Calibration of tuning parameters in the FRAPCON model

This paper uses methodology developed for Bayesian computer model calibration to constrain uncertainties in four tuning parameters in FRAPCON-3 (Geelhood et al., 2011a), a computational model that calculates steady-state fuel behavior at high burnup in light water reactors. The statistical methodology, described in Higdon et al. (2008), combines fission gas release measurements from 42 different experiments to constrain the uncertainties in four tuning parameters, as well as in additional statistical parameters describing measurement uncertainty, as well as discrepancy between model and reality. This demonstration provides a proof of principle for this methodology, suggesting its utility for more difficult parameter estimation problems, involving larger numbers of experiments and tuning parameters.

► We combine model runs with experimental results to estimate model parameters. ► We describe statistical methodology to carry out this estimation. ► We give an example using 42×36 FRAPCON model runs and 42 physical experiments.