Complexity, heterogeneity and scalability of injection induced seismicity from decameter-scale stimulation experiments

Abstract

Deep geothermal energy resources are vastly available. The conversion of these resources to base-load electrical energy is seen as a promising and sustainable substitute for nuclear based power production and to balance out weather-dependent solar and wind based power production. Enhancing the permeability of rock formations at depths where temperatures are sufficiently high for electrical power production is the main difficulty. Hydraulic stimulation is used to enhance permeability in fractured rock mass and to build a heat exchanger, through which fluids circulate between an injection and a production well. During stimulation treatment seismic events are induced. Above a certain magnitude, these cause threats to local communities and infrastructure. The avoidance of induced earthquakes, and the development of a suitable heat exchanger are the major challenges. Intermediate-scale in-situ experiments performed in underground laboratories are key to better understand the seismo-hydromechanical processes involved in the hydraulic stimulation treatment. Its main advantage is the accessibility of the rock mass at the intermediate 10 - 100~m - scale, allowing for a detailed characterization, instrumentation and monitoring of the rock mass. These scaled experiments are able to bridge the small 1 cm laboratory scale and the 1000 m field scale. Decameter-scale injection experiments were conducted in the framework of the In-situ Stimulation and Circulation (ISC) project in the crystalline rock mass of the Grimsel Test Site (GTS), Switzerland. The experimental volume approximates dimensions of 20 m x 20 m x 20 m and lies below an overburden of ~480 m. In the initial phase of the project, the experimental volume was characterized in great detail with regard to geology, the in-situ stress field, hydrology and geophysical properties. Based on the geological characterization, two shear zone sets were targeted in six stimulation experiments with the aim of inducing hydraulic shearing. An additional six stimulation experiments were conducted in intact rock aiming at inducing hydraulic fracturing. To capture the seismo-hydromechanical rock mass response of the injections, a comprehensive monitoring system was installed along tunnel walls surrounding the experimental volume and in dedicated boreholes drilled through the experimental volume. In order to gain insights into the effects of geology on the seismo-hydromechanical rock mass response, the same injection protocol was used for the six hydraulic shearing experiments and the six hydraulic fracture experiments, respectively. Fluid pressure and deformation were recorded with a dense monitoring network in boreholes surrounding the injection locations. Seismicity was monitored using sensitive in-situ acoustic emission sensors installed in boreholes and at the tunnel walls. This doctoral thesis covers the seismological analysis of twelve performed stimulation experiments, starting with a broad analysis of comparative character and moving towards a more detailed analysis. The detailed analyses focus on the single injection experiment whose seismic events outnumbered by fare those of all other experiments. High variability in seismic response in terms of seismogenic indices, b-values, spatiotemporal evolution were observed during both experimental series, which we attribute to local heterogeneities. The seismic response was enhanced for experiments targeting the highly conductive brittle-ductile shear zone set, while the injectivity increase on these structures was only marginal. The seismic responses between the hydroshearing and hydrofracturing experiments was similar, presumably because the induced hydraulic fractures linked to the pre-existing fracture network which was stimulated during the hydroshearing experiments. The induced fracture dislocation during the hydroshearing experiments was mostly aseismic. The observed increased seismic response of certain structures suggests that the stimulation of short borehole intervals with limited fluid volumes may be an effective way to limit induced seismic hazards if these structures can be avoided. The careful analysis of induced seismicity, source mechanism and mechanical responses of one single hydraulic shearing experiment revealed in detail the meter-scale complexity of hydraulic stimulation. The accurate location of seismicity led to the identification of four distinct clusters, three of which were attributed to the stimulation of fractures in the damage zone. The inferred source mechanisms varied by cluster, suggesting an already prevailing heterogeneous stress field that is further modified by stimulation. Stress redistribution, caused by the injection, induced an initially aseismic tensile-dominated fracture, leading to the seismicity of the fourth cluster. Observed streaky patterns of seismicity suggested that the injected fluid was contained in conduits along the existing fracture planes. Applying cluster analysis and a master event relocation to the induced seismicity of the same stimulation experiment revealed a large number of repeating earthquakes. Source areas within repeater families are likely to overlap (assuming an appropriate stress drop), but nonetheless show a distinct propagation direction. This spatial evolution has a larger extent compared to the dislocation distance estimated from strainmeters and borehole offsets in the experimental volume. Thus, the propagation of the repeating earthquakes is more likely to be produced by multiple closely grouped asperities that are reactivated in sequence, instead of a single asperity that is displaced over the course of the experiment. Reoccurrence times of repeating earthquake families showed an injection-pressure dependency, especially those repeater families located in close proximity to the injection interval. Statistically, repeater families lead to breaks in scaling in the cumulative frequency-magnitude distribution. The results presented shed light on the high value of scaled experiments performed in-situ in underground laboratories to investigate complex seismo-hydromechanical processes involved in the hydraulic stimulation treatment. The variability observed in seismic responses during injections at the decameter-scale questions the ability to precisely forecast induced seismicity during hydraulic stimulation treatments. Complementing the various, high-resolution observations from the decameter in-situ scale with numerical modeling improved the fundamental understanding of the stimulation process. The analysis, strategies and results obtained during the 12 stimulation experiments are highly relevant for the development of further scaled experiments.