code_saturne User's Forum

Hi Yvan,

you are right. I postprocessed the courant numbner and the interface in the region of the shaft seems to be the problem. I have to fix it changing the mesh in this region.

Thanks a lot!

Sebastian

Hello,

We'll also need to check the method of computation of the CFL number. Making the mesh a bit finer near the interface may be a good idea, but your calculation might actually be OK.

Best regards,

Yvan

Hello Yvan,

i did not anwsered yet because I optimized the mesh and wanted to do a test first. The shaft region was meshed very very fine compared to the rest. I fixed it making it a bit coarser. Up to 14 time steps the courant seems to be constant low. But beginning with the 15th time step it is still increasing. Actually, I do not know what i could improve to fix it. The Aspect ratio of the volume looks fine. There are also no warped or ent-angled faces. Do you have any idea what I can do now?

Best regards

Hello Yvan,

I could find the problem. It was again the shaft But now the courant number is good. So, your algorithm seems to be ok.

Thank you and best regards,

Sebastian

Hello Yvan,

I´d like to ask you an general question about checkpoints. I`m using checkpoints on a Blue Gene/Q system with help of cs_user_scripts.py. Is it possible to "fuse" the results to one (e.g. in ParaView)?

Best regards,

Sebastian

Hello,

I'm not sure I understand what you want to do. Are you using successive checkpoint / restarts ?
What results do you want to merge ? Checkpoints, or postprocessing ?
I assume it is postprocessing, as it is more useful.
In this case, no, there is no "immediate" way of doing this, though if you are using the EnSight output format, you could easily do this with a script: you need to shift the numbers of output suffixes for successive output directories, and concatenate the "time step" sections of the index (".chr") file.
To be safe, test this on a copy first.

It would be useful to add this functionality as a Code_Saturne script in the future, as this is a recurring need...

Regards,

Yvan

Hi,

yes, you understood it right. This is what I want to do. Iḿ using EnSight, so such a script would be nice. I do not understand it yet. Could you explain it a bit more in detail what I would have to do??

Thanks,

Sebastian

Hello,

For each calculation, in the postprocessing directory, using the EnSight gold format, transient variables are written as a sequence of files, one per variable per time step, starting at 1
(they are numbered by output, not by time step). So for example if you have pressure.00001, pressure.00002, ... pressure.00015 in the first directory, and pressure.00001, pressure.00002, ... pressure.00010 in the second, you need to rename the series from the second directory to pressure.000016, pressure.000017, ... pressure.00025, then place them in the first. Do this fo all transient variables... You also need to merge the time step descriptions in the .case file (which is pretty self explanatory, you'll see what to do if you open the .case file with an editor) to use this.

To avoid losing computation case strurctures, you could do all this in a new directory, copying/merging the .case file and using symbolic links for the data.

Regards,

Yvan

Hello Yvan,

thank you very much. It worked. I`ve another question, too. I would like to compute shear rate and shear stress in code_saturne. Therefore, i used the attached subroutines. I want to compute them for the whole boundary as well as certain curved faces, which I defined in output control -> meshes. For those faces I defined an additional writer. Coming to the problem. If I use a "transient connectivity" for both writer the computation crashes with the last time step. Using "transient connectivity" with a certain periodic output for the results writer and "fixed mesh" for the additional output writer it crashes, too. In case i`m using "fixed mesh" for both writers the computation is successful. In addition Charles told me that there is a problem with the XL-Compiler on the BlueGene/Q using the subroutines. Could you maybe have a look? I attached the subroutines as well as the XML.

Best regards,

Sebastian