Experimental parameter uncertainty in PEM fuel cell modeling

Predictability of proton exchange membrane fuel cell (PEMFC) models has suffered from significant uncertainty in material properties with experimental data on several transport coefficients being scattered over orders of magnitude. In this study, we determine the most critical transport parameters for which a more accurate experimental characterization is required to enable reliable performance prediction. First, we incorporate a comprehensive set of material parameterizations from the literature into a recently developed macro-homogeneous two-phase membrane-electrode assembly model. This computational model demonstrates the large spread in performance prediction resulting from the scattered experimental data. Membrane transport properties induce the largest spread in the fuel cell performance curve: the diffusivity of dissolved water, the protonic conductivity and the electro-osmotic drag coefficient. Second, we conduct extensive forward uncertainty propagation analyses. These include a global sensitivity analysis in which a broad range of operating conditions and material properties is covered. By introducing the concept of condition numbers to fuel cell modeling, we measure the propagation of uncertainty through the model explicitly. We list the parameters with the highest impact on predicted fuel cell properties. These are the membrane hydration isotherm, the electro-osmotic drag coefficient, the membrane thickness and the diffusivity of dissolved water. In conclusion, the most critical model parameters are the membrane transport properties, which suffer from the largest scatter in available experimental data. This calls for a better experimental characterization of the ionomer to enhance the predictability of PEMFC models. Particularly, a more precise parametrization is required for the interplay between the different water transport mechanisms and protonic conductivity. Acknowledgements: This work was supported by the Swiss National Science Foundation (project no. 153790, grant no. 407040_153790); the Swiss Commission for Technology and Innovation (contract no. KTI.2014.0115); the Swiss Federal Office of Energy; and through the Swiss Competence Center for Energy Research (SCCER Mobility)