Advancing continuum and discontinuum models of brittle rock damage and rock-support interaction

Abstract

Stress-induced damage and spalling in underground mines continue to remain a major hazard to mining personnel. In order to reduce worker injuries caused by falls of ground, it is necessary to develop techniques through which the mechanisms of pillar damage can be accurately interpreted and supports can be designed to control the anticipated displacements along pillar boundaries. With advancement in numerical modeling techniques, it is now possible to study such rock mechanics problems in detail, but there is a need for improvement in the available approaches before they can be utilized as tools for the design of ground support schemes. The goal of this research is, therefore, to advance continuum and discontinuum models to better reproduce both observed pillar damage mechanisms and the interaction between supports and (unsupported) ground. The associated findings contribute significantly towards improving our understanding of the phenomenological capabilities of continuum and discontinuum modeling approaches and their applications in different mining scenarios. The contents of this thesis can be broadly sub-divided into two sections – continuum (using Itasca’s FLAC3D software) and discontinuum (using Itasca’s UDEC software) analyses of pillar damage and rock-support interaction. In the first section, a rock yield criterion is developed that considers both brittle fracturing at low confinement and shearing at higher confinement. When implemented in FLAC3D pillar models, results consistent with the empirical trend of pillar strength as a function of width to height ratio for granite, conglomerate, and coal were obtained. In terms of site-specific case studies, pillar displacement and stress data from two different sites could be reproduced using this yield criterion. Subsequent investigations on rock-support interaction revealed that continuum models tend to underestimate the effect of support on otherwise unsupported ground, and accordingly is limited in its potential application as a support design tool. In the second portion of this thesis, the Voronoi Bonded Block Modeling (BBM) approach is employed, which represents a material domain by an aggregate of polygonal blocks. Laboratory-scale modeling of a granitic rock was pursued to understand how decisions related to model setup affect the ability of such grain-based models to reproduce various deformation mechanisms. Ultimately, a BBM representation utilizing different elastic properties for the different mineral grains along with inelastic grain properties was necessary to match the pre- and post-peak attributes of the granite under consideration. The input properties from the laboratory-scale models, however, were found to not be directly applicable to pillar-scale simulations because much larger blocks sizes were used in the latter case and blocks sizes are known to exert a significant influence on the macroscopic behavior of BBMs. Accordingly, pillar-scale BBMs were calibrated independently, although the findings regarding model setup decisions from the grain-based BBM are, in part, transferable across scales. For example, the pillar BBMs required an inelastic block representation to simulate the damage process within the confined sections of the pillars, similarly to how the laboratory-scale model also required inelastic grains to replicate the high confinement attributes observed in laboratory triaxial tests. The analysis of rock-support interaction was conducted by comparing the lateral displacements along the pillar edges without support and with various support patterns for both the polygonal and the triangular (also called Trigon) block geometries. The polygonal BBM produced behavioral differences that were closer to empirical field-data assembled from hard rock mines in comparison to the Trigon models. Coal pillars were simulated with elongated, inelastic Voronoi blocks to account for the anisotropic cleating of coal. For the case of the West Cliff longwall mine, this model representation was able to reproduce the pillar displacements at two neighboring sites that had different support patterns by modifying the support in a BBM calibrated to one site to match the support at the adjacent site. This is perhaps the first study to quantitatively demonstrate that BBMs can replicate the influence of rock reinforcement on ground behavior in rock undergoing spalling. Lastly, as BBMs are computationally intensive and cannot easily be applied at the mine-scale in 3D, an integrated modeling approach was established using two different mining case studies. In this approach, the larger-scale stress distribution was assessed using FLAC3D, and the deformation behavior near excavation boundaries (with and without support) under the expected loading path was estimated using a BBM. This study, as a whole, has demonstrated the capabilities of continuum and discontinuum modeling approaches under a variety of conditions, and the proposed approach for studying ground-support interaction has the potential for practical application in the context of site-specific support design.