code_saturne User's Forum

Hello,

I have a problem when I use a restart file in case of FSI problem.

I am studying oscillation of a cylindre in a fluid flow. If the computation is restarted during the oscillation there is discontinuities in the solution of the displacement and the velocity of the cylinder. The solution is not the same if computation is restarted or not.

Have I forgotten to do something?

Thank you in advance

Erwan

Hello,

I suspect a bug, and probably an old one. I recall you have been using Code_Saturne for some time. Did you ever try a restart with FSI on versions older than 3.0 ? The bug is probably older that 3.0, but if you have any reference of a "working" version, I am interested.

Do you have any messages in the "listing" indicating that some fields/data arrays are not found in the restart file ?

An interesting test would also be to check if reading auxiliary restart files (ileaux in cs_user_parameters.f90) or not makes a difference.

If you have a small enough, non confidential test case, I can try to reproduce/debug this.

Regards,

Yvan

There was no problem with the 1.3.3 version. I think the version 2.0 (stable) is OK. I am testing it, and I should have the answer tommorow.

I have the following warning in the listing file :
@
@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
@
@ @@ WARNING: WHEN READING THE AUXILIARY RESTART FILE
@ =======
@
@ The max number of structures NBMOMX supported by
@ the writing format of the suite file is
@ NFMTST = 99
@ There is here a greater number of structures
@ NSTRMX = 200
@ If the effective number of structures is greater,
@ these will not be reread.
@
@ The run will continue.
@
@ Check the subroutine lecamx.
@
@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@

Only one structure has been defined.

Please find the xml and mesh files attached.

Regards,

Erwan

Hello,

I downloaded your files, and will take a look within a few days. How many iterations did you need to run before restart ?

Regards,

Yvan

Thank you.

Usually i use 40000 iterations.

I have just observed something with the version 2.00 of Code_saturne. I have the same error if i use the gui to parametrize the FSI problem. It's OK if I use fortran file ....
I will try with only fortran files for the version 3.0.4

Regards,

Erwan

Yvan Fournier wrote:Hello,

I downloaded your files, and will take a look within a few days. How many iterations did you need to run before restart ?

Regards,

Yvan

Hello,

Another user encountered the same issue (http://code-saturne.org/forum/viewtopic ... 8071#p8071), but your files were most useful in fixing the bug.

For some reason, bug reports in 2013 and 2014 tend to "cluster" (a bug remains undetected for months or years, then several users encounter it within a few weeks).

In any case, 3.0.5 is out today, and should fix this bug (using usstr1, you can work around this, as per the updated example), but with the GUI, there was a real bug.

Regards, and thanks for the detailed report.

Yvan