The Effect of Turbulence Model Uncertainty on Scramjet Isolator Flow Field Analysis

The goal of this work was to perform an uncertainty and sensitivity analysis of commonly used turbulence models in Reynolds-averaged Navier-Stokes flow solvers due to epistemic uncertainty arising from ambiguities in the model coefficients. The uncertainty analysis was applied to simulations of a model scramjet isolator. The Menter-BSL, Menter-SST, Spalart-Allmaras, and Wilcox-2006 k − ω turbulence models were examined and simulations were carried out using the VULCAN flow solver, developed and maintained at the NASA Langley Research Center. Non-intrusive polynomial chaos theory was used for efficient uncertainty propagation and Sobol indices were employed to establish relative sensitivities of the flow solution to closure coefficients. The results obtained were compared to experimental data as well as to previous work focusing on different flow problems. Sets of closure coefficients that contribute most to solution uncertainty for each turbulence model were identified, which warrant further investigation as more knowledge about the effects of these coefficients is expected to reduce the uncertainty in the numerical design of scramjet isolators.