PWR Core Loading Pattern Design Assisted by Artificial Intelligence

Abstract

Core design is the a priori study of the behavior of a reactor core throughout a cycle between two fuel reloads. There is a growing interest in using Artificial Intelligence (AI) tools to accelerate this type of calculation. Coupled thermal-hydraulic/neutronic calculations allow access to many variables of special interest to develop a digital twin (metamodel), which can be used, for instance, for pattern optimization, since it presents restrictions from the point of economics, safety and licensing. This work presents a neural network (NN) trained to obtain a metamodel, which will be used to determine different optimal configurations of the fuel elements based on certain criteria.