Tiger-3D: 2D/1D Coupled Whole-core Transport Code Based on Large-Scale Parallel Computation

Wenbin Wu,Qing Li,Kan Wang

2015-01-01

Abstract:With the fast advance of computing technology, a number of researchers [1-4] have recently developed 2-D/1-D coupled methods for 3-D neutron transport. Typical reactors are much heterogeneous in the radial direction while relatively homogenous in the axial direction. 2-D/1-D methods take advantage of this geometric feature, thus obtaining good accuracy and efficiency. In 2-D/1-D methods, 3-D problems are divided axially into homogenous 2-D planes, which are solved by 2-D method of characteristics. Then 2-D MOC is coupled with 1-D diffusion or transport solvers in the framework of 3-D coarse mesh finite difference (CMFD) formulation. Generally, 2-D planes are solved independently, and this axial parallel degree is adopted in most 2-D/1-D methods. However, the number of 2-D planes is about 30 for typical reactors, which is too small to utilize large-scale clusters. Some codes like DeCART adopt parallel features of angles in 2-D MOC simultaneously. However, the number of cores being used is still much smaller compared with that modern supercomputers could provide. It’s therefore desired to develop a 2D/1-D method based on large-scale parallel computation. In this work, except for the axial parallel degree, spatial domain decomposition is radially adopted. Large-scale parallel computation is realized by combining parallel degrees of both directions of 2D/1-D method. In this summary, we provide basic theory and preliminary numerical results.