A Lumped Model of Water and Heat Transport in a PEMFC Fuel Cell and the Associated Methods for Estimating the Effective Transport Parameters

Abstract

The fine and quantitative modelling of the coupled phenomena of water and oxygen transport and heat and charge transfer in PEMFCs is difficult from a practical point of view because the interfaces between the different layers are heterogeneous and difficult to characterize. They depend in particular on the mechanical stress field. Moreover, the two-phase flows in the porous layers and feeding channels are by nature stochastic and very difficult to predict. This fine modelling requires a perfect knowledge of the geometry, mechanical and thermal properties, wettability, etc... of the materials, their interfaces and their heterogeneity in space and time. Most of these parameters are unknown and so numerous that it is impossible to estimate them independently of each other by simple experiments. Based on this observation, our philosophy is to develop simple models whose effective transport parameters can be estimated by experiments carried out in situ, for real operating conditions [1]. These models are lumped models that depend on few parameters which integrate the properties of the interfaces. In this presentation, we propose a one-dimensional coupled water and heat transport model (adapted to a differential cell), associated with methods for estimating the unknown water transport and heat transfer parameters. The estimation of the parameters is performed in situ, for an imposed flow field geometry and clamping force. Original experiments are developed to decorrelate the effective diffusion coefficients in the gas phase, adsorbed phase and heat transport properties. The estimation of the parameters is based on simple water balances, i.e. a measurement of the quantities of water leaving the anode and cathode. The model and the experimental set-up are presented. The inverse technique of parameter estimation is detailed. Different materials are implemented in order to observe the resulting impact on the estimated effective transport coefficients. Acknowledgements The authors gratefully acknowledge the financial support of the Lorraine Université d’Excellence program.