Project Details
The AutomateD VAriaNce reducTion Generator (ADVANTG) software is developed, maintained, and distributed by the Reactor and Nuclear Systems Division (RNSD) of the Oak Ridge National Laboratory (ORNL) to extend the capabilities of the Los Alamos National Laboratory MCNP code in continuous-energy Monte Carlo simulations of radiation shielding and related applications. ADVANTG automates the process of generating space- and energy-dependent variance reduction parameters that enable MCNP simulations to obtain uniformly converged tallies over arbitrary regions of interest including the whole problem domain. Weight-window bounds and biased source distributions are efficiently generated based on 3-D discrete ordinates solutions of the adjoint transport equation using the parallel SCALE Denovo solver. Recent applications of ADVANTG include: the estimation of light water reactor (LWR) pressure vessel fluence (illustrated below), full-field dose-rate distributions throughout the ITER fusion tokamak facility, and high-resolution space- and energy-dependent neutron flux distributions for activation analyses of LWR containment structures for decommissioning planning.
The parallel Multi-Step X and Neutron Activation Gamma Source Sampler (MSX/NAGSS) software performs voxel-by-voxel or cell-by-cell activation analysis and gamma source transport using the SCALE Origen activation/depletion solver and MCNP. When used with ADVANTG, MSX also provides the capability to generate variance reduction parameters that efficiently optimize the neutron simulation to obtain well-converged end-of-history photon responses at locations of interest. MSX/NAGSS performs activation calculations in parallel without the need for input/output file processing by linking directly to the Origen C++ library. In a recent application, MSX/NAGSS was used to perform material activation and gamma source calculations on a 10^8-voxel mesh using neutron fluxes estimated on the 175-group VITAMIN J structure.
The ORNL Transformative Neutronics (ORNL-TN) upgrade to MCNP5 was developed to enable high-resolution mesh tallies on extremely complex geometry models for US ITER. ORNL-TN implements an efficient shared-memory mesh tally capability for multithreaded simulations. In simulations performed on RNSD cluster, this capability increased in-memory mesh tally capacity from order 10^8 to 10^10 space/energy bins. ORNL-TN also provides fast initialization of highly detailed geometry models, such as those converted from multiple CAD models, to streamline modeling and analysis efforts. In applications using models with more than 10^5 geometry cells, initialization time was reduced from several hours to a few seconds.