Variance Reduction Techniques Using Adjoint Monte Carlo Method in Shielding Problem

Kohtaro UEKI; P. N. STEVENS

doi:10.3327/jnst.16.117

Author information

Keywords: shielding, neutron flux, Monte Carlo method, forward Monte Carlo method, adjoint Monte Carlo method, iterative forward-adjoint Monte Carlo method, importance function, event value, point value, path-length biasing, an-gular biasing, MORSE code, variance reduction, fractional standard deviation, efficiency, accuracy

JOURNAL FREE ACCESS

1979 Volume 16 Issue 2 Pages 117-131

DOI https://doi.org/10.3327/jnst.16.117

Details

Abstract

The event value Wg (r, Ω) has been derived in a form which can be obtained directly from existing adjoint Monte Carlo computer codes. It is demonstrated that the event value and the point value functions obtained from the adjoint Monte Carlo calculation can be used as the path-length biasing and the angular biasing in the forward Monte Carlo calculation, respectively. The iterative forward-adjoint Monte Carlo method using the source biasings is employed to reduce the standard deviation in the shielding problem. In addition, the effectiveness of the source biasing schemes is investigated in the same problem. The results indicate a significant reduction in the standard deviation and a substantial improvement in the efficiency of the Monte Carlo shielding calculations.

Corresponding author

Register with J-STAGE for free!