Uniaxial Compression Simulation using Discrete Elements and Parallel Processing
Sanches, Luiz Carlos Facundo1; Carrion, Ronaldo2; Lima, Carlos Héracles Moraes1
1 São Paulo State University - UNESP; 2 University of São Paulo - USP
doi:10.20906/CPS/CILAMCE2017-0377
Resumo
This paper studies macroscopic properties in geomaterials samples using Discrete Element Method (DEM) and parallel computing. In an uniaxial compression experiment, a sample of geomaterial is slowly compressed by a piston until failure occurs. The peak stress at which failure of the sample occurs is known as Unconfined Compressive Strength (UCS). Analytical relationship between the microphysical parameters and the macroscopic properties are obtained by conducting a series of simulations to tune the microphysical parameters until desired macroscopic properties. For each particle, the interaction law is used in conjunction with the momentum balance principle so as to specify a set of governing equations to describe its interactions and motion. By solving this equations, we obtain the final state of rest of these particles. In our numeric simulation, a rectangular prism of particles is placed between two piston walls which are compressed at constant speed. Then, we monitor the positions, forces and the number of broken bonds. These data is used to measure the macroscopic elastic properties of the particle model. Results show that algorithms have high scalability in CPU-based multi-thread parallel computing and it is possible to obtain some macroscopic properties of geomaterials.
Palavras-chave: Computational Geomechanics; Uniaxial Compression Simulation ; Geomaterials; Discrete Elements ; Parallel Processing