Fission Source Rejection

User Support
1

Running into some trouble scaling up a problem to more procs. The fuel geometry is graphite spheres with thin shells on the outside that have fuel in them.

Everything works fine at 100 MPI procs. At 500 I got errors about some ranks not having any fission sites - so I upped the particles and all seemed well. At 1000 MPI ranks I get this:

ERROR: More than 95% of external source sites sampled were rejected. Please
check your external source definition.

For now I have literally just gone into source.cpp and commented out that error… and everything seems to be running fine. But: I don’t know the repercussions of doing such a thing (seems like it shouldn’t matter, other than the first iteration is going to waste a lot of time).

Below you’ll find my current settings.xml… let me know if there is something else I should try:

<?xml version='1.0' encoding='utf-8'?> eigenvalue 100000 50 20 -14.0 -14.0 -9.54 14.0 14.0 19.7 material 1 10000000 -1.4 -1.4 -1.4 1.4 1.4 1.4

Thanks for your help!

Derek

2

Hi Derek,

That check is mostly there in case a user inadvertently puts a source completely outside of the geometry. If you wanted to get rid of that check, a better way would be to change EXTSRC_REJECT_FRACTION in include/openmc/constants.h. However, there’s an easier fix for your particular case. Your spatial distribution for the starting source is type=“fission”, which means it only accepts source sites if they appear in a fissionable material (generally a good idea). If you were to change that to type=“box”, it would accept source sites in non-fissionable materials. Of course your source for the first generation will be very unrealistic, but unless you’re very concerned about the number of inactive generations, it shouldn’t be much of an issue.

Best,
Paul

3

Thanks Paul - makes sense!