Genesis of Failure in Granular Materials: Bridging Physics and Engineering Perspectives

Mehdi Pouragha

When subjected to mechanical loading, granular materials undergo various deformational regimes, often culminating in a strength threshold as failure. In the field of continuum mechanics, the classical theory of elasto-plasticity has been successfully augmented with the concept of ``critical state'', resulting in constitutive relations that are capable of capturing some of the most intricate aspects of the granular media. However, the advances over the past 50 years in micro-imaging, together with the development novel numerical simulation tool such as Discrete Element Modelling (DEM), painted a different picture in which granular media is thought of as an assembly of individual distinct particles that interact with each other through numerous discrete interactions. This immediately sparked new connections with ongoing topics in statistical physics where systems comprising large number of interacting bodies have been studied. This study  outlines a micromechanical framework that attempts to bridge between the notion of phase transition in physics and failure in engineering mechanics of granular media. By expanding upon the idea of jamming transition, a so-called ``Stable Evolution State, (SES)'' is introduced  as a manifold in the space of  micromechanical variables that signifies the onset of irreversible deformations. We argue that SES suggests a coherent framework for unifying of jamming transition and the concepts of yield limit and critical state in elasto-plasticity theory. Moreover, our detailed study of the contact network topology reveals different mechanisms contributing to the evolution of microscopic fabric and the rule they play in reaching failure. A statistical analysis of meta-stable micro-avalanches casts the path towards failure as a self-organized criticality with the irreversible deformations conceived as temporary excursions beyond SES limit. Finally, the outlines of a new constitutive modelling framework is described that is developed primarily based on the concept of SES.