Improved reliability in rock mass characterisation for underground support design - Discrete fracture network model and site observation

Conventional empirical methods for rock mass characterisation can only offer limited understanding and qualitative statements of rock mass condition based on past experience in similar geological conditions. Quantifiable measures; such as possible block size, block shape, fracture trace length, etc; are not directly obtained. When possible, determination of these parameters often involves a convoluted process. It should be acknowledged that these factors influence the behaviour of a rock mass during excavation and sole reliance on empirical methods potentially constrains our understanding of rock mass behaviour during excavation. This can lead to less than optimum designs or in more severe cases rock mass failure, if the governing failure mechanism is not properly understood. The study in this paper attempts to evaluate the applicability of a DFN approach as a forward modelling tool in estimating rock mass condition using data from a completed underground project to verify the proposed methodology. This paper demonstrates the DFN models ability in estimating ranges of potential fracture intensities and fracture trace lengths to be encountered on site during excavation. The DFN predictions have been compared to the actual data mapping data to assess the effectiveness of the method. The DFN models exhibit 'resemblance ', albeit not an exact match, to the actual tunnel face conditions encountered. The results and comparisons described in this paper also demonstrate that adopting DFN approach can allow more objective, realistic, detailed, and site-specific rock mass characterisation and risk assessment work to be completed when undertaking the design, by reducing the influence of personal bias, experience and engineering judgement. Lastly, the study results also indicate that the DFN approach, utilising more realistic rock mass models, also offers opportunities in optimising underground support design compared to traditional approaches. Despite the benefits, the inherent limitations of DFN models highlighted at the end of the paper, still need to be taken into consideration when preparing a design.