No abstract

No abstract

In the nuclear reactor core, particularly in that of the Liquid Metal Cooled Fast Breeder Reactor, it is required that local incidents be detected in their earliest possible phases. For this purpose, it is important to design an efficient in-core monitoring system. Systems that are too complicated, however, are liable to cause unwarranted diminution of plant availability. In this paper are considered two problems, that represent different approaches to this thesis: one to find a means of selecting the optimum locations for the detectors, and the other to find the same for optimizing the sampling rates of the signals from a large number of in-core local detectors. These two approaches are studied as "decision under uncertainty" problems, using the minimax principle of the two-person zero-sum game theory, the idea being to minimize loss in the worst event. This involves an exhaustive search procedure in the former approach, and a linear programming technique in the latter. It is found that the methods used in this study are effective for designing an efficient in-core monitoring system, and that advantages worthy of note can be expected for them in practical applications, this despite the fact that the values of the parameters used in the numerical examples may not necessarily be reprsentative of actual conditions.

No abstract

No abstract