Abstract
Granular materials exist in various forms in nature, ranging from pure sand or silt composed of silica-based minerals to alloys and composites mixed with different substances. The complexity of assessing the behavior of granular materials at both macro and micro scales has posed challenges in accurately quantifying their behavior. However, advancements in experimentation, computational mechanics, and statistical analysis tools have provided the necessary sophistication to carry out research on complex materials obtaining insights on their response. This study aims to investigate the multiscale behavior of granular materials by combining experimentation, statistical interpretation, and numerical simulations. Microscale testing and analysis yield contact properties of nonconforming or bonded granular material contacts, providing insights into stiffness characteristics, frictional behavior, and morphological features. Statistical interpretation, such as using Bayesian probability, helps to quantify uncertainties in micromechanical test data. Macroscale testing, such as Bender element testing on triaxial apparatus, provides a quantitative understanding of bulk material behavior. Numerical analysis, such as discrete element simulations, offers a qualitative understanding of material behavior through computed micro and mesoscale variables. By integrating these studies, multiscale insights into the behavior of granular materials can be achieved, contributing to the knowledge base in this area of research. Therefore, the present study aims at following a multi-directional approach (experimental, statistical, and numerical-based studies) to contribute majorly to the understanding of the granular material.The following are the five types of studies carried out:
In the first series of studies, the mechanical behavior of contact between two granular particles, whether unbonded or bonded, was investigated. Compression and shearing tests were conducted using simple and complex load paths to gain insights into the contact response. A wide range of materials, including Leighton Buzzard sand, Solani sand, and saprolite grains from landslide colluvium were characterized and tested. Additionally, a short study focused on assessing the influence of cyclic loading and soil type on the normal and shear contact behavior of a natural sand grain-geofiber composite interface, using natural bagasse fiber was carried out. Furthermore, a custom-built micromechanical apparatus for cemented grains was employed to investigate the three-dimensional contact behavior of cemented shale rock granules, with specific emphasis on the effects of bond type and aspect ratio on the resulting bond failure envelope.
In the second series of studies, a robust Bayesian statistical probability-based algorithm was developed for curve fitting the shear stiffness reduction curves of nonconforming geomaterial contacts. For nonconforming contacts, four analytical curve fitting models were developed, including three modified hyperbolic models from the soil dynamics discipline and a semi-empirical modified version of the Mindlin-Deresiewicz model for tangential force-displacement curves. Additionally, Markov Chain Monte Carlo (MCMC) simulation was employed to derive a 95% confidence interval for the selected curve fitting model. This work demonstrated the application of statistical probability tools for data analysis and interpretation of micromechanical tests in a more systematic manner.
In the third study, a multiscale experimental study integrated with numerical simulations was conducted to examine the mechanics of polydisperse granular mixtures consisting of coarse-grained particles mixed with varying percentages of fines. The study involved macroscale bender element tests on isotopically compressed granular samples to investigate the stiffness variation with changes in size ratio and fines content. Empirical equations were developed to predict stiffness based on experimental data, incorporating the concept of equivalent void ratio to represent the influence of void ratio on stiffness. A parameter called the size disparity indicator (ISD) was introduced to consider the coupled effects of size disparity and fines content on stiffness. Microscale assessment of individual contacting grains revealed that stiffness is affected by changes in grain size, even when the size ratio is less than 6, contrary to the observations from macroscale tests. Furthermore, numerical simulations using the discrete element method (DEM) were conducted on the polydisperse granular mixtures to demonstrate the coupled effect of size ratio and fines content on the formation of the structural matrix, thereby influencing the force transfer among contacts and resulting in varying stiffness behavior.
In the fourth series of studies, the Discrete Element Method (DEM) was used to simulate consolidated drained tests on quartz sand and cemented sands to investigate their small and small-to-medium strain shear modulus behavior. For quartz sands, a two-stage calibration methodology was developed, involving calibration at both the particle scale and macroscale of the DEM model. The significance of the chosen non-linear contact model and the effect of contact properties, such as particle Young's modulus (E) and friction (μ), on the shear modulus behavior were discussed. Micro and mesoscale variables, including contact force network (CFN), fabric anisotropy (ac), coordination number (Zm), and contact distribution and orientation were used to understand the contact force contribution and distribution among grains in the DEM model. Similarly, for cemented grains, a micromechanical-based DEM approach was followed. Contact data derived from grain-scale tests on cemented sand was used to develop the DEM model. The study examined the effect of contact properties, such as bond strength, cementation, and heterogeneity of cementation, on the small and small-to-medium strain shear modulus of structured/cemented sand derived at macroscale using microscale variables such as fabric anisotropy (ac), bond breakage pattern, contact force orientation and distribution, and energy accumulation and dissipation.
In the fifth series of studies, a multiscale assessment was conducted using 3-dimensional discrete element analysis to investigate the interaction problem of an impactor colliding with the surface of a block. The study focused on the behavior of a proppant simulant (impactor) colliding with a block (analog rock) considering different variants of clastic rock. The analysis considered variations in bond and contact stiffness, as well as bond strength, using the linear parallel bond contact model. The interaction between the rock and the impactor was defined by the Hertz-Mindlin contact model, which accurately represented the force-displacement response for a grain-block system. A parametric study was performed to assess the impact of impactor size and collision velocity on the damage characteristics of the base rock. The analysis emphasized the evaluation of macroscale parameters such as stress and energy transfer, mesoscale parameters such as penetration depth and contact formation, and microscale parameters such as contact force network and fracture network.
| Date of Award | 4 Sept 2024 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Kostas SENETAKIS (Supervisor) |