A Quantitative Comparison of Loading Pattern Optimization Methods for In-Core Fuel Management of PWR

The performance of several loading pattern(LP) optimization methods was quantitatively compared through a benchmark problem of PWR LP optimization. The simulated annealing(SA) method, the genetic algorithms(GA) method, the direct search(DS) method based on assembly multiple shuffling and the binary exchange(BE) method based on fuel assembly binary exchange were investigated as candidates for the optimization techniques. Hybrid strategy which combined different optimization methods was newly proposed, and the performances of two different new hybrid methods, which combined DS with BE and GA with BE were examined. From the results of the LP optimization benchmark problem, the superiority and inferiority of each method were clarified. Furthermore, it was demonstrated that the GA+BE hybrid strategy performed best among these methods. By combining GA with BE, the weaknesses of these two methods were compensated for with each other and the optimization performance was improved significantly. Therefore, the GA+BE hybrid method is quite effective for the LP optimization problems of PWR.