Control on the standard deviation

User Support
1

Hi,

I’m currently performing criticality calculations to determine the critical mass of a system and compare it with results from other codes. I’m using a dichotomy approach, where the fissile mass is iteratively adjusted to reach a target keff value.

Within this framework, I developed a custom function that explores different moderation conditions (through the H/X ratio), evaluates keff + 3σ at each point, and refines both the fissile mass and the moderation ratio to converge toward a limit condition.

I was wondering: is there a way to implement an adaptive stopping condition in OpenMC based on a target standard deviation on keff? Ideally, I would like to allow larger uncertainties (e.g. 1000 pcm) for early iterations when masses are high, then tighten the threshold (e.g. to 100 pcm) once I get closer to the critical configuration. This would help reduce simulation time while maintaining accuracy where it matters.

Let me know if something like this is feasible or if you would recommend another approach.

Best regards.

2

If you are using the criticality search function have you thought about using python to make a loop that first runs at lower counting statistics (1/10, or 1/20th overall particles) and possibly a looser tolerance and then as it completes and loosely converges on criticality you have it rerun the criticality search with normal statistics (whatever you want the variance criticality search to be) but make the initial guess the result of the previous criticality search. If you get the correct settings I imagine it would work how you are requesting.

1 Like