Getting standard deviation on atoms in depletion simulation

Hi all

I was asked today if one can I get a standard deviation for the activity or atoms in a depletion simulation?

The standard deviation we are interested in is not related to the uncertainty on the volume from a stochastic volume calculation but from the standard deviation on the reaction rate.

This question got me thinking how would one go about adding such a feature. Perhaps one aspect that would be needed is to make use of uncertainties.ufloat for the nuclide numbers in a material?

I think the openmc.ReactionRates is currently a numpy.ndarray with extra attributes contains a reaction rate for each material, nuclide and reaction. I’m wondering if the std_dev could be added to this class

Hi Jon,

Having the uncertainty on the reaction rate is one thing - and certainly helpful - but propagating that to the number density/activity is an active area of research, (e.g. Uncertainty Quantification of the Decay Gamma Source in Mesh-Based Shutdown Dose Rate Calculations: Fusion Science and Technology: Vol 79, No 1 or “Depletion Perturbation Theory Sensitivity Coefficients in Monte Carlo Simulations”, Benjamin Murphy & Christopher Perfetti (PHYSOR 2022))


Thanks Paul

I was originally asking as I’ve seen some FLUKA simulations recently that have an error bar on the shut down dose values and activity values for a multi pulse irradiation. But perhaps that error bar is not found in the same way or I have misunderstood what it represents.

I see UW and yourself are leading the way in this area :tada:

I didn’t have access to the paper but found the thesis

I guess for OpenMC it might be useful to get some stats on depletion runs in general as this will help us know if the tally is converged and perhaps put in triggers to end the simulation if it meets std_dev or relative error goals.

Currently I am planning to add neutron flux tally to and check that the flux tally converges for a few energy groups to get an indication of the convergence of the depletion simulation, which is not very convenient.