Improvement in Estimation of Distribution Algorithm (EDA) for Fuel Loading Pattern Optimization in AHWR
The initial core of AHWR-LEU consists of 444 fuel channel locations, and it is proposed to load two differentially reactive clusters (namely, type-1 (more reactive) and type-2) in these locations for flux flattening. This loading pattern optimization problem (LPO) has been solved by applying genetic algorithm (GA) and estimation of distribution algorithm (EDA). It has been observed that genetic algorithm (GA) is more efficient for AHWR initial core LPO problem as compared to EDA. To improve the efficiency of EDA, different modifications have been suggested in this paper. As a first modification, the tournament selection method (similar as in GA) was applied in EDA for diversity. In the second modification, we have tried to improve the EDA by restricting the random numbers in a specified range without compromising randomness.
KeywordsLoading pattern optimization (LPO) Evolutionary algorithm AHWR Genetic algorithm
- 1.A. Thakur, B. Singh, N.P. Pushpam, V. Bharti, U. Kannan, P.D. Krishnani, R.K. Sinha, Fuel cycle flexibility in AHWR with the use of Th-LEU fuel, in Proceedings of International Conference on Future of HWRs, Ontario, Canada, 2011Google Scholar
- 13.A. Kumar, R. Srivenkatesan, Nodal expansion method for reactor core calculation. Technical report no. BARC-1249 (1984)Google Scholar
- 14.A. Thakur, B. Singh, U. Kannan, N.P. Pushpam, A. Kumar, P.D. Krishnani, R. Srivenkatesan, Initial core optimization of AHWR, in Proceedings of National Symposium on Radiation Physics (NSRP-18), Udaipur, India, pp. 139–141, 2010Google Scholar