Using thermal scattering 'c_D_in_D2O' provokes segmentation fault error

User Support
1

Hi all,

I am running a model with heavy water. When I add thermal scattering for deuterium in the D2O, the simulation crashes and produce a segmentation fault. The error messages I receive can vary. This is one of the message I receive:

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:

2

Hi Julien,

A few questions for you:

  • What version of OpenMC are you running and how did you install it?
  • What library are you using and how was it produced?

Also, note that we do have a short section in the user’s guide on handling segmentation faults. If you recompile OpenMC with debug flags on, you should be able to get a traceback that gives us more information on what’s going wrong.

Best,
Paul

3

Hi Paul,

I use use version 0.10.0.
I use jeff3.2 and it was downloaded and converted by the script openmc-get-jeff-data.

I have recompiled OpenMC (with MPI) and HDF5.

However, it would still crash when using c_D_in_D2O.

Now, c_D_in_D2O was just positioned after the c_H_in_H2O in my material.xml. When I put it before c_H_in_H2O or when I just delete the c_H_in_H2O from the xml file, there are no segmentation fault anymore. There are almost no H2O compared to D2O in my moderator so I am fine with that fix for now.

Could you try reproduce this error and see if you also have the same problem?

Thanks a lot.

Julien

4

Hi Julien,

Since the last release, a bug has been identified when using multiple S(a,b) tables (described a little bit in the last bullet point here). If you need to use multiple S(a,b) tables, I would recommend reinstalling OpenMC using the latest “develop” branch which already has a fix for this issue. If you do decide to update, you can find pregenerated data libraries (including JEFF 3.2) here.

Best,
Paul