In this thesis a previously unknown mechanism of failure in multilayered slope profiles is identified. In some conditions this mechanism does not confirm to the known failure models (relating to circular failure) used in slope stability analysis. For this reason, major failures have occurred in the artificial cuts despite the fact that the limit equilibrium methods suggest that these cuts would be stable. The limit equilibrium methods were originally created to apply to earth dam walls. In the open pit mining environment, where we face inhomogeneous and inclined multilayered structures, the assumptions of these limit equilibrium methods appear to be inapplicable (e.g. assumption for the equal shear strength along the failure surface). Analysis starts with a general picture of the stress state in the highwall slope, given extant geological conditions and rock properties. The study then focuses on a comparison of the crack-tip stress changes in the rockmass with and without inclusions at the microscopic level. Basing some assumptions on binocular microscope observations of grain structures, it is possible to measure the size of the different inclusions and show that the microscopic carbon flakes present in the rock fabric make a major contribution to the failure process in a mudstone layer in the slope. The approach adopts the fracture-process zone ahead of a crack tip as the controlling parameter of flaw propagation in rock. Flaw coalescence, which is poorly accounted for in current fracture models, is attributable to two phenomena: the flaw propagation due to high level of applied stress; and the linking of fracture-process zones due to the small distance between neighbouring flaws. A condition of flaw coalescence is given based on these two mechanisms. This development allows defining of two zones along the failure surface (frictional and cohesive). In the slope-stability field the shear strength of the rock along the failure plane is a composite function of cohesive and frictional strength. For instance, the relaxation stress normal to bedding, induced by overburden removal, provides an investigation method for the determination of the weakest minerals, which may act as flaws for fracture propagation in low-porosity rock. A method has been developed to determine the critical stress for tensile fracture propagation due to the rock structure and the stress reduction normal to bedding. A proposed failure mechanism is based on the polygonal failure surfaces theory developed by Kovari and Fritz (1978), Boyd’s field observations (1983), Stead and Scoble’s (1983) analyses, Riedel (1929) Shear Fracture Model, Tchalenko and Ambraseys (1970), Gammond’s (1983) and Ortlepp (1997) observations for natural shear failures, computer modelling by McKinnon and de la Barra (1998), the results of many laboratory experiments reported by Bartlett et al. (1981) and the author’s experience. The proposed failure mechanism evaluates stability of the artificial slope profile due to the embedded weak layer structure, layer thickness, layer inclination and depth of the cut. On the basis of the observations and the above-mentioned modified fracture model, the slope profile is divided into two blocks; passive and active blocks. With this new model, it is possible to calculate slope safety factors for the slope failure cases studied in the industry. It has been found that, whereas the conventional slope stability models predict stable conditions, the new model suggests that the slope is only marginally stable (i.e. that failure can be expected). / Thesis (PhD(Mining Engineering))--University of Pretoria, 2007. / Mining Engineering / unrestricted