Research on a Hybrid Stochastic-Deterministic Method for Neutron Transport Analysis

Abstract

To solve the neutron transport equation, the deterministic method discretizes the space-angle-energy phase-space to obtain the algebra equations. It is superior in computing speed, but short at resonance energy range self-shielding treatment due to multi-group approximation. In contrast, the stochastic or Monte Carlo method transforms the differential-integral equation into an integral form to construct the sampling space for continuous phase-space simulation. It can guarantee the computing accuracy with sufficient samples, but requires large computing time to simulate these samples. To combine the advantages of these two methods, a hybrid deterministic-stochastic method in incident-neutron energy is investigated in this paper. Considering the geometry handling ability, the Method Of Characteristic (MOC) was employed to treat the fast and thermal energy ranges in which cross-sections slowly vary along with the incident-neutron energy. Meanwhile, the Monte Carlo method was employed to deal with the middle energy range where severe resonance effect appears. Different energy ranges are coupled through the neutron scattering and fission contributions between each other. Encouraging conclusions can be demonstrated by the numerical results. (1) The multi-group data library for deterministic calculation in fast and thermal energy ranges and the ACE data library for stochastic calculation in resonance energy range are consistent if they are made from the same evaluated data library. (2) The transport correction employed by the deterministic calculation in fast energy range can handle the strong anisotropic scattering effect in the investigated hybrid method in this paper. (3) The hybrid method can combine the advantages of the deterministic and stochastic methods.