MG (multigrid) method is one of the most promising solvers for large scale problems. Hexahedral mesh generation and its adaptation are not enough to use for practical applications, because its mesh generation is very difficult and still labor intensive work by hand. We have developed hexahedral local refinement technique controlled by posterior error estimation. We have proposed a MG technique for unstructured hexahedral meshes with local mesh refinement. In this paper, the proposed technique is evaluated to check its performance and severe analyses of bending cantilevers. Performance of MG for unstructured hexahedral meshes is compared with that of the PCG (preconditioned conjugate gradient) through several benchmark examples of 3-D static elastic analysis. Proposed MG is faster than PCG for all problems as number of freedoms increases. Finally limitation of the proposed technique is presented.
[1]
Jun'ichi Shinbori,et al.
Multigrid technique for hexahedral mesh using unstructured local mesh refinement
,
2004
.
[2]
F. Weiler,et al.
Octree-based Generation of Hexahedral Element Meshes
,
2007
.
[3]
O. C. Zienkiewicz,et al.
A simple error estimator and adaptive procedure for practical engineerng analysis
,
1987
.
[4]
Wolfgang Hackbusch,et al.
Multi-grid methods and applications
,
1985,
Springer series in computational mathematics.
[5]
Yoshitaka Wada,et al.
Effective adaptation technique for hexahedral mesh
,
2002,
Concurr. Comput. Pract. Exp..