Power distribution bias from equilibrium xenon effects when insufficient neutron histories per cycle are used in Monte Carlo simulation of CANDU6

Abstract

This study examines the impact of insufficient neutron histories per cycle (N) on power distribution accuracy in Monte Carlo (MC) simulations of CANDU reactors when using equilibrium-xenon algorithms. We demonstrate that using an insufficient N with equilibrium xenon introduces significant bias into the power distribution. The observed bias is systematic with a strong flattening affect where power is suppressed in high-power bundles and increased in low-power bundles by up to several percent beyond the real physical flattening effect of xenon on the three-dimensional power distribution. At least 5 x 106 to 107 N must be used for criticality simulation with the 1/8th symmetric CANDU6 model studied to suppress the spatial power tilts and eliminate the bias. This recommended N is orders of magnitude larger than N used in previous full-core Monte Carlo simulations of CANDU reactors. A preliminary analysis of the CANDU bundle power tally variances shows there is an underestimation of the real variance by approximately a factor of two.