Classification of damages on borehole walls by images

Classification of damages on borehole walls by images Cite as: AIP Conference Proceedings 2196, 020031 (2019); https://doi.org/10.1063/1.5140304 Published Online: 12 December 2019 A. Conci, M. B. Hernandez, J. Salas Cuno, D. Popovici, G. Jiga, M. Biondi, and E. W. Clua ARTICLES YOU MAY BE INTERESTED IN Thermosets synthesized by metathesis: Multiscale lifetime approach AIP Conference Proceedings 2196, 020034 (2019); https://doi.org/10.1063/1.5140307 The influence of heat and steam curing on the properties of one-part fly ash/slag alkali activated materials: Preliminary results AIP Conference Proceedings 2196, 020038 (2019); https://doi.org/10.1063/1.5140311 Stereological analysis of short basalt fiber composites AIP Conference Proceedings 2196, 020040 (2019); https://doi.org/10.1063/1.5140313 AIP Conference Proceedings 2196, 020031 (2019); https://doi.org/10.1063/1.5140304 © 2019 Author(s). 2196, 020031 Classification of Damages on Borehole Walls by Images A. Conci1, a), M. B. Hernandez1,b), J. Salas Cuno1,c), D. Popovici3,d), G. Jiga4,e), M. Biondi2,f), E. W. Clua1,g) 1 Computer Institute, Federal Fluminense University - UFF, Niterói, Brazil 2 Computational & Dimensional Metrology Lab., UFF, Niterói, Brazil. 3 Department of Electrical Engineering, Polytechnic University from Bucharest, Romania 4 Department of Strength of Materials, University Politehnica of Bucharest, Romania a) Corresponding author: [email protected] [email protected], [email protected], d)[email protected],e)[email protected], f) [email protected], g)[email protected] b) c) Abstract. Estimation of the stresses acting on the borehole wall is a very important aspect in oil industry. This work presents an image processing based tool for analyzing the well safety from borehole imaging instruments, because despite the advances on the development of acquisition tools, the final interpretation remains heavily dependent on the skill, experience and alertness of a human. Existing computational tools for the most used equipment fail to detect all fracture types and do not characterize a number of important occurrences on the borehole. This work presents an approach to help in the characterization of damages from ultrasound data. A number of techniques are combined with the amplitude and transit time data available to enable distinctions between fractures, stratifications, axial displacements, porous and breakouts. INSPECTION TOOL AND DEFECT IDENTIFICATION Borehole stability problems can lead from waste of effort up to loss of part of the well, greatly increasing drilling costs in the petrol industry [1]. The changes in the different layers done by sediment of rocks and borehole defects can be identified by ultrasound from tools such as the Ultrasonic Borehole Imager (UBI). A typical UBI acquisition consists of an amplitude data and a borehole radius deviation data. Figure 1 (a) shows sample of this data as images. Normally, the software tool presets them in false color, using hues of yellow and red with different saturations and intensity, where dark colors represent opposite things for amplitude and radius. It means black means low amplitude and large radii. Combinations of amplitude and radius can indicate borehole rugosity, enlargements of the radius or axial deviation among others. Three additional curves are included in the tool data on the default presentation: the minimum, average and maximum values of the measurement. These curves can be used by experts to quickly address intervals with borehole issues. Combinations of all these elements correspond to specific well problems. In order to illustrate this statement, breakouts, keyhole wears, shear displacements, fractures and rugosity are presented in Fig 1(b-d) as they can see in the UBI imaging, respectively. Combining information turns possible to reconstruct a cross section of the well at specific depths and, this combination is essential for the interpretation of well bore deformations. For example, increased horizontal stress can cause soil compression failure so rock can rupture and its fragments can induce cavitations and well enlargements [2]. This consequence is known as a breakout and can be seen in figure 1 (e). Another example occurs when the rotary drill pipe rests on a point of the well, gradually using this point more, as shown in Fig. 1 (f) and producing a smooth modification: called a keyhole wear [3]. When the displacement of the axial or bore center is significant, see Fig. 1 (g), it indicates shear displacements. Such occurrences affect the stability of the hole and The 9th International Conference on Structural Analysis of Advanced Materials - ICSAAM 2019 AIP Conf. Proc. 2196, 020031-1–020031-4; https://doi.org/10.1063/1.5140304 Published by AIP Publishing. 978-0-7354-1947-6/$30.00 020031-1 their identification is extremely important to avoid wall instability, but until the advent of image analysis from the data there was no effective way to identify them. Although, occurrences identification could be more simple in cross-sectional graphs (Fig 1, e to g), or on 3D (Fig 1 h), in images they appear as a combination of textures (based on the amplitude and radius of the original false color image, as can be seen in Fig. (a-d) making correct diagnosis of drill hole anomalies, using the default UBI visualization tool, very complex. Therefore, the implementation of UBI data processing tools is very important: our solutions to develop them are commented on in the next section. (a) (e) (b) (f) (c) (g) (d) (h) FIGURE 1. UBI image sample (a), some defect as image (b-d) and they in the borehole circular shape modification for breakout (e), keyhole wear (d), and shear displacement (g). IMPLEMENTED SOFTWARE TOOLS The UBI tool measures an ultrasonic wave reflected at the borehole wall. Amplitude and radius image data can be analyzed visually by specialized geologist but, the complexity of the analysis and difficulty of interpretation are greatly facilitated by using software such the set of programs implemented by our group to help on evaluating well stability. When using the images provided by the equipment (i.e. the commercial software) the false colors used can promote difficulty especially because the colors related to minimum and maximum values present opposite meaning. In our implementation we use directly the measured values and a correspondence among datum and gray level, that is the darker pixel represent the low value of the sensor and the higher value is represented by the lighter shade of gray. A sample of these values directly represented as gray level can be seen in Fig. 2 (a). In this sample there is an axial eccentricity, which can be noted after correction each image line by eliminating the low pass signals by Fast Fourier Transform (FFT) as presented in Fig. 2 (b). Figure 2 (c) shows the original image after processed by using Gaussian filter in each line. Porous and breakouts are important aspect to detect and count [4-6]. For these, the borehole data are transformed in a way that only the faults are in white as can be seen, in Fig. 3 (a). This is done by using the Otsu threshold algorithm, for binary transformation by finding an adequate level in the data [6]. The Otsu’s algorithm is an adaptive thresholding that consider for threshold definition the computation of between-class and within-class variances. With this threshold the grey colour image, Fig. 3 (b), is transformed down to black or up to white. Consecutively, the number of elements in the image for each size can be counted (named granulometric 020031-2 distribution of the results) as in Fig. 3 (c). In case de appearance of the gaps is inadequate the image can continue on processing by a morphological opening operation. Figure 3 (a) shows this option in case of use structural elements of circular shape and size 3 and 5. Moreover, break outs, keyhole wear and shear can be detected by considering K-Means Clustering, which is completely automatic process [4]. Figure 3 (d) shows experimentation for the same original raw image after 50 iterations considering 5 and 7 classes clustering and then using different color for each class (producing the two false color images showed). (a) (b) (c) FIGURE 2. Original image (a) and it after using FFT based filtering (b) and Gaussian filter (c). Porous and breakouts are important aspect to detect and count [4-6]. For these, the borehole data are transformed in a way that only the faults are in white as can be seen, in Fig. 3 (a). This is done by using the Otsu threshold algorithm, for binary transformation by finding an adequate level in the data [6]. The Otsu’s algorithm is an adaptive thresholding that consider for threshold definition the computation of between-class and within-class variances. With this threshold the grey colour image, Fig. 3 (b), is transformed down to black or up to white. Consecutively, the number of elements in the image for each size can be counted (named granulometric distribution of the results) as in Fig. 3 (c). In case de appearance of the gaps is inadequate the image can continue on processing by a morphological opening operation. Figure 3 (a) shows this option in case of use structural elements of circular shape and size 3 and 5. Moreover, break outs, keyhole wear and shear can be detected by considering K-Means Clustering, which is completely automatic process [4]. Figure 3 (d) shows experimentation for the same original raw image after 50 iterations considering 5 and 7 classes clustering and then using different color for each class (producing the two false color images showed). (b) (a) (c) (d) FIGURE 3. Original image (a) and it after Otsu threshold and opening by using circular structural elements of size 3 and 5. Histogram representation of the original image content (b). Granulometric distribution after threshold (c). UBI data presented in gray levels and its representation in 5 and 7 classes using false color (d). 020031-3 Fractures and beddings (stratifications or shallow fractures) are important issues but very similar in appearance on UBI amplitude or radius textures. Inclinations of the lines detected, as can be seen in Fig.4 (a), are used in order to make distinctions between fractures (red) and beddings (blue) [5]. Many of the previous idea are based on detection of harmonics forms by using Hought transform. These have been implemented as well; see Fig.4 (b, c). (a) (b) (c) FIGURE 4. Line detected grouped by its angles (a) Hough transform for fractures (b) and bedding (c). CONCLUSIONS In this work, possible analyses by UBI (i.e. considering ultrasound) are presented. Although, the grades related to the improvement (in inspection correctness rates and time) achieved with the developed tool are confidential, most of the development algorithms were very well evaluated by the interested users and are now included in inspection programs. In order to promote continuous improvement on such systems we are constantly considering new ways to analysis the UBI data and combine theirs aspects. Despite the fact that breakouts, keyhole wear and shear be almost uniquely detectable by ultrasound, there are other cheaper tools such as the Oil Base Micro Imager (OBMI) and full bore Formation Micro Imager (FMI), based on micro resistivity imaging available for the oil industry [7]. Therefore, considering that, other types of physical information can improve the characterization of soil damage from digital data a fusion of information is the main objective of next developments. ACKNOWLEDGMENTS UBI, FMI and OBMI and commercial products of ©Schlumberger. A.C. is partially supported by FAPERJ (CNE) and CNPq Brazilian Agency projects Pq. 305416/2018-9 and Univ. 402988/2016-7. M.B.H and J.S.C. thank CAPES for the financial support. E.W.C. partially supported by Petrobras and CNPq Brazilian Agency. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. Erge, O., Karimi Vajargah, A., Ozbayoglu, M.E. and van Oort, E., Frictional pressure loss of drilling fluids in a fully eccentric annulus”. J. Natural Gas Science Engineering, 26, pp. 1119-1129 (2015). GhasemiKafrudi, E. and Hashemabadi, S., Numerical study on cuttings transport in vertical wells with eccentric drillpipe, Journal of Petroleum Science and Engineering, 140, 85-96 (2016). N. F. Hurley and T. Zhang, “Method for characterizing a geological formation traversed by a borehole,” Oct. 22 2009, US Patent App. 12/384,945. Hamerly, G.; Drake, J. Accelerating Loyd's algorithm for k-means clustering. Partitional Clustering Algorithms. pp. 4178 (2015). M. Macedo A. Conci, On the Detection of Generic Conic Form Parameters Using Hough Transform. In: Procd, SIBGRAPI 2007. N. Otsu, A threshold selection method from gray-level histograms. IEEE Trans. Sys., Man., Cyber. 9 (1): 62-66 (1979). Stanier, S. S.; Blaber, J.; Take, W. A.; White, D. J. Improved image-based deformation measurement for geotechnical applications. Canadian Geotechnical Journal, v.53, n. 5, 2016. Schlumberger Inc., Advanced borehole imaging independent of mud type, ©Schlumberger, SMP-5871, (2002) https://www.slb.com/ . 020031-4