Stochastic approach in the optimization of A nuclear power system

The paper deals with a CANDU-CANDU (Th/Pu)-LMFBR (PuO₂) nuclear power system which evolves in a finite time interval. Its initial evolution is only in the CANDU variant, and subsequently in the variants CANDU (Th/Pu) and LMFBR (PuO₂) by the use of Pu produced in the system. It is assumed that the fuel burn-up in the LMFBR (PuO₂) reactors is a random value, as it is governed by an a priori determined field of probability. The resources of natural uranium and Pu which severely influence the development of the system are also random, as they cannot be definitely known, and moreover they are actually governed by another field of probability already known. Under these conditions, the set of optimal solutions and associated optimal values represented by the nuclear electric powers released in the system at the end of the considered time interval have to be derived. Concomitantly, the distribution of the optimal value, its average value and standard deviation can be evaluated. This type of stochastic approach to nuclear power system optimization is much more valid than the deterministic approach, as it supplies information of interest for the decision-makers engaged in the solution of a nuclear power policy.