Task Introduction

Task Introduction#

Variance reduction techniques are essential for efficiently simulating deep-penetration shielding problems where analogue Monte Carlo would require prohibitively many particles. In this task you will use OpenMC to apply survival biasing, weight windows and FW-CADIS to shielding and sphere problems.

A detailed description of each method can be found in the OpenMC documentation. OpenMC also supports methods of generating weight windows including the Magic Method and FW-CADIS.

The examples cover all methods for completeness, however if you only have time to learn one method then the FW-CADIS approach is recommended.

Learning Outcomes

  • Survival biasing prevents particles from being killed by absorption, instead reducing their weight, which improves statistics in deep-penetration problems.

  • Weight windows control particle weights across the geometry, splitting particles in important regions and killing them in unimportant regions.

  • Weight windows can be generated iteratively (per-run or per-batch) using the Magic method.

  • FW-CADIS (Forward-Weighted Consistent Adjoint Driven Importance Sampling) automates weight window generation using an adjoint flux calculation to optimise for a specific tally.