Prediction of Hypersonic Ablation Using Direct Simulation Monte Carlo and Deep Learning

Hypersonic vehicles are subjected to severe aerodynamic thermal environments during atmospheric reentry, often accompanied by intense ablation phenomena. Accurate simulation of the ablation process on the vehicle’s surface is a challenging problem in hypersonic aerodynamics. The ablation process involves coupled factors such as rarefied gas effects, high-temperature thermochemical nonequilibrium effects, material thermal response, and ablation recession, which are significantly increasing the complexity of the problem. Based on the improved open-source kernel SPARTA, it has been preliminarily demonstrated that the Direct Simulation Monte Carlo (DSMC) method is feasible for modeling the ablation of some typical aerodynamic shapes. This study will further combine the energy balance equation of the ablation surface with material properties, and establish a general coupling ablation model suitable for the DSMC method. For the hypersonic reentry of a blunt body, the aerodynamic heating flux on the surface was calculated, and the image of the gas–solid boundary receding over time was obtained using the ablation model. Additionally, deep learning methods were employed to model the dynamic changes in the flowfield and shape characteristics during the ablation process. The results show that the ablation model based on the DSMC method proposed in this study has the potential to achieve real-time dynamic simulation of the aerodynamic thermal environment and vehicle’s shape changes during the ablation stage of reentry. Besides, the deep learning model can significantly improve the efficiency of ablation prediction.