TY - JOUR
T1 - OpenMP-Based Parallel Two-Dimensional Discontinuous Deformation Analysis for Large-Scale Simulation
AU - Yu, Pengcheng
AU - Peng, Xinyan
AU - Chen, Guangqi
AU - Guo, Longxiao
AU - Zhang, Yingbin
N1 - Funding Information:
This work was supported by the JSPS KAKENHI (Grant Nos. JP15K12483, JP16F16056, and JP15H01797), the National Natural Science Foundation of China (Nos. 41977213, 41672286, and 51408511), the SKLGP open fund (SKLGP2018K009), the China Scholarship Council (CSC), and the Japanese Government (MEXT) Scholarship Program. Also, this work was partly supported by the cor-esearch project conducted by Kyushu University and W-NEXCO. The financial support are gratefully acknowledged.
PY - 2020/7/1
Y1 - 2020/7/1
N2 - Discontinuous deformation analysis (DDA) has gained wide acceptance in geotechnical engineering. The simulation of large-scale engineering problems using DDA is computationally expensive. Thus, improving the calculation efficiency of DDA is of significance. The parallel technique based on open multiprocessing (OpenMP) is implemented into DDA. Previous work usually focused on one or two specific subroutines. In contrast, in this paper, all the time-consuming subroutines that have the possibility to do the calculation simultaneously, including contact detection, matrix assembly, simultaneous equation solver, postjudgment of contacts, and block information update, are modified and incorporated with parallel implementation based on OpenMP. The parallel degree can be up to 98%. The computational efficiency of the accelerated DDA is demonstrated by several large-scale simulations with different block quantities running on a different number of threads. As an application, the accelerated DDA is used to efficiently analyze a large-scale landslide, which can efficiently provide information for landslide disaster prevention and mitigation. In conclusion, the accelerated DDA exhibits high performance in the analysis of large-scale problems.
AB - Discontinuous deformation analysis (DDA) has gained wide acceptance in geotechnical engineering. The simulation of large-scale engineering problems using DDA is computationally expensive. Thus, improving the calculation efficiency of DDA is of significance. The parallel technique based on open multiprocessing (OpenMP) is implemented into DDA. Previous work usually focused on one or two specific subroutines. In contrast, in this paper, all the time-consuming subroutines that have the possibility to do the calculation simultaneously, including contact detection, matrix assembly, simultaneous equation solver, postjudgment of contacts, and block information update, are modified and incorporated with parallel implementation based on OpenMP. The parallel degree can be up to 98%. The computational efficiency of the accelerated DDA is demonstrated by several large-scale simulations with different block quantities running on a different number of threads. As an application, the accelerated DDA is used to efficiently analyze a large-scale landslide, which can efficiently provide information for landslide disaster prevention and mitigation. In conclusion, the accelerated DDA exhibits high performance in the analysis of large-scale problems.
UR - http://www.scopus.com/inward/record.url?scp=85078874152&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85078874152&partnerID=8YFLogxK
U2 - 10.1061/(ASCE)GM.1943-5622.0001705
DO - 10.1061/(ASCE)GM.1943-5622.0001705
M3 - Article
AN - SCOPUS:85078874152
SN - 1532-3641
VL - 20
JO - International Journal of Geomechanics
JF - International Journal of Geomechanics
IS - 7
M1 - 04020083
ER -