Failure localization in rock is observed ubiquitously on geological scales in the form of
fault or earthquake damage structures. Similar failure processes are observed in
confined compression tests carried out on laboratory-scale rock samples. At an
intermediate scale, seismic activity is often associated with the formation of so-called
burst fractures that are intermittently formed and exposed in the vicinity of deep
level mining operations. Computational modeling can assist the understanding of the
complex nature of these failure processes. The present study investigates the question
of how the properties of macroscopic shear band features are controlled by
microscopic constitutive behavior. The computational approach that is used is to
consider the formation of shear band structures by selectively mobilizing members of
an assembly of randomly oriented cracks that are modeled as displacement
discontinuity elements. Particular issues that are addressed are the question of
whether the microscopic failure processes are self-similar to the macroscopic
processes, and how the density of the discontinuity assembly affects the
localization patterning. It appears that the use of slip or tension-weakening
constitutive models yields equivalent “macro” results that are independent of
the “micro” mesh density for a given mesh type. If the intrinsic junction
coordination of the mesh is altered, it is found that the equivalent macro
dilation angle is changed. This has important implications in determining
whether a particular distinct element or lattice model with an intrinsic junction
structure is capable of replicating the observed failure behavior of a given
rock type. A dimensionless parameter group is suggested as a measure of
the intrinsic coordination number for a random crack model of rock micro
structure.
