Giovanni Nicodemo
PhD thesis title: Machine learning enhanced multy-scale modelling of nuclear fuel microstructural evolution
Academic Tutor: Lelio Luzzi
Academic Supervisor: Davide Pizzocri
PhD cycle: 39° (see all student profiles of the same cycle > LINK)
BSc: Energy Engineering, Politecnico di Milano
MSc: Nuclear Engineering, Politecnico di Milano
3-month exchange at the CEA research centre in Paris-Saclay, France, from 01/07/2024 to 30/09/2024. The aim of the activity was the development of a methodology to surrogate with machine learning CALPHAD simulation results and integrate them into the SCIANTIX meso-scale code.
6-month exchange at the CEA research centre in Cadarache, France, from 01/10/2025 to 31/03/2026. The activity will involve atomistic calculations on nuclear fuel and the use of machine-learning methods for calibrating models and integrating them into mesoscale codes
Thesis abstract
The development of meso-scale models to describe nuclear fuel microstructure behaviour during irradiation is a key task for the design and licensing of new nuclear fuel concepts. Current approaches consider the coupling of thermochemistry solvers with Fuel Performance Codes, leading to an increased computational burden. To overcome such limitations, I am developing and implementing a method for data assimilation of material properties into mesoscale codes, and I am working on a methodology to integrate the results of thermochemical solvers into mesoscale codes and evaluate uncertainty on material properties, using artificial neural networks.
My PhD project has been funded by the following projects:
PATRICIA https://cordis.europa.eu/project/id/945077/results/it
PUMMA https://cordis.europa.eu/project/id/945022
TRANSPARANT https://cordis.europa.eu/project/id/101166386/it
Personal interest in my research theme
I am particularly interested in how AI methodologies can help in the development of nuclear codes, aiding the transfer of information through a novel approach that goes beyond classical coupling between different codes.
