Simulation of High-Speed Cold Wall Flow on a Flat Plate Including Application of the Wray-Agarwal Turbulence Model

The accurate and efficient simulation of high-speed cold wall external aerodynamic flow is necessary for modern hypersonic vehicle design. The effect of wall cooling in the near-wall region of turbulent boundary layers heavily impacts the overall aerodynamic heating on the surface, thus impacting the vehicle design with the need for sufficient and optimal thermal protection. The development of accurate RANS one-equation turbulence models is appealing for this type of flow due to the computational efficiency compared to costly experiments. The Spalart-Allmaras (SA) turbulence model is a widely applicable model for external flows but has shown to have some inaccuracies for complex flow regimes including compressible high-speed flows. The original SA model is investigated here for the selected high-speed cold wall zero-pressure-gradient flat plate cases, along with the SA-Catris model which employs compressibility corrections developed by Catris and Aupoix. Additionally, another recently developed one-equation turbulence model is investigated, the Wray-Agarwal (WA) model. This model has shown promising results for subsonic and supersonic flow cases and the extension of this model to the hypersonic cold wall flows is evaluated. ANSYS Fluent and SU2 codes are used for computations using the SA model, the results using the WA are computed only from ANSYS Fluent, and the results using the SA-Catris are obtained using SU2. Comparisons are made to DNS data from Duan et al. as well as the computational results obtained using the SA-Catris model in US3D by Candler et al. SA results computed in SU2 showed better accuracy when compared to the US3D SA-Catris data, and overall, the SU2 results with both the SA and the SA-Catris turbulence models were found to be most accurate when compared to the DNS data. The results using the WA model compared well with those from the other turbulence models for the coupling of the thermal and velocity fields but failed to capture the boundary layer velocity profile. As the wall-to-recovery temperature ratio is reduced (i.e. cold wall flow), each model failed to capture the peak temperature point when compared to the DNS data, as well as underpredicted the buffer region of the velocity profiles.