2014SNA + MC 2013 - Joint International Conference on Supercomputing in Nuclear Applications + Monte Carlo
|Number of page(s)||8|
|Section||2. Computational Science: b. Specific Computational Methods and Methodologies|
|Published online||06 June 2014|
Generation and Testing of XS Libraries for VVER Using APOLLO2 and TRIPOLI4
1 INRNE, Sofia 1784, Bulgaria
2 Varna Free University, Varna 9027, Bulgaria
* e-mail: email@example.com
MOC based calculation schemes with APOLLO2 were used to generate few-group cross-section libraries for VVER-1000 at the nodal and pin level. This paper presents an overview of the testing of the schemes and the libraries, as well as the computational aspects. Two major ameliorations are considered: application of new developments in APOLLO2 and multicore computation for an acceptable trade-off between accuracy and efficiency. Two-level Pij-MOC industrial calculation schemes were tested against TRIPOLI4 reference results. Benchmarking of the schemes shows that the higher-order linear surface method of characteristics (LS MOC) is an efficient option for cross-section library generation. There is a significant potential for further refinement of the MOC energy mesh and the MOC parameters with the progress in distributed computing.
A multi-parameter cross-section library for MSLB analysis with homogenized nodes was tested in 2D core simulation with COBAYA3 vs. whole-core TRIPOLI4 solutions on the CEA CCRT HPC system. Pin-by-pin cross-sections and interface discontinuity factors of Black Box Homogenization type were tested in diffusion calculations with COBAYA3 pin-by-pin against transport reference solutions. Good agreement is displayed.
Key words: APOLLO2 MOC / calculation schemes / cross-section libraries / multicore computing
© Owned by the authors, published by EDP Sciences, 2014