Instability of overhanging cliffs depends mainly on rock mass structure and on tensile stresses that develop at the base of the slope. In this paper we present stability analysis of a 34m high overhanging cliff, transected by closely spaced horizontal beddings and three sets of vertical joints. The upper third of the cliff is cantilevered and extrudes more than 11m beyond the toe of the slope, giving rise to eccentric loading at the base of the slope and buildup of tensile stresses within the rock mass. Field observations suggest that the vertical joints which transect the entire cliff form "tension cracks" at the back of the cliff, but their distance from the face is uncertain. Yet, the nature of deformation depends upon the exact location of the vertical tensile crack. The stability of the cliff under different geometrical configurations was studied using continuous 2-D FEA and 2-D Discontinuous Deformation Analysis. Both FEA and DDA are shown to agree with field observations. Based on computational results rock bolt reinforcement was added to the DDA model. Optimal reinforcement scheme was determined using kinematical based criterions.
The stability of rock slopes is typically controlled by the rock mass structure, namely by the orientation, extent and density of the discontinuities in the slope. In a rock slope where the discontinuities do not dip out into the excavation space deformation and failure may be controlled by complex processes such as tilting, sliding, and block rotation. In cases where the rock mass contains subvertical joints columns of massive rock blocks may form and rotational movement may ensue . Rotational failures can be broadly classified into two categories: 1) slumping ? backward rotation; and 2) toppling ? forward rotation . Slumping occurs where sub-vertical joints dip towards the excavation space but do not "daylight" whereas toppling occurs where sub-vertical joints dip into the rock. A comprehensive review of failure modes in rock slopes is presented by Goodman and Kiefer .
Stability analyses for rock slopes can be broadly classified in to two main categories: 1) Limit Equilibrium Methods (LEM) and 2) Numerical Methods. For failure in rotation several LEM techniques are available: Janbu simplified method , Bishop?s method  and Wittke?s method  can be used for slumping, the Goodman and Bray method  can be used for toppling. The main disadvantages of LEM are: 1) kinematics is not accounted for; and 2) the failure mode is assumed in advance, prior to analysis.
Numerical methods such as the Finite Element Method (FEM) or the discontinuous Discrete Element Method (DEM) are effective in the solution of complex problems and are also capable of simulating coupled processes.