Advanced computational tools for PEM fuel cell design: Part 2. Detailed experimental validation and parametric study

This paper reports on the systematic experimental validation of a comprehensive 3D CFD-based computational model presented and documented in Part 1. Simulations for unit cells with straight channels, similar to the Ballard Mk902 hardware, are performed and analyzed in conjunction with detailed current mapping measurements and water mass distributions in the membrane-electrode assembly. The experiments were designed to display sensitivity of the cell over a range of operating parameters including current density, humidification, and coolant temperature, making the data particularly well suited for systematic validation. Based on the validation and analysis of the predictions, values of model parameters, including the electro-osmotic drag coefficient, capillary diffusion coefficient, and catalyst specific surface area are determined adjusted to fit experimental data of current density and MEA water content. The predicted net water flux out of the anode (normalized by the total water generated) increases as anode humidification water flow rate is increased, in agreement with experimental results. A modification of the constitutive equation for the capillary diffusivity of water in the porous electrodes that attempts to incorporate the experimentally observed immobile (or irreducible) saturation yields a better fit of the predicted MEA water mass with experimental data. The specific surface area parameter used in the catalyst layer model is found to be effective in tuning the simulations to predict the correct cell voltage over a range of stoichiometries.