Assessment of alternative formulations for the specific-dissipation rate in RANS and variable-resolution turbulence models

Abstract

The current study concerns the investigation of alternative formulations for the specificdissipation rate (!) in the framework of k-! based Reynolds averaged Navier-Stokes model (RANS). Later, the study is also extended to include the partially-averaged Navier-Stokes (PANS) model which is capable of performing turbulence computations at any degree of physical resolution ranging from fully modeled (RANS) to completely resolved (direct numerical simulations, DNS) simulations. We implement and assess two alternative variables to improve the robustness and accuracy of the hybrid-unstructured DLR TAU code. The evaluation of the feasibility of these variables is based on the model performance in simulating a range of canonical as well as industrial test cases. Further, we extend the applicability of the newly implemented RANS turbulence models to massively separated flows by reformulating them to provide closure to the unresolved scales of the PANS model and assess their merits as a sub-filter stress closure model. Results comparing the PANS model computations with detached eddy simulations (DES) performed on the same grid are reported to provide a holistic analysis of the PANS methodology with a different closure model considered in this study.