code_saturne User's Forum

Hello,

I have an error message with CS_2.1.4 and radiative calculation.

I run a radiative calculation with CS_2.0.0, there is no problem.
But the same mesh with CS_2.1.4 bug after 4 iterations.

Radiative parameters:
- Discrete ordiantes method
- 128 directions for angular discretisation
- absorption coefficient = 0m-1

I have an error file with "Signal SIGFPE (exception en virgule flottante) intercepté !"
I have read this could be a division by zero because of the mesh, but the same mesh works with CS_2.1.0.
Futhermore, when I run a calculation without radiative with CS_2.1.4 and there is no bug.
The problem is not with my mesh.

I didn't see, in the user guide, specific modification between the 2 versions for radiative calculation.

Do you know if there is a tutorial or a Flow modeling examples which use radiative.(In order to test my CS_2.14 installation)

Thanks
Best Regards
Alexandre

Hello,

Do you use user subroutines, or just the XML ? As version 2.1 is an intermediate development release and has not been tested as much as 2.0, you might have found a bug.

Could you post your case setup (mesh xml, and user subroutines) for both versions 2.0 and 2.1, so we may compare ?

Best regards,

Yvan

Hi,
Sorry I can't give you the mesh.
But I will adapt the tutorial "Full domain" in order to make a radiative calculation.
Or do you know if there is a tuto or a Flow modeling examples which use radiative.

Thanks
Alexandre

Hi,
I have adapt tuto Full domain.
I still have a problem with CS_2.1.4.

CS_2.1.4 is not installed on the same machine.

Thanks again for your help.
Alexandre

Hello,

I have found the bug, and it also seems to also affect version 2.0 (even though it does not cause a crash, it may lead to incorrect boundary conditions, so probably incorrect results).

We were ready to release 2.1.6, 2.0.5 (and 2.2.0 later in the week); but I'll check with colleagues if they are ok with the fix tomorrow and get the fix in those releases.

In the meantime, line 1025 of src/gui/cs_radiative_transfer.c, if you replace (*ivart) by (*ivart -1), you should fix this bug (but you should do the same at line 1037 of src/base/cs_gui_radiative_transfer.c in version 2.0 before you compare results).

Best regards,

Yvan

Hello,

Thanks for your help.

I will make the modifications in the CS_2.1.4.
I will let you know.

Thanks
Alexandre

Hi,

I would like to know if the bug is fix in the version 2.2.1.
Thanks

Alexandre

Hello,
Yes the bug is fixed.
SVN log:

Merge rev. 3227 from trunk (fix array index for radiative transfer using GUI).

Regards,
JF

Ok, thank you

Regards,
Alexandre