Within the core of nuclear reactors, the fluid-structure interaction (de Lambert et al. 2021) and the intense neutron flux lead to a progressive deformation of the fuel assemblies. When it becomes important, it can lead to additional costs and to a lengthening of fuel unloading and reloading operations. It is therefore preferable to adopt a fuel management system that avoids excessive deformation and interactions between fuel assemblies. However, the prediction of fuel assembly bow and interactions is uncertain (Wanninger, Seidl, and Macián-Juan 2018). Indeed, the initial uncertainties propagate over several successive power cycles, of about 12 months each, through the coupling of nonlinear, nested, and multidimensional thermal-hydraulic and thermomechanical simulations.

The objective of the thesis is to propose a methodology to control, or even reduce, uncertainties in the calculation of the deformations of assemblies within a nuclear reactor core. The methodology implemented will have to deal with the multidimensional and nonlinear character of the physical phenomena to be modeled/simulated, by exploiting the different levels of simulation available in order to improve the computational cost/uncertainty compromise.