Integrated studies on new NAF adjuster rods, source convergence, and numerical power tilts in full-core Monte Carlo CANDU-6 model

Abstract

A new adjuster rod design for the CANDU-6 is proposed utilizing NaF powder sandwiched between concentric Al tubes. The reactivity worth and flux shaping capability was assessed with full-core Monte- Carlo model of an equilibrium CANDU-6 core, and the new design achieves similar performance as conventional stainless steel adjusters. Elemental sodium is the primary absorber in the proposed rod, and the long-lived activation product 24Na generates delayed photoneutrons (PNs) in the heavy water moderator from the high-intensity 2.754 MeV gamma ray. At full power, the saturation activity of 24Na produces 2.3×1015n/s, exceeding by over a factor of 10 the PN yields of long-lived fission products 88Kr and 140La, and is equivalent to a delayed neutron factor of 1.03 × 10-5. The study shows it is possible to create an intense neutron source decoupled from the microscopic fission process large enough to influence the kinetic parameters of large lower reactors through neutron activation of materials in static reactivity devices. The availability of multiple Monte Carlo (MC) codes and computing hardware advances has made full-core MC simulation a tractable problem. To shift research emphasis from predicting keff to the local flux distribution, there are practical challenges when applying MC codes and models to full-core problems, such as determining key simulation parameters like inactive cycles, active cycles, number of histories (N), and initial source distribution. This paper investigates the power tilt phenomena in a 1/8 symmetry CANDU-6 full-core model developed in the Monte Carlo code MCS. Also, Central Limit Theorem (CLT) study is performed to understand the variation in bundle powers due to power tilts, with 100 low N cases initialized with different random seeds and the results show that the xenon-free 1/8 CANDU model simulation is well-converged using simulation parameters of 120 inactive cycles, 500 active cycles, and 10 million N. However, when simulated with equilibrium xenon core, there is strong autocorrelation of the CLT bundle power errors with systematically overestimated low power bundles and underestimated high power bundles. Through running 25 million N simulation, 5-10 million particles/cycle is confirmed as optimal N for 1/8 CANDU-6 model with minimal numerical power tilts contaminating the fundamental mode. However, CLT approach is exclusively applicable to xenon free condition, which xenon induces systematic biases in power tilt, further verification is necessary to determine if the observed autocorrelation is solely due to xenon.