Research on Shielding Deep Penetration Calculation Based on MC Variance Reduction Techniques

Abstract

In the study of shielding deep penetration of particle transport, analogy calculation (AN) can lead to a long calculation time and significant error. This paper takes the long vacuum-concrete cylinder as the geometric model, and the particle flux after concrete shielding is taken as the calculation objective. Based on various variance reduction techniques of MC, the application of Biasing the Source, Ring Detector, Geometry splitting, and Russian roulette, Forced Collision, PD card, DXTRAN Sphere, Exponential Transform, and Weight Window techniques in the model are analyzed. And the performance analysis of single and combined variance reduction can reduce the error of the calculation result and improve the calculation efficiency. For the model in this paper, under the same number of particles, the calculation results are compared with those of the non-variance reduction technique: After using a single variance reduction experiment, both IMP and the weight window have a positive effect on the calculation results, and the weight window is better than other techniques to reduce variance, the error in the calculation results can be reduced to 13.37%, and the quality factor FOM increased by nearly 15 times; and in various combination experiments based on IMP and weight window, the variance reduction effect of weight window combination is generally better than that of IMP combination experiment, the best effect of reducing variance is the combination of weight window with DXTRAN and PD card, which can increase FOM to 400 times and reduce the error to 2.77%. The results show that through the analysis and combination of different methods, compared with the single variance reduction techniques, the combination of IMP and the weight window other variance reduction techniques can effectively improve the FOM and reduce the error to achieve the purpose of variance reduction, but the combination of the weight window has better calculation efficiency and objectivity compared with other variance reduction techniques.