Uncertainty analysis of fuel rod steady state thermal behaviour by the probabilistic response surface method

Effects of uncertainties involved in the modelling of fuel rod thermal behaviour and materials data are studied by the probabilistic response surface method. A fuel rod irradiated at the Halden boiling water test reactor is used as an example. The statistical calculations are performed with the PROSA-1 and the fuel behaviour analyses with GAPCON-THERMAL -2 code. Statistical calculations exhibited increasing uncertainty of the estimated consequences with increasing burnup. Differences between a simple and complex response surface increased also with burnup. These differences are largest at the low probability tail areas of the distributions. The two dominating statistical parameters were uncertainties associated with the fission gas release model and fuel thermal conductivity. Other parameters as well as cross-term variations had only minor effects on results.