A monte carlo-deterministic hybrid method for fast reactor analysis

Abstract

This thesis investigates a hybrid Monte Carlo-Deterministic method for the analysis of fast reactors. In this method, the effective multi-group cross sections and high-order Legendre scattering cross sections are generated by using the collision estimator in the MCNP5 code in combination with an added data generation module. In both the new hybrid and conventional methods, the cross section data is based on a simple RZ homogenous core model. As a computational model, a 300MWe SFR (sodium-cooled fast reactor) TRU burner core has been introduced. The generated multi-group cross sections are used by a 3-D diffusion core analyzer to calculate the resulting k-eff and assembly power distribution. A whole core calculation of the heterogeneous core model is performed using MCNP5 in order to determine the reference solution for the evaluation of the analysis methods. For an in-depth verification of the hybrid method in fast reactors, the new method was also applied to both unrodded and heavily rodded cores. In the case of rodded core, a new core modeling named RRZ was proposed for a better modeling of the self-shielding effect of the control assembly in the fast reactor. Additionally, the nodal equivalence theory was successfully applied to fast reactor analysis for the first time in this work. To apply the nodal equivalence theory, a simple 1-D spectral geometry was developed to determine the flux discontinuity factor of a control assembly region on the interface between fuel and control assembly regions. The generated group-wise DF values were used to correct the cross sections of the control assembly region. Moreover, the sensitivity of the DF to the 1-D spectral geometry model was also evaluated in this work. It is concluded that the hybrid method works well for the analysis of the SFR core. Particularly, the special application of the nodal equivalence theory greatly improves the accuracy of the new hybrid method for fast reactor analysis.