A full-stage parallel architecture of three-dimensional discontinuous deformation analysis using OpenMP

Xinyan Peng, Guangqi Chen, Pengcheng Yu, Yingbin Zhang, Hong Zhang, Longxiao Guo

Research output: Contribution to journalArticle

Abstract

It can be computationally expensive to conduct large-scale engineering computation using three-dimensional discontinuous deformation analysis (3-D DDA). Improving the efficiency of 3-D DDA is thus imperative, and an available approach is to incorporate parallelism technique into 3-D DDA. In this paper, a full-stage parallelization of 3-D DDA is proposed, in which all the main subroutines that have the potential to be executed in parallel are implemented with parallelization using OpenMP. The correctness of the parallel 3-D DDA is examined by a block sliding example, showing good convergence between simulation results and analytical solutions. The computation efficiency is subsequently investigated through a set of masonry structures, and the parallel 3-D DDA exhibits good efficiency performance according to the comparison results. Afterwards, the application of the parallel 3-D DDA to large-scale colliding problem is carried out, demonstrating around 5 times enhanced efficiency with 6 threads used. Conclusively, the parallel 3-D DDA provides a more efficient approach to analyze large-scale problems than the serial conventional computation.

Original languageEnglish
Article number103346
JournalComputers and Geotechnics
Volume118
DOIs
Publication statusPublished - Feb 2020

Fingerprint

Digital differential analyzers
Parallel architectures
masonry
sliding
engineering
simulation
analysis
Subroutines

All Science Journal Classification (ASJC) codes

  • Geotechnical Engineering and Engineering Geology
  • Computer Science Applications

Cite this

A full-stage parallel architecture of three-dimensional discontinuous deformation analysis using OpenMP. / Peng, Xinyan; Chen, Guangqi; Yu, Pengcheng; Zhang, Yingbin; Zhang, Hong; Guo, Longxiao.

In: Computers and Geotechnics, Vol. 118, 103346, 02.2020.

Research output: Contribution to journalArticle

Peng, Xinyan ; Chen, Guangqi ; Yu, Pengcheng ; Zhang, Yingbin ; Zhang, Hong ; Guo, Longxiao. / A full-stage parallel architecture of three-dimensional discontinuous deformation analysis using OpenMP. In: Computers and Geotechnics. 2020 ; Vol. 118.
@article{cc1567d28ae34d07a6ec42b520103103,
title = "A full-stage parallel architecture of three-dimensional discontinuous deformation analysis using OpenMP",
abstract = "It can be computationally expensive to conduct large-scale engineering computation using three-dimensional discontinuous deformation analysis (3-D DDA). Improving the efficiency of 3-D DDA is thus imperative, and an available approach is to incorporate parallelism technique into 3-D DDA. In this paper, a full-stage parallelization of 3-D DDA is proposed, in which all the main subroutines that have the potential to be executed in parallel are implemented with parallelization using OpenMP. The correctness of the parallel 3-D DDA is examined by a block sliding example, showing good convergence between simulation results and analytical solutions. The computation efficiency is subsequently investigated through a set of masonry structures, and the parallel 3-D DDA exhibits good efficiency performance according to the comparison results. Afterwards, the application of the parallel 3-D DDA to large-scale colliding problem is carried out, demonstrating around 5 times enhanced efficiency with 6 threads used. Conclusively, the parallel 3-D DDA provides a more efficient approach to analyze large-scale problems than the serial conventional computation.",
author = "Xinyan Peng and Guangqi Chen and Pengcheng Yu and Yingbin Zhang and Hong Zhang and Longxiao Guo",
year = "2020",
month = "2",
doi = "10.1016/j.compgeo.2019.103346",
language = "English",
volume = "118",
journal = "Computers and Geotechnics",
issn = "0266-352X",
publisher = "Elsevier BV",

}

TY - JOUR

T1 - A full-stage parallel architecture of three-dimensional discontinuous deformation analysis using OpenMP

AU - Peng, Xinyan

AU - Chen, Guangqi

AU - Yu, Pengcheng

AU - Zhang, Yingbin

AU - Zhang, Hong

AU - Guo, Longxiao

PY - 2020/2

Y1 - 2020/2

N2 - It can be computationally expensive to conduct large-scale engineering computation using three-dimensional discontinuous deformation analysis (3-D DDA). Improving the efficiency of 3-D DDA is thus imperative, and an available approach is to incorporate parallelism technique into 3-D DDA. In this paper, a full-stage parallelization of 3-D DDA is proposed, in which all the main subroutines that have the potential to be executed in parallel are implemented with parallelization using OpenMP. The correctness of the parallel 3-D DDA is examined by a block sliding example, showing good convergence between simulation results and analytical solutions. The computation efficiency is subsequently investigated through a set of masonry structures, and the parallel 3-D DDA exhibits good efficiency performance according to the comparison results. Afterwards, the application of the parallel 3-D DDA to large-scale colliding problem is carried out, demonstrating around 5 times enhanced efficiency with 6 threads used. Conclusively, the parallel 3-D DDA provides a more efficient approach to analyze large-scale problems than the serial conventional computation.

AB - It can be computationally expensive to conduct large-scale engineering computation using three-dimensional discontinuous deformation analysis (3-D DDA). Improving the efficiency of 3-D DDA is thus imperative, and an available approach is to incorporate parallelism technique into 3-D DDA. In this paper, a full-stage parallelization of 3-D DDA is proposed, in which all the main subroutines that have the potential to be executed in parallel are implemented with parallelization using OpenMP. The correctness of the parallel 3-D DDA is examined by a block sliding example, showing good convergence between simulation results and analytical solutions. The computation efficiency is subsequently investigated through a set of masonry structures, and the parallel 3-D DDA exhibits good efficiency performance according to the comparison results. Afterwards, the application of the parallel 3-D DDA to large-scale colliding problem is carried out, demonstrating around 5 times enhanced efficiency with 6 threads used. Conclusively, the parallel 3-D DDA provides a more efficient approach to analyze large-scale problems than the serial conventional computation.

UR - http://www.scopus.com/inward/record.url?scp=85075250779&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85075250779&partnerID=8YFLogxK

U2 - 10.1016/j.compgeo.2019.103346

DO - 10.1016/j.compgeo.2019.103346

M3 - Article

AN - SCOPUS:85075250779

VL - 118

JO - Computers and Geotechnics

JF - Computers and Geotechnics

SN - 0266-352X

M1 - 103346

ER -