Open Cut Mine Wall Stability Analysis Utilising Discrete Fracture Networks

Identification of geotechnical hazards in pit walls is an ongoing operational requirement in open pit mining. Improvements in techniques for hazard identification have significant safety implications as open cut operations get deeper and the risks of wall failure increase. The hazards of interest for this 12 month project were those relating to structurally controlled failures such as plane or wedge failures. The goal of the project was to develop a methodology based on the principles of uncertainty quantification for the identification and prediction of these hazards in open cut high walls.

The main outcome of this work has been the development of a quantitative risk analysis methodology for automated prediction and analysis of geotechnical hazards. The methodology has shown itself to be amenable to implementation as a software tool.

Three technologies had been identified in previous research as being potentially useful for the project, namely 3D image analysis, discrete fracture network (DFN) generation and polyhedral (rock block) modelling. The work program addressed the need to investigate the development and modification of these technologies and associated algorithms for suitability for the specific geological and structural conditions present in open cut coal mining. To aid in this development, 3 mine sites were visited for data acquisition and development and validation of the proposed methodology. The sites showed large variability in terms of structural characteristics and previously identified stability issues. Thanks to the generosity of the project sponsors, one site in particular ('site A') proved extremely useful for this research in that several follow up visits were arranged allowing multiple highwall strips to be analysed. A valuable and detailed structural data set has been compiled for this site and much of the analysis in this report has benefitted from this.

The detailed methodology allows the geotechnical practitioner to determine uncertainties associated with the prediction of high wall stability issues. Further, this research has allowed several important conclusions to be made regarding the suitability of the technologies for this methodology. These include:

· Detailed digital mapping of 3D images has allowed large areas of highwall to be mapped for metre-sized structures in several hours;

· 3D image analysis utilising artificial lighting and surface colouring has allowed the identification of structure sets not easily apparent due to both ambient lighting and/or dense bedding;

· 3D image analysis bridging tools have been used successfully to bridge structural traces mapped separately but belonging to the same larger structure. An algorithm for partially automating this procedure has been developed;

· Guidelines for importing discontinuities for DFN generation to replicate observed failure frequency have been developed;

· DFN generation for folded sedimentary structures has been developed;

· A methodology for attaching uncertainty to 'deterministic' structures has been developed; and

· Various sources of uncertainties have been identified and mitigation methods proposed.

The main recommendations resulting from this work include the continuation of industry trials to aid development of a software implementation of the methodology, and extension of the data set compiled for site A.