Statistical Methods Applied to Fuel Rod Bow Analysis

Abstract

An extensive data base was used to arrive at an empirical “best estimate” model for predicting fuel rod bow as a function of burnup. The best estimate rod bow predictions were bounded (accounting for uncertainties) at an upper 95 percent value. This global bound or tolerance value was subsequently revised to be representative of the bow value at reactor operating conditions. Conservative predictions for bow were finally implemented at several high burnups in a statistically derived Departure from Nucleate Boil Ratio (DNBR) penalty function. It was found that penalties for DNBR due to bowing are insignificant even at higher burnups and, therefore, do not seriously impact the design thermal margin.