Wall-Modeled Large-Eddy Simulation and RANS for Workshop on Smooth-Body Separation

View Video Presentation: https://doi.org/10.2514/6.2021-1553.vid The accurate calculation of turbulent flow separation from a smooth body remains a challenging problem for Reynolds-averaged Navier-Stokes (RANS) methods. This work explores the applicability and accuracy of RANS and wall-modeled large-eddy simulation (WMLES) methods toward smooth-body flow-separation prediction. The flow conditions follow from the announcement for the upcoming Large Eddy Simulation Workshop on Smooth-Body Separation at AIAA SciTech 2022. A Mach 0.2 duct flow is subjected to a smooth backwards facing step that induces flow separation. The current work focuses on the lowest Reynolds number condition of interest (Reδ = 9600). Ongoing efforts are examining the medium (Reδ = 32000) and high (Reδ = 96000) Reynolds numbers. The performance of an algebraic equilibrium wall model and an algebraic non-equilibrium wall model (based on an adaptive algebraic law describing the logarithmic layer of the boundary layer) are explored with several levels of grid resolution. The WMLES results are compared to results of a wall-resolved k-ε RANS calculation. The results demonstrate that the flow separation and attachment locations are very sensitive to the numerical model and grid resolution. Under the conditions of interests, the RANS calculation does not exhibit flow separation; however, the WMLES calculations capture the expected flow separation. With increasing grid resolution, the WMLES simulations first predict a relatively large separation, which then decreases in size, before increasing again with fine grid resolution. Comparing to wall-resolved simulations in the workshop will assist in characterizing the accuracy of the WMLES results.