UNCERTAINTY QUANTIFICATION FOR FULL-CORE STEADY-STATE PWR CORE SIMULATION WITH SCALE AND PARCS

摘要

Accurate computational modeling of a nuclear reactor requires systematic analysis to include the quantification of uncertainties attributable to nuclear data, simplifications in models, and boundary conditions. It is important to methodically propagate uncertainty through the modeling process to learn the downstream physics better as well as provide margins for the designers. Currently, methods based on sampling and perturbation theory are widely used in this area. The research here demonstrates the methods developed with the SCALE and PARCS code systems and its application to the Organisation for Economic Co-operation and Development (OECD) Uncertainty Analysis in Modeling (UAM) benchmark exercise I-3. The uncertainty from nuclear input is sampled by SCALE/Sampler to generate perturbed libraries, and the perturbed library files are used for SCALE/POLARIS to perform lattice analysis and prepare cross section files for PARCS. Statistical analysis is then performed on PARCS results to determine the mean and standard deviation values on the core design parameters such as k_(eff) and power peaking factor. Using this methodology, the uncertainty from nuclear data input is propagated through the core design process. In this work, the reactor studied is the Three Mile Island Unit 1 reactor. The uncertainty of the k_(eff) for this reactor was found to be about 500 pcm. For power distributions, it was found that this reactor had an axial power distribution that was not very sensitive to the nuclear input uncertainties and a radial power distribution with higher uncertainties on the inner region of the core.