VALIDATION AND UNCERTAINTY QUANTIFICATION OF A TWO-PHASE, MULTIDIMENSIONAL PEMFC COMPUTER MODEL USING HIGH-RESOLUTION SEGMENTED CURRENT COLLECTOR DATA

We present work towards validating a multidimensional computer model capable of simulating two-phase, non-isothermal transport in PEMFCs under a wide variety of conditions. The specific hardware used to gather the experimental data is a state-of-the-art 10x10 segmented bipolar plate attached to a 50 cm(2) single cell. The five-way serpentine flow field, bipolar plates and membrane-electrode assembly are all resolved in the model. The data has been collected at Los Alamos National Laboratories under several sets of operating conditions, including different relative humidity of 25%, 50%, 75% and 100% RH and temperatures of 80 and 60 C. Current best practices for model validation are applied, including uncertainty quantification (UQ). Variability in measured data is incorporated by included uncertainty bounds on the data (using either interval bounds or statistical confidence intervals). Sensitivity analysis of model input parameters on predictions is performed using an interface to the the DAKOTA toolkit. The model is calibrated using cell polarization data with uncertainty and then used to predict the local current distribution data.