Speaker
Jia Li
(School of Nuclear Science and Technology)
Description
In order to control the global sample frequency, GVR method is deemed to be a practical way. But it is common that GVR method needs too many steps of weigh window iteration and it may fall into a long-history problem. We introduce a novel approach that is GVR method combined with reduced density in model, which could improve the calculation efficiency of GVR method in the following two aspects. One is that if the density were reduced by a reasonable factor, more neutrons would transport in the regions where far away from plasma. As a result, more flux results and more feasible GVR weight window would be obtained. It accelerates the iteration of GVR method by decreasing the steps of iteration. Secondly, using the reduced density weight windows could lead to smoother gradients in the neutron flux results. Thus, the gradients of weight windows are smoother accordingly. This is important as the very steep gradients in importance map that may lead to over splitting phenomenon or a long-history problem, which has detrimental effects on the parallel efficiency. Through some simulation experiments, this novel methods perform well, ensuring fewer steps of iteration and less possibility for long- history problem.
Co-authors
Jia Li
(School of Nuclear Science and Technology, University of Science and Technology of China, Hefei, China)
Xingchen Nie
(School of Nuclear Science and Technology, University of Science and Technology of China, Hefei, China)
