Kinetic Monte Carlo modeling of the interface heterogeneous catalysis effect based on machine learning

Abstract

<p indent="0mm">Heterogeneous catalysis in a high-enthalpy environment has a remarkable influence on the aerodynamic heating of hypersonic aircraft, and modeling this multiscale effect is an important prerequisite for the lightweight and low-redundancy design of an aircraft thermal protection system. A mesoscale interface heterogeneous catalysis model applicable to macroscale flow field calculation was established by the Monte Carlo method. By the radial basis function neural network algorithm, a machine learning-based reaction kinetics analytical surrogate model was constructed, by which fast and accurate prediction of the catalytic recombination coefficient was realized. The results showed that the machine learning model predicted a recombination coefficient similar to the kinetic Monte Carlo method while reducing the magnitude of the calculation time to that of the traditional macroscale method. Coupled with the macroscale CFD solver, the proposed model can efficiently predict surface heat flux with a multiscale heterogeneous catalysis effect. Modeling of microscale results by the machine learning method and its application in macroscale simulation not only covers the heterogeneous catalytic kinetics but also improves computational efficiency. This modeling provides a theoretical basis for multiscale modeling of complicated high-temperature effects on hypersonic aircraft surfaces.