Heterogeneous 3-D SN transport reactor calculations using ATTILA

Abstract of the technical presented to: 
Joint International Conference on Supercomputing in Nuclear Applications and Monte Carlo 2010 (SNA + MC2010)
Tokyo, Japan
October 17-20, 2010

Prepared by:
Robert Pawel Rulko^a, Mohamed Belal^b, Djordje Ilija Tomasevic^b
a Canadian Nuclear Safety Commission, Ottawa, Ontario, Canada
^b Nuclear Energy Corporation of South Africa,  Pretoria 0001, South Africa

Abstract

The Canadian Nuclear Safety Commission (CNSC) is preparing to license diverse reactor technologies (CANDU, LWR, small/research reactors), if required. This implies the need to strengthen its means of independent nuclear design verifications. To this end, the CNSC has acquired the Attila SN transport code to serve as an effective tool in the design review of new reactors, licensing analysis of CANDU and small/research reactors (NRU, MNR, and SLOWPOKE).

In this paper, we are presenting 3-D large scale parallel benchmark calculations of a small PWR with MOX using Attila SN transport code and their comparison to MCNP Monte Carlo. Our benchmark is that of Nam Zin Cho et al [Nam Zin Cho et al, Whole-Core Heterogeneous Transport Calculations and Their Comparison with Diffusion Results, Trans American Nucl Soc, Vol.83, Washington D.C, November 12-16 (2000)]. Our SN calculations performed on a dual core four processors HP desktop workstation (xw9400 WS, 64Opteron AMD) transform Cho's et al benchmark into a new 3-D hexagonal geometry heterogeneous benchmark. In addition, they provide an evaluation of Attila code in complex calculations of power reactor core with MOX and hence confirm its potential for use as a reference reactor design tool as well as a benchmarking tool in licensing applications.

In this benchmark, we computed the keff using Attila in configuration with Control Rods IN as well as generated the assembly and pin powers. The pins selected were fuel and burnable absorber pins extending diagonally from core center across different MOX enrichment zones and the neighbouring UOX assembly, i.e., the pins placed in strong transport boundary layer effect zones. As a reference solution, the Monte Carlo MCNP calculations were obtained. The MCNP model was identical to Attila model to assure consistency.

The results show that the full core parallel heterogeneous 3-D SN transport calculations of a power reactor are feasible on a small desktop workstation with four processors. Our keff results are of benchmark quality and are within 0.06 percent (60ppm) relative difference to MCNP reference result (0.9990). The assembly and pin power results are equally impressive. These results validate the Attila code for nuclear design work and licensing verification. In addition, details of generating the 3D Attila SN solution of the benchmark, performance of solver numerical options and convergence performance of Attila's linearly discontinuous finite element scheme subject to different acceleration techniques are given.

To obtain a copy of the abstract's document, contact the CNSC. When contacting the CNSC, please provide the title and date of the abstract.