Channel geometry optimization of a polymer electrolyte membrane fuel cell using genetic algorithm

The study presents the use of genetic algorithm (GA) to optimize the bipolar plate channel geometry of a polymer electrolyte membrane fuel cell (PEMFC). Contrary to previous optimization techniques, laborious fuel cell design steps, including boundary setting, mesh generation, and numerical computation at every design parameter variation step, are avoided by developing an automated program via Matlab and Comsol Multiphysics software. GA with Matlab automatically checks the optimality of the present fuel cell design with a performance index obtained from Comsol Multiphysics. If a global optimal is not reached, the new geometry set of a fuel cell is generated according to GA rules toward better fuel cell performance. The new set is then fed back to Comsol Multiphysics to have the new performance index calculated through updated boundary conditions, element and mesh generations, and numerical analysis. This automated optimization technique not only saves numerous calculations, but also obtains the global optimal result of a given fuel cell geometry. Therefore, it provides a fast and efficient optimization process and renders optimal results. In this study, two channel and rib geometry arrangements are studied: one with a symmetric anode and cathode channel arrangement, wherein channels and ribs face each other; and another with an asymmetric arrangement, wherein a channel faces a rib and vice versa. First, the two-dimensional (2D) CFD model is used to obtain the optimal result in order to speed up the optimization calculation, slightly sacrificing the model accuracy. Afterwards, the three-dimensional CFD model is utilized and experimental verification is made with the same geometries to support the validation of the 2D optimization result.