Mathematics

Nowadays, the design of next-generation nuclear reactors relies on models and numerical simulations to analyze both nominal and accidental conditions. This requires the integration of models into a multi-physics and multi-scale environment capable of describing the most significant phenomena. One of the challenges in such a complex numerical framework is the modeling of turbulence, due to the harsh operating conditions reached in the reactor vessel and the specific properties of coolants (e.g., liquid metals). Another critical aspect is the estimation and propagation of uncertainties in the adopted numerical models, which must treat as stochastic variables all phenomena that cannot be represented in detail. In this context, data-driven methods can provide significant support and form the basis for the development of digital twins for nuclear reactors, with the aim of achieving safety standards never before reached through data analysis and real-time control.