Evolution of stress and seismicity in fractured geothermal reservoirs

Geothermal power has the potential to contribute a good share of future energy needs. The International Energy Agency estimates that by 2050, geothermal will deliver about 3.5% of the world-wide power production. The technology of enhanced geothermal systems (EGS) will play an important role in the geothermal future. To develop the hot rock in the subsurface, wells are drilled to three kilometers and more. By high pressure injection of fluids, pre-existing fractures are enlarged or new fluid pathways are created. Currently, the EGS technology is still under research and development, with only small power plants and several obstacles on the path to future large scale application with power plants in the order of 100MW. Besides reduction of drilling costs and maintaining long-term production, the mitigation and control of induced seismicity has become a major challenge for the future of EGS. At the Deep Heat Mining Project Basel, a magnitude 3.4 event was induced during the development phase. Following a comprehensive risk assessment study, the project was finally abandoned and large investments were lost. The goal of this thesis is an improved understanding of seismicity induced in the surrounding of EGS. I analyze seismicity induced during the development of the EGS at Soultz-sous-Forets (France). The project in Soultz was initiated in 1987 as the European pilot site and has more than 25 year of experience in developing an EGS. During this period, a database which is unique worldwide was assembled, making Soultz the most advanced EGS project so far. The main part of this thesis consists of an analysis of the stimulation of well GPK2, conducted in June and July 2000. During the six days of stimulation, 23 400m3 of water were injected under pressures as high as 14.5 MPa. The surface seismometer network allowed localization of 7215 events. Focal mechanism solutions of 715 events with M > 1 could be obtained. This dataset, courtesy of Universite de Strasbourg and GEIE Heat Mining, Soultz, forms the basis of the studies. The PhD project was conducted partly within the FP7 GEISER (Geothermal Engineering Integrating Mitigation of Induced Seismicity in Reservoirs) project, funded by the European Commission and it benefited largely from fruitful collaborations with Universite de Strasbourg, Geowatt AG, Zuerich, GFZ German Research Centre for Geosciences, Potsdam and CSIRO Earth Science and Resources Engineering, Perth. In the first study, I analyze whether interaction of seismicity by static stress transfer plays a significant role on the spatio-temporal evolution of seismicity. I follow an analytical approach to compute the displacement field of a rectangular earthquake source. Through stacking of several sources, realistic slip distributions are obtained. The analysis reveals seemingly random distributed patches of stress increase and stress decrease of less than ± 1 MPa, except for very localized areas. Since the fracture planes have varying orientations, they form a volumetric fracture network. About 60% of hypocenters are found in areas with increased Coulomb stress where their potential for failure was increased by static stress transfer. A different behavior is observed for slippage of neighboring asperities on larger fault zones. Here, failure of asperities leads to a direct stress increase in adjacent asperities, which are then more likely to fail. This is exemplified on a cluster of events occurring on the largest fault zone in Soultz, after shut-in of the well GPK2. Subsequently, the peculiar behavior of seismicity and the hydraulic regime following shut-in of the well GPK2 is highlighted and investigated by further analysis of focal mechanism solutions. An increase of the thrust faulting component following shut-in is observed. The changes of the stress field are derived from spatio-temporally resolved inversions of focal mechanism solutions. A very strong reduction of the maximum horizontal stress and an increase of the minimum horizontal stress is revealed, which leads to a change of the stress regime from a transitional strike-slip / normal faulting regime to a pure normal faulting regime. This is the reason, why almost only normal faulting events are observed during the stimulation. These stress changes are not compatible with the coseismic stress changes. Thus, a large proportion of aseismic movements during the stimulation of well GPK2 is proposed. In the next study I focus on time-dependency in geomechanics in connection with stress transfer. In a numerical finite element model a time-dependent failure criterion combined with a damage mechanics approach is used to study time-dependency of borehole breakout development. The failure criterion is based on the observation of creep failure at loads considerably lower than the short-term strength of the rock samples. It is shown that breakouts grow both in width and depth. Through progressive stress transfer to neighboring elements peak loads at the wellbore wall are reduced and breakout growth slows down. The rate of breakout growth follows Omori's law, which is used in seismology to describe the decline of aftershock activity following a main shock.