ASSESSMENT OF ROCK MASS CHARACTERISTICS AND SLOPE STABILITY USING 3D NUMERICAL MODELING AND DISCRETE FRACTURE NETWORK IN GARHWAL HIMALAYA, INDIA

Abstract

Rock mass exhibits wide heterogeneity due to the existence of micro to mesoscale discontinuities. Mesoscale fractures control the behavior of the rock mass and microscale fractures influence the mechanical properties of the intact rock or intact rock bridges. The formation of regional geological structures such as faults and folds contributes to the complexity of rock masses due to intense shearing and fracturing. The concept of damage applied to intact rock and rock mass relates to the degradation of their strength properties. Fracture propagation occurs within the natural material because of the applied stresses; still their distribution may be halted by the eradication of applied stresses or the existence of a weak interface-like joint that arrests the propagating crack. Due to heterogeneity in the rock mass, there is high variability in joint characteristics; researchers suggest to uses statistical distributions to cope with such uncertainties. Stochastic discrete fracture networks (DFNs) provide a genuine approach to build realistic 3D fractures system based on the statistical distributions of discontinuity parameters. The circular window mapping can help to provide unbiased joint trace-related data from 2D outcrops. The 2D data can be converted into 3D using stereological techniques backed by probabilistic approaches. Remote sensing techniques such as LiDAR and photogrammetry made it possible to render the data with high precision potential not accessible with conventional methods. Developed high-resolution 3D point cloud models help to provide structural data such as trace length and orientation. The DFN must be validated with the actual mapped data and can be used to examine the rock mass characteristics. Different field-based methods exist in the literature and are used to characterize rock mass conditions. The Geological Strength Index (GSI) is a simple and commonly adopted method with broad applicability in rock engineering. The traditional approaches are limited to 2D exposures for mapping purpose, but block formation or a joint intersection is a 3D parameter. This work addresses the incorporation of the discrete fracture network (DFN) for estimating the GSI of the rock mass. The work compares the results of DFN generated using aggregate and disaggregate approaches in block size distribution (BSD) and Rock Quality Designation (RQD) measurement for a fractured rock slope. The calculated BSD and RQD using DFN and field-estimated joint conditions parameters are used to estimate the GSI of the rock mass. A machine learning-based python GUI tool was developed to find GSI from block volume and joint condition parameters. Using machine learning to predict GSI from input parameters has led to systematically digitizing the standard GSI chart. Conventional rock mass classification methods such as GSI, Jv (Joint density) and RQD focus on block size formed by the intersection of discontinuities. Characterizing the size and shape of the congregation of blocks together in the rock mass provides a comprehensive insight for studying its engineering properties. The shape of blocks can help to provide information related to the overall jointing pattern in the rock mass and even potential failure modes. Numerous methods exist in the literature to classify the shape of blocks using simple mathematical relations or numerical modeling techniques. Among them, the Block Shape Classification Method (BSCM) has been widely employed to understand the nature of block shapes present in the rock mass. The existing block shape classification method (BSCM) considers two factors: α describes the shortening of the minor principal axis of the block, and β describes the elongation of the major axis. The parameter β used the average angular relation between the chords greater than the median, considering the average angle could produce skewed results towards elongated blocks. This study proposed a Modified Block Shape Classification method (MBSCM), where parameter β is provided with a new definition and procedure for calculation. To estimate the elongation index (β), the maximum angular extension between the chords was considered, and the parameter α remained unchanged in the modified approach. The developed method was validated with synthetic rock masses of known shapes (i.e., cubic, elongated, elongated platy) constructed in the 3DEC (Three-dimensional distinct element code). Two case studies were conducted on the Himalayan slopes to demonstrate the new method’s applicability. The DFNs were generated for both slopes to find the block-related data formed by the intersection of the fracture network. The rock masses were classified based on the proportion of different shapes and sizes of blocks. The shape of blocks determined using a modified method matched the visual inspection and geometrical characteristics of blocks in the field. The slopes were also classified using the existing method to compare the outcomes. The result shows that the existing method categorized about 5% more elongated blocks than the proposed modified approach. An open-source python-based computer software (MBSCT.exe) was developed for the modified approach and was successfully used to plot the classification diagrams directly by importing the raw data file. Further, a study was conducted to understand the role of regional tectonic structure on rock mass characteristics and block kinematics. High variability in fracture intensity and trace length has been found as one moves away from the hinge zone due to damage caused by the regional syncline in Garhwal Himalaya. The high in-situ stresses in the fold hinge zone during folding propagated the fractures utmost and resulted in a high frequency of mean-sized fractures. The high fracture intensity and nearly equal joint traces intersected comparatively smaller-sized blocks. Further, at sites 2 and 3, away from the hinge zone, the mean joint spacing was comparatively high due to low fracture intensity. The lower fracture intensity is a result of declined in-situ stresses along the limb during folding that halted the fracture propagation and resulted in a comparative larger-sized block. The increase in GSI value was observed away from the hinge region. The DFNs for all three sites were imported into 3DEC to analyze the block kinematics. The study concludes that the anomalous high fracture intensity due to intense fracturing in the fold damage zone reduced the rock mass strength, providing the kinematic freedom for block movement resulting in slope failure. Whereas, at sites 2 and 3, due to low fracture intensity, the increase in block size could have provided kinematic stability to the blocks.