Transcript Time lapse vsp - West Virginia University

Time-lapse seismic simulations of potential CO2 induced changes in
acoustic properties of coals for MMV
NETL ORD Work Planning Session, Hyatt Regency, Pittsburgh: May 3-4,2010
Thomas H. Wilson and Art Wells
Abstract
We are involved in site characterization and MMV efforts at pilot tests in
the San Juan and Appalachian basins. In the San Juan Basin, CO2 injection
began July 30th of 2008 and continued through August 14th of 2009. During
the 12 month injection period approximately 256 MMCF, equivalent to nearly
18,407 short tons of CO2 (319 MMcF) were injected into the Fruitland coals.
Time-laps vertical seismic profiles (VSP) were collected to evaluate the
potential of this method to image the CO2 flood: one on June 3rd and 4th of
2008, and the post injection survey, September 17th, 2009 (a month after CO2
injection was completed). Processing of these data are still in progress.
WVU is also involved in another coal sequestration effort here in the
Appalachian Basin through ZERT with CONSOL and NETL. Time lapse
seismic imaging of the flood front will also be evaluated on that site.
Some published studies suggest that acoustic impedance of coal should
increase because of coal swelling. Xue and Ohsumi (2005), for example,
make detailed measurements of swelling strain and waveform traveltime
changes for the Kushiro Coal, Hokkaido. They note a 10% increase in Pwave velocity at 2.5 MPa (~362psi) and perhaps up to 12.7% at 12 MPa
(supercritical). Nishimoto et al. (2008) report only 2.2% increase in Vp at
12MPa under supercritical conditions.
McCrank (2009) notes that CO2 injection into the Ardley Coal, Alberta
produces a 10% reduction in velocity attributed to increased coal plasticity
after a 9 month CO2 soak.
In this study we calculate AVO variations in CMP gathers using full
solutions to the Zoepritz. CMP gathers are used as a proxy for the VSP, with
short to long offsets corresponding roughly to upper to lower borehole
sensor locations. Two possible scenarios are modeled: 1) CO2 injection
reduces coal velocity and 2) CO2 injection increases coal velocity. The
results suggest that in both cases significant time lapse response occurs
due to relative delay or advance in the pre-post arrival times.
Objective
Evaluate potential AVO and time lapse response to CO2 injection.
Present simulations for alternative cases in which CO2 increases and
decreases coal zone velocity. Determine the potential for time lapse
AVO observations in CMP and VSP records.
Zero offset VSP used as TD
curve to adjust sonic sonic and
density travel time data used to
generate AVO synthetics.
Structure – base of Fruitland
0’
3000’
monitor
Baseline - monitor
Original and modified logs for the fast
model (above). Synthetic AVO response
computed for the fast case (above right)
13000
Amplitude
12000
11000
10000
9000
8000
7000
0
500
1000
1500
2000
2500
3000
Offset
Post Injection AVO (0.6035 s)
13000
Amplitude
12000
11000
10000
Post Injection travel
time advance
9000
8000
7000
0
500
1000
1500
2000
2500
3000
Offset
Pre-Post (0.604s)
monitor
1000
Baseline - monitor
800
3000’
Velocity of
individual
coals
decreased by
10%
Case 2
Amplitude
0’
baseline
600
400
200
0
-200
-400
0
500
1000
1500
2000
2500
3000
3500
Offset (feet)
Post Injection AVO (0.605s)
15000
15000
14000
14000
13000
13000
Amplitude
Amplitude
AVO Response - Pre Injection (0.6045s)
12000
11000
10000
12000
11000
10000
9000
9000
8000
0
500
1000
1500
2000
2500
8000
3000
0
500
Offset (feet)
1000
1500
2000
2500
3000
Offset (feet)
Pre-Post (0.605s)
0
Amplitude
-500
Post Injection travel
time delay
B.
A.
SWP Vertical Seismic Profile
Case 1
Velocity in coal
section
increased 10%
Fruitland Formation
reservoir zone
A) Migrated stack line passing through the injection well (COM A ING 1) and two production wells (COM A
300 and STATE COM). B) Enhanced view reveals potential faults and fracture zones within the Fruitland
Formation and Kirtland Shale. The VSP was run in the center COM A ING 1 well.
AVO Response - Pre Injection (0.6045 s)
baseline
The Kirtland Shale
primary seal (caprock)
-1000
-1500
Three offset VSPs and one zero offset
VSP were collected at the site prior to
CO2 injection and 1.3 years later
following completion of injection. The
source point locations are shown on the
QuickBird image at left (green squares).
Presence of archaeologically sensitive
areas at the site limited our choice of
offsets. The image at left also shows
locations of wells, NETL tracer and soil
gas sample points and tiltmeters
Conclusions
Time lapse responses in common midpoint (CMP) gathers were
computed as a proxy to estimate vertical seismic profile (VSP)
response for two cases. In one case, the velocity is increased by 10%
throughout the entire Fruitland coal section including 10 to 15 foot
zones above and below the Fruitland coals. The rationale for this case
is that coal swelling in response to CO2 injection increases Vp (Xue
and Ohsumi, 2005). The swelling strain is assumed to produce a
pressure halo that increases velocity of intervals sandwiched between
the coals and intervals bounding the coal zones.
Amplitude variation with offset (AVO) was evaluated for the positive
event just below the top of the upper Fruitland coal at 0.6045 s. The
amplitude variation for this event drops with offset (upper graph at
left). The post-injection drop (middle left) is greater. Thus the baseline
minus monitor difference increases with offset. The post-injection
arrival time is advanced 1 ms.
In the second case, coal velocity is decreased 10%. The decrease is
limited to the coals. In this case the post-injection zero offset
amplitude is much greater as expected. The travel time is delayed in
this case by 0.5 ms. The post injection AVO drop is also much greater
so that an increase in amplitude with offset is again observed in the
baseline minus monitor difference (see graphs lower part of left panel).
The increase in amplitude with offset is nearly the same for both high
velocity and low velocity cases.
In this study, time lapse differences in two CMP attributes were
evaluated: 1) AVO, and 2) travel time delay or advance. Travel time
delay or advance is a discriminating attribute whereas the difference in
AVO is not. The simulations suggest that differentiation between the
two cases can be obtained in CMP gathers or VSP through crossequalization of seismic response above the injection zone followed by
careful analysis of travel time differences between events in the
baseline and monitor surveys arising from within and beneath the
injection zone.
Source point locations
Baseline
Monitor
Baseline - Monitor
Comparison of monitor and
baseline VSPs for source
offset D northwest of the
injection well. Displays
represent intermediate stage
processing results (Jitendra
Gulati with Schlumberger).
Baseline and monitor surveys
were cross-equalized before
differencing.
Although subtle, the initial
interpretation suggests that
interval velocities may have
increased in response to CO2
injection.
References
• Henthorn, B., Wilson, T., and Wells, A., 2007, Subsurface Characterization of a Carbon
Sequestration Pilot Site: San Juan Basin, NM: Annual AAPG Convention, Proceedings CD. See
also http://www.searchanddiscovery.net/documents/2007/07047henthorn /index.htm &
http://www.geo.wvu.edu/~wilson/netl/ HenthornWilson&Wells -07AAPG.pdf
• McCrank, M., 2009, Seismic detection and characterization of a CO2 flood in Ardley Coals,
Alberta, Canada: M. S. Thesis, Department of Geoscience, Calgary, Alberta, 191p.
• Xue, Z., and Ohsumi, T., 2005, Experimental studies on coal matrix swelling due to carbon
dioxide adsorption and its effect on coal permeability: Shigen-to-Sozai, vol 121, no.6, p. 231239 (in Japanese with English abstract and figure captions).
• Xue, Z., and Ohsumi, T., 2003, Laboratory measurements on swelling in coals caused by
adsorption of carbon dioxide and its impact on permeability of coal: Coal & Safety, no. 23, p
36-43.
-2000
-2500
-3000
0
500
1000
1500
2000
2500
3000
3500
Offset (feet)
Comparisons of synthetic baseline and monitor CMP gathers showing time lapse responses for two cases: 1)
increased velocity in the coal section and 2) decreased velocity within individual seams. Close up views of the
Fruitland coal section highlight differences observed in both cases. AVO plots are presented for both cases. The
CMP gathers are used as a proxy for the VSP response: the response at longer CMP offsets corresponds to
deeper phones in the borehole VSP.
Acknowledgements
This technical effort was performed in support of the National Energy Technology Laboratory’s on-going research in carbon sequestration under the RDS
contract DE-AC26-04NT41817-6060404000 and URS subcontract No. 2010-SC-RES-30033-023. We’d like to thank Dave Wildman and Donald Martello, our
DOE-NETL project managers, for their support and advice on these efforts; Scott Reeves and Brian McPherson of the Southwest Regional Partnership for their
help in facilitating our involvement in the Partnership’s activities on their San Juan Basin carbon sequestration pilot test and for allowing us to use data collected
as part of the pilot effort; and Ryan Frost.Tom Cochrane. Bill Akwari and Craig Hartline of Conoco Phillips for helping facilitate many of the activities on the site.
Appreciation is also extended to Dwight Peters, Marcia Coueslan, Jitendra Gulati, Les Nutt and Ric Smith with Schlumberger for critical assistance in the design
of the logging and VSP acquisition, analysis and processing efforts at the site. We also want to thank Bill O’Doud (NETL) SWP project manager.