Book Description

Neutron transport through a special case stochastic mixture is examined, in which spheres of constant radius are uniformly mixed in a matrix material. A Monte Carlo algorithm previously proposed and examined in 2-D has been implemented in a test version of MCNP. The Limited Chord Length Sampling (LCLS) technique provides a means for modeling a binary stochastic mixture as a cell in MCNP. When inside a matrix cell, LCLS uses chord-length sampling to sample the distance to the next stochastic sphere. After a surface crossing into a stochastic sphere, transport is treated explicitly until the particle exits or is killed. Results were computed for a simple model with two different fixed neutron source distributions and three sets of material number densities. Stochastic spheres were modeled as black absorbers and varying degrees of scattering were introduced in the matrix material. Tallies were computed using the LCLS capability and by averaging results obtained from multiple realizations of the random geometry. Results were compared for accuracy and figures of merit were compared to indicate the efficiency gain of the LCLS method over the benchmark method. Results show that LCLS provides very good accuracy if the scattering optical thickness of the matrix is small (≤ 1). Comparisons of figures of merit show an advantage to LCLS varying between factors of 141 and 5. LCLS efficiency and accuracy relative to the benchmark both decrease as scattering is increased in the matrix.