I’ve been working on the ITER 1D benchmark and trying to run it through OpenMC to then compare it with MCNP outputs.
The geometry of the problem is quite simple, it is many concentric cylinders bounded at the top and bottom by reflective planes.
After converting the MCNP input to OpenMC I could get the simulation to run fine. I then wanted to create a weight window as a means of variance reduction. But on the addition of the mesh tally the simulation stopped working and I encountered an issue/issues.
One approach used a cylindrical shell source, which had essentially the same dimensions as the ‘plasma’ cell of the geometry. After this sim reached batch 3 it would get stuck there indefinitely without outputting an errors. Just as an attempt to debug or explore what was going on, I changed the reflecting boundaries at the top and bottom of the cylinder to vacuum’s. Following this change, the simulation would run to completion
In a different approach, I used cell rejection to create my source. This approach would get stuck on batch 27 and would remain there indefinitely. However, when the bounding planes were changed to reflective, this simulation would still get stuck.
I found this issue, and tried changing the tally estimator to ‘collision’ and it seems to be running after that.
Cheers,
Jude