Particulate material fabric characterization from volumetric images by computational geometry

Abstract

This study develops computational geometry techniques to characterize fabric anisotropy from volumetric images. Although numerous laboratory test results have demonstrated the importance of fabric anisotropy on the macro-mechanical behavior of granular soils, the fabric characterization methods are underexplored. The advances of X-ray Computed Tomography (X-ray CT) techniques enable researchers to scan three-dimensional (3D) volumetric images of granular soils to observe fabric. Taking the advantage of X-ray CT, this study integrates principal component analysis, 3D watershed analysis, and Delaunay triangulation technique to compute key parameters for fabric characterization, including particle long axes, contacts and contact normals, branch vectors, void ratios, coordination number, and average branch vector length. The developed methodologies are validated through the loose and dense natural river sand specimens (containing particles from sub-angular to well-rounded). The developed computational geometry techniques may facilitate soil experimentalists to explain the observed soil behavior in laboratory tests induced by fabric anisotropy, enable constitutive modelers to integrate fabric in constitutive models, and enable discrete element modelers to calibrate their simulation models.