TY - GEN
T1 - COARSE MESH GENERATION FOR CMFD ACCELERATION IN SARAX-LAVENDER CODE
AU - Xu, Haoxiang
AU - Zheng, Youqi
AU - Xiao, Bowen
AU - Wu, Hongchun
N1 - Publisher Copyright:
© 2024 by ASME.
PY - 2024
Y1 - 2024
N2 - A coarse mesh generation algorithm is implemented in SARAX-LAVENDER code based on mesh merging, aiming at providing the foundation for a CMFD acceleration scheme. Originated from Navigation Mesh construction method in Recast code, this algorithm is slightly altered to improve performance on a reactor geometry. The algorithm can handle regular triangular node in structured geometry as well as Delaunay triangle mesh in arbitrary unstructured geometry. Such an algorithm holds several advantages as follows: (1) No extra effort is required during regular triangle or Delaunay triangle mesh generation procedure. (2) The coarse mesh generated by triangle mesh consists of only convex polygons. (3) The boundary surfaces between coarse mesh can be exactly preserved by existing triangular node. (4) The maximum edge number of generated polygons could be easily adjusted to adapt different geometry. When set to 3, the algorithm could transfer triangle mesh directly into coarse mesh format without mesh merging, making it easy to implement a two-level spatial CMFD acceleration scheme. (5) The algorithm naturally adapts the geometry after domain decomposition, which is required for parallel computing. Coarse mesh number ranges from one sixth to one third of original triangle mesh number in hexagonal, rectangle or arbitrary unstructured geometry. Similar reduction could be achieved in a MPI parallel computing condition with CPU number from 2 up to 64.
AB - A coarse mesh generation algorithm is implemented in SARAX-LAVENDER code based on mesh merging, aiming at providing the foundation for a CMFD acceleration scheme. Originated from Navigation Mesh construction method in Recast code, this algorithm is slightly altered to improve performance on a reactor geometry. The algorithm can handle regular triangular node in structured geometry as well as Delaunay triangle mesh in arbitrary unstructured geometry. Such an algorithm holds several advantages as follows: (1) No extra effort is required during regular triangle or Delaunay triangle mesh generation procedure. (2) The coarse mesh generated by triangle mesh consists of only convex polygons. (3) The boundary surfaces between coarse mesh can be exactly preserved by existing triangular node. (4) The maximum edge number of generated polygons could be easily adjusted to adapt different geometry. When set to 3, the algorithm could transfer triangle mesh directly into coarse mesh format without mesh merging, making it easy to implement a two-level spatial CMFD acceleration scheme. (5) The algorithm naturally adapts the geometry after domain decomposition, which is required for parallel computing. Coarse mesh number ranges from one sixth to one third of original triangle mesh number in hexagonal, rectangle or arbitrary unstructured geometry. Similar reduction could be achieved in a MPI parallel computing condition with CPU number from 2 up to 64.
KW - CMFD
KW - NavMesh
KW - arbitrary unstructured geometry
KW - mesh merging
UR - https://www.scopus.com/pages/publications/85209540870
U2 - 10.1115/ICONE31-135744
DO - 10.1115/ICONE31-135744
M3 - 会议稿件
AN - SCOPUS:85209540870
T3 - Proceedings of 2024 31st International Conference on Nuclear Engineering, ICONE 2024
BT - Student Paper Competition
PB - American Society of Mechanical Engineers (ASME)
T2 - 2024 31st International Conference on Nuclear Engineering, ICONE 2024
Y2 - 4 August 2024 through 8 August 2024
ER -