Optimization of geometric parameters for design a high-performance ejector in the proton exchange membrane fuel cell system using artificial neural network and genetic algorithm

In this study, a CFD model is adopted for investigating the effects of the four important ejector geometry parameters: the primary nozzle exit position (NXP), the mixing tube length (Lm), the diffuser length (Ld), and the diffuser divergence angle (θ) on its performance in the PEM fuel cell system. This model is developed and calibrated by actual experimental data, and is then applied to create 141 different ejector geometries which are tested under different working conditions. It is found that the optimum NXP not only is proportional to the mixing section throat diameter, but also increases as the primary flow pressure rises. The ejector performance is very sensitive to the mixing tube length while the entrainment ratio can vary up to 27% by change in the mixing tube length. The influence of θ and Ld on the entrainment ratio is evident and there is a maximal deviation of the entrainment ratio of 14% when θ and Ld vary from 2° to 8° and 6Dm to 24Dm, respectively. To make sure the correlation of all geometric parameters on the ejector performance, the artificial neural network and genetic algorithm are applied in obtaining the best geometric.