code_saturne User's Forum

Dear all,

I am doing the simulation about gas chemistry for in CS9.0. When I run the model, meteo file can be read normally, but chemistry profiles file data only can be read in the third hours .

However, before when I run this case in CS8.2, all the hourly data of chemistry profile can be read. I have checked referenced source files, they are basically same. Is that any difference between this two versions or any options I need to be added in the cs_user_parameter?

I have attached my run_solver.log and chemistry file.

Thanks,
Sophie

Hello,

I am not aware of any voluntary change in the settings. So this may be a bug. Could you post a setup.xml file so we can compare the behavior between versions here an check this ?

Best regards,

Yvan

Hi Yvan,
please see the attached cs_user_parameters.c and setup.xml.

Thanks!
Sophie

Hello,

The code complains it is missing some spack file. Coul you post species.spack.dat ?

Regards,

Yvan

Here are c code files and spack file.

Thank you!

Hello,

I think you forgot to post the files.

Regards,

Yvan

Here are c code files and spack file.

Thank you!

Hi Yvan,

I also did some tests with chemistry mode=1,2,3. Even I modified the chemical species accordingly, all of them will report error like
"
Call stack:
1: 0x2ae9a57ed7d3 <cs_f_read_chemistry_profile+0x1409> (libsaturne-9.0.so)
2: 0x2ae9a581087a <cs_atmo_fields_init0+0x35a> (libsaturne-9.0.so)
3: 0x2ae9a53acab4 <cs_f_ppiniv0+0xf1> (libsaturne-9.0.so)
4: 0x2ae9a51b8345 <cs_initialize_fields_stage_1+0x875> (libsaturne-9.0.so)
5: 0x2ae9a52c2919 <cs_time_stepping+0x339> (libsaturne-9.0.so)
6: 0x2ae9a4e1557b <main+0x74b> (libcs_solver-9.0.so)
7: 0x2ae9aa385555 <__libc_start_main+0xf5> (libc.so.6)
8: 0x401449 <> (cs_solver)
End of stack

Abort(1) on node 6 (rank 6 in comm 0): application called MPI_Abort(MPI_COMM_WORLD, 1) - process 6"

So I think it's still because the file cannot be read. Do you have any data example foe chemistry? I only find meteo data in SRC folder.

Thanks,
Sophie

Hello,

The issue is due to a change in v9.0. In dimensions.f90, replacing
by:
Should fix the issue.

As this broke compatibility and consistency, the previous behavior (i.e. no need for "bind" option) is restored in v9.1, and I just fixed the compatibility in v9.0 (available in v9.0 git branch now, will be in next v9.0.2 bug fix release), so as to allow both options.

So you can either modify your code or upgrade to the latest bug fixes on the v9.0 git branch.

Best regards,

Yvan