Integrated mobile system for stratigraphic rock analysis

Abstract

A method for classifying debris particles includes acquiring, using a mobile device, a digital image of rock particles generated while drilling a subterranean wellbore. The digital image may be processed to generate a segmented image of individual rock particles identifying the rock particles and extract a plurality of geometric properties from selected ones of the identified rock particles depicted in the segmented image. The plurality of geometric properties includes a size feature and a shape feature. At least one of the size features may be evaluated to identify caving particles among the identified rock particles. A spalter particle shape classification may be generated by evaluating a plurality of shape features to mark each of the identified spalter particles with a corresponding spalter shape.

Description

[0001] Cross - reference to related applications

[0002] This application claims the benefit of European Application No. 22306870.1, filed on December 14, 2022, entitled "INTEGRATED MOBILE SYSTEM FORFORMATION ROCK ANALYSIS", the disclosure of which is hereby incorporated herein by reference. Background of the Invention

[0003] Rock particles are generated during drilling operations in oil and gas exploration and production, geothermal, and scientific exploration. These rock particles are typically classified as cuttings or cavings. Cuttings particles are particles directly generated by the cutting action of the drill bit when the drill bit breaks the formation rock. Cavings particles are those particles that are not directly generated by the cutting action of the drill bit and are typically particles spalled from unstable portions of the wellbore.

[0004] It will be appreciated that the rock particles generated during drilling (e.g., cuttings and cavings particles) are abundant both in volume and quantity and can provide one of the least - costly and most - abundant data sources for understanding and characterizing subsurface rock and formation properties. For example, the shape of cavings particles is sometimes used as an indicator of the formation rock fracture mechanism. Additionally, the size and shape of cavings particles can be used to update the mechanical earth model (MEM) of the subsurface formation, including formation stress, pore pressure, and formation strength. In some formation types, large cavings particles may be indicators of abnormal pressure zones or wellbore enlargement that can lead to drilling fluid losses (circulation losses), drill pipe sticking, or even partial collapse of the wellbore.

[0005] There is a need in the industry for improved methods for evaluating cavings particles, and in particular for methods that automate or partially automate the measurement process and thus reduce labor costs and shorten the turnaround time for interpretation. Brief Description of the Drawings

[0006] To more fully understand the disclosed subject matter and its advantages, reference is now made to the following description taken in conjunction with the accompanying drawings, in which:

[0007] Figure 1 An example rig is depicted that includes an example mobile system for evaluating rock particles obtained during a drilling operation.

[0008] Figure 2A And Figure 2B (collectively referred to as Figure 2) depicts a block diagram of an example mobile system for evaluating cavings particles.

[0009] Figures 3A - 3F(Collectively referred to as FIG. 3) depicts the operation of a mobile App employed by the mobile architecture shown in FIG. 2.

[0010] Figure 4 Depicts the geometric relationship between the field of view, viewing angle, camera focal length, and height of a camera sensor.

[0011] Figure 5 Depicts a flowchart of an example method for evaluating debris particles.

[0012] Figure 6 Depicts an example method for acquiring digital images of debris particles using a mobile device.

[0013] Figure 7 Depicts an example shape classification for debris particles. DETAILED DESCRIPTION

[0014] Embodiments of the present disclosure include systems and methods for evaluating debris particles. An example method includes acquiring a digital image of rock particles generated during the drilling of a subterranean wellbore using a mobile device. The digital image can be processed to generate a segmented image identifying individual rock particles of the rock particles, and multiple geometric properties can be extracted from selected identified rock particles depicted in the segmented image. The multiple geometric properties include size features and shape features. At least one of the size features can be evaluated to identify debris particles among the identified rock particles. A debris particle shape classification can be generated by evaluating the multiple shape features in order to label each of the identified debris particles with a corresponding debris shape.

[0015] Figure 1 Depicts an example rig 20 deployed at a well site, which includes a mobile system 100 for evaluating debris particles removed from circulating drilling fluid. The rig 20 can be positioned above a subterranean formation (not shown). The rig 20 can include, for example, a derrick and hoisting equipment (also not shown) for raising and lowering a drill string 30, which extends into a wellbore 40 and includes, for example, a drill bit 32 and one or more downhole measurement tools 50 (e.g., logging-while-drilling tools and / or measurement-while-drilling tools).

[0016] The rig 20 further includes a surface system 80 for controlling the flow of drilling fluid used on the rig (e.g., when drilling a wellbore 40). In the depicted example rig, the drilling fluid 35 is pumped downhole (as depicted at 92) via a mud pump 82. The drilling fluid 35 may be pumped, for example, through a riser 83 and a mud hose 84 on its way to the drill string 30. The drilling fluid typically exits the drill string 30 at or near the drill bit 32 and creates an upward flow 94 of mud through the wellbore annulus (the annular space between the drill string and the wellbore wall). The drilling fluid then flows through a return conduit 88 and a solids control device 85 (such as a shale shaker) to a mud pit 81. Drill cuttings generated during drilling and caving debris particles spalled from the surface of the wellbore may be transported to the surface in the upward flow 94 of the drilling fluid and may be removed from the fluid at the solids control device 85. It will be understood that the terms drilling fluid and mud are used synonymously herein.

[0017] The well site may include a mobile system 100 configured to evaluate images of rock particles, e.g., to determine the size and / or shape of the particles, as described in more detail herein. The system 100 may include a mobile device that includes a digital camera (such as a cell phone or a tablet computer) located at the rig site and communicates with other computer processing devices deployed at the rig site and / or offsite. The disclosed embodiments are not limited in this regard. The system 100 may include computer hardware and software configured to automatically or semi-automatically evaluate images of rock particles. To perform these functions, the hardware may include one or more processors (e.g., microprocessors), which may be connected to one or more data storage devices (e.g., a hard disk drive or solid state memory). As is known to those of ordinary skill in the art, the processor may further be connected to a network, e.g., to receive images from a networked camera system (not shown) or another computer system. Of course, it will be understood that the disclosed embodiments are not limited to the use or configuration of any particular computer hardware and / or software.

[0018] Although Figure 1 an onshore rig 20 is depicted, it will be understood that the disclosed embodiments are equally applicable to onshore or offshore rigs. As is known to those of ordinary skill in the art, an offshore rig typically includes a platform deployed at the top of a riser extending from the seabed to the water surface. The drill string extends downward from the platform, through the riser, and into the wellbore through a blowout preventer (BOP) located on the seabed. The disclosed embodiments are not limited in these respects.

[0019] Figure 2A and Figure 2B (collectively referred to as FIG. 2) depict example block diagrams of the system 100 and the system 100', including a mobile architecture for evaluating rock particles obtained during a drilling operation. As Figure 2AAs shown, system 100 includes a mobile device 110 (such as a smartphone or tablet) that communicates mobilely or wirelessly with a backend processing module 120 (e.g., via the Hypertext Transfer Protocol HTTP or via a standard local area network wireless connection). The mobile device 110 includes a digital camera that can be used by, for example, a mud logging engineer, a geoscientist, or other rig personnel to capture images of rock particles (e.g., cuttings and / or debris) removed from the circulating drilling fluid. The mobile device 110 further includes a software application (App) configured to run on the mobile device and guide the user through the process of capturing and analyzing digital images (described in more detail below). For example, the App can be configured to send the captured images to the backend module 120 for evaluation and data storage, and to receive and display the evaluation results from the backend module 120.

[0020] Continuing to refer Figure 2A , the backend module 120 can include a Representational State Transfer (REST) module 122 for interfacing with the mobile device 110 via the HTTP protocol (e.g., for receiving uploaded images from the mobile device and downloading processed data to the mobile device). The REST module can be configured to include web endpoints for evaluating and processing the received digital images. The backend module 120 can further include various processing modules for segmenting the images, extracting various particle properties from the images, and classifying individual particles in the images. For example, in the depicted embodiment, the processing modules can include: an image segmentation module 132 configured to process the digital image and generate a segmented image; an image property extraction module 134 configured to extract properties from selected segmented particles of the segmented particles, e.g., including size and shape related features; a debris identification module 136 configured to identify debris particles as particles having at least one size related feature greater than a threshold; and a shape classification module 138 configured to process the extracted properties and group each of the identified debris particles into corresponding shape classifications or categories (as described in more detail below with respect to Figure 5 .

[0021] The backend module 120 can further include a database manager 140 configured to interface with a database 150 to store the original images, the segmented images, and the property and classification data generated by the processing modules. The system 100 can further include a dashboard 160 that shows a list of the processed images stored in the database 150. In an alternative embodiment, the database 150 can further communicate with a global subsurface database 170 that can include well logging data for all wells in a particular oil field or region.

[0022] It will be appreciated that the backend module 120, the database 150, the dashboard 160, and the global subsurface database 170 can be deployed on almost any web server, such as including cloud services (such as Google Cloud Platform, Microsoft Azure) or on-premises servers located, for example, at an oil field or a service provider. Additionally, the backend module can be deployed on almost any networked personal computer or workstation. The disclosed embodiments are not limited to any particular server or hosting architecture.

[0023] In Figure 2B , a processing module 120' is deployed in the mobile device and is accessible to the App. For example, the App can be configured to access the module 120' for image evaluation and temporary data storage and to receive and display the evaluation results. As described above with respect to Figure 2A , the module 120' can include various processing modules for segmenting an image, extracting various particle properties, and classifying individual particles in the image. For example, in the depicted embodiment, the on-board processing module can include: an image segmentation module 132', configured to process a digital image and generate a segmented image; an image property extraction module 134', configured to extract properties from selected segmented particles of the segmented particles, such as including size and shape-related features; a debris identification module 136', configured to identify debris particles as particles having at least one size-related feature greater than a threshold; and a shape classification module '138, configured to process the extracted properties and group each of the identified debris particles into a corresponding shape classification or category (as described in more detail below with respect to Figure 5 . The database manager 140' can communicate with the on-board memory 142 to temporarily store images, segmented images, extracted properties, and classification data, and communicate with the external database 150 as described above.

[0024] It will be appreciated that the system 100' can be advantageously used in remote locations with limited mobile communication networks or other access to a larger computer network including an external backend. In such an implementation, digital images can be remotely acquired and evaluated. The images and evaluation results can then be updated to an external database (or other location with more robust mobile or wireless communication) at a later time. On the other hand, the system 100 can be advantageously used in locations with robust mobile networks or other high-bandwidth communication channels, where the acquired images can be quickly uploaded to and downloaded from a backend server with more robust processing capabilities.

[0025] Continuing to refer to FIG. 2, it will be appreciated that the modules 120, 120' of the systems 100, 100' can include computer hardware and software configured to automatically or semi-automatically evaluate images of rock particles. To perform these functions, the hardware can include one or more processors (e.g., microprocessors), which can be connected to one or more data storage devices (e.g., hard disk drives or solid state memories).

[0026] Continuing to refer to FIG. 2 and further referring to Figures 3A - 3F (collectively referred to as FIG. 3), the mobile device 110 can run a software App configured to enable a user to capture digital images of rock particles and upload the images for subsequent processing. When launched, the App can be configured to display a home page 202( Figure 3A ), which allows the user to capture a digital image of the prepared rock particles using the camera tab 204, or load a pre-existing image using the gallery tab 206. After capturing or selecting a digital image 212 (as Figure 3B shown), the user can click on the analysis tab 214 to load the image into a panel and send it to a processing module for analysis. The App can indicate that the image is being uploaded and / or processed, as Figure 3C shown by 222 in. When the analysis is complete, the results can be automatically saved to the memory 142 and / or the database 150 and displayed on the App (e.g., received or downloaded by one of the modules 120, 120' and displayed).

[0027] ​ depicts an example segmented image 232 downloaded to the App. Note that each of the debris particles is outlined using a polyline. Although not shown, the outlines can be colored to indicate different particle shapes, e.g., a first color indicates a first shape classification, a second color indicates a second shape classification, and so on. The App can be further configured to enable the user to display a summary of the particle shape classifications by clicking on the category tab 234. The summary of the particle shape classifications can be in the form of a list or a chart, e.g., ​ the pie chart shown. In the example depicted, the summary shows the number of particles in each shape classification. In the depicted embodiment, three out of the 22 particles in the segmented image are classified as plate-shaped, three are classified as spike-shaped, and 16 are classified as angular. The App can still be further configured to enable the user to display the particle properties (e.g., size and shape-related features) by clicking on the property tab 236. These properties can be shown as, for example, a two-dimensional scatter plot (cross plot) of the selected properties, e.g., as ​The drawn particle roundness is compared to the particle rectangularity shown. Of course, the disclosed embodiments are not limited to showing any particular property or showing properties on a cross plot, but may include a list of properties, a histogram, or any other suitable display. Additionally, individual particle instances (segmented particles) may be shown in the display to provide an indication of the particle properties of each of the particles in the image.

[0028] As described above with respect to ​ stated, the App can be configured to prompt the user to take a digital image of the prepared rock particles. It will be appreciated that mobile devices are generally not configured for calibrating image acquisition, and the acquired images may be uncalibrated. Thus, the App can be further configured to provide instructions or a tutorial for the image acquisition process to improve the image quality. For example, the user can be instructed to deploy the rock particles on a tray or paper background having a specific color. The App can further instruct the user to optimize the external lighting, or can also automatically employ the camera flash to ensure consistent image illumination. The App can still further be configured to instruct the user to hold the mobile device at a specific distance (or within a range of distances) from the particle. Additionally, in mobile devices including a depth sensor (e.g., an optical depth sensor such as lidar), the mobile device can be configured to instruct the user to move the mobile device closer to or farther from the particle until the distance between the mobile device and the rock particle is within a predetermined range. The App can also be configured to disable the camera unless it is within a predetermined distance range.

[0029] In some applications, it may be important to measure the size characteristics (e.g., length, width, and / or area) of the particles when extracting geometric properties. For example, these size characteristics can be measured based on the size of the individual pixels in the image. For example, the pixel size can be determined using an integrated depth sensor (e.g., a lidar sensor) in the mobile device or by including a known reference object in the image. In mobile device embodiments including a depth sensor, the App can be configured to access the depth sensor when taking an image to measure the distance between the camera and the particle. Referring to ​ it, the field of view FOV of the image can be calculated based on the measured distance d and the known focal length f and sensor height h (size) of the camera or according to the viewing angle θ of the camera, e.g., as follows:

[0030] FOV = 2d tanθ

[0031]

[0032] Then, the pixel size can be determined by dividing the field of view by the number of pixels N along the height of the camera sensor, e.g., as follows:

[0033]

[0034] It will be appreciated that a pixel sizing routine can be employed to determine the global pixel size (the pixel size over the entire image) or the local pixel size (e.g., the pixel size of an individual particle in the image). The main difference between the global pixel size and the local pixel size lies in the distance measurement. The global pixel size calculation can use the average distance or the distance from the center position of the image, while the local pixel size calculation can use different distance measurements for each individual particle obtained from, for example, a lidar depth map.

[0035] In other embodiments, the App can be configured to determine the pixel size based on an object of known size in the image. In such an embodiment, the App can instruct the user to include the size-determining object in the field of view and further ask the user to input the size of the object (e.g., diameter or other dimension). The size-determining information can then be transmitted or uploaded to the processing modules 120, 120' together with the image. The processing module can also be configured to segment and identify the size-determining object (e.g., in the case of a coin, the object can be segmented and then identified as having a roundness greater than a threshold). The pixel size can be determined based on the known size of the identified object (e.g., dividing the known diameter of the coin by the number of pixels along the diameter of the identified coin). It will be appreciated that the image can include multiple size-determining objects (e.g., one at each corner of the image, or one at the upper end and another at the lower end of the image) to compensate for image distortion or the tilt of the camera relative to the image plane. Additionally, in some embodiments, the backend can be configured to identify various standard objects such as various coins or rulers (e.g., using machine learning image identification routines) and determine the pixel size based on the known size of the identified object (e.g., the known diameter of the identified coin).

[0036] In certain embodiments, the App can be further configured to check the orientation of the mobile device before taking a digital image. For example, the App can be configured to access and process accelerometer measurements made in the device to determine the orientation of the device relative to the horizontal direction. The App can be configured to show a warning light on the user interface or even disable the camera when the deviation from the horizontal direction exceeds a threshold (e.g., 10 degrees). Additionally, the App can be further configured to indicate to the user along which axis to rotate the device to achieve an acceptable horizontal orientation.

[0037] Now turning to ​ , a flowchart of an example method 300 for evaluating rock particles is depicted. Method 300 includes acquiring, at 302, a digital image of rock particles separated from drilling fluid using a handheld mobile device. These particles can include, for example, cuttings and / or debris particles generated when drilling a wellbore and can be transported to the surface in the circulating drilling fluid, as described above with respect to ​As described. In an example implementation, at 304, the digital image can be uploaded to a backend server via a mobile or wireless network, for example, as described above with respect to FIGS. 2 and 3. At 306, the acquired image can be processed by a segmentation module to obtain a segmented image. The segmentation module can be configured to, for example, identify multiple individual particles in the uploaded image. At 308, the segmented image can be processed to extract geometric properties from each of the multiple selected particles in the segmented image. For example, geometric properties can be extracted from each of the segmented particles or a subset of the segmented particles. The selected particles can be evaluated, for example, particle by particle, to extract geometric properties (e.g., size and shape-related features).

[0038] In one example implementation, the geometric properties can include size and shape-related features. The size-related features can be evaluated at  310 to identify sloughing particles among the segmented particles (e.g., to distinguish cuttings and sloughing particles). For example, the identification can be based on one or more size-related feature thresholds, where cuttings particles are identified and classified as having size features less than the threshold, while sloughing particles are identified and classified as having size features greater than the threshold. In one example implementation, sloughing particles can be identified as particles having a diameter (e.g., maximum diameter or average diameter) greater than a diameter threshold (e.g., greater than about 3 mm). In other implementations, sloughing identification can utilize particle area, perimeter, volume, or other thresholds.

[0039] The shape-related features can be processed at 312 to classify each of the identified sloughing particles according to the corresponding particle shape (e.g., and thereby generate a particle shape classification). In an example implementation, at 314, the segmented image, the extracted properties, and the shape classification can then be automatically downloaded to a handheld mobile device, where the results can be displayed to the user. In an implementation where a backend module is employed on the mobile device as depicted ​ in the segmented image, the extracted properties, and the shape classification can be automatically displayed after processing is complete.

[0040] Continuing to refer ​ to, it will be appreciated that method 300 can advantageously achieve automatic (or semi-automatic) classification of rock particle shapes with no (or only limited) manual intervention. For example, the manual input for the method may be limited to taking an image at 302 and 304 and uploading the image to the backend server. Segmenting the image at 306, extracting geometric properties at 308, identifying sloughing particles at 310, classifying the sloughing particles at 312, and downloading the segmented image, the extracted properties, and the particle classification to the mobile device at 312 can be performed automatically without manual intervention.

[0041] ​ Depicted for separating rock particles from drilling fluid and at ​An example method 320 of collecting digital images at 302. For example, using the above ​ The example drilling rig 20 is shown drilling through the subsurface formation of interest to form a borehole at 322. Rock cuttings and fallout particles are introduced into the upwardly flowing drilling fluid ( ​ 94 in ) is transported to the ground. These particles can be collected, for example, using a shale shaker or other solid separation / control equipment at 324. The particles obtained from the shale shaker can be prepared for image analysis at 326, for example, by washing and then drying in an oven. At 328, the particles can be further optionally screened or sieved to remove small or large particles. In an example embodiment, the particles can be screened to primarily retain landslide particles. In other embodiments, the sampled particles may include both rock cuttings and landslide particles. It will be appreciated that rock cuttings and landslide particles are typically distinguished based on size in the industry, with particles having a size less than a threshold being classified as rock cuttings, and particles having a size greater than a threshold being classified as landslides.

[0042] At 330, the prepared rock particles can be placed in a tray or on a sheet of paper, as described above with respect to FIG. 2 . In an exemplary embodiment, the tray or paper can have a predetermined color, such as a high-contrast (bright) background color, to enhance subsequent particle identification and segmentation in the captured image, for example, pure magenta (e.g., RGB values of 255, 0, 255), pure blue (e.g., RGB values of 0, 0, 255), pure green (e.g., RGB values of 0, 255, 0), etc. Generally, such colors do not exist in nature and thus help the instance segmentation model avoid detecting the tray's background as part of the particles. The tray can be placed on a table or other horizontal surface, where a digital image can be captured at 332 using a mobile device as described above. It will be appreciated that the image can be captured under non-calibrated lighting conditions (e.g., under ambient light or using a camera flash as also described above) or under calibrated lighting conditions using external lighting and a light sensor.

[0043] Reference again ​, in an example implementation, the segmentation module can employ a Mask Region-based Convolutional Neural Network (Mask R-CNN), such as that disclosed in U.S. Patent Application Serial No. 17 / 647,407, which is incorporated herein by reference in its entirety. Mask R-CNN can be configured to identify individual particles in a digital image and thereby generate a segmented image at 306. For example, Mask R-CNN can produce, for instance, bounding boxes and mask images. A bounding box can be defined as a set of xy coordinates in an image that indicate the region of the image containing the object of interest. The bounding box can include a confidence score for each object of interest, which ranges from 0 to 1 (e.g., a larger value indicates a higher associated confidence). The mask image can indicate (e.g., highlight or otherwise delimit) the regions of interest having a confidence score above a threshold.

[0044] It will be appreciated that Mask R-CNN is a model architecture belonging to the category of supervised learning, which means that it generally requires a training data set including images and corresponding labels. For example, the model can be trained using images including cuttings and / or debris particles of various sizes, shapes, colors, and lithology types. It will further be appreciated that the R-CNN model can be continuously retrained during a drilling operation or between oilfield drilling operations. For example, segmentation errors can be identified and corrected, and then the segmentation errors can be used to generate labeled training images that can be used to retrain (or further train) R-CNN.

[0045] In an example implementation, the segmented image can depict multiple rock particles on a high-contrast tray. Individual particles can be identified, for example, by particle contours or other demarcation lines. Additionally, each identified particle can be identified by a corresponding set of pixels in the image. In other words, the segmented image can include pixel-by-pixel segmentation, where each pixel in the image is assigned to the background or a single individual particle.

[0046] In some operations, debris particles can be placed on a tray or holder having a black (or dark) background grid prior to imaging. It will be appreciated that such a grid can sometimes interfere with image segmentation. The segmentation procedure can be modified to accommodate the grid. For example, in one method, image processing techniques such as morphological closing can be used to remove the grid from the image prior to segmentation. In another method, the image can be used to train the segmentation model, where cuttings and / or debris particles are placed on a holder including a black (or other dark) grid (as opposed to a background having a bright color). The training can include images with grids of various sizes and colors. Then, image segmentation at 306 can use the grid-trained segmentation model.

[0047] Depending on whether the image includes a background grid, it may be advantageous to employ different segmentation models at 306. For example, an image that does not include a background grid may employ a segmentation model trained using images that do not include a grid, while an image that includes a background grid may employ a segmentation model trained using images that include a background grid. In some embodiments, the processing at 306 may further include evaluating the image to determine whether the image includes a background grid, and then selecting an appropriate segmentation model based on the evaluation. Of course, the disclosed embodiments are not limited in this respect.

[0048] Further references ​ , the geometric properties extracted at 308 may include substantially any size-related features and / or shape-related features of the particle. For example, the extracted geometric properties may include at least one size-related feature, such as equivalent particle diameter, area, perimeter, maximum axis, minimum axis, and particle aspect ratio (ratio of maximum axis to minimum axis). These size-related features may be extracted, for example, by pixel measurements. For example, the area of a particle may include the number of pixels in a cross-section of the particle. Similarly, the maximum and minimum axes and perimeter may be defined as the number of pixels along an axis. These features may also be converted to conventional metric dimensions based on the calculated pixel size. In addition, the geometric-related properties may further include the spatial relationship of the pixels grouped in each particle to extract particle circularity, roundness, rectangularity, solidity, elongation, eccentricity, compactness, convexity, curling, orientation, and / or convex hull area. It will be appreciated that these shape-related features are well defined in digital image processing applications, for example, in Open CV ( ​ )、ImageJ User Guide( ​ ) and Shape Analys is and Measurement( ​ ​ ​ ) is defined in .

[0049] Still further reference ​ The extracted size-related features can be processed at 310 to identify avalanche particles in the image. As described above, identification can include comparing the extracted size features of each segmented particle to one or more size feature thresholds. Particles exceeding the thresholds can be identified as avalanche particles, while the remaining particles can be identified as rock debris particles. As described above, the size-related features can include particle diameter, perimeter, area, etc.

[0050] The extracted shape-related features can be processed using the shape classification module 138, 138' (Figure 2) to classify the identified avalanche particles into corresponding particle shapes at 312. The shape classification can include essentially any suitable shape, however, avalanche particles are typically classified into four different categories: spike-shaped, angular-shaped, plate-shaped, and block-shaped.​ Depicts an example of the shape classification of these four debris particles.

[0051] It will be appreciated that the shape of debris particles is sometimes used as an indicator of the formation fracture mechanism. For example, spiky debris can be regarded as an indication of tensile fracture, while angular debris can be regarded as an indication of shear fracture. Similarly, plate-like debris can be regarded as an indication of slip fracture caused by bedding planes, and blocky debris can be regarded as an indicator that the formation fracture is caused by pre-existing fractures. In addition, a mixed group of debris particles having these shapes may indicate multiple simultaneous fracture mechanisms. Advantageous embodiments may classify each of the particles into one of these four categories and may further classify or summarize the amount or percentage of each shape in a group of particles (e.g., collected in one or more images).

[0052] For example, particle shapes can be assigned based on any one or more of the geometric properties listed above. In one embodiment, the shape classification can employ a heuristic approach where thresholds are applied to the calculated geometric properties in a hierarchical manner. In one example heuristic approach, particles exceeding a first threshold of a first property can be assigned to a first classification. Among the remaining particles (particles not exceeding the first threshold), particles exceeding a second threshold of a second property can be assigned to a second classification. Among the particles not exceeding the first threshold or the second threshold, particles exceeding a third threshold of a third property can be assigned to a third classification. The remaining particles (not exceeding at least one of the first threshold, the second threshold, and the third threshold) can be assigned to a fourth classification. By way of example only, debris particles can be classified as spiky when the elongation feature and / or the aspect ratio feature is greater than a threshold (indicating that the maximum axial dimension of the particle is significantly greater than the minimum axial dimension). Among the remaining (non-spiky) particles, particles having a rectangularity feature exceeding a threshold can be classified as plate-like. Those particles that are neither spiky nor plate-like but have a roundness feature exceeding a threshold can be classified as blocky. The remaining particles not meeting any of these criteria can be classified as angular.

[0053] In another embodiment, avalanche particles can be classified according to the "position" of the particles in the multidimensional space of the geometric properties extracted. As described above, a set of geometric properties can be calculated (for example, for each of selected particles). This set of geometric properties can include substantially any number of size and shape-related features, for example, including at least 4 features (for example, at least 6, 8, 12 or 16 features). The shape classification module can be configured to classify particles according to the values of those geometric properties, for example, the values of those geometric properties cause similar particles to gather in the above-mentioned multidimensional feature space. Alternatively (and / or additionally) particles can be classified based on the nearest neighbor classification in the multidimensional space of the color and texture features extracted. In an exemplary embodiment, each classification of particles can be assigned based on clustering. In such an embodiment, multiple groups of particles located in the same cluster (or local area of hyperspace) can be assigned the same classification.

[0054] In yet other embodiments, the shape classification module can be configured to use a trained machine learning model. In such an embodiment, the classification model can be trained using geometric properties extracted from labeled avalanche particles. Once trained, the model can be used to infer avalanche shape classification from the segmented images and the corresponding extracted geometric properties.

[0055] Continue to refer ​ , the extracted size-related features can be further processed at 310 and / or 312 to classify the size of the identified avalanche particles. For example, each identified avalanche particle can be classified as a small, medium or large avalanche particle (e.g., in addition to having a spike-like, plate-like, block-like or angular shape). Particle size classification can utilize size-related features, such as equivalent particle diameter, area, perimeter, maximum axis, minimum axis, the sum of the maximum axis and minimum axis, or other similar size-related features. For example, particle size can be classified by comparing one or more size-related features with corresponding thresholds. In one embodiment, when a first size-related feature (such as an equivalent particle diameter) exceeds a first threshold, the rock particle can be classified as a large avalanche particle. When the first size-related feature is less than a second threshold, the particle can be classified as a small avalanche particle, and when the first size-related feature is greater than the second threshold but less than the first threshold, the particle can be classified as a medium avalanche particle. Advantageous embodiments can further classify or summarize the amount or percentage of particles in each size classification.

[0056] As described above, the particle size and shape classification and / or summary can provide an indication of the formation fracture mechanism and the overall integrity and / or stability of the wellbore to the user. In some embodiments, the App can be further configured to highlight certain breakage size and / or shape combinations that are strong indicators of wellbore instability, or to highlight images with a large percentage of large breakage particles.

[0057] Referring again to FIG. 2, the system outputs (images, segmented images, geometric properties, and particle classifications) can be stored in one or more databases 150, 170. The databases can be configured to include multiple types of data, including numerical data (such as geometric properties), string data (such as user and well names), and binary data (such as images and segmented images). Metadata and outputs can be sorted in tabular format, such as in JSON, and image data can be stored in a folder system. In another embodiment, an SQL database and Blob storage (such as Amazon S3) can be used in place of the folder system.

[0058] The database can be linked to the breakage dashboard 160, which provides a web-based visualization of the database. Data from the database can be accessed in the dashboard using almost any connected device (including mobile devices and the App). For example, by clicking "Refresh", the data in the dashboard can be updated to include the latest data in the database. The dashboard interface can be presented in tabular format, with columns similar to those in the database, and can include clickable hyperlinks to download the input image, segmented image, and other processed data such as particle classifications and particle properties.

[0059] It will be appreciated that the disclosed embodiments are not limited to classifying only breakage particles. Instead, the disclosed systems and methods can be configured to evaluate substantially any rock particles (such as cuttings and / or breakage particles) collected during drilling, and can be further configured to distinguish between cuttings and breakage particles (e.g., based on a particle size threshold or as having different non-breakage shape classifications). Additionally, the disclosed embodiments can be further configured to classify cuttings and / or breakage particles according to formation lithology, porosity, and substantially any other formation property or descriptor that can be derived from digital images.

[0060] In an example implementation, the processing module may further include a color and texture feature extraction module configured to extract color and texture related features from the identified particles. Example color related features may include the average (such as mean, median or mode) red, green and blue intensities or the distribution or standard deviation of red, green and blue intensities and / or the average brightness of each particle. Color related features may further include histograms, variances, skewness and / or kurtosis of red, green and blue intensities. The extracted texture related features may quantify various spatial relationships and / or directional variations of pixel colors and / or brightness in each particle. The extracted texture related features may include, for example, edge detection, pixel-to-pixel contrast, correlation and / or entropy. Additionally, in some implementations, the texture related features may be extracted using image texture filters (such as, for example, Gabor filters, etc.), autoencoders or other deep learning-based techniques. Further, the directional variations may be evaluated, for example, with respect to symmetry and used to generate a spectrum, which may be further compared with a reference spectrum to assign a texture classification to each particle, such as homogeneous, inhomogeneous, granular, laminated particles, etc. The texture features may be further evaluated, for example, to characterize the grain size or grain size distribution of the particles in the formation or rock particles. For example, the grain size may be identified as fine, medium, coarse or having an average size and size distribution.

[0061] It will be appreciated that color, texture and geometric feature extraction may utilize trained machine learning algorithms or any other deep learning algorithms. Such algorithms may be trained, for example, using the extracted color and texture features of different rock particle types (such as, lithology types), sizes, shapes, colors, etc., and may utilize an image database including images of rock particles. For example, such a database may be maintained at the backend or may be accessed by a backend module.

[0062] The extracted color and / or texture features can be optionally processed to further classify the rock particles according to the corresponding lithology types. For example, as described in U.S. Patent Application Serial No. 17 / 647,412, which is incorporated herein by reference in its entirety. Generally speaking, the lithology of a rock particle is a description of its physical properties visible in outcrops, hand-held or core samples, or with a low-power microscope. Lithology can refer to a detailed description of these physical properties or an overview of the overall physical properties of the rock. In the second sense, the lithology of a rock refers to the type of rock or the overall (or macroscopic) identification or classification of the rock. Example lithologies or lithology types in this second sense include sandstone, limestone, slate, shale, basalt, coal, anhydrite, dolomite, gypsum, clay, flint, granite, and the like. Thus, the color and / or texture features represent physical properties of the lithology that can be measured and quantified from a digital image and evaluated to determine the lithology (in the second sense) of the formation. For example, a trained model can be used to evaluate the color and / or texture features extracted from selected particles in the segmented image to classify the lithology types of the selected particles.

[0063] For example, based on the "position" of the particles in the multi-dimensional space of the extracted color and texture features, the extracted color and texture features can still be further processed to estimate the formation porosity. For example, a set of color and texture features can be calculated (e.g., for each of the selected rock particles). The calculated set of color and texture features can include a large number of features, such as including 16 or more features. The backend module can be configured to correlate the color and texture features with the formation porosity such that in practice, the module assigns a porosity value to the rock particles based on which set of feature values (or in other words, based on the position of the rock in the aforementioned multi-dimensional color / texture feature space).

[0064] It will be understood that the present disclosure includes numerous embodiments. These embodiments include, but are not limited to, the following embodiments.

[0065] In a first embodiment, a method for classifying the shape of caving particles includes: using a mobile device to acquire a digital image of rock particles generated when drilling a subterranean wellbore; processing the digital image to generate a segmented image that identifies individual rock particles among the rock particles; extracting a plurality of geometric properties from selected identified rock particles depicted in the segmented image, the plurality of geometric properties including size features and shape features; evaluating at least one of the size features to identify caving particles among the identified rock particles; and generating a caving particle classification by processing the plurality of shape features so as to label each of the identified caving particles with a corresponding caving shape.

[0066] The second implementation may include the first implementation, wherein the processing of the digital image to generate a segmented image, the extraction of a plurality of geometric properties, the evaluation of at least one of the size features, and the processing of the plurality of the shape features are performed by the mobile device.

[0067] The third implementation may include the first implementation, wherein the processing of the digital image to generate a segmented image, the extraction of a plurality of geometric properties, the evaluation of at least one of the size features, and the processing of the plurality of the shape features are performed by a processor external to the mobile device; and the method further includes uploading the digital image from the mobile device to the external processor and downloading the caving particle shape classification from the processing module to the mobile device.

[0068] The fourth implementation may include any one of the first to third implementations, wherein the acquisition of the digital image includes: drilling a subterranean wellbore; collecting the cuttings particles from the circulating drilling fluid; preparing the rock particles for imaging; and taking a digital photograph of the rock particles using the mobile device.

[0069] The fifth implementation may include any one of the first to fourth implementations, wherein the evaluation of at least one of the size features includes identifying and labeling the identified rock particles as caving particles when at least one of the size features is greater than a corresponding size feature threshold.

[0070] The sixth implementation may include any one of the first to fifth implementations, wherein the processing of the plurality of shape features employs a heuristic method, wherein thresholds are applied to the extracted shape features in a hierarchical manner.

[0071] The seventh implementation may include the sixth implementation, wherein the processing of the plurality of shape features includes: assigning the caving particles exceeding a first shape feature threshold to a first shape classification; assigning the caving particles exceeding a second shape feature threshold and not exceeding the first shape feature threshold to a second shape classification; assigning the caving particles exceeding a third shape feature threshold and not exceeding the first shape feature threshold or the second shape feature threshold to a third shape classification; and assigning the caving particles not exceeding the first shape feature threshold, the second shape feature threshold, or the third shape feature threshold to a fourth shape classification.

[0072] The eighth implementation may include the seventh implementation, wherein: the first shape feature threshold is an elongation threshold; the second shape feature threshold is a rectangularity threshold; and the third shape feature threshold is a roundness threshold.

[0073] The ninth implementation may include any one of the first to eighth implementations, wherein the evaluating at least one of the size features further includes labeling each of the identified spall particles with a corresponding size.

[0074] The tenth implementation may include any one of the first to ninth implementations, wherein the spall particle classification includes the segmented image, the extracted geometric properties, and the shape classification summary.

[0075] In an eleventh implementation, an integrated mobile system for evaluating spall particles includes: a mobile device including a digital camera configured to acquire digital images of rock particles generated when drilling a subterranean wellbore; and a processing module configured to receive the digital images from the digital camera, the processing module including: an image segmentation module configured to process the digital images and generate segmented images; a geometric property extraction module configured to extract size features and shape features from selected rock particles among the rock particles in the segmented images; a spall identification module configured to process at least one of the size features to identify spall particles among the rock particles; and a shape classification module configured to generate a spall particle shape classification by processing a plurality of the shape features to label each of the identified spall particles with a corresponding spall shape.

[0076] The twelfth implementation may include the eleventh implementation, wherein the processing module is in the mobile device.

[0077] The thirteenth implementation may include the eleventh implementation, wherein the processing module is external to the mobile device; the mobile device is configured to upload the digital images to the processing module via a mobile or wireless communication link; and the processing module is configured to download the spall particle shape classification to the mobile device.

[0078] The fourteenth implementation may include any one of the eleventh to thirteenth implementations, further including: a database manager communicating with a database configured to store the digital images, the segmented images, the extracted geometric properties, and the spall particle shape classification; and a dashboard configured to provide a web-based visualization of the database.

[0079] The fifteenth implementation may include any one of the eleventh to fourteenth implementations, wherein the shape classification module is configured to process the extracted geometric properties using heuristics, and thresholds are applied to the extracted shape features in a hierarchical manner such that the identified debris particles exceeding a first shape feature threshold are assigned to a first shape classification; the identified debris particles exceeding a second shape feature threshold and not exceeding the first shape feature threshold are assigned to a second shape classification; the identified debris particles exceeding a third shape feature threshold and not exceeding the first shape feature threshold or the second shape feature threshold are assigned to a third shape classification; and the identified debris particles not exceeding the first shape feature threshold, the second shape feature threshold, or the third shape feature threshold are assigned to a fourth shape classification.

[0080] In a sixteenth implementation, a method for classifying the shapes of debris particles generated when drilling a subterranean wellbore includes: acquiring, using a mobile device, a digital image including both cuttings particles and debris particles generated when drilling the subterranean wellbore; processing the acquired digital image to generate a segmented image that identifies individual particles among the cuttings particles and the debris particles depicted in the digital image; extracting size features and shape features from the segmented cuttings particles and segmented debris particles in the segmented image; processing at least one of the extracted size features to identify the debris particles; and generating a debris particle shape classification by processing a plurality of the shape features in order to label each of the identified debris particles with a corresponding debris shape.

[0081] The seventeenth implementation may include the sixteenth implementation, wherein: the mobile device further measures a distance between the mobile device and the cuttings particles and the debris particles; the extracted size features are in pixels; and processing the at least one of the extracted size features further includes: (i) processing the distance to calculate a pixel size in the digital image, (ii) multiplying the pixel size by the at least one of the extracted size features to obtain a particle size, and (iii) comparing the particle size with a particle size threshold to identify the debris particles.

[0082] The eighteenth implementation may include the sixteenth implementation, wherein the digital image further includes a size determination object; the extracted size feature is in pixels; and processing the at least one of the extracted size features further includes: (i) determining the pixel size of the size determination object, (ii) dividing the known size of the size determination object by the pixel size of the size determination object to obtain the pixel size in the digital image, (iii) multiplying the pixel size by the at least one of the extracted size features to obtain the particle size, and (iv) comparing the particle size with a particle size threshold to identify the debris particles.

[0083] The nineteenth implementation may include any one of the sixteenth to eighteenth implementations, wherein the processing of the plurality of shape features uses a heuristic method, wherein thresholds are applied to the extracted shape features in a hierarchical manner, wherein: the debris particles exceeding a first shape feature threshold are assigned to a first shape classification; the debris particles exceeding a second shape feature threshold and not exceeding the first shape feature threshold are assigned to a second shape classification; the debris particles exceeding a third shape feature threshold and not exceeding the first shape feature threshold or the second shape feature threshold are assigned to a third shape classification; and the debris particles not exceeding the first shape feature threshold, the second shape feature threshold, or the third shape feature threshold are assigned to a fourth shape classification.

[0084] The twentieth implementation may include any one of the sixteenth to nineteenth implementations, wherein the first shape feature threshold is an elongation threshold; the second shape feature threshold is a rectangularity threshold; and the third shape feature threshold is a roundness threshold.

[0085] In a twenty-first implementation, a method for evaluating rock particles includes: using a mobile device to acquire a digital image of rock particles generated when drilling a subterranean wellbore; processing the digital image to generate a segmented image that identifies individual rock particles among the rock particles; extracting a plurality of properties from selected identified rock particles depicted in the segmented image, the plurality of properties including at least one of color features, texture features, size features, and shape features; and evaluating at least one of the plurality of properties to determine at least one of the porosity and lithology of the formation penetrated by the wellbore.

[0086] Although the integrated mobile system for formation rock analysis has been described in detail, it should be understood that various changes, substitutions, and alterations can be made herein without departing from the spirit and scope of the present disclosure as defined by the appended claims.

Claims

1. A method for classifying the shape of caving debris particles, the method comprising: Using a mobile device to collect digital images of rock particles generated when drilling a subterranean wellbore; Processing the digital images to generate a segmented image that identifies individual rock particles among the rock particles; Extracting a plurality of geometric properties from selected identified rock particles depicted in the segmented image, the plurality of geometric properties including size characteristics and shape characteristics; Evaluating at least one of the size characteristics to identify caving debris particles among the identified rock particles; And Generating a caving debris particle classification by processing a plurality of the shape characteristics so as to label each of the identified caving debris particles with a corresponding caving debris shape.

2. The method according to claim 1, wherein the processing of the digital images to generate a segmented image, the extraction of the plurality of geometric properties, the evaluation of at least one of the size characteristics, and the processing of the plurality of the shape characteristics are performed by the mobile device.

3. The method according to claim 1, wherein The processing of the digital images to generate a segmented image, the extraction of the plurality of geometric properties, the evaluation of at least one of the size characteristics, and the processing of the plurality of the shape characteristics are performed by a processor external to the mobile device; and The method further comprises uploading the digital images from the mobile device to the external processor and downloading the caving debris particle shape classification from the processing module to the mobile device.

4. The method according to claim 1, wherein the collecting of the digital images comprises: Drilling a subterranean wellbore; Collecting the rock particles from the circulating drilling fluid; Preparing the rock particles for imaging; And Taking digital photographs of the rock particles using the mobile device.

5. The method according to claim 1, wherein the evaluating of at least one of the size characteristics comprises identifying and labeling the identified rock particles as caving debris particles when the at least one of the size characteristics is greater than a corresponding size characteristic threshold.

6. The method according to claim 1, wherein the processing of the plurality of shape characteristics employs a heuristic method, wherein thresholds are applied to the extracted shape characteristics in a hierarchical manner.

7. The method according to claim 6, wherein the processing of the plurality of shape characteristics comprises: Assigning the caving debris particles exceeding a first shape characteristic threshold to a first shape classification; Assigning the caving debris particles exceeding a second shape characteristic threshold and not exceeding the first shape characteristic threshold to a second shape classification; Assigning the caving debris particles exceeding a third shape characteristic threshold and not exceeding the first shape characteristic threshold or the second shape characteristic threshold to a third shape classification; and Assigning the caving debris particles not exceeding the first shape characteristic threshold, the second shape characteristic threshold, or the third shape characteristic threshold to a fourth shape classification.

8. The method according to claim 7, wherein: The first shape characteristic threshold is an elongation threshold; The second shape characteristic threshold is a rectangularity threshold; and The third shape feature threshold is a roundness threshold.

9. The method according to claim 1, wherein said evaluating at least one of said size features further comprises labeling each of the identified debris particles with a corresponding size.

10. The method according to claim 1, wherein said debris particle classification comprises said segmented image, the extracted geometric properties, and a shape classification summary.

11. An integrated mobile system for evaluating debris particles, the system comprising: a mobile device comprising a digital camera configured to capture digital images of rock particles generated during the drilling of a subterranean wellbore; and a processing module configured to receive the digital images from the digital camera, the processing module comprising: an image segmentation module configured to process the digital images and generate a segmented image; a geometric property extraction module configured to extract size features and shape features from selected rock particles among the rock particles in the segmented image; a debris identification module configured to process at least one of the size features to identify debris particles among the rock particles; and a shape classification module configured to generate a debris particle shape classification by processing a plurality of the shape features so as to label each of the identified debris particles with a corresponding debris shape.

12. The system according to claim 11, wherein the processing module is in the mobile device.

13. The system according to claim 11, wherein: the processing module is external to the mobile device; the mobile device is configured to upload the digital images to the processing module via a mobile or wireless communication link; and the processing module is configured to download the debris particle shape classification to the mobile device.

14. The system according to claim 11, the system further comprising: a database manager in communication with a database configured to store the digital images, the segmented images, the extracted geometric properties, and the debris particle shape classification; and a dashboard configured to provide a web-based visualization of the database.

15. The system according to claim 11, wherein the shape classification module is configured to process the extracted geometric properties using a heuristic method, wherein thresholds are applied to the extracted shape features in a hierarchical manner such that the identified debris particles exceeding a first shape feature threshold are assigned to a first shape classification; the identified debris particles exceeding a second shape feature threshold and not exceeding the first shape feature threshold are assigned to a second shape classification; the identified debris particles exceeding a third shape feature threshold and not exceeding the first shape feature threshold or the second shape feature threshold are assigned to a third shape classification; and the identified debris particles not exceeding the first shape feature threshold, the second shape feature threshold, or the third shape feature threshold are assigned to a fourth shape classification.

16. A method for classifying the shapes of debris particles generated when drilling an underground wellbore, the method comprising: Using a mobile device to collect digital images including both cuttings particles and debris particles generated when drilling the underground wellbore; Processing the collected digital images to generate a segmented image that identifies individual particles among the cuttings particles and the debris particles depicted in the digital images; Extracting size features and shape features from the segmented cuttings particles and segmented debris particles in the segmented image; Processing at least one of the extracted size features to identify the debris particles; And Generating a debris particle shape classification by processing a plurality of the shape features so as to label each of the identified debris particles with a corresponding debris shape.

17. The method according to claim 16, wherein: The mobile device further measures the distance between the mobile device and the cuttings particles and the debris particles; The extracted size features are in pixels; And Processing the at least one of the extracted size features further comprises: (i) processing the distance to calculate the pixel size in the digital image, (ii) multiplying the pixel size by the at least one of the extracted size features to obtain a particle size, and (iii) comparing the particle size with a particle size threshold to identify the debris particles.

18. The method according to claim 16, wherein: The digital image further includes a size determination object; The extracted size features are in pixels; And Processing the at least one of the extracted size features further comprises: (i) determining the pixel size of the size determination object, (ii) dividing the known size of the size determination object by the pixel size of the size determination object to obtain the pixel size in the digital image, (iii) multiplying the pixel size by the at least one of the extracted size features to obtain a particle size, and (iv) comparing the particle size with a particle size threshold to identify the debris particles.

19. The method according to claim 16, wherein the processing of the plurality of shape features employs a heuristic method, wherein thresholds are applied to the extracted shape features in a hierarchical manner, wherein: Debris particles exceeding a first shape feature threshold are assigned to a first shape classification; Debris particles exceeding a second shape feature threshold and not exceeding the first shape feature threshold are assigned to a second shape classification; Debris particles exceeding a third shape feature threshold and not exceeding the first shape feature threshold or the second shape feature threshold are assigned to a third shape classification; and Debris particles not exceeding the first shape feature threshold, the second shape feature threshold, or the third shape feature threshold are assigned to a fourth shape classification.

20. The method according to claim 19, wherein: The first shape feature threshold is an elongation threshold; The second shape feature threshold is a rectangularity threshold; and The third shape feature threshold is a roundness threshold.

Images