Experimental Performance of a Membrane Desorber with a H2O/LiCl Mixture for Absorption Chiller Applications

For absorption cooling cycles using water as a refrigerant, H2O/LiCl mixtures are suitable for replacing conventional H2O/LiBr mixtures. In addition, membrane devices can be used to develop compact and lighter absorption systems, and they can operate with H2O/LiCl mixtures. The present paper describes an experimental evaluation of a membrane desorber/condenser operating at atmospheric pressure. Two operation modes were analyzed: continuous cycle operation and intermittent operation. For the first operation mode, the maximum desorption rate was 3.49 kg/h·m2, with a solution temperature of 90.3 °C and a condensation temperature of 25.1 °C. The lowest desorption rate value was 0.26 kg/h·m2, with a solution temperature of 75.4 °C and a condensation temperature of 40.1 °C. In the second mode, after three operating hours, the refrigerant fluid produced, per 1 m2 of membrane area, 7.7, 5.6, 4.3, and 2.2 kg, at solution temperatures of 90.3, 85.3, 80.4, and 75.4 °C, respectively. A one-dimension heat and mass transfer model is presented. The calculated values of desorption rate and outlet temperatures were compared with the experimental data; a square correlation coefficient of 0.9929 was reached for the desorption rate; meanwhile, for the outlet solution temperatures and the outlet cooling-water temperatures, a square correlation coefficient up to 0.9991 was achieved. The membrane desorber has the advantages of operating at atmospheric-pressure conditions, high condensation temperature, the ability to use different saline solution working mixtures, and different operation methods. These advantages can lead to new absorption systems.


Introduction
H 2 O/LiBr is the most used mixture in absorption chillers for air-conditioning applications. Water, as a refrigerant, shows convenient features: it has a high latent heat of evaporation; is a natural refrigerant; and is abundant, low cost, and harmless to the environment [1]. Absorption cycles using water as a refrigerant perform better than ammonia-based cycles, and a rectifier is unnecessary for them [2]. Among the water-based mixtures, H 2 O/LiCl mixtures are an interesting alternative showing some advantages compared with H 2 O/LiBr mixtures: triple-state point (solid, liquid, and vapor forms), long-term stability, and lower cost [3]. In addition, H 2 O/LiCl mixtures show better hygroscopic properties than other aqueous solutions, such as H 2 O/CaCl 2 [4], although the smaller chemical potential of LiCl relative to LiBr leads to a larger absorber area [5]. On the other hand, Ahamad et al. [6], through a simulation analysis, found that at similar operating system temperatures, the solution concentrations in the desorber and absorber were considerably lower for the H 2 O/LiCl mixture than for the H 2 O/LiBr mixture, considerably reducing the risk of crystallization in these components. Additionally, they found that a slightly higher maximum Coefficient Of Performance (COP) could be reached with the The literature reviewed demonstrates the suitability of H 2 O/LiCl mixtures with membrane devices, especially for absorption cooling applications. The aim of the present study was to demonstrate the suitability of the H 2 O/LiCl mixture for the operation of a membrane desorber for absorption-cooling-system applications. This paper presents an experimental evaluation of a membrane desorber with a H 2 O/LiCl mixture at atmospheric-pressure conditions. The membrane device uses the air gap membrane distillation (AGMD) configuration. Two operation modes were analyzed: (i) continuous cycle operation and (ii) intermittent operation. Furthermore, the effects of the solution temperature and cooling-water temperature on the desorption rate were analyzed, and a one-dimension heat and mass transfer model was constructed.

Air Gap Membrane Distillation (AGMD) Configuration
Membrane distillation is a thermal separation process using a hydrophobic membrane as an interphase contactor between two fluids at different temperatures, where the separation occurs by the vapor mass transfer. At relatively low operating pressures, the hydrophobicity of the membrane prevents the liquid phase from wetting the membrane pores, and vapor is the only phase to cross the membrane. The temperature difference between the two sides of the membrane produces the driving force for the vapor mass transfer from the fluid at a higher temperature (hot side) to the one at a lower temperature (cold side) [23]. A schematic diagram of the AGMD configuration is shown in Figure 1. Some of the advantages of the AGMD configuration are as follows: the air gap serves as a thermal insulation layer and reduces the heat loss from the membrane; latent heat is recovered without an external heat exchanger; thermal efficiency is higher; it is suitable for pilot testing plants; and the AGMD modules can be scaled up and made with polymeric corrosion-resistant materials [24]. branes and Accurel membranes was carried out assuming that they had the same membrane physical characteristics. The authors found that the mass transfer coefficient with Oxyphan membranes was 4.2 times higher than that with the second type. The water vapor mass flux was 2.1 times higher with the Accurel membranes compared with the Oxyphan ones.
The literature reviewed demonstrates the suitability of H2O/LiCl mixtures with membrane devices, especially for absorption cooling applications. The aim of the present study was to demonstrate the suitability of the H2O/LiCl mixture for the operation of a membrane desorber for absorption-cooling-system applications. This paper presents an experimental evaluation of a membrane desorber with a H2O/LiCl mixture at atmospheric-pressure conditions. The membrane device uses the air gap membrane distillation (AGMD) configuration. Two operation modes were analyzed: (i) continuous cycle operation and (ii) intermittent operation. Furthermore, the effects of the solution temperature and cooling-water temperature on the desorption rate were analyzed, and a one-dimension heat and mass transfer model was constructed.

Air Gap Membrane Distillation (AGMD) Configuration
Membrane distillation is a thermal separation process using a hydrophobic membrane as an interphase contactor between two fluids at different temperatures, where the separation occurs by the vapor mass transfer. At relatively low operating pressures, the hydrophobicity of the membrane prevents the liquid phase from wetting the membrane pores, and vapor is the only phase to cross the membrane. The temperature difference between the two sides of the membrane produces the driving force for the vapor mass transfer from the fluid at a higher temperature (hot side) to the one at a lower temperature (cold side) [23]. A schematic diagram of the AGMD configuration is shown in Figure 1. Some of the advantages of the AGMD configuration are as follows: the air gap serves as a thermal insulation layer and reduces the heat loss from the membrane; latent heat is recovered without an external heat exchanger; thermal efficiency is higher; it is suitable for pilot testing plants; and the AGMD modules can be scaled up and made with polymeric corrosion-resistant materials [24].

Methodology
The membrane desorber/condenser assessment was performed by varying the main system's operating parameters, such as LiCl mass flow rate (ṁLiCl), inlet solution temperature (TLiCl,in), and inlet cooling-water temperature (Tcw,in). Additionally, two operation modes were analyzed. In the first, a continuous operation absorption cycle was assumed,

Methodology
The membrane desorber/condenser assessment was performed by varying the main system's operating parameters, such as LiCl mass flow rate (ṁ LiCl ), inlet solution temperature (T LiCl,in ), and inlet cooling-water temperature (T cw,in ). Additionally, two operation modes were analyzed. In the first, a continuous operation absorption cycle was assumed, which means that the initial LiCl concentration was constant. In the second case, intermittent operation mode was assumed, meaning that the desorption/condensation and the evaporation/absorption processes in an absorption system separately occurred and that the LiCl concentration increased with operation time.

Experimental Setup
The experimental setup was mainly integrated by a membrane desorber/condenser unit, heating system, and cooling system, as shown in Figure 2. The membrane desorber/condenser unit functioned to separate part of the water contained in the LiCl solution as previously described. The unit comprised two Nylamid support plates, neoprene gaskets, a metallic mesh, an aluminum cooling plate, and a PTFE membrane, as shown in Figure 3.
the LiCl concentration increased with operation time.

Experimental Setup
The experimental setup was mainly integrated by a membrane desorber/condenser unit, heating system, and cooling system, as shown in Figure 2. The membrane desorber/condenser unit functioned to separate part of the water contained in the LiCl solution as previously described. The unit comprised two Nylamid support plates, neoprene gaskets, a metallic mesh, an aluminum cooling plate, and a PTFE membrane, as shown in Figure 3.   the LiCl concentration increased with operation time.

Experimental Setup
The experimental setup was mainly integrated by a membrane desorber/condenser unit, heating system, and cooling system, as shown in Figure 2. The membrane desorber/condenser unit functioned to separate part of the water contained in the LiCl solution as previously described. The unit comprised two Nylamid support plates, neoprene gaskets, a metallic mesh, an aluminum cooling plate, and a PTFE membrane, as shown in Figure 3.   [25].
The characteristics of the hydrophobic membrane are shown in Table 1. The characteristics of the hydrophobic membrane are shown in Table 1.  ) 175 µm The heating system comprises a heating bath with temperature control, a stainlesssteel plate heat exchanger (PHE), a pump, and a Coriolis mass flowmeter. This system is in charge of supplying the necessary heat to the LiCl solution in the PHE before the solution enters the desorber unit.
The cooling system comprises a circulating cooling bath with temperature control and an integrated pump. This system provides the cooling to the desorber/condenser unit to condense the water in a vapor phase passing through the membrane. A detailed description of the membrane desorber/condenser unit was already reported by Ibarra et al. [25]. Figure 4 shows a photograph of the experimental setup.
The heating system comprises a heating bath with temperature control, a stainlesssteel plate heat exchanger (PHE), a pump, and a Coriolis mass flowmeter. This system is in charge of supplying the necessary heat to the LiCl solution in the PHE before the solution enters the desorber unit.
The cooling system comprises a circulating cooling bath with temperature control and an integrated pump. This system provides the cooling to the desorber/condenser unit to condense the water in a vapor phase passing through the membrane.
A detailed description of the membrane desorber/condenser unit was already reported by Ibarra et al. [25]. Figure 4 shows a photograph of the experimental setup. For the continuous operation mode, 16 experimental test runs were carried out; meanwhile, 4 test runs in intermittent operation mode were completed. The operating temperatures were selected on the basis of cooling absorption systems operating with renewable thermal energies [12,13] or low-grade heat sources [26]. The uncertainties of the measured variables and instruments used in the experimental test runs are shown in Table  2, and Tables 3 and 4 report the experimental operating conditions for each operation mode.  For the continuous operation mode, 16 experimental test runs were carried out; meanwhile, 4 test runs in intermittent operation mode were completed. The operating temperatures were selected on the basis of cooling absorption systems operating with renewable thermal energies [12,13] or low-grade heat sources [26]. The uncertainties of the measured variables and instruments used in the experimental test runs are shown in Table 2, and  Tables 3 and 4 report the experimental operating conditions for each operation mode.

Heat and Mass Transfer Model
A conceptual diagram of the desorber/condenser unit is shown in Figure 5. The hot saline solution is in direct contact with one side of the porous membrane. The refrigerant fluid evaporates at the entrance of the pores because of the hydrophobic nature of the membrane while a concentration profile is created: the salt concentration of the bulk fluid (X bulk ) increases until a higher concentration at the membrane interface (X mem ). The working fluid phase change occurs according to the vapor-liquid equilibrium of the H 2 O/LiCl mixture. The vapor permeates the membrane through the pores by the Poiseuille-Knudsen mechanism, crosses the air gap by diffusion, and condenses on the refrigerated plate. The heat transfer through the system is driven by the temperature difference between the hot saline solution channel and the cooling-water channel. The vapor mass transfer simultaneously occurs in the same direction with the heat transfer. The coupling of these phenomena is called the Soret effect. In order to estimate the global performance parameters to scale and design a membrane desorber/condenser, a previously described one-dimension heat and mass transfer model was used [27] and adapted; it includes the H2O/LiCl mixture equilibrium conditions to describe the vapor water desorption. The following assumptions were made: (1) The desorber/condenser unit operates at steady state conditions.
(2) Thermophysical properties are constant. Energy conservation was considered in the different regions inside the experimental device [28,29].
The desorption heat flux (QD) transferred from the hot saline solution to the membrane is given by: (1) In order to estimate the global performance parameters to scale and design a membrane desorber/condenser, a previously described one-dimension heat and mass transfer model was used [27] and adapted; it includes the H 2 O/LiCl mixture equilibrium conditions to describe the vapor water desorption. The following assumptions were made: (1) The desorber/condenser unit operates at steady state conditions. Energy conservation was considered in the different regions inside the experimental device [28,29]. The desorption heat flux (Q D ) transferred from the hot saline solution to the membrane is given by: The heat transferred to the membrane: Inside the air gap: In the permeate film: At the condensing plate: In the cooling stream: where h is the convective heat transfer coefficient, J w is the desorption rate, λ w is the latent heat of vaporization of water, k is the thermal conductivity, and δ refers to the thickness. Subscripts mem, gap, and p refer to the membrane, air gap, and condensation plate, respectively. Subscripts 1 to 7 are the numerical notation in Figure 5. Q D can be simplified in a mathematical expression as a function of the H 2 O/LiCl mixture temperature and the cooling-water stream temperature (T 1 and T 7 , respectively). The equation system, integrated by Equations (2)-(6), was solved to dismiss the intermediate temperatures (T 2 to T 6 ). Thus, Equation (7) is proposed: where the ψ, ω, k eq , and h eq are defined, respectively, as follows: k eq = k mem k gap δ gap k mem + δ mem k gap (10) The convective heat transfer coefficients h 1 and h 6 were estimated from empirical correlations. In the H 2 O/LiCl mixture channel, the correlation proposed by Shah and London [30] was used to calculate the Nusselt number, which is valid for a fully developed flow in thin rectangular channels. On the other hand, in the cooling-water channel, the correlation reported by Khayet [31] was used. The correlation proposed by Bird et al. [32] was used to calculate the convective heat transfer coefficient for the condensate film (h 4 ).
The temperatures for each zone inside the desorber/condenser device (T 2 to T 6 ) were calculated from Equations (1)-(6). However, T 1 and T 7 , considered as the bulk temperatures of H 2 O/LiCl solution and cooling-water stream, respectively, were assumed to be the experimental inlet and outlet temperatures, respectively. Stream inlet temperatures (T LiCl,in and T cw,in ) are input parameters for the mathematical model, and the outlet temperatures were calculated by a numerical procedure, as a function of the heat flux in both streams: where Cp is the heat capacity,ṁ is the mass flow rate, and Q l is the heat loss to the environment. The ratio between the mass fraction at the liquid-membrane interface (X m ) and the mass fraction in the bulk flow (X b ) is defined as follows [22,33]: where ϕ bl is the convective mass transfer coefficient at the boundary layer, which was calculated by the correlation proposed by Woods et al. [33]; and ρ LiCl is the H 2 O/LiCl solution density. The driving force of the desorption rate (J w ) is the partial pressure difference between the interface de evaporation (p 1 ) and condensate film (p 4 ). However, the mass transfer flux is restricted by a global mass transfer coefficient (K ov ): The global mass transfer coefficient involves the boundary layer mass transfer coefficient at the liquid-membrane interface (K bl ), the mass transfer coefficient on the membrane (K mem ), and the mass transfer coefficient for the air gap (K gap ). It is expressed as: The mass transfer coefficient for the membrane is expressed as: The mass transfer coefficient for the air gap is defined as: The boundary layer mass transfer coefficient is defined as: In these equations, M w is the molecular mass of water; R is the universal gas constant; P and p lm are the total operating pressure and the logarithmic-mean pressure of air, respectively; δ mem , ε, and τ are the thickness, porosity, and tortuosity of the membrane, respectively; D vap is the water vapor/air binary mass diffusion coefficient; r p is the membrane mean pore radius; δ gap is the air gap thickness; and p b and p bl are the vapor partial pressure at the bulk layer and the boundary layer, respectively. The thermodynamic properties of the H 2 O/LiCl solution were calculated by using the correlations reported by Chaudhari and Patel [34] and Wimby and Berntsson [35].
In the experimental device, the channel length is small; therefore, the temperature difference between the inlet and the outlet is smaller than the temperature difference across the hot saline solution channel and cooling-water channel. Thus, a one-dimensional model assumption allows a general description of the system for engineering design purposes.

Continuous Cycle Operation
The experimental desorption rate (J w ) was measured on the basis of the mass of distillate water produced by the membrane desorber after a defined time period, under different operating conditions. As demonstrated in previous reports with LiBr aqueous solutions [25,27], the desorption rate is mainly affected by the solution temperature rather than by the condensation temperature. For instance, at a solution temperature of 90.3 • C at the lowest condensation temperature, the desorption rate was 2.6 times higher than at 75.4 • C for the same condensation temperature. Meanwhile, the desorption rate at the highest condensation temperature was 9.4 times higher at the solution temperature of 90.3 • C than at 75.4 • C. As can be seen in Figure 6, the influence of the condenser temperature in the desorption rate was greater at the lowest solution temperature because at 75.4 • C the desorption rate was 5.2 times higher at the lowest condenser temperature than at the highest temperature; meanwhile, at the solution temperature of 90.3 • C, the desorption rate was just 1.4 times higher at the same condenser temperatures. The maximum desorption rate was 3.49 kg/h·m 2 at a solution temperature of 90.3 • C and a condensation temperature of 25.1 • C. On the other hand, the lowest desorption rate value was 0.26 kg/h·m 2 at a solution temperature of 75.4 • C and a condensation temperature of 40.1 • C. Because the vapor mass transfer driving force depends on the temperature difference between both sides of the membrane, as this temperature difference increases, the desorption rate also increases. ]  In Table 5, a comparison of the experimental results of this work with what was reported in the literature is presented. The desorption rates with the H2O/LiCl mixture are slightly lower than those reported with the H2O/LiBr mixture under similar operating conditions and with the same membrane configuration; however, the desorption rates are in a similar range to those reported with the H2O/LiBr mixture. These results show the suitability of using the H2O/LiCl mixture in a desorption process with membrane devices for absorption cooling systems. Table 5. Comparison of the experimental desorption rates for membrane desorbers reported in the literature.

Reference
Configuration dp (μm) Working Mixture  In Table 5, a comparison of the experimental results of this work with what was reported in the literature is presented. The desorption rates with the H 2 O/LiCl mixture are slightly lower than those reported with the H 2 O/LiBr mixture under similar operating conditions and with the same membrane configuration; however, the desorption rates are in a similar range to those reported with the H 2 O/LiBr mixture. These results show the suitability of using the H 2 O/LiCl mixture in a desorption process with membrane devices for absorption cooling systems.

Intermittent Operation
For this analysis, the refrigerant production (or water distilled) was defined as the quantity, in kilograms, of the refrigerant fluid produced per 1 m 2 of membrane area. After three operating hours, at a constant condensation temperature of 30.1 • C, the refrigerant fluid produced was 7.7, 5.6, 4.3, and 2.2 kg, at solution temperatures of 90.3, 85.3, 80.4, and 75.4 • C, respectively. Figure 7 shows that the refrigerant produced decreased with operation time and showed a nonlinear dependence, particularly with high solution temperatures. As previously mentioned, the mass transfer driving force increases at higher temperature differences; thus, the refrigerant production was higher at the highest solution temperature. On the other hand, as a function of operation time, the amount of refrigerant contained in the solution decreased and the LiCl concentration increased, as is shown in Figure 8, so the refrigerant production decreased. Moreover, this behavior was observed in a previous study [46].

Intermittent Operation
For this analysis, the refrigerant production (or water distilled) was define quantity, in kilograms, of the refrigerant fluid produced per 1 m 2 of membrane are three operating hours, at a constant condensation temperature of 30 Figure 7 shows that the refrigerant produced decreased wit tion time and showed a nonlinear dependence, particularly with high solution t tures. As previously mentioned, the mass transfer driving force increases at high perature differences; thus, the refrigerant production was higher at the highest temperature. On the other hand, as a function of operation time, the amount of ref contained in the solution decreased and the LiCl concentration increased, as is s Figure 8, so the refrigerant production decreased. Moreover, this behavior was o in a previous study [46].

Model Validation
On the basis of the one-dimensional heat and mass transfer model, the desorption rate (J W ), outlet solution temperatures (T LiCL,out ), and outlet cooling-water temperatures (T cw,out ) were calculated. Figure 9 shows the comparison of theoretical and experimental refrigerant desorption rates. A square correlation coefficient (R 2 ) of 0.9929 was reached, which means that the mathematical model provided values close to the experimental data. Figures 10 and 11 show the theoretical T LiCl,out as a function of the experimental T LiCl,out and the theoretical T cw,out as a function of the experimental T cw,out , respectively. In both cases, an R 2 value up to 0.9991 was achieved. This validation is essential for design heat networks to improve the heat efficiency of the absorption chillers.

Model Validation
On the basis of the one-dimensional heat and mass transfer model, the des rate (JW), outlet solution temperatures (TLiCL,out), and outlet cooling-water temp (Tcw,out) were calculated. Figure 9 shows the comparison of theoretical and exper refrigerant desorption rates. A square correlation coefficient (R 2 ) of 0.9929 was r which means that the mathematical model provided values close to the experimen Figures 10 and 11 show the theoretical TLiCl,out as a function of the experimental TL the theoretical Tcw,out as a function of the experimental Tcw,out, respectively. In both c R 2 value up to 0.9991 was achieved. This validation is essential for design heat n to improve the heat efficiency of the absorption chillers.     One way to improve the mathematical model is to divide the channels into vertical control volumes, where each one has a uniform temperature. This means considering several vertical one-dimensional heat transfer paths in parallel.

Conclusions
A membrane desorber/condenser unit was experimentally evaluated, under laboratory conditions, with a H2O/LiCl solution. Two operation modes were analyzed: continuous cycle operation and intermittent operation. In the first case, four solution temperatures (90. 3  One way to improve the mathematical model is to divide the channels into vertical control volumes, where each one has a uniform temperature. This means considering several vertical one-dimensional heat transfer paths in parallel.

Conclusions
A membrane desorber/condenser unit was experimentally evaluated, under laboratory conditions, with a H 2 O/LiCl solution. Two operation modes were analyzed: continuous cycle operation and intermittent operation. In the first case, four solution temperatures (90.3, 85.3, 80.4, and 75.4 • C) and four condensation temperatures (25.1, 30.1, 35.1, and 40.1 • C) were evaluated. According to the results, the desorption rate was principally affected by the solution temperature rather than by the condensation temperature; however, the condensation temperature effect on the desorption rate was notable at the lowest solution temperature. The maximum desorption rate was 3.49 kg/h·m 2 at a solution temperature of 90.3 • C and condensation temperature of 25.1 • C.
On the other hand, the lowest desorption rate value was 0.26 kg/h·m 2 at a solution temperature of 75.4 • C and condensation temperature of 40.1 • C. The same four solution temperatures as the first operation case were analyzed in the second mode at a constant condensation temperature of 30.1 • C. Mathematical model validation was carried out, and a square correlation coefficient of 0.9929 was reached for the desorption rate; meanwhile, for the outlet solution temperatures and the outlet cooling-water temperatures, a square correlation coefficient up to 0.9991 was achieved. After three operating hours, the refrigerant fluid produced, assuming a 1 m 2 of membrane area, was 7.7, 5.6, 4.3, and 2.2 kg with solution temperatures of 90.3, 85.3, 80.4, and 75.4 • C, respectively. The refrigerant production increased as the solution temperature increased; however, the refrigerant production decreased with operation time because the LiCl concentration increased. On the basis of the experimental results, the H 2 O/LiCl mixture can be used to replace the conventional H 2 O/LiBr mixture for absorption cooling systems that integrate membrane devices.

Data Availability Statement:
The data presented in this study are available on request from the corresponding.