code_saturne User's Forum

Hello,

I have recently upgraded to v6.1.0, and I think I have a problem in restarting a simulation when deforming mesh is acitvated.
I used to do it with v5, and never had problems.
Nevertheless, with v6.1 it looks as if the mesh saved inside the "checkpoint" directory is the original (i.e., at t=0 seconds, not deformed) mesh: restarting from the checkpoint causes negative volumes as soon as the new simulation is started.
Looking at the mesh in the checkpoint with Paraview, I can see that it is the original underformed mesh... whereas wall positions have moved during the simulation...
I have activated "transient coordinates" in the postprocessing through the GUI.

Do I miss something in the setting?

Thank you very much for your help.

Best regards,
Daniele

Hello,

The mesh saved in "checkpoint" should be the original/non deformed mesh (this mostly allows not re-running the mesh import step), but the code should (in theory) also read the vertex deformations in the auxiliary restart mesh (checkpoint/auxiliary.csm) and apply them before starting the computation.

If you have a very small test case illustrating the issue, I can test it, otherwise I will try to reproduce the issue on a case from our validation suite. The advantage of using your test case would be to have the exact combination of options which might cause this.

Best regards,

Yvan

Hello,

Thanks a lot for your help.
I reproduced the behavior on a small test case, that I have attached.
The test I did: run a simulation for 150 time steps, stop it and create the checkpoint, restart from the checkpoint. The restart fails.
Attached you will find:
- the setup.xml (the setup used to run the first part of the simulation, at the end of which the checkpoint is created);
- the mesh (cgns);
- the checkpoint directory;
- the cs_user_physical_properties.f90 used to define the mesh viscosity;
- the listing of the restarted simulation, showing the error.

If you have time to check if you have the same behavior, that would be great.

Thank you very much in advance.
Best regards,
Daniele

Hello,

There is an error in your cs_user_physical_properties.f90 file (you should use iloc instead of icel as per the loop variable inside the last loop), but otherwise I also reproduce the issue.

I'll keep you informed when I have that debugged.

Best regards,

Yvan

Hello,

There was a bug introduced in 6.0 when changing the mesh update algorithm. If pushed a fix in the master, v6.1, and v6.0 branches (so v6.2, v6.1.2, and v6.0.5 should be OK, and if you pull from Git, fixes are available now).

Thanks for the bug report and test case.

I also attach the corrected user subroutine.

Best regards,

Yvan