Multiscale characterisation of diffuse granular failure

Abstract

We study the evolution of structure inside a deforming, cohesionless granular material undergoing failure in the absence of strain localisation – so-called diffuse failure. The spatio-temporal evolution of the basic building blocks for self-organisation (i.e. force chains and minimal contact cycles) reveals direct insights into the structural origins of failure. Irrespective of failure mode, self-organisation is governed by the cooperative behaviour of truss-like 3-cycles providing lateral support to column-like force chains. The 3-cycles, which are initially in scarce supply, form a minority subset of the minimal contact cycle bases. At large length-scales (i.e. sample size), these structures are randomly dispersed, and remain as such while their population progressively falls as loading proceeds. Bereft of redundant constraints from the 3-cycles, the force chains are initially just above the isostatic state, a condition that progressively worsens as the sample dilates. This diminishing capacity for redistribution of forces without incurring physical rearrangements of member particles renders the force chains highly prone to buckling. A multiscale analysis of the spatial patterns of force chain buckling reveals no clustering or localisation with respect to the macroscopic scale. Temporal patterns of birth-and-death of 3-cycles and 3-force chains provide unambiguous evidence that significant structural reorganisations among these building blocks drive rheological behaviour at all stages of the loading history. The near-total collapse of all structural building blocks and the spatially random distribution of force chain buckling and 3-cycles hint at a possible signature of diffuse failure.

Keywords:

• granular materials
• diffuse failure
• force chain buckling
• contact cycles
• complex networks

Acknowledgements

AT and SP was supported by the US Army Research Office Grant (DAAD-W911NF1110175), the Australian Research Council Discovery Projects (DP0986876 and DP120104759), the Melbourne Energy Institute Major Research Projects and Initiatives, Development Support Fund, and the Gilbert Riggs PhD Scholarship (SP). We thank David M. Walker, Jacques Desrues and Pierre-Yves Hicher for insightful discussions.