Modeling detector response curves for a high-fidelity uranium measurement for use in simulations

Abstract

Enrichment measurements using identiFinder2 (‘HM5’) detector are performed on two Material Testing Reactor (MTR) fuel assemblies - one with low-enriched uranium plates (LEU) and another with high-enriched uranium plates (HEU). The effectiveness of International Atomic Energy Agency (IAEA) inspection plans for verifying nuclear material strata, in the form of defect detection probability (DP), is evaluated using statistical models. These models use defect identification probability (IP) curves, which represent the probability that a measured item is identified as a defective item. This paper describes a new modeling procedure that converts the experimental measurements into IP curves and employing such experimentally derived IP curves within DP simulations will better represent the experimental conditions like material type, material distribution, and plate configuration. A comparison of experimental performance curves to a simple statistical model (Gaussian, 15% relative standard deviation RSD) shows that the DP results from the modeled response of HM5 measurements better captures the experimental conditions. This result highlights a need for further research into experimental error variables test model development as use of a simple model does not adequately capture true performance in either the LEU or HEU cases.