Experimental and numerical study of stochastic branching noise in nuclear reactors

Noise analysis applied to nuclear reactor physics is a powerful tool to investigate a reactor's kinetic parameters, and more generally underlying physical processes determining core behavior. The kinetic parameters are the coefficients of simplified time-dependent neutron population equations, the so-called point kinetics equations. Experimentally determined kinetic parameters aid to validate codes and to potentially evaluate nuclear data. This thesis focuses on improvements to the kinetic parameter and uncertainty quantification, the experimental techniques, and the direct simulation of noise experiments. In particular, the notion of spatial effects, i.e. effects that render point kinetics assumptions inaccurate for noise measurements, was investigated. This was achieved by drawing on experiments conducted in the zero power reactor CROCUS.

New detection instrumentation for noise experiments was developed for CROCUS, namely current mode amplifiers for neutron detectors to allow for higher detection efficiency and thus precision; and a scintillation-based gamma detection array, called LEAF, to enable the study of gamma noise. A set of reference neutron and gamma noise experiments were conducted to exemplify an improved method of parameter and uncertainty estimation based on bootstrapping. It lead to a full uncertainty budget showing a relative uncertainty of 3.6% and 1.3% on the prompt decay constant with neutron and gamma noise, respectively. Moreover, reporting of full distributions of kinetic parameters was shown to be required in order to provide an accurate representation of the experimental results. Gamma noise was shown to be superior in terms of precision by a factor of at least two compared to neutron noise when determining the prompt decay constant.

To study spatial effects and the spatial extent of the noise field of CROCUS, a set of experiments with varying detector locations, reactivity, and particle type were conducted. The neutron noise field study showed that measurements are limited to the immediate proximity of the core, and that CROCUS is reliably modeled as a point kinetic reactor for static configurations. Nonetheless, a systematic trend was shown when comparing to code predictions, pointing to a weak spatial effect biasing neutron noise measurements at a distance. The observable gamma noise field using high efficiency detectors was shown to extend beyond neutron noise limits, enabling ex-vessel measurements. The prompt decay constant was determined at about 0.9 meters to the core center with comparable accuracy to that of gamma in-core reference experiments using less-efficient detectors. Gamma correlations were be observed outside of the reactor cavity in front of an experimental channel, enabling a prompt decay constant determination at 7 meters to the core center.

To complement the experimental investigation and to study the underlying physics, a simulation methodology to estimate noise responses was developed. Specifically, the explicit simulation of noise experiments using analog Monte Carlo transport coupled to a fission model was studied. An established code with an existing fission library coupling, TRIPOLI-4, was used as reference. In addition, a newly developed coupling of a fission library to Serpent 2 allowed for code-to-code comparison. The full methodology, verification, and validation to the aforementioned CROCUS experiments is presented.