Characterization of stress heterogeneity around a fault zone based on inversion of hydraulic fracturing tests

Characterization of the in-situ state of stress is critically important in many geoscience applications, including understanding fault mechanics. In-situ stresses can exhibit strong spatial heterogeneities, due to the influence of factors such as surface topography, slip along faults and fractures, as well as lithological contrasts. Our understanding of these factors has been limited by our inability to characterize the full stress tensor and its variability at high spatial resolution. Additionally, studying the relationship between fault mechanics and the heterogeneous stresses has been prevented by the lack of in-situ observations of fault slip in rock volumes well-characterized in terms of stress. The FEAR project provides a unique opportunity to tackle these gaps in our knowledge. As part of this project, a series of hydraulic stimulation experiments are performed in a fractured granitic rock mass intersected by major faults in ETH’s BedrettoLab. The induced seismicity and hydromechanical processes are monitored using a dense sensor network.

To characterize the stress field in the rock mass of interest, a detailed hydraulic fracturing campaign was performed in three vertical and eight inclined boreholes. We developed a new stress inversion method that can infer an arbitrarily inclined primary stress tensor from hydraulic fracturing tests performed in arbitrarily inclined boreholes. The method uses a grid search approach to invert the generalized Kirsch Solution and allows us to quantify the uncertainty of the solution (i.e. its sensitivity to error in the measured input data) both in terms of principal stress magnitudes and orientations. The required input data are fracture orientation from image logs, shut-in pressure, breakdown and fracture reopening pressure.

Applying our inversion technique to the data collected in the BedrettoLab resulted in 32 stress tensor solutions (including uncertainty) corresponding to different locations within the rock volume, as well as 14 additional data points of the S₃ magnitude. Our results show that S₃ is (sub-)horizontal in the entire rock volume, and its azimuth ranges from N147.8 to 211.4°E. The rock mass can be divided into two domains based on the stress regime: a normal faulting domain in the SSE portion of the rock volume and a strike-slip faulting domain in the NNW portion. Potential causes for the observed abrupt transition from normal to strikes-slip faulting may be compliance contrasts within the rock volume as well as fault slip along different geological structures. The normal faulting domain extends a few meters into the northern side of a major, SSW-ENE oriented fault, and it is unclear whether the transition is related to this fault.

Our high-resolution stress dataset will enable us to investigate the causes of the observed stress heterogeneity using numerical modeling tools, and to determine which faults are likely to slip and open during hydraulic stimulations. Once available, the experimental data of hydraulic stimulations will be compared to our predictions. This will provide an unprecedented opportunity to study the relationship between in-situ stresses and fault dislocation, ultimately resulting in an improved understanding of earthquake physics in general.