APPARATUS AND METHOD FOR ANALYSING DRILLING FLUID

Abstract

A system and method of analysing drilling cuttings using image data output from a hyperspectral imaging device and at least one optical camera, includes generating a hyperspectral imaging data set including a plurality of lines of hyperspectral data derived from line images taken by the hyperspectral imaging device positioned along a drilling fluid cuttings path, obtaining tracking information in respect of particles of interest from the output of the at least one optical camera, correcting the position of pixels associated with particles of interest in the plurality of lines of hyperspectral imaging data based on the obtained tracking information to generate corrected hyperspectral imaging data, and analysing the corrected hyperspectral imaging data to characterise the cuttings.

Claims

1. A method of analysing drilling cuttings using image data output from a hyperspectral imaging device and at least one optical camera, comprising: generating a hyperspectral imaging data set comprising a plurality of lines of hyperspectral data derived from line images taken by the hyperspectral imaging device positioned along a drilling fluid cuttings path; obtaining tracking information in respect of particles of interest from the output of the at least one optical camera; correcting the position of pixels associated with particles of interest in the plurality of lines of hyperspectral imaging data based on the obtained tracking information to generate corrected hyperspectral imaging data; and analysing the corrected hyperspectral imaging data to characterise the cuttings.

2. The method as claimed in claim 1, further comprising, distinguishing between background and particles of interest in the optical camera output of a portion of the drilling fluid cuttings path that includes the hyperspectral imaging line position, and differentiating between particles of interest and background in the hyperspectral imaging data based on the step of distinguishing.

3. The method as claimed in claim 1, further comprising tracking movement of particles in the optical camera output to obtain the tracking information and associating the particles of interest with the tracking information.

4. The method as claimed in claim 1, further comprising obtaining depth information, wherein the particles of interest are distinguished from the background using the depth information.

5. The method of claim 1, wherein differentiating between particles of interest and background in the hyperspectral imaging data comprises masking particles of interest in the optical data based on the step of distinguishing, and applying the mask to the hyperspectral imaging data to differentiate between particles of interest and background in the hyperspectral imaging data.

6. The method of claim 1, wherein associating the particles of interest with tracking information comprises determining the speed of movement associated with pixels in the optical camera output.

7. The method of claim 1, wherein the portion of the drilling fluid cuttings path is at least a portion of a shaker table.

8. The method of claim 7 wherein the capture of images by at least one of the hyperspectral camera and the optical camera are synchronised with the frequency of movement of the shaker table.

9. A method of analysing drilling cuttings using output from a hyperspectral camera and at least one optical camera, comprising: generating hyperspectral imaging data comprising a line of hyperspectral imaging data derived from a line image taken by the hyperspectral camera positioned along a drilling fluid cuttings path at a first time; performing a mineralogy analysis on the data of the hyperspectral line; projecting the line of hyperspectral data onto an image from the optical camera output of a portion of the drilling fluid cuttings path that includes the hyperspectral imaging line position and corresponds to the first time; classifying the mineralogy of the cuttings in the optical camera image along the projected line, and determining the morphology of the cuttings in the optical camera image.

10. The method of claim 9, further comprising generating an image including the mineralogy and morphology information.

11. A system for analysing drilling cuttings using output from a hyperspectral camera and at least one optical camera, comprising a processing unit configured to: generate a hyperspectral imaging data set comprising a plurality of lines of hyperspectral imaging data derived from line images taken by a hyperspectral imaging device positioned along a drilling fluid cuttings path; obtain tracking information in respect of particles in the output of the at least one optical camera; correct the position of pixels associated with particles of interest in the plurality of lines of hyperspectral imaging data based on the obtained tracking information to generate corrected hyperspectral imaging data; and analyse the corrected hyperspectral imaging data to characterise the cuttings.

12. The system of claim 11, wherein the processing unit is further configured to distinguish between background and particles of interest in image data from the optical camera of a portion of the drilling fluid cuttings path that includes the hyperspectral imaging line position, and to differentiate between particles of interest and background in the hyperspectral imaging data based on the distinguished background and particles of interest in the image date from the optical camera

13. The system of claim 11, wherein the processing unit is further configured to track movement of particles in the optical camera output to obtain the tracking information and associate the particles of interest in the hyperspectral imaging data with the tracking information.

14. The system of claim 11, wherein the processing unit is configured to distinguish the particles of interest from the background using depth information.

15. The system of claim 11, wherein the processing unit is configured to differentiate between particles of interest and background in the hyperspectral imaging data by masking the distinguished particles of interest in the optical data, and applying the mask to the hyperspectral imaging data.

16. The system of claim 11, wherein the tracking information comprises speed of movement associated with pixels in the optical camera output.

17. The system of claim 11, wherein the portion of the drilling fluid cuttings path is at least a portion of a shaker table.

18. A system for analysing drilling cuttings using output from a hyperspectral camera and at least one optical camera, comprising a processing unit configured to: generate hyperspectral imaging data comprising a line of hyperspectral imaging data derived from a line image taken by the hyperspectral camera positioned along a drilling fluid cuttings path at a first time; perform a mineralogy analysis on the data of the hyperspectral line; project the line of hyperspectral data onto an image from the optical camera output of a portion of the drilling fluid cuttings path that includes the hyperspectral imaging line position and corresponds to the first time; classify the mineralogy of the cuttings in the optical camera image along the projected line; and determine the morphology of the cuttings in the optical camera image.

19. A computer program embodied on a non-transitory computer readable medium and comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method of claim 1.

20. A computer program embodied on a non-transitory computer readable medium and comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method of claim 9.

Description

DRAWINGS

[0025] Embodiments of the invention will now be described in more detail, and by way of example only, with reference to the drawings, in which:

[0026] FIG. 1 is a flow diagram of a method in accordance with an embodiment;

[0027] FIG. 2 is a schematic representation useful in a comparison for mapping between the 2D optical camera and a 1D hyperspectral inspection;

[0028] FIG. 3 is a schematic representation of the outputs of the HSI and optical cameras in respect of a single tracked cuttings particle;

[0029] FIG. 4 is a flow diagram of a method in accordance with an embodiment;

[0030] FIG. 5 shows schematically a system in accordance with an embodiment.

DESCRIPTION

[0031] The invention combines hyperspectral imaging and one or more (high-speed) camera(s) for purposes of describing drilling cuttings and cavings, for example, while traversing a shale shaker or shaker table.

[0032] Apparatus according to an embodiment are shown schematically in FIG. 5. Drilling fluid returned from the downhole drilling process moves along return path 5 in the direction of shale shaker or shaker table 7. A single shaker table 7 is shown for convenience in FIG. 5. The return path continues after the shaker table 7 along path 11 to a disposal area for the cuttings. Typically the shale shaker includes a plurality of vertically arranged shaker tables with apertures of decreasing size down the stack so as to remove progressively finer particles. In this case path 11 may be in the form of a chute with the cuttings caught by each of the tables 7 being deposited down the chute 11. The upper shaker table 7 or scalping table removes the largest cuttings/ cavings particles.

[0033] A short wave infrared 1D (line) hyperspectral camera 9 is positioned at suitable location above the shaker table 7 so as to be positioned for capturing a line of hyperspectral data as the drilling fluid and any cuttings pass beneath the hyperspectral camera 9. A 2D high speed optical camera system 6 is also located in order to capture a 2D image of the shaker table 7 and any drilling fluid and cuttings on the shaker table 7. In an embodiment the camera system 6 is a stereoscopic camera system shown schematically by the two camera sensors 6 in FIG. 5. The camera system 6 may include more or alternative sensors, including optical or video cameras, single or multi-stereo-cameras, night vision cameras, IR, LIDAR, RGB-D cameras, or other recording and/or distance-sensing equipment. The camera system 6 provides a 2D capture feed of at least a portion of the shaker table including the line covered by the hyperspectral camera 9. The data acquired by the camera system 6 (and the hyperspectral camera 9) is delivered to a processing unit 10 including the computer vision (CV) capability for analysing the 2D feed and extracting both the height information (for identifying cuttings particles and separating out the background) and the speed/motion information required for tracking the individual identified particles. The CV processing unit 10 may also include shape recognition software for determining the morphology of the identified particles. In an embodiment, Illumination is provided by lamp or lamps 8.

[0034] Computer Vision (CV) or Machine Vision software is well known, for example packages such HALCON™ from MVTec, are known, and allow development of applications for blob analysis, morphology, matching, measuring, and identification for example. Applying computer vision techniques to the output from the camera system 6 allows the signals captured using the two types of sensors (hyperspectral camera and computer vision (CV)) to be synchronised, making it possible to correlate the measurements from the sensors. Examples of applications/ benefits of correlating measurements: [0035] 1. A 3D depth map from the CV system can be used to identify parts of the shaker covered by fluid only (no cuttings/cavings). This provides the HSI system with the ability to identify and (spectrally) define the fluid and/or shaker bed. In other words the depth information allows categorisation of the points in the HSI data that include cuttings from those that are just fluid or shaker table. The mineralogy identification algorithm can use this information to achieve better results since it is better able to distinguish between cuttings/cavings and the background. This procedure also allows the HSI system to provide more accurate (less contaminated) spectral data to the algorithm for mineralogical identification. [0036] It should be noted that the larger particles will generally have less drilling fluid sticking to them. Identifying the larger particles, which have greater available surface area, also improves the confidence in the mineralogical identification. [0037] 2. Using the CV system to track particles as they move across the line scan makes it possible to correlate 1D HSI measurements of the same piece of rock. CV allows for tracking of cuttings (pixels representing at least a portion of an individual cutting entity) over the 2D area of interest. [0038] Correlating as described above provides the ability to link cuttings and cavings in the returned drilling fluid to geological formations by linking to other drilling sensor or data analysis systems; in particular, the bit depth and the flow rate of the drilling fluid. [0039] Determining the morphology of cuttings and the mineralogy allows labelling of the contents of the cuttings in the 2D image. In an embodiment the morphology and the mineralogy is determined from the HSI data, albeit the optical (e.g. 2D video) is used to enhance the morphology determination from the HSI data. Alternatively or additionally, morphology can be determine separately from the 2D output. [0040] 3. The 2D camera(s) allow a 2D (continuous) hyperspectral image to be constructed from a 1D HSI output, whilst correcting for different movement speeds along the shaker across the 1D scan line (using motion detector by CV system of identified pixels). [0041] The hyperspectral camera of the embodiment is a 1D line Short Wave Infra-Red device. The 2D stereo camera feed is processed by the data processing unit or video processor to identify shapes of individual entities in the 2D frame. Since the signals are synchronised, the hyperspectral analysis can be mapped into 2D space (by time stamping the 1D hyperspectral image). The mineralogy from the hyperspectral system can then be mapped onto the (shape identified) cutting entities. In this manner, the morphology and mineralogy of an entity can be linked in accordance with the method of FIG. 1.

[0042] The steps of the method of FIG. 1 are described in more detail with reference in addition to FIG. 2: [0043] In step S1 of FIG. 1, a line of hyperspectral data taken by the HSI camera is classified by known techniques to determine the mineralogy present in the cuttings traversing the hyperspectral line at t=t1. At step S2 the 1D hyperspectral line is projected at time t=t1 onto a 2D image taken by optical (video) cameras of the shaker table 2 including the cuttings and cavings particles 1. The projection is required since the pixels in the 1D HSI image do not map one to one onto the 2D image taken by the camera system 6. [0044] Whilst CV can be used for shape detection in the 2D image it is easier to perform boundary detection in the HSI data. Thus by clustering the HSI data and performing boundary detection, using the 2D depth map as additional data, the object boundaries can be more accurately determined. [0045] In step S3 the cuttings and cavings particles 1 identified in the 2D image are classified with the mineralogy information determined in step S1. In step S4 the morphology of the cuttings and cavings particles 1 identified in the 2D image are determined using known CV techniques. Finally, in step S5 the mineralogy & morphology of each identified cuttings particle 1 are linked to obtain a complete picture, for example allowing a labelled representation of the cuttings particles on the shaker table to be displayed.

[0046] Whilst the above discussion and the flow chart in FIG. 1 imply an order to the steps, in fact the invention is not so limited and certain steps can be carried out in different orders so long as they do not rely on previous steps. In particular, the morphology of the cuttings/cavings can be determined before or in parallel with the previous steps.

[0047] A 2D image can be obtained from a succession of 1D lines of hyperspectral data as discussed below in relation to FIG. 3. In the case of a 1D hyperspectral imaging device, the shapes of the cuttings will be distorted owing to the different speed of movement of the different cuttings entities across the cuttings flow path across the 1D hyperspectral imaging location. Thus the cuttings particles 1 represented schematically in FIG. 2, are represented as having corresponding particles 1a with notional shapes that differ to the actual particles 1 as they appear on the shaker table 2.

[0048] As discussed above, particles (cuttings/cavings) recorded and presented in the HSI data appear deformed, for example due to the movement of the shaker table. CV provides the ability to identify the movement of each particle (in particular each pixel or group of pixels identified as associated with a single cuttings particle), thus also the speed of movement of each cuttings particle 1. Applying the speed information obtained from the CV data, to the particles identified in the HSI data, creates a new (2D ‘continuous’ HSI dataset constructed from a 1D line scanner), with corrected shapes.

[0049] Referring to FIGS. 3 and 4, FIG. 3 shows schematically images of the HSI domain on the left and the CV domain on the right, whereas FIG. 4 indicates the process steps of the HSI domain on the left and the process steps in the CV domain on the right.

[0050] The HSI scanner generates a set of successive HSI lines 3. The successive or series of HIS lines include position errors, which would result in deformed shapes of particles (particle pixels appear where they are located after a movement of the particle) (HSI domain).

[0051] Computer vision techniques, are used to distinguish between the background and particles of interest. In an embodiment a depth map is constructed using the optical camera output (CV domain). Additional sensors could be used instead of the camera system 6 to provide the depth information and the camera system could be used simply to provide an optical image and to allow object tracking. Conveniently, however, the camera system can also provide the depth information. Where an addition sensor system is used to provide the depth map, this must be synced and correlated with the optical camera output. Such systems are known from WO-A-2016077521. This document also discusses different ways of distinguishing between cuttings and background that may be used in the present invention, including, background subtraction/change detection either on the shaker bed or when falling off the shaker bed, identifying objects that move at approximately constant velocity as relating to cuttings, distinguishing the texture difference between cuttings and background, reflectivity and colour properties, persistence and/ or tracking techniques. These and other known techniques may be used to distinguish between cuttings and background in the present invention.

[0052] Various techniques may be used to optimize the background removal process, for example, by synchronising the capture of the 2D and 1D image with the movement of the shaker, the moving shaker screen is in the same location on every image. For example, if the shaker vibrates at a particular frequency, the refresh rate of the cameras is set accordingly, for example, a multiple of the frequency.

[0053] If in addition, a picture is taken when the shaker screen is at stand-still (or pumps are off during drilling connection) then the background can be better defined by using the still shaker screen picture as a reference picture, in other words a base line for the algorithms can be based on the reference picture. It is then less likely that the shaker screen spectrum is mixed with rock or drilling fluid spectrum. Additionally or alternatively, images captured when only mud passes by (immediately after pumps on or not drilling), can be used for comparison to improve detection reliability.

[0054] The motion of the particles is tracked using standard CV techniques to track individual particles (pixels or groups of pixels) over frames and determine speed of movement of the particle (CV domain). In FIG. 3 the position of the particle in successive time stamped frames (t1 to t4) is shown schematically; the arrows indicate the movement of the particle between frames.

[0055] By distinguishing the background from the particles of cuttings and cavings and tracking the particles over time it is possible to mask areas of interest in the 2D image that is determined to relate to cuttings/cavings (CV domain).

[0056] By applying the mask to the 1D HIS data the data relating to cuttings/ caving can be better distinguished from the background (fluid or shaker screen). From the HSI data the mineralogy detection/identification algorithm performed on each of the HSI lines 3 can be guided towards an enhanced mineralogy detection/identification result (HSI domain).

[0057] A full HSI analysis is performed on each of the spectral lines 3 to determine spectral identity (HSI domain).

[0058] Using the movement speed information determined in the CV domain and mapped on to the HSI data, speed correction can be applied to each line scan HSI data, to replace pixels to their new (corrected shape) location (HSI domain).

[0059] The plural spectral lines 3 with the corrected position data can be stitched together to provide a ‘continuous’ 2D HSI image that in a newly created representation 4 of spectrally labelled data with the location of each cutting/caving on a location determined by their first point of passing through the scan line with corrected 2D shapes.

[0060] By virtue of the method of the embodiment an improved virtual HSI cuttings/cavings representation can be provided with information relating to mineralogy and morphology of the cuttings/cavings.

[0061] Statistics can be extracted relating to shape, size distribution from the data and applied (labelled) in the representation. (HSI domain).

[0062] As a result of the above techniques, particularly, the combination of the data from the optical camera system and the HSI data, the cuttings and cavings in the drilling fluid can be more accurately described in terms of morphology and mineralogy. The mineralogy of each cutting particle can be better described, in particular, the distribution of minerals in the particles as a whole and in particular particles using the 2D representations of the individual particles from the HSI data.

[0063] In an embodiment, Illumination is provided by lamp or lamps 8. Data acquisition can be optimized by matching illumination and spectral characteristics with the specific hyperspectral camera 9. The lamp 8 may be a halogen or (thermal) infrared lamp for example. The lamp 8 may be adjusted by optimizing the bulb shape and/or reflector width and curvature so as to obtain maximum intensity along the HSI measuring line with a uniform distribution. Can also prevent overheating; if the light source causes an increase in temperature of the atmosphere above an ignition point then it can be dangerous.

[0064] The performance of data acquisition by the sensors 6, 9, may also be improved by selecting shaker screens with respect to sieve mesh and colour. In particular, so that the background can be distinguished more easily during the background removal process.

[0065] The mudflow speed may be modulated during measurements of data, for example intermittently paused to allow for a clearer picture.

[0066] Data acquisition may also be improved by adding means of spraying the shaker screen with a cleaning agent, like diesel or base oil, allowing for pictures of solids with less adhered fluids.