平行子結構有限元素計算之效率研究
Date Issued
1999
Date
1999
Author(s)
DOI
882211E002018
Abstract
In the past two decades, finite element
method has become a common computational
approach for solving complicated engineering
problems. In recent years, the scale and
complexity of engineering computations have
increased significantly. Therefore, considerable
research efforts have been devoted to the area
of parallel finite element computations for
obtaining more computing power and for
achieving better computational efficiency.
However, several issues still need to be
addressed before parallel finite element
analysis can be easily and feasibly used for
solving general large-scale and computationintensive
engineering problems.
In this work, the focus is placed on
improving the efficiency of a popularly used
parallel finite element approach, namely
parallel substructure method. Several issues
related to the efficiency of parallel substructure
method are carefully investigated in this work.
These issues include: (1) mesh partitioning for
balance of computational loads among
processors and for minimization of interprocess
communication, (2) substructure nodal
renumbering for minimizing condensation
computations using general sparse matrix
technique. In addition, parallel multi-level
substructure method and parallel equation
solvers for solution of interface degrees of
freedom using concurrent processors are
discussed. Several examples are used in this
work for studying the efficiency of the
improved parallel substructure method
proposed.
Subjects
Parallel computing
Finite element analysis
Parallel substructure method
Automatic mesh partitioning
Substructure Nodal Renumbering
Publisher
臺北市：國立臺灣大學土木工程學系暨研究所
Type
report
File(s)![Thumbnail Image]()
Loading...
Name
882211E002018.pdf
Size
75.02 KB
Format
Adobe PDF
Checksum
(MD5):5a20593b2d8c8b53b95744dce5ad55c9