Determination of clusters stimulation efficiency using offset well monitoring with fiber optics and artificial intelligence surface data processing of the stimulated well

Abstract

Embodiments presented provide for determination of stimulation efficiency, In embodiments, offset well monitoring is performed with fiber optics and the data gains from the fiber optics are processed through an artificial intelligence system to calculate stimulation efficiency. A method is provided to calculate at least one value related to a hydraulic fracturing job being performed at a wellbore. The method may comprise measuring a formation strain rate at an offset wellbore. The method may further comprise obtaining pressure, rate, concentration data related to the wellbore. The method may further comprise calculating a most probable fracture parameter using a machine learning based algorithm. The method may further comprise recalculating the formation strain rate at the offset wellbore based upon a strain rate propagation model. The method may further comprise performing iterative calculations.

Description

DETERMINATION OF CLUSTERS STIMULATION EFFICIENCY USING OFFSET WELL MONITORING WITH FIBER OPTICS AND ARTIFICIAL INTELLIGENCE SURFACE DATA PROCESSING OF THE STIMULATED WELLCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of U.S. Provisional Application No. 63 / 678, 187, entitled "DETERMINATION OF CLUSTERS STIMULATION EFFICIENCY USING OFFSET WELL MONITORING WITH FIBER OPTICS AND ARTIFICIAL INTELLIGENCE SURFACE DATA PROCESSING OF THE STIMULATED WELL" filed August 01 , 2024, the disclosure of which is hereby incorporated herein by reference in its entirety.FIELD OF THE DISCLOSURE

[0002] Aspects of the disclosure relate to determining stimulation efficiency of a wellbore intervention activity. More specifically, aspects of the disclosure use artificial intelligence coupled with data from fiber optics to estimate stimulation efficiency through offset well monitoring.BACKGROUND

[0003] A non-uniform slurry distribution over perforation clusters causes lower production rates in hydrocarbon recovery wellbores. That causes interest in the evaluation of the non-uniform ity degree of clusters stimulation. To get well resolved data on clusters distributions; however, the measuring cable (fiber optic line) must be located behind the wellbore casing. This configuration is not always possible; and as a result for most wells, this service is not available. There is another opportunity to obtain data related to a wellbore, for cable located at the treated well, during fracturing process. This technology has several disadvantages first, the cable cannot operate for a long time and might be eroded or cut by the flowrate near the perforation within one or two stages; and second, the data received with the cable within a wellbore has a lower signal-to-noise ratio so that essential filtering may be required to receive slurry distribution over clusters.

[0004] Another application of distributed acoustic sensor use, hereinafter “DAS”, is for strain monitoring in an offset well. In these cases, low frequency DAS data at the cablelocated in an offset well are analyzed to determine the treated well fracture parameters, such as height, width, length, orientation, etc. This technology may be applied at multiple stages and on many wells without impact on cable integrity. The results provided give reasonable data on the fracture parameters. This technique generally requires advanced interpretation and is difficult to be implemented as an automatic service as the inversion results are not unique.

[0005] Artificial intelligence has allowed getting quantitative information about clusters stimulation uniformity based only on the surface data (pressure, rate, and concentration), wellbore completion, and deploying a trained neural network-based model. The training for the artificial intelligence network uses surface data on pressure rate and concentration to match coefficients in the neural network. This provides a quantitative prediction of the clusters stimulation non-uniform ity devised from the noise energy received with a cable behind casing DAS data.

[0006] Currently, there is no conventional method that uses low frequency, distributed, acoustic, sensing measurement at the offset well, surface data measured on the treated well, and a machine learning based model (for example, trained neural network-based model), for the semi-quantitative or even quantitative fractures growth determination.

[0007] There is a need to provide an apparatus and methods that are easier to operate than conventional apparatus and methods and that use low frequency, distributed, acoustic sensing measurements at an offset well.

[0008] There is a further need to provide apparatus and methods that do not have the drawbacks discussed above including specific arrangement configurations that are not present in all wells.

[0009] There is a still further need to reduce economic costs associated with operations and apparatus described above with conventional tools and overall economic costs associated with stimulation activities by being able to accurately characterize stimulation efficiency.SUMMARY

[0010] So that the manner in which the above recited features of the present disclosure can be understood in detail, a more particular description of the disclosure, briefly summarized below, may be had by reference to embodiments, some of which are illustrated in the drawings. It is to be noted that the drawings illustrate only typical embodiments of this disclosure and are therefore not to be considered limiting of its scope, for the disclosure may admit to other equally effective embodiments without specific recitation. Accordingly, the following summary provides just a few aspects of the description and should not be used to limit the described embodiments to a single concept.

[0011] In one example embodiment, a method is provided to calculate at least one value related to a hydraulic fracturing job being performed at a wellbore. The method may comprise measuring a formation strain rate at an offset wellbore. The method may further comprise obtaining pressure, rate, concentration data related to the wellbore. The method may further comprise calculating a most probable fracture parameter using a machine learning based algorithm. The method may further comprise recalculating the formation strain rate at the offset wellbore based upon a strain rate propagation model. The method may further comprise performing iterative calculations during the hydraulic fracturing job.

[0012] In another example embodiment, an article of manufacture is provided, having a non-volatile memory, a set of instructions encoded onto the memory that may be read and performed by a computer, the set of instructions comprising a method. The method recited by the article of manufacture further configured to recite calculating a value from the hydraulic fracturing job at the wellbore and feeding the computer model data obtained from measuring a formation strain rate of the wellbore and obtaining pressure rate concentration data related to the wellbore. The non-volatile memory configured to store a set of instructions executable on a computing arrangement, the set of instructions at least partially containing a method for calculating a most probable fracture parameter using a machine learning based algorithm. The set of instructions contained in the article ofmanufacture may comprise recalculating the formation strain rate at the offset wellbore based upon a strain rate propagation model. The set of instructions contained in the article of manufacture may comprise performing iterative calculations during the hydraulic fracturing job.BRIEF DESCRIPTION OF THE DRAWINGS

[0013] So that the manner in which the above recited features of the present disclosure can be understood in detail, a more particular description of the disclosure, briefly summarized above, may be had by reference to embodiments, some of which are illustrated in the drawings. It is to be noted; however, that the appended drawings illustrate only typical embodiments of this disclosure and are therefore not be considered limiting of its scope, for the disclosure may admit to other equally effective embodiments.

[0014] FIG. 1 shows cluster flowrates distribution and respective non-dimensional slurry distribution values.

[0015] FIG. 2 depicts graphs of non-dimensional slurry distributions.

[0016] FIG. 3 shows the strain rate parameters determination based on non-dimensional slurry distribution model constraints.

[0017] FIG. 4 illustrates a method in one example embodiment of the disclosure.

[0018] To facilitate understanding, identical reference numerals have been used, where possible, to designate identical elements that are common to the figures (“FIGS”). It is contemplated that elements disclosed in one embodiment may be beneficially utilized on other embodiments without specific recitation.DETAILED DESCRIPTION

[0019] In the following, reference is made to embodiments of the disclosure. It should be understood; however, that the disclosure is not limited to specific describedembodiments. Instead, any combination of the following features and elements, whether related to different embodiments or not, is contemplated to implement and practice the disclosure. Furthermore, although embodiments of the disclosure may achieve advantages over other possible solutions and / or over the prior art, whether or not a particular advantage is achieved by a given embodiment is not limiting of the disclosure. Thus, the following aspects, features, embodiments and advantages are merely illustrative and are not considered elements or limitations of the claims except where explicitly recited in a claim. Likewise, reference to “the disclosure” shall not be construed as a generalization of inventive subject matter disclosed herein and should not be considered to be an element or limitation of the claims except where explicitly recited in a claim.

[0020] Although the terms first, second, third, etc., may be used herein to describe various elements, components, regions, layers and / or sections, these elements, components, regions, layers and / or sections should not be limited by these terms. These terms may be only used to distinguish one element, components, region, layer or section from another region, layer or section. Terms such as “first”, “second” and other numerical terms, when used herein, do not imply a sequence or order unless clearly indicated by the context. Thus, a first element, component, region, layer or section discussed herein could be termed a second element, component, region, layer or section without departing from the teachings of the example embodiments.

[0021] When an element or layer is referred to as being “on,” “engaged to,” “connected to,” or “coupled to” another element or layer, it may be directly on, engaged, connected, coupled to the other element or layer, or interleaving elements or layers may be present. In contrast, when an element is referred to as being “directly on,” “directly engaged to,” “directly connected to,” or “directly coupled to” another element or layer, there may be no interleaving elements or layers present. Other words used to describe the relationship between elements should be interpreted in a like fashion. As used herein, the term “and / or” includes any and all combinations of one or more of the associated listed terms.

[0022] Some embodiments will now be described with reference to the figures. Like elements in the various figures will be referenced with like numbers for consistency. In the following description, numerous details are set forth to provide an understanding of various embodiments and / or features. It will be understood; however, by those skilled in the art, that some embodiments may be practiced without many of these details, and that numerous variations or modifications from the described embodiments are possible. As used herein, the terms “above” and “below”, “up” and “down”, “upper” and “lower”, “upwardly” and “downwardly”, and other like terms indicating relative positions above or below a given point are used in this description to more clearly describe certain embodiments.

[0023] Aspects of the current disclosure provide a combination of a machine learning based model for a stage uniformity prediction based on the surface data (pressure, rate, concentration) and a strain rate propagation model, based on the hydraulic fracturing process monitoring from an offset well with use of a fiberoptic cable.

[0024] The neural network-based model, predicting non-dimensional slurry distribution (hereinafter NSD) is trained on cable behind casing, clamped on tubing and / or in stimulated wellbore, fiberoptic data and surface data on pressure, rate, and concentration. Further in the inference mode, the model quantitatively predicts the stage stimulation uniformity using only surface data. The NSD model provides the quantitative distribution of a slurry placement along the stage. In aspects of the disclosure, the model outputs a slurry placement distribution, sorted from the highest rate interval (i.e., an interval, accepting most of slurry) to lowest rate interval (i.e., accepting lowest amount of slurry), without telling a user the intervals’ locations within the stage.

[0025] The strain rate model predicts the fracture geometry of the wellbore, corresponding to all the intervals along the stage based on the fiberoptic from an offset well. This is performed from the inverse problem solution, i.e., the model converts cable response data to the fracture geometries. As will be understood, the signal at an offsettypically suffers from poor resolution unless the fracture tips are sufficiently close to the offset well.

[0026] The combination of both methods allows solving the inverse problem (i.e., fractures geometry determination based on cable measurements) constrained with the output from the NSD model, decreasing the uncertainty level for the strain rate model results. This allows sorting the slurry distribution over intervals to the correct order (i.e., depth-wise) at least when the fracture tip approaches the monitoring well. In embodiments, the technology provides at least two deliverables: fractures distribution at any time moment and stimulation uniformity; and the stimulation uniformity is predicted with higher robustness than when using only data with machine learning algorithm.

[0027] The NSD model is a neural network-based model, trained on cable behind casing, clamped on tubing and / or in stimulated wellbore, fiberoptic data and surface data on pressure, rate, and concentration. The model may also be trained based on the cable inside the wellbore if the signal is properly filtered to allow distinguish signal from different clusters. The following approach may be used (but not limited by the given approach) to provide dimensionless physical data for the proper network training. In one example embodiment, instead of pressure being used for calculations, a ratio of pressure to the hydrostatic depth may be used. In other instances, instead of rate, the Reynolds number v2MD(v D), or the estimated friction over2phydrostatic depth can be used, where v is a flow velocity, MD is a measured depth, and D is a well diameter. Here the constants such as g = 9.8^, or clean density (1000 kg / m3), are omitted, and will be eliminated after scaling. The difference between the clean fluid density and the slurry will also be learned by the model during training process. The features may be generated in various ways providing similar results if there is enough data and the network is correctly trained.

[0028] In embodiments, the target functions might be represented in different ways. The easiest is a time series array, the nondimensional slurry distribution. This parameter is a non-uniform ity indicator, which is expressed through the individual clusters flow frates as:

[0029] Here n is the number of clusters, qtare the individual clusters flowrates, the value varies from 1 (fully uniform distribution) to zero (only one cluster is stimulated). Different distributions and the respective NSD values are shown in Fig. 1 . It is also seen that NSD by itself parameter doesn’t provide exact prediction of the flowrate’s distribution as different distributions may have the same NSD.

[0030] Example embodiments of the system have been successfully trained, validated, and tested on the separate data. The typical results are shown in FIG. 2. Here the NSD varies in the range 0.6 - 0.95 for most of the time, though with some sagging, which is qualitatively reproduced by the model.

[0031] In a similar way, the target may be represented by two-, three-, or more dimensional tensors, simulating clusters distribution. As the total flowrate is known, the two-dimensional distribution depends on one parameter only and is exactly determined by the NSD. The three-dimensional distribution contains two independent variables, so contrary to the NSD model, the multidimensional model (nd model) predicts flowrate distribution over n clusters more qualitatively. The true distribution of flowrate over n clusters in this case may be approximated with a three-dimensional one, so that an arbitrary number of clusters is quantitatively described by the three fractures; at the top, in the middle, and at the bottom of the stage. The reduction from the true numbers of clusters to (for example) three clusters is performed for the inverse model, speed increase, and uncertainty minimization. This will refer to the three effective fractures developing, which parameters will be determined from the strain rate model.

[0032] The use of optical fibers as distributed deformation sensors enables the monitoring and interpreting of hydraulic fracturing induced strain in a nearby monitoring well. A fiber optic cable is placed into a monitoring well, located near the well to be stimulated, and from which the hydraulically stimulated fracture network will be initiated. From the low-frequency strain signal received on the fiber optic cable, a user can trackthe low-frequency deformation induced by the main hydraulic fracture propagation and reactivation of secondary fractures, and pursue interpreting the latter for the 4D evolution of the hydraulically stimulated fracture system and its potential interference with nearby monitoring wells. A specific application for the industry’s focus is a control of frac hits occurring when hydraulic fractures either traverse or strongly interact with nearby monitoring wells. This phenomenon is frequent in United States land fields where well and fracture positioning are often very dense.

[0033] Mathematically, a quantitative procedure of inferring the strain sources, such as fractures, from the remote deformation data includes modeling the induced strain (a forward model), adjusting the model’s parameters to match the real data (field data matching or inversion), and characterizing the resulting solution ambiguity (uncertainty analysis). All these elements may become complicated when all the associated factors are considered. In modern fracturing, multiple clusters are being stimulated within a single stage, and, as a result, a propagating stack of fractures arises. Thus, the strain pattern observed in a monitoring well is a superposition of individual frac contributions (neglecting possible natural fault re-activation and other factors). One of the inversion uncertainty factors to resolve is differentiation between individual fractures within the stack. The nd model can provide an important constraint here by providing metrics of fracture stack volume breakdown. On the other hand, when individual fracture approaches the monitoring well, fracture geometry distribution may become more visible within strain rate patterns as individual fractures may give rise to their own extrema along the cable length.

[0034] In FIG. 3, the interdependence of the strain rate and the NSD model models is shown. Contrary to strain rate, the NSD model does not provide depth-ascribed data, the clusters distribution is sorted from the cluster with the biggest rate to the cluster with the smallest rate with no depth reference. For definitional purposes, such distributions will be described as “unsorted”. The strain rate model does have resolution as fractures which develop from neighbor clusters tend to not intersect each other. This resolution is very poor; however, when the fractures are far from an offset well, it increases in strain rate during fracture development, and clusters may be resolved. To make reasonableprediction possible for all the cases of fracture driven interactions, the (n) dimensional nd model results are combined at each time step with the strain rate model. At every time step, the nd model provides slurry rates among two, three, or more effective clusters, and the order of the clusters are sorted from the largest to smallest instantaneous slurry rates, but not according to their depths. These data present constraints on the fractures volume change. If the leak-off at all fractures is assumed to be the same, then the nd model deliverables apply constrains on the relative fractures volume change at given time moment:

[0035] If the additional data on the reservoir is available, the more complex, non-uniform leak off over the stages can be considered. These constraints, applied at every time step, significantly reduce ambiguity of the strain rate inverse problems solution, i.e., determination of the fractures parameters (length, height, and width) over time.

[0036] Simultaneously, using accumulated rates distribution between the clusters, this approach allows determining stimulation uniformity. Mutual consideration of nd model and strain rate inversion problems can work two-ways and facilitate both fracture geometry determination and stimulation uniformity; the stimulation uniformity result is more robust than when using only pumping data and machine learning algorithm.

[0037] The approach might be upscaled to more than one well monitored. In this case, data from a cable from one offset well collects the information from several wells stimulated consequently (zipper frac) or simultaneously (simul frac). Then, the inverse task is solved for more than one well fractures parameters determination using more than one constraint. For example, slurry rates distribution may be determined for all the wells separately using their own data and the same machine learning algorithm. The deliverable are fractures parameters for all stimulated wells and stimulation efficiencies. The technology maybe also upscaled in such way, that several offset wells are used, each of them with it its own cable for the monitoring of one or several wells.

[0038] Referring to FIG. 4, a method 400 of calculating cluster stimulation efficiency using offset well monitoring with fiber optics and artificial intelligence is illustrated. The method may comprise, at 402, measuring a formation strain rate of the wellbore. The method may further comprise, at 404, obtaining pressure rate concentration data related to the wellbore. The method may further comprise, at 406, calculating a most probable fracture parameter using a machine learning based algorithm. The method may further comprise, at 408, recalculating the formation strain rate for the wellbore based upon a strain rate inversion model. The method may further comprise, at 410, performing iterative calculations during the hydraulic fracturing job.

[0039] Example embodiments of the claims are described. The example embodiments should not be considered limiting. In one example embodiment, a method to calculate at least one value related to a hydraulic fracturing job being performed at a wellbore. The method may comprise measuring a formation strain rate at an offset wellbore. The method may further comprise obtaining pressure rate concentration data related to the treatment wellbore. The method may further comprise calculating a most probable fracture parameter using a machine learning based algorithm. The method may further comprise recalculating the formation strain rate for the wellbore based upon a strain rate propagation model. The method may further comprise performing iterative calculations during the hydraulic fracturing job.

[0040] In another example embodiment, the method may be performed wherein the measuring the formation strain rate includes measuring through the use of at least one fiber optic cable.

[0041] In another example embodiment, the method may be performed wherein the at least one fiber optic cable is placed in a monitoring well.

[0042] In another example embodiment, the method may be performed wherein at least two monitoring wells are used.

[0043] In another example embodiment, the method may be performed wherein the calculating the most probable fracture parameter includes sorting the fracture parameter according to volume.

[0044] In another example embodiment, the method may be performed wherein the pressure rate concentration data is used to predict width, length and height of the fracture.

[0045] In another example embodiment, the method may be performed wherein the performing iterative calculations during the hydraulic fracturing job continues until the end of the hydraulic fracturing, during a shut-in and / or flow-back.

[0046] In another example embodiment, the method may be performed wherein the recalculating the formation strain rate for the wellbore based upon the strain rate inversion model includes constraining a solution by fracture rates.

[0047] In another example embodiment, the method may be performed wherein the recalculating the formation strain rate for the wellbore based upon the strain rate inversion model precedes a step of transposing the sequence of fractures in order to obtain a best match with the strain rate.

[0048] In one example embodiment of the disclosure, an article of manufacture configured with a non-volatile memory, the non-volatile memory configured to store a set of instructions configured to be read and actions performed on a computer is disclosed. The set of instructions may comprise a method to calculate at least one value related to a hydraulic fracturing job being performed at a wellbore, comprising measuring a formation strain rate of the wellbore. The method may further comprise obtaining pressure rate concentration data related to the wellbore. The method may further comprise calculating a most probable fracture parameter using a machine learning based algorithm. The method may further comprise recalculating the formation strain rate forthe wellbore based upon a strain rate inversion model. The method may further comprise performing iterative calculations during the hydraulic fracturing job.

[0049] In another example embodiment, the method performed by the article of manufacture may further comprise calculating the most probable fracture parameter includes sorting the fracture parameter according to volume.

[0050] In another example embodiment, the method performed by the article of manufacture may be performed wherein the method contained in the non-volatile memory may be performed wherein the recalculating the formation strain rate for the wellbore based upon the strain rate inversion model precedes a step of transposing the sequence of fractures in order to obtain a best match with the strain rate.

[0051] In another example embodiment, the method performed by the article of manufacture may be performed wherein the article of manufacture is a universal serial bus device.

[0052] In another example embodiment, the method performed by the article of manufacture may be performed wherein the article of manufacture is configured to be run upon at least one of a cloud computing device, a personal computer, and a web-enabled computing arrangement.

[0053] The foregoing description of the embodiments has been provided for purposes of illustration and description. It is not intended to be exhaustive or to limit the disclosure. Individual elements or features of a particular embodiment are generally not limited to that particular embodiment, but, where applicable, are interchangeable and can be used in a selected embodiment, even if not specifically shown or described. The same may be varied in many ways. Such variations are not to be regarded as a departure from the disclosure, and all such modifications are intended to be included within the scope of the disclosure.

[0054] While embodiments have been described herein, those skilled in the art, having benefit of this disclosure, will appreciate that other embodiments are envisioned that do not depart from the inventive scope. Accordingly, the scope of the present claims or any subsequent claims shall not be unduly limited by the description of the embodiments described herein.

Claims

CLAIMSWhat is claimed is:1 . A method to calculate at least one value related to a hydraulic fracturing job being performed at a wellbore, comprising: measuring a formation strain rate at an offset wellbore; obtaining pressure rate concentration data related to the treatment wellbore; calculating a most probable fracture parameter using a machine learning based algorithm; recalculating the formation strain rate for the wellbore based upon a strain rate propagation model associated with the most probable fracture parameter from the machine learning based algorithm; and performing iterative calculations during the hydraulic fracturing job to match the calculated formation strain rate with the measured formation strain rate.

2. The method according to claim 1 , wherein the iterative calculations are also performed during a shut-in and / or a flow-back to match the calculated formation strain rate with the measured formation strain rate.

3. The method according to claim 1 , wherein the measuring the formation strain rate includes measuring through the use of at least one fiber optic cable.

4. The method according to claim 2, wherein the at least one fiber optic cable is placed in a monitoring well.

5. The method according to claim 2, wherein the monitoring well is at least two monitoring wells.

6. The method according to claim 1 , wherein the calculating the most probable fracture parameter includes sorting the fracture parameter according to volume.

7. The method according to claim 1 wherein the pressure rate, and slurry concentration data is used to predict a width, a length and a height of the fracture8. The method according to claim 1 , wherein the performing iterative calculations during the hydraulic fracturing job continues until the end of the hydraulic fracturing job, during a shut-in and / or a flow-back.

9. The method according to claim 1 , wherein the recalculating the formation strain rate for the wellbore based upon the strain rate inversion model includes constraining a solution by fracture rates.

10. The method according to claim 1 , wherein the recalculating the formation strain rate for the wellbore based upon the strain rate inversion model precedes a step of transposing the sequence of fractures in order to obtain a best match with the strain rate.11 .An article of manufacture configured with a non-volatile memory, the non-volatile memory configured to store a set of instructions configured to be read and actions performed on a computer, the set of instructions comprising a method to calculate at least one value related to a hydraulic fracturing job being performed at a wellbore, comprising: measuring a formation strain rate at an offset wellbore; obtaining pressure rate concentration data related to the treatment wellbore; calculating a most probable fracture parameter using a machine learning based algorithm; recalculating the formation strain rate for the wellbore based upon a strain rate propagation model associated with the most probable fracture parameter from the machine learning based algorithm; andperforming iterative calculations during the hydraulic fracturing job to match the calculated formation strain rate with the measured formation strain rate.

12. The article of manufacture according to claim 11 , wherein the iterative calculations are also performed during a shut-in and / or a flow-back to match the calculated formation strain rate with the measured formation strain rate.

13. The article of manufacture according to claim 11 , wherein the method contained in the non-volatile memory further comprises calculating the most probable fracture parameter includes sorting the fracture parameter according to volume.

14. The article of manufacture according to claim 11 , wherein the method contained in the non-volatile memory may be performed wherein the recalculating the formation strain rate for the wellbore based upon the strain rate inversion model precedes a step of transposing the sequence of fractures in order to obtain a best match with the strain rate.

15. The article of manufacture according to claim 11 , wherein the article of manufacture is a universal serial bus device.

16. The article of manufacture according to claim 11 , wherein the article of manufacture is configured to be run upon at least one of a cloud computing device, a personal computer, and a web-enabled computing arrangement.

17. The article of manufacture according to claim 11 , wherein the machine learning algorithm is trained prior to use.

18. The article of manufacture according to claim 17, wherein the training is performed using synthetic data.

19. The article of manufacture according to claim 17, wherein the training is performed on actual wellbore data.

20. The article of manufacture according to claim 17, wherein the machine learning algorithm has multiple nodal layers.

Images