Pourrahmani, HosseinVan Herle, Jan2022-06-202022-06-202022-06-202022-06-0910.1002/er.8218https://infoscience.epfl.ch/handle/20.500.14299/188537WOS:000808179500001Water management of the proton exchange membrane fuel cell (PEMFC) is an important parameter to improve efficiency, especially in cold weather conditions. The generated water of the electrochemical reactions inside PEMFC may stay inside the cell and fill the pores in the gas diffusion layer (GDL), which degrades the cell and reduces the lifetime. The formation of water in PEMFC has a direct relationship with the capillary pressure that can be determined by the GDL's wettability and microstructure. Furthermore, wettability is dependent on the material, which can be changed by the contact angles of the GDL. Using the computational fluid dynamic (CFD) methods, different values of the GDL's contact angle have been examined to observe their corresponding effects on the maximum GDL liquid removal. Higher values of the GDL liquid removal result in better water management. Using the simulation data, an artificial neural network (ANN) model was trained in the current density of 0.4 (A/cm2$$ \mathrm{A}/\mathrm{c}{\mathrm{m}}<^>2 $$) and the voltage of 0.604 (V) to predict the GDL liquid removal in 5700 points and to develop the optimization analysis. Results indicated that the highest GDL liquid removal of 0.225 (kg/m3s$$ \mathrm{kg}/{\mathrm{m}}<^>3\mathrm{s} $$) can be obtained by the GDL's contact angle of 102.63 degrees. Highlights PEMFC three-dimensional simulation model with seven path serpentine flow field. The impacts of GDL contact angle on the maximum GDL liquid removal. Training an ANN model to calculate the maximum GDL liquid removal. At the contact angle of 102.63 degrees, maximum GDL liquid removal is 0.225 [kg/m3s$$ \mathrm{kg}/{\mathrm{m}}<^>3\mathrm{s} $$].Energy & FuelsNuclear Science & Technologyartificial neural networks (ann)computational fluid dynamic (cfd)gas diffusion layer (gdl)proton exchange membrane fuel cell (pemfc)wettabilitythermal managementtext::journal::journal article::research article