Fuel reloads optimization for TRIGA research reactor using Genetic Algorithm coupled with neutronic and thermal-hydraulic codes

Abstract This paper presents a case study of applying Genetic Algorithm (GA) coupled with Monte Carlo N-Particle Transport (MCNP) and PARET codes for a thermal-hydraulic and safety analysis to optimize the fuel reload for the TRIGA Mark II Moroccan research reactor. Based on the radial distribution of the 238U burnup ratio inside the reactor core, the five most burned fuel elements were replaced by others fresh fuel elements (12 % wt of uranium) using the Multi-Objective Genetic Algorithms (MOGA) method. Three aspects for the fuel reload optimization were considered in this study including 1) maximization of the effective multiplication factor (Keff), 2) minimization of maximum Centre Fuel Temperature (CFT) and 3) maximization of the Departure from Nuclear Boiling Ratio (DNBR). The GA programming process developed in this work was adapted to handle the constraints concerning the safety limits for the successive core configurations (CCs) automatically generated by the code. MOGA method works with an elitist selection based on the Binary Tournament Selection (BTS) method, a modified two-point crossover and a simple mutation operator. The results obtained indicate that the MOGA can successfully find an optimal CC with a Keff of 1.03498, a maximum CFT of 554 °C and a DNBR of 2.94 when five fresh fuel elements are inserted. The variation of neutron fluxes with respect to radial distance for the best CC and the fresh core was illustrated.