Problems with an overexpanded nozzle simulation
 

Hey everyone!

I'm trying to do a validation of a rocket nozzle simulation (data from this paper), currently on the 21 NPR case, but ran into some difficulties.
The simulation does not really want to converge even with first order discretisation as continuity and energy RMS oscillates around 1e-2. There is also a "turbulent viscosity limited to..." in each iteration and a reversed flow on outlet message. The cap shock position and flowfield shape seems right, y+ on the nozzle contour is below 0.5.

The model is a ~120k unstructured mesh, the problem is axisymmetric, the turbulence model is SST, boundaries: pressure-inlet at nozzle, pressure-inlet at the front of the outer area, pressure-outlet at the rear, axis on the bottom and free-slip wall on top. I attempted both first-and second order discretisations, results are similar.

I'd really appreciate some feedback :)

Pictures say more than words:
Attachment 38581 Attachment 38582Attachment 38583Attachment 38584Attachment 38585