Exploring percolation of strong network in granular assemblies: the critical roles of particle shape and loading history dependency

Abstract

The mechanical response of granular media subjected to loading-unloading-reloading is crucial for evaluation of safety of engineering works. The evolution of complex network may provide a new insight into the irreversibility of granular materials under loading-unloading-reloading cycle. In this study, we analyze network-based metrics of strong networks in granular materials with different shapes under loading, unloading and reloading path by using DEM. From a macroscopic perspective, samples of non-spherical particles demonstrate a higher degree of irreversibility than those of spherical particles during the unloading after approaching the peak stress. Evident macroscopic hysteresis loop and the corresponding topological features are investigated for 5 stages with different stress states or stress history. Although contact force distributions are similar, the coordination number and clustering coefficient of the strong network vary at a given stress level during loading-unloading cycles. For the same stress states during unloading-reloading process, the largest cluster in strong contact network percolate with a smaller force threshold, demonstrating that the irreversible structural changes to a more unstable state. The gap in strong network topology measures between initial loading and unloading or reloading states is greater in samples of non-spherical shapes. In addition, the importance of contact topology, contact orientation distribution and weak network participation for granular samples with non-spherical shapes should be noticed during loading-unloading-reloading cycles.