Error while running the same model for different particles per batch

[Shurahbil]

Hi everyone,
Today my friend was running a model in openmc for different number of particles per batch but both the model had 250 batches.At first,he was running the model with 10000000 particles per batch with 250 batches.The main purpose of this was to see the plot of shannon entropy and keff but this happened(please check the screenshot)

After 108 batches I got a warning and after 130 batches the simulation stopped.I am attaching the file here 10000000 particles per batch
When my friend got the first warning,I decided to run the simulation myself.Since I didnt have a powerful processor like him,I decided run 100,000 per batch.Everything else was the same.I didnt have any errors like him while running the simulation.Then I asked my friend to do the same.I told him to run the simulation for 100,000 batches instead of 10000000.The simulation ran successfully.Here is all the input and output files 100,000 particles per batch

Why did he get the error while running 10000000 ???
Thanks in advance

[paulromano]

Hi @Shurahbil. I wouldn’t worry too much about the warning about not being able to local a particle after crossing a lattice boundary. This can happen sometimes if a particle hits the exact corner of a lattice and is not able to figure out where it’s supposed to go next. We have settings for the allowed number of “lost” particles like this (Settings.max_lost_particles and Settings.rel_max_lost_particles) and the defaults are very low so that you can be assured that simulation results are not affected. If too many particles are lost because of geometry issues, the simulation will abort altogether.

The error you are seeing on the run with 10M particles per batch is not related to the geometry. In this case, it may indicate that something is wrong with the cross section data itself (JEFF 3.3). It’s hard to know what exactly is wrong because, as you’ve discovered, shorter runs appear to go through just fine. I’ll keep investigating and let you know if I find anything out.

[Shurahbil]

Thanks for your help