Seismic time-lapse reservoir monitoring

Traditional methods of monitoring reservoir behavior, including reservoir simulation and history-matching with production rates and pressure, can produce nonunique solutions for reservoir behavior in the interwell regions. In some instances, the uncertainty can be significant, and additional information is needed to optimize production and improve estimates of ultimate recovery. In many cases, the effect of the changing reservoir pressure and/or saturation on seismic data can be used to map the changing pattern of these reservoir properties by obtaining seismic data repeatedly during production of the reservoir. With care, seismic data obtained for other purposes (such as regional exploration) can sometimes be used for time-lapse seismic monitoring, but new data are often obtained from seismic experiments designed particularly to monitor the reservoir. The desire to minimize differences in acquisition parameters between surveys has led, in some cases, to permanent installation of sensors in the oilfield. Because most sensors deployed in this manner are deeply buried and/or cemented, this also has the effect of removing many of the sources of random seismic noise.

Experiments and studies

Experiments

Many seismic time-lapse monitoring experiments have been conducted offshore, where the wells are few and very far apart, and interwell information is especially important. Other experiments have taken place in unusual or expensive production scenarios, such as:

• Steamflooding operations in heavy oil,
• CO₂ flooding,

or

• Thermal recovery.

Because of the need for careful calibration, seismic time-lapse experiments usually include some detailed borehole work, although meaningful results can sometimes be obtained and interpreted even in the absence of good borehole data.

Studies

Three-dimensional (3D) seismic time-lapse studies [occasionally, although ambiguously, referred to as four-dimensional (4D) seismic] use two or more migrated 3D seismic images obtained months or years apart. These can consist of straightforward stacked data volumes or stacks created from partial offsets if AVO aspects are considered. They may also consist of inverted volumes obtained from stacked full-offset or partial-offset data. The comparison can be made in any number of ways, including simple visual inspection. But, it is important to recognize that differences can occur in seismic data even without changes in reservoir properties because of variations in acquisition or processing of the data sets. For example, 3D seismic data acquired from a towed-streamer marine experiment will contain an imprint that results from the direction traveled by the ship. If the experiment is repeated with the ship traveling along a different direction, even though the map grid covered is identical, the seismic rays that are gathered and stacked in each bin will have traveled through different overburden bodies in the two experiments (Fig. 1), resulting in some subtle but noticeable differences. In addition, there are many other small and sometimes uncontrollable differences between most pairs of experiments that must be removed. The process of matching seismic data from multiple experiments is called “cross equalization” and must be carried out taking care not to remove the differences being sought. Usually, seismic data from areas where changes are not anticipated, such as the shallow section, are used to control the cross-equalization process.

• 

  Fig. 1—Rays with different overburden in 3D azimuths. When seismic lines are shot in different directions, the seismic traces may not be identical, even for common reflection points, because of the different overburden encountered. This figure shows two different source-receiver lines with a common reflection point. The reflection characteristics of the horizon observed from a survey shot in one direction may be different from those observed from a survey oriented in a different direction.

The observations made on seismic time-lapse studies frequently include changes in amplitude and changes in time, although any attribute can be used, including inversion results. Changes in amplitude can often be used to directly monitor fluid migration because the reflection character changes as a result of replacing oil/water with gas, as shown in the example in Fig. 2. Other changes in reservoir properties must always be considered, such as effective pressure acting on the rock frame, and it is not always possible to separate these effects using stacked data alone. The use of offset stacks or elastic impedance volumes helps reduce this ambiguity, separating the changes that seem to be caused by fluid substitution from those caused by pressure change, and a seismic petrophysical model is required in the interpretation.

• 

  Fig. 2—Example of time-lapse seismic monitoring of a reservoir from the Gulf of Mexico. Where the amplitude difference is colored in red, gas is expected to have been released or injected into the reservoir. The wells bear this out, with a well colored in red indicating a gas injector or producer with high GOR; a well in blue has watered out, and a well in green produces oil (after Johnston et al.).

The change in seismic velocity between separate monitoring experiments will also result in a change of two-way travel time to reflectors that lie beneath the producing reservoir. This velocity-induced “sag” or “pull-up” may be monitored and provides an indication of the spatial location of reservoir changes, even in reservoirs too thin to image directly (Fig. 3). Because of this effect, interpreters should take great care in the use of direct difference volumes (obtained by simple subtraction of seismic volumes obtained at different times) in the analysis of changes below the uppermost-affected area on the seismic section.

• 

  Fig. 3—Example of seismic time-lapse velocity sag effects. The presence of steam (or other fluid and pressure changes) can change the seismic velocity of the fluid-filled rock enough to result in differences in arrival time for rays passing through the steamed zone. This map shows the change in travel time between two layers encompassing the steamed interval, indicating the movement of steam between two time-lapse seismic experiments. The greatest changes have taken place in the circled areas (from Eastwood et al.).

Fluid or pressure changes may occur outside of the reservoir being produced, and these can sometimes be observed on seismic time-lapse studies, even if they were not anticipated. Such changes can include:

• Variation in the fluid and rock velocities because of changes in pore pressure (therefore, also in effective or differential pressure)
• Changes in rock stress because of deformation of the overburden and sideburden surrounding the reservoir
• Changes in fluid saturation in nearby, unproduced, hydrocarbon reservoirs because of changes in pore pressure (dropping below bubble point) that have been communicated through the aquifer.

Development and application

Originally, seismic time-lapse monitoring was strictly a qualitative subject, and changes observed visually were related in a heuristic way to the reservoir production. As the seismic technology matured, and greater accuracy was assigned to the differences observed, there was an increasing effort to incorporate the results into more quantitative studies. Initially, output from reservoir simulators was used to provide input to Gassmann fluid-substitution schemes to compare with seismic observations. Then, some pressure effects on the rock frame were included. The comparison between predicted seismic changes and those observed was sometimes used to update the original reservoir model, just as history-matching is used to improve the initial model. Currently, there is an effort to fully link the reservoir simulation and its history-matching capability with the data provided by seismic time-lapse monitoring, guiding the simulator (or the engineer) in the interwell regions and further constraining the initial model. These efforts are in some cases related to work on geomechanical modeling of reservoirs for the inclusion of deformation in simulation.

Noteworthy papers in OnePetro

Ayeni, G. and Biondi, B. Time-Lapse Seismic Imaging by Linearized Joint Inversion – A Valhall Field Case Study. Presented at the 2012/11/4/.

Gosselin, O., Aanonsen, S.I., Aavatsmark, I. et al. History Matching Using Time-lapse Seismic (HUTS). Presented at the 2003/1/1/. http://dx.doi.org/10.2118/84464-MS.

Khazanehdari, J., Yi, T., and Curtis, T. Production History-Matching Using Time-Lapse Seismic. Presented at the 2005/1/1/. http://dx.doi.org/10.2118/97100-MS.

Lafet, Y., Roure, B., Doyen, P.M. et al. Global 4-D Seismic Inversion And Time-lapse Fluid Classification. Presented at the 2009/1/1/.

Maa, F., Sollie, R., and Hokstad, K. Prestack Calibration of Time-lapse Seismic Data. Presented at the 2000/1/1/.

Ross, C.P. and Altan, S. Time-Lapse Seismic Monitoring: Repeatability Processing Tests. Presented at the 1997/1/1/. http://dx.doi.org/10.4043/8311-MS.

Tura, A. Time-lapse Seismic: Are We There Yet? Presented at the 2003/1/1/.

Veire, H.H., Borgos, H.G., and Landrø, M. Stochastic Inversion of Pressure And Saturation Changes From Time-lapse Seismic Data. Presented at the 2003/1/1/.

External links

Use this section to provide links to relevant material on websites other than PetroWiki and OnePetro

See also

PEH:Reservoir_Geophysics

Reservoir geophysics overview

Seismic imaging and inversion

Borehole seismic applications

Passive seismic monitoring

Hydraulic fracture monitoring

Pore pressure prediction using seismic