Pore-scale parametric sensitivity analysis of liquid water transport in the gas diffusion layer of polymer electrolyte membrane fuel cell

Abstract

Water management in the gas diffusion layer (GDL) is crucial for improving the performance of polymer electrolyte membrane fuel cells (PEMFCs). In this study, a modified stochastic method that fully took into account the fiber deflection and layered structure in the thickness direction was proposed to reconstruct the realistic porous structures of three-dimensional (3D) GDL. The liquid water transport within the GDL was then numerically studied using the 3D Lattice Boltzmann method (LBM). Subsequently, the contributions of different parameters (porosity, fiber diameter, thickness, wettability, and pressure difference) to the transport process of liquid water were elucidated by implementing a multi-parametric sensitivity analysis (MPSA). The numerical results indicate that the liquid water penetration process can be divided into initial injection, selective percolation, and diffusion stages. With the decreases in porosity or increase in contact angle, the flow is inclined to transform from gentle flow to accumulation and secondary breakthrough owing to the increased percolation resistance. Further accumulation of the liquid within GDL can be identified when the evolution time progresses from the breakthrough point to the creep point. Moreover, for GDL, the degree of influence of the structural parameters on the breakthrough velocity is ranked in the order of GDL porosity, fiber diameter, GDL thickness, pressure difference, and GDL wettability. The total liquid saturation is dominated by GDL wettability and is insensitive to GDL porosity and fiber diameter owing to the conflict between the percolation resistance of the pores and the percolation time within GDL. The current work consolidates the understanding of the influencing mechanisms of different structural parameters on liquid water transport characteristics, contributing to the optimal design of the gas diffusion layer.