Hydraulic Fracturing Behavior of Opalinus Shale: A Framework, Experimentation & Insights

Hydraulic fracturing is a pivotal technology that made it possible to tap into previously inaccessible energy resources. Shales constitute the majority of unconventional reservoirs and have complex structures characterized by layered, fine-grained minerals. Pressure-volume and micro-seismic data are commonly used to assess hydraulic fracture jobs, but little is known about the produced fracture geometry and how to optimize it. Laboratory experimentation as done in the presented research can be used to address this lack of knowledge. Specifically, this thesis examines fracture behavior in shale; assesses the overall fracture geometry and gains a better understanding of fracture interaction with bedding planes. Opalinus Shale was used as the material of study, which visually had two distinct, alternating layers: a tough light layer and a soft dark layer. The following framework was set to investigate the complex rock material and hydraulic fracture process. First, the mineralogy and micro-structure were assessed through a variety of techniques, including X-Ray diffraction, scanning electron microscopy with energy-dispersive X-Ray spectroscopy and focused ion beam milling, and Raman spectroscopy. This showed that dark layers are clay-rich while the light layers are a heterogeneous mix of quartz, calcite, and other minerals. The used hydraulic fracturing liquid (hydraulic oil) and its interaction with the shale were thoroughly characterized. Capillary calculations showed a strong liquid affinity to the shale, resulting in significant and rapid capillary rise that minimizes the distance between the liquid front and the advancing fracture tip. Next, a novel series of indentation tests coupled with scratch tests were conducted to characterize the stiffness, hardness, creep compliance, and toughness along the transverse isotropic principal directions of the two distinct layers comprising this shale. Test results were used to approximate the radius of plasticity and fracture energy in each of the principal directions of each layer. Finally, a conceptual model was developed to quantify edge-to-edge and face-to-face fracture energy components, showing that fractures were more likely to propagate perpendicular to bedding. A novel technique was developed to homogenize and characterize the seismic properties of highly heterogeneous materials. Based on P-wave arrival times, an elliptical velocity model was constructed that defines velocities along and normal bedding, and the seismic plane of isotropy. This technique can prove very useful as it can be extended to field scale measurements and can improve acoustic event localization which depends on accurate velocity estimation. To do all this, a highly instrumented and controlled experimental apparatus was designed and built to subject the shale specimens to a quasi-true triaxial stress state simulating subsurface stresses and to pressurize a pre-cut artificial crack for hydraulic fracture propagation. This novel apparatus allows one to simultaneously capture images and acoustic emission data along with an array of other measurements instrumental for hydraulic fracture analysis. Despite the simple anisotropic stresses being applied to the test specimens, the fracture behavior was far more complicated than fractures propagating along the direction of maximum principal stress. The results showed the crucial role rock fabric plays in determining hydraulic fracture behavior. Acoustic emissions were also analyzed spatially and temporally, and insights such as focal mechanism frequency and relative proportionality were gained from these observations. This thesis serves as a solid step towards gaining a comprehensive understanding of hydraulic fracture behavior. The thesis also can contribute to the interpretation of field observations, and presents a valuable workflow for specimen characterization and data analyses. Last but not least, the described experiments can serve as a basis for predictive fracture models.Ph.D