A comparative analysis of predictive models for estimating the formability of stamped titanium bipolar plates for proton exchange membrane fuel cells

The microchannels on the bipolar plates (BPPs) of a proton exchange membrane fuel cell (PEMFC) have various functions. Increasing the microchannel depth leads to an increase in the efficiency of the fuel cell. In this paper, metallic BPPs of the PEMFC made of commercial pure titanium (CP-Ti) with an initial thickness of 0.1 mm are manufactured using the stamping process. First, after performing the process using experimental tests and finite elements (FE) simulation, the accuracy of the numerical results is confirmed. In the following, using the FE model and considering process parameters including die clearance, forming speed, and sheet/die friction coefficient, a set of experiments is implemented using the response surface method (RSM). The results of the filling rate are obtained from the experiments and a regression model is established to predict the filling rate of the microchannel. Then, four meta-heuristics methods including artificial bee colony optimization (ABCO), genetic algorithm (GA), artificial neural network (ANN), and adaptive neuro-fuzzy inference system (ANFIS) are used to estimate the formability of the CP-Ti BPPs. According to the results, the ANFIS model with an error of 0.0062 × 10−3 % compared to experiments is more reliable to predict the filling rate in the stamping of the CP-Ti BPPs.