Energy recovery from two-stage SWRO plant using PRO without external freshwater feed stream: theoretical analysis.

PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 1 Energy recovery from two-stage SWRO plant using PRO 1 without external freshwater feed stream: theoretical analysis1 2 3 Khaled Touatia,b* 4 aDepartment of Systems Engineering and Automatic Control, University of Valladolid, 5 47011, Valladolid, Spain. Tel: +34 983423162; Fax: +34 98342316. 6 bLaboratory of Natural Water TreatmentWater Researches and Technologies Center, 7 Techno-park Borj Cedria, BP 273, 8020, Soliman, Tunisia. Tel: + 216 79325122 / 8 79325199; Fax: + 216 79325802.*Corresponding author: [email protected] 9 10 Jacobo Salamancaa 11 aDepartment of Systems Engineering and Automatic Control, University of Valladolid, 12 47011, Valladolid, Spain. Tel: +34 983423162; Fax: +34 983423161. 13 ferna[email protected] 14 15 Hamza Elfilb 16 bLaboratory of Natural Water TreatmentWater Researches and Technologies Center, 17 Technopark Borj Cedria, BP 273, 8020, Soliman, Tunisia. Tel: + 216 79325122 / 79325199; 18 Fax: + 216 79325802. [email protected] 19 Fernando Tadeoa 20 aDepartment of Systems Engineering and Automatic Control, University of Valladolid, 21 47011, Valladolid, Spain. Tel: +34 983423162; Fax: +34 983423161. 22 ferna[email protected] 23 24 25 26 27 28 29 30 31 1 Funded by Mineco Project DPI2014-54530-R and FEDER funds PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 2 1 Abstract: 2 Research into pressure retarded osmosis (PRO) as a method to extract energy from salinity 3 gradients is on the rise. Seawater Reverse Osmosis (SWRO) is now a leading technology in 4 the desalination industry worldwide, in both small and large scale applications, due to the 5 remarkable improvements in membrane performance and associated energy efficiency. 6 Nonetheless, SWRO desalination is inherently more energy intensive when compared to 7 conventional fresh water treatment. The integration of PRO with SWRO systems is studied 8 in terms of energy consumption and effluent changes. For this, two novel integration 9 designs are evaluated, with SWRO-PRO specific energy consumption being modeled using 10 SWRO conditions at the thermodynamic restriction, and a developed PRO model. The 11 results show lower SWRO energy consumption for both configurations, with a reduction in 12 consumption of 12% to 18%, depending on the RO recovery ratios. Lastly, the effect of the 13 initial flow ratio on the dilution factor has been studied. To do so, the dilution was modeled 14 and studied for different operating conditions. It was found that detrimental effects severely 15 reduce the dilution, especially the internal concentration polarization, which induces a 16 decrease of energy recovery when using the PRO process. 17 Keywords: Pressure Retarded Osmosis; Seawater Reverse Osmosis; Energy recovery; 18 Modeling; Dilution Factor. 19 20 21 22 23 24 25 26 27 28 PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 3 Energy recovery from two-stage SWRO plant using PRO 1 without external freshwater feed stream: theoretical analysis. 2 3 1. Introduction 4 Desalination technologies are quickly growing, combining engineering and science to develop 5 innovative means for drinkable water production [1]. In fact, several countries, especially in 6 the Middle East, already depend on seawater desalination as the main source of drinking 7 water [2,3], so desalination plants have seen considerable expansion over the past decade: the 8 desalination capacity is expected to reach about 100 million cubic meters per day at the end of 9 2016 [2]. The main challenge in desalination is the use of energy: As a drinking water 10 treatment technology, seawater desalination requires more energy than conventional fresh 11 water treatment methods [3]. However, the power consumption is frequently inaccurately 12 represented when compared to other treatment technologies that provide safe and reliable 13 public water supply [4, 5]. Typically, the energy consumption represents 44% of the total 14 water cost of a SWRO plant [6]. Another challenge in the desalination industry is the handling 15 of reject brine, which is the highly concentrated by-product of the desalination process [7,8]. 16 Osmotic Energy systems were proposed as a solution due to the fact that it has been found to 17 be very promising, with the potential of reducing the cost of seawater desalination as well as 18 the environmental impact from brine discharge [9,10,11]. In other words, as a SWRO system 19 produces high water concentration (brine) and PRO uses solutions with high concentrations, it 20 would be beneficial to hybridize the two processes due to the reliance of the two membrane 21 processes on the concentration of solutions. In addition, The SWRO-PRO system has several 22 advantages. Compared to an optimized stand-alone SWRO system (SWRO with a recovery 23 energy device), SWRO energy consumption is reduced by energy production using PRO 24 [12]. Another advantage of this system is that the brine generated during the SWRO process 25 is diluted back to seawater concentration, thus minimizing the adverse environmental 26 impact that seawater RO brine disposal can have on marine ecology/habitats. Furthermore, 27 the impaired water and SWRO product water are in separate circuits, so there is no contact 28 between impaired water and drinking water. Compared to a stand-alone PRO system (i.e., 29 river-to-sea PRO), PRO energy production in the SWRO-PRO system is augmented by the 30 higher concentration of the draw solution (SWRO brine). Another key advantage of the 31 SWRO-PRO system is that the influent draw solution is pre-treated by the RO pre-treatment 32 system. Thus, the brine entering the PRO sub-system is relatively free of foulants. This 33 PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 4 draw solution pre-treatment in the SWRO-PRO system avoids additional energy 1 expenditure that would be necessary in a stand-alone PRO system. As the energy and 2 chemical costs of pre-treating seawater is a substantial operating expenditure in SWRO 3 desalination [13], the fact that SWRO-PRO capitalizes on the pre-treated brine reduces 4 some pre-treatment concerns, which can be significant in the PRO process. As a 5 consequence of these encouraging advantages, the integration of PRO into SWRO has 6 attracted the attention of many researchers and several studies are currently involved in 7 investigating the feasibility of SWRO-PRO systems [14,15,16,17,18]. In 2010, Japan 8 launched the Megaton water system. As part of the project, a prototype SWRO-PRO hybrid 9 plant was built and operated. Recycled water was supplied from a regional sewage treatment 10 facility and concentrated brine from a SWRO plant, using PRO hollow fiber modules. The 11 prototype PRO plant got the maximum output power density of 13.8 W/m2 at 30 bars of 12 hydraulic pressure difference, corresponding to 38% permeation of pure water into the brine 13 [14]. Another study was carried out using an experimental pilot system, designed and 14 constructed to investigate the reduced SWRO energy consumption by its integration with 15 PRO [15,16]. The experiments showed that the enhanced power densities for the RO-PRO 16 system ranged from 1.1 to 2.3 W/m2 and indicated that future RO-PRO systems may reduce 17 the specific energy consumption requirements for desalination by 1 kWh/m3 [16]. Another 18 work investigated the feasibility of a RO desalination system powered by a stand-alone PRO 19 unit [17]. Unfortunately, the study did not take into consideration the effect of the 20 concentration polarization and the salt reverse flux on the performance of the SWRO-PRO 21 system, which would affect the results. Seawater brine from the TuaSpring desalination plant 22 and wastewater retentate from the NEWater plant were used in [18] for energy recovery. 23 Experiments gave a power density of 4.6 W/m2 at 20 bar obtained with seawater brine as the 24 draw solution and waste water as the feed solution. A recent paper investigated two SWRO-25 PRO designs based on the positions of the pressure exchangers and the pump, then studied as 26 a function of SWRO recovery [19]. 27 However, most of the previously studied SWRO-PRO cases deal with conventional 28 integration design, where one-stage SWRO brine is the PRO draw solution and a river 29 water/wastewater effluent is the PRO feed solution. This integration methodology reveals 30 that the amount of energy produced is relatively low and requires a feed solution from a low 31 salinity water source, which increases the energy cost (pre-treatment, installations, etc…). 32 Moreover, previous studies have been based on supplying the PRO with external freshwater 33 sources. However, it is well known that the world is facing a serious fresh water scarcity 34 PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 5 problem [20]. Then, the use of fresh water for energy recovery should be avoided. 1 Furthermore, the use of wastewater effluents shows low energy recovery performance, due 2 to the severe membrane fouling and the necessity for pre-treatment, which increases the 3 energy consumption [18]. In this study, for the first time, an operational two-stage SWRO 4 unit is integrated with PRO to reduce its energy consumption. Herein, the use of an external 5 feed water source is avoided, and an alternative location of the PRO inside the process is 6 proposed. To do so, two novel SWRO-PRO designs have been proposed in terms of the 7 PRO location and the nature of the feed solution. To accomplish this objective, a model for 8 the PRO system has been developed, taking into consideration the membrane 9 characteristics, reverse salt flux, concentration polarization, and pressure drop in the 10 membrane module. The SWRO specific energy consumption is calculated by considering 11 the SWRO at its thermodynamic limit, using models from the literature [21]. Then, the 12 SWRO-PRO model was developed by combining the SWRO and PRO models. In addition, 13 the results of the model were presented for both configurations considered and compared to 14 the ideal case. Lastly, the dilution factor of the draw solution was modeled and investigated 15 as a function of PRO detrimental effects. 16 17 2. Material and method 18 2.1. Description of the SWRO desalination plant 19 To investigate the feasibility of PRO integration with the SWRO unit, a case study is 20 considered here. A Reverse-Osmosis based desalination unit is used, intended for producing 21 water for the electrolyzation process, and which is composed of the elements presented in 22 Fig.1. This plant was developed by SETA S-L as a part of the H2OCEAN project [22]. The 23 desalination unit is based on two independent lines, divided into two stages, which can be 24 connected or disconnected, as required for maintenance or operational requirements. In order 25 to reduce the number of spare parts required, both lines have the same components. The 26 procedure is started by a pre-treatment composed of three steps: chlorination to avoid organic 27 matter, ultra-filtration to block metals and particles in suspension, and finally, a backwash to 28 eliminate foulants accumulated on the membrane surfaces during the process. 29 The first pass of the SWRO unit consists of a chemical treatment applied to remove the 30 residual chlorine from the pre‐treatment, then bisulphite is used (to remove the oxidants 31 dissolved in the water and provide a bacteriostatic effect) and, finally, antifouling is used to 32 mitigate salt precipitation on the membrane surfaces to avoid the increase in energy 33 PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 6 consumption. As a safety system to avoid damaging the membrane, a micro filter is installed 1 just before the High Pressure Pump (HP) with a degree of 5 microns of filtration. To feed the 2 first-pass membranes, an HP is required; the water produced is stored in a water tank, whereas 3 the brine goes to an Energy Recovery system (Pressure exchanger PX). In routine operational 4 conditions, the recovery factor is arbitrarily selected to be 45%. 5 The Second RO Pass consists of two phases, it is first dosed with an antifouler designed to 6 avoid salt precipitation and a micro filter is installed with a degree of 5 microns of filtration, 7 as a safety system to avoid membrane fouling. Second, to feed the second-pass, the RO 8 requires an HP to pressurize the water before it enters the membrane. 9 The reverse osmosis recovery is around 70%. The water produced is stored in a DEMI water 10 tank, whereas the brine goes to an energy recovery system before being reused in the 11 proposed osmotic energy recovery system or being returned to the Ultra-Filtration Tank. 12 13 14 15 16 17 18 19 20 21 22 Fig.1: Two-stage Reverse Osmosis desalination unit 23 24 2.2. SWRO-PRO configurations and integration methodologies 25 2.2.1. First SWRO-PRO configuration 26 A simplified presentation of the first SWRO-PRO configuration is illustrated in Fig.2: the 27 seawater feed solution (Qsw) is first pre-pressurized in the pressure exchanger PX prior to 28 entering the desalination process. Exiting the first stage SWRO sub-system (RO1) are two 29 streams: a freshwater permeate stream (Qp) and a concentrated brine stream (QR1). QSR1 is 30 then depressurized to approximately half its pressure to reach an adequate condition for the 31 1st stage RO 2nd stage RO Pressure exchangers Pre-treated seawater Brine to the sea To UF tank To industrial process PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 7 PRO process [7]. The permeate of the RO1 feeds the second stage RO sub-system (RO2). To 1 recover the brine energy, an isobaric or turbocharged device could be used; alternatively, a 2 turbine could be employed to convert it into electrical energy. Following this 3 depressurization, the brine stream enters the PRO sub-system as a high salinity (draw) 4 solution (QR1 = QD). The feed solution for the PRO sub-system (QF = QR2) is the retentate of 5 the second stage. Through osmosis, the pressurized draw solution extracts water from the 6 impaired water source under isobaric conditions, resulting in a diluted draw solution (QDR). 7 The energy stored in the diluted draw solution is then exchanged with the seawater RO feed 8 prior to discharge in order to recover its potential energy and increase the energy savings of 9 the SWRO-PRO system. The PRO feed solution bleed QFR is rejected to the sea. 10 11 12 13 14 15 16 17 18 19 20 Fig. 2: First configuration SWRO-PRO integration: Standard with brine mixture. Darker 21 colors correspond to more concentrated solutions and the thickness of each arrow denotes the 22 approximate flow rate. 23 24 2.2.2. Second SWRO-PRO configuration 25 In the second configuration, the feed solution entering the PRO sub-system was changed. The 26 retentate of the second stage (QR2) was mixed with an additive pre-treated seawater flow 27 (Qad); the sum embodies the PRO feed solution. The amount of seawater flow is chosen to 28 RO1 RO2 PRO P X Qs Qp QR1 Q’ p QR2 QDR QFR ERD Q D T h e QF = QR2 PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 8 guarantee the condition that the feed and draw solution flows are equal (QF = Qad+QR2 = QD). 1 For this, a controllable valve (V) is placed to provide the desired amount of Qad. The draw 2 solution of the PRO is the brine of the first RO stage; this brine passes through the energy 3 recovery device (ERD) to adjust its pressure, when necessary, to a suitable applied pressure 4 value (which is theoretically the optimum pressure value to be applied to the draw solution in 5 the PRO process). It should be noted that the functioning of the ERD in this configuration is 6 studied later, in section 2.4.3. The exiting PRO draw solution flow is then conducted to the 7 PX to recover the pressure to the feed SWRO flow (Qsw). The PRO feed solution bleed QFR is 8 rejected to the sea. Fig.3 illustrates the second SWRO-PRO configuration. It should be 9 pointed out that the choice of seawater as an added flow to the entering PRO feed solution is 10 based on the fact that the SWRO plant is placed near the sea, and no river water is close to the 11 SWRO facility. 12 13 14 15 16 17 18 19 20 21 Fig. 3: Second configuration SWRO-PRO integration: standard with second brine-seawater 22 mixture. Darker colors correspond to more concentrated solutions and the thickness of each 23 arrow denotes the approximate flow rate. 24 25 2.3. Modeling the ideal energy consumption of the SWRO-PRO system 26 2.3.1. Two-stage Reverse Osmosis (SWRO) 27 The theoretical energy consumption is the minimum amount of energy required to produce a 28 desired volume of permeate. More precisely, the theoretical Specific Energy Consumption 29 RO1 RO2 PRO PX Qad Q p QR1 Q’p Q R QD Q D T h e se ERD Qsw QF =QR2+Qad T h e QFR (V) PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 9 (SEC) gives, for a given recovery rate, the absolute minimum amount of energy by assuming 1 that the efficiency of every component (pumps, motors, Energy Recovery Devices, etc.) is 2 100%. At the theoretical limit of constant-pressure operation, the one-stage reverse osmosis 3 (RO) system will operate with an applied hydraulic pressure equal to the final osmotic 4 pressure of the brine exiting the RO module. Thus, the minimum specific energy of 5 desalination for a one-stage RO process, SERO,desal, is equal to the final brine osmotic 6 pressure [21,30]: In our case, considering two RO stages, the total theoretical energy 7 consumption of the two-stage RO system (𝑆𝑆𝑆𝑆𝑆𝑆𝑅𝑅𝑅𝑅1−𝑅𝑅𝑅𝑅2,𝑡𝑡ℎ𝑒𝑒𝑒𝑒) is the sum of the energy 8 consumption of each stage: 9 10 𝑆𝑆𝑆𝑆𝑆𝑆𝑅𝑅𝑅𝑅1−𝑅𝑅𝑅𝑅2,𝑡𝑡ℎ𝑒𝑒𝑒𝑒= 𝜋𝜋𝑠𝑠𝑠𝑠�1 1−𝑌𝑌1+𝑌𝑌1 1−𝑌𝑌2� (1) 11 12 where 𝑌𝑌1 and 𝑌𝑌2 are the recoveries of RO1 and RO2, respectively. The salt rejection coefficient 13 Rs for both RO stages is defined as: 14 15 𝑅𝑅𝑠𝑠1= 1 −𝐶𝐶𝑝𝑝 𝐶𝐶𝑠𝑠𝑠𝑠 (2) 16 17 𝑅𝑅𝑠𝑠2= 1 −𝐶𝐶𝑝𝑝′ 𝐶𝐶𝑝𝑝 (3) 18 19 where 𝑆𝑆𝑝𝑝 and 𝑆𝑆𝑝𝑝′ are the salt concentration in the permeate solutions of the first and second 20 stages, respectively. For simplicity, the salt rejection coefficients are considered equal 21 (Rs1=Rs2=Rs). Throughout the study, a linear relation between salt concentration and osmotic 22 pressure, based on the van't Hoff equation, is assumed (π=βRTC). This assumption makes the 23 development of simple analytical expressions easy, without a significant compromise in 24 accuracy, because a concentration relevant to seawater (Csw=35g/L) is low enough for the 25 van't Hoff equation to be reasonably accurate. Fig. S1 (Supplementary material) describes the 26 minimum separation energy for a two-stage SWRO as a function of Y1 according to Eq. (1), 27 which is achievable only in an ideal, reversible, thermodynamic process. In this case, since 28 the majority of the energy is consumed by the first stage, the recovery of the second stage is 29 maintained equal to 70% throughout the study. It should be pointed out that the minimum 30 specific energy does not include the energy required to generate excess pressure in the 31 module, or the energy spent on pre-treatment or post-treatment. As can be seen in Fig. S1, the 32 PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 16 full-scale membranes suffer from several drawbacks, such as concentration polarization, 1 fouling, reverse salt diffusion, and pressure drops, which make, for instance, high dilution 2 unreachable. Thus, in SWRO-PRO hybrid systems, three parameters should be properly 3 controlled, namely: the recovery ratio, the dilution factor, and the PRO entering flow ratio. 4 5 Fig.7: Ideal specific energy consumption of SWRO-PRO system as a function of PRO 6 dilution for several RO1 recoveries. Fig.7.A represents the model results for the SWRO-PRO 7 system operating under the first configuration. Fig.7.B represents the model results for the 8 SWRO-PRO system operating under the second configuration. 9 10 2.4. Modeling the energy consumption of the SWRO-PRO system under realistic conditions 11 2.4.1. PRO model 12 Real life SWRO-PRO plants would be less energy efficient than theoretically calculated, due 13 to electrical conversion losses and dissipation. The amount of additional energy required 14 depends on the specific recovery strategy. In addition, the produced energy is proportional to 15 the mixing volume. Consequently, dilution of the draw solution is inevitable. Therefore, the 16 osmotic pressure difference decreases and reduces the performance of PRO. Besides that, 17 previous works had shown that PRO membrane performance is limited by several limiting 18 factors [26,27]. In fact, with a realistic membrane and imperfect hydrodynamics, three 19 phenomena occur to reduce the trans-membrane water flux, as shown in Fig.S4. First, the 20 porous support layer induces internal concentration polarization (ICP), which detrimentally 21 enhances πF,m by increasing the solute concentration at the feed-membrane interface, thus 22 reducing the trans-membrane driving force. Second, without perfect hydrodynamics in the 23 draw solution flow channel, dilutive external concentration polarization (ECP) develops, 24 which lowers πD,m and reduces the driving force. Lastly, because the membrane is no longer 25 1 2 3 4 010 20 30 40 50 Specific energy ( kWh/m3) DF (%) Y1 = 30% Y1 = 40% Y1 = 50% Y1 = 60% A ‐6 ‐4 ‐2 0 2 4 020 40 60 80 100 Specific energy ( kWh/m3) DF (%) Y1 = 30% Y1 = 40% Y1 = 50% Y1 = 60% B DFmax PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 17 perfectly selective, reverse salt diffusion (RSD) takes place, resulting in uncontrolled mixing 1 and, therefore, reduced energy extraction in the process. The water flux across the membrane, 2 Jw, can be defined in terms of the membrane water permeability coefficient, A; the osmotic 3 pressure at the draw side of the membrane active layer, πD,m; the osmotic pressure at the feed 4 side of the membrane active layer, πicp; and the hydraulic pressure difference across the 5 membrane: 6 𝐽𝐽𝑠𝑠=𝐴𝐴�𝜋𝜋𝐷𝐷,𝑚𝑚−𝜋𝜋𝑖𝑖𝑖𝑖𝑝𝑝−∆𝑃𝑃�=𝐴𝐴(∆𝜋𝜋𝑚𝑚−∆𝑃𝑃) (13) 7 where Δπm is the osmotic pressure difference across the membrane active layer. Based on film 8 theory, equations have been developed to determine the concentration on either side of the 9 membrane with reverse salt flux and concentration polarization accounted for [25, 26]: 10 𝑆𝑆𝐷𝐷,𝑚𝑚=�𝑆𝑆𝐷𝐷,𝑏𝑏+𝐽𝐽𝑠𝑠 𝐽𝐽𝑠𝑠�𝑒𝑒𝑒𝑒𝑒𝑒�−𝐽𝐽𝑠𝑠 𝑘𝑘𝐷𝐷�−𝐽𝐽𝑠𝑠 𝐽𝐽𝑠𝑠 (14) 11 𝑆𝑆𝐷𝐷,𝑚𝑚=�𝑆𝑆𝐷𝐷,𝑏𝑏+𝐽𝐽𝑠𝑠 𝐽𝐽𝑠𝑠�𝑒𝑒𝑒𝑒𝑒𝑒�𝐽𝐽𝑠𝑠 𝑘𝑘𝐹𝐹�−𝐽𝐽𝑠𝑠 𝐽𝐽𝑠𝑠 (15) 12 𝑆𝑆𝑖𝑖𝑖𝑖𝑝𝑝=�𝑆𝑆𝐷𝐷,𝑏𝑏+𝐽𝐽𝑠𝑠 𝐽𝐽𝑠𝑠�𝑒𝑒𝑒𝑒𝑒𝑒�𝐽𝐽𝑠𝑠 𝑘𝑘𝐹𝐹�𝑒𝑒𝑒𝑒𝑒𝑒(𝐽𝐽𝑠𝑠𝐾𝐾)−𝐽𝐽𝑠𝑠 𝐽𝐽𝑠𝑠 (16) 13 where CD,b is the bulk draw concentration, CF,b is the bulk feed concentration, and kD and kF 14 are the mass transfer coefficient in the draw and feed sides, respectively. K is the solute 15 diffusion coefficient. The water flux and the salt flux are related by the following relation 16 [27]: 17 𝐽𝐽𝑠𝑠 𝐽𝐽𝑠𝑠=𝐵𝐵 𝐴𝐴𝐴𝐴𝑅𝑅𝑇𝑇�1 + 𝐴𝐴∆𝑃𝑃 𝐽𝐽𝑠𝑠� (17) 18 19 where B is the salt permeability coefficient, and ΔP is the hydraulic pressure applied on the 20 draw solution side. Assuming the van’t Hoff relationship between osmotic pressure and 21 concentration [25]: 22 23 𝐽𝐽𝑠𝑠=𝐴𝐴��𝜋𝜋𝐷𝐷,𝑏𝑏+𝐵𝐵 𝐴𝐴�1 + 𝐴𝐴∆𝑃𝑃 𝐽𝐽𝑠𝑠��𝑒𝑒𝑒𝑒𝑒𝑒�−𝐽𝐽𝑠𝑠 𝑘𝑘𝐷𝐷�−�𝜋𝜋𝐷𝐷,𝑏𝑏+𝐵𝐵 𝐴𝐴�1 + 𝐴𝐴∆𝑃𝑃 𝐽𝐽𝑠𝑠��𝑒𝑒𝑒𝑒𝑒𝑒(𝐽𝐽𝑠𝑠𝐾𝐾)𝑒𝑒𝑒𝑒𝑒𝑒�𝐽𝐽𝑠𝑠 𝑘𝑘𝐹𝐹�−∆𝑃𝑃� (18) 24 25 The variations along the membrane's length are often neglected in models designed to 26 simulate bench scale systems because the effect is difficult to observe over very small 27 membrane samples [26]. However, the passage of water through the membrane was followed 28 PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 18 by friction, which should cause some heating of solutions. If the water flow had been created 1 due to the pressure drop on either side of the membrane, this would have led to the loss of part 2 of the energy released. These include flow rates, concentrations and hydraulic pressures, as 3 well as all other variables that are dependent on them. Their local values should be considered 4 for accurate modeling. These spatial variations can be accounted for either by taking an 5 average of inlet and outlet variables, or by considering the membrane as a finite difference 6 model [26,27]. To develop a model for full-scale PRO applications, the flat sheet membrane 7 area is divided into segments perpendicular to the water flow, evaluating flow conditions at 8 specific points along the membrane module. In this case, feed and draw solution flows are 9 assumed to be in a co-current flow mode. Fig.S3 illustrates a single segment of a flat sheet 10 PRO membrane. The permeate flow through the PRO membrane Qw can significantly dilute 11 the draw solution concentration CD, which results in reduced flux performance compared to 12 the small size test membrane. Such a dilution effect needs to be explicitly accounted for in a 13 flat sheet module. Due to this variation, three main parameters are evaluated at each point 14 along the membrane: the water flux, the pressure and the concentrations. The model 15 development steps are presented in the Supplementary material. 16 17 2.4.2. SWRO model 18 The energy consumption of each RO stage in a two-stage RO plant, at the limit of the 19 thermodynamic restriction and in the absence of energy recovery, is developed in [21]: 20 21 SECRO1=Rsπsw ηP1Y1(1−Y1) (19) 22 23 SECRO2=Rsπp ηP2Y2(1−Y2) (20) 24 25 where Rs is the salt rejection, and 𝜂𝜂𝑃𝑃 is the pump conversion. The specific energy cost for RO, 26 in the presence of an energy recovery device (ERD), operating at the limit of the 27 thermodynamic restriction, is: 28 29 𝑆𝑆𝑆𝑆𝑆𝑆𝑅𝑅𝑅𝑅1 𝐸𝐸𝑅𝑅𝐷𝐷 = 𝑅𝑅𝑠𝑠𝜋𝜋𝑠𝑠𝑠𝑠�1−𝜂𝜂𝐸𝐸𝐸𝐸𝐷𝐷[1−𝑌𝑌1] 𝜂𝜂𝑃𝑃𝑌𝑌1[1−𝑌𝑌1]� (21) 30 31 where 𝜂𝜂𝐸𝐸𝑅𝑅𝐷𝐷 is the efficiency of the energy recovery device. 32 PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 19 2.4.3. ERD energy recovery 1 The recovery of energy from SWRO systems has been a major factor in the reduction of the 2 cost of desalinated seawater, to a point where it is beginning to offer a challenge to 3 conventional sources. In our case, an ERD is placed after the first RO stage to recover energy 4 from the rejected brine. As mentioned previously, the ERD reduces the pressure of the brine 5 to a suitable value for PRO operational applied pressure ( 𝛥𝛥𝑃𝑃≈∆𝜋𝜋𝑚𝑚 2). This condition is a key 6 parameter for the optimum performance of the PRO process. In the second configuration, the 7 osmotic pressure of the PRO feed solution is relatively high at low RO1 recoveries, which is 8 not the case for the first configuration. Therefore, the energy recovered using ERD depends 9 on the osmotic pressure that enters the PRO module. As πF changes with the recovery, the 10 contribution of ERD in the second configuration also varies regarding the osmotic pressure of 11 the PRO feed solution. Then, the energy recovered by ERD in the second configuration will 12 be much lower than in the first one. Subtracting Eq. (21) from Eq.(19), the energy recovered 13 by the ERD at the thermodynamic restriction is expressed as follows: 14 15 𝑆𝑆𝑆𝑆𝐸𝐸𝑅𝑅𝐷𝐷 1 = −𝑅𝑅𝑠𝑠𝜋𝜋𝑠𝑠𝑠𝑠�𝜂𝜂𝐸𝐸𝐸𝐸𝐷𝐷[1−𝑌𝑌1] 𝜂𝜂𝑃𝑃𝑌𝑌1[1−𝑌𝑌1]� (22) 16 17 𝑆𝑆𝑆𝑆𝐸𝐸𝑅𝑅𝐷𝐷 2 = −𝑅𝑅𝑠𝑠[𝜋𝜋𝑠𝑠𝑠𝑠−𝜋𝜋𝐷𝐷]�𝜂𝜂𝐸𝐸𝐸𝐸𝐷𝐷[1−𝑌𝑌1] 𝜂𝜂𝑃𝑃𝑌𝑌1[1−𝑌𝑌1]� (23) 18 19 The subscripts 1 and 2 refer to the first and second configuration, respectively. The osmotic 20 pressure πF in Eq.(23) highlights the effect of the PRO feed osmotic pressure. As can be seen 21 in Fig.S5, the contribution of ERD to the energy recovery is much higher in the first 22 configuration. In fact, in usual RO recovery (40% ≤ Y1≤ 60%), the ERD energy recovery is 23 between 1.32 and 2 kWh/m3. Thus, the ERD effectively reduces the SEC by almost 50%. This 24 result is well-known in the literature [21]. Using the second configuration, the ERD 25 contribution is not significant due to the high pressure required for the PRO process. At high 26 recovery, the ERD energy becomes slightly significant due to the dilution of the added 27 seawater by the rejected water of the second RO stage. 28 29 30 31 32 PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 20 3. Results and discussion 1 3.1. Modeling parameters 2 The simulations in this study are based on the characteristics of a cellulose acetate flat-sheet 3 membrane used in our previous work, where a power density of 6.2 W/m2 at 13 bar has been 4 obtained in lab-scale tests using 1.026 M NaCl solution as the draw solution and 8.55mM 5 NaCl solution as the feed solution [25]. The characteristics of the membrane and the 6 hypothetical module parameters are summarized in Table S1. The initial seawater flow was 7 chosen arbitrarily to be 256m3/h, which is the real amount of treated water in the SETA-8 SWRO plant studied here. The sweater concentration is chosen to be 0.6M (35g/L). The 9 added flow rate Qad is considered to be pre-treated before being involved in the process. It 10 should be noted that the effect of organic fouling is not considered in this study, so the energy 11 produced may be lower in the presence of feed water charged with natural organic matter 12 (NOM). 13 3.2.PRO model results 14 The SEPRO model was investigated by studying the response of the PRO sub-system after the 15 variation in the effect of the operating conditions (draw and feed solution concentrations). As 16 can be seen in Fig.8, the increase in the draw solution concentration leads to an increase in the 17 energy recovery by the sub-system. On the other hand, the increase in the feed solution 18 concentration is followed by a decrease in the energy. Lastly, the increase in the draw solution 19 temperature improves the performance of the PRO due to the improvement in the effective 20 osmotic pressure and also the intrinsic membrane parameters, such as the water permeability 21 coefficient. These results are in agreement with the model behavior using a lab-scale PRO 22 membrane. Fig.9 illustrates the SEPRO in Eq. (S1), corresponding to the PRO model results for 23 the proposed configurations when diluting RO1 brine back to the sea, as a function of the 24 SWRO recovery using the cellulose acetate flat-sheet membrane with the characteristics 25 presented in Table S1. When recovery increases, the draw solution concentration increases, 26 and the SEPRO also increases. The minimum SEPRO recoveries are 0 kWh/m3 for both 27 configurations. This magnitude increases with RO1 recovery to reach an SEPRO production of 28 0.702 kWh/m3 at 77% of recovery. Compared to the theoretical SEPRO production (Fig. 5), the 29 model produces a specific energy remarkably far from the ideal case. 30 31 PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 21 1 Fig.8 PRO specific energy production (SEPRO) modeled as a function of feed and draw 2 solution concentrations under optimum PRO hydraulic pressure. Feed and draw solution 3 flows are considered equal. For (A): the feed solution concentration is 8.55mM. For (B): the 4 draw solution is 1.026M. T=25°C. 5 6 7 Fig.9: PRO specific energy production (SEPRO) modeled as a function of RO1 recovery for a 8 cellulose acetate flat-sheet membrane under optimal ΔP for both proposed configurations. 9 10 3.3.SWRO-PRO model results 11 The total energy consumption of the SWRO-PRO system for both configurations is studied 12 here. As mentioned previously, the study only takes into account the effect of the recovery 13 ratio of the first RO stage, so that of the second stage is considered constant (Y2=70%). 14 According to Fig.10 and Fig.11, the SESWRO-PRO consumption is much lower for the ideal case 15 than for the model results for both configurations. This behavior is due to the fact that the 16 ideal SEPRO production is noticeably higher than the model SEPRO production. Because of the 17 higher starting value of the model SESWRO-PRO and the lower rate at which the model SERO-PRO 18 0 0,4 0,8 1,2 1,6 0,4 0,8 1,2 1,6 2 -SEPRO (kWh/m3) Draw solution concentration (M) A 0 0,2 0,4 0,6 0,8 0 0,1 0,2 0,3 0,4 -SEPRO (kWh/m3) Feed solution concentration (M) B ‐0,8 ‐0,7 ‐0,6 ‐0,5 ‐0,4 ‐0,3 ‐0,2 ‐0,1 0 020 40 60 80 100 SEPRO (kWh/m3) Y 1 (%) configuration 1 configuration 2 PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 22 consumption decreases, the model of the SWRO-PRO system is not able to reach energy 1 neutrality (SESWRO-PRO = 0). Fig.10 shows the variation of the modeled energy consumption 2 SECSWRO-PRO with the applied pressure ΔP under two RO1 recovery ratios without using an 3 ERD. It can be seen that SECSWRO-PRO, when Y1= 40%, is higher than the case when Y1 = 50% 4 for both configurations. Moreover, at a low recovery ratio, the performance of the second 5 configuration is better than the first one in terms of energy generation and low applied 6 pressure. When the recovery ratio increases, the second configuration shows a better 7 performance than the first, due to the decrease in the PRO feed solution concentration. As a 8 comparison, for (A), the maximum energy recovery is 2.45 % and 4 % when Y1= 40% and 9 50%, respectively. For (B), the energy recovery is 6% and 8% when Y1= 40% and 50%, 10 respectively. Fig. 16 shows a comparison between the total energy consumption of the SWRO 11 plant with the presence of PRO and ERD. It can be seen that the maximum SESWRO-PRO 12 consumption point for each RO recovery is the point at which there is no PRO sub-system 13 contribution (DF = 0). In other words, this is the maximum SERO consumption for each RO 14 recovery. For configuration (A), the contribution of the PRO sub-system is very limited at a 15 low RO1 recovery ratio (Y1< 30%), where more than 96% of the energy recovery is 16 accomplished by ERD. This is due to the low amount of rejected water coming from the 17 second RO stage, which constitutes the feed solution for the PRO sub-system. This result 18 corresponds, theoretically, to less than 20% of dilution. In addition, the high concentration of 19 the draw solution may induce severe concentration polarization, which reduces the 20 performance of the membrane. Contrary to configuration (A), the contribution of PRO is 21 considerable at a low recovery ratio because the amount of PRO feed solution is provided 22 directly from the pre-treated seawater and remains nearly constant at this range of RO1 23 recovery ratio (Y1< 20%). The contribution of the PRO in the total energy recovered is around 24 45%. This high value is because of the limited contribution of ERD in this case due to the 25 high PRO feed solution concentration. Also, when the PRO feed flow ratio was maintained 26 equal to unity, the experimental dilution factor increased. For example, when Y1= 40%, the 27 energy recovered corresponds to DF= 40%, which is double the dilution in the first 28 configuration case. This result clearly shows the strong relationship between the feed flow 29 ratio and the dilution factor in the PRO process. To summarize, the notable difference 30 between the ideal case and the model case is related to the membrane performance. In fact, to 31 reach high values of dilution, an improvement in the water permeability across the membrane 32 is fundamental. Moreover, avoiding the reverse salt diffusion by reducing the salt 33 permeability of the active layer material enhances the dilution factor. Lastly, improving the 34 PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 23 inner structure of the PRO membrane support layer, to reduce the effect of the internal 1 concentration polarization and optimize the operating conditions in order to minimize the 2 external concentration polarization, is a challenge that can guarantee a better PRO dilution 3 and therefore more recovered energy. 4 5 Fig.10: SWRO-PRO specific energy without using ERD, as a function of the applied pressure 6 ΔP for different RO recoveries for the model case. (A) represents the result for the first 7 configuration and (B) for the second configuration. 8 9 10 Fig. 11: SWRO-PRO specific energy consumption as a function of the RO1 recovery for the 11 proposed configurations. Pump efficiency, membrane salt rejection, and ERD efficiency were 12 taken as 90%, 98%, and 95%, respectively. 13 14 15 4. Dilution factor in realistic conditions 16 The dilution factor reflects the amount of feed water that is mixed with the draw water to 17 produce energy. This parameter is strongly dependent on the membrane performance and the 18 2,2 2,4 2,6 2,8 3 3,2 010 20 30 40 50 60 SECSWRO-PRO (kWh/m3) ΔP (bar) Y1 = 40% Y1 = 50% A 2,2 2,4 2,6 2,8 3 3,2 010 20 30 40 SECSWRO-PRO (kWh/m3) ΔP (bar) Y1 = 40% Y1 = 50% B 0 1 2 3 4 5 6 7 8 9 020 40 60 80 100 Specific energy (kWh/m3) Y 1 (%) First configuration Second configuration SWRO without ERD PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 24 driving force, Δπ. In other words, increasing the water flux across the membrane leads to the 1 increase in the dilution. In the ideal case with a perfect membrane, the water flux depends 2 only on the osmotic pressure difference. Then, to achieve high dilutions, the concentrations of 3 feed and draw solutions should be optimized. In realistic conditions, the maximum dilution is 4 not reachable because of several limiting factors, such as the concentration polarization (CP) 5 and reverse salt diffusion (RSD), which reduce the performance of the membrane. The 6 realistic dilution factor is investigated in the current section. 7 The maximum amount of feed flow that crosses the membrane to be mixed with the draw 8 solution, ∆𝑄𝑄𝑚𝑚𝐷𝐷𝑚𝑚 , can be calculated using the following equation [32]: 9 10 ∆𝑄𝑄𝑚𝑚𝐷𝐷𝑚𝑚=�𝐶𝐶𝐷𝐷,𝑏𝑏−�𝐶𝐶𝐹𝐹,𝑏𝑏 �𝐶𝐶𝐷𝐷,𝑏𝑏+∅ 1−∅�𝐶𝐶𝐹𝐹,𝑏𝑏𝑄𝑄𝐷𝐷 (24) 11 12 Eq. (24) is applicable only for an ideal membrane with perfect hydrodynamics. The 13 rearrangement of Eq. (24) using Eqs.(5) and (6) gives the maximum dilution factor: 14 15 𝐷𝐷𝐷𝐷𝑚𝑚𝐷𝐷𝑚𝑚=𝑟𝑟�𝐶𝐶𝐷𝐷−�𝐶𝐶𝐹𝐹 �𝐶𝐶𝐷𝐷+1 𝐷𝐷�𝐶𝐶𝐹𝐹 (25) 16 17 where 𝑆𝑆𝐷𝐷 and 𝑆𝑆𝐷𝐷 are the draw and feed solution concentrations. To introduce the impact of 18 detrimental effects, Eq.(25) is modified by substituting CD and CF by Eqs.(14),(15) and (16) 19 according to the variation of the effective feed and draw concentrations described in Fig.S4 20 (Supplementary material). Table S2 summarizes the different equations of DF for each case 21 studied. The realistic dilution factor can be determined using the parameters described in 22 Table S1 and the modeling results of section 3.2. To investigate the effect of each detrimental 23 effect, DFmax is modeled under several operating conditions. In fact, the draw solution is 24 assimilated to the seawater RO brine concentration (1.2 M of NaCl) and the feed solution is 25 modified for each studied case from freshwater to river water concentration. The modeling 26 results are presented in Fig.18. The maximum dilution is obtained using an ideal membrane 27 with no reverse salt diffusion JS (RSD), concentration polarization, and an advanced state-of-28 the-art membrane with perfect hydrodynamics. In this case, the water flux Jw is directly 29 proportional to the osmotic pressure difference = 𝜋𝜋𝐷𝐷,𝑏𝑏− 𝜋𝜋𝐷𝐷,𝑏𝑏 . The maximum dilution 30 achievable in this case is 92%. When considering only the effect of the external concentration 31 polarization, which is caused by non ideal hydrodynamics in the draw solution flow channel, 32 PREPRINT Renewable Energy, Volume 105, May 2017, Pages 84–95 DOI: 10.1016/j.renene.2016.12.030 25 which induces the dilutive External Concentration Polarization (dilutive ECP) at the draw side 1 and the concentrative External Concentration Polarization (concentrative ECP) at the feed 2 side, the values of the dilution factor decrease by almost 8% from the ideal case. This 3 behavior is caused by the decrease in the effective osmotic pressure difference from 𝛥𝛥𝜋𝜋 to 4 𝛥𝛥𝜋𝜋𝐸𝐸𝐶𝐶𝑃𝑃=𝜋𝜋𝐷𝐷,𝑚𝑚− 𝜋𝜋𝐷𝐷,𝑚𝑚 (see Fig.S4). With the existence of RSD, the decrease in the DF is 5 around 24% from the ideal case. In fact, RSD causes the drop in the driving force due to the 6 penetration of salt from the draw to the feed side, which induces a decrease of Jw. To 7 emphasize the effect of the concentration polarization and internal concentration, the 8 concentration of river water (0.015M) is considered. Under these conditions, the decrease in 9 the DF reaches 52 %. As can be clearly seen, most of the DF reduction is due to the internal 10 concentration polarization with a contribution of 28%. ICP, caused by the accumulation of 11 salt at the active layer/support layer boundary, decreases the driving force from 𝛥𝛥𝜋𝜋 to 12 𝛥𝛥𝜋𝜋𝑚𝑚=𝜋𝜋𝐷𝐷,𝑚𝑚− 𝜋𝜋𝑖𝑖. Consequently, a severe drop occurs in Jw; therefore, DF decreases 13 drastically. As the energy is directly proportional to the dilution, the DF loss from the realistic 14 PRO operation may have a significant impact on the net specific energy recovered from the 15 process. The optimization of the membrane performance is a critical operation to guarantee 16 high energy recovery. It should be pointed out that this study does not consider the effect of 17 organic fouling. In fact, it was shown in a previous study that fouling severely reduces the 18 water flux and therefore the dilution. This energy loss caused by membrane limitations, in 19 conjunction with the pre-treatment energy requirements, pumping energy costs, inefficiencies 20 in the hydro-turbine and pressure exchanger, may limit the amount of energy harvested from 21 the PRO process. Of course, the increase in the concentration difference between the feed and 22 draw solution concentrations may provide a sizable amount of energy using the PRO process. 23 Nevertheless, the increase in the PRO entering solution concentration should be well chosen 24 and treated to avoid the accentuation of the concentration polarization and the reverse salt 25 diffusion, which lead to the decrease in the driving force; the osmotic pressure difference. 26 27