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:
Paper 96
Finite Element High-Performance Code for Seismic Wave Propagation in Heterogeneous Media C.J. Martins
Department of Civil Engineering, Federal Center for Technological Education, Belo Horizonte, Brazil C.J. Martins, "Finite Element High-Performance Code for Seismic Wave Propagation in Heterogeneous Media", in , (Editors), "Proceedings of the Second International Conference on Parallel, Distributed, Grid and Cloud Computing for Engineering", Civil-Comp Press, Stirlingshire, UK, Paper 96, 2011. doi:10.4203/ccp.95.96
Keywords: parallel finite element method, seismic wave propagation, MPI.
Summary
Little attention was given to using the finite element method by geophysicists until the mid-1980s; the works of Kuo and Teng [1] and Brooker and Small [2] stand out. Starting in the end of the 1980s, important contributions were made in the area of seismic wave propagation, for example Zhu et al. [3], Priolo and Seriani [4] and Abbound et al. [5]. Recent studies shows the importance of the finite element method as an acoustic and elastic modelling tool as an alternative to the finite difference method that is generally used in this type of analysis.
In order to evaluate the performance and efficiency achieved by the computer codes developed that utilize parallel processing resources based on MPI libraries, finite element analyses were carried out on a cluster of computers using the parallelization technique called domain decomposition. Analysis time was 0.1 seconds with an interval of 0.2 milliseconds, without seismogram impressions and snapshots (avoiding writing to the disk in order to avoid altering the analysis performance). The results for total analysis time obtained in function of the total number of degrees of freedom for each of the analyses carried out. It was observed that the linear behaviour of the results shows appropriate use of parallelization resources in problems with high degrees of freedom. References
purchase the full-text of this paper (price £20)
go to the previous paper |
|