Abstract
A novel method for assessing the reliability of 37- element CANDU [Canada deuterium uranium (reactor)] fuel was developed. The conceptual approach follows the principle of “best estimate plus uncertainty” (BEPU) where reliability or probability of failure to meet a fuel performance criterion is adapted as a measure of safety with due consideration of uncertainties. In this study, fuel performance was predicted using the industry standard code, ELESTRESS, which models fuel behavior during steady-state operation [1]. The outputs of the code were construed against failure criteria derived from industry norms to determine the probability of failure. In total, 105 independent iterations were made for each of 12 radially divided zones of a 480-channel CANDU reactor modelled after the Darlington Nuclear Generating Station cores [2]. The Monte Carlo simulation method is applied to analyze this problem. Probability distributions of fuel element related variables were estimated from real data supplied by Cameco Fuel Manufacturing (CFM) [3]. The inputs for element rating were simulated from a pool of core-following data generated using a 3D diffusion code, the Reactor Fuelling Simulation Program (RFSP) [4]. The results of the simulations predict zero fuel failures for all zones. The prediction of the method was compared to a deterministic, “limit of envelope” (LOE) benchmark, where all input parameters are assumed to be at their limits simultaneously. The benchmark analysis produced an output that is significantly closer to performance limits than those obtained from the proposed Monte Carlo simulation method.
