Kinetic modeling of oxidative dehydrogenation of propane (ODHP) over a vanadium–graphene catalyst: Application of the DOE and ANN methodologies

Abstract

In this research the application of design of experiment (DOE) coupled with the artificial neural networks (ANN) in kinetic study of oxidative dehydrogenation of propane (ODHP) over a vanadium–graphene catalyst at 400–500°C and a method of data collection/fitting for the experiments were presented. The proposed reaction network composed of consecutive and simultaneous reactions with kinetics expressed by simple power law equations involving a total of 20 unknown parameters (10 reaction orders and 5 rate constants each expressed in terms of a pre-exponential factors and activation energies) determined through non-linear regression analysis. Because of the complex nature of the system, neural networks were employed as an efficient and accurate tool to model the behavior of the system. Response surface methodology (RSM) and ANN methods were constructed based upon the DOE's points and were then utilized for generating extra-simulated data. The three data sets including the original experimental data, those simulated by the ANN and RSM methods were subsequently used to fit power law kinetic rate expressions for the main ODHP and side reactions. The results of kinetic modeling with simulated data sets from the ANN and RSM models compared with collected experimental data. Both methods were able to satisfactorily fit the experimental data for which the ANN data set showed the best fitting amongst them all.