max residuals near max turbulence intermittency
 

Hello everyone,

I am simulating the flow in a diffuser with SST transitional model in transient mode. The boundary layer parameters are y+<1 and expansion 1.1. The timestep is 0.0001s. The inlet velocity is about 20m/s. The flow is with both laminar and turbulence separation.

I monitor the pressure, turbulence energy and velocity in three points of freestream, laminar and turbulence separation region. The attachment is some results. We can observe developing oscillation in these pictures, which I think non-physical. http://kissengerxu.spaces.live.com/photos/cns!5B651E09CD31334B!131/

Then I adopt double precision solver, decrease and increase the timestep, however the problem remains. Fortunately when I switch second order backward euler to first order, the developing oscillation begins to be weaker.

I also check the residual distribution(P-mass, u-mom...) and find that maximum residuals are usually located near the maximum turbulence intermittency.

How can I solve the problem besides by switching to first order backward euler?

Thank you!

Xiaofei