Core loading pattern optimization of a typical two-loop 300 MWe PWR using Simulated Annealing (SA), novel crossover Genetic Algorithms (GA) and hybrid GA(SA) schemes

• SA and GA based optimization for loading pattern has been carried out.
• The LEOPARD and MCRAC codes for a typical PWR have been used.
• At high annealing rates, the SA shows premature convergence.
• Then novel crossover and mutation operators are proposed in this work.
• Genetic Algorithms exhibit stagnation for small population sizes.

A comparative study of the Simulated Annealing and Genetic Algorithms based optimization of loading pattern with power profile flattening as the goal, has been carried out using the LEOPARD and MCRAC neutronic codes, for a typical 300 MWe PWR. At high annealing rates, Simulated Annealing exhibited tendency towards premature convergence while at low annealing rates, it failed to converge to global minimum. The new ‘batch composition preserving’ Genetic Algorithms with novel crossover and mutation operators are proposed in this work which, consistent with the earlier findings (Yamamoto, 1997), for small population size, require comparable computational effort to Simulated Annealing with medium annealing rates. However, Genetic Algorithms exhibit stagnation for small population size. A hybrid Genetic Algorithms (Simulated Annealing) scheme is proposed that utilizes inner Simulated Annealing layer for further evolution of population at stagnation point. The hybrid scheme has been found to escape stagnation in bcp Genetic Algorithms and converge to the global minima with about 51% more computational effort for small population sizes.