| Through a series of related projects, the work described in this dissertation strives to better constrain, describe, and explain in situ stress heterogeneity. Although each project has specific goals, approaches, and outcomes, combined they represent important progress in the ongoing effort to understand stress heterogeneity at a variety of scales.; In the first study we analyze multi-scale variations in the direction of maximum horizontal compressive stress as a function of depth in four scientific research boreholes located in a variety of tectonic environments. Our results provide insight into the mechanisms controlling in situ crustal stress heterogeneity over scales from centimeters to kilometers. We show that the orientation of the maximum, horizontal compressive stress determined from stress-induced wellbore failures displays scale-invariant, fractal distributions with spectral exponents between 1 and 2. The scaling of the stress variations is remarkably similar to the spatial scaling of earthquakes as a function of fault size in the individual study areas. Consequently, we suggest that wellbore stress heterogeneity is controlled by slip on a fractal distribution of active faults in the surrounding crust. The observed correlation between the amount of stress heterogeneity and local fault behavior may prove useful in models of dynamic earthquake rupture, where many of the key parameters including stress and fault strength appear to vary spatially.; In the second study we develop two models---a two-dimensional, analytical model and a three-dimensional, numerical model---to explain the rotation of in situ stresses resulting from a pore pressure change on one side of an impermeable boundary (for example, near an impermeable fault in a depleted reservoir). Our models show clearly that near the boundary depletion will induce the maximum horizontal compressive stress to become more parallel to the boundary, whereas injection will have the opposite effect. We find a strong interaction between in situ stress magnitudes, rock property contrasts, reservoir shape, and boundary orientations. While the analytical model provides a good match to the field observations, the numerical model shows that for certain ratios of pore pressure change to differential horizontal stress, the analytical model may overpredict the amount of expected stress rotation. Our results can help forecast changes in stress directions throughout the life of a reservoir, which is critical for practices like hydraulic fracturing. Our work will have important implications for the frictional stability of faults and fractures both in and around the reservoir.; In the final study we utilize data from twenty-one borehole tensor strainmeters recently deployed as part of the NSF EarthScope Plate Boundary Observatory to show that the large-magnitude, long-term strain trends in the data result at least in part from poroelastic deformation driven by the wellbore stress concentration. Our results allow us to derive a model-based approach for removing the trends, offering a significant improvement over the traditional approach that uses mathematical regression on each gauge individually. In addition, from the long-term strain accumulation we can derive information about the physical properties of the rock surrounding the strainmeter as well as the local in situ stress state, which can be compared to independent stress indicators such as earthquake focal mechanisms and stress-induced wellbore failures nearby. In a general sense, this work provides insight into time-dependent borehole deformation that will have implications not only for the strainmeter community but also in wellbore engineering and other scientific applications. |
