American Nuclear Society
Home

Home / Store / Journals / Electronic Articles / Nuclear Science and Engineering / Volume 141 / Number 3 / Pages 190-217

Object-Oriented Three-Dimensional Fine-Mesh Transport Calculation on Parallel/Distributed Environments for Advanced Reactor Core Analyses

Masahiro Tatsumi, Akio Yamamoto

Nuclear Science and Engineering / Volume 141 / Number 3 / Pages 190-217

July 2002

Format:

Price:$30.00
Member Price:$27.00
Member Savings:$3.00

This paper describes a comprehensive study on the feasibility of advanced reactor core analyses within the framework of pin-by-pin multigroup transport calculations using a prototype of the object-oriented parallel core calculation code SCOPE. The SCOPE code enables the coupling of the diffusion theory method and the SPN transport theory method. The formulation of the method coupling and its verifications with benchmarks are presented.

Quantitative estimation of the pin-cell homogenization effects within the octant core geometry of a three-loop-type pressurized water reactor (PWR) was performed. Comparisons between results by heterogeneous and homogeneous calculations revealed the effects on pin-cell homogenization in large-scale geometry. In order to preserve the neutronic property in the heterogeneous calculation within the framework of pin-cell homogenized pin-by-pin calculations, the applicability of the homogenized cross section corrected by the superhomogénéisation SPH method was studied. It was found that the pin-by-pin nine-group calculation by the SP3 transport theory method with the SPH-corrected cross sections gave good accuracy for the pin power distribution approximately <1% of the root-mean-square error. The calculation accuracy of the transport calculation and the effectiveness of the method coupling were also demonstrated through analyses of the initial core of an identical three-loop-type PWR.

With fine-grained parallelism, the identical convergence property was obtained regardless of the number of processors. Parallel performance was almost scalable up to eight processors, 93% with eight processors in three-dimensional nine-group fine-mesh transport calculations with meshes of 180 × 180 × 30.

 
 
 
Questions or comments about the site? Contact the ANS Webmaster.
advertisement