code_saturne User's Forum

Hello,
I am struggling with LES on CS 6.0.0.
I have two inlets and one outlet and even when I choose 1m/s for each inlet (normally 30 m/s) my velocity increase strongly to 100 m/s and way upper after few iterations. I precise the geometry is a tube for one of the inlet, so it should not increase that much in this part. All the other parameters are default.
This mesh/geometry works well with URANS/RANS (with few oscillations) but it seems I can't use LES.
I think the problem comes from my mesh, which is very huge and I get from someone else, I wish I have not to re-do it because of its size.
Is there a way to solve this, without modifying the mesh? Else, is there a test case that proves that the mesh is the heart of the problem?

Thank you,
All the best
CB

Hello,

LES might be a bit more sensitive to the mesh. What type of elements du you use/what is the mesh quality ? If you use tetrahedra, there might be a few additional precautions to take, and we have seen in some cases adding 2% upwind to the velocity solution (not too much to avoid loss of precision, just a bit for stability) may help. There may be other suggestions, but I need to check with the turbulence experts.

Best regards,

Yvan

Hello,

Currently, my mesh is pretty bad I think and I have 2% of bad cell offset. It is a mix of plenty of elements, I attached the salome report on my mesh.
I tried the blending option: spots with a great increase in velocity still appear and made my simulation diverge.
I will be happy to see the feedback from your colleagues.

Best regards,
Camille B.

Hello,

In many cases, robustness is quite sensitive to the outlet condition and mesh, so if you have a tetrahedral mesh, extruding a layer or 2 at the outlet may help in many cases. Not a sure solution, but easy to try.

Best regards,

Yvan