ADVANTG Improvements and New Capabilities

Abstract

This report describes the progress and conclusions of a project to upgrade the ADVANTG (AutomateD VAriaNce reducTion Generator) code developed at the Oak Ridge National Laboratory. This effort was primarily motivated by a desire to make the code available to safeguards and non-proliferation users for modeling source/detector problems. The code improvements made in this project also extend the utility of ADVANTG for analyzing MCNP models. ADVANTG uses the Consistent Adjoint Driven Importance Sampling (CADIS) and Forward-Weighted CADIS (FW-CADIS) methods to generate variance reduction parameters for MCNP models. ADVANTG was originally intended for use by radiation-transport experts at ORNL. Although the ADVANTG prototype was very effective at speeding up difficult Monte Carlo problems, it would have been difficult to document and support for external users. The prototype also required significant user burden, making its benefit unclear for moderately difficult problems. Under this project, the ADVANTG prototype has been vastly improved, making accessible for a much larger set of users and easier to maintain. The new version features better integration with MCNP model features to reduce user burden; better interfacing with Denovo to improve the quality of ADVANTG’s variance-reduction parameters while reducing time-to-solution; and other improvements to simplify model debugging and remove unmaintained libraries. The updated version (ADVANTG 3) is available to nonproliferation/safeguards users in a limited beta release. Results generated using ADVANTG 3 are available in the companion report, “ADVANTG Results for a Suite of Radiation Detection, SNM Movement Detection, and Safeguards Test Problems.”