A New Global Variance Reduction Technique Based on Geometry and Energy Splitting/Roulette
Abstract With the increase of computer resources and the application of global variance reduction (GVR) method, it is a trend to obtain global distribution using the Monte Carlo (MC) method in deep-penetration shielding calculation. GVR technique uses biased source and weight window to decrease the MC calculation tally error for deep-penetration problems. However, excessive splitting of large weight particles out of the source region is time-consuming for problems with significant fluence rate variation. A new GVR technique, which performs space and energy splitting/roulette based on the importance of the phase space, is proposed in this paper to avoid excessive splitting of large weight particles. The improved GVR technique is applied to the H. B. Robinson Unit 2 (HBR-2) benchmark and CAP1400 dose rate distribution calculation. Numerical results show that the new GVR technique shows excellent performance for deep-penetration shielding calculation.