Engineering behavior of coarse-grained granular soils is significantly affected by particle shape and fabric. In the last two decades, two-dimensional (2D) image-based methods have been developed for automating particle shape and fabric measurements. Recently, more accurate three-dimensional (3D) image-based methods, such as X-ray Computed Tomography (X-ray CT), have become popular for analyzing 3D particle shape and fabric of granular soils.

X-ray CT captures volumetric images of soils, in which the soil particles are contacting with each other. Segmentation of contacting particles is the prerequisite for subsequent analysis. Therefore, this research developed an improved watershed analysis for addressing this issue. After identifying individual particles and their contacts, this study developed 3D computational geometry techniques for determining long axes, contact normals, and branch vectors for fabric characterization. Then, this study compared 2D and 3D fabric characterizations and explored the relationship between particle shape and soil fabric.

X-ray CT is an ideal technique to capture 3D particle geometries for particle shape characterization. However, Due to the high costs of device, maintenance, and operation, X-ray CT is not available for common soil laboratories. Therefore, this research developed a simple, low-cost, and fast 3D imaging technique, structured light, for analyzing 3D particle geometries. Realistic particles have been widely used in the discrete element method (DEM) simulations. This research developed a particle generation technique method by integrating spherical harmonics, Gassuan mixture model, expectation-maximization, and Dirichlet process. This technique can analyze a few of soil particles to extract their morphological properties, which were used to generate as many realistic particles as necessary. These realistic particles can be used in DEM to facilitate realistic particle-based DEM simulations.