Friction and Pressure-Dependence of Force Chain Communities in Granular Materials

Abstract: Granular materials transmit stress via a network of force chains. Despite the importance of these chains to characterizing the stress state and dynamics of the system, there is no common framework for quantifying their their properties. Recently, attention has turned to the tools of network science as a promising route to such a description. In this paper, we apply community detection techniques to numerically-generated packings of spheres over a range of interparticle friction coefficients and confining pressures. In order to extract chain-like features, we use a modularity maximization with a recently-developed geographical null model \cite{Bassett2015}, and optimize the technique to detect branched structures by minimizing the normalized convex hull of the detected communities. We characterize the force chain communities by their size (number of particles), network strength (internal forces), and normalized convex hull ratio (sparseness). We find the that the first two exhibit an approximately linear correlation and are therefore largely redundant. For both pressure $P$ and interparticle friction $\mu$, we observe crossovers in behavior. For $\mu \lesssim 0.1$, the packings exhibit more sensitivity to pressure. In addition, we identify a crossover pressure where the frictional dependence switches from having more large/strong communities at low $\mu$ vs. high $\mu$. We explain these phenomena by comparison to the spatial distribution of communities along the vertical axis of the system. These results provide new tools for considering the mesoscale structure of a granular system and pave the way for reduced descriptions based on the force chain structure.