A decomposition principle has been proposed for refueling optimization in fast breeder reactors (FBRs). In refueling optimization a total multistage decision problem with nonlinear programming is decomposed into many partial problems with solvable size and linear programming by taking advantage of the FBR physics. First, the problem is decomposed into determinations of the number of refueling subassemblies in concentric annular core zones and the subassembly-by-subassembly refueling patterns. Second, the latter process is further decomposed into the determination of the refueling patterns in the equilibrium cycles and the transition cycles. Third, the simultaneous determination of the refueling patterns throughout multicycles over the plant lifetime is decomposed into a consecutive cycle-by-cycle determination. Fourth, the linear programming problems are decomposed into a sequence of smaller ones by using a decomposition algorithm for solving large-size programming.

The number of fresh fuel subassemblies added at each cycle in the initial loading core through the equilibrium cycles is optimized in concentric annular refueling zones of the core. The optimization is carried out so as to maximize the average discharge burnup subject to nuclear and thermal constraints by using linear programming with an application of a revised simplex method.

The refueling pattern at each cycle is determined by treating each fuel subassembly separately and by minimizing power peaking subject to the number of refueling subassemblies determined in the previous step. After optimizing the refueling patterns in the equilibrium cycles, the transition cycle patterns are determined cycle by cycle to match the equilibrium cycle patterns by means of an implicit enumeration method. An explicit formulation is worked out for the implicit enumeration method, which makes it possible to determine the refueling patterns cycle by cycle.