Exploring the structural and spatial evolutions of contact networks in granular systems via the image-based recognition technique

Abstract

Contact network, a mesoscale structure of granular matters, can be specified as numerous contact loops that are sensitive to changes in loading and deformation. The mechanical behaviors of granular matters can be reflected by the structural and spatial characteristics of contact loops, yet surprisingly little is known about the spatial issue. To overcome this deficiency, several indexes are proposed to quantify the spatial distribution of contact loops, and the image-based recognition technique, an alternative technique to the Delaunay triangulation method, is developed to extract contact loop information from contact network images. Furtherly, three numerical specimens with different initial densities are compressed biaxially to explore relationships between macro-mechanical behaviors and meso-structures of contact networks. The structural evolution of contact networks is most active in the dense specimen, and the conversion of low- and high-order loops leads to macroscopic volumetric dilatancy. Additionally, the distribution of high-order loops along boundaries proves the existing conclusion that strong contacts developed more near the loading boundary. Limited by the inter-locking effect in low-order loops, even the micro-slips developing also cause a significant weakening in network homogeneity. Contrarily, the evolution from multiple slip bands to the single shear band has little effect on the network homogeneity.