A Quantitative Characterization Method for Clastic Rock Particle Size Based on CT Scan Structural Images

Abstract

This invention belongs to the field of shale oil and gas exploration and development technology, and discloses a method for quantitative characterization of clastic rock particle size based on CT scan structural images. The method first acquires the geological environment of the sedimentary diagenesis of the target layer, determines image acquisition parameters, and performs multi-scale image acquisition to obtain grayscale images. Then, it uses the equivalent sphere algorithm to obtain characterization parameters and constructs a digital model based on screening criteria. Next, it utilizes machine learning to construct a standard model and a digital result model. Finally, it identifies the dominant reservoirs and target fracturing intervals of the oil and gas layer. This method, through multi-scale image acquisition and machine learning, improves the accuracy of the three-dimensional digital model of clastic rock particle size and structural characteristics, enabling more accurate prediction of oil and gas exploration and development, improving exploration and development efficiency, reducing costs, and effectively solving the problem of inaccurate measurements in existing technologies.

Description

Technical Field

[0001] This invention belongs to the field of shale oil and gas exploration and development technology, specifically relating to a method for quantitative characterization of clastic rock particle size based on CT scan structural images. Background Technology

[0002] Quantitative characterization of clastic rock particle size has long been a world-class technical challenge that oil and gas exploration and development projects have needed to address but have not yet fundamentally solved.

[0003] Currently, the main techniques for characterizing the particle size of clastic rocks are manual qualitative measurement and scanning electron microscopy (SEM). Manual measurement can only identify particles larger than millimeters and can only measure particles displayed on two-dimensional cross-sections. SEM can only detect particles smaller than micrometers, and due to its limited field of view, it can only depict particle development characteristics within a very small area of ​​a two-dimensional slice, and cannot effectively characterize the spatial variation of particles. Current methods can only obtain random probability values ​​and cannot effectively characterize the true three-dimensional spatial situation of clastic rock particle size, resulting in low accuracy in obtaining information about clastic rock particle development, and consequently, low accuracy in oil and gas exploration and development.

[0004] Therefore, there is an urgent need to provide a three-dimensional quantitative characterization method for the particle size development of clastic rocks to improve the accuracy and efficiency of oil and gas exploration and development. Summary of the Invention

[0005] Therefore, it is necessary to provide a three-dimensional quantitative characterization method for the particle size of clastic rocks based on CT scan structural images, in order to solve the technical problem that the measurement in the existing technology is inaccurate, resulting in low accuracy of the obtained clastic rock particle development status, and thus low accuracy of oil and gas exploration and development prediction.

[0006] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution:

[0007] A method for quantitative characterization of clastic rock particle size based on CT scan structural images includes:

[0008] Step 1: Obtain the geological environment of the sedimentary rock formation of the target layer, and determine the image acquisition parameters of the clastic rock based on the geological environment;

[0009] Step 2: Collect rock images based on the clastic rock image acquisition parameters, perform multi-scale image acquisition, and obtain grayscale images of clastic rock fragment size and structure;

[0010] Step 3: Identify the size and structure of clastic rock particles in grayscale images, and apply the equivalent sphere algorithm to obtain the characterization parameters of clastic rock particle size and structure;

[0011] Step 4: Obtain the screening criteria for the size and structure characterization parameters of clastic rock particles. Based on the screening criteria, determine different particle sizes and structure characterization parameters from the size and structure characterization parameters of clastic rock particles, and then construct a digital model of the size and structure composition of clastic rock particles.

[0012] Step 5: Construct a standard machine learning model for the relationship between clastic rock particle size and structure based on micron-sized CT scan structural images;

[0013] Step 6: Based on the standard machine learning model for the size and structure of clastic rock particles, obtain a digital result model of the size and structure of clastic rock particles through machine learning;

[0014] Step 7: Based on the digital model of clastic rock particle size and structure, determine the dominant oil and gas reservoirs and the target fracturing intervals.

[0015] As a further optimization of the present invention, in step 1, the geological environment includes: sedimentary facies type, hydrodynamic conditions of the sedimentary environment, and diagenetic type; the determined clastic rock image acquisition parameters include: CT scan resolution of 0.5-100μm, scan angle of 0-360°, and one image acquired every 0.5-1°.

[0016] As a further optimization of the present invention, in step 2, when performing multi-scale image acquisition, three scales are divided: macro scale 1-10mm, meso scale 100μm-1mm, and micro scale 1-100μm; images are acquired in order from macro to micro. During the acquisition process, multi-scale image registration is achieved through three-dimensional coordinate calibration and SIFT feature matching to obtain grayscale images of clastic rock fragment particle size and structure.

[0017] As a further optimization of the present invention, in step 3, the obtained grayscale images of clastic rock particle size and structure are binarized to segment the clastic particle region; the volume V of each particle is calculated, according to the formula for the volume of a sphere, V = ... πr 3 This leads to the equivalent sphere radius r = This is used as a characterization parameter for particle size; the sphericity and flatness of the particles are calculated as structural characterization parameters, where: sphericity S = (4πr 2 ) / A, where A is the particle surface area; flatness F=b / a, where a and b are the lengths of the major and minor axes of the particle.

[0018] As a further optimization of the present invention, in step 4, the screening criteria for the size of clastic rock particles are as follows: noise particles with a volume of <10 pixels are excluded, and particles with a particle size range of 5μm-2mm are retained; the screening criteria for the structure of clastic rock particles are as follows: particles with sphericity >0.4 and flatness 0.3-1.5 are analyzed.

[0019] As a further optimization of the present invention, in step 5, a standard machine learning model for the size and structure of clastic rock particles is constructed using the U-Net convolutional neural network algorithm, specifically including: data preparation and preprocessing, U-Net model construction, model training, and model evaluation and optimization.

[0020] As a further optimization of the present invention, step 6 specifically includes: inputting the newly acquired grayscale images of clastic rock particle size and structure into a standard model, outputting the predicted particle size and structure characterization parameters, post-processing the output results, and finally constructing a digital result model of clastic rock particle size and structure.

[0021] As a further optimization of the present invention, in step 7, the dominant reservoir is determined as follows: the dominant reservoir index RQI = 0.4D. avg +0.3(1 / S o +0.3P, S o =σg / Dg; where Davg is the average equivalent sphere radius, So is the particle size sorting coefficient, σg is the geometric standard deviation, Dg is the geometric mean particle size, and P is the porosity; when RQI > 0.6, it is determined to be a dominant reservoir.

[0022] As a further optimization of the present invention, in step 7, the target fracturing segment is determined by combining rock mechanics parameters and geostress data, selecting an advantageous reservoir segment with an average particle size of 100-300 μm, a particle size standard deviation of <50 μm, and a permeability of <1 mD as the target fracturing segment.

[0023] Compared with existing technologies, the quantitative characterization method for clastic rock particle size based on CT scan structural images proposed in this invention exhibits the following significant advantages and beneficial effects:

[0024] 1. Improve the comprehensiveness and accuracy of image acquisition: Image acquisition parameters are determined by obtaining the geological environment of the sedimentary diagenesis of the target layer. A high-resolution micron-scale CT scanner is selected, with a scanning resolution of 0.5-100 μm and a scanning angle of 0-360°, acquiring one image every 0.5-1°. This comprehensively captures information from different layers of clastic rocks. Image acquisition is conducted at three scales: macroscopic, mesoscopic, and microscopic. Multi-scale image registration is achieved using 3D coordinate calibration and SIFT feature matching. This allows for the acquisition of clastic rock fragment size and structure grayscale images covering everything from the overall core to the fine grain structure, providing a rich and accurate data foundation for subsequent precise analysis.

[0025] 2. Precise Characterization of Particle Features: After binarizing the grayscale image, the equivalent sphere algorithm is applied to deduce the equivalent sphere radius based on the sphere volume formula, serving as a parameter representing particle size. Simultaneously, sphericity and flatness are calculated as structural characterization parameters. This method can accurately quantify the features of debris particles from both size and structure dimensions, providing a more comprehensive and accurate reflection of the true state of particles compared to traditional manual qualitative measurements and scanning electron microscopy qualitative detection.

[0026] 3. Construct a reliable digital model: Establish clear screening criteria to exclude noisy particles with a volume of <10 pixels, retain particles with a diameter range of 5μm-2mm, and analyze particles with a sphericity >0.4 and a flatness of 0.3-1.5. Based on this, construct a digital model of the size and structural composition of clastic rock particles, effectively remove interference information, and make the model more realistically reflect the actual composition and structure of clastic rocks, thereby improving the reliability and accuracy of the model.

[0027] 4. Enhanced Model Predictive Capability: A standard machine learning model is constructed using the U-Net convolutional neural network algorithm. Through steps such as data preparation and preprocessing, model building, training, evaluation, and optimization, the model's ability to learn and predict the characteristics of clastic rock particles is improved. Inputting newly acquired images into the standard model and performing post-processing to construct a digital result model enables more accurate prediction of clastic rock characteristics under different conditions.

[0028] 5. Accurately identify key strata: Based on the digital result model, and using the dominant reservoir index formula (RQI=0.4Davg+0.3(1 / S)... o By using the method of 0.3P to identify the dominant reservoir, and combining rock mechanics parameters and geostress data, the dominant reservoir segments with an average grain size of 100-300μm, a grain size standard deviation of <50μm, and a permeability of <1mD are selected as the target segments for fracturing. This provides precise target positioning for oil and gas exploration and development, improves exploration and development efficiency, and reduces costs. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 Based on CT scan structural images and grayscale maps of clastic rock fragments and structural development.

[0031] Figure 2 To obtain a digital quantitative model of the size and structural development of clastic rock particles through machine learning.

[0032] Figure 3 To obtain the digital quantitative distribution results of clastic rock particle size and structural development through machine learning. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0034] Figure 1 The image shows a grayscale image of the development of clastic particles and structures in clastic rocks obtained through CT scans. Figure 2 and Figure 3 The images show the digital quantitative model diagram and distribution results of clastic rock particle size and structural development obtained through machine learning.

[0035] This invention provides a method for quantitative characterization of clastic rock particle size based on CT scan structural images, comprising:

[0036] Step 1: Obtain the geological environment of the sedimentary rock formation of the target layer, and determine the image acquisition parameters of the clastic rock based on the geological environment.

[0037] Specifically, the geological environment includes: sedimentary facies type (e.g., fluvial, lacustrine, marine), hydrodynamic conditions of the sedimentary environment (e.g., water flow velocity, wave intensity), and diagenetic type (e.g., compaction, cementation, dissolution). The determined image acquisition parameters for clastic rocks include: using a high-resolution micron-sized CT scanner (e.g., the German Zeiss Xradia 620 Versa), with a CT scan resolution of 0.5-100 μm, a scanning angle of 0-360°, and acquiring one image every 0.5-1°.

[0038] Step 2: Collect rock images based on the clastic rock image acquisition parameters, perform multi-scale image acquisition, and obtain grayscale images of clastic rock fragment size and structure.

[0039] Specifically, the German Zeiss Xradia 620 Versa micro-CT scanner was used, equipped with a 110kV microfocus X-ray source, a detector resolution of 2048×2048 pixels, a scanning angle of 0-360°, a step size of 0.5°, an exposure time of 500ms / frame, and grayscale image quantization of 16 bits (0-65535, corresponding to a density of 0.2-3.0g / cm³). 3Multi-scale scanning of clastic rocks was performed: macroscopic scale (1-10 mm, whole core scan, resolution 50-100 μm), mesoscopic scale (100 μm-1 mm, local area magnification, resolution 5-20 μm), and microscopic scale (1-100 μm, fine grain scanning, resolution 0.5-5 μm). Images were acquired in order from macroscopic to microscopic. During the acquisition process, multi-scale image registration was achieved through 3D coordinate calibration (accuracy ±1 μm) and SIFT feature matching to obtain grayscale images of clastic rock grain size and structure, such as... Figure 1 As shown.

[0040] Step 3: Identify the size and structure of clastic rock particles in grayscale images, and apply the equivalent sphere algorithm to obtain the characterization parameters of clastic rock particle size and structure.

[0041] Specifically, the acquired clastic rock particle size and structural grayscale images are binarized to segment the clastic particle regions; the volume V of each particle is calculated using the sphere volume formula V = ... πr 3 This leads to the equivalent sphere radius r = This is used as a characterization parameter for particle size; the sphericity and flatness of the particles are calculated as structural characterization parameters, where: sphericity S = (4πr 2 ) / A, where A is the particle surface area; flatness F=b / a, where a and b are the lengths of the major and minor axes of the particle.

[0042] Step 4: Obtain the screening criteria for the size and structure characterization parameters of clastic rock particles. Based on the screening criteria, determine different particle sizes and structure characterization parameters from the size and structure characterization parameters of clastic rock particles, and then construct a digital model of the size and structure composition of clastic rock particles.

[0043] Specifically, the screening criteria for clastic rock fragment size are as follows: exclude fragments with a volume < 10 pixels (corresponding to an actual volume < (resolution)). 3 For example, when the resolution is 1μm, the corresponding actual volume is <10μm. 3 Noise particles were selected, retaining particles with a diameter range of 5μm-2mm (covering the main particle size levels of clastic rocks). Clastic rock particle structure screening criteria: particles with sphericity >0.4 and flatness 0.3-1.5 were analyzed, excluding extremely irregular or flaky particles.

[0044] Step 5: Based on machine learning of micron-sized CT scan structural images, construct a standard machine learning model for the relationship between clastic rock particle size and structure.

[0045] Specifically, a U-Net convolutional neural network was used. A single-channel grayscale image patch of 128×128×128 voxels was input, processed through 5 layers of downsampling (3×3 convolutional kernels, stride 2) and 5 layers of upsampling (deconvolution + skip connections), outputting a particle boundary probability map. The training data consisted of 100 core samples, divided into training, validation, and test sets in a 7:2:1 ratio. Preprocessing steps included grayscale normalization, random rotation (±15°), and Gaussian noise addition (σ=0.05). The Adam optimizer was used with a learning rate of 0.001, weight decay of 1e-5, and a batch size of 16. The Dice loss and cross-entropy loss were jointly optimized, with a validation set Dice coefficient ≥0.95 used as the convergence condition, to construct a standard machine learning model for clastic rock particle size and structure.

[0046] Step 6: Based on the standard machine learning model for the size and structure of clastic rock particles, obtain a digital result model of the size and structure of clastic rock particles through machine learning.

[0047] Specifically, newly acquired grayscale images of clastic rock particle morphology and structure are input into a standard model. The model extracts abstract features from the images through an encoder, and the decoder maps these features back to the original image size, outputting predicted particle size and structural characterization parameters. Post-processing of the output results involves a clustering analysis algorithm that groups particles with similar characteristics based on features such as particle size, sphericity, and flatness, ultimately constructing a complete digital result model. Figure 2 and Figure 3 As shown.

[0048] Step 7: Based on the digital model of clastic rock particle size and structure, determine the dominant oil and gas reservoirs and the target fracturing intervals.

[0049] Specifically, the dominant reservoir is determined as follows: Dominant Reservoir Index (RQI) = 0.4D. avg +0.3(1 / S o +0.3P, S o =σg / Dg; where Davg is the average equivalent sphere radius, So is the particle size sorting coefficient, σg is the geometric standard deviation, Dg is the geometric mean particle size, and P is the porosity; when RQI > 0.6, it is determined to be a dominant reservoir.

[0050] ;

[0051] ;

[0052] Where, d i Let n be the equivalent radius of the particle. i Let N be the number of particles, and N be the total number of particles.

[0053] Determination of target fracturing intervals: Combining rock mechanical parameters (elastic modulus 60-80 GPa, Poisson's ratio 0.2-0.3) and geostress data (horizontal principal stress difference <5 MPa), the advantageous reservoir intervals with an average particle size of 100-300 μm, a particle size standard deviation of <50 μm, and a permeability of <1 mD are selected as the target fracturing intervals.

[0054] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0055] The contents not described in detail in this specification are existing technologies known to those skilled in the art.

Claims

1. A method for quantitative characterization of clastic rock particle size based on CT scan structural images, characterized in that, include: Step 1: Obtain the geological environment of the sedimentary rock formation of the target layer, and determine the image acquisition parameters of the clastic rock based on the geological environment; Step 2: Collect rock images based on the clastic rock image acquisition parameters, perform multi-scale image acquisition, and obtain grayscale images of clastic rock fragment size and structure; Step 3: Identify the size and structure of clastic rock particles in grayscale images, and apply the equivalent sphere algorithm to obtain the characterization parameters of clastic rock particle size and structure; Step 4: Obtain the screening criteria for the size and structure characterization parameters of clastic rock particles. Based on the screening criteria, determine different particle sizes and structure characterization parameters from the size and structure characterization parameters of clastic rock particles, and then construct a digital model of the size and structure composition of clastic rock particles. Step 5: Construct a standard machine learning model for the relationship between clastic rock particle size and structure based on micron-sized CT scan structural images; Step 6: Based on the standard machine learning model for the size and structure of clastic rock particles, obtain a digital result model of the size and structure of clastic rock particles through machine learning; Step 7: Based on the digital model of clastic rock particle size and structure, determine the dominant oil and gas reservoirs and the target fracturing intervals; In step 2, when acquiring multi-scale images, three scales are defined: macroscopic scale 1-10mm, mesoscopic scale 100μm-1mm, and microscopic scale 1-100μm. Images are acquired in order from macroscopic to microscopic. During the acquisition process, multi-scale image registration is achieved through three-dimensional coordinate calibration and SIFT feature matching to obtain grayscale images of clastic rock particle size and structure. In step 3, the acquired grayscale images of clastic rock particle size and structure are binarized to segment the clastic particle regions; the volume V of each particle is calculated using the formula for the volume of a sphere. Inversely deduce the equivalent sphere radius This is used as a characterization parameter for particle size; the sphericity and flatness of the particles are calculated as structural characterization parameters, where: sphericity S = (4πr 2 ) / A, where A is the particle surface area; flatness F = b / a, where a and b are the lengths of the particle's major and minor axes; In step 4, the screening criteria for clastic rock fragment size are as follows: noisy particles with a volume of <10 pixels are excluded, and particles with a diameter range of 5μm-2mm are retained; the screening criteria for clastic rock fragment structure are as follows: particles with sphericity >0.4 and flatness 0.3-1.5 are analyzed. In step 7, the dominant reservoir is determined as follows: the dominant reservoir index RQI = 0.4D. avg +0.3(1 / S o +0.3P, S o =σg / Dg; where D avg S is the average equivalent sphere radius. o RQI is the particle size sorting coefficient, σg is the geometric standard deviation, Dg is the geometric mean particle size, and P is the porosity; when RQI > 0.6, it is determined to be a dominant reservoir. Where, d i Let n be the equivalent radius of the particle. i Where is the number of particles, and N is the total number of particles; In step 7, the target fracturing segment is determined as follows: Based on rock mechanics parameters and geostress data, a superior reservoir segment with an average particle size of 100-300 μm, a particle size standard deviation of <50 μm, and a permeability of <1 mD is selected as the target fracturing segment.

2. The method for quantitative characterization of clastic rock particle size based on CT scan structural images according to claim 1, characterized in that, In step 1, the geological environment includes: sedimentary facies type, hydrodynamic conditions of the sedimentary environment, and diagenetic type; the determined clastic rock image acquisition parameters include: CT scan resolution of 0.5-100μm, scan angle of 0-360°, and one image acquired every 0.5-1°.

3. The method for quantitative characterization of clastic rock particle size based on CT scan structural images according to claim 1, characterized in that, In step 5, the U-Net convolutional neural network algorithm is used to construct a standard machine learning model for the size and structure of clastic rock particles. This includes: data preparation and preprocessing, U-Net model construction, model training, and model evaluation and optimization.

4. The method for quantitative characterization of clastic rock particle size based on CT scan structural images according to claim 1, characterized in that, In step 6, the newly acquired grayscale images of clastic rock particle size and structure are input into the standard model, and the predicted particle size and structure characterization parameters are output. The output results are then post-processed to finally construct a digital result model of clastic rock particle size and structure.

Images