MONITORING MICROSEISMIC EVENTS

Abstract

A system (100) for monitoring a subterranean structure comprises an array (10) with n acoustic sensors capable of detecting P-waves and/or S-waves from the subterranean structure and a central controller (120) for receiving a signal (X) from the sensors. The system further comprises a lookup table (20) comprising a pre-computed travel time curve (24) expressed as relative arrival times of a signal from a location (L.sub.m) to each of the sensors (1−n); a comparison unit for comparing the received signal (X) with the pre-computed travel time curve (24), and means for raising an alarm if the received signal (X) matches the precomputed travel time curve (24). Preferably, the alarm is raised if a computed semblance value (26, 27) exceeds a predefined threshold. The system may monitor several locations (L.sub.m) in parallel using a fraction of the computer resources and time required by prior art techniques.

Claims

1-14, (canceled)

15. A system for monitoring microseismic events in a subterranean structure, the system comprising: an array with a plurality of acoustic sensors capable of detecting P-waves and/or S-waves from the subterranean structure; a central controller for receiving a signal from the sensors; a lookup table comprising a pre-computed travel time curve expressed as relative arrival times of a signal from a predetermined location to each sensor in the plurality of sensors; a comparison unit for comparing the received signal with the pre-computed travel time curve; and means for raising an alarm if the received signal matches the pre-computed travel time curve.

16. The system according to claim 15, comprising several locations.

17. The system according to claim 15, wherein the comparison unit computes a semblance for each location and compares the semblance to a predefined threshold value for the location.

18. The system according to claim 16, wherein the comparison unit computes a semblance for each location and compares the semblance to a predefined threshold value for the location.

19. The system according to claim 17, wherein the threshold value is based on historical data and semblance values.

20. The system according to claim 17, wherein the semblance is computed at predefined intervals.

21. The system according to claim 15, further comprising means for further action, wherein the means are responsive to a raised alarm.

22. The system according to claim 21, wherein the means for further action includes means for further analysis.

23. The system according to claim 21, wherein the means for further action includes means for stopping an injection.

24. A method for monitoring microseismic events in a subterranean structure comprising the steps of: selecting a discrete number of locations in the subterranean structure; computing a travel time curve for each discrete location by estimating travel times from the location to each sensor in an array with n acoustic sensors and subtracting a fixed value from every travel time; comparing the travel time curve with a continuous signal received by the sensors; and raising an alarm if the received signal matches the pre-computed travel time curve.

25. The method according to claim 24, wherein the comparison includes computing a semblance for each location and comparing the semblance to a predefined threshold for the location.

26. The method according to claim 25, wherein the semblance is computed using a fixed window length around the pre-computed travel time curve.

27. The method according to claim 24, wherein the travel times and signal regards P-waves.

28. The method according to claim 24, wherein the travel times and signal regards S-waves.

29. The method according to claim 24, wherein the alarm is assigned a severity level depending on one or more factors selected from the group comprising noise level, signal-to-noise-ratio, the number of indications detected by independent sensors and the number of indications detected by independent arrays.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The invention will be described in greater detail by means of an exemplary embodiment with reference to the accompanying drawings, in which:

[0028] FIG. 1 illustrates a structure with discrete regions of interest and a sensor array;

[0029] FIG. 2 illustrates a system layout;

[0030] FIG. 3 illustrates a pre-computed travel time curve as a function of sensor number;

[0031] FIG. 4 illustrates comparing a incoming signal with the pre-computed curve; and

[0032] FIG. 5 is a flow chart illustrating a method according to the invention.

DETAILED DESCRIPTION

[0033] The drawings are schematic and simplified to illustrate the invention. Thus, they are not to scale, and numerous details are omitted for clarity.

[0034] An aim of the present invention is to provide a monitoring system for early warning of a seismic event occurrence in a region of particular interest. In the following, the system is described in the context of injecting waste into a subterranean structure offshore. However, the invention is not limited to that particular application.

[0035] Such a monitoring system can be used in so called passive and active mode. In active mode a seismic source is deployed on the surface and fired in a regular grid in order to image the subsurface by recorded reflections. In passive mode the system records data 24/7 comprising noise and signal stemming from the water column, installations on the surface as well as the subsurface. Methods for processing and interpreting in the subsurface are considered. In particular, the invention outlines a work-flow and procedure in order to provide a first look real-time processing scheme getting monitoring results in real-time, e.g. for control of an injection process. Such injection can cause intended and unintended fracturing of rock formations. As noted in the introduction, Unwanted or unplanned fracturing may open flow paths for fluids to surface or to highly undesirable injection outside the intended zone, and thereby create a situation with negative impact on production and/or significant harm to the environment. Thus, the system and method should provide ample information in order to detect unwanted fracturing.

[0036] In order to provide a useful tool providing information in real-time, a first indication/warning of a significant fracture event is provided so that a detailed data-analysis may be performed and/or some other action can be taken, e.g. stopping injection pending further analysis.

[0037] Fracturing of rocks leads to so called microseismic events, detectable by acoustic sensors. The principles for measurements of microseismic events are well known and are exploited commercially in, for example, shale-gas or oil exploration as noted above. Therefore, these principles are only referred to herein to the extent they are relevant for the present invention.

[0038] In the present example of injection into a reservoir, existing reservoir models are used to predict the confinement boundaries for the subterranean structure or reservoir. The monitoring objectives are related to detection if these predicted confinement boundaries are breached or passed. Particular risk thresholds or intervals can be defined for detailed monitoring in order to provide warning indications. In most cases these boundaries or intervals are limited in extent.

[0039] So, in case of fracturing into a risk zone, the fractures generate sound waves traveling to surface. In most cases applying injection, an earth-model for the subsurface being affected is available. This earth-model includes information of layering and the speed of P-waves and S-waves in each layer. In the following, the existence of such a model is assumed.

[0040] The model can also be used to predict the travel time for waves going from the fracture point to each sensor using, for example, ray trace modeling.

[0041] FIG. 1 illustrates a subterranean structure with an array 10 of seismic sensor modules 110 at a seafloor. The sensors 110 are well known acoustic devices capable of detecting P-waves and/or S-waves, and are not discussed in greater detail herein. Every seismic sensor module 110 is assigned a unique number, here represented by a channel number k in the set of integers 1−n. If desired, the array 10 can comprise subterranean sensors, e.g. sensors deployed in wellbores.

[0042] In the following description and claims, the term ‘location’ denotes a region of particular interest within the subterranean structure, for example a developing crack or a fault. FIG. 1 shows examples of two such locations, L.sub.l and L.sub.m, respectively.

[0043] The locations L.sub.l and L.sub.m are two of a small number of regions of particular interest. In other words, at least one location L.sub.m is assumed in the present invention. Further, each location L.sub.m is determined from a geophysical model of the subterranean structure. The model is known in advance, and is based on surveys, logs, core samples and any other data by methods known in the art.

[0044] Denoting a travel time from location L.sub.m to sensor number k as t.sub.mk, all travel times from location L.sub.m to every sensor number k are computed from the model. For example, all travel times t.sub.11, t.sub.12, . . . t.sub.1n from L.sub.l to the sensors 1−n are computed. For clarity, only t.sub.11 to sensor 1 and t.sub.1n to sensor n are shown in FIG. 1.

[0045] Similarly, travel times t.sub.mk from location L.sub.m to all channels k, here all sensors 110, are computed from the model, but only travel times t.sub.ml, t.sub.mk and t.sub.mn are shown in FIG. 1. More specifically, these travel times are mean travel times from a central point 21 within location L.sub.m. The dotted lines from point 22 illustrate that all travel times t vary within a time window w.sub.mk (FIG. 3) due to the extent of the location L.sub.m or uncertainty in the model or both. In some applications, the sensors 110 depicted in FIG. 1 can be a subset of all sensors that are actually available.

[0046] FIG. 2 illustrates a system 100 for monitoring a subterranean structure. The main parts of the system 100 are an array 10 of sensor modules 110 connected through an umbilical 101 to a central controller 120. The system 100 is deployed to monitor a subterranean structure as waste from an installation 201 is injected into the structure through a well 203. The installation 201 can be a platform or a subsea facility at the seafloor. The central controller 120 is depicted within the installation 201 for simplicity, and can be located elsewhere. Accordingly, the umbilical 101 is not necessarily connected to the installation 201.

[0047] The umbilical 101 supplies power to the sensor modules 110 and convey signals between the array 10 and the central controller 120 through a control unit 102 at the seafloor. A communication line 103 connects the sensor modules 110 to the control unit 102. More particularly, the array 10 comprises two symmetric branches of sensor modules 110 connected to the communication line 104. The branches are connected at their distal ends by an inline connector 104 and at their proximal ends to separate connectors in the control unit 102. The sensor modules 110 are enumerated 1−n as in FIG. 1. It is understood that each sensor module 110 may contain, for example, a hydrophone providing a P-wave signal and a geophone providing an S-wave signal. Accordingly, each channel 1−n may or may not represent a signal from exactly one sensor. However, mapping sensors to channels is trivial.

[0048] FIG. 3 illustrates computed travel time.sub.mk as a function of sensor number for a location L.sub.m, i.e. a region that has been determined to be of particular interest as explained above. For example P- or S-wave will arrive at sensor k from source m at time t.sub.mk. The relative arrival times at the sensors, i.e. the shape of a travel time curve 24, may indicate whether a P-wave or S-wave from location L.sub.m is received by the array of sensors 1−n. The travel time curve 24 is an imaginary curve through the dataset recorded by a set of selected sensors 1−n. When shape of P- or S-wave arrival times from event m at all selected nodes coincides with the shape of the computed travel time curve for event m, an alarm should be raised. A window around travel times can be one fixed value independent of source and sensor type. However, setting different windows for different sources and/or different sensors is/are possible according to the invention.

[0049] The resulting travel time curve 24 is stored in a lookup table 20 associated with the central controller 120 for future use. The lookup table 20 can, for example, be implemented as a mask in a buffer, either in a memory in a general computer or in dedicated hardware as further explained below.

[0050] As noted above, the system 100 may also be operated in an active mode where an active source provides a signal at known points in space and time. This mode may require a source vessel as described in the introduction, and can, for example, be used to update or verify the model, calibrate the sensor array, etc. Also as indicated above, the active mode is less suited for continuous monitoring than the passive mode. So far, the travel times and associated travel time curves are computed from the established model. FIG. 4 illustrates a physical signal X(k, i) arriving at the array 10 of sensors 110. The sensor numbers 1−n in FIG. 3 is replaced with channel numbers k ∈ {1−n} as explained above. The signal on each channel k is sampled to provide data points indexed with integer i. Thus, integer i indexes the signal on each channel k along a time axis t. W is an integer representing the number of sampled data points in a period similar to the window w.sub.mk in FIG. 3. However, W is preferably the same integer for all channels 1−n as indicated by the boundaries drawn on both sides of the curve X(k, i).

[0051] The signal received through the channels 1−n most likely comprises noise. Noise can be removed using any known technique, e.g. tau-p transform. The output from such processing would typically appear as small oscillations around a zero value with high amplitude wavelets from an incoming signal at the respective sensors as shown in FIG. 4. In particular, an event in location L.sub.m would be expected to arrive at the sensors 1−n at relative arrival times matching the pre-computed travel time curve 24, cf. FIG. 3. As shown in FIG. 4, the signal wavelets arrive at the respective sensors, i.e. in the channels 1−n at times relative to the arrival time 25 in channel 1. Here, a sum of the wavelets would be significantly greater than a sum of noise reduced signals. More particularly, ‘the sum of wavelets’ could be a sum of positive signals or a sum of squared signals, i.e. energy, received in the intervals W displaced relative to each other according to the pre-computed curve 24 in FIG. 3.

[0052] In a preferred embodiment, a comparison unit computes a semblance sum over the channels 1−n along the pre-computed travel time curve 24 at regular intervals. If the semblance sum is computed over channels 1−n carrying noise, a value below the threshold 26 is obtained. If a signal from location L.sub.m is present, the semblance will exceed the threshold value 26 significantly, as illustrated by the peak 27, under certain known conditions. It is understood that the comparison unit corresponds to the final step in a traditional semblance analysis, hence the name. Furthermore, such a peak 27 appears even if the individual ‘wavelets’ are below the noise level in all channels 1−n, as known from semblance theory. Thus, as used herein, the semblance is expressed as:

[00001]$S_c=\frac{\underset{i=j-W/2}{\overset{j+W/2}{.Math.}}.Math.{\left[\underset{k=1}{\overset{n}{.Math.}}.Math.X_{k,i}\right]}^2}{\frac{1}{n}.Math.\underset{i=j-W/2}{\overset{j+W/2}{.Math.}}.Math.\underset{k=1}{\overset{n}{.Math.}}.Math.{\left(X_{k,i}\right)}^2}$

where the signal X, indices i, k and fixed integers n and W are as explained above. The sampled signal X(k, i) received through the channels 1−n can represent a P-wave or an S-wave as noted above. If desired, the semblance S may be normalized, i.e. multiplied by a fixed factor.

[0053] The semblance curve 26, 27 may be calculated for a defined number of locations or focal points, as well as for both P (pressure wave) and S (shear wave) arrivals.

[0054] The required operations can be done very efficiently using well known computational algorithms. For example, the sampled data points can be obtained by a single AND operation between a first-in, first-out buffer containing the signal X(k, i) in n parallel columns and a fixed buffer containing a mask for the pre-computed curve 24. These buffers can be implemented in a generalized computer or in dedicated hardware associated with the central controller 120 (FIG. 2). The remaining additions and multiplications to achieve the semblance are also cheap operations in terms of time and computer resources.

[0055] Once the semblance for a location L.sub.m is computed, a simple comparison with a predetermined threshold value suffices to determine whether a micro seismic event may have occurred in location L.sub.m, and if so, raise an alarm and/or perform some predetermined action such as obtaining further data from injection pumps and any other available data source, stopping an injection of waste pending further analysis etc.

[0056] In contrast, traditional methods typically involve suppressing noise, running STA/LTA detector, picking seismic phases, determining the location and nature of the event to see if an alarm should be raised. These methods may involve moment tensor inversion and other operations demanding large computer resources and/or time. Compared to such methods, the present invention uses a well defined model to identify one or more location(s) of interest, disregarding any other region. Further, several semblances can be computed in series or parallel requiring a fraction of the time and/or computing resources needed for prior art methods. Hence, several locations can be monitored simultaneously and in real-time with the system and method of the invention, at the possible cost of an ability to detect and analyse an event in any region in a monitored structure.

[0057] FIG. 5 illustrates the main steps of a method 500 according to the invention. Start procedures 501 include obtaining an earth model, deploying sensor modules 10 and other initialisation. In the first step 510 of the method according to the invention, one or a few locations L.sub.m are selected. The selection is based on an existing model of the structure to be monitored.

[0058] In step 520, a travel time curve is computed as described above; one curve per location L.sub.m.

[0059] Branching 530 illustrates a waiting loop. When the system is out of operation, or perhaps in an active mode, there will be no incoming signal and a system implementing the method will wait for an incoming signal. During operation, there will be a continuous signal X from the sensor modules 110 (FIGS. 1 and 2) through n channels k (FIG. 4).

[0060] In step 540, possible arrivals in incoming signal X are compared to the travel time curve(s). This step may involve noise reduction by known techniques. In a preferred embodiment, this step includes computing the semblance S as discussed above.

[0061] If arrival times of P- or/and S-waves match the pre-computed travel time curve 24, the method proceeds to step 560. If there is no match, the method returns to check for incoming signal, and then compare a next window of the signal X with the curve 24. Thus, step 540 is routinely performed at regular intervals, e.g. with a period of W/2. The period W/2 amounts to continuous monitoring, as is readily seen from FIG. 4.

[0062] In step 560 an alarm is raised. This happens whenever the signal X matches the pre-computed curve(s) 24, e.g. if the calculated semblance value exceeds a threshold value as shown at peak 27 in FIG. 4 and explained above.

[0063] The monitoring loop represented by steps 530-550 in the method 500 may comprise additional steps for checking more than one location. Alternatively, several such monitoring loops 530-550 may be run in parallel in hardware or software, for example one process per location. The details of implementation are considered obvious to those skilled in the art.

[0064] The end procedures 570 may include further actions based on the alarm, for example acquiring further data or stopping injection pending further analysis. For example, available information such as well-head and downhole pressures or pump volumes may provide key information in order to perform preventive actions.

[0065] The alarm can be assigned a value for severity or reliability. For example, the reliability or severity of an alarm would increase if an event was indicated by most or all sensors in an array, or by two arrays for P-waves and S-waves respectively. The noise level, either in absolute terms or as a signal-to-noise-ratio, also impact such a reliability attribute. In particular, if the microseismic event is detected with a high level of background noise, the uncertainty increases and the reliability decreases.

[0066] While the invention has been described above with reference to examples and certain embodiments, the scope of the invention is determined by the accompanying claims.