Year
2010
Abstract
An array of detection technologies are deployed to combat illicit tra?cking of special nuclear ma- terial (SNM). A widely-used detection architecture employs radiation portal monitors (RPMs) in which one or more radiation detectors are installed at checkpoints. The e?ectiveness of these systems may be computationally estimated by radiation transport analysis of an array of threat scenarios. The variables that de?ne scenarios, such as the source, container and cargo type, and detector technology, create a problem space that is computationally intractable by brute force methods. We present a method that decomposes the problem into submodels, each capturing a ma jor variable in the problem. Green’s functions are generated for each submodel and recombined to simulate the full transport problem. In addition, the Green’s functions are parameterized with respect to key values in the submodels. By sampling many scenarios and applying alarm algorithms to the computed detector responses, families of receiver operator characteristic (ROC) curves may be derived and a holistic picture of detector e?ectiveness emerges. Parametric studies are presented to demonstrate the capabilities of this method.