Non-parametric 3D conditional DFN modeling for probabilistic stability analysis of rock wedges in open pit mine

Mechanical behavior of the rock mass is predominately governed by the presence of discontinuities, especially at surface and near-surface excavations. Given the inherent uncertainty in characterizing the geometrical properties of rock discontinuities, precise and sound simulation of fracture systems becomes crucial for reliably predicting rock mass behavior in engineering geology applications. This paper presents a non-parametric discrete fracture network (DFN) approach to simulate fracture networks within the rock mass, offering an alternative to conventional methods that rely on predefined statistical distributions. The methodology employs bootstrapping techniques to generate a three-dimensional DFN model that better captures the heterogeneity and spatial complexity of natural fracture systems. High-resolution fracture data were collected using unmanned aerial vehicle photogrammetry, providing the basis for the DFN model development. For fracture size characterization, a non-parametric approach was employed to estimate the cumulative distribution function of fracture diameters from observed trace data, modeling the network as a Poisson point process (disc model). Additionally, the P₃₂ fracture intensity parameter was estimated using direct calculation and sequential Gaussian simulation, allowing the construction of a detailed block model. To demonstrate the practical application of this approach, the methodology was applied to a geological sector of the Golgohar open-pit mine, Iran. The developed model was subsequently utilized to evaluate the mine wall stability through probabilistic kinematic stability analysis. This study demonstrates effectiveness of non-parametric modeling in geomechanical applications, offering an advanced tool for analyzing and predicting rock mass behavior.