A discrete fracture network approach for the design of rock foundation anchorage

Abstract

A major consideration in the design of high-capacity tiedown anchors for dams, bridges and tower foundations includes the tensile resistance of the rock mass to pullout typically as a result of overturning moments or hydrostatic uplift. Rock mass pullout capacity for installed anchors is developed from the tensile strength and fracture propagation properties of intact rock, the orientation and physical properties of the discontinuities and anchor confinement at depth. The typically assumed but conservative design approach is to assume that only the dead weight of a uniformly shaped inverted ‘rock pull out cone’ provides resistance to anchor pullout with an assumed initiation point and breakout angle. It is appreciated that the ‘dead weight’ cone assumption may be valid if the discontinuities are continuous and the geometry allows for a discrete block to form in the area of anchorage. However, if these persistent joints are not ubiquitously developed this failure mechanism may not represent the conditions across a foundation and can lead to a very conservative anchor designs. A less conservative approach is to consider the deadweight and the tensile strength across the surface area of the assumed pullout cones. Strength estimates are often estimated by a designer using correlations based on the Hoek–Brown failure criteria or the Barton Q-system. However, fundamental assumption of these empirical methods is that the discontinuities are sufficiently closely spaced such that the rock mass can be considered homogenous and isotropic – and this often not the case (Hoek, Carter and Diederichs 2013). In reality, at the scale of many rock anchor pullout problems, the strength and deformation properties of the rock mass are anisotropic, controlled by a few persistent discontinuities and small ‘rock bridges’ that exist between naturally occurring joints. To accommodate scale effects and the variability of jointing, this paper introduces a reliability-based design (RBD) approach for anchors that uses discrete fracture networks (DFNs) combined with numerical models. With this approach, limitations of the current practice for anchor design are addressed and opportunities are identified to optimise rock anchor designs provided that sufficient information is available to produce a representative DFN. Typically, this requires geological mapping to characterise the rock mass and verify the conditions at discrete anchor locations and in situ anchor testing can be carried out during construction.