Multiphysics modeling is crucial for nuclear reactor analysis, yet uncertainty propagation across different physical domains—such as thermal, mechanical, and neutronic behavior—remains underexplored due to its complexity. This PhD project aims to address this challenge by developing innovative methods for integrating uncertainty quantification into multiphysics models.

The key objective is to propose optimal modeling approaches tailored to different precision requirements. The project will explore advanced techniques such as reduced-order modeling and polynomial chaos expansion to identify which input parameters most significantly impact reactor system outputs. A key aspect of the research is the comparison between "high-fidelity" models, developed using the CEA reference simulation tools, and "best-estimate" models designed for industrial use. This comparative analysis will highlight how these errors propagate through different models and simulation approaches.

The models will be validated using experimental data from SEFOR, a sodium-cooled fast reactor. These experiments provide valuable benchmarks for testing multiphysics models in realistic reactor conditions. This research directly addresses the growing need for reliable, efficient modeling tools in the nuclear industry, aiming to improve reactor safety and performance.

The candidate will work in a dynamic environment at the CEA, benefiting from access to advanced simulation resources and opportunities for collaboration with other researchers and PhD students. The project offers the possibility of presenting results at national and international conferences, with strong career prospects in nuclear reactor design, safety analysis, and advanced simulation.