Assessment and Extension of Model Error Regression Methodology in Aerodynamics Applications

Quantifying the error and uncertainty in computational physics simulations plays an important role in assessing the credibility of these simulations. While standard approaches exist to compare simulations with experimental data for validation, the utility of these comparisons is typically limited to the subset of the parameter space for which experimental data are available. However, computational physics models are often applied in regions of the parameter space in which experimental data do not exist. In these regions, estimates of error and uncertainty for computational models are still important. In this paper, we apply, analyze, and extend the methodology proposed by the American Society of Mechanical Engineers VVUQ 20 subcommittee for two aerodynamics applications. These applications include a subsonic airfoil on which aerodynamic coefficients are predicted and a hypersonic sharp cone on which wall heat flux is predicted. The results support the usefulness of the methodology and suggest that the proposed extensions are helpful and even necessary in some contexts. Recommendations for the application of the proposed extensions based on the two applications of the present study are given. Thus, this study provides a thorough evaluation of the base methodology as well as useful extensions and recommendations for its application.