Bridging the gap between prediction and real-time diagnosis of water failures in proton exchange membrane fuel cell stacks via gas distribution characterization

Abstract

Water failures and consequent relatively low voltages of certain cells in high-power fuel cell stacks severely threaten the operating stability and durability, which passively depend on real-time diagnosis with the stack regarded as a black box due to the poor knowledge in the triggers and the interactions among cells. This paper identifies a strong correlation between water-failure risks and gas maldistribution among cells, enabling risk assessment, failure prediction, and pre-operation optimization. By means of in-situ characterization, the condition parameters of gas pressure and operating temperature are proven to have minimal impact on gas distribution among cells, while an increase in inlet gas humidity induces gas redistribution attributed to uneven water accumulation. The synchronously-identified average mass transfer coefficients of the cathode change in a predictable manner with varying condition parameters. Meanwhile, under near-flooding conditions, cell voltage fluctuations, fuzzy indicators of cathode flooding, increase evidently as the distributed gas flow rate decreases. In an elaborate step-current experiment, gas flow interactions between adjacent cells are observed during water accumulation and flooding interior certain cells. Thus, the complex flooding behavior of specific cells in large stacks can be predicted through gas distribution characterization. This theory is applied in practice to a 100-cell stack, where 8 high-risk cells are accurately identified.