On the Feasibility of Neutron Noise Diagnostics in Fast Gen-IV Reactors

Abstract

This master’s thesis studies neutron noise, i.e. the temporal fluctuations from the mean neutron flux, in nuclear reactors. Neutron noise can be used to identify and localise faults and anomalies in such systems through a technique called neutron noise diagnostics, potentially providing early warning and leading to increased safety and operational availability. Due to the limited amount of neutron flux instrumentation available in a typical power reactor, the spatial shape of the noise is the key quantity to characterise the type of noise present. This study differentiates itself from prior research on neutron noise diagnostics, which mostly focused on Light-Water Reactors (LWR), whereas this work investigates the properties of the neutron noise in both LWRs and in Liquid Metal-cooled Fast Reactors (LMFR). More specifically, we compare the axial shape of the neutron noise of: Pressurised Water Reactors (PWR), Boiling Water Reactors (BWR), Sodium-cooled Fast Reactor (SFR) and Lead-Cooled Fast Reactors (LFR). The former two are LWRs, and the latter two are LMFRs. Each system’s deviation from point-kinetics is analysed, which is essential to evaluate the capability of noise diagnostics. To this end, a 1D numerical neutron transport solver was developed using linear multi-group diffusion theory with a variable number of delayed neutron families. The solver uses group constants generated by the Monte Carlo code Serpent-2 for each modelled system. The static neutron flux is described by an eigenvalue problem, which is solved by the power iteration method, accelerated using Wieland’s technique. The induced noise arises from an absorber of variable strength-type perturbation in the macroscopic absorption cross-section, and the resulting source-type problem is solved by matrix inversion. Our computations show that the induced neutron noise of the LWRs deviates more from point-kinetics, and their axial shape is thus more informative than that of the LMFRs. Also, perturbations closer to the reactor’s periphery induce a neutron noise that is more informative. The point-kinetic behaviour of the LFR is strongly dictated by the active height of the core, more so than by the fast spectrum. We also show that reactors fuelled by plutonium are more point-kinetic than those fuelled by uranium. For the results to become more conclusive, more work is needed regarding the types of reactors studied, with more analysis aimed towards what parameters affect the point-kinetic nature of the noise. Studies can also be made in higher spatial dimensions that more accurately take into account the complex geometries of nuclear reactors and incorporate burn-up calculations when generating the group constants.