Improving Continuum Models for Excavations in Rockmasses Under High Stress Through an Enhanced Understanding of Post-Yield Dilatancy
MetadataShow full item record
In recent decades, the field of rock engineering has seen an increased use of numerical modelling tools to aid in the analysis and design of underground excavations. Although complex numerical methods have been developed to explicitly capture the discontinuum mechanical processes which dominate rock and rockmass deformation, continuum methods represent a viable alternative due to their relative advantages in model runtimes and parametric simplicity. This thesis aims to advance the state-of-art and state-practice techniques for continuum models and to demonstrate their ability to replicate behaviours observed in-situ. With recent advances in the understanding of rockmass strength, it is the post-yield phenomenon of rock dilatancy which has remained the domain of greatest uncertainty in constitutive models. Particularly in the case of brittle rockmasses, where deformation is dominated by crack extension and dilation, the post-yield evolution of dilatancy in parallel with strength parameters can have a very significant impact on progressive failure processes. By improving our understanding of this evolution, continuum models for brittle rock can be improved. In the case of weaker rockmasses which can be adequately represented using a conventional strain-softening approach, an existing model for dilatancy can be applied. As is shown in this thesis, the use of a constant dilation angle to approximate the dilatant behaviour of the rockmass in this case is appropriate; accordingly, parameter selection guidelines are presented. In the case of brittle rockmasses, a mobilized dilation angle model which incorporates the strain and confinement dependencies of dilatancy is required. Through an analysis of laboratory testing data, a new model for the dilation angle is developed. Then, using five case studies (both from the literature and novel to this thesis), the applicability of the model to in-situ rockmass behaviour is demonstrated. Finally, the implications of these model results for understanding support-rockmass interaction and strainbursting phenomena are discussed.