INTRODUCTION

Scanning arriving cargo containers for illicit nuclear material is a current goal of the U.S. Homeland Security Department. This goal addresses shipments by trucks, airplanes, large cargo ships, and small vessels. One method to detect fissionable material currently under study by the U.S. Department of Energy and other agencies is active interrogation-a system that uses a radiation source, such as a collimated beam of neutrons or photons, to scan cargo containers and detect the products of induced fissions from any fissionable material. Getting particles to the fissionable material and then detecting the fissions can be extremely difficult because of attenuation in the materials themselves and the large distances between the source, the fissionable material, and the detector(s).

Computer simulation is essential to designing effective detection systems, a process that requires evaluating a wide variety of sources, detectors, designs of cargo containers, and possible materials that could be used to conceal nuclear material. Analogous to the physical radiation detection difficulties described above, computer simulations of radiation transport have difficulty calculating how many particles travel from the source to the fissionable material and then back to the detector, compared with the number of particles that interact elsewhere in the container. To reduce the statistical uncertainties in these simulations, prohibitively long calculation times are required. Long compute times severely limit the number of different parameters that can be explored in evaluating and designing active interrogation systems. If Monte Carlo simulations of active interrogation systems could be performed hundreds of times faster, dramatic improvements in design would be realized and optimization studies using more-detailed models could be completed.

This paper describes some preliminary work in applying automated variance reduction using hybrid Monte Carlo/deterministic methods. The MAVRIC sequence[1], part of ORNL's SCALE package of codes used for criticality, shielding, and reactor analysis, uses a coarse-mesh discrete ordinates calculation to determine the space- and energy-dependent importance parameters for a detailed Monte Carlo simulation. The CADIS[2] method is used to compute both the target weight windows and a consistent biased source, both functions of space and energy. The MAVRIC sequence is automated -handling the calculations for the variance reduction parameters with only minor additional input from the user - and highly capable in terms of accelerating traditional source-detector problems.

DESCRIPTION OF THE ACTUAL WORK

RESULTS

REFERENCES

1. SCALE: A Modular Code System for Performing Standardized Computer Analyses for Licensing Evaluations, ORNL/TM-2005/39, Version 6, Vols. I-III, January 2009. Available from Radiation Safety Information Computational Center at Oak Ridge National Laboratory as CCC-750.
2. J. C. WAGNER and A. HAGHIGHAT, "Automated Variance Reduction of Monte Carlo Shielding Calculations Using the Discrete Ordinates Adjoint Function," Nuclear Science & Engineering 128, 186 (1998).