Hello,
Some default settings were improved in v7.0, and some finer robustnesss oriented settings are possible (such as using a different gradient reconstruction algorithm for different variables), but otherwise, robustness of v6 and v7 should be about the same.
In your case, if you use a prescribed velocity instead of a pressure difference, what pressure difference do you have at the end ? Did you try initializing a computation by prescribing velocity and restarting with a pressure difference ? No guarantee it will work, but easy to try.
Best regards,
Yvan
Divergence for simple geometry room
Forum rules
Please read the forum usage recommendations before posting.
Please read the forum usage recommendations before posting.
-
- Posts: 4085
- Joined: Mon Feb 20, 2012 3:25 pm
Re: Divergence for simple geometry room
Hello Yvan,
I tried what you said: prescribing pressure difference which are obtained in a simulation with using velocity inlet.
However, it did not work.
Therefore, I thought I can oversome the problem if I put fluctuating pressure value (time history of pressure) instead of one static value in cs_user_boundary_conditions.f90, but this also did not work.
At the moment, a solution for this is to using hexa aound inlet and outlet where divergence happes, but idealy I hope we can do that by tetra only mesh.
Best regards,
Tsubasa
I tried what you said: prescribing pressure difference which are obtained in a simulation with using velocity inlet.
However, it did not work.
Therefore, I thought I can oversome the problem if I put fluctuating pressure value (time history of pressure) instead of one static value in cs_user_boundary_conditions.f90, but this also did not work.
At the moment, a solution for this is to using hexa aound inlet and outlet where divergence happes, but idealy I hope we can do that by tetra only mesh.
Best regards,
Tsubasa
-
- Posts: 4085
- Joined: Mon Feb 20, 2012 3:25 pm
Re: Divergence for simple geometry room
Hello,
Yes, this confirms our own experience.
Ideally, we should not need hexahedra, but with the classical code_saturne algorithm, improving robustness relative to the boundary mesh quality is not easy. With methods such as CDO, it is better, but this is still work in progress (an not accessible using the GUI).
So for the medium term, using hexahedra at outlets (which is easy to do with mesh extrusion) is a safe and practical solution.
Best regards,
Yvan
Yes, this confirms our own experience.
Ideally, we should not need hexahedra, but with the classical code_saturne algorithm, improving robustness relative to the boundary mesh quality is not easy. With methods such as CDO, it is better, but this is still work in progress (an not accessible using the GUI).
So for the medium term, using hexahedra at outlets (which is easy to do with mesh extrusion) is a safe and practical solution.
Best regards,
Yvan