The Application of Probabilistic Keyblock Methods for Support Design in Blocky Rock Masses

Support design in jointed or blocky rock masses is often approached by the use of either 'rules of thumb', rock mass classification systems such as the Q-system (Barton, Lien and Lunde, 1974) or RMR (Bieniawski, 1973) or conventional wedge analysis. Generally, this requires that a number of key simplifying assumptions are made and the support system is designed to cater for either a mean, most likely, or worst case scenario depending on the risk tolerance. The design block size is often a major uncertainty and often a worst case block size based on excavation span is used. Thompson and Windsor (2007) state that it is inappropriate to base support designs on the excavation size as these size blocks rarely occur if discontinuity trace lengths are considered. Probabilistic keyblock methods can be used to determine the most likely design block sizes. A methodology is presented whereby JBlock software (Estherhuizen, 1996) is used to estimate potential block sizes as part of the support design process. Using information such as the spacing, orientation and length of discontinuities, it is possible to simulate blocks in the walls of an excavation (Esterhuizen and Streuders, 1998). Keyblock analysis methods (Goodman and Shi, 1985) are used to evaluate whether blocks are removable and whether the chosen support will be sufficient to ensure stability. Data from two mines is used to demonstrate how probabilistic keyblock analyses can be used in support design. Discontinuity data was used to generate 10 000 removable blocks and the design block size statistics were determined. For one mine this information was used to scale blocks in a conventional. Unwedge analysis whilst for the other mine, the point estimate method (Harr, 1987) was applied. The choice of appropriate design acceptance criteria in terms of factor of safety and probability of failure is also considered.