Acceleration and Parallelization of the Method of Characteristics for Lattice and Whole-Core Heterogeneous Calculations

Abstract

The CRX code based on the method of characteristics extended to treat anisotropic scattering was accelerated with the coarse mesh rebalance (CMR) method for inner iteration and the coarse mesh/coarse group rebalance(CGR) method for outer iteration involved with the eigenvalue search. A parallelization scheme in angular domain rather than in spatial domain was also implemented for further reduction of the computing time. To effectively handle heterogeneous large scale problems, the original ray tracing scheme in the CRX code was modified to modular ray tracing so that all lattice cells have the same ray distribution for each direction. This modification reduces effectively the computer memory and computing time in the ray tracing. Therefore, it is possible to treat the large scale problems such as multi-assembly problems and whole core problems. The acceleration and parallelization methods were applied to several fuel assembly problems and a whole core problem to show their effectiveness. The numerical results show that they are effective in reducing the number of iterations and the computing time.