Click to View Poster (2.27 MB)
Long Period Long Duration Seismic Events During Hydraulic Stimulation of a Shale Gas Reservoir*
Indrajit Das1 and Mark D. Zoback2
Search and Discovery Article #40761 (2011)
Posted June 30, 2011
*Adapted from e-poster presentation at AAPG Annual Convention and Exhibition, Houston, Texas, USA, April 10-13, 2011
1
2
Geophysics, Stanford, Stanford, CA ([email protected])
Geophysics, Stanford, Stanford, CA
Abstract
We investigate two classes of unusual events of long duration and relatively low frequencies observed during hydraulic fracturing
operations in a gas shale reservoir. Multiple stages of hydraulic fracturing operations in five sub-parallel wells were monitored with an
array of seismometers deployed in the central horizontal well. When this well was fractured, the array was deployed in a vertical well.
Some of the unusual seismic events recorded are clearly tube waves that propagate along the monitoring well from the heel toward the
toe with a velocity along the borehole of ~1.5km/s. The tube waves propagate down the well from surface, but origin of the tube
waves is not known. The other unusual events are similar in appearance to non-volcanic seismic tremor sequences. They are of 10-50
seconds in duration and are observed in the frequency band of 10-40 Hz, which is much lower than the characteristic frequency band
of microearthquakes (100-300 Hz). Complex but coherent wave trains are observed in both the horizontal and vertical arrays. These
wave trains have very slight moveouts corresponding to apparent velocities ranging from 25 km/s to 9 km/s. The moveout recorded on
the vertical array indicates that they are not caused by a surface noise source; rather they result from a source in the reservoir.
Although it is difficult to resolve any clear P- and S-wave arrivals, one possible source of these low frequency, long duration events is
sub-seismic slow slip on pre-existing faults. The first of these unusual events were observed in the later part of the very first hydrofrac
stage of the first experiment and then in almost all the following stages in the five wells. One interesting observation is that they occur
before the pumping starts in some stages and even after the pumping stops in other stages. In our ongoing work we will be trying to
locate these events using waveform cross-correlation and double-difference tomography.
Copyright © AAPG. Serial rights given by author. For all other rights contact author directly.
References
Nadeau, R. M. and A. Guilhem, 2009, Nonvolcanic Tremor Evolution and the San Simeon and Parkfield, California Earthquakes:
Science, v. 325/5937, p. 191-193. doi:10.1126/science.1174155
Obara, K., 2002, Nonvolcanic Deep Tremor Associated with Subduction in Southwest Japan: Science, v. 296/5573, p. 1679-1681.
doi:10.1126/science.1070378
Peng, Z. and J. Gomberg, 2010, An integrated perspective of the continuum between earthquakes and slow-slip phenomena: Nature
Geoscience, v. 3, p. 599-607. doi:10.1038/ngeo940
Shelly, D.R., G.C. Beroza, S. Ide, and S. Nakamuta, 2006, Low –Frequency earthquakes in Shikoku, Japan, and their relationship to
episodic tremor and slip: Nature, v. 442, p. 188-191. doi.10.1038/nature04931.
Vermylen, J.P. and M.D. Zoback, 2011, Hydraulic Fracturing, Microseismic Magnitudes, and Stress Evolution in the Barnett Shale,
Texas, USA: SPE, 140507-MS, 15 p. doi:10.2118/140507-MS
Long Period Long Duration Seismic Events during Hydraulic
Stimulation of a Shale Gas Reservoir.
Indrajit Das and Mark D. Zoback
Stanford University
Motivation: How exactly does stimulation by hydraulic fracturing occur ?
 Typical induced micro-earthquakes during hydraulic stimulation of a shale gas
reservoir are of magnitudes ~ -2 to -4 and correspond to maximum patch size of
~0.6 m and a displacement of < 0.1 mm.
 Do these micro-earthquakes change the reservoir permeability sufficiently to
give the observed recovery rates or are there other modes of deformation
responsible for the stimulation ?
 Can these processes be observed in the seismograms used for microseismic
monitoring and analysis ?
2
Hydraulic Fracturing and Micro-seismic Monitoring Program
5 Wells, total 40 stages, ~4500 located
micro-earthquakes, ~100 per stage,
3 different fracturing programs
Perforation Shots
• “Simul-Frac” or Simultaneous Fracturing of A & B.
• “Zipper-Frac” or Alternate Sequential Fracturing of D & E.
• Conventional Single Well Fracturing in C.
Analyzing spectrograms as a good way of detecting coherent signals.
High Frequency energy spikes on spectrograms matches exactly with reported micro-earthquakes.
Spectrogram of Axial component of ‘Simul-Frac’ stage 7
Hz
500
0
0s
1000 s
500
Hz
Spike in Spectrogram
Reported micro-earthquake
0
60 seconds
Axial (A)
Channels
1
Transverse 1 (T1)
Transverse 2 (T2)
9
0.4 seconds
0.4 seconds
Long Period Long Duration (LPLD) events in the same spectrograms
Long period (below 80 Hz) long duration (10-100 s) events
detected in the spectrograms
LPLD waveforms after band-pass filtering from 10-80 Hz
100
Hz
350 s
0
6000 s
Tectonic tremor waveforms from Vancouver Island,
Central Range in Taiwan and the SAF (Peng and
Gomberg, 2010)
100
Hz
0
350 s
350 s
Identifying LPLD events: 1) Band-pass filtering method
Axial
Transverse 1
Transverse 2
Original Signal
60 seconds
60 seconds
60 seconds
60 seconds
60 seconds
60 seconds
Noise
Spectra of Signal
and Noise
10
100
400 Hz 10
100
400 Hz 10
100
Signal bandpassed from 10-80
Hz
60 seconds
60 seconds
60 seconds
400 Hz
Identifying LPLD events: 2) Wavelet Decomposition & Reconstruction Method
Levels
0
8
7
6
5
4
3
2
1
8
7
6
5
4
3
2
1
500 seconds
Reconstructing Signal using
energy above 6th Level
“Approximations”
( Low Frequencies)
“Details”
( High Frequencies)
500 seconds
Original Signal
500 seconds
500
Reconstructed
Signal
seconds
Levels
Wavelet
Decomposition
Energy Coefficients
Wavelet Decomposition Method slightly better than Band-Pass Filtering
Original Signal
1000 seconds
After Band Pass
Filtering
Micro-earthquake
1000 seconds
After Wavelet
Decomposition and
Reconstruction
1000 seconds
Gets rid of micro-earthquakes and spikes better
0.2 seconds
LPLD events in the frequency domain
Frequency Amplitude Spectrum of LPLD events, Micro-earthquake and Noise.
LPLD events are deficient in high frequency energy relative to
micro-earthquakes and enriched in low frequency energy.
LPLD events in the time domain.
Microearthquakes detected within some LPLD events. Coincidental or Causal ?
LPLD events in time series
Meq. within LPLD event in stacked time series
A
T1
T2
LPLD events in spectrogram
Meq. within LPLD event in stacked spectrogram
100
Hz
0
1000
Hz
0
A
100
1000
Hz
0
T1
1000
Hz
0
T2
Hz
0
100
Hz
0
180 seconds
2 seconds
Meq. shown in all 3 components
Impulsive arrivals with finite move-outs but no clear P or S phases
LPLD events in the
horizontal array
6 seconds
LPLD events in the
vertical array
6 seconds
0.8 seconds
7 ms move-out
0.8 seconds
29 ms move-out
Move-out direction in vertical array proves source
is in the reservoir !
Exact Move-outs obtained from waveform cross-correlation
Move-out across the array is obtained by cross-correlating the LPLD signal
envelope in the first channel with the rest of the channels
Cross- Correlogram
LPLD Signal
Cross-correlating 1st channel
with the rest of the channels
90 Seconds
-0.2 s
From the move-out we can get the apparent velocity across the array.
From apparent velocity we can get the approximate angle of arrival.
0s
Lag time
0.2 s
Density and Orientation of Natural Fractures in the Reservoir
Well C
II
Position of Sensors in well C during “Simul-Frac”
I
Heel
Toe
11
All stages in well C colored differently.
10
9
8
7
6
5
4
3
80
Number of natural fractures per stage in well C
0
210
Number of Microearthquakes per stage
0
Stages 6-9
Rose Diagram showing fracture orientations
Fracture density and orientations from well C has been
extrapolated to well A, B, C & D for subsequent analysis
Stages 1-5
2
1
Where do the LPLD events occur ?
Events while
Stage 7 pumping stopped
Stage 7
(0-1500s )
Stage 8
Stage 8
(0-1000s )
Stage 9
Stage 8
(8000-8300s)
Stage 9
(1300-2700s)
c
-89° -78°
-75° -78°-79°
-81° -80° -81°
-79°
69°
62° 64°
Each LPLD event is composed of multiple adjacent
sources all within a narrow range of angles.
-81
-79°
-83°
-79°
-89°
Where do the LPLD events occur ? ( conceptual model)
SHmax
Well C
Stages 6-9
Perfs ( Stage 7, 8 & 9 )
* * *
Microseisms ( Stage 7,8 & 9)
Hydro-frac
- 89
Pre-existing Faults
- 75
LPLD events
62
69
Well C
SHmin
Well B
Well A
Source of LPLD events seems to be along
two preexisting natural fractures
crosscutting a couple of hydro-frac planes.
When do the LPLD events occur ?
Stages with maximum number of LPLD events ( stage 7 & 8) are also the stages with the highest
pumping pressure and highest natural fracture density.
Well A
Well B
5500
ISIP (Psi)
3500
10 Stages
1
Conclusions
 Long period long duration seismic events , similar in appearance to tectonic tremor observed in subduction
zones and strike slip margins, was identified during hydraulic stimulation of shale gas reservoir.
 These LPLD events lack clear P and S phases but have coherent arrivals with finite move-outs. Some of the
LPLD episodes have micro-earthquakes within them although it is not yet known whether they are coincidental
or causal?
 From the moveout direction of all the events in the vertical array it was confirmed that they are coming from
the reservoir.
 Analysis of stage 7,8 and 9 of well A-B suggests that all LPLD events maybe coming from two faults
crosscutting a couple of hydrofrac planes.
 Maximum number of LPLD events were detected in the stages with the highest pressure during pumping and
the highest natural fracture density.
 All observations suggests that the long duration signals are generated by slow shear slip on a few preexisting
natural fractures due to the high fluid pressure in the reservoir. We believe this process might be contributing
significantly to the stimulation.