Application of Linear Propagation of Errors to Fuel Rod Temperature and Stored Energy Calculations

Linear propagatlon of errors evaluates modeling uncertainty by approximating a function of interest by first-order Taylor's series expansions and then approximating the variance of the function by the variance of the linear approximation. This report discusses uncertainty analysis for different nuclear fuel rod designs, the process of model validation, and the effect of cracked pellet fuel models upon temperabre uncertainty. Using a postulated power history, the uncertainty for the predicted thermal response of boiling water reactor (BWR) and pressurized water reactor (PWR} fuel rods was evaluated. Beginning-of-life (BOL) relative uncertainty for BWR and PWR fuel rods is approximately the same. while different end-of-fife {EOL} thermal response results in different EOL uncertainty. Determining the validity of modeling relative to reality is discussed in qualitative terms. Validity is dependent upon verifying that the code correctly implements the model and that satisfactory agreement is found between the model and measurements. Fuel modeling codes are now using cracked pellet fuel models, which result in decreased fuel surface temperature. Estimated stored energy is lowered; but its relative uncertainty is increased. In general, however, the absolute upper uncertainty bound for stored energy is lower for a cracked pellet model than for a solid pellet model.