Abstract
Prediction of rock slope stability has been one of the biggest concerns for open pit mining. Unrealistic assessment of slope stability impacts the safety of the project and can result in damages to workers or property. Poor predictions of rock slope stability can be caused by uncertainties and spatial variability of natural fracture systems that have significant consequence in the context of rock mass failure mechanisms.
Geometric Discrete Fracture Network (DFN) models are built, and restricted and insufficient field structural data are statistically exploited in terms of observed parameter distribution at the mine-scale level. The descriptive statistics of fracture intensities (P10), orientation, and length were used to govern the generation of a DFN model using a disaggregate approach. To assess ultimate model reliability, the DFN model was validated by comparing the orientation, linear intensity, and length mapped in the field to one of the simulated DFN models. The calibrated DFN model was then used in a two-dimensional Distinct Element Model (DEM) to evaluate the stability of a critical pit wall, where several block failures were reported at an inter-ramp and bench scale level.
Using the shear strength reduction (SSR) technique, the DFN-DEM technique used in this work was able to correctly estimate the slope stability condition. The values of the critical strength reduction factor (SRF) generated from each numerical simulation were greatly reliant on the DFN geometry configuration and trace length. A series of models with various DFN configurations were simulated to measure the uncertainties associated with the assumptions set throughout the DFN-DEM modelling process. Over 30 DFN realizations, the slope models were predominantly unstable (mean SRF˂1.0). According to kinematic study performed at the same critical slope, the mechanical behaviour of DEM rock slopes was proven to be more realistic than standard limit equilibrium (LE) models. For the simulated DEM slope models over 30 DFN-DEM runs, the probability of failure (PoF) averaged 70% with a mean safety factor of 0.93, whereas utilizing the LE method, the PoF averaged 40% with an average FoS values of 0.91 and 1.08 (best-case scenarios not included) respectively with scenarios without tensions cracks and with tension cracks estimated using the LE approach.
The DFN-DEM technique used in this research project is capable of identifying accumulated shear zones, simulating the slope stability state with high accuracy, and providing vital information about probable collapse processes. The critical shear strain zone, displacement field (vertical and horizontal) and velocity vectors were used to characterize the probable failure process, demonstrating differences in how the failing surface was created depending on the DFN configuration especially. Additionally, while adopting the distinct element approach, it is critical to define representative block design (rounding and edge minimum length), as well as the meshing size, since these factors impact the representation and location of fracture traces in the DEM models.
Keywords: Discrete Fracture Network (DFN), Distinct Element Modelling (DEM), Strength Reduction Factor (SRF), Slope Stability Analysis