Communications – electrical: acoustic wave systems and devices – Seismic prospecting – Land-reflection type
Reexamination Certificate
1999-02-04
2002-08-20
Lobo, Ian J. (Department: 3662)
Communications, electrical: acoustic wave systems and devices
Seismic prospecting
Land-reflection type
C367S046000
Reexamination Certificate
active
06438069
ABSTRACT:
BACKGROUND OF THE INVENTION
This invention relates to oil and gas reservoir management and, more specifically, to time lapse reservoir seismic signal processing.
Reservoir characterization and monitoring in the oil and gas field are important parts of reservoir management and hydrocarbon production. Effective reservoir management is a major goal of energy producing companies as they try to reduce finding costs, optimize drilling locations and increase financial returns. One technique that is attempted in this endeavor is time-lapse (also known as 4D) seismic monitoring. As fluids are extracted, swept, or injected through production and recovery, changes in the effective elastic properties of the reservoir rocks occur. The ability to monitor reservoir changes as a function of time by the use of seismic methods can lead to better location of production and infill wells, the possibility of locating unswept zones, and more efficient field maintenance, thus raising the overall value of the production lease.
In a two-dimensional approach, seismic monitoring has been examined in crosswell procedures. However, the repeated results have only been qualitatively compared, and two dimensional time-lapse surveys, so far, do not contain the type of information desired in modern reservoir management (for example, see Paulsson, et. al, 1994,
The Leading Edge
, incorporated herein by reference). Time-lapse 3D techniques have also been attempted, but they involve complex modeling procedures and require a great deal of processing without using direct data available in the survey itself. See, e.g., U.S. Pat. No. 4,969,130, incorporated herein by reference.
One problem in time lapse processing is that many conditions change over time, not just the changes in the state of the reservoir. For example, the locations of the source and receiver in the second survey will necessarily be different from those in the first. Further, the tide in a second marine survey may be higher or lower, as may the temperature of the air and water. Likewise, the specific characteristics of the sources and receivers used in the second survey will be different. Other differences, besides changes in reservoir state also occur, such as differences in the manner in which the two surveys are processed. Thus, there is a need for a method of dealing with the two surveys whereby processing differences do not detrimentally affect the result of the comparison.
For example, in gathering seismic data, a source is used to generate seismic waves which reflect from the reflectors in the earth (e.g. layer boundaries) and are received at receivers. In some cases, the source signature is a spike, although, in reality, it is not perfect. During its journey through the strata and reflectors, the signal shape is changed, and the reflection signal received at the receivers is, therefore, no longer a spike, or even close. Deconvolution is the process by which the shape of the reflection signal is “whitened” to recreate the spike shape of the data.
In another example, a broad, band-limited signal is used, which is zero phase. Deconvolution is used in such a case to remove the distortions caused by the earth.
In still another example, in performing the deconvolution in the frequency domain, all frequency samples are multiplied to bring them to an equal level, following the assumption that the source is a minimum phase signal, immediately rising to a peak and then dying. This is accomplished by autocorrelating the trace in the time domain multiple times, at a series of lag samples, which results in a generally symmetrical wavelet. The power spectrum of the wavelet is then analyzed, to determine what multipliers are needed at each frequency to flatten the frequency spectrum. This process is performed on a windowed basis, both along each trace, and across the record (as used herein, the term “record” refers to, alternatively for example, a common receiver record, a CMP record, a common shot record, a stacked trace record, etc.) The autocorrelation is performed on various windows, and the results are averaged to give the spectrum. From that spectrum, the operators needed to flatten the spectrum are chosen. The operators are then applied to all of the traces used in the input. Typically, the window is about 10 times the length of the operator to be generated, measured in number of samples. The deconvolution process and the design of a deconvolution operator are well known in the art, and it is not limited to the frequency domain example, above. It is also routinely performed in the time domain. See, e.g., Yilmaz,
Investigations in Geophysics
, Vol. 2,
Seismic Data Processing
, Society of Exploration Geophysicists (1987) and references cited therein.
In performing deconvolution, it is important to design a deconvolution operator dependent upon the data of the survey, in order to account for the specific source signature, and other equipment distortions that occur in the data. Therefore, the data in each survey has been adjusted through the use of a specific optimum deconvolution operator which is not applicable to other surveys. The result of this difference in the use of the separate deconvolution operators in time-lapse surveys is that structure appears in the difference records when two records are subtracted. This result is undesirable. However, to date, no one has proposed a practical solution to the problem.
SUMMARY OF THE INVENTION
It is an object of the present invention to address the above problems.
It has been found that, contrary to earlier beliefs, a single deconvolution operator can be used on multiple sets of data, not only without detrimental results, but improving the quality of the processing of time-lapse comparisons of seismic surveys. Accordingly, in one aspect of the present invention, a method of deconvolution of multiple sets of seismic data from the same geographic area is provided, the method comprising: designing of a deconvolution operator dependent upon data from at least two of the sets of seismic data, wherein the at least two sets of seismic data were recorded at different times or calendar dates; applying the deconvolution operator in a deconvolution process to both of the at least two sets of data; and conducting further time-lapse processing to form a difference record.
According to one embodiment of the invention, the conducting of further time-lapse processing comprises: providing a first reflection event (for example, a wavelet) in the first seismic survey data set having a corresponding second reflection event in the second seismic survey data set, wherein the first reflection event and the second reflection event represent an unchanged portion of geologic structure in or near the reservoir and wherein the first reflection event is represented by a first set of event parameters and the second reflection event is represented by a second set of event parameters. Next, an acceptance threshold difference function between the first set of event parameters and the second set of event parameters is provided. Then, a crossequalization function is determined to apply to the second set of event parameters.
According to another aspect of the invention, the crossequalization function is determined such that, upon application of the crossequalization function to the second set of event parameters, a crossequalized set of event parameters is defined, and the difference between the first set of event parameters and the crossequalized set of event parameters is below the threshold difference function. Next, the crossequalization function is applied to a third reflection event, the third reflection event being related to the second data set, wherein a crossequalized third reflection event is defined, wherein the third reflection event has a corresponding fourth reflection event in the first data set, and wherein the third and fourth reflection events represent a changing portion of the reservoir.
Comparison of the crossequalized third reflection event to the fourth reflection event by subtracting the crossequalized third reflec
Altan Mehmet Suat
Ross Christopher Philip
Lobo Ian J.
PGS Data Processing, Inc.
LandOfFree
Method for time lapse reservoir monitoring does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for time lapse reservoir monitoring, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for time lapse reservoir monitoring will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2963057