Quantifying Model Discrepancy in Time-Dependent, Coupled Analyses

Model discrepancy in coupled, time-dependent analyses is addressed with emphasis on aerothermoelastic response prediction for hypersonic aircraft. Time-dependent temperature data from historic hypersonic wind-tunnel experiments are used to calibrate both aerodynamic heating and heat transfer model errors and improve prediction confidence of coupled aerothermal predictions through time. Three different model discrepancy resolutions are considered (global, stepwise, and partitioned) for model error calibration, and appropriate model discrepancy forms in the presence of evolving coupled physics are also discussed. The three model discrepancy resolutions are compared using the model reliability metric, where the partitioned model discrepancy approach is observed to yield higher prediction confidence in both nonlinear and extrapolation regions.