Logo image
使用蟻群最佳化演算法自動搜尋壓水式反應器爐心佈局
Thesis

使用蟻群最佳化演算法自動搜尋壓水式反應器爐心佈局

林栢楓
Masters, 國立清華大學, 核子工程與科學研究所
2009

Abstract

蟻群演算法 壓水式反應器 排序蟻群系統 燃料佈局 Ant Colony Algorithm pressurized power reactor MMAS RAS Ant-Q loading pattern
In a pressurized water reactor (PWR), the fuel assemblies must be reloaded after operating a period of time. Loading pattern (LP) was designed by engineer based on experience. Because the computer capability is greatly improved in recent years, it is possible to search loading pattern using some algorithms. Ant Colony System is a metaheuristic method which is efficient to combinatorial optimization problem. In this study, three kinds of algorithms, i.e., Rank-Based System (RAS), Max-Min Ant System (MMAS), Ant-Q was adopted to search loading pattern of PWR. Maanshan nuclear power plant cycle 17,18 and 19 was applied to demonstrate the capability of the algorithms. Safety and economy must be considered in loading pattern design. The main safety constraints include hot channel factor (FΔH) and moderator temperature coefficient (MTC) and the economic consideration is cycle length and cost of fuel. The LP was mirror symmetry in quarter-core and quarter-core rotational symmetry in the full core. The search procedure contained two steps. The first step was to permute fuel assemblies (FA) in 1/8 region and the second step was to perform rotation configuration of these FAs. The developed program was coded with C. SIMULATE-3 code was used to calculate core status. In Maanshan Cycle 19 case, most of the loading patterns searched by RAS, MMAS and Ant-Q satisfied the safety limits. Generally speaking, MMAS had better performance. The results of Maanshan Cycle 17, 18 also satisfied safety limits, which showed this tool can be applied to other cycles. In addition, the design whose four FAs were placed by lower enrichment FAs also reached the target cycle length. Also, parallel computing was tested to reduce computation time.

Metrics

1 Record Views

Details

Logo image