Você está na página 1de 24

High-resolution analysis of microseismicity

related to hydraulic stimulations in the Berlín


Geothermal Field, El Salvador
Grzegorz Kwiatek1, Fatih Bulut 1
Marco Bohnhoff 1, Georg Dresen1

1) GFZ Potsdam, Section 3.2: Geomechanics and Rheology


Motivation

• Evaluation of systematic relations between the seismic


activity and fluid injection
• Characterization of fluid-induced seismicity and larger
magnitude events - occurrence in time and space, source
processes.

AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Berlín Geothermal field
• Located on the flancs of dormant volcano Cerro Tecapa
• 8 production wells + 10 injection wells (closed fluid circulation)

1km

Rivas et al., 2000; Oates et al. 2004

AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Stimulation and passive seismic monitoring
• 3 stimulations campaings (each lasting ~3 weeks) performed at two depth intervals in TR8A
well to enhance reservoir to the north.
• 12 Borehole sensors, Natural frequencies: 1, 4.5, 5.5 and 30Hz, 24 sampling frequencies
(3000Hz-24Hz), triggered mode, seismic events located assuming isotropic velocity model

S N

SHmax

Injection well

1k
m

Rivas et al., 2000; Oates et al. 2004


AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Stimulation campaigns
•Not clear correlation between injection and seismic activity (hardly visible relations between
injections and seismicity)

AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Results of monitoring
• 581 events recorded in 16 months
Whole dataset (2002–2004) including 3 stimulations
of TR8A
During injection into TR8A
• Weak clustering, low location quality.
• Source parameters not displaying any
particular pattern

S N

1km

Data from Oates et al., internal Shell report


AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Summary of the industrial monitoring

• Weak clustering – possibly low location quality


• Hardly visible relations between injection rate, location, and
seismicity
• Source parameters not displaying any particular pattern in
response to injection

Original seismic data need to be refined!!!

AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Double-difference relocation technique
• Groups of very similar waveforms observed - coming from the same spot?
• Double-difference relocation applied (Waldhauser & Ellsworth, 2000) with 1D
velocity model: 200,000 cross correlations performed, 393 events relocated.

AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Original dataset before double-difference
1.Weak clustering observed

1km Wells
Whole dataset
During injection into TR8A

AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Relocated dataset
1.Shifts of locations into the
geothermal field
2.Clustering around injection
wells

1km Wells
Whole dataset
During injection into TR8A

AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Relocated dataset
1.Shifts of locations into the geothermal field
2.Clustering around injection wells

Wells
Whole dataset
During injection into TR8A
AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Cluster analysis
• The measure of „similarity” between two events depends on interevent distance and
average cross-correlation coefficient between waveforms available.
• Events in families/clusters display similar rupture process

TR8A Cluster

AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Cluster analysis
• 9 families selected
• Events in families display similar
rupture process
• We focus on family 1 – related to
injection into TR8A

1km

AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Source parameters using spectral fitting
• Inversion for moment, source radius and attenuation

Ground velocity spectrum Ground velocity spectrum

AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Source parameters using spectral fitting
• Trade-off between source radius and attenuation resulted in unstable inversion

AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Spectral ratio refinement
• We applied modified spectral ratio technique (e.g.
Imanishi and Ellsworth, 2006)
• Applied to clusters of data (similar location and travel
paths from EQ to receiver)
• Propagation effects effectively supressed by forming
spectral ratios

ECGS Workshop 2012, Luxembourg, October 3-5, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Spectral ratio refinement
• Strong improvements in the quality of source parameters (M0, r0, Ds)

Initial spectral fitting method After spectral ratio refinement

AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Spatial and temporal behavior for stimulation 1
• Migration of seismicity outside of injection wells over the fault plane F1
• Migration stops immediately after the shut-in
• Shut-in resulted in an outburst of the larger magnitude events

̶ Well head pressure


̶ Injection rate

SHmax

250m

AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Spatial and temporal behavior for stimulation 2
• Kaiser effect observed
• Weak migration outside of injection zone
• „Leaping” type of activity due to even higher injection rates and activation of another fault
plane F2 (?)

̶ Well head pressure


̶ Injection rate

250m

AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Spatial and temporal behavior for stimulation 3
• No migration
• Clustering around of injection well (or another fault plane?)
• Triggered activity on previous faults
• Another fault plane activated (F4?)
̶ Well head pressure
̶ Injection rate

250m

AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Static stress drop
Static stress drop increases with the distance for events located on the fault plane for I1/I2
Similar feature observed in Basel (Goertz-Allmann et al., 2011) attibuted to the pore pressure
perturbations

AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Results

• We refined locations of induced earthquakes using Double-Difference method.


– Clear spatial and temporal patterns detected in response to injection
– Seismicity accumulates in highly permeable zone over pre-existing fault planes
• We refined source parameters using spectral ratio method
– Huge improvement in quality of source parameters due to supressed propagation
effects
– Seismicity seem to follow constant stress drop scaling
– Larger events after shut-its, decrease in magnitudes with time of injection, stronger
events on edges of the enhanced reservoir
– Increase of static stress drop with distance

AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Conclusions

• Analysis could be possible after the refinement of the dataset


• Migration of seismicity along existing fault planes due to increasing pore pressure
• Clustering of seismicity after shut-in or when the injection rate is not high enough
• Larger magnitude events observed after shut-in possibly triggered by shear stress
redistribution over the fault plane
• Seismicity occurrs in a certain area only if there is an increase in the load (well head
pressure, injection rate) - Kaiser effect
• Static stress drop seem to reflect the pore pressure perturbations (?)
• Low strength of rocks and high permeability responsible for relatively low seismicity level in
comparison to other geothermal sites

AGIS2012, Karlsruhe, November 26-28th, 2012 Introduction ▶ Site overview ▶ Data refinement ▶ Results ▶ Conclusions
Thank you for your attention!

We would like to acknowledge Julian Bommer, Steve Oates and Gunter Siddiqi for many valuable comments regarding
data processing and injection operations performed in Berlin HFR site. We would like to thank Sabrina Andrae and Oliver
Germer for improving the quality of the original catalog. Pedro Antonio Santos is acknowledged for providing additional
data and comments that helped us to finalize the manuscript. We thank NORSAR (Volker Oye, Julie Albaric) for providing
the dataset and comments regarding data preprocessing. We would like to thank Patricia Martinez-Garzon for providing
translations of a few papers and the geological profile.

AGIS2012, Karlsruhe, November 26-28th, 2012

Você também pode gostar