Hello CFD users !

I have a problem for modelling the boundary layer of a 3D wing. My wing is in a wind tunnel. I did many simulations using a structured mesh. They all converge. But now, I decided to reduce the wind tunnel length and to refine the mesh near the wall. The convergence curves look good until the 20th iteration where they start to become unstable. So, my simulation doesn't converge anymore. Is this due to the refinement of the mesh near the wall or to the shortening of the wind tunnel ? I use a SST turbulence model. My inlet is at 15m/s, the chord of the wing is 6in and the length of the shortened wind tunnel is about 10 times the chord. Has anyone any idea ?

Thanks !

Gilles

change the scheme of solver! use upwind!

Hi,

Rogerio - Upwinding? That sounds like a bad idea to me. Try these suggestions:

http://www.cfd-online.com/Wiki/Ansys...gence_criteria

Glenn Horrocks

Yes, it happenned to me! When i´m studying Tutorial 8> Supersonic Flow Over a Wing...

Thanks for the reply. I did a simulation using upwind and it seems converging. But with that kind of scheme, there could be numerical diffusion and I have to take it into acount. An another solution (that I didn't try yet) is given by Glenn's link. I used, for the simulation the Automatic TimeScale option. I will try to use a physical timescale that is equal to about 0.9 times the time of residence.