D3.1 – Evaluation of systematic relations between the seismic response to fluid injection and depth, injection pressure, crustal stress state, and local structural geology

Evaluation of systematic relations between the seismic response to fluid injection and depth, injection pressure, crustal stress state, and local structural geology. The analysis of microseismic data from various EGS stimulations (Table 1) shows that a clear relationship exists between fluid injection and seismic response. A widely observed feature is that seismicity occurs first close to the injection well and then gradually propagates further away from the injection well. Detailed knowledge on the spatio-temporal distribution of the seismicity is strongly dependent on the properties of the physically installed seismic network, on the uncertainty in the assumed velocity model and on the applied methods for event location. For example, Kwiatek et al. (section3) re-analysed the Berlin, El Salvador dataset using the double-difference location algorithm and, although using the same data, they candraw new conclusions as compared to previous studies. One of their observations is that the Brune stress drops of events close to the injection point are smaller than the Brune stress drops of events that are further away from the injection point. A similar trend was observed by Goertz-Allmann et al. (2011, section 10) in the Basel, Switzerland, dataset. This data is further investigated by Bachmann et al. (section 11) and Goertz-Allmann and Wiemer (section 12). The seismic network has a strong influence on the quality of the event locations and obviously, seismic networks that comprise deep downhole sensors can detect and locate dramatically more events as compared to networks that solely consist of surface stations. The seismic network in Basel has six deep borehole stations with one three-component sensor deployed in each borehole, in addition to an extensive surface seismic network.Kraft and Deichmann (section 13)investigate how to improve the catalogue of about 3500 events using cross-correlation methods, solely using high-quality downhole data. They also compare estimated focal mechanisms from the downhole network with focal mechanisms estimated by the surface network.In yet another paper on Basel, by Zhao et al. (section 14),the authors study the 19 largest events (LME) during and after the injection sequence and determine moment tensors from full waveforms. The analysis results in similar double-couple components as found by Deichmann and Ernst (2009), and in addition it reveals significant isotropic components during the early injection phase. Most of the events in the later stage are dominated by the double-couple components. The locations of the events with high isotropic components also coincide with previously found regions of high b-values and low Brune stress drops (Goertz-Allmann et al., 2011). It is therefore likely that a correlation between bvalues, stress drops and isotropic components exists. Such relationship may help to explain discriminating between induced events (re-opening of existing fractures and creation of new flow paths) and triggered events (small stress perturbations on critically stressed faults result in failure). Today’s EGS reservoirs are mainly developed between 2 and 5 km depth (Table 1) and we do not observe a relation between the seismic activity and the depth of the reservoir, neither in the amount of recorded seismicity nor in the maximum measured magnitude. However, the depth of the reservoir needs to be taken into account when it is linked to the injection pressure and to the crustal stress state. In cases where several stimulation phases were conducted at the same well, seismicity seems to be reduced or is even absent until the stress level of previous stimulations is exceeded. This so-called Kaiser-effect has been pointed out by Baisch et al. (2010) for the Cooper Basin and is also observed by Kwiatek et al. (section 3) in the Berlin, El Salvador dataset. Calo et al. (section 8) observe that a previous stimulation may change the stress field for subsequent stimulations. They show that at theSoultz-sous-Forêts site, France, the stress fieldaround the injection well might not be fully restored, which affects the seismicity patterns of secondary injection intervals. In the beginning of most EGS projects, the information on local structural geology and the knowledge of seismic velocities and densities is often poor. Especially the shear-wave velocities are often extrapolated from simple P-wave velocity models, resulting in high uncertainties in the event locations.Albaric et al. (section 4) discuss the importance to obtain a good understanding of the local structural geology already during the first stimulation phase.

In their paper about a first hydraulic stimulation at the ParalanaEGS in Australia, the authors take advantage of several seismic 2D lines and some information from borehole logs. With this information, they construct a 3D velocity model for P- and S-wave velocities, which is then used to locate the microseismic events. They further conduct double-difference relocation fora subset of events. Finally they analyse focal mechanisms of selected events and find that the resulting double-couple components are in agreement with the regional compressive stress field.Jousset et al. (section 5) apply ambient seismic noise analysis to retrieve information about the structural model of the reservoir. Their study area is the Bouillante geothermal field, Guadeloupe, French Antilles, where generally little seismicity is associated to the geothermal exploitation. After implementation of the results from the ambient seismic noise tomography, the authors locate the few recorded microseismic events and discuss the source parameters in light of the reservoir model.

Due to active processes within geothermal reservoirs, changes of the physical properties over time are likely. Cooling of the rock, dissolution and precipitation processes after long term fluid circulation, seismic and aseismic deformation of the reservoir during stimulation and afterwards during circulation, are all processes that change rock properties. Detailed analysis of waveform data from The Geysers, California, USA, (Gritto and Jarpe, section 9 and section 15) reveal temporal changes of the Vp/Vs velocity ratio over time. The authors observe e.g. that an increase in Vp/Vs ratio along with a decrease of Vp and Vs velocities is indicative for fluid-filled fractured rock.Calo et al. (2011, section 6 and 7) conducted timelapse double-difference tomography at the Soultz-sous-Forêts site and found that aseismic deformation close to the injection site might have a significant contribution to the change in velocity values. In section 8, Calo et al. present event relocations from the GPK4 stimulations at Soultz-sous-Forêts, derived from double-difference methods. The temporal evolution of the relocated seismicity is grouped around the open–hole section of the injection well and the authors interpret this pattern with respect to the in-situ stress field.