Reynolds stress on SWBLI in isolator using different turbulent models

Shock wave/boundary layer interactions (SWBLI) is an unavoidable problem in supersonic and hypersonic flow. The inherent defects caused by Reynolds-averaged method make the RANS (Reynolds-averaged Navier-Stokes) turbulent models unable to reveal the impact of small-scale recirculation zone on shock wave. Meanwhile, the LES (Large Eddy Simulation) turbulent model is still unaffordable for engineering since a huge amount of calculation. Therefore, the DES (Detached Eddy Simulation) turbulent model is a more economical choice which is a hybrid model combing the RANS in main flow and the LES in the boundary layer. In this paper, the study of Reynolds stress on SWBLI in isolator is conducted using four different commonly used RANS turbulent models (including Standard k - ε model, RNG k - ε model, standard k - ω model and SST model). The results could be used as a guide to choosing the RANS in core flow of the DES model. It was found that, in the boundary layer, the Reynolds Stress u″u″ is usually under predicted and the Reynolds Stress u″v″ is over predicted. The maximum difference was more than 50%. Averaged velocity ṽ was far more over predicted, and the maximum difference was about 500%. In main flow area, the SST model and the Standard k - ε model were more accurate, and its results could basically reflect the changes of averaged velocity and Reynolds stress. Analysis suggests that Reynolds stress difference in u″v″ caused its difference in averaged velocity, especially velocity ṽ in the y direction. This is because the over prediction of u″v″ enlarges the recirculation zone. Comparing the four turbulent models simulation results, it was suggested that, in main flow, the SST model had enough accuracy in Reynolds averaged velocity, Reynolds stress and the shock wave intermittent resolution. It is most suitable main flow area turbulent model in the DES model.