RU2444031C2 - Method of generating numerical pseudocores using borehole images, digital rock samples, and multi-point statistics - Google Patents

Abstract

FIELD: physics. ^ SUBSTANCE: methods of creating a numerical pseudocore model involving: a) obtaining logging data from a reservoir having depth-defined intervals of the reservoir, and processing the logging data into at least one interpretable borehole image data having unidentified borehole image data; b) examining one of the interpretable borehole image data, other processed logging data or both to generate the unidentified borehole image data, processing the generated unidentified borehole image data into the interpretable borehole image data to generate warped fullbore image data; c) collecting one of a core from the reservoir, the logging data or both and generating a digital core data from one of the collected core, the logging data or both such that generated digital core data represents features or structures of one or more depth-defined interval of the reservoir; and d) processing generated digital core data, interpretable borehole image data or the logging data to generate realisations of the numerical pseudocore model. ^ EFFECT: creation of three-dimensional numerical cores from computed X-ray tomography data and formation micro-image logs, and performing flow modelling in these numerical cores to understand fluid-flow paths and recovery factors in the selected reservoir. ^ 32 cl, 21 dwg

Description

BACKGROUND

FIELD OF THE INVENTION

The invention generally relates to a method that uses the underlying Multipoint Statistics (MPS) algorithm to generate numerical pseudocores from digital rock or core images and well image diagrams. More specifically, the present description of the invention relates to the creation of three-dimensional (3D) numerical cores according to computer x-ray tomography (CT scans, CT tomograms) and the formation of microimages (FMI) of the geological section of the well, and performing flow modeling in these numerical cores to understand the routes fluid flow and return coefficients in the selected manifold.

State of the art

Electrical and acoustic instruments for imaging the well are widely used for geophysical studies in underground wells, to identify and map the boundaries between the layers of the rock, for example the boundaries of the layers, and to visualize and orient faults and faults. Since electrical instruments for geophysical research are shoe-type devices with a fixed arrangement of electrodes, gaps with a lack of information between the measuring shoes are a common place. Electrical and acoustic diagrams typically have intervals with poor data quality due to non-functioning electrodes, insufficient shoe pressure, heterogeneities, cuttings, deviations from centering of instrumentation, or poor acoustic reflections.

Digital rock models are built from two-dimensional (2D) thin sections, images obtained using a scanning electron microscope (SEM), computer-generated packages of spheres, images obtained using a scanning laser confocal microscope, and various types of images obtained by computerized tomography ( CT), for example, traditional computer microtomograms and synchrotron computer microtomograms. Computed tomography images (CT, X-ray tomograms) represent the most common approach. Computed x-ray tomograms are two-dimensional (2D) cross sections obtained using an x-ray source that rotates around a sample. The density is calculated from the attenuation coefficients of the x-ray radiation. Tomograms from a series of cross sections are used to construct three-dimensional (3D) images of the sample. Due to the high density contrast between the rock and the pores filled with fluid, images of computer tomograms can be used to visualize the rock-pore system. Resolution varies from submillimeter to micron scale, depending on the device used.

Multipoint statistics methods (MPS) are used to create model simulations of spatial fields of properties of geological structures and reservoirs for modeling an underground reservoir. These methods are conditional modeling, for which they use known results, such as those measured in wells and called fixed or “solid” data, which are given absolute preference in modeling. In multipoint statistics (MPS), one-dimensional (1D), two-dimensional (2D) or three-dimensional (3D) “training images” are used as quantitative templates for modeling property fields of an underground object. Simulation using Multipoint Statistics (MPS) extracts geological structures from training images and attaches them to data locations. These structures can either a priori be geological interpretations, or they can be conceptual models.

Multipoint Geostatistics (MPS) is a new progressive geostatistical approach. It allows specialists modeling the collector to introduce their previous experience, interpretations or conceptual models into the process of modeling the reservoir using training images. These training images are numerical images of structures / features that appear to exist in the reservoir under study. When training images are available, multipoint geostatistics (MPS) can extract curved structures or complex features from training images and tie them to the collector locations where the samples / observations are collected, leading to more realistic reservoir models. The introduction of training images in reservoir modeling is a reference point in the design process. It should be noted that there are two components in the application of multipoint geostatistics (MPS): training images (conceptual models) and real data (observations). The two parts are typically separate. However, in a practical application, the generation of demonstrative training images, especially in three-dimensional (3D) representation, turned out to be a bottleneck in the implementation of multipoint geostatistics (MPS) methods. Generating a continuously variable training image is even more difficult than creating a unique training image.

There are various types of electrical and acoustic instruments for scanning wells used for geophysical exploration of wells to determine the location and map the boundaries between the layers of the rock, for example, the boundaries of the layers, and to visualize and orient faults and folds.

For example, an electrical scan of a well can be carried out in a water-based drilling fluid (electrically conductive), for example, using a Schlumberger reservoir microscanner (FMI), which is based on inclinometry technology that has become commercially available since the late 1950s x years. Equipment for electric scanning of wells is essentially a downhole tiltmeter complex design. Scanning devices have microresistive electrodes placed along the perimeter of the well on shoes that are pressed against the wall of the well. The evolutionary trend from reservoir dipsticks to well images manifested itself in the transition from several electrodes to the complex layout of electrodes on numerous shoes. See Hurley, NF, 2004, Borehole Images, in Asquith, G., and Krygowski, D .: and see Basic Well Log Analysis, 2nd Edition, AAPG Methods in Exploration Series (Methods of the American Association of Petroleum Geologists in Surveying), No. 16, pp. 151-164. The tools are first lowered into the well with the shoes closed. At the beginning of logging, either four, six, or eight shoes are pressed against the well wall. The number of shoes depends on the device for logging. Electric current is sent to the rock through the electrodes, and sensors measure the current after it interacts with the formation. Raw data includes multiple readings from electrodes, a caliper reading from individual shoes or pairs of shoes, and accelerometer and magnetometer readings along the x, y, and z axes. The angle of inclination of the wellbore and the orientation of the shoe 1 (device) are determined by magnetometers. The sampling frequency for the electrode and accelerometer readings is very high, typically 120 counts / foot (400 counts / meter).

The surface area of the well is a function of the width of the electrode blocks, the number of shoes and the diameter of the well. In general, in typical wells, readings are taken to obtain an image of 40 to 80% of the surface of the well. The missing parts of the well look like white stripes between the shoes.

Images of wells are created by correlating color maps to various depth areas or ranges of resistivity values. Then the color pixels are placed in their proper geometric position around the well. By agreement, areas of low resistivity, such as shale or fluid-filled fractures, are shown in dark colors. High resistivity areas such as sandstones and limestones show shades of brown, yellow, and white.

Two main types of processed well images are available: static and dynamic. Static images are those that have one contrast setting that applies to the entire well. They give informative ideas about the relative changes in the resistivity of the rocks throughout the well. Dynamic images, which have a variable contrast, which is placed in the movable window of the program, provide an improved overview of such features as cavities, cracks and layer boundaries. Dynamic images reveal subtle structural features of the rock that have very low resistivity values, such as shales, and very high resistivity values, such as carbonates and crystalline rocks.

Another example of electrical images of wells can be implemented in oil-based drilling mud (non-conductive), in particular when high resistivity values of drilling fluids (higher than 50 ohm · m) typical of oil-based drilling fluids are unsuitable for many electrical images of wells. Since 2001, for the oil-based drilling fluids, Schlumberger's Micro-Well Well Filler for Oil-Based Drilling Fluids (OBMI) has been available on the market. This device generates images of the well by passing an electric current into a formation of two large electrodes on each shoe, which has a high voltage (about 300 Volts). On each of the 4 shoes there is a series of closely spaced disk electrodes installed in two rows of five pieces. Well images are generated from the potential difference (voltage drop) between closely spaced electrodes. Wide gaps corresponding to unscanned sections of the well are common between shoes. Another aspect of well images can be realized while drilling, for example, “logging while drilling”, or hereinafter referred to as “LWD”. Examples of Schlumberger logging equipment are GeoVision Resistivity (GVR) and Azimuthal Density Neutron (ADN) (azimuthal density neutron sensor). The GVR resistive device uses rotating electrodes and the work is done in a water-based drilling mud. The AND device generates images based on azimuthal density readings and works in any drilling fluid. When the device rotates during drilling, a complete scan of the entire surface of the well is obtained, without gaps.

Another aspect of well images may be acoustic images of wells, also known as downhole television cameras, which are based on technology first developed in the 1960s. See an article by Zemanek, J., Glenn, EE, Norton, LJ, and Caldwell, RL, 1970, Formation evaluation by inspection with borehole televiewer: Geophysics, Volume 35, p. 254 -269.

The Ultrasonic Downhole Scanner (UBI) is Schlumberger's premier acoustic instrument for open hole applications. The UBI device, which is placed centered on the axis of the well, has a rotating transducer that emits and detects sound waves that are reflected from the wall of the well. The image is formed by recording and processing both the amplitude and the transit time of the acoustic signal. Typically, well coverage reaches 100%, with no gaps in the images. However, poor quality images may be obtained when the device is not centered, or the borehole wall is not uniform.

As characteristic indicators in the pictures of well images, among other things, petrophysical facies, such as voids, and spots of resistivity and conductivity can appear. In particular, the authors of Dehghani et al. 1999 surveys suggest that there are zones of increased porosity and permeability near caverns. See an article by Dehghani, K., Harris, P.M., Edwards, K.A. and Dees, W.T., 1999, Modeling a vuggy carbonate reservoir (AAPG Bulletin, Volume 83, pp. 19-42.

Authors Dehghani et al. (1999) used thin sections, images obtained using a scanning electron microscope (SEM), and mini-permeability measurements of the rock to confirm their concept. Schindler (2005) and Tanprasat (2005) used image analysis from core photographs made using fluorescent paints to show that clusters of small caverns mainly exist near large cavities. See papers by Schindler, J., 2005, Quantification of vuggy porosity, Indian Basin field, New Mexico (Quantitative Cavernous Porosity, Indian Basin Field, New Mexico): unpublished master's thesis, Colorado School of Mines (Colorado School of Mines), Golden, Colorado; and Tanprasat, S., 2005, Petrophysical analysis of vuggy porosity in the Shu'aiba Formation of the United Arab Emirates: Unpublished Master's thesis, Colorado School of Mines, Golden, Colorado. Such small caverns turn out to be below the resolution level of the device for creating images of the well, so that they appear more in shaded areas than in the form of discrete cavities in the geological section. If this is a common place for cavernous carbonate rocks, the electrical and acoustic images of the wells should have zones or halos of high conductivity or low amplitude (dark) near the cavities. Indeed, this feature is usually observed, for example, as shown in FIG. 3. High conductivity zones, surrounding caverns, and areas of increased small-scale porosity, known as conductivity spots, form the basis for part of the Schlumberger BorTex software package cited by Russel et al. (2002) and Hassall et al. (2004). See authors Russell, SD, Akbar, M., Vissapragada, B., and Walkden, GM, 2002, Rock types and permeability prediction from dipmeter and image logs: Shuaiba reservoir (Aptian), Abu Dhabi (Rock types and permeability prediction by readings of the tiltmeter and pictures of the geological section: Shuayba field (Aptian), Abu Dhabi): AAPG Bulletin, Volume 86, pp. 1709-1732; and see the work of Hassall, JK, Ferraris, P., Al-Raisi, M., Hurley, NF, Boyd, A. and Allen, DF, 2004, Comparison of permeability predictors from NMR, formation image and other logs in a carbonate reservoir (Comparison of permeability predictions from nuclear magnetic resonance data, imaging and other core images in a carbonate reservoir): Preprint 88683 of the Society of Petroleum Engineers (SPE) presented at the 11th International Petroleum Exhibition and Conference, Abu Dhabi, United Arab Emirates, 10-13 October.

Delhomme (1992) demonstrated the importance of mapping electrically resistive and non-resistive areas in well images. See author, Delhomme, JP, 1992, A quantitative characterization of formation heterogeneities based on borehole image analysis: Trans. 33 ^rd Symposium SPWLA (Proceedings of the 33rd Symposium of the Society of Professional Downhole Logging Analysts), article T. However, his approach was unsatisfactory due to the gaps between the shoes. He was unable to draw clear contours around areas with high or low resistivity due to the uncertainty of the profiles. Full circular images of the borehole wall (FIG. 4) allowed the authors of the present invention to draw clear contours around resistive and / or non-resistive regions in the borehole images. Such regions gave important measures of heterogeneity, especially in carbonate reservoirs. These areas are generally much larger than the digital images of the rock or cores that the inventors have generated, for example, from CT tomograms of the rocks. As a result, the inventors of the present invention required well images to determine whether they can count on detecting inhomogeneities from decimeter to meter in their flow models.

Areas with characteristic features in the images of the geological section of the well, such as caverns, and spots of resistivity and conductivity are here referred to as petrophysical facies. Other authors, such as Leduc et al. (2002) and Mathis et al. (2003), referred to as the related to the structure of the field of electrofacies. See the work of the authors Leduc, JP, Delhaye-Prat, V., Zaugg, P., and see the work of the author Mathis, B., 2002, FMI * based sedimentary facies modeling, Surmont Lease (Athabasca, Canada) (abs.) (Modeling sedimentary facies based on the readings of a reservoir electric microscanner, Surmont Lease field (Lake Athabasca, Canada) (abstract): CSPG Annual Convention, Calgary, Alberta, Canada, 10 pp .; and see the work of the authors Mathis, B., Leduc, JP and Vandenabeele , T., 2003, From the geologist's eyes to synthetic core descriptions: Geological log modeling using well-log data (abs.) (Descriptions of synthetic cores from a geologist's point of view: modeling a geological section using data x well logging (abstract)): AAPG Annual Meeting (Annual Seminar of the American Association of Petroleum Geologists (AAPG), Salt Lake City, Utah, 7 p.

To determine the petrophysical facies, structural features represented by various colors, for example, black, brown, and white (FIG. 4), could be used. Such facies have complex three-dimensional (3D) shapes. Conductivity spots, if any, are zones with increased porosity and permeability, and form areas of continuous flow between the caverns.

There are many examples in the published literature of numerical images of rocks constructed using methods (or digital rock models for rocks and pores) that include reconstructions made from two-dimensional (2D) thin sections or images obtained using a scanning electron microscope (SEM), of electrofacies derived from well logs generated on a computer of spherical packages, laser scanning confocal microscopy data and various types of CT tomograms (traditional, micro CT tomograms and si synchrotron computerized mikrotomogramm).

Authors Bakke and Oren (1997), Oren et al. (1998) and Oren and Bakke (2002) developed a method by which three-dimensional (3D) mesh structures are constructed from thin two-dimensional (2D) slices. Using the “Numerical Rocks” method (see the website http://www.numericalrocks.com/), three-dimensional (3D) pore models are calculated from thin two-dimensional (2D) sections. See authors Bakke, S., and Oren, P.-E., 1997, 3-D pore-scale modeling of sandstones and flow simulations in the pore networks (Three-dimensional modeling of pore scale in sandstones and flow simulations in pore network structures) : Preprint 35,479 of the Society of Petroleum Engineers (SPE), European 3-D Reservoir Modeling Conference (Stavanger, Norway, April 16-17, pp. 136-149; Oren, P.-E., Bakke, S., and Arntzen, O.J., 1998, Extending predictive capabilities to network models: SPE Journal, Volume 3, p. 324; and Oren, P.-E. and Bakke, S., 2002, Process based reconstruction of sandstones and prediction of transport properties: Transport in Porous Media, Volume 46, pp. 311-343. This company also uses pore models built from micro-CT scans. Authors Bakke et al. (2002) successfully applied this method for white crystalline dolomites. The articles by Duey (2008) and Suicmez and Touati (2008) summarize the results of a variety of pore network structures in sandstones processed using the Numerical Rocks method. See author's work. Duey, R., 2008, Quick analysis answers Heidrun question: Hart Energy Publishing, LP, available on the Internet at http://www.eandp.info/index2 .php? area = article & articleId = 767, March 27, 2008; and authors Suicmez, V.S. and Touati, M., 2008, Pore network modeling: A new technology for SCAL predictions and interpretations: A new approach for forecasting and interpretation using the interactive core analysis program (SCAL): Saudi Arabia Oil and Gas, issue 5, p. 64-70. Authors Wu et al. (2006) presented a method for generating three-dimensional (3D) numerical models of rocks from two-dimensional (2D) thin sections using a third-order Markov network. See article by Wu, K., Van Dijke, MIJ, Couples, GD, Jiang, Z., Ma, J., Sorbie, KS, Crawford, J., Young, I. and Zhang, X., 2006, 3D stochastic modeling of heterogeneous porous media - Applications to reservoir rocks (Three-dimensional stochastic modeling of an inhomogeneous porous medium - Application to reservoir rocks): Transport in Porous Media, Volume 65, pp. 433-467. Awwiller (2007) developed a method that models more complex sandstones than those described by Oren and Bakke (2002). Patent application US 2007/0203677 A1 (below) by Awwiller (2007) relates to this work. Authors Okabe and Blunt (2004, 2005) generated three-dimensional (3D) images from two-dimensional (2D) thin slices using multi-point statistics. See authors Okabe, H., and Blunt, MJ, 2004, Prediction of permeability for porous media reconstructed using multiple-point statistics: Physical Review E, Volume 70, p.066135- 1-10; and see an article by Okabe, H. and Blunt, M.J., 2005, Pore space reconstruction using multiple-point statistics: Journal of Petroleum Science and Engineering, Volume 46, pp. 121-137. The authors of Tomutsa and Radmilovic (2003) used ion beam pinching to create multiple two-dimensional (2D) consecutive sections, which they used to build three-dimensional (3D) pore models of submicron scale. See authors Tomutsa, L., and Radmilovic, V., 2003, Focused ion beam assisted three-dimensional rock imaging at submicron scale (International Symposium of the Soc. of Core Analysis, Pau, France, September 21-24, article SCA2003-47.

Authors Dvorkin et al. (2003) described Digital Rock Physics technology, which consists of pore-scale numerical models derived from: (a) two-dimensional (2D) thin sections and a model of statistical indicators, or (b) CT tomograms . See authors Dvorkin, J., Kameda, A., Nur, A., Mese, A. and Tutuncu, AN, 2003, Real time monitoring of permeability, elastic moduli and strength in sands and shales using Digital Rock Physics (Monitoring in real-time permeability, elastic modulus and strength in sands and shales using Digital Rock Physics): preprint 82246 of the Society of Petroleum Engineers (SPE), presented at the European Conference of the Society of Petroleum Engineers (SPE) on disturbances in the performance of reservoirs, The Hague, Netherlands, May 13-14, 7 pp. They built three-dimensional (3D) hypothetical models rock and conducted flow modeling using the method of lattice Boltzmann equations. US Pat. No. 6,516,080 (below) relates to this work.

The authors of Leduc et al. (2002) and Mathis et al. (2003) (both cited above) generated “synthetic cores” from a limited number of described cores, traditional open-hole logs, and images of geological sections of wells. In traditional open-hole logs, cluster analysis was used. Electrofacies, which are depth intervals based on well logs with similar lithology, are determined using traditional open-hole logs and structural analysis of well images. Hypothetical cores were calculated on a computer for wells from which cores were not taken using "factor tables". United States Patent 6011557 (below) refers to this work.

Authors Vahrenkamp et al. (2008) described mini-models, that is, models of collectors whose dimensions are less than 1.0 m ³ and give pseudo-properties for volume sensors in models on a collector scale. See authors Vahrenkamp, VC, Creusen, A., Tull, S., Farmer, A., Mookerjee, A. and Al Bahry, A, 2008, Multi-scale heterogeneity modeling in giant carbonate field, northern Oman (abs.) (Multiscale Simulation of Inhomogeneity in a Giant Carbonate Field, Northern Oman (abstract): GeoArabia, Volume 13, No. 1, p. 248. Mini-models are filled using "fundamental rock types" (PRT), which "cover and classify the entire range of types pore size, pore size, pore size distribution, capillary filling pressures, relative permeability values, etc. ”Foundational rock types (PRT) are organized in “Rock Type Associations” (RTAs), which are based on “sedimentary tissue” derived from images of the well’s geological section. Rock type associations (RTAs) are distributed in the reservoir using images of the well’s geological section and observed lamination, facies models, and seismic data. .

Authors Bryant et al. (1993) and Behseresht et al. (2007) described digital rock models, which are computer-generated dense packings of spheres with random periodicity. See authors Bryant, S., Mellor, D., and Cade, C., 1993, Physically representative network models of transport in porous media: American Institute of Chemical Engineers Journal, Volume 39, No. 3, pp. 387-396; and see the work of the authors Behseresht, J., Bryant, S.L. and Sepehrnoori, K., 2007, Infinite-acting physically representative networks for capillarity-controlled displacements: Preprint 110581 of the Society of Petroleum Engineers (SPE) presented at the Society of Engineers Annual Technical Conference and Exhibition -nefteyanikov (SPE), Anaheim, California, November 11-14, 15 pp. Other authors, such as Bosl et al. (1998) and Holt (2001) generated similar digital rock models for flow experiments. See authors Bosl, WJ, Dvorkin, J. and Nur, A., 1998, A study of porosity and permeability using lattice Boltzmann simulation: Geophysical Research Letters, Volume 25 pg. 1475-1478; and see the author’s work Holt, R.M., 2001, Particle vs. laboratory modeling in in situ compaction (Comparison of laboratory simulation): Physics and Chemistry of the Earth, Part A: Solid Earth and Geodesy, Volume 26, Issue 1-2, pp. 89-93.

Authors Fredrich et al. (1995) and Fredrich (1999) created three-dimensional (3D) images of rocks using laser scanning confocal microscopy. See authors Fredrich, J.T., Menendez, B. and Wong, T.F., 1995, Imaging the pore structure of geomaterials: Science, vol. 268, pp. 276-279; and see the author's work Fredrich, JT, 1999, 3D imaging of porous media using laser scanning confocal microscopy with application to microscale transport processes (Physics and Chemistry: Physics and Chemistry of the Earth, Part A: Solid Earth and Geodesy, Volume 24, Issue 7, pp. 5151-561. Authors O'Connor and Fredrich (1999) conducted flow experiments on these numerical rock images using the methods of lattice Boltzmann equations. See Authors O'Connor, R.M. and Fredrich, J.T., 1999, Microscale flow modeling in geologic materials: Physics and Chemistry of the Earth, Part A: Solid Earth and Geodesy, Volume 24, Issue 7, pp. 611-616.

In the most common approach, various types of CT tomograms are used to generate pore network structures. The authors are Vinegar (1986), Wellington and Vinegar (1987) and Withjack et al. (2003) summarized the methodology and discussed various applications of X-ray computed tomography. See author's work Vinegar, H.J., 1986, X-ray CT and NMR imaging of rocks (X-ray computed tomography and nuclear magnetic resonance imaging): JPT (Journal of Petroleum Technology), pp. 257-259; see authors work Wellington, S.L. and Vinegar, H.J., 1987, X-ray computerized tomography: JPT, pp. 885-898; and see authors work Withjack, EM, Devier, C. and Michael, G., 2003, The role of X-ray computed tomography in core analysis: Preprint 83467 of the Society of Petroleum Engineers (SPE) Presented at the Joint Session of the Western Regional Section of the Society of Petroleum Engineers (SPE) and the Pacific Section of the American Association of Petroleum Geologists (AAPG), Long Beach, CA, May 19-24, 2003, 12 pp. Authors Siddiqui and Khamees (2005) and Siddiqui et al. (2005) emphasized the use of three-dimensional (3D) images of cores and fragments of cuttings obtained using traditional and micro-CT tomograms. See the work by Siddiqui, S. and Khamees, AA, 2005, Data visualization challenges for displaying laboratory core and flow data in three-dimensions. (Preprint 106334 Society of Engineers- Oil Workers (SPE), presented at the Technical Symposium of the Society of Petroleum Engineers (SPE) in Saudi Arabia, May 14-16, 9 pp .; and see the work by Siddiqui, S. and Khamees, AA, 2005, Data visualization challenges for displaying laboratory core and flow data in three-dimensions (Three-Dimensional Visualization of Laboratory Core Data and Flows): Preprint 106334 of the Society of Engineers Oil Workers (SPE), presented at the Technical Symposium of the Society of Petroleum Engineers (SPE) in Saudi Arabia, May 14-16, 9 pp. Authors Coles et al. (1996), Fredrich et al. (2006) and Fredrich et al. (2007) used synchrotron computer microtomography to build numerical three-dimensional (3D) models of pore network structures in natural and synthetic sandstones. See papers by Coles, M.E., Hazlett, RD, Muegge, RL, Jones, KW, Andrews, B., Dowd, B., Siddons, P., Peskin, A., Spanne, P., and Soll, WE , 1996, Developments in synchrotron X-ray microtomography with applications to flow in porous media: Preprint 36531 of the Society of Petroleum Engineers (SPE) presented at the Annual Technical Conference and Exhibition of the Society of Engineers Oil Workers (SPE), Denver, Colorado, pp. 413-424; see authors Fredrich, J.T., DiGiovanni, A.A. and Noble, D.R., 2006, Predicting macroscopic transport properties using microscopic image data: Journal of Geophysical Research B: Solid Earth, Volume 111, Issue 3; and see the work of the authors Fredrich, J.T., Haney, M.M. and White, JA, 2007, Predicting petrophysical properties using 3D image data (abs.). website http://www.aapg.org. To model permeability, they used the methods of lattice Boltzmann equations.

Multipoint (or multiple-point) statistics (MPS) methods are a new family of spatial statistical interpolation algorithms proposed in the 1990s that are used to generate conditional models of fields of discrete variables, such as geological facies. See the work by Guardiano, F. and Srivastava, R.M., 1993, Multivariate geostatistics: beyond bivariate moments (Geostatistics-Troia, edited by A. Soares. Dordrecht, The Netherlands, Kluwer Academic Publications, Volume 1, pp. 133-144. A training image is a numerical model of a previously geological structure that contains facies structures and relationships that are assumed to exist in real reservoirs. Training images are conceptual in nature and can be as simple as a hand-drawn map, or they can be created using computer equipment. The original algorithm of multipoint statistics (MPS), proposed by the authors of Guardiano and Srivastava (1993), constructs a multipoint distribution function of conditional probability (CPDF) by scanning the training image again and again for each nodal point of the model. Due to limitations due to the central processing unit of the computer (CPU), this time-consuming algorithm is not practiced now.

The author of Strebelle (2002) introduced the concept of the enumeration tree, which saves all replicas of paintings found within the template throughout the training image. See the author’s work Strebelle, S., 2002, Conditional simulation of complex geological structures using multiple point statistics. Mathematical Geology, Volume 34, pp. 1-22. The fundamental algorithm of multipoint statistics (MPS), proposed by Strebelle (2002), called SNESIM, was used in many cases to simulate reservoirs and became a kind of reference apparatus for modeling river channel sediments, being combined with rotational and affine transformations (Zhang, 2002; Caers and Zhang, 2004; Strebelle and Zhang, 2004). See author's paper Zhang, T., 2002, Multiple-point simulation of multiple reservoir facies: Unpublished Master's thesis, Stanford University, California, 163 pp.; see authors Caers, J. and Zhang, T., 2004, Multiple-point geostatistics: A quantitative vehicle for integration of geologic analogs into multiple reservoir models, in the book Integration of Outcrop and Modern Analogs in Reservoir Modeling, edited by M. Grammer, PM Harris and G.P. Eberli, AAPG, Article 80, pp. 383-394; and see the work by Strebelle, S. and Zhang, T., 2004, Non-stationary multiple-point geostatistical models, in the book Geostatistics, edited by Leuangthong, O. and Deutsch, CV, vol. 1, pp. 235-244.

The fundamental algorithm of multipoint statistics (MPS) is orders of magnitude faster than the original algorithm of the authors Guardiano and Srivastava (1993), but it requires the use of a large amount of random access memory (RAM) of the computer, especially in three-dimensional (3D) representation for large training image. See the work by Guardiano, F. and Srivastava, R.M., 1993, Multivariate geostatistics: beyond bivariate moments (Geostatistics-Troia, edited by A. Soares. Dordrecht, The Netherlands, Kluwer Academic Publications, Volume 1, pp. 133-144. This limitation in terms of random access memory (RAM) in three-dimensional (3D) representations requires a compromise solution, which can lead to improper reproduction of the shape of three-dimensional (3D) objects. The limitation of random access memory (RAM) also prevents the authors of the present invention from considering too many categories or classes at the same time, thereby limiting the underlying multipoint statistics (MPS) algorithm by modeling single-valued variables.

In order to handle both unambiguously and continuously variable training images, and to reduce the consumption of random access memory (RAM) and improve form reproduction in three-dimensional (3D) applications, a multipoint statistics (MPS) algorithm such as FILTERSIM (Zhang 2006a) was created. See author's paper Zhang, T., 2006a, Filter-based training image pattern classification for spatial pattern simulation: Unpublished Master's thesis, Stanford University, California, 153 pages. FILTERSIM Algorithm applies a set of local filters to the training image, which can be either single-valued or continuous, for grouping local images into image classes. Then, image modeling is performed based on this classification. The filter is a local template (window) with a set of densities associated with the location on the template of each pixel. The application of the filter to the local image results in an estimated filter metric, the estimated metric looks like a numerical summary of this local image. A set of filters by default or specified by the user is formed so that each filter can register various aspects of the training image that are visible within the template. These filters are used to transform the training image into the space of filter evaluations. This quantification of the image reduces image size. By dividing this estimated space with a limited size, similar instructional patterns are sorted based on their filter rating indicators.

The fundamental algorithm of multipoint statistics (MPS) is orders of magnitude faster than the original algorithm of the authors Guardiano and Srivastava (1993), but it requires the use of a large amount of random access memory (RAM) of the computer, especially in three-dimensional (3D) representation for large training image. This limitation in terms of random access memory (RAM) in three-dimensional (3D) representations requires a compromise solution, which can lead to improper reproduction of the shape of three-dimensional (3D) objects. The limitation of random access memory (RAM) also prevents the authors of the present invention from considering too many categories or classes at the same time, thereby limiting the underlying multipoint statistics (MPS) algorithm by modeling single-valued variables. The fundamental multipoint statistics (MPS) algorithm identifies exact replicas of a data transformation event, builds a collector model one pixel at a time, converts multipoint data into an event, and does not allow any filtering or averaging of the images identified in the training image.

In order to handle both unambiguously and continuously variable training images, and to reduce the consumption of random access memory (RAM) and improve form reproduction in three-dimensional (3D) applications, a multipoint statistics (MPS) algorithm such as FILTERSIM (Zhang 2006a) was created. The FILTERSIM algorithm applies a set of local filters to a training image, which can be either single-valued or continuous to group local images into image classes. Then, image modeling is performed based on this classification. The filter is a local template (window) with a set of densities associated with the location on the template of each pixel. The application of the filter to the local image results in an estimated filter metric, the estimated metric looks like a numerical summary of this local image. A set of filters by default or specified by the user is formed so that each filter can register various aspects of the training image that are visible within the template. These filters are used to transform the training image into the space of filter evaluations. This quantification of the image reduces image size. By dividing this evaluation space with a limited amount, similar training templates are sorted based on their filter rating indicators.

The FILTERSIM algorithm starts with sorting local training images in a filter evaluation space of reduced size. The simulation is carried out along a sequential path in the modeling space by determining which class of images is most similar to a local data transformation event, extracting a specific image from the image class, and then patching the quantized picture into an image on modeling sites. A random modeling trajectory and selection of image classes provides the possibility of various model implementations, nevertheless, all of them are due to the same initial data. Due to the reduction in size provided by filter summarization of each image, and due to the grouping of images into classes, the algorithm is fast and rational in terms of RAM conditions.

The underlying multipoint statistics algorithm (MPS) and the FILTERSIM algorithm are capable of processing absolute, or so-called “hard” boundary conditions from data obtained in wells or rock outcrops, and conditional or “soft” boundary conditions from seismic data, facies probability fields and rotational and affine (or large-scale) transformations of conditional coordinate grids. All these data are used in the process of stochastic modeling to generate one-dimensional (1D), two-dimensional (2D) or three-dimensional (3D) maps of geological facies or rock characteristics. Since there is a random component included in multi-point statistical (MPS) modeling, the individual implementations of the property fields created by the multi-point statistics (MPS) algorithms are different, but many implementations provide geologists and oil and gas field development engineers with improved quantitative estimates of the spatial distribution and uncertainties of geological facies in the simulated reservoir volume. Moreover, these algorithms accept both hard and soft input boundary conditions for processing (Zhang, 2006a). See authors Zhang, T., Switzer, P., and Journel A., 2006b, Filter-based classification of training image patterns for spatial pattern simulation: Mathematical Geology, Volume 38, p. .63-80.

There are six default directional two-dimensional (2D) filters that are commonly used in the FILTERSIM algorithm (Zhang, 2006a; Zhang et al., 2006b) (both cited above). There are three types of filters: averaging filter, gradient filter and curvature filter, and each type of filter is used for both horizontal and vertical directions. Averaging filters are focused on the localization of signs; gradient filters are used to identify the boundaries of signs by highlighting the contrast of various signs (first-order difference); filters of curvature reveal the difference of the second order of features.

To reflect the large-scale structure, multigrid modeling is used. This gradually models each level of the multiple grid from a larger to a smaller resolution with a finer model grid limited by previous model values obtained for rougher grids. At each simulation level, zoomed filters are applied that are superimposed on the corresponding grid (Zhang, 2006b).

There are two types of training images: one with a very limited number of categories, and another with continuous variables, such as the petrophysical characteristics of the reservoir. Multipoint geostatistical methods require the use of one-dimensional (1D), two-dimensional (2D) and three-dimensional (3D) grids of training images as primary conceptual geological models that contain pictures of the studied spatial parameters. The forms of various structural features appearing in the images are supposed to represent a model of real geological structures, with each category typically representing different geological facies or different types of geological bodies. It is usually required that the training images contain “stationary” pictures, that is, the pictures must be independent of their location in space (invariant with respect to any movement) and must be periodically repeated in the space of the training image. In the case of training images used for geological modeling, this stationarity may consist, but is not limited to, in the stationarity of the orientation of the geological object (where directed objects / structural features do not rotate within the image) and stationarity of the geological scale (where the size of objects / features in the image does not change within the image) (Caers and Zhang, 2004). See authors Caers, J. and Zhang, T., 2004, Multiple-point geostatistics: A quantitative vehicle for integration of geologic analogs into multiple reservoir models, in the book Integration of Outcrop and Modern Analogs in Reservoir Modeling, edited by M. Grammer, PM Harris and G.P. Eberli, AAPG, article 80, pp. 383-394.

A problem implicitly associated with these Multipoint Statistics (MPS) algorithms is how to generate training images. It is assumed that training images model or reproduce real geological structural features and should be as far as possible derived from existing geological interpretable images. This requires research by methods of statistics and image processing that would allow the use of images from any source, for example hand-drawn sketches, aerial reconnaissance photographs, satellite photographs, seismic volumes, models of geological objects, physical scale models or models of developing geological processes. Compared with the creation of continuously variable training images, the generation of uniquely (categorically) variable training images is simpler. To generate training images with unambiguous variables, an approach based on real objects is usually used. An approach based on the characteristics of the region, combined with the added desirable boundary conditions, can be used to generate training images with continuous variables (Zhang et al., 2005). See authors Zhang, T., Nair, N., Jennings, J.W. and Bryant, SL, 2005, Models and methods for determining transport properties of touching-vug carbonates: Preprint 96027 of the Society of Petroleum Engineers (SPE) presented at the Annual Technical Conference and Exhibition of the Society of Engineers- Oil Workers (SPE), Dallas, Texas, October 9-12, 9 pp.

To perform multi-point statistics (MPS) simulations, training images must be stationary. However, in many applications of reservoir modeling, geological deposits do not exhibit stationarity of patterns / features that reflect reservoir heterogeneity and deposition anisotropy.

SUMMARY OF THE INVENTION

An object of the present invention is to provide a method for generating numerical (digital) pseudocores using well images, digital rock images, and multipoint statistics.

The present invention relates to a method for generating three-dimensional (3D) computer models of rocks and pores, known as numerical pseudocores. The method uses full circular images of the borehole wall, digital rock images and multipoint statistics (MPS) to reproduce three-dimensional (3D) pseudo-core for the logging interval, where the real core was not taken, but there are wellbore images obtained from the logging. Digital core images are used as instructions for reproducing three-dimensional (3D) features of pseudocores using multi-point geostatistics, and the final pseudocores are limited to full circular images of the borehole wall.

Other features and advantages of the invention will become more apparent from the following detailed description in conjunction with the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention is further shown in the following detailed description, with reference to numerous drawings, by way of non-limiting examples of illustrative embodiments of the present invention, in which like reference numerals indicate similar parts in the drawings, in which:

FIG. 1a shows a training image presented as a three-foot (or one meter) interval of a well image diagram, in particular, in one pass of a formation micro-scanner (FMI) in a cavernous carbonate formation, according to at least one embodiment of the invention;

FIG. 1b illustrates the result of Multipoint Statistical Modeling (MPS) using the FILTERSIM algorithm for the same training image from FIG. 1a, according to at least one embodiment of the invention;

FIG. 2a shows the same training image from FIG. 1, according to at least one embodiment of the invention;

FIG. 2b illustrates the result of the first implementation of the training image from FIG. 1a and is consistent with the original measurement data of the cavernous formation of Figure 1 in such a way that the measurement data are accepted for processing in the first (FIG. 2a), second (FIG. 2b) and third (FIG. .2c) implementations that represent subtle variations of the simulated areas, according to at least one embodiment of the invention;

FIG. 2c illustrates the result of the second implementation of the training image from FIG. 1a and corresponds to the original measurement data of the cavernous formation from FIG. 1a, according to at least one embodiment of the invention;

FIG. 2d illustrates the result of a third implementation of the training image from FIG. 1a and corresponds to the original measurement data of the cavernous formation from FIG. 1a, according to at least one embodiment of the invention;

FIG. 3a shows a static image of caverns appearing as dark areas in images obtained by logging during drilling (LWD) of a cavernous carbonate, so that the conductivity spots are shown in brown and the non-conductive areas are shown in white, according to at least one embodiment the implementation of the invention;

FIG. 3b shows a dynamic image of voids that appear as dark areas in images obtained during logging during drilling (LWD) of cavernous carbonate, so that the conductivity spots are shown in brown and non-conductive areas are shown in white, according to at least one embodiment the implementation of the invention;

FIG. 4 shows a full circular image of a borehole wall with cavernous porosity from FIG. 1a, which shows the contours (green lines) that outline regions of lower resistivity in an electrical image, according to at least one embodiment of the invention;

FIGS. 5a and 5b show a flowchart of a method for generating numerical pseudocores using well images, digital rock images, and multipoint statistics, according to at least one embodiment of the invention;

FIG.6 shows a training image from FIG.1A in a translucent three-dimensional (3D) X-ray computer tomogram (CT) tomogram of a plate sample of cavernous carbonate (approximately 6 inches (152.4 mm) high and 4 inches (101.6 mm) in diameter) in such a way that the cavities appear in red to gray color and the rock skeleton is illustrated in blue, according to an embodiment of the invention;

FIG. 7 shows images of a well bent into a cylindrical shape, respectively, of a well diameter (well diameter of approximately 8.5 inches (215.9 mm) and interval length of approximately 3 feet, or 1 meter), at a specific depth. The training image from the CT-tomograms of FIG. 6 is made on an appropriate scale and correctly positioned at its proper depth in the center of the well images; in which the caverns are illustrated in red and the rock skeleton is shown in light blue, according to embodiments of the invention.

FIG. 8 illustrates the result of generating a numerical bifacial pseudo-core from a training image and a full circular image of the well wall, which corresponds to FIG. 7 in such a way that the caverns are illustrated in red and the skeleton of the rock is illustrated in dark blue, according to embodiments of the invention;

FIG. 9 illustrates the result of generating a numerical pseudo-core from a training image and a full circular image of the well wall, which corresponds to FIG. 7 in such a way that the cavities are invisible, and the rock skeleton is shown in gray. The randomly selected section (see the upper part of FIG. 9) shows that this is a three-dimensional (3D) model, with many pores in each section (the approximate interval length is 1 foot, or 0.3 meters, and the borehole diameter is approximately 8, 5 inches (215.9 mm)), according to embodiments of the invention;

FIG.10 illustrates the result of generating a numerical pseudocore from a training image and a full circular image of the well wall, which corresponds to FIG.6. Conductivity spots (illustrated in red) resemble the contours around the conductivity spots, as shown in FIG. 4. It should be noted that three petrophysical facies are shown: pores, or caverns (illustrated in green), conductivity spots (illustrated in red), and rock skeleton (illustrated in blue), according to embodiments of the invention;

11 shows a numerical pseudo-core, as in FIG. 10, with an invisible rock skeleton that highlights the three-dimensional (3D) interconnected nature of pores or caverns (illustrated in green) and conductivity spots (illustrated in red).

FIG. 12 shows a numerical pseudo-core, as in FIG. 10, and a complex three-dimensional (3D) shape of pores or caverns (illustrated in green), and both the conductivity spots and the rock skeleton are invisible, according to embodiments of the invention;

FIG. 13 shows numerical pseudo-cores that can be reselected or transformed into a radial grid, in which radial grids can be layered based on layers observed in well images or other well diagrams, according to embodiments of the invention;

FIG. 14a shows a relative permeability curve that is used for a conductivity spot in a flow simulation; in which there are three different relative permeability curves for each of the three rock types in the pseudocore model: such as cavities, conductivity spots and an impermeable rock skeleton, according to embodiments of the invention; and

FIG. 14b shows the result of simulating a flow through a numerical pseudo core, so that a number of microinjectors of water surround the outer diameter, a number of microproducts surround the inner diameter of the pseudocore: colors represent oil saturation (So), where the formation of heterogeneous languages and breakthroughs are shown in this flow model, according to options the implementation of the invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

DETAILED DESCRIPTION OF THE INVENTION

The details shown here are given only as an example and for the purpose of illustrative discussion of embodiments of the present invention, and presented in such a plan, which is supposed to be the most useful and easily understood description of the principles and conceptual aspects of the present invention. In this regard, no attempt has been made to show the structural details of the present invention in more detail than is necessary for a fundamental understanding of the present invention. The description, together with the drawings, makes obvious to those skilled in the art how certain forms of the present invention can be practiced.

The present invention relates to a method for generating three-dimensional (3D) computer models of rocks and pores, known as numerical pseudocores. The method uses full circular images of the borehole wall, digital rock images and multipoint statistics (MPS) to reproduce three-dimensional (3D) pseudo-core for the logging interval, where the real core was not taken, but there are wellbore images obtained from the logging. Digital rock images are used as instructions for reproducing three-dimensional (3D) features of pseudocores using multi-point geostatistics, and the final pseudocores are due to full circular images of the borehole wall.

According to an aspect of the invention, the fundamental idea is to use a training image: the authors of the present invention directly use the data themselves [in creating a complete circular picture, this is the initial two-dimensional (2D) incomplete image (training image with continuous variables), which covers> 60% of the entire area, then as in the reproduction of a pseudo-core, the CatScan approach (computerized axial tomographic scan) is directly used as a three-dimensional (3D) training full-time image, which is subsequently combined with the data of a full circular image of the borehole wall]. Thus, the whole process of applying multipoint statistics (MPS) as a whole becomes computer-controlled. This advantage should be emphasized in our patent application and conditions.

Well images form oriented electrical and acoustic maps of the rocks and fluids that the well encounters. Full circular images of the borehole wall are full 360-degree views of the borehole wall. By design, most acoustic and logging tools during drilling create full circular images of the borehole wall. Many pictures of geological sections based on resistivity data require simulated full circular images of the borehole wall, since there are gaps between the shoes, and some electrodes or shoes may be damaged. In the process of modeling, algorithms with continuous variables developed within the framework of multipoint statistics (MPS) are used. Full circular images of the borehole wall provide data on the physical location of structural features such as layer boundaries, pores and spots of conductivity and resistivity within the rock on the borehole wall. These outline complex three-dimensional volumes known as petrophysical facies.

Digital images of rocks or cores can be generated from two-dimensional (2D) thin sections, images obtained using a scanning electron microscope, images obtained using a confocal microscope, or computer-generated packages of spheres. However, many digital images of rocks or cores are generated from computed tomograms (CT scans) of rocks and fluids. Tomographic images are created when the scanner passes x-rays at various angles through the rock to the receiver. X-ray attenuation is converted to density, and three-dimensional (3D) images of rocks and pores are built from successive sections. According to an aspect of the invention, each digital core image, for example, a CT scan, is directly perceived as a training image. These are discrete variable training images with characteristic features of the breed (white color) or pores (black color) in each pixel of the image. The training image can have any form of borders or contain any number of chaotic cavities.

Numerical pseudocores are created using an algorithm with discrete variables within the framework of multipoint statistics (MPS). Integer values, such as rock (0), pores (1), and conductivity spots (2), are assigned to each petrophysical facies. Digital images of the rock or core are used as training images, that is, quantitative patterns used to model property fields. Full circular images of the borehole wall cover a numerical pseudocore with cylindrical outlines for modifying the models. Each numerical pseudo-core fully follows digital images of the rock or core and full circular images of the borehole wall. Numerical pseudocores can be referenced to form models suitable for simulating fluid flow. The capillary pressure and relative permeability curves are derived from conceptual models, special core analysis, or proven thin-scale pore network modeling techniques.

FIG. 1a is a training image defined by a three-foot (one-meter) interval of a well image diagram obtained with one pass of a formation microscanner (FMI) in a cavernous carbonate formation. Caverns are pores that are filled with a water-based drilling fluid, and they appear as dark spots with low resistivity. Noteworthy intervals between the shoes in FIG.1A. For illustration, a pixelated, user-defined 3 × 3-template (bottom) moves along the training image, detecting patterns and forming filter estimates for areas that make up the immediate environment of each measured pixel. This forms the basis for multi-point statistical (MPS) modeling using the FILTERSIM algorithm. The full circular image of the borehole wall in FIG. 1b shows the implementation of the FILTERSIM algorithm, in which the entire image of FIG. 1a is used as a training image. The implementation is determined in such a way that it perfectly agrees with the original measurement data. Orientations are shown along the vertices of the images. Abbreviations: N = North; E = east; S = south; and W = West. There are no vertical exaggerations. The diameter of the bit is 8.5 inches (21.5 cm). In particular, FIG. 1b illustrates the result of Multipoint Statistics (MPS) using the FILTERSIM algorithm for the same training image from FIG. 1a. Thus, using the FILTERSIM algorithm, the authors of the present invention determine filter score indicators for each training image using a suitable pixelated template. These estimated indicators digitize the patterns and their probabilities as observed in the measurement data, then generate implementations of complete circular images of the well wall using the FILTERSIM algorithm of multipoint statistics (FIG. 1a).

FIGS. 2b, 2c, and 2d show three realizations of a full circular image of a well wall in a cavernous carbonate formation, in which FIG. 2a shows the same training image from FIG. 1a. It should be noted that the measurement data are of primary importance in each implementation, and there are subtle variations in the simulated areas. FIG.2b illustrates the result of the first implementation of the training image from FIG.2a and is consistent with the original measurement data of the cavernous formation from FIG.1a. FIG.2c illustrates the result of the second implementation of the training image and is consistent with the original measurement data of the cavernous formation of FIG.1a. FIG. 3d illustrates the result of the third implementation of the training image from FIG. 2a and is consistent with the original measurement data of the cavernous formation from FIG. 1a.

FIG. 3a shows a static image of caverns appearing as dark structural features in logging while drilling (LWD) images (see authors Xiao, L., Jun, C., Duo, YS, Han, SY, Xia, WH, and Xi, WY, 2007: Fully integrated solution for LWD resistivity image application a case study from Beibu Gulf, China (Fully integrated solution for applying the resistivity pattern in logging while drilling in a specific case of Beibu Bay, China): First Indian Regional Conference of the Society of Professional Downhole Analysts logging diagrams (SPWLA), Horizontal Formation Assessment for Basic wells, Mumbai, March 19-20, 10 pp.), so that the conductivity of the brown spots illustrated and non-conductive portions shown in white. FIG. 3b shows a dynamic image of voids appearing as dark areas in images obtained during logging during drilling (LWD) of cavernous carbonate, so that the conductivity spots are shown in brown and non-conductive areas are shown in white. Authors Dehghani et al. (1999) suggest that zones of increased porosity and permeability exist near caverns. See an article by Dehghani, K., Harris, P.M., Edwards, K.A. and Dees, W.T., 1999, Modeling a vuggy carbonate reservoir (AAPG Bulletin, Volume 83, pp. 19-42.

FIGURE 4 of the full circular image of the borehole wall with cavernous porosity (the same as Figures 1 and 2) shows the contours (green lines) that outline areas with lower resistivity in the electrical image. Orientations are shown along the top of the image. Abbreviations: N = North; E = east; S = south; and W = West. There are no vertical exaggerations. The diameter of the bit is 8.5 inches (21.5 cm).

FIG. 5 shows a flowchart of a method for generating numerical pseudocores using well images, digital rock images, and multipoint statistics. Step 100 includes collecting depth intervals of the wellbore image diagrams from the reservoir, which can be used as training images. Training images are oriented; two-dimensional (2D) scalar arrays of continuously variable numerical values, with gaps between the shoes and / or areas that require restoration, so that the shoes represent the measured values, and the gaps are parts of the well that are not reflected in the images.

Again, involving step 100 in FIG. 5, the original full circular images of the well wall were generated using certain logging tools, such as acoustic devices and logging devices while drilling. In other cases, full circular images of the borehole wall should be generated. This is especially true for resistivity-based diagrams that typically have gaps between the shoes.

Again, involving step 100 in FIG. 5, according to an aspect of the invention, for selecting a depth interval of a well image diagram, for example, the interval may have a measurement depth of 1, 3 or 10 feet (0.3, 1 or 3 meters). The user may wish to select a thick or thin interval, depending on the observed degree of layering, fracturing and other factors of heterogeneity.

Full circular images of the borehole wall can be generated using the FILTERSIM algorithm within Multipoint Statistics (MPS), as noted in step 100 in FIG. 5. In the FILTERSIM algorithm, filter estimates are used to group and then simulate patterns in the intervals between the shoes where there is no measurement data. The single implementation method comes down to randomly occupied pixel positions, and derived from a set of filter estimates to select a suitable picture for an arbitrary site. Measurement data are of primary importance, as they represent conditional modeling. The paintings adjacent to the edges of the shoes are consistent with the patterns observed on real shoes. The frequency distribution of the colors of the simulated pixels, which is a continuous variable, is exactly consistent with the frequency distribution of the measured colors.

Step 200 of FIG. 5 includes obtaining at least one core sample from the reservoir and then generating a digital core image from the obtained core sample. The obtained core adequately reflects the indicative features and / or structure of one or more deep intervals of the well and / or reservoir. For example, digital rock or core images, such as CT scans, can be used as training images. Training images are three-dimensional (3D) arrays of discrete numerical values. For example, in a two-facies model, the rock has a numerical value of 0, and the pores have a numerical value of 1. In a three-facies model, the rock has a numerical value of 0, pores have a numerical value of 1, and conductivity spots have a numerical value of 2. There are no restrictions on the number of facies. The contours of individual facies bodies (volumes) can have any shape or size.

Step 300 of FIG. 5 includes pre-modeling the collected well image image data, such as generated full circular images of the well wall.

Step 300a in FIG. 5 represents a selection of a depth interval of a well image diagram. Complete circular images of the borehole wall are generated using the FILTERSIM algorithm within Multipoint Statistics (MPS), in which MPS modeling extracts geological structures from training images and attaches them to data locations. The FILTERSIM algorithm uses filter estimates to group and then simulate patterns between the shoes where there is no measurement data.

Step 300b of FIG. 5 represents rounding into petrological facies that identify, for example: pores or cavities; spots of conductivity or resistivity in the rock on the well wall; and layer boundaries or rock skeleton. The single implementation method comes down to randomly occupied pixel positions, and derived from a set of filter estimates to select a suitable picture for an arbitrary site. Measurement data are of primary importance, as they represent conditional modeling. The paintings adjacent to the edges of the shoes are consistent with the patterns observed on real shoes. The frequency distribution of the colors of the simulated pixels, which is a continuous variable, and is exactly consistent with the frequency distribution of the measured colors.

Step 300c in FIG. 5 represents collapsing the full circular images of the borehole wall into Scaled Cylindrical Shapes. In particular, for current interpretations, it is difficult to examine well images in a three-dimensional (3D) representation. Therefore, it is common to cut a well along geographic north and then scan the cylinder to produce a two-dimensional (2D) image. In highly curved and horizontal wells, it is common practice to dissect the image of the well along the top of the well. Planar structural features that divide a cylindrical borehole into parts appear as sinusoidal waves in a two-dimensional (2D) form. To generate three-dimensional (3D) numerical pseudocores, the authors of the present invention need to collapse two-dimensional (2D) images of the well to their original three-dimensional (3D) form. To do this, the authors of the present invention must know the diameter of the well. The authors of the present invention can easily determine this from the diagrams of the diameter of the well, which are read from the original image of the well. Images should be on a 1: 1 scale, i.e. there is no vertical exaggeration.

Step 400 in FIG. 5 presents a preliminary simulation of the obtained digital core image. In step 400a, a conversion from 2 facies to 3 facies is performed so that pores or cavities, conductivity spots, and the skeleton of the rock are identified in a digital rock image. For example, if conductivity spots were not identified in a digital rock image (training image), then they can be modeled by a similar transformation of pore facies with a fixed number of volumetric image elements (voxels). In this way, large-scale heterogeneity can be extracted in a numerical pseudocore. Typically, the resolution of the volumetric image elements of a digital rock image or training core image is finer than the resolution of a full circular image of the borehole wall. To ensure that the resulting pseudo-core model will have signs on a scale consistent with the full circular image of the well wall, the training image is selected approximately in accordance with the ratio of the resolution of the digital image of the rock to the resolution of the full circular image of the well wall.

Step 500 in FIG. 5 represents the generation of numerical pseudo-core implementations by combining structures borrowed from the collected digital core analysis data and the collected well image diagram data. Step 500a represents the application of the MPS SNESIN algorithm to generate realizations of numerical pseudocores for 2 or more facies so as to modify implementations to match the sizes and shapes of facies observed in training images of the digital rock image and full circular images of the borehole wall. More than 2 facies can be simulated in cases where conductivity spots were mapped (Figures 3 and 4) in full circular images of the borehole wall and digital rock images. If conductivity spots were not identified in the digital image of the rock (training image), then they can be modeled by a similar transformation of the pore facies with a fixed number of volumetric image elements (voxels). In this way, large-scale heterogeneity can be extracted in a numerical pseudocore. The radial size and height of a numerical pseudocore is limited only by the amount of computer memory available for use.

Step 600 in FIG. 5 presents a re-sampling of numerical pseudocores for a radial grid. To conduct a flow study in a numerical pseudo-core, the Cartesian (rectangular) coordinate grid of the numerical pseudo-core model generated during the previous steps is converted into a radial coordinate grid in the form of cylindrical coordinates. Each Cartesian element of the volumetric image is assigned constant porosity or permeability according to its type of rock (skeleton, cavity or conduction spot). For each Cartesian cell in the radial coordinate network, since it consists of many Cartesian elements of the volumetric image, average values of porosity and permeability are generated. The average porosity is obtained by arithmetic averaging of the entire porosity of the Cartesian elements of the three-dimensional image inside the cell; permeability is obtained by performing geometric averaging.

Step 700 in FIG. 5 represents performing simulations of flow in an area close to the well. Numerical simulations of fluid flow, for example, water flooding, are carried out on the constructed pseudocore to evaluate important parameters, such as water cut, oil recovery coefficient and recovery coefficient. A reference table of capillary pressure and relative permeability for various facies in a numerical pseudocore gives values that are entered into a flow simulation program, for example, the Ecllipse package. This is a key step in quantifying the effect of carbonate rock heterogeneity on fluid flow based on a pseudocore model. Capillary pressure and relative permeability can be obtained from the data from the interactive core analysis program (SCAL) or the capillary pressure values for mercury injection (MICP, Mercury Injection Capillary Pressure) (if available) for cores of the same rock type. More generally, a method for modeling a pore network structure can be used to evaluate the aforementioned parameters of various types of rocks.

Step 800 in FIG. 5 provides a definition of a process for improving wettability effects in evaluation parameters, such as water cut, oil recovery coefficients, and recovery factor.

FIG. 6 shows a training image from FIG. 1a in a translucent three-dimensional (3D) view of a CT tomogram of a plate sample of cavernous carbonate (approximately 6 inches (152.4 mm) high and 4 inches (101.6 mm) in diameter), thus that the caverns appear in colors from red to gray, and the rock skeleton is illustrated in blue. FIG. 6 also shows the selected training image, CT tomogram of cavernous carbonate (author Gowelly, S., 2003, 3-D analysis of vug connectivity, Indian Basin field, New Mexico Mexico): unpublished master's thesis, Colorado School of Mines (Colorado School of Mines), Golden, Colorado). It should be noted that FIG. 1a shows a three-foot (one-meter) interval of a well image diagram, from the same depth and in the same formation, and that the caverns are large disordered pores visible to the naked eye. Moreover, in the image diagram of the well, the cavities appear as dark spots, since they are filled with water-based drilling mud, and they are electrically conductive.

FIG. 7 shows images of a well rolled into a cylindrical shape corresponding to the diameter of the well (well diameter of approximately 8.5 inches (215.9 mm) and interval length of approximately 3 feet, or 1 meter), at a specific depth. The training image from the CT tomograms of FIG. 6 is made in the proper scale and correctly positioned at its proper depth in the center of the well images, in which the caverns are illustrated in red and the rock skeleton is shown in light blue. As noted above, the images of the well are folded into a three-dimensional (3D) cylindrical shape so that full circular images of the well wall are not shown, since this allows us to present a view of the relative scales of the training image (CT tomogram) (in the center of the cylinder) and the image of the diagram. For a picture of a numerical pseudocore model (see FIG. 8 below), it will be noted that a training image and full circular images of the borehole wall are used.

FIG. 8 illustrates the result of generating a numerical pseudo-core from a training image and a full circular image of the well wall, which corresponds to FIG. 7 in such a way that the caverns are illustrated in red and the rock skeleton is illustrated in dark blue. It should be noted that the interval length can be 3 feet (1 meter) and the borehole diameter can be 8.5 inches (22 cm).

FIG. 9 illustrates the result of generating a numerical pseudo-core from a training image and a full circular image of the well wall, which corresponds to FIG. 7 in such a way that the cavities are invisible, and the rock skeleton is shown in gray. Further, a randomly selected section (see the upper part of FIG. 9) shows that this is a three-dimensional (3D) model, with many pores in each section (the approximate interval length is 1 foot, or 0.3 meters, and the borehole diameter is approximately 8.5 inches (215.9 mm)).

FIG.10 illustrates the result of generating a numerical pseudocore from a training image and a full circular image of the well wall, which corresponds to FIG.6. A numerical pseudo-core (see FIG. 9) shows the application of a circular similar transformation with an 8-pixel radius around each pore. Conductivity spots (illustrated in red) are similar to the contours around the conductivity spots, as shown in FIG. 4. Therefore, this model has 3 petrophysical facies: pores or cavities (illustrated in green), conduction spots (illustrated in red) and rock skeleton (illustrated in blue). Conductivity spots create a three-dimensional (3D) connection between pores or cavities, and make it possible to identify heterogeneity, which is inherent in most carbonate rocks.

11 shows a numerical pseudo-core, as in FIG. 10, with an invisible rock skeleton that highlights the three-dimensional (3D) interconnected nature of pores or caverns (illustrated in green) and conductivity spots (illustrated in red). It should be noted that the interval length can be 3 feet (1 meter) and the borehole diameter can be 8.5 inches (22 cm).

FIG. 12 shows a numerical pseudo-core, as in FIG. 10, and a complex three-dimensional (3D) shape of pores or caverns (illustrated in green), and both the conductivity spots and the rock skeleton are invisible. It should be noted that the interval length can be 3 feet (1 meter) and the borehole diameter can be 8.5 inches (22 cm).

FIG. 13 shows numerical pseudocores that can be reselected or transformed into a radial grid, in which radial grids can be layered based on layers observed in the images of the well or other geological sections of the well. 13 also shows a radial coordinate grid that is generated to study flow characteristics in a numerical pseudocore model. It should be noted that the inside (approximately 4 inches (101.6 mm) in diameter) of the original numerical pseudo-core was drilled to allow placement of microproducers. Microinjectors can be placed around the outer boundary of the pseudocore.

FIG. 14a shows a relative permeability curve that is used for a conductivity spot in a flow simulation; in which there are three different relative permeability curves for each of the three rock types in the pseudocore model: such as cavities, conductivity spots and an impermeable rock skeleton. In particular, FIG. 14a shows the relative permeability curve that was used for the conductivity spot in the flow simulation. It should be noted that the pseudocore for both FIG. 14a and FIG. 14b may have a height of approximately 1 foot (0.3 meters), with an outer diameter of approximately 8.5 inches (22 cm) (or the core width is 8.5 inches (22 cm) and an inner diameter of approximately 4 inches (101.6 mm)).

FIG. 14b shows the result of simulating a flow through a numerical pseudo core, so that a number of microinjectors of water surround the outer diameter, a number of microproducts surround the inner diameter of the pseudocore: colors represent oil saturation (So), where the formation of heterogeneous languages and breakthroughs are shown in this flow model, according to options the implementation of the invention. It should be noted that FIG. 14b shows the oil saturation profile of the flooded numerical pseudocore.

Further, while the present invention has been described with reference to exemplary embodiments, it is understood that the terms that have been used here are descriptive and illustrative terms rather than restrictive terms. Changes can be made within the scope of the attached claims as now claimed and as modified, without going beyond the scope and meaning of the present invention in its aspects. Although the present invention has been described here with reference to specific means, materials and embodiments, the present invention is not intended to be limited to the details disclosed herein; rather, the present invention extends to all functionally equivalent structures, methods, and applications, such as those falling within the scope of the appended claims.

Claims (32)

1. A method of forming a numerical model of a pseudocore, comprising the steps of:
a) obtain well data from the reservoir, which include depth intervals of the reservoir, and process the well data into a portion of the data of at least one interpreted image of the well having unidentified image data of the well;
b) examine one of the data parts of at least one interpreted image of the well, other processed log data or both to generate unidentified image data of the well, then process the generated unidentified image data of the well into the data part of at least one interpreted image of the well so as to generate data of at least one minimized full circular image of the well wall;
c) obtain one of the at least one core from the reservoir, a portion of the log data, or both, and generate digital core data from one of the obtained at least one core, a portion of the log data, or both, so that the generated digital core data represent one of the features or structures of one or more deep reservoir intervals; and
d) process part of the generated digital core data, part of the data of at least one interpreted image of the well or part of the log data to generate an implementation of the numerical model of the pseudocore. 2. The method according to claim 1, in which part of the data of at least one interpreted image of the well is a training image, which is a two-dimensional (2D) scalar arrays of continuously changing numerical values. 3. The method according to claim 1, in which part of the data of at least one interpreted image of the well consists of processed source data, which consists of measured values and unmeasured values. 4. The method according to claim 1, in which the unidentified image data of the well represent one of the unmeasured values of the data part of the at least one interpreted image of the well or data gaps in the data part of the at least one interpreted image of the well. 5. The method according to claim 4, in which the data spaces are selected from the group consisting of at least one damaged shoe in the collector, at least one damaged area in the collector of at least one shoe with inappropriate pressure pressing the shoe against the wall of the well in the manifold of at least one shoe with difficult contact with the wall of the well in the manifold or at least one idle shoe in the manifold. 6. The method according to claim 1, in which the other processed logging data is selected from the group consisting of one of the sets of logging data while drilling or logging data recorded by the tool lowered into the well on the cable, or a combination thereof. 7. The method according to claim 6, in which the data logs recorded by the device, lowered into the well on the cable, include at least one of the logs obtained using gamma radiation, density, acoustic, neutron logging, cavernograms or resistance logs. 8. The method according to claim 1, in which the data of at least one minimized full circular image of the borehole wall is one of a two-dimensional (2D), three-dimensional (3D) representation, or together. 9. The method according to claim 1, in which the generation of at least one full circular image of the well wall comprises the steps of:
1) select the depth interval of the well image diagram from the depth intervals of the reservoir, process the depth interval of the well image diagram using the Multipoint Statistics (MPS) algorithm, in which the Multipoint Statistics (MPS) algorithm provides modeling that extracts unidentified geological structures from at least a portion of the data , one interpreted image of the well to identify data locations in the data part of at least one interpretation uemogo borehole image;
2) process the identified data locations using filter estimates to group and then model the patterns in unidentified well image data, resulting in unidentified well image data representing one of the unmeasured values of a portion of at least one interpreted well image or data spaces in parts of at least one interpreted image of the well;
3) identify the petrophysical facies of the reservoir, the petrophysical facies selected from the group consisting of cracks, pores, caverns, spots of conductivity in the rocks in the wall of the borehole of the reservoir, resistive sections of rocks in the wall of the borehole of the reservoir, the boundaries of the layers or skeleton of the rock; and
4) the unidentified image of the well is processed into a part of at least one interpreted image of the well to generate at least one full circular image of the well wall, then, at a known diameter of the well, at least one full circular image of the well wall is collapsed into the original three-dimensional (3D) form, as a result of which the numerical pseudo-core is modified so as to correspond to one of the folded parts of the full circular image of the well wall or part, at least one interpretable borehole image. 10. The method according to claim 1, in which the logging data is selected from the group consisting of one set of logging data having multiple depths of research, analog models of core analysis data from one or more reservoirs, or a combination thereof. 11. The method according to claim 1, in which the generation of digital core data consists in identifying petrophysical facies from one of the collected at least one core or part of the log data, the petrophysical facies being selected from the group consisting of cracks, pores, cavities, spots of conductivity in the rocks in the wall of the borehole of the reservoir, resistive sections of rocks in the wall of the borehole of the reservoir, the boundaries of the layers or skeleton of the rock. 12. The method according to claim 11, in which the identification of petrophysical facies includes modeling by similar conversion of other petrophysical facies of the collected at least one core using a fixed number of volumetric image elements. 13. The method according to claim 1, in which at least one core is obtained from at least one other collector. 14. The method according to claim 1, in which the second core of at least one core is obtained from one or more collectors. 15. The method according to claim 1, in which step (d) includes plotting a digital file of generated implementations of a pseudo-core numerical model in the form of one of a digital medium or paper medium. 16. The method according to claim 1, in which generating an implementation of a numerical pseudo-core model in step (d) includes using a Multipoint Statistics (MPS) algorithm to generate implementations of numerical pseudo-core for two or more petrophysical facies to modify implementations to fit sizes and shapes facies observed in part of at least one interpreted image of the well, part of the generated digital core data and the generated data of at least one minimized full circular image well wall fermentation. 17. The method according to clause 16, in which two or more petrophysical facies selected from the group consisting of collected at least one core, part of the log data, or both, in which the petrophysical facies selected from the group consisting of cracks , pores, caverns, conductivity spots in the rocks in the wall of the reservoir well, resistive sections of rocks in the wall of the reservoir well, layer boundaries or rock skeleton. 18. The method according to clause 15, further containing a stage in which:
f) re-sampling the numerical pseudo-kernels of step (d) with conversion to a radial coordinate grid. 19. The method according to p, in which the radial coordinate grid is provided for the study of flow in a numerical model of the pseudocore and includes the steps in which:
1) construct a cylindrical coordinate grid having one or more shapes or one or more layers;
2) make repeated sampling with the transformation of the numerical model of the pseudo-core from Cartesian coordinates, each volumetric image element of which has constant porosity or constant permeability in accordance with the type of rock of the volumetric image element, into cylindrical coordinates, in which each cell of the cylindrical coordinate grid includes many Cartesian elements of the volumetric image of the numerical model of the pseudocore;
3) assign to each cylindrical cell the average porosity and average permeability based on the included multiple Cartesian elements of the volumetric image; and
4) assign to each cylindrical cell of the curves the relative permeability and capillary pressure based on the dominant rock type of the included multiple Cartesian volumetric image elements using the lookup table for one of the values of capillary pressure, relative permeability, or both, for different facies in a numerical pseudocore, which is introduced into a flow modeling program, thereby for digitizing the effect of rock heterogeneity on a fluid flow based on a pseudo-model and with the involvement of one of these special program analysis of core (SCAL), the capillary pressure for the mercury discharge values (MICP), the simulation of the pore network structure, or combinations thereof, one or more cores having the same rock type. 20. The method according to claim 19, in which the associated type of rock is selected from the group consisting of cracks, pores, caverns, conductivity spots in the rocks in the wall of the reservoir well, resistive sections of the rocks in the wall of the reservoir well, layer boundaries or rock skeleton. 21. The method according to p, optionally containing phase, in which:
f) perform flow simulations for one or more geographically related reservoirs, such as an area near the well. 22. The method according to item 21, in which the simulation of the stream includes one of a single-phase or multiphase flow so that the simulation of the flow is performed on the calculated numerical model of the pseudo-core of step (e) to evaluate one or more parameters. 23. The method according to item 22, in which one or more parameters is selected from the group consisting of porosity, permeability, capillary pressure, relative permeability, water cut of the reservoir, oil recovery coefficient or recovery factor. 24. The method according to item 21, further containing a stage in which:
g) conduct a sensitivity analysis to flow related parameters to determine the process for one of the water flooding, injection of gas into the formation, injection of steam into the formation, or oil recovery with enhanced oil recovery (EOR). 25. The method according to paragraph 24, in which the parameters related to the flow are selected from the group consisting of porosity, permeability, capillary pressure, relative permeability, water cut of the reservoir, oil recovery coefficient or recovery factor. 26. The method according to item 21, further containing a stage in which:
g) define a process for evaluating the wettability effects in the evaluated parameters, in which the evaluated parameters include one of: capillary pressure, relative permeability, water cut, oil recovery coefficient or recovery coefficient. 27. A method of obtaining a numerical model of a pseudocore, comprising the steps of:
a) obtain the log data from the reservoir, which includes the depth intervals of the reservoir, and process the log data into part of the data of at least one interpreted image of the well;
b) examine one of the data parts of at least one interpreted image of the well, other processed log data, or both to generate at least one minimized full circular image of the well wall;
c) obtain one of the at least one core from the reservoir, a portion of the log data, or both, and generate digital core data from one of the obtained at least one core, from a portion of the log data, or both, so that the generated digital core data represents one of the features or structures of one or more depth intervals of the reservoir; and
d) process part of the generated digital core data, part of the data of at least one interpreted image of the well, part of the log data or a combination thereof to generate an implementation of a numerical model of the pseudocore. 28. The method according to item 27, in which part of the interpreted data of at least one image of the well is a training image, which is a two-dimensional (2D) scalar array of continuously changing numerical values. 29. The method according to item 27, in which the other processed log data is selected from the group consisting of sets of log data during drilling or log data recorded by the tool, lowered into the well on the cable, or a combination thereof. 30. The method according to item 27, in which the logging data is selected from the group consisting of logging data sets having multiple depths of research, or analog core data models from one or more reservoirs or some combination thereof. 31. The method according to item 27, in which the generation of digital core data consists in identifying petrophysical facies from one of the collected at least one core or part of the log data, moreover, the petrophysical facies is selected from the group consisting of cracks, pores, caverns, spots of conductivity in the rocks in the wall of the borehole of the reservoir, resistive sections of rocks in the wall of the borehole of the reservoir, the boundaries of the layers or skeleton of the rock. 32. The method according to p, in which the identification of petrophysical facies includes modeling by similar conversion of other petrophysical facies collected at least one core by a fixed number of volumetric image elements.

Images