Membrane-Based Osmotic Heat Engine with ... - ACS Publications

Apr 3, 2015 - Engineering Surface Energy and Nanostructure of Microporous Films for Expanded Membrane Distillation Applications. Chanhee Boo , Jongho ...
0 downloads 0 Views 1MB Size
Download PDF
Article pubs.acs.org/est

Membrane-Based Osmotic Heat Engine with Organic Solvent for Enhanced Power Generation from Low-Grade Heat Evyatar Shaulsky,† Chanhee Boo,† Shihong Lin, and Menachem Elimelech* Department of Chemical and Environmental Engineering, Yale University, New Haven, Connecticut 06520-8286, United States S Supporting Information *

ABSTRACT: We present a hybrid osmotic heat engine (OHE) system that uses draw solutions with an organic solvent for enhanced thermal separation efficiency. The hybrid OHE system produces sustainable energy by combining pressure-retarded osmosis (PRO) as a power generation stage and membrane distillation (MD) utilizing low-grade heat as a separation stage. While previous OHE systems employed aqueous electrolyte draw solutions, using methanol as a solvent is advantageous because methanol is highly volatile and has a lower heat capacity and enthalpy of vaporization than water. Hence, the thermal separation efficiency of a draw solution with methanol would be higher than that of an aqueous draw solution. In this study, we evaluated the performance of LiCl−methanol as a potential draw solution for a PRO−MD hybrid OHE system. The membrane transport properties as well as performance with LiCl−methanol draw solution were evaluated using thin-film composite (TFC) PRO membranes and compared to the results obtained with a LiCl−water draw solution. Experimental PRO methanol flux and maximum projected power density of 47.1 L m−2 h−1 and 72.1 W m−2, respectively, were achieved with a 3 M LiCl−methanol draw solution. The overall efficiency of the hybrid OHE system was modeled by coupling the mass and energy flows between the thermal separation (MD) and power generation (PRO) stages under conditions with and without heat recovery. The modeling results demonstrate higher OHE energy efficiency with the LiCl−methanol draw solution compared to that with the LiCl−water draw solution under practical operating conditions (i.e., heat recovery 50 °C) are more efficient in terms of thermal separation than inorganic salts.18 However, loss of draw solute by reverse permeation of draw solution species, as well as process complexity in the draw solute recovery stage, pose operational challenges.19,20 Organic solvents, such as methanol, are highly volatile, having lower enthalpy of vaporization than water.21,22 Hence, thermal separation efficiency of a draw solution with methanol as a solvent is expected to be higher than that for aqueous draw solutions. In this study, we present a PRO−MD hybrid OHE system that utilizes a draw solution of an inorganic salt dissolved in methanol. The performance of a lithium chloride−methanol (LiCl−methanol) draw solution was evaluated using a commercial TFC membrane in PRO operation and compared to that with an aqueous draw solution (i.e., LiCl−water). In addition, solvent flux and corresponding projected maximum power densities achieved by LiCl−methanol with varying solution concentrations are presented. The overall energy efficiency of the hybrid OHE system was modeled by coupling the mass and energy flows between the thermal separation (MD) and power generation (PRO) stages. The implications of these results for the use LiCl−methanol for power generation by OHE are evaluated and discussed.

sheet PRO membrane coupon with an effective membrane area of 20.0 cm2 was loaded in a custom built acryl cell in PRO configuration (i.e., active layer facing draw solution). All experiments were conducted for 1 h with a crossflow velocity of 21.4 cm/s. Temperatures of both draw and feed solutions were maintained at 25.0 ± 0.5 °C. Solvent (water or methanol) flux, Jsolvent, across the membrane was measured by recording the increase in draw solution weight at 1 min intervals. Data were collected after the solvent flux had stabilized (always within 10 min) and averaged over a 50 min period. The increase in draw solution weight was converted to volume by dividing the weight data by the solvent density at 25.0 °C (i.e., 0.997 g/cm3 for water and 0.791 g/cm3 for methanol). Reverse salt flux, Js, was determined by measuring the total concentration of draw solute in the feed solution after the 1 h experiment, following the procedure described in our previous work.29 Because the initial solute concentration in the feed is zero (pure methanol or water feed), a mass balance on the solute during a PRO run yields C F(VF0 − Jsolvent A m t ) = Jsolute A m t

(1)

where CF is the solute concentration in the feed after 1 h experiments, VF0 is the initial volume of the feed solution, Jsolvent is the average solvent flux, Am is the membrane area, and t is time. LiCl concentration in the feed solution was measured using a conductivity meter (Oakton Instrument, Vernon Hills, IL). The conductivities of LiCl in both water and methanol as a function of solution concentration exhibited a good linear relationship (R2 > 0.99). Modeling Methanol Flux and Projected Power Density. Methanol fluxes produced by LiCl−methanol draw solution at different concentrations in PRO mode were modeled using the following equation, which accounts for the effects of internal concentration polarization (ICP), external concentration polarization (ECP), and reverse draw solute permeation30

■

MATERIALS AND METHODS Preparation of Lithium Chloride−Methanol Draw Solution. ACS grade lithium chloride (LiCl, Fisher Scientific, Pittsburgh, PA) and pure methanol (J. T. Baker, purity of 99.8%) were used to prepare the LiCl−methanol draw solution. LiCl is relatively soluble in methanol and its thermodynamic properties in methanol has been well-characterized in previous studies.23−25 Dissolution of LiCl in methanol is exothermic. Hence, the temperature of the LiCl−methanol solution increases during its preparation, resulting in a change in the mixture volume. To achieve the intended molar LiCl concentration in methanol, we confirmed the final volume (i.e., 1.0 L) of the mixture at 25.0 °C. PRO Membrane. A commercial thin-film composite (TFC) membrane, obtained from Hydration Technology Innovation (HTI, Albany, OR), was used for the PRO experiments. This proprietary membrane designed for forward osmosis (FO) consists of a thin, dense polyamide selective layer cast on a polysulfone support embedded in a woven polyester mesh.17 Prior to experiments, membranes were immersed in 25% isopropanol for 30 min to completely wet the pores of the membrane support layer and then rinsed thoroughly with DI water. Membranes used for experiments with LiCl−methanol draw solutions were pre-wetted in methanol for 1 h before use. Membrane Characterization. Properties of the TFC-PRO membrane were characterized using the FO characterization method developed by Tiraferri et al.26 using two different draw solutions (i.e., LiCl−methanol and LiCl−water). The solvent and reverse salt fluxes were measured at three different draw concentrations (0.25, 0.5, and 1.0 M) in FO configuration (i.e., active layer faces the feed solution), and data were fitted to the FO transport equations. Thereafter, least-squares nonlinear regression was performed to obtain the membrane properties. More details on the membrane characterization protocol are described in the Supporting Information. Measurement of PRO Solvent Flux and Reverse Salt Flux. A laboratory scale cross-flow unit was used to measure the solvent flux and reverse salt flux in PRO mode.27,28 A flat-

PRO JMethanol

⎫ ⎧ ⎛ J PRO ⎞ ⎛ J PRO S ⎞ ⎪ πD,b exp⎜ − Methanol ⎟ − πF,b exp⎜ Methanol ⎟ ⎪ k ⎪ ⎪ ⎝ ⎠ ⎝ D ⎠ = A⎨ − ΔP ⎬ PRO PRO ⎤ ⎡ ⎛ ⎞ ⎛ ⎞ J S J ⎪ ⎪ B ⎟ − exp⎜ − Methanol ⎟⎥ ⎢exp⎜ Methanol PRO ⎪ 1 + JMethanol ⎪ D k ⎝ ⎠ ⎝ ⎠ ⎦ ⎣ ⎩ ⎭ (2)

where πD,b and πF,b (πF,b is equal to zero for pure methanol feed) are the draw and feed solution bulk osmotic pressures, respectively, k is the solute mass transfer coefficient, D is the diffusion coefficient of the draw solutes, and ΔP is the applied hydraulic pressure. The pure methanol permeability coefficient, A, solute permeability coefficient, B, and structural parameter of the support layer, S, are intrinsic membrane properties and determined from the FO characterization method described above. The projected membrane power density, W, defined as the power generated per unit membrane area as a function of applied hydraulic pressure, was calculated using PRO W = JMethanol ΔP

(3)

which assumes 100% efficiency of the pressure exchanger and turbine in the OHE system. B

DOI: 10.1021/es506347j Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

■

RESULTS AND DISCUSSION Properties of LiCl−Methanol Draw Solution. Methanol is a highly volatile solvent, having a lower heat capacity, enthalpy of vaporization, and boiling point than water (Table 1).31,32 Hence, a draw solution with methanol solvent is

The variation in the osmotic pressure with the concentration of the LiCl−methanol draw solution calculated using eq 6 is presented in Figure 1. We also present the osmotic pressure of

Table 1. Relevant Thermodynamic Properties of Water and Methanol solvent

heat capacity at 25 °C, cp,solva (kJ kg−1 K−1)

enthalpy of vaporization at boiling point, hvapa (kJ kg−1)

boiling pointb (°C)

methanol water

2.53 4.18

1104 2260

64.7 99.9

a

From the CRC handbook of chemistry and physics.21 bFrom Perry’s chemical engineers’ handbook, 7th ed.22 Figure 1. Osmotic pressures generated by LiCl−methanol and LiCl− water draw solutions as a function of solution concentration at 25.0 °C.

expected to have a better thermal separation efficiency than one with a water solvent. Lithium chloride (LiCl) is relatively soluble in polar protic solvents, such as water and methanol. The maximum solubilities of LiCl in methanol and water at 25 °C are 10.0 and 19.7 mol/kg, respectively.33,34 The high solubility of LiCl in polar protic solvents is attributed to the strong ion−dipole attractive force between Li+ ions and water or methanol molecules, which have a high dipole moment.35 Because the Li+ ion has the smallest ionic radius in the group of alkali metals, it possesses relatively strong electronegativity and thus has a high affinity for the polar solvent molecules. The osmotic pressure of the LiCl−methanol draw solution is determined using an isopiestic method based on experimentally measured vapor pressure. For electrolyte solutions, the solvent activity coefficient, as, is correlated with its vapor pressure according to36,37 ⎛ p⎞ (B − V s*)(p − p*) ln as = ln⎜ ⎟ + s RT ⎝ p* ⎠

the LiCl−water solution as determined by an OLI Stream Analyzer (OLI System, Inc. Morris Plains, NJ). The results indicate that the LiCl−methanol solution generates high osmotic pressures comparable to those generated by LiCl− water over the concentration range from 0.25 to 3.0 M. More details on the determination of the osmotic coefficients for the LiCl−methanol draw solution are given in the Supporting Information. Membrane Transport Properties with Methanol as Solvent. Transport properties of the TFC-PRO membrane, solvent and solute permeability coefficients (A and B, respectively), and structural parameters (S) were determined for LiCl−methanol and compared to those determined for LiCl−water. As presented in Figure 2 (blue bar with diagonal

(4)

where p and p* are the vapor pressures of the liquid phase at electrolyte concentrations m and 0 (pure solvent), respectively, Bs is the second virial coefficient of methanol vapor, Vs* is the molar volume of pure liquid methanol, R is the gas constant, and T is the absolute temperature. The osmotic coefficient, ϕ, for a solution of 1:1 electrolyte is given by36,37 ϕ=−

ln as 2mMs

(5)

where m is the solution concentration in molality, and Ms is the molar mass of the solvent. Previous studies have determined the osmotic coefficient of LiCl−methanol solutions at specific solution concentrations and temperatures based on experimentally measured vapor pressures.25,38 To obtain the osmotic coefficients at varying solution concentrations and temperatures relevant to our study, we used the correlation model developed by Pitzer and Mayorga.37,39 Once we determine the osmotic coefficient, ϕ, the osmotic pressures of the LiCl−methanol solution at different concentrations are calculated using the modified van’t Hoff equation: π = 2ϕMRT

Figure 2. Membrane transport parameters determined from the threestep FO membrane characterization method using LiCl−water and LiCl−methanol draw solutions. Solvent flux and reverse salt flux were measured at three different draw concentrations (i.e., 0.25, 0.5, and 1.0 M), and an Excel-based calculation was carried out to determine pure solvent permeability, A, solute permeability coefficient, B, and structural parameter, S. The experiments were conducted in FO mode (i.e., draw solution facing support layer) using pure methanol or DI water as a feed solution. Crossflow velocity was set at 21.4 cm/s in both membrane channels, and temperatures of feed and draw solutions were maintained at 25.0 ± 0.5 °C. More details on the three-step FO membrane characterization method are given in the Supporting Information.

(6)

where M is the molar concentration of the LiCl−methanol solution. C

DOI: 10.1021/es506347j Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology Table 2. Relevant Thermodynamic Properties of Lithium Chloride (LiCl) in Methanol and Water solvation numberb

a

−1

stokes radiusc (Å)

solution

maximum solubility at 25 °C (mol kg )

Li

Cl

Li+

Cl−

mutual diffusion coefficientd (× 10−9 m2 s−1)

LiCl−methanol LiCl−water

10.0 19.7

7 6−7

4 6

3.78 2.30

2.91 1.21

0.725 1.30

a

−

+

From Turner et al.34 bFrom Ulich44 and Impey et al.45 cFrom Hartely and Raikes46 dFrom Kumar et al.54

pattern), the solvent permeability of methanol is ∼55% of that of water. Solvent transport through a dense, salt-rejecting membrane like FO and PRO is governed by the solutiondiffusion mechanism, where solvent solubility and diffusivity in the membrane selective (active) layer determine the membrane permeability.40 Solubility is a thermodynamic parameter indicating the amount of solvent sorbed to the membrane, whereas diffusivity is a kinetic parameter indicating how fast the solvent passes through the membrane.41 Because of the higher polarity of water molecules compared to methanol, sorption or solubility of water molecules in the polyamide layer is expected to be higher than methanol.42,43 In addition, water molecules diffuse through the membrane selective layer more readily than methanol because of their smaller molecular size (18 g/mol for water and 32 g/mol for methanol). Thus, the observed lower solvent permeability of methanol compared to water is attributed to its lower solubility as well as lower diffusivity in the membrane selective (active) layer. LiCl membrane permeability data (Figure 2, red bar with horizontal pattern) show that LiCl in methanol is less permeable through the PRO membrane compared to that in water. Table 2 presents solvation numbers for Li+ and Cl− ions and their corresponding Stokes radii in methanol and water as determined by molecular dynamic calculations.44,45 Although the solvation numbers for Li+ are comparable in water and methanol, and those for Cl− in methanol are smaller than in water, both ions have larger Stokes radii in methanol.46 The larger Stokes (or solvated) radii for Li+ and Cl− in methanol compared to those in water are attributed to the larger solvent molecules attached to the ions. Thus, the lower solute permeability of LiCl in methanol compared to that in water can be explained by the fact that Li+ and Cl− ions, with larger solvated ion sizes, are better rejected by the membrane. The structural parameter, S, is an important membrane property that determines the extent of the performance-limiting phenomenon of internal concentration polarization (ICP). One way to reduce the effect of ICP is to improve the wetting of the membrane support layer.47,48 Better support layer wetting increases the effective support layer porosity, thereby reducing the severity of ICP. The results from PRO membrane characterization show that the structural parameter, S, determined with LiCl−methanol is lower than that measured with LiCl−water (Figure 2, green bar with checked pattern). Because methanol has surface tension that is much lower than water (i.e., 22.07 and 71.99 mN/m for methanol and water, respectively, at 25.0 °C), the polysulfone support layer of the PRO membrane is better wetted by methanol, resulting in lower S. Our results are also in line with previous studies showing that membrane soaking in liquids of low surface tension (e.g., isopropyl alcohol) enhances wetting of the membrane support layer.49 Membrane Performance with LiCl−Methanol Draw Solution. Solvent flux and reverse salt flux with LiCl− methanol and LiCl−water were evaluated in PRO configuration

(i.e., porous support layer facing the feed solution and active layer facing the draw solution) without applied hydraulic pressure and compared at the same draw solution concentration (i.e., 1.0 M) in Figure 3. Water flux and reverse draw salt

Figure 3. Solvent flux, reverse salt flux, and reverse flux selectivity for LiCl−water and LiCl−methanol draw solutions obtained from PRO experiments. The PRO experiments were conducted for 1 h using pure methanol or DI water as a feed solution and 1 M LiCl−water or 1 M LiCl−methanol as a draw solution, respectively. Crossflow velocity was set at 21.4 cm/s, and temperature was maintained at 25.0 ± 0.5 °C for both feed and draw solutions. Reverse flux selectivity is determined by dividing the solvent flux with reverse salt flux as measured from the PRO experiments.

permeation with 1.0 M LiCl−water were 37.78 L m−2 h−1 and 0.54 mol m−2 h−1, respectively. The results were comparable to previously reported values using NaCl−water (most common aqueous draw solution) and the same type of TFC membrane.19 The measured water flux and reverse draw salt flux with LiCl−water are reasonable given the comparable properties of LiCl and NaCl aqueous draw solutions (i.e., osmotic pressure and salt diffusivity). Solvent (methanol) flux with 1.0 M LiCl−methanol draw solution was 26.25 L m−2 h−1, which is lower than that obtained with the 1.0 M LiCl−water draw solution. As presented in eq 2, solvent flux in PRO is dependent on the intrinsic membrane transport properties, including solvent and salt permeabilities of the membrane active layer and the structural parameter of the membrane support layer. Solvent flux increases with higher solvent permeability, lower salt permeability, and lower structural parameter. As discussed earlier and also shown in Figure 2 (red bar with horizontal pattern), LiCl membrane permeability in methanol is lower compared to that in water because of the larger solvated ion sizes of Li+ and Cl− in methanol. In addition, the structural parameter of the PRO membrane with LiCl−methanol is lower compared to that with LiCl−water (Figure 2). Consequently, the lower solvent flux with LiCl−methanol compared to that with LiCl−water is attributed to the significantly lower solvent permeability of methanol (Figure 2, blue bar with diagonal pattern), which D

DOI: 10.1021/es506347j Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology overwhelms the benefits of better wetting of the membrane support layer with methanol and reduced reverse salt permeation. Reverse salt flux selectivity (RSFS), defined as the ratio of forward solvent flux to reverse salt flux, is a key parameter determining the efficiency of a draw solution.29,50 RSFS values for LiCl−methanol and LiCl−water draw solutions were calculated using the measured solvent and reverse salt fluxes at a draw solution concentration of 1.0 M (Figure 3). Despite the solvent flux with LiCl−methanol being lower than that with LiCl−water, LiCl−methanol had a higher RSFS due to significantly reduced reverse draw salt permeation. The results suggest that LiCl−methanol can be a viable draw solution for use in the OHE to obtain high performance without significant loss of draw salt. Solvent Flux and Projected Power Density with High Draw Solution Concentrations. Methanol fluxes with high LiCl−methanol concentrations (2 and 3 M) obtained from PRO experiments are presented in Figure 4. We also modeled

Figure 5. Plots of modeled methanol flux, JPRO Methanol, and power density, W, as a function of applied hydraulic pressure, ΔP, for 1, 2, and 3 M LiCl−methanol draw solutions. Open symbols represent measured experimental PRO methanol fluxes at zero hydraulic pressure (data presented in Figure 4). The methanol fluxes were determined using eq 2 as a function of applied hydraulic pressure, and the corresponding power densities were calculated using eq 3.

pressure using eq 2 and then determined the projected power densities using eq 3. The projected power densities with LiCl− methanol increase with increasing ΔP until a maximum value of Wpeak is attained when the applied hydraulic pressure is approximately half the osmotic pressure difference between feed and draw solutions. As shown in Figure 5, maximum power densities of 8.37, 31.34, and 72.14 W m−2 were predicted for the 1, 2, and 3 M LiCl−methanol draw solutions, respectively. The corresponding applied hydraulic pressures were 23, 59, and 114 bar, respectively. We note that the predicted peak power densities are comparable to recently reported experimental values using aqueous NaCl draw solutions and a similar PRO membrane.17 The LiCl−methanol produces lower solvent flux compared to aqueous NaCl draw solutions. However, the possibility of PRO operation at relatively high applied hydraulic pressure due to the higher osmotic pressure of LiCl−methanol and the lower reverse LiCl permeation in methanol than that in NaCl−water enable power densities comparable to those obtained with aqueous NaCl draw solutions. Modeling Osmotic Heat Engine Efficiency. A schematic diagram of the evaluated PRO-MD hybrid OHE system is shown in Figure S1 of the Supporting Information. The energy efficiency of an OHE is obtained by solving the mass and heat transfer equations describing the operation of the PRO and MD stages as shown in our recent publication.16 The overall energy efficiency of a hybrid PRO-MD system is defined as16 P ηOHE = qH (7)

Figure 4. Experimental and modeled methanol fluxes as a function of LiCl−methanol draw solution concentration. Open squares indicate methanol fluxes obtained from PRO experiments, and the dashed line indicates modeled methanol fluxes. The PRO experiments were conducted for 1 h using pure methanol feed solution and LiCl− methanol draw solutions with three different concentrations (i.e., 1, 2, and 3 M). Crossflow velocity was set at 21.4 cm/s, and temperature was maintained at 25.0 ± 0.5 °C for both feed and draw solutions. Modeled methanol flux was calculated from eq 2 using A = 1.15 L m−2 h−1 bar −1, B = 0.33 L m−2 h−1, S = 252 μm, D = 0.725 × 10−9 m2 s−1, and k = 11.5 × 10−6 m s−1. Osmotic pressure of LiCl−methanol was determined using an isopiestic method and Pitzer and Mayorga correlation model (Table S2, Supporting Information).

the methanol flux using eq 2 (under zero applied hydraulic pressure, ΔP = 0) with the membrane transport properties (Figure 2) and osmotic pressures of LiCl−methanol (Figure 1). Measured PRO methanol fluxes were 37.83 and 47.13 L m−2 h−1 for the 2 and 3 M LiCl−methanol draw solutions, respectively. The model results (dashed line) are in good agreement with the experimental data. We used the same diffusion coefficient for LiCl (i.e., 0.725 × 10−9 m2/s, Table 2) in modeling the methanol flux for all draw solution concentrations due to limited published data on the dependence of LiCl diffusivity on concentration. The slightly overpredicted methanol fluxes at high LiCl concentrations are likely due to the variation of LiCl diffusivity in methanol at higher draw solution concentrations. Achievable power densities with 1, 2, and 3 M LiCl− methanol draw solutions are shown in Figure 5. We first calculated the methanol fluxes as a function of applied hydraulic

where P is the power generated by the turbine (TB) in the PRO system, and qH is the heat required to regenerate the distillate stream (S4) from the mixed PRO effluent stream (S1) in the MD system (Figure S1 of the Supporting Information). The applicability of this equation is independent of the extent of latent heat recovery by the heat exchanger (HX) in the MD system. To achieve maximum energy output in the PRO stage, we match the applied hydraulic pressure, ΔP, with the osmotic pressure of the draw solution exit stream from the PRO stage, πs1. In this case, the theoretical maximum extractable power (Pmax) becomes E

DOI: 10.1021/es506347j Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology V V = πS1ΔQ PRO = πS1 Pmax = ΔP ΔQ PRO

and therefore, θ is close to 1 (or 100% heat recovery); in this case, the OHE exhibits maximum energy conversion efficiency. More details on the determination of maximum feed recovery, γmax, T*H, and heat capacities of LiCl−methanol and LiCl−water solutions, cSp1, are given in the Supporting Information. Using eqs 11 and 12, we can express the OHE energy conversion efficiency as a function of the extent of latent heat recovery, θ (Figure 6). The results clearly demonstrate that the

γQ S

1

ρsolv

(8)

where is the volumetric trans-membrane flow rate in PRO, γ is the feed recovery rate in the MD system (γ = QS4/QS1, with Q being the mass flow rate), and ρsolv is the density of the pure solvent. We note that ΔP lower than πS1 may be used in realistic operations for higher PRO solvent flux and, correspondingly, higher membrane power density. However, fixing ΔP at πS1 in our analysis allows us to evaluate the effect of latent heat recovery in the MD stage on the overall energy conversion efficiency of the OHE by assuming the energy efficiency of the PRO stage is fully maximized. The energy efficiency of the thermal separation stage (i.e., MD) with latent heat recovery is quantified by the specific heat duty, β, which indicates the amount of heat required to generate a unit mass of distillate is16 QVPRO

β=

qH ΔQ MD

=

qH QS

4

=

cpS1(TH − TS2) γ

(9)

where ΔQMD is the trans-membrane vapor flow rate in MD, S which is equal to the distillate flow rate, QS4, cp1 is the heat capacity of the stream exiting the PRO stage, TH is the temperature of the heat source, and TS2 is the temperature of the stream exiting the heat exchanger (HX). By combining eqs 7, 8, and 9, the overall energy conversion efficiency of the OHE can be obtained as πS1 γ ηOHE = S1 ρsolv cp (TH − TS2) (10)

Figure 6. Energy efficiencies of the PRO-MD hybrid OHE system with LiCl−methanol and LiCl−water draw solutions as a function of the extent of heat recovery at draw solution concentrations of (a) 1.0, (b) 2.0, and (c) 3.0 M. The temperature of the heat source, TH, and heat sink, TC, are assumed to be 318.15 and 298.15 K, respectively. The Carnot efficiency represents the highest energy conversion efficiency achieved by a heat engine (ηCarnot = 1 − TC/TH).

OHE energy efficiency is higher with the LiCl−methanol draw solution than LiCl−water for any practical latent heat recovery, θ, by the HX (Figure 6). Because methanol has lower solvent density than water, the maximum extractable power, Pmax, in the PRO stage becomes higher for the LiCl−methanol than the LiCl−water draw solution. Also, heat capacity of the draw solution, cSp1, is significantly lower for LiCl−methanol than for LiCl−water (Table S3, Supporting Information), resulting in a lower specific heat duty of LiCl−methanol than LiCl−water for most values of θ. Only when a very high degree of latent heat recovery is possible (e.g., θ > 90%, which is not practical in real systems) does LiCl−water becomes a more favorable working solution than LiCl−methanol, as the temperature difference between TH and TH* starts to play a more important role in the overall OHE efficiency. As discussed above, maximum energy output from the PRO system can be achieved when the applied hydraulic pressure is equal to the osmotic pressure of the draw solution exit stream from the PRO system. Hence, the maximum hydraulic pressure applied in PRO becomes a performance-limiting factor that determines the overall OHE energy efficiency. To date, the highest hydraulic pressure applied in an experimental PRO system was ∼50 bar,17 which corresponds to LiCl−methanol and LiCl−water concentrations between 1 to 2 M. Within this concentration range, the degree of heat recovery allowing higher OHE efficiency with LiCl−water as the working solution, rather than LiCl−methanol, is impractically high (θ > 90% as shown in Figure 6a and b). Consequently, from a pure energy-efficiency perspective, an OHE based on LiCl− methanol is more energy-efficient than that based on LiCl− water in any practical scenario.

If we assume a maximum feed recovery rate, γmax, in the MD system, the overall OHE efficiency can be expressed as πS1 γ ηOHE = maxS ρsolv cp1 (TH − TS2) (11) We note that maximum feed recovery rate, γmax, occurs when the vapor pressure difference between feed and permeate flows is completely exploited within a module in the MD operation with a high permeate (distillate) flow rate relative to the feed flow rate. Because the overall OHE efficiency described by eq 11 is dimensionless and independent of the scale of the system (i.e., the flow rate, QS1, does not affect ηOHE), the extent of latent heat recovery is only dependent on the heat transfer kinetics in the heat exchanger, which results in different temperatures of the stream entering the heat source, TS2. Therefore, a parameter θ is defined to quantify the extent of heat recovery. θ=1−

TH* − TS2 TH* − TC

(12)

where T*H is the thermodynamic maximum temperature of the stream exiting the HX (S2) in the MD operation with a heat source (“H” in Figure S1, Supporting Information, temperature of TH), and TC is the temperature of the heat sink (“C” in Figure S1, Supporting Information). The defined θ indicates that when heat recovery is absent, TS2 = TC and θ = 0; in this case, the OHE efficiency attains its minimum value. On the contrary, when heat recovery is maximized, TS2 approaches T*H, F

DOI: 10.1021/es506347j Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Environmental Science & Technology

■

Implications and Challenges. Our study demonstrates that an osmotic heat engine (OHE) using draw solutions with an organic solvent yields higher energy conversion efficiency compared to one using an aqueous draw solution. Although the OHE energy efficiency with LiCl−methanol draw solution, evaluated based on our experimental and modeling results, is promising, further research efforts are needed to advance the technology beyond conceptualization. Several studies have indicated that PRO performance is compromised due to membrane deformation under pressurized PRO operation.17,51−53 Therefore, a mechanically robust PRO membrane that can withstand high hydraulic pressures is key for successful OHE application. Such robust PRO membranes will allow the system to be the most beneficial by using high draw solution concentrations. In addition, an MD membrane that can resist wetting by the low surface tension organic solvent (methanol) is essential for realization of the proposed OHE. Although methanol is advantageous over water in terms of thermal separation efficiency, it is more expensive than water and more difficult to incorporate into an OHE system due to its high volatility (i.e., low boiling point and high vapor pressure). On the other hand, using methanol for OHE could be advantageous in cold regions with very lower temperatures (e.g., North and South Poles and in space), because lower freezing point of methanol (−97.6 °C) makes the OHE operation possible without risk of the solvent freezing. Lastly, a detailed technoeconomic analysis that takes into account system scale-up issues will inform the economic viability of the OHE as a device for converting waste heat to sustainable energy.

■

ASSOCIATED CONTENT

A schematic diagram of the PRO-MD hybrid OHE system (Figure S1); details of the membrane characterization procedure and a complete data set (Table S1); calculation of osmotic coefficient and variables used (Table S2); calculation of heat capacity (includes Table S3) and partial vapor pressure of working solution; discussion on thermodynamically limited MD operation regimes to explain maximum feed recovery, γmax, and thermodynamic maximum and minimum temperature of the stream exiting the heat exchanger, TH* and TC*, respectively (includes Tables S4 and S5). This material is available free of charge via the Internet at http://pubs.acs.org.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel. +1 (203) 4322789. Author Contributions †

E.S. and C.B. contributed equally to this work.

Notes

The authors declare no competing financial interest.

■

REFERENCES

(1) McGinnis, R. L.; Elimelech, M. Global challenges in energy and water supply: the promise of engineered osmosis. Environ. Sci. Technol. 2008, 42 (23), 8625−8629. (2) Post, J. W.; Veerman, J.; Hamelers, H. V. M; Euverink, G. J. W.; Metz, S. J.; Nymeijer, K.; Buisman, C. J. N. Salinity-gradient power: Evaluation of pressure-retarded osmosis and reverse electrodialysis. J. Membr. Sci. 2007, 288 (1−2), 218−230. (3) Solomon, S.; Plattner, G. K.; Knutti, R.; Friedlingstein, P. Irreversible climate change due to carbon dioxide emissions. Proc. Natl. Acad. Sci. U.S.A. 2009, 106 (6), 1704−1709. (4) Cox, P. M.; Betts, R. A.; Jones, C. D.; Spall, S. A.; Totterdell, I. J. Acceleration of global warming due to carbon-cycle feedbacks in a coupled climate model. Nature 2000, 408 (6813), 184−187. (5) CO2 emission from fuel combustion, Highlights, 2013 Edition. International Energy Agency 2013. (6) Chu, S.; Majumdar, A. Opportunities and challenges for a sustainable energy future. Nature 2012, 488 (7411), 294−303. (7) Nam, J. Y.; Cusick, R. D.; Kim, Y.; Logan, B. E. Hydrogen generation in microbial reverse-electrodialysis electrolysis cells using a heat-regenerated salt solution. Environ. Sci. Technol. 2012, 46 (9), 5240−5246. (8) Lund, J. W.; Freeston, D. H.; Boyd, T. L. Direct utilization of geothermal energy 2010 worldwide review. Geothermics 2011, 40 (3), 159−180. (9) Waste Heat to Power Systems; U.S. Environmental Protection Agency: Washington, DC, 2012; Combined Heat and Power Partnership. (10) Holman, J., Perspective: Waste heat to power − Still waiting for a breakthrough. IDC Energy Insights: Renewable Energy Strategies: Perspective: Farmingham, MA2011, No. EI229456. (11) Lund, J. W.; Bjelm, L.; Bloomquist, G.; Mortensen, A. K. Characteristics, development and utilization of geothermal resources − a Nordic perspective. Episodes 2008, 31 (1), 140−147. (12) Chen, H. J.; Goswami, D. Y.; Stefanakos, E. K. A review of thermodynamic cycles and working fluids for the conversion of lowgrade heat. Renewable Sustainable Energy Rev. 2010, 14 (9), 3059− 3067. (13) McGinnis, R. L.; McCutcheon, J. R.; Elimelech, M. A novel ammonia-carbon dioxide osmotic heat engine for power generation. J. Membr. Sci. 2007, 305 (1−2), 13−19. (14) Logan, B. E.; Elimelech, M. Membrane-based processes for sustainable power generation using water. Nature 2012, 488 (7411), 313−319. (15) Achilli, A.; Childress, A. E. Pressure retarded osmosis: From the vision of Sidney Loeb to the first prototype installation - Review. Desalination 2010, 261 (3), 205−211. (16) Lin, S. H.; Yip, N. Y.; Cath, T. Y.; Osuji, C. O.; Elimelech, M. Hybrid pressure retarded osmosis-membrane distillation system for power generation from low-grade heat: thermodynamic analysis and energy efficiency. Environ. Sci. Technol. 2014, 48 (9), 5306−5313. (17) Straub, A. P.; Yip, N. Y.; Elimelech, M. Raising the bar: Increased hydraulic pressure allows unprecedented high power densities in pressure-retarded osmosis. Environ. Sci. Technol. Lett. 2013, 1 (1), 55−59. (18) McGinnis, R. L.; Hancock, N. T.; Nowosielski-Slepowron, M. S.; McGurgan, G. D. Pilot demonstration of the NH3/CO2 forward osmosis desalination process on high salinity brines. Desalination 2013, 312, 67−74. (19) Boo, C.; Khalil, Y. F.; Elimelech, M. Performance evaluation of trimethylamine−carbon dioxide thermolytic draw solution for engineered osmosis. J. Membr. Sci. 2015, 473, 302−309. (20) Lu, X.; Boo, C.; Ma, J.; Elimelech, M. Bidirectional diffusion of ammonium and sodium cations in forward osmosis: Role of membrane active layer surface chemistry and charge. Environ. Sci. Technol. 2014, 48 (24), 14369−14376. (21) Macdonald, F.; Lide, D. R. CRC handbook of chemistry and physics: From paper to web. Abstr. Pap., Am. Chem. Soc. 2003, 225, U552−U552.

S Supporting Information *

■

Article

ACKNOWLEDGMENTS

We acknowledge support received from the Advanced Research Projects Agency-Energy (ARPA-E), U.S. Department of Energy, via Grant DE-AR0000306 and from the National Science Foundation under Award Number CBET 1232619. G

DOI: 10.1021/es506347j Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology (22) Perry, R. H.; Green, D. W.; Maloney, J. O. Perry’s chemical engineers’ handbook, 7th ed.; McGraw-Hill: New York, 1997. (23) Tomasula, P.; Czerwienski, G. J.; Tassios, D. Vapor-pressures and osmotic coefficients − electrolyte-solutions of methanol. Fluid Phase Equilib. 1987, 38 (1−2), 129−153. (24) Oh, S. K. Total pressure measurements for lithium chloride plus methanol plus ethanol at 303.15 K. J. Chem. Eng. Data 1997, 42 (6), 1082−1084. (25) Bonner, O. D. The Colligative Properties of Certain Electrolytes and Nonelectrolytes in Methanol. J. Solution Chem. 1987, 16 (4), 307− 314. (26) Tiraferri, A.; Yip, N. Y.; Straub, A. P.; Castrillon, S. R. V.; Elimelech, M. A method for the simultaneous determination of transport and structural parameters of forward osmosis membranes. J. Membr. Sci. 2013, 444, 523−538. (27) Mi, B.; Elimelech, M. Chemical and physical aspects of organic fouling of forward osmosis membranes. J. Membr. Sci. 2008, 320 (1− 2), 292−302. (28) Yip, N. Y.; Elimelech, M. Influence of natural organic matter fouling and osmotic backwash on pressure retarded osmosis energy production from natural salinity gradients. Environ. Sci. Technol. 2013, 47 (21), 12607−12616. (29) Phillip, W. A.; Yong, J. S.; Elimelech, M. Reverse draw solute permeation in forward osmosis: Modeling and experiments. Environ. Sci. Technol. 2010, 44 (13), 5170−5176. (30) Yip, N. Y.; Tiraferri, A.; Phillip, W. A.; Schiffrnan, J. D.; Hoover, L. A.; Kim, Y. C.; Elimelech, M. Thin-film composite pressure retarded osmosis membranes for sustainable power generation from salinity gradients. Environ. Sci. Technol. 2011, 45 (10), 4360−4369. (31) Koo, J.; Kleinstreuer, C. Viscous dissipation effects in microtubes and microchannels. Int. J. Heat Mass Transfer 2004, 47 (14−16), 3159−3169. (32) Patil, K. R.; Tripathi, A. D.; Pathak, G.; Katti, S. S. Thermodynamic properties of aqueous-electrolyte solutions 0.1. vapor-pressure of aqueous-solutions of Licl, Libr, and Lii. J. Chem. Eng. Data 1990, 35 (2), 166−168. (33) Janz, G. J.; Tomkins, R. P. T. Nonaqueous electrolytes handbook; Academic Press: New York, 1972. (34) Turner, W. E. S.; Bissett, C. C. The solubilities of alkali haloids in methyl, ethyl, propyl, and isoamyl alcohols. J. Chem. Soc. 1913, 103, 1904−1910. (35) Atkins, P. W.; Shriver, D. F. Inorganic chemistry, 4th ed.; W.H. Freeman: New York, 2006; Vol. xxi, p 822. (36) Barthel, J.; Neueder, R.; Kunz, W. Osmotic coefficients of nonaqueous electrolyte-solutions at thermodynamic and Mcmillan− Mayer level. Pure Appl. Chem. 1993, 65 (5), 889−894. (37) Safarov, J. T. Study of thermodynamic properties of binary solutions of lithium bromide or lithium chloride with methanol. Fluid Phase Equilib. 2005, 236 (1−2), 87−95. (38) Aznar, M.; Telles, A. S. Prediction of electrolyte vapor-liquid equilibrium by UNIFAC-Dortmund. Braz. J. Chem. Eng. 2001, 18 (2), 127−137. (39) Pitzer, K. S.; Mayorga, G. Thermodynamics of electrolytes 0.2. Activity and osmotic coefficients for strong electrolytes with one or both ions univalent. J. Phys. Chem. 1973, 77 (19), 2300−2308. (40) Wilf, M.; Awerbuch, L.; Bartels, C.; Mickley, M.; Pearce, G.; Voutchkov, N. The guidebook to membrane desalination technology: reverse osmosis, nanofiltration and hybrid systems: process, design, applications, and economics. Balaban Desalination Publications: L’Aquila, Italy, 2007; Vol. xii, p 524. (41) Mulder, M. Basic principles of membrane technology, 2nd ed.; Kluwer Academic: Dordrecht; Boston, 1996; p 564. (42) Razumovskii, L. P.; Markin, V. S.; Zaikov, G. Y. Diffusion of alcohols into polyamides. Vysokomol. Soedin., Ser. A 1985, 27 (4), 863− 867. (43) Skirrow, G.; Young, K. R. Sorption, diffusion and conduction in polyamide-penetrant systems 0.1. sorption phenomena. Polymer 1974, 15 (12), 771−776.

(44) Ulich, H. Ionic mobilities in non-aqueous solvents. Trans. Faraday Soc. 1927, 23, 388−393. (45) Impey, R. W.; Sprik, M.; Klein, M. L. Ionic solvation in nonaqueous solvents: the structure of lithium ion and chloride in methanol, ammonia, and methylamine. J. Am. Chem. Soc. 1987, 109 (20), 5900−5904. (46) Hartley, H.; Raikes, H. R. The mobilities of the elementary ions in methyl alcohol. Trans. Faraday Soc. 1927, 23 (0), 393−396. (47) Arena, J. T.; Manickam, S. S.; Reimund, K. K.; Freeman, B. D.; McCutcheon, J. R. Solute and water transport in forward osmosis using polydopamine modified thin film composite membranes. Desalination 2014, 343, 8−16. (48) Wei, J.; Qiu, C. Q.; Tang, C. Y. Y.; Wang, R.; Fane, A. G. Synthesis and characterization of flat-sheet thin film composite forward osmosis membranes. J. Membr. Sci. 2011, 372 (1−2), 292− 302. (49) Song, X. X.; Liu, Z. Y.; Sun, D. R. D. L. Nano gives the answer: Breaking the bottleneck of internal concentration polarization with a nanofiber composite forward osmosis membrane for a high water production rate. Adv. Mater. 2011, 23 (29), 3256. (50) Shaffer, D. L.; Werber, J. R.; Jaramillo, H.; Lin, S.; Elimelech, M., Forward osmosis: Where are we now? Desalination (0). (51) She, Q. H.; Jin, X.; Tang, C. Y. Y. Osmotic power production from salinity gradient resource by pressure retarded osmosis: Effects of operating conditions and reverse solute diffusion. J. Membr. Sci. 2012, 401, 262−273. (52) She, Q. H.; Hou, D. X.; Liu, J. X.; Tan, K. H.; Tang, C. Y. Y. 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, 170−182. (53) Kim, Y. C.; Elimelech, M. Adverse impact of feed channel spacers on the performance of pressure retarded osmosis. Environ. Sci. Technol. 2012, 46 (8), 4673−4681. (54) Kumar, P.; Varanasi, S. R.; Yashonath, S. Relation between the diffusivity, viscosity, and ionic radius of LiCl in water, methanol, and ethylene glycol: A molecular dynamics simulation. J. Phys. Chem. B 2013, 117 (27), 8196−8208.

H

DOI: 10.1021/es506347j Environ. Sci. Technol. XXXX, XXX, XXX−XXX