Gas Crossover Predictive Modelling Using Artificial Neural Networks Based on Original Dataset Through Aspen Custom Modeler for Proton Exchange Membrane Electrolyte System

Proton exchange membrane electrolyzer cell (PEMEC) will play a central role in future power-to-H2 plants. Current research focuses on the materials and operation parameters. Setting up experiments to explore operational accident scenarios about safety feasibility is not always practical. This paper focuses on building mathematical and prediction models of hydrogen and oxygen mixing scenarios of PEMEC. A mathematical model of the PEMEC device was customized in the Aspen Custom Model (ACM) software and integrated various critical Physico-chemical phenomena as the original data set for the prediction model. The results of the mathematical simulation verified the experimental results. The prediction model proposes an artificial neural network (ANN) framework to predict component distribution in the gas stream to prevent hydrogen-oxygen explosion scenarios. The presented approach by training ANN to 1000 sets of hydrogen-oxygen mixing simulation data from ACM is applicable to bypass tedious and non-smooth systems of equations for PEMEC.