Parallelization of D.E.M. codes for fine powders
- Author
- Francisco Antonio Gilabert Villegas (UGent) , Ramon Garcia-Rojo and Sean McNamara
- Organization
- Abstract
- The aim of this work is to implement a powerful computational code to explore the consequences of the competition between cohesion and external pressure in the assembling process (aggregate formation) and the compression of a cohesive granular packing. Since both processes are extremely time consuming running on a single processor, a DEM code has been parallelized to simulate these stages. The numerical setup is arranged as follows: a parallelepiped-shaped space will contain the grains, where periodical boundary conditions at the edges will be considered (wall effects are neglected). To carry out the computations, this parallelepiped will be divided in nP = nD α cells, where nP is the number of processes, nα is the number of spatial divisions on axis α and D is the spatial dimension of the domain decomposition (in our case D = 1, 3 was used). In each cell a travelling boundary or halo will be defined. This halo must contain the necessary information required by the neighbor cells (number of shared neighbor grains, contacts, forces, etc.). Using this code, the behaviour of the system such as the variations of the volume fraction with the applied pressure will be studied as well as the improvement of performance in contrast to the serial version.
- Keywords
- Cohesive powders, parallelization, domain decomposition, packing, MPI, DEM, constitutive modeling
Downloads
-
(...).pdf
- full text (Accepted manuscript)
- |
- UGent only
- |
- |
- 690.09 KB
Citation
Please use this url to cite or link to this publication: http://hdl.handle.net/1854/LU-8741011
- MLA
- Gilabert Villegas, Francisco Antonio, et al. “Parallelization of D.E.M. Codes for Fine Powders.” Science and Supercomputing in Europe, Annual Report Book, Transnational Access Meeting (TAM 2006), edited by Paola Alberigo et al., Pan-European Research Infrastructure on High Performance Computing (HPC-Europa), 2006, pp. 681–86.
- APA
- Gilabert Villegas, F. A., Garcia-Rojo, R., & McNamara, S. (2006). Parallelization of D.E.M. codes for fine powders. In P. Alberigo, G. Erbacci, & F. Garofalo (Eds.), Science and Supercomputing in Europe, Annual Report Book, Transnational Access Meeting (TAM 2006) (pp. 681–686). Bologna, Italy: Pan-European Research Infrastructure on High Performance Computing (HPC-Europa).
- Chicago author-date
- Gilabert Villegas, Francisco Antonio, Ramon Garcia-Rojo, and Sean McNamara. 2006. “Parallelization of D.E.M. Codes for Fine Powders.” In Science and Supercomputing in Europe, Annual Report Book, Transnational Access Meeting (TAM 2006), edited by Paola Alberigo, Giovanni Erbacci, and Francesca Garofalo, 681–86. Bologna, Italy: Pan-European Research Infrastructure on High Performance Computing (HPC-Europa).
- Chicago author-date (all authors)
- Gilabert Villegas, Francisco Antonio, Ramon Garcia-Rojo, and Sean McNamara. 2006. “Parallelization of D.E.M. Codes for Fine Powders.” In Science and Supercomputing in Europe, Annual Report Book, Transnational Access Meeting (TAM 2006), ed by. Paola Alberigo, Giovanni Erbacci, and Francesca Garofalo, 681–686. Bologna, Italy: Pan-European Research Infrastructure on High Performance Computing (HPC-Europa).
- Vancouver
- 1.Gilabert Villegas FA, Garcia-Rojo R, McNamara S. Parallelization of D.E.M. codes for fine powders. In: Alberigo P, Erbacci G, Garofalo F, editors. Science and Supercomputing in Europe, Annual Report Book, Transnational Access Meeting (TAM 2006). Bologna, Italy: Pan-European Research Infrastructure on High Performance Computing (HPC-Europa); 2006. p. 681–6.
- IEEE
- [1]F. A. Gilabert Villegas, R. Garcia-Rojo, and S. McNamara, “Parallelization of D.E.M. codes for fine powders,” in Science and Supercomputing in Europe, Annual Report Book, Transnational Access Meeting (TAM 2006), Barcelona, Spain, 2006, pp. 681–686.
@inproceedings{8741011,
abstract = {{The aim of this work is to implement a powerful computational code to explore the consequences of the competition between cohesion and external pressure in the assembling process
(aggregate formation) and the compression of a cohesive granular packing. Since both processes are extremely time consuming
running on a single processor, a DEM code has been parallelized
to simulate these stages. The numerical setup is arranged as
follows: a parallelepiped-shaped space will contain the grains,
where periodical boundary conditions at the edges will be considered (wall effects are neglected). To carry out the computations, this parallelepiped will be divided in nP = nD α cells, where
nP is the number of processes, nα is the number of spatial divisions on axis α and D is the spatial dimension of the domain
decomposition (in our case D = 1, 3 was used). In each cell
a travelling boundary or halo will be defined. This halo must
contain the necessary information required by the neighbor cells
(number of shared neighbor grains, contacts, forces, etc.). Using this code, the behaviour of the system such as the variations
of the volume fraction with the applied pressure will be studied
as well as the improvement of performance in contrast to the
serial version.}},
author = {{Gilabert Villegas, Francisco Antonio and Garcia-Rojo, Ramon and McNamara, Sean}},
booktitle = {{Science and Supercomputing in Europe, Annual Report Book, Transnational Access Meeting (TAM 2006)}},
editor = {{Alberigo, Paola and Erbacci, Giovanni and Garofalo, Francesca}},
isbn = {{8886037171}},
keywords = {{Cohesive powders,parallelization,domain decomposition,packing,MPI,DEM,constitutive modeling}},
language = {{eng}},
location = {{Barcelona, Spain}},
pages = {{681--686}},
publisher = {{Pan-European Research Infrastructure on High Performance Computing (HPC-Europa)}},
title = {{Parallelization of D.E.M. codes for fine powders}},
year = {{2006}},
}