Non-Intrusive Reduced Order Modeling of Aerothermal Field Data Including Ablation Shape Change

A methodology for constructing non-intrusive reduced order models of aerothermal field data for ablating reentry vehicles is presented. Parameterization of deformations using modal decomposition is demonstrated, and resulting modal amplitudes are used as field surrogate inputs. Results are presented demonstrating that including displacement modes in the aerothermal surrogate results in lower test set prediction errors for varying numbers of training trajectories. A generative sampling procedure to potentially reduce the required number of coupled training trajectories is also presented.