GPU-based collision analysis between a multi-body system and numerous particles

Jung, Hye–Young; Jun, Chul-Woong; Sohn, Jae Hak

doi:10.1007/s12206-013-0226-4

Cited by 6 publications

(1 citation statement)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As a stepping stone towards minimising this computational effort, recent studies have succeeded in initiating efficient solutions either by developing state-of-the-art contact detection algorithms [81][82][83][84] or utilising multi-core [85][86][87] processors, heterogeneous CPU 2 -GPU 3 architecture [87][88][89][90], supercomputers [91] or parallelised GPU clusters [92]. A modern day personal desktop-possessing two or four cores-is able to simulate, for example, two-minutelong, fully three-dimensional simple soft-sphere (linear force model) DEM simulation comprising 10,000-200,000 particles within a couple of hours.…”

Section: Computational Speedupmentioning

confidence: 99%

Numerical modelling of granular flows: a reality check

2015

View full text Add to dashboard Cite

Discrete particle simulations provide a powerful tool for the advancement of our understanding of granular media, and the development and refinement of the multitudinous techniques used to handle and process these ubiquitous materials. However, in order to ensure that this tool can be successfully utilised in a meaningful and reliable manner, it is of paramount importance that we fully understand the degree to which numerical models can be trusted to accurately and quantitatively recreate and predict the behaviours of the real-world systems they are designed to emulate. Due to the complexity and diverse variety of physical states and dynamical behaviours exhibited by granular media, a simulation algorithm capable of closely reproducing the behaviours of a given system may be entirely unsuitable for other systems with different physical properties, or even similar systems exposed to differing control parameters. In this paper, we focus on two widely used forms of granular flow, for which discrete particle simulations are shown to provide a full, quantitative replication of the behaviours of real industrial and experimental systems. We identify also situations for which quantitative agreement may fail are identified, but important general, qualitative trends are still recreated, as

show abstract