: Predictions of key blocks in underground excavations are uncertain, subject to the natural variation of rock properties. This study illustrates how stochastic fracture concept could be applied to make such predictions more realistic. The methodology was tested on the rock cavern sited in crystalline basement in southeast Sweden. Fracture mapping in boreholes was used to evaluate input variables to the stochastic model. Prior to model derivation, spatial pattern of fracture intensities was investigated by nonparametric statististics, variogram and spectral analysis. Fracture model incorporating fracture orientations, size, location pattern and termination mode was generated. Subsequently, the key block statistics along the cavern was computed. Probabilistic predictions of the number of block and block weight were obtained. To illustrate the value of the predictions made, the block analysis was done for two different tunnel orientations. Some noticeable differences in distribution parameters were observed. The presented methodology offers the possibility to optimize tunnel location and predict rock support.
One of the major hazards to proper functioning of an underground object is related to sliding of rigid rock blocks formed by intersecting rock discontinuities. Work done by Warburton (1981) and Goodman & Shi (1985) belong to the most important contributions in identification of unstable rock blocks from given orientation of fractures with respect to dimensions, shape and orientation of an underground facility.