I try to use a restart point with syrthes coupling for having stationnary temperature fields. Only with CS I have realistic results with a frozen velocity field and compute the temperature in fluid with large time steps.
If I try to do the same calculation including syrthes coupling I have no more realistic temperature fields. It seems that the frozen velocity field is not taken in account in the temperature distribution (it seems the fluid is not moving like a solid phase). however I can visualize the frozen velocity in the fluid domain.
The coupling case is running fine without restart for small time steps.
Yes the restart file is also done for syrthes. Maybe my time step size is too large with the restart calculation. I have 0.01 sec for a Courant number less to 50 with Naviers Stockes and I change for 5 sec for the temperature calculation with velicity and pressure frozen fields. Is there a rule to limit the time step size in such case?
No, there is no specific recommendation. A restart should “just work”. I have not tested restarts with Syrthes coupling in a specific manner recently, but users who run major studies should use it regularly (as most computation queues are limited to 2 or 3 days, and long unsteady computations on large meshes require more than that), and have not reported issues.
If you have a small test case you can post, I’ll try to run it.
I post my case but I can not add the restart point because it’s larger than 10MB even compressed.
I have done 200 iterations of 0.01 sec to have the restart point and continue until this point for several hundreds iterations of 0.5 sec (you can see my files), I have the same problem than for larger time steps of 5 sec.