The 3D characterization of particle kinematics representing the changes in fabric observed during water-driven collapse of loess is, to the best of our knowledge, performed for the first time at the micron scale. An apparatus is specially designed to perform collapse tests on loess specimens of several millimeters in size and to capture, using X-ray tomography, particle-scale microfabric features of the sample in initial, loaded and flooded (i.e. collapsed) states. Although similar approaches have been used to assess problems such as microstructural evolution in sand, the application of these methods to assess loess collapse behaviour at the micron scale constitutes the novelty of this work. Individual particles within the specimen in the initial and deformed configurations are identified and tracked through an iterative segmentation technique and the particle tracking method ‘ID-track’. This allowed determination of particle displacement and rotation. The displacement field within the collapsed sample is found to be less uniform compared to that within the unwetted sample under loading. Coefficient of heterogeneity is defined to quantify the level of non-uniformity of the particle displacements within the deformed sample, revealing higher heterogeneity in the deformation of the collapsed sample compared to that of the unwetted sample under loading. It is also found that particles tend to rotate around axes perpendicular to, rather than parallel to, the direction of major principal stress, and the rotation magnitudes appears to be controlled largely by the size of particles while their morphology plays a minor role. The influence of porosity on the collapse process is quantified significant heterogeneous volumetric strains are observed at the single particle scale. It is also shown that the evolution of particle-to-particle contacts is much more complex than previously stated. The micron-scale investigations of individual particle kinematics following loading and wetting offer exciting new avenues for visualization and an enhanced capability for quantification of loess collapse processes.
Read full abstract