Investigation on the Spatial Variation in Fracture Orientation and its Impact on Hydraulic Behavior of Rock Mass

The consideration of spatial variation in fracture distribution within a rock mass is crucial for enhancing the reliability of Discrete Fracture Network (DFN) modeling because, in nature, fracture properties exhibit non-uniformly distributed characteristics. Few studies have explored the degree of spatial variation within rock masses and its relationship with hydraulic performance. This study investigates the spatial variation of fracture orientation using outcrop data collected from various regions. Twenty-one trace maps were divided into several sub-domains, and the spatial variation between them was analyzed using a correlation coefficient. We find that trace maps exhibit spatial variation in fracture orientation. The degree of variation significantly depends on the region from which the outcrop data originates. It is less influenced by the scale of outcrop maps or the quantity of fractures within the maps. Furthermore, hydraulic analysis was conducted for two trace maps exhibiting high and low spatial variation. The results reveal that high spatial variation leads to larger discrepancies in directional permeability. These findings emphasize the importance of considering spatial variation in fracture orientation when assessing hydraulic behavior in rock masses for application problems. 1. INTRODUCTION Fracture networks are essential when evaluating the hydraulic behavior of rock masses since fluid flow mainly occurs through fractures rather than rock matrix (Min et al., 2004). Discrete Fracture Network (DFN) modeling, based on statistical characteristics derived from outcrop data, is commonly employed to investigate the permeability of rock masses (Davy et al., 2018). Numerous studies have analyzed the hydraulic performance of rock masses using DFN (Min et al., 2004; Baghbanan and Jing, 2007; Youn and Gutierrez, 2011). Monte Carlo simulations have been utilized to assign fracture properties for generating DFN models, followed by permeability calculations for various model sizes. The concept of the equivalent discrete fracture network (E-DFN) has been also explored to propose a method for generating DFN with low computational cost (Ren et al., 2017; Ma et al., 2020). Recently, the spatial distribution characteristics of fractures have emerged as an important consideration. Fracture distribution in real rock masses often exhibits spatial variation or spatial correlation due to factors such as deformation zones or changes in stress regimes (Selroos et al., 2022). These features should be taken into account when modeling DFN for sites that require high precision such as nuclear waste repositories, yet existing methods cannot reflect these characteristics.