Which turbulence model is suitable for simulating high temperature(1000K) supersonic flow? I am using K-epsilon now. But it is underpredicting the axial velocity and mach number.

Morshed

Which K-epsilon model are you employing ? If it is some recent model and its not performing well, try the K-omega family of models.

Also, not predicting the axial velocity can be related to a diffusive discretization scheme. If you're velocities are 'slowing' down more quickly than you would like along the axial direction, perhaps change the discretization from 2nd order to a compressive scheme (i.e. upwinding - which has its own diffusion problems).

Thanks to both of you Harish and JK