Assessment of classical and novel turbulence models for strong shock-wave/boundary-layer interactions

Shock-wave/boundary-layer interactions (SWBLIs) have been a hot and long-standing issue in the aerospace field. In strong interactions, the boundary layer separates under the influence of adverse pressure gradients, and the interaction length is a key parameter describing this physical process. Despite significant progress in direct numerical simulation (DNS) and large eddy simulation (LES), accurately predicting the characteristic length of strong interactions remains a considerable challenge. In engineering practice, the widely used Reynolds-averaged Navier–Stokes (RANS) method relies mainly on turbulence models, whose inherent uncertainties often lead to discrepancies between simulation and experimental data. In this study, three classical turbulence models and two recently proposed models are assessed using fifteen test cases involving strong SWBLIs, covering a range from supersonic to hypersonic regimes. These cases include two classical two-dimensional SWBLIs configurations. Particular attention is given to the quantitative evaluation of interaction length and peak wall heat flux. The results indicate that the performance of the turbulence models exhibits a strong dependence on the Mach number. For instance, the widely used shear stress-transport (SST) k–ω model significantly overestimates both the interaction length and the heat flux, whereas the Spalart–Allmaras (SA) model demonstrates comparatively superior performance in cases with Mach numbers less than 8.0. These findings strongly suggest that, before conducting RANS-based SWBLIs studies, the reliability of the chosen turbulence model should be carefully validated against representative cases within the target research regime, and the model's applicability to the intended flow conditions and geometric configuration should be critically assessed.