OPTIMIZATION OF LOAD-FOLLOW OPERATIONS OF A 1300MW PRESSURIZED WATER REACTOR USING EVOLUTIONNARY ALGORITHMS

¹ DEN - Service d’études des réacteurs et de mathématiques appliquées (SERMA) CEA, Université Paris-Saclay, F-91191 Gif-sur-Yvette, France
² Université du Littoral Côte d’Opale (ULCO), France

valentin.drouet@cea.fr
jean-michel.do@cea.fr
verel@univ-littoral.fr

Published online: 22 February 2021

Abstract

Because of the increase of intermittent renewable energies, load follow operations for French PWR will be crucial in the years to come. The goal of this study is to make realistic changes to the plant operations in order to optimize load-following without disrupting the plant. 6 discrete parameters were considered among the overlaps, speeds and maneuvering bands of the control rods. A simulator oriented model of the pressurized reactor based on APOLLO3® is used. It includes 3D neutronics calculations with point kinetics and a 0D model of the secondary system. The operating mode (G mode) was modeled, to account for the human operator of the power plant. Two objectives were considered so as to both minimize the volume of effluents generated during the transient, and maximize the core axial stability. The reaction of the power plant to a load follow transient varies greatly during the operating cycle, because of fuel depletion effects. Therefore, 4 burnup points are considered and the objectives are computed for each point and then reduced to two “whole cycle” objectives. A biobjective massively parallel asynchronous master worker evolutionary algorithm based on AMS-MOEA/D was implemented. It is a highly exploratory algorithm, suitable for black-box problems with an important computing time. The analysis of the performances of this algorithm shows that it is able to find a diversified Pareto front, with solutions that greatly improve the load follow operations for all burnup points compared to the standard control rod parameters.