Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 95
PROCEEDINGS OF THE SECOND INTERNATIONAL CONFERENCE ON PARALLEL, DISTRIBUTED, GRID AND CLOUD COMPUTING FOR ENGINEERING
Edited by: P. Iványi and B.H.V. Topping
Paper 13

A Parallel Computation of a Characteristic Curve Method in a Domain Decomposition System

Q. Yao1, M. Ogino2 and H. Kanayama2

1Department of Intelligent Machinery and Systems, Graduate School of Engineering,
2Department of Mechanical Engineering, Faculty of Engineering,
Kyushu University, Fukuoka, Japan

Full Bibliographic Reference for this paper
Q. Yao, M. Ogino, H. Kanayama, "A Parallel Computation of a Characteristic Curve Method in a Domain Decomposition System", in P. Iványi, B.H.V. Topping, (Editors), "Proceedings of the Second International Conference on Parallel, Distributed, Grid and Cloud Computing for Engineering", Civil-Comp Press, Stirlingshire, UK, Paper 13, 2011. doi:10.4203/ccp.95.13
Keywords: characteristic curve method, finite element simulation, domain decomposition system, degrees of freedom.

Summary
To implement the characteristic curve method, a global-wise element information is required to determine the position of one particle in some time step, however, in the parallel domain decomposition system, the whole domain is split into several "parts" and one processor element (PE) only contains the element information of the current "part". For each "part", a domain decomposition is performed by the current PE and that is to say, the "part" is further divided into many subdomains; another difficulty lies in the fact that during the computation, the particle is not limited within one "part", therefore, exchanging the data between different PEs is also necessary, which requires the processor elements to communicate during the subdomain-wise computation.

In this research, a new algorithm to perform the parallel computation is developed; without increasing the computation complexity, this method reduces the memory consumption and the algorithm is thus enabled to solve large scale problems of over 10 million degrees of freedom (DOF).

References
1
H.A. van der Vorst, "Iterative Krylov Preconditionings for Large Linear Systems", Cambridge University Press, UK, 2003.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description