Large-Scale Coupling Numerical Simulation of Two-Phase Flow and Electrochemical Phenomena in Alkaline Water Electrolysis

Abstract

The alkaline water electrolysis is one of the methods of large-scale hydrogen production, and it is required to improve its efficiency. Experimental approach and numerical studies have been conducted to improve the efficiency [1,2]. Abdelouahed et al. [3] confirmed that bubbles strongly influence the flow field pattern and mix the local electrolyte near the electrodes. Ehrl et al. [4] conducted a numerical simulation of electrochemical systems with natural convection and solved both multi-ion transportation and flow in the cell. In these studies, macroscopic (mm scale) hydrodynamical behavior have been discussed, however, the size of the bubbles in water electrolysis can be micro scale and microscale flow can be induced. We have been studied about the microscale hydrodynamical effect on the alkaline water electrolysis with two-phase flow, electrochemical, and electromagnetic coupling numerical simulations. In our previous studies [1,2], numerical simulations of flow around the aligned bubbles with a cyclic boundary condition are conducted and it is revealed that the rising bubbles suppress the cell overpotential and this suppression is enhanced by a bubble atomization, however, there is a strong limitation to the bubble movement and it had been difficult to discuss the bubble-bubble interaction and the influence of that on the efficiency of the alkaline water electrolysis cell. From the background mentioned above, in this study, large-scale coupling numerical simulations are conducted to elucidate the influence of the bubble-bubble interactions on the cell overpotential. Strength of the interaction depends on dispersibility of bubbles. So, we also discuss the influence of the dispersibility of multiple bubbles on the cell overpotential.In this study, electrochemical, two-phase flow, electromagnetic field coupling numerical simulation are conducted with the same procedure as previous studies [1,2]. The analysis area is expanded to calculate the bubble-bubble interaction and the number of grids is 16 million. Graphical Processing Unit (GPU) are used in this calculation for Large-scale parallel analysis to realize this simulation. The average applied current density is 400 mA/cm2, the bubble size of oxygen bubbles is 500μm and four bubbles are set close to the anode at initial condition. The numerical simulations are conducted with changing the dispersibility from 0.25 to 0.92. Dispersibility is 1 when bubbles are equally spaced and is 0 when bubbles are concentrated and in touched.Figure 1 shows the influence of the dispersibility (D) on the cell overpotential. From D = 0.25 to 0.43, the overpotential is almost decreased with the increase in D, and it increased with the increase in D from D = 0.43 to 0.92. Fig.2 shows concentration distribution, iso-surface (1.7 mol/L) and bubbles position with dispersibility of (a) D = 0.25, (b) D = 0.43 and (c) D = 0.92. In the case of low dispersibility (D = 0.25, Fig.2 (a)), the overpotential is high and some bubbles are combined and remain close to the anode. These bubbles close to the anode, make shielding effect and increase the ohmic loss. The anodic activation overpotential is also high because fewer bubbles reduce the flow and decrease the concentration near the anode. With the increasing in D, the overpotential is suppressed (D = 0.43, Fig.2 (b)). In this case of optimum dispersibility, some bubbles locate far from the anode. This is because, the upper bubble leaves from the anode because of the lift force against the anode [5] and induces repulsive force to the lower bubble [6]. This upper bubble far from the anode, reduces shielding effect and suppresses the ohmic loss. The anodic activation overpotential is also suppressed because the interaction between the bubbles enhance the flow which facilitate the ion transportation and increase the concentration near the anode. In the case of high dispersibility (D = 0.92, Fig.2 (c)), many bubbles remain close to the anode. These bubbles make shielding effect and increase the ohmic loss. The anodic activation overpotential is also increased because there is little interaction between bubbles to generate strong flow and the concentration near the anode are decreased. In conclusion, this study reveals that the interaction between bubbles affects the cell overpotential, and bubbles in optimum dispersibility, the cell overpotential can be suppressed. Acknowledgement This study is partially supported by Nissin Sugar Found and Japan High Performance Computing and Networking plus Large-scale Data Analyzing and Information Systems. Figure 1