ABSTRACT

The realistic modeling of rock slope failure processes is a complex task. Fractures affect almost any aspect of the hydro-mechanical response of the rock mass dramatically. Discrete Fracture Network (DFN) is a numerical approach to numerically generate synthetic rock masses and study how the presence of fractures in a rock volume influences its behavior. However, to further expand their applicability, DFN models need to be meshed in order to be coupled with other numerical models, e.g., stress and strain evaluations, fracturing and failure analysis. This process is highly sensitive to the quality of the generated mesh as a "bad mesh" it could lead to numerical instabilities in the consequent numerical models. There are two main approaches to achieve a quality mesh: 1) optimizing an existing mesh 2) pre-conditioning mesh. In this paper, we discuss challenges in meshing DFN models, introduce solutions, and demonstrate how effective they could be.

1. FOUNDATIONS OF DISCRETE FRACTURE NETWORK MODELING

Discrete Fracture Network (DFN) modeling is based on stochastic modeling in which parameters of individual fractures are defined following probabilistic distribution functions (PDF) such as Gaussian (normal), exponential and uniform PDFs (Fadakar-A, 2017). The arrangement of fractures in the space defines the network with an associated topology. The modeling process often involves multiple simulations (e.g., Monte Carlo), i.e., resulting in a set of equally probable realizations; hence, the uncertainties associated with the input parameters, as well as with the results (i.e., the output) are effectively encapsulated. Since DFN models are numerical and statistical simulations, the process of modeling is heavily affected by the computing power. Practically, depending on the scale of the problem of interest, a DFN process may include one single fracture up to over several billion fractures in a single model. Furthermore, favoring realistic three-dimensional (3D) DFN models introduces new and often difficult challenges, primarily due to the complexity of associated geometry and the topology complications. In a DFN model intersections between fractures defines the topology i.e., inter-connections and connectivity (Fadakar-A et al., 2011–14).

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