Deterministic neutronics calculations usually rely on a two-step approach, called lattice and core steps. In the first one, the multigroup cross-sections are reduced (condensed over a few energy groups and homogenized over assembly-size regions) using a small subset of the whole system geometrical model (typically, a single subassembly representative of a repeated pattern) in order to reduce the dimensionality of the core calculation step. When those reduced cross-section sets are used for core sensitivity analyses, the impact of the lattice step is usually neglected. For some quantities of interest, this can lead to important discrepancies between the computed sensitivities and the actual ones, since lattice transport calculations are key for carrying the fine-energy local neutron spectrum information and resonance self-shielding effects. There can be an additional concern when those sensitivity calculations are used to provide feedback on nuclear data evaluations, or in the case of similarity studies. In order to address this issue, several approaches are available, such as direct calculations or perturbation theory studies, each representing different trade-offs in terms of cost or complexity.
The goal of this PhD is therefore to explore the state of the art of the domain, ranging from the most brute force approach to the ones based on perturbation theory, with the possibility to propose new methodologies. The implementation of the chosen methodologies in new generation codes (such APOLLO3) will allow eventually to improve the accuracy of sensitivity calculation.
The doctoral student will be based in a reactor physics research unit at CEA/IRESNE in Cadarache, which hosts many students and interns. Post-graduation perspectives include research in nuclear R&D labs and industry.