Design and test of parallel discrete element method program of shared memory type

Changsheng Li,Hongwei Yin,Chun Liu,Shenyang Cai

DOI: https://doi.org/10.13232/j.cnki.jnju.2017.06.018

2017-01-01

Abstract:With the help of such date structure and changing the neighbor search algorithm based on grid,the lock-free parallel computing of discrete element method was achieved,and an efficient program of discrete element method was written in the C language.As a test case,a rock specimen was simulated under uniaxial compression by using the solid lattice model,and the entire process of shear failure was observed.The parallel test results showed that the speedup ratio basically remained unchanged when the number of particles reached tens of thousands or more.It reflected the phenomenon that the speedup of parallel has nothing to do with the scale of the problems.The parallel efficiency was about 0.85 and the time costs linear increases with the increase of number of particles,when we used four threads.The parallel computing of discrete element method was realized based on procedure-oriented language,which could make full use of existing multi-core resources and improve computing speed greatly and be useful to simulate the complex geotechnical engineering and geological model required more particles.