Hello,
I already sent them a message a few days ago, bu I guess I'll need to ask again.
Best regards,
Yvan
Artefacts
Forum rules
Please read the forum usage recommendations before posting.
Please read the forum usage recommendations before posting.
-
Yvan Fournier
- Posts: 4168
- Joined: Mon Feb 20, 2012 3:25 pm
Re: Artefacts
Hello,
I already sent them a message a few days ago, but I guess I'll need to ask again.
Best regards,
Yvan
I already sent them a message a few days ago, but I guess I'll need to ask again.
Best regards,
Yvan
Re: Artefacts
Hi Yvan,
A gentle reminder for contacting the guys who resolve the artifacts issue of the LES simulation by using some post-processing filters.
A gentle reminder for contacting the guys who resolve the artifacts issue of the LES simulation by using some post-processing filters.
-
Yvan Fournier
- Posts: 4168
- Joined: Mon Feb 20, 2012 3:25 pm
Re: Artefacts
Hello,
I had a brief discussion with my colleagues, but it seems the solution they had we purely for postprocessing (FFT for a HIT test case).
So there is no directly available solution here it seems. We we guess may happen is that the numerical diffusivity is higher in the tetrahedral regions than in the hexahedral regions, causing your artifacts. A workaround for this would be to add explicit filters, but the LES models in code_saturne only include the implicit filtering by the mesh.
If the mesh could be reworked so as to replace the tetrahedral regions with hexahedra (possibly at the cost of a non-conformal joining, not ideal either for LES, but probably bettet), that would help. Otherwise, you would need to modify the turbulent viscosity (with a user-defined function) so as to implement an explicit filter, but we have not done it as far as I know on code_saturrne, so this is uncharted territory.
Best regards,
Yvan
I had a brief discussion with my colleagues, but it seems the solution they had we purely for postprocessing (FFT for a HIT test case).
So there is no directly available solution here it seems. We we guess may happen is that the numerical diffusivity is higher in the tetrahedral regions than in the hexahedral regions, causing your artifacts. A workaround for this would be to add explicit filters, but the LES models in code_saturne only include the implicit filtering by the mesh.
If the mesh could be reworked so as to replace the tetrahedral regions with hexahedra (possibly at the cost of a non-conformal joining, not ideal either for LES, but probably bettet), that would help. Otherwise, you would need to modify the turbulent viscosity (with a user-defined function) so as to implement an explicit filter, but we have not done it as far as I know on code_saturrne, so this is uncharted territory.
Best regards,
Yvan