In this project, we have described new developments in laboratory studies and numerical modeling that have contributed to our understanding of caprock integrity and its response to large-scale CO₂ injection. The main focus has been on understanding the controls on initiation and/or activation of fractures in caprocks, specifically mudstones and shales. The series of investigations have focused on different facets of this problem, covering mechanical, chemical, hydrological and multiphase flow impacts. New data have been collected and novel numerical methods have been derived. These studies have ranged from core-scale to basin-scale investigations.

The results from this study, although extensive, are not exhaustive. Many of the results were quite fundamental, yet with important implications for our ability to predict the impact of CO₂ injection on caprock integrity. The value of such a wide-ranging study is that it allows us to propose recommendations for future research and innovation that target needed data and models.

• Mechanical properties of fractured media: Data obtained in this study, as in many others, are restricted to clean fractures (either natural or artificially created) within a single “uniform” material. The mechanical properties of composite fractures are largely unexplored. Particularly with regard to faults, there is a need for composite fractures that represent the rock types found within a fault core, i.e. sands, clays, shales, basement, etc. The standard suite of testing also does not explore the impact of field conditions on fractures, i.e. exposure to continue stress and also CO₂ Fractures that slip in the field will continue to be exposed to stress after failure, and data are needed to quantify not just the initial failure point, but also repeated failure curves in a sequence. For example, more studies are needed to quantify how compaction and/or dilation after a shear event enhances or reduces mechanical integrity. For CO₂ exposure, studies are needed to understand whether clay smear in the fault core become more brittle or ductile when CO₂ enters the system.
• Geochemical aspects of fracture leakage: Intriguing evidence from small laboratory and theoretical studies point to fast-evolving geochemical reaction along fracture surfaces as having impact on flow behavior if a fracture becomes activated. Little is known about the extent to which sub-microscale (surface processes) affect meter-scale processes. Chemical impacts need to be expanded from small-scale experiments of isolated processes (calcite, salt precipitation, swelling) to large-scale hydro-chemical-thermal-mechanical simulation studies of a flowing fracture under realistic input and boundary conditions. Studies are needed at scale to determine if fractures will clog/disintegrate, or swell/shrink when exposed to reactive CO₂-laden fluids. For example, larger-scale simulation studies can be designed to model the calcite deposits observed in shale outcrops, such as the Kimmeridge clay. Benchmarked simulations can be used to understand the conditions when geochemical reactions in faulted shales are important and if they impact fault flow properties.
• CO₂ impact on mechanical stability: As the mudcrack numerical experiments have showed, change in fluid saturation can have a significant impact on the stress environment, leading to significant fracturing in dried out clays. These experiments were for air-water systems at the ground surface, but it raises important questions about whether the impact of CO₂ fluid saturation at depth may be important for the mechanical stability of caprocks. More research is needed to explore the possible saturation-dependent impacts of CO₂ on shales, particularly in shallow, unconsolidated zones. The possible impact of chemistry on mechanical stability is also a largely unexplored area of research. More investigations are needed to understand how CO₂ exposure affect mechanical properties of a fracture, i.e. how cohesion/friction parameters change with exposure to a reactive fluid. Here, the impacts may be due not just to CO₂ but also to other reservoir treatments that the reservoir may have experience in a previous EOR setting (e.g. low-salinity, seawater injection).
• Moving across scales from a single fracture to multi-fracture systems and quantification of effective parameters: There is a very large gap between our understanding of single-fracture deformation and flow to large-scale behavior of fracture systems connected to pressure dynamics within a CO₂ storage reservoir. We currently have all of the components: high-resolution datasets of single fracture deformation and flow, state-of-the-art numerical methods for a meter-scale multi-fracture network subject to changes in stress and pressure, and robust reservoir simulators for the larger scale system. However, there are critical gaps at the interface between these various components. First, more sophisticated analysis is needed of existing data to build constitutive models for fracture permeability. These need to be incorporated into multi-fracture flow models at the 10-meter scale to simulate the dynamics of a fracture network subject to realistic boundary conditions, ideally provided by the reservoir simulator. But in the end, complex fracture systems (most likely associated with faults) must be reduced to only the essential information and processes when fully coupled into a reservoir system. More research is needed to bridge the gap between fine-scale detailed fracture characterization and effective fracture/fault models for use in large-scale simulation studies, thus improving existing workflow for fault and fracture seal analysis.
• Large-scale reservoir simulation and fracture/fault leakage: Simulation technology at the reservoir scale is quite mature, and very large-scale methods such as VE have advanced quite significantly in recent years. The main research should be directed towards handling of fracture/fault fluid migration within very coarse grids and streamlining software with other methodologies such as needed for uncertainty quantification and optimization. For fault fluid migration, improvements beyond the standard approach in commercial simulation, which tunes transmissibility factors to get the desired cross-fault flow but are not able to handle the more complex dynamics of vertical/along-fault fluid flow. Semi-analytical solutions are available that have taken the same approach for handling sub-scale leakage along abandoned wells, and these should be explored to see if they can be practically and efficiently implemented into reservoir simulations. Analytical solutions can be adapted to account for time-dependent permeability changes with fault deformation (with relations developed from the upscaling workflow and not through modeling of fault deformation itself).
• Geophysical monitoring: The importance of integration of various geophysical data (seismic, EM, gravity) has been long emphasized for reliable and efficient monitoring strategy of CO₂ This concept has been well demonstrated for a synthetic data study. In the near future, we believe, a field-scale demonstration of applicability of all learnings from the synthetic data study should be realized so that we can better evaluate the true feasibility for practical cases, not just for desktop study cases.
• Uncertainty characterization and quantification: It is well understood that uncertainty is a main concern in all aspects of CO₂ For caprock integrity, the uncertainty problem has significant implications for predicting caprock failure, quantifying subsequent leakage and understanding its large-scale impacts. First, the characterization of uncertainty should be addressed in the laboratory, with several repeat experiments required to build statistical distributions. Secondly, it is often assumed that propagation of uncertainty requires an ensemble approach where many realizations are needed to explore the extensive parameter space. In order to reduce the computational burden of uncertainty quantification, more efficient methods need to be developed, most likely by tailoring existing stochastic or multi-step ensemble methods to the fracture initiation/activation and leakage problem.