Nozzle flow simulation in inviscid and viscous
 

Hello people! I'm using Fluent to calculate a Laval nozzle flow in 2D with air, with coupled implicit solver in a steady case. I'm trying to see the case of subsonic-supersonic with a normal shock wave. The inviscid model gives a normal shock wave inside the nozzle but when I use the Spallart-Allmaras model for viscosity (used in the Fluent tutorial for compressible flow, someone can tell me why this model?), I have some weird results. If I use a courant number of 5 (default value), I do not have any shock wave and the residuals oscillates around 1e-01. If I begin with a small Courant number (0.01) and then I increase it to 40 gradually, I have a normal shock wave, far upstream in the divergent part of the nozzle compared to the inviscid case. The residuals are under 1e-3. Moreover, there is a recirculation zone just after the shock. Can someone tell which result to believe? I am very confused... Thanks in advance for your help.