Assessing sensitivity and uncertainty in rod ejection accidents: A combined PCE-POD approach

Abstract

This study investigates the sensitivity and uncertainty of outcomes related to rod ejection accidents in nuclear reactors, focusing on the impact of perturbations in neutron parameters. By perturbing the macroscopic cross-sections of fuel assemblies, we assess how these variations influence reactor dynamics during transient events. We employ Polynomial Chaos Expansion (PCE) combined with Proper Orthogonal Decomposition (POD) to create an efficient analytical framework for quantifying uncertainty. Using Monte Carlo simulations as a benchmark, we demonstrate that PCE can accurately estimate the mean and variance of global peaking parameters with significantly fewer samples than traditional methods. Our findings reveal that only a limited number of POD modes can capture the majority of the energy distribution, facilitating a more streamlined analysis. This research enhances our understanding of reactor behavior under uncertainty and contributes to improving safety assessments in nuclear engineering.