Lattice Boltzmann simulation of water transfer in gas diffusion layers with porosity gradient of polymer electrolyte membrane fuel cells with parallel processing on GPU

Abstract

This study used the lattice Boltzmann method (LBM) to evaluate water distribution in the gas diffusion layer (GDL) of cathode PEM fuel cells (PEMFCs) with porosity gradient. Due to the LBM's capability of parallel processing with a GPU and the high volume of computing necessary, especially for small grids, the GPU parallel processing was done on a graphics card with the help of CUDA to speed up computing. The two-phase flow boundary conditions in the GDL are similar to the water transfer in the GDL of the PEMFCs. The results show that capillary force is the main cause of water transfer in the GDL, and gravity has little effect on the water transfer. Also, the use of GPU parallel processing on the graphics card increases the computation speed up to 17 times, which has a significant effect on running time. To investigate the gradient of porosity of GDLs with different porosity gradients, but the same average porosity coefficient and the same particle diameter have been evaluated. The simulation results show that the GDL with a 10% porosity gradient compared to the GDL with uniform porosity results in a 20.2% reduction in the amount of liquid water in the porous layer. Hence, increasing the porosity gradient of the GDL, further decreases the amount of liquid water in the porous layer. So, for the GDL with a porosity gradient of 14% this decrease is 29.8% and for the GDL with porosity gradient 18.5% this decrease is 38.8% compared to the GDL with uniform porosity.