Application and comparison of global variance reduction methods employed in Monte Carlo neutron transport for CFETR

Abstract Background: For the detailed Monte Carlo radiation transport, which possesses global characteristics, traditional MCNP Localized Variance Reduction (LVR) methods can hardly perform well. Global Variance Reduction (GVR) method is an effective way. Purpose: This study aims to decrease the global variance of Monte Carlo radiation transport calculations for Chinese Fusion Engineering Testing Reactor (CFETR). Methods: According to different information of neutron distribution, six kinds of GVR methods were implied and compared in a cylinder CFETR neutronics model. Global figure of merit (FOM_G), standard deviation (σ) and the percentage of scoring were analyzed. Based on predecessors' research, other kinds of new GVR methods (energy, tracks) were applied in this calculation. Results: In the point of GVR methods, FOM_G increased dramatically, spatial variation in relative errors was more gradual, and the scoring was higher, compared with AN. The GVR method based on neutron flux achieved the global calculation efficiency of 6.43 times that of AN. Apart from that, the GVR method based on neutron energy can also be a practical GVR kind, achieving the global calculation efficiency of 5.11 times that of AN. Conclusion: The GVR method based on neutron flux can performed best that has a well function of global variance reduction, achieving the global calculation efficiency. (authors)