Rapid Analysis of the SNM Smuggling Threat Space for Active Interrogation Using a Green's Function Approach

essential for ensuring the security of the United States from radiological threats. Radiation transport simulation is a widely-used tool for evaluating the probability of detection given some SNM, smuggler strategy, detector system, and alarm algorithm. Inherent in these calculations is the tradeoff between the speed and the fidelity of the computation. Full Monte Carlo (MC) radiation transport captures the true physics and geometric richness of the system; however, doing this in a reasonable amount of time for a spanning set of threats requires vast computing resources. Simplifications such as 1-D deterministic approximations can reduce the computation time drastically, but may miss important physical and geometric phenomena. We present a method which uses Green’s Functions, which are computed once up front and are then stored for future use, to rapidly analyze many possible scenarios with the fidelity approaching that of full 3-D MC transport but with a computational time on the order of seconds. Using this technique, we model an active interrogation (AI) photon source incident on a cargo container containing some SNM and analyze the time dependent flux of neutrons at the detector. We present an illustrative application of our technique on the neutron background from cosmic radiation. First, we model cosmic interactions in the atmosphere to obtain the neutron flux as a function of both energy and direction at 150m above the ground. We break the subsequent transport of the neutrons into several distinct regions: neutrons incident on the ground per neutron at 150m, neutrons reflecting back and exiting the ground per neutron incident on the ground, neutrons in the detector per neutron leaving the ground, as well as the direct path of neutrons into the detector per neutron at 150m. We are able to vary parameters such as ground composition, air composition and humidity, detector height above ground, and detector type. With this method we have achieved a 7.3% root mean square difference between our spectral neutron flux and that obtained from a full MCNPX calculation.