Coupling the Mass Transport Mechanisms of the Porous Transport Layer and Flow Channel of a Proton Exchange Membrane Fuel Cell

Abstract

A coupled model was created that linked the mass transport mechanisms of the channel and the Porous Transport Layer of a Proton Exchange Membrane Fuel Cell. A Volume of Fluid solver in the open-source Computational Fluid Dynamics software, openFOAM, was used to model the two-phase flow in the channel. A Pore Network Model designed for Proton Exchange Membrane Fuel Cells and found in literature was implemented into openFOAM to model the two-phase flow in the Porous Transport Layer. The Pore Network Model was recast into a form suitable for the Finite Volume Method used in openFOAM allowing for coupling between these two components. A coupling strategy called Dirichlet-Neumann partitioning was used for pressure on the boundary that separates the two domains. A Neumann boundary condition for pressure was applied in the channel and a Dirichlet boundary condition for pressure was applied in the Pore Network Model. The value of pressure was passed from the channel to the Pore Network iteratively until convergence was achieved. It was determined that the air pressure in the Porous Transport Layer remains constant before breakthrough therefore allowing for the Pore Network Model only to be solved before breakthrough using a linear pressure distribution along the channel boundary to capture the convective effects of the channel. The effects of convection on water transport in the Porous Transport Layer was investigated and the convective effects were significant enough to cause mass conservation violations when the air inlet velocity was altered from 5-10 m/s. It was determined that the domain widths often used in literature are too small to conserve mass and near wall effects permeate towards the domains center. Two common Porous Transport Layer materials were recreated and it was determined that within the expected range of pressure gradients significant movement of water downstream was observed for the SGL10BA material with less significant movement for the Toray090 sample. It was determined that for post-breakthrough simulations water is needed in the channel to properly set the injection velocity. The numerical stability of these post-breakthrough simulations proved to be extremely fragile.