Enhanced geothermal systems (EGS) hold the promise of using the ubiquitous heat energy of Earth. However, enhanced geothermal systems typically requires opening – so-called “stimulation” – of fluid flow channels to enhance the permeability of the new reservoir. The natural by-products of this engineered process are earthquakes. Seismic activity related to enhanced geothermal systems have seriously affected or even terminated some geothermal projects such as Basel, Switzerland in 2006 or Pohang, South Korea in 2018. Therefore, implementing of safe stimulation strategies is critical for public acceptance of any future project developing enhanced geothermal system.
In the collaborative project which outcomes were published in Science Advances we show that high-precision, near–real-time passive seismic monitoring and analysis of seismic data allowed for safe stimulation of the world’s deepest Enhanced Geothermal System project so far located in in Espoo, Finland. This page intends to present a quick overview of the project, stimulation and resulting study, as well as provide future updates on research related to the project.
If you are interested in the technical details of the project and study, see:
Kwiatek, G., T. Saarno, T. Ader, F. Bluemle, M. Bohnhoff, M. Chendorain, G. Dresen,
P. Heikkinen, I. Kukkonen, P. Leary, M. Leonhardt, P. Malin, P. Martínez-Garzón, K. Passmore, P. Passmore, S. Valenzuela, and C. Wollin (2019). Controlling fluid-induced seismicity during a 6.1-km-deep geothermal stimulation in Finland, Sci Adv 5, no. 5, eaav7224, doi 10.1126/sciadv.aav7224. [ Article Page ]
St1 Deep Heat project
The St1 Deep Heat project site is located in Helsinki suburbian area at the Aalto University in Espoo, approximately 6 km away from Helsinki city center. This pioneering project aims to provide a sustainable baseload for the campus area heating network by extracting the hot water from the depth of over 6 km. The intended configuration is to have a well doublet, where one well will be used for (cold) water injection and second well for (hot) water extraction. In the first stage of the drilling, a 6.4 km measured length well OTN-3 was drilled, with the last 1000 meters of the inclined open-hole section. The last part of the well targeted permeable geological formations at 5-6 km depth with temperatures up to 120C.
The project area is characterized by simple geology with very thin sedimentary layer of 10 m that superimposes thick Precambrian basement. Complex small tectonic structures are ubiquitous in the basement rocks, including broad and steeply dipping damage zones detected in well logs, that could likely serve as future fluid paths for the future geothermal heat plant operations.
Before this could happen, it was necessary to hydraulically stimulate the future geothermal reservoir by injecting fluids into the OTN-3 well in order to enhance its permeability. This stimulation was performed in Summer 2018. In the following stage, a new well will be drilled into the created damage zone in order to establish the well doublet and to start the commercial exploitation of the heat.
Preparation for reservoir stimulation
The stimulation monitoring network was composed of 12 sensors located at depth 2.6 km in adjacent borehole OTN-2, approximately 3 km above the injection area. This was completed by additional 12 sensors located up to 11 km away in shallow boreholes. The purpose of this network was to track the evolution of induced seismicity during hydraulic stimulation in the near-real-time. In addition, the separate network of 17 surface geophones was installed in locations critical from perspective of ground motions. This network was used to measure the ground motions and provide a feedback to traffic light system and stimulation engineers.
Five stimulation stages were selected using borehole logs and the continuous stimulation was performed in Summer 2018 and lasted 49 days. In total, 18,500 m3 of fresh water was injected at well head pressures of up to 90 MPa and injection rates of up to 800 l/minute. This amount of fluids injected into reservoir was significantly larger than that injected in Basel, Switzerland, 2006 (11,500m3) and slightly less than in the Cooper Basin, Australia (20000m3). In both of this sites, seismic events with magnitudes above 3.0 occurred. However, the “red light” for the traffic light system in St1 Deep Heat was only magnitude 2.1. If earthquake with such a magnitude would occur, the stimulation would have been stopped.
Similarly to other geothermal projects, a traffic light system has been established……
Stimulation and traffic light system operation
The system for processing and interpretation of passive seismic data in near-realtime developed and implemented by fastloc GmbH detected and located more than 6000 earthquakes in the vicinity of project site. This included the initial hypocenter location and magnitude estimate which were available up to 5 minutes after the earthquake occurrence. In case of larger events, the magnitude estimate and hypocenter location was verified by the seismologist. The catalog of seismic events was updated continuously during injection operations and provided both to the Traffic Light System operation as well as injection engineers. This two parties were deciding on how to modified injection strategy in response to occurrence of seismicity.
By the end of the project at 18,500 m3 of fresh water injected, the maximum observed magnitude was 1.9, which was below the red alert level of 2.1 set up by the local authorities.
Passive seismic data reprocessing
The resulting data were post-processed by GFZ Potsdam group from Section 4.2: Geomechanics and Scientific Drilling together with industrial and university partners that took part in the St1 Deep Heat stimulation.
The first step consisted of enhancement of the original industrial catalog the catalog by re-analysis of the waveform data in attempt to search smaller seismic events. This resulted in another 50,000 of small events detected using simple pattern matching algorithm with much smaller magnitudes. The completeness of the seismic catalog was down to the ML -1.2 and the largest observable events was ML 1.9. Translating it to the fracture length scale, this means that the observed fracture sizes that could be monitored using installed seismometers ranged from approx. 10 to 100 m.
The selected over 2,000 high quality and largest earthquakes were manually reviewed and extensively reprocessed. This included review of the waveform data and improvement of hypocenter locations using so-called double-difference relocation algorithm. This increased the precision of earthquake hypocenter locations to <60 m for the 95% of the catalog. This allowed the project team to investigate the details and spatial and temporal evolution of seismic activity in response to injection operations.
Seismicity and injection operations
The seismicity overall appeared concurrently in three four major zones, regardless of the stimulated part (stage) of the open hole section of OTN-3 well. This is an interesting feature by itself giving a hint on the strength of rocks at the reservoir depth. As the packer failure hypothesis was rejected (packer is a device that isolates some section of the well in order to concentrate the injection in the specific area), the project team concluded that the injected fluids were bypassing the stage packers entering formation at certain discrete intervals. This would mean that the vicinity of the injection well is seriously damaged, likely due to the existence of damage zones and fractures induced by drilling of the ONT-3 well.
The seismicity was observed to propagate outside of injection well OTN-3 along south-east to north-west direction, which was expected as it this is direction subparallel to the direction of maximum horizontal stress. Also, the downward migration of earthquakes was observed in the largest zone of seismicity with the progress of stimulation campaign.
Controlling induced seismicity
It was quickly realized already in the first injection phase at the bottom of the well OTN-3 that the seismic energy release is proportional to the hydraulic energy, or equivalently product of pressure and volume. Also, it was identified that any stop in injection leads to quick reduction in seismic activity. The following second injection phase was performed at maximum well head pressure of around 90 MPa and with a long intervals of injection lasting up to 3 days of continuous injection. This resulted in acceleration of seismic energy release and led ultimately to a series of relatively large earthquakes that prematurely stopped this phase of injection. It was obvious for the project team that a redefinition of the injection strategy is required.
At this moment project team realized that the maximum observable magnitude increase with cumulative injected fluid volume. Surprisingly to the team, the trend of this magnitude evolution with injection already seemed to follow the recently proposed fracture mechanics-based model presented in 2017 by Galis et al.. While naively extending the observed trend to the planned total volume of 20,000m3, it was a concern of project team that by the end of injection the red alert M2.1 event will likely occur.
Galis et al. claimed that the maximum magnitude of the so-called arrested rupture depends on the amount of energy available stored for rupture propagation. Therefore, the project team decided to reduce somehow the amount of hydraulic energy stored in the geothermal system due to fluid injection.
At first, the well head pressure was reduced to values below 90 MPa in attempt to reduce the hydraulic input energy rate into the developed geothermal system. This strategy was applied in Stage 3 of injection and seemingly did not improve the situation. Towards the end of this short stimulation stage another series of larger events occurred. Therefore, in the following step, the injection plan was progressively changed. The injection was still performed at lower injection pressure, but the injection times were progressively reduced and the resting periods in-between stimulations were enhanced. At the end of stimulation, the injection was performed over the day (12 hours) and stopped during the night time (12 hours of waiting time). This strategy visibly stabilized the seismic energy release with respect to hydraulic energy input. Till the end of the project, the red alert threshold was not exceeded, as the maximum observed magnitude reached M1.9 in the stimulation stage 4.
Why the project was successful in a nutshell
There seem to be at least three factors contributing to the success of first injection campaign at the St1 Deep Heat project. Successful control of maximum magnitude was likely due to:
- Adaptive injection strategy guided by near-realtime passive seismic monitoring and data analysis that was performed in ample time. The decisions made during the stimulation based on observation of evolution of maximum magnitude, seismic and hydraulic energy led to the effect limitation of the hydraulic energy input rate into the developed reservoir.
- Possibly favorable geological basement structures and favorable stress conditions in the reservoir, as it is suggested by the initial analysis of passive seismic data (see next section)
- Close and direct collaboration between seismologist involved in the project (ASIR LLC, fastloc GmbH) with site operator (St1 Deep Heat), Traffic Light System team (ARUP Geohazards) and local authorities.
Second look into the data
The double-difference relocated data provides no evidence for alignment of seismicity along a large fault. Instead, hypocenters show that injection led to seismicity occurring in the the distributed fracture network. This is supported by correlation between location of damage zones in well logs, geological data, and hypocenter distribution.
- Outcrop data/compact micro imager/lithology: Indication of damage zone
- Coincides with focal mechanisms orientations
- Focal mechanisms seem to be well aligned with the stress field
- Statistically significant drop-off in the number of events above M>1.5
4 No faults large enough to sustain larger events?
4 Faults can’t store enough elasting energy to support a runaway rupture?
- Finally, the seismic injection efficiency is of order of 10 to the power of -3. While comparing with empirical data from other experiments, this seismic efficiency would suggest we indeed re-activate a small-scale fracture network.We tested whether the seismic activity display properties suggesting significant stress perturbation due to fluid injection.The interevent time ratio statistics suggests the seismicity display poissonian characteristics with no triggering.We also tested whether earthquake catalog display signatures of strong spatial and temporal clustering. Following the space-time clustering method we separated the catalog into background and clustered seismicity finding that only approximately 12 % of the events are displaying clustering properties. This would suggest minor interactiong between events, and limited stress transfer.
- While looking into temporal distribution of seismicity we also observe quick dissipation of injected hydraulic energy. This graph shows the magnitude of seismic events as a function of time. We calculated the b-value in moving time window finding out that after phase 1 of injection, the observed b-value is stationary. This would suggest that there is no significant change in the deviatoric stress with progressing injection, and, consequently, the seismic hazard is guided by GR a-value.
- Project stopping M2.1 earthquake was successfully avoided by adaptive injection operations using near-realtime monitoring of induced earthquake rates, locations, magnitudes, and evolution of seismic and hydraulic energy
- Fluid injection was likely performed into a (complex?) (size-limited?) fracture/fault network leading to low stress perturbation. No major faults were found in the reservoir.
- Successful operation required close cooperation of seismologists, site operator, TLS team and local authorities during stimulation.
- The outcome of the St1 DH project indicate a possible approach allowing to manage induced seismicity in similar geothermal projects.