Date of Award

12-2018

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Civil Engineering

Committee Member

Qiushi Chen, Committee Chair

Committee Member

Ronald D. Andrus

Committee Member

Stephen Moysey

Committee Member

Hai Huang

Abstract

Granular materials, such as sand, biomass particles, and pharmaceutical pills, are widespread in nature, industrial systems, and our daily life. Fundamentally, the bulk mechanical behavior of such materials is governed by the physical and morphological features of and the interactions among constituent particles at the microscopic scale. From a modeling standpoint, the particle-based discrete element method (DEM) has emerged as the most prevalent numerical tool to model and study the behavior of granular materials and the systems they form. A critical step towards an accurate and predictive DEM model is to incorporate those physical and morphological features (e.g., particle size, shape, and deformability) pertaining to the constituent particles. The main objective of this dissertation is to approach an accurate characterization and modeling of the grading- and shape-dependent behavior of granular materials by developing DEM models that incorporate realistic physical and morphological features of granular particles. Revolving around this objective, three studies are presented: image-based particle reconstruction and morphology characterization, grading and shape-dependent shearing behavior of rigid-particle systems, and granular flow of deformable irregular particles. The first study presents a machine learning and level-set based framework to re- construct granular particles and to characterize particle morphology from X-ray computed tomography (X-ray CT) imaging of realistic granular materials. Images containing detailed microstructure information of a granular material are obtained using the X-ray CT tech- nique. Approaches such as the watershed method in two dimensions (2D) and the combined machine learning and level set method in three dimensions (3D) are then utilized and implemented to segment X-ray CT images and to numerically reconstruct individual particles in the granular material. Based on the realistic particle shapes, particle morphology is characterized by descriptors including aspect ratio, roundness, circularity (2D) or sphericity (3D). The particle shapes or morphology provide important constraints to develop DEM models with particle physical and morphological features conforming to the specific granular material of interest. In the second study, DEM models incorporated with realistic particle sizes and shapes are developed and applied to study the shearing behavior of sandy soils. The particle sizes and shapes are obtained from realistic samples of JSC-1A Martian regolith simulant. Irregular-shape particles are represented by rigid clumps based on the domain overlapping filling method. The effects of particle shape irregularity on the shearing behavior of granular materials are investigated through direct shear tests, along with the comparisons from spherical particles with or without rolling resistance. The micro-mechanisms of shape irregularity contributing to the shear resistance are identified. The last study investigates the effects of particle deformability (e.g., compression, deflection or torsion), together with particle sizes and shapes, on the granular flow of flexible granular materials. A bonded-sphere DEM model is implemented with the capability of embodying various particle sizes and irregular shapes, as well as capturing particle deformability. This approach is then applied to simulate and study the behavior of flexible granular materials in cyclic compression and hopper flow tests. The effects of particle size, shape and deformability on the bulk mechanical behavior are investigated on the basis of the DEM simulation results. The importance of particle deformability to the DEM simulations of flexible granular materials is demonstrated.

Share

COinS