Solving benchmark problems is a significant step in the validation of numerical simulation programs. The Experimental Breeder Reactor II (EBR-II) is a famous benchmark for sodium-cooled fast reactors (SFR), with a complicated core configuration and spatial distribution of nuclide density, hence the modeling difficulty and computational cost of its fine numerical models are relatively high. Therefore, simplified model that mixing the spatial distribution of nuclide density by the component type is adopted in many studies on EBR-II benchmark calculation.

This study aims to contrast the difference between the results of the fine model and the simplified model, evaluating the rationality of the simplification.

In this study, both the fine model and the simplified one were built using LoongSARAX, a neutronic numerical program for fast reactors developed by Xi'an Jiao Tong University. Some approximations were applied to these two models, i.e., one-dimensional homogenization adopted for the half-worth driver assembly to handle its complex radial geometry and the super-assembly method adopted in the cross-section generation of poison elements. Finally, deviation between neutron physical properties in two models was evaluated in terms of computation time consumption, effective multiplication factor, neutron flux density distribution.

The evaluation results show that the spatial distribution of fuel nuclide density presents strong asymmetry and strong non-uniformity in the simplified model, and calculation time spent in the simplified model is one-tenth of that in the fine model. Compared with the fine model, the effective multiplication factor (k_eff) is 1.383×10^-2 lower than in the fine model and the spatial distribution of neutron flux is lower in the center and higher in the outer core whereas the maximum relative deviation between neutron flux in two models is 4.25%.

This study demonstrates that the simplified model has a much lower calculation cost but limited numerical accuracy in k_eff and neutron flux, hence it is still necessary to adopt the fine model when necessary.