Proper-Orthogonal-Decomposition-Based Leveraging of Reentry Vehicle's Surface Environment

In the early stage of a reentry vehicle design are often necessary tools able to perform mission and trajectory trade studies. Many works in the literature present tools able to interpolate from a numerical database of high-fidelity simulations to a target free-stream condition. In this context, the work explores the capability of a reduced-order model in extending a limited database of computational fluid dynamics simulations to the full coverage of the design space. This results in a fast physic-based tool able to generate load history experienced on a vehicle's surface during the reentry flight. The reduced-order model is based on proper orthogonal decomposition coupled with a Gaussian process for interpolations. Main results are the history of the pressure and skin-friction coefficient of a reference trajectory related to a specific vehicle's control points. This output will be compared with a simple Gaussian metamodel based directly on the computational fluid dynamics data of such control points. A detailed cross-validation analysis of the model that provides a loss function map in the design space can be considered as a guide to in-fill the database with further computational fluid dynamics simulations, keeping the number of computational fluid dynamics runs at a minimum value to limit the computational budget.