Implied models approach for turbulence model form physics-based uncertainty quantification

Abstract

A new physics-based uncertainty quantification methodology is developed for understanding model form uncertainty. This framework provides insights into the reasons and conditions under which model assumptions are wrong and is used to study model form uncertainty in two-equation Reynolds-Averaged Navier-Stokes turbulence models in wall-bounded flows. Through this analysis, the shortcomings of these models are understood with an emphasis on the role of anisotropy. Error cancellation is shown to largely benefit two-equation turbulence models in wall-bounded flows, leading to models that provide correct predictions but lack physical correctness.