Structural optimization and performance study of a biomimetic spiral baffle flow channel for PEMFC based on a BO-RBF surrogate model

Abstract

The bipolar plate is a key component of a proton exchange membrane fuel cell (PEMFC), and the flow channel structure on its surface directly affects the cell performance. Inspired by the spiral valve structure in the intestine of the Greenland shark, a biomimetic spiral baffle flow channel (BSBFC) is proposed in this study to achieve efficient mass transfer under low flow resistance. To perform the structural optimization and performance analysis of the BSBFC, a numerical model for PEMFC was first established via COMSOL. To reduce the computational cost of optimization, a surrogate model for predicting PEMFC net power density was developed based on a BO-RBF neural network. On this basis, Sobol sensitivity analysis was employed to quantify the influence of structural parameters on net power density, and the response surface method was used to analyze the interactions among parameters. Finally, a genetic algorithm (GA) was introduced for global optimization to obtain the optimal biomimetic spiral baffle flow channel (OBSBFC), which was comprehensively compared with a single straight flow channel (SSFC) and rectangular baffle flow channel (RBFC). The results show that, compared with the SSFC and RBFC, the OBSBFC increases the current density by 6.16 % and 3.05 %, and raises the net power density by 5.86 % and 5.32 %, respectively, and reduces the average liquid water saturation by 6.03 % and 4.38 %.