Issue |
2014
SNA + MC 2013 - Joint International Conference on Supercomputing in Nuclear Applications + Monte Carlo
|
|
---|---|---|
Article Number | 05201 | |
Number of page(s) | 3 | |
Section | 5. Poster Session: b. Computational Science | |
DOI | https://doi.org/10.1051/snamc/201405201 | |
Published online | 06 June 2014 |
Cartesian Meshing Impacts for PWR Assemblies in Multigroup Monte Carlo and Sn Transport
Georgia Institute of Technology, 770 State Street, Boggs Bldg, 3rd Floor, Room 3-58, Atlanta, GA 30332
* Corresponding Author, E-mail: kevinlee@gatech.edu
Hybrid methods of neutron transport have increased greatly in use, for example, in applications of using both Monte Carlo and deterministic transport to calculate quantities of interest, such as flux and eigenvalue in a nuclear reactor. Many 3D parallel Sn codes apply a Cartesian mesh, and thus for nuclear reactors the representation of curved fuels (cylinder, sphere, etc.) are impacted in the representation of proper fuel inventory (both in deviation of mass and exact geometry representation). For a PWR assembly eigenvalue problem, we explore the errors associated with this Cartesian discrete mesh representation, and perform an analysis to calculate a slope parameter that relates the pcm to the percent areal/volumetric deviation (areal corresponds to 2D and volumetric to 3D, respectively). Our initial analysis demonstrates a linear relationship between pcm change and areal/volumetric deviation using Multigroup MCNP on a PWR assembly compared to a reference exact combinatorial MCNP geometry calculation. For the same multigroup problems, we also intend to characterize this linear relationship in discrete ordinates (3D PENTRAN) and discuss issues related to transport cross-comparison. In addition, we discuss auto-conversion techniques with our 3D Cartesian mesh generation tools to allow for full generation of MCNP5 inputs (Cartesian mesh and Multigroup XS) from a basis PENTRAN Sn model.
Key words: 3D / Cartesian / Sn / Multigroup / PWR / Discrete Ordinates / MCNP5 / PENTRAN / Mesh / eigenvalue / pcm
© Owned by the authors, published by EDP Sciences, 2014