Microstructure reconstruction of the gas diffusion layer and analyses of the anisotropic transport properties

•Numerical and XCT 3D reconstruction are used to simulate Toray and Freudenberg GDLs.•Anisotropic transport properties and conductivity are obtained by Pore Scale Model.•Liquid water permeability and saturation are solved by Lattice Boltzmann Method.•Toray GDL has strong anisotropic properties due to fiber orientation and resin.•All properties of the two studied GDLs are correlated to the compression strain. The gas diffusion layer (GDL) is a key component in a proton exchange membrane fuel cell and a comprehensive understanding of its transport properties is imperative for improving the performance and durability of a fuel cell. In this study, two microscopic reconstruction methods, stochastic numerical and X-ray computed tomography (XCT) reconstruction, are employed to generate 3D microstructure of two different types of GDL. The stochastic numerical reconstruction method simulates all available phases, including the pores, carbon fibers, binders, and PTFE, to generate the 3D microstructure of the GDL in comparison with experimentally obtained GDL using the XCT method. The porosity and pore size distribution of the reconstructed GDL are compared and analyzed. Pore scale model is employed to obtain the effective and anisotropic transport properties including gas diffusivity, electrical and thermal conductivity as a function of porosity caused by different compression strain. Furthermore, Lattice-Boltzmann method is used to determine the anisotropic liquid water permeability and saturation as a function of capillary pressure. The results show significant anisotropic transport properties of Toray GDL due to its disc-shape binder existed at the intersection of the fibers. On the other hand, Freudenberg GDL shows mostly isotropic transport properties due to its uniformly distributed carbon fiber without binder. The combined results not only provide a reliable framework for investigating GDL microstructure, but also provide useful property correlations for fuel cell simulations and operations.