Experimental Investigation and Modelling by Machine Learning Techniques for Hydrogen Generation by Reacting Aluminium with Aqueous NaOH Solution

Abstract

Hydrogen gas is projected as the future fuel for energy generation due to high energy density, and almost zero COx and NOx emissions during combustion. Production of H2 gas by the chemical reaction of aluminium and aqueous NaOH solution appears to be a simple and cost effective technique. Accurate forecasting of the reaction kinetics based on multi-variable interactions is necessary for large scale in-situ production and control of hydrogen gas. In the present study, the kinetics of the reaction between Al and aq. NaOH to produce H2 at different temperatures and concentrations of NaOH is investigated. Data obtained from these investigations are used to predict the reaction rate and correlate the simulated and experimental results using Machine Learning Techniques, viz, Artificial Neural Network (ANN) and Multiple Linear Regression (MLR). The simulation model obtained by the MLR technique is found to be less accurate in terms coefficient of correlation than the ANN model while comparing with the experimental results. The coefficient of correlations (Rcc) and average absolute relative error % are found to be 0.998 and 1.377% respectively during simulation by ANN. During the ANN simulation, 97.8 % of data points lie within a 95 % level of confidence indicating very good simulation capability. The results of the investigation reveal the application of Machine learning technique for the accurate prediction of hydrogen gas generation by a chemical reaction.