Partially averaged Navier-Stokes (PANS) model was adopted to numerically investigate the supersonic ramped-cavity flow. The objective of current study is to assess the capability of PANS method in the prediction of unsteady supersonic turbulent flow. Comparison of PANS

Reynolds averaged Navier-Stokes (RANS) and detached eddy simulation (DES) results with the available experimental data shows that: (1) velocity profiles

wall-pressure and skin-friction coefficient distributions predicted by PANS model with variable unresolved-to-total ratio of kinetic energy are consistent with DES results; and (2) both of PANS results with variable unresolved-to-total ratio of kinetic energy and DES results are in good agreement with the experimental data and superior to those predicted by RANS; and (3) unsteady Raynolds averaged Navier-Stokes (URANS) model predicts two dimensional steady geometries of free shear layer

while DES and PANS resolve more abundant flow structures; and (4) PANS model with constant unresolved-to-total ratio of kinetic energy cannot return to RANS model in the near-wall region despite it can resolve more scales of motion compared with RANS; and the predicted velocity profiles of upstream turbulent boundary layer mismatch the log law

thus downstream flow simulation deviates a lot from the experiment.