Eisner Styles Et Al 20131

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Main Menu

Lessons learned from hydraulic stimulation of the Bowland Shale


Leo Eisner*, IRSM ASCR, Miroslav Halló, Eva Janská, Ivo Opršal, Petr Matoušek, Seismik s.r.o., Huw Clarke
and Peter Turner, Cuadrilla Resources, Tim Harper, Geosphere Ltd.,Peter Styles, Keele University

Summary been known for long term, that low pressure injections
induce or trigger natural seismicity (Healy et al, 1968).
Shale gas development has caused an energy revolution in Hydraulic fracturing uses usually much more modest
the USA over the past decade. However, the transfer of amounts of fluids than is used in salt water disposal wells
technology abroad has been generally slow and in case of and hence it was thought to cause only low magnitude
Europe negligible. One of the aspects slowing down shale (unfelt) seismicity. The only known exceptions were large
development is induced or triggered seismicity. This study injections for geothermal development in hard (e.g.
summarizes main findings from the analysis of the passive granitic) rocks (e.g., Häring et al, 2008). Year 2011 brought
seismic data available for hydraulic stimulation of Bowland several new cases where felt seismicity was most likely
Shale in Preese Hall well, UK, in 2011. We show that induced by relatively modest hydraulic fracturing into
seismicity was most likely induced by hydraulic fracturing sedimentary formations of Ardmore shales in Oklahoma,
and that unusually large seismic events occurred already USA (Holland, 2013), the Bowland Shale, United Kingdom
during the hydraulic fracturing. These events can be used to (Eisner et al, 2011) and Horn River shale in British
provide warning and modify hydraulic fracturing Colombia, Canada (BC Oil and Gas Commission, 2011).
treatments in future to avoid induced seismicity which
would be socially unacceptable.

Figure 1 Injection rates (red and blue lines) and seismicity (violet and green symbols) observed in the Blackpool area in the vicinity of the
Preese Hall injection well. The red curve represents rate of the injected volume and the blue curve represents rate of flow back volume
from the well head (BPM are barrels per minute). Violet dots are seismic events detected by regional seismic stations (more than 80 km
away), the green triangles represent events detected by two local stations installed at the end of April, 2011. M is a local magnitude
relative to the two largest events detected on regional network.

This study will discuss in detail lessons learned from the


Introduction
Bowland Shale fracturing, one of the best known case
Felt seismicity induced by hydraulic fracturing of the first studies in public domain. The Blackpool area, UK is
gas shale in the UK attracted significant public interest seismically very inactive even to UK standards. The nearest
worldwide and caused suspension of hydraulic fracturing in moderate historical earthquake is more than 150 km away
the UK for shale gas for more than 18 months. Felt induced from the Preese Hall well site and dates back to 1957. The
seismicity caused by an injection of fluids is known since nearest minor earthquake is more than 15 km away from
injection in the Rocky Mountain Arsenal site. And it has the well site and dates back to 19th century.

© 2013 SEG DOI http://dx.doi.org/10.1190/segam2013-0239.1


SEG Houston 2013 Annual Meeting Page 4516
Main Menu

Felt seismicity and hydraulic stimulation in Bowland shale, UK

Shortly after hydraulic stimulations in the Preese Hall well, possibly different from the BGS reported event) but no
the British Geological Survey (BGS) had reported an other seismic events with magnitude greater or equal to
earthquake with local magnitude (ML) 2.3 on 1st of April ML0.25 in the vicinity of the Preese Hall well were found
2011 at 2:34 a.m. This event was located 1.8 km from the during four and half months after the injection (June 30,
well head of this first shale gas well in the UK. The second 2011 through November 25, 2011). We have also found no
stage of hydraulic fracturing was terminated at seismic events with M > 0 and waveforms similar to the
approximately 4 p.m. of the previous day. While such reported events during one year and three months before
earthquake was considered unusual for this area, the March 30, 2011, based on regional network. All of the
possibility of severely mislocated event (nearest operating observations are consistent with very low seismicity in the
station was more than 80 km away) shed only slight area and there is no evidence that seismicity would occur in
suspicion of connection between the earthquake and the future.
hydraulic fracturing. The operator proactively installed two
(BGS) stations (and even additional two Keele University
short period stations) which allowed better monitoring of
possible aftershocks. However, no aftershocks were
detected with the local stations as the seismicity subsided
(Figure 1). Prior to stage 3 the well was open for flowback.
The full size hydraulic fracturing was resumed on May 26-
27, 2011. BGS reported another earthquake ML1.5
approximately 1 km away from the wellhead on May 27 th,
at 0:48 a.m. approximately 10 hours after the termination of
the hydraulic fracturing on the previous day. At this point
the operator voluntarily suspended operations and
organized international scientific investigation group to
find out what had happened.

Seismicity and hydraulic stimulation

Careful investigation revealed total fifty-two seismic events


ranging between ML-2 and ML2.3, with waveforms similar
to the seismic events of April 1, 2011 and May 27, 2011
reported by the British Geological Survey. Fig. 1 shows
injection activity, timing and magnitudes of the detected
seismic events illustrating the causal relationship between
injection and seismicity. The seismicity closely followed
injections on 31st of March and 26th and 27th of May with
remarkably low seismic activity in the period between and Figure 2: Top: particle velocity at regional station KESW
after injections. Only two weak events (M -1.2 and -0.2) of two strongest events from April 1 and May 27, 2011.
were found after May 27, 2011, indicating an extremely Bottom: particle velocity on local station HHF station for
rapid decline in observed seismicity after the end of the the five strongest events between May 26 and May 27,
injections. Similarly only three weak events were observed 2011. The magnitude and origin times of events are shown
between the two injection periods and no event was in the figure legend. Each trace is normalized to its
detected during the stimulation of the stage 3. However, the maximum.
detection threshold was greatly improved by installation of
the local stations on April 12, 2011. Thanks to the local
stations we are able to benchmark detection from the Figure 2 illustrates that reported events of April 1, 2011
regional stations and we know the catalogue from the and May 27, 2011 show great similarity on the five
regional stations is complete down to ML 0. regional stations that recorded both of them (actually Fig. 2
shows only station due to space restriction), limiting the
All detected events were identified through cross- relative distance between the two events to less than 120
correlation detection (e.g., Frohlich et al, 2011) as only two meters as determined form the P- to S-wave arrivals time
local stations were available until the end of June 2011. At difference (bottom panel). Similarly the other events
the end of June (June 28 and 29) two additional local (bottom plot of Figure 2) recorded on local station show
stations were added. With these additional local stations we great similarity of the waveforms for both direct and
attempted to detect arbitrary events (with waveforms scattered arrivals. This observation suggests two important

© 2013 SEG DOI http://dx.doi.org/10.1190/segam2013-0239.1


SEG Houston 2013 Annual Meeting Page 4517
Main Menu

Felt seismicity and hydraulic stimulation in Bowland shale, UK

conclusions: the seismic events were caused by similar Additionally we were able to determine the source
sense of motion on similarly oriented fault(s) and the mechanism of event of August 2, 2011 from direct P- and
events were very close to each other in space (waveform S-wave amplitudes. We fitted pure shear source by least
similarity implies similarity of both source mechanisms and squares method. The mechanism illustrated in Figure 4 can
locations). be characterized as strike-slip failure and its orientation is
consistent with the regional stress direction determined
The array of four local stations operating from the end of from the borehole breakouts ranging from 180o to 195o
June 2011 was used to locate strongest aftershock event of east. This stress would most likely cause shear motion on
August 2, 2011 which is also very similar to the BGS the 70o dipping fault plane with a strike of 40o. Faults with
reported event of May 27, 2011. The BGS reported event of this orientation do not penetrate overburden, which
May 27, 2011 and event of August 2, 2011 are separated by suggests that seismicity is constrained to depths below 2.5
less than 100 meters in the direction HFF station; however, kilometers.
the relative position of these two events is less constrained
in comparison to the August 2, 2011 event, which was very
weak. Figure 3 shows the location of the event of August 2,
2011 relative to the injection intervals of the Preese Hall
well. It is 300–400 meters east of the injection intervals
with uncertainty of approximately 150 meters. The depth of
the event is 330–360 meters below perforations with
uncertainty of 250 meters. Figure 3 shows two locations of
the event based on two velocity models constructed from
measured sonic logs. The eastern location corresponds to a
simple isotropic velocity model, while the western location
was obtained in a VTI model with 20% anisotropy. The
anisotropic model fits significantly better the S-wave
arrivals, although both locations are within the uncertainty
of the other location. The depth is consistent between the
two models.

Figure 4: Equal area projection source mechanism plot


(lower hemisphere projection) of August 2, 2011 event. The
two fault planes have strike, dip and rake 40o; 70o; and -150o
respectively for the first fault plane and strike, dip and rake
299o, 62o and -23o for assumingly auxiliary fault plane.

We observe a small number of weak events relative the


number of large events (low b-value of 0.8±0.3). While this
value may be questioned because of small number of
events observed, the recorded events on the local stations
provide an excellent insight into the completeness of the
catalogue from regional network during the hydraulic
fracturing as shown in Figure 1. Note that events of May 26
and 27 greater than ML0.25 were detected on both local and
regional stations with very similar magnitudes. The
magnitude differences on average account for less than
ML0.1 and most likely result from noise variations on the
regional stations with very low signal to noise ratios. This
observation provides a much greater confidence in the
observed low b-value. The observed b-value is broadly
consistent with b-values observed at other sites of induced
felt seismicity (e.g. Healy et al, 1968, Holland, 2011),
hence we can conclude that the low b-value is not unusual
Figure 3: 3D view (top) and side view (bottom) of location for a felt seismicity induced by hydraulic fracturing.
of the August 2, 2011 event. The well trajectory is
represented by blue line. Perforation intervals (red circles)
represent injection intervals (deeper interval represents
Stage 2) and black triangles are positions of the seismic
stations used in this study. The two resulting locations of
this event are shown as red stars. More western location
represents VTI model and the eastern location represents a
simple isotropic velocity model.

© 2013 SEG DOI http://dx.doi.org/10.1190/segam2013-0239.1


SEG Houston 2013 Annual Meeting Page 4518
Main Menu

Felt seismicity and hydraulic stimulation in Bowland shale, UK

Furthermore, the recorded seismograms allows us to show


that the events with magnitude greater than 0.25 can be Conclusions and Discussion
reliably detected even on noisy stations and events with
magnitude greater than -0.6 can be reliably detected on Prevailing evidence suggests that the felt seismicity in the
more quiet stations in the proximity of the injection well. vicinity of Preese Hall was induced by hydraulic fracturing
Hence, we conclude that seismic monitoring during of the Bowland Shale. The observed seismicity is spatially
hydraulic fracturing by at least four local stations in the limited to a very small volume of several hundred meters at
appropriate geometry, can detect seismicity larger than most activating single fault system. The seismicity decays
magnitude ML0.25 provide early warning, and prevent the extremely rapidly after injections. The most likely activated
induction of significant seismic events potentially fault orientation is consistent with known deep faults that
following the precursory weaker activity (Häring et al do not penetrate the overburden, and the location of
2008). aftershocks suggests the induced seismicity was below the
injection intervals.

The observed unusual seismicity starts tens of minutes after


screen out during the Stage 2. At least 6 unusually large
seismic events are observed during the hydraulic fracture
stimulation of stage 3 on March 31, 2011. These events can
be used as warning of unusual geological setting requiring
further action from the operator in the future. However,
these events were neither felt on the surface nor were
reported by BGS. As the released seismic moment is
Figure 5: Wellhead pressure (WHP) in the Preese Hall proportional to injected volume (McGarr, 1976), the most
exploration well. Events were detected by KESW. appropriate action seems to be reduction of the injected
Displayed period: March 31st, 2011. Figure illustrates that volume. It makes also a common sense, if the injected fluid
at least six seismic events were detectable during the caused unusual seismicity, do not continue with injection.
hydraulic fracturing (when the pressure is above 7000 psi) There is some evidence that draining fluids and reducing
and could serve as warning. pressure in the injection well also reduces seismicity as can
be seen in no seismicity observed on Stage 3 and weaker
seismicity observed after Stage 5 (see Figure 1). Last but
Finally Figure 5 shows detailed picture of proppant rate, not least, screening out may be followed by unusually large
total injected volume, surface pressure and seismicity seismicity and monitoring effort should be increased during
during the stage 2 on March 31, 2011. The seismic events and after such screening out.
detected during this stage are the first observed seismic
events in the Preese Hall well area. We note that at least six Significant discussion of this case study has been devoted
events exceeded magnitude ML0.2 during the injection and to casing deformation observed in the treatment well after
can serve as a warning. As stated above such events would the Stage 2. However, this deformation occurred above the
be easily detectable on the local stations even if they were stimulated intervals while seismicity seems to be
noisy. Furthermore, we notice unusual pressure rise during consistently located deeper, suggesting that such
the first injection of proppant. The pressure rising deformation is not directly related to the fault activation.
exponentially as the proppant concentration increases
suggests screening out - one of the common reasons of Finally we conclude that local seismic stations can detect
screening out during hydraulic fracturing is interaction of unusual induced seismicity during the future hydraulic
hydraulic fracture with pre-existing natural fault. During stimulations and provide most reliable warning for
the screen-out, the fault drains the injected fluids, including operators. If reduction in stimulated volume and perhaps
fracture pad and restricted proppant blocks further growth drainage of the fluids is implemented, shale gas reserves
of the hydraulic fracture. The first observed seismic event can be safely developed with socially acceptable
with ML0.5 occurs approximately 27 minutes after the microseismicity.
screening out. This is a very large seismic event for
hydraulic fracturing and it is reasonable to assume that Acknowledgments
smaller undetectable microseismic events preceded this
event. In fact detailed detection using stacked signal from We would like thank Cuadrilla Resources for support and
regional stations supports this assumption with at least cooperation in this study.
three additional events occurring in this time interval. We
consider this good evidence of fracture interaction with a
pre-existing natural fault.

© 2013 SEG DOI http://dx.doi.org/10.1190/segam2013-0239.1


SEG Houston 2013 Annual Meeting Page 4519
Main Menu

http://dx.doi.org/10.1190/segam2013-0239.1

EDITED REFERENCES
Note: This reference list is a copy-edited version of the reference list submitted by the author. Reference lists for the 2013
SEG Technical Program Expanded Abstracts have been copy edited so that references provided with the online metadata for
each paper will achieve a high degree of linking to cited sources that appear on the Web.

REFERENCES
British Columbia Oil and Gas Commission, 2012, Investigation of observed seismicity in the Horn River
basin, www.bcogc.ca/node/8046/download, accessed 13 June 2013.
Eisner, L., E. Janská, I. Opršal, and P. Matoušek, 2011, Seismic analysis of the events in the vicinity of
the Preese Hall well: Cuadrilla , http://www.cuadrillaresources.com/wp-
content/uploads/2012/02/Geomechanical-Study-Appendix-3-2.11.2011.pdf, accessed 13 June 2013.
Frohlich, C., C. Hayward, B. Stump, and E. Potter, 2011, The Dallas-Fort Worth earthquake sequence:
October 2008 through May 2009: Bulletin of the Seismological Society of America, 101, 327–340,
http://dx.doi.org/10.1785/0120100131.
Häring, M., U. Schanz, F. Ladner, and B. Dyer, 2008, Characterisation of the Basel 1 enhanced
geothermal system: Geothermics, 37, 469–495, http://dx.doi.org/10.1016/j.geothermics.2008.06.002.
Healy, J. T., W. W. Rubey, D. T. Griggs, and C. B. Raleigh, 1968, The Denver earthquakes: Science, 161,
301–1,310.
Holland, A., 2013, Earthquakes triggered by hydraulic fracturing in south-central Oklahoma: Bulletin of
the Seismological Society of America, 103, 1784–1792, http://dx.doi.org/10.1785/0120120109.
McGarr, A., 1976, Seismic moments and volume changes: Journal of Geophysical Research, 81, 1487–
1494, http://dx.doi.org/10.1029/JB081i008p01487.

© 2013 SEG DOI http://dx.doi.org/10.1190/segam2013-0239.1


SEG Houston 2013 Annual Meeting Page 4520

You might also like