A poly‐superellipsoid‐based approach on particle morphology for DEM modeling of granular media

Abstract

Summary Particle morphology plays a key role in affecting physical and mechanical behaviors of granular media. While various mathematical approaches and shape descriptors have been proposed to describe the morphological properties of granular particles, it remains a challenge to effectively incorporate them for efficient discrete modeling of granular materials. This study presents a new poly‐superellipsoid‐based approach for three‐dimensional discrete element method (DEM) modeling of non‐spherical convex particles. A uniform mathematical description of 3D poly‐superellipsoidal surface is employed to represent a realistic granular particle, which is shown to be versatile and effective in reproducing a wide range of shape features (including elongation, flatness, angularity, and asymmetry) for real particles in nature. A novel optimization approach based on hybrid Levenberg‐Marquardt (LM) and Gilbert‐Johnson‐Keerthi (GJK) algorithms is further developed for efficient and robust contact detection in DEM simulation of poly‐superellipsoidal assemblies. Simulations of granular packing and triaxial compression tests show that the proposed approach is generally robust and efficient for both dynamic and quasistatic modeling of granular media.

A thread‐block‐wise computational framework for large‐scale hierarchical continuum‐discrete modeling of granular media

“…34 The present parallelism framework opens a number of exciting future opportunities. First, GoDEM can be readily extended to the three-dimensional without modification to the implementation; moreover, for simulations where particle shape does matter to the computational cost, 52,53 such as nonspherical particle shape 17 and particle crushing, 54 the presented framework can be readily adapted with only minor revisions of the CPU-based algorithms. For example, the authors developed an open-source code, SudoDEM (https://sudodem.github.io), for DEM modeling of nonspherical particles, which can be readily implemented in this framework.…”

“…Interested readers are referred to the literature for discrete element modeling of nonspherical particles. 17,42…”

“…The open-source CPU-based DEM code, SudoDEM, 17,42,49 developed by the authors, is employed as a baseline to benchmark the proposed thread-block-wise algorithms on GPU. SudoDEM is available at its project page online, and the Python snippets and data sets involving in this section are available on the Github repository (https://github.com/SwaySZ/ ExamplesSudoDEM) for interested readers.…”

“…Despite the rather simple contact models used at grain scales, DEM has been demonstrated to be capable of capturing a rich spectrum of characteristics of a granular material and offering physically sound interpretations on macroscopic observations. [13][14][15][16] More recent progresses in both theoretical and methodological aspects further enable us to consider particle shape 17 and particle roughness 18 with improved confidence, and to tackle multiphase, multiphysics granular problems based on coupling with other computational methods including the computational fluid dynamics (CFD) 19 and the lattice Boltzmann method (LBM). 20 Notwithstanding the various merits, DEM has its unresolved issues.…”

3D DEM simulations of basic geotechnical tests with early detection of shear localization

“…The problem is further simplified for a case in which particles are idealized by perfect spheres; however (not in this article), more complicated shapes can also be easily used [45,46]. For example, the novel so-called poly-superellipsoid has been proposed in an open-source DEM code SudoDEM [47]. The particles are considered as perfectly rigid bodies, but with smooth (soft) contacts (so-called overlaps).…”