Parallel Genetic Algorithm for Fuel Bundle Burnup Optimization of Pressurized Heavy Water Reactor

Abstract

In this study, parallel genetic algorithm (GA) was applied to solve an optimization problem of nuclear fuel management pertaining to a 220 MWe pressurized heavy water reactor (PHWR). The study aims at finding appropriate values of reference discharge burnups of two zones of the reactor core which gives maximum fuel economy, while satisfying operational and safety-related constraints. This is a multiobjective optimization problem with four objectives and four constraints. Penalty functions-based GA methodology was employed to convert the multi-objective optimization problem to an unconstrained optimization problem by defining a penalty function. Master–slave parallelization method was used to parallelize the GA in order to reduce the computational time by exploring the capabilities of high performance computing cluster. This approach used a single population and evaluation of the individuals of the population was carried out in parallel. In this study, we have run the parallel program by varying the number of CPU cores (processes) and obtained significant performance improvement.

Keywords: parallel genetic algorithm (GA), master–slave GA, nuclear fuel management, Indian PHWR, high performance computing

 Cite this Article

Ramachandran S, Jayalal ML, Jehadeesan R et al. Parallel Genetic Algorithm for Fuel Bundle Burnup Optimization of Pressurized Heavy Water Reactor. Journal of Nuclear Engineering and Technology. 2016; 6(1): 11–24p.