[SOLVED] Code Saturne 2.3.1: issues /gradient calculation

[laurent]

Hi

Following the installation of Code Saturne 2.3.1, i re-ran a test that used to run well on Saturne 2.0.4
I may have found 1 issue:
-> everything runs fine when using "iterative handling of non-orthogonalities" for the gradient calculation method (albeit a message reading "run_solver.sh[10]: module: not found [No such file or directory]") but the calculation can not run when using all the methods with "least squares method"

I have this message in the listing:
"/tmp/saturne-2.3.1/installer/code_saturne-2.3.1/src/mesh/cs_mesh_quantities.c:1663: Fatal error.

The option selection indicator for the cocg computation
for vector gradients can take the following values:
0: do not compute cocg dimensionless matrices
1: compute cocg matrices
and not 0."


I am attaching the listing of a calculation that runs OK (***-1312) and calculations that doesn't run (***-1401)


Have a nice day

[Jacques Fontaine]

Hello,
I attached a fix, you can recompile Code_Saturne with it or add this files onto your SRC folder.
Could you tell me if the fix works on your test case?

Thanks for the feedback.

Regards,

[laurent]

Hi

I tried the fix, it didn't seem to work BUT i was a bit in a hurry so i would prefer to take sometime to do things properly before saying it doesn't work.
Could have a closer look in about 1week and sadly i have another question/issue? on other topic

Sorry for the bad news on a friday!
Regards

[Jacques Fontaine]

By the way could you attach your case here?
ty

[laurent]

Yes, but next week as i am currently away from the office.
Have a nice day and thanks for the follow-up

Regards

[laurent]

Hi

Here is the case discussed above

[laurent]

Hi

I tried my case again with Code Saturne 2.3.2 and there are no issues anymore with the "least square gradient".

There is still a warning with /_6/run_solver.sh but it doesn't stop the calculation
(_6 being apparently the new folder where Saturne 2.3 stores the post-pro results instead of date-time: is it normal?)

Have a nice day

[Yvan Fournier]

Hello,

Good to know the issue has dissapeared.

But the case directory should be date_time, and not _6 (unless you changed the default behavior in the runcase by adding "--id _6" to the "code_saturne run" command).

Do you always have this behavior ?

Best regards,

Yvan