A hybrid atomistic-empirical numerical twin to predict the fuel microstructure evolution
Like other industrial sectors, the nuclear industry is developing 'digital twins', applications that simulate the behaviour of an industrial component, such as a car, plant, reactor or, in our case, a fuel rod. The proposed work is part of this approach and contributes to the development of a digital twin of a grain that can be used to simulate its microstructural evolution under irradiation.
The microstructure of nuclear fuel (uranium oxide) is severely damaged during irradiation in a reactor: the atoms produced by the fission of uranium nuclei displace the atoms in the material in a cascade, creating irradiation defects (vacancies and interstitials) whose aggregation leads to the gradual appearance of cavities and dislocation loops. These extended defects influence the volume of the material, its creep and its retention of fission gases. The physical model of the phenomenon is cluster dynamics: a set of kinetic equations representing the chemical reactions of defect aggregation by diffusion in the material.
Most of the model parameters are derived from atomistic calculations (defect formation and migration energies, irradiation damage). However, some of them have not been calculated and are practically impossible to measure directly. The approach proposed for this internship is twofold:
a.Fitting the missing parameters using the model to simulate the results of experiments (already available in the laboratory) in which the microstructure is affected, such as transmission electron microscopy characterisation of dislocations and voids (size, concentration). Innovative techniques will be used:
•Sensitivity analysis to determine which parameters affect the measured values
•Optimisation (genetic algorithms) to fine-tune these parameters. The URANIE™ platform developed by the CEA will be used for the statistical analysis of the data.
•Kinetic Monte Carlo for the damage simulation
b.Validate this fitted model by comparing its results with measurements from other experiments. The model will then be applied to new situations, such as irradiation of fuel material to predict fission gas release, or the density of loops and dislocation lines.
This modelling project involves a variety of tasks:
•Interpretation of experiments
•simple computational development of optimisation scripts for URANIE
•simulation of experimental or industrial situations
This internship offers the candidate the opportunity to contribute to the development of numerical physics applied to multiscale modelling, taking a central position and a synthetic point of view. It is also an opportunity to discover for oneself how microscopic computational approaches ultimately help to solve complex practical problems.
R. Skorek, Étude Par Dynamique d’Amas de l’influence Des Défauts d’irradiation Sur La Migration Des Gaz de Fission Dans Le Dioxyde d’uranium, PhD Thesis, Univ. Aix-Marseille, 2013.
E. Gilabert, D. Horlait, M.-F. Barthe, P. Desgardin, M.-L. Amany, G. Carlot, M. Gérardin, S. Maillard, and T. Wiss, D2.2 - Behaviour of Fission Gases and Helium in Uranium Dioxide, EC report, 2020.
Master's degree or equivalent in materials physics, modelling or numerical physics
Create candidate area
