Evaluation of rans closure models with high-fidelity les datasets for hypersonic shock boundary layer interactions

Reynolds-averaged Navier-Stokes (RANS) and hybrid Large-Eddy Simulation / Reynolds-averaged Navier-Stokes (LES/RANS) simulations of a Mach 8.2 impinging shock / turbulent boundary layer interaction are reported in this work. Both the Menter SST RANS model and the LES/RANS strategy yield accurate predictions of the extent of the interaction and the associated wall pressure and heat transfer distributions. Results from the scale-resolving techniques are ‘data mined’ to extract quantities relevant to RANS-level modeling, including eddy viscosities characteristic of momentum and heat transfer, turbulent Prandtl number distributions, and Reynolds shear-stress structure factors. The results echo those of our earlier study of a Mach 7 cylinder-flare interaction (SciTech 2020) in that substantial variations in the structure factor and the inverse turbulent Prandtl number occur within the region of shock interaction – RANS models generally assume that both of these quantities are constants. An alternative algebraic form for the turbulent heat flux due to Bowersox (2009) is also evaluated and is shown to provide modest improvements in the prediction of enthalpy variance production. In an attempt to model the structure factor better, an equation learning neural net strategy is implemented and tested.