Application of Real, Adjoint and Bilinear Weighting for Collapsing Group Constants Used in Space Neutron Diffusion Problems

摘要

Conventional collapsing for group cross sections used in multigroup nuclear reactor calculations is usually performed using normal (real; direct) flux weighting. The application of more advanced collapsing procedures using in an appropriate manner real, adjoint and bilinear weighting was in the past restricted in general to fundamental mode problems. Although the principles have been published for more than ten years, there seems to exist little recent experience on the merits and possible difficulties of these improved procedures for multidimensional diffusion problems for practical purposes, e.g. in the nuclear design and analysis of large Liquid Metal Fast Breeder Reactors (LMFBRs). The present work indicates the nature of the problems which could possibly be encountered in applying these procedures by tracing them back to the known close correspondence between group collapsing and synthesis methods. It tries to explain certain somewhat unusual features of the collapsed group constants obtained by adjoint and bilinear weighting and describes the experience gained in representative 1-dim. and 2-dim. test cases. It could be shown for criticality and perturbation calculations that in general it is advantageous to apply these improved collapsing methods if the necessary precautions are taken. Compared to the conventional collapsing procedures these improved procedures are especially useful for multidimensional problems because their application is well suited for that purpose. In the present study it could be proven that they are favorable with respect to computer time and storage needed due to the fact that the necessary number of coarse groups can be kept fairly small without deteriorating too much the accuracy and reliability of the coarse group results compared to reference results of corresponding fine group calculations with uncollapsed group constants. (ERA citation 08:039730)