Estimation of oil-sands reservoir property changes using seismic

attributes of time-lapse 3D seismic and 3D P-SV converted-wave data

Toru Nakayama1 , Akihisa Takahashi2 , Leigh Skinner3 and Ayato Kato4,
1

Japan Petroleum Exploration Co., Ltd., Japan,

2
JGI, Inc., Japan,
3
Japan Canada Oil Sands Limited, Canada,
4
Japan Oil, Gas and Metals National Corporation, Japan
ABSTRACT
Reservoir characterization was conducted using time-lapse 3D P-P seismic data volumes
and a 3D P-SV converted-wave data volume in the Hangingstone steam-assisted gravity
drainage (SAGD) operations area in Alberta, Canada. The objective of the study was to
characterize oil-sands reservoir properties affected by steam-injection using the seismic data
volumes.
The time-lapse surveys were acquired in February, 2002 and in March, 2006 with nearly
identical field acquisition parameters. One major difference between the surveys is the
receiver type; analog geophone arrays were used in 2002 and three-component digital sensors
were used in 2006. In addition to basic time-lapse analysis using the two 3D P-P volumes, 3D
P-SV data analysis was conducted using the 2006 seismic data
We also proposed a rock physics model based on the P- and S-wave velocities
measurements data of oil-sands core plugs under various pressure and temperature conditions
in a laboratory. This rock physics model was used to predict velocity changes of the oil sands
caused by any variations of pore pressure, temperature, fluid saturation and fluid phase
changes expected during SAGD operations.
Around the active SAGD well pairs, the time-lapse 3D seismic data show significant
differences in seismic character within the oil-sands reservoir. In order to evaluate seismic
response changes of the time-lapse surveys, we calculated a time-difference map between the
reservoir bottom horizons of the 2002 and the 2006 P-P seismic data, and trace-shapesimilarity and amplitude-difference maps within the oil-sands reservoir between the two P-P
seismic data volumes on a trace-by-trace basis. These seismic attribute maps clearly show the
seismic response changes around the active SAGD well pairs.
From our time-lapse study including synthetic seismograms analysis, the seismic response
changes can be quantitatively explained by P-wave velocity decrease of the oil-sands layers
due to the injection of steam. In addition, our result suggests that a larger area would be
influenced by pressure than by temperature.
A Vp/Vs ratio map approximately corresponding to the oil-sands reservoir zone was also
calculated using the seismic interpretation horizons picked on the 2006 P-P and P-SV
volumes. The Vp/Vs ratio map shows the Vp/Vs ratios around high productivity SAGD well
pairs were lower than the ones in low productivity SAGD well pairs.
In conclusion, the time-lapse 3D seismic data and 3D P-SV converted-wave data along with
the rock physics model are useful for qualitatively and quantitatively estimating the oil-sands
reservoir properties.

Introduction
Japan Canada Oil Sands Limited (JACOS) has been operating oil sands reservoirs using an in-situ
steam-assisted gravity drainage (SAGD) method since 1997 in the Hangingstone area which is
approximately 50km southwest of Fort McMurray, Alberta, Canada. In the SAGD process, two
horizontal wells vertically separated by 5m are positioned where high pressure steam is injected
through the upper well (injector) heating the oil sands, reducing viscosity and increasing the mobility
of the bitumen. After steam chamber development, movable bitumen and over-saturated steam
gravitationally move towards and are recovered from the lower well (producer) (Takahashi, 2010).
Steam chamber growth is strongly dependent on the reservoir facies adjacent to the SAGD well pairs
and steam chambers spread irregularly through the reservoir zone. Time-lapse 3D seismic monitoring
was conducted in the JACOS Hangingstone SAGD operations area to delineate steam chamber
growth and to characterize oil-sands reservoir properties affected by steam-injection.
Time-lapse 3D Seismic Surveys
The time-lapse surveys were acquired in February, 2002 and in March, 2006 with nearly identical
field acquisition parameters. One major difference between the surveys is the receiver type; analog
geophone arrays were used in 2002 and three-component digital sensors were used in 2006. In
addition to basic time-lapse analysis using the two 3D P-P volumes, 3D P-SV data analysis was
conducted using the 2006 seismic data (Nakayama et al., 2008).
Figure 1 shows a map of the two 3D seismic surveys and SAGD well locations in the field. The
production of the five eastern SAGD well pairs (A, B, C, D and E) had commenced prior to the
acquisition of the first 3D seismic in February, 2002. After acquisition of the 2002 3D seismic survey,
steam injection commenced at 10 western additional SAGD well pairs (H through Q) in four stages
prior to 2006. The second 3D seismic survey was conducted in March, 2006 over the northern area of
the 2002 baseline survey, where the active fifteen SAGD well pairs existed at that time.

Figure 1 Map of the time-lapse 3D seismic survey and the SAGD well locations. Black solid lines
represent SAGD well paths.
Core Velocity Measurement and Rock Physics Model
Kato et al. (2008) proposed a rock physics model based on the P- and S-wave velocities
measurements data of oil-sands core plugs under various pressure and temperature conditions in a
laboratory. Their rock physics model was used to predict velocity changes of the oil sands caused by
any variations of pore pressure, temperature, fluid saturation and fluid phase changes expected during
SAGD operations (Figure 2). The rock physics model was also used for our interpretations of the
time-lapse 3D seismic data and the 3D P-SV converted-wave data.

Figure 2 Sequential P- & S- wave velocity and Vp/Vs ratio changes induced by steam injection.
Sequential reservoir condition changes are represented by 23 steps. Pore pressure changes from steps
1 to 5, and then is followed by temperature changes from steps 6 to 23. The P- and S-wave velocities
at the initial condition (step 0) are assumed to be 2.4 km/s and 1.0 km/s, respectively.
Time-lapse 3D Seismic Data Analysis and Interpretation
The 2002 baseline data were used as a reference data set, and the 2006 monitoring data were
calibrated to the 2002 baseline data in order to remove differences between them that were not
production-related response changes. The calibration process consisted of band-pass filter, trace
scaling, phase-and-amplitude correction and static time correction.
Figure 3 shows the final migration sections of the North-South line from the 2002 and calibrated 2006
seismic data. The 2006 section of Figure 3 shows time delays of seismic events at the reservoir
bottom (Top Devonian horizon) and below the reservoir around the active SAGD well pairs which
pass through the left half of each section. Figure 3 also shows remarkable seismic response changes
within the oil sands reservoirs (between around Top Wabiskaw and Top Devonian horizons) near the
active SAGD well pairs.

Figure 3 Interpreted North-South line. Top Devonian is regarded as the reservoir bottom (Base
McMurray) and Top Wabiskaw is approximately 5m shallower than the reservoir top (Top
McMurray). The location of this seismic section is described as the red dotted line in Figure 1.

Figure 4 shows a time-difference map between the Top Devonian horizons of the 2002 and the 2006
3D seismic data. As reservoir thickness did not change with the steam injection, positive traveltime
differences in Figure 4 represent P-wave velocity decreases within the reservoir. Significant P-wave
decreases in the oil-sands layer were observed around the SAGD well pairs in the western part of the
survey area. As previously stated, the production of the five eastern well pairs had commenced prior
to the 2002 baseline survey recording. As a result, the reservoir elastic property changes between the
two data sets are relatively small in the eastern areas compared to the western areas.
Trace shape similarity and amplitude differences within the reservoir between the two data volumes
were calculated on a trace-by-trace basis. Figure 5 can be considered trace shape similarity maps
showing maximum cross-correlation values within the reservoir on each trace location. Figure 5
shows low cross-correlation areas (in green to magenta) around the 10 western SAGD well pairs.
These areas of low trace shape similarity indicate areas of large production-related trace shape
changes. In advance of the trace shape similarity calculations, horizon-based time-shift correction was
applied to the 2006 monitoring survey data where the time of the 2006 interpreted horizons were
adjusted to the corresponding 2002 horizons in a stretch and squeeze manner in order to remove
production-related time delays.

Figure 4 Time differences (ms) between the 2002 and 2006 Top Devonian horizons. The positive time
areas show that the 2006 Top Devonian horizon is deeper than the 2002 Top Devonian horizon in the
two-way time domain.

Figure 5 Maximum cross-correlation values within the reservoir on each trace location. Warm colors
mean lower "maximum cross-correlation" values and cold colors represent higher values.

It is noted that there are some inconsistencies between the edges and the shapes of the anomaly areas
observed on Figures 4 and 5. From the oil sands rock physics model, large velocity decreases should
occur in the high temperature zone in the steam chamber, and smaller velocity decreases are predicted
in the high pore pressure zone. Thermal conductivity of the oil sands is so low that the heat front is
relatively close to the steam chamber front. Conversely, the pressure fronts are expected to spread
wider and faster than the heated zones. As a result, a larger area would be influenced by pressure than
temperature. Large time delays of the Top Devonian horizon and large trace shape changes in the
reservoir are expected in the extent of the steam chamber zones, and small time delays of the Top
Devonian horizon and small trace shape changes in the reservoir are expected in the extent of the
increased pore pressure and lower temperature zones. We currently consider that Figure 4 shows
anomalous areas caused by the combination of high temperature and high pore pressure, and Figure 5
mainly represent areas heated during the time between the two 3D seismic surveys.
Conclusions
The time-lapse seismic survey and analysis were conducted in the JACOS Hangingstone SAGD
operations area to delineate steam-affected areas. The two seismic P-P volumes acquired in 2002 and
2006 show large differences in seismic responses within the reservoir around the SAGD well pairs.
From our time-lapse data analysis, the differences of the seismic responses between the time-lapse
seismic volumes can be quantitatively explained by P-wave velocity decrease of the oil sands layers
due to steam-injection. In addition, the data suggests that a larger area would be influenced by
pressure than by temperature. Our analysis of both time-lapse 3D seismic and 3D P-SV data along
with the rock physics model can be used to monitor qualitatively and quantitatively the rock property
changes of the inter-well reservoir sands in the field
Acknowledgments
This study was conducted jointly between Japan Canada Oil Sands Limited (JACOS) and Japan Oil,
Gas and Metals National Corporation (JOGMEC). The authors thank JOGMEC, JACOS and JAPEX
for permission to publish these data.
References
Kato, A., Onozuka, S. and Nakayama, T. [2008] Elastic property changes in a bitumen reservoir
during steam injection. The Leading Edge, 27(9), 1124-1131.
Nakayama, T., Takahashi, A., Skinner, L. and Kato, A. [2008] Monitoring an oil-sands reservoir in
northeast Alberta using time-lapse 3D seismic and 3D P-SV converted-wave Data. The Leading Edge,
27(9), 1158-1175.
Takahashi, A. [2010] Oil-sands reservoir characterization for optimization of field development.
Heavy Oils: Reservoir Characterization and Production Monitoring, SEG, 227-234.