Multicomponent, time lapse seismology has great potential for monitoring production processes in reservoirs. The reason is simply the presence of fractures. Shear waves are much more sensitive than P waves to the presence of fractures or microfractures and the fluid content within the fracture network. Fractures introduce seismic anisotropy into a reservoir, causing two shear modes to propagate with different velocities and therefore resulting in different arrival times. The arrival time difference is referred to as shear wave splitting or birefringence and is a critical parameter for estimating fracture density (see Martin and Davis, 1987).
Shear wave splitting is the key to monitoring production processes. At Vacuum Field the behavior of the fluid property changes associated with CO2 flooding caused changes in the velocities of the split shear waves passing through the reservoir interval. Fluid properties change in response to CO2 and oil entering a miscible phase. Shear wave splitting is useful to identify areas of anomalous reservoir pressure. Shear wave splitting and velocities are extremely sensitive to the local stress field because all rocks and especially carbonates contain incipient networks of microfractures at a state of near-criticality (Zatsepin and Crampin, 1997).
Weyburn Field is the current site of experimental monitoring efforts designed to enhance the resolution of multicomponent seismic data to monitor production processes. At Weyburn CO2 flooding is occurring in a thin 30 m (100 ft) fractured carbonate reservoir. There amplitudes of split shear waves are being used to observe and monitor production processes associated with the flooding.
Vacuum Field is located 20 miles west of Hobbs on the Northwestern Shelf of the Permian Basin in the state of New Mexico. This field produces oil from the San Andres (divided into upper and lower) and Grayburg (dolomite and sandstone) formations at an average depth of 4500 feet. The geological setting of these formations is that of the rim of a broad carbonate shelf province to the north and northwest and of a deeper intracratonic basin on the southeast and east. Dominant reservoir homogeneities include faults, fracture zones, anhydrite plugged zones linked to old fracture systems, and the layered structure of the reservoir itself.
The San Andres reservoir has been under production since 1940. In 1980 a water flooding program provided an increase of production from an average of 3000 BOPD to a peak average production of 17000 BOPD. After 10 years of water injection, production declined to an average of 4000 BOPD and it was decided to start an infill drilling campaign and to prepare for a tertiary CO2 flood. In 1995, the first CO2 injection program took place at a single well (see Talley et al, 1998) and in 1997 CO2 injection was expanded to a series of six injectors. Due to this tertiary recovery program, production is projected to increase up to an average of 6000 BOPD by 2003.
Multicomponent Seismic Data Acquisition and Processing
During the six-well CO2 injection program a baseline multicomponent seismic survey was acquired in December 1997, prior to the start of injection which began in April, 1998. The monitoring survey was conducted in December 1998, eight months after the start of injection. Approximately one billion standard cubic feet of CO2 was injected during this time interval.
Because time-lapse effects are subtle, the baseline and monitor survey(s) were designed to maximize the signal-to-noise ratio of the data as well as its repeatability. Because seismic anisotropy is the key to reservoir monitoring, maintaining high fold within the useable offset range that Alford rotation is applicable is essential. In processing, linear processes were used that are surface consistent, thereby preserving the integrity of the signal between the various surveys.
Multicomponent seismic data revealed the presence of faults with 10 to 20 feet of vertical throw at the reservoir level. Recurrent movement on these faults created fracture zones in the reservoir. The open fracture systems are conduits for fluid movement vertically and laterally in the reservoir. The faults, in some locations, are partially sealing due to the presence of anhydrite, cataclasis or offset of flow units. The six-well injection introduced approximately 1 billion scf of CO2 by the time of our monitoring survey. The injection rates were designed to keep the static reservoir pressure constant so the anisotropy anomalies are largely associated with fluid saturation change and not effective stress changes.
The time-lapse seismic interpretation of the seismic data shows a differential seismic anisotropy anomaly between the baseline and the seismic monitoring survey that coincides with the tertiary flood bank (Figure 1). This anomaly is measured over the entire reservoir interval and is shown as a velocity anomaly where the velocity VS1 decreased and the VS2 increased. Spatially, this anomaly suggests a strong fracture control on the permeability. Carbon dioxide has a high mobility ratio and can move over relatively large distances particularly within a fracture network.
The offset water injectors to the south were used to contain the CO2 bank. Areas around these injectors show areas where the S2 velocity decreased due to local overpressuring and the propping open of fractures due to overpressuring.
The greatest need is to monitor and control the areal and vertical distribution of CO2 in the reservoir to maximize contact with the oil and to optimize sweep efficiency so that oil is not bypassed in the tertiary recovery operation. By producing timelapse anisotropy differences we have been able to track the tertiary flood bank and produce a spatial image of the bank. This enables us to monitor the lateral sweep efficiency of the reservoir. The vertical sweep efficiency is handled through amplitude differentials of split shear waves. S2 amplitude difference anomalies between the pre and post surveys occur dominantly in the Lower San Andres. This is highly encouraging because shear wave anisotropy may give us higher vertical resolution, enabling us to visualize changes approaching the flow unit scale.
Weyburn Field, located on the northeast flank of the Williston Basin in southeast Saskatchewan, Canada is currently being monitored utilizing 4-D multicomponent seismic technology. Approximately 1000 wells, including 137 horizontal wells with 284 lateral legs, have been used to recover 24% of the 1.4 billion barrels of oil originally in place. Pan Canadian, the operator, has converted 19 patterns of horizontal wells to CO2 injection. Injection of 3 to 7 mmcf/day/well has occurred since early October 2000. The goal of CO2 flooding is to increase production by at least 15% incremental oil.
Weyburn Field produces from the Mississippian Midale Beds of the Charles Formation. The reservoir is made up of two parts: the uppermost Marly dolomite and the lowermost Vuggy limestone. The reservoir is overlain by an anhydrite unit that forms the top and updip lateral seal to the reservoir. The combination of the overlying anhydrite and the porous upper part of the reservoir provides an acoustic impedance contrast and a seismic reflector coincident with the reservoir interval. The reservoir zone generally averages 100 ft (30 m) in thickness, has a temperature of 145° F (63° C), and a pressure of approximately 3000 psia (20.7 MPa).
The most porous unit is the Marly, averaging 26% porosity. The Marly averages 30 ft (9 m) thick in the study area. Permeability of this zone is low, averaging 10 md. Horizontal wells drilled since 1991 in Weyburn Field have targeted the Marly as a zone of bypassed pay. These wells have substantiated the belief that the Marly unit has not been as effectively swept as its underlying counterpart, the Vuggy. The Vuggy averages 70 ft (21 m) in thickness with an average porosity of 11% and permeability of 15 md. The flow capacity of the formation is the product of permeability and net thickness. The Marly has a relatively low flow capacity relative to the Vuggy and correspondingly low sweep efficiency. The potential for bypassed oil in the Marly is greater with CO2 flooding than it is with waterflooding because of the comparatively high mobility of CO2. The specific goal of our research is to create a high-resolution measurement system for monitoring the sweep of the Midale reservoir at Weyburn Field.
A baseline survey was acquired in Fall 2000 prior to CO2 injection. The areal extent of the survey encompasses approximately 3.4 square miles (9 square kms) and monitors at least four injection patterns. The surface data was shot with a fine grid of source and receivers to maximize the quality of seismic data for poststack and pre-stack analysis of amplitude. These amplitudes will enable determination of attribute volumes of bulk density, compressional and shear wave velocities, and fracture density/permeability on a high resolution basis as input to the dynamic reservoir characterization process. A repeat or monitoring seismic survey is scheduled for Fall 2001.
If we can monitor bypass areas laterally and vertically in the reservoir by 4-D, multicomponent seismology we can effectively manage the reservoir. The bottom line is to lower costs, and increase recovery. Multicomponent 4-D seismology enhances the accuracy of the reservoir model. An enhanced reservoir model enables improved forecasting which allows for better operational response. Better operations and improved reservoir management positively affects the bottom line. The technology pays for itself through the additional percentage of incremental recovery that can be added by increased recovery efficiency.
The authors thank industry sponsors of the Colorado School of Mines Reservoir Characterization Project for supporting this research.
About the Author(s)
Thomas L. Davis is currently Professor of Geophysics at Colorado School of Mines. At Mines he guides the leading edge research of the Reservoir Characterization Project, whose mission is to develop and apply time-lapse (4- D), multicomponent (3-C and 9-C) seismology to improved recovery. He has been an organizer of technical conferences, workshops and Continuing Education programs for the SEG. Tom was the SEG’s Second Vice President in 1989, Distinguished Lecturer in Spring, 1995 and Technical Program Co-Chairman in 1996. He received the C. J. Mackenzie Award from the Engineering College of the University of Saskatchewan in 1997, the Milton B. Dobrin Award from the University of Houston in 1998 and the Colorado School of Mines Dean’s Excellence Award in 1999.
Martin, M. A. and Davis, T. L., 1987, Shear wave birefringence: A new tool for evaluating fractured reservoirs: The Leading Edge, v. 6, no. 10, p. 22 - 28.
Talley, D. J., Davis, T. L., Benson, R. D., and Roche, S. L., 1998, Dynamic reservoir characterization of Vacuum Field: The Leading Edge, v. 17, no. 10, p. 1396 - 1402.
Zatsepin, S. V., and Crampin, S., 1997, Modelling the compliance of crustal rock-I. Response of shear-wave splitting to differential stress, Geophysical Journal International, v. 129, p. 477 - 494.