In many cases, fluid injections into the subsurface have been recognised to cause seismic activity. Understanding the driving physical processes of these man-made earthquakes and their occurrence in time and space is of significant scientific and public interest. Not only does a deeper insight into the physics of fluid-induced seismicity yield fundamental knowledge on seismogenic processes in the brittle Earth's crust but it is also a crucial point for reservoir characterisation and injection performance. In the end, these are inevitable steps towards seismic hazard assessment and risk mitigation.
The focus of this thesis is on spatio-temporal migration patterns of fluid-induced seismicity at different scales. In a first study, I work with seismicity induced by the rather local effect of single-borehole, high-pressure fluid injections. This part is followed by two studies on earthquakes caused by the regional impact of large-volume injections through numerous wells under gravity. Whereas high-pressure, single-borehole injections are typically used for hydraulic fracturing operations to enhance shale gas recovery from unconventional reservoirs and for hydraulic stimulations of geothermal heat production sites, gravity-driven, large-volume injections are part of wastewater disposal which has been performed throughout the last decade in the central U.S. In the context of this work, wastewater is a saline fluid, co-produced with natural oil and gas, which is re-injected into the subsurface at a later stage. Apart from the actual injection phase, I also consider seismic events that occur in the postinjection interval of high-pressure fluid injections and in times of decreasing wastewater disposal volumes. I assess the governing question which parameters and physical processes control features of fluid-induced seismicity in time and space by joining different methods.
By means of analytical solutions, I derive a novel scaling law for postinjection-induced seismicity in the first study. My findings suggest that the spatio-temporal evolution of the seismically active zone depends on the index of non-linearity and the Euclidean dimension of pore-fluid pressure diffusion. I combine these results with numerical modelling of non-linear pore-fluid pressure diffusion in 3D. Based on the numerical pressure solutions, I generate catalogues of synthetic seismicity to validate the novel scaling relation. The subsequent successful application to different borehole injection case studies demonstrates that the relation can be used to estimate the two driving parameters of spatio-temporal features of seismicity induced by high-pressure fluid injections.
An analytical approach, combined with numerical modelling, also forms the basis of the second study. Using known relations for reservoir-induced seismicity, I present a new first-principle model for wastewater disposal-induced earthquakes observed in the crystalline basement in Central Oklahoma, U.S., called underground reservoir-induced seismicity (URIS). The model consists of the following physical mechanisms; a normal stress acting on the seismogenic basement induced by the mass of the disposal fluid added to the pore-space of the target injection formation, the diffusion of pore-fluid pressure in the basement, and poroelastic coupling which contributes to pore-fluid pressure- and stress changes in the basement.
I implement the novel conceptual model in a numerical finite element model, solving for poroelastic pressure- and stress changes. The obtained values are then used to calculate failure criterion stress changes and to generate synthetic clouds of seismicity. My findings demonstrate that the URIS model captures the observed time- and depth-distribution of earthquakes located in the study area both during constant and in times of declining injection volumes.
In the third study, I combine results from a time-dependent 2D cross-correlation with numerical modelling solutions to explain spatio-temporal patterns of wastewater disposal-induced seismicity in southern Kansas which seems to migrate away from the high-volume disposal area with time. The cross-correlation reveals that the majority of earthquakes preferably occurs towards the east-northeast of the disposal wells. This feature may be explained by the directional migration of poroelastic stresses and pore-fluid pressure diffusion, probably caused by a large-scale, fault-induced anisotropic character of the basement permeability. Two-dimensional numerical modelling of poroelastic pressure- and stress changes for the study area suggests that the observed shift of the seismically active zone is guided by the high permeability of the injection formation and that the depth-migration of pore-fluid pressure and poroelastic stresses is driven by the basement permeability.
In general, the presented study verifies that poroelastic fluid-rock interaction may be crucial to explain seismic activity far away from large-volume disposal wells. Furthermore, it proposes that a volume reduction lowers seismicity rates only locally and that a state-wide reduction of the seismicity may require many more years.
Overall, the findings of the three works contribute to a general understanding of driving processes and parameters of fluid-induced seismicity across different scales. Thus, the presented novel model approaches may ultimately be used in future studies to assess and mitigate the risk posed by anthropogenic earthquakes.
