Direct shear failure of a synthetic rock containing discontinuous joints

Direct shear tests were used to establish the shear behaviour of continuous planar-joints, discontinuous stepped-joints and discontinuous open-joints. The joints were cast in a Synthetic rock made of plaster, sand and water and tested under normal stresses that ranged from 50 kPa to 3.5 MPa. A total of 88 direct shear tests were carried out. The shear behaviour of both the continuous and discontinuous joints was found to be dependent on the normal stress. At normal stresses below the magnitude of tensile strength, approximately 1.8 MPa, the joint behaviour of the continuous planar joints was essentially elastic-perfectly plastic. Implying that the peak strength of the planar joint was essentially the same as the residual strength. At normal stresses above the magnitude of the tensile strength, continuous and discontinuous joints displayed either strain weakening or brittle behaviour. Hence at these normal stresses the peak strength exceeds the residual strength. No single failure envelope could be used to describe the shear of the joints, even the continuous planar joints. A primary reason for this non-unique failure envelope was the large dilation that occurred at high normal stresses. This dilation was attributed to grain crushing, and the roughness resulting from this crushing and gouge formation as shearing occurred.;The Phase2 elasto-plastic finite element software was used to simulate a number of the direct shear tests. The properties for the synthetic rock were established from uniaxial compressive strength and Brazilian tensile strength tests. The Phase2 simulations were in reasonable agreement with the laboratory direct shear tests for the continuous planar-joints and the discontinuous stepped-joints when the normal stress was less than the magnitude of the tensile strength. However, at normal stresses above the magnitude of the tensile strength, the agreement between Phase2 and the laboratory tests was reduced. There was essentially no agreement between the Phase2 results and the discontinuous open-joint laboratory results. These findings suggest that the material properties for a continuum model may have to be calibrated to the laboratory results that were determined following the stress path simulated in the continuum model.