Glutenite reconstruction prediction permeability numerical simulation calculation method based on two-dimensional image and computer readable storage medium
By using semantic segmentation and finite element modeling based on deep neural networks, the problem of large calculation errors in sandstone and conglomerate permeability in traditional methods has been solved, achieving more accurate permeability prediction and more realistic model construction.
Patent Information
- Application Number
- CN202511191057.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-22
- Publication Date
- 2025-12-12
AI Technical Summary
Traditional methods use simple spheres to replace particles in modeling, which cannot accurately take into account the huge differences in particle size between large and small particles in sandstone and conglomerate, as well as the complex morphology of large particles. This leads to inaccurate reconstruction of the seepage model, resulting in a large error in the predicted permeability calculation.
Semantic segmentation based on deep neural networks is used to divide the two-dimensional image of sandstone into a first grain block and a second grain block. The two blocks are modeled and coupled separately to construct a finite element planar model for seepage simulation and to calculate and predict the permeability.
The accuracy of permeability calculation has been improved, the actual effects of large and small particles have been fully considered, the rationality and reliability of the model have been enhanced, and the seepage characteristics of sandstone and conglomerate have been comprehensively reflected.
Smart Images
Figure CN121120983A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of sandstone permeability prediction, and particularly relates to a sandstone reconstruction permeability prediction numerical simulation calculation method based on a two-dimensional image and a computer readable storage medium. BACKGROUND
[0002] In the fields of oil exploration and geological research, it is crucial to accurately obtain the predicted permeability of rocks, which is of great significance for evaluating oil and gas reserves and predicting oil reservoir recovery efficiency. With the development of technology, using two-dimensional rock images to estimate the numerical value of predicted permeability has become an important research method.
[0003] In the past, when reconstructing a seepage model based on a two-dimensional rock image, a relatively simple and general method is usually used. That is, a model is built based on the particle size distribution in the two-dimensional rock image, and a sphere is used to replace the particles in the model. This method can achieve relatively good results when dealing with rocks with good particle size sorting, and can accurately simulate the seepage capacity. Because in rocks with good particle size sorting, the particle size is relatively uniform, and the use of a sphere to approximate the particles has little effect on the overall seepage characteristics, the model built based on this can reflect the actual seepage situation to a certain extent.
[0004] However, for sandstone, the situation is much more complex. The sandstone has poor sorting, and the particle size of large particles can differ by several orders of magnitude compared to small particles. In this case, if the traditional modeling method of replacing particles with spheres is still used, it will cause a large error in calculating the predicted permeability. For example, large particles play a key role in supporting the framework in sandstone, and their size and distribution have a decisive influence on the flow channel of fluid. Due to the large difference in particle size between large particles and small particles, the simple replacement of spheres cannot accurately reflect the real role of large particles in the seepage process, resulting in a large deviation between the calculated seepage path and flow and the actual situation.
[0005] At the same time, the morphology of large particles in sandstone is also complex. Large particles are not simple regular spheres, and their morphological characteristics have a significant impact on the predicted permeability. The shape, surface roughness, and arrangement of large particles will significantly affect the flow characteristics of fluid in rock pores. If the gravel of large particles is replaced by a sphere, the influence of these complex morphological characteristics on the predicted permeability cannot be accurately simulated, resulting in a large deviation in the calculation results of the predicted permeability.
[0006] In summary, the traditional method of reconstructing a seepage model based on a two-dimensional rock image by replacing particles with spheres has obvious limitations when it comes to sandstone, which has poor sorting and complex morphology of large particles. SUMMARY
[0007] The technical problem solved by the present application is that the related technology mentioned in the background art has at least one defect: in the related technology, because a simple sphere is used to replace particle modeling, the difference in particle size between large particles and small particles of the sandstone is too large, the shape of the large particles is complex, and other factors cannot be accurately considered, resulting in inaccurate reconstruction of the seepage model, and further causing large calculation error of the predicted permeability, and the present application provides a sandstone reconstruction predicted permeability numerical simulation calculation method based on a two-dimensional image and a computer readable storage medium.
[0008] The technical solution adopted by the present application to solve its technical problem is: a sandstone reconstruction predicted permeability numerical simulation calculation method based on a two-dimensional image is constructed, comprising the following steps:
[0009] S1, based on a deep neural network, performing semantic segmentation on a two-dimensional image of the sandstone to divide the two-dimensional image of the sandstone into a first particle block and a second particle block, the particle size of the first particle block being larger than the particle size of the second particle block;
[0010] S2, dividing the second particle block to obtain a plurality of second particle block units, modeling each second particle block unit until a three-dimensional reconstruction seepage model of the second particle block based on discrete elements is obtained;
[0011] S3, in the process of converting the two-dimensional image of the sandstone into a finite element plane model, for the first particle block, the original shape of the finite element plane model of the first particle block is retained, for the second particle block, the finite element plane model of the second particle block is obtained by processing the three-dimensional reconstruction seepage model, the finite element plane model of the first particle block is coupled with the finite element plane model of the second particle block to obtain the finite element plane model of the two-dimensional image of the sandstone, and a seepage simulation experiment is performed in the finite element plane model of the two-dimensional image of the sandstone to calculate the predicted permeability of the original two-dimensional image of the sandstone.
[0012] In some embodiments, step S1 comprises:
[0013] S11, obtaining a plurality of two-dimensional images of the sandstone to be studied;
[0014] S12, processing the obtained two-dimensional images of the sandstone;
[0015] S13, establishing a two-dimensional image semantic segmentation network, then using an image labeling tool to manually label the first particle block and the second particle block in each processed two-dimensional image of the sandstone to obtain a training data set, training the two-dimensional image semantic segmentation network through the training data set to obtain a trained two-dimensional image semantic segmentation network;
[0016] S14, performing semantic segmentation on the two-dimensional images of the sand-shale rock to be studied by using the trained complete two-dimensional image semantic segmentation network, so as to divide the two-dimensional images of the sand-shale rock into the first particle block and the second particle block.
[0017] In some embodiments, in step S2, modeling each second particle block unit includes:
[0018] statistically counting the particle size distribution and porosity in each second particle block unit, and modeling each second particle block unit according to the particle size distribution and the porosity.
[0019] In some embodiments, step S2 includes:
[0020] S21, dividing the second particle block to obtain a plurality of second particle block units;
[0021] S22, counting the number, proportion and particle size distribution range information of particles of different sizes in each second particle block unit, obtaining a histogram of the particle size distribution in each second particle block unit, and calculating the porosity in each second particle block unit according to the distribution relationship between particles and pores in each second particle block unit;
[0022] S23, modeling each second particle block unit by using a discrete element method according to the particle size distribution and porosity information counted in each second particle block unit, and after completing the discrete element modeling of each second particle block unit, splicing and integrating the discrete element models of all second particle block units according to the positional relationship of the second particle block units in the two-dimensional image of the sand-shale rock to gradually build a three-dimensional reconstruction percolation model of the second particle block based on discrete elements.
[0023] In some embodiments, in step S22, counting the number, proportion and particle size distribution range information of particles of different sizes in each second particle block unit includes:
[0024] recognizing the outline of each particle by using an image edge detection algorithm and measuring the particle size, and aggregating the particle sizes of all particles in each second particle block unit to obtain the number, proportion and particle size distribution range information of particles of different sizes in each second particle block unit.
[0025] In some embodiments, in step S22, calculating the porosity in each second particle block unit according to the distribution relationship between particles and pores in each second particle block unit includes:
[0026] Within each of the second particle block units, the particle portion is marked with a first color and the porous portion is marked with a second color. The porosity within the second particle block unit is then determined by calculating the proportion of the area of the second color within the entire area of the second particle block unit.
[0027] In some embodiments, step S3 includes:
[0028] S31. Using image vectorization technology, the first grain block and the second grain block in the two-dimensional image of the sandstone are converted into vector images respectively;
[0029] S32. Import the vector diagrams of the first particle block and the second particle block into the finite element software for modeling. During the finite element modeling process, for the first particle block, retain the original form of the finite element plane model of the first particle block. For the second particle block, use the three-dimensional reconstructed seepage model. Through data mapping and transformation, map the particle information and pore structure in the three-dimensional reconstructed seepage model to the finite element plane model of the second particle block.
[0030] S33. Couple the finite element plane model of the first particle block with the finite element plane model of the second particle block to obtain the finite element plane model of the two-dimensional image of the sandstone.
[0031] S34. Perform a seepage simulation experiment on the finite element plane model of the two-dimensional image of the sandstone and calculate the predicted permeability of the original two-dimensional image of the sandstone.
[0032] In some embodiments, step S31 includes:
[0033] S311. Perform edge detection on the two-dimensional image of the sandstone and conglomerate;
[0034] S312. Extract the complete contour information of the first particle block and the second particle block using a contour tracking algorithm;
[0035] S313. Convert the complete outline information of the first particle block and the second particle block into vector graphics.
[0036] In some embodiments, the first particle block is the block containing particles with a particle size ≥ 2 mm, and the second particle block is the block containing particles with a particle size < 2 mm.
[0037] The present invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the numerical simulation calculation method for predicting permeability of sandstone and conglomerate based on two-dimensional images as described in any of the preceding claims.
[0038] By implementing this invention, the following beneficial effects are achieved:
[0039] This invention, through semantic segmentation based on deep neural networks, can accurately distinguish between the first particle block (large particle block) and the second particle block (small particle block), avoiding simplistic treatment of complex particle morphologies. In subsequent modeling and calculation, the first particle block (large particle block) and the second particle block (small particle block) are fully considered, making the seepage path and flow rate calculations more realistic. This not only improves calculation accuracy and provides reliable data for related fields, but also makes the finite element planar model more consistent with the actual structure of sandstone and conglomerate, improving the model's rationality and reliability, and more comprehensively showcasing the seepage characteristics of sandstone and conglomerate. This can improve the accuracy of predictive permeability calculations for sandstone and conglomerate based on two-dimensional images. Attached Figure Description
[0040] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings:
[0041] Figure 1 A flowchart illustrating an embodiment of the present invention's numerical simulation calculation method for predicting permeability of sandstone and conglomerate based on two-dimensional images is shown.
[0042] Figure 2 Two-dimensional images of several sandstone and conglomerate rocks to be studied are shown;
[0043] Figure 3 This demonstrates the use of a two-dimensional image semantic segmentation network for... Figure 2 Binary images obtained by segmenting two-dimensional images of several sandstone and conglomerate rocks to be studied;
[0044] Figure 4 The first particle block is shown in Figure 2 A schematic diagram of a two-dimensional image of the sandstone and conglomerate to be studied;
[0045] Figure 5 The second particle block is shown in Figure 2 A schematic diagram of a two-dimensional image of the sandstone and conglomerate to be studied;
[0046] Figure 6 The second particle block is shown in Figure 2 A schematic diagram of pores on a two-dimensional image of the sandstone and conglomerate to be studied;
[0047] Figure 7 A histogram of particle size distribution within a certain second particle block unit is shown;
[0048] Figure 8 It shows the basis Figure 7 The three-dimensional reconstructed seepage model based on discrete element method was obtained by considering the particle size distribution and porosity of the second particle block.
[0049] Figure 9 It shows Figure 2 A schematic diagram of the three-dimensional reconstructed seepage model of the second particle block of samples 1-5, the distribution of fluid velocity in the model, and the distribution of fluid pressure in the model;
[0050] Figure 10 It shows Figure 2 A schematic diagram of the three-dimensional reconstructed seepage model of the second particle block of samples 6-10, the distribution of fluid velocity in the model, and the distribution of fluid pressure in the model;
[0051] Figure 11 It shows Figure 2 The finite element planar model of sample 1 in the sample;
[0052] Figure 12 It shows Figure 11 A schematic diagram of the visualization process of seepage simulation experiment using the finite element planar model of sample 1. Detailed Implementation
[0053] To provide a clearer understanding of the technical features, objectives, and effects of the present invention, specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.
[0054] It should be noted that the flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.
[0055] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.
[0056] To address the technical problem in related technologies where the use of simple spheres to replace particles in modeling fails to accurately account for factors such as the significant differences in particle size between large and small particles in sandstone and conglomerate, and the complex morphology of large particles, leading to inaccurate seepage model reconstruction and thus large errors in predicted permeability calculation, this invention provides a numerical simulation calculation method for predicted permeability of sandstone and conglomerate based on two-dimensional images. This method can improve the accuracy of predicted permeability calculation results for sandstone and conglomerate based on two-dimensional images.
[0057] like Figure 1As shown, some embodiments of the present invention disclose a numerical simulation calculation method for predicting permeability of sandstone and conglomerate based on two-dimensional images, including steps S1, S2 and S3, as follows:
[0058] S1. Based on deep neural networks, semantic segmentation is performed on the two-dimensional image of sandstone and conglomerate to divide the two-dimensional image of sandstone and conglomerate into a first grain block and a second grain block. The grain size of the first grain block is larger than that of the second grain block, that is, the first grain block is a large grain block and the second grain block is a small grain block.
[0059] S2. Divide the second particle block into several second particle block units, and model each second particle block unit until a three-dimensional reconstructed seepage model of the second particle block based on discrete elements is obtained.
[0060] S3. In the process of converting the two-dimensional image of sandstone into a finite element planar model, for the first grain block, the original shape of the finite element planar model of the first grain block is retained. For the second grain block, the finite element planar model of the second grain block is obtained by processing the three-dimensional reconstructed seepage model obtained in step S2. The finite element planar model of the first grain block is coupled with the finite element planar model of the second grain block to obtain the finite element planar model of the two-dimensional image of sandstone. Seepage simulation experiments are carried out in the finite element planar model of the two-dimensional image of sandstone to calculate the predicted permeability of the original two-dimensional image of sandstone.
[0061] This invention, through semantic segmentation based on deep neural networks, can accurately distinguish between the first particle block (large particle block) and the second particle block (small particle block), avoiding simplistic treatment of complex particle morphologies. In subsequent modeling and calculation, the first particle block (large particle block) and the second particle block (small particle block) are fully considered, making the seepage path and flow rate calculations more realistic. This not only improves calculation accuracy and provides reliable data for related fields, but also makes the finite element planar model more consistent with the actual structure of sandstone and conglomerate, improving the model's rationality and reliability, and more comprehensively showcasing the seepage characteristics of sandstone and conglomerate. This can improve the accuracy of predictive permeability calculations for sandstone and conglomerate based on two-dimensional images.
[0062] In some embodiments, the first particle block is the block containing particles with a particle size (diameter) ≥ 2 mm, and the second particle block is the block containing particles with a particle size (diameter) < 2 mm.
[0063] In some embodiments, step S1 includes S11, S12, S13, and S14, as follows:
[0064] S11, obtain as follows Figure 2Two-dimensional images of several sandstone and conglomerate rocks to be studied are shown, such as electron micrographs, CT slices, or thin sections of cast bodies.
[0065] S12. Process the acquired two-dimensional image of the sandstone and conglomerate.
[0066] Step S12 includes:
[0067] S121. Convert the acquired two-dimensional image of sandstone and conglomerate to grayscale to facilitate subsequent processing.
[0068] S122. A filtering algorithm is used to denoise the two-dimensional image of sandstone and conglomerate after grayscale processing, wherein the filtering algorithm includes median filtering and / or Gaussian filtering.
[0069] S123. Normalize the two-dimensional image of the sandstone and conglomerate after noise reduction.
[0070] S13. Establish a two-dimensional image semantic segmentation network (such as U-Net two-dimensional image semantic segmentation network, etc.), and then use image annotation tools (such as Labelme software, etc.) to manually annotate the first and second grain blocks in the two-dimensional images of each processed sandstone and conglomerate to obtain a training dataset. Train the two-dimensional image semantic segmentation network using the training dataset to obtain a fully trained two-dimensional image semantic segmentation network.
[0071] S14. Using a well-trained two-dimensional image semantic segmentation network, semantic segmentation is performed on several two-dimensional images of sandstone and conglomerate to be studied, so as to divide the two-dimensional images of sandstone and conglomerate into a first grain block and a second grain block.
[0072] Please refer to Figure 3 , Figure 4 and Figure 5 , Figure 3 This demonstrates the use of a two-dimensional image semantic segmentation network for... Figure 2 The image shows a binary representation of several two-dimensional images of sandstone and conglomerate to be studied, segmented into gray first grain blocks and black second grain blocks. Figure 4 The first particle block is shown in Figure 2 A schematic diagram of a two-dimensional image of the sandstone and conglomerate to be studied. Figure 5 The second particle block is shown in Figure 2 A schematic diagram of a two-dimensional image of the sandstone and conglomerate to be studied.
[0073] In some embodiments, step S2 involves modeling each second particle block unit, specifically including:
[0074] The particle size distribution and porosity within each second particle block unit are statistically analyzed, and each second particle block unit is modeled based on the particle size distribution and porosity.
[0075] In some embodiments, step S2 involves dividing the second particle block into several second particle block units, statistically analyzing the particle size distribution and porosity within each second particle block unit, and modeling each second particle block unit based on the particle size distribution and porosity until a three-dimensional reconstructed seepage model of the second particle block based on discrete element method is obtained, including S21, S22, and S23, as detailed below:
[0076] S21. Divide the second particle block to obtain several second particle block units.
[0077] The process involves using a segmentation algorithm to divide the second particle block. The chosen segmentation algorithm should be able to reasonably determine the size and shape of the second particle block unit based on factors such as image resolution and the distribution characteristics of the second particle (small particles), ensuring that each second particle block unit contains sufficient particle information without losing local details due to excessive size. For example, an adaptive mesh generation algorithm can be used based on the density distribution of the second particle (small particles), making the mesh size smaller in densely populated areas than in sparsely populated areas. In other words, the mesh size is relatively small in densely populated areas and relatively large in sparsely populated areas.
[0078] S22. Within each second particle block unit, statistically analyze the number, proportion, and distribution range of particles of different sizes to obtain a histogram of particle size distribution within each second particle block unit. Figure 7 A histogram of particle size distribution within a certain second particle block unit is shown. Simultaneously, based on the distribution relationship between particles and pores within each second particle block unit, the porosity within each second particle block unit is calculated.
[0079] In step S22, the number, proportion, and distribution range of particles of different sizes are statistically analyzed within each second particle block unit. Specifically, this includes:
[0080] By using an image edge detection algorithm, the outline of each particle is identified and its particle size is measured. The particle size of all particles in each second particle block unit is summarized to obtain information on the number, proportion, and distribution range of particles of different sizes in each second particle block unit.
[0081] In step S22, the porosity within each second particle block unit is calculated based on the distribution relationship between particles and pores within that unit. This specifically includes:
[0082] Figure 6 The second particle block is shown in Figure 2A schematic diagram of pores on a two-dimensional image of the conglomerate to be studied. Within each second grain block unit, the grain portion is marked with the first color and the pore portion is marked with the second color. The porosity within the second grain block unit is then determined by calculating the proportion of the second-colored area within the entire second grain block unit area.
[0083] S23. Based on the particle size distribution and porosity information obtained statistically within each second particle block unit, the discrete element method (DEM) is used to model each second particle block unit. After completing the DEM modeling of each second particle block unit, the DEM models of all second particle block units are stitched together and integrated according to the positional relationship of their original second particle blocks in the two-dimensional image of the sandstone and conglomerate, gradually constructing a three-dimensional reconstructed seepage model of the second particle block based on the DEM. Figure 8 It shows the basis Figure 7 The three-dimensional reconstructed seepage model based on discrete element method was obtained by considering the particle size distribution and porosity in the particles.
[0084] In this process, it is necessary to ensure that the particles between adjacent second-grain blocks can maintain reasonable physical connections and interactions at the splicing points, so as to truly reflect the three-dimensional spatial distribution and interrelationships of the second grains (small grains) in the conglomerate.
[0085] In some embodiments, the numerical simulation calculation method for predicting permeability of sandstone and conglomerate based on two-dimensional images further includes:
[0086] S4. Import the constructed three-dimensional reconstructed seepage model of the second particle block into the software (such as COMSOL Multiphysics software), set the physical properties, boundary conditions and initial conditions of the fluid; start the simulation calculation (such as CFD simulation calculation based on Darcy's law, etc.) to simulate the flow process of the fluid in the three-dimensional pore network, and then perform visualization analysis on the simulation results to obtain at least one of the following: predicted permeability, fluid velocity distribution in the model, fluid pressure distribution in the model and streamline information of the three-dimensional reconstructed seepage model.
[0087] Please refer to Figure 9 and Figure 10 , Figure 9 It shows Figure 2 The diagram shows the three-dimensional reconstructed seepage model of the second particle block of samples 1-5, the distribution of fluid velocity in the model, and the distribution of fluid pressure in the model. Figure 10 for Figure 2 A schematic diagram of the three-dimensional reconstructed seepage model of the second particle block of samples 6-10, the distribution of fluid velocity in the model, and the distribution of fluid pressure in the model.
[0088] In some embodiments, step S3 includes S31, S32, S33, and S34, as follows:
[0089] S31. Using image vectorization technology, the first and second grain blocks in the two-dimensional image of sandstone and conglomerate are converted into vector images respectively.
[0090] In step S31, the first and second grain blocks in the two-dimensional image of the sandstone and conglomerate are converted into vector images, specifically including:
[0091] S311. Perform edge detection on the two-dimensional image of sandstone and conglomerate;
[0092] S312. Extract the complete contour information of the first particle block and the second particle block using a contour tracking algorithm;
[0093] S313. Convert the complete outline information of the first particle block and the second particle block into vector graphics.
[0094] S32. Import the vector diagrams of the first particle block and the second particle block into the finite element software for modeling. During the finite element modeling process, for the first particle block, retain the original shape of the finite element plane model of the first particle block. For the second particle block, use the three-dimensional reconstructed seepage model obtained in step S2. Through data mapping and transformation, map the particle information and pore structure in the three-dimensional reconstructed seepage model to the finite element plane model of the second particle block.
[0095] S33. Couple the finite element planar model of the first particle block with the finite element planar model of the second particle block to obtain the finite element planar model of the two-dimensional image of the sandstone.
[0096] S34. Conduct seepage simulation experiments in the finite element plane model of the two-dimensional image of the sandstone and conglomerate, and calculate the predicted permeability of the original two-dimensional image of the sandstone and conglomerate.
[0097] Please refer to Figure 11 and Figure 12 , Figure 11 It shows Figure 2 The finite element planar model of sample 1 in the sample, Figure 12 It shows Figure 11 The diagram shows the visualization process of the seepage simulation experiment using the finite element planar model of sample 1. The predicted permeability of each sample calculated through the seepage simulation experiment is shown in Table 1.
[0098] Table 1 Predicted permeability of each sample
[0099]
[0100] Some embodiments of the present invention also disclose a computer-readable storage medium storing a computer program that, when executed by a processor, implements the numerical simulation calculation method for predicting permeability of sandstone and conglomerate based on two-dimensional images as described in any of the above embodiments, which will not be repeated here.
[0101] By implementing this invention, the following beneficial effects are achieved:
[0102] (1) Improving the accuracy of predicted permeability calculation: Traditional methods, which use spheres to model particles, cannot accurately account for the significant differences in particle size between large and small particles in sandstone and conglomerate, as well as the complex morphology of large particles, resulting in large errors in predicted permeability calculation. This invention, through semantic segmentation based on deep neural networks, can accurately distinguish between the first particle block (large particle block) and the second particle block (small particle block), avoiding the simplification of complex particle morphologies. In subsequent modeling and calculation processes, the influence of these factors on seepage characteristics is fully considered, making the calculated seepage path and flow rate closer to the actual situation, effectively improving the accuracy of predicted permeability calculation and providing more reliable data support for fields such as petroleum exploration and geological research.
[0103] (2) More Reasonable Model Construction: In traditional methods, spheres are used for modeling, which fails to reflect the crucial skeletal support role of large particles in sandstone and conglomerate, as well as their decisive influence on fluid flow channels. In the technical solution of this invention, the second particle block (small particle block) is divided and its particle size distribution and porosity are statistically analyzed before modeling, which can more accurately reflect the distribution characteristics of small particles. At the same time, the first particle block (large particle block) retains its original morphology, realistically restoring the shape, surface roughness, and arrangement of large particles, making the constructed finite element planar model more consistent with the actual structure of sandstone and conglomerate, greatly improving the rationality and reliability of the model.
[0104] (3) Comprehensive reflection of seepage characteristics: Since the morphological characteristics of large particles have a significant impact on the predicted permeability, the traditional spherical substitution method cannot accurately simulate the influence of these complex characteristics on the predicted permeability. This invention utilizes the coupling method of the finite element planar model of the second particle block (i.e., the finite element planar model using a three-dimensional reconstruction of the seepage model) with the finite element planar model of the first particle block to comprehensively consider the flow characteristics of fluid in particles of different sizes and complex pore structures, including the obstruction and guidance effect of large particle morphology on fluid, and the influence of the relationship between small particles and pores on seepage, thus more comprehensively reflecting the seepage characteristics of sandstone and conglomerate, and providing a powerful tool for in-depth research on the seepage mechanism of sandstone and conglomerate.
[0105] It is understood that the above embodiments only illustrate some implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can freely combine the above embodiments or technical features without departing from the concept of the present invention, and can also make several modifications and improvements, all of which fall within the protection scope of the present invention. That is, the embodiments described "in some embodiments" can be freely combined with any of the preceding and following embodiments. Therefore, all equivalent transformations and modifications made within the scope of the claims of the present invention should be covered by the claims of the present invention.
Claims
1. A numerical simulation method for predicting permeability of sandstone and conglomerate based on two-dimensional image reconstruction, characterized in that, Includes the following steps: S1. Based on a deep neural network, semantic segmentation is performed on the two-dimensional image of sandstone to divide the two-dimensional image of sandstone into a first grain block and a second grain block, wherein the grain size of the first grain block is larger than the grain size of the second grain block. S2. Divide the second particle block into several second particle block units, and model each second particle block unit until a three-dimensional reconstructed seepage model based on discrete elements is obtained for the second particle block. S3. In the process of converting the two-dimensional image of the conglomerate into a finite element planar model, for the first particle block, the original shape of the finite element planar model of the first particle block is retained. For the second particle block, the finite element planar model of the second particle block is obtained by processing according to the three-dimensional reconstructed seepage model. The finite element planar model of the first particle block and the finite element planar model of the second particle block are coupled to obtain the finite element planar model of the two-dimensional image of the conglomerate. Seepage simulation experiments are performed in the finite element planar model of the two-dimensional image of the conglomerate to calculate the predicted permeability of the original two-dimensional image of the conglomerate.
2. The numerical simulation calculation method for predicting permeability of sandstone and conglomerate based on two-dimensional image reconstruction according to claim 1, characterized in that, Step S1 includes: S11. Obtain two-dimensional images of several sandstone and conglomerate rocks to be studied; S12. Process the obtained two-dimensional image of the sandstone and conglomerate; S13. Establish a two-dimensional image semantic segmentation network, and then use an image annotation tool to manually annotate the first grain block and the second grain block in each processed two-dimensional image of the sandstone and conglomerate to obtain a training dataset. Train the two-dimensional image semantic segmentation network using the training dataset to obtain a fully trained two-dimensional image semantic segmentation network. S14. The two-dimensional image semantic segmentation network is trained to perform semantic segmentation on several two-dimensional images of the sandstone and conglomerate to be studied, so as to divide the two-dimensional images of the sandstone and conglomerate into the first grain block and the second grain block.
3. The numerical simulation calculation method for predicting permeability of sandstone and conglomerate based on two-dimensional image reconstruction according to claim 1, characterized in that, In step S2, each second particle block unit is modeled, including: The particle size distribution and porosity within each second particle block unit are statistically analyzed, and each second particle block unit is modeled based on the particle size distribution and porosity.
4. The numerical simulation calculation method for predicting permeability of sandstone and conglomerate based on two-dimensional image reconstruction according to claim 3, characterized in that, Step S2 includes: S21. Divide the second particle block to obtain a number of second particle block units; S22. In each of the second particle block units, the number, proportion and distribution range of particles of different sizes are statistically analyzed to obtain a histogram of particle size distribution in each of the second particle block units. At the same time, based on the distribution relationship between particles and pores in each of the second particle block units, the porosity in each of the second particle block units is calculated. S23. Based on the particle size distribution and porosity information obtained statistically within each second particle block unit, the discrete element method is used to model each second particle block unit. After completing the discrete element modeling of each second particle block unit, the discrete element models of all second particle block units are spliced and integrated according to the positional relationship of their original second particle blocks in the two-dimensional image of the sandstone and conglomerate, and the three-dimensional reconstructed seepage model of the second particle block based on discrete element is gradually constructed.
5. The numerical simulation calculation method for predicting permeability of sandstone and conglomerate based on two-dimensional image reconstruction according to claim 4, characterized in that, In step S22, the number, proportion, and distribution range of particles of different sizes are statistically analyzed within each of the second particle block units, including: By using an image edge detection algorithm, the outline of each particle is identified and its particle size is measured. The particle sizes of all particles in each second particle block unit are summarized to obtain information on the number, proportion, and distribution range of particles of different sizes in each second particle block unit.
6. The numerical simulation calculation method for predicting permeability of sandstone and conglomerate based on two-dimensional image reconstruction according to claim 4, characterized in that, In step S22, based on the distribution relationship between particles and pores within each second particle block unit, the porosity within each second particle block unit is calculated, including: Within each of the second particle block units, the particle portion is marked with a first color and the porous portion is marked with a second color. The porosity within the second particle block unit is then determined by calculating the proportion of the area of the second color within the entire area of the second particle block unit.
7. The numerical simulation calculation method for predicting permeability of sandstone and conglomerate based on two-dimensional image reconstruction according to claim 1, characterized in that, Step S3 includes: S31. Using image vectorization technology, the first grain block and the second grain block in the two-dimensional image of the sandstone are converted into vector images respectively; S32. Import the vector diagrams of the first particle block and the second particle block into the finite element software for modeling. During the finite element modeling process, for the first particle block, retain the original form of the finite element plane model of the first particle block. For the second particle block, use the three-dimensional reconstructed seepage model. Through data mapping and transformation, map the particle information and pore structure in the three-dimensional reconstructed seepage model to the finite element plane model of the second particle block. S33. Couple the finite element plane model of the first particle block with the finite element plane model of the second particle block to obtain the finite element plane model of the two-dimensional image of the sandstone. S34. Perform a seepage simulation experiment on the finite element plane model of the two-dimensional image of the sandstone and calculate the predicted permeability of the original two-dimensional image of the sandstone.
8. The numerical simulation calculation method for predicting permeability of sandstone and conglomerate based on two-dimensional image reconstruction according to claim 7, characterized in that, Step S31 includes: S311. Perform edge detection on the two-dimensional image of the sandstone and conglomerate; S312. Extract the complete contour information of the first particle block and the second particle block using a contour tracking algorithm; S313. Convert the complete outline information of the first particle block and the second particle block into vector graphics.
9. The numerical simulation calculation method for predicting permeability of sandstone and conglomerate based on two-dimensional image reconstruction according to any one of claims 1-8, characterized in that, The first particle block is the block containing particles with a particle size ≥ 2 mm, and the second particle block is the block containing particles with a particle size < 2 mm.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the numerical simulation calculation method for predicting permeability of sandstone and conglomerate based on two-dimensional images as described in any one of claims 1-9.