Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
Water flooding of cathodic gas diffusion layers (GDLs) in proton-exchange membrane fuel cells at high current densities or low temperatures limits efficient operation due to disturbed transport of reactants to the catalytic sites or products away into the channels. We utilize tomography-based direct pore-level simulations to provide quantitative insights into the transport characteristics of partially saturated GDLs with and without hydrophobic surface treatment to eventually guide the design of better GDLs. High-resolution (voxel size of 1.3 μm) computed tomography images of two different types commercial Toray TGP-H-120 GDLs, one of them untreated and the other treated by hydrophobic coatings, at different water saturation levels were taken. These images were then digitally processed to precisely segment the gas, water, and fiber phases. The digitalized phase information was used in direct pore-level numerical flow simulations to determine effective relative diffusivity of the gas phase, relative permeability of the gas and liquid water phase, and tortuosity in the gas phase. Mathematical morphology operations are used to calculate size distributions of the liquid water phase and the gas phase at different saturation levels for a better understanding of the pore occupation by water at different capillary pressure. Percolation simulations are used to provide information on the connectivity of the gas and liquid phases. The results were validated with available experimental data and semi-empirical correlations. Power law expressions provide a good level of accuracy for curve fitting. The hydrophobic coating does not affect the effective permeability and effective diffusivity of air; however, it improves the water permeability significantly. The quantitative results presented provide insights and guidance for designing GDLs with better suited transport behavior.
François Gallaire, Alessandro Bongarzone