Under cyclic loading, sand will undergo a solid-liquid phase transition during the liquefaction. This study utilizes discrete element method (DEM) to investigate the stage characteristics of sand macroscopic stress-strain response during the solid-liquid phase transition. The microscopic mechanism of sand solid-liquid phase transition is elucidated from the perspective of contact network. The results indicate that based on the sand flowability, the liquefaction process can be divided into solid phase, solid-liquid transition phase, and liquid phase stages. The strong contact network within the sand is the primary contributor to its effective stress, and the degradation of the originally well-connected strong contact network are the reasons for the sand solid-liquid phase transition. A parameter xi c has been proposed to measure the connectivity of the strong contact network. The weak contacts between particles dominates the sliding and rolling between particles, which is the reason for the macroscopic deformation and flow of sand.

The contact network of granular materials is often divided into strong and weak subnetworks, which play different roles in micromechanics. Within the strong contact network, there exists the largest connected component, that is, the largest cluster, which may connect system boundaries and could be the most important structure in force transmission of the whole system. This paper concerns the particular features of the largest cluster in the strong contact network of granular materials, by considering the combining effects of loading path and particle shape. A series of true triaxial tests with various intermediate principal stress ratios are conducted for granular assemblies of different shaped particles using the discrete element method (DEM). Both the macroscopic stress-strain responses and the microscopic topological changes of the contact network are investigated. It is found that both particle shape and loading path will influence the shear strength and the topological features of the strong network. The threshold zeta$\zeta $ (the ratio to the average force) is used to distinguish the strong and weak networks, and a critical threshold can be identified by comparing the network-based metrics. The largest cluster within the strong network approaching the critical threshold can span the boundaries in each direction with minimum contacts, which occupies a small portion of particles and contacts but transmits a considerable portion of the applied stress. In addition, the similar contribution weight of the largest cluster to the deviatoric stress is identified for granular materials with different particle shapes.