Intake design for atmosphere breathing electric propulsion: the role of gas surface interaction with DSMC simulations and potential machine learning applications

Abstract This study investigates the performance of an Atmosphere Breathing Electric Propulsion intake in Very Low Earth Orbit. In the first part, two gas surface interaction models, the Maxwell model and the Cercignani-Lampis-Lord model, are compared to assess their impact on intake performance, accounting for more or less realistic accommodation coefficients and surface interactions. In the second part, the high-fidelity Direct Simulation Monte Carlo results from the Maxwell model are used as training dataset for the Gaussian Process Regression algorithm. Thus, a surrogate model is developed for predicting intake collection efficiency and compression ratios across varying altitudes (160–220 km), conic slope angles ( $$12^{\circ}$$ to $$20^{\circ}$$ ), and gas surface interaction conditions, which indirectly accounts for the effect of surface contamination by atomic oxygen. This surrogate model aims at enabling rapid evaluation of intake performance while capturing trends due to altitude, geometry, and surface interactions. This work represents an initial framework that can be extended in future studies to include additional geometric and operational parameters, supporting accelerated atmosphere breathing electric propulsion intake design and mission analysis.