Accelerating DSMC Simulations of Micrometeoroid Entry using a Gaussian Process Surrogate Model

Earth is impacted by tens of thousands of tons of meteoric material every year, with the bulk of these impactors being micrometeoroids, or meteoroids on the order of a gram, or smaller. Micrometeoroids are primarily studied using radar, where their properties are inferred from the signal-to-noise ratio of radar signals that are reflected off the surrounding plasma. The entry of micrometeoroids into the atmosphere is a complex physical process, which includes ablation, rarefied hypersonic flow, and plasma formation. To better understand this phenomenon, we formulate micrometeoroid entry as a fluid-structure interaction problem in rarefied flow. We develop a physics-informed Gaussian Process (GP) surrogate model to predict the velocity distribution functions (VDFs) of air near the surface of a micrometeoroid. With this approach, we can accurately calculate the surface fluxes (momentum flux and heat flux) by taking moments of the GP-generated VDFs. This model bridges the gap toward high-fidelity multiphysics simulations by allowing slow-time-step structural solvers to couple with a rapid surrogate that provides real-time boundary conditions. Ultimately, this framework enables accurate fluid-solid interface coupling without the prohibitive cost of full DSMC at every structural increment. This approach will enable the computational investigation of a wide variety of meteoric phenomena, including differential ablation and fragmentation. These also include shock wave and vapor cone development, which result from collisionality due to the presence of ablated meteoric material. By improving our understanding of these physical processes, our simulations provide valuable insight into the formation of radar-observable plasma signatures, enabling the correlation of radar cross section to plasma density and meteoroid mass, ultimately enhancing the interpretation of radar-based measurements.