WO2016202403A1

WO2016202403A1 - Method for determining the seismic signature of a drill bit acting as a seismic source

Info

Publication number: WO2016202403A1
Application number: PCT/EP2015/063836
Authority: WO
Inventors: Jakob B.U. Haldorsen
Original assignee: Read As
Priority date: 2015-06-19
Filing date: 2015-06-19
Publication date: 2016-12-22

Abstract

A method for determining the seismic signature of a drill bit acting as a seismic source while drilling. The method comprises the steps of recording compressional and shear waves generated by the drill bit by using at least one three- component sensor; separating recorded waves by decomposing data into three components; determining time delay between compressional waves and fast and slow moving shear waves on each component by measuring arrival times on said sensor as well as polarization of the waves; combining determined time delays and polarization of waves for determining the location of the seismic source, and determining the seismic signature by combining recorded data and the determined location of the seismic source.

Description

Method for determining the seismic signature of a drill bit acting as a seismic source Introduction

The invention comprises a method for determining the seismic signature of a drill bit acting as a seismic source while drilling. The method allows data to be recorded while drilling, where the data are processed like normal seismic data generated by a surface source.

Background

There are a number of publications describing commercial and semi-commercial

technologies for using noise generated by a working drill bit as a seismic source.

US-2013/0128693A1 (Geiser) is an example where a drill bit is used as a seismic source for performing velocity analysis. Another example is US-7,512,034B2 (Haldorsen) describing processing methods for acquiring drill noise seismic data.

An important limitation and a major obstacle for commercial success of drill-noise related seismic solutions has been the reliance on extracting a pilot signal from axial vibrations transmitted through drill pipe from drill bit to the surface. This limitation is particularly important in offshore work where Polycrystalline Diamond Compact (PDC) shear bits are commonly used. Formation signals generated by a PDC bit, in particular horizontally polarized compressional or shear signals, may couple poorly to the axial vibration in the drill string. In deviated wells, using any type of drill bit, the drill-string transmission is substantially lost due to friction between the drill string and the formation.

The present invention will overcome and eliminate these limitations. This is achieved by combining source-estimation technology known from earthquake seismology to determine source mechanisms using acoustic waves generated during hydraulic-fracturing operations. Using single receiver stations, this technique involves detection of two or three mutually orthogonally polarized seismic waves, and using essentially the estimation method for these elemental waves as outlined in: Haldorsen, J. B. U., Brooks, N. J., and Milenkovic, M., "Locating Microseismic Sources using Migration-Based Deconvolution"; Geophysics, 78, September/October, 2013. This publication describes a method for finding micro-seismic hypocenters from data recorded by arrays of triaxial motion sensors. The method

reconstructs the elastic time-series signatures for possible micro-seismic sources at any point in 3D space by using full-waveform migration of the recorded vector wave field. The imaging condition for the migration is based on a semblance-weighted deconvolution between two or more reconstructed modal source signatures, requiring similarity and simultaneity of the reconstructed signatures. This imaging condition eliminates the need for an absolute timing of the data, giving optimum resolution for the location of the micro- seismic sources which is better than correlation-based approaches and ensures numerical stability by adapting to the signal and noise conditions of the data. Because the method eliminates time-consuming phase picking and traditional event association, it is well suited for fully or semi-automated data processing.

Seismic waves may possibly be overlapping in time, but as they propagate at different velocities, they will be seen to be arriving at different times. Using more that one set of sensors one can determine more detailed parameters for the source of the seismic waves.

According to the present invention, knowledge of the location of the drill bit and its compressional- and shear waves signatures of the working drill bit is used to record data while drilling and processing the data as ordinary seismic data generated by a surface source with an extended source signature.

The seismic signature of a working drill bit can be determined by using elements of moment-tensor analysis, involving both directly transmitted compressional and shear waves. This is a technique known from earthquake seismology, as well as from the monitoring of micro-earthquakes and hydraulic-fracturing operations for tight hydrocarbon reservoirs.

The purpose of the present invention is to gain improved knowledge of seismic signatures and the location of the drill bit allowing seismic data recorded while drilling to be processed like normal seismic data generated by a vibrating surface source with an extended source signature, e.g. a Vibroseis source. Subsidiary, the detailed source signature and the radiation pattern will give information about the mechanism by which the working drill bit crushes the formation rock, knowledge that can be used to improve the construction of the drill bit and increase the effectiveness of the drilling.

Short description of the invention

The present invention is a method for determining the seismic signature of a drill bit acting as a seismic source while drilling. The method comprises the steps of:

- recording compressional and shear waves generated by the drill bit by using at least one three-component sensor;

separating recorded waves by decomposing data into three components;

determining time delay between compressional waves and fast and slow moving shear waves on each component by measuring arrival times on said sensor as well as polarization of the waves;

combining determined time delays and polarization of waves for determining the location of the seismic source, and

determining the seismic signature by combining recorded data and the determined location of the seismic source.

Further features of the method are defined in the claims. Detailed description of the invention

The invention will now be described in detail with reference to the figures.

Figure 1 illustrates the nine different couples required to obtain equivalent forces for a generally oriented displacement discontinuity in anisotropic media.

Figure 2 shows synthetic, mixed fast and slow orthogonal waves rotated by the azimuth and polar angles (φ, Θ) = (50°, 40°).

Figure 3 shows a map of the energy in the six off-diagonal elements of the covariance matrix (described below) for the three traces shown in Figure 2 as a function of 3-D rotation angles for the coordinate system.

Figure 4 shows the three orthogonal waveforms appearing from the data in Figure 2, after a 3D rotation by the angles that minimized the energy in the off-diagonal terms in the covariance matrix. It can be seen that the three elemental modes of the wave field are completely separated by the rotation corresponding to the minimum indicated in Figure 3.

Figure 5 shows data acquired while performing a hydraulic fracturing operation.

Figure 6 shows the decomposition of the data from Figure 5 into three components by minimizing the off-diagonal terms in their covariance matrix. The separation discussed above has three equivalent solutions. For the purpose of Figure 6, we are placing the component with the least energy on component number 3.

Figure 7 shows the raw waveforms recorded on a vertical array consisting of 12 three- component receivers during a hydraulic fracturing operation.

Figure 8 shows the data in Figure 7 after rotating the coordinate system to show 3 orthogonal wave-field components.

Figure 9 shows the image of the seismic source obtained from the separated data shown in figure 8, after applying a semblance- weighted deconvolution filter based on the compressional signal, and migration.

In a drilling situation, stress seen by the rock is a combination of stress introduced by the rotating bit and stress naturally present in the rock and imposed from the environment. When this stress exceeds the elastic limit of the rock, the rock fails.

Rock failure will be associated with a displacement discontinuity, and will lead to the emission of a seismic signal that in turn may be recorded at receivers. The recorded signal contains information about the character of the seismic source that generated the recorded signal comprising a combination of waves. Different failure mechanisms will generate different combinations of compressional and shear waves with different radiation patterns.

A rotating drill bit will set off a sequence of micro-earthquakes when crunching rock. The drill bit will thus act as a seismic source. From recordings of propagating waves by one or more three-component seismic sensors, the hypocenter and the signature of the seismic source generating so called micro-earthquakes can be determined. When the signature is determined seismic data can be processed like normal seismic data generated by a vibrating surface source with an extended source signature.

A seismic moment tensor relates to the effects of displacement of discontinuities in 3D space and gives the displacement at any point in space when convolved with a derivative of the Green's tensor describing the medium between the rupture point and the observation point. The seismic moment tensor is a quantity that depends on the source strength and orientation of the rock failure, and it characterizes all information about the source that can be learned from observing waves whose wavelength is much longer than the linear length of the rock failure.

The seismic moment tensor represents force couples acting in one direction with an arm in another direction. For three components of forces and three components of arms, there are nine different force couple as indicated in Figure 1.

Figure 1 illustrates the nine different couples required to obtain equivalent forces for a generally oriented displacement discontinuity in anisotropic media. The nine combinations are annotated by the orientation of direction and offset of the attack points for the forces, e.g. (y,x) means that the force couple acts in y-direction and the attack points are displaced in the x direction.

In earthquake seismology, the measurements are mostly inverted for a source mechanism involving pure shear-slip faults. As opposed to earthquake seismology where volume changes normally cannot occur, the objective for hydraulic fracturing is to open up new venues for the oil and gas to flow. This calls for a more complex moment-tensor inversion, also including sources with volumetric components, such as tensile cracks (along the diagonal in the 3x3 representation in Figure 1).

The diagrammatic representation in Figure 1 could conceivably be applied to either roller- cone or PDC bits, with a somewhat different direct identification of the force couples.

Initially one may want to implement the tools to do a full moment tensor analysis.

The method for determining the seismic signature of a drill bit acting as a seismic source while drilling comprises several steps. The first step is recording compressional- and shear waves generated by a drill bit by using at least one three-component sensor.

Using arrays with several three-component sensors will make the determination of the source parameters more accurate. Using more receiver stations enables more accurate estimates of more source parameters that could influence the radiation pattern of the generated seismic waves. This would give a less biased estimate of the source signature.

Knowledge of the position of the drill bit could be used to constrain the solution. At the least, one can say that two subsequent micro tremors or micro earthquakes are located close to each other in space.

Source parameters such as time series and source polarization for finite time intervals, ranging from a few seconds to several minutes, can be combined and used to condition the seismic sensors to compress the source signature to a spike using a process generally known as signature deconvolution, e.g.: Haldorsen, J., D. Miller, and J. Walsh, 1994, "Multichannel Wiener deconvolution of vertical seismic profiles": Geophysics, 59, 1500-151 1 , and for a noise-optimized version of deconvolution: Chen, C.-W., D.E. Miller, H.A. Djikpesse, J.B.U. Haldorsen, and S. Rondenay, 2010, "Array- conditioned deconvolution of multiple component teleseismic recording", Geophys. J. Int., 182, 967-976, for the application on earthquake data.

In essence, the method for estimating the compressional and shear signatures from recorded data, and from these estimating the location of the micro-seismic sources, is described by Haldorsen, J. B. U., Brooks, N. J., and Milenkovic, M., "Locating Microseismic Sources using Migration-Based Deconvolution", Geophysics, 78, September/October, 2013.

However, knowing the location of the seismic source, in this case the drill bit, means that one can constrain the estimation of the source signatures. Then, knowing both the location and the seismic signatures of the working drill bit, the recorded data can be processed like ordinary seismic data generated by a surface source with an extended source signature.

In the following, the method for separating orthogonal waves rely heavily on the work of Haldorsen, Brooks and Milenkovic (2013), ref. Haldorsen, J. B. U., Brooks, N. J., and Milenkovic, M., Locating Microseismic Sources using Migration-Based Deconvolution: Geophysics, 78, September/October, 2013.

It is expected that a working drill bit will generate a mixture of compressional and shear waves propagating at different velocities, with the compressional component travelling typically 70 percent faster than the shear component.

When a shear wave travels through an anisotropic layer, the shear wave is split, with a faster shear wave polarized parallel to the dominant fracture orientation or the maximum stress direction, and a slower shear wave polarized perpendicular to the fracture/maximum-stress direction.

Provided they have the same generating seismic source, the compressional, fast shear and slow shear all have similar waveforms but will be polarized in mutually orthogonal directions.

As mentioned, the first method step according to the invention is recording compressional and shear waves. As the receivers axes are unlikely to be aligned to the polarization of the compressional and shear waves, each of the components of the recorded wave field will contain mixtures of compressional and shear modes.

The second step is separating the recorded waves by determining time delay between compressional waves and fast and slow moving shear waves. This is done by measuring arrival times on said sensor as well as polarization of the waves.

At its most basic level, the analysis performed on the waveforms recorded on one three- component sensor should be able to separate these three components of the wave field. Although one receiver station with a three-component sensor may be sufficient for determining some of the source parameters, more accurate estimates of the source

parameters could be obtained by using two or more receiver stations, comprising arrays of three-component sensors. Undetermined parameters may significantly bias the estimates of the parameters one tries to determine. To get information about radiation patterns, one would like to have more than one array of three-component receivers.

The time delay between the fast and the slow shear waves gives an indication of the degree of the anisotropy of the formation, and the time delay between all three components, the compressional- and the two share waves, will give an indication of the distance to the seismic source.

Using At for the difference in arrival time, and As for the difference in slowness (inverse velocity) of two components of the wave-field, the distance from the receiver to the source is given by:

Figure 2 shows a synthetic example of recordings containing three orthogonal receivers both recording a mix of three mutually orthogonal waveforms arriving at slightly different times and with polarization not coinciding with the axis of the receivers.

The synthetic data comprises mixed fast and slow orthogonal waves rotated by the azimuth and polar angles (φ, Θ) = (50°, 40°). Each of the three wave-field components contains a 30 Hz Ricker wavelet, one starting at around 40 ms, the other two are delayed by additional 24 and 34 ms.

The "covariance matrix" for the three mutually orthogonal Carthesian components of the wave field, and it is a 3-by-3 matrix with nine elements, each of which is the correlation function between two of the components. The "diagonal elements" in the 3-by-3 matrix contain the autocorrelation functions for the three components, and the "off-diagonal" elements contain the cross-correlation functions. If each of the three components have different arrival times at the receiver (as they will for compressional, vertical and

horizontally polarized components generated by a single source), the cross-correlation functions between the compressional and shear modes will have a peak at the arrival-time difference between the modes.

Figure 3 shows a map of the energy in the six off-diagonal elements in the covariance matrix for the three traces shown in Figure 2 as a function of 3D rotation angles for the coordinate system. A rotation by the angles associated with the least off-diagonal energy gives the three traces shown in Figure 4.

Figure 4 shows the three orthogonal waveforms appearing from the data in Figure 2, after a 3D rotation by the angles that minimized the energy in the off-diagonal terms in the covariance matrix, as indicated in Figure 3. Figure 4 shows very good agreement with the input parameters. Of all equivalent right-handed permutations of axis all giving the same total power of the off-diagonal elements of the covariance matrix, we have arbitrarily chosen the one with the least energy on the 3^rd axis. Alternatively, one could choose the solution that places the fastest component on axis number 1.

This wave-field decomposition technique has been tested on data acquired while performing hydraulic fracturing operation.

Figure 5 shows data acquired while performing a hydraulic fracturing operation. The single- level three-component data that has been oriented by minimizing the energy on component number 3.

Figure 6 shows the result of decomposition of the data from Figure 5 into three components by minimizing the off-diagonal terms in their covariance matrix, while placing the component with the least energy on component number 3. The data have been separated with a possible compressional arrival on component 2 at around 0.14 s, and the shear arrivals on component 1 at around 0.35 s, while component 3 appears to be virtually unchanged, indicating that there is little or no observable shear-wave splitting for the particular rock formation and acquisition geometry.

A separation in time of 210 ms between wave field number 2 and wave field number 1 , assuming the propagation velocities are 4000 and 2000 m/s, used in Equation (1) gives a distance to the source of 840 m, along the direction of the polarization vector for the fastest of the two compressional components .

In Figures 7 to 9, we follow this one step further with data acquired in a different hydraulic fracturing operation.

Figure 7 shows raw wave-forms recorded on a vertical array consisting of 12 three- component receivers during a hydraulic fracturing operation.

Figure 8 shows the data in Figure 7 after rotating the coordinate system to show three orthogonal wave-field components by the method described above, i.e. ordered according to energy with the most energetic shear component in component 1 , the least energetic compressional component in component 3. Component 2 appears to contain a wave field with a somewhat slower propagation velocity than the primary shear component.

Figure 9 shows the image of the seismic source obtained from the separated data shown in Figure 8. According to the method described by Haldorsen et al., 2013, the image is obtained by calculating, for both compressional and shear waves, the propagation times and ray angles at the receiver from each of the points (x,z) in the horizontal range from 0 to 600 m and depths from 1700 to 2300 m, The three mutually orthogonal recordings in Figure 7 were projected onto the ray (for compressional) and perpendicular to the ray (for shear), shifted back in time by the estimated travel times. Each of the projected and time-shifted wave forms should now contain a receiver-specific estimate of the compressional and shear waves that possibly could have been generated at the point (x,z), at a time reference of the generation of the acoustic energy at the point (x,z). By summing the estimates over all receivers, one is left with three time series: one which we call the "compressional source signature" as it involves a projection along the propagating rays, typical for a compressional wave, and two which we call the "shear source signatures" as they involve projections perpendicular to the propagating rays, typical for shear wave fields. The image in Figure 9 is the value at time equal to zero of the semblance weighted deconvolution of the shear signatures by the compressional signature. The result of the deconvolution is a measure of the coherence of the signals. Mathematically, with respect to time, one may say that the semblance-weighted deconvolution process measures the timing of the shear signatures relative to the timing of the compressional signature (as was illustrated by Equation (1)). The value of the de-convolved signatures at time equal to zero measures their simultaneity. Absolute coherency and simultaneity of the compressional and shear signatures will result in de-convolved time series with a perfect spikes at time equal zero.

Considering that the source in this case was the working drill bit, the image in Figure 9 demonstrates that a semblance-deconvolution operator based on the estimated compressional wave field gives an adequate focusing on the location of the source.

In a practical example, one may want to apply different tools for separating the incoming waves, such as maximum correlation methods or Singular Value Decomposition (SVD); however, Figures 2 - 9 demonstrate that a separation is indeed possible. An assumption that waveforms are similar may add better conditioning for the wave-field separation. With more than one receiver station, the difference between source signals estimated from adjacent receiver stations could conceivably be attributed to either background noise, or to radiation patterns associated with the source mechanism.

Knowing the position of the drill bit can be used to further constraining of the solution and possibly adjusting the velocities or the rotation angles in order to optimally focus the receivers on the location of the drill bit. This may be possible even without direct knowledge of the source location, assuming that two subsequent tremors are located close to each other in space. Again, more receivers would allow more source parameters to be determined and with greater accuracy.

By using the method according to the invention, giving improved knowledge of the source signature of the working drill bit, one can either, like what is conventionally done with Vibroseis data, correlate these signatures into the recorded data, or use the semblance- weighted deconvolution method described by Haldorsen, Borland and Heijna (2004), ref. Haldorsen, J. B. U., Borland, W., and Heijna, H. B., 2004, "Using harmonics as signal for Extended High-Frequency Processing of Vibrator VSP Data", Submitted to the IPTC, joint AAPG, SEG, and EAGE technology meeting, November 21-2, 2005 in Doha, Qatar.

Claims

1. A method for determining the seismic signature of a drill bit acting as a seismic

source while drilling, comprising the following steps:

separating recorded waves by decomposing data into three components;

2. The method according to claim 1 , where recording of compressional and shear waves generated by the drill bit is performed by using arrays of two or more three- component sensors for improved estimates.

3. The method according to claim 1 or 2, where the data are decomposed into three

components by minimizing off-diagonal terms in a co variance matrix.

4. The method according to claim 1 or 2, where the time delay between fast and slow moving share waves are used for indicating degree of the anisotropy of the formation.

5. The method according to claim 1 or 2, where the time delay between compressional waves and fast and slow moving shear waves are used for indicating distance to the seismic source.

6. The method according to claim 1 or 2, where the position of the drill bit is used as an input for focusing the receivers on the exact location of the drill bit.