| Summary: | Proton exchange membrane (PEM) fuel cells are promising alternatives to conventional power sources mainly because of potential environmental impacts. Although PEM fuel cells have been considered for various applications, there are still certain technical challenges toward large-scale commercialization of this type of energy system. This study concentrates on analyzing and modeling the experimentally measured two-phase flow pressure drop signatures in fuel cell flow channels. PEM fuel cells produce water during operation which results in liquid-gas two-phase flow inside their flow channels. Due to small length scale of the flow channels, the two-phase flow in a PEM fuel cell is mainly dominated by capillary forces. These forces tend to hold droplets which eventually increase the pressure drop along the flow channels. This study concentrates on pressure drop analysis which is critical in realizing time-dependent changes in the current density and quantifying water accumulation. In this way, a prediction model for pressure drop signatures is presented based on Auto Associative Kernel Regression. This model has a great potential for real-time monitoring and diagnostic in PEM fuel cells. The experimental data was collected through an ex-situ test section by injecting water and supplying air at different flow rates into two parallel flow channels. Keywords: Proton exchange membrane (PEM) fuel cell, Peak analysis, Empirical model decomposition, Auto Associative Kernel Regression
|
|---|
