Dynamic model of oxygen starved proton exchange membrane fuel-cell using hybrid analytical-numerical method

One of the primary life-limiting factors in PEM fuel-cells arises from performance degradation resulting from transfer (crossover) leaks. Transfer leaks result in oxygen starvation and models of fuel cells under oxygen starved conditions would allow for detection of fault inception. This paper develops a unified fuel-cell model for when the fuel-cells can either deliver power (termed driving-mode, and for when the cell absorb power (termed driven-mode) for higher leak rates. The model captures the gradient of the reactants both in the GDL and in the flow channel in addition to capturing the various electro-chemical effects. The response of the model under normal conditions is first validated for normal operation against previously published experiments. The response of the model under oxygen-starved conditions is then validated against simulated leaks in three different cell architectures: a Ballard 9-cell Mk1100 stack where hydrogen is injected into one cell, and a Ballard 10-cell Mk902 stack and 20-cell Mk903 stack where hydrogen is injected into the upstream cathode flow. Finally, the response of the model is also validated against an actual leaky Mk902 cell. The model generally agrees well with the measured cell voltage data for all the above experiments.