MPI parallelization of unstructured mesh adaptation

被引：0

作者：

Lepage, C. Y. ^{[1
]}

St-Cyr, A. ^{[2
]}

Habashi, W. G. ^{[3
]}

机构：

[1] Newmerical Technol Int, 680 Sherbrooke St,W, Montreal, PQ H3A 2M7, Canada

[2] Natl Ctr Atmospher Res, Boulder 80305, CO USA

[3] McGill Univ, Comp Fluid Dynam Lab, Montreal, PQ, Canada

来源：

COMPUTATIONAL FLUID DYNAMICS 2004, PROCEEDINGS | 2006年

关键词：

D O I：

10.1007/3-540-31801-1_105

中图分类号：

O3 [力学];

学科分类号：

08 ; 0801 ;

摘要：

A 3D hybrid unstructured anisotropic mesh adaptation module is parallelized using MPI on distributed-memory architectures. Attention is given to the parallelization issues of the adaptation algorithm, with focus on domain decomposition, dynamic load-balancing, and treatment of the nodes on the frozen interfaces between partitions to achieve a uniformly adapted high-quality converged mesh. A cell agglomeration technique is proposed for the treatment of mesh periodicity, internal faces, and Y+ adaptation in columns of prisms.

引用

页码：727 / +

页数：2