We correlate the failure modes observed in the field with those identified during a model testing program (Sliding on a single plane, stepping, multi-plane stepping, failure through intact rock, sliding and rotation of blocks). Step failure occurs when the major compressive principal stress is nearer the vertical and this mode controls the surface slope of the individual benches for the 800m-high west wall of the Chuquicamata mine; with separation on a continuous joint dipping back into the slope and sliding or toppling on the non persistent joint dipping into the excavation. Mapping of the benches provided an opportunity to do the difficult task of assessing the percentage of joint persistence, using the distribution of non-persistent joint spacings. Having determined the persistence of the joints and the size of the bridges, the potential for a deeper-seated failure in the lower slope of the mine can be assessed. Here, the major principal compressive stress is nearer the horizontal, and can lead to a through-going failure along non-persistent joints and intervening bridges in the lower slope. Additional laboratory tests on model rock slopes prepared with block assemblies illustrate the failure mechanisms.
A complex failure mechanism can develop in a joint mass where the persistence of an adversely oriented joint system is less than the overall slope height. At high values of the confinement stress (σ2/σ>0.07, where ere is the unconfined compressive strength of the intact material), failure occurred through a planar surface, partly along the joints and partly through the intervening intact material. Cording and Jamil (1997) report the results of a model-testing program to evaluate the strength characteristics of non-persistent rock joints. The model rock mass was 610mm high by 305mm wide by 203mm deep and consisted of a sand-cement material having an unconfined compressive strength of 5 to 13 MPa. Patterns of non-persistent joints were inserted in the sample prior to set of the model material using paper strips to form the joint. Not only could rock bridges between the joint segments be modeled with this process, but the resulting joints were tight, without the openness and imperfect fit that is typical of models formed by an assemblage of individual model rock blocks. Biaxial loading was applied to the sample end and sides using hydraulic rams loading a pyramid of simply supported triangular steel pieces so that a uniform load was distributed to each of the 75mm-wide platens set side-by-side along the sample sides and end. The material was free to deform and fail into the sides of the model on which the minimum principal stress acted. The joints consisted of a single set of parallel joints. The main parameters varied in the tests are illustrated in Figure 1. Joint length, Lj, with respect to the length of the rock bridge, L,., was varied, as was the spacing between joints, d, and the orientation of the joints with respect to the principal stress axes, β. The joint step angle (γ) remained 90° in all the tests.