Incorporation of the Stefan-Maxwell Multicomponent Diffusion Model into a Pore Network Model of the PEMFC Electrode

Sunday, 5 October 2014: 13:40
Sunrise, 2nd Floor, Galactic Ballroom 8 (Moon Palace Resort)
M. Aghighi (McGill University) and J. T. Gostick (Mcgill University)
Polymer electrolyte membrane fuel cells are one of the best candidates to replace internal combustion engines. The key requirement for commercial success of PEMFCs is to demonstrate optimal performance at high current density. However, because of liquid water generation, the power density is reduced by mass transport limitations at the cathode. Accordingly, precise modeling of mass transfer inside the porous structure of fuel cell electrodes is crucial.

    Ordinary diffusion is the most commonly considered diffusion model for PEMFC in the literature. In most of these studies, the gas transport is considered as a binary system of oxygen diffusing through nitrogen (and sometimes water vapor). However, realistic simulation of fuel cell operation requires simultaneous modeling of both oxygen and water vapor transport in a stagnant film of nitrogen. In multicomponent diffusion, the fluxes of all of the species are important to consider since they might affect the diffusive transport of the other species.  

    A pore network modeling has been developed using the Stefan-Maxwell approach, to simulate the diffusion of gases mixtures inside fuel cell electrodes. Pore network models (PNM) provide an alternative approach for the continuum modeling in porous media. Rather than using finite element models of transport in the pore space, PNMs use pore-to-pore nodal balances to model species transport. They also enable the structural properties of the porous material to be incorporated directly into the model, rather than through constitutive relationships. 

    The SM model is notoriously difficult to solve numerically for pore networks, but some simplifications and solution schemes have been proposed in the literature [1,2] based on decomposing Jacobian matrix of the equations, which are less computationally expensive. In this work, these methods have been evaluated to determine an appropriate solution algorithm for the pore network. These advantages come at the expense of rigorous transport phenomena calculations since some simplification to the SM model are made. 


1.          Wood, J., L. Gladden, and F. Keil, Chemical Engineering Science, 2002. 57(15): p. 3047-3059.

2.          Rieckmann, C. and F.J. Keil, 1997. 36(8): p. 3275-3281.


This work was funded by AFCC and the NSERC CRD program.