Boiling water reactor (BWR) bottom reflector calculations in lattice codes such as CASMO are presently used only to produce accurate boundary conditions for core interfaces in nodal diffusion codes. Homogenized cross-section constants and discontinuity factors are calculated in one dimension (1-D) without the explicit presence of the control rod absorber. If the spatial flux in a BWR bottom reflector is required, for example, for depletion calculations of withdrawn control rods, the homogenization of the reflector must be based on a representation of the three-dimensional (3-D) geometry and material composition that is as true as possible.

This paper investigates differences in cross-section and discontinuity factors from 1-D calculations in CASMO with 3-D Monte Carlo calculations of a realistic bottom reflector model in MCNP5. The cross-section and discontinuity factors from CASMO and MCNP5 are furthermore implemented in the nodal diffusion code SIMULATE5 to investigate the effect on the neutron fluxes in the bottom reflector.

The results show that for the case investigated, the 1-D homogenization in CASMO5 produces a 26% overestimation of the homogenized thermal absorption cross section in the reflector and a 62% underestimation of the homogenized fast absorption cross section. These cross-section differences have essentially no impact on the neutron flux in the core but cause a 4.5% and 12.3% underestimation of the thermal and fast neutron flux, respectively, in the reflector region.