OFFSHORE RESERVOIR MONITORING SYSTEM AND METHOD FOR ITS OPERATION

Abstract

An offshore reservoir monitoring system (100) comprises a vertical array (110) with multiple seismic receivers (120) less than 10 m apart. During operation, the vertical array is deployed in a shallow borehole (111) in a seabed (10) away from noise at a seafloor (11). The dense spacing of receivers (120) ensures an adequate number of sensors (120) in the shallow borehole (111) and a spatial sampling rate appropriate for suppressing coherent noise in the shallow layers under the seafloor (11). Vertical arrays (110) can be added to the system (100) at any time.

Claims

1-13. (canceled)

14. An offshore reservoir monitoring system comprising: a vertical array with multiple seismic receivers configured for installation in a seabed; and a recording node for recording data from the multiple seismic receivers, wherein the spacing between adjacent seismic receivers is less than 10 m.

15. The system according to claim 14, wherein several vertical arrays are deployed in dedicated boreholes in the seabed.

16. The system according to claim 15, wherein the distance between vertical arrays is larger than the spacing between adjacent seismic receiver levels.

17. The system according to claim 14, wherein a seismic receiver comprises three orthogonal geophones.

18. The system according to claim 14, wherein a seismic receiver comprises a hydrophone.

19. The system according to claim 14, wherein the vertical array is a fully sealed system with fibre-optic seismic sensors.

20. A method for operating a system according to claim 14, comprising the steps of: installing the system; recording data from the seismic sensors on the recording node; performing an active survey at first predetermined intervals; monitoring microseismic events between active surveys; and harvesting data from the recording node at second predetermined intervals using an underwater vessel.

21. The method according to claim 20, wherein installing the system includes providing the borehole by a technique selected from a group consisting of flush drilling, percussion drilling and drilling with a drill bit.

22. The method according to claim 20, wherein recording data includes signal processing in the recording node.

23. The method according to claim 20, wherein recording data includes deriving a vector gradient of a wave field.

24. The method according to claim 20, wherein harvesting data includes retrieving the recording node from the seabed.

25. The method according to any claim 20, further comprising the step of sampling data for suppressing coherent noise.

26. The method according to any claim 20, further comprising the step of sampling data for separating up-going and down-going wave fields.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The invention will be explained with reference to exemplary embodiments and the accompanying drawings, in which:

[0041] FIG. 1 illustrates a vertical array according to the invention;

[0042] FIGS. 1a-c illustrate vector differences;

[0043] FIG. 2 illustrate a system with several vertical arrays;

[0044] FIG. 3 illustrates a field wide three-dimensional receiver array;

[0045] FIG. 4 illustrates a method according to the invention

DETAILED DESCRIPTION OF A PREFERRED EMBODIMENT

[0046] The drawings are schematic and not to scale. For ease of understanding, numerous details known to the skilled person are omitted from the drawings and following description.

[0047] The problem at hand may loosely be described as measuring a property u(x, t)=(u(t), v(t), w(t)) at discrete points (X, Y, Z, t) in space and time. A formal discussion of known representations of the seismic equations is beyond the scope of this disclosure. However, we note that the problem may be described as determining a boundary condition or final state, hereinafter a wavefield for short, sampled at discrete points (X, Y, Z, t) in space-time. Hence, the Nyquist-Shannon sampling theorem is central to ensure proper reconstruction of the full wavefield in spatial and temporal directions, and appears in some examples below.

[0048] The focal point mechanism describes the nature of a microseismic event, e.g. a dip slip, strike slip, dilation or a combination, and is typically derived from observation of first motions, e.g. whether the first arriving P wave breaks up or down. Alternatively, the moment tensor and slip vector at the focal point may be computed using digitally sampled wavefields and/or their transforms in the frequency-wavenumber (f-k) domain. Either way, the observations regard P and S-waves arriving at different spatial locations as known in the art.

[0049] FIG. 1a illustrates a system 100 according to the invention installed in and on a seabed 10 under a body of water 12. The system 100 comprises a vertical array 110 with a cable 112 connected to multiple seismic receivers 120, a connector 130 and a seabed recording node 140 on the seafloor 11, i.e. the interface between the seabed 10 and the body of water 12.

[0050] Specifically, the vertical array 110 is installed in a shallow borehole, e.g. extending 10 to 100 m downward from the seafloor 11. The seabed 10 may comprise sediments such as sand and clay above a layer of solid rock. In all circumstances, rotary flush drilling can be used to drill the hole for the sensor array. The sensor array is then installed in the hole and fixed in place using cement, grout or by forcing the collapse of the formation around the hole. In some instances, e.g. for a shorted sensor array 110 and lower sediment shear strengths, percussion or push drilling is a fast and cost-effective way to install the sensor array 110 in the seabed 10. In this case a specially designed sensor array is required. Deployment rates an order of magnitude better may be achieved. A small number of vertical arrays 110 may be installed around a well or platform for fraction of the cost of a traditional PRM system with associated infrastructure.

[0051] The design of the seismic receivers 120 depends on intended use. For example, it may be desirable to include the system 100 in a global or regional receiver array for earthquakes. In this case, most receivers 120 might comprise standard geophones configured to detect frequencies in the high frequency range 0.5-1000 Hz, i.e. signals with periods two seconds or less. The remaining receivers 120 might contain geophones designed to detect low frequencies associated with earthquakes, e.g. signals with periods 20 seconds or more. Moreover, the incoming signals may be band-pass filtered such that the signal is zero outside a specified range. Signals having a Fourier transform satisfy the Nyquist-Shannon criterion and may be completely restored if sampled above the Nyquist frequency. For example, signals in the range 0.5-1000 Hz may be restored completely in the time domain if sampled at a frequency above 2(10000.5) Hz2 kHz. Similarly, the spatial sampling rate required for wavefield reconstruction is less for long wave signals, so sparsely distributed low frequency geophones may still provide adequate sampling of the wavefield.

[0052] The sensors detecting particle motion may include inexpensive MEMS accelero-meters integrating over time to provide particle velocities in three spatial dimensions. Other geophones may rely on different known principles. Further, hydrophones are optional, but may provide valuable additional information on P-waves. Most hydrophones may be inexpensive electrostatic devices, and the remaining fraction, e.g. 10%, may be high-grade.

[0053] The design of receivers 120 also depends on the technique selected for deployment. For example, fibre optic sensors withstand the shocks involved in percussion drilling while fragile instruments with movable parts do not. As noted in the introduction, DAS sensors use the principle of Rayleigh backscattering, and are at present not sensitive enough for use in the system 100, although this could change in the future. However, fibre-optic sensors based on Michelson interferometry to measure phase changes are sensitive enough. The vertical array 110 could for example be a fully sealed system with distributed sensors separated by Fibre-Bragg gratings or semi-silvered surfaces. This embodiment would have no moving parts and be suitable for percussion installation. A potential disadvantage would be the increased power required to drive laser diodes for extracting data from these sensors.

[0054] From the previous paragraphs, it should be clear that the actual design of the receiver array 110 and choice of sensors in the receivers 120 must be left to the skilled person.

[0055] Grout, cement or concrete 111 in the borehole fixes the position of the array 110 and ensures proper acoustic contact with the seabed 10. The cable 112 connects the receivers 120 to the recording node 140 via the connector 130. The vertical separation or spacing of the receivers 120 is typically about 1 to 5 m, although spacing up to 10 m is possible. The short spacing ensures accurate spatial sampling of the wavefield. In comparison, the sensor spacing in a production well is typically tens of metres. Crossline spacing of hundreds of metres between seafloor nodes are common.

[0056] The vertical array 110 acts as an antenna to amplify the signal and directionally tune into signals from a particular part of the deeper subsurface. Specifically, the temporal phase shifts between adjacent receivers indicate a vertical angle, and the sum or average of the signals amplify the desired signals and suppress random noise thereby improving the SNR.

[0057] In contrast to land systems using one or two levels, the system 100 will be sampling the seismic events over much smaller intervals allowing the signal and noise wavefields to be characterised. In addition to making the sensor array much cheaper to manufacture, it will also have a major impact on the installation.

[0058] In FIG. 1a, each vertical array 110 is connected to one recording node 140 through a connection 130. The connection 130 allows disconnecting the node 140, e.g. for uploading data and maintenance at a mother vessel on the surface and/or replacing the node 140 with a freshly synchronised and fully charged recording node 140. The node 140 can last on the seabed for at least 1 year and is also able to transmit the data periodically to a receiving probe using an optical link enabling the data to be downloaded=from the node whenever desired.

[0059] In some embodiments, each recording node 140 may be connected to several vertical arrays 110 or other seabed sensors. Moreover, the signal processing may be divided between the receivers 120 and the recording node 140 in any suitable manner. For example, a typical receiver 120 convert an analogue sensor signal to a digital output, but receivers providing an analogue signal for an A/D-converter in the recording node 140 are not excluded. Similarly, processing to reduce storage requirements are typically performed in the recording node, but does not exclude signal processing in the receivers 120. For example, an accelerometer in the receiver 120 may present a time integrated and sampled signal that can be stored in the recording node 140 without further processing or after additional processing by the recording node 140. Either way, the signal processing is performed at the recording node 140 as opposed to in a central computer facility. Examples of other signal processing that may be performed at the recording node 140 include stacking, identifying a microseismic event in a continuous flow of data and low-pass filtering/anti-aliasing.

[0060] FIG. 1b-1d illustrate use of vector differences to estimate horizontal and vertical gradients. For convenience, bold face denotes vector variables throughout this disclosure. Specifically, FIG. 1b shows vectors A, B and C received at three different receivers 120. The vectors form slightly different angles with the vertical, and typically represent vector phases received at three different receivers 120, each receiver 120 comprising three orthogonal geophones. This representation may characterise any wavefield of interest. Alternatively, the vectors A, B and C may represent incoming rays, e.g. obtained by temporal phase shifts. The latter representation may be useful in the high frequency range, i.e. frequencies above 0.5 Hz as defined above.

[0061] Thus, in FIGS. 1c and 1d, the vector differential BA between the first two vertical receivers may represent some vector response representing a property parallel to an incoming wave front. Similarly, the differential CB between the next receiver pair gives the vector response to the next incoming wave front. Linear combinations of the differentials can be used to derive the vertical wavefield gradient as C2B+A and the horizontal wavefield gradient as CB+k(BA). If using a single sensor string proves too difficult or inaccurate to calculate the gradient of the wavefield, sensors from different vertical arrays that are placed near each other may be used.

[0062] In general, the vertical differential is easier to calculate than the horizontal but with an array of 6 or 7 three-component receivers, the statistics should be strong enough to obtain an accurate estimate for the scalar value k. The distance between the receivers would have to be precise but could be much less than a (seismic) wavelength apart. The orientation of the array would also have to be tightly controlled. In the more general case, it is assumed that the acoustic impedance of the sediments around the receivers changes as well. Essentially this means that a model of the near surface velocity must be built so that these corrections can be made. This information should be provided by direct arrival times or by using the myriad of seismic traces to model the near surface acoustic contrasts. However it does require sub-millisecond sampling.

[0063] In a different approach, consider a smooth function (x) and the Taylor expansions of (xx):

[00001]$\begin{array}{cc}\left(x+.Math..Math.x\right)=\left(x\right)+\frac{}{x}.Math..Math..Math.x+\frac{1}{2}.Math.\frac{{}^2.Math.}{x^2}.Math.{\left(.Math..Math.x\right)}^2+\frac{1}{6}.Math.\frac{{}^3.Math.}{x^3}.Math.{\left(.Math..Math.x\right)}^3+O\left({\left(.Math..Math.x\right)}^4\right).& \left(1\right)\\ \left(x-.Math..Math.x\right)=\left(x\right)-\frac{}{x}.Math..Math..Math.x+\frac{1}{2}.Math.\frac{{}^2.Math.}{x^2}.Math.{\left(.Math..Math.x\right)}^2-\frac{1}{6}.Math.\frac{{}^3.Math.}{x^3}.Math.{\left(.Math..Math.x\right)}^3+O\left({\left(.Math..Math.x\right)}^4\right)& \left(2\right)\end{array}$

where O((x).sup.4) means terms of order (x).sup.4 and higher. The requirement of a smooth function is for example satisfied by a (sum of) plane wave(s) and is not limiting in most applications. Incidentally, a plane wave has a Fourier transform and may be set to zero outside a limited frequency range, e.g. 0.5-1 kHz, as required by the Nyquist-Shannon sampling criterion.

[0064] Subtracting (2) from (1) and solving for

[00002]$\frac{}{x}$

yields:

[00003]$\begin{array}{cc}\frac{}{x}=\frac{1}{2.Math..Math..Math.x}\left[\left(x+.Math..Math.x\right)-\left(x-.Math..Math.x\right)\right]+\frac{1}{3}.Math.\frac{{}^3.Math.}{x^3}.Math.{\left(.Math..Math.x\right)}^2+O\left({\left(.Math..Math.x\right)}^4\right)& \left(3\right)\end{array}$

The leading term after the square brackets is proportional to (x).sup.2, and defines a maximum error in the approximation

[00004]$\begin{array}{cc}\frac{}{x}\frac{1}{2.Math..Math..Math.x}\left[\left(x+.Math..Math.x\right)-\left(x-.Math..Math.x\right)\right]& \left(4\right)\end{array}$

Thus, if x is small, (x).sup.2 is negligible and equation (4) is an acceptable approximation.

[0065] Similarly, adding (2) to (1) and solving for

[00005]$\frac{{}^2.Math.}{x^2}$

yields:

[00006]$\begin{array}{cc}\frac{{}^2.Math.}{x^2}\frac{1}{{\left(.Math..Math.x\right)}^2}\left[.Math.\left(x+.Math..Math.x\right)-2.Math.\left(x\right)+\left(x-.Math..Math.x\right)\right]& \left(5\right)\end{array}$

Equation (5) also has an error term of order (x).sup.2.

[0066] The derivation of equations (4) and (5) illustrate the benefits and limitations of centred approximation, which is widely used to estimate functions representing physical properties. Moreover, the variable x is arbitrary and may for example represent a spatial direction or time. Similarly, the function (x) may represent a particle displacement u(x, t) or its velocity du/dt. Plane waves are already mentioned. Yet another possibility is a convolution of body waves, i.e. a slowly varying function without the wiggles of individual P-waves or S-waves, rather than a wave front. Thus, if x is replaced with the spatial variable Z shown in FIGS. 2 and 3 and Z represent the distance between adjacent receivers 120 in a vertical array, it is easily seen that equations (4) and (5) provide accurate estimates for the first and second spatial and temporal derivatives of such convolutes.

[0067] The fine sampling prevents aliasing of data arriving at higher angles rather than ultra-high frequency waves travelling along the array.

[0068] Another feature of the array 100 is that, when a tight grid of shots is used, each receiver can be recalculated as a virtual source that does not depend on knowledge of the near surface. Although the number of traces generated will be relatively low, this could be used for tomographic imaging in the near surface that may be highly effective for characterising near-surface velocity changes (for example in palaeo-channels) and so help to building a better image at depth.

[0069] FIG. 2 illustrates a system 100 with several vertical arrays 110 extending from the seafloor 11. The distance between the vertical arrays 110 in a direction X along the seafloor 11 is substantially longer than the vertical spacing of receivers in the arrays 110.

[0070] Under conditions illustrated in a previous example, a virtual receiver may be assumed between each physical receiver or vertical array 110. More specifically, assume some function (x) at a virtual receiver and measured values (xx) at adjacent receivers. Then, (x) is the average:

(x)=[(x+x)+(xx)](6)

Estimates for spatial and temporal first derivatives may be obtained from equation (3). While substituting equation (6) into (5) always yield second derivatives of zero, centred approximation may still be useful e.g. for ray tracing algorithms.

[0071] FIG. 3 illustrates a full scale system 100 and a Cartesian coordinate system X, Y, Z in which several vertical arrays 110 are arranged in a row along the X-axis, a plurality of rows are arranged in the Y-direction and each vertical array 110 comprises multiple 4C-sensors 120 arranged in the Z-direction. There are no wired or wireless connections between the recording nodes 140, and the dash-dot lines just illustrate an imaginary grid on the seafloor 11.

[0072] Due to the noise reduction and the estimate of higher spatial frequencies, the distance between receiver stations is expected to be about the same as for deep water node surveys (approximately 400-500 m), so approximately 100 vertical arrays and 100 seabed nodes could cover a medium-to-large North Sea field covering 25 square kilometres.

[0073] An underwater vessel 300 of any known kind is able to travel through the water column between the nodes 140 and a mother vessel on the surface. Equipment 310 on the underwater vessel 300 is able to receive data from each recording node 140, e.g. over a fast short range optical link. The underwater vessel 300 is also able to disconnect a recording node 140 from an associated connector 130 and retrieve the node 140 to the surface for transferring data to a central computer system, maintenance and recalibration as known in the art.

[0074] FIG. 4 illustrates a method 400 for operating the system 100.

[0075] Step 410 includes all steps required prior to operations, e.g. installing the vertical arrays 110 in shallow boreholes in the seabed 10 as described previously. While not shown explicitly in FIG. 4, is should be understood that vertical arrays 110 may be added to the system during operation. This allows the system 100 to grow, for example from a pilot system similar to the embodiment in FIG. 2 to a full scale system as shown in FIG. 3.

[0076] Decision 420 determines when it is time to mobilise a source vessel and perform a monitoring survey, i.e. perform an active survey 430.

[0077] In step 430, airgun shots will be fired in an active survey to provide an active dataset for analysis and/or to calibrate the sensor array. During a survey 430, e.g. an instance in a 4D time lapse series, the vertical array(s) 110 continuously record the seismic wavefield in the subsurface. Such scheduled active surveys 430 confirm or adjust changes made to the ground model during a previous period of passive monitoring 440.

[0078] Step 440 illustrates that the system 100 performs passive monitoring between active surveys 430. The order of steps 430 and 440 is arbitrary in the sense that step 430 may be performed before passive monitoring 440 in a pilot installation, whereas a new vertical array 110 may be deployed anytime between active surveys 430 and hence start with step 440.

[0079] Steps 430 and 440 both include a step 401: Recording data. Hence, most or all seismic receivers 120 should have a dynamic range allowing for tiny passive seismic signals from natural and man-made microseismic events nearby in step 440 as well as more powerful reflections from an active source during active seismic acquisition in step 430. Recording 401 starts as soon as the node 140 is connected to the sensor array 110 and will continue until the battery runs out or the recording node 140 is replaced.

[0080] Decision 450 implements a schedule for when to send data to the surface.

[0081] Step 460 includes transmitting recorded data to the underwater vessel 300, e.g. over a high-speed optical link every few weeks or months. Step 460 also includes disconnecting the entire recording node 140 from the connector 130 at regular intervals, e.g. every 1-2 years, for uploading data to a central digital storage, maintenance, calibration and recharging or replacing batteries.

[0082] Decision 470 illustrates that the system 100 is designed to operate until the end of life for the field, e.g. as opposed to a 4D OBS array with recording nodes deployed on the seafloor for the duration of an active survey.

[0083] Step 480 includes any required cleanup and decommissioning work, e.g. removing a recording node 140 while leaving the vertical array(s) 110 in the seabed 10.

[0084] While the invention has been described by examples with reference to specific embodiments, the scope of the invention is defined by the following claims.