Since Discrete Element Method (DEM) was first introduced for modeling micromechanical interactions of granular materials back in late 1970s, significant progress has been made to improve the performance of DEM algorithms. For example, a variety of approaches have been developed to simulate triaxial tests using DEM to better understand the fundamental mechanical behavior of granular materials. Nevertheless, potential error accumulation over the necessary large number of timesteps as a part of the explicit time integration may undermine the simulation accuracy. This paper presents the development, implementation and validation of a computing scheme that is based on real-time data fusion between a sensing mechanism and real time (SMART) computing. This computing framework consists of: (1) real-time data acquisition of particle kinematics through a wireless instrumentation called “SmartRocks” that are embedded at discrete locations in a granular assembly, and (2) a built-in data-fusion-based algorithm using the Kalman filter to integrate the prediction generated by DEM and the measurements reported by “SmartRocks.” To evaluate the performance of the SMART computing algorithm, laboratory large-scale triaxial tests on ballast specimens were conducted and the results were compared to traditional DEM-only and SMART computing simulations. It is concluded the SMART computing improved the simulation accuracy over the DEM-only simulations in terms of the deviatoric stress vs. axial strain, volumetric strain vs. axial strain, and final deformed specimen shape, and hence can be used to model large-scale triaxial tests with high fidelity.
All Science Journal Classification (ASJC) codes
- Geotechnical Engineering and Engineering Geology
- Computer Science Applications