Simulation always stops after exactly 50 batches and writes statepoint.50.h5

User Support
1

Summary

I have a very simple Godiva bare-sphere model (HEU metal, radius ~8.71 cm, density 18.74 g/cm³) written in Python using the OpenMC API.No matter what I set for settings.batches, settings.inactive, settings.particles, settings.no_batch_limit = True, settings.trigger_max_batches = 1000, or settings.trigger = None, the simulation always stops after exactly 50 batches and writes statepoint.50.h5. The final combined k-eff is printed, but only ~30–40 active batches are performed (depending on inactive count).Key observations so far

  1. Maximum batch is always 50

    • The last batch line is always 50/1 …

    • Statepoint file is consistently statepoint.50.h5

    • No error message — simulation finishes normally and prints results / timing statistics

  2. Same behavior in multiple versions

    • OpenMC 0.14.0 (conda-forge binary)

    • OpenMC 0.15.3 (previously compiled from source in a different env)

  3. k-effective estimate at batch 50 is identical

    • Example output at batch 50:
      50/1 0.82559 0.84084 +/- 0.00378

    • This value does not change even when I increase settings.particles from 10,000 → 100,000 → 500,000 or change inactive from 20 → 100 → 200.

    • Final combined k-eff is always ~0.84076 ± 0.00281 (at batch 50)

  4. Changing parameters has no visible effect on early batches

    • Increasing particles per batch, inactive batches, or max batches does not make the simulation continue past batch 50.

    • The printed k and uncertainty at batch 50 remain exactly the same across all tests. Questions

        • Is there a hidden/internal convergence criterion in 0.14.0 that cannot be disabled via Python API?

        • Is batch 50 a hard-coded limit in some conda builds or older versions?

        • How can I force OpenMC to run the full requested number of active batches (120+) without early exit?

2

Hi,

There is no such thing, you should be able to adjust these values freely. I would suggest you are either not properly updating the settings.xml file within you settings declaration, or you have a model within the folder you are operating out of. If there is a model it always runs first when you are calling openmc.run, therefore possibly rerunning a model that has batches set to 50 and gens set to 1. You can verify easily by looking into the folder you are operating out of, or by looking at the top of the outputs when running openmc.run and see if it states that there is a model.xml that its reading. If either or these are not the issue, I would post your settings declaration section and I can take a closer look.