Discrete Fracture Network (DFN) models can represent discontinuities at a determined scale within a rock mass. A constrained DFN is inherently a more accurate representation of an actual fracture network. Traditionally, deterministic DFN models are constrained to fracture locations. Site conditions affect the scale at which DFN models should be analyzed, the expected variability of intensity, and govern rules that are applied to the determination of in situ block geometry. This paper discusses the analysis of three semi-deterministic DFN models constrained to different input parameters relevant to mine planning and design. A surface study of over 200 km2, an underground excavation, and multiple trace maps of an active face of an open pit mine are considered in this study. Constraints to fracture location, size, orientation, intensity, waviness, and termination are applied to all three models. Analysis of these models considers individual fracture orientation, size, and location with respect to mapped data. Fracture intensities are verified against input parameters and the variability in intensity is assessed. For each DFN model, it is shown that the P32 fracture intensities follow a Power Mean relation. When comparing the coefficient of variation to the sample size, sampled over the entire volume of interest, a Power Law relation is shown to exist. Unfractured blocks can also be located by assessing the sampled volumes with a fracture intensity of 0 for a given DFN. The permitted concavity of identified blocks is controlled by a density parameter that can be modified based on observed site conditions. The scale of analysis for fracture network variability is determined by considering fracture intensity, clustering and the size of the DFN model. A means of customizing DFN analysis to site conditions is presented that also allows for the assessment of emergent properties which can be used to validate stochastic DFN models.

This content is only available via PDF.
You can access this article if you purchase or spend a download.