Radiation Transport and Criticality Group

Monte Carlo methods are used to compute fluxes or dose rates over large areas using mesh tallies. For problems that demand that the uncertainty in each mesh cell be less than some set maximum, computation time is controlled by the cell with the largest uncertainty. This issue becomes quite troublesome in deep-penetration problems, and advanced variance reduction techniques are required to obtain reasonable uncertainties over large areas.[1] Oak Ridge National Laboratory has developed a new sequence, MAVRIC, which will be available with the release of SCALE 6. In this sequence, a methodology called Consistent Adjoint-Driven Importance Sampling (CADIS) has been incorporated for effective variance reduction. The CADIS methodology was developed to quickly and automatically determine the biased source distribution, as well as all of the values for the weight windows over a rectangular mesh and a given energy group structure.[2-3] The method first determines the approximate adjoint particle flux, usually using a discrete ordinates code. The source for this calculation is the detector energy-group response for the process of interest (e.g., dose rate) at the location(s) of interest. The resulting adjoint flux at each location and energy is equated to the importance of particles and is combined with the source distribution to generate the biased source and weigh