Abstract

This paper discusses a new computational strategy for the analysis of inelastic processes in granular rocks subjected to varying levels of confinement. The purpose is to provide a flexible and efficient tool for the analysis of failure processes in geomechanical settings. The proposed model is formulated in the framework of Lattice Discrete Particle Models (LDPM), which is here calibrated to capture the behavior of a high-porosity rock widely tested in the literature: Bleurswiller sandstone. The procedure required to generate a realistic granular microstructure is described. Then, the micromechanical parameters controlling the fracture response at low confinements, as well as the plastic behavior at high pressures have been calibrated. It is shown that the LDPM model allows one to explore the effect of fine-scale heterogeneity on the inelastic response of rock cores, achieving a satisfactory quantitative performance across a wide range of stress conditions. The results suggest that LDPM analyses represent a versatile tool for the characterization and simulation of the mechanical response of granular rocks, which can assist the interpretation of complex deformation/failure patterns, as well as the development of continuum models capturing the effect of micro-scale heterogeneity.

1. INTRODUCTION

An accurate knowledge of the engineering properties of rocks is crucial for a variety of geomechanical problems, ranging from wellbore stability, to failure in rock slopes, underground excavations, and crustal faults [1]. While strength and deformation properties are usually obtained from a limited number of in situ and/or laboratory tests, their determination is invariably affected by considerable heterogeneities [2]. Such lack of homogeneity impacts engineering conclusions at all length scales and requires appropriate theoretical and computational tools.

Advanced numerical modeling represents a useful tool to explore how mechanical processes interact across length scales. Considerable advances in this area have based on Finite Element computations, where heterogeneities can be incorporated at both sample and site scales [3-5]. Nevertheless, to capture realistically the path-dependent response of geomaterials, continuum formulations tend to be characterized by a large number of parameters. If such constants lack clear connections with measurable attributes (e.g., grain size and sorting), their calibration becomes poorly constrained. Furthermore, the tendency of rock samples to undergo strain localization [6, 7] further prevents the validation and/or implementation of continuum models, requiring a direct link between strain localization and microstructural attributes [8].