Pressure drop after restarting the simulation

[Stéphane Blanchet]

Dear support and users,

I noticed a sudden pressure drop after restarting a simulation (about 2Pa, I attached a screenshot of the pressure on one probe). I used the same parameters as the previous calculation.

Is this something I should be worried about? I mean, the pressure drop is not so big, but the phenomenon is quite odd I think.

Any ideas?

My best regards,

Stéphane

[Yvan Fournier]

Hello,

I do not believe you should be worried: for Code_Saturne computing incompressible flows, what is important is the pressure difference relative to a an (arbitrary) reference pressure, and not the absolute pressure.

Did you restart using the same number of processors, or in case of parallelism, run on the same machine ? If this is not the case, local face numbers may be different, and the reference pressure point chosen differently, so the behavior you see is normal. See  https://code-saturne.info/products/code-saturne/issues/30 with a proposed patch (testing for the restart pressure modification would require both the initial calculation and the restart to use the patch).

If you are running on the same number of processors with the same partitioning, there should not be a pressure difference when restarting with both main and auxiliary restart files, but it is possible the some minor piece of information is missing from the restart file and recomputed at restart, leading to the difference you observe (which remains only a minor issue, but which we still try to avoid, so your calculation options could be of interest for us i this case).

Best regards,

  Yvan

[Stéphane Blanchet]

Hello,

I ran the simulation on the same machine and using the same number of processors.

I can send you the files you are interested in, just tell me which files you need.

Best Regards,

Stéphane

[David Monfort]

Hello Stéphane,
The easiest for us would be the whole test-case (user files, xml file, runcase and mesh) if it is possible for you. This way we shall be able to run exactly the same simulation.
David

[Stéphane Blanchet]

Hello,
You can download the test case here:   http://filex.uhp-nancy.fr/get?k=sHH12IGANoIZ5snDeud
I did some batch calculations, but I cleaned the runcases I send you in order to be more "readable".
Let me know if you need something else.
Stéphane

[David Monfort]

Thanks for the attachment. We'll try to have a look at it in the next few days...
David

[Stéphane Blanchet]

Ok, thank you very much.
Stéphane