2002 Fiscal Year Final Research Report Summary
Development of parallel computational system for 3-D wind flow; around ground with structures using CAD/GIS data
| Project/Area Number |
12650482
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
構造工学・地震工学
|
|---|
| Research Institution | CHUO UNIVERSITY |
|---|
Principal Investigator |
KASHIYAMA Kazuo Chuo University, Faculty of Science and Engineering,Professor, 理工学部, 教授 (10194721)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
HIRANO Hirokazu Chuo University, Faculty of Policy Studies, Professor, 総合政策学部, 教授 (80256023)
|
|---|
| Project Period (FY) |
2000 – 2002
|
| Keywords | CAD / GIS / wind flow / stabilized finite element methc / parallel computing |
|---|
| Research Abstract |
A finite element modeling system for ground considering structures and a parallel finite element system for 3-dimentional wind flow have been developed in this study.
In the finite element modeling system, the CAD and GIS data were combined using common data format (DXF) and it was used for mesh generation for ground and structure. In order to express the geographical features accurately, a side of triangular element expressed not only the contour lines but also the valley and mountain lines. The Delaunay triangulation method was applied for making an unstructured mesh and the multi-boxcel method was applied for making a structured mesh. By using the present system, an accurate finite element mesh considering the complicated configuration of ground and structures.
In the parallel finite element system for wind flow, the unsteady Navier-Stokes equation based on Boussinesq approximation was used for the governing equation of wind flow and the SUPG and PSPG.(pressure-stabilizing/Petrov-Galerkin) methods were used stabilization method. The P1/P1 element was employed for the discretization in space. Parallel implementation based on domain decomposition method and MPI was carried out on a distributed memory parallel cluster. The computed results were compared with the experimental results. From the results, the numerical result obtained by the present method is good in agreement with the experimental results. Furthermore, the good parallel performance was obtained from the results of parallel performance.
|
Research Products
(13 results)
