Turbulence modeling for shock wave/turbulent boundary layer interactions

Randolph Pascal Lillard, Purdue University

Abstract

Accurate aerodynamic computational predictions are essential for the safety of space vehicles, but these computations are of limited accuracy when large pressure gradients are present in the flow. The goal of the current project is to improve the state of compressible turbulence modeling for high speed flows with shock wave / turbulent boundary layer interactions (SWTBLI). Emphasis is placed on models that can accurately predict the separated region caused by SWTBLI. These flows are classified as nonequilibrium boundary layers because of the very large and variable adverse pressure gradients caused by the shock waves. The Lag model was designed to model these nonequilibrium flows by incorporating history effects. Standard one- and two-equation models (Spalart Allmaras and SST) and the Lag model are run and compared to the new model. The focus of this work is thus to introduce a new model that builds on the success of the Lag model, but uses the Reynolds Stress Tensor (RST) as the lagged variable. This new model, the Reynolds stress tensor lag model (lagRST), is assessed against multiple wind tunnel tests and correlations as well as other models. The basis of the Lag and lagRST models is to preserve the accuracy of the standard turbulence models in equilibrium turbulence, when the Reynolds stresses are linearly related to the mean strain rates, but create a lag between mean strain rate effects and turbulence when nonequilibrium effects become important, such as in large pressure gradients. The effect this lag has on the results for SWTBLI and massively separated flows is determined. These computations are done with a modified version of the OVERFLOW code. This code solves the Reynolds Averaged Navier Stokes (RANS) equations on overset grids. It was used for this study for its ability to input very complex geometries into the flow solver, such as the Space Shuttle in the full stack configuration. The model was successfully implemented within two versions of the OVERFLOW code, v2.0aa and v2.2c. Results show a substantial improvement over the baseline models for transonic separated flows. The results are mixed for the SWTBLI flows assessed. Separation predictions are not as good as the baseline models, but the over prediction of the peak heat flux downstream of the reattachment shock that plagues many models is reduced.

Degree

Ph.D.

Advisors

Lyrintzis, Purdue University.

Subject Area

Aerospace engineering

COinS