Simulate fluid transport in gas diffusion layers of PEM fuel cells using lattice Boltzmann method and x-ray computed tomography

摘要

Polymer electrolyte fuel cell (PEMFC) is a devise to convert chemical energy to electricity by harvesting the electrons released in oxidation of hydrogen. The performance of PEMFC is affected by many factors, and one of them is gas flow in the porous gas diffusion layer (GDL) and catalyst layer (CL). The main objective of this PhD project is to investigate the impact of micron-scale pore structures of GDL on fluid flow using a combination of numerical modelling and imaging technology with a view to improve fuel cell design. X-ray computed micro-tomography was developed to visualize 3D structures of GDL at scales of one microns, and focused ion beam (FIB) scanning electron microscope was developed to visualize the 3D structure of the CL at a scale of a few nano-metres. The 3D structures were then combined with the lattice Boltzmann equation (LBE) model to investigate the gas flow through the GDL. The simulated microscopic velocity not only reveals the detailed gas flow paths in the GDL, but also provides a way to estimate the macroscopic transport parameters, including anisotropic permeability, diffusivity and tortuosity, some of which are difficult to measure experimentally. The attraction of the LBE methods is their flexibility in dealing with different microscopic forces and complicated boundaries. Different LBE methods have been developed in the literature, including single-relaxation time LBE model and multiple-relaxation time LBE model, with the former being claimed to be superior to the later. In this project, I thoroughly investigated the performance of the single-relaxation time LBE and the multiple-relaxation time LBE for simulating single-phase flow in GDL and other porous media. The results showed that, using only two thirds of the computational time of multiple-relaxation LBE model, the single-relaxation time LEB model could give reasonable results when the relaxation time was unity. For unity relaxation time, the fluid viscosity can be recovered by adjusting the size of the time step. This is significant for 3D simulations which are computation-demanded. Practical applications need to stack the fuel cells and to avoid gas leakage, in which the GDLs will be non-uniformly compressed. The impact of the compression on gas flow and hence fuel cell performance was also investigated. The by-product of fuel cells is water generated at the cathode; how to drain the water is a critical issue in fuel cell design. Based on the 3D x-ray images, I simulated the movement of liquid water through GDL from the catalyst layer to the channel with a view to investigate the impact of making GDL hydrophobic on water flow pattern. Another contribution of this thesis is gas flow in the catalyst layer in which the averaged pore sizes is less than one micron and the Knudsen number cannot be neglected. The pore geometry of the pore in catalyst layer was simplified into a bundle of tubes with various diameters that can be calculated from the pore-size distribution. A model for gas flow in each tube is then simulated; the results show that the permeability of the catalyst layer is not constant but varies with Knudsen number, meaning that the permeability of catalyst layer for oxygen, water vapour, nitrogen and hydrogen is different. Assuming a constant permeability for all the gases, as used in the available fuel cell models in literature, could give rise to significant errors. The work presented in this thesis improved our understanding of gas flow processes in fuel cells, and would offer a tool to help fuel cell design.