Force Value diverge while supersonic simulation
 

I am running simulation to calculate drag of certain body at supersonic condition.

rhoCentralFoam is used, and it shows good convergence in laminar setup.

However, when I apply turbulence models such as k-w SST, force seem to diverge, increase constantly while whole simulation time domain and result in unrealistic strength.

I checked all boundary condition for newly added terms, and reduced time step to reduce oscillation. But since it is divergence, not oscillation, nothing helped.

How can I get rid of this divergence and get correct force data??