cgns issue with v6.1.0

Questions and remarks about code_saturne usage
Forum rules
Please read the forum usage recommendations before posting.
Post Reply
daniele
Posts: 148
Joined: Wed Feb 01, 2017 11:42 am

cgns issue with v6.1.0

Post by daniele »

Hello,

Maybe someone can help in solving the weird problem I have...
I run code_Saturne v6.1.0 on two different clusters. The first difference that I find is that in one cluster the CGNS format required by Saturne is HDF, whereas on the other cluster Saturne wants the ADF... Is this linked to the installation options retained?

The real issue I have is that the CGNS HDF mesh is correctly read by the Saturne preprocessor (on one cluster). On the other hand, Saturne gives the following boundary condition error when I run the mesh CGNS ADF (on the other cluster):

Code: Select all

First face with boundary condition definition error
  (out of 788794)
  has boundary condition type 0, center (-0.0557381, 0.04092, -0.0079226)

cs_boundary_conditions.c:363: Fatal error.

Some boundary condition definitions are incomplete or incorrect.
Looking at the ERROR file, all the boundary faces are of type 0...

I tried to double check, ro recreate the mesh, but could not solve this issue.

Anyone has an idea of what could cause the problem?

Thank you very much.
Best regards,
Daniele
Yvan Fournier
Posts: 4070
Joined: Mon Feb 20, 2012 3:25 pm

Re: cgns issue with v6.1.0

Post by Yvan Fournier »

Hello,

This seems to be a CGNS installation, and possibly version issue (CGNS may be installed with or without HDF5 support, explaining the different default).

Could you post a "preprocessor.log" for each case ?

Regards,

Yvan
daniele
Posts: 148
Joined: Wed Feb 01, 2017 11:42 am

Re: cgns issue with v6.1.0

Post by daniele »

Hello Yvan,

Thanks for your help.
You're right, doing a tkdiff between the two preprocessor.log, the only difference is the CGNS format support.

For the 6.1.0 version requiring HDF and working correctly:

Code: Select all

  code_saturne version 6.1.0   (built Fri Jul 17 16:01:17 2020)

  CGNS 3.4.0 file format support
  MED 4.0.0 (HDF5 1.10.5) file format support
For the 6.1.0 version requiring ADF and giving the BC problem:

Code: Select all

code_saturne version 6.1.0   (built Mon Jun  8 14:33:57 2020)

  CGNS 3.2.1 file format support
  MED 3.0.8 (HDF5 1.8.20) file format support
To note that the BC error does not appear in the preprocessor.log, but in the listing.
I used to work with ADF format with previous Saturne versions, and never had such problems. Do you see any way to solve the issue, other than re-installing Saturne with the same CGNS format support?

Thank you.
Best regards,
Daniele
Yvan Fournier
Posts: 4070
Joined: Mon Feb 20, 2012 3:25 pm

Re: cgns issue with v6.1.0

Post by Yvan Fournier »

Hello,

Could you post the "listing" (run_solver.log) and setup.log file for each case ? Once the file is converted by the preprocessor, there should be no difference. So if the rest of the preprocessor.log is the same, I would expect no issue.

Best regards,

Yvan
daniele
Posts: 148
Joined: Wed Feb 01, 2017 11:42 am

Re: cgns issue with v6.1.0

Post by daniele »

Hello,

I have attached the two listing files, that I have renamed "listing_ok" and "listing_error".
You are right, I should have noticed it, there is something inconsistent in the "listing_error": why does the code look for calculation parameters and perform variables initialization, if it is only a preprocessing calculation?

I therefore made a test: I wrote manually the BC list inside the GUI, and then saved the setup as "mesh preprocessing only". I don't have errors anymore, but the code performs iterations as if it was a "standard computation"!!
What is strange is that the directory created by the code is called "preprocessor_...".

The good thing is that the calculation works fine...

I had never faced this weird behavior before. Have you?

Thank you very much.
Best regards,
Daniele
Attachments
listing_ok.txt
(9.93 KiB) Downloaded 123 times
listing_error.txt
(16.14 KiB) Downloaded 131 times
Yvan Fournier
Posts: 4070
Joined: Mon Feb 20, 2012 3:25 pm

Re: cgns issue with v6.1.0

Post by Yvan Fournier »

Hello,

Do you use cs_user_scripts.py ? The required option could be removed in a user setting, so that could be an explanation.

Otherwise, could you post the XML files for each case ?

Regards,

Yvan
daniele
Posts: 148
Joined: Wed Feb 01, 2017 11:42 am

Re: cgns issue with v6.1.0

Post by daniele »

Hello,

No, I don't use cs_user_scripts.py.
Attached the setup.xml.

Thanks in advance.
Best regards,
Daniele
Attachments
setup_ok.xml
(4.74 KiB) Downloaded 129 times
setup_error.xml
(8.58 KiB) Downloaded 121 times
daniele
Posts: 148
Joined: Wed Feb 01, 2017 11:42 am

Re: cgns issue with v6.1.0

Post by daniele »

Hello Yvan,

Actually, I think there is something wrong with the specific v610 installation I am testing. I have just found out that user routines are not taken into account by the code: they are compiled, but they are not used... :cry:

Best regards,
Daniele
Post Reply