Factors governing mass transfer during membrane electrodialysis regeneration of LiCl solution for liquid desiccant dehumidification systems

Abstract This study investigates the mass transfer mechanisms and the performance of membrane electrodialysis (ED) for regenerating lithium chloride (LiCl) solution commonly used in liquid desiccant dehumidification systems. Experiments were conducted using an ED experimental system while numerical simulation was performed using COMSOL Multiphysics. The results showed that the water flux transfer due to osmosis and electro-osmosis during ED regeneration of LiCl liquid desiccant was significant and could not be ignored. The water flux due to osmosis and electro-osmosis is directly associated with the osmotic gradient and the applied current between the cathode and anode, respectively. The average flux of water from the spent solution to the regenerated solution decreased from 0.292 to 0.161 g/s m 2 when the initial concentration of the solutions in the spent and regenerated tanks increased from 18 to 30% (wt/wt) with the same applied current of 12 A and the same solution flow rate of 100 L/h. On the other hand, the salt flux due to osmosis was insignificant. The average salt flux transfer was 0.0053 g/s m 2 when the initial concentration difference between the regenerated and the spent channels was 25% (wt/wt). Simulations were conducted to elucidate the relationship between the concentration profile of LiCl solution along the membrane surface and the concentration polarization in the ED channel with respect to the circulation flow rate and applied current. Overall, the results suggest that the concentration difference between the regenerated and spent LiCl solutions should be minimized for an optimum ED performance.

suggest that the concentration difference between the regenerated and spent LiCl solutions 1 should be minimized for an optimum ED performance. The increase in energy demand for air conditioning and climate changes are some of the 3 most significant challenges facing the building sector (Pérez-Lombard, Ortiz, & Pout, 2008). The building sector accounts for around 40% of total world energy usage, of which about 5 50% is attributed to heating, ventilation and air-conditioning (HVAC) systems (Duan, Zhan, can only migrate through cation-exchange membranes (CEMs) and anion-exchange 10 membranes (AEMs), respectively.

11
The use of ED for liquid desiccant regeneration was first proposed by (Li & Zhang, 2009), 12 who used a single stage photovoltaic-electrodialysis (PV-ED) system. A double stage PV-ED There are four major mass transfer mechanisms in ED, namely electro-osmosis, osmosis, 11 ion migration and diffusion. Electro-osmosis and osmosis are responsible for water transport.  The research method used to investigate the mass transfer mechanisms of ED for liquid 1 desiccant regeneration is illustrated in Fig. 1. It mainly consists of three steps: i) the 2 experimental design and experimental tests; ii) the ED characteristics identification, and the 3 modelling system setup and validation; and iii) the experimental and numerical investigation 4 of the mass transfer mechanisms of ED for liquid desiccant regeneration.

5
A range of experimental tests, as summarized in Table 1, were first designed and 6 conducted based on a lab-scale ED experimental setup, which will be introduced in Section 3.

7
Based on the experimental data collected, the ED characteristics such as the membrane water  The water flux (J w,os ) and salt flux (J salt,Dif ) through the ED channels due to osmosis and 18 diffusion without current apply can be determined using Eqs.
(1) and (2), respectively. When 19 an electric current is applied, both osmosis and electro-osmosis can affect the water flux from 20 the spent channels to the regenerated channels and the total water flux (J w,tot ) transferred can 21 be determined by Eq. (3).

22
The net salt mass flux (J salt,net ) through the ED membranes with an applied current is a 23 combination of the effects of the ions migration and diffusion, and can be calculated by Eq.
(4). The mass flux from the spent channels to the regenerated channels of the ED stack (J ED ) 1 can therefore be determined by Eq. (5).
2 J w,os where J salt,I and J w,eos are respectively the salt flux and water flux due to applied current, and regenerated solutions at the exit of the ED stack can be determined by Eqs. (7) and (8), where C G,exit and C S,exit are the concentrations of the solutions at the exits of the regenerated 2 channels and spent channels respectively, and ̇1 and ̇2 are the volumetric flow rates of the 3 inlet solutions into the regenerated and spent cycles, respectively. The geometric dimensions of the ED cell were specified based on the manufacturing data of 8 the experimental setup to be introduced in Section 3. The key parameters used in the model 9 are summarized in Table B.1 in Appendix B.  (Tanaka, 2010), in which the water transport number due to the electro-osmosis and the water 10 permeability factor of the membranes due to osmosis were obtained from the experiments (i.e.
where B is the water permeability factor of AEMs and CEMs which is calculated based on 16 overall water transfer from the low concentration side to the high concentration side of the 17 membrane wall, is the osmotic pressure, c is the molar concentration of the solution, n is the 18 moles number of the solute in the solution (e.g. n=2 for LiCl), R is the gas constant, T is the 19 absolute solution temperature, i is the applied current density, t w is the water transport number 20 of membranes, and the subscripts G and S represent the regenerated and spent channels, 21 respectively.

22
The heat transfer between the spent and regenerated channels of the ED cells was not 23 considered in this study. In addition, it was assumed that the regenerated and spent channels 24 have the same hydrodynamic characteristics as the same solution flow rates were used in both 1 flow channels.   Table B.1 in Appendix B.

16
Lithium sulfate (Li 2 SO 4 ) solution was used as the electrode rinsing solution in the ED stack.

17
Two extra cation CMV membranes were used at the cathode and anode to prevent sulfate 18 from migrating into the spent and regenerated solutions. The regenerated, spent and rinsing 19 electrolyte solutions were circulated through three 3.5 L transparent PVC tanks, respectively.

20
The solutions were circulated through the ED stack by three magnetic drive pumps. Further 21 details of this ED system are available elsewhere (Guo et al., 2016).

22
A portable density meter (30PX Densito) was used to measure the density and temperature  Table B.2 in Appendix B.
where m is the number of the independently measured variables.   Fig. 3 shows the osmotic pressure differences between the spent and regenerated channels, 1 and the fluxes of water and LiCl through the membranes at various concentration differences 2 between the two solutions (i.e. Experimental cases 1-5). There was no electric potential 3 between the cathode and anode in order to eliminate any electro-osmosis effect.

4
The osmotic pressure difference between the spent and regenerated channels decreased 5 with operating time since the concentration difference between both channels decreased 6 continuously over time (Fig. 3a). As shown in Fig. 3b), the average water flux from the spent 7 to the regenerated solutions within 2 hours operation was about 0.23 g/s.m 2 when the initial 8 concentration difference between the regenerated and spent solutions was 25% (wt/wt), 9 whereas the average water flux decreased to 0.026 g/s.m 2 when the initial concentration 10 difference between the regenerated and spent solutions decreased to 5% (wt/wt). For this 11 group of the experiments, the maximum increase of the regenerated and spent solution 12 temperature during the 2 hours test was 3.9 o C. This temperature increase is mainly due to the 13 heat rejection from the solution pumps and the variation in the room temperature.
14 Salt flux transfer due to the concentration difference between the regenerated and spent 15 solutions was generally small when comparing to that of water flux transfer (Fig. 3c). The 16 average salt flux transfer was 0.0053 g/s.m 2 when the initial concentration difference between 17 the regenerated and the spent solutions was 25% (wt/wt).

18
Overall, water transfer due to osmosis is more significant than salt transfer. It is therefore 19 necessary to minimize the concentration difference between the regenerated and spent 20 solutions to control the negative impact of osmosis on the ED performance for liquid 21 desiccant regeneration.

Ion transfer due to applied current 22
The salt flux transfer during ED regeneration was analyzed using the third experiment set 23 (i.e. Experiments cases 10-18). Table 2   The average concentration difference of the regenerated solution between the entrance and 7 the exit of the ED stack (∆C ���� ) during the whole test period for each case is also shown in Table   8 2. However, it is hard to identify which factor has a higher impact on the ED regeneration 9 performance. Basically, the regeneration performance is related to the salt mass transfer from 10 the spent channels to the regenerated channels and this salt mass transfer is directly associated 11 with the applied current.  29.031% (wt/wt). In the regenerated channel, the concentration of the bulk solution was 13 always lower than that at the membrane walls, while the concentration of the bulk solution in 14 the spent channels was higher than that at the membrane walls (Fig. 7). boundary layers on the CEM was larger than that on the AEM and this was because the 24 mobility of Cl − ions was larger than that of Li + ions (Strathmann, 2004). As a result, in the regenerated channel, the concentration on the AEM surface was less than that on the CEM 1 surface. The concentration at the membrane walls increased with the increase of the applied 2 current. However, in the spent channel, the concentration at the membrane wall decreased 3 with the increase of the applied current. The maximum concentration at the CEM wall in the 4 regenerated channel was about 30.3% (wt/wt), while the minimum concentration at the same 5 membrane wall in the spent channel decreased to 27.7% (wt/wt) when the applied current was 6 12 A.  The effects of four key mass transfer mechanisms, namely osmosis, diffusion, electro-2 osmosis, and ion migration on water and ion transport during the regeneration of LiCl 3 desiccant solution using electrodialysis (ED) were experimentally and numerically 4 investigated. Ion flux, the thickness of the concentration polarization boundary layer, and salt 5 concentration profile along ED membranes were numerically investigated using Nernst-Plank 6 equations, and computational fluid dynamics using COMSOL Multiphysics software. The 7 results showed that water transport due to osmosis is more significant than salt transport. The 8 average water fluxes transferred from the spent to the regenerated solutions within 2 hours 9 operation were 0.23 and 0.026 g/s.m 2 when the initial concentration differences between the 10 regenerated and spent solutions were 25 and 5% (wt/wt), respectively. The average salt flux 11 transfer was 0.0053 g/s.m 2 when the initial concentration difference between the regenerated 12 and the spent channels was 25% (wt/wt). The flux of water transfer increased with the 13 increase of the applied current but it was decreased with the increase of the solution initial  The results also showed that, unlike conventional desalination applications using ED, 22 water transport due to osmosis and electro-osmosis during the regeneration of LiCl liquid 23 desiccant could not be neglected. Both the experiment and the numerical results showed that the concentration difference between the regenerated and spent LiCl solutions should be 1 minimized in order to achieve a better ED regeneration performance.   Energy Conversion and Management,88,[218][219][220][221][222][223][224][225][226][227][228][229][230] Marszal, A. J., Heiselberg, P., Bourrelle, J., Musall, E., Voss, K., Sartori, I., & Napolitano, A.    Waugaman, D., Kini, A., & Kettleborough, C. (1993). A review of desiccant cooling systems.      Step 1 Step 3 Fig. 2. Comparison between the model simulated concentration difference between the 3 entrance and exit of the ED stack and that derived from the experimental data.      Main characteristics of the ED and inputs of the ED model.