An advanced nuclear reactor analysis methodology for heterogeneous cores.

摘要

We have developed an extension of the present-day reactor-analysis methodology that systematically accounts for the effects that different neighbors have on a given assembly's few-group constants. The extension is applicable for use in both core-level diffusion and quasidiffusion calculations. The basis of our extension is the use of branch calculations that generate the effect of unlike neighbors on a given assembly's group constants. These new branch calculations, for the effects of neighboring assemblies, fit into the same branching framework used in present reactor-analysis methods for many other parameters, such as temperature, power, burnup, void, and boron concentration. We perform assembly-level calculations to estimate the change in a given assembly's group constants due to the adjacent assemblies.;One extension of present methods is to use superposition of the effects of neighboring assemblies to reduce the number of branch calculations that are needed to estimate the effects of all possible neighbor permutations. We have identified two 'basic' four-assembly configurations to isolate the effects of a single unlike-neighboring assembly on a given assembly's constants. We have shown that this superposition approximation produces very accurate estimates of the group constants in other configurations.;In addition, we also use energy-, angle-, and position-dependent albedos to simulate the presence of the unlike neighbors in our branch calculations. We have developed and tested a procedure for efficiently estimating these albedos. We have shown that our albedos produce accurate estimates of an assembly's group constants in both of the 'basic' colorsets of the superposition approximation. We have implemented these albedos and used the resulting few-group cross sections in a full-core analysis, using both diffusion and quasidiffusion, to demonstrate substantial improvement in the core eigenvalue and the reconstructed pin-by-pin power distribution.