Patent Application
20120116680
The invention relates to the monitoring and estimation, with a high degree of resolution, of the location of subsurface microseismic sources associated with fluid movement in hydrocarbon-containing reservoir rock.
The accurate monitoring of fluid movement in hydrocarbon reservoirs is essential for increasing the recovery from the reservoir rock reserves. Direct measurement of fluid flow in the reservoirs, if at all possible, is a complex and expensive task. As a result, indirect methods have been investigated. One useful approach is to measure source locations of seismic events, which are often triggered by rock fractures resulting from fluid movement. These microseismic hydraulic fracture events are also referred to as “hydro-fracs”. An important step in locating the seismic sources are the calculation of Green's functions, which can be used to map the recorded seismic energy to where it was generated to thereby identifying the source locations.
Reservoir management based on microseismic signals emanating from reservoir rock requires the precise estimation of the microseismic event source locations to optimally guide fluid-injection rates and injection locations. A problem encountered in the field is that the required precision of the microseismic source location is beyond the seismic resolution of a typical passive seismic array located on the surface. The narrow aperture of the surface receivers, high signal/noise ratios, and the relatively long wavelengths of the source prevent a precise seismic estimate of where the rock is being fractured.
A sophisticated mathematical analysis, or analytical tool, referred to as a propagator function, can be applied to an area of interest in order to map the recorded seismic energy to its source location. It has been known in the art to utilize Green's functions in connection with the analysis of available data.
The following patent documents disclose various methods relating to the use of Green's functions in modeling seismic data.
Brandsberg-Dahl et al. USP 2008/0130411 disclose a method utilizing a measured (near-exact) Green's functions that requires vertical seismic profile (“VSP”) data to image the surface seismic data. The methods of Brandsberg-Dahl et al. are described as being applicable to instances where the velocity profile is approximately one-dimensional and where there are lateral velocity variations. Imaging of surface seismic data and “self-imaging” of VSP data is disclosed as being possible without first having to estimate a velocity model. The measurements obtained from the VSP data can also be used as a tool for calibrating computed Green's functions and migration operators, e.g., in a Kirchoff migration or other type of migration.
Etgen U.S. Pat. No. 6,049,759 discloses a process that is based upon the observation that an underlying 3-D velocity model consisting of horizontally layered constant-velocity media (i.e., a so-called v(z) velocity model) results in prestack migration of a collection of common offset traces that can be written as a convolution of a migration operator with the seismic data. Based on this observation, Etgen discloses that the speed of the migration may be increased by transforming both the migration operator and the seismic data to another domain, such as a Fourier transform, where the convolution between the two can be calculated more efficiently. The process starts with specification of a velocity model that does not vary with respect to the x or y (horizontal) coordinates. It is critical to this instant embodiment that this model be horizontally layered if the migration result is to be exact. In a preferred embodiment disclosed by Etgen, two Green's function arrays are computed from the information contained within the velocity model. The two arrays theoretical travel times and resulting amplitudes of a wave front that originates as a point source on the surface is measured remotely at various radial distances and depths within the specified velocity model. The Green's function arrays are used in the calculation of the migration operator. Etgen discloses that the Green's functions need be calculated only once for the entire migration, as the velocity model is constrained to be horizontally layered. The choice of a particular migration approach (e.g., Kirchhoff migration, finite difference migration, an F-K algorithm, a finite element algorithm, or a reflectivity algorithm) determines the equations that are used to calculate the values within the operator volume. The migration operator is then applied to a common offset seismic data volume, and a discrete 3-D Fourier transform of the common offset data volume is calculated. The migration then continues by operating on a set of individual constant frequency slices taken from the volume of 3-D Fourier transform coefficients. Etgen process must be based on the velocity model. In addition, the use of Green's function in Etgen is a precursor to another migration operator, such as Kirchhoff migration.
Marmalyevskyy, et al. U.S. Pat. No. 7,110,323 discloses a method for using non-primary reflections to identify and characterize steeply dipping sub-vertical events. Secondary reflections or duplex waves are migrated while the effects of primary reflections are suppressed. Substantially vertical events are located by gathering common source or receiver traces for processing. Wave fields of these gathers are continued downward to the level of the base boundary, and then a seismic image of sub-vertical events is formed at each discrete depth level. The migration of non-primary reflections is generally based on the Kirchoff transformation whereby the Green's function is repeatedly recalculated along the subject fault according to the kinematics of the subject non-primary reflections, while the velocity model is that used for conventional migration procedures.
He, et al. U.S. Pat. No. 5,798,982 discloses a method for constructing an impedance model of a subsurface volume including obtaining a 3-D seismic image of the subsurface volume from observed seismic reflection trace data, constructing an a priori impedance model of the subsurface volume based on estimated impedance trends within the subsurface volume, and creating a model of the seismic reflection trace data for the subsurface volume based on the a priori impedance model by combining the a priori impedance model with a seismic source function or wavelet which is time/depth dependent. The model and the observed seismic reflection trace data are compared and the a priori impedance model is modified to decrease the variation. This process is directed to obtaining different information and, as in the prior reference, there is no application of Green's function in He, et al. as a propagator function.
The following additional patent references describe the use of various types of functions that are used to provide models to provide migrated seismic data, without specific reference to Green's function: Martinez et al. U.S. Pat. No. 6,826,484; Thomsen U.S. Pat. No. 6,128,580; Reshef et al. U.S. Pat. No. 6,904,368; Kelly U.S. Pat. No. 6,687,617; Carrazzone, et al. U.S. Pat. No. 5,583,825; Lee et al. U.S. Pat. No. 7,388,808; Meng U.S. Pat. No. 6,519,532; Ober et al. U.S. Pat. No. 6,021,094; Tal-Ezer U.S. Pat. No. 6,819,628; Ren U.S. Pat. No. 7,035,737; Kelly U.S. Pat. No. 7,039,526; and Ren et al. U.S. Pat. No. 6,466,873.
From the above, it will be understood that it is well known that a seismic image can be obtained by correlating Green's functions, as well as various other functions, with seismic data. In accordance with the prior art, the Green's function could be obtained by two methods:
a. by forward modeling with known a velocity model, such as ray-tracing, finite-difference extrapolation, phase-shift, and others; or
b. by direct measurement based on classical VSP data, such as was described above in the Brandsberg-Dahl, et al. US2008/0130411 disclosure.
The known methods of the prior art in some instances produce results lacking in the desired level of accuracy in mapping the microseismic event source locations due to the assumptions made in processing the data utilized and/or the methods require complex, and therefore expensive, arrays of equipment to collect the data required to practice the method. One particular limitation is the requirement of developing a velocity model.
It is therefore an object of the present invention to provide an improved method for deriving and use of Green's functions as the propagator function in order to more accurately image or map the location of microseismic events that originate in a volume of reservoir rock.
It is another object of this invention to overcome prior art resolution limitations that are the result of sparse passive recording geometry.
A further object of this invention is to provide a method of utilizing all the natural arrivals in the data for a given reservoir region in order to estimate the locations of microseismic sources with super resolution.
In accordance with the present invention, the Green's function is derived from the seismic data itself, that is, no velocity model is necessary. The process of the invention provides a seismic image for identifying the location of a subsurface microseismic hydro-frac event initiated by fluid movement in a predetermined region of the reservoir rock with high resolution and includes the steps of:
a. collecting seismic data generated by a drill bit during drilling of a well bore in the reservoir and correlating the seismic data with the corresponding drill bar vibration data;
b. obtaining an estimate of local velocity in the reservoir surrounding the well bore;
c. calculating the natural Green's function(s) for the earth from the seismic-while-drilling data of step (a) using the formula:
G(x+dx|x′)
where x′ is the seismic-while-drilling source location;
d. extrapolating the phase shift migration from data taken along the surface using a migration operator e−iωτ, where ω is frequency and τ is time, as an extrapolation operator to extrapolate the traces to greater depths;
e. collecting seismic data corresponding to one or more microseismic events occurring in the predetermined region of the reservoir;
f. applying the Green's function(s) obtained in step (c) to estimate the location of the microseismic events associated with the data of step (e) in accordance with the following formula:
m(x+dx)=G(x+dxIx)*d(x′)dx′ [1]
where d(x′) represents passive seismic data in the frequency domain along the surface x′, and m(x′) represents the image of the point scatterer, which is equivalent to zero-offset migration of post stack data, and x+dx represents the unknown location of the hydro-frac source and the integration is over the trace locations on the surface; and
g. displaying the results of solving the formula of step (f) as images corresponding to the location of the one or more microseismic events.
The “natural” Green's function is constructed, e.g., by correlating the drill bar vibration corresponding to the pilot signal for the drill bit with the seismic waves generated by the drill bit while drilling in the region of interest. The Green's function(s) are also derived by summing correlated traces between master traces and the remaining traces in shot gathers. In this method, all of the natural arrivals in the data are used. When additional measurements become available, the data can be used to further refine and/or confirm the reliability of the Green's functions that have been applied. The additional measurement can also be used as a cross-check for the method of the invention. The additional measurements can also be used in combination with the method of the invention to improve the accuracy of the images generated.
In accordance with this invention, the band-limited Green's function G(x,z,t|x′z′,0) can also be estimated from VSP (vertical seismic profile), RVSP (reverse VSP), or surface seismic data and applied to estimate the location of hydro-frac source locations. These Green's functions are used as the migration kernels and provide estimates of the hydraulic fracturing source locations that are more accurate than the diffraction limit method because all of the natural arrivals in the data are utilized.
The Green's function G(x,z,t|x′,z′,0) can be estimated from surface seismic data using several alternative methods. The Green's function can be obtained by recording the drill bit pulses on the surface, which is equivalent to RVSP data. The pilot signal for the drill bit is recorded at the drill bit and then correlated with the observed data to get the impulse-like Green's function. The Green's function can also be approximated by a VSP-type experiment. To estimate the Green's functions at locations near the drill bit, wavefield extrapolation methods can be used that are similar to the wave field extrapolation of seismic data when a local velocity model is of the present invention is obtained by conventional well logging methods. However, in the method of the present invention no local velocity model is required.
Another method for estimating the Green's function is by correlating every trace in a surface seismic shot gather with a master trace in the same shot gather and then summing the resulting correlation trace for all shot gathers to yield G(x,z,t|x′,z′,0). This Green's function G(x,z,t|x′, z′,0) is for a scatterer buried at (x′, z′) as if it were a buried source shooting into surface geophones at (x,z) and listening time t. The estimated Green's function resembles that of a shot gather for an RVSP experiment, except that it is freely available from surface seismic data rather than from an expensive RVSP experiment. This procedure can be used to complement common focus point (CFP) methods that are based on certain assumptions about seismic statics. Such assumptions are not required in the practice of the method of the present invention.
As used herein the term “shot gather” includes the energy or waves emanating from microseismic hydraulic fracture events in the reservoir rock.
The invention will be described in further detail below and with reference to the attached drawings and illustrations in which:
In accordance with this invention, Green's function G(x,z,t|x′,z′,0) is estimated from various data sources, including RVSP, VSP, SWD or surface data, and used to provide images of the estimated locations of hydro-frac sources with a high degree of resolution. These Green's functions can be used as the migration kernels to estimate hydro-frac source locations with greater accuracy than the diffraction limit, because all of the natural arrivals in the data are utilized.
Green's function can be estimated from surface seismic data using two different methods. The first method is to use the drill bit as a seismic source as is known in the seismic-while-drilling (SWD) methods. The seismic data are recorded on the surface and correlated with the recorded pilot signal to give the impulse-like Green's function. Evaluation of the Green's function far from the drill bit is achieved by employing a wavefield continuation method using a local velocity model.
2. The second method estimates the Green's function by correlating a trace in the surface seismic shot gather with a master trace in the same shot gather and then summing up the resulting correlation trace for all shot gather yields G(x,z,t|x′,z′,0). This Green's function G(x,z,t|x′,z′,0) is for a scatterer buried at (x′,z′) as if it were a buried source shooting into surface geophones at (x,z). The estimated Green's function resembles that of a shot gather for an RVSP experiment, except that it is freely available from surface seismic data rather than from an expensive RVSP experiment. This method can be used as a complement to common focusing point (CFP) methods that have certain assumptions about seismic statics. Such assumptions are not required by interferometric redatuming.
Once the Green's function is estimated from the data, the source is located by seeking the maximum value of the migration image m(x′,z′,t) for all trial source locations (x′,z′) in the model and time shift values t; the summation is over all trace indices g and over the temporal variable t′, Here d(g,0,t) represents the passive data recorded on the surface at location (g,0) where the Cartesian coordinate system's origin is at the surface and z=0 defines the free surface. Also, G(g,0,t|x′,z′,0) represents the recorded Green's function for a source at (x′,z′) and receiver at (g,0). For brevity, the notation here is illustrated for a 2D model, but this method is equally applicable to a 3D model with an areal recording array on the surface. For a 3D application, the G(g,0,t|x′,z′,0) representation is replaced with the 3D representation G(x,y,z,t|x′,y′, z′,0) and d(x,z=0,t) by d(x,y,z=0) where the y axis is in and out of the page and perpendicular to the x-z plane.
In accordance with the present invention, the Green's function G(x,z,t|x′,z′,0) of a subterranean region is estimated from (1) vertical seismic profile (VSP) data, (2) reverse vertical seismic profile (RVSP) data, (3) seismic-while-drilling (SWD) data, or (4) seismic surface data. The estimated Green's function is then used in a migration image function, and the maximum value of the migration image provides the estimated location(s) of the hydro-fracturing source(s). The migration image m(x′,z′,t) can be represented as: m(x′,z′,t)=sum_g sum_t′G(g,0,t+t′|x′,z′,0) d(g,0,t′). This value is determined for all trial source locations (x′,z′) in the model and time shift values t. The summation is over all trace indices g and over the temporal variable t′. In the equation, d(g,0,t) represents the passive data recorded on the surface at location (g,0) where the Cartesian coordinate system's origin is at the surface and z=0 defines the free surface. Also, G(g,0,t|x′,z′,0) represents the recorded Green's function for a source at (x′,z′) and a receiver at (g,0).
In prior art migration techniques, such as standard Kirchhoff migrations, the model is based upon a narrow receiver aperture and coarse geophone sampling which results in imperfect cancellation of neighboring ellipses in the migration image. Consequently, artifacts appear as “wide smiles.” However, in accordance with the present invention, a reverse time migration is used with Green's function as the operator, i.e., full-trace migration, where data is summed along traveltime curves associated with all events. This provides super resolution capabilities, which is also known as multi-arrival super resolution, because it effectively extends the aperture of the recording experiment.
In accordance with the invention, and in order to address these limitations of the prior art, the Green's function G(x|x′) of the earth is obtained from the drill bit as will be described in more detail below, or VSP experiment, and this Green's function is used to image the hydro-frac source location to better than the diffraction limits of the seismic recording, i.e., less than 5* L horizontal resolution where L is aperture width and z is depth. Most importantly, for the practice of the method of the present invention, no global velocity model is needed.
The Green's function obtained from a seismic-while-drilling (SWD) experiment is used to estimate the location of the hydro-frac sources using the following formula.
m(x+dx)=G(x+dxIx)*d(x′)dx′ [1]
where d(x′) represents passive seismic data in the frequency domain along the surface x′, and m(x′) represents the image of the point scatterer, which is equivalent to zero-offset migration of post stack data, and x+dx represents the unknown location of the hydro-frac source and the integration is over the trace locations on the surface.
The Green's function G(x+dxx′) in equation 1 can be estimated from the measured SWD Green's function G(x|x′), where x′ is the SWD source location by a typical extrapolation operation. For example, phase shift migration takes data along the surface and extrapolates the traces to deeper depths by an exponential-like extrapolation operator. A similar procedure can be applied to a series of SWD Green's functions obtained from a series of sources along the well trajectory, except that the sources are extrapolated away from the well bore. This extrapolation requires a local estimate of the velocity around the well, which are easily obtained from the well logs or other similar procedures known to the art.
Referring to
Utilization of equation 1 overcomes the resolution limitations inherent in a sparse passive recording geometry. This is because equation 1 uses the natural Green's function of the earth, which includes all events, such as multiples and diffractions. Using this method provides super resolution, which is also referred to as multi-arrival super resolution capabilities because it effectively extends the aperture of the recording process or protocol.
With further reference to the illustrations of
With continuing reference to the illustration on the right-hand side of
A further advantage of the method of the invention that improves image resolution is the inclusion of evanescent waves in imaging the data. An evanescent wave is a nearfield standing wave with an intensity that exhibits exponential decay with distance from the boundary at which the wave was formed. Evanescent waves are a general property of wave-equations, and can, in principle, occur in any context to which a Green's function wave-equation applies. Evanescent waves are formed at the boundary between two media having different wave motion properties and are most intense within one-third of a wavelength from the surface of formation. Evanscent waves are formed when seismic waves traveling in a region of an underground geological formation undergo total internal reflection at the formation boundary because they strike it at an angle greater than the critical angle. The physical explanation for the existence of the evanescent wave is that the pressure gradients of acoustical waves cannot be discontinuous at a boundary, as would be the case if there were no evanescent wave-field.
Both synthetic and field RVSP data have been used to confirm the accuracy of the method of the present invention. The synthetic data were generated for the Red Sea salt body model shown in
The synthetic RVSP salt body data were generated by a 2D finite-difference solution to the acoustic wave equation for 1000 surface geophone receivers spaced at 20 m intervals. The RVSP point sources for a vertical well were buried at depths starting at 3100 m, and spaced at 20 m intervals to deeper depths; in total, 50 shot gathers were generated for this RVSP configuration as shown by the vertical white bar in
Equation 1 is used to image the location of an unknown point source located somewhere along the well. In the examples, the point source location is at the depth of 3500 m along the well. The full-trace migration images are compared to the early-arrival migration images, and a time window is used to mute out all arrivals except the earliest arrivals. For example, a 60 ms early-arrival migration operator includes only the arrivals within a 60-ms duration time window centered about the first onset of energy in the trace. The 3D image of the point scatterer is shown in the inset in
The early-arrival migration operator will have less resolution than the full-trace operator, because it only accounts for the direct wave field and does not include all of the scattering events. This predicted result is validated by the point source images shown in
Referring to
Typically, the signal/noise ratio of an image is enhanced by {N}1/2, where N is the number of traces with additive random noise. With super stacking, it is theoretically possible to increase the signal/noise ratio by a factor of {N+M}1/2, or more, where M is the number of strong reflection events in the migration operator shown in
Referring now to
With reference to
The super fold property is also confirmed by the results shown in
Referring now to
Using reciprocity, the VSP data was transformed into virtual RVSP data. The reservoir site shot gather is referred to below as RVSP data. On this basis, the super resolution properties of the site VSP data can be compared and analyzed with the RVSP synthetic data that was described above.
Referring now to
It will be noted that the images shown in
Synthetic simulations demonstrate the increase in spatial resolution and noise reduction in imaging point source locations with semi-natural Green's functions. These Green's functions are obtained by combining RVSP, VSP, or recorded data with numerical data modeled in a local region about the source point. Using the entire trace as the migration operator produces results that are similar to full-trace migration, except that the velocity model is not needed. Similar to increasing the receiver aperture in the data, spatial resolution and fold are increased by widening the time window in the semi-natural Green's function. A spatial resolution to ¼ of the seismic wavelength appears to be attainable using the apparatus and setup described above. Resolution of up to 1/30th of a wavelength is attainable with use of small-scale scatterers near the well, and the reception of evanescent energy.
Although the invention has been described with respect to specific illustrative embodiments, further modifications and alternatives will be apparent from the description to those of ordinary skill in the art, and the scope of invention is therefore to be determined by the claims that follow.
