A shale digital core reconstruction method for hydrogen flow behavior simulation
Through Markov chain Monte Carlo method and X-ray whole rock analysis combined with scanning electron microscopy identification, the shale digital core model considering mineral composition is reconstructed, solving the accuracy and calculation difficulty of hydrogen flow simulation in the prior art The problem has achieved more accurate shale hydrogen reservoir design and safety assessment.
Patent Information
- Application Number
- CN202310498546.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-06
- Publication Date
- 2024-12-17
- Estimated Expiration
- 2043-05-06
AI Technical Summary
The existing digital core reconstruction method ignores the component characteristics of the pore structure when simulating shale gas reservoirs, resulting in unclear understanding of the microscopic migration mechanism of hydrogen, long calculation time and low accuracy, and lack of consideration of shale pore type and whole minerals. Method of ingredients.
Markov chain Monte Carlo method combined with X-ray whole rock mineral analysis and scanning electron microscope organic/inorganic pore recognition to obtain the pore characteristics and mineral composition of the core, and reconstruct the three-dimensional digital core through multiple directions of two-dimensional scanning images. and calibrate the mineral components of inorganic pores.
A more accurate digital core model was obtained, which could more conveniently simulate the flow of hydrogen-containing gas in the core, which improved the accuracy of injection and procurement cycle design and reservoir capacity evaluation of shale hydrogen storage, and enhanced storage safety assessment.
Smart Images
Figure CN116793919B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of indoor simulation of oil and gas reservoirs, and in particular to a shale digital core reconstruction method for simulating hydrogen flow behavior. Background Art
[0002] Driven by the "dual carbon" goals, the low-carbon green energy pattern and paradigm are evolving in all directions. Hydrogen energy, as the most promising clean energy, is also undergoing a transformation from "grey hydrogen" to "blue hydrogen" to "green hydrogen". With the development of hydrogen production industry technology, hydrogen production capacity has increased, resulting in a surge in demand for large-scale hydrogen storage facilities. Shale gas reservoirs have the advantages of native sealing, large storage capacity, wide distribution, and complete injection and production facilities, making them a strong competitive place for large underground hydrogen storage. However, research on large-scale underground hydrogen storage has mostly focused on salt caverns, aquifers, and abandoned sandstone oil and gas reservoirs, and there are few reports on experiments and research on shale hydrogen storage.
[0003] The flow behavior, microscopic migration mechanism, and storage safety of hydrogen in shale are the many bottlenecks faced by the construction of shale hydrogen storage. As a mature method for constructing pore network models, digital core technology has been widely used in the simulation research field of various fluids at the microscopic pore scale. Traditional digital core reconstruction methods focus on the true restoration of pores and rock skeletons. The main methods for extracting and reconstructing pore structures are simulated annealing, sequential simulation, process method, and Monte Carlo method. Shale gas reservoirs have the characteristics of a wide pore scale span and strong heterogeneity, which makes the reconstruction methods that are effective for sandstone have problems such as long calculation time, high calculation difficulty, and low accuracy of pore feature reconstruction when applied to shale gas reservoirs.
[0004] The Markov chain Monte Carlo method is an improvement on the probability distribution Monte Carlo method and has been proven to be effective in reconstructing the pore structure of shale gas reservoirs. However, traditional digital core reconstruction ignores the compositional characteristics of the pore structure itself. The neglected compositional characteristics have a significant impact on the microscopic adsorption and diffusion mechanism of hydrogen, which will lead to the hidden danger of unclear understanding or incomplete consideration of the mechanism of hydrogen microscopic migration.
[0005] Currently, there is no digital core reconstruction method that takes into account the pore type and full mineral composition of shale. Summary of the invention
[0006] To solve at least one of the above problems, the present invention proposes a shale digital core reconstruction method for simulating hydrogen flow behavior. The digital core model finally obtained can simulate the flow of hydrogen-containing gas in the core more accurately and conveniently.
[0007] The technical solution of the present invention is as follows: a shale digital core reconstruction method for hydrogen flow behavior simulation comprises the following steps:
[0008] S1. Obtaining a two-dimensional scanning image of the core in at least three directions, and obtaining pore characteristics of the core, wherein the pore characteristics include pore types;
[0009] S2, obtaining the mineral composition of the core;
[0010] S3, based on the two-dimensional scanning image of the core and the pore characteristics of the core, the Markov chain Monte Carlo method is used to obtain the reconstructed digital core intermediate;
[0011] S4. According to the content of minerals in the core, the mineral composition of inorganic pores is calibrated using the Markov chain Monte Carlo method to obtain the reconstructed digital core.
[0012] The beneficial effect of the present invention is that the present invention combines X-ray whole-rock mineral analysis, scanning electron microscope organic / inorganic pore identification and Markov chain Monte Carlo method to calibrate the pore structure and composition of the digital core, so that the final digital core model can more accurately and conveniently simulate the flow of hydrogen-containing gas in the core. The final result is of great practical significance for the reasonable injection and production cycle design, storage capacity, and safety and airtightness evaluation of shale hydrogen storage. BRIEF DESCRIPTION OF THE DRAWINGS
[0013] Figure 1 This is an example of obtaining a reconstructed digital core intermediate in S3. DETAILED DESCRIPTION
[0014] In order to have a clearer understanding of the technical features, purposes and beneficial effects of the present invention, the technical solution of the present invention is now described in detail below, but it should not be understood as limiting the scope of implementation of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of the present invention.
[0015] The present invention is further described below in conjunction with embodiments.
[0016] In the following embodiments, the “voxel” is a basic point that constitutes a three-dimensional space, similar to a pixel point in a two-dimensional image.
[0017] A shale digital core reconstruction method for hydrogen flow behavior simulation includes the following steps:
[0018] S1. Obtain two-dimensional scanning images of the core in at least three directions, and obtain pore characteristics of the core, wherein the pore characteristics include pore types.
[0019] Specifically, in this step, a variety of methods can be used to obtain the two-dimensional scanning image, but based on actual accuracy, reliability and convenience, SEM (scanning electron microscope) is usually used for acquisition. Since the method of using SEM to obtain a two-dimensional scanning image of the core is a common method in the art, its specific operation will not be described in detail here.
[0020] At the same time, since the core is a three-dimensional structure, it is necessary to obtain at least two-dimensional scanning images of the core in three directions, so that the digital core finally reconstructed is a three-dimensional image. From the perspective of the difficulty of actual modeling, it is preferred to use the Cartesian three-dimensional coordinate system as the extension direction of the three faces to obtain three two-dimensional scanning images of the core, and one axis of the Cartesian three-dimensional coordinate system passes through the center line of the core. Of course, those skilled in the art can obtain different two-dimensional scanning images in different ways and directions according to actual conditions.
[0021] After obtaining the two-dimensional scanning image of the core, the threshold segmentation method is used to distinguish the skeleton and pores in each two-dimensional image, and finally the pore types of the core including pore size distribution, organic pores, and inorganic pores can be obtained. The threshold segmentation method is a common method for processing images, so its specific operation will not be described in detail here.
[0022] S2, obtaining the mineral composition of the core;
[0023] In this process, the core is ground into powder and its mineral composition is analyzed using an X-ray diffractometer, and the mineral composition of the core can finally be obtained.
[0024] Of course, there are other methods in the art to obtain the mineral composition of the core, such as chromatography, etc., but considering the wide range of use, economy and accuracy, an X-ray diffractometer is selected in this embodiment to measure the components of the core.
[0025] S3, based on the two-dimensional scanning image of the core and the pore characteristics of the core, the Markov chain Monte Carlo method is used to obtain the reconstructed digital core intermediate;
[0026] Specifically, the process mainly includes the following steps:
[0027] First, the two-dimensional scanning image features of the core in multiple directions are extracted to determine the pore stationary probability distribution function in each image; in this embodiment, the two-dimensional scanning image is first binarized, and then the binarized image is traversed and sampled with 2, 3 and 5 neighbors to finally determine its pore probability density function.
[0028] In order to further enable those skilled in the art to understand the operation of traversing and sampling an image with multiple neighbors, an example is given below using two neighbors:
[0029] 2-neighbor sampling mainly includes the joint probability of 2 voxels. Each voxel has two situations: skeleton and pore. Therefore, the probability density function of 2-neighbor sampling is shown in Table 1:
[0030] Table 12 Probability distribution of neighbor sampling
[0031]
[0032] In the figure, the black blocks represent pores and the white blocks represent skeletons. According to the two-dimensional scanning image, the corresponding probability density function can be obtained.
[0033] Similarly, for 3-neighbor sampling, there are 8 cases in total, and for 5-neighbor sampling, there are 32 cases in total. All can be obtained through 2D scanning images.
[0034] like Figure 1 As shown, the first direction in the three-dimensional direction is selected for one-dimensional reconstruction: based on the porosity of the core, the state of the first voxel in this direction is randomly assigned, and the voxel in this direction is predicted based on the Markov chain Monte Carlo method.
[0035] For the Markov Chain Monte Carlo method, any point in the three-dimensional voxel model satisfies the following conditions:
[0036] p(x ijk |{x lmn :0<l<i,0<m<j,0<n<k})=p(x ijk |x i-1,j,k ,x i,j-1,k ,x i,j,k-1 ) Formula 1
[0037] In the formula, x ijk represents the voxel at the three-dimensional coordinates i, j, and k; p represents the probability distribution function.
[0038] The above formula indicates that for a point (i, j, k) in a three-dimensional voxel model, its state x ijk It is only related to the states of the neighboring points of the point in three directions. According to the voxels of different boundaries, the state estimation in three directions is carried out respectively. The joint probability is the average value of the sum of the probabilities in the three directions, which is finally shown in Formula 2:
[0039] p(x ijk )=α[p(x i )+p(x j )+p(x k )] Formula 2
[0040] In the formula, p(x i ) is the state probability distribution function from the i direction; p(x j) is the state probability distribution function in the j direction; p(x k ) is the state probability distribution function from the k direction; i, j, k are the three directions for obtaining the two-dimensional scanning image of the core; α represents the coefficient, which is usually taken as Represents the average value of the state probability distribution function in three directions.
[0041] Select the second direction in the three-dimensional direction to carry out two-dimensional reconstruction: based on the one-dimensional reconstruction, the voxels on the two-dimensional plane are predicted based on the Markov chain Monte Carlo method;
[0042] The third direction of the three-dimensional core is selected to carry out three-dimensional reconstruction: based on the two-dimensional reconstruction and based on the Markov chain Monte Carlo method, the voxels in the three-dimensional space are predicted, and finally the reconstructed digital core intermediate is obtained.
[0043] In the above reconstruction process, neighbor sampling is needed to predict it. Generally speaking, for one-dimensional reconstruction, the specific composition (skeleton or pore) of the first voxel needs to be randomly predicted based on the actual porosity of the core; for the remaining subsequent voxels, two-neighbor sampling is usually used to predict them. For two-dimensional reconstruction, the first voxel can be predicted by three-neighbor sampling, and the subsequent voxels can be predicted by five-neighbor sampling. For three-dimensional reconstruction, five-neighbor sampling in three directions is usually used for prediction. Of course, those skilled in the art can also use the remaining multi-neighbor sampling for prediction.
[0044] However, in actual situations, after the reconstruction of the digital core is completed, the porosity parameters of the reconstructed digital core may have certain errors with the porosity parameters of the actual core. When the error is too large, the coefficient α needs to be corrected. The value of the corrected coefficient α is as follows:
[0045]
[0046] In the formula, Δα is the correction value of coefficient α, is the porosity after reconstruction, is the average porosity of three scanning surfaces.
[0047] Although theoretically, as long as the scanning equipment has high enough accuracy, or the core is divided into multiple blocks and scanned separately, the scanned images can be pieced together into a two-dimensional image in the same direction. However, both methods have corresponding defects: either the cost is relatively high or the workload is large. Therefore, in this embodiment, a low-resolution image (i.e., a large scanning area) can be scanned first, and then several high-resolution images (i.e., a small part of the low-resolution image is scanned), and then the low-resolution image and the high-resolution image are superimposed.
[0048] In this embodiment, the low-resolution image and the high-resolution image are superimposed by using an improved superposition algorithm to reconstruct images of different resolutions, and the organic pores are represented by 0, the inorganic pores are represented by 1, and the skeleton is represented by 2 to rewrite the digital core data volume. The specific superposition algorithm is shown in the following formula:
[0049]
[0050] S4. According to the content of minerals in the core, the mineral composition of inorganic pores is calibrated using the Markov chain Monte Carlo method to obtain the reconstructed digital core;
[0051] The first direction in the three-dimensional direction is selected for one-dimensional reconstruction: based on the mineral composition of the core, the mineral composition of the first voxel in this direction is randomly assigned a state, and the mineral composition of the voxel in this direction is predicted based on the Markov chain Monte Carlo method;
[0052] The second direction in the three-dimensional direction is selected to carry out two-dimensional reconstruction: based on the one-dimensional reconstruction, the mineral composition of the voxels on the two-dimensional plane is predicted based on the Markov chain Monte Carlo method;
[0053] The third direction in the three-dimensional core is selected to carry out three-dimensional reconstruction: based on the two-dimensional reconstruction and the Markov chain Monte Carlo method, the mineral composition of the voxels in the three-dimensional space is predicted, and finally the reconstructed digital core is obtained.
[0054] In this step, similar to S3, the prediction of voxels needs to be based on the corresponding probability density function, but the difference from S3 is that the probability density function in this step is the percentage of the content of each mineral, and its probability is relatively certain.
[0055] For rock cores, they usually contain multiple minerals. Therefore, when reconstructing, one of the minerals can be reconstructed and calibrated first. When the mineral is completely calibrated in the three-dimensional direction, the other mineral can be calibrated. When all minerals are completely calibrated, the digital rock core is reconstructed.
[0056] In some cases, we only need to study the effect of one or several minerals on the flow of hydrogen. Therefore, we can also calibrate these minerals according to the above method.
[0057] S5. Extract the pore structure of the reconstructed digital core model using the maximum sphere method.
[0058] In fact, after step S4, the reconstruction of the digital core is completed. However, in order to make the pores of the digital core closer to the pores of the real core, the pores are corrected using the maximum sphere method.
[0059] The maximum sphere method requires that each maximum sphere is tangent to the skeleton particles or boundaries in the digital core and ensures that the maximum spheres are not subsets of each other. down 2 and R up 2 Represent the square of the lower and upper radii of the largest sphere, where R up 2 Specifically, it represents the center of the sphere C(x c ,y c ,z c ) to the nearest skeleton voxel V(x g ,y g ,z g ) squared:
[0060] R up 2 =dist 2 (C,V g )=(x c -x g ) 2 +(y c -y g ) 2 +(z c -z g ) 2 (5)
[0061] R down 2 Then the center of the sphere C(x c ,y c ,z c ) to any pore element V(x,y,z):
[0062] R down 2 = max{dist 2 (V,C)|dist 2 (V,C)<R up 2} (6)
[0063] First, expand the search outward from the face, edge, and vertex of any single voxel, stop when encountering a skeleton voxel or image boundary, and take the smallest square distance L 2 As a follow-up control search range.
[0064] According to L 2 , find the square of the distance between the lower limit and the upper limit corresponding to each voxel, that is, R down 2 With R up 2; Inscribe the largest sphere that meets the conditions. Remove the partially inscribed spheres that are completely contained, and select the sphere with the largest local radius (radius r maxi ) starts to search for another adjacent local maximum ball (with a radius of r maxj ) Stop. The entire digital core is traversed by taking the entire pore throat as a connected pore throat, thereby completing the construction of the real pore network.
[0065] The present invention can reconstruct the digital core of shale considering the mineral composition and pore type; it has important practical significance for the reasonable injection and production cycle design, storage capacity, and safety and airtightness evaluation of shale hydrogen storage under the background of "dual carbon". The digital core reconstruction method of shale used in this paper can be applied to the simulation study of the microscopic migration mechanism of hydrogen and shale, and improve the authenticity and accuracy of the simulation of the microscopic migration mechanism of hydrogen.
[0066] In fact, the digital core obtained by the method of the embodiment of the present invention can be used not only for hydrogen flow analysis, but also for flow analysis of other gases that can be adsorbed in the core, such as carbon dioxide, natural gas, etc. However, from the perspective of actual use, the adsorption degree of hydrogen is greater and has more practical research significance.
[0067] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the embodiments of the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.
Claims
1. A shale digital core reconstruction method for hydrogen flow behavior simulation, characterized in that: The following steps are involved: S1. Obtaining a two-dimensional scanning image of the core in at least three directions, and obtaining pore characteristics of the core, wherein the pore characteristics include pore types; S2, obtaining the mineral composition of the core; S3, based on the two-dimensional scanning image of the core and the pore characteristics of the core, a Markov chain Monte Carlo method is used to obtain a reconstructed digital core intermediate; this step includes the following sub-steps: Extract the two-dimensional scanning image features of the core in multiple directions and determine the pore stationary probability distribution function in each image; The first direction in the three-dimensional direction is selected for one-dimensional reconstruction: based on the porosity of the core, the state is randomly assigned to the first voxel in this direction, and the voxel in this direction is predicted based on the Markov chain Monte Carlo method; Select the second direction in the three-dimensional direction to carry out two-dimensional reconstruction: based on the one-dimensional reconstruction, the voxels on the two-dimensional plane are predicted based on the Markov chain Monte Carlo method; The third direction of the 3D core is selected to carry out 3D reconstruction: based on the 2D reconstruction and based on the Markov chain Monte Carlo method, the voxels in the 3D space are predicted, and finally the reconstructed digital core intermediate is obtained; In the above steps, prediction is made based on the following formula: p(x ijk )=α[p(x i )+p(x j )+p(x k )]; Where: p(x i ) is the state probability distribution function from the i direction; p(x j ) is the state probability distribution function in the j direction; p(x k ) from the state probability distribution function in the k direction, i, j, k are the three directions for obtaining the two-dimensional scanning image of the core, and α represents the coefficient; When the reconstructed digital core intermediate is obtained, the coefficient α can be corrected by the following formula: In the formula, Δα is the correction value, is the porosity after reconstruction, is the average porosity of the three scanned surfaces; S4. According to the content of minerals in the core, the mineral composition of inorganic pores is calibrated using the Markov chain Monte Carlo method to obtain a reconstructed digital core. This step includes the following sub-steps: The first direction in the three-dimensional direction is selected for one-dimensional reconstruction: based on the mineral composition of the core, the mineral composition of the first voxel in this direction is randomly assigned a state, and the mineral composition of the voxel in this direction is predicted based on the Markov chain Monte Carlo method. The second direction in the three-dimensional direction is selected to carry out two-dimensional reconstruction: based on the one-dimensional reconstruction, the mineral composition of the voxels on the two-dimensional plane is predicted based on the Markov chain Monte Carlo method; The third direction in the three-dimensional direction is selected to carry out three-dimensional reconstruction: based on the two-dimensional reconstruction and based on the Markov chain Monte Carlo method, the mineral composition of the voxels in the three-dimensional space is predicted, and finally the reconstructed digital core is obtained; Between S3 and S4, the following steps are also included: using an improved superposition algorithm to superimpose reconstructed images with different resolutions, the improved superposition algorithm is as follows: In the formula, 0 represents organic pores, 1 represents inorganic pores, and 2 represents the skeleton.
2. The method according to claim 1, characterized in that In S1, two-dimensional scanning images of the core in three mutually perpendicular directions are obtained based on SEM.
3. The method according to claim 1, characterized in that: In S1, a threshold segmentation method is used to distinguish pores and skeletons in the two-dimensional scanning image, so as to obtain the pore characteristics of the core, and the pore characteristics of the core include pore size distribution, organic pores and inorganic pores.
4. The method according to claim 1, characterized in that: In S2, the mineral composition of the core is obtained based on X-ray diffraction mineral analysis.
5. The method according to claim 1, characterized in that After calibrating the mineral components of the inorganic pores, the method further includes: extracting the pore structure of the reconstructed digital core model by using the maximum sphere method.