A New 0D Physics-Based Model for Proton-Exchange Membrane Fuel Cell Performance Simulation

This paper presents a new physics-based 0D steady-state model for simulating the performance of proton-exchange membrane fuel cells (PEMFC), including phenomena such as temperature gradients, transport of membrane water, liquid water and gases. The model adopts an explicit formulation resulting in extremely low computational times, enabling sensitivity analyses, parameter estimation and cross-validation, to an extent which has not been previously reported for PEMFC models. Validation against polarization curves, high-frequency resistance, crossover current density, and roughness factor measurements for fifteen operating conditions demonstrates very good predictive accuracy. With 18 estimated parameters, the model achieves median root-mean square errors of 14.1 mV and 3.6 mΩ cm 2 for cell voltage and high-frequency resistances on testing datasets. However, overfitting becomes evident when the number of estimated parameters is increased. Beyond performance prediction, this work underscores the importance of robust parameter estimation and cross-validation in PEMFC modelling. It also highlights the potential of lightweight 0D physics-based models for accurate performance prediction and interpretation of experimental data.