A new local variance reduction method based on anti-forward Monte Carlo calculation

Monte Carlo (MC) method is widely used in radiation shielding calculations with the advantages of high fidelity geometry modeling, complex radiation source description and continuous-energy cross sections. The deep-penetration problem prevents the further application of MC method in radiation transport calculation. It is difficult to obtain reliable results for a certain tally due to the poor particle statistics. Thus, effective local variance reduction (LVR) technique must be applied to bias particles toward tally region. In this paper, combined with weight window technique, a new local variance reduction method based on anti-forward calculation has been proposed. The detector tally in forward run is set as source position in anti-forward run. Anti-forward calculation flux has been used to construct weight window parameters. From the numerical test calculations, a speedup of 420 times has been achieved compared to the analog simulation. More computational time has been spent on one Monte Carlo source particle, but more tally information has been collected. The goal of local variance reduction has been achieved. The proposed local variance reduction technique could be a useful tool in MC deep-penetration shielding calculation.