Understanding Flooding Phenomena in Mini-Channel of Proton Exchane Membrane Fuel Cells

International audienceProper management of the water produced in the Proton Exchange Membrane Fuel Cell (PEMFC) is necessary to improve the system performance and lifetime. Indeed, the accumulation of liquid water in the GDL (Gas Diffusion Layer) or in the gas supply channels can reduce the supply of reactants to the active layer. The development of PEMFCs with more compact cells, with thinner channels and reaching higher current densities will make this system even more sensitive to flooding. On the contrary, the absence of humidification in a cell will cause a decrease in the proton conductivity of the ionomers and a degradation of the materials.In order to improve the predictability of the flooding of PEMFCs, and to better understand the liquid water transport phenomena in the cell, a channel-rib pore network model is under development. It relies on previous work done at CEA-LITEN and CNRS/IMFT[1], [2]. This model has been validated against experimental data, as shown in Figure 1. It will eventually allow predicting the behaviour of the system locally as a function of the local operating conditions and local properties of the gas diffusion layer. This model, once coupled to a second model developed at the scale of a cell, which accounts for heterogeneities between inlet and outlet of the cell [3], will allow a more efficient prediction of water distribution in the cell and especially the occurrence of flooding. This coupling is necessary since even though the cell model has already shown good agreement with experimental data for some operating conditions as seen in Figure 2, it still needs more accuracy in other cases, for example at higher current density (Figure 3).A first step towards the objective of a coupled model is to obtain for given operating conditions effective transport parameters at the rib-channel scale, which can be used as inputs to the cell model. A simple transport model to represent the transport of species in the GDL is Fick's law, which allows to consider a binary mixture (vapour and oxygen in our case) whose species would only be subject to diffusion. However, there is a pressure gradient across the GDL, imposed by the difference between the flow of O2 consumed and the flow of water vapour produced. Of course, this gradient can be assumed small outside of flooding conditions. However, in the case where the GDL is highly saturated with water, the pressure gradient within the GDL can increase. The convective fluxes due to the pressure gradient will then be higher. Therefore, the gas transport within the GDL will be studied mainly by two models: a binary diffusion model based on Fick's law, and a diffusion/convection model based on Young and Todd's model[4]. They will be compared to define if a simple and low computational time consuming model can be sufficient for our study or not