Two Internships : Multiscale modeling of fission gas behavior toward a digital twin of nuclear fuel experiments and Atomic descriptors for AI-driven modeling of nuclear fuels

2-Atomic descriptors for AI-driven modeling of nuclear fuels The properties of several nuclear fuels depend on how atoms are distributed over the crystal lattice and displaced from their lattice sites. This internship contributes to the development of an artificial intelligence method designed in our laboratory to describe this atomic distribution and these displacements, with the aim of improving the prediction of fuel properties asks • Generate atomic configurations of these systems temperatures using at several molecular dynamics, and analyze atomic displacements around equilibrium positions. • Evaluate several existing descriptors, in particular their sensitivity to small displacements and whether atomic positions can be reconstructed from them. • Propose different ways of integrating these descriptors into PULSE and, where appropriate, test some of them on a model system. What you’ll learn • Atomic-scale simulation and machine learning materials. methods applied to • Statistical physics and materials physics for applications. nuclear energy • Scientific methodology, from literature review to critical analysis of results. • Scientific programming in Python within a research team. This internship provides experience in research at the interface between nuclear materials physics and artificial intelligence, including atomic-scale simulations, statistical physics, and machine learning. 1-Multiscale modeling of fission gas behavior toward a digital twin of nuclear fuel experiments The in-reactor performance of nuclear fuel depends on numerous physical phenomena spanning several length scales, from atomic-level mechanisms to microstructural evolution at the grain scale. This internship involves the numerical simulation of the phenomena involving fission gas, with the aim of improving the understanding and prediction of their effects on macroscopic fuel behavior Tasks power contributes to • Run phase-field simulations of gas bubble evolution in UO2, compare the results with experimental data, and test alternative assumptions. modeling • Perform calculations at smaller length scales (rate theory, molecular dynamics) to estimate selected input parameters of the phase-field model, in particular those that control the amount of gas reaching the grain boundaries. • Implement in Inferno the improved models and parameters from these studies. What you’ll learn • Practical experience with numerical simulation methods at different length scales. these these • Knowledge of the physical mechanisms governing fuel behavior under irradiation. • A scientific approach, from formulating hypotheses to testing them against experiments. • Collaborative development of scientific software in Python, with version control (Git) and testing. This internship provides an opportunity to take part in nuclear fuel research and to develop skills in materials physics, numerical simulation, and software development.