Finite element simulation and ANFIS prediction of dimensional error effect on distribution of BPP/GDL contact pressure in a PEM fuel cell

Distribution of contact pressure between the bipolar plate and gas diffusion layer considerably affect the performance of proton exchange membrane fuel cell. In this regard, an adaptive neuro-fuzzy inference system (ANFIS) is developed to predict the contact pressure distribution on the gas diffusion layer due to dimensional errors of the bipolar plate ribs in a proton exchange membrane fuel cell. Firstly, the main data set of input/ output vectors for training and testing of the ANFIS is prepared based on a  nite element simulation of the contact between the bipolar plate and gas diffusion layer. An experimental procedure is used to validate the simulation results. Then, the ANFIS is developed and validated using the randomly selected data series for network testing. The applied ANFIS model has ten inputs made up of the dimensional errors of the bipolar plate ribs (e1 … e10). The standard deviation of contact pressure distribution (Pstd) on the gas diffusion layer is the unique output of the ANFIS model. To select the best ANFIS model, the average errors of various architectures two different data series of training and testing of the main data set are calculated. Results indicated that the developed ANFIS has an acceptable performance in predicting the contact pressure distribution for the cited fuel cell model. The proposed integrated prediction model is feasible and effective for the dimensional tolerances considered. This method can reduce computing time and cost considering the acceptable accuracy of the obtained results, and can be used to analyze the effects of dimensional errors of bipolar plate on the performance of proton exchange membrane fuel cell.