Fast predictions of cold start performances of fuel cells using data-driven assisted models

Abstract

The existence of supercooled water in the porous electrode of Proton exchange membrane fuel cell (PEMFC) leads to the difficulty in the modelling. In this work, a cold start model of PEMFCs considering the existence of supercooled water is established, which is used to examine how the supercooled water affects the cold start performances. We find that due to the transport of supercooled water in the porous electrode, the ice can be formed in the entire porous electrode, which is different from the model not considering the supercooled water where the ice can only store in the catalyst layer. Accordingly, when the cold start is failed, the ice blockage positions are highly dependent on the structure of porous electrodes. We then propose a theoretical formula for the fast predictions of cold start time of PEMFCs, with the aid of data-driven models. Using this data-driven assisted model, we elucidate how the current density and structure of porous electrodes affect the cold start performance. The insights provided in this work help to optimize the cold start performance of PEMFCs.