I employ the Discrete Element Method to analyze the micromechanical response of granular rocks to unstable failure. Calibrated granular models of sandstone and granite are subjected to biaxial experiments under confining pressures of 0–50 MPa, leading to the development of shear fractures through interaction of emergent microcracks occurring in shear and tensile modes. I document the mode and energy associated with microcracks to analyze fracture growth patterns and quantify energy release. Shear fracture growth in my sandstone model occurs through cooperative interaction between shear and tensile microcracks, with shear microcracks accounting 4-44% of total microcracks and 31-92% of fracture energy. Shear fracture growth in my granite models occurs through coalescence of tensile microcracks, which account for 96-98% of total microcracks and 87 -93% of fracture energy. My model results show that fracture energy increases with confining pressure, accounting for 10-15% of the total input mechanical energy in sandstone vs. 16-47% in granite. I estimate that the work done against friction from intergranular and fracture sliding accounts for 69-86% of total input energy in sandstone models and 46-81% in granite models. My results indicate that frictional deformation is a significant term in the energy budget during rock deformation. I employ the Discrete Element Method to analyze statistical indicators of critical failure during shear fracture growth in calibrated models of sandstone. To investigate the precursory signatures of critical failure, I document the location, mode and stress associated with emergent tensile and shear microcracks during biaxial experiments under confining pressures of 0-45 MPa. I employ the documented microcracking activity to calculate temporal trends of microcracking variance, fraction of shear microcracks to total microcracks, fractal dimension of microcrack locations, acoustic energy variance and seismic b-value. My results indicate that each parameter is a function of strain-to-failure and can be treated as an indicator of critical point. Pre-failure damage evolution in our sandstone model is characterized by: 1)increase in microcracking variance of one to two orders of magnitude, 2)peak shear microcrack fraction 3) peak fractal dimension of microcrack locations, 4) increase in acoustic energy variance of an order of magnitude, and 5) peak b-values. We employ the five microcracking indicators and confining pressure as inputs for an artificial neural network (ANN) to predict critical failure. Over confining pressures of 0-45 MPa, our ANN architecture exhibits good prediction capability for stress-to-failure and strain-to-failure for our sandstone model. Our machine learning approach reveals that microcracking variance, seismic b-value and fractal dimension are the most important indicators of critical failure. Thus, we develop an integrated analysis of microcracking indicators to understand signatures of critical point and combine them with machine learning to predict failure in granular rock. I model mudstone permeability during consolidation and during fluid injection by simulating porous media flow using the lattice Boltzmann method. I define the mudstone structure using clay platelet thickness, aspect ratio, orientation and pore widths. Over the representative range of clay platelet lengths (0.1 – 3 µm), aspect ratios (length/thickness=20-50) and porosities (0.07 – 0.80), my permeability results show good correlation with natural mudstone datasets. Over =0.32-0.58, the porosity-permeability trends of two mudstone models of heterogenous mineralogy match experimental datasets well. I extend my methodology to evaluate how mudstone permeability might evolve during microfracture network growth or macrofracture propagation upon fluid injection. My results suggest that the growth of a distributed microfracture network results in greater vertical permeability increase than a single macrofracture upon fluid injection in compacted mudstones. My modeling approach provides a simple means to estimate permeability during burial and compaction or fluid injection based on knowledge of porosity and mineralogy.