Comparison of Optimizing Loading Patterns on the Basis of SA and PMA Algorithms
Optimization of loading patterns is a very important task from economical point of view in a nuclear power plant. The optimization algorithms used for this purpose can be categorized basically into two categories: deterministic ones and stochastic ones. In the Paks nuclear power plant a deterministic optimization procedure is used to optimize the loading pattern at BOC so that the core would have maximal reactivity reserve.
To the group of stochastic optimization procedures belong mostly simulated annealing (SA) procedures and genetic algorithms (GA). There are new procedures as well which try to combine the advantages of SAs and GAs. One of them is called population mutation annealing algorithm (PMA).
In the Paks NPP we would like to introduce fuel assemblies including burnable poison (Gd) in the near future. In order to be able to find the optimal loading pattern (or near-optimal loading patterns) in that case we have to optimize our core not only for objective functions defined at BOC, but at EOC as well. For this purpose I used stochastic algorithms (SA and PMA) to investigate loading pattern optimization results for different objective functions at BOC and EOC.
In this paper (and at the symposium) I would like to present these results.