Search

Abstract: This paper is concerned with the Indian design of a 220MWe Pressurized Heavy Water Reactor having Natural Uranium fuel and heavy water as moderator and coolant. At the beginning of life, it is necessary to flatten the power by loading some Thorium bundles to achieve a nearly full power operation. The determination of best possible locations of Thorium bundles, which maximize fuel economy as well as safety, is a complex combinatorial optimization problem. About two decades ago, an optimum configuration of Thorium bundles was successfully arrived at by using a gradient based method and this pattern was actually loaded in the Indian PHWR at Kakrapar which went critical in 1992 [Balakrishnan, K., Kakodkar, A., 1994. Optimization of the initial fuel loading of the Indian PHWR with Thorium bundles for achieving full power. Annals of Nuclear Energy 21, 1–9]. Here, the same problem is revisited for two reasons. Firstly, computational techniques based on completely different philosophy namely “Genetic Algorithm” (GA) and “Estimation of Distribution Algorithm” (EDA) have been used. Secondly, the enormous increase in computing power during the last two decades is expected to provide a more exhaustive search. Indeed, it has been possible to find out many feasible Thorium configurations of comparable merit. Our results are similar with the result of the earlier BARC study but provide a range of additional configurations. As in earlier BARC work, we find that one can get from 95% to 97% full power without violating various safety aspects such as maximum bundle power, maximum channel power, channel outlet temperature and worth of the two shutdown systems. In the present work, the number of Thorium bundles which can be loaded range from 22 to 34. One of the outcomes of this study is that the computational techniques suitable for this type of problems have been identified and developed. Further studies involving the use of some other evolutionary methods and problems such as optimization of depleted Uranium loading are in progress. [Copyright &y& Elsevier]