pressure retarded osmo-sis: an experimental and theoretical investigation, J. Membr. Sci. 343 (2009) 42–52.

[19] Q.H. She, Y.K.W. Wong, S.F. Zhao, C.Y.Y. Tang, Organic fouling in pressure retardedosmosis: experiments, mechanisms and implications, J. Membr. Sci. 428 (2013)181–189.

[20] R.W. Field, J.J. Wu, Mass transfer limitations in forward osmosis: are some potentialapplications overhyped? Desalination 318 (2013) 118–124.

[21] M. Kostoglou, A.J. Karabelas, Comprehensive simulation of flat-sheet membrane el-ement performance in steady state desalination, Desalination 316 (2013) 91–102.

[22] D.E. Wiley, D.F. Fletcher, Computational fluid dynamics modelling of flow and per-meation for pressure-driven membrane processes, Desalination 145 (2002)183–186.

[23] Y.C. Kim, M. Elimelech, Potential of osmotic power generation by pressure retardedosmosis using seawater as feed solution: analysis and experiments, J. Membr. Sci.429 (2013) 330–337.

[24] J.R. McCutcheon, R.L. McGinnis, M. Elimelech, A novel ammonia–carbon dioxide for-ward (direct) osmosis desalination process, Desalination 174 (2005) 1–11.

[25] J.W. Post, J. Veerman, H.V.M. Hamelers, G.J.W. Euverink, S.J. Metz, K. Nymeijer, C.J.N.Buisman, Salinity-gradient power: evaluation of pressure-retarded osmosis and re-verse electrodialysis, J. Membr. Sci. 288 (2007) 218–230.

[26] K.S. Spiegler, Y.M. El-Sayed, The energetics of desalination processes, Desalination134 (2001) 109–128.

[27] T. Thorsen, T. Holt, The potential for power production from salinity gradients bypressure retarded osmosis, J. Membr. Sci. 335 (2009) 103–110.

[28] N.Y. Yip,M. Elimelech, Thermodynamic and energy efficiency analysis of power gen-eration from natural salinity gradients by pressure retarded osmosis, Environ. Sci.Technol. 46 (2012) 5230–5239.

[29] A. Zhu, P.D. Christofides, Y. Cohen, Effect of thermodynamic restriction on energycost optimization of RO membrane water desalination, Ind. Eng. Chem. Res. 48(2008) 6010–6021.

[30] A. Achilli, A.E. Childress, Pressure retarded osmosis: from the vision of Sidney Loebto the first prototype installation—review, Desalination 261 (2010) 205–211.

[31] E. Sivertsen, T. Holt, W. Thelin, G. Brekke, Modelling mass transport in hollow fibremembranes used for pressure retarded osmosis, J. Membr. Sci. 417–418 (2012)69–79.

[32] M.H. Sharqawy, L.D. Banchik, J.H. Lienhard V, Effectiveness-mass transfer units(ε-MTU) model of an ideal pressure retarded osmosis membrane mass exchanger,J. Membr. Sci. 445 (2013) 211–219.

[33] L.D. Banchik, M.H. Sharqawy, J.H. Lienhard V, Limits of power production due to fi-nite membrane area in pressure retarded osmosis, J. Membr. Sci. (2014).

[34] S. Lin, A.P. Straub, M. Elimelech, Thermodynamic limits of extractable energy bypressure retarded osmosis, Energy Environ. Sci. 7 (2014) 2706–2714.

[35] J.R. McCutcheon, M. Elimelech, Influence of membrane support layer hydrophobic-ity on water flux in osmotically driven membrane processes, J. Membr. Sci. 318(2008) 458–466.

[36] N.Y. Yip, M. Elimelech, Performance limiting effects in power generation from salin-ity gradients by pressure retarded osmosis, Environ. Sci. Technol. 45 (2011)10273–10282.

[37] K.S. Spiegler, O. Kedem, Thermodynamics of hyperfiltration (reverse osmosis):criteria for efficient membranes, Desalination 1 (1966) 311–326.

[38] K.L. Lee, R.W. Baker, H.K. Lonsdale, Membranes for power generation by pressure-retarded osmosis, J. Membr. Sci. 8 (1981) 141–171.

[39] J. Kim, M. Park, S.A. Snyder, J.H. Kim, Reverse osmosis (RO) and pressure retardedosmosis (PRO) hybrid processes: model-based scenario study, Desalination 322(2013) 121–130.

[40] N.Y. Yip, A. Tiraferri, W.A. Phillip, J.D. Schiffman, L.A. Hoover, Y.C. Kim, M. Elimelech,Thin-film composite pressure retarded osmosis membranes for sustainable powergeneration from salinity gradients, Environ. Sci. Technol. 45 (2011) 4360–4369.

[41] G.Z. Ramon, B.J. Feinberg, E.M.V. Hoek, Membrane-based production of salinity-gradient power, Energy Environ. Sci. 4 (2011) 4423–4434.

[42] A. Tiraferri, N.Y. Yip, W.A. Phillip, J.D. Schiffman, M. Elimelech, Relating performanceof thin-film composite forward osmosis membranes to support layer formation andstructure, J. Membr. Sci. 367 (2011) 340–352.

[43] G. Han, P. Wang, T.-S. Chung, Highly robust thin-film composite pressure retardedosmosis (PRO) hollow fiber membranes with high power densities for renewablesalinity-gradient energy generation, Environ. Sci. Technol. 47 (2013) 8070–8077.