In a fusion reactor, the fuel for the fusion reaction is a mixture of deuterium and tritium, two isotopes of hydrogen. Tritium is radioactive, with a half-life of 12.3 years, and is produced in the CANDU reactor, so tritium for fusion machines must be produced on site. This is the purpose of the breeding blanket (BB), where lithium nuclei interact with neutrons produced by the fusion reaction to create tritium. The Eurofusion design for BBs is a water-cooled lithium-lead (WCLL) design in which the lithium-lead (LiPb) circulates inside a Eurofer structure cooled by water tubes (also in Eurofer). To protect these BBs from the very high-temperature plasma, a tungsten (W) armouring acts as the first wall material. Understanding the tritium cycle in a complex system such as a fusion reactor is essential for both safety and plasma fuel supply. It is currently simulated using a system code such as EcosimPro, which can be used to assess the quantity retained in the wall and transported from the BBs to the cooling system. In their current version, these system codes only take into account the diffusion of tritium in the materials, forgetting about the trapping of tritium around defects. The aim of this project is to extend the multidimensional and multiphysics finite element modelling tool FESTIM (co-developed by LSPM & IRFM) to model tritium transport in BBs. This will help assess the impact of trapping on tritium transport and permeation and provide better input data for the system codes.

Start of thesis: Oct 2022

Type of thesis: CEA/IRFM

Supervisors: J. Mougenot (Dir), Y. Charles (Dir), E. Hodille (CEA IFRM)

Abstract: The modelling of damage processes and their transition to failure in mechanical structures is increasingly based onvariational approaches , which can be interpreted as modelling based on damage laws regularised by introducing the gradient of the internal damage variable as an additional state variable (now called phase fields). They have proved to be quite effective in describing the responses of various mechanical structures even in damage-induced softening regimes, and where strain/damage localisation phenomena occur. Despite their many successes, the applicability of these damage gradient models to strong (non-linear) thermo-mechanical coupling has yet to be developed .Moreover, current work deals only with simple 1 or 2-D geometries. A second objective of this thesis is to extend this theory to 3D geometries.

Start of thesis: November 2022

Type of thesis: joint thesis with the Ecole Nationale d’Ingénieurs de Sfax (ENIS), Tunisia