Improvements in the long characteristics method and their efficiency for deep penetration calculations

The efficiency of the general-purpose, discrete-ordinates, long-characteris tics, combinatorial geometry transport code MCCG3D for deep penetration cal culation of the void duct problem in the OECD benchmark is examined, and th e ability of the code to generate increasingly accurate solutions for bench marking purposes is demonstrated. This is made possible by the built-in abi lity of MCCG3D to work with very fine space meshes and a very large number of quadrature angles, thus reducing ray effects. Using integration along ch aracteristics, MCCG3D is able to eliminate the numerical transverse dispers ion which contaminates other transport solutions based on discrete ordinate s. Other assets of MCCG3D are a versatile combinatorial-geometry descriptio n, the ability to calculate local angular fluxes with very high accuracy an d the recently added capability for calculating point fluxes. All these fea tures make MCCG3D an outstanding numerical tool for benchmarking. The used combination of options provides very impressive results, both for generatin g benchmark quality solutions (1-% accuracy) and rather fast "reasonable ac curacy" solutions. (C) 2001 Elsevier Science Ltd. All rights reserved.