Development and validation of a reduced-order control-oriented model for hydrogen proton-exchange membrane fuel cells

Hydrogen proton-exchange membrane fuel cells are one of the technologies that can contribute to achieve the decarbonization targets set in the transport sector. However, such systems suffer from long-term degradation when subjected to highly dynamic operation, especially if low oxygen concentration conditions are reached internally in the cathode stream. Model-based control strategies can help to detect in advance and prevent the appearance of local low oxygen concentration conditions, enlarging the fuel cell stack durability. The purpose of the current research is to develop and validate a reduced-order control-oriented model for proton exchange membrane fuel cells. The model is structured to provide detailed one-dimensional resolution in the axial direction of the advective flow (along the gas feeding channels), so that local composition and electrochemical performance can be evaluated. Instead, mean values are considered in the transversal direction for each characteristic layer of the membrane electrode assembly (bipolar plates, gas diffusion layers, catalyst layers and membrane) to reduce computational cost. Moreover, the main novelty from this model is the development of explicit solutions for mass and species conservation equations, enabling faster than real time computation that is required for model-based control purposes. Once defined, the model is calibrated and validated against a set of data extracted from the literature. The results show that the model reproduces the polarization curves at different working conditions, and also resolves the local inhomogeneities along the cell in active species concentration and electrochemical losses that appear under low cathode stoichiometry. As these local phenomena drive both degradation and efficiency, the physics-based, reduced-order formulation with explicit local-property solution proposed avoids the cause-conflation issues associated to 0D models commonly used in control platforms. • PEMFC model with 1D resolution in gas channel and lumped layers is proposed. • Model validated with polarization curves for varying gas mixtures and stoichiometry. • Grid independence reached with 10 control volumes: 2.4× faster than real time. • Low stoichiometry causes local variations of current density and O2 concentration. • Cathode oxygen depletion drives cell performance due to higher activation losses.